Introduction
Parkinson’s disease (PD) is after Alzheimer’s disease the second most frequent neurodegenerative disease, with both syndromes being probably triggered by a combination of genetic predisposition and cumulative effects of environmental and physiological stressors, manifesting progressively with old age [
1]. The resulting clinical movement deficit in PD is largely explained by the selective atrophy of dopaminergic neurons in the midbrain. The accumulation of ubiquitinated cytoplasmic aggregates (“Lewy bodies”) containing proteins such as α-synuclein and the appearance of oxidative stress in the affected tissues are established hallmarks of PD pathology. Rodents treated with the neurotoxins 6-OH-dopamine or the mitochondrial inhibitor 1-methyl-4-phenylpyridinium (MPP
+) exhibit characteristic disease symptoms and were crucial to establish therapies. Over the past decade, the identification of more than a dozen mutant genes responsible for monogenic PD syndromes [
2,
3] and another dozen PD candidate genes with significant risk association in GWAS studies [
4‐
6] has greatly advanced our knowledge about pathogenesis, but the function of most corresponding gene products is still enigmatic and the pathways in common for both autosomal dominant and recessive PD variants remain elusive.
The
PARKIN gene encoded at the PARK2 locus was identified particularly early as a frequent cause of young-onset autosomal recessive PD, and its protein sequence clearly codes for an ubiquitin E3 ligase [
7]. Initial assumptions about its role for polyubiquitination and proteasomal protein degradation [
8] are now being challenged, and its relevant substrates and its relevant loss of function are being discussed controversially, e.g., regarding mitochondrial DNA damage, autophagy, and apoptosis [
9‐
13]. One isolated report with particularly convincing data demonstrated a role of PARKIN for receptor tyrosine kinase (RTK) endocytosis and signaling, implicating it in growth regulation [
14].
PARKIN expression can be upregulated in dependence on ATF4, a transcription factor triggered by mitochondrial stress and the unfolded protein response [
15].
Some PARKIN functions were shown to be downstream of Pten-induced protein kinase 1 (PINK1), a mitochondrial outer membrane kinase encoded at the autosomal recessive PARK6 locus [
16,
17]. PINK1 provides protection against ROS-induced damage, and its mRNA level increases with progressive cell age [
18]. Furthermore, PINK1 is transcriptionally induced in response to stress via FOXO3a and is stabilized against proteolytic degradation when the mitochondrial membrane potential dissipates [
19,
20]. In
Drosophila melanogaster the loss of PINK1 and PARKIN results in increased mitochondrial fission [
21‐
23]. The role of PINK1 in mitochondrial dynamics of mammalian cells, however, is more complex and controversial as reviewed recently [
24]: PINK1 was described to act as a fission factor [
23] as well as a fusion factor [
25,
26], while our PINK1-knockout mouse model demonstrated no effect on the mitochondrial morphology of unstressed cells [
27]. PINK1 and PARKIN appear to act together also in the degradation of damaged mitochondria (mitophagy) [
28‐
31]. It was proposed that the accumulation of PINK1 protein in dysfunctional mitochondria is a prerequisite for the relocalization of PARKIN from the cytosol to the mitochondrial outer membrane [
19,
28,
29]. There is ongoing debate whether this is caused via direct phosphorylation and which other proteins interact with PINK1 and PARKIN [
29,
32,
33]. It is also unclear which environmental and physiological stress events play the prominent role for the PINK1/PARKIN pathway. The relevance of stress was particularly investigated for the polyubiquitin–proteasome degradation of proteins [
17,
27,
34,
35], but recently, the mutation effects could be maximized when potent mitochondrial uncouplers like CCCP were used as stressor [
36].
In view of the regulation of PINK1 in response to a more physiological challenge namely cytokine deprivation [
20], we were interested whether the physiological competition for nutrients and growth factors which control autophagy [
37] has a role in the regulation of PARKIN. A previous study reported PARKIN expression to be downregulated in SH-SY5Y neuroblastoma cells by proteasomal stress, but unchanged after oxidative stress and ER stress, while an independent study observed PARKIN upregulation in some neuroblastoma cell lines after ER stress [
38,
39]. Our observations indicate that trophic deprivation and nutrient starvation cause distinct regulations of PARKIN and PINK1 in neuronal cells.
Material and methods
Bafilomycin A1, rapamycin (all Sigma-Aldrich, Taufkirchen, Germany) and LY294002 (Jena Bioscience, Jena, Germany) stocks were prepared with DMSO and stored at −20°C until use. All cell culture media, l-glutamine, penicillin G–streptomycin solution, and minimal essential medium with Earle’s salts (MEM) nonessential amino acids were purchased from Invitrogen (Karlsruhe, Germany), DMSO and PBS were from Sigma-Aldrich (Taufkirchen, Germany).
Cell culture
SH-SY5Y cells were purchased from The European Collection of Cell Cultures (Sigma-Aldrich, Taufkirchen, Germany) and cultured in Roswell Park Memorial Institute (RPMI) 1,640 medium containing 2 g/l
d-glucose, 2 mM
l-glutamine and 10% FCS. HeLa cells were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (Braunschweig, Germany, DSM-No. ACC57) and were cultured in MEM supplemented with 10% FCS and 1% MEM nonessential amino acids. Primary cortical neurons were isolated from 1- to 4-day-old mice as described before [
27] and were kept in Neurobasal medium with 10 mM
l-glutamine for 10–20 days before performing experiments. Mouse embryonic fibroblasts (MEF) were isolated from mouse embryos with standard procedures and cultivated in DMEM with 4.5 g/l
d-glucose supplemented with 15% BGS (Fisher Scientific, Schwerte, Germany). All cells were kept at 37°C with 5% CO
2 and 95% air.
Starvation experiments
SH-SY5Y cells were starved of trophic factors by incubating them in their normal growth medium (RPMI 1640 containing 2 g/l d-glucose and 2 mM l-glutamine) but without FCS (serum starvation). The starvation effect was maximized by incubating SH-SY5Y cells and primary neurons in HBSS (Invitrogen, nutrient starvation). HBSS contains 1.26 mM CaCl2, 0.493 mM MgCl2 × 6H2O, 0.407 mM MgSO4 × 7H2O, 5.33 KCl, 0.441 KH2PO4, 4.17 mM NaHCO3, 137.93 mM NaCl, 0.338 mM Na2HPO4, 0.0266 mM Phenol Red, and 5.56 mM (1 g/l) d-glucose but no amino acids.
To analyze mRNA levels by real-time reverse transcriptase quantitative PCR (qPCR), 0.5 × 106 SH-SY5Y cells were seeded in a six well, while 2.5 × 106 SH-SY5Y cells were seeded in a 10-cm dish to investigate protein expression, always 20 h prior to experimental start. Medium was removed, cells washed with PBS and covered with 2 ml (for RNA) or 10 ml (for protein) of control medium or starvation medium as indicated in the figure captions. Neurons were starved without washing them with PBS. Afterwards cells were incubated for the appropriate time at 37°C, 95%, air and 5% CO2. Neuron starvation experiments were performed in two to four individual neuron cultures for each genotype and time point.
RNA isolation, cDNA synthesis, and quantitative real-time PCR
Supernatant medium was collected and spun down (5 min, 500×
g, 4°C). Cell pellets were combined with adherent cells, and both were washed twice with PBS. Lysis of cells and isolation of total RNA was carried out utilizing the RNeasy mini kit (QIAGEN, Hilden, Germany). For first-strand synthesis, 1 μg of total RNA was digested with DNase I and reverse transcribed with SuperScript III reverse transcriptase utilizing oligo(dT)
20 and random primers (all Invitrogen, Karlsruhe, Germany). Transcript changes of 30 ng cDNA per sample were analyzed in a 20-μl reaction volume with qPCR in a StepOnePlus Real-Time PCR System and the appropriate TaqMan gene expression assays (all Applied Biosystems, Darmstadt, Germany):
SGK1 (Hs00178612_m1),
FOXO3a (Hs00921424_m1),
PINK1 (Hs00260868_m1), and
PARKIN (Hs01038318_m1). Mean of expression changes was normalized to mean of TATA box-binding protein (TBP: Hs99999910_m1) as an internal “housekeeping” control. For analysis of mRNA levels from mouse neurons, TaqMan gene expression assays
pink1 (Mm00550827_m1),
parkin (Mm00450186_m1), and
tbp (Mm00446973_m1) were used. Relative expression changes were calculated with the 2
−ΔΔCt method [
40] utilizing Microsoft Excel 2007 software, whereupon the ΔCt of the corresponding control served as calibrator. The resulting 2
−∆∆Ct values of
n experiments (indicated in the figures) were averaged for the appropriate time and treatment.
SDS-PAGE and western blotting
To determine PARKIN and LC3 protein expression, adherent cells were collected by scraping in 5 ml PBS supplemented with Complete EDTA free protease inhibitor (PI) cocktail (Roche Diagnostics GmbH, Mannheim, Germany), centrifuged for 5 min at 500×
g, 4°C and combined with floating cells, which were likewise spun down. Cell pellets were washed twice with PBS-PI (5 min, 4°C). The resulting cell pellets were frozen in liquid nitrogen and stored at −80°C until further analysis. Proteins were extracted with 2× SDS lysis buffer (137 mM Tris/HCl pH 6.8, 4% SDS, 20% glycerol) and freshly added PhosSTOP phosphatase inhibitor and PI cocktail (all Roche Diagnostics GmbH, Mannheim, Germany). Samples were sonicated (5 s, 3 cycles × 10%, 40% power) with a Sonopuls UW2070 ultrasonic homogenizer to advance cell lysis, to destroy DNA and to homogenize the samples. Afterwards samples were spun down (20 min, 16,100×
g, RT). Subsequently protein concentration was determined with the BC assay protein quantitation kit (VWR, Darmstadt, Germany). To investigate PARKIN expression, 30 μg protein per sample, prepared with 6× Laemmli sample buffer, were boiled for 5 min at 95°C and separated on a 10% SDS gel. Proteins were transferred to nitrocellulose membrane (Whatman, Dassel, Germany) by wet blotting (50 V, 90 min). Membranes were blocked for 1 h with 5% skim milk powder/TBS–Tween (0.05%) and probed with primary monoclonal (clone PRK8) mouse anti-PARKIN [
41] antibody (New England Biolabs, Frankfurt, Germany) at titer 1:1,000 overnight at 4°C. Specificity of anti-PARKIN antibody was controlled utilizing PARKIN-knockout (ko) and wild-type (wt) mouse brain samples. Quantification of PARKIN and equal protein loading was controlled by normalization to GAPDH (monoclonal (6C5) antimouse, 1:15,000; Merck, Darmstadt, Germany). Primary antibodies were detected with IRDye 800CW conjugated goat antimouse IgG secondary antibody (1:20,000, 1 h, RT) on the Odyssey Infrared Imaging System (both LI-COR, Bad Homburg, Germany). Band intensities were measured with the appropriate Odyssey software, and protein expression ratios were calculated with Microsoft Excel 2007 software. To investigate autophagosome formation 30 μg protein from each sample were separated on a NuPage 12% Bis–Tris precast gel utilizing 1× NuPage MOPS SDS running buffer (both Invitrogen, Karlsruhe, Germany). LC3 protein was detected through chemiluminescence on a PVDF membrane (Bio-Rad Laboratories, Munich, Germany). Membranes were probed with a monoclonal antimouse LC3 (clone 2G6) antibody (Nanotools, Tenningen, Germany) at titer 1:500. Primary antibody was detected with a secondary ECL antimouse IgG HRP-linked antibody (Amersham Biosciences, Glattbrugg Switzerland) at titer 1:15,000 on ECL hyperfilm (Amersham Biosciences). The ratio of LC3-II/GAPDH expression was determined by quantification of the western blot with ImageJ.
Constructs and transfection
GFP-PARKIN has been described previously [
42]. To obtain cherry-GFP-PARKIN an existing cherry-GFP fusion protein was digested with BspEI and BamH1 to remove the insert. Into these sites the PARKIN ORF was cloned after digesting GFP-PARKIN with BspEI and BamH1. All constructs were verified by sequencing. SH-SY5Y cells and MEF were transfected by electroporation with the cell line transfection kit V respective with the MEF transfection kit and the Nucleofector II (all Lonza, Basel, Switzerland) according to the manufacturer’s instruction. HeLa cells were transfected with Effectene (Qiagen, Hilden, Germany) likewise according to the manufacturer’s instruction.
Confocal laser scanning microscopy and immunostaining
LAMP-1 and LAMP-2 were detected with anti-LAMP-1 CD 107A and anti-LAMP-2 CD 107B antibody, while p62/SQSTM1 was detected with anti-p62 Ick ligand as primary antibody (all BD Bioscience, Heidelberg, Germany). Cy2-conjugated Affini Pure Donkey antimouse IgG (H+L) antibody served as secondary antibody (Dianova, Hamburg, Germany). All microscopic analyses were performed with a Leica TCS SP5 confocal laser scanning microscope equipped with the appropriate filters and a HCX PL APO lambda blue 63.0×, 1.40 OIL UV objective that was controlled by the LAS AF scan software (version 1.8.2) (Leica Microsystem, Wetzlar, Germany). Pictures were visualized with IMARIS 6.0.0 (BITPLANE Scientific solutions), and no deconvolution was performed. Images were maximum image projections apart from the insets in Fig.
3c, d which represent single slices.
Statistics
Raw data generated with Microsoft Excel 2007 were transferred to GraphPad Prism 4.03 software to calculate mean, SEM, and
P values. For qPCR significance was calculated by unpaired
t test between mean of 2
−∆∆Ct values under different conditions, e.g., nonstarved (RPMI + FCS) vs. starved (HBSS − FCS) for the corresponding time. Experimental variation was whenever possible (qPCR) calculated by averaging ΔΔCt values of the controls of
n experiments and generating 2
−∆∆Ct values. SEM is indicated by error bar in the figures. One-way ANOVA followed by Tukey posttest was used to calculate statistics for PARKIN protein expression change (Fig.
2b) and mRNA changes of primary neurons (Fig.
5).
P values smaller than 0.05-fold were considered to be significantly different: *
P ≤ 0.05, **
P ≤ 0.01, and ***
P ≤ 0.001.
Discussion
In this study we investigated the involvement and the regulation of parkinsonism-related genes in response to serum and nutrient starvation. Our expression analyses document a prominent role for PARKIN in growth factor signaling: during unavailability of serum and nutrients, neural SH-SY5Y cells induce PARKIN more than other published pathway components [
20] in a phasic pattern for mRNA levels at 8–24 h after starvation, in a process dependent on PI3K-Akt-mTOR signaling and autophagy. These data are supported by alterations of
PARKIN mRNA levels after application of pharmacological drugs targeting the PI3K-Akt-mTOR pathway, as we found a significant upregulation of the
PARKIN transcript in unstarved cells after specific inhibition of the PI3K-Akt-mTOR phosphorylation cascade either through LY294002 or through rapamycin. The PI3K-Akt signaling is involved in a number of different pathways including cell growth, proliferation, cell cycle control, cell migration and invasion, survival, protein, glycogen and fatty acid synthesis [
52]. While the upregulation of
PARKIN mRNA in unstarved cells after LY294002 treatment is most probably due to LY294002’s ability to mimic starvation, it could be also/additionally a compensatory antiapoptotic mechanism. As Akt blocks Bax conformation changes and mitochondrial translocation [
53,
54] and PARKIN has antiapoptotic properties [
55‐
58], an increase of the
PARKIN transcript could therefore compensate for the LY294002-mediated inhibition of Akt.
Our results are compatible with strong biochemical evidence in peripheral cell types showing the ubiquitin E3 ligase PARKIN able to modulate EGF receptor trafficking and promote PI3K-Akt signaling through a direct interaction between the PARKIN ubiquitin-like (Ubl) domain and the ubiquitin-interacting motif of the protein Eps15, a component of the RTK endocytosis machinery [
14], indicating that PARKIN function affects RTK signaling and that a dysregulation could be relevant for the nervous system.
Published data about transcriptional dysregulation in PD are scarce and usually not controlled by time course analyses, in spite of the intense investigations into PD pathogenesis and into the dozen novel PD genes over the past decade [
59]. A genome-wide survey of the brain transcriptome from the MPP
+-treated mouse model of PD identified an upregulated expression of the immediate early gene SGK1 as the prominent effect [
46]. SGK1 transcription is strongly stimulated by serum or specific growth factors such as insulin but also by other stressors such as high glucose concentrations or oxidative stress and is activated by PI3K signaling [
60]. SGK1 exerts an antiapoptotic effect through its inhibitory phosphorylation of FOXO3a and BAD [
61]. The transcription factor FOXO3a is induced by nutrient restriction and oxidative stress, is inhibited by Akt-signaling, and mediates the protective upregulation of antioxidant proteins such as MnSOD and later the proapoptotic regulation of FasL, Bim, and Bcl-6 [
62]. FOXO3a was observed to be sequestered into Lewy bodies, the pathognomonic protein aggregates in neurons affected by PD [
44]. Thus, SGK1 and FOXO3a represent credible examples of transcriptional dysregulation in PD, both being implicated in stress-dependent PI3K-Akt growth signaling.
Our analysis whether
PARKIN expression levels depended on the presence of serum growth factors showed similarly strong effects early on for
FOXO3a and
SGK1, later for
PARKIN and
PINK1 in dopaminergic neuronal cells (Fig.
1). This effect could be maximized to a twofold
PINK1 transcript induction and sixfold
PARKIN induction upon the additional removal of nutrients from the culture medium (Fig.
2). This transcript induction served not only to compensate the ongoing PARKIN protein sequestration in autophagosomes and lysosomes (Fig.
3), but was sufficient to elevate the PARKIN protein steady-state levels threefold after 36 h in a sustained fashion, suggesting that PARKIN is specifically needed during this period, hours after maximal LC3-II conversion and induction of lysosomal degradation (compare the time courses for PARKIN protein and LC3-II in Figs.
2b and
3a, f). A degradation of PARKIN by lysosomes was implicated after CCCP treatment [
28,
29,
63], and our data show that inhibition of the autophago-lysosomal pathway during starvation, e.g., by bafilomycin A1 significantly reduced
PARKIN mRNA levels, indicating a similar fate for PARKIN during starvation. However, CCCP-induced PARKIN degradation was dependent on PINK1, while starvation-triggered
PARKIN induction became significant before the
PINK1 induction (Fig.
2), and appeared independent from PINK1 (Fig.
5), suggesting only partially shared upstream regulation mechanisms and implying that PARKIN participates in different pathways in stress response depending on the specific stressor.
Thus, serum deprivation and nutrient starvation result not only in the known early expression regulation of
FOXO3a and
SGK1 during the early period of growth factor signaling and autophagosome formation (until 12 h in our assay, see Fig.
3f) but also in a later strong upregulation of
PARKIN and
PINK1 levels, which could exert protective antiapoptotic effects [
64]. The data are compatible with the previously reported dependence of
PINK1 expression on the activity of the phosphatase PTEN, which antagonizes the PI3K-Akt-mTOR kinase signaling cascade [
65]. Cell type-specific regulation of PARKIN had already been noted after ER or proteasomal stress [
38,
39], but a time course correlation of this effect to cellular processes or an investigation into the mechanisms involved was not attempted, except an identification of ATF4 as transcription factor of
PARKIN [
15].
In conclusion, our systematic expression studies of dopaminergic neural cells documented in a time course over 72 h after starvation and treated with drugs to modulate specific pathways all consistently point to an important role of PARKIN and PINK1 in the serum- and nutrient-dependent signaling, suggesting that they act closely together in this pathway. It would be interesting to investigate whether additional parkinsonism-relevant genes are involved in the same signaling network; similar expression profiling studies could be useful to screen novel disease-associated candidate genes and elucidate their putative roles in stress response pathways.