Regulation of ER-Mitochondria tethering in an in vivo animal model of Parkinson's disease

Mitochondria form a tubular, reticulated network which shape is controlled by opposing fusion and fission events (Bereiter-Hahn and Voth, 1994). The mitofusins 1 and 2 (Mfn1 and Mfn2) are conserved, dynamin-like GTPases embedded in the outer mitochondrial membrane (OMM) that mediate mitochondrial fusion in coordination with OPA1 (Rojo et al., 2002; Santel and Fuller, 2001; Wong et al., 2000). Mitochondrial shaping proteins have pleiotropic functions. In particular, while Mfn1 seems primarily involved in organellar docking and fusion, Mfn2 is enriched at contact sites between ER and mitochondria where it is implicated in the formation of molecular linkers that are capable of organelles tethering (Chen et al., 2012; de Brito and Scorrano, 2008). Recent works attributed to these points of close contact between the OMM and the nearby ER, called MAMs (mitochondria-associated ER-membranes) or MERCs (mitochondria-ER contacts), an important role in the propagation of cellular signals, including those that control lipid metabolism, calcium (Ca2+) homeostasis and cell death (Rowland et al., 2012; Rizzuto et al., 1998; Vance, 1990). Indeed, aberrations in ER-mitochondria juxtaposition have been described in cellular models of different neurodegenerative diseases, including Alzheimer's, Huntington's and Parkinson's disease (Krols et al., 2016; Calì et al., 2013; Ottolini et al., 2013; Area-Gomez et al., 2012; Calì et al., 2012; Panov et al., 2002). Although the exact cause for neuronal loss is not clear Parkin, an E3-ubiquitin ligase mutated in familiar Parkinson's Disease (PD) is selectively recruited to dysfunctional mitochondria and promotes their elimination via autophagy, a process known as mitophagy (Narendra et al., 2008). PINK1, a protein kinase, also a PD related gene, is required for Parkin recruitment and stress induced mitophagy (Ziviani et al., 2010). In several model systems, Parkin selectively ubiquitinates the mitochondrial outer membrane profusion protein Mfn1 and Mfn2 and fly homologue Marf (Tanaka et al., 2010; Ziviani et al., 2010; Gegg et al., 2010). Accordingly, lack of Parkin or PINK1, which operates upstream Parkin in the same pathway, results in impaired ubiquitination of Mfn and increased levels of Mfn (Ziviani et al., 2010). Given that Parkin affects Mfn steady state and ubiquitination levels, we propose to (i) address the ubiquitination levels of Mfn2 and whether Parkin downregulation affects it; (ii) investigate whether Parkin regulates ER-mitochondria tethering by impinging on Mfn2 steady state and ubiquitination levels; (iii) evaluate the physiological significance of ER-mitochondria interaction in an in vivo animal model of Parkinson's disease. Our hypothesis is that Parkin dependent Mfn ubiquitinatination controls ER-mitochondria tethering, thus impinging on Ca2+ transfer and Ca2+ homeostasis, which dysregulation has been described in a number of molecular pathways leading to PINK1 and Parkin loss of function dependent neurodegeneration (Calì et al., 2013; Ottolini et al., 2013; Calì et al., 2012). In order to address the previously listed hypothesis we analysed the pattern of ubiquitination of Mfn2 in mouse embryonic fibroblasts (MEFs) upon downregulation of Parkin. To this aim, we (i) immunoprecipitated Mfn2 with specific anti Mfn2 antibody and performed western blotting analysis with specific anti HA antibody in cells overexpressing HA tagged Ubiquitin; (ii) measured the degree of tethering between ER and mitochondria in control and Parkin downregulating cells. We used two independent approaches to measure ER-mitochondria tethering: we first measured the percentage of ER co-localizing with mitochondria by using Mander's coefficient of co-localization upon volume-rendered 3D reconstruction of z-axis stacks of confocal images of cells expressing organelles targeted fluorescence probes (mito-RFP and ER-YFP, respectively) (Rizzuto et al., 1998). Secondly, we took advantage of a FRET based probe (Naon et al., 2016) to measure ER-mitochondria proximity. In this sensor, called FEMP, FRET intensity is inversely proportional to the distance between the two fluorophores (mito-YFP and ER-CFP) that are appropriately targeted to the two compartments. (iii) We investigated the physiological significance of ER-mitochondria tether in an in vivo animal model of PD that lacks PINK1 expression. To this aim we used the fruitfly Drosophila melanogaster, which has many advantages. First, fly mutants deriving from loss of function mutations of PINK1 have been extensively characterized and cause a robust phenotype represented by age-related degeneration of DA neuron loss and locomotor deficits (Poole et al., 2008; Clark et al., 2006; Park et al., 2006; Yang et al., 2006; Wang et al., 2006). Secondly, a variety of genetic modifications and epistasis experiments can be easily performed in vivo to dissect molecular pathways. Our results showed that Parkin downregulation reduced Mfn ubiquitination and ER-mitochondria tethering in MEFs. Interestingly, we found that the pattern of Mfn2 ubiquitination and ER-mitochondria tethering is also impaired in CMT type 2A disease-associated Mfn2 mutations (Mfn2R94Q, Mfn2P251A and Mfn2R280H respectively). Although indirectly, these findings strongly suggested that ubiquitination of Mfn2, rather than its steady state levels, is important in the regulation of ER-mitochondria tethering. To identify the precise site of Mfn2 ubiquitination and directly link lack of ubiquitination with impaired ER-mitochondria tether, we took advantage of a bioinformatics approach to identify among species-highly conserved lysine (K) residues. We identified twenty Lysine residues that were conserved between human, mouse and fly. We compared these residues with those identified by a mass spectrometry-based study published in 2014 (Bingol et al., 2014) that identified Parkin-dependent ubiquitination sites. We identified six Lysine residues that were likely to represent good candidates for Parkin-dependent ubiquitination of Mfn2. We generated non-ubiquitinatable mutants for those sites by substituting Lysine (K) with Arginine (R), a common procedure to impair ubiquitination and investigated the pattern of ubiquitination of the non-ubiquitinatable Mfn2 mutants by western blotting. Expression of non-ubiquitinable mutant K416R resulted in impaired Mfn2 ubiquitination. Of note this mutant was also unable to correct ER-mitochondrial contacts when expressed in Mfn2 KO MEFs and only partial restored ER-mitochondrial Ca2+ transfer. In summary, our results provided strong evidences that Mfn2 ubiquitination is a prerequisite for ER-mitochondria physical and functional interaction and that K416 in the HR1 domain of Mfn2 is a genuine site for Parkin dependent ubiquitination. A number of studies have shown impaired Ca2+ homeostasis in cellular models lacking PINK1 or Parkin (Heeman et al., 2011; Sandebring et al., 2009). Although it is not clear why dopaminergic neurons specifically degenerate in PD, it is tempting to hypothesis that impaired Ca2+ homeostasis resulting from impaired Ca2+ cross talk at ER-mitochondria interface could lead or contribute to degeneration. Elegant studies have shown that artificial tether between ER and mitochondria can be used to modulate Ca2+ transfer (Csordas et al., 2010; Csordas et al., 2006). With that in mind, we addressed whether expressing an ER-mitochondria synthetic linker in a well-established in vivo Drosophila model of PINK1 loss of function could ameliorate PINK1 KO phenotypes by impinging on ER-mitochondria cross talk. We therefore generated a number of fly lines expressing the synthetic linker driven by a neuron-specific driver in the fly wing neurons. This linker was generated by Csordas et al. (Csordas et al., 2006) and consists of a monomeric fluorescent protein (RFP) fused to the outer mitochondrial membrane targeting sequence at the N terminus and fused to the ER targeting sequence at the C terminus. We could observe a well-defined and easily quantifiable RFP-fluorescence spots throughout the L1 vein of the fly wing that perfectly matched the morphology seen when expressing mito-GFP or ER-GFP alone in the wing neurons (Vagnoni and Bullock, 2016), which indicated that the synthetic linker was appropriately expressed. Interestingly, we found an amelioration of PINK1 KO climbing ability upon expression of the artificial synthetic linker. This result strong indicates that restoration of proper ER-mitochondrial communication in PINK1 KO background can be beneficial in ameliorating the phenotype associated to an in vivo animal model of PD, paving the way for novel approaches for medical intervention.