Full sequencing and haplotype analysis of MAPT in Parkinson disease and REM sleep behavior disorder

Jiao Li1,2, Jennifer A. Ruskey, MSc,2,3, Isabelle Arnulf, MD, PhD,4, Yves Dauvilliers, MD, PhD,5,

Michele T.M. Hu, MBBS, FRCP, PhD,6,7, Birgit Högl, MD,8, Claire S. Leblond, PhD,2,9, Sirui Zhou, PhD,2,3, Amirthagowri Ambalavanan, PhD,2,3,, Jay P. Ross, BSc,2,9, Cynthia V. Bourassa, MSc,2,3, Dan Spiegelman, MSc,2,3, Sandra B Laurent,2,3, Ambra Stefani, MD,8, Christelle Charley Monaca, MD, PhD,10, Valérie Cochen De Cock, MD, PhD,11,12, Michel Boivin, PhD,13,14, Luigi Ferini-Strambi, MD, PhD,15, Giuseppe Plazzi, MD, PhD,16,17, Elena Antelmi, MD, PhD,16,17, Peter Young, MD,18, Anna Heidbreder, MD,18, Catherine Labbe, PhD19, Tanis J. Ferman, PhD19, Patrick A. Dion, PhD,2,3, Dongsheng Fan, MD, PhD,1, Alex Desautels, MD, PhD,20,21, Jean-François Gagnon, PhD,20,22, Nicolas Dupré, MD, MSc,23,24, Edward A. Fon,2,3, Jacques Y. Montplaisir, MD, PhD,20,25, Bradley F. Boeve, MD,26, Ronald B. Postuma, MD, MSc,2,3,27, Guy A. Rouleau, MD, PhD,2,3,9 Owen A. Ross, PhD28,29 and Ziv Gan-Or, MD, PhD,2,3,9.

Affiliations :1Department of Neurology, Peking University Third Hospital, Beijing 100191, People R China 2Montreal Neurological Institute, McGill University, Montréal, QC, H3A 0G4, Canada, 3Department of Neurology and neurosurgery, McGill University, Montréal, QC, H3A 0G4, Canada, 4Sleep Disorders Unit, Pitié Salpêtrière Hospital, Centre de Recherche de l’Institut du Cerveau et de la Moelle Epinière and Sorbonne Universities, UPMC Paris 6 univ, Paris, 75013, France, 5Sleep Unit, National Reference Network for Narcolepsy, Department of Neurology Hôpital-Gui-de Chauliac, CHU Montpellier, INSERM U1061, Montpellier, 34000, France, 6Oxford Parkinson’s Disease Centre (OPDC), University of Oxford, Oxford, OX1 2JD, United Kingdom, 7Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, OX1 2JD, United Kingdom, 8Sleep Disorders Clinic, Department of Neurology, Medical University of Innsbruck, Innsbruck, 6020, Austria, 9Department of Human Genetics, McGill University, H3A 0G4, Montréal, QC, Canada, 10University Lille north of France, Department of clinical neurophysiology and sleep center, CHU Lille, Lille, 59000, France, 11Sleep and neurology unit, Beau Soleil Clinic, Montpellier, 34070, France, 12EuroMov, University of Montpellier, Montpellier, 34095, France, 13GRIP, École de psychologie, Université Laval, Québec city, QC, G1V 0A6, Canada, 14Institute of Genetic, Neurobiological and Social Foundations of Child Development, Tomsk State University, Tomsk, 634050, Russia, 15Department of Neurological Sciences, Università Vita-Salute San Raffaele, Milan, 20132, Italy, 16Department of Biomedical and Neuromotor Sciences (DIBINEM), Alma Mater Studiorum, University of Bologna, Bologna, 40126, Italy, 17IRCCS, Institute of Neurological Sciences of Bologna, Bologna, 40139, Italy, 18Department of Sleep Medicine and Neuromuscular Disorders, University of Muenster, 48149, Germany, 19Department of Psychiatry and Psychology, Mayo Clinic Jacksonville, FL, USA, 20Centre d’Études Avancées en Médecine du Sommeil, Hôpital du Sacré-Cœur de Montréal, Montréal, QC, H4J 1C5, Canada, 21Department of Neurosciences, Université de Montréal, Montréal, H3T 1J4, Canada, 22Département de psychologie, Université du Québec à Montréal, Montréal, QC, H3C 3P8, Canada, 23Division of Neurosciences, CHU de Québec, Université Laval, Quebec City, QC, Canada, 24Department of Medicine, Faculty of Medicine, Université Laval, Quebec City, QC, Canada, 25Department of Psychiatry, Université de Montréal, Montréal, QC, H3T 1J4, Canada, 26Department of Neurology, Mayo Clinic Rochester,

MN, USA, 27Department of Neurology, Montreal General Hospital, Montréal, QC, H3G 1A4, Canada, 28Department of Neuroscience, Mayo Clinic Jacksonville, FL, USA, 29Department of Clinical Genomics, Mayo Clinic Jacksonville, FL, USA,

Correspondence:Ziv Gan-OrMontreal Neurological Institute, McGill University1033 Pine Avenue, West,Ludmer Pavilion, room 327Montreal, QC, H3A 1A1Phone: +1-514-398-5845e-mail: ziv.gan-or@mcgill.ca

Word count: 1673 Running title: MAPT in PD and RBDKeywords: REM sleep behavior disorder, Parkinson disease, Genetics, MAPTConflict of interest disclosure: All authors report no conflict of interests.Funding sources: This work was financially supported by the Michael J. Fox Foundation and the Canadian Consortium on Neurodegeneration in Aging (CCNA).

Abstract

Background: MAPT haplotypes are associated with PD, but their association with REM-sleep

behavior disorder (RBD) is unclear.

Methods: Two cohorts were included: A) PD (n=600), RBD (n=613) patients and controls (n=981),

B) DLB patients with RBD (n=271) and controls (n=950). MAPT-associated variants and the entire

coding sequence of MAPT was analyzed. Age-, sex- and ethnicity-adjusted analyses were performed

to examine the association between MAPT, PD and RBD.

Results: MAPT-H2 variants were associated with PD (ORs 0.62-0.65, p=0.010-0.019), but not with

RBD. In PD, the H1 haplotype OR was 1.60 (95%CI 1.12-2.28, p=0.009), and the H2 OR was 0.68

(95%CI 0.48-0.96, p=0.03). The H2/H1 haplotypes were not associated with RBD.

Conclusions: Our results confirm the protective effect of the MAPT-H2 haplotype in PD, and

define its components. Furthermore, our results suggest that MAPT does not play a major role in

RBD, emphasizing different genetic background than in PD in this locus.

Introduction

Rapid eye movement (REM) sleep behavior disorder (RBD) is characterized by loss of muscle

atonia and enactment of dreams during REM sleep. RBD will progress, in most cases, to an overt

synucleinopathy; either Parkinson’s disease (PD), dementia with Lewy bodies (DLB) or, rarely,

multiple system atrophy (MSA).1 Multiple genetic variants have been implicated in PD2, 3 and some

in DLB and MSA,4 yet the potential role of most of them in RBD is still unknown.

Common genetic variation at the MAPT locus represent the second strongest genetic

association in recent genome-wide association studies (GWAS) in PD,2 and may also have a minor

role in DLB.5, 6 Recent studies in MSA also highlighted the MAPT H2 haplotype, although the small

sample size precluded a genome-wide significant p-value.7, 8 Studies in PD focusing on the H1 and

H2 MAPT haplotypes demonstrated that these haplotypes are associated with increased and

decreased risk for PD, respectively, 9-11 and GWAS confirmed that the MAPT H2 haplotype was

strongly associated with PD risk.2

MAPT sub-haplotype analysis and a sequencing study recently performed in DLB, suggested

that a rare sub-haplotype, H1G, and a rare coding variant, p.A152T, were associated with DLB.5, 6

However, a previous study in a large DLB cohort showed no evidence of association of the MAPT

locus with DLB.12 Overall, it seems that MAPT haplotypes play an important role in PD, while their

role in MSA and DLB is still not clear, and could be minor. Since RBD patients may convert to

either PD or DLB, two studies have been recently performed to determine the association between

MAPT and RBD.13, 14 Both studies suggested a possible association, however, both studies were

small and thus underpowered.

In the current study, we aimed to perform a thorough genetic study of the MAPT locus and its

association with RBD (idiopathic RBD and RBD in DLB) and PD, using targeted next generation

sequencing of the entire coding region of MAPT, and in-depth haplotype analysis, in larger cohorts

with RBD.

Methods

Study population

Two Independent cohorts were included (Supplementary Table 1), and more details on their

recruitment and diagnosis is in the Supplementary file: A) The Montreal Neurological Institute

(MNI) cohort included 600 PD patients, 613 RBD patients and 981 controls, all unrelated, of

European ancestry. B) The Mayo cohort included 271 patients with clinical DLB who were

diagnosed with RBD and 950 controls. All participants in both cohorts signed an informed consent

at enrollment to the study, and the respective ethical review boards approved the study protocols.

Genotyping

In cohort A, eight MAPT SNPs were genotyped (Table 1), including six MAPT locus haplotype-

tagging SNPs and two additional MAPT SNP, which were previously reported to be associated with

PD and RBD.13, 14 In cohort B, the six haplotype-tagging SNPs were previously genotyped and

reported. Further details on the genotyping are in the Supplementary file.

Targeted next-generation sequencing

A subset of cohort A, including samples from 525 PD patients, 342 RBD patients and 825 controls

was sequenced using targeted next generation sequencing (NGS). The coding sequences of 51 PD-

related genes (Supplementary Table 2), including MAPT, were captured using molecular inversion

probes (MIPs), as was previously described.15 Further details on alignment, filtering and analysis

are in the Supplementary file.

Statistical analysis

Full details on the statistical analysis can be found in the Supplementary file. To account for

potential bias that may have occurred in cohort A, due to the different populations in patients and

controls (despite all being of European origin), a principal component analysis (PCA) was

performed. Goodness-of-fit chi-square test with one degree of freedom was performed to examine

deviation from Hardy-Weinberg equilibrium (HWE) in the controls for each variant. Adjusted and

unadjusted regression models were performed to examine the association between the tested SNPs

and haplotypes in the MAPT locus and PD or RBD. Burden analysis was done as described in the

Supplementary file. All analyses were performed with PLINK 1.9 or R.

Results

Association of MAPT SNPs with PD and RBD

Data on population stratification and adjustment is in the supplementary file. Genotyping success

rate of the eight selected MAPT SNPs was 100 % in both cohorts, and all variants were in Hardy-

Weinberg equilibrium in the control group. An additional seven common variants with allele

frequencies > 0.05 were identified in the subset of samples from cohort A (525 PD patients, 342

RBD patients and 825 controls) that underwent targeted next generation sequencing of MAPT.

Logistic regression models, with and without adjustment for age, sex and population principal

components, were performed (Table 1).

Of the 15 common variants (Table 1), seven H2-haplotype variants were in almost full LD

(rs12185268, rs8070723, rs1800547, rs62063786, rs62063787, rs17651549, rs10445337), all

associated with a reduced risk for PD (ORs 0.62-0.65, p=0.010-0.019, age, sex and ethnicity

adjusted). The two other H2-haplotye SNPs that were previously reported to be associated with

RBD, rs12185268 and rs1800547,13, 14 were not associated with RBD in the current study, nor were

any of the other MAPT SNPs. In cohort B, rs7521 was nominally associated with DLB-RBD (OR

1.24, 95% CI 1.02-1.52, p=0.035, Supplementary Table 3), however, it was not associated with

RBD in cohort A. Burden analysis did not identify association between rare MAPT variants and PD

or RBD (see Supplementary file).

Analysis of MAPT haplotypes and association with risk for PD and RBD

The H1 haplotype was associated with an increased risk for PD (OR 1.60, 95% CI 1.12-2.28,

p=0.009), and the H2 haplotype was associated with a reduced risk for PD (OR 0.68, 95% CI 0.48-

0.96, p=0.03). However, the H1 and H2 haplotype were not associated with RBD both in cohort A

of idiopathic RBD patients and in cohort B with DLB-RBD. With the data from the targeted NGS,

we demonstrate that the H2 haplotype includes eight coding variants (Supplementary Figure 2)

which are all in full or almost full LD: p.P202L (rs63750417), p.D285N (rs62063786), p.R370W

(rs17651549), p.S447P (rs10445337), p.P493P (rs1052551), p.T540T (rs62063845), p.A562A

(rs1052553), p.N590N (rs17652121).

Subsequently, we analyzed the H1 sub-haplotypes to determine if any of them is associated

with PD or RBD (Table 2). A total of 20 H1 sub-haplotypes with frequency > 0.01 were identified

in cohort A, and 16 H1 sub-haplotype with frequency >0.01 were identified in cohort B. Two sub-

haplotypes, H1J and H1Z, were nominally associated with PD in the un-adjusted analysis, but lost

significance when adjusted for age, sex and principal components. Similarly, the H1H and H1O

were nominally associated with RBD in cohort A in the unadjusted model (data not shown), but lost

significance after the adjustment for age, sex and principal components. In cohort B, the previously

reported DLB-associated haplotype H1G was associated with DLB-RBD (OR 2.85, 95% CI 1.30-

6.27, p=0.009, Supplementary Table 4). However, it was also associated with DLB without RBD

(from cohort B, data not shown) and in cohort A, this haplotype had identical frequencies of 0.011

in RBD patients and controls (p=0.78, Table 2), suggesting that the association is driven by DLB

and not by RBD.

Discussion

The current study increases our understanding of the role of MAPT haplotypes in PD and RBD by:

a) We confirm the association between MAPT H1/H2 haplotypes in PD, with an increased risk for

H1 and decreased risk for H2. b) MAPT is not associated with RBD, which provides further

evidence for RBD having a different genetic background than PD, at least partially. c) Our targeted

NGS analysis defines the H2 haplotype, including several coding MAPT variants (nonsynonymous

and synonymous, Supplementary Figure 2), although probably not expressed in the CNS.16

The lack of association of MAPT with RBD suggests that RBD has a genetic background that

does not fully overlap with that of PD. While GBA mutations are associated with both PD and

RBD,17-19 previous studies had demonstrated that LRRK2 pathogenic mutations that may cause PD

are not associated with RBD,20, 21 and that PD patients with LRRK2 mutations had little or no

RBD.22, 23 In addition, the APOE 4 haplotype, which is strongly associated with Alzheimer’s

disease and DLB, was also not associated with RBD.24 Therefore, RBD may represent a subtype of

synucleinopathy with its own genetic background, but this hypothesis should be tested in larger

cohorts by using whole-genome methods. Whether these genetic associations and lack thereof

suggest lack of role for tauopathy in RBD, should be examined by pathological studies of PD and

DLB patients with and without RBD, as well as in idiopathic RBD.

The association of the H1/H2 haplotypes with PD is well defined across various populations,2,

10, 11. Since this haplotype includes multiple, potentially regulatory non-coding variants, as well as

coding synonymous and nonsynonymous variants, these data alone do not allow us to identify the

specific variant that affects the risk within this haplotype. Exons 4a and 6, in which the

nonsynonymous variants are found, are thought to not be expressed in the CNS.16 However, since

the PNS is involved early at the course of PD, and may even be where PD begins, we cannot rule

out a role for these variants in PD.25

The current study has several limitations. The differences in age and sex of patients and

controls can potentially bias results of age-related effects of genetic factors. However, we accounted

for this potential bias by adjusting for both in the regression models. Similarly, there could be an

effect of the different European populations, therefore we further adjusted for the two top principal

components of ethnicity as detected by PCA. Another potential limitation is the study size, despite

being the largest genetic study on RBD, to the best of our knowledge. It is still possible that an

association could only be observed in a larger cohort of patients and controls. Therefore,

subsequent, larger studies are needed to further examine this association.

Overall, our data confirm the association of MAPT with PD, and suggest that MAPT genetic

variants have minor or no role in RBD. Further studies are needed to replicate these results, and to

examine the specific effects of variants within the H2 haplotype, but also other haplotypes, on PD

risk and mechanism.

Acknowledgements

We thank the patients and controls for their participation in the study. This work was financially

supported by the Michael J. Fox Foundation (MJFF) and by the Canadian Consortium on

Neurodegeneration in Aging (CCNA). This study was also funded by the Monument Trust

Discovery Award from Parkinson’s UK and supported by the National Institute for Health Research

(NIHR) Oxford Biomedical Research Centre based at Oxford University Hospitals NHS Trust and

University of Oxford, and the Dementias and Neurodegenerative Diseases Research Network

(DeNDRoN). GAR holds a Canada Research Chair in Genetics of the Nervous System and the

Wilder Penfield Chair in Neurosciences. JFG holds a Canada Research Chair in Cognitive Decline

in Pathological Aging. JL is supported by the China Scholarship Council. We thank Daniel

Rochefort, Pascale Hince, Helene Catoire, Pierre Provencher, Cathy Mirarchi and Vessela Zaharieva

for their assistance. We thank the Quebec Parkinson’s Network and its members (http://rpq-qpn.ca/)

for their collaboration. We thank all the members of the International RBD Genetics Consortium

(IRBDGC).

Author Roles1. Research project: A. Conception – JL, ZGO B. Organization – JL, JAR, IA, YD, MTMH, BH,

CSL, AS, CCM, VCDC, MB, LFS, GP, EA, PY, AH, CL, TJP, PAD, AD, JFG, ND, EAF, JYM, BB, RBP, GAR, OAR, ZGO. C. Execution - JL, JAR, SZ, AA, JPR, CVB, DS, SBL, CL.

2. Statistical Analysis: A. Design – JL, SZ, AA, ZGO B. Execution – JL, ZGO C. Review and Critique - IA, YD, MTMH, BH, SZ, AA, JPR, AS, CCM, VCDC, MB, LFS, GP, EA, PY, AH, CL, TJF, PAD, DF, AD, JFG, ND, EAF, JYM, BB, RBP, GAR, OAR.

3. Manuscript Preparation: A. Writing of the first draft – JL, ZGO B. Review and Critique – JL, JAR, IA, YD, MTMH, BH, CSL, SZ, AA, JPR, CVB, DS, SBL, AS, CCM, VCDC, MB, LFS, GP, EA, PY, AH, CL, TJF, PAD, DF, AD, JFG, ND, EAF, JYM, BB, RBP, GAR, OAR, ZGO.

Full financial disclosure for the previous 12 months:JL - nothing to discloseJAR - nothing to discloseIA – Received research funding from FUI (French Investissement Bank)YD - received funds for seminars, board engagements and travel to conferences by UCB Pharma, Jazz, Theranexus, Flamel and Bioprojet.MTMH –received a Parkinson's UK innovation grant “Biomarkers for Dementia and Cognitive decline in Parkinson's disease,” Parkinson's UK Monument Discovery Award “Targeting the earliest pathways to Parkinson's disease,” EU small‐ or medium‐scale focused research project (STREP) Award “Virtual, physiological and computational neuromuscular models for the predictive treatment of Parkinson's Disease‐NoTremor,” and MJFF Biomarker Grant “Development of potential diagnostic biomarkers in Parkinson's disease.”BH – Received funding from the Government of Tyrol translational research funds and from the Austrian Science funds FWF, received honoraria as a consultant from Axovant Mundipahrma and Benevolent Bio, and speaker Honoraria from Otsuka, Janssen Cilag, Lilly, UCB, Abbvie and Nutricia.CSL – nothing to discloseSZ - nothing to disclose AA - nothing to disclose JPR - nothing to disclose CVB - nothing to disclose DS - nothing to disclose SBL - nothing to disclose AS – has received support for research from AxovantCCM – is on the advisory board and received travel and consultancy fees from UCB Pharma.VCDC – has received travel grants from Eole, Orkyn, SOS oxygeneMB – nothing to discloseLFS - nothing to discloseGP – Served on the advisory boards of UCB pharma, Jazz pharmaceuticals and Bioproject.EA – has received a grant from the European Academy of Neurology PY – has received speaker honoraria from Loewenstein Medical GmbH & Co.KG, Neuro Consil, Düsseldorf, UCB Pharma, Genzyme GmbH, Bayer Vital. Served on the advisory boards of Vanda Pharmaceuticals Germany, Genzyme Europe.AH- Has received speaker honoraria from UCB Pharma, Vanda, received travel support from Bioproject. Has served on the advisory board of UCB Pharma CL – nothing to discloseTJF – nothing to disclosePAD - Supported by research grants from CIHR, CCNA, the International Essential Tremor Foundation and ALS Canada-Brain Canada.DF – nothing to discloseAD – nothing to discloseJFG – Supported by research grants from CIHR and Parkinson’s Society Canada.ND - Supported by research grants from Fondation du CHUQ - Kilimandjaro à Québec, CIHR, ALS

Canada & Brain Canada, CQDM/ Brain Canada, FRQS and the W. Garfield Weston Foundation.EAF - Supported by research grants from Canadian Institutes of Health Research (CIHR), Michael J Fox Foundation, ALS Canada-Brain Canada Hudson Team grant, CQDM/ Brain Canada - Focus on Brain Program, Brain Canada, Fonds de Recherche du Québec (FRQS), Canadian Consortium on Neurodegeneration in Aging (CCNA), and National Parkinson Foundation (NPF).JYM - has served on the advisory boards for Sanofi-Aventis, Servier, Merck, Jazz Pharmaceutical, Valeant Pharmaceutical, and Impax Laboratory; has received honoraria from Valeant and Otsuka Pharmaceutical; has received grant support from GlaxoSmithKline and MerckBFB – has received funding from the National Institute on Aging, the National Institute of Neurological Disorders and Stroke, Roivant Sciences, C2N Diagnostics, and Axovant Sciences.RBP – has received personal compensation for travel, speaker fees, and consultation from Boehringer-Ingelheim, GE Health Care, Novartis, Roche, Theranexus and Teva Neurosciences, and is funded by grants from the Fonds de Recherche du Quebec - Sante, the Michael J. Fox Foundation, the W. Garfield Weston Foundation, and the Canadian Institutes of Health Research.GAR – Supported by research grants from CIHR, CCNA, the International Essential Tremor Foundation and ALS Canada-Brain Canada.OAR – has received research support from the National Institutes of Health (P50-NS072187; R01-NS078086), Michael J Fox Foundation and the Department of Defense (W81XWH-17-1-0249).ZGO – has received consultation fees from Genzyme and from Lysosomal Therapeutics Inc. Supported by grants from the Michael J Fox Foundation.

References

1. Postuma RB, Iranzo A, Hogl B, et al. Risk factors for neurodegeneration in idiopathic rapid eye

movement sleep behavior disorder: a multicenter study. Ann Neurol 2015;77(5):830-839.

2. Chang D, Nalls MA, Hallgrimsdottir IB, et al. A meta-analysis of genome-wide association

studies identifies 17 new Parkinson's disease risk loci. Nat Genet 2017;49(10):1511-1516.

3. Gan-Or Z, Dion PA, Rouleau GA. Genetic perspective on the role of the autophagy-lysosome

pathway in Parkinson disease. Autophagy 2015;11(9):1443-1457.

4. Nussbaum RL. Genetics of Synucleinopathies. Cold Spring Harb Perspect Med 2017.

5. Labbe C, Heckman MG, Lorenzo-Betancor O, et al. MAPT haplotype H1G is associated with

increased risk of dementia with Lewy bodies. Alzheimers Dement 2016;12(12):1297-1304.

6. Labbe C, Ogaki K, Lorenzo-Betancor O, et al. Role for the microtubule-associated protein tau

variant p.A152T in risk of alpha-synucleinopathies. Neurology 2015;85(19):1680-1686.

7. Labbe C, Heckman MG, Lorenzo-Betancor O, et al. MAPT haplotype diversity in multiple

system atrophy. Parkinsonism Relat Disord 2016;30:40-45.

8. Sailer A, Scholz SW, Nalls MA, et al. A genome-wide association study in multiple system

atrophy. Neurology 2016;87(15):1591-1598.

9. Lill CM, Roehr JT, McQueen MB, et al. Comprehensive research synopsis and systematic

meta-analyses in Parkinson's disease genetics: The PDGene database. PLoS Genet

2012;8(3):e1002548.

10. Wider C, Vilarino-Guell C, Jasinska-Myga B, et al. Association of the MAPT locus with

Parkinson's disease. Eur J Neurol 2010;17(3):483-486.

11. Zabetian CP, Hutter CM, Factor SA, et al. Association analysis of MAPT H1 haplotype and

subhaplotypes in Parkinson's disease. Ann Neurol 2007;62(2):137-144.

12. Bras J, Guerreiro R, Darwent L, et al. Genetic analysis implicates APOE, SNCA and suggests

lysosomal dysfunction in the etiology of dementia with Lewy bodies. Hum Mol Genet

2014;23(23):6139-6146.

13. Fernandez-Santiago R, Iranzo A, Gaig C, et al. MAPT association with REM sleep behavior

disorder. Neurol Genet 2017;3(1):e131.

14. Gan-Or Z, Girard SL, Noreau A, et al. Parkinson's Disease Genetic Loci in Rapid Eye

Movement Sleep Behavior Disorder. J Mol Neurosci 2015;56(3):617-622.

15. Ross JP, Dupre N, Dauvilliers Y, et al. Analysis of DNAJC13 mutations in

French-Canadian/French cohort of Parkinson's disease. Neurobiol Aging 2016;45:212 e213-212

e217.

16. Caillet-Boudin ML, Buee L, Sergeant N, Lefebvre B. Regulation of human MAPT gene

expression. Mol Neurodegener 2015;10:28.

17. Gan-Or Z, Amshalom I, Kilarski LL, et al. Differential effects of severe vs mild GBA mutations

on Parkinson disease. Neurology 2015;84(9):880-887.

18. Gan-Or Z, Mirelman A, Postuma RB, et al. GBA mutations are associated with Rapid Eye

Movement Sleep Behavior Disorder. Ann Clin Transl Neurol 2015;2(9):941-945.

19. Sidransky E, Nalls MA, Aasly JO, et al. Multicenter analysis of glucocerebrosidase mutations

in Parkinson's disease. N Engl J Med 2009;361(17):1651-1661.

20. Barber TR, Lawton M, Rolinski M, et al. Prodromal Parkinsonism and Neurodegenerative Risk

Stratification in REM Sleep Behavior Disorder. Sleep 2017;40(8).

21. Fernandez-Santiago R, Iranzo A, Gaig C, et al. Absence of LRRK2 mutations in a cohort of

patients with idiopathic REM sleep behavior disorder. Neurology 2016;86(11):1072-1073.

22. Ehrminger M, Leu-Semenescu S, Cormier F, et al. Sleep aspects on video-polysomnography in

LRRK2 mutation carriers. Mov Disord 2015;30(13):1839-1843.

23. Pont-Sunyer C, Iranzo A, Gaig C, et al. Sleep Disorders in Parkinsonian and Nonparkinsonian

LRRK2 Mutation Carriers. PLoS One 2015;10(7):e0132368.

24. Gan-Or Z, Montplaisir JY, Ross JP, et al. The dementia-associated APOE epsilon4 allele is not

associated with rapid eye movement sleep behavior disorder. Neurobiol Aging 2017;49:218

e213-218 e215.

25. Poewe W, Seppi K, Tanner CM, et al. Parkinson disease. Nat Rev Dis Primers 2017;3:17013.

Figure 1

Table 1.Allele frequencies of MAPT variants genotyped with TaqMan assays or identified in targeted next generation sequencing.

Minor allele frequency PD vs. Control RBD vs. ControlPD

(n=600)RBD

(n=613)Controls (n=981)

OR (95% CI)a pa OR (95% CI)a pa

SNPs genotyped using TaqMan assays

rs12185268 0.213 0.237 0.252 0.64 (0.45-0.92) 0.015 0.90 (0.60-1.35)

0.603

rs1467967 0.347 0.300 0.328 1.15 (0.90-1.46) 0.2670.96 (0.72-

1.28)0.771

rs242557 0.336 0.345 0.335 1.10 (0.86-1.40) 0.4551.01 (0.76-

1.34)0.936

rs1800547 0.213 0.237 0.249 0.65 (0.46-0.93) 0.018 0.97 (0.66-1.43)

0.883

rs3785883 0.158 0.179 0.155 0.93 (0.69-1.26) 0.6270.93 (0.66-

1.32)0.692

rs2471738 0.222 0.194 0.203 1.31 (1.00-1.73) 0.0531.21 (0.87-

1.67)0.262

rs8070723 0.219 0.240 0.253 0.65 (0.46-0.93) 0.018 0.86 (0.57-1.30)

0.471

rs7521 0.468 0.466 0.454 0.95 (0.75-1.20) 0.6660.93 (0.70-

1.23)0.595

Variants identified in targeted NGS

rs63750417 0.207 0.232 0.2440.72 (0.44-1.16) 0.176

1.14 (0.69-1.88)

0.605

rs63750072 0.053 0.058 0.0530.96 (0.44-2.06) 0.910

1.23 (0.61-2.44)

0.566

rs62063786 0.216 0.232 0.247 0.63 (0.43-0.90) 0.0120.90 (0.59-

1.36)0.613

rs62063787 0.216 0.232 0.246 0.64 (0.44-0.91) 0.015 0.91 (0.60-1.38)

0.666

rs17651549 0.216 0.232 0.248 0.65 (0.45-0.93) 0.019 0.94 (0.62-1.42)

0.757

rs2258689 0.184 0.207 0.200 0.91 (0.68-1.21) 0.5130.97 (0.69-

1.35)0.852

rs10445337 0.216 0.232 0.247 0.62 (0.43-0.89) 0.010 0.90 (0.59-1.35)

0.596

SNPs, single nucleotide polymorphisms; PD, Parkinson’s disease; RBD, rapid eye movement sleep behavior disorder; OR, odds ratio; CI, confidence interval.a Adjusted for age, sex, and the two major components in the population stratification principal component analysis.

Table 2.Haplotype analysis of MAPT H1 sub-haplotypes in PD and RBD patients.

Allele frequency PD vs. Control RBD vs. ControlHaplotype Haplotype

structure aPD RBD controls OR (95% CI) b p b OR (95% CI) b p b

H2 GAGGGCGG 0.221 0.246 0.259 0.68 (0.47-0.96) 0.030 0.94 (0.63-1.41) 0.768H1B AGGAGCAA 0.157 0.133 0.140 1.29 (0.93-1.78) 0.123 1.06 (0.73-1.55) 0.763H1C AAAAGTAG 0.084 0.082 0.084 1.31 (0.88-1.95) 0.184 1.23 (0.76-1.99) 0.401H1D AAAAGCAA 0.056 0.067 0.065 0.85 (0.50-1.45) 0.553 0.98 (0.55-1.73) 0.931H1E AAGAGCAA 0.123 0.110 0.111 0.89 (0.60-1.33) 0.566 0.91 (0.56-1.49) 0.713H1F AGGAACAA 0.016 0.014 0.019 0.50 (0.17-1.49) 0.213 0.56 (0.16-1.89) 0.347H1G AGAAACAA 0.013 0.011 0.011 2.85 (0.84-9.72) 0.094 0.76 (0.11-5.12) 0.779H1H AAGAACAA 0.055 0.066 0.048 0.88 (0.52-1.50) 0.649 0.96 (0.52-1.79) 0.905H1I AGAAGCAA 0.030 0.036 0.039 0.66 (0.33-1.32) 0.238 0.65 (0.28-1.49) 0.309H1J AAGAGCAG 0.027 0.026 0.017 2.27 (0.87-5.91) 0.093 2.16 (0.65-7.20) 0.210H1L AAGAACAG 0.011 0.014 0.013 0.59 (0.21-1.68) 0.321 0.43 (0.11-1.68) 0.223H1M AGAAGCAG 0.021 0.023 0.028 0.48 (0.19-1.23) 0.125 0.35 (0.11-1.12) 0.077H1O AAAAACAA 0.010 0.021 0.013 0.71 (0.13-3.81) 0.685 1.56 (0.38-6.36) 0.535H1P AGGAGTAG 0.018 0.015 0.015 2.10 (0.69-6.39) 0.191 2.54 (0.71-9.04) 0.150H1Q AAAAGTAA 0.015 0.017 0.016 0.63 (0.21-1.87) 0.401 0.82 (0.29-2.28) 0.699H1U AAAAGCAG 0.040 0.040 0.035 1.70 (0.89-3.26) 0.111 1.48 (0.66-3.31) 0.341H1V AGGAATAG 0.014 0.011 0.016 1.38 (0.50-3.78) 0.531 0.88 (0.22-3.57) 0.857H1W AGGAGCAG 0.018 0.017 0.019 0.80 (0.27-2.37) 0.688 1.11 (0.32-3.83) 0.869H1X AGAAATAG 0.017 0.010 0.014 0.74 (0.23-2.34) 0.602 0.78 (0.20-3.12) 0.730H1y AAAAATAG 0.011 0.015 0.011 1.45 (0.40-5.23) 0.570 1.15 (0.28-4.71) 0.846H1Z AGAAGTAG 0.045 0.028 0.030 1.72 (0.85-3.48) 0.132 0.94 (0.39-2.31) 0.899

PD, Parkinson’s disease; RBD, rapid eye movement sleep behavior disorder; OR, odds ratio; CI, confidence interval.a Eight SNPs defining the haplotypes are given in the 5’ to 3’ order as follows: rs12185628, rs1467967, rs242557, rs1800547, rs3785883,

rs2471738, rs8070723, rs7521.b Adjusted for age, sex, and the two major components in the population stratification principal component analysis.

Supplementary material

Full sequencing and haplotype analysis of MAPT in Parkinson disease and REM sleep behavior disorder

Jiao Li1,2, Jennifer A. Ruskey, MSc,2,3, Isabelle Arnulf, MD, PhD,4, Yves Dauvilliers, MD, PhD,5,

Michele T.M. Hu, MBBS, FRCP, PhD,6,7, Birgit Högl, MD,8, Claire S. Leblond, PhD,2,9, Sirui Zhou, PhD,2,3, Amirthagowri Ambalavanan, PhD,2,3,, Jay P. Ross, BSc,2,9, Cynthia V. Bourassa, MSc,2,3, Dan Spiegelman, MSc,2,3, Sandra B Laurent,2,3, Ambra Stefani, MD,8, Christelle Charley Monaca, MD, PhD,10, Valérie Cochen De Cock, MD, PhD,11,12, Michel Boivin, PhD,13,14, Luigi Ferini-Strambi, MD, PhD,15, Giuseppe Plazzi, MD, PhD,16,17, Elena Antelmi, MD, PhD,16,17, Peter Young, MD,18, Anna Heidbreder, MD,18, Catherine Labbe, PhD19, Tanis J. Ferman, PhD19, Patrick A. Dion, PhD,2,3, Dongsheng Fan, MD, PhD,1, Alex Desautels, MD, PhD,20,21, Jean-François Gagnon, PhD,20,22, Nicolas Dupré, MD, MSc,23,24, Edward A. Fon,2,3, Jacques Y. Montplaisir, MD, PhD,20,25, Bradley F. Boeve, MD,26, Ronald B. Postuma, MD, MSc,2,3,27, Guy A. Rouleau, MD, PhD,2,3,9 Owen A. Ross, PhD28,29 and Ziv Gan-Or, MD, PhD,2,3,9.

Supplementary full methods

Study population

Two Independent cohorts were included in the current study (Supplementary Table 1):

A) The Montreal Neurological Institute (MNI) cohort included 600 PD patients, 613 RBD

patients (261 of which were included in a previous study that included analysis of one MAPT

SNP1) and 981 controls, all unrelated, of European ancestry. PD patients and controls were

recruited from clinics across Québec, Canada, including the Quebec Parkinson’s Network

(http://rpq-qpn.ca/) and from Montpellier, France, and were mainly of French-Canadian and

French origin. PD patients were diagnosed by movement disorders specialists from the

participating clinics, based on the UK Parkinson’s Disease Brain Bank Criteria (without

exclusion of patients with familial history of PD). RBD patients were recruited by collaborators

from the International RBD Study Group from Europe and Canada, as part of the International

RBD Genetics Consortium (IRBDGC). RBD diagnosis was confirmed with polysomnography,

according to the International Classification of Sleep Disorders, version 2 (ICSD-2) criteria2. Of

the RBD patients, 342 were of French-Canadian or French origin.

B) The Mayo cohort series consists of 271 patients with clinical DLB who were diagnosed with

RBD (termed hereafter DLB-RBD) confirmed with polysomnography according to the ICSD-2

criteria or diagnosed with probable RBD using the Mayo Sleep Questionnaire (which was

demonstrated to have sensitivity of 98% and specificity of 74% in a cohort of DLB patients 3),

and 950 controls. Further details on this cohort and their recruitment were previously described.4

All participants in both cohorts signed an informed consent at enrollment to the study, and the

respective ethical review boards approved the study protocols.

Genotyping

In cohort A, a total of eight SNPs from the MAPT locus were genotyped, including six MAPT

locus haplotype-tagging SNPs (rs1467967, rs242557, rs3785883, rs2471738, rs8070723, rs7521)

and two additional MAPT SNP (rs12185268, rs1800547), which were previously reported to be

associated with PD and RBD.1, 5 Genotyping was performed using TaqMan SNP genotyping

assays on The Applied Biosystems™ QuantStudio™ 7 Flex Real-Time PCR System (Applied

Biosystems, Foster City, CA) according to the manufacturer's instructions. In cohort B, the six

haplotype-tagging SNPs were previously genotyped and reported, without distinction between

presence or absence of RBD in DLB. Genotyping was performed as previously described.4

Targeted next-generation sequencing

A subset of cohort A, including samples from 525 PD patients, 342 RBD patients and 825

controls was sequenced using targeted next generation sequencing (NGS). The coding sequences

of 51 PD-related genes (Supplementary Table 2), including MAPT, were captured using

molecular inversion probes (MIPs), as was previously described.6 Targeted NGS was performed

using the Illumina HiSeq 2500 platform (Innovation Genome Center, McGill University, and

Genome Quebec). Sequencing data were aligned against the human genome (GRCh37 assembly)

using Burrows-Wheeler Aligner,7 and the variant calling and annotation were done using

Genome Analysis ToolKit (GATK)8 and ANNOVAR9 tools, respectively. For the current study,

data on MAPT was extracted. However, to control for possible population stratification bias, data

on all common variants in these genes was used (see section on principal component analysis). In

the variant analysis, variants with sequencing depth of < 20X or with genotype quality < 60 were

excluded.

Principal component analysis

To account for potential bias that may have occurred in cohort A, due to the different populations

in patients and controls (despite all being of European origin), a principal component analysis

(PCA) was performed. Data on 164 common SNPs from the 51 PD-related genes was extracted

for all 1692 study participants that underwent targeted NGS. These data were compared with the

same variants in the 1000 Genome Project (www.internationalgenome.org) in the CHB (n=103),

CEU (n=99), JPT (n=104) and YRI (n=108) samples (Supplementary Figure 1). Ancestry and

population stratification were assessed using the “smartpca” module implemented in the

EIGENSOFT package10 in R, and the principal components of ancestry were determined.

Statistical analysis

Association of MAPT SNPs with PD and RBD

Goodness-of-fit chi-square test with one degree of freedom was performed to examine deviation

from HWE in the controls for each variant. To examine the association between the tested SNPs

in the MAPT locus and PD or RBD, a binary logistic regression model was performed. To

account for the possible effects of age, sex and ethnicity potential differences, the regression

model was performed with and without age, sex and the 2 first major principal components of

ancestry as covariates. In the combined analysis of RBD patients from cohort A and DLB-RBD

patients from cohort B, the regression model was adjusted for age, sex, and center. This analysis

was performed separately for the PD and RBD cohorts, and all analyses were performed using

PLINK 1.9.11

Association of MAPT haplotypes with PD and BRD risk

Haplotypes were determined, and assignment of the H1/H2 haplotypes and H1 sub-haplotypes

was done as was previously described. Haplotypes with frequency of <0.01 were excluded from

the analysis. Logistic regression over the haplotypes was performed to test for association with

PD and RBD, with and without age, sex and the 2 principal ancestry components as covariates.

This analysis was done separately for PD and RBD from cohort A, and for the RBD vs. controls

from cohort B. In the combined haplotype analysis of RBD patients from cohort A and DLB-

RBD patients from cohort B, the regression model was adjusted for age, sex, and center. All

analyses were performed with PLINK.

Burden analysis of functional MAPT variants in PD and BRD

To examine whether there is a burden effect of coding variants that affect the protein sequence,

an optimized Sequence Kernel Association Test (SKAT-O)12 was performed using R. First, we

examined the burden of all non-synonymous variants, followed by an analysis of all variants

predicted to be deleterious by the SIFT (Sorting Intolerant From Tolerant) algorithm.13

Supplementary Table 1.

Study population.  Cohort A Cohort B Population PD RBD Controls DLB-RBD ControlsN 600 613 981 271 950NGS, N 525 342 825 - -Male:Female Ratioa 1.81 3.92 1.11 4.77 0.74Age (Mean±SD)b 65.4±9.8 67.4±8.6 44.1±14.6 63.6±13.3 65.1±12.7Age (Range)c 30.0-91.0 32.0-87.0 19.0-83.0 15.0-91.0 18.0-92.0N, number; NGS, next generation sequencing – number of samples that went through targeted NGS; SD, standard deviation; PD, Parkinson’s disease; RBD, rapid eye movement sleep behavior disorder; DLB-RBD, patients with dementia with Lewy bodies (DLB) and RBD.a Data on sex were not available for 8 PD, 42 RBD and 84 control individualsb Data on age were not available for 16 PD, 17 RBD and 46 control individuals c Data on DLB-RBD age was not available for 13 patients.

Supplementary Table 2. Fifty one genes that were fully sequenced using molecular inversion probes. Common variants in these genes were used to demonstrate ancestry by performance of principal component analysis (see supplementary Figure 1)

ACMSD FGF20

LAMP3

PSAP SREBF1

ATP13A2

GAK LRRK2 RAB25 STK39

BST1 GBA MAPT RAB7L1

STX1B

C18orf8 GCH1 MCCC1

RIT2 SYT11

CCDC62

GIGYF2

NPC1 SCARB2

TMEM163

DDRGK1

GPNMB

PARK2 SETD1A

TMEM175

DGKQ HIP1R PARK7 SIPA1L2

UCHL1

DNAJC13

HTRA2

PINK1 SLC41A1

USP25

EIF4G1 INPP5F

PLA2G6

SMPD1 VPS13C

FAM47E

ITGA8 PM20D1

SNCA VPS35

FBXO7

Supplementary Table 3.Frequencies of MAPT SNPs in patients with DLB and RBD vs. controls, and a combined analysis with RBD patients and controls from cohort A.

Minor allele frequency DLB-RBD vs. controls RBD+DLB-RBD vs. combined controls

DLB-RBD Controls RBD + DLB-RBD Combined controls OR (95% CI) a p a OR (95% CI) b p b

n=271 n=950 n=884 n=1931rs1467967 0.323 0.319 0.307 0.324 1.12 (0.89-1.39) 0.331 0.98 (0.84-1.13) 0.746rs242557 0.393 0.350 0.360 0.343 1.14 (0.93-1.40) 0.196 1.06 (0.92-1.23) 0.387rs3785883 0.181 0.171 0.179 0.163 0.99 (0.75-1.29) 0.928 1.03 (0.86-1.23) 0.762rs2471738 0.214 0.203 0.200 0.203 1.02 (0.79-1.31) 0.881 1.03 (0.87-1.23) 0.697rs8070723 0.214 0.249 0.232 0.251 0.79 (0.62-1.02) 0.068 0.89 (0.76-1.05) 0.171rs7521 0.493 0.456 0.474 0.455 1.24 (1.02-1.52) 0.035 1.09 (0.95-1.25) 0.210

DLB-RBD, patients with dementia with Lewy bodies (DLB) and rapid eye movement sleep behavior disorder (RBD); OR, odds ratio; CI, confidence interval.a Adjusted for age and sex.b Adjusted for age, sex and center.

Supplementary Table 4.Haplotype analysis of MAPT H1 sub-haplotypes in DLB-RBD patients and controls, and combined analysis with RBD patients and controls from cohort A.

Allele frequency DLB-RBD vs. controls RBD+DLB-RBD vs. combined controls

Haplotype Haplotype structure a

DLB-RBD Controls RBD + DLB-RBD Combined controls OR (95% CI) b p b OR (95% CI) c p c

n=271 n=950 n=884 n=1931H2 AGGCGG 0.221 0.263 0.241 0.260 0.78 (0.60-1.00) 0.050 0.89 (0.75-1.04) 0.147H1B GGGCAA 0.187 0.179 0.173 0.175 1.25 (0.93-1.67) 0.134 1.08 (0.89-1.31) 0.438H1C AAGTAG 0.131 0.120 0.119 0.117 1.15 (0.82-1.62) 0.422 1.07 (0.84-1.36) 0.585H1D AAGCAA 0.092 0.070 0.078 0.069 1.35 (0.89-2.04) 0.155 1.11 (0.83-1.48) 0.479H1E AGGCAA 0.087 0.095 0.094 0.094 0.94 (0.63-1.40) 0.754 0.95 (0.73-1.24) 0.711H1F GGACAA . . 0.009 0.012 . . 0.49 (0.18-1.30) 0.153H1G GAACAA 0.031 0.015 0.016 0.012 2.85 (1.30-6.27) 0.009 2.22 (1.11-4.43) 0.024H1H AGACAA 0.046 0.046 0.051 0.042 1.04 (0.63-1.72) 0.879 1.16 (0.80-1.69) 0.439H1I GAGCAA 0.035 0.037 0.040 0.041 1.24 (0.61-2.54) 0.557 0.95 (0.62-1.44) 0.799H1J AGGCAG 0.016 0.011 0.014 0.009 1.70 (0.62-4.67) 0.302 1.61 (0.77-3.35) 0.204H1L AGACAG 0.024 0.026 0.031 0.027 0.77 (0.38-1.58) 0.473 0.95 (0.61-1.49) 0.827H1M GAGCAG 0.015 0.029 0.021 0.029 0.41 (0.16-1.05) 0.062 0.53 (0.31-0.90) 0.019H1O AAACAA 0.023 0.021 0.023 0.017 1.11 (0.44-2.77) 0.823 1.41 (0.75-2.66) 0.287H1P GGGTAG 0.011 0.017 0.010 0.014 0.56 (0.17-1.89) 0.350 0.94 (0.41-2.13) 0.882H1R AGGTAG 0.016 0.013 0.017 0.016 1.49 (0.53-4.18) 0.449 1.51 (0.80-2.86) 0.206H1U AAGCAG 0.027 0.027 0.025 0.024 0.79 (0.36-1.72) 0.553 1.01 (0.70-1.46) 0.958H1V GGATAG . . 0.010 0.014 . . 0.55 (0.26-1.18) 0.126H1x GAATAG 0.018 0.015 0.012 0.013 1.03 (0.39-2.76) 0.953 0.88 (0.45-1.75) 0.722H1y AAATAG 0.018 0.016 0.016 0.014 1.09 (0.39-3.04) 0.869 1.08 (0.52-2.22) 0.835

DLB-RBD, patients with dementia with Lewy bodies (DLB) and rapid eye movement sleep behavior disorder (RBD); OR, odds ratio; CI, confidence interval.a Six SNPs defining the haplotypes are given in the 5’ to 3’ order as follows: rs1467967, rs242557, rs3785883, rs2471738, rs8070723, rs7521.b Adjusted for age and sex.c Adjusted for age, sex and center.

Supplementary Figure 1.

To examine our population structure, we extracted on common variants found in the sequencing of the 51 genes detailed in Supplementary Table 2. These variants were used against data from the 1000 genome project to examine the origin of our population. As clearly evident, our PD (pink dots), RBD (orange dots) and control (grey dots) populations segregated with the European ancestry (CEU, red dots, mostly covered by the dots of our population). The two major components were then used as covariates in the regression model.

Supplementary Figure 2.

The structure of the MAPT H2 haplotype, which includes eight coding variants, four of which are amino acid substitutions (p.P202L - rs63750417, p.D285N - rs62063786, p.R370W - rs17651549, p.S447P - rs10445337, p.P493P - rs1052551, p.T540T - rs62063845, p.A562A - rs1052553 and p.N590N - rs17652121). The intronic areas depicted here do not represent their real size. Of note, all four nonsynonymous variants are located in exons 4a and 6, which are probably expressed only in the peripheral nervous system. However, since the periphery, such as the gut nervous system, may be involved in early PD, a role for these variants in PD susceptibility cannot be excluded.

References

1. Gan-Or Z, Girard SL, Noreau A, et al. Parkinson's Disease Genetic Loci in Rapid Eye Movement Sleep Behavior Disorder. J Mol Neurosci 2015;56(3):617-622.2. Thorpy MJ. Classification of sleep disorders. Neurotherapeutics 2012;9(4):687-701.3. Boeve BF, Molano JR, Ferman TJ, et al. Validation of the Mayo Sleep Questionnaire to screen for REM sleep behavior disorder in an aging and dementia cohort. Sleep Med 2011;12(5):445-453.4. Labbe C, Heckman MG, Lorenzo-Betancor O, et al. MAPT haplotype H1G is associated with increased risk of dementia with Lewy bodies. Alzheimers Dement 2016;12(12):1297-1304.5. Fernandez-Santiago R, Iranzo A, Gaig C, et al. MAPT association with REM sleep behavior disorder. Neurol Genet 2017;3(1):e131.6. Ross JP, Dupre N, Dauvilliers Y, et al. Analysis of DNAJC13 mutations in French-Canadian/French cohort of Parkinson's disease. Neurobiol Aging 2016;45:212.e213-212.e217.7. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 2009;25(14):1754-1760.8. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 2010;20(9):1297-1303.9. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res 2010;38(16):e164.10. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 2006;38(8):904-909.11. Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 2015;4:7.12. Lee S, Emond MJ, Bamshad MJ, et al. Optimal unified approach for rare-variant association testing with application to small-sample case-control whole-exome sequencing studies. Am J Hum Genet 2012;91(2):224-237.13. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc 2009;4(7):1073-1081.