iPSCs were generated from dermal fibroblasts isolated from two PARK2 patients carrying parkin mutations and two control subjects using retroviruses carrying Oct4, Sox2, Klf4, and c-Myc to reprogram the cells as previously described [¹⁴,¹⁵]. The PARK2 patients were a 71-year-old female (PA) with a homozygous deletion of parkin exons 2–4 and a 50-year-old male (PB) with a homozygous deletion of exons 6 and 7 (Table  1 and Additional file 1A and B). Patient PA died 1 year after enrollment in the study at the age of 72. A previously-established human iPSC clone from control subject A, 201B7 (B7), was also used [¹⁵]. In addition, the following human embryonic stem cell (hESC)-like iPSC clones were selected for detailed analysis: three controls (B7 and YA9 from control A, and WD39 from control B), three from patient PA (PA1, PA9 and PA22), and four from patient PB (PB1, PB2, PB18 and PB20) (Figure  1A and Additional file 2A and B).

The PARK2 iPSCs expressed pluripotent hESC markers (Figure  1A and Additional file 2A-C) and formed teratomas containing all three germ layers (Additional file 2D). All of the retroviral transgenes were silenced in each clone (Additional file 2E). The iPSCs derived from PA and PB retained the corresponding homozygous parkin deletions and exhibited genomic stability (Figure  1B-D; Additional file 3A and B; and Table  1). All of the clones differentiated into neurons, including tyrosine hydroxylase (TH)-positive neurons, through a process of embryoid body and neurosphere formation (Figure  1A). Thus, all of the lines were successfully reprogrammed into a pluripotent state and were suitable for further analysis.

Because increased levels of oxidative stress have been documented in other PD models [⁷,¹⁰,¹⁶,¹⁷], we examined oxidative metabolism in the iPSC clones by measuring the cellular levels of reduced glutathione (GSH). GSH reacts with reactive oxygen species (ROS) and is catalyzed by glutathione S-transferase [¹⁸]. Consistent with previous results from patient-derived cells [¹⁶], the levels of GSH in PARK2 iPSC-derived neurospheres were significantly lower than those in control iPSC-derived neurospheres (Figure  2A). We also examined ROS production using 2’, 7’-dichlorodihydrofluorescin (DCF) fluorescence to measure the levels of intracellular oxidants. The DCF fluorescence intensity in the PARK2 iPSC-derived neurons was significantly higher than that in control iPSC-derived neurons (Figure  2B and C), which indicated an increased level of oxidative stress. A recent study showed that, in PARK2 iPSC-derived neurons, monoamine oxidase (MAO)-A and -B levels and oxidative stress levels are increased, as is spontaneous DA release [¹⁰]. Here, we found no significant differences in MAO-A and -B expression levels between PARK2 and control neurons (Additional file 4A and B).

The Nrf2 pathway plays a cytoprotective role under conditions of ROS accumulation. Recent studies show that activation of the Nrf2 pathway reduces oxidative stress and provides partial protection from MPTP-mediated neurotoxicity [¹⁹]. Elevated Nrf2 expression was observed in the postmortem brain of a PD patient [²⁰]. These data suggest a putative link between the Nrf2 pathway and PD, and prompted a closer investigation of this signaling pathway in control and PARK2 iPSC-derived neurons [^19-21]. The expression of Nrf2 pathway proteins, such as Nrf2 and NADH quinone oxidoreductase (NQO1), was significantly increased in PARK2 iPSC-derived neurons (Figure  2D and E). These data are in line with previous reports [^19-21], and suggest that the Nrf2 cytoprotective pathway may be activated in PARK2 iPSC-derived neurons to prevent further damage from oxidative stress. Taken together, these data demonstrated an increased level of oxidative stress accompanied by activation of the Nrf2 pathway in PARK2 neurons.

Increased oxidative stress (which affects anti-oxidant defense systems) and mitochondrial dysfunction are implicated in the pathogenesis of PD [¹,¹³,^21-23]. Furthermore, ROS accumulation causes both oxidative damage and mitochondrial dysfunction in the substantia nigra (SN) of parkin-deficient mice [⁷]. However, the exact mechanism of mitochondrial pathogenesis associated with PARK2 is controversial. For example, while Drosophila parkin mutants show abnormal mitochondrial morphology, parkin-knockout mice do not [⁷,²⁴]. In addition, while a greater degree of mitochondrial branching is observed in fibroblasts derived from PARK2 patients, the detailed morphology of the mitochondria in these cells has not been characterized [²⁵]. To investigate these mitochondrial abnormalities in more depth, we performed a detailed morphological analysis of mitochondria in PARK2 iPSC-derived neurons using electron microscopy. Mitochondria in PARK2 neurons from both patients showed a highly electron-dense matrix and swollen mitochondrial cristae within the inner mitochondrial membrane (IMM) (Figure  3A, black arrowheads). The perikaryal volume density of the abnormal mitochondria was significantly increased in PA and PB iPSC-derived neurons relative to control clones (Figure  3B). Furthermore, the density of normal mitochondria decreased (Figure  3B). Importantly, both abnormal and normal mitochondria were observed in PARK2 neurons (Figure  3A, white arrowheads). Abnormal mitochondria were observed in 87.8% of iPSC-derived neurons from PA, and 79.5% of iPSC-derived neurons from PB. These data indicated that abnormal mitochondrial morphology was a feature of most PARK2 iPSC-derived neurons from these patients. In addition, abnormal tubulovesicular structures were observed adjacent to the Golgi cisternae in PARK2 iPSC-derived neurons (Figure  3A). These abnormal mitochondrial and tubulovesicular structures were not observed in PARK2 fibroblasts or in undifferentiated iPSCs (Additional file 5A and B). These histological abnormalities represent novel PARK2-related neuronal pathologies.

PARKIN is involved in the mitochondrial fission/fusion system and is recruited to depolarized mitochondria to promote mitophagy [⁵,^26-29]. In iPSC-derived neurons containing a mutation in PINK1 (a protein kinase upstream of PARKIN), PARKIN is not recruited appropriately to mitochondria [³⁰]. We hypothesized that PARKIN-deficient human neurons would show aberrant removal of depolarized mitochondria. To examine the turnover of damaged mitochondria, we treated iPSC-derived neurons with carbonyl cyanide m-chlorophenyl hydrazine (CCCP), which triggers the loss of mitochondrial membrane potential and results in the removal of damaged mitochondria. The intensity of TMRE, a mitochondrial membrane potential-dependent dye, clearly decreased in both control and PARK2 iPSC-derived neurons treated with CCCP, which indicated a reduced mitochondrial membrane potential in both sets of neurons (Additional file 6). To determine the extent to which the damaged mitochondria were eliminated, we measured the area of the IMM after CCCP treatment. Compared with untreated cells, there was a dramatic loss of IMM area in the treated control neurons, but not in the treated PARK2 neurons (Figure  3C, left four columns; Figure  3D, black bars). To assess whether lysosomes were involved in the CCCP-induced elimination of mitochondria, we treated cells with Bafilomycin (Baf) A₁, an inhibitor of the vacuolar type H(+)-ATPase. Baf A₁ attenuated the CCCP-dependent reduction in the IMM area in control neurons (Figure  3C, right four columns; Figure  3D, white bars). To confirm that the abnormal turnover of damaged mitochondria was characteristic of neuronal cells, PARK2 fibroblasts and undifferentiated iPSCs were treated with CCCP. CCCP-treated PARK2 fibroblasts and undifferentiated iPSCs exhibited the same mitochondrial dynamics as CCCP-treated control cells (Additional file 5C-E). Together, these data indicated aberrant degradation of mitochondria damaged by CCCP treatment in PARK2 iPSC-derived neurons.

These results support a recently proposed working model for PD, in which damaged mitochondria accumulate due to a disruption in PARKIN-mediated mitochondrial quality control [²⁸]. The electron microscopy data, which showed a mixture of abnormal and normal mitochondria, indicated that PARKIN-mediated mitochondrial quality control is compromised, even in young PARK2 iPSC-derived neurons. In these cells, residual normal mitochondria may have compensated for the damaged ones. Thus, while our findings suggest that the PARKIN-dependent mechanisms that regulate mitochondrial homeostasis are disrupted in PARK2 cells, further detailed analyses are required to fully understand the mechanism underlying this disruption and the implications for PD.

LBs are pathological neuronal inclusions composed principally of α-synuclein. They are typically associated with PD and certain forms of dementia [¹,¹³,³¹]. Although LBs are generally thought to be absent from PARK2 patients [¹,¹³,³¹], rare cases of LB formation in the brains of PARK2 patients have been reported recently [¹²,³²,³³]. The PARKIN protein co-localizes with LBs in some patients with sporadic PD [³⁴], and a functional interaction between PARKIN and α-synuclein is indicated by both in vitro and in vivo findings [^35-37]. These results suggest that PARKIN-pathway may contribute to LB formation in PD patients.

We were able to conduct a histopathological analysis of postmortem brain tissue from patient PA. Hematoxylin and eosin staining of the SN revealed low levels of brown-black melanin pigment compared with healthy SN tissue (Figure  4A and A’). Surprisingly, LBs accumulated in the SN and other areas of the brain in patient PA (Figure  4B and Table  2). Furthermore, α-synuclein and pα-synuclein immunoreactive puncta and neurites were observed in the areas where LBs were present (Figure  4B). TH/pα-synuclein double-positive neurons were also detected in the SN (Figure  4C). Of note, α-synuclein-positive/TH-negative or pα-synuclein-positive/TH-negative neurons in the SN and other areas of the brain tissue from patient PA’s brain were observed (Table  2). These data suggested that α-synuclein accumulated not only in TH+ neurons, but also in other types of neurons. Postmortem tissue from the brain of the father of patient PB was also examined. The father carried a homozygous deletion of exons 6 and 7 of the parkin gene (Figure  1B, Additional file 1B and 7A), similar to patient PB. There was no evidence of LBs or α-synuclein-positive neurons in the autopsied brain tissue of the father (Figure  4D). Thus, since the genetic background of patient PB and his father are likely to be very close (Additional file 1B and 7A), these results are probably reflective of a specific phenotype of patient PB, which was different from that in patient PA (Figure  4A-D).

To determine whether iPSC-derived neurons recapitulated the in vivo phenotypes of the corresponding cell donors, we next examined α-synuclein accumulation in PARK2 iPSC-derived neurons. To rule out the possibility that α-synuclein expression in undifferentiated PARK2 iPSCs was increased by multiplication of the SNCA gene, genomic aberrations acquired during the process of iPSC establishment, or by repeated passage of the cells, the SNCA gene copy number in iPSCs was quantified by genomic qPCR. A comparison with control iPSCs showed that iPSCs from both PA and PB carried the normal number of SNCA gene copies (Additional file 8A). Moreover, immunostaining for α-synuclein did not reveal any increase or decrease in α-synuclein protein levels in PARK2 iPSCs (Additional file 8B). As a control for LB formation, we generated iPSC-derived neurons from a 106-year-old woman (designated Cent1-8), since previous work suggested that aging is a predisposing factor for LB formation in PD patients [³¹,³⁸]. Since α-synuclein was also expressed in non-neural cells, triple labeling for α-synuclein, βIII-tubulin, and TH was performed to ensure that only neurons were examined (Figure  5A, asterisks). The proportion of α-synuclein-positive iPSC-derived neurons that were also positive for βIII-tubulin from PB was similar to that in the controls (including Cent1-8); however, the proportion was significantly higher in PA. These results were consistent with the in vivo phenotypes of the cell donors based on analysis of postmortem brain tissue (1629, 357, 805, 3747, and 4330 iPSC-derived βIII-tubulin+ neurons in control A, control B, Cent1-8, PA and PB respectively; Figure  5A-C, arrows and arrowheads). Thus, the increase in α-synuclein expression levels seen in PARK2 iPSC-derived neurons cannot be attributed solely to the effects of aging, but associated with the disease phenotype.

The obvious LB-formation was observed in the postmortem brain of PA patient, who showed a late onset at 61 years, corresponding to the enhanced α-synuclein accumulation in the iPSC-derived neurons from the same patient. Thus, it is likely that early-stage LB formation was recapitulated in vitro in iPSC-derived neurons. Furthermore, the present findings are consistent with recent work by several groups, which suggest that the age of onset of PARK2 in patients with LB formation (41 on average) is later than in patients without LB formation (below 40) [¹²,³²,³³]. The earlier onset in patient PB (at 28 years) than in PA (at 61 years) would be consistent with the finding of lower α-synuclein accumulation in PB iPSC-derived neurons compared with PA iPSC-derived neurons. On the other hand, and in contrast to the observations of brain tissue from PA, analysis of brain tissue from the father of patient PB, in whom the onset of PD was 39 years of age, revealed no evidence of LB formation (Figure  4D). Importantly, PA iPSC-derived neurons showed significantly more α-synuclein accumulation than PB iPSC-derived neurons (Figure  5A and C). These results suggest that the extent of α-synuclein accumulation is an important factor in LB formation. Then, how can we explain the difference of α-synuclein accumulation between PA and PB patients-derived neuronal cells? It is possible that PA is a rare example of PARK2 complicated by sporadic PD. Although both PA and PB iPSCs showed a normal SNCA gene copy number, it is possible that PA-derived cells acquired an unknown gene mutation relating to LB formation. Thus, we cannot rule out the possibility that other factors may affect LB formation in PARK2 patients. Further analyses will be required to identify these putative factors. Although iPSC clones from sporadic and familial PD patients were recently established [¹⁷,³⁰,^39-42], this report is the first to demonstrate that the phenotype of PD-specific iPSC-derived neurons replicates the in vivo phenotype seen in postmortem brain tissue from the corresponding cell donor.

For control A, human dermal fibroblasts (HDFs) from the facial dermis of a 36-year-old Caucasian female (Cell Applications Inc.) were used to establish iPSCs (201B7; Passage 20–29, YA9; Passage 15–24). The 201B7 iPSCs were kindly provided by Dr. Yamanaka [¹⁵]. A skin-punch biopsy from a healthy 16-year-old Japanese female obtained after written informed consent (Keio University School of Medicine) was used to generate the control B iPSCs (WD39; Passage 8–17). PA iPSCs (PA1, 9, and 22; Passage 10–19) and PB iPSCs (PB1, 2, 18, and 20; Passage 8–17) were generated from a 71-year-old Japanese female patient and a 50-year-old Japanese male patient, respectively, using the same methods used to generate control B iPSCs. The maintenance of HDFs, lentiviral production, retroviral production, infection, stem cell culture and characterization, and teratoma formation were performed as described previously [¹⁴,¹⁵]. All of the experimental procedures for skin biopsy and iPS production were approved by the Keio University School of Medicine Ethics committee (Approval Number: 20-16-18) and Juntendo University School of Medicine Ethics committee (Approval Number: 2012068). hESCs (KhES-1; Passage 29–38 (kindly provided by Dr. Norio Nakatsuji) were cultured on feeder cells in iPS culture media [⁴³].

Neural differentiation of iPSCs was performed as previously described [⁴⁴] with slight modifications (Okada et al., manuscript in preparation). Briefly, iPSC colonies were detached from feeder layers and cultured in suspension as EBs for about 30 days in bacteriological dishes. EBs were then enzymatically dissociated into single cells and the dissociated cells cultured in suspension in serum-free media (MHM) [⁴⁴] for 10 to 14 days to allow the formation of neurospheres. Neurospheres were passaged repeatedly by dissociation into single cells followed by culture in the same manner. Typically, neurospheres between passages 3 and 8 were used for analysis. For terminal differentiation, dissociated or undissociated neurospheres were allowed to adhere to poly-L-ornithine- and fibronectin-coated coverslips and cultured for 10 days.

For immunocytochemical analysis, cells were fixed with phosphate buffered saline (PBS) containing 4% paraformaldehyde (PFA) for 30 min at room temperature (RT). The cells were analyzed by immunofluorescence staining using antibodies to the following proteins: β-III-tubulin (1:1000, Sigma), NANOG (1:100, ReproCELL), OCT3/4 (1:200, Santa Cruz Biotechnology), SSEA-4 (1:200, Millipore), TRA-1-60 (1:200, Millipore), TH (1:100, Millipore), α-synuclein (1:500, Invitrogen), pα-synuclein (1:1000, Wako), cleaved-Caspase3 (1:500, Cell Signaling) and ComplexIII (C-III)-core I (1:200, Invitrogen). Cells were washed with PBS after incubation with the primary antibody, followed by incubation with an Alexa Fluor 488-, Alexa Fluor 555-, or Alexa Fluor 647-conjugated secondary antibody (1:500, Invitrogen). Images were obtained using Apotome (Zeiss) or LSM-710 confocal (Zeiss) microscopes.

Genomic DNA was purified from HDFs and iPSCs using a DNeasy kit (Qiagen). The PCR conditions used have been previously described [²,⁴²].

RNA isolation and reverse transcription (RT)-PCR were performed as previously described [⁴⁴]. The amount of cDNA was normalized to β-actin mRNA. Real-time RT-PCR was performed on a ABI PRISM Sequence detection System 7900HT (Applied BioSystems) using SYBR premix ExTaq (Takara). Primers for the detection of Oct4, the transgenes Oct4-tg, Sox2-tg, Klf4-tg and c-Myc-tg, and MAO-A, and -B have been previously described [¹⁰,¹⁵].

To assess teratoma formation, iPSCs were injected into the testis of 8-week-old NOD/SCID mice (OYG International) as previously described [¹⁴]. Eight weeks after transplantation, tumors were dissected and fixed with 4% PFA in PBS. Paraffin-embedded tissue was sectioned and stained with H&E. Images were obtained using a BZ-9000 (Keyence) microscope.

Genomic DNA was restricted, labeled, and purified using the Agilent Oligo CGH Microarray Kit (Agilent Technologies) according to the manufacturer’s protocol. Labeled genomic DNA was processed for hybridization on a 4x 180K microarray (Agilent Technologies). Processing was performed as instructed by the manufacturer. The genomic analysis was performed using Agilent Genomic Workbench ver. 6.0 software (Agilent Technologies).

Reduced GSH levels were measured according to the kit manufacturer’s protocol (GSH-Glo Glutathione Assay; Promega). Chymotrypsin-like proteasome activity was measured using a Cell-Based Proteasome-Glo Assay according to the manufacturer’s instructions (Promega). Briefly, neural cells (1.0 × 10⁴) derived from neurospheres were seeded in triplicate into a white 96-well plate (Nunc). Prepared reagent (100 μl) was added to each well. After incubation for 10 min at RT, luminescence intensity was recorded. ROS levels were determined by measuring DCFH-DA fluorescence (Invitrogen). Briefly, neurons were incubated with 5 μM DCFH-DA and Hoechst (1:2000) for 30 min at 37°C, after which they were washed with PBS and then incubated in differentiation media. Fluorescence was measured by an In Cell Analyzer 2000 system (GE Healthcare Biosciences).

Differentiated neurons were harvested in MAPK lysis buffer containing proteinase inhibitor, and protein concentrations were measured by BCA assay (Thermo Scientific). Samples were diluted to yield equivalent protein concentrations and then 4 μg was denatured by the addition of 4X sample buffer (Invitrogen) supplemented with β-mercaptoethanol followed by boiling. Samples (7 μl/lane) were loaded onto a 4–20% SDS-polyacrylamide gradient gel. Membranes were incubated in blocking solution with the indicated primary antibodies at 4°C overnight. Immunoreactive proteins were detected with horseradish peroxidase (HRP)-conjugated secondary antibodies and then visualized by chemiluminescence (Pierce, Rockford, IL, USA) according to the manufacturer’s instructions. Quantification of band intensities was performed using an RAS4000 system. The primary antibodies used were anti-NQO1 (1:1000, Abcam), anti-NRF2 (1:1000, Santa Cruz Biotechnology) and β-actin (1:5000, Cell Signaling).

Neurons were cultured with 30 μM CCCP (Sigma-Aldrich) or DMSO, with or without 5 μM Baf A₁ (Sigma-Aldrich), for 48 h. The cells were then fixed and stained for βIII-tubulin and C-III Core I, and counterstained with Hoechst. To quantify the IMM area of the neurons, the cytoplasmic area was extracted as shown in Figure 3C. The C-III Core I-positive signals within the extracted area were then converted to gray-scale and digitized. The IMM area was quantified from the digitized values using Image J software.

iPSC-derived neurons were incubated with 1nM TMRE (Invitrogen) for 15 min at 37°C and then observed under an Olympus IX81 microscope.

Cells were fixed with 2% glutaraldehyde/2% PFA in 0.1 M phosphate buffer (PB) (pH7.2), post-fixed with 1% OsO4 in 0.1 M PB (pH 7.2), blocked and stained with a 2% aqueous solution of uranyl acetate, dehydrated with a graded series of ethanol, and then embedded in Epon 812 (TAAB). Coverslips were detached and the embedded samples were placed under a stereomicroscope to identify the cells of interest. Ultrathin sections were cut with a Leica UC6 or UC7 ultramicrotome (Leica Microsystems) and then stained with uranyl acetate and lead citrate. Samples were observed with a Hitachi H7100 or HT7700 electron microscope.

Morphometric analysis was used to measure the volume density of mitochondria in the neuronal perikarya as previously described [⁴⁵]. Briefly, electron micrographs of neurons (n = 20, 23, 41, and 44 for control A (B7), control B (WD39), PA9 and PB2, respectively) were obtained at a magnification of ×7000. After enlarging to three times the original magnification, point-counting was carried out to determine the volume density using a double-lattice test system with 1.5 cm spacing. Mitochondria were classified as normal, abnormal, or undetermined. The abnormal mitochondria were defined as those with irregularly arranged cristae, or with a high electron-dense matrix. The volume density (Vv) of each type of mitochondrion was expressed as percent volume according to the following formula: Vv = (Pi/Pt) × 100 (%), where Pi is the number of points falling on each mitochondrial structure and Pt is the number of points falling on the neuronal perikarya.

The ethical committee of the Kitasato University School of Medicine and Juntendo University School of Medicine reviewed and approved the protocol for analysis of autopsied brain tissue. Patients and control subjects were informed of the study and gave written informed consent. Brain tissue from patient PA was obtained following her death at age 72; brain tissue from the father of patient PB was obtained when he died at age 70 [⁴⁶]. Tissue was fixed with 10% formalin and then embedded in paraffin. Midbrain sections (6 μm thick) were cut, deparaffinized with xylene, and then rehydrated in ethanol. After being boiled and treated with H₂O₂, sections were subjected to immunofluorescence staining with antibodies to the following proteins: α-synuclein (1:500, Invitrogen), pα-synuclein (1:1000, Wako), and TH (1:1000, Calbiochem). After washing with PBS, sections were incubated with a biotinylated secondary antibody (1:500; Vector Laboratories Inc.) at RT for 1 hr followed by incubation with an avidin-biotin peroxidase complex (Vector Laboratories Inc.) for 1 hr. Immunoreactive proteins were visualized using 3,3-diaminobenzidine (DAB; Wako Pure Chemical Industries) and nuclear fast red staining. For immunofluorescence, FITC-conjugated and Cy3-conjugated secondary antibodies (1:500; Jackson Immunoresearch Laboratories) were used. Images were obtained using a BIOREVO (Keyence) and a confocal laser-scanning LSM710 (Zeiss) microscope.

Values represent the mean ± SEM. The Mann–Whitney U-test was used to evaluate differences between groups. A P value of < 0.05 was considered significant.

IX-X, motor cranial nerves IX-X; LC, Locus Coeruleus; SN, Substantia Nigra; nbM, nucleus basal of Meynert; Amy, Amygdala; Ent, Entorhinal cortex; T, Temporal lobe; F, Frontal lobe; P, Parietal lobe.