|
Replication of the Association between Copy Number Variation on 8p23.1 and Autism by Using ASD-specific BAC Array.
Vol. 8(1) 1927, March 2010To discover genetic markers for autism spectrum disorder (ASD), we previously applied genomewide BAC array comparative genomic hybridization (arrayCGH) to 28 autistic patients and 62 normal controls in Korean population, and identified that chromosomal losses on 8p23.1 and on 17p11.2 are significantly associated with autism. In this study, we developed an 8.5K ASDspecific BAC array covering 27 previously reported ASDassociated CNV loci including ours and examined whether the associations would be replicated in 8 ASD patient cell lines of four different ethnic groups and 10 Korean normal controls. As a result, a CNV-loss on 8p23.1 was found to be significantly more frequent in patients regardless of ethnicity (p＜0.0001). This CNV region contains two coding genes, DEFA1 and DEFA3, which are members of DEFENSIN gene family. Two other CNVs on 17p11.2 and Xp22.31 were also distributed differently between ASDs and controls, but not significant (p=0.069 and 0.092, respectively). All the other loci did not show significant association. When these evidences are considered, the association between ASD and CNV of DEFENSIN gene seems worthy of further exploration to elucidate the pathogenesis of ASD. Validation studies with a larger sample size will be required to verify its biological implication. Autism spectrum disorder (ASD) is neurodevelopmental syndrome with a broad spectrum of phenotypes including profound deficits in social relatedness and communication, repetitive behaviors, and restricted interests (Rutter, 2005). There is accumulating evidence which suggests substantial genetic contribution to ASD development (Folstein and RosenSheidley, 2001; Veenstra VanderWeele and Cook, 2004). For example, the concordance rate of autism was significantly higher in monozygotic twins than in dizygotic twins (Folstein and RosenSheidley, 2001; VeenstraVanderWeele and Cook, 2004). Although previous studies including linkage analyses have revealed a number of potential ASDassociated loci, genetic and phenotypic heterogeneity of the disease has made it difficult to replicate the results as well as to identify ASDassociated genetic mechanisms (Folstein and RosenSheidley, 2001; Klauck, 2006; VeenstraVanderWeele and Cook, 2004). Recently a new type of human genetic variation named copy number variant (CNV) has been suggested as one of the major sources of genomic diversity ranging from normal phenotypic heterogeneity to disease susceptibility (Freeman et al., 2006; Iafrate et al., 2004; Sebat et al., 2004). Indeed, lines of evidence have been reported which demonstrated the association between CNV and ASD (Cho et al., 2009; Cook and Scherer, 2008; Glessner et al., 2009; Marshall et al., 2008; Sebat et al., 2007). In our previous study, we identified 38 copy number variable regions in ASD patients, two of which (CNVs on 8p23.1 and 17p11.2) were found to be significantly associated with ASD (Cho et al., 2009). Using 27 CNV loci previously reported to be associated with ASD including ours, we developed an 8.5K ASDspecific BAC array and screened whether the association between those CNVs would be replicated in patient cell lines of four different ethnic groups. We used DNA extracted from Blymphocyte cell lines of 8 ASD patients of four different ethnicities; Caucasian, Chinese, Asiatic Indian, and Hispanic (Coriell Institute for Medical Research, Camden, NJ, USA). General characteristics of the 8 ASD patients are shown in Table 1. As normal control subjects, we used 10 normal Korean adults without any sign of ASD and other genetic disorders. Using the D5500A PUREGENE DNA Isolation kit (QIAGEN, Valencia, CA, USA), we extracted genomic DNA from the blood of controls according to manufacturer’s instructions. DNA quantity and quality were checked by a NanoDrop assay and 1% agarose gel electrophoresis. This study was performed under the approval of the Institutional Review Boards of The Catholic University of Korea (CUMC07U025). In total, 846 BAC clones representing the 27 candidate CNV regions were selected from the DNA BAC library (Macrogen, Seoul, Korea) and the BAC clone DNA was extracted as described elsewhere (Cho et al., 2009; Chung et al., 2004). In brief, each BAC clone was isolated, grown in 500ml of culture media, pelleted, from which DNA was extracted using an alkaline lysis method. All selected clones were bi-directionally sequenced using the ABI PRISM 3700 DNA Analyzer (Applied Biosystems, Foster City, CA, USA), and their sequences were blasted and mapped according to their linear positions. Extracted BAC DNA was dissolved in 50% DMSO with a concentration of 400~500ng/ul and spotted in triplicate onto the glass by the Genemachines OmniGrid 100 arrayer (Digilab Genomic Solutions, Holliston, MA, USA). We used Corning UltraGAPS amine coating slides (Corning, Acton, MA, USA) and Telecam SMP4 pins (Arrayit Corporation, Sunnyvale, CA, USA) for DNA spotting. We followed processes for a general contact type spotter. Array-CGH was performed as described elsewhere using the MAUI hybridization station (BioMicro Systems, Salt Lake City, Utah, USA) (Joo et al., 2008). In one tube, Cy3labeled sample and Cy5labeled reference DNAs were mixed together and 50μg human Cot I DNA (HybMasker, ConnectaGen, Seoul, Korea), 20μl of 3 M sodium acetate and 600μl of cold 100% ethanol were added to precipitate DNA. The obtained pellet was resuspended in 40μl of hybridization solution containing 50% formamide, 10% dextran sulfate, 2 SSC, 4% SDS and 200 ug of yeast tRNA. Hybridization was performed in slide chambers for 48 hr at 37C. After the hybridization, slides were washed serially at room temperature in solution 1 (2X salinesodium citrate (SSC), 0.1% SDS) for 10 min (1X), in solution 2 (0.1X SSC, 0.1% SDS) for 10 min (2X) and in solution 3 (0.1X SSC) for 1 min (3X) followed by rinsing in distilled water for 10 sec. Finally, the slides were spindried for 2 min at 1000 rpm. All the experiments were duplicated and dye swapping was done for more reliable interpretation. Arrays were scanned and analyzed using the GenePix 4200A twocolor fluorescent scanner (MDS Analytical Technologies, Sunnyvale, CA, USA) and the images were processed using the MacViewer 1.6 imaging software (Macrogen, Seoul, Korea). Signal intensity ratios (test/reference) were measured and converted to log2 scale. Background corrected signal intensity ratios were normalized using Lowess normalization. For defining CNVs, we set the cutoff values of signal intensity ratios at above 0.25 (log2 scale) for copy number gain and at below −0.25 for copy number loss according to our previous study (Cho et al., 2009; Jeong et al., 2008). We also analyzed dyeswap experiment results to get more reliable CNV call. If a CNV call is true, its test/reference intensity ratio value must be inverted by switching the dyes. Only the original ratios and inverted dye swap signals were observed, the according CNV was called as a true one. We used Stata version 10.0 (Stata Corporation, College Station, Tx) and performed the chi-square or Fisher’s exact test to compare the distribution of CNV regions between cases and controls. False discovery rate (FDR) was used for multiple comparisons correction. p values less than 0.05 was considered statistically significant. We developed a ASD-specific BAC array named ‘MacArray Karyo Autism Prototype’ which covers a total of 27 CNV regions that have been reported to be associated with autism in previous studies including ours (Cho et al., 2009; Klauck, 2006; Schellenberg et al., 2006; Sultana et al., 2002; Vorstman et al., 2006). A total of 846 BAC clones representing the candidate 27 CNV regions were selected for designing ASDspecific BAC array (Fig. 1). Locus specificities of selected clones were improved by removing multiple locibinding clones under standard fluorescence in situ hybridization (FISH). The information about those targeted 27 genomic regions and 846 BAC clones is available in Supplementary data 1. Among the 846 BAC clones, 77 spots (15 loci) showed CNVs in at least 2 study participants. Details of the CNV profiles of the 8 ASDs and 10 normal controls for the 77 spots are available in supplementary data 2. Of the 77 spots, a CNVloss on 8p23.1 (BAC168_A06) was found to be significantly more frequent in ASD patients (7 out of 8 ASD patients versus none in controls, p＜ 0.0001) (Table 2). Two other CNVs were distributed differently between patients and controls, but not significant. A CNVloss on 17p11.2 (BAC130_G23) was found in 3 out of 8 ASD patients, but not in controls (p=0.069). A CNVgain on Xp22.31 (BAC231_F19) was found in 4 out of 10 controls, but not in ASDs (p=0.092) (Table 2). The most significant CNV region (BAC168_A06, 8p23.1) contains two coding genes, DEFA1 and DEFA3, which are members of DEFENSIN gene family. Interestingly, the copy number loss CNV on 8p23.1 was identified in all four different ethnic groups and also consistent in duplicated experiments. Fig. 2 shows the examples of CNVs on 8p23.1 and 17p11.2 in ASD patients, which are duplicated and dye swapped and no CNVs in normal control individuals. It has been widely accepted that ASD has a strong genetic background (Folstein and Rosen-Sheidley, 2001; Rutter, 2005; Veenstra-VanderWeele and Cook, 2004). Although genomic loci or genes reported to be associated with ASD were not easily replicated in subsequent studies for many reasons such as disease heterogeneity per se and relatively high noise levels of genomics technology used, some of the ASDassociated CNVs have been successfully replicated. For example, CNVs in NRXN1, NLGN4 and SHANK3 genes, which had been identified in target gene studies, were replicated in genomewide CNV studies for ASD (Glessner et al., 2009; Marshall et al., 2008). Given that SNPs or other genetic markers were not very consistent in ASD, this suggests that structural variation could be a more consistent and important marker in ASD and facilitate the elucidation of genetic mechanisms behind the disease. In our previous study, we identified 38 CNV regions in 28 Korean ASD patients and two of them (CNVs on 8p23.1 and 17p11.2) were significantly associated with ASD. In this study, we designed and prepared an ASDspecific BAC array for CNV analysis and examined whether the significant CNV markers would be replicated in independent ASD patients of diverse ethnic origins. The ASD-specific BAC array contains the two significant CNV regions identified in our previous study and also 25 putative ASD-associated regions reported by previous linkage studies (Cho et al., 2009; Klauck, 2006; Rutter, 2005; Schellenberg et al., 2006; Sultana et al., 2002; Vorstman et al., 2006). However, all 25 candidate regions suggested by linkage analyses were not replicated in our study population. This data seems coherent with previous observations that positive findings from one linkage study often fail to replicate in the other observation (Glessner et al., 2009), but it could be due to small sample size of the study. In this study, the association of copy number loss CNV on 8p23.1 with ASD was consistently replicated in diverse ethnic groups. 8p23.1 locus is known to be a frequent region of DNA structural variation in human (Hollox et al., 2008a). DEFENSIN gene family, which is a well-known component of innate immunity, is clustered in this region (Ganz, 1999; Hollox et al., 2008a). CNV of DEFENSIN family, especially betadefensin, has been reported as a risk factor for different diseases such as psoriasis, Crohn’s disease and cystic fibrosis (Hollox, 2008b; Hollox et al., 2008c), which suggests that CNV of DEFENSIN family is likely to be associated with autoimmune diseases. Immunological dysfunction has been suggested to be associated with autism (Folstein and Rosen-Sheidley, 2001; Rutter, 2005) and the copy number loss on 8p23.1 was reported to be associated with behavioral problems or mental slowness (Pettenati et al., 1992). When this evidence is considered, the association between ASD and CNV of DEFENSIN gene seems worthy of further exploration to elucidate the pathogenesis of ASD. Validation studies with a larger sample size will be required to verify its biological implication.
|
