Introduction

Parkinson’s disease (PD) was first described in 1817 by James Parkinson1, who wrote about patients with shaking palsy, i.e., involuntary tremulous motion with lessened muscular power. PD occurs in approximately two of 1000 people and is highly correlated with aging, affecting about 1% of the older population above 60 years of age2,3 The main neuropathological symptoms are α-synuclein-containing Lewy bodies and loss of dopaminergic (DA) neurons in the substantia nigra pars compacta. PD patients experience movement difficulties with three cardinal signs: tremor, rigidity, and bradykinesia. Some of the main non-motor symptoms of PD include loss of smell, depression, sleep disorders, and dementia, but a wide range of other symptoms such as excessive saliva4 and susceptibility to melanoma11). Some of these pathways were not previously known to be associated with PD, but they were dysregulated in most of our PD lines, showing new possible cellular defects that may be targeted for treatment. For the DA neurons derived from the PD patient with the Parkin mutation, there were not many significantly dysregulated pathways. These did include “transport vesicle membrane.”

A reduction in synaptic activity is observed in neurons derived from sPD patients

We continued the study by differentiating neurons from a PD patient with no PD-causing mutations (UKERfAY6-X-001). We recorded n = 20 control neurons (two patients: 40102 and UKERfO3H-X-001) and n = 33 neurons from the sPD patient. There was no significant change in the total evoked action potentials (Fig. 3A for averages and 3B, 3C for representative traces). The average of the sodium currents is presented in Fig. 3D. The average over the slow potassium currents is presented in Fig. 3E and the average of the fast potassium currents is presented in Fig. 3F. There was no significant difference between the sPD and the control neurons in any of the sodium, slow potassium, and fast potassium currents. Representative EPSC traces are presented in Fig. 3G, H. There were no significant differences in the mean amplitude (Fig. 3I) or the mean rate (Fig. 3J) of EPSCs. However, when counting the number of neurons that had synaptic activity (see “Methods”), we did see a significant reduction in this number in the sPD neurons (p = 0.0013, Fig. 3K). Furthermore, analysis of the decay time constant of the synaptic events revealed a faster decay in sPD neurons, with an average of 2.3 ± 0.1 ms for control neurons and 1.5 ± 0.07 ms for sPD neurons (p = 9e−12, the distribution is presented in Supplementary Fig. 3). Spike shape analysis did not reveal any significant changes in the spike height (Supplementary Fig. 4A). The spike was significantly narrower in the sPD neurons (control 6 ± 0.8 ms, sPD 3.6 ± 0.2 ms, p = 0.0035, Supplementary Fig. 4B). The spike threshold was significantly more negative in the sPD neurons (control −18.4 ± 1.3 mV, sPD −24.6 ± 1 mV, p = 0.0006, Supplementary Fig. 4C). There was no significant change in the fast AHP (Supplementary Fig. 4D) or the capacitance (Supplementary Fig. 4F), and the input conductance was larger in sPD neurons (control 0.44 ± 0.12 nS, sPD 0.89 ± 0.16 nS, p = 0.05, Supplementary Fig. 4G). The cumulative distribution of amplitudes of EPSCs was similar between the control and sPD neurons (Supplementary Fig. 4E).

Fig. 3: A reduction in the number of neurons with synaptic activity is observed in neurons derived from a sPD patient.
figure 3

A No significant change was observed in the excitability measured by the total evoked potentials in the neurons derived from the sPD patient compared to controls. B Representative example of evoked action potentials in a control neuron. C Representative example of evoked action potentials in a neuron derived from a sPD patient. D No significant changes were observed in the sodium currents. E No significant changes were observed in the slow potassium currents. F No significant changes were observed in the fast potassium currents. G Example recording of synaptic currents in a neuron derived from a healthy control (the lower plot presents a zoom in on the segment denoted in the black dashed lines in the upper graph). H Example recording of synaptic currents in a neuron derived from a patient with sPD (the lower plot presents a zoom in on the segment denoted in the black dashed lines in the upper graph). I No significant change was observed in the mean amplitude of the synaptic currents in the sPD neurons. J No significant change was observed in the mean rate of the synaptic currents in the sPD neurons. K A significant reduction in the number of neurons that had synaptic activity was observed in sPD neurons compared to controls. L Signaling network analysis with the top enriched KEGG pathways for the sPD patient-derived neurons compared to the controls.

Full size image

Three sPD lines (UKERfRJO-X-001, UKERfAY6-X-001, and UKERfM89-X-001) were pooled together for the analysis of gene expression, plus four control (lines 40102, UKERf1JF-X-001, UKERf33Q-X-001, and UKERfO3H-X-001). The top 10 GO terms, KEGG, and MSigDB dysregulated pathways for the DA neurons derived from sPD patients are presented in Supplementary Fig. 15, and the full list is shown in Supplementary Tables 3–5. The signaling network analysis with the top enriched KEGG pathways for the sPD neurons compared to the control neurons is presented in Fig. 3L. There were a few dysregulated pathways that were synapse-related such as regulation of synaptic structure, regulation of synaptic organization, regulation of synaptic plasticity, and more. The most dysregulated pathways were related to nasopharyngeal carcinoma, with almost 250 affected genes and a false discovery rate (FDR) of 1.6e–105. Other highly dysregulated pathways were related to the cilium. Like the other genetic PD lines, the extracellular matrix was highly affected, and more dysregulated pathways that were repeatedly affected in our PD lines were focal adhesion, collagen processes, PI3K-Akt, protein digestion and absorption, and pathways related to reactive oxygen species and metabolic processes. When performing ICC for two ECM-proteins fibronectin and collagen IV, we have observed reduced puncta size in the sPD neurons compared to the control neurons (see Supplementary Fig. 11). Several hypoxia-related pathways were affected as well.

A reduction in synaptic activity is observed in neurons derived from more sPD patients using a different protocol and selecting a subset of the neurons

We next analyzed data that were acquired using a different differentiation protocol (see “Methods”, second DA differentiation); in addition, only neurons that were defined as type 5 neurons (see definition in “Methods”) were used for the analysis. Strikingly, despite the different methods used both in differentiating the neurons and in analyzing the data, the main phenotype of a reduction in synaptic activity was present in the sPD neurons as well.

We recorded from control individual (line 40102) and two sPD patients (UKERfRJO-X-001, UKERfR66-X-001). It should be noted that one of the patients had a heterozygous missense mutation in the EIF4G1 gene (UKERfRJO-X-001). The number of total evoked action potentials was not different between control and sPD neurons (averages presented in Fig. 4A and representative traces are presented in Fig. 4B, C). The averages of the sodium currents are presented in Fig. 4D, the averages of the slow potassium currents are presented in Fig. 4E, and the averages of the fast potassium currents are presented in Fig. 4F. Representative traces of synaptic activity are presented in Fig. 4G (control) and 4H (sPD). The average amplitude of the synaptic currents was increased in sPD neurons, but not significantly (Fig. 4I). The average rate of synaptic currents was significantly reduced in the sPD neurons (control 3 ± 0.6 Hz, sPD 1.4 ± 0.2 Hz, p = 0.0037, Fig. 4J). The cumulative distribution of the amplitude of synaptic currents was slightly right-shifted in the sPD neurons. The capacitance was reduced in the sPD neurons, indicating smaller cells (control 64 ± 38 pF, sPD 45 ± 33 pF, p = 0.02). Performing action potential shape analysis, we did not observe any significant changes in the spike height (Supplementary Fig. 5A) or the spike width (Supplementary Fig. 5B). The amplitude of the fast AHP was larger in the sPD neurons (control −5.7 ± 0.8 mV, sPD −10.9 ± 0.6 mV, p = 0.0000035, Supplementary Fig. 5C), and the threshold for evoking an action potential was more depolarized (control −43 ± 0.8 mV, sPD −39.9 ± 0.6 mV, p = 0.0016, Supplementary Fig. 5D).

Fig. 4: Reduced synaptic activity in neurons derived from two sPD patients compared to control neurons.
figure 4

A No significant change was observed in the excitability measured by the total evoked potentials in the neurons derived from the sPD patient compared to controls. B Representative example of evoked action potentials in a control neuron. C Representative example of evoked action potentials in a neuron derived from a sPD patient. D Sodium currents are increased in the sPD neurons compared to the controls. E The slow potassium currents are reduced in the sPD neurons compared to the control neurons. F The fast potassium currents are reduced in the sPD neurons compared to the control neurons. G A representative trace of synaptic currents in a control neuron (the lower plot presents a zoom-in on the segment denoted in the black dashed lines in the upper graph). H A representative trace of synaptic currents in an sPD neuron (the lower plot presents a zoom in on the segment denoted in the black dashed lines in the upper graph). I The mean amplitude of the synaptic currents was increased, but not significantly, in the sPD neurons. J The mean rate of synaptic currents was significantly reduced in the sPD neurons. K The cumulative distribution of the sPD neurons is slightly right-shifted, indicating large amplitudes of synaptic currents. L The capacitance of sPD neurons was significantly reduced in sPD neurons compared to control neurons.

Full size image

A reduction in synaptic activity is observed in neurons derived from an edited iPSC line with an inserted A53T mutation in the SNCA gene

We next performed experiments on neurons derived from a healthy human subject whose fibroblasts were reprogrammed into iPSCs and on an engineered line, in which the A53T mutation was edited into the SNCA gene (isogenic lines). The healthy line and the edited mutated line were differentiated into DA neurons by a differentiation technique that is proprietary to Fujifilm Cellular Dynamics Inc (CDI) (iCell DopaNeurons85). Having an edited healthy line allowed us to observe a similar genetic background, thereby measuring the neuronal changes that occurred specifically due to the A53T mutation. iCell DopaNeurons (CDI) were previously shown to have a protein expression pattern that supported a midbrain lineage DA phenotype85. Using whole-cell patch-clamp, we measured the functional features of A53T neurons compared to the control neurons. Fourteen A53T and 17 control neurons were recorded. The total number of evoked potentials in 17 depolarization steps (see “Methods”) was not significantly different between A53T neurons (39 ± 5) and control neurons (43 ± 6) (Fig. 5A and representative traces in Fig. 5B, C). Using voltage clam** we measured the sodium/potassium currents in the neurons (representative traces are shown in Fig. 5D, E). Sodium currents were not significantly different, except for the opening of the sodium channels; at −30 mV, control neurons displayed a sodium current of 1.6 ± 0.7 pA/pF, whereas A53T neurons displayed a sodium current of 10.4 ± 2.5 pA/pF (p = 0.0012, Fig. 5F). Slow and fast potassium currents were not significantly different between A53T neurons and controls (Fig. 5G, H). Spike parameters were not altered in the A53T neurons (Supplementary Fig. 6A–D). The cumulative distribution curve of the amplitude of synaptic currents was left-shifted in A53T compared to controls, indicating lower amplitudes of synaptic currents (Fig. 5I). Representative example recordings of EPSCs for control neurons (Fig. 5J) and an A53T neuron (Fig. 5K) are shown. The mean amplitude of synaptic currents was significantly reduced in A53T neurons (16.7 ± 1.2 pA) compared to control neurons (20.6 ± 0.3 pA, p = 2e−6, Fig. 5L). The mean rate of synaptic events was significantly reduced in A53T neurons (0.2 ± 0.06 Hz) compared to control neurons (1 ± 0.4 Hz, Fig. 5M, p = 0.05). Cell capacitance was smaller, but not significantly, in A53T neurons (20.2 ± 1.6 pF) compared to control neurons (23 ± 1.4 pF, Supplementary Fig. 6E). The input conductance was decreased, but not significantly, in A53T neurons (Supplementary Fig. 6F). Further imaging of neurons stained for TH showed that the A53T neurons were smaller and had fewer, less arborized neurites (Supplementary Fig. 7A–H). We similarly analyzed immunostaining images for the other PD lines and did not find changes in neurites’ lengths (see Supplementary Fig. 8). The mean percentage of neurons with neurite beading was 3.7 ± 0.6% in control cultures and 28 ± 1.8% in A53T cultures (p = 2e−10, Supplementary Fig. 7I–K). Increased expression of the SNCA gene was observed in the A53T mutant neurons, and also in the LRRK2 mutant neurons, and the neurons derived from the patient with the triplication in the SNCA gene (Supplementary Fig. 9).

Fig. 5: Reduced synaptic activity in neurons derived from an engineered line of iPSCs with an A53T mutation in the SNCA gene compared to isogenic controls.
figure 5

A No significant change in the excitability of the A53T neurons compared to control neurons. B Example recording of action potentials in current-clamp mode of a control neuron. C Example recording of action potentials in current-clamp mode of an A53T neuron. Sodium currents of A53T neurons open at a lower depolarization potential than control neurons. D Example recording of sodium and potassium currents in voltage-clamp mode in a control neuron. E Example recording of sodium and potassium currents in voltage-clamp mode in an A53T neuron. F No significant changes were observed in the sodium currents of the A53T neurons compared to controls. G No significant changes were observed in the slow potassium currents in the A53T neurons. H No significant changes were observed in the fast potassium currents in the A53T neurons. I The cumulative distribution of the amplitudes of EPSCs is left-shifted in A53T neurons compared to control neurons, indicating lower amplitudes in the A53T neurons. J Representative trace of the synaptic currents in a control neuron (the lower plot presents a zoom-in on the segment denoted in the black dashed lines in the upper graph). K Representative trace of the synaptic currents in an A53T neuron (the lower plot presents a zoom-in on the segment denoted in the black dashed lines in the upper graph). L The average amplitude of synaptic currents was significantly reduced in A53T neurons compared to control neurons. M The average rate of synaptic currents was significantly reduced in A53T neurons compared to the control neurons. N Signaling network analysis with the top enriched KEGG pathways for the A53T mutant compared to the controls.

Full size image

The top 10 affected KEGG and MsigDB pathways and GO terms are presented in Supplementary Fig. 16, and the full list is given in Supplementary Tables 3–5. The signaling network analysis with the top enriched KEGG pathways of the A53T mutant neurons compared to the controls is presented in Fig. 5N. There were many dysregulated synapse-related pathways such as dopaminergic synapse, postsynaptic membrane, presynaptic membrane, chemical synaptic transmission, and more. Similar to the other PD lines presented in this study, there were many extracellular matrix-related affected pathways as well as pathways related to focal and cell adhesion, collagen processes, protein digestion and absorption, and PI3K-Akt signaling pathways. It is interesting to note that similar findings of synaptic defects, both at the electrophysiological and morphological level, as well as dysregulation of pre- and post-synaptic factors and transsynaptic adhesion molecules identified by transcriptome analysis, were previously reported44.

Protein aggregates and altered morphology in the A53T SNCA mutated iDopaNeurons compared to control neurons

We next delved deeper to look for additional morphological and cellular alterations that occurred in the engineered A53T iDopaNeurons neurons compared to controls. The neurons were stained for neuronal markers TUJ1 and MAP2, and both A53T and control neurons expressed these markers (Fig. 6a–d). Misfolded protein aggregates were observed in A53T neurons but rarely seen in control neurons (Fig. 6e–j). The number of aggregates was quantified using the aggresomal kit and imaging in ultra-high resolution, 60.8 ± 7.1% of the A53T neurons displayed protein aggregates (n = 99), compared to 2.2 ± 0.6% of the control neurons (n = 96, p < 0.0001, Fig. 6q). To assess the number of synapses, we co-stained for synapsin1 and the post-synaptic marker PSD95. The number of puncta pairs Syn1/PSD95 density was significantly reduced in A53T cultures; 2.3 ± 0.2 pairs in 10 µm were observed in control cultures, whereas only 1.5 ± 0.16 (n = 15 neurites) pairs in 10 µm were observed in A53T cultures (p = 0.009, Fig. 6k–p, r). To assess how many of the misfolded protein aggregates were α-synuclein positive, we co-stained α-synuclein/aggresomes in the A53T neurons. Some of the aggregates contained α-synuclein protein (Fig. 6s–u).

Fig. 6: Protein aggregates in A53T dopaminergic neurons.
figure 6

a Representative image showing expression of neuron-specific class III beta-tubulin/ 4′,6-diamidino-2-phenylindole (TUJ1/DAPI) in control dopaminergic neurons. Scale bar 10 µm. b MAP2/DAPI expression in control dopaminergic neurons. Similar representative images of A53T dopaminergic neurons expressing TUJI/DAPI (c) and MAP2/DAPI d. e–g Control dopaminergic neurons exhibit almost no protein aggregates, as can be seen in the aggresomes and merged image with synapsin::EGFP/DAPI on the right. h–j A53T dopaminergic neurons display high levels of protein aggregates that is observed in the aggresomes staining. k–p Immunostaining for postsynaptic density protein 95 (PSD95 in green) and synapsin1 (syn1) reveals a drastic decrease in the ratio of synapsin1 to PSD95, indicating fewer synapses. q Quantification of protein aggregates shows a drastic increase in the number of neurons harboring protein aggregates in the A53T mutated neurons compared to controls. r Quantification of the ratio of syn1/PSD95 shows a drastic and significant decrease in A53T dopaminergic neurons compared to controls. s–u α-Synuclein staining and aggresomes reveal co-localization of some of the protein aggregated with α-synuclein. Error bars represent standard deviations in this figure.

Full size image

Seeking commonly affected pathways in PD neurons

PD patients share similar symptoms despite completely different genes causing the disease, or even when there is no defined genetic cause. Therefore, we sought to find common pathways between neurons derived from PD patients with different mutations. We started by pooling the RNA sequencing results from all our lines with PD-causing mutations—SNCA duplication, SNCA triplication, LRRK2, and Parkin—and we looked for differential expression compared to the four control lines. The 10 top terms enriched among up and down-regulated genes are presented in Fig. 7a–d (GO and KEGG pathways). The full list, as well as MsigDB dysregulated pathways, are given in Supplementary Tables 3–5. A clear picture emerges of pathways that were strongly affected in neurons derived from all PD lines with mutations; some of them were not previously known to be associated with PD. Collagen-related pathways and the extracellular matrix were very strongly affected in the PD lines, as was confirmed by ICC experiments at the protein level for Fibronectin and collagen IV (Supplementary Fig. 11). Focal adhesion was another pathway that repeated throughout the monogenic PD lines, as well as the PI3K-Akt signaling pathway, pathways related to cancer and oxidoreductase activity, and protein digestion and absorption. Importantly, synapse-related pathways such as synapse, synaptic membrane, postsynapse, and more were commonly dysregulated in our monogenic PD lines. In the up-regulated pathways, we found cell adhesion molecules (CAM), which are molecules that interact with the extracellular matrix and may be a compensation mechanism for the reduced collagen and other extracellular matrix-related genes. We also looked for overrepresented GO terms for genes that were common in the monogenic PD lines and the sPD lines and these are presented in Fig. 7e, f. The extracellular matrix is commonly implicated in both monogenic and sPD. Other dysregulated pathways were collagen pathways, focal and cell adhesion, PI3K-Akt signaling pathway, protein digestion and absorption, and a few pathways related to hypoxia. Pathways related to oxidative stress such as reactive oxygen species and oxidative stress were also dysregulated. Age-related pathways such as brain up, Alzheimer’s disease up, and cellular senescence were also commonly dysregulated in both monogenic and our sPD lines. Pathways that might be related to synapse function that were dysregulated included cell-cell junction, axon guidance, cell junction assembly, vasculature development, and regulation of neuron development projection.

Fig. 7: Pathways affected in monogenic (samples of all mutations pooled together) PD and pathways that are affected in neurons with a low rate of EPSCs.
figure 7

a Top downregulated GO terms in the monogenic PD compared to controls. b Top downregulated KEGG pathways in monogenic PD compared to controls. c Top up-regulated GO terms in the monogenic PD compared to controls. d Top up-regulated KEGG pathways in monogenic PD compared to controls. GO terms for upregulated genes in the monogenic PD compared to controls. e Commonly down-regulated GO terms in the monogenic neurons and the sPD neurons compared to the control neurons. f Commonly downregulated KEGG pathways in the monogenic neurons and the sPD neurons compared to the control neurons. g Downregulated GO terms in DA neurons that have a low rate of EPSCs. h Downregulated KEGG pathways in DA neurons that have a low rate of EPSCs. i Upregulated GO terms in DA neurons that have a low rate of EPSCs. j Upregulated KEGG pathways in DA neurons that have a low rate of EPSCs.

Full size image

Seeking pathways related directly to a reduction in synaptic transmission rate

The common electrophysiological phenotype that we observed in all PD lines was a drastic and significant reduction in the rate of synaptic current events. Therefore, we were interested to determine the affected pathways when the differential expression was taken relative to the EPSC rate of the investigated cell line (see “Methods”). The 10 top upregulated pathways for neurons with reduced EPSC rates are presented in Fig. 7g–j and the entire list of dysregulated GO terms and KEGG and MsigDB pathways are presented in Supplementary Tables 3–5. As expected, many synaptic pathways were dysregulated when comparing neurons that had a high-frequency rate of synaptic activity vs. neurons with a low rate of synaptic events. These included, for example, dopaminergic synapse, glutamatergic synapse, GABAergic synapse, serotonergic synapse, aminergic neurotransmitter loading into synaptic vesicles, filopodium, axon guidance, and many more. Other dysregulated pathways were similar to those that were dysregulated in PD, probably since PD lines exhibited a reduced synaptic event rate. These included many extracellular matrix pathways, focal adhesion, CAMs, melanosome, lysosome, cilium, mitochondrial pathways, endoplasmic reticulum-related pathways, and several cancer-related pathways.

Discussion

PD affects the lives of nearly one million people in the US and is the most prevalent movement disorder. The hallmarks of PD are aggregates of the α-synuclein protein that are more specific to areas in the brain with a high density of DA neurons. In some PD cases, Lewy bodies appear in those DA-dense areas. These Lewy bodies are composed of protein aggregates whose main component is the α-synuclein protein. However, it is not clear if the Lewy bodies are causing the neuronal cell death that is observed in high-density DA neuron areas or is a side effect of other processes that occur and are the actual triggers for neurodegeneration. Recent studies suggest that degeneration starts as micro aggregates of the α-synuclein protein form, long before Lewy bodies appear. Strikingly, there are several different PD-causing mutations, but most patients do not have a known genetic origin and are considered idiopathic or sporadic. Finding phenotypes that are common to PD early in the lifetime of neurons (and the patients) will help to understand disease mechanisms and develop an effective treatment. Furthermore, the use of patient-derived neurons helps to mitigate the problem of the lack of animal models for sporadic PD.

Here we report a neurophysiological phenotype that is common to DA neurons derived from PD patients for the following mutations: the A53T SNCA mutation, SNCA copy number variations (duplication as well as triplication), LRRK2, Parkin, and importantly, sPD patients. Using whole-cell patch-clamp, we found that neurons derived from PD patients all exhibited a significant reduction in the rate of synaptic activity. It is known that neurons that are plated sparsely have a low survival rate86,87; neurons need interactions with their neighboring neurons for survival. Therefore, the low connectivity that we observed in the PD neurons may exacerbate neuronal death caused by different mechanisms. Several labs have shown that reprogramming of adult cells into iPSCs erases aging signatures and epigenetic modifications81,82, and therefore the neurons in our cultures are considered to be young, and even pre-natal neurons. Consequently, the observed phenotype of a reduction in synaptic activity in these very young neurons indicates an early process and a biological predisposition that is present in the patients’ DA neurons, probably long before patients exhibit any motor deficits. Similar findings were described in SNCA mutations in mice42,43,45,47,48. In these mice, the earliest phenotype that was observed was a reduction in synaptic activity, followed by protein aggregates that, after weeks and sometimes months, evolved into the loss and death of DA neurons and motor dysfunction. Overall, our results imply that neuronal cultures derived from human patients using iPSC techniques are a good model for studying PD progression, starting with this prodromal phenotype in the neurons. Moreover, this phenotype may have important implications for the early diagnosis and prediction of disease onset.

The α-synuclein gene plays a major role in PD. The SNCA gene was only dysregulated in the neurons derived from the patients with the LRRK2, A53T, and SNCA triplication mutations, with a very large increase in the neurons derived from the patient with the SNCA triplication (Supplementary Fig. 9). It is interesting to note that the neurons derived from the patient with the SNCA duplication do not significantly overexpress SNCA, perhaps due to compensation mechanisms. However, we revealed shared dysregulated pathways analyzing gene expression of neurons derived from PD patients with different genetic mutations as well as the sporadic form of the disease and this was often consistent with, and supportive of the electrophysiology. Among these dysregulated pathways and genes, we found the CAMs. Several CAM families have been shown to localize at the synapses88,89 and to influence the assembly and function of synapses in the CNS. The PI3K-Akt signaling pathways were also overrepresented among downregulated genes in both monogenic and sPD, and they have been shown to recruit PSD-95 to synapses90. The FoxO pathway, another commonly overrepresented pathway among downregulated genes in our PD lines, has also been shown to play an important role in synaptic growth, synaptic vesicle exocytosis91, and the promotion of synaptic plasticity92. The latter two pathways have been shown to interact93. Proteoglycans and collagen fibers also exhibited dysregulation in both our monogenic and sPD lines. These make up the extracellular matrix, which more and more evidence suggests is a part of the tetrapartite synapse. The extracellular matrix-related pathways were strongly dysregulated in almost all our PD lines. Overall, we saw the dysregulation of many genes and pathways that affected synaptic formation. We hypothesize that this gene expression dosage dysregulation prevents neurons derived from PD patients from establishing effective synapses, measured in our electrophysiological experiments as a reduced rate of synaptic events. We hypothesize that PD neurons have a reduced ability to form operative synapses, which reduces the rate of synaptic events and triggers homeostatic mechanisms that take place; a vast dysregulation of genes associated with synaptic transmission-related pathways then occurs, as we observed when analyzing gene expression in our PD lines. It is interesting to note that, in those PD lines where there was a stronger decrease in the synaptic event’s rate, we also observed more abundant synaptic-related pathways that were overrepresented among upregulated genes. In the LRRK2 and the SNCA duplication, where the reduction in the synaptic rate was not as severe, we detected fewer synaptic pathways that were significantly enriched among upregulated genes, further stressing the importance of electrophysiology for the detection of subtler changes. Overall, our results strongly suggest that PD pathology starts at the synapse (as our in vitro neurons are young, and therefore reflect early brain events), in agreement with previous reports and hypotheses regarding the role of disrupted synapses in PD27,94,95. Studies have shown that loss of synaptic terminals exceeds the loss of DA cell bodies74 and that α-synuclein aggregates at the presynaptic terminals before forming Lewy bodies18,28; these results are supported by our findings of young and rejuvenated DA neurons that already exhibit synaptic deficits.

To summarize, our work demonstrates an early phenotype that is common to neurons from several PD mutations as well as sPD patients. It also demonstrates that there are common pathways that are affected in and common to PD patients (mutation-driven or sporadic). Importantly, these findings reveal that PD can be studied using iPSC-derived neuron technology, allowing us to trace the disease progression step by step. The affected pathways that we have identified through the analysis of gene expression should now be considered important targets for further research.

Methods

A written informed consent was provided by all the participants in the study to take part in the study.

Ethics

The study was approved by the Salk institute with the following approvals: IRB 09-0003 and ESCRO 07-009.

Human patients

Supplementary Tables 1 and 2 (see Supplementary) present the clinical features of the human patients who participated in this study. The first cohort, listed in Supplementary Table 1, was diagnosed by Dr. Juergen Winkler, and the second cohort (Supplementary Table 2) was diagnosed by Dr. Alexis Brice. A written informed consent was provided by all the participants in the study to take part in the study.

Dopaminergic (DA) neurons

DA neurons were generated from iPSCs based on a previously described protocol84 with modifications102 and sjdbOverhang value of 100. Raw or TPM (Transcripts per million) gene expression was quantified across all gene exons with HOMER103 using the top-expressed isoform as a proxy for gene expression, and differential gene expression was carried out on the raw counts using the edgeR104 package version 3.28.1. For each disease type, differentially expressed genes were defined as having a false discovery rate (FDR) <0.05 when comparing two experimental conditions. We also separated monogenic subjects from sPD and compared each group separately to controls using the exact test function in edgeR. We then combined all subjects (monogenic, sPD, and control) and treated the EPSC rate as a continuous variable; we conducted differential expression analysis using the GLM model in edgeR. In both cases, differentially expressed genes were defined as having an FDR < 0.05. A GO enrichment test and KEGG pathway analysis were performed using the program DAVID Bioinformatics Resources 6.8105. Overrepresentation of GO terms and KEGG pathway was determined by FDR < 0.05. MsigDB106 overrepresentation analysis was carried out using HOMER findGO.pl using the corrected Benjamini & Yakutieli method for multiple testing correction107.

Immunocytochemistry and cell imaging

Neuronal cultures were fixed with 4% paraformaldehyde for 15 min at room temperature and then treated with PBS containing 0.1% Triton X-100. After a 15 min PBS wash, cells were blocked with 5% BSA in PBS for 1 h, then incubated with the primary antibody in PBS at 4 °C overnight and the next day after a few PBS washes with secondary antibodies for 1 hour at ambient temperature. This process was followed by a 10 min incubation with DAPI and a final set of PBS washes. The coverslips were mounted on glass slides using PVA-DABCO. Primary antibodies used were rabbit anti-Tuj1, (1:500, Covance), chicken anti-MAP2 (1:400, Abcam), rabbit anti-αSynuclein (1:500, Invitrogen), rabbit anti-TH (1:500, Pel-Freez), mouse anti-PSD95 (1:500, Life Tech), rabbit anti-Synapsin I (1:500, Calbiochem), anti-Fibronectin (1:200, Sigma). Corresponding Alexa FluorTM secondary antibodies were then used (1:1000). For detecting protein aggregates, the PROTEOSTAT® Aggresome Detection kit was used according to the manufacturer’s instructions.

Confocal z-stacks were acquired with a Zeiss LSM 880 Airy scan microscope (Carl Zeiss, Microimaging Inc.) using 405 nm Diode laser, 488 nm Argon, and 543 nm HeNe lasers with a Plan NeoFluar 40×/1.3 oil DIC or a Plan-Apochromat 63×/1.4 oil DIC objective.

Imaging of cell morphology

Neurons stained for TH were imaged and analyzed with Neurolucida SW (MBF Bioscience), where neurites and soma were manually traced.

Calcium imaging

To image calcium transient neuronal cultures were incubated for 60 min in the recording solution (128 mM NaCl, 4 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 45 mM sucrose, 10 mM glucose, and 10 mM HEPES; pH is titrated to 7.4) in the presence of 4 μg/ml cell-permeant Fluo5-AM (Abcam). Cultures were then placed in fresh recording solution and imaged. The calcium transients were imaged at 10 Hz similar to previous experiments described108.

Statistical analysis

Unless otherwise stated, p values were calculated using a two-sample t-test (two-tailed).