What Can We Learn About Epilepsy from Genome Sequences

Transcription

1 What Can We Learn About Epilepsy from Genome Sequences David Goldstein, Ph.D. Professor & Director Center for Human Genome Variation Duke University American Epilepsy Society Annual Meeting

2 Disclosure Biogen Idec, Knome, Roche - Scientific Advisory Boards Vertex, Shire Pharmaceuticals, SAEC Consultant Bill and Melinda Gates Foundation, Eisai Inc., Gilead Sciences Inc., Henry M Jackson Foundation, NIH, SAIC- Frederick Inc., University of Massachusetts Medical School, Translational Genomics Research Institute, UCB Grant/Research Support BioMarin Pharmaceutical Inc., Biogen Idec Inc., Regeneron Pharmaceuticals, Abbott Laboratories Honoraria Merck & Co USA- IP rights Duke University - Employee American Epilepsy Society Annual Meeting 2012

3 Learning Objectives Understanding how genetic risk factors for epilepsy are identified in next generation sequence data American Epilepsy Society Annual Meeting 2012

4 Epi4K Organizational Chart Administrative Core overall Coordination of Center s activities Sequencing, Biostatistics and Bioinformatics Core all NGS and initial identification of sequencing variants Phenotyping and Clinical Information Core - verifying and archiving the phenotypic data associated with every DNA sample Epi4K Consortium., Epi4K: gene discovery in 4,000 genomes, Epilepsia, 2012 Aug;53(8):

5 Epi4K Project 1- Epileptic Encephalopathies (EE) Infantile Spasms (IS) 1 in 3000 live births and onset between 4-12 months of life Characteristic chaotic interictal & EEG pattern of hypsarrhythmia, the sine qua non of the syndrome 50-60% of IS cases have developmental brain malformations, tuberous sclerosis complex, chromosomal syndromes and metabolic conditions Patients may evolve into LGS Onset between 1-8 years Lennox-Gastaut syndrome (LGS) Characterized by mixed seizure types and intellectual disabilities Cause unknown in about 25-35% cases, symptomatic of structural or metabolic abnormalities

6 Epi4K Project 1- Participants Contributed by EPGP The Epilepsy Phenome/Genome Project: an actively enrolling 5-year NINDSfunded study to consent 5,250 participants in families with epilepsy with detailed phenotyping, medical records, EEG, and MRI data available for all. All Project 1 samples are from the IS-LGS arm of EPGP (Elliott Sherr). DNA (Coriell Institute) Phenotyping Interviews Medical Records EEG & MRI Scientific Core Review IS/LGS 447 families enrolled IGE/LRE 1073 first degree pairs enrolled PMG/PVNH 187 families enrolled IS/LGS cryptogenic or FCD Proband with epilepsy, both biological parents enrolled with blood sample Both first degree relatives have idiopathic IGE/LRE Onset <45 PMG or bilateral PVNH Proband with epilepsy, both biological parents enrolled with blood sample No confirmed genetic syndrome

7 Epi4K Project 1 Hypothesis and Strategy Hypothesis Significant number of IS and LGS cases of unknown cause are due to dominant de novo mutations Research Strategy Conduct whole exome sequencing on 300 family trios (half reported today) that include patients with cryptogenic IS/LGS and their unaffected parents (samples from EPGP collection) Screen de novo point mutations and de novo CNVs Evaluate rare inherited variants

8 Number of IS/LGS probands Identified and confirm 180 de novo mutations in 165 trios Number Exonic of de or novo splice exonic site or splice site mutations per proband de novo mutations per trio

9 Do we see more than expected under the null? Likelihood analysis by Andrew Allen to make statistical estimates about architecture: (η)proportion of the exome sequence that can carry diseaseinfluencing mutations (η) (γ) the relative risk given the presence of at least one deleterious mutation versus having none By assuming a mutation rate and a model: Let the probability that a de novo mutation is deleterious with respect to the disease phenotype be decomposed into the product of two components: C, the probability that a de novo coding variant is deleterious with respect to the protein in which it is found, and, the probability that a de novo is found in a gene related to the disease phenotype.

10 Exome Variant Server N = 4300 EA AA (13,006 chromosomes sequenced to average median depth of 111x) n = 1.87M variants (of the missense/lgd variants, 86.9% have <0.1% frequency) HGNC (CCDS) = genes with >70% CCDS bases 10x average coverage. Tennessen et al. Science (N=2440 samples) Frequency Spectrum: most variants are very rare Participating Groups: Seattle GO - University of Washington, Seattle, WA Broad GO - Broad Institute of MIT and Harvard, Cambridge, MA WHISP GO - Ohio State University Medical Center, Columbus, OH Lung GO - University of Washington, Seattle, WA WashU GO - Washington University, St. Louis, MO Heart GO - University of Virginia Health System, Charlottesville, VA ChargeS GO - University of Texas Health Sciences Center at Houston

11 Motivation for intolerance score A central challenge in interpreting genomes is determining which genes, when mutated, elevate disease risk. Until now it has not been feasible to develop a quantitative assessment of intolerance of genes to functional variation that then relates this to the genes known to cause disease. An example score that assesses intolerance of a gene to functional variants, conditioned by frequency spectrum, measures the departure of a gene from the average number of common functional mutations found in genes with similar mutation burden: Y = Number of damaging coding variants >0.1% MAF Regressor (X) = Total number of coding variants Score (S) = Studentized residuals from regressing Y on X

12 Fn(x): Cumulative Percentage explained Mendelian disease genes are preferentially found with the intolerant group of genes To assess predictive utility, a score is plotted across possible disease backdrops from an independent disease database (OMIM). OMIM-All: All OMIM (non-cancer) Mendelian disease genes OMIM-HI: OMIM Haploinsufficiency disease genes OMIM-Rec: OMIM Recessive inheritance disease genes (exc. HI) Percentile Scores OMIM_clean OMIM_HI OMIM_Rec nonmim

13 Fn(x): Cumulative Percentage explained Known EE genes are intolerant to common functional variation Of the 14 OMIM EE genes, ARX was not sufficiently covered in EVS data to generate a tolerance score. To the remaining 13 known EE genes, KCNT1, a newly identified 1,2 but not yet OMIM listed EE gene, was added for a score comparison: 100 Phenotype OMIM HGNC Score MIM number Phenotype EIEE 5 SPTAN1 [D] 0.30% [REF] 1,2 EE NEW KCNT1 [D] 1.60% EIEE 11 SCN2A [D] 1.80% EIEE 13 SCN8A [D] 2.30% Dravet SCN1A [D] 4.00% EIEE 9 PCDH19 [D] 10.40% EIEE 12 PLCB1 [R] 11.50% EIEE 4 STXBP1 [D] 14.80% EIEE 7 KCNQ2 [D] 15.80% EIEE 2 CDKL5 [D] 15.80% EIEE 10 PNKP [R] 20.80% LGS MAPK10 [D] 23.70% EIEE 8 ARHGEF9 [R] 44.60% EIEE 3 SLC25A22 [R] 63.00% 0 Example, 10% = most intolerant ~1700 (10%) genes of 16,956 assessed genes with CCDS transcript(s). Percentile Scores 1 Barcia, G. et al. De novo gain-of-function KCNT1 channel mutations cause malignant migrating partial seizures of infancy. Nat Genet 44, , doi: /ng.2441 (2012). 2 Heron, S. E. et al. Missense mutations in the sodium-gated potassium channel gene KCNT1 cause severe autosomal dominant nocturnal frontal lobe epilepsy. Nat Genet 44, , doi: /ng.2440 (2012).

14 g de novo mutations enriched in intolerant genes 70 ) P< Epi4K ID ASD Siblings Score intolerant gene (<25%tile): 25% of lowest Score capture 54% of OMIM HI genes and 86% of known early EE genes.

15 Gene Disorder CACNA1A familial hemiplegic migraine, episodic ataxia with and without seizures (Ophoff et al., 1996, Yamazaki et al., 2011), autism (O Roark et al., 2012) CHD2 Autism (Endele et al., 2010), intellectual disability with seizures (Raunch et al., 2012) GABRB3 autism (Iossifov et al., 2012) FLNA periventricular heterotopia (Robertson et al., 2006) GRIN1 intellectual disability (Hamdan et al., 2011) GRIN2B Autism (O Roark et al., 2012), intellectual disability (Endele et al., 2010) HNRNPU multiple congenital abnormalities with seizures (Need et al., 2012) IQSEC2 intellectual disability (Raunch et al., 2012, Shoubridge et al., 2010) Infantile spasms (Morleo et al., 2008) MTOR hemimegalencephaly (somatic mutation) (Lee et al., 2012) NEDD4L photosensitive epilepsy (Dibbens et al., 2007) TNK2 implicated in infantile epilepsy (Hitomi et al., manuscript submitted) HNRNPH1 RNA binding protein Likely genes

16 TNK2 mutation causes epilepsy by blocking binding to NEDD (Chantal DePondt) (a) TNK2 716V (b) TNK2 716M EGF EGF EGFR a b Reduced expression EGFR a b NEDD TNK2 ERK NEDD TNK2 ERK Degradation Epilepsy

17 Methods Mock WT Variant pcdna3.1(+)-tnk2 pcdna3.1 (+) (empty vector) pcdna3.1(+)-tnk2 pcdna3.1 (+)-NEDD4L- 294P pcdna3.1(+)-tnk2 pcdna3.1 (+)-NEDD4L- 294R COS-7 (Lysis) TNK2 NEDD TNK2 U b NEDD TNK2 Immunoprecipitatio n Protein-G sepharose anti-tnk2 (Monoclonal) Protein-G sepharose anti-tnk2 (Monoclonal) Protein-G sepharose anti-tnk2 (Monoclonal) Western blotting anti-tnk2 (Polyclonal) Anti-NEDD4L Anti-Ubiquitin anti-tnk2 (Polyclonal) Anti-NEDD4L Anti-Ubiquitin anti-tnk2 (Polyclonal) Anti-NEDD4L Anti-Ubiquitin

18 NEDD4L 294P (WT) NEDD4L 294R (Variant) TNK2 V716M EGF EGFR a b Downregulation EGF EGFR a b EGF EGFR a b ERK ERK P ERK P NEDD TNK2 Ub Ub Ub TNK2 NEDD Ub Epilepsy NEDD Ub TNK2 Epilepsy Degradation NEDD4L mutation is dominant negative Together, NEDD4L and TNK2 implicate the EGF/ERK signalling pathway

19 Isolated unexplained genetic conditions Need et al 2012 Studying 12 families with unknown congenital disorders (different features in each case) One child with severe early onset epilepsy had a de novo mutation in HNRNPU that disrupts splicing of the gene Vandana Shashi Anna Need

20 Synonymous Mutation in HNRNPH1 results in skipping of intron 12 HNRNPH1 WT HNRNPH1 DExon12 HNRNPH1 WT STOP Q435Q (Exon) HNRNPH1 DExon12 STOP (Exon) 11 13

21 Key conclusions to date De novo mutations responsible for at least 18% of the cases studied Implications for genetic diagnostics Newly identified genes are converging on pathways RNA binding proteins HNRNPU and HNRNPHI implicated in epileptic encephalopathies for first time Dysregulation of EGF/ERK signaling implicated in encephalopathies for first time Functional analyses are a key part of interpretation (and ultimate utility)

22 Addition analysis support in SBB Andrew Allen, Ph.D. (Biostatistician), Slave Petrovski, Ph.D., Elizabeth Ruzzo Functional work in SBB Yuki Hitomi, Ph.D. NINDS Officers Administrative officer: Randall Stewart, Scientific officer: Katrina Gwinn, Liaison to advisory committee: Brandy Fureman