Pre-formed fibrils of α-synuclein cause mitochondrial dysfunction and impair mitophagy
The pathological hallmark of idiopathic and most genetic forms of PD is the misfolding and subsequent deposition of α-synuclein into insoluble, beta sheet rich aggregates deposits referred to as Lewy pathology, resulting in the accumulation of posttranslationally modified α-synuclein. One of the most prominent modifications is the phosphorylation at serine 129 (pS129) (Rocha et al., 2018) which is also widely used to detect disease associated α-syn by immunohistochemical and biochemical methods. High-resolution studies revealed that mitochondrial fragments are an integral part of Lewy Body structures, suggesting an interaction of misfolded α-synuclein with mitochondrial structures (Shahmoradian et al., 2019). Furthermore, patient-derived A53T SNCA iPSCs show delayed mitophagy (Devi et al., 2008; Ryan et al., 2018; Wang et al., 2019). Consistent with these findings, we found pS129 α-synuclein associated nearly exclusively with the mitochondrial fraction of cortical brain extracts in human PD patients (Fig. 1A-C). To assess whether pathogenic α-synuclein induced mitochondrial dysfunction we seeded cultured primary neurons with α-synuclein preformed fibrils (PFFs) (Volpicelli-Daley et al., 2011), which lead to pS129 positive α-synuclein accumulation (Figs. 1D-E and S1A-B). We observed concentration- and time-dependent defects in mitochondrial respiration (Figs. 1F-I and S1C-F), impaired mitophagy (Figs. 1J), and a dose-dependent accumulation of pUb (Fig. 1K). Interestingly, a chronic, low dose of the mitochondrial uncoupler CCCP also led to impaired mitophagy in neurons (Fig. 1J) in contrast to the widely used high doses which increase mitophagy (CITATIION). This result suggests that stimulating chronic low levels of mitophagy via mitochondrial uncoupling leads to stalled mitophagy in which PINK1 is partially activated but unable to complete the mitophagic process; as a result, pUb that would otherwise be turned over when the mitophagic process is completed, builds up in the cell. These data support a model in which α-synuclein pathology increases mitochondrial dysfunction and impairs mitophagy, leading to accumulation of pUb (Fig. 1L).
The static observation that pUb is elevated in PFF challenged neurons could indicate reduced or increased PINK1 function, resulting in a decreased or increased rate of mitophagy respectively. This is an often-observed conundrum in autophagy-related processes when measuring adaptor proteins at single time points (CITATION). Parallel measurements of decreased mitophagy levels in this model clearly show that the pUb accumulation is a result of stalled mitophagy, and that increased mitochondrial dysfunction and impaired mitophagy is a caused of α-synuclein aggregation (Figs. 1L), leading to an increase in pUb. Conversely, alleviation of mitochondrial stress and stalled mitophagy by further activation of PINK1 would result in lower pUb and small molecule activators of PINK1 could provide a strategy to resolve stalled mitophagy induced by aggregated α-synuclein restoring the neuronal homeostasis.
Identifying and qualifying small molecule activators of PINK1
Our experiments establish that reduced rates of mitophagy and unresolved mitochondrial damage are key impairments in α-synuclein pathobiology and support pharmacological activation of PINK1 as a potential avenue to mitigate α-syn-induced cellular impairments. Our initial approach to activate PINK1 with neo-substrates (Hertz et al., 2013) led us to the discovery of kinetin, which activates PINK1 in cells and relieves mitochondrial mutations in flies and mice in a PINK1-dependent manner (Osgerby et al., 2017; Tsai et al., 2022). However, because kinetin has low potency and poor pharmacokinetics (PK) and brain penetrance, we could not detect an effect in PD models in vivo (Orr et al., 2017). To overcome these limitations and to discover novel small molecule PINK1 activators with drug-like properties, we synthesized and screened small molecules derived from the structural core of kinetin (Fig. 2A-B). First, active compounds were discovered by measuring activity in a cell-based assay for mitophagy in which a pH sensitive protein (keima) is localized on mitochondria (mKeima) and a characteristic shift in the absorption/excitation spectrum is observed upon initiation of mitophagy (Figs. 2C and S2C) (Lazarou et al., 2015). We tested each compound in a 7-point concentration curve in the presence of a low concentration (1 µM) of FCCP and oligomycin (FO). In cell culture models, low levels of FO are necessary to trigger mitochondrial stress and stabilize PINK1; this dose was selected because it did not robustly trigger mitophagy on its own.
In order to rule out nonspecific, additive mitochondrial toxicity as the mechanism of action of a compound showing activity in the mKeima assay, we counter-screened active compounds for mitochondrial toxicity in a galactose/glucose cell growth assay (Arroyo et al., 2016; Gohil et al., 2010; Marroquin et al., 2007). We used a 20% decrease in growth rate in galactose-rich media relative to glucose-rich media as a cutoff for mitotoxicity (Figs. 2D and S2A). A subset of these active, non-mitotoxic compounds were then evaluated for initial developability using the following criteria: solubility, permeability, brain efflux, in vitro liver microsome clearance in multiple species, CYP screening, plasma protein binding, and hERG inhibition (Fig. 2). Several compounds that fulfilled developability criteria were tested in mouse pharmacokinetic (PK) and tissue distribution studies. The compound MTK458 showed good potency, no observable mitotoxicity, attractive oral pharmacokinetics (PK), and high brain penetrance. To further confirm that MTK458 is not impairing mitochondria with a more subtle effect, we measured mitochondrial respiration rates in HeLa cells treated with MTK458 for 1 hour. MTK458 did not affect basal respiration, maximal respiration, or spare respiratory capacity (Figs. 2E and S2D).
We tested MTK458 in successive assays for PINK1 pathway activity in the presence of low concentrations of FO. In HeLa cells expressing YFP-Parkin and mito-Keima (YPMK), MTK458 increased pUb as assessed by a custom pUb ELISA assay (Fig. 2F) and mass spectrometry (Figure S2B). Next, we monitored Parkin activation by PINK1 through its cytosolic-to-mitochondrial translocation via live-cell imaging. MTK458 accelerated localization of YFP-Parkin to the mitochondria with a clear dose-responsiveness. (Figs. 2G and S2E). Thus, MTK458 increases early stage (pUb), mid-stage (Parkin recruitment to mitochondria), and late-stage (mitophagy, Fig. 3H) processes of the PINK1/Parkin cascade working in a dose-dependent and PINK1-dependent manner.
PINK1 also phosphorylates Parkin at S65, and downstream of ubiquitin and Parkin phosphorylation, the mitofusin proteins (e.g., MFN1/2) and some outer mitochondrial membrane proteins (e.g., VDAC) are degraded. Consistent with PINK1 activation, MTK458 increases pS65 Parkin and decreases MFN1, MFN2, and VDAC (Figures S3A and S3C-G). PINK1 activation also results in phosphorylation of Rab proteins, specifically Rab8A, 8B and 13, at the highly conserved residue of serine 111 (Lai et al., 2015). The phosphorylation of the Rabs is not catalyzed by PINK1 directly, but is abolished in PINK1 knockout cells, indicating this phosphorylation site can be used as a downstream proxy for PINK1 activity. Consistent with being a PINK1 activator, MTK458 also increases the pS111 Rab8A signal in cells treated with a low dose of FO (Figures S3B and S3H). In the absence of any mitochondrial stressor MTK458 does not induce any of the aforementioned PINK1 biomarkers (Figure S3I), further displaying that the induction and increased activation of mitophagy pathways through MTK458 is selective for dysfunctional states of mitochondria.
MTK458 shows direct PINK1 binding
Although PINK1 from non-mammalian species has been utilized for structural studies (Gan et al., 2022), purification of human PINK1 for direct binding assays or crystallization has not yet been achieved. In order to demonstrate direct PINK1 engagement we developed a novel direct binding assay for PINK1 in human cells based on the nanoBRET (bioluminescence resonance energy transfer) system (Machleidt et al., 2015), using a tracer molecule based on the structure of MTK458. In this system, PINK1 was N-terminally tagged with NanoLuc luciferase (NL-PINK1), and MTK458 was labeled with the nanoBRET 590 dye. With this approach, a BRET signal only results if the labeled MTK458 is within 100 angstroms of NL-PINK1 (Fig. 3A). We observed a concentration dependent increase in BRET signal in cells expressing NL-PINK1 and treated with the MTK458-derived nanoBRET tracer (Fig. 3B), suggesting that MTK458 binds directly to PINK1. As a control, MTK458 did not bind an unrelated but luciferase-tagged kinase, GSK3 (GSK3B-NL), Fig. 3B). When we used a non-specific kinase binding tracer K8 (Promega) induced a dose-dependent increase in BRET ratio with the GSK3B-NL, and less signal with NL-PINK1 (Figure S4A), suggesting that MTK-458 promotes mitophagy through direct and specific binding of PINK1.
MTK458 stabilizes the PINK1/TOM complex and opposes PINK1 inactivation
We next explored the mechanism by which MTK458 potentiates PINK1 activity. Previous work with kinetin and the active metabolite KTP suggested that modification to a triphosphate form is required for activity (Hertz et al., 2013). However, MTK458 cannot be ribosylated (data not shown), so despite similarities in structure, we postulated that MTK458 must act via a new mechanism. Activation of PINK1 is believed to involve dimerization, auto-phosphorylation in trans at Ser228, and formation of a high molecular weight (HMW) complex with components of the mitochondrial translocase of the outer membrane (TOM) proteins (Lazarou et al., 2012; Okatsu et al., 2012; Rasool and Trempe, 2018). To investigate the effect of MTK458 on PINK1 dimerization, we used a split-nanoLuc protein fragment complementation system whereby cells were transfected with two species of PINK1, one fused with SmBiT and the other with LgBiT (Fig. 3C). When PINK1 dimerizes, the SmBiT and LgBiT proteins assemble into a functional nanoLuc protein that can generate a luminescence signal. MTK458 increased PINK1 dimerization in a concentration dependent manner (Fig. 3D). Importantly, MTK458 or low-dose F/O alone (t = 0 point is + F/O) did not stimulate PINK1 dimerization when applied separately, while FO priming in combination with MTK458 treatment resulted in a robust PINK1 dimerization as evidenced by increased luminescence signal in this assay. Next, we used Phos-tag SDS-PAGE (Kinoshita et al., 2009) and blue-native gel electrophoresis (Lazarou et al., 2012) to test the effect of MTK458 on PINK1 phosphorylation and complex formation, respectively. We observed an increase in phospho-PINK1 by MTK458 in Phos-tag SDS-PAGE (Figs. 3E-F and S4B). When lysates from cells exposed to high F/O were treated with lambda protein phosphatase, the intensity of the phospho-PINK1 band and pUb bands decreased (data not shown). Besides phosphorylation, addition of MTK458 increased the total amount of PINK1 in the active, HMW complex (Figs. 4G-H and S4C), suggesting that MTK458 treatment increased the levels of total and phosphorylated PINK1.
In the absence of mitochondrial stress, PINK1 is rapidly destabilized and degraded (Jin et al., 2010). We found that MTK458 does not activate PINK1 without a mitochondrial stressor, whereas it potentiates both PINK1 autophosphorylation and complex formation with low-dose mitochondrial stress. Based on this finding, we hypothesized that MTK458 stabilizes the active PINK1 complex and therefore delays its inactivation. To investigate the effect of MTK458 on PINK1 complex stability after removal of mitochondrial toxins, we performed FCCP washout studies in SK-OV-3 cells, which express endogenous levels of Parkin and downstream components of the PINK1/parkin pathway (Kakade et al., 2022) (Fig. 4A). SK-OV-3 cells were transiently treated with FCCP alone or combined FCCP/MTK458 for 2h, then the FCCP was removed by washing the cells three times with FBS-containing medium (“washout”). After the washout, the cells were treated with eitherMTK458 or DMSO control (Fig. 4B). Exposure to FCCP induced high PINK1 and pUb levels, which rapidly decreased after washout in the DMSO condition (Figs. 4C-E and S4D). However, if the cells were co-treated with MTK458 during the transient FCCP treatment, the high PINK1 and pUb levels were sustained after the washout as detected by both immunoblotting and mass spectrometry (Ordureau et al., 2014) (Figs. 4C-E and S4D). The high molecular weight PINK1 complex was also sustained by MTK458 even after FCCP is removed (Figures S4E-F). Importantly, MTK458 treatment did not interfere with mitochondrial repolarization, suggesting that the PINK1 complex-potentiating effect is not driven by a compound-driven effect on mitochondrial membrane potential (Figures S4G-H). Taken together, our data supports a model where MTK458 potentiates and prolongs the stability of the active PINK1 complex (Fig. 4A), but does not initiate complex stabilization without mitochondrial depolarization.
MTK458 drives clearance of pathologic α-synuclein in vitro
Alpha-synuclein aggregation induces mitochondrial dysfunction coupled, reduced rates of mitophagy and therefore accumulation of pUb in primary neurons (Fig. 1K). Therefore, we wanted to test whether treatment with MTK458 would increase PINK1 activity in models of proteinopathy induced mitochondrial dysfunction. To test this, we utilized two independent proteinopathy models, an inducible mitochondrial proteinopathy model and the aggregated α-synuclein seeding model (PFFs) noted above.
First, we utilized a cell-based model, which expresses a deletion mutant of ornithine transcarbamylase (ΔOTC) (Fig. 5A) (Moisoi et al., 2014). This mutanyields detergent-insoluble, intra-mitochondrial ΔOTC protein aggregates within the mitochondrial matrix that can be cleared by PINK1/Parkin-mediated mitophagy (Burman et al., 2017; Moisoi et al., 2014). In HeLa cells expressing doxycycline-inducible ΔOTC and YFP-Parkin, enhancing PINK1 activity with MTK458 treatment resulted in the robust clearance of ΔOTC as measured by either immunofluorescence or Western blotting (Figs. 5B-C), demonstrating that clearance of intra-mitochondrial aggregates can be enhanced by increased PINK1-mediated mitophagy driven by MTK458.
Having shown that MTK458 treatment could reduce artificially induced intra-mitochondrial aggregates in non-neuronal cells, we next tested whether PINK1 activation could ameliorate PD-patient relevant pathology in mouse and human neurons in vitro. Primary mouse hippocampal neuron cultures from PINK1wt (5E-H) or PINK1 knockout (KO) (5I) mice seeded with α-synuclein PFFs on DIV7 (Volpicelli-Daley et al., 2014) were allowed to seed and further develop detergent insoluble pS129 α-synuclein aggregates prior to the addition of MTK458 on DIV9 and DIV12 (Fig. 5D). On DIV14, pS129 α-synuclein aggregates were detectable by immunoblotting from the insoluble fraction (Figs. 5E-G) and immunofluorescence (Figs. 5H-I). MTK458 treatment led to the clearance of pS129 α-synuclein aggregates (12–250 kDa) in a dose and PINK1 dependent manner (Figs. 5E-I).
We further tested the effect of MTK458 in iPSC-derived neurons from patients carrying the A53T-α-synuclein mutation associated with familial PD (Figure S5A). This line carries an A53T α-synuclein mutation causing the derived DA neurons to accumulate pS129 α-synuclein without the addition of exogenous PFFs, and additionally serves to bridge primary mouse neuron studies with human neurons. To first test if global induction of mitophagy driven could reduce pS129 α-synuclein, cells were treated with FCCP to activate PINK1 by depolarization (Figures S5B-D). FCCP alone reduced pS129 α-synuclein pathology in these cells, but at the expense of an increase in mitochondrial stress throughout the cell, as evidenced by stabilization of PINK1 and increased pUb levels (Figures S5B-D). In contrast, MTK458 treatment for 10 days reduced α-synuclein pathology and the mitochondrial stress marker pUb (Figures S5B-D). We hypothesize that mitochondrial depolarization is being triggered by a-synuclein pathology in the absence of FCCP and PINK1 stabilization is further induced in cells treated with MTK458, but at lower levels as compared to FCCP treatment and more selectively on impaired mitochondria. Consistent with the model we proposed above, activation of PINK1 in the patient-derived iPSC neurons reduced protein aggregate load and ultimately drove a reduction in pUb (Figures S5B-D), indicative of clearance of damaged mitochondria. Our results in various PD cell models suggest that pharmacological augmentation of PINK1 activity can ameliorate proteinaceous pathology in vitro .
MTK458 drives clearance of pathologic α-synuclein in vivo
To test if PINK1 activation could rescue α-synuclein pathology in vivo, we utilized a widely adopted pre-clinical model for PD in which α-synuclein PFFs are injected unilaterally into the striatum of mice, leading to progressive spread α-synuclein pathology (Fig. 6A) (Luk et al., 2012). Microdialysis studies with MTK458 in the mouse striatum showed a similar unbound plasma and brain exposures at equilibrium (unbound partition coefficient, Kp u,u ~1) (data not shown) showing that MTK458 has excellent mouse pharmacokinetics and high brain penetrance. Consistent with our finding in mouse primary neuron cultures that PFF injection led to profound α-synuclein pathology in the striatum as evidenced by aggregated and pS129 α-synuclein after 12 weeks of incubation accompanied by an increase in pUb (Figs. 6B-D and 6G). PFF, but not PBS injection, increases brain pUb (Fig. 6G), and central (TREM2) and peripheral (IL6, CXCL1) inflammatory markers (Figures S6A-C). Daily oral administration of MTK458 in these mice led to dose dependent decrease (up to ~ 50%) in α-synuclein pathology in 3-month studies (Figs. 6B-D). MTK458 also rescued an activity deficit in freely moving PFF-seeded mice as assessed by home cage monitoring (Figs. 6E-F) (Lim, M.A., et al 2017 Frontiers in Pharmacology). The dose-response rescue in pathology matched the rescue in motor activity. The increase in TREM2, IL-6, and CXCL1 inflammatory markers were also attenuated by MTK458 (Figures S6A-C), which is in line with a proposed role for PINK1 in inflammation (Sliter et al., 2018).
Consistent with the results from our iPSC-derived neuron experiments, MTK458 treatment decreased pUb in the brains of PFF seeded mice (Fig. 6G), suggesting a reduction in mitochondrial stress level due to the clearance of damaged mitochondria and pS129 α-synuclein aggregates. Unexpectedly, we did not see an increase in plasma pUb levels in the PFF-challenged mice as compared to PBS challenged mice. However, a decrease in plasma pUb was observed in mice treated with MTK458 (Fig. 6H) showing global PINK1 pathway engagement.
Since plasma concentrations of pUb in naïve, wild-type animals are measurable and greater than in PINK1 KO animals (Chin et al., 2023), we hypothesized that we might be able to detect a decrease in pUb after a short-term MTK458 treatment in animals. Such a change would be useful to measure target engagement of PINK1 activator compounds in animals or in patients. To test this hypothesis, we dosed naïve, wild-type Sprague-Dawley rats for five days with either a vehicle control or 50 mg/kg MTK458 and found a significant decrease in pUb compared to the vehicle-treated or pre-dosed rats (Figs. 6I-J). The magnitude of the plasma pUb lowering effect (ROC = 1.00) (Figs. 6K and S6E-H) suggests it may be useful as a specific and sensitive pharmacodynamic biomarker for MTK458 treatment.