Phosphorylation of Optineurin by protein kinase D regulates Parkin-dependent mitophagy

Abstract

Degradation of damaged mitochondria, a process called mitophagy, plays a role in mitochondrial quality control and its dysfunction has been linked to neurodegenerative pathologies. The PINK1 kinase and the ubiquitin ligase Parkin-mediated mitophagy represents the most common pathway in which specific receptors, including Optineurin (Optn), target ubiquitin-labeled mitochondria to autophagosomes. Here, we show that Protein Kinases D (PKD) are activated and recruited to damaged mitochondria. Subsequently, PKD phosphorylate Optn to promote a complex with Parkin leading to enhancement of its ubiquitin ligase activity. Paradoxically, inhibiting PKD activity enhances the interaction between Optn and LC3, promotes the recruitment of Parkin to mitochondria, and increases the mitophagic function of Optn. This enhancement of mitophagy is characterized by increased production of mitochondrial ROS and a reduction in mitochondrial mass. The PKD kinases may therefore regulate Optn-dependent mitophagy by amplifying the Parkin-mediated degradation signals to improve the cell response against oxidative stress damage.

Keywords: Cell biology; Molecular biology.

Graphical abstract

Figure 1

Optineurin is a substrate of protein kinases D (A) Alignment of PKD phosphorylation consensus motif LXRXXS with that of known human protein substrates of PKD kinases (Rin1 - Ras and Rab Interactor 1, HDAC5 - Histone Deacetylase 5, Hsp27 - Heat Shock Protein 27) and potential targeted Ser residues at position 342 and 415 of the human Optn protein are shown. (B) Optn was immunoprecipitated from HEK293T cells overexpressing VSV-tagged Optn wt or its S342A mutated form in the presence of constitutively active PKD (PKD-CA) as indicated. Immunoprecipitates were resolved on SDS-PAGE and analyzed by Western blot using specific anti-pS342 Optn antibodies, anti-VSV (control) and anti-PKD antibodies. (C) Phosphorylation status of Optn (on S342) and PKD (on S916) were determined by Western blot in uninduced HeLa cells overexpressing VSV-Optn and after stimulation by oxidative stress (H₂O₂) or by PKD activators: PMA and ionomycin. Quantification of the signals acquired from 2 western blot experiments is depicted in the adjacent graph as the ratio of phosphorylated Optn to total Optn protein levels. Data are represented as average +/− standard deviation. (D) Schematic representation of human Optn protein showing its structural domains: coiled-coil regions (CC), LC3-Interacting Region (LIR), Leucine zippers (LZ), zinc fingers (ZF), putative helical domains (HLX) and Ub-binding domain. Localization of identified PKD interaction domain as well as PKD phosphorylation sequence LERKNS are shown. Conserved Leucine and Arg residues of the PKD consensus site and potential phosphorylated serine 342 residue are indicated in blue and red, respectively. (E) Distribution of GFP-tagged PKD isoforms and Optn were monitored by immunofluorescence microscopy in untreated HeLa cells. Bars = 10 μm. Preferential localization of Optn and PKD is indicated by arrows.

Figure 2

Protein kinases D are activated and recruited to damaged mitochondria during mitophagy (A) Total cell lysates from HeLa cells expressing empty vector or BFP-tagged Parkin harvested at different time after mitophagy induced by the combination of oligomycin and antimycin A (OA), were analyzed by Western blot using anti-pS916 PKD, PKD, Parkin and p62 antibodies. The quantification of the ratio of phosphorylated PKD to total PKD protein levels obtained from 3 independent western blot experiments is illustrated in the graph below. Data are represented as average +/− standard deviation. (B) Changes in localization of GFP-tagged PKD1 were monitored by immunofluorescence microscopy in non-induced (NI) HeLa cells stably expressing Parkin or following OA treatment for 2, 4 or 6h. Mitochondria were labeled using MitoTracker. Bars = 10 μm. Arrows indicate round shaped structures resembling autophagosomes, surrounding or located close to mitochondria aggregates.

Figure 3

The S342 phosphorylation regulates Optn localization relative to Parkin (A) Total cell lysates from HeLa cells expressing BFP-tagged Parkin and transfected with VSV-Optn were harvested at different time after mitophagy induction by OA and analyzed by Western blot using anti-pS342 Optn, Optn and p62 antibodies. Quantification of the signals acquired from 3 independent western blot experiments is depicted in the adjacent graph as the ratio of phosphorylated Optn to total Optn protein levels. Data are represented as average +/− standard deviation. ∗p < 0.05 (OA vs. NI: 2h: n = 3, Δ = −3.40, t = −11.9, ddl = 2.33, p = 0.013; 4h: n = 3, Δ = −2.26, t = −10.09, ddl = 2.56, p = 0.014; 6h: n = 3, Δ = −2.963, t = −11.29, ddl = 2.40, p = 0.013; 8h: n = 3, Δ = −2.032, t = −9.593, ddl = 2.63, p = 0.015). (B) Localization of Optn wt, phosphodeficient (S342A and S177A) or phosphomimetic mutants (S342D and S177D) of Optn and Parkin relative to mitochondria, were monitored by immunofluorescence microscopy in HeLa cells that were either left uninduced (NI) or stimulated by CCCP (Carbonyl cyanide m-chlorophenylhydrazone, respiratory chain uncoupler). Bars = 10 μm. Images were analyzed by quantification of fluorescence intensity (arbitrary unit) of each channel over the distance indicated using line scans (Fiji software). (C) The average statistical distance between Optn and Parkin was determined on images such as those presented in B, using the SODA (“Statistical Object Distance Analysis”) statistical tool of the ICY software (Quantitative Image Analysis Unit, Institut Pasteur). The coupling percentage was calculated when both proteins are recruited to the surface of the mitochondria during mitophagy and is represented as a function of the distance (in pixels) between two fluorescent points. Data are represented as average +/− standard deviation. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (S342A vs. WT, 1 pixel: n = 6, Δ = 1.67, t = 5.80, ddl = 9.61, p = 0.006; 2 pixels: n = 6, Δ = 3.557, t = 6.323, ddl = 6.80, p = 0.009; 4 pixels: n = 6, Δ = 7.03, t = 5.22, ddl = 5.28, p = 0.044; 5 pixels: n = 6, Δ = −6.3967, t = −28.665, ddl = 5.00, p < 0.001; S177A vs. WT, 1 pixel: n = 6, Δ = 4.27, t = 14.84, ddl = 9.61, p < 0.001; 2 pixels: n = 6, Δ = 8.183, t = 14.546, ddl = 6.80, p < 0.001; 4 pixels: n = 6, Δ = 8.471, t = 6.29, ddl = 5.28, p = 0.019).

Figure 4

Parkin and Optn form a complex regulated by PKD kinases (A) Optn was immunoprecipitated from HeLa cells overexpressing Parkin and transiently transfected with PKD1, as indicated. After transfection, cells were left untreated or treated with OA for 6h before lysis. Immunoprecipitates and Input (10% of the lysate used to perform the IP) were resolved on SDS-PAGE and analyzed by western blot using anti-Parkin, anti-VSV (control) and anti-PKD antibodies. The molecular masses (kDa) are represented on the left of each immunoblot. (B and C) HeLa cells expressing Parkin were transiently transfected with VSV-tagged Optn wt or Optn-S342A, as indicated. After transfection, cells were treated with PKDi and OA, as indicated, lysed, and subjected to VSV (B) or Parkin (C) immunoprecipitation. Immunoprecipitates and Input (total lysate) were resolved on SDS-PAGE and analyzed by western blot using anti-Parkin and anti-VSV (control) antibodies. The molecular masses (kDa) are represented on the left of each immunoblot. (D) Quantifications of the immunoprecipitation signals obtained from 2 independent experiments, as shown in B (left panel) and C (right panel), respectively, are illustrated as the ratio of Parkin/Optn following immunoprecipitation to total protein levels detected in total cell extracts (Input). Data are represented as average +/− standard deviation. ND, not determined. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (in D: PKDi vs. wt n = 2, Δ = 0.708, t = 10.3, ddl = 3.00, p = 0.004; S342A vs. wt n = 2, Δ = 1.041, t = 15.11, ddl = 3.00, p = 0.001; in E: PKDi vs. wt n = 2, Δ = 0.515, t = 4.82, ddl = 2, p = 0.001).

Figure 5

PKD inhibition enhances Optn-LC3 complex formation Representative images of Optn-LC3 complexes visualized by in situ proximity ligation assays (PLA) during mitophagy are shown HeLa cells expressing BFP-Parkin were pretreated with PKDi or left untreated as indicated and induced with OA for 4 and 8 h. Images were acquired using a confocal spinning-disk microscope. Bars = 10 μm. Analyses were conducted on five images from two independent experiments, each containing 10–15 cells identified by the detection of actin protein using Alexa Fluor 488 Phalloidin. The number of Optn-LC3 PLA dots per cell was determined using Cell Profiler software, and statistical significance was determined. Data are represented as boxplot. n.s., not significant. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (PKDi n = 59 vs. NI n = 71: Δ = −0.494, t = −0.315, ddl = 117 p = 1.000; PKDi+OA4h n = 32 vs. OA4h n = 47: Δ = −11.0, t = −3.90, ddl = 68.0, p = 0.003; PKDi+OA8h n = 53 vs. OA8h n = 38: Δ = −13.83, t = −3.96, ddl = 87.6, p = 0.002).

Figure 6

Inhibition of PKD kinases affects Parkin activity (A and B) HeLa cells stably expressing Parkin and transiently expressing HA-Ubiquitin were treated with OA, following 45 min pretreatment with PKDi as indicated, and lysed. Ubiquitin-modified complexes were immunoprecipitated from lysate. Immunoprecipitates and Input (total lysate) were resolved on SDS-PAGE and analyzed by Western blot using specific anti-pS65 Ub antibodies (A) or anti-panUb antibodies (B). (C) HeLa cells overexpressing Parkin and transiently transfected with VSV-tagged Optn wt or its S177A, S342A or D474N mutated forms were treated with PKDi and/or OA and lysed. Total lysate was analyzed by Western blot using anti-pS65 Ub and Optn antibodies. Phosphorylated S65-Ubiquitin levels were quantified in stimulated conditions (OA 2h) from 2 independent western blot experiments and depicted in the adjacent graph as the ratio of phosphorylated-S65Ub to actin protein levels presented in Figure S5B. Data are represented as average +/− standard deviation. n.s., not significative. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (WT: OA + PKDi vs. OA n = 2, Δ = 0.734, t = 6.63, ddl = 8.00, p = 0.002; S177A: OA + PKDi vs. OA n = 2, Δ = 0.946, t = 8.54, ddl = 8.00, p < 0.001; S342A: OA + PKDi vs. OA n = 2, Δ = −0.411, t = −3.71, ddl = 8.00, p = 0.069; D474N: OA + PKDi vs. OA n = 2, Δ = 0.0973, t = 0.879, ddl = 8.00, p = 0.980). (D) Total lysates from HeLa cells overexpressing Parkin pre-treated with PKDi or BX795 (Bx, TBK1 inhibitor) and harvested at different time after mitophagy induction were analyzed by Western blot using anti-pS65 Parkin, Parkin, pS342 Opn, p62, S65-Ub and Actin antibodies.

Figure 7

Protein kinase D inhibition affects mitophagy (A) HeLa cells expressing mitochondria-targeted Keima and BFP-Parkin were pre-treated for 45 min with PKDi or BX795 (Bx) before induction with OA for 1 and 3 h. The analysis was conducted by flow cytometry, and the percentages of cells positive for Keima 561nm (pH4) are showcased using a boxplot graph. Statistical analyses were conducted on 9 independent replicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (PKDi vs. untreated: 1h n = 9, Δ = −7.32, t = −6.00, ddl = 73.0, p < 0.001; 3h n = 9, Δ = −18.3, t = −15.0, ddl = 73.0, p < 0.001; Bx vs. untreated: 1h n = 9, Δ = 0.182, t = 0.149, ddl = 73.0, p = 1.000; 3h n = 9, Δ = 16.796, t = 13.779, ddl = 73.0, p < 0.001). (B) HeLa cells expressing mitochondria-targeted Keima and BFP-Parkin were transfected with control siRNA (siNT) or siRNA directed against the three isoforms of PKD (siPKD1/2/3) or TBK1 and left 72h before the 3 h induction with OA. The analysis was conducted via flow cytometry, and the percentages of cells positive for Keima 561nm (pH4) are showcased using a boxplot graph. Statistical analyses were conducted on 8 independent replicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (siPKD vs. siNT: NI n = 8, Δ = −10.2, t = −4.49, ddl = 42.0 p < 0.001; OA n = 8, Δ = −16.4, t = −7.17, ddl = 42.0, p < 0.001; siTBK vs. siNT: NI n = 8, Δ = 2.88, t = 1.26, ddl = 42.0, p = 0.804; OA n = 8, Δ = 8.007, t = 3.508, ddl = 42.0, p = 0.013). (C) HeLa cells expressing mitochondria-targeted Keima, BFP-Parkin with either VSV-tagged Optn wt, Optn-S342A or PKD1 were harvested before and after stimulation by OA. The analysis was conducted by flow cytometry, and the percentages of cells positive for Keima 561nm (pH4) are showcased using a boxplot graph. Statistical analyses were conducted on 8 independent replicates. n.s. not significative ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (PKDi vs. untreated: NI n = 8, Δ = −0.716, t = −0.377, ddl = 56.0 p = 1.000; OA n = 8, Δ = 7.71, t = −4.06, ddl = 56.0, p = 0.004; S342A vs. WT: NI n = 8, Δ = −2.67, t = −1.41, ddl = 56.0, p = 0.851; OA n = 8, Δ = −14.2, t = −7.49, ddl = 56.0, p < 0.001). (D) Parental and triple knock-out (TKO, deficient for Optn, NDP52 and Tax1BP1) HeLa cells both expressing BFP-Parkin and mtKeima were transfected with Optn wt and pre-treated for 45 min with PKDi before induction by OA as indicated. The analysis was conducted by flow cytometry, and the percentages of cells positive for Keima 561nm (pH4) are showcased using a boxplot graph. Statistical analyses were conducted on 10 independent replicates. n.s. not significative ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (WT-PKDi vs. WT: OA n = 10, Δ = −10.0, t = −4.19, ddl = 72.0 p = 0.002; OA + Optn n = −22.0, Δ = 7.71, t = −9.18, ddl = 72.0, p < 0.001; TKO-PKDi vs. TKO: OA n = 10, Δ = −3.67, t = −1.53, ddl = 72.0, p = 0.786; OA + Optn n = 10, Δ = −15.08, t = −6.3038, ddl = 72.0, p < 0.001).

Figure 8

PKD inhibition enhances mitochondrial ROS production and reduces the global mitochondrial mass (A) Measurement of mitochondrial ROS production in HeLa cells expressing BFP-Parkin, pretreated with PKDi or Bx (Bx795) and induced with OA as indicated. Cells were incubated with 500 nM MitoSOX for 30 min and subjected to cytometry analysis. Quantification of MitoSOX fluorescence was performed in gated positive Parkin expressing cells by cytometry and data analyzed using FlowJo. Data are represented as boxplot. Statistical analyses were conducted on 9 independent replicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (PKDi: n = 9 NI: Δ = −29.7, t = −1.90, ddl = 16.0, p = 0.877; OA30min: Δ = −343, t = −16.82, ddl = 15.0, p < 0.001; OA1h: Δ = −436, t = −15.11, ddl = 13.4, p < 0.001; OA2h: Δ = −317.56, t = −10.274, ddl = 12.8, p < 0.001; OA3h: Δ = −326.89, t = −9.731, ddl = 12.4, p < 0.001; OA4h: Δ = −305.1, t = −9.290, ddl = 12.3, p < 0.001; Bx: n = 9 NI: Δ = −7.89, t = −496, ddl = 16.0, p = 1.000; OA30min: Δ = −171.6, t = −9.03, ddl = 15.7 p < 0.001; OA1h: Δ = −239, t = −10.665, ddl = 15.9, p < 0.001; OA2h: Δ = −193.1, t = −7.21, ddl = 14.4, p < 0.001; OA3h: Δ = −123.00, t = −4.7121, ddl = 15.2, p = 0.018; OA4h: Δ = −84.8, t = −3.15, ddl = 14.4, p = 0.250). Mitochondrial membrane potential (B) and total mitochondria mass (C) were measured in SHSY5Y cells pre-treated with PKDi or CRT and harvested at different time after stimulation with OA as indicated. Cells were incubated with MitoTracker Green FM or MitoSOX for 30 min before FACS analysis. Quantification of the TRMR and MitoTracker signals was performed by flow cytometry, and the Mean Fluorescent Intensities (MFI) of MitoTracker and MitoSOX are showcased using a boxplot graph. Statistical analyses were conducted on 6 independent experiments by performing a pairwise comparison of the same time points of mitophagy induction in the presence or absence of PKDi or CRT. Data are represented as boxplot. MitoTracker MFI ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (PKDi: n = 9 NI: Δ = −30.6, t = −0.866, ddl = 18.0, p = 1.000; AO 30min: Δ = 888, t = 18.75, ddl = 15.4, p < 0.001; AO 1h: Δ = 1188, t = 31.40, ddl = 14.9, p < 0.001; AO 2h: Δ = 422.0, t = 15.49, ddl = 17.2, p < 0.001; AO 3h: Δ = 273.3, t = 6.567, ddl = 13.0, p < 0.001; OA 4h: Δ = 510, t = 25.6, ddl = 12.69, p < 0.001; CRT: n = 9 NI: Δ = −373, t = −12.5, ddl = 16.2, p < 0.001; OA 30min: Δ = 709, t = 26.1, ddl = 11.17, p < 0.001; OA 1h: Δ = 568, t = 23.0, ddl = 16.82, p < 0.001; OA 2h: Δ = 228.9, t = 9.142, ddl = 15.08, p < 0.001; OA 3h: Δ = 835, t = 35.9, ddl = 17.16, p < 0.001. OA 4h: Δ = 608.4, t = 27.676, ddl = 15.97, p < 0.001; MitoSOX MFI ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 (PKDi: n = 9 NI: Δ = −30.8, t = −13.00, ddl = 10.33, p < 0.001; AO 30min: Δ = −176, t = −56.8, ddl = 11.89, p < 0.001; AO 1h: Δ = −237.8, t = −91.4, ddl = 13.27, p < 0.001; AO 2h: Δ = −316.2, t = _39.87, ddl = 8.62, p < 0.001; AO 3h: Δ = −339.9, t = −61.80, ddl = 9.38, p < 0.001; AO 4h: Δ = −359.2, t = −65.63, ddl = 14.45, p < 0.001; CRT: n = 9 NI: Δ = −9.56, t = −5.94, ddl = 13.41, p = 0.003; OA 30min: Δ = −60.67, t = −36.002, ddl = 13.94 p < 0.001; OA 1h: Δ = −62.33, t = −27.28, ddl = 9.58, p < 0.001; OA 2h: Δ = −122.3, t = −58.6, ddl = 15.89, p < 0.001; OA 3h: Δ = −95.11, t = −31.45, ddl = 13.08, p < 0.001. AO 4h: Δ = −77.78, t = −21.37, ddl = 12.59, p < 0.001).

Figure 9

PKD inhibition enhances Parkin recruitment to damaged mitochondria (A) Live imaging microscopy experiments were conducted using HeLa cells expressing BFP-Parkin and mitochondrial Keima (red signal). Representative images were captured every 30 min for up to 2 h in the absence or presence of a PKD inhibitor. Images were acquired every 15 min over a total of 4 h using a confocal spinning-disk microscope outfitted with a 100× objective lens. Bars = 10 μm. Parkin recruitment to damaged mitochondria was quantified by calculating the Pearson coefficient between Parkin and Keima signals. Quantification was carried out on data collected from five independent experiments for each condition. Data are represented as average +/− standard deviation. (B) Schematic model of the Optn-dependent mitophagy process. Following mitochondrial stress, PINK1 kinase activates Parkin through phosphorylation (pSer65-Parkin) and also phosphorylates ubiquitin chains (phospho-Ser65-Ubiquitin) on the mitochondrial surface. Accumulation of phospho-Ser65-Ubiquitin on damaged mitochondria can promote retention of pSer65-Parkin that can in turn provides additional ubiquitin molecules for phosphorylation by PINK1, creating a feedforward mechanism. Our data suggest that, when a local mitochondrial ROS threshold is reached, activation of the PKD kinases can enhance Parkin activity through their ability to favor Parkin-Optn complexes. Paradoxically, we found that inhibition of PKD activity leads to higher mitophagy efficiency as depicted by the red arrow. We hypothesize that formation of Optn-Parkin complexes might prevent interaction of Optn and LC3 by steric hindrance (the identified PKD-Optn interaction region encompassing the LIR domain of Optn), leading to reduced or delayed autophagosome formation. Due to the transient nature of PKD-mediated phosphorylation of Optn, disrupting the Optn-Parkin complex over time could promote the interaction between Optn and LC3. Subsequently, TBK1 kinase, activated and recruited during mitophagy induction, increases the affinity of Optn for the autophagosomal marker family members LC3 in order to elongate and mature the autophagosome vesicle, completing the entire mitophagy process.