Mutant HTT (huntingtin) impairs mitophagy in a cellular model of Huntington disease

Mitochondrial dysfunction in ST-Q111 cells

Mitochondrial dysfunction and impairment of the proteolytic systems have been extensively linked to neurodegenerative diseases, including HD [4]. Neurons are particularly vulnerable to autophagic impairment as well as to mitochondrial dysfunction, mostly due to their particularly high-energy dependence and to their post-mitotic nature. The precise and proper degradation of dysfunctional mitochondria by mitophagy is essential for maintaining control over mitochondrial quality and quantity in neurons. Here, we focus on the role of HTT protein in the mechanism of mitophagy and the consequences of the polyQ tract expansion. To this end, we used differentiated immortalized striatal neurons derived from a knock-in mouse expressing human exon 1 carrying a polyQ tract with 7 glutamines STHdh-Q7 (ST-Q7, control) or 111 glutamines STHdh-Q111 (ST-Q111, mutant) inserted into the endogenous mouse Htt gene. To validate our model, we first verified the presence of mitochondrial dysfunction with different approaches. The ratio between non-depolarized mitochondria, detected with a mitochondrial probe that is sensitive to the mitochondrial membrane potential (TMRM), and total mitochondria, detected with a mitochondrial membrane potential-insensitive mitochondrial probe (MitoTracker Deep Red, MTDR), indicates the number of healthy mitochondria remaining after depolarization. After 24 h of mild (rotenone 1 μM for 4 h) or acute depolarization (CCCP 10 μM for 24 h), this ratio was higher in control cells compared to mutant cells (Figure 1A). This finding is compatible with a reduced capacity for the turnover of damaged mitochondria in mutant cells and; therefore, further analysis is required. We evaluated mitochondrial function by measuring the oxygen consumption rate (OCR) of differentiated striatal cells under basal conditions and in response to cellular respiration stress. Under basal conditions, both differentiated striatal cell lines had similar respiration levels (Figure 1B and C). Inhibition of adenosine triphosphate (ATP) synthase with oligomycin caused an expected decrease in respiration in both ST-Q7 and ST-Q111 cells, with no significant difference between the two cell types. The addition of the uncoupling agent CCCP collapsed the proton gradient and caused an increase in the OCR until the maximum respiration rate was achieved. ST-Q111 cells clearly showed a lower rate of maximal respiration and a lower spare respiratory capacity compared with control cells (Figure 1B and C), indicating that mitochondria from ST-Q111 cells are bioenergetically less-active and have a decreased ability to respond to increased energy demand. When cells were initially subjected to mild depolarization by exposure to rotenone for 4 h and then left to recover for 24 h, mitochondria from mutant ST-Q111 cells exhibited lower levels of basal respiration. After the addition of the different inhibitors and uncoupling agents, mutant cells showed reductions in ATP production, maximal respiration, and spare respiratory capacity compared to control cells (Figure 1D), suggesting that inefficient mitophagy triggered in response to cellular stress leads to the accumulation of damaged mitochondria. Consequently, the remaining pool of mitochondria present in the cell is not fully healthy, thus, contributing to the general mitochondrial dysfunction observed in HD. One of the main consequences of the cell’s inability to properly eliminate damaged mitochondria is the increase in ROS. These can be detected by the fluorescent probes CellROX Green (Figure 1E and F) and CM-H₂DCFDA (Figure 1G). Our data showed increased ROS levels in ST-Q111 cells compared with ST-Q7 cells, in part, due to the higher levels of damaged mitochondria present in mutant cells. In conclusion, differentiated ST-Q111 cells after depolarization are more prone to accumulate damaged mitochondria and to display altered mitochondrial respiration and higher levels of ROS.

Mitophagy is affected in ST-Q111 cells

Following the confirmation that our model, in accordance with other HD models [26–30], presents mitochondrial dysfunction, we next sought to verify if the mitophagy was indeed affected in differentiated striatal ST-Q111 cells and could at least in part, explain the mitochondrial dysfunction observed in HD. Altered autophagy has been widely described in vivo and in vitro models of HD, as well as in postmortem brain samples from patients [21,31–36]. Indeed, HTT protein has been shown to be implicated in the cargo recognition step during selective autophagy [18,19,21]. Due to its ability to act as a scaffold, HTT brings together different proteins required for autophagy to take place. However, the presence of the polyQ tract can alter these interactions and destabilize different protein complexes, thus affecting the efficiency of selective autophagy processes, including mitophagy. To study the consequences of glutamine expansion in HTT’s role in mitophagy, we used different approaches to monitor the autophagy-dependent degradation of mitochondria. First, we studied mitochondrial clearance by quantifying the colocalization of mitochondria within lysosomes, as the last step of the mitophagy process. Under basal conditions, the number of lysosomes loaded with mitochondrial material in the presence of lysosomal inhibitors was similar between control and mutant striatal cells (Figure 2A). However, after mild and acute mitochondrial depolarization, control cells showed a clear increase in the number of lysosomes colocalizing with mitochondria, but we observed no increase in the mutant ST-Q111 cells (Figure 2A). This finding indicates that mHTT cells are less efficient at engulfing and digesting damaged mitochondria following depolarization. To analyze mitochondrial clearance in an alternative way, we quantified mitophagy flux by detecting the mitochondrial pool with the MTDR probe under basal conditions and after induction of mitophagy by rotenone and CCCP, in the presence and absence of a lysosomal inhibitor (60 μM chloroquine) to avoid mitochondrial degradation within lysosomes. The increase of the MTDR signal in each condition due to the presence of chloroquine is representative of the number of mitochondria degraded by mitophagy (Figure 2B). ST-Q7 cells showed a significant increase of MTDR signal in the presence of lysosomal inhibitors (chloroquine) in all three conditions, indicating that mitophagy flux is active. However, in QT-111 cells, the increase of the MTDR signal in the presence of chloroquine is lower and, after induction of mitophagy with rotenone or CCCP, is not statistically significant, indicating that mitophagy flux is deficient in these cells.

As another measure of mitochondrial degradation, we quantified mitochondrial (mt)DNA as a reporter of mitochondrial mass using an anti-DNA antibody that preferentially binds to mtDNA [37]. ST-Q7 cells displayed a loss of DNA intensity upon rotenone and CCCP treatment, which is indicative of mitochondrial removal in response to damage. However, we observed no significant changes in ST-Q111 cells, further implying that mitochondria were not being degraded with the same efficiency as ST-Q7 cells upon stress-induction (Figure 2C). These results indicate that induced mitophagy is impaired in differentiated ST-Q111 cells, and suggest that mHTT can directly or indirectly affect the efficiency of this process.

The polyQ tract in mHTT affects the initiation of mitophagy

To gain insight into the mechanism by which mHTT affects mitophagy, we analyzed the different steps of the mitophagy process, on the one hand, the induction of autophagy and, on the other, the labeling and subsequent recognition of the targeted mitochondria. First, by focusing on the parallel activation of the ULK1 and BECN1-containing class III phosphatidylinositol 3-kinase (PtdIns3K) complexes, we studied if mHTT could affect the initiation of the autophagy machinery. The initiation process is coordinated by ULK1, which forms a complex with ATG13 and RB1CC1/FIP200 to regulate the initial step of autophagy induction in mammalian cells [38–40]. To induce autophagy, ULK1 must be activated and released from the MTOR complex 1 (MTORC1), losing the MTOR-mediated inhibitory phosphorylation at Ser757 [41]. Simultaneously, AMP-activated protein kinase (AMPK) activates ULK1 through physical interaction [42] and via phosphorylation of the Ser555, Ser777, and Ser317 residues [41,43]. Previous studies showed that HTT plays a key role as a scaffold protein by binding active ULK1 at its C-terminal domain, competing with MTORC1 for ULK1 binding, and promoting autophagy initiation [18,19]. To understand how the presence of the mutant polyQ tract might affect this step, we analyzed the interaction of ULK1 with MTOR and HTT. To this end, we used the proximity ligation assay (PLA) since immunoprecipitation with the endogenous proteins was technically not feasible due to the low levels of these proteins. PLA allows a highly sensitive and highly specific quantification of protein-protein interaction under different conditions. PLA methodology was initially validated by quantifying a well-known HTT-SYVN1/HRD1 interaction [44] (Fig. S1). We next studied the interactions of ULK1 by PLA and observed that in differentiated ST-Q111 cells, the ULK1-MTOR interaction was more stable after depolarization compared to that seen in ST-Q7 cells, where the ULK1-MTOR interaction decreased upon depolarization to permit ULK1 activation (Figure 3A and negative controls in Fig. S2). These results suggest that, in mutant cells, ULK1 remains more inactive and bound to MTOR than in control cells. Similarly, the ULK1-HTT interaction increased in ST-Q7 cells after depolarization, but this interaction was not induced in ST-Q111 cells after depolarization, confirming that ULK1 has a reduced affinity for mHTT (Figure 3B). Using PLA, we also detected a lower amount of active ULK1 (phosphorylated at S317) bound to HTT in ST-Q111 cells after depolarization (Figure 3C). In order to determine if the changes in ULK1-MTOR and ULK1-HTT interactions could affect ULK1’s phosphorylation state, we analyzed MTOR-dependent and AMPK-dependent phosphorylations. MTOR-dependent Ser757 inactivating phosphorylation showed a tendency to decrease after mitochondrial depolarization that, although was non-significant, would agree with ULK1 activation and autophagy induction. However, ST-Q111 cells maintained high levels of Ser757 phosphorylation upon depolarization, indicating that ULK1 remains inactivated(Figure 3D). We also analyzed ULK1’s AMPK-dependent phosphorylations at Ser555 and Ser317, both of which are more abundant after mitochondrial depolarization in both ST-Q7 and ST-Q111 cell lines, with no significant changes between cell lines (Figure 3D), indicating that AMPK-dependent phosphorylations are not negatively affected by mHTT. Together, these results suggest that the extended polyQ tract in mHTT might be interfering with ULK1 activation and the shift from the MTORC1 to the HTT scaffolding complex, a necessary step to initiate autophagy and consequently that of mitophagy as well.

Following its activation, ULK1 phosphorylates BECN1, a component of the autophagy initiation complex or PtdIns3K complex formed by BECN1, PIK3R4/VPS15, PIK3C3/VPS34, and other proteins and responsible for autophagosome formation [45]. Recently it was shown that the deubiquitinase ATXN3 (ataxin 3), which is altered in spinocerebellar ataxia type 3, interacts through its wild-type (WT) polyQ domain with BECN1 and protects BECN1 from proteasomal degradation. However, BECN1 binds not only to ATXN3 but also to a greater extent to expanded polyQ tracts such as the ones present on mHTT, thus, competing with the physiological binding of ATXN3 and promoting the degradation of BECN1 [46]. Lower levels of BECN1 are related to a reduced initiation of complex assembly and lower macroautophagy rates [46]. Using PLA, we investigated whether mHTT could affect mitophagy at this step by impairing the formation of the PtdIns3K complex. At basal levels, the presence of mHTT downregulated the BECN1-PIK3R4/VPS15 interaction. Upon depolarization, differentiated ST-Q7 cells exhibited high levels of BECN1-PIK3R4/VPS15 interaction, which is indicative of assembly of the PtdIns3K initiation complex, while differentiated ST-Q111 cells displayed downregulation of the formation of this complex (Figure 3E and S2). These results further confirm that expanded polyQ tracts present in mHTT impair the formation of the autophagy initiation complex in striatal cells.

Mitophagy in differentiated striatal cells does not affect the polyUB labeling of mitochondria

Mitophagy can degrade mitochondria through various pathways depending on the cell type and conditions [4]. Some of these pathways may be alternative, but others could be redundant and might compensate for each other under certain physiologic or pathologic conditions [4].

Mitophagy can be achieved by a UB-mediated process with the collaboration of E3 ubiquitin ligases and the presence of non-mitochondrial receptors like SQSTM1/p62, NBR1, OPTN, and CALCOCO2 containing an UB-associated (UBA) and an LC3-interacting region (LIR) domain in their structure. Alternatively, UB-independent mitophagy can take place with the participation of mitochondrial receptors like BNIP3 (BCL2/adenovirus E1B 19 kDa protein-interacting protein), BNIP3L (BCL2/adenovirus E1B 19 kDa protein-interacting protein 3-like, also known as NIX) and FUNDC1 (FUN14 domain-containing protein 1). These receptors can bind directly to LC3 through their LIR domain in response to different conditions like hypoxia or metabolic stress [4].

We further investigated the different alternative pathways to elucidate the effect of mHTT in mitophagy in differentiated striatal cells. First, we analyzed the effect of mHTT in the best-characterized mitophagy mechanism, mediated by the action of PINK1 (PTEN induced kinase 1) protein. Depolarized mitochondria are labeled with phosphorylated polyUB at serine 65 (polyUB-p-S65) tagged on a variety of substrates present on the outer mitochondrial membrane, and these tags act as the signal to be recognized by the mitophagy machinery. E3 UB ligases mediate this tagging process, along with PRKN/PARKIN, the most well-characterized UB ligase protein involved in this step. However, PRKN requires activation by the kinase PINK1, which phosphorylates both PRKN and also UB at their serine 65. Thus, and according to many studies, the first step of this mitochondrial labeling is the retention of full-length PINK1 protein on the mitochondrial surface and the subsequent recruitment and activation of PRKN protein [47–50]. Active PRKN ubiquitinates the substrates on the outer mitochondrial membrane, which feeds a positive feedback loop, amplifying the “eat-me” signals [37]. However, recent studies have shown that this scenario might not be common during neuronal mitophagy since PRKN cannot be detected in mitochondria in several neuronal models [51–54], indicating that other alternative E3 UB ligases might take the role of PRKN [55]. Consequently, neuronal mitophagy might take place in a slower process compared to that in other cell types or models that exhibit high levels of PRKN translocated to the mitochondria [56].

According to this situation, we first investigated the possible role of PINK1 and PRKN upon mitophagy induction in our cellular model. We observed that PINK1 protein levels remained steady in differentiated ST-Q7 cells upon mild or acute depolarization. In differentiated ST-Q111 cells, basal PINK1 protein levels slightly increased, but again no changes were observed upon rotenone- or CCCP-treatment (Figure 4A). Depolarization induced by various effectors did not alter PRKN protein levels in total cellular homogenates. Importantly, we were unable to detect PRKN in isolated mitochondrial fractions (Figure 4B). PRKN translocation to mitochondria was also not detected by confocal immunofluorescence in the presence of basal or depolarization-induced conditions in differentiated ST-Q7 and ST-Q111 cells (Figure 4C). This scenario is similar to that observed in neuronal cell lines where endogenous PRKN translocation to the mitochondria was not detectable [51–54]. To test if the lack of mitochondrial-associated PRKN after depolarization was due to the lower levels of PRKN protein, we artificially overexpressed PRKN protein tagged to a fluorescent protein (EGFP). Even with the overexpression of EGFP-PRKN protein, we did not observe a clear translocation of EGFP-PRKN within a short time with mild depolarization (1 μM Rot 4 h). Only after 24 h with a strong depolarizing agent (10 μM CCCP) was it possible to observe a partial translocation of EGFP-PRKN from a diffuse cytosolic distribution to a mitochondrial localization (Figure 4D).

Nevertheless, EGFP-PRKN mitochondrial translocation was only detectable in less than 50% of ST-Q7 (42.7% ± 11.8), and ST-Q111 (46.3% ± 11.2) differentiated striatal cells (Figure 4D), as previously reported [57]. This result contrasts with other studies where PRKN overexpression led to a clear mitochondrial localization in other cell lines, including neuronal cells [51,52,58,59]. Based on these findings, we analyzed endogenous and EGFP-PRKN translocation to mitochondria in two human immortalized neuroblastoma cell lines, BE(2)-M17 and SH-SY5Y. Similarly, in differentiated striatal cells, endogenous PRKN did not translocate to mitochondria in either cell line upon mild or acute depolarization (Fig. S4A). In contrast, artificial overexpression of EFGP-PRKN under strong depolarizing conditions led to nearly all cells showing EGFP-PRKN translocation to the mitochondria (Fig. S4B). According to these results, and in agreement with other studies focused on neuronal mitophagy, endogenous PRKN levels in our differentiated cellular model seemed too low for clear detection of PRKN translocation to the mitochondria. Nevertheless, mitophagy still takes place since we still detected mitochondria engulfed by autophagosomes and fused with lysosomes (Figure 2A).

Next, we analyzed if, despite the absence of PRKN in mitochondria upon depolarization, the labeling of mitochondria with polyUB chains could still take place, probably by the action of alternative E3 UB ligases. To this end, we analyzed the polyUB labeling of total homogenate and isolated mitochondrial fractions detected with anti-polyUB antibody (Figure 5A). We observed no significant differences between control and mHTT samples for both homogenate and isolated mitochondrial fractions after depolarization. When studying only polyUB-p-S65 both in total homogenate and isolated mitochondria, we observed a tendency toward increased levels of this mitophagy-related labeling (Figure 5A) in both control and mHTT cells upon depolarization. These results indicate that although we could not detect PRKN translocation to mitochondria, ubiquitin chains are phosphorylated at Ser65 in the presence of both control and mHTT. Therefore, mitophagy is taking place in differentiated striatal cells through a UB-dependent mechanism, although this UB-labeling step seems to be PRKN-independent and HTT-independent. However, since other forms of mitophagy independent of ubiquitin labeling [4,60] could also be participating, we analyzed UB-independent mitophagy mediated by the receptors BNIP3, BNIP3L, and FUNDC1 to determine if they could also be assisting in the induced mitophagy in our cellular model. We analyzed the levels of these three receptors in mitochondria before and after depolarization. FUNDC1 protein could not be detected in striatal cell lines under basal conditions or after depolarization treatment (data not shown). We analyzed levels of BNIP3 and BNIP3L proteins in the isolated mitochondrial fractions (Figure 5B). We detected both proteins, as members of the BH3-only protein family, as monomers and oligomers. Quantification of total levels of these proteins did not increase after mitophagy induction by depolarization, suggesting that, under these conditions, these proteins are probably not participating in mitophagy. These findings are in accordance with other studies, where other stress-associated conditions like hypoxia and metabolic stress trigger BNIP3/BNIP3L/FUNDC1-mediated mitophagy, but not depolarization [4,61]. In conclusion, in striatal cells, mitophagy after depolarization seems to be UB-dependent (although without the intervention of PRKN); however, none of these steps seem to be impaired by the presence of mHTT.

The polyQ tract in mHTT affects the recruitment of mitophagy receptors during mitophagy

Once we had shown that despite the absence of PRKN, mitochondria are still labeled with the polyUB-p-S65, we analyzed the next step in mitophagy mediated by the UB tag. After the labeling of damaged mitochondria with polyUB-p-S65 chains, a set of receptors are able to recognize and bind the mitochondria and simultaneously recruit LC3-II to form an autophagosome around the mitochondria.

Previously, we and others showed that HTT could also interact with the selective autophagy receptor protein SQSTM1/p62 [18,19,21]. In addition to SQSTM1/p62, different protein receptors for selective mitophagy like CALCOCO2 (calcium binding and coiled-coil domain), OPTN (optineurin), NBR1 (NBR1, autophagy cargo receptor) and TAX1BP1 (Tax1 binding protein 1) have been identified [62]. In a recent study using the combined knockout of all five mitophagy receptors, it was shown that, although all of the receptors can intervene during mitophagy, only OPTN and CALCOCO2 are essential [37], with the others probably having redundant roles. First, and in agreement with previous data [18,19,21] and with the computer-based predictions, we used immunoprecipitation to confirm the interaction of HTT with the mitophagy receptors SQSTM1/p62, NBR1, OPTN, and CALCOCO2 (Fig. S3). Once we validated the interactions, we used PLA to investigate and quantify the efficiency of the interaction of HTT with the essential mitophagy receptors OPTN and CALCOCO2 (Figure 6A and B). The OPTN-HTT interaction became reduced after depolarization only when HTT was carrying the mutant polyQ tract (Figure 6A and S2). Moreover, we could not detect the CALCOCO2-HTT interaction in the presence of the mutant polyQ tract (Figure 6B). The lack of interaction between CALCOCO2 and mHTT was not due to the absence of CALCOCO2 protein but rather due to the total loss of interaction since CALCOCO2 protein could still be detected by immunofluorescence in mHTT cells before and after depolarization (Fig. S5). These results show that the interaction between the two essential mitophagy receptors and HTT is strongly influenced by the number of glutamines in the polyQ tract. In the case of CALCOCO2, the interaction is totally abolished when mHTT is present.

The polyQ tract in mHTT affects the interaction of LC3-II with mitophagy receptors

As the last step of substrate recognition during mitophagy, receptors interact with and recruit LC3-II, thereby allowing elongation of the autophagosome membrane around mitochondria. We used PLA to analyze the interaction between the two essential mitophagy receptors (OPTN and CALCOCO2) with LC3 and observed that, upon depolarization, the OPTN-LC3 and CALCOCO2-LC3 interactions increased in differentiated ST-Q7 cells (Figure 7A and 7B). Conversely, the presence of the mutant polyQ tract impaired the OPTN-LC3 interaction and decreased the CALCOCO2-LC3 interaction, probably as a consequence of the failure recruitment of receptors to the mitophagy site (Figure 6A and B). We also used PLA to examine interactions between the SQSTM1/p62 and NBR1 autophagy receptors with LC3 and observed the same effect; mHTT dramatically impaired SQSTM1/p62 and NBR1 interactions with LC3 (Figure 7C and D). These results show the impact of the presence of mHTT on the recruitment of LC3-II to the mitophagy initiation site, a key step to promote the elongation of the autophagosome membrane to engulf tagged mitochondria.

Reagents

Rotenone (R-8875), CCCP (carbonyl cyanide 3-chlorophenyl hydrazone, C2759), oligomycin (75351), antimycin A (A8674) and the lysosomal inhibitors chloroquine (C6628) and bafilomycin A₁ (B1793) were from Sigma-Aldrich. MitoTracker™ Deep Red FM (Molecular Probes, M22426), CM-H₂DCFDA (Life Technologies, C6827), Cell Rox Green Reagent (Thermo Fisher Scientific, C10444) and TMRM (Tetramethylrhodamine methyl ester, Sigma-Aldrich, T668). Sodium ortho-vanadate (Sigma-Aldrich, S6508), NaF (Sigma-Aldrich, 201154), sodium pyrophosphate (Sigma-Aldrich, 221368), sodium 2-glycerophosphate (Sigma-Aldrich, G9422), benzamidine (Sigma-Aldrich, 12072), PMSF (Sigma-Aldrich, P7626), iodoacetamide (Sigma-Aldrich, I1149), MG132 (Enzo BML, PI102).

Cell culture and treatments

Plasmid transfection

For plasmid transfection, Lipofectamine 3000 (Thermo Fisher Scientific, L3000015) was used. Cells were transfected with 0.8 µg (EGFP-PRKN) of DNA following the manufacturer’s instructions.

Immunofluorescence microscopy

Cells were fixed in 4% paraformaldehyde for 30 min at RT. Following the incubation with a blocking solution containing normal goat serum 3% (Vector Laboratories, S100), Triton X-100 0.1% (Sigma, T9284) in PBS (Thermo Fisher, 70013065) for 1 h at RT, the corresponding primary antibodies were used diluted in the same blocking solution overnight at 4°C. Then, cells were incubated 1 h at RT with the corresponding Alexa secondary antibodies, diluted in blocking solution. Nuclei were stained with 10 µM Hoechst 33342 (Invitrogen, H3570) in PBS. When indicated, 100 nM MitoTracker™ Deep Red FM prepared in Opti-MEM medium was added before proceeding with the protocol. Coverslips were mounted onto superfrost ultra plus slide (Menzel-Gläser, Ref.10417002) using DAKO Fluorescent Mounting Medium (DAKO, S3023). Images were acquired using an Olympus FV1000 confocal microscope (Tokyo, Japan). Intensity analysis was performed by drawing a region of interest to each cell and measuring its intensity using ImageJ 1.50a (NIH, USA).

Immunoblot detection

For biochemical studies, cells were detached and centrifuged at 800 g for 5 min at 4°C. The pellet was resuspended in PBS, centrifuged two more times, and resuspended in radioimmunoprecipitation assay buffer or RIPA buffer supplemented with proteases inhibitors (HALT Protease inhibitor cocktail EDTA free, Thermo Scientific, 1862209) and 1 mM PMSF (Sigma-Aldrich, P7626). For ubiquitin detection, cells where resuspended in UB-buffer (50 mM Tris/HCl pH 7.5, 1% Triton X-100, 1 mM sodium ortho-vanadate, 50 mM NaF, 5 mM sodium pyrophosphate, 0.27 M sucrose, 10 mM sodium 2-glycerophosphate, 0.2 mM PMSF, 1 mM benzamidine, 1 mM EGTA, 1 mM EDTA, 100 mM iodoacetamide, 25 mM MG132) in order to avoid deubiquitination. Homogenates were incubated on ice for 15 min and sonicated. Total protein concentration was determined by the bicinchoninic assay method with BSA as a standard protein. Cellular protein extracts were heated at 95°C for 5 min after adding loading buffer 6x. Proteins were resolved by SDS-PAGE on different percentages of polyacrylamide gels, ranging from 7% to 15%. Protein All Blue Standards ladder (Biorad, 161–0373) was used as a ladder. Resolved proteins were transferred onto nitrocellulose membranes (GE Healthcare, 10600002) or PVDF membranes (Whatman, 10485289) in the case of polyUB immunodetection, and blocked with 5% nonfat milk powder (Sigma-Aldrich, 70166) in PBS for 1 h at RT. Membranes were incubated with the corresponding primary antibodies diluted in 4% bovine serum albumin (Sigma-Aldrich, A4503) in PBS overnight at 4°C. Then, we proceeded to the incubation with the corresponding secondary antibodies coupled with horseradish peroxidase and diluted in 5% milk in PBS for 1 h at RT. Finally, proteins were visualized using either West Pico SuperSignal Substrate or SuperSignal West Femto (Thermo Fisher Scientific, 34080 and 34095, respectively) on an ImageQuant RT ECL imaging system (GE Healthcare). Immunoblots were quantified by densitometry using ImageJ 1.50a.

Mitochondrial enrichment

Mitochondria were isolated from striatal cells, as previously described on Frezza et al. [92]. Briefly, cells were washed with PBS, detached, and centrifuged at 600 g at 4°C for 5 min. The pellet was resuspended in ice-cold IBc buffer (10 mM Tris-MOPS [Sigma-Adrich, M5162], 1mM EGTA/Tris [Sigma-Aldrich, E4378] and 0.2 M sucrose [Sigma-Aldrich, S9378] in H₂O at pH 7.4). To homogenize, we used a Teflon pestle operated at 1600 rpm and stroked the cell suspension placed in a glass potter during 30–40 times. The cell suspension was centrifuged at 600 g for 10 min at 4°C, and the supernatant was further centrifuged at 7000 g for 10 min at 4°C. Pellet was washed with ice-cold IBc buffer and centrifuged again at 7000 g for 10 min at 4°C. Finally, the pellet containing mitochondria was resuspended in an adequate volume of IBc buffer, and mitochondrial protein concentration was determined by the bicinchoninic assay method with BSA as a standard protein.

Mitochondrial membrane potential

Differentiated ST-Q7 and ST-Q111 cells were incubated with 50 nM TMRM or 10 nM MitoTracker™ Deep Red FM probes for 30 min in a cell incubator. Cells were washed twice with pre-warmed PBS and recollected with a cell scraper in the medium. Using a FACSAria flow cytometer (BD Biosciences), 10,000 cells were acquired, and fluorescence detected with the indicated lasers and detectors. Membrane potential was analyzed as mean fluorescence intensity (MFI) and represented as fold-change compared to the corresponding controls using the FCS Express (v3 De Novo Software^TM).

Seahorse XF24 mitochondrial respiration analysis

Oxygen consumption rate (OCR) was measured using the Seahorse XF24 equipment (Seahorse bioscience Inc., North Billerica, MA, USA). ST-Q7 and ST-Q111 cells were seeded in Seahorse XF24 Cell Culture Microplate (Seahorse Biosciences, 100777–004), differentiated and treated with the corresponding treatments. Cells were rinsed once and incubated in 700 μl of XF assay buffer (DMEM without NaHCO₃, 2 mM glutamax; pH 7.4, 5 mM glucose, 1 mM sodium pyruvate), then equilibrated for 1 h at 37°C in a non-CO₂ incubator. All medium and solutions of mitochondrial complex inhibitors were adjusted to pH 7.4 on the day of assay. Following three baseline measurements of OCR, mitochondrial complex inhibitors were sequentially injected into each well. Three OCR readings were taken after the addition of each inhibitor and before the automated injection of the subsequent inhibitor. Mitochondrial complex inhibitors, in order of injection, included oligomycin (1 μM), CCCP (0.5 μM each time), antimycin A (1 μM), and rotenone (1 μM). OCR was automatically calculated, recorded, and plotted by Seahorse XF24 software version 1.8 (Seahorse Bioscience, Billerica, MA, USA). At the end of each assay, cells were lysed with ice-cold RIPA buffer and the protein content estimated by the bicinchoninic assay method with BSA as a standard protein.

Quantification of reactive oxygen species (ROS) by fluorescent probes

Cells were washed twice with pre-warmed PBS and either 50 μM of CM-H₂DCFDA probe or 50 μM CellROX was added in medium and incubated for 30 min in a cell incubator. Using a FACSAria flow cytometer, 10,000 cells were acquired, and fluorescence was read at 488 nm laser with a 530/30 emission filter. ROS was analyzed as mean fluorescence intensity (MFI) and represented either as MFI or as a percentage compared to the corresponding controls using the FCS Express v3 software. For immunofluorescence analysis, 5 μM of CellROX was added to the medium and incubated for 30 min in a cell incubator, before proceeding with the immunofluorescence protocol. Images were acquired using an Olympus FV1000 confocal microscope.

Mitophagy flux assay

Differentiated ST-Q7 and ST-Q111 cells were treated with the corresponding reagents (vehicle, rotenone, CCCP, chloroquine). Cells were incubated with 10 nM MitoTracker™ Deep Red FM probe for 30 min in cell incubator before harvesting. Cells were washed twice with pre-warmed PBS and collected with a cell scraper in the medium. Using a FACSAria flow cytometer, 10.000 cells were acquired and fluorescence was read at 640 nm red laser with a 660/20 emission filter. Mitophagy flux was analyzed as previously described [93] as a mean fluorescence intensity (MFI) in the presence and absence of 60 μM chloroquine using the FCS Express v3 software.

Immunoprecipitation

Differentiated ST-Q7 and ST-Q111 cells were harvested in PBS using and centrifuged at 800 g at 4ºC for 5 min. Pellet was resuspended in IP buffer containing 0.01% Triton in PBS and protease inhibitors. 1:2 mixture of Dynabeads protein A and protein G (Thermo Scientific, 10,001 and 10,003, respectively) was incubated with the corresponding antibody, together with 5 mM BS3 (bis[sulfosuccinimidyl]suberate, Sigma-Aldrich, S5799), to crosslink the antibody to the Dynabeads, in IP buffer and let rocking 30 min at RT. The supernatant was discarded using a magnetic particle concentrator (DynaMagTM-2, Invitrogen, 4211) and BS3 was inactivated with 50 mM Tris-HCl pH 7.4 for 15 min. Dynabeads were washed three times with IP buffer and incubated with the corresponding cell lysate for 90 min at 4ºC. After collecting the flow-through, Dynabeads were washed three times with IP buffer, and samples were incubated in elution buffer containing 0.2 M glycine pH 2.5, LB6x and 16 mM dithiothreitol (DTT) 10 min at 70ºC and immunoprecipitated samples were collected. Immunoprecipitates (IP), whole-cell extracts (input), and flow-through were analyzed by standard immunoblotting.

Proximity ligation assay (PLA)

PLA was performed in 3% formaldehyde-fixed cells (Sigma-Aldrich, F1635) following the manufacturer’s instructions. Briefly, samples were blocked for 30 min at RT and incubated with specific primary antibodies overnight. Secondary antibodies anti-mouse PLUS and anti-rabbit MINUS conjugated with oligonucleotides were added to the reaction (Duolink PLA Probe anti-mouse PLUS and anti-rabbit MINUS, Duolink, Sigma-Aldrich, DUO92001 and DUO92005, respectively) and incubated for 1 h at 37°C. Ligation reagents were added and incubated for 30 min at 37°C. Amplification solution containing nucleotides and fluorescently-labeled oligonucleotides, and polymerase was added and incubated for 100 min at 37°C (Detection Reagent RED, Duolink; Sigma-Aldrich, DUO92008). The proximity ligation signal is observed as fluorescence spots and was analyzed using an Olympus FV1000 confocal microscope. The corresponding negative control experiments were performed under identical experimental conditions. Quantification was performed by analyzing the number of dots per cell and the area fraction that these dots occupy per cell using ImageJ 1.50a.

Statistical analysis

The values were expressed as the mean ± standard error of the mean (SEM). The significant differences (*p < 0.05, **p < 0.01, ***p < 0.001) when comparing two groups were determined by a two-tailed unpaired Student’s t-test. When comparing more than two groups, significant differences were determined by one-way or two-way ANOVA followed by Tukey’s post hoc test. Statistical analyzes were performed using GraphPad Prism 6.