Summary
Genetic backgrounds and risk factors among individuals with Parkinson’s disease (PD) are highly heterogenous, limiting our ability to effectively detect and treat PD. Here we connect several potential PD risk genes and elements to one biological pathway. Elevation of Fe2+-levels causes Ca2+-overflow into the mitochondria, through an interaction of Fe2+ with mitochondrial calcium uniporter (MCU), the Ca2+-import channel in the inner mitochondrial membrane, and resultant MCU oligomerization. This mechanism acts in PD neuron models and postmortem brains. Miro, a Ca2+-binding protein, functions downstream of Ca2+-dysregulation, and holds promise to classify PD status and monitor drug efficacy in human blood cells. Polygenetic enrichment of rare, non-synonymous variants in this iron-calcium-Miro axis influences PD risk. This axis can be targeted by multiple ways to prevent neurodegeneration in PD models. Our results show a linear pathway linking several PD risk factors, which can be leveraged for genetic counseling, risk evaluation, and therapeutic strategies.
Main Text
Introduction
Parkinson’s disease (PD) is a leading cause of disability, afflicting the aging population. The dopamine (DA)-producing neurons in the substantia nigra are the first to die in PD patients. A bottleneck that hinders our ability to effectively detect and treat PD may be the presence of highly heterogenous genetic backgrounds and risk factors among different patients. More than 90% of the PD cases are considered sporadic with no known causal mutations. Genome-wide association studies (GWAS) have identified over 90 risk loci (Diaz-Ortiz et al., 2022). Functional studies on known causal genes of familial patients and from cellular and animal PD models have pointed to multiple “cellular risk elements”, such as mitochondrial damage, lysosomal dysfunction, immune system activation, neuronal calcium mishandling, and iron accumulation (Angelova et al., 2020; Apicco et al., 2021; Belaidi and Bush, 2016; Buttner et al., 2013; Kim et al., 2020; Lee et al., 2018; Surmeier et al., 2017; Tabata et al., 2018; Verma et al., 2017; Vuuren et al., 2020). These distinct genetic and cellular risk factors may confer individual heterogeneity in disease onset, but also suggest that there are networks and pathways linking these “hubs” in disease pathogenesis. Identifying their connections could be crucial for finding a cure for PD.
Mitochondria are the center of cellular metabolism and communication. Ions such as calcium and iron, are not only essential for diverse mitochondrial functions but can be stored inside the mitochondria to maintain cellular ionic homeostasis. Ion channels in the plasma and mitochondrial membranes coordinate for ion uptake, transport, and storage. For example, calcium ions enter the cell via voltage- or ligand-gated calcium channels across the cell surface. Inside the cell, they are taken up by mitochondria through the outer mitochondrial membrane (OMM) channel, VDAC, and the inner mitochondrial membrane (IMM) channel, mitochondrial calcium uniporter (MCU) (Baughman et al., 2011), and extruded into the cytosol through the IMM transporter, NCLX (Palty et al., 2010). MCU is a multimeric holocomplex consisting of additional regulatory subunits, such as essential MCU regulator (EMRE), mitochondrial calcium uptake 1 (MICU1), MICU2, and MCUb (Fan et al., 2018; Fan et al., 2020; Lambert et al., 2019). Channels complementary to these major mitochondrial calcium channels also exist (Patron et al., 2022). It remains a mystery regarding the relation of calcium and iron ions in PD mechanisms and their contribution to disease susceptibility.
Identifying the cellular causes to neuron death will not only provide more effective disease management but also shed light on molecular signatures shared by a subset of people affected by the disease. A convenient, cost-effective method to spot the vulnerable population, even before the symptom onset, will be extremely valuable for early intervention and preventive medicine. It will improve the efficacy of clinical trials for testing experimental drugs, by facilitating patient stratification and serving as a pharmacodynamic marker.
Understanding disease-causing cellular paths will also help us zoom in on rare genetic variants that contribute to disease etiology but otherwise are difficult to discover through GWAS. Integrating studies of risk variants with disease models and human tissues could establish a causal link of a biological pathway to a disease with complex traits such as PD, with the promise of identifying more effective druggable targets and biomarkers. In this work, we harness the power of combining human genetics, cellular and in vivo models, and patient’s tissues, and identify an iron-calcium-Miro axis in PD. Iron accumulation causes mitochondrial calcium overload via promoting MCU oligomerization and its channel activity, which may consequently disturb cellular calcium homeostasis. Miro, a calcium-binding protein, acts downstream of calcium dysregulation in PD models. Functional and genetic impairments in this axis may increase PD risk and indicate PD status.
Results
A high-content Miro1 screening assay identifies a network of Ca2+-related drug hits for PD
Miro is an OMM protein essential for mediating mitochondrial motility and safeguarding their quality. Human Miro1 and Miro2 are paralogs with high sequence similarity. We have previously shown that in fibroblasts or neurons derived from sporadic and familial PD patients, Miro1 degradation upon mitochondrial depolarization is delayed, consequently slowing mitophagy and increasing neuronal sensitivity to stressors (Hsieh et al., 2019; Hsieh et al., 2016; Shaltouki et al., 2018). This Miro1 phenotype would serve as an excellent readout for screening small molecules that promote Miro1 degradation following depolarization. Many compounds in the commonly used screening libraries have well-defined roles and targets, and some show efficacy to treat certain human diseases. This rich information may allow us to reveal cellular pathways underlying the Miro1 phenotype in PD. To this end, we established a sensitive immunocytochemistry (ICC)-based assay that was suitable for high-throughput screening (Figures 1A, S1, S2, S3A, more details in Method). We performed the primary screens at the Stanford High-Throughput Bioscience Center (HTBC) using 3 drug libraries in a sporadic PD fibroblast line. Overall, we identified 35 actives (1.92% primary hit rate) that reduced Miro1 following mitochondrial depolarization (Figure S1, Table S1A-B). To validate the results of the high-content assays, we retested 34 out of the 35 positive Miro1 reducers identified at the Stanford HTBC in our own laboratory using fresh compounds and our confocal microscope. We confirmed that 15 compounds reliably reduced Miro1 protein levels following mitochondrial depolarization in PD fibroblasts (Figure S2, Table S1C). Next, we performed a pathway analysis using a knowledge graph-based tool to reveal the potential cellular pathways connecting Miro1 to each hit compound. Strikingly, we discovered intracellular Ca2+ as a primary shared factor in the hit drug-Miro1 network (Figure 1B, Table S2). Two drugs, Benidipine and Tranilast, could directly inhibit plasma membrane Ca2+-channels. Benidipine is a blocker of voltage-gated Ca2+-channels (L-, N-, T-type), and Tranilast has been proposed to inhibit ligand-gated Ca2+-channels (TRPV2) (Darakhshan and Pour, 2015). Dysregulation of Ca2+ homeostasis has been widely reported in PD models (Angelova et al., 2020; Apicco et al., 2021; Buttner et al., 2013; Kim et al., 2020; Lee et al., 2018; Surmeier et al., 2017; Tabata et al., 2018; Verma et al., 2017). Our results suggest a direct link of the Ca2+-binding protein, Miro, to Ca2+-mediated abnormality in PD.
Validation of a role for Benidipine in Miro degradation
To understand in detail the relation of Miro with the Ca2+-pathway in Parkinson’s pathogenesis, we further examined Benidipine’s role in Miro protein stability. Using the same ICC method as in Figure S2, we found that Benidipine reduced Miro1 in a dose-dependent manner in PD fibroblasts treated with carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) (Figure S3B). To exclude the possibility of any artifacts caused by our ICC method, we verified our results using an entirely different approach to detect Miro1 response to depolarization. We measured Miro1 and additional mitochondrial proteins by Western blotting. We also depolarized mitochondria with a different uncoupler, carbonyl cyanide m-chlorophenyl hydrazone (CCCP) (Hsieh et al., 2019), instead of FCCP. We detected Miro1 and mitochondrial markers at 6 and 14 hours after CCCP treatment. We have previously demonstrated that in healthy control fibroblasts following CCCP treatment, Miro1 is degraded earlier (6 hours) than multiple other mitochondrial markers (14 hours) (Figure S3C) (Hsieh et al., 2019; Hsieh et al., 2016), consistent with the observation of proteasomal degradation of Miro1 prior to mitophagy (Chan et al., 2011; Hsieh et al., 2019; Hsieh et al., 2016; Wang et al., 2011). Using this alternative method, we confirmed that both Miro1 degradation and damaged mitochondrial clearance were impaired in the PD cell line we used for screens. Importantly, Benidipine promoted Miro1 degradation after 6 hours following CCCP treatment without affecting the matrix protein ATP5β and facilitated mitochondrial clearance as was evidenced by the degradation of both Miro1 and ATP5β at 14 hours post-treatment (Figure S3C). Interestingly, Miro2 was also resistant to depolarization-induced degradation in PD cells (Hsieh et al., 2019) and Benidipine rescued its phenotype (Figure S3C). This result suggests that Miro1 and Miro2, which share the Ca2+-binding EF-hands, are functionally redundant in the Ca2+-dependent regulation. We confirmed that Benidipine did not affect Miro1 messenger RNA (mRNA) expression detected by reverse transcription quantitative real-time PCR (RT-qPCR) under basal and depolarized conditions in PD cells (Figure S3D). Neither did Benidipine alter the basal ATP levels (Figure S3E), nor the mitochondrial membrane potential measured by TMRM staining (Figure S3F). Collectively, we have demonstrated that Benidipine, a Ca2+-channel blocker, specifically promotes Miro degradation upon depolarization in PD fibroblasts using multiple methods.
Benidipine rescues Parkinson’s phenotypes in human neuron and fly models of PD
We have previously shown that reducing Miro rescues Parkinson’s phenotypes in cellular and in vivo models (Hsieh et al., 2019; Hsieh et al., 2016; Li et al., 2021; Shaltouki et al., 2018), suggesting that small molecules that lower Miro protein levels could represent an effective therapeutic approach for PD. From our high-throughput screens we have discovered that Benidipine, which targets the Ca2+ pathway, promotes Miro degradation upon depolarization in skin cells of a PD patient (Figures 1, S1-3). We next tested whether Benidipine was useful for alleviating Parkinson’s phenotypes in two independent models: the human neuron and fly models. We examined Benidipine using induced pluripotent stem cells (iPSCs) from one familial patient with the A53T mutation in SNCA (encodes α-synuclein) and its isogenic wild-type control (Hsieh et al., 2019; Hsieh et al., 2016; Li et al., 2021; Shaltouki et al., 2018). We differentiated iPSCs to neurons expressing tyrosine hydroxylase (TH), the rate-limiting enzyme for DA synthesis as previously described (Figure S4A) (Hsieh et al., 2019; Hsieh et al., 2016; Li et al., 2021; Shaltouki et al., 2018). These patient-derived neurons display increased expression of endogenous α-synuclein (Shaltouki et al., 2018). We identified DA neurons by TH-immunostaining and cell death by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)-staining (Li et al., 2021; Shaltouki et al., 2018) (Figures 2A-B, S4A). We have previously shown that iPSC-derived DA neurons from PD patients are more vulnerable to stressors than those from healthy controls (Hsieh et al., 2019; Hsieh et al., 2016; Li et al., 2021; Shaltouki et al., 2018). The treatment of the complex III inhibitor, Antimycin A, at 10 μM for 6 hours caused acute neuronal cell death leading to the loss of TH and the increase of TUNEL signals in neurons derived from the PD patient (Hsieh et al., 2019; Li et al., 2021; Shaltouki et al., 2018) (Figure 2A-B). Notably, Benidipine treatment at 10 μM for 30 hours significantly rescued this stressor-induced neuron death (Figure 2A-B).
In order to cross-validate the neuroprotective effect of Benidipine in vivo, we fed Benidipine to a fly model of PD, which expressed the pathogenic human α-synuclein protein with the A53T mutation (α-syn-A53T) in DA neurons driven by TH-GAL4 (Hsieh et al., 2019; Li et al., 2021; Shaltouki et al., 2018). These flies exhibit PD-relevant phenotypes such as age-dependent locomotor decline and DA neuron loss (Hsieh et al., 2019; Li et al., 2021; Shaltouki et al., 2018). Importantly, feeding PD flies with 2.5 μM Benidipine from adulthood prevented DA neuron loss in aged flies (Figure 2C) and improved their locomotor ability (Figure 2D). Taken together, Benidipine, which eliminates the Miro1 defect in PD fibroblasts, rescues PD-related phenotypes in human neuron and fly models.
The EF-hands of Miro play a role in causing Parkinson’s-relevant phenotypes
Having demonstrated a link between Miro and Ca2+ in multiple models of PD, we next determined whether Ca2+ directly interacted with Miro to contribute to phenotypes in these models. To achieve this goal, we made GFP-tagged human Miro1 protein in both the wild-type (Miro1-WT) form and in a mutant form where two point mutations were introduced in the two EF-hands of Miro1 (Miro1-KK) to block Ca2+-binding (Wang and Schwarz, 2009). We expressed GFP-tagged Miro1 (WT or KK) and Mito-dsRed that labeled mitochondria in iPSC-derived neurons from the PD patient and isogenic control, described earlier. We chose the distal segment of the axon for analysis (Hsieh et al., 2016). We have previously shown that following 100 μM Antimycin A treatment that triggers mitophagy, Miro1 and mitochondria are sequentially degraded in wild-type neurons (Hsieh et al., 2019; Hsieh et al., 2016; Shaltouki et al., 2018). We observed the same mitochondrial events in isogenic control axons transfected with GFP-Miro1-WT.
Within 25 min GFP-Miro1-WT was partially degraded, and within 58 min mitochondrial clearance was induced (Figure 3A-C). In contrast, in PD neuron axons transfected with GFP-Miro1-WT, the degradation rates of both Miro1 and damaged mitochondria upon Antimycin A treatment were slowed (Figure 3A-C), consistent with our previous studies (Hsieh et al., 2019; Shaltouki et al., 2018). Notably, GFP-Miro1-KK significantly rescued these phenotypes in PD axons: it expedited the degradation rates to the control level (Figure 3A-C). These data suggest that Miro1 directly binds to Ca2+ to mediate mitochondrial phenotypes in PD neurons, at least in part.
To confirm the Miro-Ca2+ relation in vivo, we generated transgenic flies carrying T7-tagged fly Miro (DMiro)-WT or DMiro-KK. DMiro is an ortholog of human Miro1 and Miro2. Both DMiro-WT and DMiro-KK were expressed at comparable levels when the transgenes were driven by the ubiquitous driver Actin-GAL4 (Figure S4B). We next crossed these transgenic flies to a fly PD model described earlier that expressed human α-syn-A53T in DA neurons driven by TH-GAL4. Consistent with the results from human neurons, DMiro-KK significantly rescued the PD-relevant phenotypes including the age-dependent DA neuron loss and locomotor decline, as compared to DMiro-WT (Figure 3D-E). Altogether, we have provided evidence showing that the Ca2+-binding domain of Miro plays a key role in causing phenotypes in human neuron and fly models of PD.
The Ca2+-uptake ability of mitochondria is enhanced in PD neurons
Our results, showing that either blocking Ca2+-entry into the cell or inhibiting Ca2+-binding to Miro rescues the Miro and neurodegenerative phenotypes in PD models (Figures 1–3), placed Ca2+ dysregulation upstream of Miro. We next dissected how Ca2+-handling was mis-regulated in neurons derived from the PD patient (Figures 2–3). We stimulated these neurons with the G-protein-coupled receptor (GPCR) agonist, thrombin, and measured cytosolic and mitochondrial Ca2+ levels with live Calcium Green and Rhod-2 staining, respectively. We found that thrombin triggered intracellular Ca2+ mobilization and elevation, which was comparable between PD and isogenic control neurons (Figure 4). However, mitochondria in PD neurons sustained significantly larger Ca2+-elevation after thrombin stimulation, as compared to control (Figure 4). These results indicate that the mitochondrial Ca2+-import ability is boosted in PD neurons.
Iron promotes the assembly of MCU oligomers
The enhanced mitochondrial Ca2+-uptake observed in PD neurons (Figure 4) led us to investigate the Ca2+-uptake channel in the IMM–MCU. The oligomerization of MCU is essential for MCU’s function to import Ca2+ into the mitochondria (Dong et al., 2017; Fan et al., 2018; Fan et al., 2020). We hypothesized that MCU oligomerization could be affected by small molecules in the mitochondrial microenvironment, particularly those with a role in PD, including Ca2+ (Figure 4), reactive oxygen species (ROS), and iron (Belaidi and Bush, 2016). To explore this possibility, we performed size exclusion chromatography (SEC), where protein complexes with higher molecular weight (MW) are eluted faster than those with lower MW, using lysates of HEK cells treated with Fe2+, Ca2+, or H2O2. Detecting MCU from cell lysates using SEC has been successfully shown (Dong et al., 2017; Lambert et al., 2019; Tomar et al., 2016). We found that Fe2+, Ca2+, and H2O2 treatment all shifted the MCU elution peaks to the earlier fractions of higher-order oligomers compared to control (Figure 5A, S4C-G; anti-MCU was validated in Figure S4C). In contrast, the elution pattern of Miro1 was largely unaltered by any of these treatments (Figure S4D-G). These results show that intracellular small molecules including iron, Ca2+ (Fan et al., 2020), and H2O2 (Dong et al., 2017) can influence the equilibrium of MCU oligomeric complexes and may consequently alter the MCU channel activity.
SEC is a sensitive method to detect protein composition dynamics, while Native-PAGE can determine the overall form and amount of a multimeric native protein. The human MCU oligomer bands from HEK cells migrated between 400-700 KDa in Native-PAGE (Figure 5B; the negative control in the left lane, MCU−/−, showed no signal) (Baughman et al., 2011; Ghosh et al., 2020; Tomar et al., 2016). Importantly, we found that Fe2+ treatment, but not Ca2+ or H2O2, resulted in an increase in the total intensity of the MCU oligomer bands (Figure 5B). Miro1 protein also oligomerized and migrated as a single band around 480 KDa in Native-PAGE, which was not significantly affected by any of these treatments (Figure 5B). These data indicate that iron not only shifts the MCU complexes to higher-order oligomers but also enlarges the total number of these complexes, and thus may have a more profound impact on the MCU activity than Ca2+ and H2O2.
We next examined additional membrane proteins that may assist the MCU function using HEK cells. By detecting total protein levels using Western blotting, we found when Fe2+ was added in media and lysis buffer, MCUb and NCLX were lowered, but not any other proteins examined (Figure S5). MCUb is an inhibitor of MCU (Lambert et al., 2019), and NCLX is an IMM exchanger believed for mitochondrial Ca2-extrusion (Lee et al., 2016; Palty et al., 2010). The reduction of both proteins could exacerbate the phenotype of mitochondrial Ca2+-overload. By evaluating oligomeric dynamic changes using SEC, we found that the OMM channel, VDAC, and MCU’s gating regulator, MICU1, showed a matching elution pattern with MCU, and the treatment of Fe2+ shifted all 3 proteins to the similar earlier fractions of higher order oligomers, which coincided with the elution peak of MIC60, a core structural protein at the crista junctions and contact sites (Zerbes et al., 2012), and IP3R1, the major ER Ca2+-channel that delivers Ca2+ to the OMM (Katona et al., 2022) (Figure 5A, C). These data are consistent with VDAC, MICU1, and MCU being associated in the same super-complexes and suggest a possible spatial reorganization of the MCU super-complexes upon iron elevation, thus allowing easier access to ER Ca2+ supply. Taken together, our results show that iron promotes MCU oligomerization and may cause Ca2+-accumulation inside the mitochondria.
Fe2+ binds to the MCU complex and acts on its Ca2+-import ability
We probed the mechanism underlying the observed impact of iron on MCU oligomerization (Figure 5A-B). One hypothesis was that the MCU complex bound to iron leading to conformational changes. To explore this possibility, we immunoprecipitated (IP) endogenous MCU from HEK cells and detected the iron concentrations in the IP samples. We found significantly more Fe2+ ions pulled down with MCU when HEK cells were treated with Fe2+, compared with other controls (Figure 6A). This result shows that the MCU complex interacts with Fe2+. To determine whether MCU directly bound to Fe2+, we switched our experiments to an in vitro setting. To circumvent the problem of precipitation caused by a fast speed of Fe2+ oxidation in vitro, we used an Fe2+ mimic, Co2+ ion (Billesbolle et al., 2020). We first confirmed that Co2+ behaved similarly as Fe2+ in our functional assays in HEK cells: Co2+ treatment increased MCU oligomerization detected by Native-PAGE (Figure 6B), just like Fe2+ (Figures 5B, 6B), and both Fe2+ and Co2+ treatment enhanced the mitochondrial Ca2+-uptake ability following thrombin application (Figure 6C). Using fluorescence-detection SEC on purified human MCU protein (Fan et al., 2020), we found that Co2+ caused the formation of higher-order oligomers of MCU, and decreased MCU protein stability with increased temperature (Figure 6D-E). These results demonstrate that the Fe2+ mimic, Co2+, directly binds to MCU, increasing its oligomerization.
We next searched for amino acid residues in the matrix domain of MCU (PDB: 5KUE) predicted to bind to Fe2+ using an in-silico program (http://bioinfo.cmu.edu.tw/MIB/) (Lin et al., 2016; Lu et al., 2012), and found 3 amino acids: 74D, 148D, and 159H. The latter 2 residues were also predicted to bind to Co2+. We mutated these 3 sites to Alanine (named “MCU-3A”). Indeed, we detected significantly less Fe2+ bound to Flag-tagged MCU-3A, as compared to MCU-WT, produced from HEK cells without endogenous MCU (MCU−/−) (Figure 6F). To determine whether these mutations were sufficient to eliminate the Fe2+-triggered oligomerization of MCU, we expressed MCU-WT or MCU-3A in MCU−/− HEK cells, treated these cells with Fe2+, and ran the IPed proteins in Native-PAGE. As expected, MCU-3A abolished MCU’s response to Fe2+ treatment: the MCU oligomer band intensity was no longer increased (Figure 6G). We then live imaged mitochondrial Ca2+-dynamics, as described in Figure 4, in these cells. We consistently observed a larger mitochondrial Ca2+-elevation following thrombin stimulation in MCU-WT-transfected HEK cells treated with Fe2+ as compared to no Fe2+-treatment, and MCU-3A blunted the peak increase (Figure 6H-I). Because Fe2+ also lowered NCLX levels (Figure S5) which could affect mitochondrial Ca2+-extrusion, we measured the mitochondrial Ca2+-efflux rate. Indeed, Fe2+ treatment slowed the efflux rate, which was prevented by MCU-3A (Figure 6J), suggesting that the Fe2+-triggered efflux delay might depend on Ca2+-overload. Altogether, our results show that Fe2+ binds to the MCU complex, promoting MCU oligomerization and its channel activity.
Iron functions upstream of calcium to mediate phenotypes of PD neurons
Our discovery of the action of Fe2+ on the MCU activity (Figures 5–6) suggested that in PD neurons, the phenotype of mitochondrial Ca2+-overload (Figure 4) might depend on iron. To confirm their causal relation, we treated PD neurons with deferiprone (DFP), an iron chelator (Munson et al., 2021). Indeed, DFP significantly reduced mitochondrial Ca2+-accumulation following thrombin stimulation (Figure 7A) and prevented cell death triggered by Antimycin A treatment (Figure 7B) in iPSC-derived neurons from the familial PD patient described earlier. We treated neurons from a second, sporadic patient with DFP and saw a similar neuroprotective effect (Figure 7B). In vivo, feeding the fly model of PD as shown earlier (Figures 2–3) with DFP consistently rescued the PD-relevant phenotypes, including age-dependent DA neuron loss and locomotor decline (Figure 7C-D). Collectively, our results show that iron functions upstream of calcium to mediate neurodegeneration in PD models.
MCU, MCUb, and NCLX are affected in PD postmortem brain
Our finding showing that iron impacts mitochondrial Ca2+-channels and transporters (Figure 5, 6B, S5) prompted us to examine these proteins in postmortem brains of people with PD, diffuse Lewy body disease (DLBD), or Alzheimer’s disease (AD) (Table S3). We homogenized the frontal cortex and ran the brain lysate in Native- or SDS-PAGE. We focused on MCU oligomers, NCLX, and MCUb, which were shown earlier specifically altered by Fe2+ treatment in HEK cells (Figure 5B, 6B, S5). We found the PD group clustered and separated from the healthy control group, with higher intensity of the MCU oligomer bands and lower intensity of both the NCLX and MCUb bands (Figure 7E), similar to the observations in HEK cells treated with Fe2+ (Figure 5B, 6B, S5). This unique clustering was not observed in the AD or DLBD group (Figure 7E, S6). Together, our data suggest that the combined functional impairment of MCU, MCUb, and NCLX may be one of the molecular signatures shared by people with PD.
Miro1 in blood cells reflects PD status and responds to drug treatment
Now we have demonstrated a functional axis of iron-calcium-Miro in neurons of our PD models and patients. We next sought evidence of its impairment in peripheral tissues of PD patients, which could serve as an excellent candidate for biomarker and pharmacodynamic marker development. Measuring the combined impairment of MCU oligomerization, MCUb, and NCLX requires a large amount of protein and Native-PAGE, which is not applicable for high-throughput screening and clinical practice. We then explored Miro1, whose slower degradation following mitochondrial depolarization was downstream of Ca2+-dysregulation in our PD models, as shown earlier (Figures 1–3). Notably, we have previously found that in skin fibroblasts and iPSCs the delay of Miro1 degradation upon depolarization distinguishes PD patients and genetic carriers from healthy controls (Hsieh et al., 2019; Hsieh et al., 2016; Nguyen et al., 2021). Although fibroblasts can be readily obtained by a skin biopsy, a blood test remains the most convenient method. We investigated whether Miro1 could be detected in peripheral blood mononuclear cells (PBMCs) acquired from a blood draw. We cultured PBMCs from a healthy donor from the Stanford Blood Center (SBC, Table S4) and depolarized the mitochondrial membrane potential using two different methods: Antimycin A plus Oligomycin (Ordureau et al., 2020), or CCCP. We found that both depolarizing approaches caused the degradation of Miro1 and additional mitochondrial markers in a time-dependent manner, detected by Western blotting (Figure S7A-B), consistent with other cell types (Hsieh et al., 2019; Hsieh et al., 2016; Nguyen et al., 2021).
To enable high-content screening, we applied an enzyme-linked immunosorbent assay (ELISA) of Miro1 (Figure S7C-D) to PBMCs from the same donor with 6-hour CCCP treatment. We saw a similar Miro1 response to CCCP using ELISA (SBC, Table S4). We then used this ELISA to screen a total of 80 healthy controls and 107 PD patients (Table S4). Miro1 Ratio (Miro1 protein value with CCCP divided by that with DMSO from the same person) was significantly higher in PD patients compared to healthy controls (Figure 7F, Table S4), indicating that Miro1 is more stable upon depolarization in PD patients.
To determine whether our method could be used to classify an individual into a PD or healthy group, we employed machine learning approaches using our dataset. We trained a logistic regression model to assess the impact of Miro1 Ratio on PD diagnosis, solely on its own or combined with additional demographic and clinical parameters (Method). Unified Parkinson’s Disease Rating Scale (UPDRS) is a tool to measure motor and non-motor symptoms of PD which may reflect disease severity and progression. Using UPDRS, our model yielded an accuracy (an individual was correctly classified as with PD or healthy) of 81.2% (p<0.000001; area under the Receiver Operator Curve (ROC)–AUC=0.822), and using Miro1 Ratio, the accuracy was 67.6% (p=0.03; AUC=0.677). Notably, if both Miro1 Ratio and UPDRS were considered, our model generated an improved accuracy of 87.8% (p=0.02; AUC=0.878), without the interference of age or sex (Method, Figure S7E-F). Therefore, our results suggest that the molecular (Miro1 Ratio) and symptomatic (UPDRS) evaluations may reveal independent information, and that combining both tests may more accurately categorize individuals with PD and measure their responses to experimental therapies.
To probe the potential utilization of this Miro1 assay in future clinical trials for stratifying patients or monitoring drug efficacy, we treated PBMCs from 4 PD patients (Table S4) with either of the two compounds known to reduce Miro1, Benidipine (Figures 1–2, S2) and Miro1 Reducer 3 (MR3) (Hsieh et al., 2019; Li et al., 2021). Miro1 protein levels upon CCCP treatment were lowered by each compound in all 4 patients (Figure 7G), showing that the Miro1 marker in PBMCs can respond to drug treatment. Collectively, our results suggest that Miro1 protein in blood cells may be used to aid in diagnosis and drug development.
Rare variants in the iron-calcium-Miro pathway are associated with PD status
After dissecting the functional impairment of this iron-calcium-Miro axis in PD, we explored its genetic contribution to PD. Earlier, we showed that chelating iron, blocking Miro’s binding to Ca2+, or preventing Ca2+-entry into the cell all alleviated parkinsonian neurodegeneration (Figures 2, 3, 7). We evaluated the genes encoding the protein targets of these approaches, which are spatially distinct and localized to three subcellular locations: (1) IMM Ca2+-channels and transporters (targeted by Fe2+), (2) the Ca2+-binding protein Miro on the OMM, and (3) plasma membrane Ca2+-channels (targeted by Benidipine and Tranilast) (Figure 7H, Table S5). By analyzing common variants within or near any of the investigated genes in GWAS reported in (Nalls et al., 2019), we did not observe significant association with PD clinical status.
We next employed the whole-genome sequencing (WGS) data from the Accelerating Medicines Partnership–Parkinson’s Disease (AMP–PD) (1,168 control; 2,241 PD), and assessed rare non-synonymous and damaging variants using burden based and SKATO methods. We discovered polygenetic vulnerability to PD at all three spatial hotspots: significant association with PD status of rare variants in selective T-type Ca2+-channel subtypes (Cav3.2, 3.3) (cell surface), Miro2 (OMM), and NCLX (IMM) (Table S5). A SKATO Test on all variants of T-type or L-type Ca2+-channel subtypes consistently showed significant association with PD status of T-type channels, which survived multiple comparison correction, but not of L-type channels (Table S5). Together, our analysis unravels genetic predisposition of this Ca2+-pathway to PD.
To functionally validate the selection of T-type Ca2+-channels from our human genetic study, we employed the same screening ICC assay described earlier (Figures S1-2) by which we discovered the non-selective pan-Ca2+-channel blocker, Benidipine, but now using 2 different specific L-type and 3 different T-type Ca2+-channel blockers. Intriguingly, we again discovered a striking selection of T-type versus L-type channels, in the connection with Miro1 in PD fibroblasts (Figure S7G): only T-type blockers promoted Miro1 degradation following depolarization, just like Benidipine, supporting the human genetic finding. Similar to Miro1, Miro2 was also resistant to depolarization-triggered degradation in PD fibroblasts (Hsieh et al., 2019) and responded to Benidipine treatment (Figure S3C). NCLX functionally interacts with MCU (Lee et al., 2016), and was coregulated with MCUb and MCU by Fe2+ (Figure 5B, 6B, S5) and PD (Figure 7E). Although it remains elusive why rare variants in these genes are selectively associated with PD status, our combined, unbiased analyses of human genetics, tissues, cell, and in vivo models corroborate the complexity and multifactorial nature of PD etiology, and indicate that the polygenetic architecture built around this iron-calcium-Miro axis might influence an individual’s risk to develop PD.
Discussion
In this work, we have established a pathway of iron-calcium-Miro dysregulation in our PD models (Figure 7H). Elevation of iron concentrations may cause mitochondrial Ca2+-overload by promoting the MCU activity and reducing NCLX levels (Figures 4–7, S5). It is possible that the initial mitochondrial Ca2+-accumulation subsequently disrupts Ca2+-homeostasis at the cellular level, eliciting responses of additional Ca2+-binding proteins, such as Miro (Figure 7H). We have shown multiple ways to target this axis. Chelating iron, reducing Ca2+-entry into the cell, or blocking Miro’s binding to Ca2+ is each neuroprotective (Figures 2, 3, 7). Of note, the impairment of this axis can be reflected in blood cells using a Miro1 assay with high content capacity (Figure 7F-G) and can be detected in the genome of PD patients (Table S5). Hence, this ionic axis may be important for PD pathogenesis and can be leveraged for better detecting and treating the disease.
Although the precise course of PD pathogenesis remains unclear, emerging evidence has demonstrated its complex and polygenetic nature. Interactions among multiple organs and cell types, systemic immune activation, and environmental triggers act uniquely in different individuals. Although GWAS and segregation studies have unveiled many PD risk or causal loci, rare genetic variants may be particularly important for conferring individual heterogeneity in disease onset and etiology. However, rare variants associated with PD risk are difficult to discover given the limitations in sample collection and the unmet need for exome-sequencing as opposed to sparse genotyping on microarrays. In our work, we have located several potential risk genes by first connecting them in the same biological pathway. In an individual prone to PD, there might be already mitochondrial malfunction causing mitochondrial proteins to release labile iron, or impairments in systemic iron uptake and circulation causing iron accumulation at the cellular level. Any predisposed genetic perturbations in this iron-calcium-Miro axis, which could be due to different rare variants in each person, would further exacerbate its dysfunction. With or without another trigger such as a virus infection or environmental insult, the tipping point for disease onset might be reached. Thus, combining complementary genetic and functional studies may help us better understand the destructive paths leading to the disease and identify network hubs for therapeutic targeting.
Screening people for genetic variants in this iron-calcium-Miro axis and detecting Miro in blood cells may help stratify a unique population of patients and at-risk individuals, who will particularly benefit from therapeutic interventions targeting this axis. For example, treating people bearing genetic variants in Ca2+-channels with Ca2+-channel blockers, iron-chelators, or both, may yield the best efficacy. Similarly, people who test positive for the Miro1 phenotype in PBMCs may respond best to drugs reducing Miro1. Integrating genetic screening of this axis with Miro1 detection in peripheral tissues may enhance the accuracy of risk evaluation and help design personalized treatment, such as a cocktail of different Ca2+-channel blockers, iron chelators, and Miro reducers, to improve prevention and treatment efficacy.
Calcium mishandling and iron accumulation have been widely observed in PD neurons (Angelova et al., 2020; Apicco et al., 2021; Belaidi and Bush, 2016; Buttner et al., 2013; Kim et al., 2020; Lee et al., 2018; Surmeier et al., 2017; Tabata et al., 2018; Verma et al., 2017; Vuuren et al., 2020). Now we have provided a mechanistic link. Fe2+ elicits mitochondrial Ca2+-overload through acting on IMM Ca2+-channels and transporters. Further investigations are needed to dissect how Fe2+ regulates MCUb and NCLX levels. One hypothesis is that these proteins are targeted by Ca2+-activated mitochondrial proteases. Our results (Figure 6J) have suggested that Fe2+-triggered Ca2+-efflux delay depends on mitochondrial Ca2+-overload. A recent study has shown that NCLX protein levels are lowered in another mitochondrial iron-dependent disease, Friedreich’s Ataxia (FA), through a possible mechanism dependent on calpain cleavage (Britti et al., 2021). Intriguingly, iron chelators have displayed promising therapeutic benefits in preclinical models of both PD (Clark et al., 2020) and FA (Llorens et al., 2019), and is currently in multiple clinical trials for treating symptoms of PD patients (Clark et al., 2020). Thus, the underlying molecular mechanisms may be shared by multiple diseases with mitochondrial iron accumulation.
The consequences to intracellular Ca2+-dysfunction could be profound and detrimental. Dissecting these downstream details in PD will help us find more powerful targets and biomarkers. Mitochondrial Ca2+-overload could overwhelm the oxidative phosphorylation system (Ashrafi et al., 2020), damaging the electron transport chain. Malfunctions of MCU and NCLX may disrupt spatially discrete Ca2+ transients and oscillations adjacent to ER membranes, leading to defects in autophagy (Zheng et al., 2022), which may consequently cause protein aggregation and lysosomal dysfunction. Intriguingly, chelating iron can induce selective autophagy pathways (Wilhelm et al., 2022), raising the question whether MCU and local Ca2+ dynamics are involved in iron-mediated autophagy. Ca2+ may be also important for safeguarding mitochondrial quality. Miro appears to be a molecular switch between distinct routes of mitochondrial quality control. Proteasomal degradation of Miro dissociates mitochondria from microtubules and allows the entire damaged mitochondria to enter the mitophagy pathway (Chan et al., 2011; Hsieh et al., 2016; Wang et al., 2011), whereas sustaining Miro on the microtubule motors is required for the biogenesis of mitochondrial derived vesicles (MDVs) that deliver a subset of mitochondrial proteins (including Miro) to lysosomes (Konig et al., 2021). Notably, MDVs occur under mild oxidative stress or at steady state, and before mitophagy which is triggered by extended damage (Konig et al., 2021; Lin et al., 2017; McLelland et al., 2014). Miro may also ride with those damaged mitochondria expelled from neurons through nanotunnels and vesicles (Ahmad et al., 2014; Davis et al., 2014; Melentijevic et al., 2017; Rosina et al., 2022; Saha et al., 2022). Perhaps it starts with only one of these biological processes impaired by Ca2+ signals, leading to Miro retention on damaged mitochondria, which further affects more Miro-dependent quality control pathways. More studies are needed to unravel the precise roles of Miro and other Ca2+-binding proteins in PD pathogenesis and how Ca2+ regulates these roles. The physical and functional interactions of these Ca2+-binding proteins with specific Ca2+-channels in plasma, mitochondrial, or additional organellar membranes, may constitute the core underlying mechanisms.
Author contributions
R.V. did drug screens. S.E.R. and R.B.A. performed pathway analysis. V.B., A.S.D., C.-H.H., L.L., and S.C. performed human cell experiments. L.L. conducted fly work. C.M.M. and L.F. did in vitro work. Y.LG. and M.D.G. analyzed human genetic data. P.N. analyzed PBMC data. X.W. conceived and supervised the project. All authors designed the experiments and wrote the paper.
Competing interests
The authors declare the following competing interests: X.W. is a co-founder, adviser, and shareholder of AcureX Therapeutics, and a shareholder of Mitokinin Inc. V.B., L.L., C.-H.H., and R.V. are shareholders of AcureX Therapeutics. P.N. is employed by Vroom Inc. Patents based on this study were filed by 25 Stanford University with X.W., R.V., V.B., L.L., C.-H.H. as inventors. The remaining authors declare no competing interests.
Data availability
Further information and reagents are available from the corresponding author.
Acknowledgements
We thank M.F. Tsai for MCU KO cells, L. Pallanck for flies, A. E. Medlock, O. Ross, and S. Han for discussion, C. Kwak, Z. T. Cook, D. M. Conradson and J. Panji for technical support, the Steinberg lab for assistance, the Wang lab members for discussion, the Stanford HTBC (RRID: SCR_017794) and D. E. Solow-Cordero for assistance with drug screens, and the SBC, the Stanford Alzheimer’s Disease Research Center (ADRC), the Department of Pathology at University of California Los Angeles (UCLA), R. N. Alcalay, Columbia University, and the Michael J. Fox Foundation (MJFF) for providing human tissues and cells. This work was supported by NINDS (RO1NS089583, RO1NS128040-01; X.W.), NIGMS (RO1GM143258; X.W.), NIA (RO1AG060747; M.D.G.), Stanford SPARK and Spectrum SPADA (X.W.), Sanofi iAward (X.W.), the Warren Alpert Foundation (X.W.), the MJFF (021146; X.W.), the Harrington Discovery Institute (X.W.), the Belgian American Education Foundation (R.V.), the Glenn Foundation Postdoctoral Fellowship (L.L.), and the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie (890650; Y.LG.).
Footnotes
updating results in figures, and correcting several typos