Isolated nigral degeneration without pathological protein aggregation in autopsied brains with LRRK2 p.R1441H homozygous and heterozygous mutations

Leucine-rich repeat kinase 2 (LRRK2) is the most common causative gene for autosomal dominant Parkinson’s disease (PD) and is also known to be a susceptibility gene for sporadic PD. Although clinical symptoms with LRRK2 mutations are similar to those in sporadic PD, their pathologies are heterogeneous and include nigral degeneration with abnormal inclusions containing alpha-synuclein, tau, TAR DNA-binding protein 43, and ubiquitin, or pure nigral degeneration with no protein aggregation pathologies. We discovered two families harboring heterozygous and homozygous c.4332 G > A; p.R1441H in LRRK2 with consanguinity, sharing a common founder. They lived in the city of Makurazaki, located in a rural area of the southern region, the Kagoshima prefecture, in Kyushu, Japan. All patients presented late-onset parkinsonism without apparent cognitive decline and demonstrated a good response to levodopa. We obtained three autopsied cases that all presented with isolated nigral degeneration with no alpha-synuclein or other protein inclusions. This is the first report of neuropathological findings in patients with LRRK2 p.R1441H mutations that includes both homozygous and heterozygous mutations. Our findings in this study suggest that isolated nigral degeneration is the primary pathology in patients with LRRK2 p.R1441H mutations, and that protein aggregation of alpha-synuclein or tau might be secondary changes. Electronic supplementary material The online version of this article (10.1186/s40478-018-0617-y) contains supplementary material, which is available to authorized users.


Introduction
Parkinson's disease (PD) is the most common neurodegenerative movement disorder. Pathologies include neuronal loss and gliosis in the substantia nigra pars compacta (SNpc), locus coeruleus (LC), and dorsal motor nucleus of the vagus nerve, as well as the appearance of Lewy pathologies [9]. Lewy pathologies are the pathological hallmark of PD, and their major component is alpha-synuclein, encoded by synuclein alpha (SNCA) [43]. Many genetic factors for PD (PARK from 1 to 23) have been detected in the past two decades [7]. LRRK2 was originally mapped as a candidate region by linkage analysis from a large family living in the Sagamihara region of the Kanagawa prefecture in Japan [13]. Following its discovery, two groups concurrently reported the mutations p.R1441G, p.Y1699C, and p.R1441C in LRRK2 from patients in England, Spain, Germany, and the United States [33,54]. LRRK2 is located on 12q12 (MIM#609007) and includes 51 exons, and encodes a large protein (2527 amino acids) that belongs to the ROCO protein family and includes seven domains: armadillo, ankyrin, leucine-rich repeat (LRR), Ras in complex proteins (Roc), C-terminal of Roc (COR), kinase, and WD40 [26]. Currently, seven missense mutations (p.N1437H, p.R1441C/G/H, p.Y1699C, p.G2019S, and p.I2020T) are considered pathogenic variants. LRRK2 p.G2019S is the most common mutation, responsible for 36% of familial PD in North African Arab-Berbers, and 28% of familial PD in Ashkenazi Jews [21]. LRRK2 p.G2385R is also known as a risk variant for the onset of PD in Asian countries such as Japan, Taiwan, and Singapore [8,14,46].
To date, heterogeneous brain pathology has been reported in 55 autopsies of patients harboring LRRK2 mutations [40]. Even in the same mutation, for instance LRRK2 p.G2019S, pathologies have been reported with and without alpha-synuclein-positive Lewy bodies or Lewy neurites, Alzheimer's disease (AD) pathology, or tau-immunopositive glial tangle pathologies [17,36,38]. Neuropathology in LRRK2 p.I2020T mutations from the original Sagamihara family also revealed heterogeneous pathologies, including pure degeneration of the SNpc without any inclusions, Lewy bodies, or multiple system atrophy (MSA) pathology [19]. It is still unclear why LRRK2 mutations have such varied pathological and clinical manifestations.
In the present study, we detected two families (families A and B) that harbored c.4332 G > A, p.R1441H mutations in LRRK2, including ten PD patients presenting slowly progressing, late-onset parkinsonism. The two families had consanguinity. Our genetic analysis of seven PD patients indicated five homozygote and two heterozygote mutations of p.R1441H. Of these, we conducted brain autopsies of three patients: two homozygotes and one heterozygote. Our findings provide a new perspective of brain pathologies in patients with LRRK2 mutations.

Subjects
This study was approved by the ethics committee of the Juntendo University School of Medicine, in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki). All participants for genetic and clinical analyses gave full written informed consent before participation. The inheritance mode was defined as autosomal dominant when the family members of at least two consecutive generations were affected, and as autosomal recessive when one same-generation sibling was affected. The diagnosis of PD was established using clinical criteria [18]. A good responder to levodopa was defined as a patient whose symptoms improved with levodopa treatment. The two families, A and B, were from Makurazaki city in the Kagoshima prefecture, located in the southwest of Japan, with a population of approximately 20,000 in 2018 (Fig. 1a). The parents of each family were first cousins and born in Makurazaki. Our definition of inheritance confirmed family A as autosomal recessive inheritance, and family B as autosomal dominant inheritance (Fig. 1b).

Genetic analyses
We collected genomic DNA using QIAamp DNA Blood Midi Kit (QIAGEN, Hilden, Germany) from eight individuals in family A, which included four patients with PD, one patient with schizophrenia without parkinsonism, and three healthy siblings. From family B we collected genomic DNA from four individuals, which included three patients with PD and one healthy sibling. We selected four patients with PD (A-II-2, A-II-6, B-III-2, and B-III-6) for whole genome sequencing (WGS). WGS was performed using TruSeq DNA PCR-Free Library Prep Kit (Illumina, San Diego, CA, USA) and paired-end sequencing (150 bp × 2) on a HiSeq X Ten (Illumina). Sequence reads from WGS were trimmed by Trimmomatic (version 0.36) [2] and aligned to the GRCh37 human reference genome using BWA (version 0.7.17) [27]. Duplicated reads were removed by Picard (https://broadinstitute.github.io/picard/). Variants calling was performed using GATK (version 4.0.1.1) [31], and variants were annotated using ANNOVAR (version 2017Jul17) [49]. LRRK2 exon 31 was sequenced using the Sanger method, previously reported by Zimprich et al. [54]. Haplotypes were constructed using genetic markers including four SNPs and eight microsatellites mapped onto the flanking region of LRRK2. These genetic markers were genotyped by Sanger method or fragment analysis using fluorescence-labeled primers, as reported previously [4].

Genetic analyses
We identified 13 consensus variants by WGS analysis (Table 1 and Additional file 1). Among them, 11 of the 13 variants were insertion/deletion, which might be misaligned false positive variant calls. The remaining two variants were MUC5B (c.7843G > A:p.G2615S) and LRRK2 (c.4322G > A:p.R1441H). Mucin 5B, oligomeric mucus/gelforming (MUC5B) has been reported as a susceptibility gene for pulmonary fibrosis [4]. Therefore, the results of WGS indicated that LRRK2 (c.4322G > A: p.R1441H) is a causative mutation for families A and B. Sanger sequencing validation identified eight symptomatic patients (A-II-2, A-II-3, A-II-5, A-II-6, A-II-7, B-III-2, B-III-6, and B-III-8) with p.R1441H, and four asymptomatic individuals (A-II-1, A-II-9, A-III-1, B-III-3) without p.R1441H, which signals complete segregation of p.R1441H in families A and B (Fig. 1b). In addition, Sanger sequencing revealed a homozygous mutation in five patients (A-II-3, A-II-5, A-II-6, B-III-6, and B-III-8) and a heterozygous mutation in three patients (A-II-2, A-II-7, and B-III-2; Additional file 1: Figure S1). There were no pathogenic mutations, as well as risk variants and haplotypes, including SNCA, PAKR16, BST1, and MAPT, related to familial PD except LRRK2 p.R1441H in our WGS reads. Haplotype analysis indicated that patients from families A and B shared a common haplotype in the region of between D12S2080 and D12S2522, which signals a founder effect (Additional file 1: Table S2).

Case presentations
The parents of family A (A-I-1 and A-I-2) and family B (B-II-1 and B-II-2) married with consanguinity. All patients indicated as black symbols in the family trees were clinically diagnosed with typical PD (Fig. 1b). A-II-7 was diagnosed as schizophrenia without parkinsonism. Age at onset of PD was 61.25 ± 9.19 (± SD) in all patients, 61.60 ± 7.23 in patients with homozygous mutations, and 68.50 ± 6.54 in patients with heterozygous mutations. There were no differences in age at onset between homozygous and heterozygous mutations (p > 0.05). The clinical history of the three autopsied cases is described in the following sections.

A-II-3
This patient had a homozygous mutation of p.R1441H in LRRK2. At 68 years of age, he noticed left-side dominant tremor in his upper and lower limbs. The attending doctor observed his rigidity, tremor, and masked face and diagnosed him with PD. Following diagnosis, his motor symptoms showed a good response to levodopa carbidopa 200 mg and cabergoline 2 mg per day. In the following few years, he remained in good condition and with good activity of daily living, without support from

Pathological analysis
The neuropathological findings of three patients (individual A-II-3, A-II-6, and B-III-2) are presented. The brain weights of A-II-3, A-II-6, and B-III-2 were 1350 g, 1350 g, and 1126 g, respectively.
Macroscopically, the substantia nigra (SN) was markedly depigmented in our cases, compared with age-matched controls (Fig. 2 a-c), but the LC was preserved. Atrophy and morphological changes were not observed in other brain regions. In H&E, KB, GFAP, and IBA1 staining, neuronal cell loss and astrogliosis were apparent in the SN (Fig. 2 d-i). Almost all microglia exhibited a ramified shape (resting form) in the SN of all three cases; small numbers of ameboid-shaped cells (reactive form) appeared only in B-III-2 ( Fig. 2 j-l). These findings were considered common pathologies of chronic nigral degeneration. In TH stains, loss of dopaminergic neurons was apparent in the SNpc, compared with age matched-controls (Fig. 3 a-d). In contrast, there was no neuronal cell loss or gliosis in the LC in any case. The remains of neurons showed no Lewy bodies in H&E sections. In anti-phosphorylated alpha-synuclein (p-alpha-synuclein) and ubiquitin immunostains, there were no Lewy pathologies (bodies, neurites, or dots) in the SN, compared with sporadic PD (Fig. 3 e-h). Moreover, p-alpha-synuclein-positive pathologies were not detected at all in the brainstem, limbic area, subcortical nuclei, white matter, or neocortex in any individual. Phosphorylated-tau (p-tau)-positive neurofibrillary tangles (NFT) and threads were limited to within the parahippocampus and hippocampus (Braak NFT stage II) [3]. Amyloid-beta-positive senile plaques were detected within the range of normal aging in the neocortex. These tau pathologies were considered primary age-related tauopathy (PART) [6]. There were no TDP43-positive pathologies or LRRK2-positive pathologies. These pathological findings were common to all three patients. In addition to these common findings, a small number of p-tau-positive structures were observed in the brainstem and cerebellar dentate nucleus of individual B-III-2, while the peripheral nerve (sural nerve) was also evaluated in A-II-6 and demonstrated preserved axons and myelin. In summary, the major pathological feature of all three individuals was isolated nigral degeneration without Lewy or other pathological inclusions.

Discussion
In the present study, we investigated two families that harbored homozygous and heterozygous mutations of p.R1441H in LRRK2. All affected members had a founder effect, suggesting a common haplotype in two different families. The two families came from a narrow district in Makurazaki city, with consanguinity. The environment incidentally generated both heterozygous and homozygous mutations of p.R1441H in two families. All PD patients presented late-onset parkinsonism with a good response to levodopa and a mild disease course, without apparent cognitive decline or dysautonomia. The clinical manifestations were similar to those in patients with sporadic PD or PD with other LRRK2 mutations. Most PD patients with LRRK2 mutations are heterozygous missense mutations with an autosomal dominant inheritance mode, which we infer as a gain-of-toxic-function pathological mechanism of LRRK2 mutations. In the families in our study, there were no differences in clinical symptoms, age at onset, or brain pathologies between patients with heterozygous and homozygous mutations. Previous reports also described no apparent differences in symptoms between homozygous and heterozygous mutations in LRRK2 p.G2019S [1,24,25]. The patients with LRRK2 p.G2019S homozygous mutations had middle-aged onset, tremor-dominant and mixed type of tremor, and rigid akinesia, similar to heterozygous patients [24]. It has also been reported that homozygous p.R1441C knock-in mice present a normal phenotype, with no dopaminergic neurodegeneration [48]. It remains unclear why homozygote and heterozygote patients with LRRK2 mutations have similar clinical manifestations.
Arginine (R) is located at position 1441within the Roc domain of the LRRK2 protein. The missense mutation in R1441 induces different amino acid changes of glycine (G), cysteine (C), or histidine (H) [30,33,54]. LRRK2 mutations of p.R1441C/G/H induce late-onset parkinsonism with good response to levodopa, and closely resemble sporadic PD or PD with LRRK2 p.G2019S mutations [20,30,34]. The p.R1441H mutations have been identified in four families from Asia, Europe, and North America; it is present in diverse ethnicities [12,30,42,53]. One rare case with LRRK2 p.R1441H has been related to progressive supranuclear palsy [42]. However, our three autopsy cases showed homogeneity in their symptoms; they had preserved cognitive function until the end of their lives, with the appearance of persistent psychosis in the advanced stage, and without dysautonomia. These findings may arise from their consanguinity, a founder effect, and their descent from a small city population.
A summary of 55 autopsied cases with LRRK2 mutations (p.G2019S, p.I2020T, p.R1441C, and others) has been previously reported [40]. Most pathologies involved SNpc degeneration associated with Lewy pathologies. However, other patients presented broad types of pathology, including pure nigral degeneration without any inclusions, MSA-like-synucleinopathy, PSP-like tauopathy, frontotemporal lobar degeneration with ubiquitin-positive-inclusions, and coexistence with TDP-43-positive inclusions. The pathological heterogeneity suggests that LRRK2 mutations may cause a cascade of effects to induce neuronal deterioration.
To our knowledge, this study is the first report of neuropathology in LRRK2 p.R1441H mutations. In addition, there have been no pathological reports of any homozygous mutations in LRRK2. Our pathological presentation of both homozygous and heterozygous mutations in LRRK2 is crucial for considering LRRK2-associated pathogenesis. Isolated nigral (SNpc) degeneration without alpha-synuclein, tau, amyloid-beta, ubiquitin, and TDP43 pathology was a prominent feature in all three of our autopsy cases. Previous studies have reported pure nigral degeneration, without inclusions of disease-specific proteins, in p.G2019S, p.R1441G, and p.I2020T mutations [15,17,19,29]. In addition, all patients showed typical sporadic PD-like parkinsonism, similar to our cases. Considering the clinicopathological relationship, their parkinsonism might be caused by neurodegeneration of the SNpc, irrespective of expression levels of alpha-synuclein pathologies. Nigral degeneration not associated with Lewy pathologies was also observed in the brains of most parkin RBR E3 ubiquitin protein ligase (PRKN) mutations and one PTEN induced putative kinase 1 (PINK1) mutation previously [10,45]. Parkin and PINK1 collaborate in mitophagy to remove damaged mitochondria and maintain mitochondrial quality [44]. There is convincing evidence linking LRRK2 with autophagy/mitophagy [5,23,32,37,39,47], with regulating roles for LRRK2 involving mTOR signaling [11,22,28,39,52] as well as Parkin/PINK1 mitophagy. In contrast, LRRK2, a large multi-domain protein, plays an important role in phosphorylation of itself and of target substrates in cellular signaling. Some mutations in the kinase domain, especially p.G2019S, are considered to augment kinase activity, and excess kinase activity is known to induce neurotoxicity [41,50]. Alternatively, mutations in the Roc-COR domain can cause a decrease in GTPase activity and are thought to play an important role in neurodegeneration [51]. In addition, in relation to alpha-synuclein and tau pathologies, several studies have shown that LRRK2 affects these protein aggregations, but alpha-synuclein-or tau-aggregation-related pathogenesis in brains with LRRK2 mutations remains poorly understood [26]. Our three cases, and twelve previously reported cases (p.R1441H and p.R1441G in the Roc domain, p.G2019S and p.I2020T in the kinase domain), all showed SNpc degeneration without Lewy or other pathological inclusions, although these enzymes act at a steady rate and the biological aberrations induced by these mutations are quite different. Recently, a Severe losses of dopaminergic neurons were present in the substantia nigra of cases with the mutation, compared with age-matched controls, using a tyrosine hydroxylase (TH) immunostain (a: age-matched control, b: A-II-3, c: A-II-6, d: B-III-2). However, sporadic PD with Lewy pathology corresponding to Braak stage 5 showed typical alpha-synuclein-positive neuronal inclusions in the substantia nigra; there was no Lewy pathology in the substantia nigra or other brain regions of cases with the mutation, as assessed with immunohistochemical analysis for alpha-synuclein (e: age-matched control, f: A-II-3, g: A-II-6, h: B-III-2). The scale bars represent a-d: 1 mm, and e-h: 200 μm, respectively linkage between Ras analogue in brain (Rab) GTPases and alpha-synuclein, LRRK2, and vacuolar sorting protein (VPS) 35 associations with PD pathogenesis was highlighted [16]. One study reported that p.R1441C in the GTPase domain enhances GTP binding and stimulates LRRK2 activity through interaction with Rab29 and the Golgi apparatus [35]. These data suggest that p.R1441H in the GTPase domain can also cause LRRK2-associated neurotoxicity induced by Rab29-mediated Golgi recruitment.

Conclusions
It was noteworthy that our three autopsied series showed homogeneous pathology. Regardless of the homozygous or heterozygous mutations of p.R1441H, the pathological findings and the clinical features were similar, suggesting that any possibility of a gene dosage effect could be excluded. Two consanguineous families with LRRK2 p.R1441H had a founder effect; the similarity might be caused by their homogeneous genetic background. Our pathological findings indicated that isolated nigral degeneration is essential pathology for LRRK2 p.R1441H mutations. When considering previous reports of neuropathologies from LRRK2 mutations, SNpc degeneration was a constant finding in all cases. Other findings of protein pathologies, such as alpha-synuclein, tau, ubiquitin, and TDP43, were inconsistent. Thus, we consider that neuronal degeneration in the SNpc is primary degeneration, caused by neurotoxicity of pathogenic LRRK2 mutants. We speculate that abnormal protein aggregation, such as of alpha-synuclein, tau, ubiquitin, or TDP43, would be downstream in the pathological cascade of LRRK2. Our three autopsy cases strongly support this theory, showing isolated nigral degeneration with LRRK2 mutation. In addition, it is important to investigate the possibility of proteopathic aggregations that have not been identified previously in LRRK2 mutated pathologies, in future pathological and biochemical studies.

Additional file
Additional file 1: Table S1. Consensus nonsynonymous variants detected by whole genome sequencing. A portion of the output from ANNOVAR is shown. Most of the variants are indel and might be misaligned variant calls. NA: not applicable. Figure S1. Sequence electropherogram of LRRK2 c.4322G > A:p.R1441H. Three representative sequences are shown. Healthy sibling A-II-1 has wild type (W/W), patient B-III-2 has heterozygous (W/M), and patient A-II-3 has homozygous (M/M) c.4322G > A variant (arrowhead) in LRRK2 exon 31.