Abstract
Mutations in SNCA, the gene encoding α-synuclein (αSyn), cause familial Parkinson’s disease (PD) and aberrant αSyn is a key pathological hallmark of idiopathic PD. This α-synucleinopathy leads to mitochondrial dysfunction, which may drive dopaminergic neurodegeneration. PARKIN and PINK1, mutated in autosomal recessive PD, regulate the preferential autophagic clearance of dysfunctional mitochondria (“mitophagy”) by inducing ubiquitylation of mitochondrial proteins, a process counteracted by deubiquitylation via USP30. Here we show that loss of USP30 in Usp30 knockout mice protects against behavioral deficits and leads to increased mitophagy, decreased phospho-S129 αSyn, and attenuation of SN dopaminergic neuronal loss induced by αSyn. These observations were recapitulated with a potent, selective, brain-penetrant USP30 inhibitor, MTX115325, with good drug-like properties. These data strongly support further study of USP30 inhibition as a potential disease-modifying therapy for PD.
Introduction
A large body of evidence implicates dysfunction of mitochondrial homeostasis as a key pathophysiological mechanism in Parkinson’s disease (PD)1,2,3,4,5,6,7,8. Thus, maintenance of a healthy pool of functioning mitochondria requires a system for selectively degrading dysfunctional mitochondria (“mitophagy”)9. Autosomal recessive PD (AR-PD) due to PARKIN deficiency10 links PD directly to a defect in mitophagy. In response to mitochondrial dysfunction, PARKIN translocates to the outer mitochondrial membrane where it interacts with PINK1 (another gene where mutations cause AR-PD11,12) to ubiquitylate mitochondrial proteins, thereby inducing fusion of mitochondria with autophagosomes, followed by autophagic degradation13,14.
We hypothesize that defective mitophagy may also exacerbate α-synuclein (αSyn) toxicity. Point mutations in the αSyn gene15,16,17,18,19,20, or duplications or triplications in the gene21,22,23, cause autosomal dominant PD (AD-PD). αSyn induces mitochondrial complex I dysfunction, potentially by directly binding to TOM20 on the mitochondrial membrane and thereby interfering with mitochondrial protein import24. In addition, dysfunctional mitochondria produce increased reactive oxygen species (ROS), consistent with increased markers of oxidative damage in PD brains, and ROS can increase αSyn accumulation, thus fueling a self-accelerating pathological loop25,26,27,28,29,30,31,32,33. However, the role of mitophagy in clearing away dysfunctional mitochondria in the setting of αSyn induced mitochondrial impairment in vivo is unknown. Indirect evidence for a possible role in this setting comes from findings that PINK1 KO rats show enhanced vulnerability to αSyn toxicity34.
Most strategies to modulate mitophagy also alter autophagy in general, or impact other steps in the autophagy-lysosome pathway35, making it difficult to study mitophagy specifically36,37. A target that could allow specific molecular manipulation of mitophagy is USP30. USP30 is a deubiquitylating enzyme (DUB) tethered to the outer mitochondrial membrane where it directly removes ubiquitin attached by PARKIN or other E3 ligases38,39,40, thereby counteracting PARKIN’s ability to promote mitophagy39,41,42. As such, siRNA-mediated depletion of USP30 rescues mitophagy in PARKIN-deficient cells and protects dopaminergic (DA) neurons in PARKIN-deficient Drosophila40,43 and human neurons in cell culture42,44. Thus, inhibition of USP30 is an attractive therapeutic strategy for restoring mitophagy to achieve neuroprotection in PD. We now report data demonstrating that disruption of USP30 in Usp30 KO mice stimulates mitophagy and results in highly significant protection against αSyn toxicity. Further, we report that these effects can be recapitulated by a potent and highly selective brain-penetrant small molecule, MTX115325, with drug-like properties. Together, these data validate USP30 as a potential therapeutic target for neuroprotection in PD.
Results
Usp30 KO mice are viable and show no overt pathology
To generate Usp30 KO in mice, we flanked the Usp30 essential exon 4 with loxP sites (conditional ready allele) and then deleted it using CRE recombinase (constitutive allele) to generate Usp30 KO mice (Fig. 1a). The successful deletion of the Usp30 gene was confirmed by lack of Usp30 mRNA in the brain, kidney, heart, skeletal muscle, spleen, liver, pancreas and testis (Fig. 1b). Furthermore, no USP30 protein was detected in the cortex of Usp30 KO mice (Fig. 1c). As previously reported45, Usp30 KO mice were born at expected mendelian frequencies (Fig. 1d). In addition, we now perform high-throughput phenotyping of over 300 phenotypic parameters and show that Usp30 KO mice have no overt pathologies (Supplementary Fig. 1 and Supplementary Data 1). To see if loss of USP30 leads to pathologies with age, we established an ageing cohort and show that USP30 loss has no detectible deleterious effects with ageing when compared to wildtype (WT) littermate controls (Fig. 1e). In fact, we noticed that compared to the WT littermate controls (C57BL/6 N background), 1-year-old Usp30 KO mice are protected from fatty liver accumulation (Supplementary Fig. 2). Taken together, these data revealed no adverse effects from USP30 loss in mice.
Knockout of Usp30 enhances mitophagy levels in dopaminergic neurons
To test the hypothesis that USP30 depletion would affect mitophagy in dopaminergic neurons, we crossbred Usp30 KO mice with mito-QC reporter mice, which have a GFP-mCherry tandem fused to a mitochondrial localization signal derived from the protein FIS146. This makes it possible to measure mitophagy in vivo as the GFP signal is quenched in the acidic environment of lysosomes during mitophagy46. Thus, red mCherry puncta without a green GFP signal reflects mitochondria fused with lysosomes during mitophagic degradation (Fig. 1f). Colocalisation of red mCherry puncta with LAMP1 was used as an alternative strategy for measuring mitophagy in brain sections where the endogenous signal of mCherry-GFP is not easily detected with confocal microscopy (Supplementary Fig. 3c).
To understand if USP30 loss can affect mitochondrial clearance, we quantified the mitophagy signals in dopaminergic neurons of mito-QC/Usp30 KO mice compared to mito-QC WT littermates at 16 weeks of age. In this scenario, in individual SN dopaminergic neurons (SNpc; TH-positive, blue; Fig. 1g), we quantified the number of mCherry positive puncta (mCherry, red) colocalized with a lysosomal marker (LAMP1, green; Fig. 1f central left panels) representing mitophagosomes fused with lysosomes. We found that mCherry puncta are significantly and specifically increased in the dopaminergic neurons of Usp30 KO mice compared with WT mice (8.8 ± 0.6 per DA neuron in WT mice and 12.7 ± 1.5 per DA neuron in Usp30 KO mice, n = 13–14, p = 0.0264; Fig. 1h). We also found the mCherry/mitophagy puncta were not changed in other peripheral tissues such as muscle (Supplementary Fig. 3a, b) but were significantly increased in cortical neurons and hippocampus neurons from Usp30 KO mice compared with Usp30 WT mice at 40 weeks of age (Supplementary Fig. 3c, d). Taken together, our results indicate that loss of USP30 in mice increases basal levels of mitophagy in DA neurons in the SNpc, cortical neurons and hippocampal neurons, but not in muscle (Suppl. Fig. 3).
Usp30 KO attenuates dopaminergic neuronal loss in the AAV-A53T-SNCA mouse model
To test whether enhancement of mitophagy in DA neurons of Usp30 KO mice is associated with protection of DA neurons from αSyn toxicity, we used a validated AAV1/2-A53T-SNCA αSyn overexpression PD mouse model that shows dopaminergic neurodegeneration and motor deficits in rat, mouse and non-human primate models47,48,49,50,51. Firstly, to determine if induction of αSyn affects mitophagy, we measured mitophagy within the SNpc in ipsilateral (AAV-A53T-SNCA injected) as compared to contralateral sites (Not-injected; NI) at 28 weeks post-delivery. To verify that the mito-QC subcellular localization is specific for mitophagy, we also used staining with the mitochondrial marker, OPA-1 and we observed similar results to data generated with the mito-QC reporter system (Supplementary Fig. 4a, b). Mitophagy puncta were significantly increased in both the contralateral and ipsilateral SNpc DA neurons of Usp30 KO mice (15.80 ± 6.837 and 14.29 ± 6.862 per DA neuron, respectively) compared with Usp30 WT mice (8.889 ± 3.833 and 9.640 ± 4.881 per DA neuron, respectively) after unilateral AAV-A53T-SNCA injection (Supplementary Fig. 4b). Interestingly, expression of mutant αSyn did not affect the basal level of mitophagy independent of USP30 loss, suggesting no direct correlation between accumulation of αSyn and USP30-dependent mitophagy control at the timepoint assessed (Supplementary Fig. 4b). To determine if loss of USP30 protects against αSyn-induced DA neuronal loss in the Usp30 KO mice, we performed TH+ neuronal counting within the SNpc in ipsilateral (AAV-A53T-SNCA injected) as compared to contralateral (NI) sites at 28 weeks after the injections (Fig. 2a). These results show that USP30 loss significantly attenuated the DA neuronal loss caused by αSyn overexpression (AAV-A53T-SNCA injection; 47.35 ± 4.614 % in WT mice, p < 0.0001; 29.47 ± 6.412 % in mito-QC mice, p = 0.0025; 66.15 ± 3.135 % in mito-QC/Usp30 KO mice, p < 0.0001; one sample t test comparing with 100%, Fig. 2b) compared with WT controls (p = 0.0043; Fig. 2b) or mito-QC mice (p = 0.0002; Fig. 2b). Thus, USP30 absence protects against αSyn-induced DA neuronal loss.
Usp30 KO inhibits development of αSyn pathology and associated motor deficits
To determine if upregulation of mitophagy in Usp30 KO mice injected with AAV-A53T-SNCA is associated with decreased αSyn pathology, we did immunostaining in the SNpc with anti-phospho S129-αSyn, a pathological form of αSyn found in Lewy bodies52. Injection of the AAV-empty vector (AAV-Ev) did not produce pathological αSyn in the ipsilateral SNpc of WT, mito-QC and mito-QC/Usp30 KO mice, respectively (Supplementary Fig. 5a). In contrast, there was robust phospho-S129- αSyn fluorescence immunostaining in the ipsilateral SNpc of AAV-A53T-SNCA injected WT mice (67.10 ± 4.899 in AAV-A53T-SNCA versus AAV-Ev, p < 0.0001; Fig. 2c, d) and mito-QC mice (76.62 ± 4.854 in mito-QC AAV-A53T-SNCA versus AAV-Ev injected mice, p < 0.0001; Fig. 2c, d). Notably, the intensity of phospho-S129 αSyn in dopaminergic neurons of the AAV-A53T-SNCA injected mice was significantly reduced in mito-QC/Usp30 KO mice (Fig. 2c, d; 21.09 ± 3.065 in mito-QC/Usp30 KO AAV-A53T-SNCA injected mice versus 67.10 ± 4.899 in WT or 76.62 ± 4.854 in mito-QC injected with AAV-A53T-SNCA, p < 0.0001).
To further understand whether USP30 depletion affects the association between pathological αSyn and mitochondria in the PD model, we analyzed the overlap of a mitochondrial marker (OPA-1, red) with phospho-S129-αSyn (green) in the ipsilateral SNpc of the AAV-A53T-SNCA mouse model (Supplementary Fig. 5b). The mitochondria visualized by OPA-1 staining were mostly visible as puncta in the ipsilateral SNpc of mito-QC/Usp30 WT mice but were visible as a dynamic network in the Usp30 KO mice (Suppl. Fig. 5b). The colocalization analysis showed roughly 80% of total mitochondria was associated with pathological S129-αSyn in the SNpc neurons of the AAV-A53T-SNCA -injected WT mice (78.09 ± 1.218 %) (Supplementary Fig. 5c). Intriguingly, the fraction of mitochondria that co-stained with S129-αSyn was dramatically decreased to around 20% in the Usp30 KO mice (16.93 ± 2.739 %) (Supplementary Fig. 5c). Furthermore, the fraction of pathological S129-αSyn located on mitochondria remains comparable in both the WT (46.39 ± 2.638 %) and Usp30 KO mice (44.79 ± 2.617 %) (Supplementary Fig. 5d). These results show that USP30 depletion reduced the accumulation of pathological S129-αSyn on mitochondria.
Usp30 KO protects against αSyn-induced motor deficits
To assess if loss of dopaminergic neurons at 28 weeks post-AAV-A53T-SNCA injection is associated with motor deficits, and to assess the impact of Usp30 KO on motor function, we measured the asymmetrical usage of forelimbs in the cylinder test, which is sensitive to asymmetric dopamine deficiency47,48,49,50,51. Unilateral injection of AAV-Ev did not affect motor function in female or male mice of all three experimental groups (WT, mito-QC and mito-QC/Usp30 KO; Fig. 2f), thus excluding any nonspecific effect from the virus or the stereotaxic injection itself. In contrast, unilateral injection of the AAV-A53T-SNCA in both WT and mito-QC mice induced motor dysfunction, as shown by less use of the forelimbs contralateral to the injection (compared to use of the ipsilateral forelimb) (Fig. 2e). Notably, Usp30 KO significantly protected against the αSyn-induced motor deficits in both female and male mito-QC/Usp30 KO mice (mito-QC AAV-A53T-SNCA versus mito-QC/Usp30 KO AAV-A53T-SNCA, p < 0.0001; WT AAV-A53T-SNCA versus mito-QC/Usp30 KO AAV-A53T-SNCA, p < 0.0001; Fig. 2e and Supplementary Videos 1–3). These results demonstrate that Usp30 KO rescues αSyn-induced motor deficits, as measured by the cylinder test.
Usp30 KO protects against αSyn-induced loss of striatal dopamine and TH+ terminals
To test the impact of Usp30 KO on αSyn-induced loss of DA neurites projecting into the striatum, we measured the density of TH+ terminals in the striatum of brain sections (Fig. 3a). The relative optical density of TH+ fibers was significantly decreased in both WT mice (36.28 ± 5.539 %; p < 0.0001, Fig. 3b) and mito-QC mice (33.26 ± 5.721 %; p < 0.0001, Fig. 3b), but not in the Usp30 KO mice (84.05 ± 5.277 %; Fig. 3b) following AAV-A53T-SNCA injection.
