A Point of Rarity in Genetic Risk for Bipolar Disorder and Schizophrenia
Total Page:16
File Type:pdf, Size:1020Kb
WEB-ONLY CONTENT Rare Copy Number Variants A Point of Rarity in Genetic Risk for Bipolar Disorder and Schizophrenia Detelina Grozeva, MSc; George Kirov, PhD, MRCPsych; Dobril Ivanov, MSc; Ian R. Jones, PhD, MRCPsych; Lisa Jones, PhD; Elaine K. Green, PhD; David M. St Clair, MD, PhD; Allan H. Young, PhD, FRCPsych; Nicol Ferrier, PhD, FRCPsych; Anne E. Farmer, PhD, FRCPsych; Peter McGuffin, PhD, FRCPsych; Peter A. Holmans, PhD*; Michael J. Owen, PhD, FRCPsych*; Michael C. O’Donovan, PhD, FRCPsych*; Nick Craddock, PhD, FRCPsych*; for the Wellcome Trust Case Control Consortium Arch Gen Psychiatry. 2010;67(4):318-327 14 RESULTS schizophrenia article. In cases, this is caused by the ex- clusion of schizoaffective cases in the current analysis. The differences with regard to controls are caused by re- COMPARISON OF BURDEN OF COPY NUMBER analysis of a small subset of them. Some of the control VARIANTS ACCORDING TO SIZE BETWEEN data were processed with the use of different reference BIPOLAR CASES, CONTROLS, AND batches, which enabled us to include more controls in SCHIZOPHRENIA CASES the current analysis. We compared our bipolar disorder cases against our set of schizophrenia cases from the same population who had CNVs THAT OCCURRED MORE OFTEN IN been examined by the same methods (n=440).14 The copy BIPOLAR DISORDER CASES THAN CONTROLS number variants (CNVs) were classified into size cat- egories as in our previous report. eTable 1 shows the Although we did not observe an overall increase of CNV results. Compared with bipolar disorder, in the schizo- burden in bipolar cases compared with controls, some phrenia sample we observed a significant excess of large individual CNVs were more common in cases than con- deletions (PϽ.001) and total large CNVs (PϽ.001) and trols. None was significantly associated after correction a trend for excess large duplications (P=.053). for multiple testing. Those CNVs for which there was an uncorrected nominal significance level of ␣Յ.05 are pre- DIFFERENCES BETWEEN SAMPLES USED sented in eTable 2. IN SCHIZOPHRENIA ANALYSIS REPORTED PREVIOUSLY AND SCHIZOPHRENIA-CONTROL ANALYSIS OF CNVs IN REGIONS PREVIOUSLY COMPARISONS IN THIS ANALYSIS REPORTED TO SHOW ASSOCIATIONS IN BIPOLAR DISORDER SAMPLES There are small differences with respect to the analyzed individuals reported herein and the ones described in our We investigated whether CNV regions previously hypoth- esized to be involved in susceptibility to bipolar disorder *Indicates research team members who played a major role in were affected by CNVs in our data. The CNVs on chro- supervising and coordinating the bipolar research described mosomes 1p34.3, 14q23.3, and 22q12.3, affecting GLUR7 herein. (official symbol, GRIK3), AKAP5, and CACNG2, respec- eTable 1. Comparisons of Rates of CNVs Between Controls and BD and SZ Casesa Size CNVs per BD to Control Ratio SZ to Control Ratio BD to SZ Ratio CNV Type Range, kb Control (P Value) (P Value) (P Value) Deletion 100-200 0.104 0.801 (.03) 0.771 1.045 200-500 0.102 0.864 1.141 0.758 500-1000 0.012 1.264 0.938 1.348 Ͼ1000 0.007 0.579 3.189 (.004) 0.181 (Ͻ.001) Total 0.225 0.85 (.01) 1.032 0.825 Duplication 100-200 0.093 0.982 0.882 1.126 200-500 0.155 1.051 0.951 1.105 500-1000 0.050 0.868 1.147 0.757 Ͼ1000 0.023 0.839 1.570 0.53 (.053) Total 0.321 0.990 1.007 0.984 Total (deletionϩduplication) 0.546 0.930 1.017 0.918 Abbreviations: BD, bipolar disorder; CNV, copy number variant; kb, kilobase; SZ, schizophrenia. a Significant differences are in boldface type. P values are not corrected for multiple testing. Those that would survive correction are the excess of large (Ͼ1 megabase) deletions in SZ compared with either controls or BD cases. (REPRINTED) ARCH GEN PSYCHIATRY/ VOL 67 (NO. 4), APR 2010 WWW.ARCHGENPSYCHIATRY.COM E1 ©2010 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 eTable 2. Regions Showing More CNVs in Cases Than Controls at a Nominal Significance of PՅ.05 BD P Value, Locusa Start bpb End bpb Type Casesc Controlsc Fisher Exact Test Gene 1q25.1 173 769 777 173 978 862 dup 5 1 .03 TNR 9q31.1 104 826 097 104 885 068 del/dup 3 0 .05 None 12p11.21 31 202 250 31 301 551 dup 19 16 .03 OVOS2 18p11.21-11.1 14 694 694 15 092 421 dup 3 0 .05 ANKRD30B 19p12 20 001 614 20 177 979 del/dup 5 1 .03 ZNF682, ZNF90, ZNF486 19p12 24 013 968 24 295 825 dup 3 0 .05 ZNF254 Abbreviations: BD, bipolar disorder; bp, base pair; CNV, copy number variant; del, deletion; dup, duplication. a Locus refers to the chromosome band. b Start bp and end bp provide the start base pair position and the end base pair position of the CNVs at the specified locus. c Numbers of individuals with BD and controls harboring the CNV. eTable 3. Percentage of Samples Carrying at Least 1 Singleton CNV in Our Study Compared With the Study of Zhang et al20 Current Study, No. (%) Study of Zhang et al, No. (%) Single CNV Type Cases With CNV Controls With CNV Cases With CNV Controls With CNV Deletion 101 (6.0) 173 (6.2) 162 (16.2) 127 (12.3) Duplication 121 (7.1) 234 (8.3) 197 (19.7) 197 (19.1) Deletionϩduplication 187 (11.0) 329 (11.7) 320 (32.0) 299 (29.0) Abbreviation: CNV, copy number variant. tively, were studied.21 No deletions or duplications were not in any control. Where several CNVs mapped to the found. The other studied locus was 3q13.3, reported to same region, they did not always fully overlap. The start harbor duplication in the GSK3B gene.22 This gene codes and end points in the table refer to the total region cov- for glycogen synthase kinase, a key player in the Wnt sig- ered by such a cluster of CNVs. The list contains several naling pathway and a target of lithium.22 No CNVs were intriguing candidate genes (eg, neuroligin 1, neuregu- detected at this locus in bipolar cases. lin 3, and the ␣-2 catenin), although we have no spe- cific evidence that any contribute to susceptibility to bi- SINGLETON EVENTS: COMPARISON polar disorder. OF OUR DATA WITH PREVIOUS STUDIES ANALYSIS OF CNVs THAT DISRUPT GENES With the use of PLINK,24 deletions-only and duplications- only data sets were generated. In these data sets, a dele- The global burden of CNVs in cases and controls with tion was still defined as a single-occurrence event even respect to genes that are deleted, duplicated, or dis- if there was duplication at the same region. The same rule rupted by CNVs was estimated by using PLINK24 1.05. was applied to duplications. Therefore, the number of Only CNVs that overlapped with at least 1 gene, on single-occurrence deletions and single-occurrence du- the basis of hg18 genomic coordinates, were taken into plications does not sum to the total number of single- account. This analysis was performed for all CNVs and occurrence events. then for the CNVs that occurred only once in the data. The power in our sample at ␣=.05 to detect a signifi- cant difference with the effect size observed by Zhang et CNVs PREVIOUSLY ASSOCIATED al20 (␦=0.11) was 95%. For comparison, the power to de- WITH SCHIZOPHRENIA tect a significant difference (␦=0.11) with their sample size was 71% (␣=.05). As mentioned in the “Comment” Table 4 in the main article shows the main chromo- section, the Zhang et al study20 used a higher-resolution somal regions with CNVs associated with schizophre- array (Affymetrix 6.0) and had a smaller sample size. Both nia and the number of the corresponding CNVs in the of these factors can lead to the observation of a higher bipolar disorder cases and controls. Herein we provide number of singleton CNVs in their study and could ex- more information on these CNVs. plain the differences from our study (eTable 3). A deletion at 15q13.3 loci has been found to be asso- ciated with schizophrenia in 2 large studies of schizo- ANALYSIS OF GENES WITHIN CNVs phrenia.12,13 Deletion in the same region was reported to FOUND ONLY IN CASES be associated with mental retardation and seizures.16 No deletions were observed in bipolar disorder, but 2 cases We provide a list of genes within all CNVs that were pres- and no controls were found to harbor duplications in this ent in cases but not in controls (eTable 4). The list in- region (Fisher exact test, P=.14). The significance (or oth- cludes not only singleton CNVs (those found once in the erwise) of duplications in this region is not yet known data set) but also CNVs found in more than 1 case, but from other studies. (REPRINTED) ARCH GEN PSYCHIATRY/ VOL 67 (NO. 4), APR 2010 WWW.ARCHGENPSYCHIATRY.COM E2 ©2010 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/26/2021 eTable 4. Genes in CNVs in Cases Only No. of Chr Start bp End bp Type NCBI Genes Observations 1 45 592 076 45 881 891 3 TESK2, LOC126661, MMACHC, PRDX1, AKR1A1, NASP, CCDC17, 1 GPBP1L1 1 46 131 935 46 268 021 3 MAST2 1 1 65 071 676 65 413 912 3 JAK1, MIRN101-1, AK3L1, AK3L2 1 1 86 324 452 86 589 985 1 COL24A1, ODF2L 1 1 88 847 985 89 032 679 3 PKN2 1 1 107 551 303 107 823 281 3 NTNG1 1 1 143 576 984 143 916 898 1 PDE4DIP, FLJ21272, SEC22B 1 1 152 936 201 153 149 168 1 KCNN3 1 1 166 729 757 167 024 066 1 XCL2, XCL1, DPT 1 1 173 980 586 174 202 365 3 LOC100128153, RFWD2 1 1 217 365 162 217 613 673 1 LOC643723, LYPLAL1 1 1 221 702 789 222 127 692 3 LOC644151, LOC653428, CAPN8, CAPN2, TP53BP2 1 1 231 361 905 231 514 154 1 PCNXL2 1 1 244 896 239 245 307 486 3 C1orf71, SCCPDH, LOC100130097, LOC149134, AHCTF1, ZNF695, 1 ZNF670 2 155 674 306 842 3 LOC727818, SH3YL1, ACP1, FAM150B 1 2 9 950 762 10 157 626 3 TAF1B, GRHL1, UNQ5830, KLF11, LOC100131506,