© 2008 Nature Publishing Group http://www.nature.com/naturegenetics B R I E F C O M M U N I C AT I O N S Amyotrophic lateral sclerosis (ALS) is a severely disabling and lethal disorder caused by progressive degeneration of motor neurons in the brain, spinal cord and brainstem. ALS affects 1–3 per 100,000 people, and the average survival time is three years from disease onset. To date, no effective treatment is available1. Familial ALS accounts for up to 10% of ALS cases, with approximately 20% of familial ALS cases linked to mutations in SOD1. Mutations in ALS2, DCTN1, VAPB and ANG have been found in rare cases of familial ALS2. Sporadic ALS accounts for 490% of ALS cases and is considered to be a multifactorial disease with an estimated heritability ranging from 0.38 to 0.85 (ref. 3). Variants in several genes, including ANG4, VEGF5, HFE6 and PON1 (ref. 7), and copy number variations in SMN1 and SMN2 (ref. 8) have been reported to be associated with ALS susceptibility. However, attempts to replicate these findings in other populations have frequently failed. For instance, sequence variations in ANG are reported to be associated with ALS in Irish and Scottish populations but rarely in English, Swedish or Italian populations. Similarly, mutations in SOD1 are found in 12%– 23% of families with ALS in the United States, the UK, Germany, Sweden and Belgium, but they are rare in families with ALS in Portugal, Switzerland and The Netherlands (F.B. and P.M.A., unpublished data). Sporadic as well as familial ALS therefore seems to be a genetically heterogeneous disease, even across European populations. Genetic variation in DPP6 is associated with susceptibility to amyotrophic lateral sclerosis Michael A van Es1,15, Paul WJ van Vught1,15, Hylke M Blauw1,15, Lude Franke2,15, Christiaan GJ Saris1, Ludo Van Den Bosch3, Sonja W de Jong1, Vianney de Jong4, Frank Baas5, Ruben van’t Slot2, Robin Lemmens3, Helenius J Schelhaas6, Anna Birve7, Kristel Sleegers8,9, Christine Van Broeckhoven8,9, Jennifer C Schymick10, Bryan J Traynor11, John HJ Wokke1, Cisca Wijmenga2,12, Wim Robberecht3, Peter M Andersen7, Jan H Veldink1, Roel A Ophoff13,14 & Leonard H van den Berg1 We identified a SNP in the DPP6 gene that is consistently strongly associated with susceptibility to amyotrophic lateral sclerosis (ALS) in different populations of European ancestry, with an overall P value of 5.04
108 in 1,767 cases and 1,916 healthy controls and with an odds ratio of 1.30 (95% confidence interval (CI) of 1.18–1.43). Our finding is the first report of a genome-wide significant association with sporadic ALS and may be a target for future functional studies. Table 1 Descriptive statistics and results for SNP rs10260404 MAFa n HWEb Cases Controls Cases Controls Cases Controls P valuec OR (95% CI)d Stage I Netherlands USA Stage I combinede 461 276 737 450 271 721 0.44 0.42 0.43 0.37 0.34 0.36 0.48 0.28 0.39 0.78 0.06 0.19 0.006 0.003 4.30
105 1.30 (1.08–1.56) 1.45 (1.13–1.86) 1.34 (1.08–1.65) Stage II Netherlands Sweden Belgium Stage II combinede 272 467 291 1,030 336 439 420 1,195 0.42 0.4 0.4 0.41 0.37 0.34 0.35 0.35 0.18 0.26 0.43 0.11 0.51 0.23 0.39 0.06 0.04 0.006 0.11 0.0002 1.26 1.31 1.21 1.26 Stages I + II combinede 1,767 1,916 0.42 0.35 0.33 0.38 5.40
108 1.30 (1.18–1.43) (1.01–1.58) (1.08–1.58) (0.96–1.51) (1.11–1.42) aMinor allele frequencies for cases and controls for each population. bHardy-Weinberg equilibrium P values for cases and controls for each population. cP values were calculated for each individual population using w2 test on allele counts. dOdds ratios (OR) were calculated for the minor allele in each population; 95% confidence intervals are shown in parentheses. eP values and ORs using data from multiple populations were calculated using the Mantel-Haenszel method. 1Department of Neurology, Rudolf Magnus Institute of Neuroscience and 2Complex Genetics Section, Department of Biomedical Genetics, University Medical Center Utrecht, Utrecht 3584 CX, The Netherlands. 3Department of Neurology, University Hospital Gasthuisberg, Leuven B-3000, Belgium. 4Departments of Neurology and 5Neurogenetics, Academic Medical Center, Amsterdam 1105 AZ, The Netherlands. 6Department of Neurology, Radboud University Nijmegen Medical Centre, Nijmegen 6525 GA, The Netherlands. 7Institute of Clinical Neuroscience, Umeå University Hospital, Umeå SE-901 85, Sweden. 8Neurodegenerative Brain Diseases Group, Department of Molecular Genetics, VIB, Antwerpen B-2610, Belgium. 9University of Antwerp, Antwerpen B-2610, Belgium. 10Laboratory of Neurogenetics, National Institute of Aging, National Institutes of Health, Bethesda, Maryland 20892, USA. 11Section on Developmental Genetic Epidemiology, National Institute of Mental Health, Bethesda, Maryland 20892, USA. 12Department of Genetics, University Medical Center Groningen, University of Groningen, Groningen 9700 RB, The Netherlands. 13Department of Medical Genetics and Rudolf Magnus Institute of Neuroscience, University Medical Center Utrecht, Utrecht 3584 CX, The Netherlands. 14Neuropsychiatric Institute, University of California, Los Angeles, California 90095, USA. 15These authors contributed equally to this work. Correspondence should be addressed to L.H.vdB. (L.H.vandenBerg@umcutrecht.nl) or R.A.O. (Ophoff@ucla.edu). Received 26 June; accepted 16 October; published online 16 December 2007; doi:10.1038/ng.2007.52 NATURE GENETICS VOLUME 40 [ NUMBER 1 [ JANUARY 2008 29 B R I E F C O M M U N I C AT I O N S 153.4 Mb 153.6 Mb 153.8 Mb 154.0 Mb Figure 1 Schematic overview of DPP6. P values from the combined analysis of the two genomewide studies are shown for all SNPs in a 900-kb region surrounding rs10260404. rs7803828, located distal to rs10260404, had a lower P value in the combined analysis of the GWAs but did not fulfill the initial criteria for SNP selection (P o 0.01 in both GWAs). Subsequent analysis of seven SNPs in the associated 50-kb locus (r2 4 0.8) showed the lowest allelic P value for rs10260404 at P ¼ 5.04
108. The Bonferroni-corrected genome-wide significance level was set at P ¼ 0.05/311,946 ¼ 1.6
107. 154.2 Mb DPP6 0.00001 Allele frequency P value rs10260404 0.0001 0.001 0.01 0.05 0.1 1 –400 –300 –200 –100 0 100 200 300 Distance to rs10260404 (kb) 400 500 Netherlands, 467 cases and 437 controls from Sweden and 291 cases and 420 controls from Belgium (Supplementary Methods 10–8 Joint analysis of USA, and Supplementary Table 1 online). Overall Dutch, Belgian, Swedish Bonferroni-corrected 10–7 power for the study is shown in Supplemenand second Dutch cohort genome-wide significance tary Table 2 online. threshold 10–6 Genotyping of these 15 SNPs was done 10–5 with Taqman technology (Supplementary 10–4 Methods). We included 100 randomly 10–3 selected individuals from our Dutch genome10–2 wide association (GWA) study sample for 10–1 TaqMan genotyping of the 15 SNPs and 1 observed a concordance rate of 499.6% between platforms. Before this analysis, we examined whether population stratification was present in the available GWA data Linkage disequilibrium (D′) from the Dutch and US sample series using 89 82 99 98 75 99 Eigenstrat and did not detect any (see 76 95 99 67 94 Supplementary Methods and Supplemen76 59 82 69 tary Figure 1a,b online). 84 49 94 Because the sample series in our study were 59 51 Coloring (r 2) derived from different populations, we calcu56 lated overall P values and odds ratios (OR) 0 0.5 1 using the Mantel-Haenszel method as well as the w2 test on allele counts for all 15 SNPs. To identify previously unknown ALS susceptibility genes, we carried Only one SNP, rs10260404, showed genome-wide significance after out a genome-wide association study using Illumina 300K Beadchips Bonferroni correction for the 311,946 SNPs tested in the first stage. (Supplementary Methods online). After stringent quality control, we The overall P value for rs10260404 was 5.04
108 (corrected carried out association analysis on 311,946 SNPs in 461 affected P ¼ 0.017) with an odds ratio of 1.30 (95% CI ¼ 1.18–1.43) using individuals (cases) and 450 healthy controls matched in age, gender the w2 test and 5.40
108 with an OR of 1.30 (95% CI ¼ 1.18–1.43) and ethnicity from The Netherlands. The overall call rate was 99.5%. using the Mantel-Haenszel method10. The association for rs10260404 Results for all 311,946 SNPs are available online (http://www. was slightly more significant under a genotypic model (Cochranalscentrum.nl/index.php?id¼GWA.) We did not observe any genome- Armitage trend test), with a P value of 3.30
108 and with an wide significant association with ALS after Bonferroni correction for increased disease susceptibility for homozygote carriers of the risk allele multiple testing. (OR ¼ 1.60 with 95% CI ¼ 1.32–1.92) compared to heterozygotes Recently, first-stage data from a genome-wide association study of (OR ¼ 1.20 with 95% CI ¼ 1.06–1.41) in a dose-dependent manner. 276 ALS cases and 271 controls from the United States was released. P values and odds ratios for each individual population are shown No genome-wide significant findings were observed in this study9. in Table 1. The minor allele frequency for rs10260404 was 42% for Considering the genetic heterogeneity of ALS and the fact that both cases compared to 35% for controls. Results for all 15 SNPs analyzed studies were conducted with relatively small sample sizes, we hypothe- in stage 2 are shown in Supplementary Table 3 online. Rs10260404 maps to a 50-kb linkage disequilibrium (LD) block on sized that signals from truly associated SNPs might be present, although weak. We therefore decided to combine both datasets and chromosome 7q36 (r2 4 0.8), within a gene encoding dipeptidyl to follow up on all SNPs that had P o 0.01 in each study peptidase 6 (DPP6; Fig. 1). Combining the two GWA sets, we found independently and unidirectional allelic association (that is, associa- several SNPs within this 50-kb block that showed association with tion in the same direction of the allele associated). Fifteen SNPs disease at P o 0.01. To rule out LD beyond this 50-kb block, we fulfilled these criteria and were analyzed in three additional indepen- re-examined 130 SNPs in a 900-kb region surrounding rs10260404 dent populations consisting of 272 cases and 336 controls from The and did not find any SNP to be associated at P o 0.01 (Fig. 1). 30 39 43 8 27 90 rs rs 90 27 82 4 rs 78 03 73 3 88 15 rs 13 52 54 04 04 rs 26 10 rs rs 10 23 97 94 Allele frequency P value © 2008 Nature Publishing Group http://www.nature.com/naturegenetics Initial GWA allele frequency P value in Dutch and USA samples VOLUME 40 [ NUMBER 1 [ JANUARY 2008 NATURE GENETICS © 2008 Nature Publishing Group http://www.nature.com/naturegenetics B R I E F C O M M U N I C AT I O N S Comparison of LD structure in this 900-kb region showed similar haplotype structure in the Dutch, US and HapMap CEPH sample datasets (Supplementary Fig. 2a online). Further examination of the associated 50-kb LD block also indicated that similar LD structure is present in both the Dutch and US population (Supplementary Fig. 2b,c). It is therefore unlikely that the initial finding of ALS association is due to genetic variation outside the 50-kb LD block containing rs10260404. To fine-map the associated 50-kb LD block and carry out haplotype analyses, we additionally genotyped all SNPs (n ¼ 6) within this block that showed an association with disease at P o 0.01 in the combined analysis of both genome-wide studies. Genotyping of these six SNPs was done with Taqman technology (Supplementary Methods). Single-SNP analysis of these six additionally genotyped SNPs did not show any SNP to be associated more significantly than rs10260404 (Supplementary Table 4 online). We then applied a recently developed multimarker indirect association method that takes advantage of the correlation structure between SNPs in the HapMap sample (weighted haplotype analysis (WHAP); http://whap.cs.ucla.edu/) using rs10260404 and the additional six flanking SNPs11. Using this imputation method, we again identified the strongest association signal for rs10260404, with P ¼ 6.69
108 (Supplementary Methods and Supplementary Table 5 online). Subsequent haplotype analysis with Haploview, using the ‘solid spine of LD’ method to define haplotypes, showed the strongest association signal for a haplotype containing the CC alleles of flanking SNPs rs10239794 and rs10260404, with a P value of 3.01
109 and an allelic OR of 1.34 (95% CI ¼ 1.17–1.54; Supplementary Fig. 3 online). Results from examining long-range LD, fine mapping (including imputation analysis) and haplotype analysis all indicated that the strongest signal for association hinges on the ‘C’ allele of rs10260404, suggesting that the underlying variation for disease susceptibility is at this site. Because the entire associated 50-kb LD block containing rs10260404 is located within intron 3 of DPP6, and there are no known genes or microRNAs nearby, we consider this to be the putative ALS-associated gene (Fig. 1). DPP6 is located on chromosome 7q36 at location 153,380,839– 154,315,627 (Build 35). It consists of 26 exons and is 954 kb in size (OMIM 126141). DPP6 (also known as DPPX) encodes a dipeptidylpeptidase-like protein expressed predominantly in the brain, with very high expression in the amygdala, cingulate cortex, cerebellum and parietal lobe (http://symatlas.gnf.org/SymAtlas). This peptidase regulates the biological activity of neuropeptides by converting precursors to active forms or vice versa12. DPP6 binds specific voltagegated potassium channels and alters their expression and biophysical properties. Notably, differential DPP6 gene expression has been linked to spinal cord injury in rats13, and DPP6 was also identified as a nervous system–specific gene with accelerated evolutionary rate in the primate lineage14. In conclusion, we identify genetic variation in the DPP6 gene that is highly associated with ALS susceptibility in a combined sample of 1,767 cases and 1,916 healthy control subjects from European descent. The identified SNP, rs10260404, is located within an intron of DPP6, and no known functional variants within the gene have been yet identified. Further study will provide insight into genetic variation at NATURE GENETICS VOLUME 40 View publication stats [ NUMBER 1 [ JANUARY 2008 this locus, its potential effect on gene function and, ultimately, its role in disease susceptibility. Identification of a common variant within DPP6 is an exciting first step in the genetic study of sporadic ALS, and it opens up new avenues for studying the molecular basis of this devastating disease. Note: Supplementary information is available on the Nature Genetics website. ACKNOWLEDGMENTS We are indebted to the individuals and their families who participated in this project. This project has been generously supported by The Netherlands Organisation for Scientific Research (NWO) and the ‘‘Prinses Beatrix Fonds’’ (L.H.vdB.). We would also like to thank H. Kersten and M. Kersten for their generous support (L.H.vdB.) as well as J.R. van Dijk and the Adessium foundation (L.H.vdB.), the US National Institutes of Health grants GM68875 and MH078075 (R.A.O.), the Kempe Foundation (P.M.A.), the Swedish Brain Research Foundation and Bertil Hållsten (P.M.A.), the Björklund Foundation for ALS Research (P.M.A.), the Interuniversity Attraction Pole Programme P6/43 (Belgian Science Policy Office) (W.R., L.V.D.B. and C.V.B.) and the E. von Behring Chair for Neuromuscular and Neurodegenerative Disorders (W.R.). C.V.B., W.R. and L.V.D.B. are supported by the Fund for Scientific Research Flanders (FWO-F), and K.S. holds a postdoctoral fellowship of the FWO-F. P.M.A. and A.B. are supported by the ‘Swedish Brain Power Foundation’. This study used data from the SNP Database at the US National Institute of Neurological Disorders and Stroke Human Genetics Resource Center DNA and Cell Line Repository (http://ccr.coriell.org/ninds/). The authors thank E. Strengman, P. Sodaar, H. Veldman, H. Yigittop, W. Scheveneels, A. D’hondt, P. Tilkin and A. Nilsson for assistance with genotyping and DNA preparation. We also thank F.G. Jennekens and G. Hille Ris Lambers for helping with the DNA sample collection. AUTHOR CONTRIBUTIONS M.A.vE., P.W.J.vV., H.M.B. and L.F. contributed equally to this study. M.A.vE., P.W.J.vV., H.M.B. and R.vS. participated in the Illumina and TaqMan SNP genotyping and data analysis. M.A.vE., J.H.V. and L.F. were involved in the design of the study, handled genotype data and performed statistical analyses. M.A.vE., H.M.B., C.G.J.S., P.M.A., L.V.D.B., S.W.dJ., A.B., R.L., V.dJ., F.B., H.J.S., K.S., C.V.B., J.H.J.W., C.W. and W.R. were responsible for DNA collection and clinical characterization of affected individuals in the study. J.C.S. and B.J.T. obtained all DNA samples from the United States and performed genotyping experiments and analysis on these samples. M.A.vE. drafted the manuscript. R.A.O. and L.H.vdB. are lead investigators and contributed equally to this work. They designed and supervised the study and contributed in the writing of the manuscript. All authors participated in the critical revisions of the manuscript. Published online at http://www.nature.com/naturegenetics Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions 1. Rowland, L.P. & Shneider, N.A. N. Engl. J. Med. 344, 1688–1700 (2001). 2. Pasinelli, P. & Brown, R.H. Nat. Rev. Neurosci. 7, 710–723 (2006). 3. Graham, A.J., Macdonald, A.M. & Hawkes, C.H. J. Neurol. Neurosurg. Psychiatry 62, 562–569 (1997). 4. Greenway, M.J. et al. Nat. Genet. 38, 411–413 (2006). 5. Lambrechts, D. et al. Nat. Genet. 34, 383–394 (2003). 6. Sutedja, N.A. et al. Arch. Neurol. 64, 63–67 (2007). 7. Saeed, M. et al. Neurology 67, 771–776 (2006). 8. Veldink, J.H. et al. Neurology 65, 820–825 (2005). 9. Schymick, J.C. et al. Lancet Neurol. 6, 322–328 (2007). 10. Skol, A.D., Scott, L.J., Abecasis, G.R. & Boehnke, M. Nat. Genet. 38, 209–213 (2006). 11. Zaitlen, N., Kang, H.M., Eskin, E. & Halperin, E. Am. J. Hum. Genet. 80, 683–691 (2007). 12. Wada, K. et al. Mamm. Genome 4, 234–237 (1993). 13. Tachibana, T., Noguchi, K. & Ruda, M.A. Neurosci. Lett. 327, 133–137 (2002). 14. Dorus, S. et al. Cell 119, 1027–1040 (2004). 31