Parkinson’s Disease in Saudi Patients: A Genetic Study

Parkinson’s disease (PD) is one of the major causes of parkinsonism syndrome. Its characteristic motor symptoms are attributable to dopaminergic neurons loss in the midbrain. Genetic advances have highlighted underlying molecular mechanisms and provided clues to potential therapies. However, most of the studies focusing on the genetic component of PD have been performed on American, European and Asian populations, whereas Arab populations (excluding North African Arabs), particularly Saudis remain to be explored. Here we investigated the genetic causes of PD in Saudis by recruiting 98 PD-cases (sporadic and familial) and screening them for potential pathogenic mutations in PD-established genes; SNCA, PARKIN, PINK1, PARK7/DJ1, LRRK2 and other PD-associated genes using direct sequencing. To our surprise, the screening revealed only three pathogenic point mutations; two in PINK1 and one in PARKIN. In addition to mutational analysis, CNV and cDNA analysis was performed on a subset of patients. Exon/intron dosage alterations in PARKIN were detected and confirmed in 2 cases. Our study suggests that mutations in the ORF of the screened genes are not a common cause of PD in Saudi population; however, these findings by no means exclude the possibility that other genetic events such as gene expression/dosage alteration may be more common nor does it eliminate the possibility of the involvement of novel genes.


Introduction
Parkinson's disease (PD) is a movement disorder that was first described in 1817 [1], with a prevalence of approximately 1-2% at age 60 [2]. It is characterized by the occurrence of four cardinal symptoms: bradykinesia, resting tremor, rigidity and postural imbalance. These motor manifestations are attributable to dopamine deficiency in the striatum due to degeneration of the dopaminergic neurons within the substantia nigra pars compacta (SNpc).
The past two decades have witnessed rapidly emerging evidence for the key role of genes in the etiology of Parkinson's disease (PD), supplanting a long-held view about the non-genetic nature of the disease. Intensive research, following the discovery of α-Synuclein (SNCA [MIM 163890]), has hitherto identified more than 16 PD related loci [3]. Despite the recent advances, only 10% of the familial cases and less than 5% of the sporadic ones can be ascribed to monogenic mutations in either autosomal recessive ( [4][5][6]. However, the phenotypic commonalities in familial and sporadic PD has led researchers to believe that both forms of the disease may share some mutual pathways. Moreover, PD is expected to impose a major socioeconomic burden on aging populations. One way to relieve this burden is by gaining clearer understanding of the genetic etiology of the disease that may aid in designing effective diagnostic and therapeutic strategies. In this study we sought to determine the genetic causes of PD in Saudi patients. Such studies are lacking with the exception of a single report of a missense mutation in PINK1 in an extended Saudi family with Early-onset PD [7].

Subjects and Methods Subjects
A total of 98 individuals with PD, of which 33 were familial [24 autosomal recessive (AR) and 9 autosomal dominant (AD)], 63 were sporadic and 2 cases with incomplete family history data, were enrolled in this study. This study was approved by the Institutional Review Board of King Faisal specialist hospital and Research Center (project RAC# 2110035). Approved written consent forms were obtained from all subjects prior to their enrollment. Neurological assessment of patients was performed by movement disorder specialists and diagnosis of PD was established according to the accepted criteria. Patients were grouped as familial (with at least one reportedly affected first-or second-degree relative) or sporadic (no family history of the disease), and as Juvenile onset (JO; age of onset (0-20) years), Early onset (EO;  years) and late onset (LO; >50 years). Demographic and clinical features of patients are summarized in Table 1. Detailed clinical features of selected familial and sporadic cases are described in S1 File.

Experimental procedures
Mutational analysis of PD genes. Peripheral blood specimens were collected from patients for genomic DNA isolation using standard protocols. The entire coding sequence, including intron/exon boundaries, for common PD-genes; SNCA, PARKIN, PINK1, PARK7/ DJ1, LRRK2 and other PD-associated genes including; UCHL1[MIM 191342], GIGYF2[MIM 612003], FBXO7 [MIM 605648], and VPS35 [MIM 601501] was investigated in patients by means of direct sequencing using ABI Prism Big Dye Terminator ready reaction cycle sequencing kit (Applied Biosystems, Foster City, CA, USA). All 98 DNA samples were sequenced for the common PD-genes (mentioned above), while 82 out of 98 were sequenced for both common and other PD-associated genes (see above). Primers and PCR conditions are available upon request. Novel non-synonymous sequence variants with pathogenic prediction were screened in 700 Saudi normal controls, whereas those with benign predicted effect were screened in around 100 ethnically matching healthy controls.
RT-PCR. Total RNA was extracted from lymphocytes using PAXgene Blood RNA Kit (PreAnalytiX GmbH, Switzerland), followed by cDNA synthesis using Reverse Transcription System (Promega, CA, USA). Direct PCR amplification of LRRK2, SCNA, PINK1, PARKIN and PARK7/DJ1 cDNA was performed using gene-specific primers and β-actin was used as an internal control. The resulting amplicons were evaluated by electrophoresis on 2% Agarose gel. For primers sequences, PCR products size and transcripts information see S2 Table. Representative bands were sequenced to confirm origin.
Detection of copy number alterations. Out of the 98 samples, 25 representative samples were screened for dosage alterations of both common and associated-PD genes using the Cyto Scan HD array (Affymetrix, Santa Clara, CA,USA) which contains 2.6 million markers for genome coverage. The data was analyzed using the Chromosome Analysis Suite version Cyto 3.0 using GRC 38/hg19 of the UCSC Genome Browser. A threshold of log2 ratios of more than 0.58 for CNV gains and less than -1 for CNV losses was used.

Results
In the present study we detected; a total of 118 different sequence variants including three pathogenic point mutations and exon/intron dosage alteration of PARKIN. Of the detected sequence variants, 9 were HGMD-listed non-synonymous, 91 were reported in dbSNP (S1 Table) and 18 were novel variants ( Table 2).

Reported variants
Nine HGMD-listed variants were detected in our patients, of which 5 were classified as disease-causing mutations and 4 as disease-associated polymorphisms. Two of the disease-causing  mutations were present in PARKIN; p.Q34R and p.T240M, two were present in PINK1; p. T313M and p.E476K and one was found in PARK7/DJ1; p.R98Q [7,[15][16][17][18][19][20][21]. As for the diseaseassociated polymorphisms: two were present in PARKIN; p.S167N and p.V380L, one was present in UCHL1; p.S18Y and another was found in LRRK2; p.N551K [22][23][24][25][26]. Genotypes and frequencies of all detected variants are described in S1 Table. All the allelic variants occurred in heterozygous state apart from two sporadic cases; one (PD-108) homozygous for p.T313M missense mutation located in PINK1 and the other homozygous for p.S18Y substitution in UCHL1. In addition, two PARKIN variants (p.V380L and p.T240M) were found to be shared in a heterozygous state in two affected siblings (FM 19) (S1A Fig). The heterozygous p.R98Q substitution in PARK7/DJ1 was detected in two affected siblings and their unaffected father (FM92) (S1C Fig) in addition to a single sporadic case. For detailed clinical features of the cases see S1 File.

Novel variants
Eighteen novel sequence variants were detected in our samples; 10 were non-coding (9 intronic and one in the 3'-UTR) and eight were exonic. Out of the eight exonic variants, one was a nonsense mutation (p.Q178X in PARKIN) four were non-synonymous (p.E195Q and p.T240A in PARKIN; p.G189R and p.G409R in PINK1) and three were synonymous substitutions (p. P322P and p.L413L in PINK1 and Q1251Q in GIGYF2). Genotypes and frequencies are described in Table 2. The novel p.G409R substitution in PINK1 was identified in a homozygous state in two affected siblings form an AR family (FM 49) (Fig 1A and 1B) (for detailed clinical features see S1 File) and was absent in more than 700 Saudi normal controls. While p.G189R substitution was identified in one sporadic case and two controls ( Table 2).
As for PARKIN novel variants, the nonsense mutation (p.Q178X) occurred in a homozygous state in a single sporadic case (PD-110) with EOPD (S6 Fig), while the two non-synonymous substitutions (p.E195Q and p.T240A) were found to be shared (in a heterozygous state) with another variant either in a different gene or within PARKIN itself. For instance, p.E195Q substitution was found to be shared with p.N551K polymorphism in LRRK2 in a single sporadic case (SP 7) (S1D Fig). Similarly, p.T240A novel variant was found to be shared with the previously reported PARKIN mutation (p.Q34R) in a single proband and his affected father (FM 6) (S1B Fig). While p.G189R was found in a heterozygous state in one sporadic case and two controls.

cDNA analysis
Gene expression evaluation was performed on a subset of patients' samples (n = 12) from which RNA was available using primers designed to amplify specific portions of LRRK2, SNCA, PINK1, PARKIN or PARK7/DJ1 transcripts. Amplicons of the correct predicted sizes corresponding to the selected genes transcripts were successfully generated in all tested samples with the exception of one sample (PD-25a) that produced a shorter transcript of PARKIN (~160bp) which appears to be missing a 137bp fragment equivalent to the size of exon 7 (S3C Fig) The presence of transcripts for the selected genes in the tested PD samples, suggest that mutation(s) in the regulatory elements or deep intronic regions of these genes are more likely to be absent.

CNV analysis
Interestingly, CNVs were observed only in PARKIN. Heterozygous and homozygous exonic loss and partial intronic loss was observed in 8 cases (S3 Table). Loss of exon 3 and parts of introns 2 and 3 was detected in two different families (FM 19 and FM 21). In  In silico structural analysis of PARKIN (p.E195Q) and PINK1 (p.G409R) novel variants Encouraged by the type of amino acid change (from non-polar amino acid to a positively charged one) (Fig 1B), the absence in~1400 chromosomes, the evolutionary conservation (S2A Fig) and the damaging prediction analysis of p.G409R substitution ( Table 2), we decided to take advantage of computational tools to explore its impact on PINK1 function and structure. We therefore opted to model only the kinase domain (aa 162-512) harboring Gly 409 residue ( Fig 1D) (see S1 File for detailed description). The In silico analysis revealed loss of 4 αhelices in PINK1 mut that could disturb the domain conformation or stability. Furthermore, the spatial distance between the P+1 binding motif (responsible for substrate binding) and the adjacent helix G, the two segments comprising the P+1 specificity pocket [28], was increased, possibly to accommodate the large side chain of Arginine (Fig 1E and 1F).
Despite the neutral prediction analysis of p.E195Q variant (Table 2), the substitution of a negatively charged amino acid with a polar uncharged one (Fig 1C), the conservation of the native amino acid throughout mammals (S2B Fig), its absence in 192 control chromosomes, and its position within the Unique Parkin domain (UPD), a Zn +2 -binding domain important for substrate binding and ubiquitination [29,30], prompted us to investigate the structural and functional consequences of this substitution. Our PARKIN structural models were based on a previous PARKIN model, spanning amino acid residues (141-465), deposited in the RCSB (http://www.rcsb.org/pdb/home/) database under the accession number (PDB-ID# 4K95) (see S1 File for detailed description). Although p.E195Q substitution caused very subtle changes in protein folding, a number of secondary structure changes have been observed (Fig 1G and 1H). These include alterations in the number and/or length of secondary structural elements compared to the predicted PARKIN WT model. Firstly, the loss of one β-strand and one α-helix ( Fig 1G). Secondly, two β-strands (within the UPD domain; aa 142-227 [30]) were shortened, one by 8 amino acids and the other by 4 amino acids (Fig 1H).

Discussion
In the current study we set out to investigate the genetic basis of PD in Saudi patients. We decided to use a more general categorization of either familial or sporadic based on the presence or absence of positive family history and subsequently screen all patients for mutations in both PD-autosomal and PD-recessive genes.
Interestingly, our sequence analysis of well-established PD-autosomal recessive (PARKIN, PINK1 and PARK7/DJ1) and PD-autosomal dominant (SCNA and LRRK2) genes in families with the corresponding mode of inheritance as well as sporadic cases, detected only three pathogenic point mutations; two of which were missense [p.G409R in PINK1 (Fig 1A and 1B) and p.T313M in PARKIN (S1 Table)], while the third was a nonsense mutation [p.Q178X in PAR-KIN, (S6 Fig)]. The lack of LRRK2 mutations, a common cause of PD in North African Arabs [31,32] and Ashkenazi Jews [33], in our AD and sporadic PD-cases suggests that PD is genetically more heterogeneous in Saudis compared to other Middle Eastern populations.
The p.G409R variant was predicted to be pathogenic by four softwares (Table 2). Our in silico protein modeling predicted that PINK1 mut lacked secondary structure elements (4 α-helices) and the substitution of Gly 409 with Arg increased the spatial distance between P+1 binding motif and the adjacent helix G (Fig 1E and 1F). Residues in P+1 binding motif and helix G are involved in forming a P+1 specificity pocket necessary for kinase-substrate interaction [28,34]. Potential implications of such changes in the secondary structures and the P+1 specificity pocket of PINK1 may include; compromised structural integrity of the domain and altered substrate recognition specificity, which may interfere with PINK1 normal kinase activity. This is in line with previous studies demonstrating the adverse effect of substitution with Val at the same residue on PINK1 kinase activity and substrate recognition [27,28,35,36]. However, functional and cellular studies are required to confirm the predicted consequences. The other pathogenic PINK1 mutation detected in this study, is the p.T313M substition previsiouly described in a Saudi and a Chinese kindred with early onset PD [7,16] and was shown to cause neuronal toxicity and abnormal mitochondrial accumulation [37]. This mutation was present in a homozygous state in one patient (PD-108) with no consanguinty or positive family history reported, however, DNA from parents was not available for carrier status assessment.
As for PARKIN novel variants, p.Q178X truncating mutation in exon 4 had the ability to bypass nonsense-mediated mRNA decay as demonestrated by the presense of PARKIN transcript (S3C Fig) and therefore, may give rise to a defective protein product missing 287 amino acid residues. The second novel PARKIN variant is p.E195Q. In silico modeling suggests that this substitution has a subtle effect on protein confirmation (Fig 1G). However, alterations in the number and/or length of secondary structures were observed (Fig 1G and 1H). The analysis revealed that PARKIN mutt had lost one β-strand and one α-helix ( Fig 1G). Moreover, a portion of the two central anti-parallel β-strands of the UPD Zn-binding fold, transformed to random coil structure (Fig 1H). Disruption of Zn 2+ coordination is one possible outcome of such a structural transition especially since the altered strands contain at least one proposed Zn 2+ coordinating residue (Cys 212 ). Proper Zn 2+ ions coordination is perquisite for the maintenance of PARKIN 3-D structure, this is supported by studies on Zn 2+ -binding domains showing that EDTA-induced-Zn 2+ removal causes protein unfolding and therefore, would be expected to interfere with its normal function [29,38,39].
The other novel PARKIN variant (p.T240A) was detected in a single familial case in a heterozygous state (FM 6, proband and affected father) (S1B Fig). This mutation was absent in 192 control chromosomes, had neutral prediction analysis ( Table 2) and was modestly conserved (S2D Fig). However, mutations at the same residue (p.T240R/M) have been speculated to alter a possible phosphorylation site for casein kinase II (CK-II) [19,40], or to disrupt PAR-KIN association with Ubiquitin-conjugating enzyme (E2) necessary for ubiquitin-dependent proteasomal degradation [41][42][43].
The previously reported missense mutations identified in this study, excluding p.T313M in PINK1, (S1 Table), are less likely to be disease-causing in our patients due to one or a combinations of the following; presence in normal controls, neutral prediction analysis, reported lack of co-segregation in familial cases or lack of/equivocal evidence for functional impact [20,[44][45][46][47][48]. Although a heterozygous variant occurring in autosomal recessive gene is unlikely to be sufficient to cause the disease by itself, it may, however, confer risk in conjunction with other mutations. In line with this, (FM 6) harboring p.Q34R mutation in PARKIN was also found to be heterozygous for p.T240A novel missense change in the same gene (S1B Fig), however, whether these variants co-segregate with the disease or not, could not be assessed as DNA samples from unaffected family members were not available.
Moreover, two affected siblings (FM 19) (S1A Fig) were heterozygous carriers for two variants in PARKIN; p.T240M and p.V380L, a polymorphysim widely reported in various ethnic groups [25,26]. Even though p.T240M was reported as a disease-causing mutation [19], its pathogenicity remains unconfirmed, since it has been predicted as neutral by two out of four programs (S1 Table) and has been detected in one control. There are conflicting reports with regard to the impact of PARKIN (p.V380L) polymorphism on PD risk, however, a recent metaanalytic study demonstrated association of this polymorphism with moderate protection against the disease [26]. Meanwhile, whether this variant exerts the same effect in Saudi population or not, is yet to be known.
Another HGMD-listed mutation reported here, is the p.R98Q mutation of PARK7/DJ1 observed in a heterozygous state in one sporadic case and two affected siblings and their unaffected father (FM 92) displaying an AR from of PD (S1C Fig). Even though it has been classified in HGMD as disease-causing, this mutation is likely to be a polymorphism as suggested by the benign prediction analysis (S1 Table) and its reported presence at a similar frequency in European PD patients and ethnically matching healthy controls [46]. Also this mutation didn't alter protein stability when expressed in mammalian cells [45,49]. In contrast, this mutation has been shown to alter PARK7/DJ1 interaction with its binding partners and to reduce its antioxidant activity [50][51][52]. Therefore, more comprehensive functional analysis is necessary to ascertain the impact of this variant.
The general thinking has been that PARKIN or PINK1-mediated PD occurs through autosomal recessive inheritance, whereby the presence of homozygous or compound heterozygous mutations is necessary to drive the disease. As most of PARKIN and PINK1 variants, detected in this study, were observed in the patients in heterozygous condition, their clinical significance would be hard to interpret especially when in vivo and in vitro functional studies are lacking. Intriguingly, although single heterozygous variants are considered insufficient per se to cause the disease, subclinical dopamine dysfunction has been shown using functional neuroimaging in asymptomatic individuals heterozygous for PARKIN/PINK1 single mutations [reviewed in [53]]. This observation points towards a possible role of PARKIN/PINK1 heterozygous mutations in the pathogenesis of PD. Beyond monogenic inheritance, the "dual hit" hypothesis, whereby a second hit (be it exposure to environmental toxins, additional mutations in other PD genes or other pathways relevant to the disease), may offer an explanation for some of the heterozygous cases [44,54].
The current study was initially designed to test only for variants in the coding sequence and intron-exon junctions, thus other genetic events such as exonic rearrangements, copy number alteration and mutations within intron or regulatory regions could be overlooked. To that end we preformed CNV analysis of PD-genes (selected in this study) on a subset of samples. CNV changes were observed in PD patients mainly in SCNA and PARKIN genes [55]. Our CNV analysis failed to detect any gene dosage alterations in PD related genes apart from PARKIN. A homozygous loss that affected exon 7 was detected in affected members in one family (S4D Fig  and S3 Table), and a homozygous deletion of exons 3 and 4 was confirmed in a sporadic case (S3 Table), deletions affecting these exons were previously reported [56,57]. A common loss that affected exon 3, the most frequently reported mutation in PARKIN [4], and parts of introns 2 and 3 was observed in 2 families. The two affected siblings from one family were heterozygous for this deletion in addition to the p.T240M point mutation in PARKIN (S1A and S4A Figs), in the second family the affected father was homozygous for this deletion while his unaffected son was heterozygous for this deletion (S4B Fig). We also, evaluated the mRNA expression of PD-common genes (S3 Fig) in 12 samples. Transcripts were present in all samples, except one case (PD-25a) harboring PARKIN exon 7 homozygous deletion (S3 Table) in which PARKIN transcript was shorter (S3C Fig). These results indicate that mutation(s) in the regulatory elements of the screened genes or in other molecules involved in their transcription process are unlikely to be present at least in the tested samples.
In general,~82% of reported mutations in PD patients are simple mutations and~18% are CNV changes [53]. Our comprehensive analysis of PD causative and related genes identified only three point mutation and 2 CNVs. Thus, it seems reasonable to conclude that mutations in the ORF of the screened genes are not a common cause of PD in Saudi population. However, although our findings do not rule out the possibility of the involvement of the screened genes in the development of the disease, as gene expression/dosage may be perturbed, it is tempting to speculate the involvement of still unidentified genes.