An Official Journal of the American Academy of Neurology Neurology.org/ng • Online ISSN: 2376-7839 Volume 3, Number 4, August 2017 Genetics

Functionally pathogenic ExACtly zero or once: Clinical and experimental EARS2 variants in vitro A clinically helpful guide studies of a novel P525R may not manifest a to assessing genetic FUS mutation in amyotrophic phenotype in vivo variants in mild epilepsies lateral sclerosis Table of Contents Neurology.org/ng  Online ISSN: 2376-7839 Volume 3, Number 4, August 2017

THE HELIX e171 Autopsy case of the C12orf65 mutation in a patient e175 What does phenotype have to do with it? with signs of mitochondrial dysfunction S.M. Pulst H. Nishihara, M. Omoto, M. Takao, Y. Higuchi, M. Koga, M. Kawai, H. Kawano, E. Ikeda, H. Takashima, and T. Kanda EDITORIAL e173 This variant alters function, but is it pathogenic? M. Pandolfo e174 Prevalence of 36 in a US Companion article, e162 population J.M. Valera, T. Diaz, L.E. Petty, B. Quintáns, Z. Yáñez, ARTICLES E. Boerwinkle, D. Muzny, D. Akhmedov, R. Berdeaux, e162 Functionally pathogenic EARS2 variants in vitro may M.J. Sobrido, R. Gibbs, J.R. Lupski, D.H. Geschwind, not manifest a phenotype in vivo S. Perlman, J.E. Below, and B.L. Fogel N. McNeill, A. Nasca, A. Reyes, B. Lemoine, B. Cantarel, A. Vanderver, R. Schiffmann, and D. Ghezzi Editorial, e173 e170 Loss-of-function variants of SCN8A in intellectual disability without seizures e163 ExACtly zero or once: A clinically helpful guide to J.L. Wagnon, B.S. Barker, M. Ottolini, Y. Park, assessing genetic variants in mild epilepsies A. Volkheimer, P. Valdez, M.E.M. Swinkels, C.A. Bennett, S. Petrovski, K.L. Oliver, and S.F. Berkovic M.K. Patel, and M.H. Meisler

e161 Abnormal expression of homeobox and e172 Clinical and experimental studies of a novel transthyretin in C9ORF72 expansion carriers P525R FUS mutation in amyotrophic lateral N.A. Finch, X. Wang, M.C. Baker, M.G. Heckman, sclerosis T.F. Gendron, K.F. Bieniek, J. Wuu, L. Kuang, M. Kamelgarn, A. Arenas, J. Gal, D. Taylor, M. DeJesus-Hernandez, P.H. Brown, J. Chew, W. Gong, M. Brown, D. St. Clair, E.J. Kasarskis, and K.R. Jansen-West, L.M. Daughrity, A.M. Nicholson, H. Zhu M.E. Murray, K.A. Josephs, J.E. Parisi, D.S. Knopman, R.C. Petersen, L. Petrucelli, B.F. Boeve, N.R. Graff-Radford, Y.W. Asmann, D.W. Dickson, e166 Brain calcifications and PCDH12 variants M. Benatar, R. Bowser, K.B. Boylan, G. Nicolas, M. Sanchez-Contreras, E.M. Ramos, R. Rademakers, and M. van Blitterswijk R.R. Lemos, J. Ferreira, D. Moura, M.J. Sobrido, A.-C. Richard, A.R. Lopez, A. Legati, J.-F. Deleuze, A. Boland, O. Quenez, P. Krystkowiak, P. Favrole, e164 Comparing sequencing assays and human-machine D.H. Geschwind, A. Aran, R. Segel, E. Levy-Lahad, analyses in actionable genomics for glioblastoma D.W. Dickson, G. Coppola, R. Rademakers, and K.O. Wrzeszczynski, M.O. Frank, T. Koyama, J.R.M. de Oliveira K. Rhrissorrakrai, N. Robine, F. Utro, A.-K. Emde, B.-J. Chen, K. Arora, M. Shah, V. Vacic, R. Norel, E. Bilal, E.A. Bergmann, J.L. Moore Vogel, J.N. Bruce, e176 UNC5C variants are associated with cerebral A.B. Lassman, P. Canoll, C. Grommes, S. Harvey, amyloid angiopathy L. Parida, V.V. Michelini, M.C. Zody, V. Jobanputra, H.-S. Yang, C.C. White, L.B. Chibnik, H.-U. Klein, A.K. Royyuru, and R.B. Darnell J.A. Schneider, D.A. Bennett, and P.L. De Jager Table of Contents continued

CLINICAL/SCIENTIFIC NOTES e167 Novel fukutin mutations in limb-girdle muscular e169 Updated nomenclature for human and mouse dystrophy type 2M with childhood onset neurofibromatosis type 1 genes M. Smogavec, J. Zschüntzsch, W. Kress, J. Mohr, C. Anastasaki, L.Q. Le, R.A. Kesterson, and D.H. Gutmann P. Hellen, B. Zoll, S. Pauli, and J. Schmidt

e168 Homozygous mutation in HSPB1 causing distal e165 Brainstem phenotype of cathepsin A–related vacuolar myopathy and motor neuropathy arteriopathy with strokes and leukoencephalopathy E. Bugiardini, A.M. Rossor, D.S. Lynch, M. Swash, Y.T. Hwang, R. Lakshmanan, I. Davagnanam, A.M. Pittman, J.C. Blake, M.G. Hanna, H. Houlden, A.G.B. Thompson, D.S. Lynch, H. Houlden, N. Bajaj, J.L. Holton, M.M. Reilly, and E. Matthews S.H. Eriksson, D.-E. Bamiou, and J.D. Warren

Podcast Video LOE classification LOE recommendation

tinyurl.com/NeurologyNG twitter.com/GreenJournal

Cover image: The binding details of the C-terminus of FUS and Trn1. P525 and Y526 of FUS (cyan) and L419, I457, W460 and D384 of Trn1 are shown in sticks. Trn1 is shown in static electric surface mode, in which the colors of red, blue, and green represent negative charge, positive charge, and hydrophobicity, respectively. See “Clinical and experimental studies of a novel P525R FUS mutation in amyotrophic lateral sclerosis.” Neurology.org/ng  Online ISSN: 2376-7839 Volume 3, Number 4, August 2017

fi VISION: Neurology® Genetics will be the premier peer-reviewed Executive Of ce, American Academy of Neurology fi Catherine M. Rydell, CAE, Executive Director/CEO journal in the eld of neurogenetics. 201 Chicago Ave Minneapolis, MN 55415 MISSION: Neurology: Genetics will provide neurologists with Tel: 612-928-6100 outstanding original contributions that elucidate the role of genetic and epigenetic variations in diseases and biological traits Editorial Office of the central and peripheral nervous systems. Patricia K. Baskin, MS, Executive Editor Kathleen M. Pieper, Senior Managing Editor, Neurology Editor Lee Ann Kleffman, Managing Editor, Neurology: Genetics Neurology: Genetics Sharon L. Quimby, Managing Editor, Neurology® Clinical Practice Stefan M. Pulst, MD, Dr med, FAAN Morgan S. Sorenson, Managing Editor, Neurology® Neuroimmunology & fl Professor and Chair, Department of Neurology Neuroin ammation University of Utah Cynthia S. Abair, MA, Senior Graphics Editor Salt Lake City, UT Andrea R. Rahkola, Production Editor, Neurology [email protected] Robert J. Witherow, Senior Editorial Associate Specialties: Genetics, movement disorders Karen Skaja, Senior Editorial Associate Kaitlyn Aman Ramm, Editorial Assistant Editor-in-Chief Kristen Swendsrud, Editorial Assistant Neurology® Andrea Willgohs, Editorial Assistant Robert A. Gross, MD, PhD, FAAN Professor of Neurology and of Pharmacology and Physiology Strong Epilepsy Center Publisher University of Rochester Medical Center Wolters Kluwer Rochester, NY Baltimore, MD [email protected]

Editorial Inquiries Publishing Staff Tel: 612-928-6400 Kim Jansen, Executive Publisher Toll-free: 800-957-3182 (US) Alexandra Lazerow, Production Team Leader, AAN Journals Fax: 612-454-2748 Steve Rose, Editorial Assistant [email protected] Stacy Drossner, Production Associate

Academy Officers Ralph L. Sacco, MD, MS, FAAN, President James C. Stevens, MD, FAAN, President Elect Ann H. Tilton, MD, FAAN, Vice President Carlayne E. Jackson, MD, FAAN, Secretary Janis M. Miyasaki, MD, MEd, FRCPC, FAAN, Treasurer Terrence L. Cascino, MD, FAAN, Past President Neurology.org/ng  Online ISSN: 2376-7839 Volume 3, Number 4, August 2017

Editor Biostatistics Neurology® Genetics Richard J. Kryscio, PhD Stefan M. Pulst, MD, Dr med, FAAN University of Kentucky Professor and Chair, Department of Neurology Lexington, KY University of Utah Christopher A. Beck, PhD Salt Lake City, UT University of Rochester [email protected] Rochester, NY Specialties: Genetics, movement disorders Sue Leurgans, PhD Editor-in-Chief Rush University Medical Center Neurology® Chicago, IL Robert A. Gross, MD, PhD, FAAN Level of Evidence Evaluations Professor of Neurology and of Pharmacology and Physiology Gary S. Gronseth, MD, FAAN Strong Epilepsy Center University of Kansas University of Rochester Medical Center Kansas City, KS Rochester, NY [email protected] Podcasts Ted M. Burns, MD Deputy Editors University of Virginia Neurology Charlottesville, VA Bradford B. Worrall, MD, MSc, FAAN Andrew M. Southerland, MD, MSc, Deputy Podcast Editor Professor of Neurology and Public Health Sciences University of Virginia University of Virginia Charlottesville, VA Charlottesville, VA Ombudsman Neurology: Genetics David S. Knopman, MD, FAAN Nicholas Elwood Johnson, MD Mayo Clinic University of Utah Rochester, MN Salt Lake City, UT Specialties: Muscle diseases, patient-centered outcomes Scientific Integrity Advisor Robert B. Daroff, MD, FAAN Associate Editors Case Western Reserve University Neurology: Genetics Cleveland, OH Alexandra Durr, MD, PhD Editorial Board Hôpital de la Salpêtrière Hilary Coon, PhD, University of Utah, Salt Lake City, UT Paris, France Giovanni Coppola, MD, UCLA, Los Angeles, CA Specialties: Spastic paraplegia, Huntington disease, ataxia Chantal Depondt, MD, PhD, Université Libre de Bruxelles, Brussels, Belgium Massimo Pandolfo, MD, FAAN Brent L. Fogel, MD, PhD, FAAN, UCLA, Los Angeles, CA Hôpital Erasme Anthony J. Griswold, PhD, University of Miami, Miami, FL Université Libre de Bruxelles Orhun H. Kantarci, MD, Mayo Clinic, Rochester, MN Brussels, Belgium Julie R. Korenberg, PhD, MD, University of Utah, Salt Lake City, UT Specialties: Epilepsy, Friedreich ataxia, stem cells Margherita Milone, MD, PhD, Mayo Clinic, Rochester, MN Davide Pareyson, MD, C. Besta Neurological Institute IRCCS Foundation, Raymond P. Roos, MD, FAAN Milan, Italy University of Chicago Medical Center Shoji Tsuji, MD, PhD, University of Tokyo Graduate School of Medicine, Chicago, IL Tokyo, Japan Specialties: Motor neuron disease, prion disease, neurovirology/neuroimmunology Dineke S. Verbeek, PhD, University Medical Center Groningen, Groningen, The Netherlands Jeffery M. Vance, MD, PhD David Viskochil, MD, PhD, University of Utah, Salt Lake City, UT University of Miami Juliane Winkelmann, MD, Technische Universität München, Helmholtz Miami, FL Zentrum Münche, Munich, Germany Specialties: Parkinson, Alzheimer disease, inherited neuropathies Juan I. Young, PhD, University of Miami, Miami, FL Level of Evidence Review Team Melissa J. Armstrong, MD, Gainesville, FL Richard L. Barbano, MD, PhD, FAAN, Rochester, NY Richard M. Dubinsky, MD, MPH, FAAN, Kansas City, KS Jeffrey J. Fletcher, MD, MSc, Ann Arbor, MI Gary M. Franklin, MD, MPH, FAAN, Seattle, WA David S. Gloss II, MD, MPH&TM, Charleston, WV John J. Halperin, MD, FAAN, Summit, NJ Jason Lazarou, MSc, MD, Toronto, Ontario, Canada Steven R. Messé, MD, FAAN, Philadelphia, PA Pushpa Narayanaswami, MBBS, DM, FAAN, Boston, MA Alex Rae-Grant, MD, Cleveland, OH Neurology.org/ng  Online ISSN: 2376-7839 Volume 3, Number 4, August 2017

Publication Information Neurology® is a registered trademark of the American Academy of Neurology (registration valid in the United States). Neurology® Genetics (eISSN 2376-7839) is an open access journal published online for the American Academy of Neurology, 201 Chicago Avenue, Minneapolis, MN 55415, by Wolters Kluwer Health, Inc. at 14700 Citicorp Drive, Bldg. 3, Hagerstown, MD 21742. Business offices are located at Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103. Production offices are located at 351 West Camden Street, Baltimore, MD 21201-2436. © 2017 American Academy of Neurology. Neurology® Genetics is an official journal of the American Academy of Neurology. Journal website: Neurology.org/ng, AAN website: AAN.com Copyright and Permission Information: Please go to the journal website (www.neurology.org/ng) and click the “©Request Permissions” icon for the relevant article. Alternatively, send an email to [email protected]. General information about permissions can be found here: https://www.lww.com/journal-permission. Disclaimer: Opinions expressed by the authors and advertisers are not necessarily those of the American Academy of Neurology, its affiliates, or of the Publisher. The American Academy of Neurology, its affiliates, and the Publisher disclaim any liability to any party for the accuracy, completeness, efficacy, or availability of the material contained in this publication (including drug dosages) or for any damages arising out of the use or non-use of any of the material contained in this publication. Advertising Sales Representatives: Wolters Kluwer, 333 Seventh Avenue, New York, NY 10001. Contacts: Eileen Henry, tel: 732-778-2261, fax: 973-215-2485, [email protected] and Elizabeth S. Hall, tel: 267-804-8123, [email protected]. In Europe: Avia Potashnik, Wolters Kluwer, tel: 144 207 981 0722; 144 7919 397 933 or e-mail: [email protected]. Careers & Events: Monique McLaughlin, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-521- 8468, fax: 215-521-8801; [email protected]. Reprints: Meredith Edelman, Commercial Reprint Sales, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-555-1212 (office), 215-356-2721 (mobile); [email protected]; [email protected]. Special projects: US & Canada: Alan Moore, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-521-8638, [email protected]. International: Andrew Wible, Senior Manager, Rights, Licensing, and Partnerships, Wolters Kluwer; [email protected]. THE HELIX What does phenotype have to do with it?

Stefan M. Pulst, MD, A recurring theme in this column and in articles , we identified a Gly263Asp amino Dr med published in Neurology® Genetics is the problem of acid substitution. This variant occurred in an amino acid interpreting DNA variants, ever more present as next- that was evolutionarily conserved, but not nearly to the generation sequencing invariably leads to the identi- same degree as amino acids changed by other verified Correspondence to fication of multiple rare variants in a given individual. KCNC3 mutations. Mutation prediction programs pre- Dr. Pulst: [email protected] This is discussed in this issue in 2 articles and an dicted this amino acid substitution to be deleterious and accompanying editorial.1–3 indeed, on introducing the variant Kv3.3 channel into Neurol Genet The Berkovic group critically examines reported var- frog oocytes, the biophysical characteristics were altered. 2017;3:e175; doi: 10.1212/ iants in autosomal dominant epilepsies.1 They make use What saved the day for us was the fact that the patient NXG.0000000000000175 oftheExACdatabaseandina2ndanalysisofthe hadaconcomitantrepeatexpansion mutation in expanded ExAC V2 plus gnomAD databases with CACNA1A,andtheKCNC3 variant did not segregate whole-exome and whole-genome sequencing data on with other affected family members. In retrospect, 123,136 individuals and 15,496 individuals, respec- a tip-off might have been that the biophysical changes tively. They find that occurrence of a variant more than in channel function, although significant, were rela- once in this database greatly reduces the probability that tively mild compared with the gain-of-function therespectivevariantisdiseasecausingwithhighpen- Phe448Leu mutation associated with early onset etrance. Their analysis also reflects on the potentially ataxia.5 Thus, significant may not always equal erroneous assignments of DNA variants in EFHC1 relevant. and SRPX2 as disease causing in familial epilepsy As Pandolfo points out, it is important to examine syndromes. the phenotype in its relationship to the variant The report by McNeil and colleagues2 describes and established phenotypes. Our colleagues in neurora- a molecular genetic case study of a patient presenting diology know this well and frequently add to their re- with an unusual leukodystrophy. The authors initially ports “clinical correlation is needed.” The same identify compound heterozygous mutations in the applies to the interpretation of in vitro assays. The EARS2 gene that led to reduction in protein levels by compound heterozygous variants in EARS2 described 70%. But as Massimo Pandolfo points out in the by McNeill et al. reduced protein levels to only 30% in accompanying editorial,3 the typical clinical and imag- the patient’s skin fibroblasts, probably not a sufficient ing phenotypes associated with mutations in EARS2 reduction to be functionally important. The next chal- did not fit the phenotype of the patient. McNeill lenge will be understanding the implications of changes et al. were aided by a prior study of this patient that in functional assays that may be significant in the con- had also shown compound heterozygous mutations in text of in vitro studies, but may not be significant at the SNORD118, encoding a small nucleolar RNA involved organismal level. Maybe biochemists and neurophysiol- in ribosome biogenesis. In addition, the phenotype ogist need to append “clinical correlation required!” associated with SNORD118 mutations actually fit the patient much better! We need to realize that the clinical STUDY FUNDING No targeted funding reported. or imaging subtleties of leukoencephalopathy, brain calcifications, and cysts vs leukoencephalopathy with DISCLOSURE thalamus and brainstem involvement and high lactate S.M. Pulst serves on the editorial boards of Journal of Cerebellum, Neuro- may escape the non-subspecialist. Molecular Medicine, Continuum, Experimental Neurology, Neurogenetics, Several years ago, we came across a similar scenario.4 and Nature Clinical Practice Neurology; receives research support from NIH, Target ALS, National Ataxia Foundation, and ISIS Pharmaceuti- By screening patients with autosomal dominant ataxias cals; has consulted for Ataxion Therapeutics and served on a speakers’ for mutations in KCNC3, the gene encoding the Kv3.3 bureau for Athena Diagnostics, Inc.; is a stockholder of Progenitor Life

From the Department of Neurology, University of Utah, Salt Lake City. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Sciences; has received license fee payments from Cedars-Sinai Medical Cen- 2. McNeil N, Nasca A, Reyes A, et al. Functionally pathogenic ter; holds patents for Nucleic acids encoding ataxin-2 binding , EARS2 variants in vitro may not manifest a phenotype in- Nucleic acid encoding Schwannomin-binding proteins, Transgenic mouse vivo. Neurol Genet 2017;3:e162. doi: 10.1212/NXG. expressing a polynucleotide encoding a human ataxin-2 polypeptide, Meth- 0000000000000162. ods of detecting spinocerebellar ataxia-2 nucleic acids, Nucleic acid encoding 3. Pandolfo M. This variant alters protein function, but is it spinocerebellar ataxia-2 to, Schwannomin-binding proteins, and Composi- pathogenic? Neurol Genet 2017;3:e173. doi: 10.1212/ tions and methods for spinocerebellar ataxia; and receives an honorarium from the AAN as the Editor of Neurology: Genetics.GotoNeurology.org/ng NXG.0000000000000173. for full disclosure forms. 4. Figueroa KP, Waters MF, Garibyan V, et al. Frequency of KCNC3 DNA variants as causes of spinocerebellar ataxia 13 REFERENCES (SCA13). PLoS One 2011;6:e17811. 1. Bennett CA, Petrovski S, Oliver KL, Berkovic SF. 5. Waters MF, Minassian NA, Stevanin G, et al. Mutations in ExACtly zero or once: clinically helpful guide to assess- voltage-gated potassium channel KCNC3 cause degenera- ing genetic variants in mild epilepsies. Neurol Genet tive and developmental central nervous system phenotypes. 2017;3:e163. doi: 10.1212/NXG.0000000000000163. Nat Genet 2006;38:447–451.

2 Neurology: Genetics EDITORIAL This variant alters protein function, but is it pathogenic?

Massimo Pandolfo, MD Identifying the causative mutation for a mendelian However, one important piece did not fit in the disease among several, or many, genetic variants rep- puzzle. In the original description of LTBL, brain resents a challenge, even more so with the increasing MRI findings were characterized in all patients by Correspondence to use of next-generation sequencing for genetic diagno- extensive symmetrical cerebral white matter abnor- Dr. Pandolfo: [email protected] sis. In 2015, the American College of Medical Genet- malities sparing the periventricular rim and by sym- ics and Genomics (ACMG) and the Association for metrical signal abnormalities of the thalami, Neurol Genet Molecular Pathology (AMP) proposed guidelines to midbrain, pons, medulla oblongata, and cerebellar 2017;3:e173; doi: 10.1212/ classify variants in genes associated with mendelian white matter. As indicated in the name of the disease, NXG.0000000000000173 diseases according to their potential pathogenicity.1 brain lactate, measured by proton magnetic resonance ACMG-AMP rules include genetic criteria such as spectroscopy (MRS), was increased. LTBL is not lim- population, segregation, de novo, allelic, and database ited to the nervous system, as it may cause liver abnor- data, along with functional criteria, including com- malities and increased lactate in all body fluids. Of putational and predictive data and functional studies. interest, clinical, MRI, and MRS follow-up of pa- When the same amino acid change is found as in an tients with LTBL allowed us to identify a “severe” established pathogenic variant, this is considered subgroup showing early postnatal onset with failure strongly supportive of pathogenicity. Functional to thrive and hypotonia then spastic tetraparesis, dys- studies showing a deleterious effect are also strongly tonia, visual impairment, and seizures, followed by supportive of pathogenicity. clinical stagnation, brain atrophy, and persisting high In this issue of Neurology® Genetics, McNeill et al.2 lactate levels; and a “mild” subgroup, showing later report the finding by whole-exome sequencing clinical onset, most often in the second half of the first (WES) of compound heterozygous variants in the year of life, with spasticity, loss of milestones, occa- EARS2 gene in a patient with leukoencephalopathy. sional seizures, and extreme irritability, followed by EARS2, encoding mitochondrial glutamyl-tRNA syn- partial clinical recovery, MRI improvement, and thetase, is mutated in the autosomal recessive leu- declining lactate levels.3 koencephalopathy with thalamus and brainstem The patient studied by McNeill et al. neither involvement and high lactate (LTBL).3 One of the showed the unique neuroimaging features of LTBL variants detected in this patient, c.328G.A nor presented the peculiar biphasic clinical course of (p.G110S), had previously been found in multiple this disease. Their patient was a 6-year-old boy with patients with LTBL. In silico analysis supported path- a progressive encephalopathy consisting of intellectual ogenicity of the other variant (c.1045G.A, disability, refractory seizures, dystonia, chorea, and p.E349K) as well. Furthermore, McNeill et al. per- spasticity, who died at the age of 16 after a progressive formed functional analyses in patient skin fibroblasts, disease course. MRI showed a diffuse leukoencephal- finding a significant decrease in EARS2 protein levels opathy with multiple calcifications affecting gray mat- to 30% of normal controls, along with an ;11% ter nuclei, juxtacortical U-fibers, subcortical and decrease in the oxygen consumption rate and periventricular white matter, brainstem, and the den- ;43% decrease in the maximal respiratory rate com- tate nucleus of the cerebellum, while MRS showed no pared with control. Mitochondrial dysfunction was increase in lactate. Autopsy showed a severe vasculop- corrected by the expression of wild-type EARS2 pro- athy causing ischemic lesions, calcifications, white tein. Taken together, these results seemed to provide matter degeneration, and cyst formation. This clini- strong evidence supporting pathogenicity of the iden- cal, imaging, and pathologic features are strongly sug- tified EARS2 variants. gestive of a condition called leukoencephalopathy, See article From the Department of Neurology, Hôpital Erasme and Laboratory of Experimental Neurology, Université Libre de Bruxelles, Belgium. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge for this editorial was waived at the discretion of the Editor. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 brain calcifications, and cysts (LCC),4 which is strictly experimental data from animal models may allow limited to the nervous system and not associated with to address the issue of how potentially pathogenic increased lactate. mutations in different genes interact and eventually So, if the phenotype of this patient was clearly dif- determine the phenotype. Recently, a study from ferent from LTBL and rather fitting a different condi- Baylor College of Medicine on more than 2,000 tion, LCC, what to do of the putative EARS2 patients receiving a molecular diagnosis after WES mutations? Does this case extend the phenotypic showed that 4.9% of them carried seemingly path- spectrum linked to EARS2 mutation? Or conversely, ogenic mutations in 2 loci.6 These patients showed are the EARS2 variants unrelated to the LCC pheno- either distinct disease phenotypes, with individual type of this patient? And, if the latter is the case, have phenotypic features attributable to only 1 of the 2 the EARS2 variants modified the patient’s phenotype? diagnoses, or overlapping phenotypes, with pheno- McNeill et al. found a partial answer to these ques- typic features attributable to either one of the diag- tions when they realized that their patient had also noses. Overlapping disease phenotypes were been independently investigated in a genetic study observed particularly when the coexisting mutations on LCC and found to carry compound heterozygous altered proteins in the same functional pathway. The mutations in SNORD118, encoding a small nucleolar peculiarity of the patient of McNeill et al., deserving RNA involved in ribosome biogenesis.5 SNORD118 additional investigation, is the apparently “pure” was also mutated in additional LCC families, identi- LCC phenotype lacking clinical features attributable fying it as a causative gene for this condition. Based to the EARS2 mutations, when an overlapping phe- on this finding, they conclude “that the presence of notype might have been expected. the SNORD118 mutations as the pathogenic driver of LCC relegates the functionally pathogenic EARS2 STUDY FUNDING variants in this case to an unknown status.” No targeted funding reported. What lessons can we draw from this report? First, DISCLOSURE rigorous characterization of the phenotype is essential Massimo Pandolfo has served on the scientific advisory boards of Apo- for the interpretation of genetic test results. This is pharma and Voyager Therapeutics; has served on the editorial boards not as trivial as it may seem because examples abound of Acta Neurologica Belgica, the Orphanet Journal of Rare Diseases, and Neurology: Genetics; holds a patent for Direct molecular diagnosis of of mutations in different genes causing highly similar Friedreich’s ataxia; has been a consultant for Biomarin; has received phenotypes and of mutations in the same gene caus- research support from Biomarin, Fonds National de la Recherche Scien- ing very different phenotypes. The very concept of tifique (Belgium), Offrez-moi-la-lune, Friedreich’s Ataxia Research Alli- “reverse phenotyping” is based on the latter point. ance, and Association Belge contre les Maladies Neuro-Musculaires; and has received royalty payments from Athena Diagnostics. Go to However, in many cases, some core clinical and para- Neurology.org/ng for full disclosure forms. clinical features may enable to predict how likely is the involvement of a specific gene. In the case REFERENCES described by McNeill et al., normal levels of lactate 1. Richards S, Aziz N, Bale S, et al. Standards and guidelines in the brain and in body fluids argued against an for the interpretation of sequence variants: a joint consensus EARS2-linked phenotype, even more than structural recommendation of the American College of Medical MRI features and atypical clinical course. The sec- Genetics and Genomics and the Association for Molecular Pathology. Genet Med 2015;17:405–424. ond, probably more unexpected lesson, is that even 2. McNeill N, Nasca A, Reyes A, et al. Functionally patho- demonstrating an abnormal function of the encoded genic EARS2 variants in vitro may not manifest a phenotype protein may not be enough to prove pathogenicity of in vivo. Neurol Genet 2017;3:e162. doi: 10.1212/NXG. a genetic variant. In a way, this is just a reiteration of 0000000000000162. the previous point: if the phenotype does not reflect 3. Steenweg ME, Vanderver A, Ceulemans B, et al. Novel the observed in vitro abnormalities, a different cause infantile-onset leukoencephalopathy with high lactate level and slow improvement. Arch Neurol 2012;69:718–722. is possible and even likely. 4. Stephani C, Pfeifenbring S, Mohr A, Stadelmann C. Late- Finally, the question whether seemingly patho- onset leukoencephalopathy with cerebral calcifications and genic EARS2 mutations in this patient may have cysts: case report and review of the literature. BMC Neurol somehow modified his phenotype remains open. 2016;16:1–12. Even if the expected increase in lactate did not 5. Jenkinson EM, Rodero MP, Kasher PR, et al. Mutations in occur, subclinical mitochondrial dysfunction may SNORD118 cause the cerebral microangiopathy leukoen- cephalopathy with calcifications and cysts. Nat Genet 2016; have worsened the consequence of the microangiop- 48:1185–1192. athy of LCC, but the reported data do not allow us 6. Posey JE, Harel T, Liu P, et al. Resolution of disease phe- to test this hypothesis. More generally, a combina- notypes resulting from Multilocus Genomic Variation. N tion of clinical observation of multiple cases with Engl J Med 2017;376:21–31.

2 Neurology: Genetics Functionally pathogenic EARS2 variants in vitro may not manifest a phenotype in vivo

Nathan McNeill, PhD* ABSTRACT Alessia Nasca, MS Objective: To investigate the genetic etiology of a patient diagnosed with leukoencephalopathy, Aurelio Reyes, PhD brain calcifications, and cysts (LCC). Benjamin Lemoine, MS Methods: Whole-exome sequencing was performed on a patient with LCC and his unaffected fam- Brandi Cantarel, PhD ily members. The variants were subject to in silico and in vitro functional testing to determine Adeline Vanderver, MD pathogenicity. Raphael Schiffmann, MD* Results: Whole-exome sequencing uncovered compound heterozygous mutations in EARS2, . . Daniele Ghezzi, MD* c.328G A (p.G110S), and c.1045G A (p.E349K). This gene has previously been implicated in the autosomal recessive leukoencephalopathy with thalamus and brainstem involvement and high lactate (LTBL). The p.G110S mutation has been found in multiple patients with LTBL. In silico

Correspondence to analysis supported pathogenicity in the second variant. In vitro functional testing showed a sig- Dr. McNeill: nificant mitochondrial dysfunction demonstrated by an ;11% decrease in the oxygen consump- [email protected] tion rate and ;43% decrease in the maximum respiratory rate in the patient’sskinfibroblasts compared with the control. EARS2 protein levels were reduced to 30% of normal controls in the patient’s fibroblasts. These deficiencies were corrected by the expression of the wild-type EARS2 protein. However, a further unrelated genetic investigation of our patient revealed the presence of biallelic variants in a small nucleolar RNA (SNORD118) responsible for LCC. Conclusions: Here, we report seemingly pathogenic EARS2 mutations in a single patient with LCC with no biochemical or neuroimaging presentations of LTBL. This patient illustrates that variants with demonstrated impact on protein function should not necessarily be considered clinically relevant. ClinicalTrials.gov identifier: NCT00001671. Neurol Genet 2017;3:e162; doi: 10.1212/ NXG.0000000000000162

GLOSSARY LCC 5 leukoencephalopathy, brain calcifications, and cysts; LTBL 5 leukoencephalopathy with thalamus and brainstem involvement and high lactate; MAF 5 minor allele frequency; MRR 5 maximum respiratory rate; OCR 5 oxygen consumption rate.

Leukodystrophies and genetic leukoencephalopathies are a heterogeneous group of rare inherit- able neurologic diseases predominantly affecting the white matter of the brain.1 Specific genetic diagnosis of these disorders was previously found in only about half of the known cases,2 but next-generation sequencing has helped to increase the rate of genetic classification up to 80%.3 Next-generation sequencing has been successfully used in the identification of rare mutations in a number of genes causing many different white matter diseases and mitochondrial disorders.4 Particularly, mutations in the nuclear-encoded aminoacyl tRNA synthetase genes (ARSs) have See editorial been implicated in numerous mitochondrial disorders that cause white matter abnormalities.5 However, the explosive increase in rare genetic variants reported in the population in recent Supplemental data at Neurology.org/ng *These authors contributed equally to this work. From the Baylor Research Institute (N.M., B.L., R.S.), Baylor Scott and White Health, Dallas, TX; Unit of Molecular Neurogenetics (A.N., D.G.), Foundation IRCCS Institute of Neurology “Besta,” Milan, Italy; Mitochondrial Biology Unit (A.R.), Medical Research Council, Cambridge, United Kingdom; Department of Bioinformatics (B.C.), University of Texas Southwestern Medical Center, Dallas; and Department of Neurology (A.V.), George Washington University School of Medicine, Children’s National Health, DC. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 years has at times misrepresented the link Sanger sequencing. The EARS2 variants were validated using between mutation and disease by incorrectly Sanger sequencing. Genomic DNA was amplified using primers 6 that targeted the variant location and subsequently sequenced labeling many genetic variants as pathogenic. using Applied Biosystems’ BigDye Terminator v1.1 Sequencing This is in part due to the lack of experimental Kit chemistry on a 3130xl sequencer (Applied Biosystems- validation of the pathogenicity of variants Thermo Fisher Scientific, Waltham, MA). In addition, the through functional in vitro or in vivo experi- EARS2 genes in 2 unrelated patients diagnosed with leukoence- phalopathy, brain calcifications, and cysts (LCC) were Sanger mentation. Thus, clinical reporting of such sequenced using primers that spanned all 9 exons and exon-intron variants is reliant on accurate information boundaries. PCR amplification was performed using HotStarTaq about the nature of the variant highlighting Master Mix (Qiagen, Hilden, Germany) according to the man- ’ the importance of aggregated population data- ufacturer s protocols. EARS2 primers are listed in table e-1 at Neurology.org/ng. bases for correct assessment of variant frequen- cies and the need for rigorous functional Skin fibroblast cell culture and complementation studies. Primary skin fibroblasts from the affected patient and controls assays. As such, mitochondrial dysfunction were cultured in 13 Dulbecco Modified Eagle Medium can be ascertained through functional mito- (Corning, Corning, NY) with 4.5 g/L glucose supplemented with chondrial respiration assays, which are thought L-glutamine and sodium pyruvate, 10% fetal bovine serum, and 1% antimyotic in a 25-cm2 flask. The medium was changed every to be helpful in predicting DNA variant path- 2 days until 90%–100% confluent, at which time the cells were ogenicity. However, functionally validated trypsinized with 0.25% or 0.05% trypsin and washed with 13 variants may not always be clinically relevant phosphate-buffered saline (Gibco-Thermo Fisher Scientific) 21 21 2 to the pathogenesis of the disease in question. without Ca /Mg then subpassaged to a 75 cm flask. Cultures were maintained in a humidified atmosphere at 37°C with 5% CO . METHODS Standard protocol approvals, registrations, 2 Fibroblasts were immortalized with pRNS-1 by transfection and patient consents. This study was approved to use human using Lipofectamine2000 (Invitrogen-Thermo Fisher Scientific) subjects by the appropriate institutional review board. Biological and selected by 100 mg/mL Geneticin G-418 (Gibco-Thermo samples from the patient and first-degree relatives were collected Fisher Scientific).13 For complementation, wild-type (wt) cDNAs subsequent to written informed consent; ClinicalTrials.gov of EARS2 from a commercial clone (OCAAo5051A02110D) identifier: NCT00001671. It included a family consisting of 2 (Source BioScience, Nottingham, United Kingdom) were cloned parents and their affected son and his 3 unaffected siblings. into the pLenti6.2/V5 TOPOVector (Life Technologies-Thermo Whole-exome sequencing. Each family member was subjected Fisher Scientific), and virions were obtained as previously to whole-exome sequencing and genetic analysis. Briefly, libraries described.14 Mutant and control immortalized fibroblasts were were constructed from genomic DNA of each family member transfected with viral supernatant and selected upon exposure using the Illumina TruSeq DNA Sample Prep methodology to 2 mg/mL Blasticidin (Life Technologies-Thermo Fisher (Illumina, San Diego, CA), and exonic targets were captured Scientific). using Illumina’s TruSeq Exome Enrichment technology accord- Western blot protein expression. Fibroblasts and immortal- ing to the manufacturer’s protocols. 2 3 50 bp paired-end ized cells from patient and controls were trypsinized, pelleted, sequencing was performed on the Illumina HiSeq 2000. and solubilized in RIPA buffer with protease inhibitors; 50 mg Bioinformatic analysis. Sequence reads were aligned to the of protein was loaded for each sample in 12% denaturing sodium human reference genome hg19 using the Burrows-Wheeler dodecyl sulfate polyacrylamide gel electrophoresis. A rabbit poly- Aligner v.0.5.9.7 PCR duplicates were marked and removed clonal against EARS2 (SAB2100641) (Sigma-Aldrich, using Picard v.1.70 (broadinstitute.github.io/picard/), and local St. Louis, MO) and a mouse monoclonal antibody against realignment, quality score recalibration, and SNP and INDEL GAPDH (#MAB374) (Millipore, Billerica, MA) were used. variant calling were performed using the Genome Analysis Mitochondrial respiration assays. Oxygen consumption rate Toolkit v.1.3.8 The variant calls were subsequently annotated by (OCR) and maximum respiratory rate (MRR) were measured Annovar.9 The gene annotations were made against the RefSeq using a SeaHorse FX-96 apparatus (Agilent Technologies, Santa database, and known variants and allele frequencies were anno- Clara, CA)15 in primary fibroblasts and immortalized fibroblasts tated with dbSNP 137, all ethnicities from the 1000 Genomes in naive conditions and after transduction with wild-type Project (April 2012, August 2015 release), the National Heart, EARS2 cDNA. Lung, and Blood Institute (NHLBI) GO Exome Sequencing Project (ESP, esp6500 release), and the Exome Aggregation Consortium (ExAC). Nonsynonymous variants were further RESULTS Clinical profile. A 6-year-old male patient annotated with pathogenicity prediction scores from SIFT,10 born to healthy nonconsanguineous parents pre- PolyPhen-2,11 LRT, and MutationTaster.12 Conservation scores sented with a progressive encephalopathy consisting from PhyloP, GERP, and PhastCons were provided by Muta- of intractable seizures, dystonia, chorea, and spasticity tionTaster and Annovar. Multiple sequence alignments were with severely impaired cognitive function. MRI performed with Clustal Omega in Uniprot (.org). Variant (figure 1, A–D) and CT (figure 1E) showed diffuse prioritization and candidate gene identification used an in-house workflow that stratified the variant annotation data and white white matter signal abnormalities and numerous matter disease association analysis with a recessive or dominant calcifications throughout the brain in gray matter model of inheritance. nuclei and juxtacortical U-fibers as well as the

2 Neurology: Genetics with citrate synthase at the upper limits of normal, Figure 1 Patient with leukoencephalopathy brain imaging which may indicate an increase in mitochondrial content. Glucose, lactate, and total protein were normal in the CSF. The patient died at 16 years of age. Autopsy showed that the entire CNS was devastated by a vas- culopathy with secondary ischemic lesions and mineralization, leading to the progressive obliter- ation of the blood vessel lumina and gliosis resulting in the presence of Rosenthal fibers. Necrosis, dystro- phic calcifications, white matter degeneration, and cyst formation were found. There were no abnormal- ities outside the CNS. The patient was diagnosed with the cerebral microangiopathy LCC.16 LCC is a rare disorder, as fewer than 50 cases have ever been reported in the literature.17

Exome sequencing and genetic analysis. Whole-exome sequencing generated an average of 3.2 Gb of sequence and 61–73 million reads per individual that mapped uniquely to the genome with a mean sequence coverage of 263. LCC is an assumed auto- somal recessive disorder due to the occurrence of sib- ling pairs, which include females, but variants MRIs (A–D) and CT (E) indicate white matter signal abnormalities, cysts, and calcifications following both an autosomal recessive and autosomal throughout the cerebral hemispheres. Diffuse cerebral white matter lesions are present, dominant inheritance pattern were interrogated. Fol- which is demonstrated by hypointense signals in the sagittal (A) and axial T1-weighted (B) images and the CT (E) and by the hyperintense signals in the axial T2-weighted (C) and axial lowing a candidate gene prioritization and filtering fluid-attenuated inversion recovery (FLAIR) images. The CT (E) shows extensive calcifica- strategy that enriched for rare (minor allele frequency tions in the subcortical white matter and along the periventricular white matter. Large cystic [MAF] of ,1%) nonsynonymous exonic/splice var- – lesions can be seen along the quadrigeminal plane and parieto-occipital regions (B D). iants that both segregated in a recessive or dominant manner and may or may not be associated with white periventricular white matter, brainstem, the dentate matter disease pathogenesis, only 2 recessive variants nucleus of the cerebellum, and the subcortical white in the EARS2 gene segregated in a compound hetero- matter of both cerebral hemispheres. White matter zygous manner, c.328G.A (p.Gly110Ser) and abnormalities were observed through increased sig- c.1045G.A (Glu349Lys). No other obvious exonic nals in the white matter in T2-weighted (figure 1C) or splice variants were observed. The presence or and fluid-attenuated inversion recovery (FLAIR) absence of the 2 variants and their segregation within (figure 1D) images and decreased signals on sagittal the family were validated with Sanger sequencing (figure 1A) and T1-weighted images (figure 1B). MR (figure e-1). The patient was compound heterozygous spectroscopy showed decrease in N-acetyl aspartate for the 2 variants, the mother was heterozygous for and no lactate peak. A number of large cysts were also the c.328G.A variant, and the father was heterozy- present in the cerebellum and the supratentorial gous for the c.1045G.A variant. Two of the siblings region (figure 1, A–D). Anatomically, lateral ven- were heterozygous for either variant. Sanger sequenc- tricles were enlarged with a missing septum pelluci- ing of the exon and exon/intron boundaries of EARS2 dum, atrophy of the corpus callosum, and basal in 2 unrelated patients with LCC revealed only com- ganglia volume loss. Routine laboratory testing re- mon polymorphisms and no rare mutations. vealed normal values and complete blood count. Extensive biochemical investigations were all normal In silico analysis of EARS2 variants. The EARS2 var- except for a CNS folate deficiency from decreased iants were analyzed in silico to determine the MAF concentrations of 5-MTHF of unknown cause. within the population, predicted pathogenicity, and Muscle biopsy demonstrated a normal histopathology evolutionary conservation. The c.1045G.A nucleo- and mitochondria that were normal in number, dis- tide resides in exon 5 of EARS2 in the anticodon tribution, and morphology with no mtDNA abnor- binding domain, resulting in a missense amino acid malities. Cellular lactate-to-pyruvate ratio was normal change from a large acidic negatively charged glutamic in cultured skin fibroblasts. Skeletal muscle showed acid to a large positively charged basic lysine at residue normal carnitine profiles and normal activity of the 349 (Grantham Score: 56). It was present at low fre- electron transport chain complexes I, II, III, and IV quencies in 1000 Genomes (MAF 0.10%), ESP (MAF

Neurology: Genetics 3 0.18%), and ExAC (MAF 0.13%). It was also present through Caenorhabditis elegans, and the E349 residue in dbSNP (rs#187662524). The variant was predicted showed amino acid class conservation through C. as benign by PolyPhen-2, damaging or disease causing elegans. The c.328G.A variant was also found in 3 by SIFT and MutationTaster, and neutral by LRT. other unrelated patients with leukoencephalopathy This position is conserved according to GERP, PhyloP, with thalamus and brainstem involvement and high and PhastCons. lactate (LTBL),18,19 whereas the c.1045G.Avari- The c.328G.A nucleotide change resides in exon 3 ant has never been reported to be associated with in the catalytic domain of EARS2, resulting in a mis- human diseases. According to the ACMG and AMP sense amino acid change from a small polar glycine to recommended guidelines for interpreting sequence a small polar serine at residue 110 (Grantham Score: variants, both variants demonstrate evidence for 56). It was not present in the 1000 Genomes, but it pathogenicity.20 was present at low frequencies in the ESP (MAF 0.03%) and ExAC (MAF 0.04%) databases. It was also Biochemical/protein studies on patient’s fibroblasts. Skin present in dbSNP (rs#201842633). Pathogenicity pre- fibroblasts from the proband were analyzed to evalu- diction programs considered this variant as damaging/ ate the effect of the identified EARS2 variants. The disease causing by SIFT, PolyPhen2, MutationTaster, total amount of EARS2 protein detected by Western and LRT. GERP, PhyloP, and PhastCons also pre- blot analysis was reduced to 30% in the patient’s dicted high conservation at this position. fibroblasts compared with controls (figure 2A). Oxy- Multiple sequence alignment (figure e-2) showed gen consumption, which depends on and reflects the that the G110 residue is invariant from humans cumulative proficiency of the whole set of

Figure 2 Functional characterization of EARS2 variants on fibroblasts

(A) EARS2 protein amount in the patient’s (Pt) and control (CT1, CT2, and CT3) fibroblasts, obtained by Western blot using an anti-EARS2 antibody. An anti- GAPDH antibody was used as a loading control. (B) Oxygen consumption analysis in the patient’s (Pt) and control fibroblasts. Histograms show OCR (B.a and

D.a) and MRR (B.b and D.b). OCR and MRR values (mean of 6–8 replicates) are expressed as picomoles of O2 per minute and normalized by cell number. p value obtained with 2-tailed Student t test, *p , 0.05; **p , 0.01. (C) EARS2 protein amount in the patient’s (iPt) and control (iCT) immortalized fibroblasts, in basal conditions and after transduction with wt EARS2 (1EARS2); Western blot analysis was performed as described in A. (D) Oxygen consumption analysis, as reported in B, performed in patient’s (iPt) and control (iCT) immortalized fibroblasts, in basal conditions and after transduction with wild-type EARS2 (1EARS2). ***p , 0.001. MRR 5 maximum respiration rate; OCR 5 oxygen consumption rate.

4 Neurology: Genetics mitochondrial respiratory chain complexes, was mea- which leads to microcystic and macrocystic parenchy- sured. Significant reductions of both OCR and MRR, mal degeneration with white matter changes second- indicating reduced electron flow through the respira- ary to brain edema rather than primary tory chain, were observed (figure 2B). These altera- demyelination.16,17,21 MRI of the patient was consis- tions were also confirmed in patient-derived tent with LCC and not LTBL. The patient’s pathol- immortalized fibroblasts (figure 2, C and D), which ogy was also consistent with LCC including were then used for complementation through trans- angiomatous-like rearrangements of microvessels duction of wild-type EARS2. Overexpression of with secondary degeneration of cellular constituents EARS2 protein in both control and mutant trans- such as gliosis and aggregates of intermediate fila- duced cells (figure 2C) was associated with the ments called Rosenthal fibers,16,22 none of which are recovery of defective respiratory parameters (OCR known to be found in LTBL. LCC and LTBL have and MRR) to normal values (figure 2D). strikingly different neuroimaging and pathologic fea- tures, suggesting that EARS2 is not involved in the DISCUSSION Using whole-exome sequencing and pathogenesis of LCC. ad hoc strategic filtering with integration of gene However, functional studies on the LCC patient’s information and disease association, 2 rare EARS2 fibroblasts clearly demonstrated a mitochondrial dys- (MIM 612799) mutations were observed to be seg- function due to abnormal mitochondrial respiration regating in a compound heterozygous manner within with significantly decreased MRR and OCR and a single family with an individual presenting with decreased EARS2 protein levels suggestive of a patho- LCC (MIM 614561; Labrune Syndrome). EARS2 is genic role of the EARS2 variants; nonetheless, LCC targeted to the mitochondria and functions as a cru- does not present with mitochondrial abnormalities. cial component of mitochondrial translation by cat- Abnormal mitochondrial respiration and decreased alyzing the ligation of glutamate to its cognate tRNA protein levels as seen in this patient have both been molecule; however, all its functions have not yet been confirmed in other patients with LTBL.18,19,23 Pa- elucidated. Previous genetic studies identified muta- tients with LTBL have seen up to a ;70% decrease tions in EARS2 as the cause of combined oxidative in the MRR,18 and our patient showed a less severe phosphorylation deficiency 12 (COXPD12), also but still a significant decrease of ;43% in the MRR. known as LTBL (MIM 614924).18 Biochemical and Despite these abnormalities, increased lactate, which neuroimaging features of LTBL associated with is a hallmark of LTBL and a consequence of patho- EARS2 mutations include characteristic lactate ele- genic EARS2 mutations,18,19 was not seen in MR vation in MR spectroscopy and body fluids, variable spectroscopy or body fluids in the present patient corpus callosum involvement, and symmetric white with LCC. matter signal abnormalities in the cerebral Recently, mutations in the SNORD118 gene white matter, thalami, brainstem, and cerebellar located in the 39 UTR region of the TMEM107 gene white matter with sparing of the periventricular rim were identified in the pathogenesis of LCC.24 (figure 3, A–D).18,19 SNORD118 encodes the box C/D snoRNA U8 Conversely, LCC is strictly a neurologic disorder important for ribosome biosynthesis. In that publica- limited to the CNS and consisting of a cerebral mi- tion, family member 1A from family F278, which was croangiopathy resulting in presumed tissue hypoxia, this study’s patient with LCC, was found to have

Figure 3 LTBL brain imaging

Axial T2-weighted (A–C) and T1-weighted (D) MRIs demonstrate T2-hyperintensities and T1-hypointensities respectively, indicating lesions in the deep cerebral white matter and periventricular white matter with sparing of the periventricular rim. Signal hyperintensities are also present in the thalami (B) and dorsal part of the midbrain (C). Modified from reference 18. LTBL 5 leukoencephalopathy with thalamus and brainstem involvement and high lactate.

Neurology: Genetics 5 compound heterozygous mutations in SNORD118: STUDY FUNDING a novel n.75A.G mutation and a very rare n.8G.C Funding provided by the Baylor Scott & White Healthcare Foundation. mutation. The presence of the SNORD118 muta- DISCLOSURE tions as the pathogenic driver of LCC relegates the N. McNeill, A. Nasca, A. Reyes, B. Lemoine, and B. Cantarel report no functionally pathogenic EARS2 variants in this case to disclosures relevant to the manuscript. A. Vanderver has served on the an unknown status; however, EARS2 as a genetic dis- scientific advisory board of Shire Pharmaceuticals; has received travel ease modifier or the additional presence of protective funding/speaker honoraria from Hunter’s Hope (family foundation for Metachromatic Leukodystrophy), United Leukodystrophy Foundation, alleles from defective EARS2 cannot be ruled out. Metachromatic Leukodystrophy Foundation, and European Leukodys- The absence of a clear LTBL phenotype in trophy Foundation; holds a patent (pending) for Sialic acid measurement a patient with apparent functionally validated patho- in CSF in the diagnosis of Vanishing White Matter disease; has been an genic EARS2 variants, and the presence of other path- unpaid consultant for Stem Cells Inc.; and has received research support from Pennsylvania Department of Health, Frontiers in Leukodystrophy ogenic variants (i.e., SNORD118 mutations) strongly Initiative (FrontLINe), NIH, H-ABC Research Fund, Calendar Research associated with the patient’s phenotype, throws doubt Contribution Fund, H-ABC Research Fund, and Aicardi-Goutieres Syn- on our ability to infer the clinical pathogenic effect of drome Research Fund. R. Schiffman has served on the scientific advisory a variant through in vitro experiments. According to boards of and has received travel funding/speaker honoraria from Protalix Biotherapeutics and Amicus Therapeutics; holds the following patents: the standards and guidelines for the interpretation of Triheptanoin diet for Adult Polyglucosan Body Disease (APBD), Use of sequence variants set in place by the American Col- tetrahydrobiopterin as a marker and a therapeutic agent for Fabry disease lege of Medical Genetics and Genomics (ACMG),20 (3), Urinary triaosylceramide (GB3) as a marker of cardiac disease, and Gene encoding a new TRP channel is mutated in mucolipidosis; has been the EARS2 variants are considered pathogenic on a consultant for Gerson Lehrman Group Councils and Guidepoint their own and should be labeled as such in clinical Global; has served on the speakers’ bureau of Genzyme Corporation; reports. The functional assays indicating mitochon- and has received research support from Protalix Biotherapeutics, Amicus drial dysfunction through respiration defects as pre- Therapeutics, and Baylor Research Foundation. D. Ghezzi has served on the editorial board of Orphanet Journal of Rare Diseases; and has received sented in this study are typically sufficient to research support from the Italian Ministry of Health, European Com- demonstrate clinical relevance of EARS2 variants18 munities, Foundation Telethon, CARIPLO Foundation Italy, and Pier- and other variants in mitochondrial-related genes.25,26 franco and Luisa Mariani Foundation of Italy. Go to Neurology.org/ng It is important that this indicates that one cannot for full disclosure forms. simply always assume that putative functionally dam- Received February 13, 2017. Accepted in final form April 18, 2017. aging variants ascertained from in vitro experimenta- tion are clinically relevant. In vitro observations about REFERENCES the deleterious effects of a given variant on biochem- 1. Vanderver A, Prust M, Tonduti D, et al. Case definition ical functionalities do not necessarily translate to and classification of leukodystrophies and leukoencephalo- in vivo pathogenicity or strict clinical causality; a sec- pathies. Mol Genet Metab 2015;114:494–500. ond, possibly more relevant, genetic or epigenetic 2. Parikh S, Bernard G, Leventer RJ, et al. A clinical approach to the diagnosis of patients with leukodystro- factor should always be taken into account, especially phies and genetic leukoencephelopathies. Mol Genet for cases with previously unreported genotype/phenotype Metab 2015;114:501–515. correlations. This further complicates the pathogenic 3. Kevelam SH, Steenweg ME, Srivastava S, et al. Update on validation of rare genetic variants and their role in disease, leukodystrophies: a historical perspective and adapted def- underscoring the need for more robust and precise inition. Neuropediatrics 2016;47:349–354. validation methods. 4. Vanderver A, Simons C, Helman G, et al. Whole exome sequencing in patients with white matter abnormalities. – AUTHOR CONTRIBUTIONS Ann Neurol 2016;79:1031 1037. N.H.M. conceived and designed the study and experiments; performed 5. Yao P, Fox PL. Aminoacyl-tRNA synthetases in medicine – sample preparation; performed bioinformatics and statistical analysis of and disease. EMBO Mol Med 2013;5:332 343. sequence data; performed bioinformatics analysis of EARS2 variants; per- 6. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of formed mutation analysis and Sanger sequencing; performed fibroblast protein-coding genetic variation in 60,706 humans. cell culture; drafted the manuscript; and edited the manuscript. Nature 2016;536:285–291. A.N. performed cell culture, Western blot, and Seahorse assays and 7. Li H, Durbin R. Fast and accurate short read alignment analyzed data. AR performed the immortalization. B.L. performed sample with Burrows-Wheeler transform. Bioinformatics 2009; preparation and whole-exome sequencing. B.C. performed bioinformatics 25:1754–1760. analysis of sequence data. A.V. acquired and assessed the patient; 8. McKenna A, Hanna M, Banks E, et al. The Genome provided an additional patient for Sanger sequencing; and edited the man- Analysis Toolkit: a MapReduce framework for analyzing uscript. R.S. conceived and designed the study and experiments; acquired and assessed the patient; and edited the manuscript. D.G. conceived and next-generation DNA sequencing data. Genome Res – designed the study and experiments; analyzed the Seahorse data and 2010;20:1297 1303. Western blot data; and edited the manuscript. 9. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput ACKNOWLEDGMENT sequencing data. Nucleic Acids Res 2010;38:e164. The authors thank the study subjects for their participation and consent. 10. Kumar P, Henikoff S, Ng PC. Predicting the effects of They also thank Dr. Marjo van der Knaap for providing an additional coding non-synonymous variants on protein function patient for Sanger sequencing. using the SIFT algorithm. Nat Protoc 2009;4:1073–1081.

6 Neurology: Genetics 11. Adzhubei IA, Schmidt S, Peshkin L, et al. A method and 19. Danhauser K, Haack TB, Alhaddad B, et al. EARS2 server for predicting damaging missense mutations. Nat mutations cause fatal neonatal lactic acidosis, recurrent Methods 2010;7:248–249. hypoglycemia and agenesis of corpus callosum. Metab 12. Schwarz JM, Cooper DN, Schuelke M, Seelow D. Muta- Brain Dis 2016;31:717–721. tionTaster2: mutation prediction for the deep-sequencing 20. Richards S, Aziz N, Bale S, et al. Standards and guidelines age. Nat Methods 2014;11:361–362. for the interpretation of sequence variants: a joint consen- 13. Litzkas P, Jha KK, Ozer HL. Efficient transfer of cloned sus recommendation of the American College of Medical DNA into human diploid cells: protoplast fusion in sus- Genetics and Genomics and the Association for Molecular pension. Mol Cell Biol 1984;4:2549–2552. Pathology. Genet Med 2015;17:405–424. 14. Zhang JC, Sun L, Nie QH, et al. Down-regulation of 21. Livingston JH, Mayer J, Jenkinson E, et al. Leukoence- CXCR4 expression by SDF-KDEL in CD34(1) hemato- phalopathy with calcifications and cysts: a purely neuro- poietic stem cells: an anti-human immunodeficiency virus logical disorder distinct from coats plus. Neuropediatrics strategy. J Virol Methods 2009;161:30–37. 2014;45:175–182. 15. Invernizzi F, D’Amato I, Jensen PB, Ravaglia S, Zeviani 22. Corboy JR, Gault J, Kleinschmidt-DeMasters BK. An M, Tiranti V. Microscale oxygraphy reveals OXPHOS adult case of leukoencephalopathy with intracranial calci- impairment in MRC mutant cells. Mitochondrion 2012; fications and cysts. Neurology 2006;67:1890–1892. 12:328–335. 23. Oliveira R, Sommerville EW, Thompson K, et al. Lethal 16. Labrune P, Lacroix C, Goutieres F, et al. Extensive neonatal LTBL associated with biallelic EARS2 variants: brain calcifications, leukodystrophy, and formation of case report and review of the reported neuroradiological parenchymal cysts: a new progressive disorder due to features. JIMD Rep 2017;33:61–68. diffuse cerebral microangiopathy. Neurology 1996;46: 24. Jenkinson EM, Rodero MP, Kasher PR, et al. Mutations 1297–1301. in SNORD118 cause the cerebral microangiopathy leu- 17. Stephani C, Pfeifenbring S, Mohr A, Stadelmann C. Late- koencephalopathy with calcifications and cysts. Nat Genet onset leukoencephalopathy with cerebral calcifications and 2016;48:1185–1192. cysts: case report and review of the literature. BMC Neurol 25. Ghezzi D, Baruffini E, Haack TB, et al. Mutations of the 2016;16:19. mitochondrial-tRNA modifier MTO1 cause hypertrophic 18. Steenweg ME, Ghezzi D, Haack T, et al. Leukoencephal- cardiomyopathy and lactic acidosis. Am J Hum Genet opathy with thalamus and brainstem involvement and 2012;90:1079–1087. high lactate “LTBL” caused by EARS2 mutations. Brain 26. Wong LJ. Mitochondrial syndromes with leukoencepha- 2012;135:1387–1394. lopathies. Semin Neurol 2012;32:55–61.

Neurology: Genetics 7 ExACtly zero or once A clinically helpful guide to assessing genetic variants in mild epilepsies

Caitlin A. Bennett, BSc ABSTRACT (Hons) Objective: To assist the interpretation of genomic data for common epilepsies, we asked whether Slavé Petrovski, PhD variants implicated in mild epilepsies in autosomal dominant families are present in the general Karen L. Oliver, MSc population. Samuel F. Berkovic, MD, Methods: We studied 12 genes for the milder epilepsies and identified published variants with FRS strong segregation support (de novo germline mutation or $4 affected family members). These variants were checked in the Exome Aggregation Consortium (ExAC), a database of genetic variation in over 60,000 individuals. We subsequently evaluated variants in these epilepsy genes Correspondence to Dr. Berkovic: that lacked strong segregation support. To determine whether the findings in epilepsies were [email protected] representative of other diseases, we also assessed the presence of variants in other dominant neurologic disorders (e.g., CADASIL). Results: Published epilepsy variants with strong segregation support (n 5 65) were absent (n 5 61) or present once (n 5 4) in ExAC. By contrast, of 46 published epilepsy variants without strong segregation support, 8 occurred recurrently (2–186 times). Similarly, none of the 45 disease- associated variants from other neurologic disorders with strong segregation support occurred more than once in ExAC. Reanalysis using the larger ExAC V2 plus gnomAD reference cohort showed consistent results. Conclusions: Variants causing autosomal dominant epilepsies are ultra-rare in the general popu- lation. Variants observed more than once in ExAC were only found among reports without strong segregation support, suggesting that they may be benign. Clinicians are increasingly faced with the interpretation of genetic variants of unknown significance. These data illustrate that variants present more than once in ExAC are less likely to be pathogenic, reinforcing the valuable clinical role of ExAC. Neurol Genet 2017;3:e163; doi: 10.1212/NXG.0000000000000163

GLOSSARY ExAC 5 Exome Aggregation Consortium; GEFS1 5 genetic epilepsy with febrile seizures plus; VOUS 5 variants of unknown significance; WES 5 whole-exome sequencing.

Recent advances in DNA technologies have increased the influence that genomic data have in epilepsy clinical practice. Finding a genetic cause can have important implications for diagnosis and prognosis, treatment and genetic counseling, as well as the psychological and financial ben- efits associated with ending the diagnostic journey.1,2 Clinical laboratories are understandably conservative in ascribing pathogenicity; however, the elevated reporting of variants of unknown significance (VOUS) makes interpretation complex and renders follow-up efforts, such as segregation analyses and functional studies, impractical at a large scale. Novel missense variants are particularly challenging. Establishing the frequency of a VOUS in control populations is currently among the most reliable considerations in variant interpretation.

Supplemental data at Neurology.org/ng From the Department of Medicine (C.A.B., S.P., K.L.O., S.F.B.), Epilepsy Research Centre; and Department of Medicine (S.P.), Royal Melbourne Hospital, University of Melbourne, Victoria, Australia. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 validated de novo variants found in our list of 12 epilepsy genes, Table 1 Genes for autosomal dominant epilepsy syndromes among sporadic cases of “mild” epilepsies, were also included. We deliberately limited our search to the literature published prior to Epilepsy syndrome Gene 2015, as this literature can realistically be considered to precede Autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE) CHRNA4 the launch of the ExAC database. A separate review was con- CHRNB2 ducted for each of the 12 genes by searching the gene name and “epilepsy.” For genes associated with both milder and more CHRNA2 severe phenotypes, such as genetic epilepsy with febrile seizures Genetic epilepsy with febrile seizures plus (GEFS1) SCN1B plus (GEFS1) and Dravet syndrome with SCN1A, only variants SCN1A found among families or individuals with the milder phenotype were included. Variants reported exclusively among the epileptic KCNQ3 Benign familial neonatal epilepsy (BFNE) encephalopathies were excluded. KCNQ2 A second list of variants without strong segregation support

Benign familial neonatal-infantile epilepsy (BFNIE) SCN2A was generated. Lack of strong segregation support was defined as segregation data based on less than 4 affected individuals or 1 GABRG2 Childhood absence epilepsy, febrile seizures, and GEFS not performed at all. We hypothesized that this group of variants Autosomal dominant epilepsy with auditory features (ADEAF) LGI1 might be more enriched for background genetic variants in epi-

Genetic generalized epilepsy (GGE) SLC2A1 lepsy genes. To determine whether findings in mild epilepsies were reflec- DEPDC5 Familial focal epilepsy with variable foci (FFEVF) tive of other autosomal dominant neurologic diseases, we evalu- ated variants with strong segregation support from 45 published families with familial hemiplegic migraine, CADASIL, The Exome Aggregation Consortium autosomal dominant Alzheimer disease or frontotemporal (ExAC) database contains exome data for more dementia. The genes examined were ATP1A2, CACNA1A, than 60,000 individuals compiled at the Broad SCN1A, NOTCH3, APP, PSEN1, PSEN2, GRN, MAPT, and TARDBP1. 3 Institute. For de novo mutations affecting es- Finally, after the study was completed, the expanded ExAC tablished epilepsy genes, absence from ExAC is V2 plus gnomAD was released and the analysis was rerun. This widely accepted as one supportive argument for effectively doubled the size of the ExAC control cohort, with pathogenicity and this is particularly true for whole-exome sequencing (WES) and whole-genome sequencing data available on 123,136 individuals and 15,496 individuals, epileptic encephalopathies and most neurode- respectively.3 velopmental disorders.4 The question is whether this utility extends to the milder epi- RESULTS Sixty-five missense or predicted splice var- lepsies where negative selection may not have iants for the milder familial epilepsies with strong seg- been as strong. regation support were identified. Of these, 61 (94%) The milder presentations and the observa- were absent from ExAC and 4 (6%) were present in tion of reduced penetrance within dominant ExAC once (a single allele was reported) (figure; table families raise the possibility that some causa- e-1 at Neurology.org/ng). By contrast, of the 46 missense or splice variants tive variants might be present in the general contained in the same publications but lacking the population at low frequencies. Here, to assess strong segregation support, 13 (28%) were present how informative ExAC might be to ascribing in ExAC. Of these, 8 variants were observed among pathogenicity to inherited variants among 2–186 carriers (figure; table e-1). These 8 variants milder epilepsies, we evaluated the frequency are listed in table 2. in ExAC of published epilepsy variants, where Of the 45 missense variants with strong segrega- de novo status or segregation in large families tion support for other neurologic disorders, 43 supported pathogenicity. (96%) were absent from ExAC and 2 (4%) were pres- ent once (figure; table e-1). METHODS A list of 12 genes was generated for the “milder” Reanalysis with ExAC V2 and gnomAD datafiles epilepsies (table 1). We excluded PRRT2, as it is prone to align- showed that the epilepsy variants in our current study ment errors in next-generation sequencing. with strong segregation support are present no more Literature search. We attempted an exhaustive PubMed search than once, with the exception of 2 variants in SCN1B for missense and splice variants deemed causative for dominant associated with febrile seizures and GEFS1 which are familial epilepsies, published in English, on the basis of strong present 3 and 4 times (table e-1). segregation support, which we defined as prior linkage analysis For variants without strong segregation support, or familial segregation in at least 4 affected family members. their frequency in ExAC V2 approximately doubled, We note that segregation in 4 individuals is inadequate to support gene discovery but, in the context of an established gene for an as expected, compared with the initial analysis and 3 appropriate phenotype, such segregation provides considerable new variants were observed multiple times in ExAC support for the role of the variant in the particular family. Sanger V2 (table 2).

2 Neurology: Genetics segregation evidence for pathogenicity were present Figure Frequency distribution of allele counts in ExAC for variants in epilepsy families no more than once in the ExAC collection of over 60,000 individuals, illustrating that these variants are ultra-rare in the general population. This pattern was also seen in the variants found in other neurologic disorders that were reported based on strong segre- gation support. However, the literature-reported epilepsy variants without strong segregation support were more likely to be present in more than 1 carrier among the ExAC collection (figure), suggesting that many of these variants (table 2) are unlikely to be penetrant disease-causing alleles, but rather back- ground variation in established epilepsy genes or possibly risk alleles contributing to a more complex genetic architecture that is not currently well understood. WES in patients with epilepsy can be a powerful diagnostic tool, but interpretation of the data remains complex. In the field of epilepsy genetics, this com- plexity is highlighted by the likely false attribution Variants with strong segregation support, including confirmed de novo mutations among sporadic common epilepsies, are shown in blue (n 5 65); variants without strong segregation of pathogenicity to 2 genes, EFHC1 and SRPX2. support are shown in red (n 5 46); and variants from other neurologic disorders are shown in SRPX2 was deemed a pathogenic gene in 2006 in 2 green (n 5 45). ExAC 5 Exome Aggregation Consortium. families with epilepsy, speech dyspraxia, and intellec- tual impairment, including one from our centre.5 The DISCUSSION The interpretation of putative epi- key family was subsequently shown to have a GRIN2A lepsy genetic variants remains complex for researchers variant, and in fact the SRPX2 variant from this fam- and clinicians alike. Here, we assessed the allele count ily appears in ExAC 47 times. Similarly, the patho- of literature-reported epilepsy variants in the ExAC genicity of EFHC1 as a gene for juvenile myoclonic database and found that variants with strong epilepsy has been disputed.6

Table 2 Eleven variants without linkage analysis or strong segregation support, which are represented in ExAC V2 plus gnomAD more than once

Total allele count

ExAC V2 Gene Syndrome Ethnicity Variant ExAC plus gnomAD Ethnicities represented in ExAC V2 plus gnomAD

DEPDC5 Childhood epilepsy with French Canadian c.2591C.T (T864M)11 8 25 Latino, European (Non-Finnish), South Asian, and centrotemporal spikes other

DEPDC5 FFEVF Israeli c.1355C.T (A452V)12 42 118 European (Non-Finnish), Ashkenazi Jewish, and other

DEPDC5 FFEVF Australian c.3311C.T (S1104L)12 1 35 European (Non-Finnish), Latino, South Asian, and African

DEPDC5 FFEVF Australian c.3217A.C (S1073R)12 1 8 European (Non-Finnish)

KCNQ2 BFNIE Italian c.188715G.A13 3 7 European (Non-Finnish), South Asian, and Latino

KCNQ3 BFIE Italian c.2338C.T (R780C)13 13 25 European (Non-Finnish), European (Finnish), South Asian, and Latino

SCN1A FS Italian c.3924A.T (E1308D)14 91 177 European (Non-Finnish), African, European (Finnish), Latino, Ashkenazi Jewish, and other

SCN1A GEFS1 Italian c.1625 G.A (R542Q)14 186 453 Ashkenazi Jewish, European (Non-Finnish), South Asian, European (Finnish), African, Latino, East Asian, and other

SCN1A Panayiotopoulos syndrome Italian c.2369A.T (Y790C)14 1 6 European (Non-Finnish) and other

SLC2A1 GGE Australian c.668 G.A (R223Q)15 2 3 European (Non-Finnish) and South Asian

SLC2A1 GGE Australian, Israeli c.179C.T (T60M)15 3 9 African, European (Non-Finnish), South Asian, and other

Abbreviations: BFIE 5 benign familial infantile epilepsy; BFNIE 5 benign familial neonatal-infantile epilepsy; FFEVF 5 familial focal epilepsy with variable foci; FS 5 febrile seizures; GEFS15genetic epilepsy with febrile seizures plus; GGE 5 genetic generalized epilepsy.

Neurology: Genetics 3 Even for widely accepted epilepsy genes, the inter- The use of large reference cohorts as a clinical tool pretation of VOUS is complex. A large contribution is not without limitations. As the minor allele fre- to false-positive epilepsy variants comes from variants quency resolution improves with additional samples identified through “secondary” studies, wherein path- added to large reference cohorts (like ExAC), it is pos- ogenicity is largely inferred based on the knowledge of sible that clinically relevant variants for the milder prior disease-causing variants within the gene. These epilepsies might begin to appear more than once. variants are often not subject to the same rigorous Indeed, our results with ExAC were supported with evaluation of pathogenicity as variants from the orig- ExAC V2 plus gnomAD where the sample size was inal discovery,6 resulting in more permissive pathoge- doubled. The only exception was 2 variants in nicity assignments. Indeed, we know that epilepsy SCN1B associated with febrile seizures and GEFS1 genes carry background variation in the general pop- which are present 3 and 4 times (table e-1). A possible ulation. For example, the variants in table 2, which explanation for this slightly elevated frequency is that are in established epilepsy genes, were published as variants causing febrile seizures (a mild phenotype pathogenic and some of these are present in ExAC at often resolving with age) are expected to have a subtler relatively high frequencies, emphasizing the caution impact on reproductive fitness than variants causing necessary when assessing pathogenicity solely on the other common genetic generalized and nonacquired basis of a known epilepsy gene. This is one of the focal epilepsies, and this is also coupled with the high- most important considerations in the interpretation er rates of febrile seizures in the general population of results from clinical gene panels. than other common epilepsies. Thus, it is unlikely The issue of variant misclassification extends well that variants with a minor allele frequency beyond the field of epilepsy to other monogenic dis- .0.0005% (approximate current ExAC V2 resolu- eases. It has recently been demonstrated that each tion) have a major contribution to autosomal domi- ExAC participant carries, on average, ;54 variants nant forms of mild epilepsies. that have previously been reported as disease causing.3 Our study and the message presented here are inde- Of these, ;41 variants have a minor allele frequency pendently supported by our recent Epi4K study, where of .1%.3 It is therefore likely that many of these we found that when comparing to a control population, variants are either benign variants or contributing to the epilepsy risk signals identified in epilepsy genes complex inheritance. among large collections of epilepsy samples were driven The frequency of VOUS in control samples is by ultra-rare variants absent in ExAC.9 Our current thus an invaluable line of evidence in interpretation, study takes an alternative approach, comparing the fre- and the ExAC database offers a control cohort of quency in ExAC of variants with and without strong unprecedented size. ExAC is not enriched for path- segregation support, to reach the same conclusion. It is ogenic variants in genes that are commonly tested in important to note that although European and some a clinical diagnostics setting, thus supporting its util- other genetic ancestries are well represented in ExAC, ity in classifying variants.7 In a recent study, the many ethnic minorities are underrepresented or not ExAC database was found to have an excess of osten- represented at all, posing additional issues in interpreta- sibly pathogenic variants for prion disease but found tion when assessing data from a patient from an ethnic that the variants with the strongest independent evi- background not well represented in these databases.10 dence of pathogenicity were absent from ExAC, which is consistent with our findings.8 The authors AUTHOR CONTRIBUTIONS attributed the excess allele frequency for variants Caitlin A. Bennett: acquisition and analysis of data and drafted the man- with weak evidence for pathogenicity to the incor- uscript. Slavé Petrovski: study concept and design; interpretation of data; and revised the manuscript. Karen L. Oliver: interpretation of data and rect attribution of pathogenicity for certain variants revised the manuscript. Samuel F. Berkovic: study concept and design; or reduced penetrance of these alleles, an interpreta- study supervision; and revised the manuscript. tion shared here. Thus, the interrogation of ExAC provides STUDY FUNDING a highly useful way of assessing the relevance of Study funded by NHMRC Program Grant 1091593. VOUS for patients with epilepsy and is likely to be widely applicable to other neurologic disorders. DISCLOSURE We have shown that disease-causing variants in epi- C.A. Bennett reports no disclosures. S. Petrovski serves on the scientific advisory board and has interest in Pairnomix and serves on the editorial lepsy are very rare, and thus variants that are present board of Epilepsia. K.L. Oliver reports no disclosures. S.F. Berkovic has more than once in the current ExAC cohort are served on the scientific advisory boards for UCB Pharma Eisai and unlikely to contribute to disease in a large way. It Janssen-Cilag; serves on the editorial boards of Lancet Neurology and is important that the inverse is definitely not true Epileptic Disorders and the Advisory Board of Brain; and may accrue — ’ future revenue on pending patent WO61/010176: Therapeutic com- we cannot infer that a variant s absence in ExAC pound that relates to discovery of PCDH19 gene as the cause of familial is sufficient evidence of pathogenicity. epilepsy with mental retardation limited to females; is one of the

4 Neurology: Genetics inventors listed on a patent held by Bionomics Inc. on diagnostic testing population database: insights of relevance to variant clas- of using the SCN1A gene, WO2006/133508, and on pending patent sification. Genet Med 2015;18:850–854. WO61/010176: Therapeutic compound that relates to discovery of 8. Minikel EV, Vallabh SM, Lek M, et al. Quantifying prion PCDH19 gene as the cause of familial epilepsy with mental retardation disease penetrance using large population control cohorts. limited to females; has received speaker honoraria from UCB; has Sci Transl Med 2016;8:1–12. received unrestricted educational grants from UCB, Janssen-Cilag, and 9. Consortium Epi4K, Epilepsy Phenome/Genome Pro- Sanofi-Aventis; and receives/has received research support from the National Health and Medical Research Council of Australia and NINDS. ject. Ultra-rare genetic variation in common epilepsies: Go to Neurology.org/ng for full disclosure forms. a case-control sequencing study. Lancet Neurol 2017; 16:135–143. Received March 16, 2017. Accepted in final form April 13, 2017. 10. Petrovski S, Goldstein DB. Unequal representation of genetic variation across ancestry groups creates healthcare REFERENCES inequality in the application of precision medicine. 1. EpiPM Consortium. A roadmap for precision medicine in Genome Biol 2016;17:157. the epilepsies. Lancet Neurol 2015;14:1219–1228. 11. Martin C, Meloche C, Rioux MF, et al. A recurrent 2. Scheffer IE. Genetic testing in epilepsy: what should you mutation in DEPDC5 predisposes to focal epilepsies in be doing? Epilepsy Curr 2011;11:107–111. the French-Canadian population. Clin Genet 2014;86: 3. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of 570–574. protein-coding genetic variation in 60,706 humans. 12. Dibbens LM, de Vries B, Donatello S, et al. Mutations in Nature 2016;536:285–291. DEPDC5 cause familial focal epilepsy with variable foci. 4. Kosmicki J, Samocha K, Howrigan D, et al. Refining the Nat Genet 2013;45:546–551. role of de novo protein truncating variants in neurodeve- 13. Zara F, Specchio N, Striano P, et al. Genetic testing in lopmental disorders using population reference samples. benign familial epilepsies of the first year of life: clin- Nat Genet 2017;49:504–510. ical and diagnostic significance. Epilepsia 2013;54: 5. Roll P, Rudolf G, Pereira S, et al. SRPX2 mutations in 425–436. disorders of language cortex and cognition. Hum Mol 14. Orrico A, Galli L, Grosso S, et al. Mutational analysis of Genet 2006;15:1195–1207. the SCN1A, SCN1B and GABRG2 genes in 150 Italian 6. Pal D, Helbig I. Commentary: pathogenic EFHC1 muta- patients with idiopathic childhood epilepsies. Clin Genet tions are tolerated in healthy individuals dependent on 2009;75:579–581. reported ancestry. Epilepsia 2015;56:195–196. 15. Arsov T, Mullen S, Rogers S, et al. Glucose transporter 1 7. Song W, Gardner SA, Hovhannisyan H, et al. Exploring deficiency in the idiopathic generalized epilepsies. Ann the landscape of pathogenic genetic variation in the ExAC Neurol 2012;72:807–815.

Neurology: Genetics 5 Abnormal expression of homeobox genes and transthyretin in C9ORF72 expansion carriers

NiCole A. Finch, MS ABSTRACT Xue Wang, PhD Objective: We performed a genome-wide brain expression study to reveal the underpinnings of Matthew C. Baker, BSc diseases linked to a repeat expansion in 9 open reading frame 72 (C9ORF72). Michael G. Heckman, MS Tania F. Gendron, PhD Methods: The genome-wide expression profile was investigated in brain tissue obtained from Kevin F. Bieniek, PhD C9ORF72 expansion carriers (n 5 32), patients without this expansion (n 5 30), and controls Joanne Wuu, ScM (n 5 20). Using quantitative real-time PCR, findings were confirmed in our entire pathologic Mariely DeJesus- cohort of expansion carriers (n 5 56) as well as nonexpansion carriers (n 5 31) and controls Hernandez, BS (n 5 20). Patricia H. Brown, MS Results: Our findings were most profound in the cerebellum, where we identified 40 differentially Jeannie Chew, BA expressed genes, when comparing expansion carriers to patients without this expansion, includ- Karen R. Jansen-West, BS ing 22 genes that have a homeobox (e.g., HOX genes) and/or are located within the HOX gene Lillian M. Daughrity, BS cluster (top hit: homeobox A5 [HOXA5]). In addition to the upregulation of multiple homeobox Alexandra M. Nicholson, genes that play a vital role in neuronal development, we noticed an upregulation of transthyretin PhD Melissa E. Murray, PhD (TTR), an extracellular protein that is thought to be involved in neuroprotection. Pathway analysis Keith A. Josephs, MD aligned with these findings and revealed enrichment for processes involved in Joseph E. Parisi, MD (anatomic) development (e.g., organ morphogenesis). Additional analyses uncovered that HOXA5 David S. Knopman, MD and TTR levels are associated with C9ORF72 variant 2 levels as well as with intron-containing Ronald C. Petersen, MD transcript levels, and thus, disease-related changes in those transcripts may have triggered the Leonard Petrucelli, PhD upregulation of HOXA5 and TTR. Bradley F. Boeve, MD Conclusions: In conclusion, our identification of genes involved in developmental processes and Neill R. Graff-Radford, neuroprotection sheds light on potential compensatory mechanisms influencing the occurrence, MD presentation, and/or progression of C9ORF72-related diseases. Neurol Genet 2017;3:e161; doi: Yan W. Asmann, PhD 10.1212/NXG.0000000000000161 Dennis W. Dickson, MD Michael Benatar, MBChB, DPhil GLOSSARY 5 5 5 5 Robert Bowser, PhD ALS amyotrophic lateral sclerosis; FTD frontotemporal dementia; FTLD frontotemporal lobar degeneration; IQR interquartile range; MND 5 motor neuron disease. Kevin B. Boylan, MD * Rosa Rademakers, PhD Amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD) are part of a disease Marka van Blitterswijk, MD, PhD* continuum. Although ALS is the most common form of motor neuron disease (MND) and re- sults in progressive muscle weakness, FTD is a frequent cause of dementia and is associated with Correspondence to changes in personality, behavior, and language. A hexanucleotide repeat expansion in chromo- Dr. van Blitterswijk: some 9 open reading frame 72 (C9ORF72) is a major genetic cause of both diseases.1,2 Emerging [email protected] or evidence suggests that C9ORF72-related diseases are characterized by a loss of C9ORF72 Dr. Rademakers: 1 1 [email protected] expression, the formation of RNA foci with flawed RNA transcripts, and the generation of dipeptide repeat proteins aberrantly translated from the repeat expansion,3,4 with both RNA foci Supplemental data at Neurology.org/ng *These authors contributed equally to this work as co–last authors. From the Department of Neuroscience (N.A.F., M.C.B., T.F.G., K.F.B., M.D.-H., P.H.B., J.C., K.R.J.-W., L.M.D., A.M.N., M.E.M., L.P., D. W.D., R.R., M.v.B.), Department of Health Sciences Research (X.W., Y.W.A.), Department of Neurology (N.R.G.-R., K.B.B.), Division of Biomedical Statistics and Informatics (M.G.H.), Mayo Clinic, Jacksonville, FL; Department of Neurology (J.W., M.B.), University of Miami, FL; Department of Neurology (K.A.J., J.E.P., D.S.K., R.C.P., B.F.B.), Mayo Clinic, Rochester, MN; and Divisions of Neurology and Neurobiology (R.B.), Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, AZ. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the NIH. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 and dipeptide repeat proteins potentially con- already been determined.10–12 To examine the cerebellar TTR 5 tributing to disease by compromising nucleo- protein, Western blots (n 10) and immunohistochemistry (n 5 13) were performed on a representative subset of samples 5–7 cytoplasmic transport. (table e-1 at Neurology.org/ng). Next, an immunoassay was used In our quest to increase our understanding to evaluate TTR protein levels in the CSF, studying 2 indepen- of C9ORF72-related diseases, we assessed the dent clinical cohorts obtained at either the Mayo Clinic (n 5 67) 5 genome-wide expression profile in brain tissue or the University of Miami (n 40, table e-1). obtained from the Mayo Clinic Florida Brain Standard protocol approvals, registrations, and patient Bank (n 5 82). Of interest, in C9ORF72 consents. All participants agreed to participate in the study, and biological samples were obtained after informed consent with eth- expansion carriers, we discovered an upregula- ical committee approval from the respective institutions. tion of genes involved in developmental pro- cesses and neuroprotection, particularly in the Methods and statistical analysis. To examine the genome- wide expression pattern, Whole-Genome DASL HT assays (Il- cerebellum, a region without substantial neu- lumina, San Diego, CA) were used, which were processed by the ronal loss that demonstrates pathologic hall- Mayo Clinic Core Facility. Validation was performed with marks of C9ORF72-related diseases,3,8 and in quantitative real-time PCR using TaqMan assays (Life Technologies, Carlsbad, CA). Western blotting was used to which abnormalities associate with neuropath- evaluate cerebellar TTR protein levels, complemented with 9,10 ologic and clinical phenotypes. Such find- immunohistochemistry to assess the presence of potential TTR ings may point toward mechanisms that could protein aggregates. Meso Scale Discovery (MSD, Rockville, MD) compensate for the harmful effects of electrochemiluminescence detection technology was used to establish a sandwich immunoassay for TTR. Cell culture ex- C9ORF72 repeat expansions. periments were then performed in an attempt to clarify under- lying mechanisms. In U251 and HepG2 cells, a loss of C9ORF72 METHODS Participant selection. From the Mayo Clinic expression was mimicked with small interfering RNAs (siRNAs, Florida Brain Bank, participants were selected for our genome- Dharmacon, Lafayette, CO), and in addition, the effect of full- wide expression study: patients with a pathologic diagnosis of length C9ORF72 and the repeat expansion itself was examined by frontotemporal lobar degeneration (FTLD) and/or MND who transfecting cells with expression vectors.13 A detailed description 5 harbored C9ORF72 repeat expansions (n 32), patients with of our methods and statistical analysis is provided in the sup- 5 FTLD and/or MND without repeat expansions (n 30), and plemental data. controls without neurologic diseases (n 5 20, table 1). To con- firm the observed upregulation of homeobox A5 (HOXA5) and RESULTS Upregulation of homeobox genes and TTR transthyretin (TTR), quantitative real-time PCR was performed, in C9ORF72 expansion carriers. We performed when expanding investigations to our entire pathologic cohort of C9ORF72 expansion carriers for whom brain tissue was available a genome-wide expression study in the cerebellum (n 5 56) as well as FTLD and/or MND patients without an and frontal cortex to identify genes involved in expansion (n 5 31) and controls without any neurologic disease C9ORF72-related diseases. First, we compared pa- (n 5 20).11 In this cohort, C9ORF72 transcript levels, the length tients with or without a repeat expansion in of the repeat expansion, and dipeptide repeat protein levels had C9ORF72. Although participants included in those groups are both affected by neurodegenerative dis- eases, this enabled us to find C9ORF72-specific dif- Table 1 Participant characteristics ferences. Second, we compared expansion carriers

C9Plus cohort C9Minus cohort Control cohort with controls without neurodegenerative diseases, Cohort/variable (n 5 32) (n 5 30) (n 5 20) allowing the detection of more general differences Genome-wide expression that could, theoretically, be due to the presence of

Sex, male 20 (63) 12 (40) 7 (35) a neurodegenerative disease. In the cerebellum, when comparing expansion car- Age at death, y 63.7 (58.4–71.7) 75.0 (64.0–81.8) 87.5 (81.8–93.0) riers to patients without expansions, we detected 40 RIN cerebellum (value) 9.4 (9.2–9.6) 9.2 (8.7–9.4) 9.3 (8.5–9.4) differentially expressed genes (table e-2). Generation – – – RIN frontal cortex (value) 9.0 (8.5 9.6) 9.1 (8.6 9.5) 8.9 (8.6 9.2) of a heat map of those genes revealed that expansion Diagnosis carriers generally cluster together (figure 1). Of interest, FTLD 12 (38) 10 (33) — our list of differentially expressed genes contained 22

FTLD/MND 10 (31) 10 (33) — genes that have a homeobox (e.g., HOX genes) and/or

MND 10 (31) 10 (33) — are located within the HOX gene cluster (table e-2). In addition to the upregulation of multiple homeobox Other ——— genes (top hit: HOXA5)thatplayavitalroleinneu- Abbreviations: FTLD 5 frontotemporal lobar degeneration; IQR 5 interquartile range; ronal development,14 we noticed a cerebellar upregula- 5 MND motor neuron disease. tion of TTR (table e-2), an extracellular protein that is Data are sample median (IQR) or n (%). Information was obtained for patients with (C9Plus) 15–19 and without (C9Minus) expansions in C9ORF72, as well as from controls. This study was thought to be involved in neuroprotection. We performed in the cerebellum and frontal cortex. then compared expansion carriers with controls and

2 Neurology: Genetics discovered 1,575 differentially expressed genes in the In the frontal cortex, a comparison between pa- cerebellum (table e-2). Again, our heat map showed tients with or without repeat expansions resulted in that expansion carriers tend to cluster together (figure the detection of 3 differentially expressed genes: 1). Of interest, our new list contained 37 of the 40 HOXA5, C9ORF72, and POU class 4 homeobox 2 (93%) genes we identified previously (table e-2), (POU4F2; table e-2). We also compared expansion including homeobox genes and TTR. carriers with controls and revealed 679 differentially We also performed gene ontology analysis and expressed genes, including C9ORF72 and TTR (table observed an enrichment for pathways involved in e-2). Again, enrichment was observed for pathways the regulation of (anatomic) development, which involved in developmental processes (table e-3). was most profound when comparing expansion car- riers with disease controls (e.g., organ morphogenesis, Associations of C9ORF72 transcripts with HOXA5 and TTR pattern specification process, regionalization, and transcripts in our overall cohort. In previously published skeletal system development, table e-3), but which studies, we investigated the levels of known C9ORF72 was also seen when comparing expansion carriers with transcript variants (variant 1 [NM_145005.6], variant 2 controls (table e-3). [NM_018325.4], and variant 3 [NM_001256054.2])

Figure 1 Expression of homeobox genes and transthyretin

C9Plus 5 patients with C9ORF72 repeat expansions; C9Minus 5 patients without C9ORF72 repeat expansions; and control 5 controls without neurologic diseases. Heat map plots of intensity values of differentially expressed genes are displayed for the cerebellum, when comparing C9ORF72 expansion carriers with patients without expansions (A, fold change above 1.2), and when comparing C9ORF72 expansion carriers with controls (A, fold change above 2.5 [more stringent to allow visualization]). Rows (samples) and columns (genes) are grouped by hierarchical clustering using Manhattan distance measure- ments; low intensities are shown as blue, and high intensities are shown as red. In our expression cohort, cerebellar expression levels of homeobox A5 (HOXA5; B) and transthyretin (TTR; C) are increased in patients with C9ORF72 repeat expansions as compared to patients without expansions or to controls. The median is represented by a solid line, and each box spans the 25th percentile to the 75th percentile (interquartile range). A Western blot is shown demonstrating higher cerebellar TTR protein levels in expansion carriers (1) than in patients without this expansion (2, D). Quantification of Western blot samples confirmed the cerebellar increase of TTR protein levels in patients with a repeat expansion as compared to patients without this expansion (E), which is displayed in a bar graph that represents the mean of the relative normalized TTR protein with the SEM, using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the loading control.

Neurology: Genetics 3 as well as 2 intronic regions (1 upstream of the repeat

0.050 expansion [intron 1a] and 1 downstream of the repeat Value

, 11

p expansion [intron 1b]). In this study, we examined the p transcripts same cohort to validate our findings related to HOXA5 and TTR, which demonstrated that their cerebellar lev- 6.92) 0.40 1.54) 0.30 1.30) 0.90 HOXA5 – – – els are indeed higher in C9ORF72 expansion carriers than in (disease) controls (table 2, figure 1). In addition, inthefrontalcortex,weshowedthatTTR levels are elevated in patients with a C9ORF72 repeat expansion ) and glyceraldehyde-3-phosphate as compared to controls (table 2). Of note, we also 21.18) 1.00 (0.43 3.55) 1.00 (0.34 1.30) 1.00 (0.66 performed a sensitivity analysis to assess whether our RPLP0 – – – findings could have been biased by differences in age at death. It is important that similar findings were observed when restricting our analysis to a subset of participants with a comparable age at death (not shown). In addition, given the low levels of TTR and HOXA5, were too low for reliable detection using quantitative real-time

Value C9Minus Control especially in (disease) controls, we also validated their p upregulation in a subset of participants using other tech-

HOXA5 niques, including digital molecular barcoding (not 6.92) 0.01 2.01 (0.35 1.54) 7.69e-05 1.68 (0.27 1.30) 5.41e-08 0.91 (0.55 shown) and previously published RNA sequencing data – – – 0.017 considered significant after Bonferroni correction for 3 comparisons). (figure e-1).20 , p

0.025 considered significant after Bonferroni correction, frontal cortex: The vast amount of C9ORF72 expression data , , as well as for controls. In the cerebellum, 2 tests were performed ( p available for this cohort then allowed us to determine whether the levels of C9ORF72 transcripts were asso- ciated with the levels of HOXA5 and TTR. In our 33.01) 1.00 (0.66 C9ORF72 – 15.42) 1.00 (0.43 26.86) 1.00 (0.34 overall cohort (expansion carriers, disease controls, – – and controls), lower cerebellar levels of C9ORF72 transcript variant 2 were associated with higher cere- bellar levels of both HOXA5 (r 520.60, p 5 3.21e- 09, Spearman test of correlation) and TTR (r 5 20.47, p 5 2.21e-06, Spearman test of correlation,

Value C9Plus Control table 3), which is not surprising given the fact that p expansion carriers demonstrate decreased levels of C9ORF72 transcript variant 2.11 More excitingly, 21.18) 0.19 5.69 (1.88 3.55) 4.09e-05 8.48 (2.69 1.30) 8.22e-10 10.98 (4.43 we noticed that higher cerebellar levels of intron- – – – containing transcripts (both intron 1a and intron 1b) are associated with higher cerebellar levels of HOXA5 transcripts (intron 1a: r 5 0.43, p 5 values below 0.050 were considered significant.

p 6.47e-05, intron 1b: r 5 0.36, p 5 0.0008, Spear- man test of correlation, table 3). In the frontal cortex, 33.01) 0.91 (0.55 transcript levels are normalized to the geometric mean of endogenous control genes ribosomal protein, large, P0 ( – transcripts using TaqMan assays in the expression cohort 15.42) 2.01 (0.35 26.86) 1.68 (0.27 lower C9ORF72 variant 2 levels were also associated – – 52 5 TTR with higher TTR levels (r 0.28, p 0.006, TTR

and Spearman test of correlation, table 3). and C9Plus vs C9Minus cohortC9Plus C9Minus C9Plus vs control cohort C9Minus vs control cohort

transcripts), and thus, Associations of C9ORF72 transcripts with HOXA5 and HOXA5 a HOXA5

TTR TTR transcripts in expansion carriers. Because we were value; Value able to validate our findings related to HOXA5 and values below 0.025 were considered significant after Bonferroni correction; in the fontal cortex, the levels of p 0.04 5.69 (1.88 3.01e-06 8.48 (2.69 p 1.24e-12 10.98 (4.43

p TTR and detect significant associations with specific C9ORF72 transcripts in our overall cohort, we then ). Information was obtained for patients with (C9Plus) and without (C9Minus) expansions in evaluated the presence of any potential associations interquartile range. TTR TTR HOXA5

5 within our cohort of C9ORF72 expansion carriers. In GAPDH Expression studies of the cerebellum, increased levels of total C9ORF72 transcripts were associated with increased HOXA5 transcripts), and thus, transcripts (r 5 0.51, p 5 8.86e-05, Spearman

TTR test of correlation, table 4), most prominently in pa- Frontal cortex TaqMan Group Cerebellum Table 2 A Kruskal-Wallis rank-sum test was performed to determine whether expression levels differed between groups (cerebellum: considered significant); when significant differences were detected, a Wilcoxon rank-sum test was used for pairwise comparisons ( Similar findings were obtained when normalizing to neuronal markers (not shown for simplicity). PCR and only 1 test was performed ( Abbreviation: IQR Data are sample median (IQR) or a dehydrogenase ( and tients with a pathologic diagnosis of FTLD (r 5 0.65,

4 Neurology: Genetics potential associations with poly(GP) and poly(GA) Table 3 Associations of HOXA5 and TTR transcripts with C9ORF72 transcripts in the overall cohort proteins, which can form abundant inclusions in the neocortical regions, hippocampus, thalamus, Overall and cerebellum.3,4,9,21–24 Although no significant asso- ciations were detected for TTR, we did observe an TaqMan Group Association Spearman r (95% CI) p Value association for HOXA5: higher levels of dipeptide Cerebellum HOXA5 Total 20.21 (20.40 to 0.02) 0.07 repeat proteins were associated with higher levels of Variant 1 20.19 (20.39 to 0.03) 0.09 HOXA5 (poly[GP]: r 5 0.52, p 5 0.0002, poly[GA]: Variant 2 20.60 (20.73 to 20.42) 3.21e-09 r 5 0.56, p 5 3.89e-05, Spearman test of correlation, Variant 3 0.06 (20.14 to 0.27) 0.56 table 4).

Intron 1a 0.43 (0.23 to 0.60) 6.47e-05 In the frontal cortex, we noted a trend between C9ORF72 variant 2 and TTR (r 520.39, p 5 Intron 1b 0.36 (0.16 to 0.54) 0.0008 0.004, Spearman test of correlation), particularly in TTR Total 20.19 (20.39 to 0.02) 0.06 the subset of patients with a pathologic diagnosis of Variant 1 20.20 (20.40 to 0.003) 0.05 FTLD (r 520.62, p 5 0.002, Spearman test of Variant 2 20.47 (20.62 to 20.28) 2.21e-06 correlation, table 4). Variant 3 20.03 (20.24 to 0.18) 0.78 In both brain regions, we did not detect significant

Intron 1a 0.23 (0.03 to 0.42) 0.03 associations with other variables, such as expansion

Intron 1b 0.23 (0.04 to 0.41) 0.03 size (table 4), disease subgroup, sex, age at onset, age at death, or survival after onset (not shown). Frontal cortex TTR Total 20.15 (20.35 to 0.07) 0.16

2 2 Variant 1 0.06 ( 0.25 to 0.14) 0.58 Cerebellar changes in TTR transcripts are reflected by Variant 2 20.28 (20.47 to 20.08) 0.006 changes in protein levels. Given the fact that TTR is an Variant 3 20.17 (20.35 to 0.03) 0.09 extracellular protein, we determined whether changes

Intron 1a 0.23 (0.02 to 0.42) 0.03 in RNA levels were reflected by changes in protein levels, which could indicate that TTR may serve as Intron 1b 0.14 (20.07 to 0.34) 0.19 a biomarker for C9ORF72-related diseases. Because Data are Spearman correlation coefficient r (95% confidence interval [CI]) or p value; our findings were most profound in the cerebellum, HOXA5 and TTR transcript levels are normalized to the geometric mean of endogenous control genes ribosomal protein, large, P0 (RPLP0), and glyceraldehyde-3-phosphate dehy- we extracted protein from this neuroanatomic region drogenase (GAPDH). In total, in our overall cohort (expansion carriers, disease controls, and and performed Western blots. As expected, we de- controls), we examined 6 different associations (total C9ORF72 transcripts, C9ORF72 tected a significant increase in cerebellar TTR protein C9ORF72 C9ORF72 transcript variant 1, transcript variant 2, transcript variant 3, intron levels in patients with a repeat expansion (mean – C9ORF72 – C9ORF72 1a containing transcripts, and intron 1b containing transcripts) for 6 each outcome, and thus, p values below 0.0083 were considered significant after Bon- 174% 34%) as compared to patients without ferroni correction. A Spearman test of correlation was used (p , 0.0083 considered sig- a repeat expansion (mean 100% 6 18%, p , 0.05, nificant after Bonferroni correction). Similar findings were obtained when normalizing to 2-sample t test, figure 1). We also performed neuronal markers (not shown for simplicity). immunohistochemistry to examine whether an aggregated form of the TTR protein was present in p 5 0.0009, Spearman test of correlation). Of inter- the cerebellum because TTR protein aggregates have est, we also noticed that elevated levels of intron- been reported in other diseases, such as familial containing transcripts were associated with elevated amyloid polyneuropathy.25 We observed diffuse levels of HOXA5 or TTR (table 4). For HOXA5, this cytoplasmic TTR staining in pyramidal neurons and association (intron 1a: r 5 0.60, p 5 1.61e-06, Purkinje cells, and in the neuropil; however, no TTR intron 1b: r 5 0.54, p 5 2.75e-05, Spearman test of deposits were detected similar to those seen in pa- correlation, table 4) was driven by patients with tients with TTR amyloidosis (not shown). a pathologic diagnosis of FTLD (intron 1a: r 5 0.72, To further evaluate TTR as a potential biomarker, p 5 0.0001, intron 1b: r 5 0.59, p 5 0.003, we determined its protein levels in the CSF. In our Spearman test of correlation). For TTR, however, the first cohort, the median TTR protein level in expan- association (intron 1a: r 5 0.40, p 5 0.003, intron sion carriers was 15.5 mg/mL (interquartile range 1b: r 5 0.43, p 5 0.001, Spearman test of correla- [IQR] 13.7–17.6) and in the remaining participants tion, table 4) was most profound in patients with 16.3 mg/mL (IQR 14.5–17.7), which was not signif- a pathologic diagnosis of MND (intron 1a: r 5 0.83, icantly different (p 5 0.29, Wilcoxon rank-sum test). p 5 0.0002, intron 1b: r 5 0.88, p 5 3.78e-05, Our second cohort revealed a median TTR protein Spearman test of correlation). level of 12.5 mg/mL in expansion carriers (IQR 11.0– Because we previously discovered associations 12.6) and 12.3 mg/mL in other participants (IQR between dipeptide repeat proteins and intron- 11.8–14.4); again, this difference did not reach sta- containing transcripts,11 we subsequently evaluated tistical significance (p 5 0.58, Wilcoxon rank-sum

Neurology: Genetics 5 6

Table 4 Associations of HOXA5 and TTR transcripts with C9ORF72 transcripts, expansion size, and dipeptide repeat proteins in expansion carriers

C9Plus cohort FTLD cohort FTLD/MND cohort MND cohort

TaqMan Group Association Spearman r (95% CI) p Value Spearman r (95% CI) p Value Spearman r (95% CI) p Value Spearman r (95% CI) p Value

Cerebellum HOXA5 Total 0.51 (0.29 to 0.68) 8.86e-05 0.65 (0.32 to 0.83) 0.0009 0.20 (20.40 to 0.65) 0.47 0.43 (20.19 to 0.86) 0.13

Variant 1 0.33 (0.08 to 0.54) 0.01 0.50 (0.11 to 0.78) 0.01 20.12 (20.68 to 0.44) 0.67 0.16 (20.45 to 0.68) 0.59

Variant 2 20.13 (20.40 to 0.15) 0.33 20.07 (20.51 to 0.41) 0.77 20.19 (20.68 to 0.36) 0.50 20.08 (20.67 to 0.50) 0.80 erlg:Genetics Neurology: Variant 3 0.38 (0.13 to 0.59) 0.005 0.35 (20.12 to 0.69) 0.10 0.41 (20.16 to 0.81) 0.12 0.22 (20.42 to 0.77) 0.44

Intron 1a 0.60 (0.37 to 0.77) 1.61e-06 0.72 (0.39 to 0.89) 0.0001 0.56 (20.02 to 0.91) 0.03 0.41 (20.21 to 0.82) 0.14

Intron 1b 0.54 (0.31 to 0.71) 2.75e-05 0.59 (0.20 to 0.82) 0.003 0.65 (0.17 to 0.88) 0.009 0.42 (20.13 to 0.80) 0.14

C9ORF72 expansion size 20.17 (20.42 to 0.10) 0.24 20.09 (20.53 to 0.36) 0.68 20.42 (20.70 to 0.02) 0.12 20.18 (20.64 to 0.44) 0.55

Poly(GP) 0.52 (0.26 to 0.72) 0.0002 0.32 (20.12 to 0.71) 0.15 0.50 (20.15 to 0.91) 0.08 0.45 (20.09 to 0.84) 0.10

Poly(GA) 0.56 (0.33–0.73) 3.89e-05 0.60 (0.20 to 0.85) 0.004 0.41 (20.26 to 0.84) 0.17 0.09 (20.59 to 0.62) 0.74

TTR Total 0.25 (20.02 to 0.50) 0.07 0.44 (0.08 to 0.70) 0.03 20.02 (20.62 to 0.60) 0.95 0.35 (20.23 to 0.83) 0.22

Variant 1 0.07 (20.23 to 0.36) 0.60 0.29 (20.13 to 0.62) 0.18 20.18 (20.74 to 0.47) 0.53 0.03 (20.52 to 0.61) 0.91

Variant 2 20.17 (20.43 to 0.12) 0.21 20.06 (20.53 to 0.42) 0.77 20.24 (20.74 to 0.36) 0.40 20.37 (20.78 to 0.21) 0.19

Variant 3 0.11 (20.19 to 0.39) 0.43 0.10 (20.34 to 0.52) 0.65 0.05 (20.61 to 0.67) 0.85 0.24 (20.36 to 0.75) 0.41

Intron 1a 0.40 (0.15 to 0.59) 0.003 0.27 (20.16 to 0.66) 0.22 0.28 (20.31 to 0.73) 0.31 0.83 (0.54 to 0.96) 0.0002

Intron 1b 0.43 (0.18 to 0.63) 0.001 0.37 (20.04 to 0.67) 0.08 0.33 (20.32 to 0.80) 0.23 0.88 (0.62 to 0.97) 3.78e-05

C9ORF72 expansion size 0.01 (20.29 to 0.30) 0.93 0.05 (20.38 to 0.45) 0.83 0.12 (20.57 to 0.66) 0.67 0.01 (20.66 to 0.64) 0.96

Poly(GP) 0.09 (20.19 to 0.36) 0.55 20.11 (20.49 to 0.30) 0.62 0.19 (20.43 to 0.66) 0.54 0.38 (20.15 to 0.76) 0.17

Poly(GA) 20.02 (20.29 to 0.26) 0.91 0.15 (20.26 to 0.54) 0.50 0.00 (20.60 to 0.63) 0.99 20.002 (20.53 to 0.57) 1.00

Frontal cortex TTR Total 20.19 (20.47 to 0.11) 0.17 20.05 (20.51 to 0.43) 0.84 20.33 (20.79 to 0.24) 0.26 20.39 (20.86 to 0.30) 0.19

Variant 1 20.03 (20.32 to 0.26) 0.82 20.09 (20.40 to 0.56) 0.69 20.12 (20.58 to 0.40) 0.70 0.01 (20.61 to 0.60) 0.96

Variant 2 20.39 (20.59 to 20.14) 0.004 20.62 (20.80 to 20.31) 0.002 20.51 (20.90 to 0.07) 0.06 20.11 (20.68 to 0.53) 0.73

Variant 3 20.03 (20.31 to 0.26) 0.81 0.32 (20.10 to 0.67) 0.14 20.33 (20.79 to 0.30) 0.26 0.08 (20.61 to 0.68) 0.80

Intron 1a 0.18 (20.08 to 0.41) 0.21 0.31 (20.14 to 0.70) 0.16 0.14 (20.39 to 0.58) 0.63 0.21 (20.46 to 0.75) 0.49

Intron 1b 20.02 (20.31 to 0.30) 0.88 20.08 (20.52 to 0.47) 0.72 20.08 (20.71 to 0.51) 0.79 0.22 (20.54 to 0.80) 0.47

C9ORF72 expansion size 0.02 (20.28 to 0.30) 0.91 20.25 (20.62 to 0.19) 0.26 0.49 (20.08 to 0.83) 0.07 20.19 (20.80 to 0.48) 0.55

Poly(GP) 0.15 (20.15 to 0.42) 0.32 0.40 (20.10 to 0.71) 0.08 0.43 (20.14 to 0.88) 0.12 0.05 (20.63 to 0.65) 0.85

Abbreviations: FTLD 5 frontotemporal lobar degeneration; MND 5 motor neuron disease. Data are Spearman correlation coefficient r (95% confidence interval [CI]) or p value; HOXA5 and TTR transcript levels are normalized to the geometric mean of endogenous control genes ribosomal protein, large, P0 (RPLP0), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). In total, within our cohort of C9ORF72 expansion carriers, we examined 16 different associations (total C9ORF72 transcripts, C9ORF72 transcript variant 1, C9ORF72 transcript variant 2, C9ORF72 transcript variant 3, intron 1a–containing C9ORF72 transcripts, intron 1b–containing C9ORF72 transcripts, C9ORF72 repeat length, poly[GP] levels, poly[GA] levels, disease subgroup, sex, age at onset, age at death, and survival after onset [using 3 different cutoff points]) for each outcome, and thus, p values below 0.0031 were considered significant after Bonferroni correction; in this table, only 9 of those 16 associations are displayed (total C9ORF72 transcripts, C9ORF72 transcript variant 1, C9ORF72 transcript variant 2, C9ORF72 transcript variant 3, intron 1a–containing C9ORF72 transcripts, intron 1b–containing C9ORF72 transcripts, C9ORF72 repeat length, poly[GP] levels, and poly[GA] levels). A Spearman test of correlation was used (p , 0.0031 considered significant after Bonferroni correction). Similar findings were obtained when normalizing to neuronal markers (not shown for simplicity). test). Of note, similar findings were obtained when The upregulation of multiple homeobox genes adjusting for possible confounders, when removing and TTR was further substantiated by the results of outliers, and when restricting our analysis to specific our gene ontology analysis that revealed enrichment (sub)groups (e.g., symptomatic participants). for developmental processes. The observed enrich- Loss of C9ORF72 expression increases HOXA5 and TTR ment for developmental processes is not surprising transcripts. Next, we performed cell culture experi- given the function of homeobox genes and TTR in 14–19 ments to determine which C9ORF72-related disease neuronal development and neuroprotection. characteristics might drive the specific upregulation Homeobox genes, for example, are involved in neu- of HOXA5 and TTR. Of interest, we observed an ronal specification and target connectivity; they play increase of HOXA5 in response to a knock-down of a key role in the identity, organization, and peripheral 14 total C9ORF72 transcripts (mean 193% 6 3%, p 5 connectivity of motor neuron subtypes. TTR is 0.0002, 2-sample t test). A comparable effect was important for the transportation of thyroid hormones seen on TTR (mean 129% 6 6%, p 5 0.01, 2- (thyroxine [T4]) and retinol (vitamin A), and in addi- sample t test) after knocking down total C9ORF72 tion, it is thought to participate in behavior, mainte- transcripts. Given our significant findings related to nance of normal cognitive processes during aging, 26 C9ORF72 transcript variant 2, we then targeted neuropeptide processing, and nerve regeneration. variant 2. It is important that knocking down of Of interest, homeobox genes and TTR are linked. variant 2 was sufficient to increase HOXA5 levels For instance, an active metabolite of vitamin A (all- (mean 274% 6 12%, p 5 0.005, 2-sample t test). trans retinoic acid) functions in maintenance of many For TTR,targetingC9ORF72 variant 2 also affected processes (e.g., brain function) and is essential for TTR levels (mean 293% 6 37%, p 5 0.03, 2- limb and organ development through homeobox – 27 sample t test). Overexpression of full-length gene mediated mechanisms. C9ORF72 or expression of 66 GGGGCC repeats, Although the relative differences we observed in however, did not seem to affect HOXA5 or TTR (not HOXA5 and TTR are a reflection of small absolute shown). Consequently, it seems that a loss of differences, the fact that their levels are barely detect- C9ORF72 expression triggers an increase in HOXA5 able in (disease) controls raises the possibility that and TTR. their expression may have been (re)activated in the adult brain. Such a (re)activation could be specific to DISCUSSION We set out to increase our under- C9ORF72-related diseases and might be driven by standing of C9ORF72-related diseases and revealed certain mechanisms underlying those diseases. It elevated levels of multiple homeobox genes (e.g., is currently unknown where the elevated HOXA5 HOXA5) and TTR. Of interest, our findings were and TTR levels are coming from; a change in cell- most profound in the cerebellum, a region without type composition, invading progenitor cells, and substantial neuronal loss that demonstrates patho- cell-autonomous effects need to be taken into logic characteristics of diseases linked to a repeat consideration. expansion in C9ORF72.3,8 In fact, in this neuroan- One of the pathologic hallmarks of C9ORF72- atomic region, associations with neuropathologic related diseases is a reduction in C9ORF72 expression and clinical features of the disease have already been levels.1 Because the most prominent decrease has reported. It has been shown, for instance, that been reported for C9ORF72 variant 2 transcripts,11 dipeptide repeat protein levels are lower in patho- one could speculate that levels of this transcript might logically diagnosed patients with MND as compared be associated with levels of HOXA5 or TTR. Indeed, to patients with FTLD,9,10 and that they are associ- in our overall cohort, we demonstrated that lower ated with the cognitive score of clinically diagnosed levels of C9ORF72 variant 2 transcripts were associ- patients with ALS.10 Moreover, an association ated with higher levels of HOXA5 and TTR between dipeptide repeat proteins and intron- transcripts. We also detected associations with containing transcripts has been described in the intron-containing transcripts; in our expansion car- cerebellum, indicating that transcripts containing riers, for instance, higher levels of intron-containing theentirefirstintronmay serve as templates for transcripts were associated with HOXA5 transcript repeat-associated non-ATG translation.11 In addi- levels, especially in patients with a pathologic diagno- tion, extensive alternative splicing and poly- sis of FTLD. In addition, HOXA5 transcript levels adenylation defects have been reported in the were associated with dipeptide repeat proteins, which cerebellum of C9ORF72 expansion carriers.20 The is in agreement with reports that describe associations fact that our findings were most prominent in between intron-containing transcripts and dipeptide the cerebellum emphasizes that this region may have repeat proteins.11 For TTR, we also observed a corre- been underappreciated and could play an important lation with intron-containing transcripts, but this role in C9ORF72-related diseases. association was most profound in patients with

Neurology: Genetics 7 a pathologic diagnosis of MND; no significant asso- carriers. Our findings may point to the presence of ciations were observed with dipeptide repeat proteins. compensatory mechanisms aiming to mitigate the Thus, although some associations were shared progression of C9ORF72-related diseases. between HOXA5 and TTR, others differed. We emphasize that in our study, as in any obser- AUTHOR CONTRIBUTIONS vational study, performing association analysis NiCole A. Finch, Matthew C. Baker, Tania F. Gendron, Kevin F. Bien- iek, Mariely DeJesus-Hernandez, Patricia H. Brown, Jeannie Chew, between 2 variables is not intended to provide infor- Karen R. Jansen-West, Lillian M. Daughrity, Alexandra M. Nicholson, mation about possible mechanisms, but rather is in- Melissa E. Murray, and Robert Bowser: acquisition of data, analysis or tended to address the initial question whether 2 interpretation of data, and revising the manuscript for content, including variables are related in any way. It is only after this ini- writing of content. Xue Wang and Yan W. Asmann: analysis or interpre- tation of data and drafting the manuscript for content, including writing tial question is addressed that further questions, such of content. Michael G. Heckman: analysis or interpretation of data, sta- as mechanism, become relevant. To determine what tistical analysis, and drafting the manuscript for content, including writ- may have driven an increase in HOXA5 and TTR, ing of content. Joanne Wuu: contribution of vital reagents/tools/patients and revising the manuscript for content, including writing of content. we already performed cell culture experiments, dem- Keith A. Josephs, Joseph E. Parisi, David S. Knopman, Ronald C. Peters- onstrating that lower levels of C9ORF72 resulted in en, Leonard Petrucelli, Bradley F. Boeve, Neill R. Graff-Radford, Dennis higher levels of HOXA5 and TTR. These effects were W. Dickson, Michael Benatar, and Kevin B. Boylan: revising the manu- observed when targeting either total C9ORF72 tran- script for content, including writing of content, contribution of vital re- agents/tools/patients, and obtaining funding. Rosa Rademakers: study scripts or C9ORF72 transcript variant 2; we cannot, concept or design, acquisition of data, analysis or interpretation of data, however, exclude nonspecific effects on other tran- drafting the manuscript for content, including writing of content, revising script variants, and further studies are warranted. the manuscript for content, including writing of content, study supervi- Additional studies could also help to learn more about sion or coordination, and obtaining funding. Marka van Blitterswijk: study concept or design, acquisition of data, analysis or interpretation mechanisms that link C9ORF72 to HOXA5 and of data, statistical analysis, drafting the manuscript for content, including TTR, particularly because little is known about the writing of content, revising the manuscript for content, study supervision function, and interaction partners, of C9ORF72. or coordination, and obtaining funding. Moreover, future studies could examine downstream targets and/or upstream regulators that might con- ACKNOWLEDGMENT The authors thank Dr. J. Jiang, Dr. C. Lagier-Tourenne, Dr. D. Edbauer, tribute to the observed differences. and Dr. D.W. Cleveland for providing used to measure dipep- It is important that TTR protein levels have been tide repeat proteins. In addition, they thank Dr. M. Prudencio and Dr. V. evaluated as a potential biomarker for ALS and V. Belzil for sharing RNA sequencing results. FTD,28–33 but findings were inconsistent, which STUDY FUNDING could, in part, be explained by the genetic, patho- Supported by the National Institutes of Health (R21 NS093118, R01 logic, and clinical heterogeneity observed in those NS080882, R35 NS097261, R35 NS097273, P50 AG016574, P01 patients. Although our results seem to indicate that NS084974, and R56 NS061867) and the ALS Therapy Alliance. cerebellar TTR protein levels are elevated in Dr. van Blitterswijk is a former recipient of the Milton Safenowitz C9ORF72 expansion carriers, we could not detect Post-Doctoral Fellowship for ALS Research from the ALS Association and is currently supported by the Clinical Research in ALS and Related significant differences in CSF TTR protein levels. Disorders for Therapeutic Development (CReATe) Clinical Research Fel- The lack of a significant difference could be due to lowship. CReATe (U54 NS092091) is part of the Rare Diseases Clinical the presence of posttranslational modifications to the Research Network (RDCRN), an initiative of the Office of Rare Diseases Research (ORDR) and National Center for Advancing Translational Sci- TTR protein that are undetectable using our immu- ences (NCATS). CReATe is funded through collaboration between noassay. Alternatively, it might be possible that the NCATS and the National Institute of Neurological Disorders and Stroke. secretory pathway is affected, hampering the secretion of TTR into the CSF. Future experiments using mass DISCLOSURE spectrometry and immunoassays with different anti- NiCole A. Finch and Xue Wang report no disclosures. Matthew C. Baker holds the following patents: US Patent No. 12/302,691 (Detecting and bodies as well as experiments investigating the secre- Treating Dementia [2008]) and US Patent No. 12/413,869 (Methods tion of TTR (e.g., in cell culture models) should be and Materials for Detecting and Treating Dementia [2009]). Michael used to test these hypotheses. In addition, future G. Heckman has served on the editorial board of Parkinsonism & Related studies should examine whether TTR protein levels Disorders. Tania F. Gendron has received speaker honoraria from Johns Hopkins; holds a patent for Methods and materials for detecting poly are associated with features of the disease (e.g., in the (GP) proteins in tissues from C9ORF72 repeat expansion carriers; has CSF or plasma) and whether they change over time, received research support from NIH, the ALS Association, and the Mus- especially because one could postulate that a single cular Dystrophy Association; and receives license fee payments for C9ORF72 repeat expansion constructs and viruses, and for antibodies time point in a clinical cohort may not reflect changes against C9ORF72 dipeptide repeat proteins. Kevin F. Bieniek reports no observed in a pathologic cohort (end-stage disease). disclosures. Joanne Wuu has received research support from the Muscular Thus, we discovered elevated levels of multiple Dystrophy Association, the NIH, the Food and Drug Administration, Eli homeobox genes and TTR, reported to be involved Lilly and Company, the University of Miami, Department of Neurology, the ALS Association, and the Department of Defense. Mariely DeJesus- in developmental processes and neuroprotection, in Hernandez holds a patent on Methods to screen for the hexanucleotide brain tissue obtained from C9ORF72 expansion repeat expansion in the C9ORF72 gene. Patricia H. Brown, Jeannie

8 Neurology: Genetics Chew, Karen R. Jansen-West, and Lillian M. Daughrity report no dis- neuromuscular disease patients, and Biomarkers for detecting and treating closures. Alexandra M. Nicholson has received research support from the joint related pain; has been a consultant for Cytonics, Inc. and Merck; Association for Frontotemporal Degeneration. Melissa E. Murray has has received research support from the NIH, ALS Association, and Target served on the editorial boards of BMC Neurology and Frontiers in Neu- ALS; and holds stock/stock options for Iron Horse Diagnostics, Inc. rology and has been a consultant for Avid Radiopharmaceuticals. Keith A. Kevin B. Boylan receives research support from the NIH, ALS Associa- Josephs receives research support from the NIH, the Dana Foundation, tion, Genentech, Cytokinetics Inc., the Mayo Foundation, Neuraltus and the Alzheimer’s Association; and is an editorial board member for Pharmaceuticals, GlaxoSmithKline, Avanir Pharmaceuticals, and Synapse Acta Neuropathologica and Journal of Neuropathology & Experimental Neu- Biomedical. Rosa Rademakers receives research support from the NIH, rology. Joseph E. Parisi has served on the Defense Health Board Health the ALS Therapy Alliance, the Consortium for Frontotemporal Degen- Care Delivery Subcommittee; has received publishing royalties from eration Research, the Mayo Clinic Udall Center of Excellence, and the Oxford University Press; and has received research support from the Florida State Alzheimer’s Disease Research grant; received honoraria for NIH. David S. Knopman has served on the scientific advisory boards lectures or educational activities not funded by industry; serves on the of the Bluefield Project, Lundbeck Pharmaceuticals, and DIAN study medical advisory board of the Association for Frontotemporal Degener- DSMB; served on a Data Safety Monitoring Board for Lundbeck Phar- ation and on the board of directors of the International Society for maceuticals and the DIAN study; was an investigator in clinical trials Frontotemporal Dementia; and holds patents on Methods to screen for sponsored by Lilly, TauRx, and Biogen Pharmaceuticals; has received the hexanucleotide repeat expansion in the C9ORF72 gene, and Detect- travel funding/speaker honoraria from the Alzheimer Conference (Seoul, ing and treating dementia. Marka van Blitterswijk receives research sup- Korea) and the Behavioral Neurology Conference (Hyderabad, India); port from the NIH; is supported by the Clinical Research in ALS and has served on the editorial board of Neurology; and has received research Related Disorders for Therapeutic Development (CReATe) Clinical support from the NIH. Ronald C. Petersen has served on the scientific Research Fellowship; and is a former recipient of the Milton Safenowitz advisory boards of Pfizer, Janssen Alzheimer Immunotherapy, Elan Phar- Post-Doctoral Fellowship for ALS research from the ALS Association. Go maceuticals, Wyeth Pharmaceuticals, and GE Healthcare; has received to Neurology.org/ng for full disclosure forms. publishing royalties from Oxford University Press; has been a consultant for Roche Incorporated, Merck, GeneTech, Biogen, and Eli Lilly; has received research support from the NIH; and has served on the National Received December 28, 2016. Accepted in final form April 18, 2017. Advisory Council on Aging. Leonard Petrucelli has received research support from the Mayo Clinic Foundation, the NIH, the ALS Associa- REFERENCES tion, Lundbeck, Biogen, Robert Packard Center for ALS Research at 1. DeJesus-Hernandez M, Mackenzie IR, Boeve BF, et al. Johns Hopkins, Target ALS, the Association for Frontotemporal Degen- Expanded GGGGCC hexanucleotide repeat in noncoding eration, and the Department of Defense; has received license fee pay- region of C9ORF72 causes chromosome 9p-linked FTD ments from Lundbeck, Biogen, and Denali; has received royalty and ALS. Neuron 2011;72:245–256. payments for the licensing of TDP-43 antibody; has served on the sci- 2. Renton AE, Majounie E, Waite A, et al. A hexanucleotide entific advisory boards of SAB, Denali SAB, and Biogen; and serves on repeat expansion in C9ORF72 is the cause of chromosome the editorial boards of the Journal of Neuroscience, Molecular Neurodegen- – eration, and PLoS One. Bradley F. Boeve has served as an investigator for 9p21-linked ALS-FTD. Neuron 2011;72:257 268. clinical trials sponsored by GE Healthcare and FORUM Pharmaceuticals; 3. Ash PE, Bieniek KF, Gendron TF, et al. Unconventional receives royalties from the publication of a book entitled Behavioral Neu- translation of C9ORF72 GGGGCC expansion Generates rology of Dementia (Cambridge Medicine, 2009); serves on the Scientific Insoluble Polypeptides specific to c9FTD/ALS. Neuron Advisory Board of the Tau Consortium; has consulted for Isis Pharma- 2013;77:639–646. ceuticals and Ionis Pharmaceuticals; and receives research support from 4. Mori K, Weng SM, Arzberger T, et al. The C9orf72 the NIH, GE Healthcare, FORUM Pharmaceuticals, C2N Diagnostics, GGGGCC repeat is translated into aggregating dipeptide- the Little Family Foundation, and the Mangurian Foundation. Neill R. repeat proteins in FTLD/ALS. Science 2013;339:1335–1338. Graff-Radford has served on the editorial board of Alzheimer’s Research & 5. Zhang K, Donnelly CJ, Haeusler AR, et al. The C9orf72 Therapy; has received publishing royalties from UpToDate; has consulted repeat expansion disrupts nucleocytoplasmic transport. for Cytox; and has received research support from TauRx, Lilly, Biogen, – Axovant, and the NIH. Yan W. Asmann reports no disclosures relevant Nature 2015;525:56 61. to this manuscript. Dennis W. Dickson has served on the editorial boards 6. Jovicic A, Mertens J, Boeynaems S, et al. Modifiers of of Acta Neuropathologica, Brain, Brain Pathology, Neurobiology of Aging, C9orf72 dipeptide repeat toxicity connect nucleocytoplas- Annals of Neurology, Neuropathology, International Journal of Clinical and mic transport defects to FTD/ALS. Nat Neurosci 2015; Experimental Pathology, and American Journal of Neurodegenerative Dis- 18:1226–1229. ease; has received travel funding and speaker honoraria from Novartis; 7. Freibaum BD, Lu Y, Lopez-Gonzalez R, et al. GGGGCC and has received research support from the NIH, the Society for PSP: repeat expansion in C9orf72 compromises nucleocytoplas- Foundation for PSP/CBD and Related Disorders, and the Mangurian mic transport. Nature 2015;525:129–133. Foundation. Michael Benatar has served on the scientific advisory boards 8. Mizielinska S, Lashley T, Norona FE, et al. C9orf72 fron- of Denali, Ra Pharmaceuticals, Alnylam, and Mitsubishi Tanabe; has totemporal lobar degeneration is characterised by frequent served on the editorial board of Journal Watch Neurology; has been con- sultant for Congressionally Directed Medical Research Program neuronal sense and antisense RNA foci. Acta Neuropathol – (CDMRP) ALS Research Program Integration Panel; has received 2013;126:845 857. research funding from the Muscular Dystrophy Association, the ALS 9. Schludi MH, May S, Grasser FA, et al. Distribution of Association, the NIH, the Food and Drug Administration, the Depart- dipeptide repeat proteins in cellular models and C9orf72 ment of Defense, Cytokinetics Inc., Alexion Pharmaceuticals, Kimmel- mutation cases suggests link to transcriptional silencing. man Estate, Eli Lilly and Company, Neuraltus, and the ALS Recovery Acta Neuropathol 2015;130:537–555. Fund; and has had involvement in legal proceedings regarding Morris 10. Gendron TF, van Blitterswijk M, Bieniek KF, et al. Cer- James LLP. Robert Bowser is a founder of Iron Horse Diagnostics, Inc. (a ebellar c9RAN proteins associate with clinical and neuro- company focused on biomarkers related to ALS that holds patents on pathological characteristics of C9ORF72 repeat expansion TTR as a biomarker for motor neuron diseases); has served on the carriers. Acta Neuropathol 2015;130:559–573. scientific advisory boards of Denali Therapeutics and Above & Beyond, LLC; has served on the editorial boards of Scientific Reports, the Interna- 11. van Blitterswijk M, Gendron TF, Baker MC, et al. Novel tional Journal of Proteomics, and the American Journal of Neurodegenerative clinical associations with specific C9ORF72 transcripts in Disease; holds patents for Biomarkers for the diagnosis and prognosis of patients with repeat expansions in C9ORF72. Acta Neu- ALS, Biomarkers to monitor drug treatment of ALS and other ropathol 2015;130:863–876.

Neurology: Genetics 9 12. van Blitterswijk M, Dejesus-Hernandez M, Niemantsver- 23. Mann DM, Rollinson S, Robinson A, et al. Dipeptide driet E, et al. Association between repeat sizes and clinical repeat proteins are present in the p62 positive inclusions and pathological characteristics in carriers of C9ORF72 in patients with frontotemporal lobar degeneration and repeat expansions (Xpansize-72): a cross-sectional cohort motor neurone disease associated with expansions in study. Lancet Neurol 2013;12:978–988. C9ORF72. Acta Neuropathol Commun 2013;1:68. 13. Gendron TF, Bieniek KF, Zhang YJ, et al. Antisense tran- 24. Mackenzie IR, Arzberger T, Kremmer E, et al. Dipeptide scripts of the expanded C9ORF72 hexanucleotide repeat repeat protein pathology in C9ORF72 mutation cases: form nuclear RNA foci and undergo repeat-associated clinico-pathological correlations. Acta Neuropathol 2013; non-ATG translation in c9FTD/ALS. Acta Neuropathol 126:859–879. 2013;126:829–844. 25. Plante-Bordeneuve V, Said G. Familial amyloid polyneur- 14. Philippidou P, Dasen JS. Hox genes: choreographers in opathy. Lancet Neurol 2011;10:1086–1097. neural development, architects of circuit organization. 26. Fleming CE, Nunes AF, Sousa MM. Transthyretin: more Neuron 2013;80:12–34. than meets the eye. Prog Neurobiol 2009;89:266–276. 15. Santos SD, Lambertsen KL, Clausen BH, et al. CSF trans- 27. Cunningham TJ, Duester G. Mechanisms of retinoic acid thyretin neuroprotection in a mouse model of brain ische- signalling and its roles in organ and limb development. mia. J Neurochem 2010;115:1434–1444. Nat Rev Mol Cell Biol 2015;16:110–123. 16. Li X, Masliah E, Reixach N, Buxbaum JN. Neuronal pro- 28. Ryberg H, An J, Darko S, et al. Discovery and verification duction of transthyretin in human and murine Alzheimer’s of amyotrophic lateral sclerosis biomarkers by proteomics. disease: is it protective? J Neurosci 2011;31:12483–12490. Muscle Nerve 2010;42:104–111. 17. Li X, Buxbaum JN. Transthyretin and the brain re-visited: 29. Brettschneider J, Lehmensiek V, Mogel H, et al. Proteome is neuronal synthesis of transthyretin protective in Alz- analysis reveals candidate markers of disease progression in heimer’s disease? Mol Neurodegener 2011;6:79. amyotrophic lateral sclerosis (ALS). Neurosci Lett 2010; 18. Li X, Zhang X, Ladiwala AR, et al. Mechanisms of trans- 468:23–27. thyretin inhibition of beta-amyloid aggregation in vitro. 30. Ranganathan S, Williams E, Ganchev P, et al. Proteomic J Neurosci 2013;33:19423–19433. profiling of cerebrospinal fluid identifies biomarkers for 19. Fleming CE, Saraiva MJ, Sousa MM. Transthyretin enhances amyotrophic lateral sclerosis. J Neurochem 2005;95: nerve regeneration. J Neurochem 2007;103:831–839. 1461–1471. 20. Prudencio M, Belzil VV, Batra R, et al. Distinct brain 31. Ruetschi U, Zetterberg H, Podust VN, et al. Identification transcriptome profiles in C9orf72-associated and sporadic of CSF biomarkers for frontotemporal dementia using ALS. Nat Neurosci 2015;18:1175–1182. SELDI-TOF. Exp Neurol 2005;196:273–281. 21. Mori K, Arzberger T, Grasser FA, et al. Bidirectional tran- 32. Hansson SF, Puchades M, Blennow K, Sjogren M, scripts of the expanded C9orf72 hexanucleotide repeat are Davidsson P. Validation of a prefractionation method translated into aggregating dipeptide repeat proteins. Acta followed by two-dimensional electrophoresis—applied Neuropathol 2013;126:881–893. to cerebrospinal fluid proteins from frontotemporal 22. Zu T, Liu Y, Banez-Coronel M, et al. RAN proteins and dementia patients. Proteome Sci 2004;2:7. RNA foci from antisense transcripts in C9ORF72 ALS 33. Gloeckner SF, Meyne F, Wagner F, et al. Quantitative and frontotemporal dementia. Proc Natl Acad Sci USA analysis of transthyretin, tau and amyloid-beta in patients 2013;110:E4968–E4977. with dementia. J Alzheimers Dis 2008;14:17–25.

10 Neurology: Genetics Comparing sequencing assays and human-machine analyses in actionable genomics for glioblastoma

Kazimierz O. ABSTRACT * Wrzeszczynski, PhD Objective: To analyze a glioblastoma tumor specimen with 3 different platforms and compare Mayu O. Frank, NP, potentially actionable calls from each. MS* Methods: Tumor DNA was analyzed by a commercial targeted panel. In addition, tumor-normal Takahiko Koyama, PhD* DNA was analyzed by whole-genome sequencing (WGS) and tumor RNA was analyzed by RNA Kahn Rhrissorrakrai, PhD* sequencing (RNA-seq). The WGS and RNA-seq data were analyzed by a team of bioinformaticians Nicolas Robine, PhD and cancer oncologists, and separately by IBM Watson Genomic Analytics (WGA), an automated Filippo Utro, PhD system for prioritizing somatic variants and identifying drugs. Anne-Katrin Emde, PhD Bo-Juen Chen, PhD Results: More variants were identified by WGS/RNA analysis than by targeted panels. WGA com- Kanika Arora, MS pleted a comparable analysis in a fraction of the time required by the human analysts. Minita Shah, MS Conclusions: The development of an effective human-machine interface in the analysis of deep Vladimir Vacic, PhD cancer genomic datasets may provide potentially clinically actionable calls for individual pa- Raquel Norel, PhD tients in a more timely and efficient manner than currently possible. Erhan Bilal, PhD ClinicalTrials.gov identifier: NCT02725684. Neurol Genet 2017;3:e164; doi: 10.1212/ Ewa A. Bergmann, MSc NXG.0000000000000164 Julia L. Moore Vogel,

PhD GLOSSARY Jeffrey N. Bruce, MD CNV 5 copy number variant; EGFR 5 epidermal growth factor receptor; GATK 5 Genome Analysis Toolkit; GBM 5 glioblas- Andrew B. Lassman, MD toma; IRB 5 institutional review board; NLP 5 Natural Language Processing; NYGC 5 New York Genome Center; RNA-seq 5 RNA sequencing; SNV 5 single nucleotide variant; SV 5 structural variant; TCGA 5 The Cancer Genome Atlas; TPM 5 Peter Canoll, MD, PhD transcripts per million; VCF 5 variant call file; VUS 5 variants of uncertain significance; WGA 5 Watson Genomic Analytics; Christian Grommes, MD WGS 5 whole-genome sequencing. Steve Harvey, BS Laxmi Parida, PhD The clinical application of next-generation sequencing technology to cancer diagnosis and treat- Vanessa V. Michelini, BS ment is in its early stages.1–3 An initial implementation of this technology has been in targeted Michael C. Zody, PhD panels, where subsets of cancer-relevant and/or highly actionable genes are scrutinized for Vaidehi Jobanputra, PhD potentially actionable mutations. This approach has been widely adopted, offering high redun- Ajay K. Royyuru, PhD dancy of sequence coverage for the small number of sites of known clinical utility at relatively Robert B. Darnell, MD, low cost. PhD However, recent studies have shown that many more potentially clinically actionable muta- tions exist both in known cancer genes and in other genes not yet identified as cancer drivers.4,5 Correspondence to Dr. Darnell: Improvements in the efficiency of next-generation sequencing make it possible to consider [email protected] whole-genome sequencing (WGS) as well as other omic assays such as RNA sequencing (RNA-seq) as clinical assays, but uncertainties remain about how much additional useful infor- mation is available from these assays. Supplemental data at Neurology.org/ng *These authors contributed equally to the manuscript. From the New York Genome Center (K.O.W., M.O.F., N.R., A.-K.E., B.-J.C., K.A., M.S., V.V., E.A.B., J.L.M.V., M.C.Z., V.J., R.B.D.); IBM Thomas J. Watson Research Center (T.K., K.R., F.U., R.N., E.B., L.P., A.K.R.); Columbia University Medical Center (J.N.B., A.B.L., P.C., V.J.); Memorial Sloan-Kettering Cancer Center (C.G.), New York, NY; IBM Watson Health (S.H., V.V.M.), Boca Raton, FL; Laboratory of Molecular Neuro-Oncology (M.O.F., R.B.D.), and Howard Hughes Medical Institute (R.B.D.), The Rockefeller University, New York, NY. B.-J.C. is currently affiliated with Google, New York, NY. V.V. is currently affiliated with 23andMe, Inc., Mountain View, CA. E.A.B. is currently affiliated with Max Planck Institute of Immunobiology and Epigenetics, Freiburg, Germany. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Aside from cost, a challenge of WGS or Tumor purity and ploidy. Tumor purity was calculated from whole-transcriptome data is the expertise and WGS data using Titan.23 In addition, purity and ploidy were calculated from the Illumina OMNI 2.5M Array using ASCAT.24 time required to interpret the full spectrum of somatic mutations. To address this challenge, RNA sequencing. We used the Illumina TruSeq stranded mes- senger RNA protocol and sequenced 100 million reads. Reads Watson for Genomics (Watson Genomic Ana- were aligned using STAR25 and Gencode genes were quantified lytics [WGA]), a cancer analytic tool, uses stan- using featureCounts.26 Ninety-five percent of reads mapped the dard variant call files (VCFs), copy number reference genome. We normalized the counts with DESeq2 and 27 variant (CNV), and differential gene expression adjusted the quantification to account for GC bias and batch effects28 between The Cancer Genome Atlas (TCGA) GBM data to return a list of recommended cancer RNA-seq and our sample. The normalized expression data are drugs. Here, we present the results of a targeted used to identify GBM subtypes.29 cancer panel along with WGS and RNA-seq in Therapeutic targets and drug recommendations. The a patient with glioblastoma (GBM). We also NYGC uses the custom clinical Tier classification system for SNVs. compare results of expert interpretation of the Tier 1 variants are clinically important variants in the cancer type tumor genome by bioinformaticians and oncol- being studied (e.g., epidermal growth factor receptor [EGFR] T790M is known to be clinically important in lung cancer30). ogists at New York Genome Center (NYGC) The same variant observed in a cancer unknown to manifest this and at collaborating institutions with those variant is classified as Tier 2 (e.g., the clinical importance of EGFR generated by WGA. T790M is unknown in GBM). Tier 3 variants are in targetable genes; however, the specific variant is not known to be targetable (e.g., an unknown mutation in EGFR). Tier 4 variants are in genes METHODS Standard protocol approvals, registrations, cataloged by COSMIC cancer census and not included in Tiers 1– and patient consents. This study was approved by multiple 3.15 All other variants are in Tier 5 and considered variants of institutional review boards (IRBs), including Rockefeller University uncertain significance (VUS). Variants in Tiers 1–4 are considered IRB and Biomedical Research Alliance of New York IRB. The potentially targetable. Variants were matched to potential treat- study was registered in ClinicalTrials.gov (NCT02725684). ments by identifying the most aberrant genes from a combination Informed written consent was obtained from the participant. of SNV, INDEL, SV, and RNA-seq data and by searching the Participant. This report describes the first participant in a multi- NYGC drug-to-gene database. Prioritization of potential treat- institutional study. NYGC-GBM-01 was a 76-year-old man with ments was based on further manual assessment including criteria GBM. DNA and RNA were extracted from snap-frozen tissue. such as strength of data supporting variants detected, FDA approval DNA from blood was obtained for comparison. The samples of drug in GBM or in another cancer type, current GBM trial for were analyzed by WGS and RNA-seq. a drug, and successful use of the drug to target the variant identified to treat GBM or other cancer types. Single nucleotide variants and INDELs. Whole-genome Watson Genomic Analytics. WGA, an IBM research proof- libraries were prepared using the Illumina TruSeq Nano of-concept environment of Watson for Genomics,31 is a cogni- DNA Sample Prep Kit and were sequenced on Illumina HiSeq tive system built on several different predictive models to analyze X instruments (Illumina, San Diego, CA). Paired-end 2 3 150 bp up to whole-genome scale molecular data. VCFs, CNV, and gene reads were aligned to the GRCh37 human reference (BWA aln expression data are input to WGA. The VCF file provided to v.0.7.8)6 and processed using a pipeline that includes marking of WGA contains the union from 3 calling algorithms each for duplicate reads using Picard tools and realignment around SNVs and INDELs specified in the Methods section. CNV data INDELs and base recalibration using Genome Analysis Toolkit are inputted as copy number log2 (T/N) ratio values per gene. (GATK) version 2.7.4.7 muTect v1.1.4,8 LoFreq v2.0.0,9 Strelka Modified Z-scores of RNA-seq normalized expression data per v1.0.13,10 Pindel,11 and Scalpel12 were used to return the union of gene are used as proxy for differential gene expression. Modified variant calls. Variants were filtered out if they were at .1% z-score per gene is calculated by subtracting the median tran- frequency in the 1000 Genomes or ExAC data sets, had more scripts per million (TPM) value (over the TCGA GBM cohort) to than 2 alleles to remove artifacts, raw frequency in the tumor was this sample’s TPM and dividing by the TCGA SD. With this lower than that in the normal, or matched a custom “blacklist” of input, WGA leverages a comprehensive database of structured known systematic errors generated by comparing normal germ- (201 sources include DrugBank, NCI, COSMIC, ClinVar, and line replicates. Remaining single nucleotide variants (SNVs) and 1000 Genomes) and unstructured (evidence extracted from lit- INDELs were annotated via snpEff,13 snpSift,13 and GATK erature using Natural Language Processing [NLP]) biological and VariantAnnotator using annotation from ENSEMBL,14 COSMIC,15 medical data. To date, WGA processed abstracts from PubMed Gene Ontology,16 and 1000 Genomes.17 and where possible, began analyzing full-text articles. In addition, Structural variation. Structural variants (SVs), such as CNVs the NLP engine is being trained to understand the approximate and complex genomic rearrangements, were detected by NBIC- 5,600 clinical trials at ClinicalTrials.gov. It is from the unstruc- seq,18 Delly,19 CREST,20 and BreakDancer.21 We prioritized tured sources that WGA maintains a current repository of drug- SVs in the intersection of callers and those with additional disease associations and biomarkers for prognosis and therapeutics, split-read evidence via SplazerS.22 SVs with split-read support in as well as matching patients to relevant clinical trials based on the matched normal or annotated as known germline variants molecular criteria. WGA identifies gene alterations most likely to (1000 Genomes call set, Database of Genomic Variants) were be important in cancer and then identifies relevant treatments that removed as likely germline variants. The predicted somatic SVs directly or indirectly target the variant. WGA also identifies VUS, were annotated with gene overlap (RefSeq, Cancer Gene Census) resistive or sensitizing markers for the drug of interest, and relevant including prediction of potential effect on resulting proteins. clinical trials.

2 Neurology: Genetics RESULTS Case report. NYGC-GBM-01 was a 76- Table 2 Number of somatic variants identified year-old man who presented with headache and dif- ficulty with ambulation. CT of the brain revealed Type of variant Count a mass in the left parietal region. He underwent initial Single nucleotide variants (SNVs) 8,449 resection for which pathology revealed a GBM, neg- Insertions and deletions (INDELs) 431 ative for the following: EGFR amplification by in situ hybridization fluorescence, EGFRvIII RNA expres- Exonic SNV 133 sion, IDH1 R132H by immunostaining, 1p36/ Exonic INDEL 16 19q13 deletion by fluorescent in situ hybridization Copy number gain 2 analysis, and MGMT methylation. Postsurgically, he Copy number loss 5 had right-sided hemineglect and right/left confusion. He became somnolent and required a re-resection and ventriculoperitoneal shunt placement. Two Tumor analysis. The metrics for the WGS are shown months after initial resection, he completed radiation in table 1. The sample had an estimated tumor therapy with 40 Gy over 3 weeks with concurrent purity of 47%–52% and ploidy of 1.99. Table 2 temozolomide 75 mg/m2 daily. He then completed 3 describes the types and number of variants identi- cycles of adjuvant temozolomide at 5 months after fied. Specifically, variant-calling pipeline analysis initial resection, after which progression was seen on identified 8,449 total somatic mutations (with MRI. He considered multiple options, but experi- 150 falling in exonic, protein-coding regions) and enced functional decline, and was no longer trial a complex landscape of amplifications and deletions. eligible. Instead, he started on the first dose of bev- Mutational signatures are an important molecular acizumab and CCNU, 7 months after initial resec- characterization of the tumor and assessing appli- tion. He further declined and died 1 month later. cability of immunotherapy,32 for example, RNA-seq A sample from the initial resection was examined identified the sample as the mesenchymal subtype of with a FoundationOne test, and a snap-frozen sample GBM.28 WGS supported this with evidence of an was received for sequencing in this study. DNA and NF1 mutation and a CDKN2A loss with a gain of RNA extraction, sequencing, and analysis required Chr 7 and a loss of Chr 10. 7 weeks at which time a tumor board meeting was The NYGC identified 6 actionable SNVs, of convened, including the treating oncologist, a neu- which 2 were Tier 3 variants (MET R755fs and ro-oncologist, and bioinformaticians, to discuss the FGFR3 L49V) and 4 were Tier 4 variants (in results of the analysis completed by the NYGC. A STAG2, PIK3R1, NF1,andERG, described in table clinical report of the findings was subsequently issued e-1 at Neurology.org/ng). In addition, 5 CNVs were to the oncologist. This tumor board meeting occurred identified, of which 2 were in genes that had SNVs after the completion of the first cycle of adjuvant te- (table 3). CREST and Pindel identified a 299-bp mozolomide. The oncologist planned on referring the intragenic deletion at the intron-exon junction of patient to clinical trials identified by the NYGC, but exon 11 in MET, as well as an amplification of at the time of progression, he was no longer trial eli- MET (log2 CNV3.64-foldtumorvsnormalampli- gible due to functional decline. fication). RNA-seq confirmed overexpression of MET (z-score 2.23) and an in-frame exon-skipping event (METex11), at an allele frequency of approxi- Table 1 Sample whole-genome sequencing mately 50% (figure 1). This observation is molecu- metrics larly analogous to the skipping of exon 14 identified Mean coverage in lung adenocarcinoma and other cancer types.33 Tumor 75.63 Although METex11 is located in the extracellular

Normal 42.43 domain, we hypothesized that this mutation could lead to an overactivation of MET and could be tar- Total reads geted by a tyrosine kinase inhibitor. By analogy with Tumor 1,920,954,396 studies of MET(D7–8),34 both would lead to the lack Normal 1,019,519,682 of transmembrane localization. We also noted that Mapped reads mislocalization of MET(D7–8) renders the variant Tumor 1,805,911,478 not targetable using antibodies. However, MET-

Normal 959,221,187 specific tyrosine kinase inhibitors could efficiently

303 coverage, % deactivate the kinase. We also identified a codon insertion in PIK3R1 Tumor 97.8 (p.R562_M563insIle/c.1686_1688dupTAT), which Normal 86.8 is a regulatory protein that interacts with and inhibits

Neurology: Genetics 3 Table 3 List of variants identified as actionable by 3 different platforms

Identified variant Identified associated drugs

Gene Variant NYGC WGA FO NYGC WGA FO

CDKN2A Deletion Yes Yes Yes Palbociclib, LY2835219 Palbociclib LY2835219 Clinical trial LEE001

CDKN2B Deletion Yes Yes Yes Palbociclib, LY2835219 Palbociclib LY2835219 Clinical trial LEE002

EGFR Gain (whole arm) Yes ——Cetuximab ——

ERG Missense P114Q Yes Yes — RI-EIP RI-EIP —

FGFR3 Missense L49V Yes VUS — TK-1258 ——

MET Amplification Yes Yes Yes INC280 Crizotinib, cabozantinib Crizotinib, cabozantinib

MET Frame shift R755fs Yes ——INC280 ——

MET Exon skipping Yes ——INC280 ——

NF1 Deletion Yes ——MEK162 ——

NF1 Nonsense R461* Yes Yes Yes MEK162 MEK162, cobimetinib, Everolimus, temsirolimus, trametinib, GDC-0994 trametinib

PIK3R1 Insertion Yes Yes — BKM120 BKM120, LY3023414 — R562_M563insI

PTEN Loss (whole arm) Yes ——Everolimus, AZD2014 ——

STAG2 Frame shift R1012 fs Yes Yes Yes Veliparib, clinical trial Olaparib —

DNMT3A Splice site 2083-1G.C ——Yes ———

TERT Promoter-146C.T Yes — Yes ———

ABL2 Missense D716N Germline NA VUS

mTOR Missense H1687R Germline NA VUS

NPM1 Missense E169D Germline NA VUS

NTRK1 Missense G18E Germline NA VUS

PTCH1 Missense P1250R Germline NA VUS

TSC1 Missense G1035S Germline NA VUS

Abbreviations: FO 5 FoundationOne; NYGC 5 New York Genome Center; RNA-seq 5 RNA sequencing; WGA 5 Watson Genomic Analytics; WGS 5 whole- genome sequencing. Genes, variant description, and, where appropriate, candidate clinically relevant drugs are listed. Variants identified by the FO as variants of uncertain significance (VUS) were identified by the NYGC as germline variants.

the functional catalytic protein, PIK3CA. Activating a PIK3CA E545K oncogenic variant, in which mutations of PIK3CA are known cancer drivers, as a combination of an MET inhibitor and a PIK3CA are loss-of-function mutations of PIK3R1.Func- inhibitor showed better sensitivity than single ther- tional studies have shown that PIK3R1 amino acid apy.37 The PIK3CA E545K variant also activates D560 is involved in hydrogen binding with PIK3CA. Taken together, these findings led us to PIK3CA and is an essential amino acid in regulat- suggest combinatorial INC280 (MET inhibitor) and ing the activity of the catalytic subunit. The muta- BKM120 (PIK3CA inhibitor) therapy for potential tion identified here is in the same helical inhibitory clinical consideration, and this suggestion would have (iSH2) domain of PIK3R1. The variant binds but made the patient eligible for a clinical trial assessing fails to inhibit PIK3CA, leading to enhanced cell efficacy of this combination (NCT01870726). survival, Akt activation, anchorage-independent cell growth, and oncogenesis.35,36 Furthermore, Watson for Genomic analysis. Data for NYGC-GBM- analysis of the crystal structure (figure 2) supports 01 were input into WGA, which produced a report the conclusion that this mutation would inhibit the summarizing actionable variants and a list of associ- functional interaction between the 2 proteins, spe- ated drugs, including some based on a pathway target cifically through N345 of PIK3CA, resulting in analysis. WGA identified 6 actionable alterations, 14 PIK3CA activation. associated drugs, 9 VUS (including FGFR3), copy A recent cell line study identified a synergistic number losses in Chr 9 (focal), 10 (armscale), and 11 effect between MET exon 14–skipping variants and (focal), and gain on Chr 7 (armscale). Both the

4 Neurology: Genetics Figure 1 Sashimi plot representing the MET exon–skipping event

Red lines indicate exon coverage and exon junctions. Numbers in red indicate the number of reads supporting these junc- tions (for instance, 1,181 reads are split between exons 10 and 12). Only junctions with more than 100 reads are repre- sented here.

NYGC and WGA identified 5 actionable alterations genes. WGA found 10 clinical trials that may be (in genes NF1, MET, CDKN2A, CDKN2B, and relevant across 6 actionable alterations. PIK3R1; table 3). WGA reported an NF1 SNV and Comparison with a panel. Table 3 also compares all annotated the variant as inactivating but did not variants and drugs identified by the NYGC and deem copy number change to be sufficient for calling FoundationOne. NYGC analysis identified 8 unique this or EGFR and PTEN. A 1-copy gain of EGFR was variants not found by the FoundationOne, including ’ below WGA s threshold for classification as a target- an exon-skipping event. The NYGC identified drugs able variant. Furthermore, this variant was shown to for 10 targets, while FoundationOne identified drugs be negative for amplification by in situ hybridization for 4. Furthermore, 6 of the variants reported as of fluorescence. However, the NYGC decided to list it unknown significance occurring in the tumor by as potentially targetable, given it is a known action- FoundationOne were germline variants. One variant able variant in GBM. Similarly, a 1-copy PTEN loss is (DNMT3A splice site 2083-1G.C) was called by reported by the NYGC but not by WGA or by the FoundationOne, but the position and base change FoundationOne. The NYGC reported this variant were different from a nearby variant identified by the because of its clinical implications; it is associated NYGC. with resistance to EGFR tyrosine kinase inhibition via AKT/mTOR pathway activation and is linked to DISCUSSION The NYGC is undertaking a WGS cetuximab resistance, but can be targeted by mTOR research study in patients with GBM to investigate inhibitors.38 For PIK3R1, the NYGC identified the efficiency and feasibility of WGS to inform ther- BKM120 as a potential therapeutic option based on apeutic options. Here, the results of NYGC WGS additional RNA-seq evidence of overexpression of and RNA-seq were compared with a clinical panel PIK3CA. WGA identified PIK3R1 as a relevant var- assay. Also, in collaboration with IBM, the NYGC iant via SNV data by WGS and used RNA-seq examined the therapeutic options identified by WGA information in pathway and drug analysis to also based on WGS and RNA-seq data. Genomic results recommend BKM120. The MET amplification and from this patient clearly displayed the diversity of associated drugs are reported by both platforms; driver events typically seen in GBM. Of interest, we however, WGA had 2 drugs for MET amplification, identified mutations in targetable genes that were not whereas the NYGC prioritized 1 therapeutic option, precise matches to known specific targetable variants, INC280, based on GBM trial data availability. The and which nonetheless suggested potential therapeu- NYGC reported 8 clinical trials associated with 5 tic options.

Neurology: Genetics 5 sequencing of both germline and tumor DNA not Figure 2 Overlay of genomic mutations on crystal structure of PIK3R1 only heightened our sensitivity for what variants might be tumor drivers but was able to rule out a number of germline variants called by the Founda- tionOne as not likely to be primary drivers of this patient’s GBM. Although we conducted WGS of this sample at roughly twice the cost of WES, the primary analysis was performed on the protein-coding region of the genome. There may be technical advantage to WGS even for assaying targeted regions. WES relies on hybridization capture of specific genes which introdu- ces intrinsic bias for each gene as a function of GC/ AT content, while WGS relies more simply on mechanical shearing of DNA prior to sequencing. Previous studies have found that for disorders caused by constitutional mutations,40,41 WGS is more sensi- tive than WES for variant detection. To assess whether WGS could detect variants not identified by WES to justify the added cost, it would require a direct comparison of the assays on the same sample. We are undertaking a study to address this question. This patient died approximately 8 months from the time of initial resection falling short of the median survival time for GBM. The oncologist recommen- ded enrollment in a clinical trial targeting PIK3 and MET alterations on recurrence on adjuvant temozo- lomide. However, the patient’s clinical decline elim- inated his ability to participate in trials. This highlights one of the challenges of the clinical appli- New York Genome Center identified GBM somatic mutations in iSH2 domain of PIK3R1 (red). cation of precision medicine technology. The identi- Known D560Y and N564D (orange) mutants in the PI3K regulatory domain (PIK3R1) fail to fication of targets and potentially useful drugs in inhibit PIK3CA and lead to enhanced cell survival, Akt activation, anchorage-independent cell growth, and oncogenesis.34 The R562-M563 insertion mutation in NYGC-GBM1 potentially a timely manner is only the first step. Drug and drug alters the functional interaction within this region, specifically through N345 (green) of trial access is crucial to determine the benefit of this PIK3CA. The potential inability of catalytic regulation due to PIK3R1 mutation can translate approach in cancer management. to therapeutic targeting with a PIK3CA inhibitor such as BKM120 and therefore the iden- Another key observation was that the WGA anal- tification of trial NCT01870726 in combination with MET inhibitor INC280. ysis vastly accelerated the time to discovery of poten- tially actionable variants from the VCF files. As Multimodal analysis (WGS and RNA-seq) previously reported, we found that WGA was able increased confidence in the identification of the to provide reports of potentially clinically actionable MET mutation; analysis of the literature of prior insights within 10 minutes, while human analysis of MET exon–skipping events suggested the plausibil- this patient’s VCF file took an estimated 160 hours ity of considering a tyrosine kinase inhibitor that of person-time. This is critical if sequencing is to be could target MET. Similarly, manual literature search brought out of the research arena and into the scaled, of the PIK3CA E545K oncogenic variant led to the real-world clinical realm. This study is an important conclusion that this was likely an activating mutation. step forward promoting human-machine interface as Moreover, manual database searches resulted in the a way to address a key bottleneck in cancer genomics. suggestion of a combinatorial treatment with an MET inhibitor and a PIK3CA inhibitor, which made AUTHOR CONTRIBUTIONS logical sense and also made the patient eligible for Kazimierz O. Wrzeszczynski: analysis and interpretation of the data and drafting and revising the manuscript. Mayu O. Frank: design and con- a clinical trial for this combination (NCT01870726). ceptualization of the study, interpretation of the data, and drafting and None of these observations were evident from the revising the manuscript. Takahiko Koyama: analysis and interpretation panel. This suggests that pursuing a more extensive of the data and revising the manuscript. Kahn Rhrissorrakrai: analysis comparison of panel and deeper sequencing (e.g., and interpretation of the data and revising the manuscript. Nicolas Rob- ine: analysis and interpretation of the data and revising the manuscript. WGS and RNA-seq) will be of interest. An added Filippo Utro: revising the manuscript. Anne-Katrin Emde, Bo-Juen point, previously noted by others,39 is that the Chen, Kanika Arora, Minita Shah, Vladimir Vacic, Raquel Norel, Erhan

6 Neurology: Genetics Bilal, Ewa A. Bergmann, Julia L. Moore Vogel, Jeffrey N. Bruce, Andrew DNA sequencing using multiple metal layer structure with different B. Lassman, Peter Canoll, Christian Grommes, Steve Harvey, Laxmi Par- organic coatings forming different transient bondings to DNA, Fabrica- ida, Vanessa V. Michelini, Michael C. Zody, Vaidehi Jobanputra, and tion of tunneling junction for nanopore DNA sequencing, Verification of Ajay K. Royyuru: revising the manuscript. Robert B. Darnell: design complex workflows through internal assessment or community based and conceptualization of the study, interpretation of the data, and draft- assessment, Field effect based nanosensor for biopolymer manipulation ing and revising the manuscript. and detection, DNA sequence using multiple metal layer structure with different organic coatings forming different transient bondings to DNA, ACKNOWLEDGMENT Integrated nanowire/nanosheet nanogap and nanopore for DNA and RNA-seq, Field effect based nanosensor for biopolymer manipulation The authors are grateful for critical input from many scientific members of and detection, Integrated nanowire/nanosheet nanogap and nanopore the NYGC and for input from the Rockefeller University and BRANY for DNA and RNA-seq, Electron beam sculpting of tunneling junction IRBs. for nanopore DNA sequencing, Charged entities as locomotive to control motion of polymers through a nanochannel, Molecular dispensers (2), STUDY FUNDING Nanopore capture system, Protein structure analysis (2), Method of This study was supported by an unrestricted grant from IBM to the identifying robust clustering, Hydrophobic moment of multi-domain NYGC, and from NYGC internal funds. proteins (3), System and program storage device of object classification utilizing optimized Boolean expressions, Apparatus, method, and DISCLOSURE product of manufacture for transforming supply chain networks using K.O. Wrzeszczynski reports no disclosures. M.O. Frank has been a con- pair-wise nodal analysis, Method and apparatus for protein structure sultant for the New York Genome Center and has received research sup- analysis, Techniques for reconstructing supply chain networks using port from the International Society of Nurses in Genetics, Rockefeller pair-wise correlation analysis, and Object classification using an opti- University, and NIH CTSA. T. Koyama holds a patent (pending) for mized Boolean expression; is an employee of IBM Corporation; is Drug scoring utilizing pathway analysis and is an employee of IBM Cor- a member of the Industry Advisory Board of International Society poration. K. Rhrissorrakrai owns stock in Johnson & Johnson, Pfizer, and for Computational Biology; and has received research support from Merck and is an employee of IBM Corporation. N. Robine reports no Pfizer and CHDI. R.B. Darnell serves on the Scientific Advisory Board disclosures. F. Utro is an employee of IBM Corporation. A.-K. Emde, at the New York Genome Center, Roundtable on Translating B.-J. Chen, K. Arora, and M. Shah report no disclosures. V. Vacic is an Genomic-Based Research for Health at National Academy of Medi- employee of and owns stock options in 23andMe, Inc. R. Norel reports cine, and Stanford Medicine Board of Fellows; holds patent nos. no disclosures. E. Bilal is an employee of IBM Corporation. E.A. Bergmann’s 7,989,203, 6602709, 6,750,029, 7,928,190, and 14/104,581; holds husband is an employee of Agenus. J.L. Moore Vogel has been a con- a patent (pending) for Use of BET inhibitors to treat neurodevelop- sultant for the New York Genome Center. J.N. Bruce reports no disclo- mental disorders and epilepsy; has received honoraria from Lennart sures. A.B. Lassman has served on the scientific advisory boards of Astra Philipson Memorial Lecture, Uppsala University, Sweden; has received Zeneca, AbbVie, Sapience Therapeutics, Genentech, Bioclinica, VBI research support from IBM, Sohn New York City Collaboration for Vaccines, Cortice Biosciences, Oxigene, Regeneron, and Novocure; has Pediatric Cancer Research, Starr Cancer Consortium, and NIH; is received travel funding/speaker honoraria from prIME Oncology and a Howard Hughes Medical Institute investigator; and owns stock in from aforementioned scientific advisory boards; has served on the edito- Amgen, Illumina, and Agios. Go to Neurology.org/ng for full disclo- rial boards of Neuro-Oncology and the Journal of Neurooncology; has been sure forms. a consultant for WebMD, the American Society of Clinical Oncology, and the American Academy of Neurology; has been a grant reviewer for Received December 12, 2016. Accepted in final form April 19, 2017. the Italian Association for Cancer Research; and has received research support from the NCI, the Radiation Therapy Oncology Group Foun- REFERENCES dation, and the James S. McDonnell Foundation. P. Canoll has served on 1. Good BM, Ainscough BJ, McMichael JF, Su AI, Griffith the editorial boards of PLoS One and GLIA and has received research OL. Organizing knowledge to enable personalization of support from NIH/National Institute of Neurological Disorders and medicine in cancer. Genome Biol 2014;15:438. Stroke, NCI, NIDA, and the James F. McDonnell Foundation (Swanson). C. Grommes has received research support from Pharmacyclics. S. Harvey 2. Griffith M, Miller CA, Griffith OL, et al. Optimizing reports no disclosures. L. Parida has served on the advisory board of the cancer genome sequencing and analysis. Cell Syst 2015; NYU Tandon School of Engineering; has served on the editorial boards 1:210–223. of BMC Bioinformatics, IEEE/ACM Transactions on Computational 3. Hyman DM, Solit DB, Arcila ME, et al. Precision medicine Biology and Bioinformatics, the Journal of Computational Biology, and at Memorial Sloan Kettering Cancer Center: clinical next- the SIAM Journal of Discrete Mathematics; holds patents for Transductive generation sequencing enabling next-generation targeted feature selection with maximum-relevancy and minimum-redundancy therapy trials. Drug Discov Today 2015;20:1422–1428. criteria, Improved metagenome mapping, Confidence interval estimation 4. Kandoth C, McLellan MD, Vandin F, et al. Mutational of species in metagenomic data, Estimating multiple parents from a matrix landscape and significance across 12 major cancer types. of F1 hybrid progeny, Systems and methods for fitting LD distributions Nature 2013;502:333–339. at genomic scales, TLASSO: Transductive Lasso for High Dimensional Data Regression Problems, Lossless compression of the enumeration 5. Zack TI, Schumacher SE, Carter SL, et al. Pan-cancer space of founder line crosses, and Efficient sorting of large dimensional patterns of somatic copy number alteration. Nat Genet data; receives publishing royalties from Chapman & Hall; and is an 2013;45:1134–1140. employee of IBM Research. V.V. Michelini is an employee of, has 6. Li H, Durbin R. Fast and accurate short read alignment received research support from, and holds stock/stock options in IBM with Burrows-Wheeler transform. Bioinformatics 2009; Corporation. M.C. Zody has received travel funding and teaching hon- 25:1754–1760. oraria from Fourth Culture, LLC., and teaching honoraria from Harvard 7. McKenna A, Hanna M, Banks E, et al. The Genome CFAR and Cold Spring Harbor Laboratory; has served as a guest member Analysis Toolkit: a MapReduce framework for analyzing of the editorial board for Annual Review of Genomics and Human Genet- next-generation DNA sequencing data. Genome Res ics; is supported in part by a grant from the Alfred P. Sloan Foundation; 2010;20:1297–1303. has received research support from IBM, NHGRI/NIH, NCI/NIH, and the Alfred P. Sloan Foundation; holds a patent for Method for inference 8. Cibulskis K, Lawrence MS, Carter SL, et al. Sensitive of HLA types from short read sequencing data; and owns stock in Ther- detection of somatic point mutations in impure and het- mo Fisher and Merck. V. Jobanputra reports no disclosures. A.K. Royyuru erogeneous cancer samples. Nat Biotechnol 2013;31: holds patents for Mixed polynucleotide and forming method thereof, 213–219.

Neurology: Genetics 7 9. Wilm A, Aw PP, Bertrand D, et al. LoFreq: a sequence- 25. Dobin A, Davis CA, Schlesinger F, et al. STAR: ultra- quality aware, ultra-sensitive variant caller for uncovering fast universal RNA-seq aligner. Bioinformatics 2013; cell-population heterogeneity from high-throughput 29:15–21. sequencing datasets. Nucleic Acids Res 2012;40:11189– 26. Liao Y, Smyth GK, Shi W. featureCounts: an efficient 11201. general purpose program for assigning sequence reads to 10. Saunders CT, Wong WS, Swamy S, Becq J, Murray LJ, genomic features. Bioinformatics 2014;30:923–930. Cheetham RK. Strelka: accurate somatic small-variant call- 27. Risso D, Schwartz K, Sherlock G, Dudoit S. GC-content ing from sequenced tumor-normal sample pairs. Bioinfor- normalization for RNA-Seq data. BMC Bioinformatics matics 2012;28:1811–1817. 2011;12:480. 11. Ye K, Schulz MH, Long Q, Apweiler R, Ning Z. Pindel: 28. Johnson WE, Li C, Rabinovic A. Adjusting batch effects a pattern growth approach to detect break points of large in microarray expression data using empirical Bayes meth- deletions and medium sized insertions from paired-end ods. Biostatistics 2007;8:118–127. short reads. Bioinformatics 2009;25:2865–2871. 29. Verhaak RG, Hoadley KA, Purdom E, et al. Integrated 12. Narzisi G, O’Rawe JA, Iossifov I, et al. Accurate de novo genomic analysis identifies clinically relevant subtypes of and transmitted indel detection in exome-capture data glioblastoma characterized by abnormalities in PDGFRA, using microassembly. Nat Methods 2014;11:1033–1036. IDH1, EGFR, and NF1. Cancer Cell 2010;17:98–110. 13. Cingolani P, Platts A, Wang le L, et al. A program for 30. Sharma SV, Bell DW, Settleman J, Haber DA. Epidermal annotating and predicting the effects of single nucleotide growth factor receptor mutations in lung cancer. Nat Rev polymorphisms, SnpEff: SNPs in the genome of Drosoph- Cancer 2007;7:169–181. ila melanogaster strain w1118; iso-2; iso-3. Fly (Austin) 31. Rhrissorrakrai K, Koyama T, Parida L. Watson for ge- 2012;6:80–92. nomics: moving personalized medicine forward. Trends 14. Aken BL, Ayling S, Barrell D, et al. The Ensembl gene Cancer 2016;2:392–395. annotation system. Database (Oxford) 2016;2016:baw093. 32. Rizvi NA, Hellmann MD, Snyder A, et al. Cancer immu- 15. Forbes SA, Bindal N, Bamford S, et al. COSMIC: mining nology: mutational landscape determines sensitivity to complete cancer genomes in the Catalogue of Somatic PD-1 blockade in non-small cell lung cancer. Science Mutations in Cancer. Nucleic Acids Res 2011;39:D945– 2015;348:124–128. D950. 33. Frampton GM, Ali SM, Rosenzweig M, et al. Activation 16. Gene Ontology C. The Gene Ontology in 2010: exten- of MET via diverse exon 14 splicing alterations occurs in sions and refinements. Nucleic Acids Res 2010;38:D331– multiple tumor types and confers clinical sensitivity to D335. MET inhibitors. Cancer Discov 2015;5:850–859. 17. Genomes Project C, Auton A, Brooks LD, et al. A global 34. Navis AC, van Lith SA, van Duijnhoven SM, et al. Iden- reference for human genetic variation. Nature 2015;526: tification of a novel MET mutation in high-grade glioma 68–74. resulting in an auto-active intracellular protein. Acta Neu- 18. Xi R, Hadjipanayis AG, Luquette LJ, et al. Copy number ropathol 2015;130:131–144. variation detection in whole-genome sequencing data 35. Backer JM. The regulation of class IA PI 3-kinases by using the Bayesian information criterion. Proc Natl Acad inter-subunit interactions. Curr Top Microbiol Immunol Sci USA 2011;108:E1128–E1136. 2010;346:87–114. 19. Rausch T, Zichner T, Schlattl A, Stutz AM, Benes V, 36. Jaiswal BS, Janakiraman V, Kljavin NM, et al. Somatic Korbel JO. DELLY: structural variant discovery by inte- mutations in p85alpha promote tumorigenesis through grated paired-end and split-read analysis. Bioinformatics class IA PI3K activation. Cancer Cell 2009;16:463–474. 2012;28:i333–i339. 37. Liu X, Jia Y, Stoopler MB, et al. Next-generation sequenc- 20. Wang J, Mullighan CG, Easton J, et al. CREST maps ing of pulmonary sarcomatoid carcinoma reveals high fre- somatic structural variation in cancer genomes with base- quency of actionable MET gene mutations. J Clin Oncol pair resolution. Nat Methods 2011;8:652–654. 2016;34:794–802. 21. Chen K, Wallis JW, McLellan MD, et al. BreakDancer: an 38. Frattini M, Saletti P, Romagnani E, et al. PTEN loss of algorithm for high-resolution mapping of genomic struc- expression predicts cetuximab efficacy in metastatic colo- tural variation. Nat Methods 2009;6:677–681. rectal cancer patients. Br J Cancer 2007;97:1139–1145. 22. Emde AK, Schulz MH, Weese D, et al. Detecting genomic 39. Jones S, Anagnostou V, Lytle K, et al. Personalized geno- indel variants with exact breakpoints in single- and paired- mic analyses for cancer mutation discovery and interpre- end sequencing data using SplazerS. Bioinformatics 2012; tation. Sci Transl Med 2015;7:283ra253. 28:619–627. 40. Belkadi A, Bolze A, Itan Y, et al. Whole-genome sequenc- 23. Ha G, Roth A, Khattra J, et al. TITAN: inference of copy ing is more powerful than whole-exome sequencing for number architectures in clonal cell populations from detecting exome variants. Proc Natl Acad Sci USA 2015; tumor whole-genome sequence data. Genome Res 2014; 112:5473–5478. 24:1881–1893. 41. Turner TN, Hormozdiari F, Duyzend MH, et al. Genome 24. Van Loo P, Nordgard SH, Lingjaerde OC, et al. Allele- sequencing of autism-affected families reveals disruption of specific copy number analysis of tumors. Proc Natl Acad putative noncoding regulatory DNA. Am J Hum Genet Sci USA 2010;107:16910–16915. 2016;98:58–74.

8 Neurology: Genetics Autopsy case of the C12orf65 mutation in a patient with signs of mitochondrial dysfunction

Hideaki Nishihara, MD, ABSTRACT PhD Objective: To describe the autopsy case of a patient with a homozygous 2-base deletion, Masatoshi Omoto, MD, c171_172delGA (p.N58fs), in the C12orf65 gene. PhD Methods: We described the clinical history, neuroimaging data, neuropathology, and genetic anal- Masaki Takao, MD, PhD ysis of the patients with C12orf65 mutations. Yujiro Higuchi, MD, PhD Results: The patient was a Japanese woman with a history of delayed psychomotor development, Michiaki Koga, MD, PhD primary amenorrhea, and gait disturbance in her 20s. She was hospitalized because of respira- Motoharu Kawai, MD, tory failure at the age of 60. Pectus excavatum, long fingers and toes, and pes cavus were re- PhD vealed by physical examination. Her IQ score was 44. Neurologic examination revealed Hiroo Kawano, MD, PhD ophthalmoplegia, optic atrophy, dysphagia, distal dominant muscle weakness and atrophy, hy- Eiji Ikeda, MD, PhD perreflexia at patellar tendon reflex, hyporeflexia at Achilles tendon reflex, and extensor plantar Hiroshi Takashima, MD, reflexes. At age 60, she died of pneumonia. Lactate levels were elevated in the patient’sserum PhD and CSF. T2-weighted brain MRI showed symmetrical hyperintense brainstem lesions. At Takashi Kanda, MD, autopsy, axial sections exposed symmetrical cyst formation with brownish lesions in the upper PhD spinal cord, ventral medulla, pons, dorsal midbrain, and medial hypothalamus. Microscopic anal- ysis of these areas demonstrated mild gliosis with rarefaction. Cell bodies in the choroid plex- uses were eosinophilic and swollen. Electron microscopic examination revealed that these cells Correspondence to Prof. Kanda: contained numerous abnormal mitochondria. Whole-exome sequencing revealed the 2-base [email protected] deletion in C12orf65. Conclusions: We report an autopsy case of the C12orf65 mutation, and findings suggest that mitochondrial dysfunction may underlie the unique clinical presentations. Neurol Genet 2017;3: e171; doi: 10.1212/NXG.0000000000000171

GLOSSARY ATPase 5 adenosine triphosphatase; COX 5 cytochrome oxidase; COXPD7 5 combined oxidative phosphorylation defi- ciency type 7; H&E 5 hematoxylin and eosin; SDH 5 succinate dehydrogenase; SPG55 5 spastic paraplegia 55.

The C12orf65 gene encodes a mitochondrial matrix protein that has a role in releasing peptides from mitochondrial ribosomes.1 To date, 2 phenotypes have been associated with C12orf65 gene mutations: combined oxidative phosphorylation deficiency type 7 (COXPD7)1 and auto- somal recessive spastic paraplegia 55 (SPG55).2 While C12orf65 defects exhibit a wide spectrum of phenotypes, the 3 primary clinical features are optic atrophy, peripheral neuropathy, and spastic paraparesis.3 Although biochemical studies suggest that these C12orf65 mutations cause mitochondrial dysfunction,1,2,4 very few pathologic analyses and no autopsy cases supporting these findings have been reported. In this article, we report an autopsy case associated with 2- nucleotide deletion in the C12orf65 gene. Our patient presented with optic atrophy, peripheral neuropathy, pyramidal signs, mental retardation, and several additional features, including long

From the Department of Neurology and Clinical Neuroscience (H.N., M.O., M. Koga, M. Kawai, T.K.), Department of Laboratory Science (H.K.), Department of Pathology (E.I.), Yamaguchi University Graduate School of Medicine, Japan; Department of Neurology and Cerebrovascular Medicine (M.T.), Saitama International Medical Center, Saitama Medical University, Japan; and Department of Neurology and Geriatrics (Y.H., H.T.), Kagoshima University Graduate School of Medical and Dental Sciences, Japan. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 (1:200, monoclonal; DAKO, Tokyo, Japan), 4 Figure 1 Family pedigree and genetic studies (1:1,000, polyclonal; Merck Millipore, Tokyo, Japan), neurofila- ment (SMI31) and Schwann/2E (1:5000, monoclonal; COSMO BIO, Tokyo, Japan), and phospho-TDP-43 (s409/410, 1:7,000; COSMO BIO).8 Peripheral nerve specimens were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer, postfixed in 1% osmium tetroxide, embedded in Epon, and stained with toluidine blue. Muscle sam- ples were snap frozen in liquid nitrogen–cooled isopentane. Serial 8-mm-thick cryosections were stained with H&E, nicotinamide adenine dinucleotide tetrazolium reductase, adenosine triphos- phatase (ATPase, preincubation at pH 4.6 and 10.8), modified Gomori trichrome, succinate dehydrogenase (SDH), and cyto- chrome oxidase (COX).

Electron microscopy. Choroid plexus specimens from formalin-fixed tissues were dissected and refixed in 2.5% glutar- aldehyde. Muscle samples were directly fixed in 2.5% glutaral- dehyde. Electron microscopy was performed as previously reported.7 Ultrathin sections were examined under a Hitachi H- 7500 transmission electron microscope.

Genetic analysis. We used the same methodology as was used in a previous study.9 The captured exome library was sequenced using a HiSeq 2000 (Illumina, San Diego, CA). Sequences were aligned to the human reference genome (NCBI37/hg19) using the Burrows-Wheeler Aligner, and variant calling was performed using SAMtools.10,11 Variants were annotated using in-house scripts, which provided the list of variants. The mutation in the C12orf65 gene was validated using Sanger sequencing on samples from the patient and her younger sister.

Arrow indicates the proband. Her parents were first cousins. There was no family history of similar illness. Sequencing chromatogram of the 2-base deletion in the C12orf65 gene of RESULTS Description of the patient. The proband the proband (IV:1; arrow) and her sister, a wild-type, heterozygous carrier (IV:2). (IV:1), a 61-year-old Japanese woman at the time of death, was born of consanguineous parents (figure 1). fingers and toes, pectus excavatum, lack of Based on the information obtained from her parents, no other family members had neurologic disorders. secondary sexual characteristics, primary The patient was delivered without any physical amenorrhea, osteoporosis, and late onset respi- abnormalities at birth. However, she showed delayed ratory insufficiency. psychomotor development and could not walk until she was 1 year and 10 months old. She was admitted METHODS Standard protocol approvals, registrations, to elementary school at age 7 and needed special and patient consents. The protocol of the studies was reviewed and approved by the Institutional Review Board of Kagoshima support education because of mental retardation. At University. The proband (IV:1) and her younger sister (IV:2) age 26, she was diagnosed with primary amenorrhea. provided written informed consent to participate in this study. According to medical records, neurologic examina- Postmortem study, histology, light microscopy, and tions revealed bilateral optic atrophy, exotropia, immunohistochemistry. An autopsy was performed 12 hours weakness in both tibialis anterior muscles, steppage after the patient’s death. After a thorough macroscopic inspec- gait, and hyperreflexia at patellar reflex. The patient tion, conventional histologic techniques were performed to make was developmentally disabled and placed in a nursing – specimens as mentioned previously.5 7 Tissue sections were then home when she was in her 30s. In her early 40s, she stained with hematoxylin and eosin (H&E). At the time of was diagnosed with osteoporosis. Leg weakness was autopsy, several small pieces of fresh brain tissue (right frontal exacerbated, and she needed aid for walking at age 55. lobe) were dissected, immediately frozen in dry ice, and stored at 280°C for future studies. At 60 years of age, she was unable to eat and was The remainder of the brain was fixed, dissected, and stained admitted to a local hospital where she was diagnosed 5–8 m as previously reported. The 3- m-thick sections were stained as having CO2 narcosis with infection. Although she with H&E for nuclei and eosinophilic structures, modified recovered from the infections, she still had hyper- Gallyas-Braak silver staining for fibrils, and Klüver-Barrera for capnia and was transferred to Yamaguchi University myelin. We performed immunohistochemical studies using the Hospital. Physical examination revealed an abnor- following monoclonal antibodies: amyloid Ab (11–28) (12B8, 1:100; IBL, Gunma, Japan), phospho-tau (AT8, 1:3,000, Inno- mally long arm span (162 cm) compared with her genetics, Ghent, Belgium), phosphorylated a-synuclein (1:7,000, height (159 cm), long fingers and toes, pes cavus, pSyn#64, monoclonal, Wako), glial fibrillary acidic protein pectus excavatum, and hypertelorism. Neurologic

2 Neurology: Genetics Figure 2 Brain MRI

(A) Sagittal image of T2-weighted brain MRI. Arrows indicate hyperintense lesions in the brainstem. (B) Axial images dem- onstrating symmetrical T2 high signals (arrows) in the bilateral thalamus, extending inferiorly along the midbrain pontine tegmentum, terminating in the ventral aspect of the medulla. examination showed bilateral optic atrophy, oph- This loss-of-function mutation in exon 2, which thalmoplegia, dysphagia, significant distal domi- encodes the C12orf65 RF-1 domain, could introduce nant muscle weakness and wasting, hyperreflexia at a premature stop codon that leads to messenger RNA patellar tendon reflex, hyporeflexia at Achilles degradation through the nonsense-mediated decay tendon reflex, and extensor plantar reflexes. Pain mechanism. The patient’s younger sister (figure 1, and temperature sensations were almost normal. IV:2) was heterozygous for the 2-base deletion, Neuropsychological evaluation was indicative of an c171_172delGA (p.N58fs) in C12orf65. Gene anal- intellectual disability (Wechsler Adult Intelligence ysis of their parents could not be performed because Scale–Revised: Verbal IQ 50, Performance IQ 47, they were deceased. We confirmed that this mutation andFull-ScaleIQ44).Arterialbloodgasanalysis was not found in public databases, including the 1000 showed hypercapnia (60.7 mm Hg), hypoxia Genomes project databases (browser.1000genomes. (65.4 mm Hg), and hyperbicarbonemia (35.6 org), the Exome Aggregation Consortium (exac. mmol/L) under ambient conditions. Lactate levels broadinstitute.org/), and the Human Genetic Varia- intheplasmaandCSFwereslightlyelevated(1.51 tion Database comprising the exome sequencing and 2.02 mmol/L, respectively). Exercise and pul- results of 1,208 Japanese individuals (genome.med. monary function tests could not be performed kyoto-u.ac.jp/SnpDB/) using same methodology as because of her respiratory and cognitive status. previously reported.9 No other causative mutations Brain MRI revealed symmetric T2 prolongation were identified among the known disease-causing involving the bilateral thalamus, which extended genes, including those related to Charcot-Marie- inferiorly along the midbrain pontine tegmentum Tooth disease, hereditary sensory and autonomic and terminating in the ventral aspect of the medulla neuropathies, hereditary motor neuropathy, and (figure 2, A and B). Nerve conduction studies amyotrophic lateral sclerosis. showed distal dominant motor sensory axonal Postmortem study, histology, light microscopy, dysfunction. Although her respiratory status tem- immunohistochemistry, and electron microscopy. General porarily improved under noninvasive positive autopsy. The patient was 159.2 cm in height and 31 kg pressure ventilation assistance, respiratory dys- in body weight. Her skeletal muscles were extremely function gradually worsened, and the patient died atrophic in a distal dominant manner. The patient of pneumonia at age 61 years. also had long fingers and toes, pectus excavatum, Genetic analysis. Exome sequencing identified a novel and pes cavus. Her organs revealed mild atrophy. homozygous 2-base deletion, c171_172delGA (p. The uterus showed extreme hypoplasia with a length N58fs), in the patient’s C12orf65 gene (figure 1). of 35 mm and a width of 20 mm.

Neurology: Genetics 3 Figure 3 Pathologic findings

(A) Gross pathology of the brainstem. Axial sections revealed symmetrical brownish lesions with cyst formation (ar- rows) at the level of the medial hypothalamus and dorsal midbrain and pons. These cystic lesions were also present in the ventral medulla and upper spinal cord. (B–D) Hematoxylin and eosin (H&E) staining of the midbrain (B), pons (C), and medulla (D). (E) Enlarged image defined by a square in B (bar 5 500 mm). These T2WI high-signal areas showed rarefaction with gliosis. (F) A photomicrograph around a cystic lesion of the inferior olivary nucleus. Numerous macro- phages and proliferation of small vessel channels were observed (H&E staining). (G) H&E staining of the patient’s choroid plexus cells. Microscopic examination demonstrated the markedly swollen and eosinophilic cytoplasm of the lateral ventricle choroid plexus cells (bar 5 10 mm). (I) Electron micrograph of the patient’s choroid plexus cells. The cells contained numerous enlarged abnormal mitochondria (bar 5 2 mm). (H, J) H&E staining (H, bar 5 10 mm) and electron micrograph (J, bar 5 2 mm) of the choroid plexus cells in normal subjects. (K, L) Immunostaining of cytochrome c oxidase of the patient’s skeletal muscle (K) and age-matched control’s (L). Cytochrome oxidase activity in whole fibers was low in the patient’sskeletalmuscle.

Brain pathology. Postfixed brain weight was 1,150 g, medulla, pons, dorsal midbrain, and medial hypothal- and the pituitary gland was enlarged. A thin mem- amus (figure 3A). Using microscopy, we found that brane with brownish discoloration was observed on these areas showed moderate-to-severe rarefaction the inner surface of the dura mater, which is consis- and mild-to-moderate gliosis (figure 3, B–E). tent with a previous subdural hematoma. The arach- Numerous macrophages, as well as proliferation of noid membrane was cloudy at the level of the frontal small vessel channels, were observed around the cystic lobes. The frontal lobes also exhibited mild atrophy. lesions (figure 3F). Ischemic neurons were not found The brainstem and cerebellum were preserved. Axial around the lesions, nor was there evidence of sections exposed symmetrical brownish lesions with occluded vessels. Microscopic examination revealed cyst formation in the upper spinal cord, ventral markedly swollen and eosinophilic cytoplasm in the

4 Neurology: Genetics DISCUSSION We have described clinical, neurora- Table Summary of autopsy findings diologic, and pathologic studies of an autopsy case Neuropathology with a homozygous mutation (c171_172delGA

Symmetrical brownish lesions with cyst Medial hypothalamus, dorsal midbrain, pons, [p.N58fs]) in the C12orf65 gene. In 2010, formation ventral part of the medulla, and upper spinal cord C12orf65 gene mutations were reported in 3 individ- Numerous abnormal mitochondria Choroid plexus cells, pituitary gland, and skeletal uals who developed Leigh syndrome, optic atrophy, muscles and ophthalmoplegia. This clinical phenotype was Axonal degeneration of the sural nerve designated COXPD7.1 In 2012, the second pheno- Cytochrome oxidase deficiency in skeletal type of the C12orf65 gene mutation was reported. muscles The individual showed autosomal recessive hereditary Hyperplasia of the pituitary gland spastic paraplegias with optic atrophy and neuropa- Microbleedings in the posterior cingulate cortex thy. This clinical condition, designated SPG55, has milder clinical presentations than COXPD7.2 Thin old subdural hematoma Because the number of mutation sites has General pathology increased recently, genotype-phenotype heterogene- Emaciation 159.2 cm in height and only 31 kg in body weight ities have been reported.3,4,12–15 The typical pheno- Malformation Long fingers and toes, pectus excavatum, and pes types of a C12orf65 mutation includes optic cavus atrophy, peripheral neuropathy, pyramidal signs, Mild atrophy of organs Heart: 320 g, left lung: 350 g, right lung: 380 g, liver: 890 g, spleen: 55 g, left kidney: 150 g, right and cognitive disorders. Although our patient showed kidney: 150 g these typical features of a C12orf65 mutation, she also Uterus hypoplasia The length was 35 mm and width was 20 mm developed endocrine abnormalities such as a lack of

Bronchial pneumonia of the left inferior secondary sexual characteristics, primary amenorrhea, lobe and osteoporosis, all of which are features of mito- – Pleural effusion chondrial diseases.16 18 The endocrine abnormalities

Aortic atherosclerosis of our patient may have been caused by mitochon- drial dysfunction because morphological abnormali- ties were also found in mitochondria of the pituitary choroid plexus cells of the lateral ventricle (figure gland cells. Although an endocrine function test was 3G). In the H&E sections, those cell bodies were not performed, these findings suggest dyspituitarism. markedly swollen and eosinophilic compared with the Disease severity and mortality are thought to be control cases. These morphological features suggested related to C12orf65 protein length and whether the the presence of underlying mitochondrial abnormal- mutant site contains the glycine-glycine-glutamine ities. Therefore, the choroid plexus cells were motif, which interacts with the large ribosomal sub- subjected to electron microscopic analysis. Ultra- unit to release the polypeptide chain from the P-site structurally, these cells contained numerous abnor- bound peptidyl transfer RNA.3,13 Previous findings mally enlarged mitochondria (figure 3I). We also suggest that pathogenesis in C12orf65-mutant pa- identified numerous large mitochondria in the pitu- tients is mediated by mitochondrial dysfunctions.1 itary gland and skeletal muscles. Additional findings Fibroblasts cultured from 2 patients with C12orf65 included hyperplasia of the pituitary gland and small mutations showed global and uniform defects in the petechial hemorrhages in the posterior cingulate translation of mitochondrial DNA-encoded proteins, cortex. resulting in a severe decrease in oxidative phosphory- Peripheral nerves and muscles. A sural nerve biopsy lation complexes I, IV, and V and a small decrease in specimen showed a mild loss of myelinated fibers complex III. Patients with SPG55 showed a relatively without myelin ovoids, which was indicative of long- larger C12orf65 protein, possibly due to a compara- standing axonal degeneration. The biceps brachii tively preserved complex V and unaffected complex muscle displayed mildly atrophic fibers and slight III activity.2 These previous data suggest that fat replacement. ATPase staining revealed mild small C12orf65 protein length is related to mitochondrial group atrophy and fiber type grouping. There were respiratory chain function and thereby disease sever- no ragged-red fibers or strongly SDH-reactive blood ity. The expected protein length in our patient was 57 vessels. COX activity in whole fibers was low com- amino acids, the shortest of those previously reported, pared with the age-matched control, and there was no and the same length as the protein studied by Heidary focal COX deficiency (figure 3, K and L). Most of the et al.12 Moreover, immunohistochemical analysis re- diaphragm muscle fibers were small in size (20–30 vealed that our patient’s cytochrome c oxidase activ- mm); however, fat replacement was minimal and large ities were extremely low, possibly pointing to group atrophy was absent. The pathologic findings mitochondrial respiratory chain complex abnormali- are summarized in the table. ties. Several patients with the COXPD7 phenotype

Neurology: Genetics 5 and a short C12orf65 protein experienced respiratory We reported an autopsy case with a confirmed failure in the first or second decade of life.1 In com- C12orf65 gene mutation. The patient, with a novel parison, our patient’s respiratory ability was well pre- homozygous loss-of-function deletion, presented served, and the prognosis was relatively good, with peripheral neuropathy, optic atrophy, mental suggesting that protein length is not always correlated retardation, pyramidal signs, and primary amenor- with disease severity. rhea. Pathologic findings revealed numerous and Until now, only 3 reports included histologic enlarged abnormal mitochondria, which might be information from C12orf65-mutant patients, 2 related to the pathogenesis of the various phenotypes. using a sural nerve specimen and the other a skin biopsy specimen. The 2 sural nerve biopsies showed AUTHOR CONTRIBUTIONS markedly reduced numbers of myelinated fibers Dr. Nishihara performed the experiments, analyzed and interpreted the data, and wrote the manuscript. Dr. Omoto, Prof. Takao, Prof. Kawano, accompanied by numerous small myelinated fiber and Prof. Ikeda performed the pathologic experiments, analyzed the data, clusters, suggesting regeneration. A decrease in the and edited the manuscript. Dr. Higuchi and Prof. Takashima performed number of unmyelinated fibers was also con- the genetic experiments, analyzed the data, and edited the manuscript. firmed.4,15 Our patient also showed large myelinated Dr. Koga and Dr. Kawai evaluated the data, provided advice about the experiments, and edited the manuscript. Prof. Kanda designed and super- fiber loss and no acute changes such as myelin vised the study, evaluated the data, and wrote the manuscript. The study ovoids. Therefore, C12orf65 mutations may cause was planned by Prof. Kanda. chronic axonal degeneration in peripheral nerves. Electron photomicrographs of the 1 skin biopsy re- ACKNOWLEDGMENT vealed enlarged mitochondria engorged by an abnor- The authors are grateful to all study individuals and their relatives. They also thank Mitsutoshi Tano for technical assistance. mally large number of densely packed cristae.12 Previous reports suggested that mitochondrial dys- STUDY FUNDING functions are related to certain features of a C12orf65 No targeted funding reported. mutation such as peripheral neuropathy, optic atro- 1,2,4,19 phy, and spastic paraplegia. Our investigation DISCLOSURE found evidence of mitochondrial abnormalities not H. Nishihara has received research grants (16H07008) from the Japan Soci- only on the skin but also in the choroid plexus, ety for the Promotion of Science and has received research support from the pituitary gland, and skeletal muscles, demonstrating Uehara Memorial Foundation and the Japan Multiple Sclerosis Society. M. Omoto reports no disclosures. M. Takao has received Grants-in-Aid that mitochondrial dysfunction was widespread and for Scientific Research on Innovative Areas (Comprehensive Brain Science could be related to the pathogenesis and observed Network, 221S0003) and the Platform of Supporting Cohort Study and clinical manifestations. Biospecimen Analysis (JSPS KAKENHI JP 16H06277) and KIBAN C (JSPS KAKENHI 26430060). Y. Higuchi, M. Koga, M. Kawai, H. Kawano, Our patient also showed symmetrical MRI abnor- and E. Ikeda report no disclosures. H. Takashima has received speaker malities in the hypothalamus over the upper cervical honoraria from Kyushu University, GlaxoSmithKline, Bayer, Eisai Co., spinal cord. The same abnormalities were reported Novartis, Tanabe Mitsubishi, Biogen Idec, Benesis Co., Takeda, Teijin, in previous cases, but histologic analysis had not been Pfizer, Otsuka, Ono, Hisamitsu, Shionogi, Asteras, FP, Boehringer, and Dainippon Sumitomo Pharma; has been a consultant for Teijin; has 12,14 performed. Pathologic analysis of the present received research support from Nervous and Mental Disorders, Research patient showed bilateral rarefaction in part of the Committee for Charcot-Marie-Tooth Disease, Neuropathy, Ataxic Dis- medial hypothalamus, mesencephalic tegmentum, ease, Applying Health and Technology, and the Ministry of Health, pontine tegmentum, ventral medulla, and upper cer- Labour, and Welfare (Japan); and has received royalty payments from Athena Diagnostics. T. Kanda has served as the Editor-in-Chief for vical spinal cord tissues, surrounded by mild gliosis. Clinical and Experimental Neuroimmunology and was on the editorial These tissues contained white matter and gray matter, advisory broad for Neuropathology and has received research grants which suggests that there is no association with spe- (Nos. 23659457, 25293203, and 26670443) from the Japan Society for the Promotion of Science (Tokyo, Japan), a Research Grant for Neuro- cific anatomical tracts and structures. In addition, the immunological Diseases from the Ministry of Health, Labour, and Wel- pathologic changes of the choroid plexuses again fare of Japan (K2002528), and an Intramural Research Grant (25-4) for allude to a diagnosis of mitochondrial disease. In fact, Neurological and Psychiatric Disorders of National Center of Neurology the characteristic MRI findings, pathologic features of and Psychiatry. Go to Neurology.org/ng for full disclosure forms. the cystic lesions, and choroid plexus pathology Received February 6, 2017. Accepted in final form May 8, 2017. strongly suggest the presence of mitochondrial abnor- malities and led to molecular analysis. Although the REFERENCES pathogenesis of bilateral rarefaction in the brainstem 1. Antonicka H, Ostergaard E, Sasarman F, et al. Mutations is unclear, the same MRI abnormalities have been in C12orf65 in patients with encephalomyopathy and reported in other mitochondrial disorders such as a mitochondrial translation defect. Am J Hum Genet 2010;87:115–122. Leber optic neuropathy and MFN2 mutation.20,21 2. Shimazaki H, Takiyama Y, Ishiura H, et al. A homozygous Symmetrical brainstem abnormalities should be con- mutation of C12orf65 causes spastic paraplegia with optic sidered a key indicator of possible mitochondrial atrophy and neuropathy (SPG55). J Med Genet 2012;49: dysfunction. 777–784.

6 Neurology: Genetics 3. Spiegel R, Mandel H, Saada A, et al. Delineation of c12orf65 gene: report of a novel mutation and review of C12orf65-related phenotypes: a genotype-phenotype rela- the literature. J Neuroophthalmol 2014;34:39–43. tionship. Eur J Hum Genet 2014;22:1019–1025. 13. Buchert R, Uebe S, Radwan F, et al. Mutations in the 4. Tucci A, Liu YT, Preza E, et al. Novel C12orf65 muta- mitochondrial gene C12ORF65 lead to syndromic auto- tions in patients with axonal neuropathy and optic atro- somal recessive intellectual disability and show genotype phy. J Neurol Neurosurg Psychiatry 2014;85:486–492. phenotype correlation. Eur J Med Genet 2013;56:599– 5. Takao M, Ghetti B, Yoshida H, et al. Early-onset demen- 602. tia with Lewy bodies. Brain Pathol (Zurich, Switzerland) 14. Imagawa E, Fattal-Valevski A, Eyal O, et al. Homozygous 2004;14:137–147. p.V116* mutation in C12orf65 results in Leigh syn- 6. Takao M, Ghetti B, Hayakawa I, et al. A novel mutation drome. J Neurol Neurosurg Psychiatry 2016;87:212–216. (G217D) in the Presenilin 1 gene (PSEN1) in a Japanese 15. Montecchiani C, Pedace L, Lo Giudice T, et al. family: presenile dementia and parkinsonism are associated ALS5/SPG11/KIAA1840 mutations cause autosomal with cotton wool plaques in the cortex and striatum. Acta recessive axonal Charcot-Marie-Tooth disease. Brain neuropathologica 2002;104:155–170. 2016;139:73–85. 7. Takao M, Mori T, Orikasa H, et al. Postmortem diagnosis 16. Finsterer J. Mitochondriopathies. Eur J Neurol 2004;11: of Fabry disease with acromegaly and a unique vasculop- 163–186. athy. Virchows Arch 2007;451:721–727. 17. Guo Y, Yang TL, Liu YZ, et al. Mitochondria-wide asso- 8. Takao M, Aoyama M, Ishikawa K, et al. Spinocerebellar ciation study of common variants in osteoporosis. Ann ataxia type 2 is associated with Parkinsonism and Lewy Hum Genet 2011;75:569–574. body pathology. BMJ case Rep 2011;2011. 18. Chen CM, Huang CC. Gonadal dysfunction in mito- 9. Higuchi Y, Hashiguchi A, Yuan J, et al. Mutations in chondrial encephalomyopathies. Eur Neurol 1995;35: MME cause an autosomal-recessive Charcot-Marie-Tooth 281–286. disease type 2. Ann Neurol 2016;79:659–672. 19. Abramov AY, Smulders-Srinivasan TK, Kirby DM, et al. 10. Li H, Durbin R. Fast and accurate short read alignment Mechanism of neurodegeneration of neurons with mito- with Burrows-Wheeler transform. Bioinformatics (Oxford, chondrial DNA mutations. Brain 2010;133:797–807. England) 2009;25:1754–1760. 20. Paulus W, Straube A, Bauer W, Harding AE. Central 11. Li H, Handsaker B, Wysoker A, et al. The sequence align- nervous system involvement in Leber’s optic neuropathy. ment/map format and SAMtools. Bioinformatics (Oxford, J Neurol 1993;240:251–253. England) 2009;25:2078–2079. 21. Boaretto F, Vettori A, Casarin A, et al. Severe CMT type 2 12. Heidary G, Calderwood L, Cox GF, et al. Optic atrophy with fatal encephalopathy associated with a novel MFN2 and a Leigh-like syndrome due to mutations in the splicing mutation. Neurology 2010;74:1919–1921.

Neurology: Genetics 7 Prevalence of spinocerebellar ataxia 36 in a US population

Juliana M. Valera, MS ABSTRACT Tatyana Diaz Objective: To assess the prevalence and clinical features of individuals affected by spinocerebel- Lauren E. Petty, MS lar ataxia 36 (SCA36) at a large tertiary referral center in the United States. Beatriz Quintáns, PhD Methods: A total of 577 patients with undiagnosed sporadic or familial cerebellar ataxia compre- Zuleima Yáñez, PhD hensively evaluated at a tertiary referral ataxia center were molecularly evaluated for SCA36. Eric Boerwinkle, PhD Repeat primed PCR and fragment analysis were used to screen for the presence of a repeat Donna Muzny, MS expansion in the NOP56 gene. Dmitry Akhmedov, PhD Rebecca Berdeaux, PhD Results: Fragment analysis of triplet repeat primed PCR products identified a GGCCTG hexanu- Maria J. Sobrido, MD, cleotide repeat expansion in intron 1 of NOP56 in 4 index cases. These 4 SCA36-positive PhD families comprised 2 distinct ethnic groups: white (European) (2) and Asian (Japanese [1] and Richard Gibbs, PhD Vietnamese [1]). Individuals affected by SCA36 exhibited typical clinical features with gait ataxia James R. Lupski, MD, and age at onset ranging between 35 and 50 years. Patients also suffered from ataxic or spastic PhD, DSc limbs, altered reflexes, abnormal ocular movement, and cognitive impairment. Daniel H. Geschwind, Conclusions: In a US population, SCA36 was observed to be a rare disorder, accounting for 0.7% MD, PhD (4/577 index cases) of disease in a large undiagnosed ataxia cohort. Neurol Genet 2017;3:e174; Susan Perlman, MD doi: 10.1212/NXG.0000000000000174 Jennifer E. Below, PhD Brent L. Fogel, MD, PhD GLOSSARY ERIS 5 Error Rate In Sequencing (ERIS); IBD 5 identically by descent; LOD 5 logarithm of the odds; MAF 5 minor allele frequency; RP-PCR 5 repeat primed PCR; SCA 5 spinocerebellar ataxia; SCA36 5 spinocerebellar ataxia 36; SNP 5 single nucleotide polymorphism. Correspondence to Dr. Fogel: [email protected] Genetic cerebellar ataxia is a clinically heterogeneous disease that progressively destroys the cere- bellum and consequently impairs balance and coordination in the affected individual.1 The diagnosis of cerebellar ataxia can be challenging because there are more than 500 genes associated with either primary or secondary ataxia, and many affected families remain undiagnosed for decades. The introduction of unbiased genomic diagnostic testing methods, such as clinical exome sequencing, into the diagnostic evaluation has greatly improved time to diagnosis,2,3 but is limited in its ability to detect certain forms of genetic mutation such as repeat expansion disorders.4,5 The spinocerebellar ataxias (SCAs) are a diverse group of neurodegenerative disorders characterized by an autosomal dominant pattern of inheritance and cerebellar symptoms, currently consisting of at least 43 distinct clinical entities, of which 11 are repeat expansion disorders.6 One of these conditions, spinocerebellar ataxia 36 (SCA36), is caused by a GGCCTG hexanucleotide repeat expansion in the first intron of the pre-mRNA processing gene, nucleolar protein 56 (NOP56).7 Supplemental data at Neurology.org/ng From the Program in Neurogenetics (J.M.V., T.D., D.H.G., S.P., B.L.F.), Department of Neurology and Department of Human Genetics (D.H.G., B.L.F.), David Geffen School of Medicine, University of California Los Angeles; The Human Genetics Center (L.E.P., J.E.B.), University of Texas School of Public Health, Houston; Fundación Pública Galega de Medicina Xenómica-SERGAS (B.Q., Z.Y., M.J.S.), Instituto de Investigación Sanitaria (IDIS), Santiago de Compostela; Genomic Medicine Group (U711) (B.Q., Z.Y., M.J.S.), Centre for Biomedical Network Research on Rare Diseases (CIBERER), Institute of Health Carlos III, Madrid, Spain; Grupo del Investigación en Genética (Z.Y.), Universidad Simón Bolívar, Barranquilla, Colombia; Department of Molecular and Human Genetics (E.B., D.M., R.G., J.R.L.) and Center (J.R.L.), Baylor College of Medicine, Houston, TX; and Department of Integrative Biology and Pharmacology (D.A., R.B.), Institute of Molecular Medicine Center for Metabolic and Degenerative Diseases (R.B.), and Cell and Regulatory Biology Program of The University of Texas Graduate School of Biomedical Sciences (R.B.), McGovern Medical School at The University of Texas Health Science Center at Houston. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure 1 Pedigrees and Southern blot analysis of the SCA36 families identified in this study

(A) Affected individuals (dark fill) and index cases (arrow) are indicated. 1Genotyped individuals; *Individuals who had exome sequencing performed; gray fill indicates at-risk individuals, currently asymptomatic. All genotyped individuals in families A and B were included in the linkage analysis. (B) Southern blot analysis of patient DNA from probands of 3 of the identified SCA36 families, a negative control (2), and a positive control (1). The 3 Kb wild-type allele (arrow) and the larger expanded allele (*) are shown. SCA36 5 spinocerebellar ataxia 36.

Individuals with this disorder show progressive, counseling both before and after completion of the study. All late-onset, ataxic symptoms affecting limb, study methods were approved by the Institutional Review Board of the University of California, Los Angeles. trunk, and/or gait stability.7 To date, SCA36 prevalence has been reported in regions of Exome sequencing and array genotyping. Whole-exome 7,8 9 8 8 sequencing was performed for 3 members of 2 families (families A Japan, Spain, France, Germany, and most and B, figure 1A) at the Human Genome Sequencing Center 10 recently, in China, but its prevalence in the (HGSC) at Baylor College of Medicine as part of the Baylor- United States is undetermined. Through com- Hopkins Center for Mendelian Genomics initiative. With 0.5 ng bined methods of exome sequencing, linkage of DNA, an Illumina paired-end pre-capture library was generated as described in the BCM-HGSC protocol (hgsc.bcm.edu/content/ analysis, and fluorescent repeat primed PCR protocols-sequencing-library-construction). Four of these libraries (RP-PCR), we molecularly identified SCA36 werepooledandthenhybridizedinsolutiontotheHGSCCORE in 4 index cases, representing less than 1% of design13 (52 Mb; NimbleGen, Madison, WI) in accordance with ’ the undiagnosed population at a tertiary referral the NimbleGen SeqCap EZ Exome Library SR User s Guide. Each capture pool, containing 4 samples, was sequenced in one lane using ataxia center. the Illumina HiSeq 2000 platform in paired-end mode, with sequencing-by-synthesis reactions extended for 101 cycles from each METHODS Patient recruitment and diagnostic evalua- end and an additional 7 cycles for the index read. On average, each tion. All patients were initially seen at our tertiary referral ataxia cen- exome yielded 9.6 Gb of data with 93% of the targeted exome bases ter for disorders of gait and balance. All patients underwent covered to a depth of 20-fold or greater. Variant calls were generated a comprehensive clinical evaluation for acquired causes of ataxia prior according to the GATK best practices (Genome Analysis Toolkit 3. 1,11 to consideration of genetic causes. For enrollment in this study, 4) using the hg19 reference genome and annotated using the SVS patients were required to have negative testing for the most common Annotation Suite (Golden Helix). Only variants causing non- 6,12 genetic ataxias worldwide, specifically SCA1, SCA2, SCA3, SCA6, synonymous protein-coding changes or splice-site variants were and SCA7, and Friedreich ataxia if clinically appropriate. Age at onset included in the analysis. Annotated variants were filtered according or mode of inheritance was not used as exclusion criteria to prevent to a minor allele frequency (MAF) using the Exome Aggregation 3 bias against phenotypic variability or inaccurate assessment of famil- Consortium (ExAC) database14 (r.0.2) (exac.broadinstitute.org) at ial association in relatives who could not be directly assessed clinically. an MAF of 0.1% for genes containing a single heterozygous variant, Standard protocol approvals, registrations, and patient and 1% for genes containing multiple heterozygous variants or consents. Written informed consent was obtained to collect homozygous variants. Variants remaining after filtering were com- DNA for genetic analysis. All patients were provided genetic pared to a list of 2,256 genes identified by the keyword “ataxia” in

2 Neurology: Genetics theOnlineMendelianInheritanceinMandatabase15 (omim.org) Hybridization with the labeled probe was performed at 65°C over- and assessed for their clinical significance as previously described.3 night. The membrane was developed by X-ray film exposure. In parallel to the exome workflow, Illumina Infinium Human Exome v1-2 array data were generated for linkage analysis from 6 RESULTS Identification of 2 SCA36 families by exome members of family A and 2 members of family B (figure 1A). sequencing and linkage analysis. We independently per- Array data quality assessment included orthogonal confirmation formed exome sequencing on 3 individuals from 2 of sample identity and purity using the Error Rate In Sequencing undiagnosed families (families A and B, figure 1A) (ERIS) pipeline developed at the BCM:HGSC. Using an “e- GenoTyping” approach, ERIS screens all sequence reads for exact with adult-onset dominant cerebellar ataxia given matches to probe sequences defined by the variant and position of the high diagnostic yield of this testing in patients interest. A successfully sequenced sample must meet quality con- previously screened for the most common genetic trol metrics of ERIS single nucleotide polymorphism (SNP) array forms of ataxia.2,3,19,20 However, we were unable to . concordance ( 90%) and ERIS average contamination rate identify any potentially clinically relevant sequence (,5%). These arrays capture more than 240,000 SNPs, which variation associated with ataxia using standard were carried forward into quality filtering, pedigree validation, 3 and linkage analyses. All insertions/deletions and nonautosomal methods. To address the possibility of a novel gene polymorphisms and all variants without an rs identifier were or a form of genetic variation not detected by exome removed. In addition, SNPs were removed for high missingness sequencing (e.g., noncoding, copy number variants, (.5%) or mapping to identical locations (duplicates) resulting in or repeat expansions), we separately performed gen- 227,103 variants for analysis. Pedigrees were validated using the otyping using a cytogenomic SNP array and sub- prePRIMUS QC pipeline to estimate pairwise kinship and were sequent linkage analysis in both families. Among reconstructed in PRIMUS.16,17 these data, we identified 2 regions on chromosome 20 Linkage analysis. To identify overlapping haplotypes shared that reached maximum theoretical LOD scores across identically by descent (IBD) by all genotyped cases and not families A and B containing a known clinical ataxia shared by genotyped controls in families A and B (figure gene, NOP56, causing disease through repeat 1A), multifamily parametric linkage analysis was conducted using a fully penetrant dominant model with no phenocopies expansion (figure e-1). Exome analysis had indicated 5 18 (f0,1,2 0,1,1) using ALLEGRO. Logarithm of the odds that families A and B did not possess any rare protein- (LOD) scores generated from linkage in these families are coding variants in any genes potentially associated unable to meet genome-wide significance thresholds because with their phenotype in any of the identified linked of power (max LOD given the information across the pedigrees regions (figure e-1). The absence of any clinically is 1.2); the purpose of these analyses was to identify candidate relevant rare variants in ataxia genes within the regions where IBD sharing was consistent with the model of inheritance for disease in both families for further analysis sequenced affected family members (figure 1A) rather than to identify regions of genome-wide significant prompted us to evaluate these 2 families for the repeat linkage. Rare (MAF ,0.1% in ExAC), nonsynonymous and expansion in NOP56 associated with SCA36.7 splice-site variants from exome sequencing falling within Identification of an expanded GGCCTG repeat in linkage regions (peaks reaching maximum LOD, ;58.6 megabases genome wide) were identified and evaluated as NOP56 in families A and B. RP-PCR analysis was used potential disease candidates (figure e-1 at Neurology.org/ng). to assess for expansion in the NOP56 gene. This method specifically detects the presence of an expan- SCA36 repeat expansion testing. DNA was isolated from the peripheral blood and purified using the Gentra Puregene Blood Kit sion through the generation of multiple products of (Qiagen, Hilden, Germany). Fluorescent RP-PCR was im- varying size from different amplification sites, leading plemented,9 and PCR products were separated then analyzed on an to a stuttering effect (figure e-2). RP-PCR identified ABI Prism 3730 XL Analyzer (Retrogen, San Diego, CA). Peak the pathogenic hexanucleotide repeat expansion in Scanner 2 software was used to assess for the presence of the SCA36 intron 1 of NOP56 in both families A and B (figure e- hexanucleotide repeat expansion in NOP56. This method specifically 2), confirming the diagnosis of SCA36. Because RP- detects expanded alleles and does not assess normal NOP56 alleles. PCR lacks the ability to quantitate expansion size, Southern blot. For confirmation of repeat expansions, Southern Southern blot analysis was used as a second confir- blot analysis was conducted on DNA samples of 3 affected individ- matory technique, detecting NOP56 repeat ex- m uals (figure 1B). Genomic DNA (10 g) was digested overnight pansions in the probands from families A and B with BsmI (New England Biolabs, Ipswich, MA). Digested DNA (figure 1B). Given that SCA36 is considered a rare was resolved on a 0.8% agarose gel, depurinated with 0.25 M HCl, denatured with 1.5 M NaCl, 0.5 M NaOH neutralized with 1.5 M repeat expansion disorder, but has been observed to NaCl, 0.5 M Tris-HCl (pH 7.0), and transferred with 203 saline- have varying prevalence in different ethnic pop- sodium citrate buffer onto a Hybond-N1 nylon membrane (GE ulations,7–10 the finding of 2 consecutive SCA36 Health care, Piscataway, NJ). DNA was immobilized on the families prompted us to evaluate our entire clinical membrane by UV cross-linking. A 464-bp probe for hybridization cohort for the disease to determine the prevalence in was synthesized from genomic DNA by PCR using primers10 a US ataxic population. We analyzed 577 index cases flanking exons 2 and 3 of NOP56. The PCR-amplified probe DNA was used as template to make a 32P-labeled probe using random using fluorescent RP-PCR. The demographics of this hexamer primers (#58875; Invitrogen, Carlsbad, CA), Klenow patient population including age, sex, ethnicity, and polymerase (#M0210S, NEB) and dNTP’s containing [32P]-dCTP. phenotype are shown in table 1. Only 2 additional

Neurology: Genetics 3 origins of these 4 families were either white, non- Table 1 Patient demographics Hispanic, European (2 families), or Asian (1 Viet-

Phenotype Percentage of cohort namese and 1 Japanese family). The clinical features

Spinocerebellar ataxia 36.0 of these 6 individuals (5 men and 1 woman, mean age at onset: 44.5 6 5.8 years) are summarized in table 2. Pure cerebellar ataxia 31.0 All affected individuals exhibited an ataxia gait, con- Multiple system atrophy 17.0 sistent with cerebellar dysfunction, on examination. Spastic ataxia 14.0 Other symptoms included limb ataxia or spasticity 4.0 and altered reflexes. Saccadic ocular pursuit was Spastic paraplegia 3.0 present in 5 of 6 individuals, predominantly saccadic

Leukodystrophy with ataxia 2.0 pursuit or overshoot, whereas hearing impairment was reported in only 1 of 6 individuals. Two of 6 Other neurologic phenotype 7.0 patients presented with mild cerebellar atrophy de- Ethnicity tected using neuroimaging. White, non-Hispanic 76.7

White, Hispanic, or Latino 11.1 DISCUSSION We evaluated 577 index cases seen at

Asian 9.1 a tertiary referral ataxia center in the United States and identified the NOP56 hexanucleotide repeat Chinese 1.2 expansion in 4 families. When compared with other Indian 1.0 reports, the prevalence of SCA36 in the United States Korean 1.0 (n 5 577, 0.7%) appears to be less frequent than in Japanese 0.8 most other countries evaluated to date, including Vietnamese 0.6 Spain (n 5 160, 6.3%),9 France (n 5 270, 4.4%),8 5 7 Taiwanese 0.4 Western Japan (n 251, 3.6%), and Eastern Japan 5 8 Indonesian 0.2 (n 231, 2.2%) and is more similar to that of China (n 5 512, 0.6%)10 and Germany (n 5 175, 0.0%)8 Unspecified 3.9 (table 3). It should be noted that our cohort was of Black 3.6 general undiagnosed patients with ataxia, unselected Sex by mode of inheritance or other features, in whom the Male 48.6 most frequent genetic etiologies had been previously

Female 51.4 excluded. As a tertiary referral center, we cannot esti-

Average age 54.5 6 17.1 y mate the number of patients with more common genetic etiologies identified in the community, and therefore not referred, so the overall prevalence in all positive cases were identified, thus bringing the total undiagnosed ataxia cases is likely lower. Consistent to 4/577 cases (0.7%). with previous reports, SCA36-positive individuals in Clinical characteristics of SCA36 in US families. SCA36 our cohort descended from European and East Asian was observed in 4 index cases from a cohort of 577 backgrounds. Despite the majority of our cohort patients (families A–D; figure 1A). An additional 2 descending from European or Asian ancestry (table positive cases were subsequently identified (1 each 1), we observe a much lower prevalence than coun- from families A and B, respectively). The ancestral tries such as Spain, France, or Japan.7–9 This may

Table 2 Clinical findings of SCA36 patients reported in this study

Age at Gait Limb Saccadic Reflexes in Hearing Cerebellar Family Ethnicity ID Sex Age, y onset, y ataxia ataxia ocular pursuit Dysarthria Fasciculation lower limbs impairment atrophy

A White, non-Hispanic III-1 M 60 45 111 2 Face [ 22 (English and German)

III-4 M 61 50 111 1 Face, tongue, [ 12 right bicep

B Asian (Vietnamese) II-2 M 59 46 12 1 1 Absent [ 21

II-3 F 55 41 112 1 Absent Normal 22

C Asian (Japanese) II-5 M 73 50 12 1 1 Absent [ 22

D White, non-Hispanic III-1 M 42 35 221 2 Absent Normal 21 (European)

15present; 25absent; [ 5 increased; SCA36 5 spinocerebellar ataxia 36.

4 Neurology: Genetics Table 3 Comparison of clinical characteristics of US SCA36 patients with other SCA36 populations globally

Overall USa Han Chinese10 Western Japan7 Spain9 France8 Eastern Japan8

Sex 5 M/1 F 3 M/2 F 8 M/6 F 20 M/24 F 8 M/12 F 4 M/4 F

Age at onset, y 44.5 6 5.8 44.8 6 3.8 53.1 6 3.5 52.8 6 7.4 50.0 6 6.9 52.3 6 8.6

Age at examination, y 59.3 6 10.0 55.8 6 4.5 67.1 6 8.5 63.8 6 12.5 61.3 6 10.0 61.0 6 8.3

Gait ataxia 100.0% 100.0% 100.0% 98.0% 100.0% 88.0%

Limb ataxia 50.0% 100.0% 93.0% 88.0% NA NA

Saccadic ocular pursuit 66.7% 100.0% 93.0% 88.0% 39.0% 63.0%

Dysarthria 66.7% 100.0% 100.0% 64.0% 61.0% 100.0%

Tongue fasciculation 16.7% 20% 71.0% 61.0% 12.0% 63.0%

Muscle fasciculation 33.3% 80.0% 64.0% NA NA NA

Hyperreflexia 66.7% 100.0% 79.0% 41.0% 67.0% 63.0%

Hearing impairment 16.7% 80.0% NA 74.0% 44.0% 88.0%

Abbreviations: NA 5 not available; SCA36 5 spinocerebellar ataxia 36. a Current study. stem partially from the inclusion criteria defining our R.G., J.R.L., and B.L.F. contributed to next-generation sequencing cohort, but likely also reflects regional variation across design and analysis. L.E.P. and J.E.B. conducted all bioinformatics anal- yses. D.A. and R.B. performed the Southern blot analysis as shown in Europe and Asia, as noted in the studies from China figure 1B. J.M.V., T.D., and B.L.F. wrote the manuscript, and all au- and Germany,8,10 for example. One of our families thors were responsible for its review and critique. (family B) identifies their family origins as emanating from Vietnam, which would be the first reported case ACKNOWLEDGMENT of SCA36 from that country. The authors thank all the patients and their families who contributed to this study. B.L.F.thanks Xizhe Wang and Stephanie Pang for technical Clinical examination of those 6 affected SCA36 assistance. patients in our cohort confirms many of the previ- ously reported SCA36 symptoms, including gait STUDY FUNDING and limb ataxia, dysarthria, increased lower limb This work was supported in part by a US National Human Genome reflexes, polyneuropathy, saccadic pursuit, and cogni- Research Institute and National Heart Lung and Blood Institute grant (U54HG006542) to the Baylor-Hopkins Center for Mendelian Ge- tive impairment (table 3). Facial or tongue fascicu- nomics, the National Institute for Neurologic Disorders and Stroke lations are typically seen in more than 50% of (grant R01NS082094 to Dr. Fogel), and the National Ataxia Foundation reported SCA36-positive cases with the exception (Young Investigator Award to Dr. Fogel). The research described was of France (12%) and were present in 2 of 6 (33%) supported by the NIH/National Center for Advancing Translational Sci- ence UCLA Clinical and Translational Science Institute grant SCA36-positive individuals in this study. Hearing UL1TR000124. Dr. Fogel acknowledges the support through donations impairment is present in more than 40% of previ- to the University of California by the Rochester Ataxia Foundation. The ously reported SCA36-positive individuals and was authors also received support from the Spanish Institute of Health Carlos III (grant PI12/00742) and FEDER funds. Z.Y. was supported by a grant also reported in 2 of 6 individuals from family B from Fundación Carolina. R.B. was supported by grants from the NIH, (figure1Aandtable2),althoughdetailedaudiologic National Institute of Arthritis and Musculoskeletal and Skin Diseases studies were not conducted on all the affected patients (R01AR059847), and the National Institute of Diabetes and Digestive in this study. In summary, SCA36 represents a small and Kidney Diseases (R01DK092590). D.A. was supported by the Amer- ican Heart Association (15POST25090134). percentage of the undiagnosed cerebellar ataxia popu- lation in the United States. Given that SCA36 is DISCLOSURE a repeat expansion disorder that would not be detected J.M. Valera, T. Diaz, and L.E. Petty report no disclosures. B. Quintáns by the high-yield genomic diagnostic tests performed has been a consultant for Genomic Consulting S.L. Z. Yáñez reports no early in a patient’sevaluation,3 repeat expansion testing disclosures. E. Boerwinkle has received research support from NIH. D. Muzny has received research support from NIH and the Simons Foun- should be considered in individuals with appropriate dation. D. Akhmedov has received research support from the American clinical phenotypes and either a sporadic or dominant Heart Association. R. Berdeaux has received research support from NIH. mode of inheritance. M.J. Sobrido has received speaker honoraria from Actelion Pharmaceut- icals; has served on the editorial boards of Human Mutation and Applied & Translational Genomics; is the CEO of Genomic Consulting; has AUTHOR CONTRIBUTIONS a private clinical neurology practice and performs genetic diagnosis of B.L.F., J.E.B., M.J.S., and J.R.L. contributed to the conception and ataxias; and has received research support from Actelion Pharmaceuticals, design of the research project, and J.M.V., T.D., and L.E.P. were respon- Instituto de Salud Carlos III (Spain), and the Asociación Galega de sible for its execution. S.P., B.L.F., and D.H.G. supervised the clinical Ataxias (AGA). R. Gibbs has received travel funding/speaker honoraria data, samples, and patient care. B.Q. and Z.Y. contributed to the RP- from Fusion Conference; has served on the editorial board of Genome PCR protocol design and control sample preparation. E.B., D.M., Research; and has received research support from NIH. J.R. Lupski has

Neurology: Genetics 5 stock ownership in 23andMe; is a paid consultant for Regeneron Phar- 5. Fogel BL, Satya-Murti S, Cohen BH. Clinical exome maceuticals; has stock options in Lasergen; is a member of the scientific sequencing in neurologic disease. Neurol Clin Pract advisory boards of Baylor Genetics Laboratories, Lasergen Inc., Regener- 2016;6:164–176. on Pharmaceuticals, and BioPontis Alliance for Rare Diseases; receives 6. Durr A. Autosomal dominant cerebellar ataxias: polyglut- publishing royalties from Humana Press; is an employee of Baylor College amine expansions and beyond. Lancet Neurol 2010;9: of Medicine (the Department of Molecular and Human Genetics at 885–894. Baylor College of Medicine derives revenue from the chromosomal mi- croarray analysis and clinical exome sequencing offered in the Baylor 7. Kobayashi H, Abe K, Matsuura T, et al. Expansion of Genetics Laboratory [bmgl.com]); has received research support from intronic GGCCTG hexanucleotide repeat in NOP56 NIH; and is a coinventor on multiple United States and European pat- causes SCA36, a type of spinocerebellar ataxia accompa- ents related to molecular diagnostics for inherited neuropathies, eye dis- nied by motor neuron involvement. Am J Hum Genet eases, and bacterial genomic fingerprinting. D.H. Geschwind has served 2011;89:121–130. on the scientific advisory board for Ovid Therapeutics; has served on the 8. Obayashi M, Stevanin G, Synofzik M, et al. Spinocerebel- editorial boards of Cell, Molecular Autism, Molecular Neuropsychiatry, lar ataxia type 36 exists in diverse populations and can be Nature, Nature Genetics, Nature Neuroscience, Neurology, Neuron, New caused by a short hexanucleotide GGCCTG repeat expan- England Journal of Medicine, Public Library of Science (PLoS), Genetics, sion. J Neurol Neurosurg Psychiatry 2015;86:986–995. Science, and Translational Psychiatry; holds patents for Peripheral Gene 9. García-Murias M, Quintáns B, Arias M, et al. “Costa da Expression Biomarkers for Autism, Genetic Risk Factor for Neurodegen- ” erative Disease, Compositions and Methods for Diagnosing and Treating Morte ataxia is spinocerebellar ataxia 36: clinical and – Brain Cancer and Identifying Neural Stem Cells, Genetic Variants genetic characterization. Brain 2012;135:1423 1435. Underlying Human Cognition: Novel Diagnostic and Therapeutic Tar- 10. Lee YC, Tsai PC, Guo YC, et al. Spinocerebellar ataxia gets, Peripheral Gene Expression Biomarkers for Autism, Brain Gene type 36 in the Han Chinese. Neurol Genet 2016;2:e68. Expression Changes Associated with Autism Spectrum Disorders, Full doi: 10.1212/NXG.0000000000000068. Biomarkers in Friedreich’s Ataxia (provisional patent application), Signal- 11. Fogel B, Perlman S. Cerebellar disorders: balancing the ing Networks Causing Neurodevelopmental Disorders In Human Neu- approach to cerebellar ataxia. In: Tuite PJ, Gálvez-Jiménez rons, Genes Dysregulated in Autism: Potential Biomarkers and N, editors. Uncommon Causes of Movement Disorders. Therapeutic Pathways, Peripheral Gene Expression Biomarkers for New York: Cambridge University Press; 2011:198–216. Autism, A Genetic Target for Treatment of Individuals with Neurocog- 12. Fogel BL, Lee JY, Lane J, et al. Mutations in rare ataxia nitive Spectrum Disorders, Neuronal Regeneration, Frataxin Knock- Down Mouse, Jakmip1 Knockout Mouse, and Cyfip1 Transgenic genes are uncommon causes of sporadic cerebellar ataxia. – Mouse; receives publishing royalties from Oxford University Press; has Mov Disord 2012;27:442 446. been a consultant for Ovid Therapeutics Ltd; has received research sup- 13. Bainbridge MN, Wang M, Yuanqing W, et al. Targeted port from Takeda Pharmaceutical Company, NIH, the Simons Founda- enrichment beyond the consensus coding DNA sequence tion, Adelson Medical Research Foundation, The Tau Consortium, and exome reveals exons with higher variant densities. Genome Ovid Therapeutics Ltd; and receives license fee payments for Mouse Biol 2011;12:R68. model of Friedreich Ataxia. S. Perlman has received research support 14. Exome Aggregation Consortium (ExAC). ExAC Browser. from the University of California, Los Angeles and the National Ataxia Cambridge, MA: 2014. Available at: http://biorxiv.org/ Foundation. J.E. Below has received research support from Sanofi Inno- content/early/2015/10/30/030338. Accessed December vation Awards Program, Tulane National Primate Research Center, and 2015. NIH. B.L. Fogel has received speaker honoraria from the American Academy of Neurology; has received travel funding from the National 15. McKusick-nathans Institute of Genetic Medicine; Johns Ataxia Foundation; has received speaker honoraria and travel funding Hopkins Medicine; National Human Genome Research from American Physician Institute for Advanced Professional Studies; Institute. Online Mendelian Inheritance in Man. Avaiable and has received research support from NIH and the National Ataxia at: http://omim.org/. Accessed December 2015. Foundation. Go to Neurology.org/ng for full disclosure forms. 16. Staples J, Qiao D, Cho MH, et al. PRIMUS: rapid recon- struction of pedigrees from genome-wide estimates of Received September 20, 2016. Accepted in final form May 10, 2017. identity by descent. Am J Hum Genet 2014;95:553–564. 17. Staples J, Ekunwe L, Lange E, et al. PRIMUS: improving REFERENCES pedigree reconstruction using mitochondrial and Y haplo- 1. Shakkottai VG, Fogel BL. Autosomal dominant spinocer- types. Bioinformatics 2016;32:596–598. ebellar ataxia. Neurol Clin 2013;31:987–1007. 18. Gudbjartsson DF, Jonasson K, Frigge ML, et al. Allegro, 2. Pyle A, Smertenko T, Bargiela D, et al. Exome sequencing a new computer program for multipoint linkage analysis. in undiagnosed inherited and sporadic ataxias. Brain 2015; Nat Genet 2000;25:12–13. 138:276–283. 19. Ohba C, Osaka H, Iai M, et al. Diagnostic utility of whole 3. Fogel BL, Lee H, Deignan JL, et al. Exome sequencing in exome sequencing in patients showing cerebellar and/or ver- the clinical diagnosis of sporadic or familial cerebellar mis atrophy in childhood. Neurogenetics 2013;14:225–232. ataxia. JAMA Neurol 2014;71:1237–1246. 20. Sawyer SL, Schwartzentruber J, Beaulieu CL, et al; 4. Fogel BL, Lee H, Strom SP, et al. Clinical exome sequenc- FORGE Canada Consortium. Exome sequencing as a diag- ing in neurogenetic and neuropsychiatric disorders. Ann nostic tool for pediatric-onset ataxia. Hum Mutat 2014; NY Acad Sci 2016;1366:49–60. 35:45–49.

6 Neurology: Genetics Loss-of-function variants of SCN8A in intellectual disability without seizures

Jacy L. Wagnon, PhD* ABSTRACT * Bryan S. Barker, BA Objective: To determine the functional effect of SCN8A missense mutations in 2 children with Matteo Ottolini, MS intellectual disability and developmental delay but no seizures. Young Park, PhD Methods: Genomic DNA was analyzed by next-generation sequencing. SCN8A variants were Alicia Volkheimer, MS introduced into the Na 1.6 complementary DNA by site-directed mutagenesis. Channel activity Purnima Valdez, MD v was measured electrophysiologically in transfected ND7/23 cells. The stability of the mutant Marielle E.M. Swinkels, channels was assessed by Western blot. MD Manoj K. Patel, PhD† Results: Both children were heterozygous for novel missense variants that altered conserved res- Miriam H. Meisler, PhD† idues in transmembrane segments of Nav1.6, p.Gly964Arg in D2S6 and p.Glu1218Lys in D3S1. Both altered amino acids are evolutionarily conserved in vertebrate and invertebrate channels and are predicted to be deleterious. Neither was observed in the general population. Both variants Correspondence to completely prevented the generation of sodium currents in transfected cells. The abundance of Dr. Meisler: Nav1.6 protein was reduced by the Glu1218Lys substitution. [email protected] Conclusions: Haploinsufficiency of SCN8A is associated with cognitive impairment. These obser- vations extend the phenotypic spectrum of SCN8A mutations beyond their established role in epileptic encephalopathy (OMIM#614558) and other seizure disorders. SCN8A should be con- sidered as a candidate gene for intellectual disability, regardless of seizure status. Neurol Genet 2017;3:e170; doi: 10.1212/NXG.0000000000000170

GLOSSARY cDNA 5 complementary DNA; HEK 5 human embryonic kidney.

Whole-exome sequencing has revealed a major role for de novo mutations in the etiology of spo- radic intellectual disability.1 Between one-third and one-half of sporadic cases may be accounted for by de novo mutations in genes required for neuronal development and synaptic transmission. The neuronal SCN8A (Nav1.6) is concentrated at the axon initial segment and at nodes of Ranvier of myelinated axons.2 Exome analysis for SCN8A mutations has thus far focused on children with seizure disorders.3 More than 150 missense mutations of SCN8A have been identified, and gain-of-function hyperactivity is the most common pathogenic mechanism for seizures. By contrast, we previously described a loss-of-function, protein truncation allele of SCN8A that cosegregated with cognitive impairment in a family without seizures.4 To follow up on that observation, we have now examined the functional effects of 2 SCN8A missense mutations identified by exome sequencing in children with intellectual disability who also did not have

*These authors contributed equally to this work as cofirst authors. †These authors contributed equally to this work as colast authors. From the Department of Human Genetics (J.L.W., Y.P., M.H.M.), University of Michigan, Ann Arbor; Department of Anesthesiology (B.S.B., M.O., M.K.P.) and Neuroscience Graduate Program (B.S.B., M.K.P.), University of Virginia, Charlottesville; Department of Medicine (A.V.), Veterans Affairs Medical Center (A.V.), and Department of Pediatrics (P.V.), Duke University, Durham, NC; and Department of Medical Genetics (M.E.M.S.), University Medical Centre Utrecht, the Netherlands. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 mutagenesis with QuikChange II XL (Agilent Technologies, SCN8A Figure 1 Location and evolutionary conservation of mutations in Santa Clara, CA) as described.5 Two independent mutagenesis individuals with intellectual disability experiments generated cDNA clones A and B for each mutation. The 6-kb open reading frame was resequenced, and clones lacking other mutations were analyzed.

Electrophysiology. Neuron-derived ND7/23 cells (Sigma Aldrich, St. Louis, MO) were cultured and transfected as described.5 Electrophysiologic recordings of fluorescent cells were performed 48 hours after transfection in the presence of 500 nM tetrodotoxin to block endogenous sodium currents. Currents were recorded using the whole-cell configuration of the patch-clamp recording technique.5

Western blot. Human embryonic kidney (HEK) 293 cells were

cultured at 37°C, transfected with Nav1.6 cDNA, and lysates were prepared and analyzed 24 hours after transfection as described5 using affinity-purified polyclonal rabbit anti-Scn8a antibody (Millipore # AB5580, lot 2784259, 1:500 dilution).

RESULTS Identification of novel missense variants of SCN8A. Patient 1 is a 7-year-old girl who experienced global developmental delay and hypotonia in early childhood. She walked and spoke her first words at 18 months. She is receiving special education services at school and is repeating the first grade due to below average academic attainment. Psychoeducational testing re- vealed receptive-expressive language disorder and bor- derline intellectual functioning with a diagnosis of social communication disorder. She did not meet cri- teria for autism spectrum disorder. (A) Four-domain structure of the voltage-gated sodium channel a subunit. p.Gly964Arg (G964R) is located in transmembrane segment 6 of domain II. p.Glu1218Lys (E1218K) is Attention-deficit hyperactivity disorder was located in transmembrane segment 1 of domain III. (B) Evolutionary conservation of residue diagnosed at 6 years and hasrespondedtomethyl- G964 in multiple species. (C) Conservation of residue E1218 in multiple species. a 5 anole; phenidate. Exome sequencing revealed the SCN8A c 5 chicken; dpara 5 drosophila “paralytic”;f5 fugu; h 5 human; jscn 5 jellyfish sodium . channel; m 5 mouse; z 5 zebrafish. Amino acids are indicated by the single-letter code; dots variant c.2890G C (p.Gly964Arg; G964R) which represent identity to the human amino acid. arose de novo and was not present in either parent. Gly 964 is located in transmembrane segment 6 seizures. Both mutations caused complete loss of domain II (D2S6) and is highly conserved of channel activity, confirming the role of loss- through invertebrate and vertebrate evolution of-function mutations of SCN8A as a cause of (figure 1, A and C). isolated cognitive impairment. Patient 2 is a 10-year-old boy who was born after a pregnancy complicated by polyhydramnios. Devel- METHODS Molecular diagnostics. Exome sequencing for opment was delayed from birth. Early ataxic gait patient 1 was performed by GeneDX (Gaithersburg, MD). In resolved with age. Behavioral problems included tem- addition to the SCN8A variant, a frameshift mutation in GJB2 per tantrums. Metabolism and brain MRI were nor- (c.167delT, p.L56RfsX26) was inherited from an unaffected par- mal. There were no dysmorphic features. Exome ent. Exome sequencing for patient 2 was performed at the labo- sequencing and analysis of 770 genes identified the ratory for DNA Diagnostics in the University Medical Center SCN8A variant c. 3652G.A (p.Glu1218Lys; Utrecht. In addition to the SCN8A variant, the PDHA1 variant (c.520G.A, p.Ala174Thr) was present in the child and an unaf- E1218K) located at the distal terminus of transmem- fected grandfather. Analysis of copy number variation and Fragile brane segment 1 in domain III (D3S1). This residue X expansion for patient 2 were negative. Procedures were is highly conserved through evolution (figure 1, A and approved by the institutional ethics standard committees. B). The variant was not present in the maternal Standard protocol approvals, registrations, and patient genome; the father was not available for testing. consents. Written consent for research was obtained from the Additional clinical features are detailed in the guardians of both patients whose variants were studied. table. Both mutations were predicted to be deleterious by in silico prediction programs. Neither mutation was Site-directed mutagenesis of Nav1.6 complementary DNA. Mutations were introduced into the tetrodotoxin-resistant previously observed in patients or in the Exome

mouse complementary DNA (cDNA) Nav1.6R by site-directed Aggregation Consortium Database.

2 Neurology: Genetics Table Clinical features of patients with intellectual disability and SCN8A mutations

Patient 1 Patient 2

Sex Female Male

Age 7 y 10 y

Nucleotide change c.2890G.C c. 3652G.A

Inheritance De novo Unknown (father unavailable)

Protein change G964R E1218K

Channel domain D2S6 D3S1

In silico predictions Deleterious Deleterious

CADD 25 32

PROVEAN 27.4 23.8

Polyphen 2 0.9 1.0

Channel activity Inactive Inactive observed

Protein Stable Unstable

Diagnosis Language disorder borderline intellectual function Intellectual disability

IQ 73 56

Social interaction Social communication disorder; minimal interaction with Temper tantrums peers; good eye contact

Development Global delay walk, talk at 18 mo Global motor delay, severe speech delay

EEG Normal Not done

Seizures None None

MRI Not done Normal

Metabolism Not done Normal

Motor development Hypotonia; motor delay improved with time Unstable gait, resolved

Other phenotypes ADHD at 6-y special education, chronic headache, No other phenotypes finger chewing

Treatment ADHD responsive to methylphenidate No medication

Abbreviation: ADHD 5 attention-deficit hyperactivity disorder.

Electrophysiologic characterization of SCN8A mutations with the predicted molecular weight of 250 kDa, G964R and E1218K. ND7/23 cells were transfected which was not present in nontransfected cells (figure with wild-type or mutant cDNAs, and sodium 2C). However, repeated transfections of the E1218K currents were recorded (figure 2, A and B). The mutant cDNA detected only a low level of protein macroscopic Na current in nontransfected cells was (figure 2, C and D), indicating that this mutation 210.0 6 1.8 pA/pF (n 5 8). Cells transfected with reduces protein stability. wild-type Na 1.6 exhibited a robust macroscopic Na v DISCUSSION The patients described exhibited current of 2102 6 12 pA/pF (mean 6 SEM) (n 5 developmental delay and cognitive impairment but 30). The current in cells transfected with mutant no history of seizures. Each carried a pathogenic mis- cDNA did not differ from nontransfected cells: sense variant of SCN8A that altered a highly con- G964R, 25.2 6 0.5 pA/pF (n 5 30) and E1218K, served amino acid residue in a transmembrane 29.8 6 1.5 pA/pF (n 5 27). To confirm the loss of segment of the channel, resulting in loss of function. activity, site-directed mutagenesis was repeated and SCN8A is one of the most conserved genes in the a second independent clone with each mutation was mammalian genome, with an unusually low rate of analyzed, with the same result. Both missense var- coding variation.6 The low frequency of frameshift iants thus result in a complete loss of channel and nonsense mutations among 60,000 individual activity. exomes was used to calculate the probability of 1.0 Expression of mutant protein. To evaluate the mecha- that SCN8A is intolerant to haploinsufficiency.7 The nism for loss of channel activity, we examined protein population data support our conclusion that haploin- abundance in transfected HEK cells. Wild-type sufficiency of SCN8A is responsible for cognitive

Nav1.6 and G946R cDNAs generated a protein impairment in these patients.

Neurology: Genetics 3 intellectual disability. Chronic reduction of neuronal Figure 2 Mutations E1218K and G964R result in loss of Na 1.6 function v activity may alter the dynamics of synaptic plasticity during maturation and lead to aberrant cerebral cir- cuitry and intellectual disability. In contrast to the loss-of-function variants described here, we previously identified 8 gain-of- function variants resulting in channel hyperactivity resulting in epileptic encephalopathy (reviewed in ref- erence 3). We also found loss-of-function missense variants in 2 patients with seizures,11,12 indicating that genetic background influences clinical outcome. Both gain-of-function and loss-of-function variants of the related sodium channel SCN2A have also been asso- ciated with seizures.13 A recent analysis of de novo mutations in more than 7,000 individuals with devel- opmental disorders identified 5 missense variants of SCN8A in patients with seizures and 2 missense var- iants of SCN8A in patients with cognitive impair- ment but no seizures.14 We would suggest that the former variants are likely to cause channel hyperac- tivity and the latter to cause loss of function. Protein truncation variants of SCN8A are underrepresented in all populations studied to date, including controls, patients with intellectual disability, and patients with seizures. It seems likely that these protein truncations are associated with distinct disorders not yet subjected to large-scale exome sequencing, such as neuromus- cular and movement disorders. The de novo mutation of SCN8A in patient 1 is consistent with the growing recognition of the role of de novo mutations in sporadic intellectual disability. In 3 earlier studies examining 142 individuals with intellectual disability, 1 de novo missense mutation of (A) Averaged current-voltage (I-V) relation for cells expressing WT (black) (n 5 30), E1218K 14–17 (red) (n 5 27), and G964R (blue) (n 5 30) Nav1.6 channels and nontransfected controls (gray) SCN8A was detected. Better estimates of the (n 5 8). Data represent mean 6 SEM. (B) Representative traces of families of Na currents quantitative contribution of SCN8A mutations to from ND7/23 cells transfected with the indicated Nav1.6 complementary DNA (cDNA). (C sporadic and inherited forms of isolated cognitive and D) Western blot of transfected human embryonic kidney 293 cells (30 mg protein) impairment will emerge from additional large-scale immunostained with rabbit polyclonal anti-Nav1.6 antibody. NT 5 nontransfected; WT 5 wild-type cDNA. Lanes A and B represent independently generated mutant cDNA clones screening of patient populations. for each mutation. AUTHOR CONTRIBUTIONS Impaired cognition in the patients is also consis- Jacy L. Wagnon provided molecular data and contributed to writing the manuscript. Bryan S. Barker provided electrophysiologic data and manu- tent with the evidence that heterozygous loss-of- script editing. Matteo Ottolini provided electrophysiologic data. Young function mutations in mouse Nav1.6 also result in Park provided molecular data. Alicia Volkheimer provided clinical infor- cognitive and behavioral deficits without spontaneous mation for patient 1. Purnima Valdez provided clinical information for patient 1. Marielle E.M. Swinkels provided clinical information for seizures. Impaired learning in water maze and eye- patient 2. Manoj K. Patel provided experimental data and edited the 8,9 blink conditioning tests and elevated anxiety in manuscript. Miriam H. Meisler initiated the study and drafted the the open-field test10 have been described. manuscript. At the cellular level, complete inactivation of mouse Scn8a reduces repetitive firing, resurgent cur- STUDY FUNDING Supported by NIH (R01 NS 34509 [M.H.M.] and R01 NS 75157 rent and persistent current in cerebellar Purkinje cells, [M.K.P.]). prefrontal cortical pyramidal cells, and hippocampal CA1 cells.2 These changes decrease the frequency of DISCLOSURE action potentials. Reduced neuronal activity is a likely J.L. Wagnon has received research funding from the NIH and the Dravet consequence of the loss-of-function mutations of Syndrome Foundation. B.S. Barker, M. Ottolini, Y. Park, A. Volkheimer, P. Valdez, M.E.M. Swinkels, and M.K. Patel report no dis- SCN2A and receptors involved in excitatory neuro- closures. M.H. Meisler has served on scientific advisory boards for the transmission that have also been associated with Dravet Syndrome Foundation and the Cute Syndrome Foundation and

4 Neurology: Genetics has received research support from the NIH. Go to Neurology.org/ng for delay but not trace eyeblink classical conditioning. Behav full disclosure forms. Neurosci 2006;120:229–240. 9. McKinney BC, Chow CY, Meisler MH, Murphy GG. Received March 10, 2017. Accepted in final form May 11, 2017. Exaggerated emotional behavior in mice heterozygous null for the sodium channel Scn8a (Nav1.6). Genes Brain Be- REFERENCES hav 2008;7:629–638. 1. Vissers LE, Gilissen C, Veltman JA. Genetic studies in 10. Levin SI, Khaliq ZM, Aman TK, et al. Impaired motor intellectual disability and related disorders. Nat Rev Genet function in mice with cell-specific knockout of sodium 2016;17:9–18. channel Scn8a (NaV1.6) in cerebellar purkinje neurons 2. O’Brien JE, Meisler MH. Sodium channel SCN8A (Nav1.6): and granule cells. J Neurophysiol 2006;96:785–793. properties and de novo mutations in epileptic encepha- 11. de Kovel CGF, Meisler MH, Brilstra EH, et al. Char- lopathy and intellectual disability. Front Genet 2013;4: acterization of an SCN8A mutation in a patient with 213. epileptic encephalopathy. Epilepsy Res 2014;108: 3. Meisler MH, Helman G, Hammer MF, et al. SCN8A 1511–1521. encephalopathy: research progress and prospects. Epilepsia 12. Blanchard MG, Willemsen MH, Walker JB, et al. De 2016;57:1027–1035. novo gain of function and loss of function mutations of 4. Trudeau MM, Dalton JC, Day JW, Ranum LPW, Meisler SCN8A in patients with intellectual disabilities and epilep- MH. Heterozygosity for a protein truncation mutation of sies. J Med Genet 2015;52:330–337. sodium channel SCN8A in a patient with cerebellar atro- 13. Wolff M, Johannesen KM, Hedrich UBS, et al. Genetic phy, ataxia and mental retardation. J Med Genet 2006;43: and phenotypic heterogeneity suggest therapeutic impli- 527–530. cations in SCN2A-related disorders. Brain Epub 2017 5. Wagnon JL, Barker BS, Hounshell JA, et al. Pathogenic Mar 4. mechanisms of recurrent epileptogenic mutations of 14. Deciphering Developmental Disorders Study. Prevalence SCN8A in epileptic encephalopathy. Ann Clin Transl and architecture of de novo mutations in developmental Neurol 2015;3:114–123. disorders. Nature 2017;542:433–438. 6. Petrovski S, Wang Q, Heinzen EL, Allen AS, Goldstein 15. Rauch A, Wieczorek D, Graf E, et al. Range of genetic DB. Genic intolerance to functional variation and the mutations associated with severe non-syndromic sporadic interpretation of personal genomes. PLoS Genet 2013;9: intellectual disability: an exome sequencing study. Lancet e1003709. 2012;380:1674–1682. 7. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of 16. Hamdan FF, Srour M, Capo-Chichi JM, et al. De novo protein-coding genetic variation in 60,706 humans. mutations in moderate or severe intellectual disability. Nature 2016;536:285–291. PLoS Genet 2014;10:e1004772. 8. Woodruff-Pak DS, Green JT, Levin SI, Meisler MH. 17. Gilissen C, Hehir-Kwa JY, Thung GT, et al. Genome Inactivation of sodium channel Scn8a (NaV 1.6) in Pur- sequencing identifies major causes of severe intellectual kinje neurons impairs learning in Morris water maze and disability. Nature 2014;511:344–347.

Neurology: Genetics 5 Clinical and experimental studies of a novel P525R FUS mutation in amyotrophic lateral sclerosis

Lisha Kuang, PhD ABSTRACT Marisa Kamelgarn, BS Objective: To describe the clinical features of a novel fused in sarcoma (FUS) mutation in a young Alexandra Arenas, BS adult female amyotrophic lateral sclerosis (ALS) patient with rapid progression of weakness and Jozsef Gal, PhD to experimentally validate the consequences of the P525R mutation in cellular neuronal models. Deborah Taylor, MS Methods: We conducted sequencing of genomic DNA from the index patient and her family mem- Weiming Gong, PhD bers. Immunocytochemistry was performed in various cellular models to determine whether the Martin Brown, MD newly identified P525R mutant FUS protein accumulated in cytoplasmic inclusions. Clinical fea- Daret St. Clair, PhD tures of the index patient were compared with 19 other patients with ALS carrying the P525L Edward J. Kasarskis, MD, mutation in the same amino acid position. PhD . . Haining Zhu, PhD Results: A novel mutation c.1574C G (p.525P R) in the FUS gene was identified in the index patient. The clinical symptoms are similar to those in familial ALS patients with the P525L mutation at the same position. The P525R mutant FUS protein showed cytoplasmic localization Correspondence to and formed large stress granule–like cytoplasmic inclusions in multiple cellular models. Dr. Zhu: [email protected] or Conclusions: The clinical features of the patient and the cytoplasmic inclusions of the P525R Dr. Kasarskis: mutant FUS protein strengthen the notion that mutations at position 525 of the FUS protein [email protected] result in a coherent phenotype characterized by juvenile or young adult onset, rapid progression, variable positive family history, and female preponderance. Neurol Genet 2017;3:e172; doi: 10.1212/NXG.0000000000000172

GLOSSARY ALS 5 amyotrophic lateral sclerosis; EGFP 5 enhanced green fluorescent protein; fALS 5 familial amyotrophic lateral sclerosis; FTD 5 frontotemporal dementia; FUS 5 fused in sarcoma; MND 5 motor neuron disease; NLS 5 nuclear locali- zation sequence; PBS 5 phosphate-buffered saline; WT 5 wild type.

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease affecting the upper and lower motor neurons of the brain and spinal cord. It causes progressive paralysis and finally death due to respiratory failure typically within 3 years of symptom onset. Approximately 10%–15% of the ALS cases are familial ALS (fALS) and caused by mutations in several different genes. Recently, mutations in the fused in sarcoma (FUS) gene have been identified in a subset of patients with fALS.1,2 The FUS gene is located on chromosome 16p11.2 and encodes a multi- domain protein with 526 amino acids. The FUS protein is predominantly localized in the nucleus in most tissues, whereas it also has a notable cytoplasmic presence in neurons.3 Most fALS mutations are located in the C-terminal nuclear localization sequence (NLS),4 causing accumulation of FUS-containing inclusions in the cytoplasm. The FUS protein binds to RNA/ DNA and is involved in a variety of RNA metabolism pathways, including transcription,5–8 splicing,8–10 nucleocytoplasmic RNA shuttling11 and RNA transport,12 mitochondrial RNA biogenesis and function,13,14 and DNA repair.15,16

From the Molecular and Cellular Biochemistry (L.K., J.G., H.Z.), Department of Toxicology and Cancer Biology (M.K., A.A., D.S.C., H.Z.), and Department of Neurology (D.T., E.J.K.), College of Medicine, University of Kentucky, Lexington; Hefei National Laboratory for Physical Sciences at the Microscale (W.G.), University of Science and Technology of China, Anhui; Department of Neurology (M.B.), University of Louisville; and Research and Development (E.J.K., H.Z.), Lexington VA Medical Center, KY. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the University of Kentucky. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 In this study, we report a novel mutation METHODS Patient information. The index patient was c.1574C.G (p.525P.R) in FUS in a 26- a 26-year-old woman who experienced subacute onset of proxi- mal upper extremity weakness over 4–6 weeks followed by pro- year-old woman with rapid onset and progres- gression. By 5 months, she had evidence of acute and chronic sion of weakness. The P525R mutant FUS denervation on EMG testing in the upper extremities, lower protein showed cytoplasmic localization and extremities, thoracic paraspinous, and sternocleidomastoid mus- formed large stress granule–like cytoplasmic cles. The forced vital capacity was 64% of predicted; the following were normal/negative: comprehensive metabolic panel, complete inclusions in multiple cellular models. The blood count, vitamin B12, hemoglobin A1C, heavy metals, cre- phenotype is similar to fALS individuals with atine kinase, aldolase, Lyme, rapid plasma reagin, antinuclear the p.525P.L mutation, suggesting that antibody, rheumatoid factor, SSa/SSb, serum protein electro- phoresis, and HIV. She was evaluated for a second opinion mutations at position 525 in FUS cause ful- (M.B.) at 6 months when her ALSFRSr score was 37/48. She was minant motor neuron disease (MND). started on riluzole 50 mg bid. By 9 months, she was receiving total care in a nursing home. Her ALSFRSr score was 24/48. On examination, she had bifacial weakness, normal tongue contour and power, and extensive Figure 1 Identification of the P525R mutation and patient pedigree weakness of neck flexion and extension. An obese body habitus with extremity edema precluded evaluation of muscle atrophy or fasciculations. She did not have antigravity power in any upper extremity muscle and profound proximal lower extremity weak- ness. There was muscle hypotonia in the upper extremities and increased tone in the lower extremities with elevated tendon re- flexes and bilateral Babinski signs. She died of respiratory weak- ness after a 12-month course. Additional history revealed that her mother was evaluated 12 years back for upper motor neuron predominant ALS (E.J.K.). By 8 months after the onset of progressive weakness, she had grade 3/5 power in the upper extremities, grade 42/5 power in the lower extremities, normal bulbar function, and corticospinal tract findings in extremities and in the bulbar region. Detailed clinical sensory testing was normal. EMG testing demonstrated fasciculations, few fibrillations, and large amplitude polyphasic motor unit potentials. She died at the age of 40 after a 6-year progressive course.

Standard protocol approvals, registrations, and patient consents. The study was approved by the Institutional Review Board of the University of Kentucky. Written informed consent was obtained from all subjects who participated in the study.

Mutation analysis (genomic DNA sequencing). Genomic DNA was extracted from peripheral blood using the Gentra Pure- gene Blood Kit (158467; Qiagen, Hilden, Germany). Mutation analysis on FUS was performed by amplifying the exon encoding position P525 and its flanking regions, followed by PCR product sequencing.

Plasmids. The enhanced green fluorescent protein (EGFP) tagged pEGFP-C3-FUS WT, R521G, P525L, and R495X plasmids were constructed in previous studies.17,18 The P525R FUS mutant was amplified using PCR and subcloned into the pEGFP-C3 (Clontech, Mountain View, CA) vector using the BglII and KpnI sites.

Cell culture and transfection (N2A, primary cortical neurons, and human skin fibroblasts). Mouse neuroblas- toma N2A cells were cultured in Dulbecco’s Modified Eagle Medium (D5796; Sigma, St. Louis, MO) supplemented with 10% fetal bovine serum, 100 unit/mL penicillin, and 100 mg/ (A) Sequencing results of amplified patient genomic DNA. The FUS mutation, c.1574C.G mL streptomycin in a humidified incubator at 37°C under 5% (p.Pro525Arg) was identified in the patient. The antisense strand electropherogram is shown CO /95% air. Transient transfection was performed using Lipo- on the top, and the reading frame depicting the corresponding amino acid substitution is 2 shown below. The normal genomic DNA was from the unaffected brother of the patient. (B) fectamine 2000 (Invitrogen, Life Technologies, Grand Island, NY). The family pedigree of the patient with ALS. Numbers indicate the age at death based on Mouse primary cortical neuron cultures were prepared as pre- 18 family history and public records. Predicted protein composition at position 525 of FUS viously reported. Briefly, neonatal mice (strain C57BL/6; based on DNA sequencing. ALS 5 amyotrophic lateral sclerosis; FUS 5 fused in sarcoma; Jackson Laboratory, Bar Harbor, ME) were killed by decapitation P 5 proline; R 5 arginine. within 24 hours of birth. After the incubation with trypsin, cells

2 Neurology: Genetics were dissociated using pipette lavage and cultured in Neurobasal supplemented with 20% fetal bovine serum, 2 mM L-glutamine, Medium (21103049; Life Technologies) with B27 supplement 100 unit/mL penicillin, and 100 mg/mL streptomycin) in tissue

(17504044; Life Technologies), L-glutamine, and penicillin/ culture plates in a humidified incubator at 37°C under 5% CO2/ streptomycin. After the treatment of 5-fluoro-29-deoxyuridine 95% air. Fibroblast cells grew from tissue fragments and were (F0503; Sigma-Aldrich) for 4 days to kill glial cells, the primary maintained under the same conditions as above. neurons were transfected with EGFP-tagged WT or mutant FUS expression plasmids using Lipofectamine 2000 (11668; Life Immunofluorescence microscopy. N2A cells or primary corti- Technologies). The primary neurons were fixed for immunoflu- cal neurons were seeded on gelatin or poly-D-lysine hydrobromide- orescence 48 hours after the transfection. coated glass coverslips. Twenty-four hours after the transfection Human skin fibroblast cell cultures were established as previ- with EGFP-FUS, cells were rinsed with 13 PBS, fixed with 4% ously described.19 Briefly, a 3-mm punch skin biopsy was ob- formaldehyde in 13 PBS, and permeabilized with 0.25% Triton tained after informed consent from the patient with X-100 in 13 PBS. Primary fibroblast cells were cultured, fixed, and symptomatic ALS and family members who were free of neuro- permeabilized similarly as above. The primary antibodies used in logic disease. Skin biopsies were washed with phosphate-buffered this study were mouse anti-FUS (sc-47711; Santa Cruz Bio- saline (PBS), minced into small pieces, and incubated in a fibro- technology, Santa Cruz, CA), rabbit anti-G3BP1 (13057-2-AP; blast growth medium (MEM [M5650; Sigma-Aldrich] Proteintech), and goat anti-TIA1 (sc-1751; Santa Cruz). The

Figure 2 The P525R mutant FUS is mislocalized in cytoplasmic inclusions in N2A cells

N2A cells were transfected with EGFP-tagged WT or mutant FUS. Cytoplasmic inclusions of mutant FUS were colocalized with stress granule markers TIA1 and G3BP1 as indicated by arrows. Scale bar, 10 mm. FUS 5 fused in sarcoma; WT 5 wild type.

Neurology: Genetics 3 secondary antibodies were Alexa Fluor 488 donkey anti-mouse variants. Sequencing results confirmed that the (A-21202; Life Technologies), Alexa Fluor 647 donkey anti- patient was heterozygous in the FUS gene at rabbit (A-31573; Life Technologies), and Alexa Fluor 568 don- c.1574C.G and that her sibling and maternal grand- key anti-goat (A-11057; Life Technologies). The samples were mother were homozygous for normal alleles in FUS mounted using Vectashield Mounting Medium (Vector Labora- tories, Burlingame, CA). Images were acquired using a Nikon A1 (figure 1A). DNA was not available from her mother confocal microscope with a 603 objective. who died prior to the study. Examination of the ped- igree using publicly available records did not identify RESULTS P525R mutation in the index patient. Com- any other members with a diagnosis of ALS, MND, mercial genetic testing of the ALS patient (Prevention or neurologic impairment (figure 1B). Genetics, Marshfield, WI) revealed 2 normal alleles in C9orf72 with 5 and 14 GGGGCC repeats. The FUS P525R mutation is mislocalized in cytoplasmic patient was heterozygous in the FUS gene at inclusions in N2A cells and primary cortical neurons. Because the P525R mutation is located in the NLS c.1574C.G (p.Pro525Arg) and heterozygous for of FUS, we first tested whether this mutation caused a predicted benign variant in the SQSTM1 gene c. FUS mislocalization in cultured N2A cells com- C350C.T (p.Ala117Val). Sequencing of 18 addi- pared with other fALS FUS mutants. The wild- tional genes implicated in ALS, frontotemporal type (WT) FUS was located in the nucleus, dementia (FTD), and ALS/FTD (i.e., ANG, ARH- GEF28, CDH13, CHMP2B, GRN, HNRNPA1, whereas the P525R mutant FUS protein mis- HNRNPA2B1, MAPT, OPTN, PFN1, PSEN1, localized from the nucleus to the cytoplasm and PSEN2, SOD1, TARDBP, TREM2, UBQLN2, formed large inclusions (figure 2). The mis- VAPB, and VCP) did not reveal any pathologic localization and inclusions are similar to those observed in P525L and R495X mutations. In comparison, the R521G mutation also caused Figure 3 The P525R mutant FUS is mislocalized in cytoplasmic inclusions in cytoplasmic localization of FUS, but the majority of mouse primary cortical neurons the mutant protein was still retained in the nucleus. In addition, cytoplasmic inclusions of the FUS P525R mutant colocalized with stress granule markers G3BP1 and TIA-1 in N2A cells (figure 2). Similarly, other ALS FUS mutants also showed coloc- alization of cytoplasmic inclusions and stress granule markers. The results are consistent with previous studies that cytoplasmic inclusions of ALS FUS mu- tants are colocalized with stress granule markers.17,20 We further examined the subcellular localization of P525R mutant FUS in mouse primary cortical neurons (figure 3). The P525R mutant FUS protein was largely localized in the cytoplasm of primary neu- rons and formed inclusions positive for stress granule marker G3BP1 in a similar fashion as P525L and R495X mutants. By contrast, the WT FUS protein was predominantly in the nucleus. The results from both N2A and primary neurons consistently indicate that P525R mutant FUS forms cytoplasmic inclu- sions reminiscent of stress granules.

Cytoplasmic inclusions of the P525R mutant FUS protein in human fibroblasts. Skin fibroblast cells derived from this patient were examined along with fibroblast cells from patients with fALS carrying the R521G mutation or healthy controls. Under normal cell cul- ture conditions, the majority of the mutant FUS pro- tein was still localized in the nucleus with visible cytoplasmic distribution. Several cytoplasmic puncta of mutant FUS were observed colocalizing with the stress granule marker G3BP1 (figure 4A). The partial Mouse primary cortical neurons were transfected with EGFP-tagged WT or mutant FUS. Cytoplasmic inclusions of mutant FUS were colocalized with the stress granule marker mislocalization in patient-derived primary fibroblasts is G3BP1 as indicated by arrows. Scale bar, 10 mm. FUS 5 fused in sarcoma; WT 5 wild type. similar to the previously published study.19

4 Neurology: Genetics patient with ALS showed little response to the arse- Figure 4 Mislocalization of the P525R mutant FUS in skin fibroblast cells derived from the patient with ALS nite treatment. The result suggests that fibroblasts carrying the P525R mutation are more vulnerable to oxidative stress compared with either the WT or the R521G mutation.

DISCUSSION Here, we report a patient with ALS carrying a novel heterozygous FUS mutation P525R (figure 1). Based on the family history, her mother also died of ALS, and we speculate that she carried the FUS P525R mutation. The index patient’s grand- mother is healthy in her 70s and has 2 WT FUS alleles; thus, we speculate that her mother likely gained a spontaneous mutation at c.1574C.T. We identified 19 ALS cases carrying the P525L mutation in the FUS gene in the literature and summarized them in table. The literature indicates that approxi- mately 70% of ALS subjects with the documented P525L mutation have no family history of ALS (table), suggesting that either spontaneous mutation event at this locus is not infrequent and/or that indi- viduals with ALS do not survive long enough to reproduce, thus appearing to be sporadic ALS. Pre- vious studies determined that the C-terminal 32 amino acids of FUS functions as an NLS and is crit- ical for its nuclear import mediated by transportin (Trn1).21–23 Indeed, many ALS FUS mutations are located in the NLS. Mutations on the very end of the C-terminus appear to cause the most rapidly pro- gressive phenotype with basophilic cytoplasmic inclu- sions in motor neurons at autopsy. At the position of amino acid 525 of FUS, P525L mutation has been linked to familial ALS and apparently sporadic cases24 (table). Patients with a P525L mutation were found to have an early disease onset and rapidly progressive disease with short survival when compared with pa- tients with other FUS mutations.21,24 In this study, (A) Relatively a small amount of mutant FUS was mislocalized in the cytoplasm and formed puncta colocalized with G3BP1 (as indicated by arrows) under normal cell culture conditions. the patient with the P525R mutation also had an (B) Skin fibroblast cells were treated with 0.5 mM sodium arsenite for 1 hour before fixation. early onset in her mid-20s and a rapid progression Under arsenite (oxidative stress) treatment, a notable amount of FUS P525R mutant mislo- over a 12-month course to death. The clinical features m calized in the cytoplasm and formed inclusions colocalized with G3BP1. Scale bar, 10 m. are similar to those carrying the truncated R495X ALS 5 amyotrophic lateral sclerosis; FUS 5 fused in sarcoma. mutation lacking the entire NLS (mean age at onset: 35 6 16 years; average survival: 16.4 6 10 months To determine the response of WT and different from disease onset).25 The novel P525R mutation in mutant FUS proteins to stress, we treated fibroblasts this case strengthens the notion that mutations at with sodium arsenite that induces oxidative stress. All position 525 of FUS are more fulminant as compared fibroblasts responded to arsenite treatment and to other FUS mutations. There is a female pre- formed numerous stress granules in the cytoplasm ponderance in the reported patients with P525L FUS as marked by the stress granule marker G3BP1 (figure occurring in 75% of recorded cases (table). We cur- 4B). However, the P525R mutant FUS protein re- rently have no explanation for this phenomenon. sponded differently from the WT or R521G mutant. The importance of the C-terminal proline residue A substantial portion of P525R mutant FUS was was reported for other NLS in which Pro525 and mislocalized into the cytoplasm and colocalized with Tyr526 are strictly conserved as the signature G3BP1 in stress granules (figure 4B). By contrast, PY-NLS.26 Figure 5, which is based on our previously WT FUS in healthy control cells and the R521G published structural analysis, shows that P525 plays mutant FUS in fibroblast cells derived from another a critical role in the binding of FUS-NLS to Trn1

Neurology: Genetics 5 6

Table ALS patients with a mutation at position P525 of FUS

Clinical involvement

Age at Survival, Family Lower motor Upper motor Case Sex onset, y moa FUS genotype Protein history Ethnicity Onset Bulbar Limb neuron neuron Atypical featuresb Autopsy Citation

1 NR 22 6 c.1574C.T p.P525L NR NR NR NR NR NR NR NR No Kwiatkowski et al.1

2 F 21 12 c.1574C.T p.P525L Yesc White Bulbar Yes Yes Yes Yes No No Chio et al.29

3 F 22 10 c.1574C.T p.P525L No White LEd No Yes Yes No No Yes Bäumer et al.30 ; erlg:Genetics Neurology: Mackenzie et al.24

4 F 18 11 c.1574C.T p.P525L No Afro- UE Yes Yes Yes Yes No Yes Bäumer et al.30 ; white Mackenzie et al.24

5 F 13 17 c.1574C.T p.P525L No NR LE No Yes Yes No Motor developmental delay; Yes Huang et al.31 mild learning disability

6 M 13 24 NR p.P525L Yes Japanese LE NR Yes NR NR No Yes Ito et al.32

7 NR 11 NR NR p.P525L Yes White NR NR NR NR NR No No Fecto et al.33

8 NR 44 (?) NR c.1574C.T p.P525L NR NR NR NR NR NR NR NR No Kwiatkowski et al1

9 NR 15 NR c.1574C.T p.P525L NR NR NR NR NR NR NR NR No Kwiatkowski et al1

10 F 13 15 c.1574C.T p.P525L Yese Japanese Limb Yes Yes NR NR Developmental delay Yes Mochizuki et al.34

11 M 26 13 c.1574C.T p.P525L No White Limb Yes Yes Yes No No No Sporoviero et al.35

12 F 45 42 c.1574C.T; c.521_5231 p.P525L No White Limb Yes Yes Yes Yes Previous diagnosis: multiple sclerosis No Sporoviero et al.35 3delGAGGTG

13 F 11 14 c.1574C.T p.P525L No NR Limb No Yes Yes Yes No No Conte et al.36

14 F 19 NR c.1574C.T p.P525L No Chinese Limb Yes Yes Yes Yes No No Zou et al.37

15 F 18 NR c.1574C.T p.P525L No NR Bulbar Yes NR NR NR No No Hübers et al.38

16 M 20 NR c.1574C.T p.P525L No NR Bulbar Yes NR NR NR No No Hübers et al.38

17 M 24 7 c.1574C.T p.P525L No NR Bulbar Yes NR NR NR No No Hübers et al.38

18 F 23 8 c.1574C.Tf p.P525L Yesg NR NR No Yes Yes Yes No Yes King et al.39

19 F 21 6 c.1574C.T p.P525L No NR Bulbar Yes Yes Yes Yes Ptosis, opthalmoplegia No Leblond et al.40

20 F 26 12 c.1574C.G p.P525R Yesg White UE Yes Yes Yes Yes No No This report

Abbreviations: ALS 5 amyotrophic lateral sclerosis; FUS 5 fused in sarcoma. a Death or 24 h/d assisted ventilation. b Prior to assisted ventilation. c ALS: mother, maternal grandfather, maternal uncle; not genotyped. d UE, upper extremities; LE, lower extremities; NR, not recorded. e ALS: sister, mother; not genotyped. f Also: TARDP p.Y374X. g ALS: mother; not genotyped. Figure 5 The binding details of the C-terminus of FUS and Trn1

P525 and Y526 of FUS (cyan) and L419, I457, W460 and D384 of Trn1 are shown in sticks. Trn1 is shown in static electric surface mode, in which the red, blue, and green represent negative charge, positive charge, and hydrophobicity, respec- tively. The hydrogen bond between Y526 and D384 is illustrated. (A) WT FUS binding is based on the structure of the FUS-NLS/Trn1 complex ( entry 4FQ3). (B) An illustration of the P525R mutant structure. If R525 should remain in the same position as P525, the side chain of R525 would insert into the surface of Trn1. The steric constrains will force R525 to adapt a different conformation. Consequently, the Y526 residue (shown in thin sticks) in the P525R mutant should not keep the same conformation as in the WT structure. FUS 5 fused in sarcoma; NLS 5 nuclear localization sequence; WT 5 wild type. through the hydrophobic interaction with L419, mutation and the endogenous expression levels of the I457, and W460 of Trn1.21 In addition, P525 puts mutant FUS protein contribute to the observation a C-terminal bend in the peptide and positions Y526 that the majority of the FUS protein remained in to form a hydrogen bond with D384 of Trn1. In the the nucleus. It is also likely that cell type difference modeling structure of P525R, the strong charge of between fibroblast cells and neurons contributes to Arg does not match the hydrophobicity of the cave the less prominent mislocalization of mutant FUS in formed by L419, I457, and W460 of Trn1. More- fibroblast cells. Nevertheless, when exposed to oxi- over, the side chain of Arg is predicted to be too dative stress by sodium arsenite treatment, FUS large to fit in the small cavity (figure 5). Thus, not P525R mutant fibroblasts showed stronger response only the hydrophobic interaction formed by P525 with substantial mislocalization and inclusion for- would be destroyed but also the relocation of resi- mation in cytoplasm as compared to WT and due525wouldpushY526awayandbreakthe R521G mutants (figure 4B). This supports the hydrogen bond between Y526 and D384. The hypothesis that the P525R mutant is more suscep- modeling analysis is consistent with previous re- tible to stress conditions such as arsenite-induced ports that the P525L mutation caused dramatic oxidative stress. decrease in the binding affinity of FUS-NLS to We identified a novel mutation in the FUS gene Trn1,21,22 leading to substantial cytoplasmic that is associated with an early onset and rapid rate of accumulation. disease progression. Mechanistically, the P525R ALS-linked FUS mutants form cytoplasmic inclu- mutation causes a cytoplasmic localization of the sions that are colocalized with stress granule markers FUS protein that is prominently colocalized with in various cell models.25,27 Using induced pluripotent stress granules, suggesting that a substantial cytoplas- stem cell–derived motor neurons, Lenzi et al.28 also mic mislocalization and a strong association with showed that ALS mutant FUS was recruited into stress stress granules could be an indicator of the degree granules. In this study, we showed that cytoplasmic of severity in FUS fALS. inclusions of FUS P525R were dominant in N2A cells (figure 2) and primary cortical neurons (figure 3). AUTHOR CONTRIBUTIONS These inclusions were colocalized with stress granule Lisha Kuang performed experiments, analyzed data, and wrote the man- uscript. Marisa Kamelgarn performed the experiments. Alexandra Arenas markers TIA1 and G3BP1, suggesting that this novel performed the experiments. Jozsef Gal performed the experiments. mutant P525R was also recruited into stress granules. Deborah Taylor performed the pedigree analysis. Weiming Gong per- However, the mislocalization and inclusion formation formed the structural analysis. Martin Brown reported the phenotype. were less prominent in fibroblast cells derived from Daret St. Clair conceptualized the study and wrote the manuscript. Edward J. Kasarskis conceptualized the study, designed experiments, ana- patients with ALS under normal experimental con- lyzed data, and wrote the manuscript. Haining Zhu conceptualized the ditions (figure 4A). It is likely that the heterozygous study, designed experiments, analyzed data, and wrote the manuscript.

Neurology: Genetics 7 ACKNOWLEDGMENT 10. Yang L, Embree LJ, Tsai S, Hickstein DD. Oncoprotein The authors thank Meghann Bruno, RN, for assistance. TLS interacts with serine-arginine proteins involved in RNA splicing. J Biol Chem 1998;273:27761–27764. STUDY FUNDING 11. Zinszner H, Sok J, Immanuel D, Yin Y, Ron D. TLS This study was in part supported by the National Institutes of Neurological (FUS) binds RNA in vivo and engages in nucleo- Disorder and Stroke grant R01NS077284, MDA grant MDA352743, ALS cytoplasmic shuttling. J Cell Sci 1997;110:1741–1750. Association grant 6SE340 and VA MERIT award I01 BX002149 (to 12. Sephton CF, Tang AA, Kulkarni A, et al. Activity-dependent H.Z.), and Crispen and Heidrich endowments (to E.J.K.). The support FUS dysregulation disrupts synaptic homeostasis. Proc Natl from the Multidisciplinary Value Program (MVP) initiative in the University Acad Sci USA 2014;111:E4769–E4778. of Kentucky College of Medicine is appreciated. M.K. and A.A. are supported 13. Morlando M, Dini Modigliani S, Torrelli G, et al. FUS by the National Institute of Environmental Health Sciences training grant stimulates microRNA biogenesis by facilitating co- T32ES007266. transcriptional Drosha recruitment. EMBO J 2012;31: 4502–4510. DISCLOSURE 14. Dini Modigliani S, Morlando M, Errichelli L, Sabatelli M, L. Kuang reports no disclosures. M. Kamelgarn has received research ’ support from the Muscular Dystrophy Association and the Amyotrophic Bozzoni I. An ALS-associated mutation in the FUS 3 - Lateral Sclerosis Association. A. Arenas reports no disclosures. J. Gal has UTR disrupts a microRNA-FUS regulatory circuitry. received research support from NIH/NINDS, the University of Nat Commun 2014;5:4335. Kentucky, and the American Cancer Society. D. Taylor and W. Gong 15. Mastrocola AS, Kim SH, Trinh AT, Rodenkirch LA, report no disclosures. M. Brown has received travel funding from NEALS Tibbetts RS. The RNA-binding protein fused in sarcoma (Northeast ALS Consortium). D. St. Clair reports no disclosures. (FUS) functions downstream of poly(ADP-ribose) poly- E.J. Kasarskis has served on the scientific board of DSMB for Cytokinetics; merase (PARP) in response to DNA damage. J Biol Chem has served on the editorial boards of Amyotrophic Lateral Sclerosis and 2013;288:24731–24741. Frontotemporal Degeneration; has been a consultant for Neuraltus Pharmaceut- 16. Wang WY, Pan L, Su SC, et al. Interaction of FUS and icals; and has received research support from Neuraltus Pharmaceuticals, HDAC1 regulates DNA damage response and repair in NINDS, NIEHS, and the ALS Association. H. Zhu has served on the – editorial boards of the Journal of Biological Chemistry and Frontiers in Biology neurons. Nat Neurosci 2013;16:1383 1391. and has received research support from the Muscular Dystrophy Association 17. Gal J, Zhang J, Kwinter DM, et al. Nuclear localization and the Amyotrophic Lateral Sclerosis Association. Go to Neurology.org/ng sequence of FUS and induction of stress granules by ALS for full disclosure forms. mutants. Neurobiol Aging 2011;32:2323.e27–2323.e40. 18. Kamelgarn M, Chen J, Kuang L, et al. Proteomic analysis Received February 28, 2017. Accepted in final form May 16, 2017. of FUS interacting proteins provides insights into FUS function and its role in ALS. Biochim Biophys Acta REFERENCES 2016;1862:2004–2014. 1. Kwiatkowski TJ Jr, Bosco DA, Leclerc AL, et al. Muta- 19. Gal J, Kuang L, Barnett KR, et al. ALS mutant SOD1 tions in the FUS/TLS gene on chromosome 16 cause interacts with G3BP1 and affects stress granule dynamics. familial amyotrophic lateral sclerosis. Science 2009;323: Acta Neuropathol 2016;132:563–576. 1205–1208. 20. Yang L, Zhang J, Kamelgarn M, et al. Subcellular local- 2. Vance C, Rogelj B, Hortobagyi T, et al. Mutations in ization and RNAs determine FUS architecture in differ- FUS, an RNA processing protein, cause familial amyotro- ent cellular compartments. Hum Mol Genet 2015;24: phic lateral sclerosis type 6. Science 2009;323:1208–1211. 5174–5183. 3. Andersson MK, Stahlberg A, Arvidsson Y, et al. The multi- 21. Niu C, Zhang J, Gao F, et al. FUS-NLS/Transportin 1 functional FUS, EWS and TAF15 proto-oncoproteins show complex structure provides insights into the nuclear target- cell type-specific expression patterns and involvement in cell ing mechanism of FUS and the implications in ALS. PLoS spreading and stress response. BMC Cell Biol 2008;9:37. One 2012;7:e47056. 4. Lagier-Tourenne C, Polymenidou M, Hutt KR, et al. 22. Zhang ZC, Chook YM. Structural and energetic basis of Divergent roles of ALS-linked proteins FUS/TLS and ALS-causing mutations in the atypical proline-tyrosine TDP-43 intersect in processing long pre-mRNAs. Nat nuclear localization signal of the fused in sarcoma protein Neurosci 2012;15:1488–1497. (FUS). Proc Natl Acad Sci USA 2012;109:12017–12021. 5. Wang X, Arai S, Song X, et al. Induced ncRNAs alloste- 23. Dormann D, Madl T, Valori CF, et al. Arginine methyla- rically modify RNA-binding proteins in cis to inhibit tran- tion next to the PY-NLS modulates Transportin binding scription. Nature 2008;454:126–130. and nuclear import of FUS. EMBO J 2012;31:4258–4275. 6. Tan AY, Riley TR, Coady T, Bussemaker HJ, Manley JL. 24. Mackenzie IR, Ansorge O, Strong M, et al. Pathological TLS/FUS (translocated in liposarcoma/fused in sarcoma) heterogeneity in amyotrophic lateral sclerosis with FUS mu- regulates target gene transcription via single-stranded tations: two distinct patterns correlating with disease severity DNA response elements. Proc Natl Acad Sci USA 2012; and mutation. Acta Neuropathol 2011;122:87–98. 109:6030–6035. 25. Bosco DA, Lemay N, Ko HK, et al. Mutant FUS proteins 7. Dhar SK, Zhang J, Gal J, et al. FUsed in sarcoma is a novel that cause amyotrophic lateral sclerosis incorporate into regulator of manganese superoxide dismutase gene tran- stress granules. Hum Mol Genet 2010;19:4160–4175. scription. Antioxid Redox Signal 2014;20:1550–1566. 26. Lee BJ, Cansizoglu AE, Suel KE, Louis TH, Zhang Z, 8. Yang L, Gal J, Chen J, Zhu H. Self-assembled FUS binds Chook YM. Rules for nuclear localization sequence recog- active chromatin and regulates gene transcription. Proc nition by karyopherin beta 2. Cell 2006;126:543–558. Natl Acad Sci USA 2014;111:17809–17814. 27. Gal J, Strom AL, Kwinter DM, et al. Sequestosome 9. Dichmann DS, Harland RM. fus/TLS orchestrates splic- 1/p62 links familial ALS mutant SOD1 to LC3 via an ing of developmental regulators during gastrulation. Genes ubiquitin-independent mechanism. J Neurochem 2009; Dev 2012;26:1351–1363. 111:1062–1073.

8 Neurology: Genetics 28. Lenzi J, De Santis R, de Turris V, et al. ALS mutant FUS 35. Sproviero W, La Bella V, Mazzei R, et al. FUS mutations proteins are recruited into stress granules in induced plu- in sporadic amyotrophic lateral sclerosis: clinical ripotent stem cell-derived motoneurons. Dis Model Mech and genetic analysis. Neurobiol Aging 2012;33:837. 2015;8:755–766. e1–837.e5. 29. Chio A, Restagno G, Brunetti M, et al. Two Italian kin- 36. Conte A, Lattante S, Zollino M, et al. P525L FUS muta- dreds with familial amyotrophic lateral sclerosis due to tion is consistently associated with a severe form of juvenile FUS mutation. Neurobiol Aging 2009;30:1272–1275. amyotrophic lateral sclerosis. Neuromuscul Disord 2012; 30. Baumer D, Hilton D, Paine SM, et al. Juvenile ALS with 22:73–75. basophilic inclusions is a FUS proteinopathy with FUS 37. Zou ZY, Cui LY, Sun Q, et al. De novo FUS gene muta- mutations. Neurology 2010;75:611–618. tions are associated with juvenile-onset sporadic amyotro- 31. Huang EJ, Zhang J, Geser F, et al. Extensive FUS- phic lateral sclerosis in China. Neurobiol Aging 2013;34: immunoreactive pathology in juvenile amyotrophic lateral 1312.e1–1312.e8. sclerosis with basophilic inclusions. Brain Pathol 2010;20: 38. Hübers A, Just W, Rosenbohm A, et al. De novo FUS 1069–1076. mutations are the most frequent genetic cause in early- 32. Ito H, Fujita K, Nakamura M, et al. Optineurin is co- onset German ALS patients. Neurobiol Aging 2015;36: localized with FUS in basophilic inclusions of ALS with 3117.e1–3117.e6. FUS mutation and in basophilic inclusion body disease. 39. King A, Troakes C, Smith B, et al. ALS-FUS pathology Acta Neuropathol 2011;121:555–557. revisited: singleton FUS mutations and an unusual case 33. Fecto F, Siddique T. Making connections: pathology and with both a FUS and TARDBP mutation. Acta Neuro- genetics link amyotrophic lateral sclerosis with frontotem- pathol Commun 2015;3:62. poral lobe dementia. J Mol Neurosci 2011;45:663–675. 40. Leblond CS, Webber A, Gan-Or Z, et al. De novo FUS 34. Mochizuki Y, Isozaki E, Takao M, et al. Familial ALS with P525L mutation in Juvenile amyotrophic lateral sclerosis FUS P525L mutation: two Japanese sisters with multiple with dysphonia and diplopia. Neurol Genet 2016;2:e63. systems involvement. J Neurol Sci 2012;323:85–92. doi: 10.1212/NXG.0000000000000063.

Neurology: Genetics 9 Brain calcifications and PCDH12 variants

Gaël Nicolas, MD, PhD ABSTRACT Monica Sanchez- Objective: To assess the potential connection between PCDH12 and brain calcifications in Contreras, MD, PhD a patient carrying a homozygous nonsense variant in PCDH12 and in adult patients with brain Eliana Marisa Ramos, calcifications. PhD Methods: We performed a CT scan in 1 child with a homozygous PCDH12 nonsense variant. We Roberta R. Lemos, PhD screened DNA samples from 53 patients with primary familial brain calcification (PFBC) and 26 Joana Ferreira, PhD patients with brain calcification of unknown cause (BCUC). Denis Moura, BSc Maria J. Sobrido, MD, Results: We identified brain calcifications in subcortical and perithalamic regions in the patient PhD with a homozygous PCDH12 nonsense variant. The calcification pattern was different from what Anne-Claire Richard, BSc has been observed in PFBC and more similar to what is described in in utero infections. In patients Alma Rosa Lopez, BSc with PFBC or BCUC, we found no protein-truncating variant and 3 rare (minor allele frequency , Andrea Legati, PhD 0.001) PCDH12 predicted damaging missense heterozygous variants in 3 unrelated patients, Jean-François Deleuze, albeit with no segregation data available. MD, PhD Conclusions: Brain calcifications should be added to the phenotypic spectrum associated with Anne Boland, PharmD, PCDH12 biallelic loss of function, in the context of severe cerebral developmental abnormalities. PhD A putative role for PCDH12 variants remains to be determined in PFBC. Neurol Genet 2017;3: Olivier Quenez, MSc e166; doi: 10.1212/NXG.0000000000000166 Pierre Krystkowiak, MD, PhD GLOSSARY Pascal Favrole, MD BCUC 5 brain calcification of unknown cause; ExAC 5 Exome Aggregation Consortium; PFBC 5 primary familial brain calcification. Daniel H. Geschwind, MD, PhD A homozygous nonsense PCDH12 variant has recently been reported in consanguineous fam- Adi Aran, MD ilies, where the affected children had congenital microcephaly, epilepsy, and profound global Reeval Segel, MD developmental disability.1 Fetal MRI and USG showed dysplastic elongated masses in the Ephrat Levy-Lahad, MD midbrain-hypothalamus-optic tract area and hyperechogenic perithalamic foci. PCDH12 enc- Dennis W. Dickson, MD odes a protocadherin associated with membrane physical stability, adhesion, and vasculature Giovanni Coppola, MD maintenance and has recently been pointed out as a candidate gene for primary familial brain Rosa Rademakers, PhD calcification (PFBC). PFBC is characterized by the presence of calcifications affecting primarily João R.M. de Oliveira, the basal ganglia, in the absence of secondary cause.2 Clinical manifestations include movement MD, PhD disorders, cognitive impairment, psychiatric disturbances, and headache, most frequently begin- ning during adulthood.2,3 Heterozygous variants causing autosomal dominant PFBC in up to – Correspondence to 50% of the families were identified in 4 genes: SLC20A2, PDGFRB, PDGFB, and XPR1.4 8 We Dr. de Oliveira: [email protected] previously searched for genes with a cerebral expression pattern similar to the PFBC major

From the Department of Genetics and CNR-MAJ (G.N., A.-C.R., O.Q.), Normandie Univ, UNIROUEN, Inserm U1245, Rouen University Hospital, Normandy Center for Genomic and Personalized Medicine, France; Department of Human Genetics (G.N.), Genome Research, Supplemental data Radboud UMC, Nijmegen, The Netherlands; Department of Neuroscience (M.S.-C., D.W.D., R.R.), Mayo Clinic, Jacksonville, FL; Department at Neurology.org/ng of Psychiatry (E.M.R., A.R.L., A.L., G.C., D.H.G.), Semel Institute for Neuroscience and Human Behavior, David Geffen School of Medicine, University of California Los Angeles; Keizo Asami Laboratory (R.R.L., J.F., D.M., J.R.M.d.O), Federal University of Pernambuco, Recife, Brazil; Fundación Pública Galega de Medicina Xenómica (M.J.S.), Clinical University Hospital of Santiago de Compostela-SERGAS, Spain; Centre National de Recherche en Génomique Humaine (CNRGH) (J.-F.D., A.B.), Institut de Biologie François Jacob, CEA, Evry; Department of Neurology (P.K.), Amiens University Hospital; Department of Neurology (P.F.), Tenon Hospital, AP-HP, Paris, France; Medical Genetics MRI Unit (R.S., E.L.-L.), Shaare Zedek Medical Center; Hebrew University-Hadassah School of Medicine (R.S., E.L.-L.); and Neuropsychiatry Department (J.R.M.d.O), Universidade Federal de Pernambuco, Recife, Brazil. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 causative gene SLC20A2 using the Allen Brain did not enable to determine whether these foci are eventually Atlas (brain-map.org/),9,10 observing a higher calcifications. Therefore, we performed a brain CT scan in- individual III-1, family B from the original pedigree.1 SLC20A2 expression in regions affected by calcifications in PFBC. PCDH12 was singled PCDH12 screening in patients with brain calcification. We included a total of 79 worldwide adult cases with brain calcifi- 10 out with the highest significant correlation, cations that were referred to 5 centers of expertise, negatively and a follow-up analysis with additional brains screened for the known PFBC causative genes (supplemental data). still shows PCDH12 as the most similar pat- Of these, 53 cases matched the clinical inclusion criteria for PFBC tern to SLC20A2, even when compared with (detailed previously in reference 3). Briefly, these cases exhibited at least bilateral basal ganglia calcifications and no secondary cause. the other known PFBC causative genes (table The remaining 26 patients were included on a neuropathologic e-1 at Neurology.org/ng). basis if they presented moderate-to-severe basal ganglia calcifica- To evaluate the potential link between tions. Note that calcifications also involved other brain regions in almost all cases and that other causes of brain calcifications could PCDH12 and brain calcifications, (1) we per- not be excluded in these patients, thereafter referred as having formed a CT scan in a patient reported to carry BCUC. All patients were screened for pathogenic variants by a homozygous nonsense PCDH12 variant and sequencing all coding exons of PCDH12 (reference transcript: (2) we screened DNA samples from patients NM_016085.3). Bioinformatics predictions were performed using direct access to Polyphen2 HumDiv,12 SIFT,13 and Mutation with PFBC or brain calcifications of unknown Taster14 tools, and the minor allele frequency (MAF) was checked cause (BCUC). at the Exome Aggregation Consortium (ExAC) website accessed in August 2016 (exac.broadinstitute.org/).15 Detailed inclusion crite- ria and sequencing methods are provided in supplemental data. METHODS CT imaging in PCDH12 homozygous variant carriers. In the original report, patients with symmetric intrauterine Standard protocol approvals, registrations, and patient growth retardation, severe microcephaly, visual impairment, dysto- consents. All patients provided written informed consent for nia, epilepsy, and profound developmental disability were shown genetic analyses. to carry a PCDH12 c.995T.A, p.R839X homozygous variant.1 This variant is considered to be pathogenic when carried at the RESULTS CT of a PCDH12 homozygous variant carrier. homozygous state following the American College of Medical CT is the reference imaging to identify brain calcifica- Genetics and Genomics and the Association for Molecular tion, so we used it to determine the nature of the Pathology recommendations.11 Brain imaging revealed mid- brain hypothalamus dysplasia and significant periventricular hyperechogenic foci identified in a patient with and/or periventricular hyperechogenicity. Fetal USG and MRI ahomozygousnonsensep.R839XPCDH12 variant.1 We identified spots of perithalamic calcification located in the posterior arms of the internal capsules Figure Brain CT imaging of a patient carrying the PCDH12 c.995T>A, p.R839X homozygous variant and in juxtacortical right white matter (figure). PCDH12 screening in patients with brain calcification. As we provided evidence that PCDH12 biallelic loss of function is associated with brain calcification and given the high level of coexpression with the PFBC major causative gene SLC20A2, we next screened this gene in a group of patients with PFBC or BCUC. Among the 79 patients with PFBC or BCUC, we did not identify any protein-truncating variant (nonsense, splice site, or frameshift insertion/deletion). How- ever, we detected 4 rare (MAF ,0.001 in ExAC) heterozygous PCDH12 missense variants in 4 unre- lated patients: c.163C.G, p.(R55G); c.440G.T, p.(S147I); c.995T.A, p.(I332N); and c.3271G.A, p.(G1091S) (table 1). Three were predicted damag- ing by at least 1 in silico tool, while variant p.R55G was predicted benign by all 3 tools. The c.440G.T, p.(S147I) variant had an MAF of 2.5e-05 in the ExAC database and was exclusively found in 3 individuals with the same ancestry as the patient (classified in ExAC as European non- Finnish). Two of the 3 in silico tools (Mutation

(A, B) Coronal sections. (C, D) Transversal sections. Spot calcifications affecting perithalamic Taster and Polyphen2 HumDiv, but not SIFT) pre- regions (white arrows, A–C) and subcortical regions (red arrows, B, D). dicted a damaging effect for this change to the protein

2 Neurology: Genetics Table 1 Rare PCDH12 variants identified in a series of 79 patients with PFBC or BCUC

Protein ExAC SIFT Polyphen2 HumDiv Mutation Taster Location (Ghrc37) cDNA changea changea frequencyb prediction prediction prediction PhyloP

chr5:141337254 c.163C.G p.(R55G) 6.1e-04 Tolerated Benign Polymorphism 21.01

chr5:141336977 c.440G.T p.(S147I) 2.5e-05 Tolerated Possibly damagingc Disease causingc 2.14c

chr5:141336422 c.995T.A p.(I332N) 1e-04 Deleteriousc Probably damagingc Disease causingc 4.48c

chr5:141325230 c.3271G.A p.(G1091S) 3.3e-05 Deleteriousc Probably damagingc Disease causingc 4.81c

Abbreviations: BCUC 5 brain calcification of unknown cause; cDNA 5 complementary DNA; ExAC 5 Exome Aggregation Consortium; PFBC 5 primary familial brain calcification. a Accession number: NM_016085.3. b ExAC minor allele frequency assessed in August 2016.15 c Values are above each threshold.

function. DNA from relatives was not available for reference imaging tool for detecting and assessing cal- segregation analysis. This variant is located in the cifications, MRI is the primary imaging tool for the second cadherin tandem repeat domain (EC2) detection of all other brain abnormalities in the (NCBI accession cd11304) and, therefore, could absence of radiation. T2* or susceptibility-weighted affect homophilic adhesive behavior and calcium- images increase the diagnostic performance of MRI dependent cell adhesion.16 for calcification compared with the other sequences. The c.995T.A, p.(I332N) and c.3271G.A, However, they can sometimes miss small calcifica- p.(G1091S) variants are both predicted damaging by tions, and they are still complementary with CT to all 3 in silico tools. The p.I332N variant was reported describe precise shape and intensity and to definitely with an overall MAF of 0.0001 in ExAC, found in 12 conclude on the differential identification with iron individuals of East Asian ancestry (the patient was born deposits.19,20 In our patient, neither T2* nor in Southeastern Asia) and 1 individual of European susceptibility-weighted images were available. non-Finnish ancestry. The p.G1091S variant has an In the original report, the efficiency of nonsense- overall MAF of 3.3e-05, found in 1 individual of mediated decay has been measured as 84%, suggesting European non-Finnish ancestry (same as the patient) a strong loss of function. The patients carrying the non- and 3 individuals of South Asian ancestry. DNA from sense PCDH12 variant in a homozygous state may still relatives was not available for segregation analysis of express little amount of the truncated protein, but no any variant. Variant p.I332N is also located in a cad- full-length PCDH12. This supports the hypothesis that herin tandem repeat domain, namely EC3. However, loss of function of PCDH12 is the mechanism leading p.G1091S variant is located in a highly conserved site to the patient’s phenotype, including brain calcification. in the cytoplasmic domain, which has a unique In a candidate gene approach, we searched for rare sequence among the cadherin family. Unlike the other PCDH12 variants in PFBC and BCUC patients and cadherins, the cytoplasmic domain of PCDH12 does found no protein-truncating variants. Three hetero- not interact with catenins, and it is involved in cellular zygous missense variants, predicted damaging by at processes other than cell junction, such as regulation least one of the tools, were identified in 2 patients of gene expression and signaling pathways.17 Clinical with PFBC and 1 patient with BCUC. Given the fact details of all 3 predicted damaging variant carriers are that biallelic loss of PCDH12 function leads to provided in the supplemental data. a severe neurodevelopmental phenotype, it is unlikely that these variants have a dominant-negative effect. DISCUSSION We show here that a homozygous However, as they are missense variants, their putative nonsense PCDH12 variant, detected in patients with effect on protein function is hard to predict, and it severe developmental delay and microcephaly,1 is remains possible that they are responsible for loss of associated with brain calcifications. This feature function, gain of function, or have a neutral effect on should therefore be added to the phenotypic spec- protein function. The frequencies of theses variants in trum of this rare disorder. The pattern of calcifica- the patients’ respective populations as estimated in tions is, however, different from the typical findings ExAC are not inconsistent with a causative effect, as in PFBC, where calcifications always affect at least they are in the same frequency ranges as other disease- both pallidum,3 and resembled to those observed in causing variants in SLC20A2.8 Because neither segrega- various neuroinfectious prenatal conditions, such as tion nor functional data are available, it is not possible TORCH infections.18 Brain calcification is a highly to conclude about their pathogenicity at this stage. informative feature on brain imaging of children with Besides PFBC, brain calcifications can be detected neurodevelopmental disorders.18 Although CT is the in other numerous distinct conditions, such as

Neurology: Genetics 3 systemic phosphocalcic metabolism disorders of in- Coppola, Rosa Rademakers, and João R.M. de Oliveira. Manuscript herited or acquired cause, in utero or postnatal infec- draft: Gaël Nicolas, Monica Sanchez-Contreras, Eliana Marisa Ramos, Roberta R. Lemos, and João R.M. de Oliveira. Critical revision: Gaël tions, interferonopathies, inborn errors of metabolism, Nicolas, Monica Sanchez-Contreras, Eliana Marisa Ramos, Roberta and other rare inherited diseases.21 Calcifications are R. Lemos, João R.M. de Oliveira, and Giovanni Coppola. Study design believed to be related to increased type-I interferon and supervision: João R.M. de Oliveira, Gaël Nicolas, Giovanni Coppola, and Rosa Rademakers. response in both in utero viral infections and interfero- nopathies.22 Several of these clinical presentations, STUDY FUNDING including TORCH in utero infections and typical The French study (G.N.) was funded by Conseil Régional de Haute Aicardi-Goutières syndrome, are similar to the ones Normandie—APERC 2014 no. 2014-19 in the context of Appel d’Offres observed in the PCDH12 homozygous carriers. In other Jeunes Chercheurs (CHU de Rouen). Mayo Clinic group (M.S.-C., conditions, mutations in OCLN and JAM3 genes, en- D.W.D., and R.R.) was supported by the Morris K. Udall Parkinson’s coding endothelial cell adhesion proteins, result in micro- Disease Research Center of Excellence (NINDS P50NS072187). J.R.M.d.O. acknowledges funding from CPNq (480255/2013-0, 440770/ 18,23,24 angiopathy associated with calcifications. Given the 2016-5, 470781/2014-9, and 307909/2012-3). Funding sources had no knownfunctionofPCDH12,we postulate that similar specific roles. mechanisms could be associated with the calcifications observed in the PCDH12 homozygous loss-of-function DISCLOSURE carriers. G. Nicolas, M. Sanchez-Contreras, and E.M. Ramos report no disclosures. R.R. Lemos has received research support from FACEPE—Fundação de PCDH12 is a protocadherin associated with AmparoaCiênciaeTecnologiadoEstadodePernambuco(Brazil). 25 membrane physical stability and adhesion. A J. Ferreira has received research support from FACEPE—Fundação de Pcdh12 knockout mouse model revealed several AmparoaCiênciaeTecnologiadoEstadodePernambuco(Brazil). — age-independent vessel impairments, such as rami- D. Moura has received research support from FACEPE Fundação de AmparoaCiênciaeTecnologiadoEstadodePernambuco(Brazil). fications of medial elastic lamellae and increased M.J. Sobrido has received speaker honoraria from Actelion Pharmaceuticals; inner diameter and circumferential mid-wall is the CEO of Genomic Consulting; has a private clinical neurology practice stress.26 PCDH12 has been widely studied as and genetic diagnosis of patients with inherited diseases; and has received a key-player cadherin involved in placental mainte- research support from Actelion Pharmaceuticals, Instituto de Salud Carlos III (Spain), and Asociación Galega de Ataxias (AGA). A.-C. Richard, nance and also a preeclampsia biomarker; however, A.R. Lopez, and A. Legati report no disclosures. J.-F. Deleuze has served little is known about its involvement in brain phys- on the editorial board of Human Genetics; is listed as an inventor on a patent iology. It is conceivable that mutations in PCDH12 for Quality assessment of induced pluripotent cells by referring to splicing signatures of pluripotency; and has received research support from Labex and SLC20A2, which share similar expression pat- GenMed from French ANR. A. Boland and O. Quenez report no disclo- terns in the brain, might lead to similar phenotypes. sures. P. Krystkowiak has served on the scientific advisory boards of Lund- Of interest, Slc20a2 knockout mice developed not beck, Allergan, IPSEN, Novartis Pharma, and Servier Euthérapie; and has only brain calcifications but also fetal growth received speaker honoraria from Lundbeck, Allergan, IPSEN, Novartis Pharma, and Servier Euthérapie. P. Favrole reports no disclosures. restriction, lower birth viability, and placental cal- D.H. Geschwind has served on the scientific advisory board of Ovid Ther- cification associated with thickened basement apeutics Inc.; serves on the editorial boards of Cell, Molecular Autism, membranes.27 In both mouse models, the placental Molecular Neuropsychiatry, Human Molecular Genetics, Neuron, Current phenotype and the vascular impairment are addi- Opinion in Genetics and Development,andTranslational Psychiatry; has acted as a reviewer for Neurology®; is listed as an inventor on the following tional putative links between SLC20A2 and patents: Peripheral Gene Expression Biomarkers for Autism (3), Genetic PCDH12, which deserve additional studies on Risk Factor for Neurodegenerative Disease, Compositions and Methods for mouse models. Diagnosing and Treating Brain Cancer and Identifying Neural Stem Cells, Genetic Variants Underlying Human Cognition: Novel Diagnostic and PCDH12 biallelic loss of function causes a severe Therapeutic Targets, Brain Gene Expression Changes Associated neurodevelopmental phenotype associated with brain with Autism Spectrum Disorders, Full Biomarkers in Friedreich’sAtaxia calcifications. Rare predicted damaging heterozygous (provisional patent application), Signaling Networks Causing Neurodeve- — PCDH12 variants were identified in patients with lopmental Disorders In Human Neurons, Genes Dysregulated in Autism Potential Biomarkers and Therapeutic Pathways, mTor—A Genetic Target PFBC or BCUC here, but whether they are associated for Treatment of Individuals with Neurocognitive Spectrum Disorders, with brain calcification or not remains to be deter- Neuronal Regeneration, Frataxin Knock-Down Mouse, Jakmip1 Knockout mined. To address this question, follow-up studies Mouse, and Cyfip1 Transgenic Mouse; receives publishing royalties from will be necessary including screening other series, as- Oxford University Press; has been a consultant for OVID Therapeutics Ltd.; has received research support from Takeda Pharmaceutical Company, sessing the segregation of rare variants and functional NIH/NIMH, NIH/NICHD, NIH/NINDS, NIH/NIA, the Simons Foun- consequences. dation, Adelson Medical Research Foundation, and the Tau Consortium; holds stock/stock options in Ovid Therapeutics; and receives license fee payments for a Mouse model of Friedreich Ataxia (licensed by UCLA). AUTHOR CONTRIBUTIONS A. Aran reports no disclosures. R. Segel has received research support from Collection and interpretation of data: Gaël Nicolas, Monica Sanchez- the Joint Research Fund (Keren Meshutefet) of the Hebrew University and Contreras, Eliana Marisa Ramos, Roberta R. Lemos, Joana Ferreira, the Hadassah and Shaare Zedek Hospitals. E. Levy-Lahad has received Denis Moura, Maria J. Sobrido, Anne-Claire Richard, Alma Rosa Lopez, research support from USAID MERC program, Israel Science Foundation Andrea Legati, Jean-François Deleuze, Anne Boland, Olivier Quenez, (Israel), Breast Cancer Research Foundation (BCRF; New York, NY), and Pierre Krystkowiak, Pascal Favrole, Daniel H. Geschwind, Adi Aran, Israel Cancer Research Fund (ICRF; New York, NY). D.W. Dickson has Reeval Segel, Ephrat Levy-Lahad, Dennis W. Dickson, Giovanni received travel funding and speaker honoraria from Novartis; has served on

4 Neurology: Genetics the editorial boards of Acta Neuropathologica, Brain, Brain Pathology, Neu- Genetics and Genomics and the Association for Molecular ropathology and Applied Neurobiology, Annals of Neurology, Neuropathology, Pathology. Genet Med 2015;17:405–424. International Journal of Clinical and Experimental Pathology,andAmerican 12. Adzhubei I, Jordan DM, Sunyaev SR. Predicting func- Journal of Neurodegenerative Disease; has received research support from the tional effect of human missense mutations using Poly- following grants: P50AG016574 (Core Leader), P50NS072187 (Center Phen-2. Curr Protoc Hum Genet 2013;Chapter 7:Unit Director), P01NS084974 (Project Leader), and P01AG003949 (Core 7.20. Leader); and has received research support from the Mangurian Founda- tion. G. Coppola receives publishing royalties from Oxford University Press 13. Ng PC, Henikoff S. SIFT: predicting amino acid changes and receives research support from Takeda Pharmaceutical Company, NIH, that affect protein function. Nucleic Acids Res 2003;31: the Consortium for Frontotemporal Dementia Research, Adelson Medical 3812–3814. Research Foundation, the Tau Consortium, CHDI, and John Douglas 14. Schwarz JM, Cooper DN, Schuelke M, Seelow D. Muta- French Alzheimer’s Foundation. R. Rademakers holds patents for Detecting tionTaster2: mutation prediction for the deep-sequencing and Treating Dementia, and for Methods and Materials for Detecting and age. Nat Methods 2014;11:361–362. Treating Dementia and has received research support from NIH and the 15. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of Consortium for Frontotemporal Dementia. J.R.M. de Oliveira has received protein-coding genetic variation in 60,706 humans. research support from the Federal Council for Research and Technology Nature 2016;536:285–291. (Brazil) and FACEPE—Fundação de Amparo a Ciência e Tecnologia do Estado de Pernambuco (Brazil). Go to Neurology.org/ng for full disclosure 16. Bouillot S, Tillet E, Carmona G, et al. Protocadherin-12 forms. cleavage is a regulated process mediated by ADAM10 pro- tein: evidence of shedding up-regulation in pre-eclampsia. – Received April 3, 2017. Accepted in final form May 31, 2017. J Biol Chem 2011;286:15195 15204. 17. Wu Q, Maniatis T. Large exons encoding multiple ecto- REFERENCES domains are a characteristic feature of protocadherin genes. – 1. Aran A, Rosenfeld N, Jaron R, et al. Loss of function of Proc Natl Acad Sci USA 2000;97:3124 3129. PCDH12 underlies recessive microcephaly mimicking 18. Livingston JH, Stivaros S, van der Knaap MS, Crow YJ. intrauterine infection. Neurology 2016;86:2016–2024. Recognizable phenotypes associated with intracranial cal- – 2. Sobrido MJ, Coppola G, Oliveira J, Hopfer S, Geschwind cification. Dev Med Child Neurol 2013;55:46 57. DH. Primary familial brain calcification. In: Pagon RA, 19. Kozic D, Todorovic-Djilas L, Semnic R, Miucin-Vukadi- Adam MP, Ardinger HH, et al, editors. GeneReviews. novic I, Lucic M. MR imaging: an unreliable and poten- Seattle: University of Washington; 1993. tially misleading diagnostic modality in patients with 3. Nicolas G, Charbonnier C, de Lemos RR, et al. Brain intracerebral calcium depositions. Case report. Neuro En- calcification process and phenotypes according to age docrinol Lett 2009;30:553–557. and sex: lessons from SLC20A2, PDGFB, and PDGFRB 20. Liu S, Buch S, Chen Y, et al. Susceptibility-weighted mutation carriers. Am J Med Genet B Neuropsychiatr imaging: current status and future directions. NMR Bi- Genet 2015;168:586–594. omed 2017:30. 4. Legati A, Giovannini D, Nicolas G, et al. Mutations in 21. Deng H, Zheng W, Jankovic J. Genetics and molecular XPR1 cause primary familial brain calcification associ- biology of brain calcification. Ageing Res Rev 2015;22: ated with altered phosphate export. Nat Genet 2015; 20–38. 47:579–581. 22. Crow YJ, Manel N. Aicardi-Goutieres syndrome and the type 5. Nicolas G, Pottier C, Maltete D, et al. Mutation of the I interferonopathies. Nat Rev Immunol 2015;15:429–440. PDGFRB gene as a cause of idiopathic basal ganglia cal- 23. Rajab A, Aldinger KA, El-Shirbini HA, Dobyns WB, Ross cification. Neurology 2013;80:181–187. ME. Recessive developmental delay, small stature, micro- 6. Keller A, Westenberger A, Sobrido MJ, et al. Mutations in cephaly and brain calcifications with locus on chromosome the gene encoding PDGF-B cause brain calcifications in 2. Am J Med Genet A 2009;149A:129–137. humans and mice. Nat Genet 2013;45:1077–1082. 24. Mochida GH, Ganesh VS, Felie JM, et al. A homozygous 7. Wang C, Li Y, Shi L, et al. Mutations in SLC20A2 link mutation in the tight-junction protein JAM3 causes hem- familial idiopathic basal ganglia calcification with phos- orrhagic destruction of the brain, subependymal calcifica- phate homeostasis. Nat Genet 2012;44:254–256. tion, and congenital cataracts. Am J Hum Genet 2010;87: 8. Lemos RR, Ramos EM, Legati A, et al. Update and muta- 882–889. tional analysis of SLC20A2: a major cause of primary familial 25. Rampon C, Bouillot S, Climescu-Haulica A, et al. Proto- brain calcification. Hum Mutat 2015;36:489–495. cadherin 12 deficiency alters morphogenesis and transcrip- 9. Hawrylycz MJ, Lein ES, Guillozet-Bongaarts AL, et al. An tional profile of the placenta. Physiol Genomics 2008;34: anatomically comprehensive atlas of the adult human brain 193–204. transcriptome. Nature 2012;489:391–399. 26. Philibert C, Bouillot S, Huber P, Faury G. Protocadherin- 10. da Silva RJ, Pereira IC, Oliveira JR. Analysis of gene 12 deficiency leads to modifications in the structure and expression pattern and neuroanatomical correlates for function of arteries in mice. Pathol Biol (Paris) 2012;60: SLC20A2 (PiT-2) shows a molecular network with poten- 34–40. tial impact in idiopathic basal ganglia calcification (“Fahr’s 27. Wallingford MC, Gammill HS, Giachelli CM. Slc20a2 disease”). J Mol Neurosci 2013;50:280–283. deficiency results in fetal growth restriction and placental 11. Richards S, Aziz N, Bale S, et al. Standards and guidelines calcification associated with thickened basement mem- for the interpretation of sequence variants: a joint consen- branes and novel CD13 and lamininalpha1 expressing sus recommendation of the American College of Medical cells. Reprod Biol 2016;16:13–26.

Neurology: Genetics 5 UNC5C variants are associated with cerebral amyloid angiopathy

Hyun-Sik Yang, MD ABSTRACT Charles C. White, PhD Objective: To determine whether common genetic variants in UNC5C, a recently identified late- Lori B. Chibnik, PhD onset Alzheimer disease (LOAD) dementia susceptibility gene, are associated with AD suscepti- Hans-Ulrich Klein, PhD bility or AD-related clinical/pathologic phenotypes. Julie A. Schneider, MD Methods: We used data from deceased individuals of European descent who participated in the David A. Bennett, MD Religious Orders Study or the Rush Memory and Aging Project (n 5 1,288). We examined whether Philip L. De Jager, MD, there were associations between single nucleotide polymorphisms (SNPs) within 6100 kb of the PhD UNC5C gene and a diagnosis of AD dementia, global cognitive decline, a pathologic diagnosis of AD, b-amyloid load, neuritic plaque count, diffuse plaque count, paired helical filament tau den- sity, neurofibrillary tangle count, and cerebral amyloid angiopathy (CAA) score. We also evaluated Correspondence to Dr. De Jager: the relation of the CAA-associated variant and dorsolateral prefrontal cortex (DLPFC) UNC5C [email protected] RNA expression. Secondary analyses were performed to examine the interaction of the CAA- associated SNP and known genetic risk factors of CAA as well as the association of the SNP with other cerebrovascular pathologies. Results: AsetofUNC5C SNPs tagged by rs28660566T was associated with a higher CAA score (p 5 2.3 3 1026): each additional rs28660566T allele was associated with a 0.60 point higher CAA score, which is equivalent to approximately 75% of the higher CAA score associated with each allele of APOE e4. rs28660566T was weakly associated with lower UNC5C expression in the human DLPFC (p 5 0.036). Moreover, rs28660566T had a synergistic interaction with APOE e4 on their association with higher CAA severity (p 5 0.027) and was associated with more severe arteriolosclerosis (p 5 0.0065). Conclusions: Targeted analysis of the UNC5C region uncovered a set of SNPs associated with CAA. Neurol Genet 2017;3:e176; doi: 10.1212/NXG.0000000000000176

GLOSSARY AD 5 Alzheimer disease; CAA 5 cerebral amyloid angiopathy; CI 5 confidence interval; DLPFC 5 dorsolateral prefrontal cortex; LD 5 linkage disequilibrium; LOAD 5 late-onset Alzheimer disease; MAF 5 minor allele frequency; MAP 5 Rush Memory and Aging Project; OR 5 odds ratio; PHFtau 5 paired helical filament tau; QC 5 quality control; RNA-Seq 5 RNA- sequencing; ROS 5 Religious Orders Study; SNP 5 single nucleotide polymorphism.

A recent study reported the association of UNC5C T835M (rs137875858A), a rare coding variant in this netrin receptor implicated in axon guidance during development,1 with late- onset Alzheimer disease (LOAD).2 As we have shown in our study of the TREM locus,3 multiple genetic variants in the same locus can have independent effects on disease susceptibility and could affect specific features of the disease. We, therefore, assessed whether common UNC5C variants influence AD-related cognitive and pathologic phenotypes in 2 large clinical pathologic cohort studies of aging and dementia. We found a set of UNC5C single nucleotide Supplemental data at Neurology.org/ng From the Departments of Neurology and Psychiatry (H.-S.Y., C.C.W., H.-U.K., P.L.D.J.), Program in Translational NeuroPsychiatric Genomics, Institute for the Neurosciences; Department of Neurology (H.-S.Y.), Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital; Harvard Medical School (H.-S.Y., H.-U.K.); Harvard T.H. Chan School of Public Health (L.B.C.), Boston; Program in Medical and Population Genetics (H.-S.Y., C.C.W., L.B.C., H.-U.K., P.L.D.J.), Broad Institute, Cambridge, MA; Rush Alzheimer’s Disease Center (J.A.S., D.A.B.) and Department of Neurological Sciences (J.A.S., D.A.B.), Rush University Medical Center, Chicago, IL; and Department of Neurology (P.L.D.J.), Center for Translational & Systems Neuroimmunology, Columbia University Medical Center, New York, NY. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/ng for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 polymorphisms (SNPs) in a single linkage dis- were performed as previously described.3 Genotype QC was done 9 equilibrium (LD) block linked to the severity with PLINK software version 1.08, and the metrics included a geno- type success rate of .95%, Hardy-Weinberg p . 0.001, and of cerebral amyloid angiopathy (CAA), a dis- amisshaptestof,1 3 1029. Closely related individuals are removed ease of small cerebral arterioles caused by with Pihat method using PLINK. EIGENSTRAT (default setting)10 b-amyloid accumulation.4 was used to generate a genotype covariance matrix, and population outliers were removed. Then, genotype imputation was done with BEAGLE software (version 3.3.2)11 on a reference map from 1,000 METHODS Subjects, standard protocol approvals, Genomes Project Consortium interim phase I haplotypes (2010– registrations, and patient consents. Our subjects were from 2 2011 data freeze). Imputed genotypes with a minor allele frequency large community-based cohort studies of older adults, the Reli- (MAF) of .0.01 and an INFO score of .0.3wereusedinthis gious Orders Study (ROS) and the Rush Memory and Aging analysis. We defined the UNC5C region as the chromosomal seg- Project (MAP).5,6 Both studies enroll older adults without known ment containing the transcribed elements of UNC5C and 6100 kb dementia, and each participant signed written informed consent of flanking DNA (chr4: 95,983,655-96,570,361 on hg19 build), and Anatomical Gift Act. All study steps were done in compliance which contained 2,269 1000 Genomes Project reference SNPs. Of with the protocol approved by the Rush University Medical note, APOE genotypes were measured through a separate sequencing Center Institutional Review Board. Overall follow-up rate and procedure as previously described,5,6 and APOE e2ande4allele autopsy rate for deceased individuals were greater than 85%. Of counts were used in this study. RNA was extracted from frozen note, these 2 cohorts were designed to be used in combined postmortem DLPFC from a subset of participants and was quanti- analyses: ROS and MAP were designed and managed by the same fied using next-generation quantitative RNA-sequencing (RNA- team of investigators at the Rush Alzheimer’s Disease Center and Seq). QC and quantile normalization of RNA-Seq Fragments Per captured shared phenotypic measures. We limited our analyses to Kilobase of transcript per Million fragments mapped (FPKM) values deceased European-descent individuals with quality-controlled for UNC5C were done as previously described,12 and 494 subjects in genome-wide genotyping data and completed autopsy (February our study had nonmissing values. 2016 Data Freeze, total n 5 1,288).

Cognitive and pathologic phenotypes. Nine AD-related Statistical analyses. Statistical analyses were performed with R cognitive and pathologic variables (diagnosis of AD dementia, 3.2.1 (r-project.org), and we combined ROS and MAP in all our global cognitive decline, pathologic diagnosis of AD, b-amyloid analyses, unless otherwise specified. To derive the number of load, neuritic plaque count, diffuse plaque count, paired helical independent genetic markers tested in this study, we used LD- 2 filament tau [PHFtau] density, neurofibrillary tangle count, and based SNP pruning (with an r threshold of 0.2, a window of 50 the CAA score) were defined as previously described.3,5–7 Square- SNPs, and a step of 5 SNPs), done with PLINK, as previously 3 root transformed values were used for analyses of continuous published. This procedure yielded 391 independent LD blocks pathologic variables (b-amyloid load, neuritic plaque count, dif- within the UNC5C region. As we tested for 9 traits, Bonferroni- a a 5 3 25 fuse plaque count, PHFtau density, and neurofibrillary tangle adjusted for the primary analysis was 1.4 10 (0.05 ÷ 3 count) to account for their positively skew deviation, as previously [391 9]). Assuming additive genetic effects, logistic regression described.3 CAA was graded on a 5-point scale (0–4) in 4 neo- models were used for dichotomous outcome variables (AD cortical regions (dorsolateral prefrontal cortex [DLPFC], angular dementia and pathologic AD), and linear regression models were gyrus, inferior temporal gyrus, and calcarine cortex) and averaged to used for others, with covariates including age at death, sex, cohort derive a CAA score (0–4).7 Of note, among the 1,288 subjects in (ROS vs MAP), and the first 3 principal components from the this study, numbers of subjects with measured values for each genotype covariance matrix derived with EIGENSTRAT. Years phenotype, who were included in each analysis, were as follows: of education was also included as a covariate in the analyses of diagnosis of AD dementia (n 5 1,257), global cognitive decline phenotypes affected by cognitive performance (AD dementia and (n 5 1,193), pathologic diagnosis of AD (n 5 1,140), b-amyloid global cognitive decline). Results from 2 genotyping platforms load (n 5 1,095), neuritic plaque count (n 5 1,132), diffuse were calculated separately and meta-analyzed with PLINK, and plaque count (n 5 1,132), PHFtau density (n 5 1,088), neuro- the consistencies across genotyping platforms were examined with 2 13 fibrillary tangle count, (n 5 1,132), and CAA score (n 5 1,107). meta-analyses I values. The results were plotted using 14 For secondary analyses, atherosclerosis, arteriolosclerosis, LocusZoom, with chromatin state tracks predicted by the 15 gross infarcts, and microinfarcts are measured as previously ChromHMM core 15-chromatin state model from the Road- 16 described.8 In brief, atherosclerosis was assessed by visual inspec- map Epigenomics Project. tion at the circle of Willis, and the severity was semiquantitatively For SNPs associated with a trait (CAA), we looked up the LD 2 graded (scale from 0 to 6), which was collapsed into 4 levels r values between the lead SNP and all other identified SNPs in 17 (none/possible, mild, moderate, and severe) for analysis. Arterio- the HaploReg database and also analyzed each of the other losclerosis was microscopically evaluated in anterior basal ganglia SNPs in a same linear model with the lead SNP to examine with a semiquantitative scale (0–7), which was collapsed into 4 whether they represent a different LD block or not. The popu- levels (none, mild, moderate, and severe) for analysis. Gross in- lation variance explained by the lead SNP was derived from farcts and microinfarcts were recorded as present vs absent. a sequential adjusted R-squared analysis with linear models: We Among the 1,288 subjects in this study, numbers of subjects with first calculated the adjusted R-squared from a linear model to measured values for each phenotype, who were included in each explain the trait of interest with age at death, sex, study cohort, analysis, were as follows: atherosclerosis (n 5 1,141), arteriolo- genotyping platform, and first 3 principal components from the sclerosis (n 5 1,130), gross infarcts (n 5 1,136), and micro- genotype covariance matrix. Then, we included the lead ’ infarcts (n 5 1,136). SNP s minor allele dosage in the model and calculated the incre- ment of the adjusted R-squared. The population variance ex- Genotyping and RNA-Seq data acquisition and processing. plained by the SNP was then compared with that of APOE e4. Genome-wide genotyping (on either Affymetrix GeneChip 6.0 or Il- To further evaluate the functional significance of the LD lumina OmniQuad Express), quality control (QC), and imputation block associated with increased CAA severity, we examined the

2 Neurology: Genetics predicted chromatin state (per the ChromHMM core 15- Affymetrix GeneChip 6.0, and 174 were genotyped chromatin state model from the Roadmap Epigenomics Pro- with Illumina OmniQuad Express. In our primary 15,16 ject ) of each SNP associated with CAA severity. Then, the analyses using additive models, 10 UNC5C intronic lead SNP was tested for the association with the UNC5C RNA SNPs were associated with a higher CAA score (p , level in a linear model, controlling for age at death, sex, study 3 25 cohort, genotyping platform, first 3 principal components from 1.4 10 ) (figure 1 and table 2). All 10 SNPs were the genotype covariance matrix, and technical covariates (RNA imputed SNPs with good imputation quality (INFO integrity score, log2(total aligned reads), postmortem interval, score .0.89; table e-1 at Neurology.org/ng), and the and number of ribosomal bases). Then, the UNC5C RNA level association was consistent across genotyping plat- was tested for its association with CAA severity, controlling for forms, shown by meta-analysis I2 5 0 for all 10 SNPs. the same set of covariates. No other associations are found between UNC5C In addition, to test whether the association between the iden- . tified UNC5C variant and CAA is independent of known genetic variants and other tested AD-related traits (p 5. 24 risk factors for CAA (APOE e4, e2, and CR1 rs6656401A),18 each 0 3 10 for all other trait-SNP pairs, well above the of the previously reported genetic risk factors was added to the threshold type I error rate of this study; figure e-1). linear model testing association between the UNC5C variant and The top CAA risk allele rs28660566T (MAF 5 the CAA score. Interaction between the UNC5C variant and 3.6%) was in LD with the other 9 identified SNPs a previously reported genetic risk factor was checked using the (r2 $ 0.36, D9 $ 0.93; table 2). To rule out their product term in the linear regression model, if the association of T the UNC5C variant and the CAA score attenuated after including effect independent of rs28660566 , each of the 9 that known genetic risk factor in the model. After the statistical SNP dosages was tested in a same linear additive interaction was found, stratified analysis was done to further test model with rs28660566T, and none were associated the relationship between 2 genetic risk factors. Of note, we with the CAA score when adjusted for rs28660566T e e assumed an additive effect of CAA severity by APOE 4, 2, or (p . 0.05). Thus, rs28660566T tags all other SNPs A CR1 rs6656401 in our models. associated with a higher CAA score, and this LD Further exploratory analyses were performed to interrogate the association of the identified UNC5C CAA variant and cere- block defines a chromosomal segment containing sev- brovascular pathology. Ordinal logistic regression models were eral UNC5C exons (figure 1). In the linear model, the used for the ordered categorical variables (atherosclerosis and presence of an rs28660566T allele corresponded to arteriolosclerosis), and logistic regression models were used for a 0.60-point higher CAA score, which is approxi- the binary variables (gross infarct and microinfarct). Age at death, mately 75% of the effect size of an APOE e4 allele sex, study cohort, genotyping platform, and first 3 principal com- on this trait (estimated effect 0.80, p 5 1.5 3 10238). ponents from the genotype covariance matrix were included as The UNC5C variant therefore has a large effect, covariates for each analysis. p values were Bonferroni adjusted for T multiple comparisons (number of testing 5 4 for these targeted but rs28660566 (MAF 3.6%) explained approxi- exploratory analyses). mately only 1.9% of the population variance given Finally, the sample size required to reproduce our result was its low MAF, whereas APOE e4 (MAF 13.8%) ex- calculated with the Genetic Power Calculator, assuming an addi- plained approximately 13.1% of the population 19 tive model and a similar effect size. variance. Of note, rs28660566T dosage was not associated with parenchymal b-amyloid load RESULTS Demographic characteristics of the sub- (p 5 0.80). jects are summarized in table 1. The total number The predicted chromatin state of a given chro- of subjects analyzed in this study was n 5 1,288. mosomal segment (table 2 and figure 1) can be help- Among the subjects, 1,114 were genotyped with ful in inferring its functional implications (e.g., enhancer, promoter, etc.). The lead SNP rs28660566T was within a chromosomal region Table 1 Demographic characteristics of the subjects identified as an enhancer in multiple brain regions,

ROS MAP Combined Nonmissing suggesting that this region may have a regulatory role.NoneoftheotheridentifiedSNPswerein Cohort size, n 614 674 1,288 NA the regulatory chromosomal region in the DLPFC. Age at death, mean (SD) 87.8 (6.9) 89.8 (6.0) 88.9 (6.5) 1,288 Of interest, rs28660566T dosage was nominally Sex, female, n (%) 383 (62.4) 456 (67.7) 839 (65.1) 1,288 associated with decreased UNC5C mRNA expres- Education, y (SD) 18.2 (3.4) 14.4 (2.8) 16.3 (3.64) 1,287 sion in the DLPFC (estimated effect 20.097, 95%

Diagnosis of AD dementia, n (%) 261 (43.4) 246 (37.5) 507 (40.3) 1,257 confidence interval [CI] 20.19 to 20.0063, p 5

Pathologic diagnosis of AD, n (%) 361 (62.7) 360 (63.8) 721 (63.2) 1,140 0.036), but the DLPFC UNC5C mRNA expression level was not associated with the CAA score (p 5 CAA, median (IQR) 0.75 (1.5) 0.75 (1.5) 0.75 (1.5) 1,107 0.75). Abbreviations: AD 5 Alzheimer disease; CAA 5 cerebral amyloid angiopathy; IQR 5 inter- APOE e4, e2, and CR1 rs6656401A have been 5 5 5 quartile range; MAP Rush Memory and Aging Project; NA not applicable; ROS Reli- previously reported to be associated with higher odds gious Orders Study. 18 Nonmissing indicates number of participants with measured values in the combined cohort, and/or severity of CAA. The association between who are included in the genetic association study for each trait. rs28660566T and the CAA score was attenuated by

Neurology: Genetics 3 Figure 1 Regional association plot of the CAA score in the UNC5C region

Regional association plots of CAA in UNC5C 6100 kb. X-axis denotes chromosomal location on , and Y-axis denotes negative log of the p value of the association between the SNP and the CAA score, adjusting for age at death, sex, study cohort, and first 3 principal components from the genetic covariance matrix. Each point corresponds to each SNP, and the color of each point indicates linkage disequilibrium r2 value between each SNP of interest and rs28660566T. Blue line in the graph represents recombination rate in cM/Mb. The color-coded tracks below the association plot show predicted functional chromatin states in inferior temporal cortex (Inf Temp), DLPFC, and angular gyrus cortex (angular), derived from the Roadmap Epigenomics project. CAA 5 cerebral amyloid angiopathy; DLPFC 5 dorsolateral prefrontal cortex; TSS 5 transcription start site; ZNF 5 zinc finger. The track below the predicted chromatin states shows the location of genes: the thick line is exon, and the thin line is intron.

approximately 13% when the APOE e4 allele count In an additional exploratory analysis to test was included in the model (estimated effect 0.52, whether rs28660566T alters the relationship 95% CI 0.29–0.75, p 5 1.2 3 1025). In fact, a sta- between parenchymal b-amyloid load and the tistical interaction between APOE e4 allele count and CAA score, we found no statistically significant rs28660566T dosage on the CAA score was present, interaction between rs28660566T and parenchymal suggesting a synergistic effect modification between b-amyloid load on their associations with the CAA these 2 genetic risk factors (estimated effect 0.44, score (p 5 0.42). Then, we also explored whether 95% CI 0.049–0.84, p 5 0.027). In a stratified anal- rs28660566T was associated with other cerebrovas- ysis, the association between rs28660566T dosage and cular pathologies (Bonferroni-corrected p value the CAA score was weaker in a subgroup with no threshold 0.0125): rs28660566T was associated with APOE e4 allele (n 5 822, estimated effect 0.38, increased odds of more severe arteriolosclerosis 95% CI 0.12–0.64, p 5 0.0046) compared with (odds ratio [OR] 1.8; 95% CI 1.2–2.8; p 5 the association in a subgroup with 1 or 2 APOE e4 0.0065) but not with atherosclerosis (OR 1.6; alleles (n 5 285, estimated effect 1.00, 95% CI 0.50– 95% CI 1.1–2.4; p 5 0.023), gross infarcts (p 5 1.50, p 5 1.2 3 1024). By contrast, the association 0.54), or microinfarcts (p 5 0.69). between rs28660566T and the CAA score did not Finally, we calculated the sample size required change by adding APOE e2orCR1 rs6656401A in to replicate our result with the Genetic Power Calcu- the model (with APOE e2: estimated effect 0.60; lator,19 assuming a similar effect size. Assuming an 95% CI 0.35–0.85; p 5 2.0 3 1026, with additive model with 4% of MAF, 389 subjects are rs6656401A: estimated effect 0.60; 95% CI 0.35– needed to detect 2% of total population variance with 0.85; p 5 2.3 3 1026). a type I error rate of 0.05 and power of 80%.

4 Neurology: Genetics Table 2 SNPs associated with the severity of CAA

Chromatin state Estimated effect SNP Chr4 POS MAF LD (r2) (95% CI) p Value Inf temp DLPFC Angular

rs28660566T 96,125,762 0.036 NA 0.60 (0.35–0.85) 2.3 3 1026 Enh Enh Enh

rs11941383T 96,151,958 0.042 0.76 0.52 (0.30–0.74) 4.1 3 1026 TxWk TXWk TxWk

rs140160653A 96,116,350 0.017 0.36 0.78 (0.44–1.13) 8.8 3 1026 Enh TxWk EnhG

rs77321857T 96,130,211 0.022 0.52 0.69 (0.38–1.00) 9.8 3 1026 Quies TxWk TxWk

rs74590670G 96,163,984 0.022 0.52 0.69 (0.38–1.00) 1.1 3 1025 Tx Tx TxWk

rs139069750G 96,148,504 0.022 0.52 0.69 (0.38–1.00) 1.2 3 1025 TxWk TxWk TxWk

rs138877862C 96,136,118 0.022 0.46 0.69 (0.38–0.99) 1.2 3 1026 TxWk TxWk TxWk

rs75985930A 96,142,169 0.022 0.46 0.69 (0.38–0.99) 1.2 3 1026 Enh TxWk TxWk

rs75083968A 96,127,609 0.022 0.52 0.69 (0.38–0.99) 1.2 3 1025 Quies TxWk TxWk

rs147682457A 96,139,245 0.022 0.52 0.68 (0.38–0.99) 1.3 3 1025 TxWk TxWk TxWk

Abbreviations: Angular 5 angular gyrus; CAA 5 cerebral amyloid angiopathy; CI 5 confidence interval; DLPFC 5 dorso- lateral prefrontal cortex; Enh 5 enhancers; EnhG 5 genic enhancers; Inf Temp 5 inferior temporal cortex; LD 5 linkage disequilibrium; MAF 5 minor allele frequency; MAP 5 Rush Memory and Aging Project; Quies 5 Quiescent/low; ROS 5 the Religious Orders Study; SNP 5 single nucleotide polymorphism; Tx 5 strong transcription; TxWk 5 weak transcription. In the combined ROS-MAP cohort, linear regression model was used to assess the association of each genotype with the CAA score, adjusting for age at death, sex, study cohort, and first 3 principal components from the population covariance matrix. Estimated effect indicates estimated CAA score increase per each minor allele of the SNP of interest. Chr4 POS shows chromosomal position of each SNP per hg19 reference map. Chromatin state shows predicted chromatin state in 3 cortical regions available from the Roadmap Epigenomics data set and ChromHMM algorithm (core 15-state model), in the chromosomal region where the SNP is located. LD(D9) 5 1andmeta-analysesI2 5 0 across genotyping platforms for all SNPs in this table.

DISCUSSION Targeted analysis of the chromosomal The functional mechanism of the UNC5C CAA region of UNC5C, a recently identified AD suscepti- severity risk locus remains to be determined. The lead bility gene, uncovered a possible new risk locus asso- SNP, rs28660566T, is located within a predicted ciated with CAA severity, which comprises a single enhancer chromosomal locus in multiple brain re- LD block captured by rs28660566T. Of interest, gions,15,16 and in our study, it was a weak eQTL rs28660566T synergistically interacts with APOE e4 associated with lower UNC5C mRNA expression in to increase CAA severity, suggesting a possibility of the DLPFC. However, the DLPFC UNC5C mRNA a common downstream pathway. level was not associated with CAA severity. Nonethe- In a recent study reporting the association of a rare less, our DLPFC RNA-Seq data lack cell-type specific UNC5C coding variant rs137875858A (UNC5C information and might not capture differential T835M) with LOAD, the authors showed that expression limited to certain tissue types such as arte- UNC5C T835M exerts its effect through increased rioles. As CAA is a disease primarily of small cortical neuronal vulnerability rather than increased amyloid and leptomeningeal arterioles,4 gene expression pro- or tau production.2 Consistent with this report, no files from isolated leptomeningeal and cortical arterio- common UNC5C variant was associated with paren- les might be necessary to further dissect the chymal b-amyloid or PHFtau accumulation in our mechanisms of UNC5C in CAA progression. study. Thus, the association of UNC5C Of note, the previously reported UNC5C T835M rs28660566T with CAA is more likely to be mediated variant (chromosome 4, position 96,091,431) is by ineffective perivascular b-amyloid drainage rather 34.3 kb away from rs28660566T, and it is not included than increased neuronal b-amyloid production. Of in the chromosomal segment defined by the LD block interest, rs28660566T was also associated with arterio- captured by rs28660566T. UNC5C T835M might be losclerosis, a disease of small deep cerebral arterioles, on a same haplotype with rs28660566T,giventhelow suggesting an implication of UNC5C in arteriolar sus- recombination frequency between the 2 loci (figure 1), ceptibility to pathologic changes, not only limited to but we cannot confirm this in ROS-MAP participants vascular b-amyloid deposition. In fact, a previous because the sequencing of the UNC5C region is not study reported that UNC5C is expressed in endothelial yet available. However, given the rarity of T835M cells, and inhibition of UNC5C attenuated netrin-1 (MAF ,0.1%), it is very unlikely that the observed effect on in vitro endothelial cell migration,20 suggest- association is driven by UNC5C T835M in our subject ing a potential role of UNC5C in vascular development population (n 5 1,107 for the CAA analysis). We also and maintenance. note that rs28660566T is not in LD with rs3846455G

Neurology: Genetics 5 (237.3 kb apart from rs28660566T), an UNC5C allele Journal of Neuropathology & Experimental Neurology;hasbeenacon- that we have recently reported to be associated with sultant for AVID Radiopharmaceuticals, Navidea Biopharmaceuticals Inc., the Michael J. Fox Foundation, National Football League, and worse late-life cognition, when adjusted for the burden National Hockey League; has received research support from AVID of multiple neuropathologies (including CAA) and Radiopharmaceuticals, NIH, and NIA; and has taken part in legal demographics.21 Thus, there are both allelic and phe- proceedings involving National Football League, National Hockey League, and World Wrestling Entertainment. D.A. Bennett has served notypic heterogeneity within the UNC5C locus, sim- on scientific advisory boards for Vigorous Minds, Takeda Pharmaceut- 3 ilar to our previous findings in the TREM locus. icals, and AbbVie; has served on the editorial boards of Neurology, The study is methodologically robust. The analyses Current Alzheimer Research,andNeuroepidemiology; and has received are performed on more than 1,100 participants from 2 research support from NIH. P.L. De Jager has served on scientific advisory boards for TEVA Neuroscience, Genzyme/Sanofi, and Celgene; community-based cohort studies, with a semiquantitative has received speaker honoraria from Biogen Idec, Source Health care Ana- assessment of CAA severity in multiple cortical regions. lytics, Pfizer Inc., and TEVA; has served on the editorial boards of the We performed a targeted gene analysis of UNC5C, Journal of Neuroimmunology, Neuroepigenetics,andMultiple Sclerosis; and has and a set of UNC5C variants showed significant received research support from Biogen, Eisai, UCB, Pfizer, Sanofi/Gen- zyme, NIH, NIA, and the National MS Society. Go to Neurology.org/ association with the CAA severity after rigorously ng for full disclosure forms. correcting for the number of hypotheses tested. Moreover, the result was consistent across 2 geno- Received March 21, 2017. Accepted in final form June 6, 2017. typing platforms. Validation efforts with indepen- dent data sets are essential to extend the evaluation REFERENCES 1. Leonardo ED, Hinck L, Masu M, Keino-Masu K, Acker- of this risk locus, and close to 400 subjects with man SL, Tessier-Lavigne M. Vertebrate homologues of C. pathologic assessment of CAA in multiple cortical elegans UNC-5 are candidate netrin receptors. Nature areas are required to replicate our findings with type 1997;386:833–838. 1 error rate of 0.05 and power of 80%. In addition, 2. Wetzel-smith MK, Hunkapiller J, Bhangale TR, et al. A experimental studies would be essential to under- rare mutation in UNC5C predisposes to late-onset Alz- ’ stand the underlying biological mechanisms. In con- heimer s disease and increases neuronal cell death. Nat Med 2014;20:1452–1457. clusion, we found an UNC5C variant to be 3. Replogle JM, Chan G, White CC, et al. A TREM1 variant associated with CAA severity, and additional studies alters the accumulation of Alzheimer-related amyloid arerequiredtoclarifytheroleofUNC5C in CAA pathology. Ann Neurol 2015;77:469–477. pathogenesis. 4. Charidimou A, Gang Q, Werring DJ. Sporadic cerebral amyloid angiopathy revisited: recent insights into patho- AUTHOR CONTRIBUTIONS physiology and clinical spectrum. J Neurol Neurosurg Hyun-Sik Yang and Charles C. White: drafting/revising the manuscript Psychiatry 2012;83:124–137. for content, study concept or design, statistical analysis, and analysis or 5. Bennett DA, Schneider JA, Arvanitakis Z, Wilson RS. interpretation of data. Lori B. Chibnik: drafting/revising the manuscript Overview and findings from the Religious Orders Study. for content, study concept or design, and analysis or interpretation of Curr Alzheimer Res 2012;9:628–645. data. Hans-Ulrich Klein: drafting/revising the manuscript for content 6. Bennett DA, Schneider JA, Buchman AS, Barnes LL, and analysis or interpretation of data. Julie A. Schneider and David A. Boyle PA, Wilson RS. Overview and findings from the Bennett: drafting/revising the manuscript for content, analysis or inter- rush Memory and aging project. Curr Alzheimer Res pretation of data, acquisition of data, and obtaining funding. Philip L. – De Jager: drafting/revising the manuscript for content, study concept 2012;9:646 663. or design, analysis or interpretation of data, acquisition of data, study 7. Boyle PA, Yu L, Nag S, et al. Cerebral amyloid angiopathy supervision, and obtaining funding. and cognitive outcomes in community-based older per- sons. Neurology 2015;85:1930–1936. ACKNOWLEDGMENT 8. Arvanitakis Z, Capuano AW, Leurgans SE, Buchman AS, The authors thank the study participants. More information regarding ob- Bennett DA, Schneider JA. The relationship of cerebral taining (ROS/MAP/ROS and MAP) data for research use can be found at vessel pathology to brain microinfarcts. Brain Pathol 2017; the RADC Research Resource Sharing Hub (radc.rush.edu). 27:77–85. 9. Purcell S, Neale B, Todd-Brown K, et al. PLINK: STUDY FUNDING a tool set for whole-genome association and The study was funded by NIH grants P30AG10161, R01AG17917, population-based linkage analyses. Am J Hum Genet R01AG036836, RF1AG15819, and U01AG046152. 2007;81:559–575. 10. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, DISCLOSURE Shadick NA, Reich D. Principal components analysis cor- ’ H.-S. Yang has received research support from Alzheimer sAssocia- rects for stratification in genome-wide association studies. tion Clinical Fellowship. C.C. White reports no disclosures. L.B. Nat Genet 2006;38:904–909. Chibnik has received research support from NIH/NIA. H.-U. Klein 11. Browning BL, Browning SR. A unified approach to geno- reports no disclosures. J.A. Schneider has served on scientific advi- sory boards for Alzheimer’s Association, Fondation Plan Alzheimer type imputation and haplotype-phase inference for large (France), the Dutch CAA Foundation, University of Washington/ data sets of trios and unrelated individuals. Am J Hum Group Health Alzheimer’s Disease Patient Registry/Adult Changes Genet 2009;84:210–223. in Thought study, New York University, AVID Radiopharmaceut- 12. Chan G, White CC, Winn PA, et al. CD33 modulates icals, Genentech, Grifols, and Eli Lilly; has served on the editorial TREM2: convergence of Alzheimer loci. Nat Neurosci boards of the Journal of Histochemistry & Cytochemistry and the 2015;18:1556–1558.

6 Neurology: Genetics 13. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Mea- 18. Biffi A, Shulman JM, Jagiella JM, et al. Genetic variation suring inconsistency in meta-analyses. BMJ 2003;327: at CR1 increases risk of cerebral amyloid angiopathy. Neu- 557–560. rology 2012;78:334–341. 14. Pruim RJ, Welch RP, Sanna S, et al. LocusZoom: regional 19. Purcell S, Cherny SS, Sham PC. Genetic Power visualization of genome-wide association scan results. Bio- Calculator: design of linkage and association genetic informatics 2010;26:2336–2337. mapping studies of complex traits. Bioinformatics 15. Ernst J, Kellis M. ChromHMM: automating chromatin- 2003;19:149–150. state discovery and characterization. Nat Methods 2012;9: 20. Lejmi E, Leconte L, Pedron-Mazoyer S, et al. Netrin-4 215–216. inhibits angiogenesis via binding to neogenin and recruit- 16. Roadmap Epigenomics C, Kundaje A, Meuleman W, et al. ment of Unc5B. Proc Natl Acad Sci USA 2008;105: Integrative analysis of 111 reference human epigenomes. 12491–12496. Nature 2015;518:317–330. 21. White CC, Yang HS, Yu L, et al. Identification of genes 17. Ward LD, Kellis M. HaploReg: a resource for exploring associated with dissociation of cognitive performance and chromatin states, conservation, and regulatory motif alter- neuropathological burden: Multistep analysis of genetic, ations within sets of genetically linked variants. Nucleic epigenetic, and transcriptional data. PLoS Med 2017;14: Acids Res 2012;40:D930–D934. e1002287.

Neurology: Genetics 7 Clinical/Scientific Notes

Corina Anastasaki, PhD UPDATED NOMENCLATURE FOR HUMAN AND the predicted amino acid sequences of the human Lu Q. Le, MD, PhD MOUSE NEUROFIBROMATOSIS TYPE 1 GENES (RefSeq NM_000267; UniProt P21359) and mouse Robert A. Kesterson, PhD (UniProt Q04690) full-length transcripts using David H. Gutmann, MD, Neurofibromatosis type 1 (NF1; OMIM 162200) is protein BLAST revealed 98% amino acid sequence PhD one of the most common neurogenetic conditions, identity (figure e-3). affecting 1 in 3,000 people worldwide. Characterized Next, we examined the previously published alter- Neurol Genet natively spliced exons: exon 9a contains a 30- 2017;3:e169; doi: 10.1212/ by a propensity to develop nervous system tumors, NXG.0000000000000169 learning and behavioral deficits, and pigmentary nucleotide sequence encoding 10 amino acids and is abnormalities, NF1 is caused by a germline sequence predominantly expressed in postmitotic brain neu- 3 alteration in the NF1 gene (OMIM: 613113; chro- rons, whereas the muscle-specific exon 48a contains mosome 17q11.2). In addition, somatic NF1 54 nucleotides that encode 18 amino acids at the 4 sequence changes have been reported in numerous carboxyl terminus of the protein. The 63- other cancers,1 extending their importance to malig- nucleotide exon 23a encodes 21 amino acids that nancies occurring in the general population. are inserted within the RAS-GTPase-activating pro- The NF1 gene, comprising 57 exons, was initially tein (GAP)-related domain of the NF1 protein (neu- identified, and its complete coding sequence was assem- rofibromin), resulting in diminished RAS-GAP 4 bled over 25 years ago.2 In the ensuing decades, 4 alter- activity. The most recently described alternatively natively spliced exons (9a, 10a-2, 23a, and 48a) were spliced exon, exon 10a-2, contains 45 nucleotides discovered and added to a large number of exons with and encodes 15 amino acids, is located at the amino temporary numerical assignments (e.g., 10a2 or 23.2). terminus of neurofibromin, and this sequence has This has led to considerable confusion and inconsisten- been hypothesized to direct intracellular membrane cies in clinical and scientific publications, which also has targeting by virtue of its predicted transmembrane critical implications for NF1 DNA sequence alteration domain.5 Using published alternatively spliced exon reporting as part of patient management, genotype- sequences (figure e-1), BLAST searches were per- phenotype correlations, and the interpretation of small formed, and these alternatively spliced exons were animal models engineered with patient-specific renumbered and named according to their location sequence changes. To provide a unified annotation sys- (figure 1)—between exons 11–12 (11alt12; formerly tem for the NF1 gene, the mouse and human genes 9a), exons 12–13 (12alt13; formerly 10a-2), exons were aligned and assembled using published sequences 30–31 (30alt31; formerly 23a), and exons 56–57 for sequential numbering and comparison. (56alt57; formerly 48a). In this regard, we propose The full-length cDNA sequences of NF1 (8457 that the nucleotides (and the encoded amino acids) of nucleotides; figure e-1 at Neurology.org/ng) and Nf1 these alternatively spliced exons be numbered (8463 nucleotides; figure e-2) genes were initially 11alt12_1–30 (1–10), 12alt13_1–45 (1–15), aligned using Nucleotide Basic Local Alignment 30alt31_1–63 (1–21), and 56alt57_1–54 (1–18). Search Tool (BLAST) (NIH) optimized for mega- Possible new alternatively spliced exons should be blast, revealing 92% sequence identity with 0.08% similarly named based on their location. mismatches (figure e-2). Excluding the alternatively 11alt12 is 100% identical to both the cDNA and spliced exons, the ubiquitously expressed 57 exons amino acid sequences of the predicted Mus musculus were consecutively numbered (figure 1 and tables transcript variant X7 (mRNA XM_006532442.3), Supplemental data at e1–e3). As such, the full-length human NF1 tran- while 30alt31 is 98.4% identical to the cDNA and Neurology.org/ng script, NF1-002 (ENST00000356175.7), contains 95.2% identical to the amino acid sequence (contain- 57 exons and encodes 2818 amino acids, while the ing a single conservative lysine-to-arginine change) of corresponding full-length mouse Nf1 transcript, Nf1- MusmusculusNf1(ENSMUST00000071325.8) 003 (ENSMUST00000108251.8), comprises 57 (figures e-4 and e-5). Notably, the 12alt13 and exons and encodes 2,820 amino acids, with 2 addi- 56alt57 human sequences were not found in either tional amino acids encoded by exon 17. Alignment of mouse transcriptome or genome databases following

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure 1 Proposed nomenclature for human NF1 and mouse Nf1 genes

(A) Schematic diagram illustrating the 57 exons of the human (top panel) and mouse (bottom panel) full-length transcripts, as well as the positions of the alternatively spliced exons (11alt12, 12alt13, 30alt31, and 56alt57). The proposed numbering system is denoted in each exon, while the prior numbering system is indicated below each exon. The RAS-GTPase-activating protein and the CRAL-TRIO domains are highlighted in gray. (B) Table detailing the nucleotides and amino acids spanning each of the 57 exons in human NF1 and mouse Nf1 genes. CRAl-TRIO domain 5 cellular retinaldehyde-binding protein and TRIO guanine exchange factor structural domain.

megablast, discontinuous megablast, or BLASTN numbering system will provide better calibration of searches (figures e-4 and e-5). Based on the absence of human genetic testing results with Nf1 preclinical 12alt13 and 56alt57 sequences in the mouse, further precision medicine models developed using geneti- studieswillberequiredtodefinetheirrolesinneuro- cally engineered mice6 and induced pluripotent stem fibromin function. It should be noted that neither cells.7 12alt13 nor 56alt57 has been reported in other reference From the Department of Neurology (C.A., D.H.G.), Washington model organisms (rat or zebrafish; data not shown). University in St. Louis, MO; Department of Dermatology (L.Q. L.), University of Texas Southwestern Medical Center, Dallas; and Taken together, we assembled and updated the Department of Genetics (R.A.K.), The University of Alabama at human NF1 and mouse Nf1 sequences to facilitate Birmingham. greater consistency in both clinical reporting and Author contributions: C.A. performed the analyses, assembled the basic research studies on NF1. The clarification of figures and tables, and helped write the manuscript. L.Q.L. and R.A.K. were involved in the design of the study and edited the the nomenclature for the NF1 gene is a necessary step manuscript. D.H.G. wrote the manuscript with C.A. and oversaw to enable standardization of exon, nucleotide, and the study at all stages. amino acid numbering, which is particularly impor- Study funding: D.H.G. is supported by a Research Program Award from the National Institute of Neurological Disorders and Stroke (1- tant for reporting patient-specific DNA sequence var- R35-NS097211-01). The authors also acknowledge support from iants. In addition, implementation of this proposed The Giorgio Foundation.

2 Neurology: Genetics Disclosure: C. Anastasaki and L.Q. Le report no disclosures. R.A. 1. Patil S, Chamberlain RS. Neoplasms associated with germ- Kesterson has served on the editorial board of Frontiers in Genetics line and somatic NF1 gene mutations. Oncologist 2012;17: and has received research support from UAB Hepato/Renal Fibrocys- 101–116. tic Diseases Core Center, UAB Comprehensive Cancer Center, UAB 2. Marchuk DA, Saulino AM, Tavakkol R, et al. cDNA cloning Center for Nutrition and Obesity Research, UAB Rheumatic Disease of the type 1 neurofibromatosis gene: complete sequence of Core Center, UAB Diabetes Research Center, and the Giorgio Foun- the NF1 gene product. Genomics 1991;1:931–940. dation. D.H. Gutmann has served on scientific advisory boards for the American Association for Cancer Research and the National 3. Geist RT, Gutmann DH. Expression of a developmentally- Institute for Neurological Disorders and Stroke Advisory Council; regulated neuron-specific isoform of the neurofibromatosis – has served on the editorial boards of Glia and Familial Cancer; holds 1 (NF1) gene. Neurosci Lett 1996;211:85 88. patents for Identification of the NF1 gene and protein and Identifi- 4. Gutmann DH, Andersen LB, Cole JL, Swaroop M, Collins cation of NF1 protein as mTOR regulator; has received research FS. An alternatively-spliced mRNA in the carboxy terminus support from the US Army Department of Defense, NCI Tumor of the neurofibromatosis type 1 (NF1) gene is expressed in Microenvironment Network, NIH, the Giorgio Foundation, and muscle. Hum Mol Genet 1993;2:989–992. ’ Children s Tumor Foundation; and receives royalty payments from 5. Kaufmann D, Müller R, Kenner O, et al. The N-terminal the University of Michigan. Go to Neurology.org/ng for full disclosure splice product NF1-10a-2 of the NF1 gene codes for a trans- forms. The Article Processing Charge was funded by the authors. membrane segment. Biochem Biophys Res Commun 2002; This is an open access article distributed under the terms of the 294:496–503. Creative Commons Attribution-NonCommercial-NoDerivatives Li- 6. Li K, Turner AN, Chen M, et al. Mice with missense and cense 4.0 (CC BY-NC-ND), which permits downloading and shar- ing the work provided it is properly cited. The work cannot be nonsense NF1 mutations display divergent phenotypes changed in any way or used commercially without permission from compared with human neurofibromatosis type I. Dis Model – the journal. Mech 2016;9:759 767. 7. Anastasaki C, Woo AS, Messiaen LM, Gutmann DH. Elu- Received February 10, 2017. Accepted in final form April 20, 2017. cidating the impact of neurofibromatosis-1 germline muta- tions on neurofibromin function and dopamine-based Correspondence to Dr. Gutmann: [email protected] learning. Hum Mol Genet 2015;24:3518–3528.

Neurology: Genetics 3 Clinical/Scientific Notes

Enrico Bugiardini, MD HOMOZYGOUS MUTATION IN HSPB1 CAUSING areflexic. Her CK level was 404 IU/L. Neurophys- Alexander M. Rossor, DISTAL VACUOLAR MYOPATHY AND MOTOR iologic studies suggested an axonal motor neurop- MRCP NEUROPATHY athy. Sensory nerve action potentials were, David S. Lynch, MRCPI nevertheless, at the lower limit of normal for ampli- Michael Swash, FRCP tude in the lower limbs (right sural 6 mV, right Alan M. Pittman, PhD Case report. A 57-year-old woman, born to parents superficial peroneal 7 mV, and normal range .5 mV), Julian C. Blake, MRCP of Gujarati Indian descent (figure, A), presented at and distal lower limb motor responses were absent. Michael G. Hanna, FRCP age 19 with pain and stiffness in her calves and a ten- Needle EMG showed prominent chronic neuro- Henry Houlden, FRCP dency to trip. In her 20s, a formal neurologic exam- genic changes with large motor units recruiting in Janice L. Holton, ination demonstrated predominantly distal lower reduced numbers but at increased firing rates to FRCPath limb weakness and normal upper limb muscle a reduced interference pattern. This EMG pattern Mary M. Reilly, FRCPI strength.1 Motor and sensory nerve conduction stud- was most pronounced distally but evident proxi- Emma Matthews, MRCP ies were normal with the exception that no motor mally in the upper and lower limbs. No low ampli- response was elicited from the extensor digitorum tude or brief polyphasic motor units were seen on Neurol Genet brevis. Fibrillations and polyphasic action potentials any occasion at re-presentation. Muscle MRI was 2017;3:e168; doi: 10.1212/ NXG.0000000000000168 were present on EMG. The creatine kinase (CK) level performed, showing widespread severe muscle fatty was 1,452 IU/L. A quadriceps muscle biopsy was replacement (figure, C). performed at age 27 from which images were available There was no relevant family history. Her mother in the records.1 At that time, muscle fiber diameters has diabetes and her father a right above knee ampu- were large ranging from 50 to 80 mm. Many of the tation for peripheral vascular disease. Neither had fibers contained single or multiple unrimmed va- neurologic complaints. However, clinical examina- cuoles that appeared empty in the modified Gomori tion of both in their 80s revealed mild distal weakness trichrome preparation (figure, B.a). There was no (MRC grade 4/5) in the upper and lower limbs with increase in endomysial connective tissue or evidence areflexia. Pinprick sensation was reduced to the mid- of inflammation, necrosis, or regeneration. There was forearm and foot in her mother. Her father had no evidence of glycogen, increased lipid, or acid phos- reduced vibration sense to the left ankle. Neurophys- phatase staining in the vacuoles (figure, B.b). The iologic testing was not possible in either. ATPase at pH 9.5 demonstrated that most of the Targeted exome sequencing of the proband’s vacuolated fibers were of type 2, and electron micros- DNA using the Agilent Focused Exome kit identified copy showed electron-dense material within vacuoles a homozygous variant (c.418C.G, p.Arg140Gly) in (figure, B.c and d).1 The overall appearances were HSPB1, which was confirmed by Sanger sequencing. those of a vacuolar myopathy without any features Both parents were heterozygous. We have reported suggesting neurogenic change, and she was diagnosed this variant previously in heterozygous form in indi- with a distal myopathy.1 viduals from 5 Indian Gujarati families with distal She re-presented at age 57 following the develop- motor neuropathy.2 ment of slowly progressive severe upper and lower limb weaknesses. On examination at age 57, there Discussion. Our patient presented at age 19 with was evidence of distal more than proximal upper clinical and biopsy features consistent with a distal limb weakness affecting wrist extension (Medical myopathy. Prominent vacuoles in type 2 fibers con- Research Council [MRC] grade 41/5), finger tained granular, electron-dense material (figure, B) extension (4 1 5), first dorsal interossei (1/5), that was interpreted to represent the product of myo- abductor pollicis brevis (3/5 right and 4/5 left), fibrillar degeneration.1 At presentation, neurophysio- and abductor digiti minimi (3/5 right and 4/5 left). logic studies did show fibrillations and polyphasic In the lower limbs, she had severe proximal weak- action potentials. However, nerve conduction studies ness (grade 2/3) with no movement at the ankles. were normal with the exception of 1 absent motor Sensory modalities were preserved except for nerve response, and overall, the clinical image was felt reduced vibration sense at the ankles. She was to represent a distal myopathy at this time. Subsequent

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure Clinical-pathologic features of patient homozygous for the HSPB1 p.Arg140Gly mutation and protein conservation between species

(A) Family pedigree: an arrow indicates the proband; half-filled indicates distal weakness in parents who were heterozygous for p.Arg140Gly mutation. (B) Biopsy of the quadriceps muscle performed at age 27; (B.a) modified Gomori trichrome staining shows variation in fiber diameter and prominent vacuoles within many muscle fibers, arrows; (B.b) Periodic acid–Schiff preparation showed no evidence of glycogen accumulation within vacuoles (arrows); (B.c) ATPase pH 9.5 demonstrates that vacuoles are predominantly in darkly stained type 2 fibers, arrows; and (B.d) ultrastructural examination of the muscle revealed electron-dense material within vacuoles (arrows). (C) Muscle MRI demonstrating severe widespread fatty infiltration of pelvic, thigh, and calf muscles with relative sparing of the adductor longus (red arrows). (D) Conservation of HSPB1 and HSPB5 (CRYAB) amino acid sequence between species. The Arg140 residue in HSPB1 is well conserved and corresponds to the position of Arg120 in the CRYAB gene.

neurophysiologic studies performed at age 57 follow- myopathy and neuronopathy have been attributed ing progressive limb weakness revealed an axonal to an HSPB1 mutation in 1 family.6 We describe motor neuropathy. Although chronic end-stage a patient with a homozygous HSPB1 mutation also myopathy may have neurophysiologic features that presenting with a distal vacuolar myopathy, motor can appear neurogenic, in this case, even the less neuropathy, and minimal sensory involvement, sup- affected proximal limb muscles failed to demonstrate porting the association of HSPB1 mutations with this any myopathic motor units or myopathic recruitment. phenotype. This expands the genetic testing indicated HSPB1 is a small heat-shock protein highly ex- in distal vacuolar myopathy. pressed in striated muscle with an important role in From the MRC Centre for Neuromuscular Diseases (E.B., A.M.R., maintaining myofibrillar structure during stress J.C.B., M.G.H., J.L.H., M.M.R., E.M.), UCL Institute of Neurol- conditions.3 ogy and National Hospital for Neurology and Neurosurgery; Depart- Mutations in HSPB1, HSPB3, and HSPB8 are ment of Molecular Neuroscience (D.S.L., A.M.P., M.G.H., H.H., 4 J.L.H.), and Division of Neuropathology (J.L.H.), UCL Institute classically associated with motor neuropathy. HSPB5 of Neurology, London; Department of Neurology (M.S.), The Royal (CRYAB) has been associated with a wide spectrum of London Hospital; and Department of Clinical Neurophysiology clinical manifestations including desmin-related myo- (J.C.B.), Norfolk and Norwich University Hospital, UK. fibrillar myopathy. The protein position of the Author contributions: Enrico Bugiardini: drafting and revising the manuscript, study concept, and interpretation of data. Alexander Arg140Gly HSPB1 mutation in our case corresponds M. Rossor: interpretation of data and revising the manuscript. David to the Arg120 HSPB5 residue mutated in this myop- S. Lynch: acquisition of data and analysis of data. Michael Swash: athy (figure, D).4 Heterozygous mutations in HSPB8 interpretation of data and revising the manuscript. Alan M. Pittman: have recently been reported causing neuropathy and acquisition of data and analysis of data. Julian C. Blake: interpretation of data and revising the manuscript. Michael G. Hanna: revising the distal myopathy with rimmed vacuoles and fibrillar manuscript and study concept. Henry Houlden: revising the manuscript aggregates in 2 families.5 Subsequently, distal and study concept. Janice L. Holton: interpretation of data and revising

2 Neurology: Genetics the manuscript. Mary M. Reilly: revising the manuscript and study receives support from the Medical Research Council (MRC), MRC concept. Emma Matthews: drafting and revising the manuscript, study Centre grant (G0601943), and the National Institutes of Neurological concept, and interpretation of data. Diseases and Stroke and office of Rare Diseases (U54NS065712); has Study funding: The research leading to these results has received served on the editorial boards of Brain, Neuromuscular Disorders; funding from the European Community’s Seventh Framework Pro- Journal of the Peripheral Nervous System;andJNNP;hasbeen gramme (FP7/2007–2013) under grant agreement no. 2012- a consultant for Servier, Acceleron, and Alnylam; and has received 305121 “Integrated European–omics research project for diagnosis research support from Wellcome Trust, UCL CBRC, Ipsen, Muscular and therapy in rare neuromuscular and neurodegenerative diseases Dystrophy Association (US), Muscular dystrophy campaign grants, (NEUROMICS).” This work is also supported by a Medical Research CLH/UCL Comprehensive Biomedical Research Centre, and CRDC. Council Centre grant and Wellcome Trust grant on Synaptopathies. E. Matthews is funded by a postdoctoral fellowship from the NIHR and Part of this work was undertaken at University College London has received research support from the Brain Research Trust and the Hospitals/University College London, which received a proportion of Muscular Dystrophy Campaign. Go to Neurology.org/ng for full disclo- funding from the Department of Health’s National Institute for sure forms. The Article Processing Charge was funded by the authors. Health Research Biomedical Research Centres’ funding scheme. This is an open access article distributed under the terms of the Disclosure: E. Bugiardini reports no disclosures. A.M. Rossor is Creative Commons Attribution License 4.0 (CC BY), which permits funded by a Wellcome Trust Postdoctoral Fellowship for Clinicians unrestricted use, distribution, and reproduction in any medium, pro- (110043/Z/15/Z). D.S. Lynch reports no disclosures. M. Swash has vided the original work is properly cited. received speaker honoraria for lectures and educational activities not Received April 10, 2017. Accepted in final form April 24, 2017. funded by industry; has served on the editorial boards of various scientific and medical journals; has held patents regarding electronic measuring devices; receives royalties from the publication of several Correspondence to Dr. Matthews: [email protected] books; has served as Chief Medical Officer for Swiss Re Life and Health; serves on the Board of Directors for MedHand AB Stockholm 1. Swash M, Schwarz MS, Thompson A, Cox E, Gray A. Distal and Malvern Arts Press; has served as an advisor to Best Doctors Inc., myopathy with focal granular degenerative change in vacuo- Europe (receives compensation); holds stock/stock options in Med- lated type 2 fibers. Clin Neuropathol 1988;7:249–253. Hand AB Stockholm; and serves as an expert witness for legal pro- 2. Rossor AM, Morrow JM, Polke JM, et al. Pilot phenotype ceedings. A.M. Pittman and J.C. Blake report no disclosures. and natural history study of hereditary neuropathies caused M.G. Hanna is supported by a Medical Research Council Centre by mutations in the HSPB1 gene. Neuromuscul Disord grant, the National Centre for Research Resources, the Myositis Support 2017;27:50–56. Group, and the National Highly Specialised Service (HSS) Depart- 3. Sugiyama Y, Suzuki A, Kishikawa M, et al. Muscle develops ment of Health UK and has been a consultant for Novartis. H. Houl- den has received research support from the Medical Research Council a specific form of small heat shock protein complex com- (MRC) UK, The BRT, The MDA USA, Muscular Dystrophy UK, posed of MKBP/HSPB2 and HSPB3 during myogenic dif- – Ataxia UK, Muscular Dystrophy UK, Rosetrees Trust, The Wellcome ferentiation. J Biol Chem 2000;275:1095 1104. Trust, and the National Institute for Health (NIHR) UCL/ 4. Benndorf R, Martin JL, Kosakovsky Pond SL, Wertheim JO. UCLH BRC. J.L. Holton has received travel funding from Neuropathy- and myopathy-associated mutations in human Merck Serono; has served on the editorial board of Neuropathol- small heat shock proteins: characteristics and evolutionary ogy Applied Neurobiology; has been an employee of University history of the mutation sites. Mutat Res Rev Mutat Res College London; and has received research support from Alz- 2014;76:15–30. heimer’s Research Trust, the Margaret Watson Memorial Trust 5. Ghaoui R, Palmio J, Brewer J, et al. Mutations in HSPB8 Grant from The Sarah Matheson Trust, Action Medical causing a new phenotype of distal myopathy and motor Research, Brain Net Europe: Support for the Queen Square – Brain Bank for Neurological Disorders, the Sarah Matheson neuropathy. Neurology 2016;86:391 398. Trust, Myositis Support Group, the Multiple System Atrophy 6. Lewis-Smith DJ, Duff J, Pyle A, et al. Novel Trust, Michael J Fox Foundation for Parkinson’sResearch, HSPB1 mutation causes both motor neuronopathy Alzheimer’s Research UK, MSA Coalition, King Baudouin Foun- and distal myopathy. Neurol Genet 2016;2:e110. dation Sophia Fund, and Medical Research Council. M.M. Reilly doi: 10.1212/NXG.0000000000000110.

Neurology: Genetics 3 Clinical/Scientific Notes

Mateja Smogavec, MD* NOVEL FUKUTIN MUTATIONS IN LIMB-GIRDLE (figure, A and D). We observed muscular atrophy with Jana Zschüntzsch, MD* MUSCULAR DYSTROPHY TYPE 2M WITH accentuation of the proximal lower extremities and Wolfram Kress, MD CHILDHOOD ONSET reduced muscle strength. Gower and Trendelenburg Julia Mohr, PhD signs were positive. Deep tendon reflexes were dimin- Peter Hellen, MD Limb-girdle muscular dystrophies (LGMDs) are ished. The patient also suffered from mild dysphagia Barbara Zoll, MD a heterogeneous group of childhood- or adult- and respiratory distress. EMG showed myopathic † Silke Pauli, MD onset inherited neuromuscular disorders, which are changes. Muscular MRI revealed a pronounced lipo- Jens Schmidt, characterized by weakness and wasting of proximal limb dystrophy of proximal upper and lower limb and limb- † MD, FEAN and axial muscles. LGMDs are divided into 8 girdle muscles (figure, B, C, E, G, and H). autosomal-dominant (LGMD1)1 and25autosomal- A multigene panel for LGMDs by next-generation Neurol Genet sequencing (NGS) on genomic DNA isolated from 2017;3:e167; doi: 10.1212/ recessive (LGMD2) forms (omim.org/phenotypicSeries/ NXG.0000000000000167 PS253600). One autosomal-recessive form is the a blood sample of our patient was performed. Coding LGMD2M (new nomenclature: Muscular Dystrophy- and flanking sequences of all relevant genes were en- Dystroglycanopathy, type C, 4, MDDGC4, MIM riched using Agilent SureSelect technology. Massively #611588), which is caused by homozygous or com- parallel sequencing was performed using the HiSeq pound heterozygous mutations in the fukutin gene 2500 Sequencing platform from Illumina. Sanger (FKTN; MIM #607440) on chromosome 9q31.2. sequencing was used for badly covered regions and The LGMD2M, as the clinical mild end of fukutin- validation of potentially pathogenic mutations. related muscle dystrophy without mental retardation, is In our patient, 2 novel missense mutations . . rare worldwide and described mostly in individuals of c.895A C; p.Ser299Arg and c.1325A G; non-Japanese descent (table e-1 at Neurology.org/ng).2–5 p.Asn442Ser of the FKTN gene (NM_001079802.1) Here, we report the second German case with an were found by NGS analysis and confirmed by Sanger LGMD2M phenotype and describe 2 previously sequencing. Both mutations disturb highly conserved unreported mutations in the FKTN gene in a German positions in the fukutin protein. Several in silico pre- female patient. diction programs (MutationTaster, PolyPhen-2, SIFT) indicated that these mutations are most likely damag- Clinical case presentation. A 32-year-old female ing. In Exome Aggregation Consortium (ExAC) and patient was admitted to our Neuromuscular Center. in Exome Server Browser, the variant c.895A.Cwas Written consent was obtained for the use of all clin- detected in heterozygous state only in one person, ical and diagnostic data. The patient was born to whereas the variant c.1325A.G had not been healthy nonconsanguineous parents. At the age of observed before. 7 years, cramping muscle pain, especially in the With only one mutation c.1325A.G present in thighs, started and was accompanied by creatine the father, a compound heterozygosity of the 2 mu- kinase (CK) elevation. A muscle biopsy record tations in our patient is strongly implied. Based on all described a chronic myopathic pattern and marked information, we predict these 2 mutations as causa- inflammation. The patient was treated with a corti- tive for the patients’ disease. costeroid, which profoundly reduced the CK level. In the following years, the patient developed a slowly Discussion. Mutations in FKTN cause a wide clinical progressive muscle weakness of both legs. At the age spectrum of myopathies: from severe congenital forms of 16 years, problems in climbing stairs and lifting of muscular dystrophy with additional extramuscular Supplemental data at heavy objects were noted. At the time of admission, symptoms to a milder form of LGMD2M, which is Neurology.org/ng the patient started to use a wheelchair for outdoor normally not associated with cognitive impairment.2 In activities. the phenotypical spectrum of FKTN mutations, the Neurologic examination revealed normal intellec- classification from congenital muscular dystrophy tual status, cranial nerve function, and sensorium. (CMD) to LGMD2M was defined according to the Inspection showed scapular winging, calf muscle onset of weakness.2 In practice, patients might display enlargement, toe walking, and lumbar hyperlordosis a phenotype well in between CMD and LGMD,2,6

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure Clinical presentation and imaging findings in the patient

Clinical presentation: atrophy of the proximal muscles in the upper extremities. Note the scapula alata (A) and the characteristic symmetric hypertrophy of gastrocnemius muscles (D). Imaging findings: MRI of the upper extremity demonstrates a moderate atrophy of the rotator cuff and the deltoid muscle (B and C). MRI of the lower extremity shows a severe symmetric atrophy with fatty degeneration of musculature of both tights (E and F). The less atrophied sartorius and gracilis muscles depict a pronounced edema (G) and a strong gadolinium enhancement representing acute disease activity (H and I). B, C, E, and F: transverse T1-weighted images, G: coronal STIR sequence; H and I: transverse fat-saturated T1-weighted images postcontrast.

pointing to an inherent difficulty of a strict classifica- Study funding: No targeted funding reported. tion. The phenotype of our patient is characterized by Disclosure: M. Smogavec reports no disclosures. J. Zschüntzsch has childhood onset, hypertrophy of calves, normal intel- received travel funding/speaker honoraria from Biotest AG and Grifols. W. Kress reports no disclosures. J. Mohr has been an ligence, and initial responsiveness to steroids. employee of CeGaT GmbH and Praxis für Humangenetik. The reduced ability of our patient to walk in her P. Hellen, B. Zoll, and S. Pauli report no disclosures. J. Schmidt teens is in line with previous reports.7 In addition, has served on scientific advisory boards for BioMarin and CSL Behring; has received travel funding/speaker honoraria from Bayer, we can provide clinical data of our patient until the Biogen, Biotest, CSL Behring, Grifols, Novartis, Octapharma, and age of 32 years. Our patient initially responded to VitalAire; has served on the editorial boards of BMC Neurology, steroid therapy, which might be explained by the American Journal of Neuroscience, Edorium Journal of profound inflammation in the muscle biopsy, com- Neurology, World Journal of Neurology, and Journal of Neu- romuscular Diseases; has been a consultant for Novartis and parable with the positive effect of steroids in CSL Behring; and has received research support from Biogen, Duchenne muscular dystrophy. CSL Behring, Novartis, Octapharma, Association francaise contre We show a rare case of LGMD2M and provide les myopathies, and Deutsche Gesellschaft für Muskelkranke. further insight into heterogeneity of phenotypes Go to Neurology.org/ng for full disclosure forms. The Article Pro- cessing Charge was funded by Institute of Human Genetics caused by mutations in FKTN and underline the Goettingen. importance of broad genetic assessment to provide This is an open access article distributed under the terms of the correct diagnosis and to facilitate individual treatment Creative Commons Attribution-NonCommercial-NoDerivatives Li- in the future. cense 4.0 (CC BY-NC-ND), which permits downloading and shar- ing the work provided it is properly cited. The work cannot be *These authors contributed equally to this work as co–first authors. changed in any way or used commercially without permission from †These authors contributed equally to this work as co–last authors. the journal. From the Institute of Human Genetics (M.S., B.Z., S.P.), Depart- Received February 8, 2017. Accepted in final form April 28, 2017. ment of Neurology (J.Z., J.S.), and Department of Neuroradiology Correspondence to Dr. Smogavec: [email protected] (P.H.), University Medical Center Göttingen; Department of goettingen.de or Prof. Schmidt: [email protected] Human Genetics (W.K.), University of Würzburg; and CeGaT GmbH und Praxis für Humangenetik (J.M.), Tübingen, Germany. 1. Nigro V, Savarese M. Genetic basis of limb-girdle mus- Author contributions: Mateja Smogavec and Jana Zschüntzsch: case cular dystrophies: the 2014 update. Acta Myol 2014;33: report study and concept and design. Wolfram Kress: analysis and 1–12. interpretation of data and critical revision of manuscript. Julia Mohr: analysis and interpretation of data. Peter Hellen: interpreta- 2. Godfrey C, Clement E, Mein R, et al. Refining genotype tion of radiologic imaging. Barbara Zoll: critical revision of the phenotype correlations in muscular dystrophies with defec- manuscript. Silke Pauli and Jens Schmidt: case report study concept tive glycosylation of dystroglycan. Brain 2007;130:2725– and critical revision of the manuscript. 2735.

2 Neurology: Genetics 3. Vuillaumier-Barrot S, Quijano-Roy S, Bouchet-Seraphin C, girdle muscular dystrophy without mental retardation. Neu- et al. Four Caucasian patients with mutations in the fukutin romuscul Disord 2009;19:352–356. gene and variable clinical phenotype. Neuromuscul Disord 6. Ceyhan-Birsoy O, Talim B, Swanson LC, et al. Whole 2009;19:182–188. exome sequencing reveals DYSF, FKTN, and ISPD muta- 4. Saredi S, Ruggieri A, Mottarelli E, et al. Fukutin gene tions in congenital muscular dystrophy without brain or eye mutations in an Italian patient with early onset muscular involvement. J Neuromuscul Dis 2015;2:87–92. dystrophy but no central nervous system involvement. Mus- 7. Riisager M, Duno M, Hansen FJ, Krag TO, Vissing CR, cle Nerve 2009;39:845–848. Vissing J. A new mutation of the fukutin gene causing late- 5. Puckett RL, Moore SA, Winder TL, et al. Further evidence onset limb girdle muscular dystrophy. Neuromuscul Disord of Fukutin mutations as a cause of childhood onset limb- 2013;23:562–567.

Neurology: Genetics 3 Clinical/Scientific Notes

Yun Tae Hwang, MBBS BRAINSTEM PHENOTYPE OF CATHEPSIN A– 6 kHz showed delayed wave V (figure, F; table e-2): Rahul Lakshmanan, RELATED ARTERIOPATHY WITH STROKES AND this was not attributable to cochlear dysfunction MBBS LEUKOENCEPHALOPATHY (given the normal wave I latency and 6–8kHztone Indran Davagnanam, detection; figure, E, tables e-1 and e-2). Furthermore, 3 MBBCh Cathepsin A–related arteriopathy with strokes and a test of spatial noise perception showed abnormal Andrew G.B. Thompson, leukoencephalopathy (CARASAL) is a recently binaural interaction (table e-3), indicating (together MBBS, PhD identified cause of adult-onset cerebral leukodystro- with the brainstem evoked responses) superior olivary David S. Lynch, MBBCh phy due to CTSA gene mutations described in 3 nuclei involvement. Peripheral vestibular assessment Henry Houlden, MBBS, Dutch and British families.1,2 The clinical phenotype with electronystagmography and caloric tests revealed PhD of CARASAL continues to be defined. Here, we subtle smooth pursuit deficits, consistent with a brain- Nin Bajaj, BMBCh, PhD report a British patient with CARASAL with brain- stem localization. Polysomnography revealed moderate Sofia H. Eriksson, MD, stem dysfunction as a leading clinical issue. periodic limb movements of sleep and (although PhD there was no dream enactment) loss of REM atonia, Doris-Eva Bamiou, suggestive of REM sleep behavior disorder. PYTHI, PhD Case description. A 48-year-old Caucasian woman Extensive investigations for metabolic and genetic 2 Jason D. Warren, MBBS, (British CARASAL case ) presented with 5 years of causes of leukodystrophy proved unrevealing until the PhD deteriorating concentration and behavioral disinhi- patient was ultimately shown to have a pathogenic bition. Recently, she had developed alternating c.973C.T, p.R325C missense mutation in the Neurol Genet right- or left-sided facial pain of fluctuating inten- CTSA gene, confirming the diagnosis of CARASAL.1 2017;3:e165; doi: 10.1212/ sity, which was ameliorated with carbamazepine. NXG.0000000000000165 The patient was found to share allele 123 at marker She also reported prominent, nonpositional vertigo, D20S838, indicating a common genetic ancestry difficulty following conversations in noisy environ- with previously reported Dutch cases.1 ments, hyperacusis, tinnitus, and hoarseness. Her sleep was disturbed by vivid nightmares and fre- Discussion. Initial descriptions of CARASAL have quent intrusive leg movements. Medical history emphasized stroke as a dominant clinical feature included migraine, hypertension, sinusitis, asthma, and relatively indolent cognitive decline, although and depression. In the family history, her father died most patients have had memory complaints at presen- at age 60 years after a stroke, and several paternal tation.1 In the series reported by Bugiani et al.,1 relatives reportedly had young onset cognitive symptoms of lower cranial nerve dysfunction (includ- decline, although no details were available. Her ing vertigo, dysphagia, dry mouth, dry eyes, central Folstein Mini-Mental State Examination score was facial paresis, or dysarthria) occurred in approximately 27/30, losing points for orientation and generation 70% of cases, and refractory hypertension was a fur- of a novel sentence, and there was bedside evidence ther clinical hallmark. MRI changes commonly of executive dysfunction and cognitive slowing, involve brainstem white matter with additional despite preserved memory and perceptual functions, involvement of the thalami and other gray matter corroborated on neuropsychometry. The general nuclei,1 as in our case (figure, A–D). The audioves- neurologic examination was unremarkable. tibular test profile here indicated a brainstem lesion, Brain MRI (figure, A–D) revealed diffuse, conflu- in addition to mild cochlear dysfunction of uncertain ent T2-weighted hyperintensity of supratentorial white provenance. Considered collectively, the available evi- matter, basal ganglia, and thalamus with extensive dence suggests a potential brainstem substrate for the Supplemental data at involvement of midbrain, pons, and medulla, includ- symptom complex of facial pain, vertigo, hearing al- Neurology.org/ng ing auditory pathways. Pure tone audiometry revealed terations, hoarseness, and sleep disorder exhibited by sensorineural hearing loss with a “cookie-bite” profile our patient and similar symptoms described in pre- most marked for midfrequencies and transient otoa- vious cases of CARASAL. The potential value of coustic emissions, consistent with mild genetic brainstem involvement in the differential diagnosis cochlear dysfunction (figure, E; table e-1 at Neurol- of CARASAL and related entities has not been pre- ogy.org/ng). Auditory evoked brainstem responses at viously emphasized.

Neurology.org/ng Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure MRI and neuro-otologic findings in the present case

Axial T2-weighted MRI sections through the brainstem (A–C) and a coronal fluid-attenuated inversion recovery MRI section through the thalami (D) are shown. Red arrows (A) indicate involvement of the lateral lemnisci; dotted white arrows (B) indicate involvement of the superior olivary nuclei; solid white arrows (C) indicate involvement of the dorsal, ventral, and inferior olivary nuclei. Pure tone audiometry plots (E) illustrate a “cookie-bite” profile of mild midfrequency hearing loss in both right (red) and left (blue) ears (threshold [dB] on y axis, abnormal .20 dB; table e-1). Auditory brainstem evoked responses (F); 3 recordings displayed for left (above) and right (below) ears showing that peaks (length of vertical latency marker indicates amplitude 0.2 mV) for wave V are consistently delayed beyond the normal range (gray oblongs) and normal latencies for waves I and III, indicating dysfunction of brainstem pathways between ventral cochlear nuclei and nucleus of the lateral lemniscus (table e-2). Note that these brainstem responses were evoked by a 6-kHz tone, which had a normal pure tone audiometric threshold for both ears (E).

At present, clinical differentiation of the cerebral Neuroscience (D.S.L., H.H.), UCL Institute of Neurology, and UCL arteriopathies exemplified by CARASAL, CARASIL, Ear Institute (D.-E.B.), University College London; Lysholm 4 Department of Neuroradiology (R.L., I.D.) and Department of Clin- and CADASIL remains challenging. Although radio- ical and Experimental Epilepsy (S.H.E.), National Hospital for logic involvement of brainstem white matter tracts is Neurology and Neurosurgery, London; and Department of Neurology frequently observed, neuro-otologic and other symp- (N.B.), Queen’s Medical Centre, Nottingham, United Kingdom. toms referable to brainstem structures are relatively Author contributions: Yun Tae Hwang: clinical assessment. Rahul 5 Lakshmanan: interpretation of MRI and preparation of the MRI uncommon in adult-onset leukodystrophies. In figure. Indran Davagnanam: interpretation of MRI and preparation CADASIL, such symptoms are generally less salient of the MRI figure. Andrew D.B. Thompson: clinical assessment. than cognitive and psychiatric decline.6 Based on the David S. Lynch: genetic analysis. Henry Houlden: genetic analysis. clinical evidence of the present case, we propose that Nin Bajaj: clinical assessment. Sofia H. Eriksson: assessment of poly- somnography. Doris-Eva Bamiou: neuro-otologic assessments and in- CARASAL should be considered in patients with vestigations. Jason D. Warren: lead clinician. All authors were adult-onset leukoencephalopathy and prominent involved in drafting and critically reviewing the manuscript. early symptoms implicating a brainstem origin. This Acknowledgment: The authors are grateful to the patient for her proposal carries the caveat that clinical experience involvement. They thank Marjo Van Der Knaap et al. at VU Uni- versity Medical Center, Amsterdam, for analysis of microsatellite with CARASAL remains limited (at present, a single markers. 1 genetic variant ). Although the true prevalence and Study funding: This work was supported by the Alzheimer’s Society mechanism remain speculative, pending further stud- (AS-PG-16-007), the NIH Research University College London ies in larger cohorts with neuropathologic correlation, Hospitals Biomedical Research Centre (CBRC 161), the Leonard brainstem features in CARASAL could reflect dual Wolfson Foundation, and the Wellcome Trust (091673/Z/10/Z). Disclosure: Y.T. Hwang has received travel funding from Guarantors effects of dystrophic white matter tracts and impaired of Brain (United Kingdom). R. Lakshmanan, I. Davagnanam, perfusion of cranial nerve nuclei due to perforant A.G.B. Thompson, and D.S. Lynch report no disclosures. H. Houlden arteriopathy: a mechanism previously proposed to has received research support from the Medical Research Council (MRC) underpin selective brainstem damage in CADASIL.7 UK, The BRT, The MDA USA, Muscular Dystrophy UK, Ataxia UK, Rosetrees Trust, The Wellcome Trust, and the National Institute for From the Dementia Research Centre (Y.T.H., J.D.W.), Department Health Research (NIHR) UCL/UCLH BRC. N. Bajaj has received of Neurodegenerative Disease (A.G.B.T.), Department of Molecular travel funding from Bial Pharma Ltd.; is an employee of NHS UK; and

2 Neurology: Genetics has received research support from the UK7T Network, the Medical 1. Bugiani M, Kevelam SH, Bakels HS, et al. Cathepsin Research Council (MRC) UK, the Michael J Fox Foundation, and A-related arteriopathy with strokes and leukoencephalop- ’ Parkinson s UK. S.H. Eriksson has served on a data safety monitoring athy (CARASAL). Neurology 2016;87:1777–1786. committee for Wellcome Trust; has received speaker honoraria from 2. Lynch SD, de Paiva ARB, Zhang WJ, et al. Clinical and Eisai Pharma and UCB Pharma; has received travel funding from genetic characterisation of leukoencephalopathies in adults. UCB Pharma; and has received research support from NIHR University Brain (in press 2017). College London Hospitals Biomedical Research Centre. D.-E. Bamiou has served on the editorial boards of Folia Phoniatrica et Logo- 3. Cameron S, Dillon H, Newall P. Development and evalu- paedica and Hearing and Balance and Communication Disor- ation of the listening in spatialized noise test. Ear Hear ders; has received research support from GNResound, EU 2006;27:30–42. EMBalance, and EU Evotion; and is an employee of University 4. Haffner C, Vinters HV. CADASIL, CARASIL, CARASAL: College London. J.D. Warren has received research support from the linguistic subtleties of cerebral small vessel disease. Neu- Wellcome Trust Senior Clinical Fellowship. Go to Neurology. rology 2016;87:1752–1753. org/ng for full disclosure forms. The Article Processing Charge 5. Ahmed RM, Murphy E, Davagnanam I, et al. A practical was funded by Wellcome Trust. approach to diagnosing adult onset leukodystrophies. This is an open access article distributed under the terms of the J Neurol Neurosurg Psychiatry 2014;85:770–781. Creative Commons Attribution License 4.0 (CC BY), which permits 6. Hardy CJ, Marshall CR, Golden HL, et al. Hearing and unrestricted use, distribution, and reproduction in any medium, pro- dementia. J Neurol 2016;263:2339–2354. vided the original work is properly cited. 7. Chabriat H, Mrissa R, Levy C, et al. Brain stem MRI Received March 3, 2017. Accepted in final form May 12, 2017. signal abnormalities in CADASIL. Stroke 1999;30: Correspondence to Prof. Warren: [email protected] 457–459.

Neurology: Genetics 3