Moyamoya Disease and the Risk of Parkinson's Disease

ABSTRACT

Objectives

Moyamoya disease (MMD) is a rare cerebrovascular disorder characterized by the progressive narrowing of arteries at the base of the brain, forming abnormal collateral vascular networks. While vascular parkinsonism is noted in MMD, its link to Parkinson's disease (PD) has not been explored. We aimed to determine whether the risk of PD is increased in patients with MMD and to identify the potential role of the RNF213 gene.

Methods

We report two cases of PD with a history of MMD associated with the Arg4810Lys variant of RNF213. Using the Korean National Health Insurance Service database, we studied the association between MMD and subsequent PD risk using Cox proportional hazards regression analysis. Moreover, we explored the interaction between α‐synuclein and RNF213 proteins by overexpressing SNCA and RNF213 in SH‐SY5Y cells. Furthermore, we confirmed the proximity of Lewy pathology and RNF213 proteins in postmortem brain tissues from three patients with PD and age‐matched controls.

Results

Among 16,830 patients aged ≥ 40 with MMD, 3415 were enrolled, excluding those with prior PD or stroke, along with a matched control group of 33,974. The analysis revealed an increased PD risk in patients with MMD (hazard ratio = 4.45 [3.40–5.82], p < 0.0001). Overexpression of RNF213 resulted in cytoplasmic inclusions, which were worsened by the co‐expression of SNCA and the Arg4810Lys variant of RNF213. Histological studies confirmed the co‐localization of α‐synuclein aggregates and RNF213 proteins in PD brain tissues.

Interpretation

This study confirms the heightened PD risk in patients with MMD and suggests a pathophysiological link through α‐synuclein and RNF213 interactions.

1. Introduction

Parkinson's disease (PD) is a neurodegenerative disease characterized by an extended latent period over decades [1]. Several studies have revealed the association between non‐degenerative diseases, such as stroke and PD, exhibiting an increased risk of PD in patients with stroke [2]. Although shared pathomechanisms such as mitochondrial dysfunction and reactive oxygen species have been suggested, the discrepancy in key features between vasculopathy and alpha‐synuclein (αSyn) aggregation remains difficult to harmonize.

Moyamoya disease (MMD) is a non‐atherosclerotic cerebrovascular disorder characterized by progressive steno‐occlusion at the terminal end of the internal carotid arteries (ICAs) and compensatory smoke‐like collateral vascular networks at the base of the brain [3, 4]. Most clinical manifestations of MMD are associated with cerebral ischemia or infarction in the affected arteries and intracerebral hemorrhage, including transient ischemic attack, headache, seizures, cognitive impairment, and various movement disorders [4, 5]. Although parkinsonism, defined as bradykinesia with resting tremor or rigidity, has rarely been reported as a clinical presentation of MMD, the reported cases were accompanied by parenchymal ischemic lesions, suggesting vascular parkinsonism [6, 7]. The incidence of MMD is approximately 10 times higher in East Asian countries such as Korea and Japan than in other regions and commonly exhibits bimodal age distributions with peaks at approximately 10 and 30–40 years [4, 8]. Although the etiology of MMD remains elusive, the p.Arg4810Lys (c.14429G>A) polymorphism of the RNF213 was identified as the major susceptibility gene for MMD in a genome‐wide association study and linkage analysis in East Asian populations [9, 10]. A rare form of homozygotes for the RNF213 polymorphism exhibits a higher rate of penetrance, earlier age of onset, and more severe disease burden than heterozygotes, suggesting a gene‐dosage effect [11]. The RNF213, or mysterin gene encodes moyamoya steno‐occlusive disease‐associated AAA+ and RING (really interesting new gene) finger protein, which uniquely encompasses both ubiquitin ligase and ATPase activities [10]. The AAA+ modules of RNF213 are involved in the dynamic change of oligomeric forms and the targeting of lipid droplets, and the mutant lacking ATP‐binding motifs or the RING finger ubiquitin ligase domain disrupts lipid droplet targeting and enhances aggregate formation [12, 13]. Although ubiquitin ligase activity is reminiscent of Parkin mutations in PD, the characteristics of RNF213 to form aggregates and interact with lipid droplets suggest a potential interaction with αSyn, based on the findings that αSyn co‐aggregates with other self‐aggregating proteins and has extensive reciprocal interactions with lipid droplets. However, whether MMD with its RNF213 variant has any pathological interaction with αSyn remains unclear.

Here, we report two patients with MMD who subsequently developed PD and provide supportive evidence for an increased risk of PD in these patients using the Korean National Health Insurance Service (KNHIS) database. To investigate the pathological association between MMD and PD, we performed in vitro studies centered on the common variant of RNF213, as well as postmortem studies.

2. Methods

2.1. Data Source

This retrospective cohort study utilized the KNHIS database to collect population‐based, nationally representative data. The health insurance system of Korea is a social insurance program in which 97% of all citizens are enrolled, with monthly fees based on income and number of household members [14]. The KNHIS also administratively manages the medical expenses of the remaining 3% of Medicaid beneficiaries with the lowest income and reimburses medical institutions. Therefore, the KNHIS database contains information on the demographics, socioeconomic status, and healthcare use of the entire population, using diagnostic codes based on the International Classification of Diseases, Tenth Revision (ICD‐10). Furthermore, the KNHIS provides a national health checkup program every 2 years for individuals aged ≥ 40 years, indicating a high rate of examinations (75.9% in 2023) [15], and the pertinent data include physical examinations, self‐administered questionnaires on health behavior, and laboratory tests.

2.2. Study Participants

The operational definition of MMD was based on the ICD‐10 diagnostic code (I675) or registration with the corresponding code (V128) in the Rare Intractable Disease (RID) program. The RID program operated in South Korea and is an administrative system that provides copayment for 90% of relative medical costs for patients with rare diseases or specific incurable diseases, including MMD and PD. The diagnostic accuracy of patients registered in the RID program is considered reliable because the disease is registered and regularly updated by attending specialists. The criteria for RID registration for MMD were neurological symptoms or signs of cerebrovascular abnormalities confirmed by brain imaging. From 2008 to 2020, individuals aged ≥ 40 years who used medical care with an operational diagnosis of MMD were enrolled in the study cohort (n = 16,830), and the date of medical use was defined as the index date (Figure S1). The following exclusion criteria were applied: patients who (1) had a history of PD before MMD diagnosis, which was defined as patients receiving medical care to treat PD for 3 years before the index date (n = 496); (2) had a previous or concurrent ischemic or hemorrhagic stroke according to the ICD‐10 diagnostic codes with claim data for brain CT and/or MRI to validate the stroke (n = 9928; Table S1); and (3) were not matched with controls for age, sex, and index date (n = 2994). For the control group, we obtained an initial pool of 168,316 individuals who did not have an MMD diagnosis and were matched by age, sex, and index date (1:10 matching). After applying identical exclusion criteria to the MMD group, 3415 patients with MMD were matched (1:10) with 33,974 controls. The mean follow‐up duration (SD) was 8.7 (3.8) years in patients with MMD, which was shorter than that in controls (9.3 [3.0] years; p < 0.0001). Among the final participants, 1598 patients with MMD and 15,644 controls had national health check‐up data within 2 years of the index date.

2.3. Study Outcome and Follow‐Up

The primary outcome was newly developed PD, which was operationally defined using ICD‐10 codes (G20). Although PD is also listed in the RID program with a code of V124, the RID program code for PD was not utilized because an individual could not be registered with two or more diseases of the RID program simultaneously, such as MMD and PD. All participants were followed up from the index date until the corresponding PD onset date, censored date, or end of the study (December 31, 2020).

2.4. In Vitro Analysis of SNCA and RNF213 Overexpression in SH‐SY5Y Cells and Immunohistochemistry of Postmortem Brain Tissues

All experimental materials, preparation, immunoblotting, immunofluorescence staining, co‐immunoprecipitation, cell viability assay, postmortem study, and statistical analysis are described in the Supporting Information Methods.

2.5. Statistical Analysis

The baseline characteristics of the study participants are described as mean (SD) for age and number (%) for categorical values. Comparisons between patients with MMD and controls were performed using Student's t‐test for continuous variables and a chi‐square test or Fisher's exact test for categorical variables. Cox proportional hazard regression models were used to evaluate the association between MMD and risk of PD. A multivariate analysis was conducted by sequentially adjusting for the following covariates, which are defined in Table S2: Model 1 included age, sex, and income; Model 2 included those of Model 1, hypertension, diabetes, dyslipidemia, ischemic heart disease, chronic kidney disease, and depression; and Model 3 included those of Model 2, smoking, alcohol drinking, physical activity, and body mass index. All adjusted hazard ratios were calculated using 95% confidence intervals. In addition, to clarify the temporal relationship between MMD and PD onset, lag periods of 1, 3, and 5 years from the index date were used in additional sensitivity analyses. All statistical analyses were performed using SAS (Statistical Analysis Software, version 9.4, SAS Institute Inc., Cary, North Carolina, USA).

2.6. Study Approval

The clinical study was approved by the Institutional Review Board of Kyung Hee University Hospital (IRB No. 2021‐02‐008; 2022‐05‐029). Written informed consent for the case study was obtained from two patients, and consent for the use of de‐identified and anonymized data was waived. Ethics approval for the postmortem study was provided by the Institutional Review Board of Seoul National University Hospital (IRB No. H2301‐095‐1396).

2.7. Data Sharing

The authors are restricted from sharing the KNHIS database, and researchers can request access to the data from the KNHIS website (https://nhiss.nhis.or.kr). All other data generated in this study are available to the corresponding author upon reasonable request.

3. Results

3.1. Clinical Vignettes

3.1.1. Case 1

A 63‐year‐old woman was hospitalized with two episodes of transient ischemic attacks, during which she experienced slurred speech and numbness on the right side of the face and arm lasting for 10 and 15 min, respectively. Magnetic resonance imaging (MRI) revealed no parenchymal diffusion‐restricted lesions. However, the ivy sign, characterized by high signal intensity, was found along both cerebral cortical sulci, indicating prominent leptomeningeal collaterals on the fluid‐attenuated inversion recovery sequence [16]. Digital subtraction angiography revealed occlusion in both supraclinoid ICAs and near the collaterals from the posterior and transdural circulation, which led to the diagnosis of MMD (Figure S2A). Technetium Tc99m ethyl cysteine dimer single‐photon emission computed tomography (SPECT) detected no abnormal cerebral perfusion before or after acetazolamide injection. Genetic testing for RNF213 polymorphism confirmed a susceptible heterozygous variant, p. Arg4810Lys. The patient took antiplatelet medications and was followed up for 80 years without significant cerebrovascular events. Seventeen years after her diagnosis of MMD, she complained of gait difficulties, and a neurological examination revealed parkinsonism, bradykinesia, rigidity on the left side, hypomimia, and abnormal postural reflex. Follow‐up SPECT revealed decreased perfusion in the temporal cortex, frontal cortex, and basal ganglia of the left hemisphere and the parietal and frontal cortex of the right hemisphere with normal perfusion reserve. ¹⁸F‐fluorodeoxy glucose (FDG) positron emission tomography (PET) was normal. Dopamine transporter imaging (¹⁸F‐N‐(3‐fluoropropyl)‐2β‐carboxymethoxy‐3β‐(4‐iodophenyl) nortropane [FP‐CIT] PET) revealed bilateral severe loss of uptakes in dopamine synaptic terminals with anteroposterior gradient and asymmetry, which was more severe on the right side (Figure S2B). Newly developed vascular lesions were not observed on follow‐up brain MRI (Figure S2C). She presented with hyposmia at diagnosis and developed constipation as the disease progressed. The patient responded well to a minimal dose of levodopa (100 mg per day), which improved her gait and overall bradykinesia. However, after 2 years, she developed visual hallucinations, preventing an increase in the dosage beyond 175 mg of levodopa per day. Four years post‐diagnosis, she began experiencing frequent falls. We diagnosed this case as probable PD according to the 2015 Movement Disorder Society PD diagnostic criteria [17].

3.1.2. Case 2

A 60‐year‐old woman visited a clinic with left‐hand tremors while answering on a phone or lifting objects. She visited another hospital at the age of 56 years because of a headache and was diagnosed with MMD with preserved cerebral perfusion and no vascular lesions on neuroimaging. At our clinic, neurological examinations revealed parkinsonism, hypomimia, bradykinesia, rest tremor, and rigidity in the left arm. Brain MR angiography exhibited severe stenosis in both distal ICA with compensatory supply from the posterior circulation, compatible with MMD (Figure S2D). Hexamethylpropylene amine oxime technetium Tc99m SPECT revealed normal cerebral perfusion at baseline and after acetazolamide injection. FP‐CIT PET revealed bilateral loss of uptake in dopamine synaptic terminals with an anteroposterior gradient and asymmetry, which was more severe on the right side (Figure S2E) without corresponding vascular lesions in the brain MI (Figure S2F). Her FDG PET was normal. She denied a family history of MMD, and genetic testing for RNF213 detected a heterozygous polymorphism (p.Arg4810Lys). While her tremor remained stationary with 1 mg rasagiline, she developed an intracerebral hemorrhage in the left occipital lobe with diffuse subarachnoid hemorrhage at 62 years of age. Her neurological status remained unchanged after the hemorrhage. At 63 years of age, the patient reported non‐motor symptoms, including constipation, hyposmia, urinary symptoms, and suspected REM sleep behavior disorder. At 65 years of age, her motor symptoms were managed with an add‐on of pramipexole at 0.375 mg per day. Given the presence of rest tremor and olfactory loss, this case is consistent with the criteria for clinically established PD.

Both patients were diagnosed with adult‐onset sporadic MMD with the heterozygous RNF213 polymorphism p.Arg4810Lys and PD. In the first patient, although SPECT exhibited decreased perfusion in the left basal ganglia, the more sensitive FDG PET was normal. In the second patient, brain imaging revealed no structural or functional abnormalities, despite vessel deformities. In both patients, dopamine transporter imaging confirmed severe dopamine deficiency at the nigrostriatal synaptic terminals, and dopaminergic medications were effective, consistent with idiopathic PD. We summarized the clinical course of the two case patients in Table S3.

Because both patients had rare diseases, investigating whether the development of the diseases was pathogenetically interconnected was necessary. Thus, we compared the incidence of PD between patients diagnosed with MMD and a control group without MMD using the KNHIS database and performed experimental studies.

3.2. The Korean NHIS Database: MMD and Subsequent PD

3.2.1. Baseline Characteristics of Study Participants

The mean age (Standard deviation [SD]) of study participants (N = 37,389) at the index date was 43.5 (7.2) years, and 65.4% were female (Table 1). Compared to the controls, the patients with MMD were older (45.5 ± 6.8 vs. 43.4 ± 7.2 years old), higher in the proportion of bottom quartile of income (25.2% vs. 17.5%), and suffered more frequently from hypertension (41.0% vs. 9.3%), diabetes (11.0% vs. 4.0%), dyslipidemia (34.6% vs. 7.7%), ischemic heart disease (22.5% vs. 5.2%), chronic kidney disease (1.3% vs. 0.6%), and depression (21.0% vs. 8.8%). Patients with MMD had fewer current smokers (8.5% vs. 14.4%) and more nondrinkers (89.9% vs. 85.6%), were more likely to engage in regular exercise more than five times a week (21.7% vs. 20.1%), but a higher proportion were obese (42.9% vs. 29.1%) compared to the controls.

TABLE 1.

Baseline characteristics of study participants.

Characteristics	Total (no. [%])	MMD status (no. [%])		p
		No	Yes
	37,389 (100)	33,974 (100)	3415 (100)
Age (year, mean [SD])	43.5 [7.2]	43.3 [7.2]	45.5 [6.8]	< 0.0001
40–59	35,376 (94.6)	32,144 (94.6)	3232 (94.6)	0.9454
≥ 60	2013 (5.4)	1830 (5.4)	183 (5.4)
Sex
Male	12,935 (34.6)	11,759 (34.6)	1176 (34.4)	0.8372
Female	24,454 (65.4)	22,215 (65.4)	2239 (65.6)
Income
< 25%	6801 (18.2)	5939 (17.5)	862 (25.2)	< 0.0001
≥ 25%	30,588 (81.8)	28,035 (82.5)	2553 (74.8)
Hypertension	4546 (12.2)	3146 (9.3)	1400 (41.0)	< 0.0001
Diabetes	1752 (4.7)	1375 (4.0)	377 (11.0)	< 0.0001
Dyslipidemia	3785 (10.1)	2603 (7.7)	1182 (34.6)	< 0.0001
Ischemic heart disease	2532 (6.8)	1764 (5.2)	768 (22.5)	< 0.0001
Chronic kidney disease	254 (0.7)	208 (0.6)	46 (1.3)	< 0.0001
Depression	3704 (9.9)	2988 (8.8)	716 (21.0)	< 0.0001

No. (%)	17,242 (100)	15,644 (100)	1598 (100)	p
Smoking status
Never‐smoker	13,536 (78.5)	12,234 (78.2)	1302 (81.5)	< 0.0001
Ex‐smoker	1310 (7.6)	1150 (7.4)	160 (10.0)
Current smoker	2396 (13.9)	2260 (14.4)	136 (8.5)
Alcohol consumption
0 day/week	14,823 (86.0)	13,386 (85.6)	1437 (89.9)	< 0.0001
1–2 day/week	1690 (9.8)	1568 (10.0)	122 (7.6)
3–4 day/week	538 (3.1)	508 (3.2)	30 (1.9)
≥ 5 day/week	191 (1.1)	182 (1.2)	9 (0.6)
Physical activity
0 day/week	5794 (33.6)	5273 (33.7)	521 (32.6)	< 0.0001
1–2 day/week	4645 (26.9)	4251 (27.2)	394 (24.7)
3–4 day/week	3308 (19.2)	2972 (19.0)	336 (21.0)
≥ 5 day/week	3495 (20.3)	3148 (20.1)	347 (21.7)
Body mass index, kg/m2
Underweight (< 18.5)	754 (4.4)	707 (4.5)	47 (2.9)	< 0.0001
Normal weight (18.5–22.9)	7514 (43.6)	6983 (44.6)	531 (33.2)
Overweight (23.0–24.9)	3737 (21.7)	3403 (21.8)	334 (20.9)
Obesity class I (25.0–29.9)	4427 (25.7)	3864 (24.7)	563 (35.2)
Obesity class II (≥ 30.0)	810 (4.7)	687 (4.4)	123 (7.7)

Abbreviations: MMD, Moyamoya disease; SD, standard deviation.

3.2.2. Risk of PD in Patients With MMD

Until 2021, 259 patients developed PD (75 PD in the MMD group and 184 in the control group). Incidence of PD was 253.4 per 100,000‐person year in patients with MMD, which is higher than that of controls, 58.4 per 100,000‐person year. Patients with MMD exhibited a 4.45‐fold higher risk of PD than the control in the crude model (95% confidence interval [CI], 3.40–5.82; Table 2; Figure 1). The high risk of incident PD in patients with MMD was consistent across models 1 to 3 (model 1: adjusted hazard ratio [aHR], 3.68 [95% CI, 2.78–4.86]; model 2: aHR, 3.01 [95% CI, 2.23–4.07]; model 3: aHR, 2.75 [95% CI, 1.76–4.29]). Applying lag times of 1, 3, and 5 years for PD incidence, the association of increased risk of PD with MMD did not change across models 1 and 2, whereas model 3 could not be applied due to the small sample size.

TABLE 2.

Risk of Parkinson's disease in Moyamoya disease.

By MMD status	No.	Event, No.	Incidence rate (per 100,000 person‐y)	HR [95% CI]
				Crude	Model 1 a	Model 2 b	Model 3 c
No lag periods
Non‐MMD	33,974	184	58.4	1 [Reference]	1 [Reference]	1 [Reference]	1 [Reference]
MMD	3415	75	253.4	4.45 [3.40–5.82]	3.68 [2.78–4.86]	3.01 [2.23–4.07]	2.75 [1.76–4.29]
1‐year lag period
Non‐MMD	33,722	176	56.2	1 [Reference]	1 [Reference]	1 [Reference]	NA
MMD	3388	64	216.4	3.99 [2.99–5.31]	3.38 [2.52–4.56]	2.69 [1.95–3.70]
3‐year lag period
Non‐MMD	33,229	173	56.0	1 [Reference]	1 [Reference]	1 [Reference]	NA
MMD	3339	47	159.4	2.96 [2.15–4.09]	2.46 [1.76–3.43]	2.00 [1.40–2.85]
5‐year lag period
Non‐MMD	26,080	114	41.5	1 [Reference]	1 [Reference]	1 [Reference]	NA
MMD	2608	33	124.8	2.41 [1.89–3.07]	2.56 [1.71–3.82]	2.14 [1.39–3.28]

Abbreviations: CI, confidence interval; HR, hazard ratio; MMD, Moyamoya disease; NA, not applicable due to the small sample size.

^a Model 1 was adjusted for age, sex, and income level.

^b Model 2 was adjusted additionally for hypertension, diabetes, dyslipidemia, ischemic heart disease, chronic kidney disease, and depression to model 1.

^c Model 3 was adjusted additionally for smoking, alcohol drinking, exercise, and body mass index to model 2.

FIGURE 1.

Cumulative incidence of Parkinson's disease in patients with moyamoya disease. MMD, Moyamoya disease; PD, Parkinson's disease.

3.3. In Vitro Experiment

3.3.1. RNF213 Aggregate Puncta Formation and Cytotoxicity in SH‐SY5Y Cells Are Enhanced by p.Arg4810Lys Mutation

To investigate the pathological role of RNF213 and its mutant p.Arg4810Lys, the neuroblastoma cell line SH‐SY5Y was transfected with RNF213 wild‐type (WT) and p.Arg4810Lys. RNF213 WT and p.Arg4810Lys mutant (MT) were robustly overexpressed when monitored by western blot using an anti‐RNF213 specific antibody (Figure 2A). The steady‐state RNF213‐Arg4810Lys expression level was slightly higher than that of RNF213 WT (Figure 2A,B). Endogenous RNF213 in SH‐SY5Y cell lysates was very low and was undetectable at the exposure time used to detect overexpressed exogenous RNF213 (Figure 2A). The immunofluorescence intensities of RNF213 WT and MT in SH‐SY5Y cells transfected with the corresponding constructs were based on the expression levels observed in western blotting, with approximately two‐fold higher expression levels of RNF213 MT compared to those of RNF213 WT (Figure 2C,D). Strikingly, RNF213 WT formed aggregate puncta when expressed in SH‐SY5Y cells (Figure 2C,E). The aggregate puncta formation was enhanced almost three‐fold by the Arg4810Lys mutation (Figure 2C,E).

FIGURE 2.

Moyamoya disease‐associated Arg4810Lys mutation of RNF213 increases its aggregate formation and cytotoxicity. (A) Representative western blots showing expression of RNF213 wild‐type (WT) and Arg4810Lys mutant (MT) in total lysates from SH‐SY5Y cells transiently transfected (48 h) with the indicated constructs. β‐actin served as an internal loading control. (B) Quantification of relative expression of RNF213 WT and MT in the panel A normalized to β‐actin (n = 3 per group). (C) Representative immunofluorescence image of RNF213 WT and MT expression in SH‐SY5Y cells transiently transfected (48 h) with the indicated constructs. RNF213 antibody was used to label exogenous RNF213 expression. Scale bar = 10 μm. (D) Quantification of relative RNF213 immunofluorescence signals in SH‐SY5Y cells transfected with mock, RNF213‐WT, and RNF213‐MT (n = 9 images from 3 separate experiments per group). (E) Quantification of RNF213 aggregate puncta numbers in SH‐SY5Y cells transfected with mock, RNF213‐WT, and RNF213‐MT (n = 18 cells from 3 separate experiments). (F) Cell viability assessment of SH‐SY5Y cells transfected with the indicated constructs (mock, RNF213‐WT, RNF213‐MT) (48 h) and treated with hydrogen peroxide (100 μM, 24 h). Cell viability was determined using s trypan blue exclusion assay (n = 6 separate experiments per group). Data in all panels are presented as the mean ± standard error of the mean. **p < 0.01, and ***p < 0.001, one‐way analysis of variance test followed by Tukey's post hoc analysis.

Overexpression of RNF213 WT caused cytotoxicity, leading to a 26% reduction in cell viability, as determined using a trypan blue exclusion assay (Figure 2F). Cytotoxicity by RNF213 MT expression in SH‐SY5Y cells was further augmented by RNF213 WT expression, resulting in 59% cell viability, compared with 74%. Synergistic interactions with oxidative stress were also examined after cotreatment with hydrogen peroxide. Hydrogen peroxide treatment increased the cytotoxicity of mock, RNF213‐WT, and RNF213‐MT cells to a similar extent in an additive manner. In conclusion, the overexpression study suggests that the accumulation of RNF213 causes aggregate puncta formation and cytotoxicity, most prominently via p.Arg4810Lys.

3.3.2. Co‐Aggregate Puncta Formation and Cytotoxicity of αSyn Are Augmented by co‐Expression of RNF213‐4810K in SH‐SY5Y Cells

We sought to determine the relationship between RNF213 expression and αSyn aggregation because RNF213 forms aggregate puncta. αSyn expression alone showed an evenly distributed αSyn immunofluorescence signal without the formation of any distinct aggregate puncta (Figure S3A,B). However, co‐expression of RNF213‐WT and αSyn led to aggregate puncta formation, in which the RNF213‐WT and αSyn signals overlapped (Figure 3A,B). Compared to the number of aggregate puncta induced by RNF213‐WT or ‐MT expression alone (Figure 2E), co‐expression of RNF213/αSyn led to modestly higher total aggregate puncta formation, respectively (Figure 3A,C). Strikingly, compared to RNF213‐WT/αSyn, RNF213‐MT/αSyn resulted in an increased tendency to form RNF213 and αSyn double‐positive aggregate puncta rather than RNF213 only aggregate (Figure 3A,B). For both RNF213‐WT and ‐MT co‐expression with αSyn, the proportion of αSyn only aggregates was minimal (Figure 3A,B), suggesting that RNF213 plays a role in inducing αSyn co‐aggregation. A more than two‐fold increase in the total number of aggregate puncta in SH‐SY5Y cells with RNF213‐MT/αSyn co‐expression was observed as compared to those with RNF213‐WT/αSyn co‐expression (Figure 3A,C). Immunostaining for ubiquitin revealed that these RNF213 mutant–associated puncta were largely positive for ubiquitin (Figure S3C), suggesting that the aggregates are ubiquitinated and may be substrates for protein quality control pathways. To further evaluate whether RNF213 interacts with α‐synuclein, we performed co‐immunoprecipitation assays in SH‐SY5Y cells co‐transfected with HA–α‐synuclein and either FLAG‐tagged wild‐type RNF213 (RNF213‐WT) or a mutant form (RNF213‐MT). Cell lysates were immunoprecipitated using an anti‐FLAG antibody, and the presence of HA–α‐synuclein in the precipitates was analyzed by immunoblotting. The results revealed that RNF213‐WT co‐immunoprecipitated with HA–α‐synuclein, indicating a physical association between the two proteins (Figure S3D). Notably, the amount of HA–α‐synuclein co‐precipitated with RNF213‐MT was substantially increased compared to RNF213‐WT (Figure S3D), suggesting that the RNF213 mutation enhances its interaction with α‐synuclein. These findings support a mutation‐dependent modulation of RNF213–α‐synuclein binding, which may have implications for aggregate formation and disease pathogenesis.

FIGURE 3.

RNF213‐Arg4810Lys enhances a‐synuclein aggregation and cytotoxicity. (A) Representative immunofluorescence images of RNF213 (WT, and MT) and a‐synuclein in SH‐SY5Y cells transfected (48 h) with the indicated combination of constructs using anti‐RNF213 and α‐synuclein antibody, respectively. Scale bar = 10 μm. (B) Analysis of aggregate puncta composition in SH‐SY5Y cells transfected with HA‐a‐synuclein (HA‐αSyn) and RNF213 (WT or MT) (n = 18 cells from 3 separate experiments). (C) Quantification of total number of aggregate puncta per cell in SH‐SY5Y cells transfected with the indicated constructs (n = 18 cells from 3 separate experiments). (D) Trypan blue exclusion cell viability assessment of SH‐SY5Y cells transfected with the indicated combination of constructs (48 h; n = 6 separate experiments per group). Data in all panels are presented as the mean ± standard error of the mean. ***p < 0.001, one‐way analysis of variance test followed by Tukey's post hoc analysis.

The synergistic interaction between RNF213 and αSyn in the induction of cytotoxicity was investigated in SH‐SY5Y cells by transfection with expression plasmids and subsequent trypan blue exclusion assays. While αSyn expression alone failed to induce cytotoxicity in SH‐SY5Y cells (Figure S3E), co‐expression of αSyn increased the cytotoxicity induced by RNF213‐WT or RNF213‐MT (Figure 3D). Co‐expression of αSyn increased RNF213‐WT‐ and RNF213‐MT‐induced cytotoxicity by 14.2% and 16.3%, respectively (Figure 3D). These results indicated the pathological role of RNF213 and its mutation in the formation of αSyn aggregate puncta and αSyn toxicity.

3.3.3. Increased RNF213 Expression and Distribution to Lewy Inclusion in Postmortem Brains From Patients With PD

We determined the pathological relevance of RNF213 in patients with PD (Table S3; Figure S4). First, we examined potential alterations in RNF213 expression and distribution patterns in postmortem brain sections (pons) from patients with PD and age‐matched healthy controls. Immunohistochemistry revealed high expression levels of RNF213 in postmortem brain sections from patients with PD compared to healthy controls (Figure 4A,B). High‐magnification images of anti‐RNF213 immunohistochemistry showed densely stained aggregate puncta distributed mostly in the cytoplasm of cells in the brains of postmortem patients (Figure 4A). Since the morphology of RNF213 distribution displayed aggregate puncta, we subsequently performed co‐immunofluorescence labeling of RNF213 with the Lewy body marker pS129‐αSyn in postmortem brain sections. Strikingly, co‐immunofluorescence revealed that most of the pS129‐αSyn positive aggregates showed RNF213 signals (Figure 4C,D), indicating a strong distribution of RNF213 in Lewy bodies.

FIGURE 4.

RNF213 expression is increased and colocalized with Lewy body in the postmortem brains of Parkinson's disease (PD) patients. (A) Representative immunohistochemistry images of RNF213 expression in postmortem brain sections (Pons) of patients with PD and age‐matched healthy subjects using anti‐RNF213 antibody. Enlarged images are shown in the bottom panel. Scale bar = 20 μm. (B) Quantification of RNF213 expression levels in the pons region of postmortem brains from healthy controls and patients with PD (n = 3 subjects per group). (C) Co‐immunofluorescence images showing expression and distribution of pS129‐αSyn and RNF213 in the pons postmortem brain sections from patients with PD and healthy controls monitored using anti‐RNF213 and pS129‐αSyn antibodies, respectively. Scale bar = 20 μm. (D) Analysis of aggregate puncta composition in the pons postmortem brain sections from patients with PD (n = 30 cells from 3 PD). Data in all panels are presented as the mean ± standard error of the mean. ***p < 0.001, Unpaired Student's t‐test.

4. Discussion

This sequential study began with clinical cases of the development of PD in patients with adult‐onset MMD harboring the RNF213 (p.Arg4810Lys) variant, confirming the increased risk of PD in patients with MMD using a nationwide cohort database. It also provided in vitro evidence of worsened aggregation and cytotoxicity by the RNF213 p.Arg4810Lys variant and SNCA co‐expression. Finally, we revealed the coexistence of RNF213 and αSyn aggregates in the postmortem brains of patients with PD.

4.1. Increased Risk of PD in MMD

In this cohort study, patients with MMD had a 4.45‐fold higher risk of developing PD than those without MMD. The proportion of patients with MMD with known risk factors for PD was significantly higher than that of those without MMD, including old age, components of metabolic syndrome, depression, and no tobacco smoking, alcohol consumption, and physical activity [18, 19, 20]. Accordingly, adjusted models were proposed to address potential confounding factors, and a positive association between MMD and incident PD was maintained. However, after adjusting for demographic factors, comorbidities, and lifestyle modifiers, the risk of developing PD in patients with MMD was slightly attenuated by 2.75 times.

Case studies suggested that parkinsonism in two patients with MMD was unlikely to be attributable to a previous stroke or cerebrovascular insufficiency, and they all harbored the RNF213 genetic variant. Although the limited methodology of the study using the KNHIS database could not exclude patients with MMD who had vascular events within the index period, the database was analyzed to determine the association between these two uncommon diseases, excluding patients with MMD who were diagnosed with stroke before the index date.

As MMD exhibits discrepancies in prevalence between Asian and Western populations, further studies are needed to confirm its association with the risk of PD in a Western cohort. Moreover, considering the bimodal age distribution and different RNF213 variant frequencies in Asian populations [9, 21], comparing the incidence of PD between pediatric and adult‐onset patients with MMD may yield significant findings.

4.2. Aggregate Formation and Cytotoxicity Associated With Arg4810Lys Mutation of RNF213

We discovered that RNF213 overexpression in neuroblastoma cell lines causes cytotoxicity and self‐aggregated puncta formation, both of which are increased by the MMD susceptibility mutation p.Arg4810Lys. Since the recessive PD gene parkin has E3 ligase activity known to be beneficial to cells, preventing neuronal death against diverse toxic stimuli, our findings are difficult to explain in terms of the E3 ligase activity of RNF213. Lipid accumulation is associated with lipotoxicity and metabolic disorders [22]. RNF213 can physically interact with lipid droplets [12] and its overexpression may increase the quantity and size of lipid droplets in cells. Abnormal interactions between lipid droplets and αSyn could underlie the gain of the toxic function of RNF213 WT and Arg4810Lys mutants in SH‐SY5Y cells [23].

4.3. Pathophysiological Interaction of RNF213 With αSyn

Coaggregation of RNF213 and αSyn was first recognized in our study. RNF213 has previously been reported to form a homohexamer and to target lipid droplets [12]. The ATPase domain of RNF213 is required for its oligomerization. Consistent with this report, overexpression of RNF213 produced aggregate puncta in SH‐SY5Y cells, and this aggregate formation was robustly enhanced by the p.Arg4810Lys mutation in the present study. Moreover, both the WT and p.Arg4810Lys RNF213 facilitated αSyn coaggregation. The mechanism by which RNF213 causes the coaggregation of αSyn remains unclear. Similar to AIMP2 or PAR in previous studies [24, 25], RNF213 aggregates likely function as seeds contributing to αSyn coaggregation. The dysregulated interaction of RNF213 with lipid droplets may influence αSyn aggregation due to the extensive interactions between lipids and αSyn [26]. Neuronal lipid droplet content reportedly increases in neurodegenerative diseases. Glucocerebrosidase mutations and defective lipid metabolism can contribute to the pathogenesis of both Gaucher's disease and PD. Investigating other RNF213 mutations regarding their association with PD and αSyn aggregation would be informative. In addition, given the observed interaction between RNF213 and α‐synuclein, it would be valuable to investigate whether RNF213 also contributes to the pathogenesis of other α‐synucleinopathies, such as multiple system atrophy.

4.4. Hypothetic Pathophysiological Links Between RNF213 Angiopathy and PD

MMD and PD are distinct disorders with different clinical manifestations; MMD is a cerebrovascular disorder, whereas PD is a neurodegenerative disorder. However, common pathophysiological links between MMD and PD may exist. Although the precise function of the RNF213 protein remains to be elucidated, it has been implicated in pathological angiogenesis with pericyte reduction and blood–brain barrier (BBB) impairment accompanied by microglial activation and elevated levels of pro‐inflammatory cytokines in mouse models [27]. Disruption of the BBB has been observed in PD and may contribute to neuroinflammation and dopaminergic neuronal loss by permitting the infiltration of inflammatory cells and potentially toxic substrates [28]. Neuroinflammation in PD is associated with microglial activation in response to neuronal damage, which promotes the release of pro‐inflammatory cytokines including TNF‐alpha, IL‐1beta, and IL‐6, reactive oxygen species, and nitric oxide, perpetuating further inflammatory processes and the degeneration of dopaminergic neurons [29, 30].

In addition, dysregulation of RNF213 inhibits endoplasmic reticulum (ER) stress [31] and attenuates T‐cell‐mediated immune responses [32]. ER stress is involved in protein folding and processing. The accumulation of misfolded αSyn in PD can lead to ER stress, which triggers the unfolded protein response and attempts to restore cellular homeostasis [33]. The disruption of ER stress can diminish the process of eliminating αSyn aggregates within dopaminergic neurons, ultimately leading to neuroinflammation, apoptosis, and cell loss [34]. Finally, emerging evidence indicates that adaptive immune responses, including T‐cell activation, play a role in PD, in which T‐cells might infiltrate the brain and potentially contribute to neuroinflammation and neuronal death [35]. Some studies have suggested that PD may have an autoimmune component with T cells targeting neuronal antigens and exacerbating neurodegeneration [36].

However, the RNF213 variant (p.Arg4810Lys) is considered a major founder variant only in the East Asian population, and its penetrance is low, with 1 affected individual out of 150 carriers of the variant [37]. Many other variants are also associated with MMD, other than RNF213 [4], suggesting that impairment of RNF213 is not a sufficient condition for the development of MMD. Therefore, the pathophysiology of both MMD and PD is commonly based on multiple hit hypotheses, in which BBB breakdown, neuroinflammation, and intracellular homeostasis are considered candidates for overlapping hits to trigger or accelerate both diseases [38, 39].

4.5. Clinical Relevance of RNF213 Expression and Distribution in PD and Its Implication for PD Diagnosis and Treatments

We report that RNF213 expression is relatively high in the postmortem brains of patients with PD as compared to age‐matched healthy controls. Given that RNF213 overexpression in SH‐SY5Y cells is responsible for the increased cytotoxicity and αSyn aggregation, RNF213 accumulation may contribute to the pathology of PD. Whether increased transcriptional activation of RNF213 or impaired degradation of RNF213 aggregates results in RNF213 accumulation remains unclear. If RNF213 expression and its aggregation contribute to αSyn aggregation in PD, agents that regulate RNF213 expression could serve as useful and novel strategies for disease‐modifying therapies for PD. To further clarify the mechanisms and quantify RNF213 expression, future studies using fresh‐frozen brain tissues are needed to perform RT‐qPCR and western blot analyses in PD‐affected regions.

4.6. Limitations

This study had several limitations. First, we attempted to find pathological links between MMD and PD from RNF213 variants, a strong genetic risk factor for MMD in East Asians; however, stratified analyses were not applicable in this nationwide cohort study because gene test results were not available. Second, patients diagnosed with ischemic or hemorrhagic stroke before PD diagnosis were excluded to avoid probable vascular parkinsonism in MMD; however, cerebral perfusion status could not be corroborated. Therefore, we could not conclude that parkinsonism in the patients with MMD purely stems from neurodegeneration, not cerebral hemodynamics. Furthermore, to ensure that the observed association was specific to PD and not influenced by parkinsonism resulting from other neurodegenerative diseases, we conducted additional analyses excluding conditions such as other tauopathies and Alzheimer's disease. These analyses demonstrated that the increased incidence of PD in patients with MMD remained consistent (see Table S5). Third, although we included multiple covariates, additional potential confounders associated with MMD treatment such as calcium channel blockers, antiplatelet drugs, and surgical procedures were not considered. Fourth, this study included patients with MMD with missing health checkup data due to the small number of patient groups; therefore, the model including health‐related habits could not be representative of all patients with MMD. Lastly, common limitations of retrospective cohort studies using claims data from South Korea exist, including possible misdiagnosis of atypical parkinsonian disorders; detection bias due to increased testing, screening, or surveillance in patients with MMD compared to controls; and issues of limited generalizability.

Overall, we found that MMD is associated with an increased risk of PD. A common gene of MMD, RNF213, plays a role in αSyn aggregation and cytotoxicity in in vitro cells and postmortem tissues, linking the rare cerebrovascular disease, MMD, and the neurodegenerative disorder, PD. Further studies are necessary to determine the relationship between the molecular mechanisms underlying progressive RNF213 vasculopathy and neurodegenerative alpha‐synucleinopathies.

Author Contributions

D.Y. and T.‐B.A. contributed to the concept and design of the study. All authors contributed to the acquisition and analysis of data. D.Y., J.‐Y.S., and R.L. contributed to drafting the text or preparing the figures. All authors approved the final version of the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Definition of ischemic or hemorrhagic stroke.

Table S2: Operational definition of covariates.^1,2

Table S3: Summary of clinical features in two cases of Parkinson's disease diagnosed after Moyamoya disease.

Table S4: Epidemiological information of Parkinson's disease and control used for pathological study.

Table S5: Risk of Parkinson's disease in Moyamoya disease after excluding other neurodegenerative diseases.¹

Figure S1: Flow diagram of study participants. MMD, Moyamoya disease; PD, Parkinson's disease; SD, standard deviation.

Figure S2: Two cases of a prolonged history of Moyamoya disease with RNF213 (p.Arg4810Lys) variant and comorbid Parkinson's disease.

Figure S3: α‐Synuclein overexpression alone fails to induce aggregate‐like puncta in SH‐SY5Y cells, but RNF213 mutant enhances α‐synuclein interaction and promotes formation of ubiquitin‐positive aggregates.

Figure S4: Pathological analysis of postmortem brain sections from patients with PD.

Funding Statement

This work was funded by National Research Foundation of Korea (NRF) ; Korea Government (MSIT) grants RS‐2024‐00349023 and RS‐2024‐00361688; Bio & Medical Technology Development Program .

Data Availability Statement

The corresponding author takes responsibility for the integrity of the data and the accuracy of the data analysis. Supporting data of the findings in this study are available upon reasonable request from the corresponding author.