ITPR2 as a susceptibility gene in sporadic amyotrophic lateral sclerosis: a genome-wide association study

Articles ITPR2 as a susceptibility gene in sporadic amyotrophic lateral sclerosis: a genome-wide association study Michael A van Es, Paul W Van Vught, Hylke M Blauw, Lude Franke, Christiaan G Saris, Peter M Andersen, Ludo Van Den Bosch, Sonja W de Jong, Ruben van ‘t Slot, Anna Birve, Robin Lemmens, Vianney de Jong, Frank Baas, Helenius J Schelhaas, Kristel Sleegers, Christine Van Broeckhoven, John H J Wokke, Cisca Wijmenga, Wim Robberecht, Jan H Veldink, Roel A Ophoff, Leonard H van den Berg Summary Background Amyotrophic lateral sclerosis (ALS) is a devastating disease characterised by progressive degeneration of motor neurons in the brain and spinal cord. ALS is thought to be multifactorial, with both environmental and genetic causes. Our aim was to identify genetic variants that predispose for sporadic ALS. Methods We did a three-stage genome-wide association study in 461 patients with ALS and 450 controls from The Netherlands, using Illumina 300K single-nucleotide polymorphism (SNP) chips. The SNPs that were most strongly associated with ALS were analysed in a further 876 patients and 906 controls in independent sample series from The Netherlands, Belgium, and Sweden. We also investigated the possible pathological functions of associated genes using expression data from whole blood of patients with sporadic ALS and of control individuals who were included in the genome-wide association study. Findings A genetic variant in the inositol 1,4,5-triphosphate receptor 2 gene (ITPR2) was associated with ALS (p=0·012 after Bonferroni correction). Combined analysis of all samples (1337 patients and 1356 controls) confirmed this association (p=3·28×10–⁶, odds ratio 1·58, 95% CI 1·30–1·91). ITPR2 expression was greater in the peripheral blood of 126 ALS patients than in that of 126 healthy controls (p=0·00016). Interpretation Genetic variation in ITPR2 is a susceptibility factor for ALS. ITPR2 is a strong candidate susceptibility gene for ALS because it is involved in glutamate-mediated neurotransmission, is one of the main regulators of intracellular calcium concentrations, and has an important role in apoptosis. Introduction Amyotrophic lateral sclerosis (ALS), a form of motor neuron disease, is a neurodegenerative disorder characterised by progressive wasting and weakness of limb, bulbar, and respiratory muscles. The disease is caused by loss of motor neurons in the spinal cord, brainstem, and motor cortex and can occur at any time in adulthood, with a median age of onset in the midfifties. About half of patients die within 3 years of symptom onset, usually because of respiratory failure.1–3 The only therapeutic strategy to slow progression of ALS is currently riluzole, which delays disease development by 3–6 months.4 Almost one tenth of cases of ALS are familial. 12–23% of familial ALS is linked to mutations in the superoxide dismutase 1 gene (SOD1),5 and rare cases are associated with mutations in the genes that encode alsin, dynactin, VAMP-associated protein B and C (VAPB), and angiogenin ribonuclease RNase A family 5 (ANG).6,7 The remaining 90% or more of ALS cases are sporadic, and are thought to be multifactorial, with both environmental and genetic components.8–10 On the basis of concordance rates in twin studies, estimates of the heritability of ALS range from 0·38 to 0·85.11 The cause of motor neuron degeneration in sporadic ALS is unknown, but the many possible mechanisms include oxidative stress, glutamate-mediated excitotoxicity, and apoptosis.1,12,13 Several candidate-gene http://neurology.thelancet.com Vol 6 October 2007 studies have investigated these possibilities, and have reported associations with variants of genes that include ANG, vascular endothelial growth factor (VEGF), hemochromatosis (HFE), and paraoxonase 1 (PON1), and with variations in copy number for the genes that encode survival of motor neuron proteins 1 and 2 (SMN1 and SMN2).6,14–20 In the candidate approach, a gene is selected on the basis of its function and is subsequently tested for association with a disease. An alternative approach, made possible by progress such as completion of the International Haplotype Map (HapMap) project and the development of high-throughput, high-density genotyping technology, is that of genome-wide association. In this approach, nearly all common variation in the genome can be screened for association with disease in an unbiased way,21,22 in contrast to the candidate-gene approach, which is based on pathophysiological hypotheses. Indeed, genome-wide association studies have successfully identified genetic risk factors for diseases that include age-related macular degeneration, diabetes mellitus types 1 and 2, coeliac disease, and breast cancer.23–26 Most variation in human DNA is caused by single nucleotide polymorphisms (SNPs). More than 10 million SNPs have been identified and these common variants are thought to contribute to disease susceptibility (the common disease, common variant Lancet Neurol 2007; 6: 869–877 Published Online September 7, 2007 DOI:10.1016/S14744422(07)70222-3 Department of Neurology, Rudolf Magnus Institute of Neuroscience (M A van Es MD, P W Van Vught MSc, H M Blauw MD, C G Saris MD, S W de Jong MD, J H J Wokke MD, J H Veldink MD, L H van den Berg MD), Complex Genetics Section, Department of Biomedical Genetics (L Franke MSc, R van ‘t Slot BSc, C Wijmenga PhD), and Department of Medical Genetics and Rudolf Magnus Institue of Neuroscience (R A Ophoff PhD), University Medical Center Utrecht, Utrecht, The Netherlands; Institute of Clinical Neuroscience, Umeå University Hospital, Umeå, Sweden (P M Andersen MD, A Birve PhD); Department of Neurology, University Hospital Gasthuisberg, Leuven, Belgium (R Lemmens MD, L Van Den Bosch PhD, W Robberecht MD); Department of Neurogenetics (F Baas PhD) and Department of Neurology (V de Jong MD), Academic Medical Center, Amsterdam, The Netherlands; Department of Neurology, Radboud University Medical Center, Nijmegen, The Netherlands (H J Schelhaas MD); Neurodegenerative Brain Diseases Group, Department of Molecular Genetics, VIB and University of Antwerp, Antwerpen, Belgium (K Sleegers MD, C Van Broeckhoven PhD); Department of Genetics, University Medical Center Groningen, The Netherlands (C Wijmenga PhD); Neuropsychiatric Institute, University of California, Los Angeles, CA, USA (R A Ophoff) 869 Articles Correspondence to: Leonard H van den Berg, University Medical Centre Utrecht, 3508 GA Utrecht, The Netherlands lberg@umcutrecht.nl For more on the HapMap project, see http://www. hapmap.org/ hypothesis). The HapMap project has shown that genetic variants that are near each other are often inherited together, meaning that SNPs are inherited in groups or haplotypes. Tag SNPs can be used to identify these haplotypes, and roughly 300 000 tag SNPs contain most of the information about patterns of genetic variation for all 10 million common SNPs. High-density, high-throughput genotyping technology has made genome-wide association studies possible by enabling researchers to test all tag SNPs for association with a disease in a single experiment.27 To identify genetic factors for sporadic ALS, we did a genome-wide association study in patients and healthy controls from The Netherlands, and we replicated our most significant findings in two independent sample series. We also investigated the possible pathological functions of associated genes using expression data from peripheral whole blood in patients with sporadic ALS and in control individuals who were included in the genome-wide association study. Such studies have previously shown that changes in gene expression in peripheral blood can be associated with diseases that do not have obvious blood phenotypes, including cerebral infarction, Huntington’s disease, Alzheimer’s disease, and Parkinson’s disease.28–31 Methods Participants We analysed populations from The Netherlands, Belgium, and Sweden. All samples from Belgian and Swedish patients were screened for SOD1 mutations, and the Swedish samples were also screened for ANG mutations. No samples with mutations in these genes were included in this study. Because no SOD1 mutation has been reported in patients with sporadic ALS or familial ALS in The Netherlands (Baas F and Andersen PM, unpublished), no SOD1 screening was done in these samples. Patients were included in the study if they fulfilled the criteria for probable or definite sporadic ALS. Diagnosis was according to the 1994 El Escorial criteria32 and was by neurologists who specialise in neuromuscular disease, and particularly in ALS. Detailed family histories were obtained for all patients, and those with a family history of ALS were excluded. Control individuals had no medical or family history of neurological disease, and were matched to patients for age, sex, and ethnic origin. Participants were excluded from the study if they had a first-degree relative in the study or if genotyping revealed 200 000 or more concordant SNPs. All participants were confirmed to have ancestry in the country of study: all four grandparents born in The Netherlands for the Dutch groups, reported Flemish descent for at least three generations for the Belgian group, and reported northern Swedish citizenship for at least three generations for the Swedish group. 870 The patients with sporadic ALS who were included in the Dutch genome-wide association study had been referred to the University Medical Centre (UMC) Utrecht, the Academic Medical Centre Amsterdam, or the University Medical Centre Nijmegen, St Radboud; the control individuals for this group were unrelated, healthy volunteers who were spouses of patients with sporadic ALS or who accompanied non-ALS patients to the UMC Utrecht neurology outpatient clinic. An independent group was recruited from an ongoing, prospective population-based study on ALS in The Netherlands. In this study, a capture-recapture design is used to identify all prevalent and incident cases in The Netherlands.33,34 Family practitioners are then asked to recruit controls matched for age, sex, and ethnic group from their patient registers. All patients in the Belgian group were referred to the University Hospital Gasthuisberg, Leuven, and the Belgian control group consisted of unrelated, healthy, Flemish people selected from individuals who had married into families that were participating in other genetic studies of neurological diseases. The Swedish cohort consisted of patients who were referred to the Umeå University ALS Clinic, and the control volunteers were spouses of patients with ALS or spouses of patients with other neurological disorders. All patients gave written informed consent, and approval was obtained from the local ethical committees for medical research. The approved protocols do not permit public release of individual genotyping data; however, we are currently seeking permission from all participants for future public release, and genome-wide data are available on request from the corresponding author. Procedures We used a three-stage genotyping design (figure 1) that reduces both the number of tests that are necessary and the chances of false-positive associations.35 In all stages, all samples were genotyped individually. In the first stage, more than 300 000 SNPs were tested for association with ALS in the Dutch genome-wide association cohort. In the second stage, the 500 most significantly associated SNPs were selected for further analysis in independent Dutch and Belgian populations. In the third stage, all SNPs with p<0·1 in stage two were genotyped in the Swedish population; the stringency condition was set at p<0·1 to allow for the genetic heterogeneity of the disease. For example, SOD1 is mutated in about a quarter of families with ALS in the USA, the UK, Germany, Sweden, and Belgium, but is mutated rarely in families with ALS in Portugal, Switzerland, and The Netherlands.36 The frequency of mutations in ANG, VAPB, and VEGF in patients with ALS is similarly variable.36 Venous blood samples were taken and DNA was extracted according to standard procedures. All firsthttp://neurology.thelancet.com Vol 6 October 2007 Articles stage genotyping was done with Illumina Infinium II HumanHap 300K SNP chips (Illumina, San Diego, CA, USA), and all second-stage genotyping was done with Illumina GoldenGate, which is a highly multiplexed PCR-based genotyping assay. All first-stage and secondstage genotyping was done at the UMC Utrecht and according to the manufacturer’s protocols. Stage-three genotyping was done with standard allelic discrimination assays from Applied Biosystems. PCR mixes consisted of 10 ng genomic DNA, 2·5 µL Taqman master mix, and 0·01 µL assay mix (both from Applied Biosystems, Foster City, CA, USA), and double-distilled H2O in a 5 µL reaction volume in 384-well plates. PCR conditions were denaturation at 95°C for 10 min, 40 cycles of denaturation at 92°C for 15 s, and annealing and extension at 60°C for 1 min. Allelic PCR products were analysed on the ABI Prism 7900HT Sequence Detection System using SDS 2.3 software (Applied Biosystems, Foster City, CA, USA). We randomly selected 100 individuals from the first two stages and genotyped them for all SNPs in stage three using the Taqman assays to ensure that the methods used in the different stages generated the same genotypes for each individual (concordance rate >99·5%). To investigate further the pathological function of genes identified as associated with ALS, we collected DNA and RNA from 126 patients with sporadic ALS and 126 age-matched and gender-matched healthy controls from the Dutch genome-wide association study. All blood samples were taken during the first visit to the neurology outpatient clinic in the UMC Utrecht. Samples were taken between 1000 h and 1200 h and before riluzole treatment was initiated. Blood for RNA isolation was drawn into PAXgene tubes and RNA was isolated with the PAXgene Blood RNA kit (Qiagen, Valencia, CA, USA). Isolation was done according to the manufacturer’s instructions, with inclusion of an optional DNase digestion step. Total RNA was quantified by spectrophotometry and the quality of total RNA was checked with an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). Concentrations of inositol 1,4,5-trisphosphate receptor type 2 (ITPR2) mRNA were measured using Illumina Sentrix humanref8 arrays. 100 ng total RNA was used for first-strand and second-strand cDNA synthesis with a MessageAmp Kit (Ambion, Foster City, CA, USA), and 24k Illumina Sentrix HumanRef-8 Expression BeadChips were hybridised and scanned according to the manufacturer’s instructions. Raw images were converted into data files with the Gene Expression Module of BeadStudio software package (Illumina, San Diego, CA, USA). The rank invariant algorithm was used for normalisation. Statistical analysis Our study provided 80% power to detect an allelic association with an odds ratio of at least 1·68, on the http://neurology.thelancet.com Vol 6 October 2007 Stage 1 (genome-wide association study) 311 946 SNPs Stage 2 500 SNPs Most strongly associated SNPs The Netherlands 461 patients 450 controls The Netherlands 272 patients 336 controls SNPs with p<0·1 Belgium 291 patients 267 controls Stage 3 17 SNPs Sweden 313 patients 303 controls Figure 1: Study design basis of a mean minor allele frequency of 0·26 and the assumption that the causal variant is typed or efficiently tagged. For each SNP, we used the PLINK data analysis toolset version 0.99r to do basic allelic χ² tests with one degree of freedom.37 Odds ratios with 95% CI were calculated for the minor allele of each SNP. All SNPs were tested in control samples for deviations from Hardy-Weinberg equilibrium, and we used Eigenstrat software (version 1.01) to detect evidence of population stratification.38 A Shapiro-Wilk W test of normality with SPSS software (version 14.0) showed that expression did not follow a normal distribution (p<0·001), so to compare ITPR2 expression between patients and control participants we used a Wilcoxon-Mann-Whitney U test. Role of the funding source The study sponsors had no role in the study design, data collection, data analysis, data interpretation, or writing of this report. The corresponding author had full access to all data in this study and had the final responsibility for the decision to submit for publication. Results Figure 1 shows the numbers of people and SNPs tested after exclusions at each genotyping stage, and table 1 shows characteristics of the study populations. In the genome-wide study, we genotyped samples from 477 patients with sporadic ALS and 472 control volunteers. We subsequently excluded 12 patients and 22 controls because the genotyping was of a poor quality (proportion of all genotypes successfully identified [call rate]<95%), and two pairs of patients were excluded because they were related (>200 000 concordant SNPs). Two samples were genotyped twice, and this gave concordance of over 99·9% for each sample. The mean call rate across all samples was 99·5%; the call rate was at least 99% for 29 8807 SNPs and at least 95% for 315 293 SNPs. A total of 284 182 806 unique genotype calls were made. Probability of deviation from the Hardy-Weinberg equilibrium was 0·05 or less for 14 498 SNPs and 0·01 or less for 3207 SNPs in control samples. The mean minor allele frequency was 0·26, and 396 SNPs had a minor allele frequency of less than 0·01. 871 Articles Total Male Female Spinal onset Bulbar onset Age at onset (years) Survival (months) Stage 1 (genome-wide study), The Netherlands Patients with sporadic ALS 461 272 (59%) 189 (41%) 318 (69%) 143 (31%) 59 (20–86) 33 (5–142) Controls 450 266 (59%) 184 (41%) ·· ·· 60 (22–87) ·· Stage 2, The Netherlands Patients with sporadic ALS 272 155 (57%) 117 (43%) 193 (71%) 79 (29%) 58 (16–83) 43 (4–196) Controls 336 192 (57%) 144 (43%) ·· ·· 59 (29–95) ·· Stage 2, Belgium Patients with sporadic ALS 291 172 (59%) 119 (41%) 212 (73%) 79 (27%) 59 (18–86) 39 (5–177) Controls 267 155 (58%) 112 (42%) ·· ·· 51 (18–92) ·· Stage 3, Sweden Patients with sporadic ALS 313 178 (57%) 135 (43%) 207 (66%) 106 (34%) 60 (20–89) 31 (9–108) Controls 303 158 (52%) 145 (48%) ·· ·· 62 (25–94) ·· Total Patients with sporadic ALS 1337 776 (58%) 562 (42%) 936 (70%) 401 (30%) 59 (16–89) 36 (4–196) Controls 1356 773 (57%) 583 (43%) ·· ·· 58 (18–95) ·· Data are number of individuals, with percentages in parentheses, and median ages and survival times, with ranges in parentheses. Age at onset is the age at first manifestation of weakness for patients and the age at inclusion for controls. Survival is time from onset of initial weakness. Table 1: Characteristics of study populations Chr Chr position MAF Controls rs2306677 rs30264 HWE Patients Allelic p value Controls Patients 12 26 527 653 0·05 0·10 0·36 0·56 0·0007 5 129 514 804 0·10 0·13 0·51 0·85 0·03 rs6559732 9 85 358 512 0·13 0·16 0·85 0·11 0·07 rs9257425 6 29 044 535 0·26 0·29 0·25 0·92 0·08 rs1565730 12 65 143 477 0·26 0·29 1·00 0·54 0·14 rs3109032 3 113 656 323 0·20 0·24 0·32 0·82 0·34 rs721363 3 113 659 820 0·20 0·24 0·38 0·82 0·37 rs3026935 1 157 384 314 0·08 0·06 0·76 0·71 0·44 rs4620270 8 125 506 992 0·13 0·16 0·28 0·76 0·56 rs2837501 21 40 499 803 0·45 0·50 0·56 0·67 0·58 rs901709 23 25 800 777 0·26 0·22 0·33 0·06 0·63 rs280199 2 121 449 946 0·29 0·24 0·84 0·65 0·79 rs4551564 1 222 304 339 0·42 0·37 0·50 0·28 0·83 rs543721 2 136 878 027 0·41 0·37 0·31 0·53 0·85 rs1007241 13 89 136 899 0·37 0·41 0·19 0·22 0·88 rs1929492 9 103 206 143 0·06 0·10 0·16 1·00 0·95 rs6948572 7 137 312 129 0·20 0·23 0·44 0·55 0·98 Chr=chromosome. MAF=minor allele frequency. HWE=Hardy-Weinberg equilibrium. Table 2: Frequencies and p values of the 17 SNPs tested for association in stage 3 For the full SNP results, see http://www.alscentrum.nl/index. php?id=GWA 872 Allelic χ² tests with one degree of freedom were done on all 311 946 SNPs that had a minor allele frequency of 0·01 or more, a call rate of 95% or more, missingness of 0·05 or less, and probability of deviation from the Hardy-Weinberg equilibrium of 0·01 or more. Results for all SNPs in stage one are available on the corresponding author’s departmental website. Table 1 and figure 1 show numbers of patients included in the second-stage genotyping after exclusions for poor genotyping and family relationships. Of the 500 selected SNPs, 485 were included for the second-stage analysis on the basis of their call rate (>95%) and minor allele frequency (>0·01). Mean call rate was 98·2% and average minor allele frequency was 0.24. Eigenstrat analysis between Dutch and Belgian controls showed no evidence of population stratification (differences in allele frequencies due to systematic differences in ancestry), so we analysed the Dutch and Belgian http://neurology.thelancet.com Vol 6 October 2007 Articles populations as one group. The webtable shows results for all SNPs in stage two. 17 SNPs had p<0·1 with the same direction of allele frequency (ie, the minor allele was more prevalent in patients in both the initial screen and replication stage, or the minor allele was more prevalent in controls in both stages), and were selected for genotyping in the population of patients and controls from Sweden. Table 2 shows these stage-three results. This analysis revealed an association with sporadic ALS for one SNP, rs2306677 (p=0·0007, and p=0·012 after Bonferroni correction for testing of 17 SNPs). Because low minor allele frequencies can bias genetic association studies, we recalculated the statistics for rs2306677 using Fisher’s Exact Test, which confirmed the initial results. Table 3 shows odds ratios and p values for the four individual populations, and for the combined sample from all three stages of genotyping. Rs2306677 lies in a gene on chromosome 12p11 that encodes ITPR2. Because the association signal maps to a 45 kb linkage disequilibrium block (r²>0·6) within ITPR2 (figure 2), this gene seems to be the ALS-associated gene. Expression of ITPR2 was greater in peripheral blood of patients than in that of controls (figure 3). Unfortunately, the correlation between genotype and expression could not be analysed, because RNA was not available from a sufficient number of patients and controls who were homozygous for the risk allele. Discussion In our three-stage genome-wide association study, rs2306677 was associated with ALS after correction for multiple testing (table 3). Re-evaluation of stage-two results for this SNP by population showed a consistent pattern of association with ALS in the two independent Dutch series and the Swedish series, but contrasting results for the Belgian population (table 3). This difference might be due to the relatively small size of the sample (which could have caused inaccurate estimates of the allele frequencies), type 1 error, population-specific environmental factors, and the genetic heterogeneity of ALS across Europe.36 For instance, sequence variations in ANG are associated with ALS in Ireland and Scotland, MAF but not in English, Swedish, or Italian populations.6 Association with ALS also seems to differ between Dutch and Belgian populations for SOD1 (Baas F and Andersen PM, unpublished) and the promoter region of VEGF.39 First-stage data from a genome-wide association study of 271 patients with ALS and 276 controls from the USA40 reported no genome-wide significant findings, probably owing to the small sample size. The study had 80% power to detect an odds ratio of 2·75, and therefore failed to detect an association for rs2306677, even though subsequent analysis of rs2306677 in this population40 revealed an association with ALS of p=0·14. Addition of this SNP data to our study would provide further evidence for an association between ALS and rs2306677 (p=1·48×10–⁶, compared with p=3·28×10–⁶ without the US data; minor allele frequencies in the combined population are 0·12 and 0·09 for patients and controls, respectively). Another genome-wide study of ALS41 used a design that was similar to ours: many SNPs were screened in the initial phase and fewer candidate SNPs were assessed in later phases, to decrease the burden of multiple testing. In that study, the 384 SNPs most strongly associated with ALS, and not all SNPs that had associations of p<0·05, were selected for further analysis. The results for all SNPs, and thus data on SNPs in ITPR2, are not currently publicly available. Rs2306677 lies in intron 42 of ITPR2. Because the association signal maps to a 45 kb linkage disequilibrium block that contains five exons within the gene, ITPR2 seems to be the ALS-associated gene (figure 2). ITPR1, ITPR2, and ITPR3 form the inositol 1,4,5-trisphosphate receptor family. ITPR2 is a calcium channel on the endoplasmic reticulum that is primarily responsible for controlling intracellular calcium concentrations in neurons (figure 4).42 After stimulation of glutamate receptors, inositol 1,4,5-trisphosphate is released and then binds to ITPR2, which is highly expressed in motor neurons.43 Altered function of ITPR2 can increase vulnerability to high intracellular calcium concentrations, which can cause apoptosis and selective degeneration of motor neurons, the hallmark of ALS.1 HWE Allelic p value Fisher’s exact p value Odds ratio (95% CI) 0·80 0·0005 0·0005 1·81 (1·29–2·55) 0·47 0·4130 0·4159 0·85 (0·57–1·26) 0·57 0·0017 0·0015 1·85 (1·25–2·73) Controls Patients Controls Patients The Netherlands (stage 1) 0·06 0·11 0·40 Belgium (stage 2) 0·10 0·09 0·50 The Netherlands (stage 2) 0·07 0·12 0·67 Sweden (stage 3) 0·05 0·10 0·36 0·56 0·0007 0·0009 2·12 (1·35–3·33) The Netherlands pooled 0·06 0·11 0·51 0·62 3·73×10–6 3·5×10–6 1·81 (1·41–2·35) Overall pooled 0·07 0·11 0·47 0·61 3·28x10–6 3·76x10–6 1·58 (1·30–1·91) See Online for Webtable MAF=minor allele frequency. HWE=Hardy-Weinberg equilibrium. P values were calculated using both the basic allele test and Fisher’s exact test. Table 3: Statistics for rs2306677 per population http://neurology.thelancet.com Vol 6 October 2007 873 Articles 0·0001 Allele frequency (p value) 0·001 0·01 0·1 1·0 26 500 000 26 550 000 26 600 000 43 52 40 58 70 69 33 85 8 2 30 46 93 86 78 95 81 96 53 13 73 75 22 51 38 75 91 99 73 8 70 40 99 98 55 75 22 37 90 79 2 2 rs11048593 rs11048586 99 28 83 98 24 61 81 84 62 rs1001452 98 34 99 13 70 73 79 rs3782294 rs11048567 rs7975326 rs7303990 rs10842745 rs2220168 rs1393403 44 19 17 99 58 18 40 39 34 77 81 78 46 9 99 86 86 58 rs905298 rs4964002 rs1825478 rs4964001 rs2306677 97 97 67 4 31 84 32 64 32 30 9 31 21 29 89 90 80 36 17 0 55 15 50 35 87 4 43 31 37 49 37 93 7 9 59 67 83 14 85 75 46 47 29 8 86 49 26 37 83 57 9 61 26 64 75 11 56 80 72 15 4 42 55 82 64 7 85 61 33 13 18 71 72 39 8 17 85 59 49 64 r² 36 85 50 17 31 0·00–0·17 57 58 4 4 37 35 22 80 67 4 18 0·17–0·33 3 62 62 45 66 5 0·33–0·50 5 3 28 89 38 47 0·50–0·67 42 18 65 61 55 0·67–0·83 9 59 33 1 0·83–1·00 31 20 75 10 6 16 72 0 3 rs11048545 40 79 87 16 22 Linkage disequilibrium pattern 74 98 rs7309048 85 27 rs7975290 rs2344158 rs1532720 5 31 rs1385342 81 rs1352388 rs10505999 rs10743585 rs12366669 Exons 7 82 65 87 13 86 83 81 83 63 16 31 64 67 74 4 29 4 37 62 72 93 65 43 86 17 94 59 7 73 91 Figure 2: Association results, marker density, and linkage disequilibrium structure for the locus that contains rs2306677 within ITPR2 in the Dutch genomewide association population Top: allelic p values for each SNP. Bottom: linkage disequilibrium pattern. Linkage disequilibrium is given as r2 between SNPs in a 150 kb region that surrounds rs2306677 (r2 is on a scale from 0 to 1; by convention, the zero before the decimal point is not shown in such plots). r2>0·8 indicates that the two SNPs are inherited together more often than can be accounted for by chance, and suggests that the alleles are close together on the DNA strand. Lower values indicate that inheritance is independent and that the variants are probably not in close proximity. For example, r2=0·99 between rs2306677 and rs905298; these SNPs also have similar p values, which is expected because they tend to be inherited together. Use of r2>0·8 reveals a 45 kb linkage disequilibrium block that contains rs2306677 and that spans from rs10505999 in intron 39 to rs905298 in intron 45 within ITPR2. Thus, the associated variant lies within this block and is rs2306677 or is a variant close to rs2306677. 874 http://neurology.thelancet.com Vol 6 October 2007 Articles Contributors RAO and LHvdB are lead investigators and contributed equally to this work. MvE, PWJvV, HMB, and LF also contributed equally. RAO and http://neurology.thelancet.com Vol 6 October 2007 p=0·00016 4 3 Normalised expression Calcium release from the endoplasmic reticulum by ITPR2 seems to be crucial to both the extrinsic death receptor pathway and the intrinsic apoptosis pathway that involves mitochondria and cytochrome c.44–47 Apoptosis, which is thought to be the final common pathway in motor neuron degeneration in ALS,12 is impaired in cells that are genetically altered to have no ITPR2; by contrast, cells that overexpress ITPR2 show increased cell death.48 Our finding of greater ITPR2 mRNA expression in the peripheral blood of patients with ALS than in that of controls suggests that some cell types might have increased susceptibility to apoptosis in ALS.44 Cells that have limited calcium-buffering capacity, such as motor neurons, might be particularly severely affected.13 Genome-wide studies are increasingly recognised as powerful, but their high sensitivity can lead to falsepositive identification. Therefore, replication of associations in well designed, adequately powered studies is essential. However, methods used for correcting for multiple testing, such as Bonferroni, tend to cause type 2 errors. Although the Bonferroni method is appropriate for epidemiological studies that investigate a few independent variables, disease-associated SNPs are unlikely to act independently of each other, and this method will probably overcorrect.8 This point is demonstrated by associations with ALS that have been replicated in many populations, such as those for HFE and PON, but that have not been detected by the genome-wide association studies. Owing to the size of the population analysed in stage one and the small number of SNPs analysed at later stages, this study is not a completely comprehensive screen for genetic susceptibility factors for ALS. Because this study lacked adequate power to detect odds ratio smaller than 1·68, larger or collaborative genome-wide efforts will probably reveal further genetic susceptibility factors for ALS. Nonetheless, this study is to our knowledge the largest genome-wide study of ALS to have been published, and it has identified a novel and plausible susceptibility gene for sporadic ALS pathogenesis. On the basis of known functions of ITPR2, involvement of this gene in the pathogenesis of ALS is plausible because: ITPR2 expression is higher in motor neurons than in other neurons;43 ITPR2 has a role in glutamatemediated neurotransmission;42 ITPR2 is one of the main regulators of intracellular calcium concentrations in neurons;49 overexpression of ITPR2 is essential for apoptosis;44–47 and patients with ALS overexpress ITPR2. Identification of a common variant within ITPR2 is one of the first of many steps in the genetic study of sporadic ALS, and it opens up new avenues for study of the molecular basis of this devastating disease. 2 1 0 ALS Control Figure 3: Normalised expression of ITPR2 in 126 ALS patients and 126 healthy matched controls The box shows the IQR, and the horizontal line within this box is the median. Horizontal bars outside the box show the range around the mean that contains 95% data, and outliers are shown as circles. Presynaptic neuron Glutamate NMDAR AMPAR mGLuR Gq PLC [Ca2+]i Apoptosis IP3 ITPR2 Ca2+ ER Postsynaptic neuron Figure 4: Pathway by which ITPR2 might contribute to cell death Released glutamate binds to metabotropic glutamate receptors (mGluR), NMDA-type receptors (NMDAR), and AMPA-type receptors (AMPAR). After glutamate binds to mGluR, a second messenger system is activated in which G-protein q subunits and phospholipase C (PLC) cause levels of inositol 1,4,5trisphosphate (IP3) to rise. Inositol 1,4,5-trisphosphate then binds to the inositol 1,4,5-trisphosphate receptor type 2 (ITPR2), which releases calcium from the endoplasmic reticulum (ER). Raised ITPR2 expression can therefore increase intracellular concentrations of calcium ([Ca2+]i). Increases in [Ca2+]i caused in this way, or by calcium influx through NMDAR or AMPAR, can lead to apoptosis. 875 Articles LHvdB designed and supervised the study, and MvE, JHV, and LF also involved in the study design. MvE, PWJvV, HMB, and RvtS participated in the laboratory-based genotyping and data analysis. LF did the data handling. CGS generated all expression data and preformed all statistical analysis of these data. MvE, JHV, and LF did statistical analysis. MvE, HMB, CGS, PMA, LVDB, SWdJ, AB, RL, VdJ, FB, HJS, KS, CVB, JHW, CW, and WR were responsible for DNA collection and characterisation of patients. MvE drafted the manuscript. All authors participated in revision of the manuscript and have seen and approved the final version. 15 16 17 18 Conflicts of interest We have no conflicts of interest. Acknowledgments We thank the patients and their families for their participation. This project has been generously supported by The Netherlands Organization for Scientific Research (NWO) (LHvdB, FB) and the Prinses Beatrix Fonds (LHvdB), the US National Institutes of Health grants GM68875 and MH078075 (RAO), the Kempe Foundation (PMA), the Swedish Brain Research Foundation (PMA), the Björklund Foundation for ALS Research (PMA), Interuniversity Attraction Pole Programme VI (Belgian Science Policy Office), and the E von Behring Chair for Neuromuscular and Neurodegenerative Disorders. AB was supported by the Swedish Brain Power Foundation. CVB is supported by the Fund for Scientific Research Flanders (FWO-F), and KS holds a postdoctoral fellowship of the FWO-F. We thank Eric Strengman, Peter Sodaar, Henk Veldman, Haci-Ali Yigittop, Wendy Scheveneels, Ann D’Hondt, Petra Tilkin, and Ann-Charloth Nilsson for assistance with genotyping and DNA preparation. We also thank FG Jennekens and G Hille Ris Lambers for helping with the DNA sample collection. References 1 Cleveland DW, Rothstein JD. From Charcot to Lou Gehrig: deciphering selective motor neuron death in ALS. Nat Rev Neurosci 2001; 2: 806–19. 2 Mitchell JD, Borasio GD. Amyotrophic lateral sclerosis. Lancet 2007; 369: 2031–41. 3 Rowland LP, Shneider NA. Amyotrophic lateral sclerosis. N Engl J Med 2001; 344: 1688–700. 4 Lacomblez L, Bensimon G, Leigh PN, Guillet P, Meininger V. Dose-ranging study of riluzole in amyotrophic lateral sclerosis. Amyotrophic Lateral Sclerosis/Riluzole Study Group II. Lancet 1996; 347: 1425–31. 5 Rosen DR, Siddique T, Patterson D, et al. Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature 1993; 362: 59–62. 6 Greenway MJ, Andersen PM, Russ C, et al. ANG mutations segregate with familial and ‘sporadic’ amyotrophic lateral sclerosis. Nat Genet 2006; 38: 411–13. 7 Nishimura AL, Mitne-Neto M, Silva HC, et al. A mutation in the vesicle-trafficking protein VAPB causes late-onset spinal muscular atrophy and amyotrophic lateral sclerosis. Am J Hum Genet 2004; 75: 822–31. 8 Hardiman O, Greenway M. The complex genetics of amyotrophic lateral sclerosis. Lancet Neurol 2007; 6: 291–92. 9 Simpson CL, Al-Chalabi A. Amyotrophic lateral sclerosis as a complex genetic disease. Biochim Biophys Acta 2006; 1762: 973–85. 10 Veldink JH, Van den Berg LH, Wokke JH. The future of motor neuron disease: the challenge is in the genes. J Neurol 2004; 251: 491–500. 11 Graham AJ, Macdonald AM, Hawkes CH. British motor neuron disease twin study. J Neurol Neurosurg Psychiatry 1997; 62: 562–69. 12 Sathasivam S, Ince PG, Shaw PJ. Apoptosis in amyotrophic lateral sclerosis: a review of the evidence. Neuropathol Appl Neurobiol 2001; 27: 257–74. 13 Van Den Bosch L, Van Damme P, Bogaert E, Robberecht W. The role of excitotoxicity in the pathogenesis of amyotrophic lateral sclerosis. Biochim Biophys Acta 2006; 1762: 1068–82. 14 Oosthuyse B, et al. Deletion of the hypoxia-response element in the vascular endothelial growth factor promoter causes motor neuron degeneration Nat Genet 2001; 28: 131–38. 876 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Goodall EF, Greenway MJ, van M, I, Carroll CB, Hardiman O, Morrison KE. Association of the H63D polymorphism in the hemochromatosis gene with sporadic ALS. Neurology 2005; 65: 934–37. Kasperaviciute D, Weale ME, Shianna KV, et al. Large-scale pathwaysbased association study in amyotrophic lateral sclerosis. Brain 2007; 130: 2292–301. Lambrechts D, Storkebaum E, Morimoto M, et al. VEGF is a modifier of amyotrophic lateral sclerosis in mice and humans and protects motoneurons against ischemic death. Nat Genet 2003; 34: 383–94. Saeed M, Siddique N, Hung WY, et al. Paraoxonase cluster polymorphisms are associated with sporadic ALS. Neurology 2006; 67: 771–76. Sutedja NA, Sinke RJ, Van Vught PW, et al. The association between H63D mutations in HFE and amyotrophic lateral sclerosis in a Dutch population. Arch Neurol 2007; 64: 63–67. Veldink JH, Kalmijn S, Van der Hout AH, et al. SMN genotypes producing less SMN protein increase susceptibility to and severity of sporadic ALS. Neurology 2005; 65: 820–25. Davey SG, Ebrahim S, Lewis S, Hansell AL, Palmer LJ, Burton PR. Genetic epidemiology and public health: hope, hype, and future prospects. Lancet 2005; 366: 1484–98. Cordell HJ, Clayton DG. Genetic association studies. Lancet 2005; 366: 1121–31. The Welcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 2007; 447: 661–78. Easton DF, Pooley KA, Dunning AM, et al. Genome-wide association study identifies novel breast cancer susceptibility loci. Nature 2007; 447: 1087–93. Klein RJ, Zeiss C, Chew EY, et al. Complement factor H polymorphism in age-related macular degeneration. Science 2005; 308: 385–89. van Heel DA, Franke L, Hunt KA, et al. A genome-wide association study for celiac disease identifies risk variants in the region harboring IL2 and IL21. Nat Genet 2007; 39: 827–29. The International HapMap Consortium. The International HapMap Project. Nature 2003; 426: 789–96. Borovecki F, Lovrecic L, Zhou J, et al. Genome-wide expression profiling of human blood reveals biomarkers for Huntington’s disease. Proc Natl Acad Sci U S A 2005; 102: 11023–28. Tang Y, Lu A, Aronow BJ, Sharp FR. Blood genomic responses differ after stroke, seizures, hypoglycemia, and hypoxia: blood genomic fingerprints of disease. Ann Neurol 2001; 50: 699–707. Maes OC, Xu S, Yu B, Chertkow HM, Wang E, Schipper HM. Transcriptional profiling of Alzheimer blood mononuclear cells by microarray. Neurobiol Aging 2006; published online Sept 18. DOI:10.1016/j.neurobiolaging.2006.08.004. Scherzer CR, Eklund AC, Morse LJ, et al. Molecular markers of early Parkinson’s disease based on gene expression in blood. Proc Natl Acad Sci U S A 2007; 104: 955–60. Brooks BR. El Escorial World Federation of Neurology criteria for the diagnosis of amyotrophic lateral sclerosis. J Neurol Sci 1994; 124(suppl): 96–107. Chao A, Tsay PK, Lin SH, Shau WY, Chao DY. The applications of capture-recapture models to epidemiological data. Stat Med 2001; 20: 3123–57. Hook EB, Regal RR. Capture-recapture methods. Lancet 1992; 339: 742. Wen SH, Tzeng JY, Kao JT, Hsiao CK. A two-stage design for multiple testing in large-scale association studies. J Hum Genet 2006; 51: 523–32. Cronin S, Hardiman O, Traynor BJ. Ethnic variation in the incidence of ALS: a systematic review. Neurology 2007; 68: 1002–07. Purcell S, Neale B, Todd-Brown K, et al. PLINK: a toolset for wholegenome association and population based linkage analysis. Am J Hum Genet 2007; 81: 559–75. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet 2006; 38: 904–09. Van Vught PW, Sutedja NA, Veldink JH, et al. Lack of association between VEGF polymorphisms and ALS in a Dutch population. Neurology 2005; 65: 1643–45. http://neurology.thelancet.com Vol 6 October 2007 Articles 40 41 42 43 44 45 Schymick JC, Scholz SW, Fung HC, et al. Genome-wide genotyping in amyotrophic lateral sclerosis and neurologically normal controls: first stage analysis and public release of data. Lancet Neurol 2007; 6: 322–28. Dunckley T, Huentelman MJ, Craig DW, et al. Whole-genome analysis of sporadic amyotrophic lateral sclerosis. N Engl J Med 2007; published online Aug 1. DOI:10.1056/NEJMoa070174. Mikoshiba K. Inositol 1,4,5-trisphosphate IP(3) receptors and their role in neuronal cell function. J Neurochem 2006; 97: 1627–33. Van Den Bosch L, Verhoeven K, De Smedt H, Wuytack F, Missiaen L, Robberecht W. Calcium handling proteins in isolated spinal motoneurons. Life Sci 1999; 65: 1597–606. Choe CU, Ehrlich BE. The inositol 1,4,5-trisphosphate receptor (IP3R) and its regulators: sometimes good and sometimes bad teamwork. Sci STKE 2006; 28: re15. Boehning D, Patterson RL, Snyder SH. Apoptosis and calcium: new roles for cytochrome c and inositol 1,4,5-trisphosphate. Cell Cycle 2004; 3: 252-54. http://neurology.thelancet.com Vol 6 October 2007 46 47 48 49 Tantral L, Malathi K, Kohyama S, Silane M, Berenstein A, Jayaraman T. Intracellular calcium release is required for caspase-3 and -9 activation. Cell Biochem Funct 2004; 22: 35–40. Wozniak AL, Wang X, Stieren ES, Scarbrough SG, Elferink CJ, Boehning D. Requirement of biphasic calcium release from the endoplasmic reticulum for Fas-mediated apoptosis. J Cell Biol 2006; 175: 709–14. Gutstein DE, Marks AR. Role of inositol 1,4,5-trisphosphate receptors in regulating apoptotic signaling and heart failure. Heart Vessels 1997; Suppl 12: 53–57. Yamamoto-Hino M, Sugiyama T, Hikichi K, et al. Cloning and characterization of human type 2 and type 3 inositol 1,4,5trisphosphate receptors. Receptors Channels 1994; 2: 9–22. 877