ORIGINAL RESEARCH published: 28 April 2022 doi: 10.3389/fragi.2022.851039 Somatic Mutations Detected in Parkinson Disease Could Affect Genes With a Role in Synaptic and Neuronal Processes Edited by: Irene Lobon 1*, Manuel Solís-Moruno 2,1, David Juan 1, Ashraf Muhaisen 3,4, Federico Abascal 5, Victoria Belancio, Paula Esteller-Cucala 1, Raquel García-Pérez 1, Maria Josep Martí 4,6, Eduardo Tolosa 4,6, Tulane University, United States Jesús Ávila 4,7, Raheleh Rahbari 5, Tomas Marques-Bonet 1,8,9,10, Ferran Casals 2,11† Reviewed by: Eduardo Soriano 3,4*† Prescott Deininger, 1 Tulane University, United States Institute of Evolutionary Biology (UPF-CSIC), Barcelona, Spain, 2Genomics Core Facility, Departament de Ciències Valentina S. Caputo, Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain, 3Department of Cell Biology, Physiology and London South Bank University, Immunology and Institute of Neurosciences, Universitat de Barcelona (UB), Barcelona, Spain, 4Centre for Networked Biomedical United Kingdom Research on Neurodegenerative Diseases (CIBERNED), Madrid, Spain, 5Cancer, Ageing, and Somatic Mutation (CASM), Wellcome Sanger Institute, Cambridge, United Kingdom, 6Department of Neurology, Hospital Clínic de Barcelona, Institut *Correspondence: d’Investigacions Biomédiques August Pi i Sunyer (IDIBAPS), University of Barcelona (UB), Barcelona, Spain, 7Centro de Biología Irene Lobon Molecular Severo Ochoa, Madrid, Spain, 8Catalan Institution of Research and Advanced Studies (ICREA), Barcelona, Spain, irene.lobon@crick.ac.uk 9 CNAG-CRG, Centre for Genomic Regulation (CRG), Barcelona Institute of Science and Technology (BIST), Barcelona, Spain, Eduardo Soriano 10 Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona, Barcelona, Spain, 11Departament de esoriano@ub.edu Genètica, Microbiologia i Estadística, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain † These authors have contributed equally to this work The role of somatic mutations in complex diseases, including neurodevelopmental and Specialty section: neurodegenerative disorders, is becoming increasingly clear. However, to date, no study This article was submitted to has shown their relation to Parkinson disease’s phenotype. To explore the relevance of Genetics, Genomics and Epigenomics of Aging, embryonic somatic mutations in sporadic Parkinson disease, we performed whole-exome a section of the journal sequencing in blood and four brain regions of ten patients. We identified 59 candidate Frontiers in Aging somatic single nucleotide variants (sSNVs) through sensitive calling and a careful filtering Received: 08 January 2022 Accepted: 16 March 2022 strategy (COSMOS). We validated 27 of them with amplicon-based ultra-deep Published: 28 April 2022 sequencing, with a 70% validation rate for the highest-confidence variants. The Citation: identified sSNVs are in genes with synaptic functions that are co-expressed with Lobon I, Solís-Moruno M, Juan D, genes previously associated with Parkinson disease. Most of the sSNVs were only Muhaisen A, Abascal F, Esteller-Cucala P, García-Pérez R, called in blood but were also found in the brain tissues with ultra-deep amplicon Martí MJ, Tolosa E, Ávila J, Rahbari R, sequencing, demonstrating the strength of multi-tissue sampling designs. Marques-Bonet T, Casals F and Soriano E (2022) Somatic Mutations Keywords: Parkinson disease, somatic mutations, brain mosaicism, neurodegenaration, somatic genome alteration Detected in Parkinson Disease Could Affect Genes With a Role in Synaptic and Neuronal Processes. Abbreviations: CNV, Copy number variant; COSMOS, Combined Or Single sample MOSaicism detection (method); PD, Front. Aging 3:851039. Parkinson disease; PIR, Position-in-reads bias score; PON, ‘Panel of normals’; SBS, Single base signatures; sSNV, Somatic single doi: 10.3389/fragi.2022.851039 nucleotide variant; VAF, Variant allele frequency. Frontiers in Aging | www.frontiersin.org 1 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease INTRODUCTION 2014) and that over a fourth carried somatic mutations affecting pathways known to contribute to tau hyperphosphorylation Somatic mutations appear during development and tissue (Park et al., 2019), demonstrating the power of exome analysis. maintenance, making every individual a mosaic of cells with The link between Parkinson disease (PD) and somatic slightly different genomes. Early mutations occurring before mutations is not as clear. Notably, only about 10% of cases gastrulation are shared by tissues of different germ layer origin can be attributed to monogenic forms (Lesage and Brice, (Lodato et al., 2015; Bae et al., 2018) and can cause disease (Kluge 2009), and at most 30% of patients have an affected first- et al., 2019; Mensa-Vilaró et al., 2019). Mutations occurring later in degree relative (Rocca et al., 2004). A study on 511 sporadic development or during adult tissue maintenance that confer a cases tested multiple brain regions with a sensitivity limit at 5% of proliferative advantage can produce clonal expansions of the cells variant allele frequency (VAF) and did not find any somatic carrying them, limited by each tissue’s growth dynamics (Lee-Six variants in SNCA, the main causative gene in early-onset familial et al., 2019; Moore et al., 2020). Notably, this type of somatic PD (Proukakis et al., 2014). On the other hand, patients showed mutations are the cause of cancer (Nowell, 1976) but have also high levels of heteroplasmic mitochondrial DNA deletions been found in healthy skin (Martincorena et al., 2015; Abyzov et al., (Bender et al., 2006) and more SNCA somatic copy number 2017), esophagus (Martincorena et al., 2018), colon (Lee-Six et al., gains in substantia nigra neurons compared to controls, which 2019), liver (Brunner et al., 2019), endometrium (Moore et al., 2020) positively correlated with age of onset (Mokretar et al., 2018). or lung (Yoshida et al., 2020). Further, organ-exclusive somatic To explore the potential link between somatic SNVs and mutations in the MTOR pathway are involved not only in cancer Parkinson disease, we sequenced the exome of five different (Guertin and Sabatini, 2007) but also in neurodevelopmental tissues—substantia nigra, striatum, neocortex, cerebellum, and disorders such as hemimegaloencephaly (Poduri et al., 2012) and blood—from ten sporadic Parkinson patients to an average focal cortical dysplasia (Lim et al., 2015). Other studies linking coverage of 60X. We developed and implemented a sSNV somatic mutations to disease did not determine whether the variants detection protocol based on both single tissue and joint were of early or late origin, including a case of congenital arrhythmia information, COSMOS, which we used to identify 59 in which the causal mutation was only tested in cardiomyocytes and candidate sSNVs. Further amplicon-based deep sequencing lymphocytes and found in both cell types (Priest et al., 2016). confirmed 27 of them, with an average of 3 sSNVS per It has been recently described that adult neurons accumulate over individual (range 0–6). Although our power was limited, the 17 somatic single nucleotide variants (sSNVs) per year, in a process confirmed sSNVs were enriched in synaptic and axonal processes independent of cell division (Abascal et al., 2021). Even before birth, and patients with more sSNVs had a weak tendency to die earlier, each human neuron carries a few hundred sSNVs (Bae et al., 2018) suggesting a potential role of these variants in the disease. and those with a frequency higher than ~2% can also be detected in Interestingly, over 76% of the variants validated in multiple tissues that originate from a different germ layer (Lodato et al., 2015), brain tissues were only called in blood in the exome data, suggesting that brain development does not heavily rely on clonal demonstrating that the study of more accessible and expansions. Other types of somatic variation have also been found in unaffected tissues may well serve to identify variants with neurons: retrotransposon mobilization is common and lower frequency in diseased organs. disproportionately impacts protein-coding loci (Baillie et al., 2011) and over three copy number variants (CNVs), mainly losses, can be found in each cell (Cai et al., 2014). While the RESULTS number of CNVs decreases in brain with age (Chronister et al., 2019), sSNV load increases (Hoang et al., 2016; Lodato et al., 2018; Dataset Abascal et al., 2021), posing questions about their relevance in We sequenced the whole exome of five different tissues from ten neuronal diversification, plasticity, and dysfunction. sporadic Parkinson patients to an average coverage of 60X. Blood The somatic variant hypothesis for neurodegenerative diseases was obtained from stored vials while the cerebellum, neocortex, states that unexplained sporadic cases could be caused by somatic substantia nigra, and striatum samples were collected during mutations, presumably in the same genes affected in familial cases autopsies (Figure 1A). Patients’ median age was 81 at the time of (Pamphlett, 2004). Supporting this theory, age-related sSNVs death, they had varying ages at disease onset, and both sexes were have been proposed to accumulate faster in neurodegeneration represented (Supplementary Table S1). As expected, germline (Lodato et al., 2018), although more research is needed (Abascal single nucleotide variant (SNV) calling resulted in clustering of et al., 2021). Besides these later-acquired mutations, earlier tissue samples by individual except for two samples from individual somatic mutations may also contribute to phenotypes and DV2, which appear to be contaminated (Supplementary Figure S1). diseases. This is suggested by the fact that individuals with Accordingly, this individual was excluded from subsequent analyses autism spectrum disorder have a higher burden of somatic but included in the sample set used to estimate background noise. mutations than their unaffected siblings, measured in blood (Dou et al., 2017). In Alzheimer, targeted sequencing of blood samples showed that somatic variants in autosomal dominant Germline Variants are Associated With genes (such as APP) could explain ~2% of cases (Nicolas et al., Parkinson 2018). Other studies on blood and hippocampus exomes from All patients were diagnosed with sporadic Parkinson because Alzheimer patients showed numerous sSNVs (Parcerisas et al., no affected first-degree relatives were known. This assessment Frontiers in Aging | www.frontiersin.org 2 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease FIGURE 1 | (A) Study overview. We sequenced the whole exome of blood and four brain regions from ten sporadic Parkinson disease patients. In this cohort, we found 1 germline SNP associated with Parkinson in OMIM and 4 via GWAS. The deleterious germline SNVs were enriched in spine apparatus and kinesin binding genes. Candidate somatic variants were carefully annotated and filtered using COSMOS, resulting in 59 candidate sSNVs. We used deep amplicon sequencing to validate them in all samples. The 27 validated sSNVs are enriched in synaptic and neuronal functions. VAF: variant allele frequency, FP: false positives. (B). Germline SNVs related to Parkinson. A bar is shown for each germline variant, with its color indicating the type of variant: missense (red), splice acceptor (yellow), or intronic (blue). Filled circles indicate individuals carrying the variant (DV2 is not included) and darker colors indicate homozygous SNVs. The affected gene is indicated on top and the dbSNP identifier and the frequency of the alternative allele in the IBS population are shown below each bar. did not include genetic analyses, so we evaluated their 0.27, and similarly, it was 0.28 among the 10 Spanish germline variants before considering their somatic Alzheimer patients included in this study (Figure 1B). mutational landscape. We used three different strategies to We made use of the greater statistical power of previous prioritize germline SNVs. First, we identified germline SNVs in GWAS studies to identify Parkinson-associated germline genes linked to Parkinson in OMIM (www.omim.org) that variants (41 variants, Supplementary Table S3) (Tan et al., were deleterious as indicated by a CADD score > 15 (Rentzsch 2010; Do et al., 2011; International Parkinson Disease et al., 2019). Only one variant met these criteria, rs17651549, a Genomics Consortium et al., 2011; International Parkinson missense mutation in the MAPT gene, which encodes the tau Disease Genomics Consortium and Wellcome Trust Case protein, predicted as deleterious by multiple methods and at a Control Consortium 2, 2011; Lill et al., 2012; Nalls et al., 2014; highly conserved position in vertebrates (Supplementary Chen et al., 2016) and found four in our cohort (Figure 1B). We Table S2). This variant, which was heterozygous in over found a splice acceptor variant in TMEM175, rs34884217, in DV3 half the patients, has been previously linked to PD by and DV4. This variant, predicted to affect nonsense-mediated different means: multivariate family-based association tests decay, has been associated with Parkinson (Nalls et al., 2014; (Wang et al., 2010), pathway analysis (Song and Lee, 2013), Heckman et al., 2017) and TMEM175 deficiency is linked to the and targeted resequencing (Spataro et al., 2015). However, in a increase of α-synuclein aggregation (Jinn et al., 2017), suggesting contradictory haplotype association analysis, it provided a a possible causal link. A missense mutation of PARK2 showed the reduced risk for PD (Li et al., 2018). This SNP is not rare highest odds ratio for Parkinson disease in Europeans in a meta- in Europe; particularly, the frequency of the alternative allele analysis (Ramakrishnan et al., 2016). The other two SNVs have in the 1000GP IBS population (Iberian populations in Spain) is been associated with the disease (Huang et al., 2004; Elstner et al., Frontiers in Aging | www.frontiersin.org 3 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease 2009; Williams-Gray et al., 2009), but conflicting results have also used VarScan 2 with lax parameters (-min-coverage 1-min- been reported (Guella et al., 2010; Federoff et al., 2012). reads2 1 -p-value 1 -min-var-freq 0.000001). The resulting call In addition, all germline SNVs were filtered by deleteriousness set is mostly composed of germline heterozygous SNVs, calls (CADD > 15 and SIFT prediction) and their frequency in the within CNVs, recurrent sequencing errors, and other artifacts 1000GP European population (<0.1). The resulting 214 variants besides the somatic SNVs. Hence, we developed a filtering affected 207 genes, with each patient carrying a median of 27 strategy to identify high confidence somatic SNVs, COSMOS deleterious and rare variants (range 20–32). An (Combined Or Single sample MOSaicism detection, available in overrepresentation enrichment analysis (Zhang et al., 2005) https://github.com/ilobon/COSMOS). COSMOS can be used to detected seven molecular function and cellular component annotate the relevant information needed to identify reliable terms significantly enriched in this gene set (FDR ≤ 0.05, somatic candidates on standard VCF files and to filter calls. Supplementary Figure S2). Remarkably, the term with the Detection of reliable variants can be performed with a single highest enrichment ratio was spine apparatus (Figure 1B), a sample approach, a joint filtering approach (using multiple derivate of the smooth endoplasmic reticulum generally present samples from the same individual), or both. in dendritic spines that seems to participate in spine remodeling in Parkinson disease models (Smith et al., 2009). Kinesing COSMOS binding and motor activity were also among the significantly Manual inspection of the calls obtained from HaplotypeCaller enriched terms, with seven out of the nine individuals carrying a and VarScan 2 allowed the identification of multiple sources of deleterious mutation in the genes driving these associations. artefacts (some of which have been addressed by other studies These genes have been previously linked to PD in different (Chen et al., 2017; Kim et al., 2019; Park et al., 2019; Solís-Moruno studies. As an example, CLSTN1 overlapped a significantly et al., 2021; Wang et al., 2021)), for which we devised filtering hypomethylated CpG (Chuang et al., 2017), was differentially approaches. The rationale was that even if our patients had the expressed (Kong et al., 2018), and carried a missense mutation same disease and could carry mutations in the same pathways or (YemniAl et al., 2019) in PD cases. Several dynein and kinesin even genes, the probability that they would bear somatic proteins also appear to be relevant in PD (Pellegrini et al., 2016). mutations in the exact same position was negligible in just 10 All of this highlights the complexity of Parkinson disease, in individuals. Hence, variants present in multiple individuals were which common and rare variants affect multiple pathways that presumed to be artefacts. seem to contribute to the phenotype. Abundant data is needed to We discarded all off-target calls because they have higher uncover these associations and the evaluation of somatic strand imbalances (69.4% of off-target calls fail the Poisson test vs. mutations has the potential to contribute to this effort. 39.5% of on-target calls) and read pair imbalances (21.8 vs. 4.4% failing the read pair ratio filter). The most frequent confounding Somatic Variant Calling factors for on-target calls were germline heterozygous SNVs, Somatic variant calling was first developed for cancer, and CNV regions, unresolved regions of the genome and regions standard approaches are based on paired tumor and normal that are more difficult to align (indels, homopolymers). samples comparison (Krøigård et al., 2016; Xu, 2018). Growing Germline heterozygous calls comprised at least 93.1% of our evidence shows that low frequency embryonic somatic mutations on-target calls. They can be easily distinguished when their VAF are present in multiple tissues, even from different germ layers is close to or higher than 50%. However, germline calls’ VAFs are (Lodato et al., 2015; Bae et al., 2018), so paired comparisons are more over dispersed when the coverage is low (Supplementary inadequate in scenarios with no or limited clonal expansions. Figure S3) and VAFs as low as 18% can result from germline Requiring a minimum of two supporting reads in our 60X data variants sequenced at 60X (and even as low as VAF = 10% for would limit our sensitivity to a mean VAF over 3%. Variants with 20X, our lower bound depth). We used a binomial test that frequency could have originated before gastrulation, be (discarding 83.2% of on-target calls) and a VAF upper limit present in multiple tissues and require joint sample calling. (76.5% of on-target calls had VAF > 40%) to identify the most On the other hand, blood clonal expansion, especially obvious germline variants, with a large overlap between the two prevalent in old age (Jaiswal et al., 2014), will result in filters (Figure 2A). These variants are common to all tissues from variants private to this tissue, requiring single sample calling. the same individual, so when multiple samples are available, In standard paired samples calling, shared calls are essentially requiring that all pass the binomial test greatly increases the discarded, which filters out germline variants and recurrent detection power, following an exponential distribution artifacts. The main obstacle unpaired single or joint sample (Supplementary Figure S4). We discarded an additional 5.5% calling have is precisely differentiating these confounding of variants using this approach (first bar in Figure 2A). In factors from true somatic variants. addition, most of these calls can also be flagged by the Germline callers are optimized to discard variants that do not number of called individuals. We generated a population and fit with set VAF expectations. Low frequency variants, such as batch specific panel for each patient using the 45 samples from those we are interested in, are considered sequencing noise or the other 9 PD individuals, and found it identified 92.1% of the contamination, and discarded. For this reason, we called variants germline variants previously discarded. Alternatively, an external in all samples using HaplotypeCaller with the ploidy parameter panel of normals (PON) consisting of 428 whole genomes set to 10, which increases its sensitivity to lower frequency sequenced at the Sanger Institute identified 82.6% of them variants, as previously shown (Wang et al., 2021). We also (Figure 2A). Frontiers in Aging | www.frontiersin.org 4 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease FIGURE 2 | Relevance of COSMOS filters in our Parkinson exome data. (A). Germline SNVs. Intersection of the main criteria identifying germline heterozygous point mutations: VAF (high variant allele frequency), Binomial (non-significant binomial test for allele depths) and BinomialInd (non-significant binomial test in at least one sample from the individual). For each intersection, the number of variants also found in the other individuals of our dataset (PD panel, filled in lavender) and an external panel (PON, purple triangle) are shown. (B). CNV regions. Intersection of DepthRange (most extreme depth variants) and SegmentalDups (segmental duplications WGAC track) with PD panel and PON (light and dark green triangles, respectively) and variants present in the 1000GP strict mask filled in green. (C). Relationship between criteria. Log2 ratio between the number of variants failing both the row and the column criteria and those failing the row criterium only. Higher log2 ratios (red) denote higher co-occurrence of criteria failures. The annotation column (green gradient) indicates the total number of positions failing each of the row criteria. Regions with CNVs can create artefactual somatic variants when indels and homopolymers is challenging, and technical and reads are collapsed, mapped into the single copy in the reference biological noise are difficult to distinguish, so we discarded close- genome. CNV callers can be used to identify the highest confidence by variants to increase our call set confidence (see Methods). Finally, copy number alterations, but more sensitive approaches are needed to variants with too few reads supporting the alternative allele cannot be detect noisy regions and obtain high-confidence somatic variant calls. distinguished from random noise, and their imbalances cannot be We excluded regions with extremely low or high coverage (keeping evaluated, so a hard cut-off was applied (details in Methods). 75% of the distribution) or overlapping the WGAC segmental The recommended best practices for calling somatic duplications track, which filtered out 60.4% of the remaining on- mutations in unpaired samples or when variants are expected target non-germline calls. We found 97.5% of these calls overlapped in all samples consists of a non-stringent variant calling followed the 1000GP strict mask, which identifies regions of the genome with by several filters (Wang et al., 2021). Many of the features we recurrently higher/lower coverage or lower mapping quality, showing identified have been previously targeted by manual filters in its suitability as an alternative. Most calls failing our CNV filters were similar contexts (Chen et al., 2017; Kim et al., 2019; Park called in multiple individuals (94.1% in our PD panel and 84% in the et al., 2019; Solís-Moruno et al., 2021; Wang et al., 2021). To independent PON) (Figure 2B). Other features useful to identify facilitate their implementation, we developed COSMOS, a these artifacts are a high number of variants within read length pipeline that can be used to consistently annotate and filter distance, an imbalance in the number of mismatches in reads the described read features on VCF files (https://github.com/ carrying each allele or in the proportion of clipped reads, and a ilobon/COSMOS). Its main advantages are that it automates the low mappability (Supplementary Figure S5). process and that it provides the option to filter variants that pass Unresolved regions of the genome or highly variable duplications all required filters in (1) single samples; (2) a minimum number can result in collapsed mapping with no significant increment of of samples, which can be different sample sets for each filter; or coverage because only reads spanning the homologous region and (3) in either single of multiple samples. carrying few variants will be mapped. These are easily identified by a biased position of the alternative allele in the reads, clustered at read ends. We defined a position-in-reads bias score (PIR) to address this Most Somatic Mutations Appear to Have a issue (see Methods). Other features targeting these artifacts are allele Pre Gastrulation Origin clipping imbalances, a high number of variants in the vicinity, more A total of 59 variants passed COSMOS filtering across samples than three haplotypes found, allele imbalances for mapping quality, from the ten Parkinson patients. 7 of which were called and read pair imbalances (Figure 2C). Local alignment around exclusively by VarScan 2. After manual inspection, we Frontiers in Aging | www.frontiersin.org 5 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease FIGURE 3 | (A). Validation of candidate variants. Number of mutations validated (Pass) in multiple or a single tissue or found to be false positives or germline heterozygous variants with amplicon sequencing data. Variants are distributed by calling-confidence tier, and colors indicate the tissue in which the variant was originally called in the exome data. (B). Mutational spectrum of somatic SNVs in Parkinson brains. Variants present in the brain of Parkinson patients, segregated by substitution and trinucleotide context. (C). Correlation with COSMIC signatures. Moderate Pearson correlations (r > 0.3) between the spectrum of brain somatic variants and the single base signatures (SBS) from COSMIC. Text describes the etiology or studies relevant for each signature. classified variants in four tiers of confidence based on the had only one candidate sSNV call, but DV4 and DV5 had 7 and 6, presence of mismatches in the region and the features of the respectively, demonstrating moderate interindividual variability. mutant supporting reads (including other variants in phase, Only 6 of the 27 variants were present in a single tissue, which was strand, orientation, region covered; examples in Supplementary always blood (Figure 3A, second panel). Interestingly, 76.2% (16 Figures S6, S7). Tier 1 had 17 high confidence variants, tier 2 out of 21) of variants validated in multiple tissues had been called consisted of the 7 VarScan calls, tier 3 contained 29 lower exclusively in blood in the exome data (Figure 3A, first panel), confidence candidate sSNVs, and tier 4 included 5 putative false explained by their higher frequency in this tissue (mean positives (Supplementary Table S4 and Methods). A median of difference of 10.8%, mean VAF in blood 11.5 vs. 0.7% in the 7 candidates were called in each individual (range 1–11), and other tissues). This was not a consequence of our calling method, perhaps unsurprisingly, given the limited coverage, 91.5% of all as 68.8% of these variants (11 out of 16) had no read supporting the candidate variants were called in just one tissue, mostly the alternative allele in any brain sample but were then validated blood (61.1% of single tissue calls) (Figure 3A). in at least 2 brain regions. This probably results from their We evaluated 58 out of the 59 variants with the rhAmpSeq random amplification in tissue maintenance of blood but targeted amplicon sequencing system, which allows the could also be a consequence of depletion in the central multiplexed amplification of multiple genomic positions. This nervous system. way, we sequenced data from all positions and all patients at a mean coverage of over 18,000X so that non-called individuals could be used as negative controls. Four of the samples could not The Spectrum of Somatic SNVs in be amplified enough and had low coverages (mean ≤ 2,300X, Parkinson Supplementary Figure S8). Variants were considered as The discovery that different mutagenic agents–such as UV light, validated when the support for the alternative allele was carcinogens or intrinsic cell processes–produce distinct higher than in other patient’s samples and if their VAF substitution patterns in a context-dependent manner led to the suggested they were not germline variants (see Methods). development of mutational signature analysis (Alexandrov et al., A total of 27 somatic SNVs were validated using these criteria. 2013). To obtain the mutational spectrum of Parkinson disease, Validation rates were 70.6% for tier 1 (12 out of 17), 57.1% for tier we combined the 21 sSNVs validated in at least one brain sample 2 (4 out of 7), 37.9% for tier 3 (11 out of 29) and 0% for tier 4 (0 and classified them by substitution and trinucleotide context out of the 5 negative control variants). Per individual, a median of (Figure 3B). The number of variants was insufficient for 4 variants were validated, and the median validation rate was 50% mutational signature deconvolution, so we calculated its (mean of 40.6%, ranging from 0 to 83%). All calls from 3 Pearson correlation with the COSMIC single base signatures. individuals (DV4, DV5, and DV9) were false positives. DV9 Since most of the identified variants are present in multiple Frontiers in Aging | www.frontiersin.org 6 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease FIGURE 4 | (A). Clustering of tissues by VAF. Frequency of the 21 somatic variants found in brain was used to cluster the tissues. Genes are shown and variants are ordered by individual, with age at death shown on top. Black tiles indicate the variant did not pass all validation criteria in amplicon sequencing but could still have support in the tissue. (B). Age correlation. Correlation between each patient’s number of potentially deleterious variants (nonsynonymous, in splicing consensus sites or in UTRs) found in brain and age at death. The line shows a non-significant fitted linear model (p-value = 0.19). (C). Functional enrichments of extended gene set. Enrichment ratio of the top 25 terms by FDR of an overrepresentation enrichment analysis of the co-expression network extended gene set (n = 177 genes). The considered databases were Gene Ontology Biological Process, Cellular Component, and Molecular Function, Human Phenotype Ontology and disease GLAD4U. tissues and are therefore of early origin, we expected to find a high analyze general tissue dynamics (Figure 4A). As expected from similarity to signatures SBS1 and SBS5, as recently found in the its clonal expansion, blood is the most distant tissue, with brain mutational spectrum of early embryonic SNVs (Coorens et al., tissues being more closely clustered. Remarkably, striatum and 2021). Indeed, we found both among the four moderately substantia nigra, both affected in Parkinson disease (Bernheimer correlated signatures (Pearson’s r > 0.3, Figure 3C). SBS1 is a et al., 1973), cluster together, which could be caused by their ubiquitous signature that results from the spontaneous closer developmental origin and/or more similar physiology. The deamination of methylated cytosines. SBS5 can be detected not presence of 3 variants with higher frequencies in all brain tissues only in most cancer samples (Alexandrov et al., 2020) but also in compared to blood indicates that our findings are not the result of de novo mutations (Rahbari et al., 2016), somatic mutations (Bae blood contamination in the other tissues. Importantly, blood was et al., 2018), and in population level variants (Mathieson and not called in the exome sequencing data for these variants, and Reich, 2017). In spite of our limited power in this analysis, the amplicon sequencing confirmed the VAF distribution in the signature SBS6 was also found to be highly correlated with the different tissues. de novo mutational spectrum (Rahbari et al., 2016). The Possible Impact of Deleterious sSNVs Variant Allele Frequency of Somatic SNVs in Brain Reconstructs Tissue Relationships We then tested the relationship between the number of We validated sSNVs in 6 out of the 9 subject individuals, ranging potentially deleterious brain variants and individuals’ age at from 1 to 6 variants per patient, a modest yet sensible number for death. We excluded synonymous sSNVs (4 variants, including an exome analysis. All of DV10’s sSNVs were only detected in the only sSNV validated in the brain samples from patient DV8, blood, but the other 21 variants (from DV1, DV3, DV6, DV7 and Supplementary Table S4) and intronic variants not affecting DV8) were detected in at least two brain tissues. We clustered the donor or acceptor sites (2 variants), leaving 11 nonsynonymous tissues based on their VAFs at these sSNVs. Because each sSNVs, 3 variants affecting splicing consensus sites, and 1 variant individual has just a few variants, we pooled them together to in a UTR region, as these are more probable to affect age at death. Frontiers in Aging | www.frontiersin.org 7 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease The age of onset was unknown for one of the four individuals To further investigate the tissue specificity of the genes included in this analysis, DV1. As age of onset is highly correlated harboring the 15 validated sSNVs, we performed a tissue with age at death (Pearson’s r = 0.9), we used it to evaluate the enrichment using TissueEnrich (Jain and Tuteja, 2019). In correlation between age and the number of sSNVs. Although accordance with the functional enrichment, cerebral cortex counts of deleterious somatic variants identified in brain were was the only significantly enriched tissue (adjusted p-value = negatively correlated with age at death (Figure 4B, Pearson’s r = 0.003 and Supplementary Figure S10). −0.81), the trend was not significant (p-value = 0.19), To explore the consequences of the validated variants, we unsurprisingly given the small data set. examined their effect on protein structures and found two interesting cases. All tissues from patient DV7 carried a mutation (p.Gly998Glu) in KIF5A (Uniprot Q12840), a Genes Carrying sSNVs in Parkinson are kinesin heavy chain protein. Using Genome3D (Lewis et al., Involved in Synaptic Processes Functions 2015), we found a DomSerf (Buchan et al., 2013) structural Out of the 21 sSNVs found in brain, 4 nonsynonymous variants prediction for the region containing this amino acid (913- were predicted to be deleterious by SIFT. Remarkably, three of 1032, confidence 100). The affected residue is in the surface of the genes carrying these variants are related to processes this small globular C-terminal domain (Supplementary Figure relevant in the brain. GRIP1 is involved in transmission S11A). This region interacts with kinesin adaptor proteins such as across chemical synapses and the regulation of neuron TRAK1 and TRAK2, which mediate cargo binding (Hirokawa projection arborization (Geiger et al., 2014). KCNK2 (or and Noda, 2008), suggesting the possible relevance of this sSNV. TREK-1) encodes a voltage-independent potassium channel In addition, all DV1’s tissues but cerebellum carried a substitution essential in securing saltatory conduction at high frequency in UBE2U (Q5VVX9), p. Pro96Ala, which is included in a on myelinated afferent nerves (Kanda et al., 2019). UBE2U, a DomSerf structure of residues 2-157 (confidence 100) and ubiquitination enzyme, is a candidate regulator of chromatin appears to be close to the active site. Although it seems responses at double strand breaks (Guo et al., 2017), which are improbable that the structure of the active site changes as a of fundamental relevance for gene expression in the brain consequence of this mutation (Supplementary Figure S11B), its (Madabhushi et al., 2015). Furthermore, ubiquitination flexibility or orientation may be affected. dysfunction has been linked to Parkinson (Geisler et al., 2014) and Alzheimer (Gómez-Ramos et al., 2015). The other variant is in DENND4A, a secondary guanine nucleotide DISCUSSION exchange factor that activates Rab-10, participating in the insulin-regulated glucose transporter GLUT4 translocation to In this study, we explored the presence of somatic mutations in a cell membranes (Sano et al., 2011). cohort of ten Parkinson disease patients and identified and We explored the biological pathways potentially affected by validated 27 somatic SNVs whose functions are enriched in the identified somatic mutations with an overrepresentation synaptic processes, suggesting a potential role of somatic enrichment analysis. For this analysis, we only included the 15 variants in Parkinson disease. Massive parallel sequencing sSNVs that result in nonsynonymous changes, could affect makes it possible to characterize somatic genetic variation and splicing consensus sites or in UTR regions and were validated study its role genetic disorders. These techniques have overcome in a brain tissue. Top enriched terms included “glial cell the main limitation of first-generation sequencing technologies projection”, “axonogenesis” and similar processes and are potentially capable of detecting somatic variants at very (Supplementary Figure S9). However, as expected from such low frequencies if sufficient sequencing depth is generated. a small set of genes, none of the enrichments were significant Thanks to this, the role of somatic mutations in a variety of (FDR ≤ 0.05). To gain power and retrieve the GO terms that diseases is becoming increasingly clear (Gleeson et al., 2000; better describe the functional context of the validated sSNVs, we Messiaen et al., 2011; Poduri et al., 2012; Priest et al., 2016; expanded the gene set by adding genes with significantly high co- Bar et al., 2017; Nicolas et al., 2018; Mensa-Vilaró et al., 2019), expression scores (above 900) as reported in the STRING including neurodegenerative (Park et al., 2019) and database (Szklarczyk et al., 2019), resulting in a total of 177 neurodevelopmental disorders (Dou et al., 2017). genes. Performing an enrichment analysis with the expanded However, somatic SNV calling from sequencing data outside gene set, we found significant results including multiple synaptic the framework of paired samples still poses a challenge. Single cell terms (top 25 in Figure 4C), showing that our validated sSNVs and single-cell-derived colony sequencing suffer from affect genes tightly connected to the protein networks associated amplification and in vitro growth artifacts, respectively with these functions. Besides the Gene Ontology databases, to test (Abascal et al., 2021), and the single-molecule approaches a more direct relationship of our expanded gene list with necessary to avoid them are still expensive (Abascal et al., phenotypes and diseases, we also included the Human 2021). Somatic detection calling in unpaired bulk sequencing Phenotype Ontology and the GALD4U disease databases. samples requires identifying bona fide mutations in a haystack of Chorea, dyskinesia and tremor were among the significantly artefacts. Multiple studies have addressed false positives and used enriched phenotypes and neurodegenerative diseases, and filters similar to those described here (Park et al., 2019; Wang Parkinson disease among the GLAD4U database enriched et al., 2021; Solís-Moruno et al., 2021; Kim et al., 2019), but it can terms (Supplementary Table S5). still be challenging to implement them in a consistent manner Frontiers in Aging | www.frontiersin.org 8 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease and apply them in a joint calling scenario. Here we present thousands of individuals would be needed to perform association COSMOS (https://github.com/ilobon/COSMOS), a studies (a GWAS-like approach). Furthermore, somatic computational approach to accurately identify somatic mutations increase with age, so an age-matched cohort would mutations from bulk sequencing data that is highly be essential to control for this effect. customizable to each experiment’s requirements. We highlight We discarded variants when multiple individuals had reads the relevance of including a panel of individuals to help identify supporting the same allele. This is because the probability of germline variants and recurrent artifacts, which are far more recurrent errors is much higher than the somatic mutation rate, common than somatic mutations at the sensitivity a reasonable especially given the number of divisions we can disentangle with sequencing coverage provides. We also demonstrated that bulk medium-coverage sequencing data. Also, this strategy leveraging multiple samples from the same individual can help assumes that the disease can be caused by a variety of increase the accuracy and sensitivity to identify sSNVs, as mutations, instead of a few recurrent events, as it is a previously suggested (Kim et al., 2019). Besides technical common disease with a low heritability component (Keller considerations, somatic mutations causing neurodegeneration et al., 2012). However, germline variants are recurrent because could be depleted in the affected organs, so using samples they appear in more fragile or tolerated regions of the genome. from unaffected tissues might be key for their discovery. Somatic mutations will surely be subject to the same processes Indeed, some of the somatic variants we validated did not when evaluated at a population level. These variants might then have enough support in the brain tissues’ exome data, and it be selected for or against in different tissues, as shown for was only the information from blood that allowed their detection. epithelia (Martincorena et al., 2015; Martincorena et al., 2018; This could result from their random amplification in tissue Lee-Six et al., 2019; Moore et al., 2020; Yoshida et al., 2020), maintenance of blood, especially in older aged individuals who changing our expectations to find a given somatic variant in a show higher levels of clonal hematopoiesis (Watson et al., 2020). particular tissue. Using sensitive and accurate sequencing Including a wider variety of unaffected tissues in future studies techniques, such as NanoSeq (Abascal et al., 2021), will be will be important to clarify this scenario. fundamental to discover such somatic recurrent events. A possible confounding factor in this type of analysis is blood Finally, exome sequencing data is limited to a narrow set of contamination, which would result in the presence of the same regions of the genome. This limited our power to observe somatic mutations at lower frequencies in other tissues. Our different cell lineages and their presence in each tissue or results show that three of the validated variants have consistently perform mutational signature deconvolution. Although exome higher VAFs in the brain tissues than in blood, demonstrating variants can be more easily interpreted, the decreasing cost of their presence in brain cells is more frequent than in blood. sequencing and recent advances in contact maps (Lu et al., 2020) Besides their synaptic functions, the mutational spectrum, make finding and interpreting non-coding variants more although underpowered, was similar to that of embryonic accessible. somatic mutations. Again, it would be useful to explore other The somatic mutations we identified in Parkinson patients are tissues, such as epithelium, which shares the advantage of the promising candidates to contribute to the disease, as they affect random amplification and absence of the putative depletion genes involved in neuronal and axonal pathways and interact without this risk of contamination. Nonetheless, positive with genes associated with the disease. Together with previous selection of sSNVs has been identified in normal skin studies, our results suggest an exciting new research path for the (Martincorena et al., 2015) and the extent to which this can study of disease, especially complex disorders such as be detrimental for their use as an outgroup tissue remains neurodegenerative diseases. This new evidence supports that unknown. not only germline point mutations, copy number variants, The main caveats of our study are the reduced number of mitochondrial variants, and the environment are relevant, but subject samples used and the lack of control individuals, for also somatic mutations could contribute to Parkinson disease, which paired tissue samples of similar quality are difficult to probably by affecting the same key pathways. The study of obtain. This prevented us from establishing that the uncovered somatic mutations in larger cohorts can help to identify the load of somatic variants related to neuronal functions is linked to relevant molecular routes, helping us understand the disease and the disease. As we focused on brain tissues, translation coupled finding potential therapeutic targets. damage could explain the enrichment in the genes that are expressed in this organ, but all validated sSNVs were present in blood. Extending this experiment to more tissues and METHODS individuals will be key establish the role of somatic mutations in Parkinson disease. In addition to somatic mutations, we found Exome Sequencing that our cohort carried germline variants linked to the disease. Tissue samples from cerebellum, neocortex, striatum and However, these were sporadic cases, suggesting that germline substantia nigra and blood were sourced from the HCB- mutations are not enough to cause the disease, and that the IDIBAPS biobank. Blood samples were obtained from stored presence of somatic mutations might contribute as an additive vials while brain samples were collected at autopsies, 6–18 h after factor that increases an individual’s susceptibility to the disease. death, and kept frozen at -80 °C. DNA extractions were carried Because of these complex interactions, and as these somatic out with the Qiagen DNeasy Blood & Tissue Kit. Genomic DNA variants could be also found in healthy controls, hundreds or samples were randomly fragmented into 150–200 bp length Frontiers in Aging | www.frontiersin.org 9 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease sequences. Adapters were ligated and the resulting templates were Depth of coverage files were obtained with GATK v3.6 purified with AgencourtAMPure SPRI beads. Libraries were DepthOfCoverage and GC content per target was calculated amplified by ligation-mediated polymerase chain reaction with GCContentByInterval. Then, CNVs were called jointly for (LM-PCR). The Exon Focus SureSelect kit from Agilent was all samples with XHMM v1.0 (Fromer et al., 2012) with standard used to capture the exome and paired-end 100 bp sequencing was parameters following its recommended best practices. Short performed on an Illumina Hiseq2000 platform. tandem repeats in hs37d5 were determined with Tandem Repeats Finder v.4.09 (Benson, 1999) with parameters 2 7 7 80 Sequencing Data Processing 10 12 500 -h. Homopolymers were then extracted based on their The resulting FASTQ files were mapped with BWA v0.7.8 mem (Li, homogeneous repeat motif. The 1000GP strict mask FASTA files 2013) to the human hs37d5 assembly. Lane-specific read groups were obtained from the 1000GP FTP (ftp://ftp.1000genomes.ebi. were added with Picard Tools v1.95 (Broad Institute, 2013) ac.uk/vol1/ftp/release/20130502/supporting/accessible_genome_ AddOrReplaceReadGroups and bams were merged by sample masks/) and were transformed into a BED file. BED files for with samtools v1.9 (Li et al., 2009). Read duplicates were WGAC segmental duplications, common dbSNP SNPs and removed with Picard Tools v1.95 MarkDuplicates mappability for 100mers for hg19—which shares coordinates REMOVE_DUPLICATES = true. Base quality score recalibration with hs37d5—were obtained from the UCSC table browser and indel realignment were applied following GATK’s best practices (Karolchik et al., 2004). This information was annotated in the (DePristo et al., 2011) with GATK v3.6 (McKenna et al., 2010). VCF files with BCFtools (Li et al., 2009). A panel of normals Secondary alignments were also excluded with samtools view -F 256. (PON) containing allele counts at each genomic position along Cram files including unmapped reads are available at ENA 428 individuals sequenced at the Sanger Institute was used to (European Nucleotide Archive) under the study accession identify recurrent errors. number PRJEB43918. All the code used to process and generate data are available in https://github.com/ilobon/ParkinsonSomatic Somatic Variant Filtering Using COSMOS We first defined a non-callable set of positions, including off- Germline Variants target calls as well as those overlapping with the 1000GP strict Germline variants were called with GATK v3.6. First, GVCFs mask, the WCAG track, mappability lower than 1, by were obtained for each sample independently with -T homopolymers or within 5 base pairs of indels. Additional HaplotypeCaller–emitRefConfidence GVCF. Then, all samples sources of artefacts were identified in the lax call sets by were genotyped together with -T GenotypeGVCFs and a standard manual inspection of the raw data with IGV (Robinson et al., hard filter was applied with -T VariantFiltration--filterExpression 2011), oftentimes evidenced by recurrent patterns in different “QD < 2.0 || FS > 60.0 || MQ < 40.0 || MQRankSum < -12.5 || individuals. To filter calls we wrote a python module, COSMOS ReadPosRankSum < -8.0”. A PCA of the hard-filtered genotypes (Combined Or Single sample MOSaicism detection), which is was performed with EIGENSOFT v7.2.1 (Price et al., 2006). available in https://github.com/ilobon/COSMOS. The first step is Information on the called variants was annotated with SnpEff to annotate all the information necessary to classify true and false 4.3t and SnpSift 4.3t (Cingolani et al., 2012) and dbNSFP was calls. Then, variants are filtered according to the user indicated used to add population frequencies, effect prediction and features and thresholds. Two different filtering approaches can be conservation scores. Overrepresentation enrichment analysis used. When multiple samples—tissues or replicates—from the was performed with WebGestalt (Zhang et al., 2005) using the same individual are available, their combined information can be Gene Ontology (GO) database (Carbon et al., 2009) for molecular used to filter calls more accurately. This is, since true and false functions and genome protein-coding genes as background. variants have partially overlapping values in the determining features, even true somatic mutations can fail some tests. Hence, Somatic Variant Calling we can take advantage of the multiple samples, and assuming the Somatic variants were called using HaplotypeCaller and VarScan variants are present in more than one of them, require than at 2 with lax parameters. Then, an extensive filtering strategy was least n samples pass each filter, allowing a different combination applied to recover high confidence somatic SNVs. For of passing samples at each feature (-combined TRUE- HaplotypeCaller somatic variant calling, GVCFs per sample nSamplesPerInd n). COSMOS can also be used to filter each were obtained with GATK -T HaplotypeCaller -ploidy 10 -A sample individually (-combined FALSE), or output the result of StrandAlleleCountsBySample -- emitRefConfidence GVCF. both approaches (-combined BOTH-nSamplesPerInd n). Then, all GVCFs were genotyped together per chromosome To filter our Parkinson exome data, we used both approaches, with -T GenotypeGVCFs -L chr -ploidy 10 -A requiring at each position either a single sample passing all filters or StrandAlleleCountsBySample to obtain somatic SNV and indel any combination of four out of the five tissues passing at every calls. For VarScan 2 (Koboldt et al., 2012) somatic variant calling, filter. The parameters we used were -c BOTH -ns 4 -ad 2 -adss 3 -vaf mpileup files were obtained with samtools mpileup per 0.5 -dp1 20 -dp2 100 -sr 2 -pr 4 -sp 0.05 -b 0.05 -nrl 4 -hap 4 -cnv NO individual. Then, single nucleotide variants were called with -pir 4,4 -vafq 0.4 -clip 0.9 -mq 0.05 -mm 0.05 -pon 0.05. In short, this VarScan v2.3.2 mpileup2snp with lax parameters: -min- requires at least 2 reads supporting the alternative allele, or 3 for the coverage 1 -min-reads2 1 -p-value 1 -min-var-freq single sample approach; a variant allele frequency lower than 0.5; a 0.000001 -output-vcf. Indels were called with mpileup2indel depth between 20 and 100; a strand count ratio <2; a pair count with the same parameters. ratio <4; non-significant p-values for a Poisson test of strand Frontiers in Aging | www.frontiersin.org 10 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease counts; a significant binomial test of allele counts; at most 4 required for this study in accordance with the national legislation variants within read length; less than 4 haplotypes; absence of and the institutional requirements. XHMM CNV call; PIR score of 4, meaning that position in reads of each allele is not biased; a variant allele frequency from direct bam read counts <0.4; an allele clipping ratio difference <0.9; non- AUTHOR CONTRIBUTIONS significant Mann-Whitney tests for mapping quality an number of mismatches per allele and a significant beta-binomial test for allele ES, IL, DJ, and JA designed the study, MM and ET provided the counts compared to the PON. More features and different samples, AM selected and prepared the samples for sequencing, IL, thresholds can be used for filtering depending on each dataset PE-C, and RG-P analyzed the data, MS-M performed the amplicon characteristics such as depth. sequencing, FA and RR provided somatic variant detection expertise, ES, DJ, FC, and TM-B supervised the research and IL wrote the Amplicon-Based Deep Sequencing manuscript with suggestions from all other authors. To validate the high confidence variants, we performed amplicon-based deep sequencing (ADS) by using rhAmpSeq technology (IDT, Coralville, United States). This is a FUNDING multiplexed strategy that amplifies all selected positions in a unified reaction. We then sequenced the amplified material from This work was supported by the grants from the Spanish each sample in a MiSeq v3 run obtaining paired end 300bp reads MINECO and MICIN (SAF2016-76340R and PID2019- to a mean coverage ~18,000X. Data were processed the same way 106764RB-C21, Excellence Unit María de Maeztu/Institute of as the exome files. To consider a variant as a validated sSNV in Neurosciences), CIBERNED (ISCIII), and by the “Fundación each sample, we required that (1) it was the second most common Reina Sofía” (Exome Project) awarded to ES, by RTI 2018- allele in that sample and (2) its VAF was higher than the mean + 2 096824-B-C22 grant from Agencia estatal de Investigación- standard deviations of all other individuals’ samples VAFs with Spanish Ministry of Science, Innovation and Universities co- sufficient coverage (defined as not lower than the mean–1 financed by FEDER, and by Direcció General de Recerca, standard deviation). We additionally classified variants passing Generalitat de Catalunya (2017SGR-702) to FC, MS-M is these filters in multiple tissues of an individual and with VAFs > supported by the Ministerio de Economía y Competitividad, 30% in all of them as germline heterozygous variants Spain (Maria de Maetzu grant MDM-2014-0370-16-3). PE-C (Supplementary material). was supported by a Formació de Personal Investigador fellowship from Generalitat de Catalunya (FI_B00122). RR is funded by Cancer Research UK (C66259/A27114). DATA AVAILABILITY STATEMENT The datasets presented in this study can be found in online ACKNOWLEDGMENTS repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ebi.ac. We are grateful to patients and their families for donating the uk/ena/browser/view/PRJEB43918. samples that made this study possible. ETHICS STATEMENT SUPPLEMENTARY MATERIAL The studies involving human participants were reviewed and The Supplementary Material for this article can be found online at: approved by Clinical Research Ethics Committee, Hospital Clinic https://www.frontiersin.org/articles/10.3389/fragi.2022.851039/ Barcelona. Written informed consent for participation was not full#supplementary-material Alexandrov, L. B., Nik-Zainal, S., Wedge, D. C., Campbell, P. J., and Stratton, M. R. REFERENCES (2013). Deciphering Signatures of Mutational Processes Operative in Human Cancer. Cel Rep. 3, 246–259. doi:10.1016/j.celrep.2012.12.008 Abascal, F., Harvey, L. M. R., Mitchell, E., Lawson, A. R. J., Lensing, S. V., Ellis, P., Bae, T., Tomasini, L., Mariani, J., Zhou, B., Roychowdhury, T., Franjic, D., et al. et al. (2021). Somatic Mutation Landscapes at Single-Molecule Resolution. (2018). Different Mutational Rates and Mechanisms in Human Cells at Nature 593, 405–410. doi:10.1038/s41586-021-03477-4 Pregastrulation and Neurogenesis. Science 359, 550–555. doi:10.1126/science.aan8690 Abyzov, A., Tomasini, L., Zhou, B., Vasmatzis, N., Coppola, G., Amenduni, M., Baillie, J. K., Barnett, M. W., Upton, K. R., Gerhardt, D. J., Richmond, T. A., De et al. (2017). One Thousand Somatic SNVs Per Skin Fibroblast Cell Set Baseline Sapio, F., et al. (2011). Somatic Retrotransposition Alters the Genetic Landscape of Mosaic Mutational Load with Patterns that Suggest Proliferative Origin. of the Human Brain. Nature 479, 534–537. doi:10.1038/nature10531 Genome Res. 27, 512–523. doi:10.1101/gr.215517.116 Bar, D. Z., Arlt, M. F., Brazier, J. F., Norris, W. E., Campbell, S. E., Chines, P., et al. Alexandrov, L. B., Kim, J., Haradhvala, N. J., Huang, M. N., Tian Ng, A. W., Wu, Y., (2017). A Novel Somatic Mutation Achieves Partial rescue in a Child with et al. (2020). The Repertoire of Mutational Signatures in Human Cancer. Hutchinson-Gilford Progeria Syndrome. J. Med. Genet. 54, 212–216. doi:10. Nature 578, 94–101. doi:10.1038/s41586-020-1943-3 1136/jmedgenet-2016-104295 Frontiers in Aging | www.frontiersin.org 11 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease Bender, A., Krishnan, K. J., Morris, C. M., Taylor, G. A., Reeve, A. K., Perry, R. H., from Whole-Exome Sequencing Depth. Am. J. Hum. Genet. 91, 597–607. et al. (2006). High Levels of Mitochondrial DNA Deletions in Substantia Nigra doi:10.1016/j.ajhg.2012.08.005 Neurons in Aging and Parkinson Disease. Nat. Genet. 38, 515–517. doi:10.1038/ Geiger, J. C., Lipka, J., Segura, I., Hoyer, S., Schlager, M. A., Wulf, P. S., et al. (2014). ng1769 The GRIP1/14-3-3 Pathway Coordinates Cargo Trafficking and Dendrite Benson, G. (1999). Tandem Repeats Finder: a Program to Analyze DNA Sequences. Development. Dev. Cel 28, 381–393. doi:10.1016/j.devcel.2014.01.018 Nucleic Acids Res. 27, 573–580. doi:10.1093/nar/27.2.573 Geisler, S., Vollmer, S., Golombek, S., and Kahle, P. J. (2014). The Ubiquitin- Bernheimer, H., Birkmayer, W., Hornykiewicz, O., Jellinger, K., and Seitelberger, F. Conjugating Enzymes UBE2N, UBE2L3 and UBE2D2/3 Are Essential for (1973). Brain Dopamine and the Syndromes of Parkinson and Huntington Parkin-dependent Mitophagy. J. Cel Sci. 127, 3280–3293. doi:10.1242/jcs. Clinical, Morphological and Neurochemical Correlations. J. Neurol. Sci. 20, 146035 415–455. doi:10.1016/0022-510x(73)90175-5 Gleeson, J. G., Minnerath, S., Kuzniecky, R. I., Dobyns, W. B., Young, I. D., Ross, M. Broad Institute (2013). Picard: A Set of Command Line Tools (In Java) for E., et al. (2000). Somatic and Germline Mosaic Mutations in the Doublecortin Manipulating High-Throughput Sequencing (HTS) Data and Formats Such Gene Are Associated with Variable Phenotypes. Am. J. Hum. Genet. 67, as SAM/BAM/CRAM and VCF. Available at: https://broadinstitute.github.io/ 574–581. doi:10.1086/303043 picard/. Gómez-Ramos, A., Podlesniy, P., Soriano, E., and Avila, J. (2015). Distinct Brunner, S. F., Roberts, N. D., Wylie, L. A., Moore, L., Aitken, S. J., Davies, S. E., X-Chromosome SNVs from Some Sporadic AD Samples. Sci. Rep. 5, et al. (2019). Somatic Mutations and Clonal Dynamics in Healthy and Cirrhotic 18012–18023. doi:10.1038/srep18012 Human Liver. Nature 574, 538–542. doi:10.1038/s41586-019-1670-9 Guella, I., Asselta, R., Tesei, S., Zini, M., Pezzoli, G., and Duga, S. (2010). Buchan, D. W., Minneci, F., Nugent, T. C., Bryson, K., and Jones, D. T. (2013). ThePDXKrs2010795 Variant Is Not Associated with Parkinson Disease in Scalable Web Services for the PSIPRED Protein Analysis Workbench. Nucleic Italy. Ann. Neurol. 67, 411–412. doi:10.1002/ana.21964 Acids Res. 41, W349–W357. doi:10.1093/nar/gkt381 Guertin, D. A., and Sabatini, D. M. (2007). Defining the Role of mTOR in Cancer. Cai, X., Evrony, G. D., Lehmann, H. S., Elhosary, P. C., Mehta, B. K., Poduri, A., Cancer Cell 12, 9–22. doi:10.1016/j.ccr.2007.05.008 et al. (2014). Single-Cell, Genome-wide Sequencing Identifies Clonal Somatic Guo, Y., An, L., Ng, H.-M., Sy, S. M. H., and Huen, M. S. Y. (2017). An E2-Guided Copy-Number Variation in the Human Brain. Cel Rep. 8, 1280–1289. doi:10. E3 Screen Identifies the RNF17-Ube2u Pair as Regulator of the Radiosensitivity, 1016/j.celrep.2014.07.043 Immunodeficiency, Dysmorphic Features, and Learning Difficulties (RIDDLE) Carbon, S., Ireland, A., Mungall, C. J., Shu, S., Marshall, B., and Lewis, S. (2009). Syndrome Protein RNF168. J. Biol. Chem. 292, 967–978. doi:10.1074/jbc.m116. AmiGO: Online Access to Ontology and Annotation Data. Bioinformatics 25, 758854 288–289. doi:10.1093/bioinformatics/btn615 Heckman, M. G., Kasanuki, K., Diehl, N. N., Koga, S., Soto, A., Murray, M. E., et al. Chen, C.-M., Chen, Y.-C., Chiang, M.-C., Fung, H.-C., Chang, K.-H., Lee-Chen, G.- (2017). Parkinson’s Disease Susceptibility Variants and Severity of Lewy Body J., et al. (2016). Association of GCH1 and MIR4697 , but Not SIPA1L2 and Pathology. Parkinsonism Relat. Disord. 44, 79–84. doi:10.1016/j.parkreldis. VPS13C Polymorphisms, with Parkinson’s Disease in Taiwan. Neurobiol. Aging 2017.09.009 39, e1–221. doi:10.1016/j.neurobiolaging.2015.12.016 Hirokawa, N., and Noda, Y. (2008). Intracellular Transport and Kinesin Chen, L., Liu, P., Evans, T. C., and Ettwiller, L. M. (2017). DNA Damage Is a Superfamily Proteins, KIFs: Structure, Function, and Dynamics. Physiol. Rev. Pervasive Cause of Sequencing Errors, Directly Confounding Variant 88, 1089–1118. doi:10.1152/physrev.00023.2007 Identification. Science 355, 752–756. doi:10.1126/science.aai8690 Hoang, M. L., Kinde, I., Tomasetti, C., McMahon, K. W., Rosenquist, T. A., Chronister, W. D., Burbulis, I. E., Wierman, M. B., Wolpert, M. J., Haakenson, M. Grollman, A. P., et al. (2016). Genome-wide Quantification of Rare Somatic F., Smith, A. C. B., et al. (2019). Neurons with Complex Karyotypes Are Rare in Mutations in normal Human Tissues Using Massively Parallel Sequencing. Aged Human Neocortex. Cel Rep. 26, 825–835. e7. doi:10.1016/j.celrep.2018. Proc. Natl. Acad. Sci. U.S.A. 113, 9846–9851. doi:10.1073/pnas.1607794113 12.107 Huang, X., Chen, P. C., and Poole, C. (2004). APOE- 2 Allele Associated with Chuang, Y.-H., Paul, K. C., Bronstein, J. M., Bordelon, Y., Horvath, S., and Ritz, B. Higher Prevalence of Sporadic Parkinson Disease. Neurology 62, 2198–2202. (2017). Parkinson’s Disease Is Associated with DNA Methylation Levels in doi:10.1212/01.wnl.0000130159.28215.6a Human Blood and Saliva. Genome Med. 9, 76. doi:10.1186/s13073-017-0466-5 International Parkinson Disease Genomics Consortium and Wellcome Trust Case Cingolani, P., Patel, V. M., Coon, M., Nguyen, T., Land, S. J., Ruden, D. M., et al. Control Consortium 2 (2011). A Two-Stage Meta-Analysis Identifies Several (2012). Using Drosophila melanogaster as a Model for Genotoxic Chemical New Loci for Parkinson’s Disease. Plos Genet. 7, e1002142. Mutational Studies with a New Program, SnpSift. Front. Gene 3, 35. doi:10. International Parkinson Disease Genomics ConsortiumNalls, M. A., Plagnol, V., 3389/fgene.2012.00035 Hernandez, D. G., Sharma, M., Sheerin, U. M., et al. (2011). Imputation of Coorens, T. H., Moore, L., Robinson, P. S., Sanghvi, R., Christopher, J., Hewinson, Sequence Variants for Identification of Genetic Risks for Parkinson’s Disease: a J., et al. (2021). Extensive Phylogenies of Human Development Inferred from Meta-Analysis of Genome-wide Association Studies. Lancet 377, 641–649. Somatic Mutations. Nature 597 (7876), 387–392. doi:10.1016/S0140-6736(10)62345-8 DePristo, M. A., Banks, E., Poplin, R., Garimella, K. V., Maguire, J. R., Hartl, C., Jain, A., and Tuteja, G. (2019). TissueEnrich: Tissue-specific Gene Enrichment et al. (2011). A Framework for Variation Discovery and Genotyping Using Next- Analysis. Bioinformatics 35, 1966–1967. doi:10.1093/bioinformatics/bty890 Generation DNA Sequencing Data. Nat. Genet. 43, 491–498. doi:10.1038/ng.806 Jaiswal, S., Fontanillas, P., Flannick, J., Manning, A., Grauman, P. V., Mar, B. G., Do, C. B., Tung, J. Y., Dorfman, E., Kiefer, A. K., Drabant, E. M., Francke, U., et al. et al. (2014). Age-Related Clonal Hematopoiesis Associated with Adverse (2011). Web-Based Genome-wide Association Study Identifies Two Novel Loci Outcomes. N. Engl. J. Med. 371, 2488–2498. doi:10.1056/nejmoa1408617 and a Substantial Genetic Component for Parkinson’s Disease. Plos Genet. 7, Jinn, S., Drolet, R. E., Cramer, P. E., Wong, A. H.-K., Toolan, D. M., Gretzula, C. A., e1002141. doi:10.1371/journal.pgen.1002141 et al. (2017). TMEM175 Deficiency Impairs Lysosomal and Mitochondrial Dou, Y., Yang, X., Li, Z., Wang, S., Zhang, Z., Ye, A. Y., et al. (2017). Postzygotic Function and Increases α-synuclein Aggregation. Proc. Natl. Acad. Sci. U.S.A. Single-nucleotide Mosaicisms Contribute to the Etiology of Autism Spectrum 114, 2389–2394. doi:10.1073/pnas.1616332114 Disorder and Autistic Traits and the Origin of Mutations. Hum. Mutat. 38, Kanda, H., Ling, J., Tonomura, S., Noguchi, K., Matalon, S., and Gu, J. G. (2019). 1002–1013. doi:10.1002/humu.23255 TREK-1 and TRAAK Are Principal K+ Channels at the Nodes of Ranvier for Elstner, M., Morris, C. M., Heim, K., Lichtner, P., Bender, A., Mehta, D., et al. Rapid Action Potential Conduction on Mammalian Myelinated Afferent (2009). Single-cell Expression Profiling of Dopaminergic Neurons Combined Nerves. Neuron 104, 960–971. e7. doi:10.1016/j.neuron.2019.08.042 with Association Analysis Identifies Pyridoxal Kinase as Parkinson’s Disease Karolchik, D., Hinrichs, A. S., Furey, T. S., Roskin, K. M., Sugnet, C. W., Haussler, Gene. Ann. Neurol. 66, 792–798. doi:10.1002/ana.21780 D., et al. (2004). The UCSC Table Browser Data Retrieval Tool. Nucleic Acids Federoff, M., Jimenez-Rolando, B., Nalls, M. A., and Singleton, A. B. (2012). A Res. 32, 493D–496D. doi:10.1093/nar/gkh103 Large Study Reveals No Association between APOE and Parkinson’s Disease. Keller, M. F., Saad, M., Bras, J., Bettella, F., Nicolaou, N., Simon-Sanchez, J., et al. Neurobiol. Dis. 46, 389–392. doi:10.1016/j.nbd.2012.02.002 (2012). Using Genome-wide Complex Trait Analysis to Quantify ’missing Fromer, M., Moran, J. L., Chambert, K., Banks, E., Bergen, S. E., Ruderfer, D. M., Heritability’ in Parkinson’s Disease. Hum. Mol. Genet. 21, 4996–5009. doi:10. et al. (2012). Discovery and Statistical Genotyping of Copy-Number Variation 1093/hmg/dds335 Frontiers in Aging | www.frontiersin.org 12 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease Kim, J., Kim, D., Lim, J. S., Maeng, J. H., Son, H., Kang, H.-C., et al. (2019). The Use Analyzing Next-Generation DNA Sequencing Data. Genome Res. 20, of Technical Replication for Detection of Low-Level Somatic Mutations in 1297–1303. doi:10.1101/gr.107524.110 Next-Generation Sequencing. Nat. Commun. 10, 1047. doi:10.1038/s41467- Mensa-Vilaró, A., Bravo García-Morato, M., de la Calle-Martin, O., Franco-Jarava, 019-09026-y C., Martínez-Saavedra, M. T., González-Granado, L. I., et al. (2019). Kluge, M. L., Graber, E., Foley, K., Hansen, L. V., Sellers, H. L., Milosevic, D., et al. Unexpected Relevant Role of Gene Mosaicism in Patients with Primary (2019). Monozygotic Twins Discordant for Congenital Adrenal Hyperplasia Immunodeficiency Diseases. J. Allergy Clin. Immunol. 143, 359–368. doi:10. Due to Mosaicism. Eur. J. Endocrinol. 182, 19–0249. doi:10.1530/EJE-19-0249 1016/j.jaci.2018.09.009 Koboldt, D. C., Zhang, Q., Larson, D. E., Shen, D., McLellan, M. D., Lin, L., et al. Messiaen, L., Vogt, J., Bengesser, K., Fu, C., Mikhail, F., Serra, E., et al. (2011). (2012). VarScan 2: Somatic Mutation and Copy Number Alteration Discovery Mosaic Type-1 NF1 Microdeletions as a Cause of Both Generalized and in Cancer by Exome Sequencing. Genome Res. 22, 568–576. doi:10.1101/gr. Segmental Neurofibromatosis Type-1 (NF1). Hum. Mutat. 32, 213–219. 129684.111 doi:10.1002/humu.21418 Kong, P., Lei, P., Zhang, S., Li, D., Zhao, J., and Zhang, B. (2018). Integrated Mokretar, K., Pease, D., Taanman, J.-W., Soenmez, A., Ejaz, A., Lashley, T., et al. Microarray Analysis provided a New Insight of the Pathogenesis of Parkinson’s (2018). Somatic Copy Number Gains of α-synuclein (SNCA) in Parkinson’s Disease. Neurosci. Lett. 662, 51–58. doi:10.1016/j.neulet.2017.09.051 Disease and Multiple System Atrophy Brains. Brain 141, 2419–2431. doi:10. Krøigård, A. B., Thomassen, M., Lænkholm, A.-V., Kruse, T. A., and Larsen, M. J. 1093/brain/awy157 (2016). Evaluation of Nine Somatic Variant Callers for Detection of Somatic Moore, L., Leongamornlert, D., Coorens, T. H. H., Sanders, M. A., Ellis, P., Dentro, Mutations in Exome and Targeted Deep Sequencing Data. PLoS One 11, S. C., et al. (2020). The Mutational Landscape of normal Human Endometrial e0151664. Epithelium. Nature 580, 640–646. doi:10.1038/s41586-020-2214-z Lee-Six, H., Olafsson, S., Ellis, P., Osborne, R. J., Sanders, M. A., Moore, L., et al. Nalls, M. A., Pankratz, N., Lill, C. M., Do, C. B., Hernandez, D. G., Saad, M., et al. (2019). The Landscape of Somatic Mutation in normal Colorectal Epithelial (2014). Large-scale Meta-Analysis of Genome-wide Association Data Identifies Cells. Nature 574, 532–537. doi:10.1038/s41586-019-1672-7 Six New Risk Loci for Parkinson’s Disease. Nat. Genet. 46, 989–993. doi:10. Lesage, S., and Brice, A. (2009). Parkinson’s Disease: from Monogenic Forms to 1038/ng.3043 Genetic Susceptibility Factors. Hum. Mol. Genet. 18, R48–R59. doi:10.1093/ Nicolas, G., Acuña-Hidalgo, R., Keogh, M. J., Quenez, O., Steehouwer, M., hmg/ddp012 Lelieveld, S., et al. (2018). Somatic Variants in Autosomal Dominant Genes Lewis, T. E., Sillitoe, I., Andreeva, A., Blundell, T. L., Buchan, D. W. A., Chothia, C., Are a Rare Cause of Sporadic Alzheimer’s Disease. Alzheimer’s Demen. 14, et al. (2015). Genome3D: Exploiting Structure to Help Users Understand Their 1632–1639. doi:10.1016/j.jalz.2018.06.3056 Sequences. Nucleic Acids Res. 43, D382–D386. doi:10.1093/nar/gku973 Nowell, P. C. (1976). The Clonal Evolution of Tumor Cell Populations. Science 194, Li, H. (2013). Aligning Sequence Reads, Clone Sequences and Assembly Contigs with 23–28. doi:10.1126/science.959840 BWA-MEM. Ithaca, NY: Cornell University. Pamphlett, R. (2004). Somatic Mutation: a Cause of Sporadic Neurodegenerative Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., et al. (2009). Diseases? Med. Hypotheses 62, 679–682. doi:10.1016/j.mehy.2003.11.023 The Sequence Alignment/Map Format and SAMtools. Bioinformatics 25, Parcerisas, A., Rubio, S. E., Muhaisen, A., Gómez-Ramos, A., Pujadas, L., Puiggros, 2078–2079. doi:10.1093/bioinformatics/btp352 M., et al. (2014). Somatic Signature of Brain-specific Single Nucleotide Li, J., Ruskey, J. A., Arnulf, I., Dauvilliers, Y., Hu, M. T. M., Högl, B., et al. (2018). Variations in Sporadic Alzheimer’s Disease. Jad 42, 1357–1382. doi:10.3233/ Full Sequencing and Haplotype Analysis of MAPT in Parkinson’s Disease and jad-140891 Rapid Eye Movement Sleep Behavior Disorder. Mov Disord. 33, 1016–1020. Park, J. S., Lee, J., Jung, E. S., Kim, M. H., Kim, I. B., Son, H., et al. (2019). Brain doi:10.1002/mds.27385 Somatic Mutations Observed in Alzheimer’s Disease Associated with Aging and Lill, C. M., Roehr, J. T., McQueen, M. B., Kavvoura, F. K., Bagade, S., Schjeide, B.- Dysregulation of Tau Phosphorylation. Nat. Commun. 10, 1–12. doi:10.1038/ M. M., et al. (2012). Comprehensive Research Synopsis and Systematic Meta- s41467-019-11000-7 Analyses in Parkinson’s Disease Genetics: The PDGene Database. Plos Genet. 8, Pellegrini, L., Wetzel, A., Grannó, S., Heaton, G., and Harvey, K. (2016). Back to the e1002548. doi:10.1371/journal.pgen.1002548 Tubule: Microtubule Dynamics in Parkinson’s Disease. Cell Mol Life Sci 74 (74), Lim, J. S., Kim, W.-i., Kang, H.-C., Kim, S. H., Park, A. H., Park, E. K., et al. (2015). 409–434. doi:10.1007/s00018-016-2351-6 Brain Somatic Mutations in MTOR Cause Focal Cortical Dysplasia Type II Poduri, A., Evrony, G. D., Cai, X., Elhosary, P. C., Beroukhim, R., Lehtinen, M. K., Leading to Intractable Epilepsy. Nat. Med. 21, 395–400. doi:10.1038/nm.3824 et al. (2012). Somatic Activation of AKT3 Causes Hemispheric Developmental Lodato, M. A., Rodin, R. E., Bohrson, C. L., Coulter, M. E., Barton, A. R., Kwon, M., Brain Malformations. Neuron 74, 41–48. doi:10.1016/j.neuron.2012.03.010 et al. (2018). Aging and Neurodegeneration Are Associated with Increased Price, A. L., Patterson, N. J., Plenge, R. M., Weinblatt, M. E., Shadick, N. A., and Mutations in Single Human Neurons. Science 359, 555–559. doi:10.1126/ Reich, D. (2006). Principal Components Analysis Corrects for Stratification in science.aao4426 Genome-wide Association Studies. Nat. Genet. 38, 904–909. doi:10.1038/ Lodato, M. A., Woodworth, M. B., Lee, S., Evrony, G. D., Mehta, B. K., Karger, A., ng1847 et al. (2015). Somatic Mutation in Single Human Neurons Tracks Priest, J. R., Gawad, C., Kahlig, K. M., Yu, J. K., O’Hara, T., Boyle, P. M., et al. Developmental and Transcriptional History. Science 350, 94–98. doi:10. (2016). Early Somatic Mosaicism Is a Rare Cause of Long-QT Syndrome. Proc. 1126/science.aab1785 Natl. Acad. Sci. U.S.A. 113, 11555–11560. doi:10.1073/pnas.1607187113 Lu, L., Liu, X., Huang, W. K., Giusti-Rodríguez, P., Cui, J., Zhang, S., et al. (2020). Proukakis, C., Shoaee, M., Morris, J., Brier, T., Kara, E., Sheerin, U. M., et al. (2014). Robust Hi-C Maps of Enhancer-Promoter Interactions Reveal the Function of Analysis of Parkinson’s Disease Brain-Derived DNA for Alpha-synuclein Coding Non-coding Genome in Neural Development and Diseases. Mol. Cel 79 (3), Somatic Mutations. Mov Disord. 29, 1060–1064. doi:10.1002/mds.25883 521–534. doi:10.1016/j.molcel.2020.06.007 Rahbari, R., Wuster, A., Lindsay, S. J., Hardwick, R. J., Alexandrov, L. B., Al Turki, Madabhushi, R., Gao, F., Pfenning, A. R., Pan, L., Yamakawa, S., Seo, J., et al. S., et al. (2016). Timing, Rates and Spectra of Human Germline Mutation. Nat. (2015). Activity-Induced DNA Breaks Govern the Expression of Neuronal Genet. 48, 126–133. doi:10.1038/ng.3469 Early-Response Genes. Cell 161, 1592–1605. doi:10.1016/j.cell.2015.05.032 Ramakrishnan, V., Alphonsa, T. A., Husain, R. A., Ahmed, S. S., Subramaniyan, K., Martincorena, I., Fowler, J. C., Wabik, A., Lawson, A. R. J., Abascal, F., Hall, M. W. and Kumar, S. (2016). Association of Rs1801582 and Rs1801334 PARK2 J., et al. (2018). Somatic Mutant Clones Colonize the Human Esophagus with Polymorphisms with Risk of Parkinson’s Disease: A Case-Control Study in Age. Science 362, 911–917. doi:10.1126/science.aau3879 South India and Meta-Analysis. Meta Gene 10, 32–38. doi:10.1016/j.mgene. Martincorena, I., Roshan, A., Gerstung, M., Ellis, P., Van Loo, P., McLaren, S., et al. 2016.09.007 (2015). High burden and Pervasive Positive Selection of Somatic Mutations in Rentzsch, P., Witten, D., Cooper, G. M., Shendure, J., and Kircher, M. (2019). normal Human Skin. Science 348, 880–886. doi:10.1126/science.aaa6806 CADD: Predicting the Deleteriousness of Variants throughout the Human Mathieson, I., and Reich, D. (2017). Differences in the Rare Variant Spectrum Among Genome. Nucleic Acids Res. 47, D886–D894. doi:10.1093/nar/gky1016 Human Populations. PLOS Genet. 13, e1006581. doi:10.1371/journal.pgen.1006581 Robinson, J. T., Thorvaldsdóttir, H., Winckler, W., Guttman, M., Lander, E. S., McKenna, A., Hanna, M., Banks, E., Sivachenko, A., Cibulskis, K., Kernytsky, A., Getz, G., et al. (2011). Integrative Genomics Viewer. Nat. Biotechnol. 29, 24–26. et al. (2010). The Genome Analysis Toolkit: a MapReduce Framework for doi:10.1038/nbt.1754 Frontiers in Aging | www.frontiersin.org 13 April 2022 | Volume 3 | Article 851039 Lobon et al. Somatic Mutations in Parkinson Disease Rocca, W. A., McDonnell, S. K., Strain, K. J., Bower, J. H., Ahlskog, J. E., Elbaz, A., Williams-Gray, C. H., Goris, A., Saiki, M., Foltynie, T., Compston, D. A. S., Sawcer, et al. (2004). Familial Aggregation of Parkinson’s Disease: The Mayo Clinic S. J., et al. (2009). Apolipoprotein E Genotype as a Risk Factor for Susceptibility Family Study. Ann. Neurol. 56, 495–502. doi:10.1002/ana.20228 to and Dementia in Parkinson’s Disease. J. Neurol. 256, 493–498. doi:10.1007/ Sano, H., Peck, G. R., Kettenbach, A. N., Gerber, S. A., and Lienhard, G. E. (2011). s00415-009-0119-8 Insulin-stimulated GLUT4 Protein Translocation in Adipocytes Requires the Xu, C. (2018). A Review of Somatic Single Nucleotide Variant Calling Algorithms Rab10 Guanine Nucleotide Exchange Factor dennd4C. J. Biol. Chem. 286, for Next-Generation Sequencing Data. Comput. Struct. Biotechnol. J. 16, 15–24. 16541–16545. doi:10.1074/jbc.c111.228908 doi:10.1016/j.csbj.2018.01.003 Smith, Y., Villalba, R. M., and Raju, D. V. (2009). Striatal Spine Plasticity in YemniAl, E. A., Monies, D., Alkhairallah, T., Bohlega, S., Abouelhoda, M., Parkinson’s Disease: Pathological or Not? Parkinsonism Relat. Disord. 15, Magrashi, A., et al. (2019). Integrated Analysis of Whole Exome Sequencing S156–S161. doi:10.1016/s1353-8020(09)70805-3 and Copy Number Evaluation in Parkinson’s Disease. Sci. Rep. 9, 3344–3349. Solís-Moruno, M., Mensa-Vilaró, A., Batlle-Masó, L., Lobón, I., Bonet, N., doi:10.1038/s41598-019-40102-x Marquès-Bonet, T., et al. (2021). Assessment of the Gene Mosaicism burden Yoshida, K., Gowers, K. H. C., Lee-Six, H., Chandrasekharan, D. P., Coorens, T., in Blood and its Implications for Immune Disorders. Sci. Rep. 111 (11), 1–11. Maughan, E. F., et al. (2020). Tobacco Smoking and Somatic Mutations in doi:10.1038/s41598-021-92381-y Human Bronchial Epithelium. Nature 578, 266–272. doi:10.1038/s41586-020- Song, G. G., and Lee, Y. H. (2013). Pathway Analysis of Genome-wide Association 1961-1 Studies for Parkinson’s Disease. Mol. Biol. Rep. 40, 2599–2607. doi:10.1007/ Zhang, B., Kirov, S., and Snoddy, J. (2005). WebGestalt: an Integrated System for s11033-012-2346-9 Exploring Gene Sets in Various Biological Contexts. Nucleic Acids Res. 33, Spataro, N., Calafell, F., Cervera-Carles, L., Casals, F., Pagonabarraga, J., Pascual- W741–W748. doi:10.1093/nar/gki475 Sedano, B., et al. (2015). Mendelian Genes for Parkinson’s Disease Contribute to the Sporadic Forms of the Disease†. Hum. Mol. Genet. 24, 2023–2034. doi:10. Conflict of Interest: The authors declare that the research was conducted in the 1093/hmg/ddu616 absence of any commercial or financial relationships that could be construed as a Szklarczyk, D., Gable, A. L., Lyon, D., Junge, A., Wyder, S., Huerta-Cepas, J., et al. potential conflict of interest. (2019). STRING V11: Protein-Protein Association Networks with Increased Coverage, Supporting Functional Discovery in Genome-wide Experimental Publisher’s Note: All claims expressed in this article are solely those of the authors Datasets. Nucleic Acids Res. 47, D607–D613. doi:10.1093/nar/gky1131 and do not necessarily represent those of their affiliated organizations, or those of Tan, E. K., Kwok, H. K., Tan, L. C., Zhao, W. T., Prakash, K. M., Au, W. L., et al. the publisher, the editors and the reviewers. Any product that may be evaluated in (2010). Analysis of GWAS-Linked Loci in Parkinson Disease Reaffirms PARK16 as this article, or claim that may be made by its manufacturer, is not guaranteed or a Susceptibility Locus. Neurology 75, 508–512. doi:10.1212/wnl.0b013e3181eccfcd endorsed by the publisher. Wang, K. S., Mullersman, J. E., and Liu, X. F. (2010). Family-based Association Analysis of theMAPT Gene in Parkinson. J. Appl. Genet. 51, 509–514. doi:10. Copyright © 2022 Lobon, Solís-Moruno, Juan, Muhaisen, Abascal, Esteller-Cucala, 1007/bf03208881 García-Pérez, Martí, Tolosa, Ávila, Rahbari, Marques-Bonet, Casals and Soriano. Wang, Y., Bae, T., Thorpe, J., Sherman, M. A., Jones, A. G., Cho, S., et al. (2021). This is an open-access article distributed under the terms of the Creative Commons Comprehensive Identification of Somatic Nucleotide Variants in Human Brain Attribution License (CC BY). The use, distribution or reproduction in other forums is Tissue. Genome Biol. 22, 1–32. doi:10.1186/s13059-021-02285-3 permitted, provided the original author(s) and the copyright owner(s) are credited Watson, C. J., Papula, A. L., Poon, G. Y. P., Wong, W. H., Young, A. L., Druley, T. and that the original publication in this journal is cited, in accordance with accepted E., et al. (2020). The Evolutionary Dynamics and Fitness Landscape of Clonal academic practice. No use, distribution or reproduction is permitted which does not Hematopoiesis. Science 367, 1449–1454. doi:10.1126/science.aay9333 comply with these terms. Frontiers in Aging | www.frontiersin.org 14 April 2022 | Volume 3 | Article 851039