Genomewide association analysis of human narcolepsy and a new resistance gene

Abstract

Human narcolepsy is a hypersomnia that is affected by multiple genetic and environmental factors. One genetic factor strongly associated with narcolepsy is the HLA-DRB1*1501-DQB1*0602 haplotype in the human leukocyte antigen region on chromosome 6, whereas the other genetic factors are not clear. To discover additional candidate regions for susceptibility or resistance to human narcolepsy, we performed a genomewide association study, using 23,244 microsatellite markers. Two rounds of screening with the use of pooled DNAs yielded 96 microsatellite markers (including 16 markers on chromosome 6) with significantly different estimated frequencies in case and control pools. Markers not located on chromosome 6 were evaluated by the individual typing of 95 cases and 95 controls; 30 markers still showed significant associations. A strong association was displayed by a marker on chromosome 21 (21q22.3). The surrounding region was subjected to high-density association mapping with 14 additional microsatellite markers and 74 SNPs. One microsatellite marker (D21S0012m) and two SNPs (rs13048981 and rs13046884) showed strong associations (P < .0005; odds ratios 0.19-0.33). These polymorphisms were in a strong linkage disequilibrium, and no other polymorphism in the region showed a stronger association with narcolepsy. The region contains three predicted genes--NLC1-A, NLC1-B, and NLC1-C--tentatively named "narcolepsy candidate-region 1 genes," and NLC1-A and NLC1-C were expressed in human hypothalamus. Reporter-gene assays showed that the marker D21S0012m in the promoter region and the SNP rs13046884 in the intron of NLC1-A significantly affected expression levels. Therefore, NLC1-A is considered to be a new resistance gene for human narcolepsy.

Figure 1.

Genomewide association analyses with 23,244 microsatellite markers. A, Flow chart of this study. a, MS = microsatellite marker. b, Sixteen MSs on chromosome 6 and 80 MSs on other chromosomes showing reproducible peak patterns and remaining significantly different between cases and controls. c, Selected randomly from the samples used in the first and second screenings. d, Eleven MSs reaching significance in both 2×2 and 2×m contingency tables. e, All samples used in first and second screenings. B, Results of genomewide screening. The figure shows P values by Fisher’s exact test based on 2×2 contingency tables (green line) or on 2×m contingency tables (red line) in the first screening. Yellow circles indicate 30 MSs that showed significant associations in both first and second screenings and in individual typing.

Figure 2.

High-density mapping with additional microsatellite markers and SNPs. A, Association analyses using high-density microsatellite markers with 220 cases and 420 controls. Unblackened circles indicate microsatellite markers used in the first and second screenings. Blackened circles indicate microsatellite markers newly developed for the high-density mapping. The dark line shows the P values calculated by Fisher’s exact test based on 2×2 contingency tables, whereas the lighter line shows those from 2×m contingency tables. B, Association analyses using SNPs with 190 cases and 190 controls. The X-axis indicates the distance from D21S0012m. SNP rs12483718 is located near D21S0241i. The Y-axis shows the P values calculated by Fisher’s exact test based on 2×2 contingency tables. Two SNPs, rs13048981 and rs13046884, showed the strongest associations in the NLC1 region. C, Variation screening and high-density association analyses in the NLC1 region. Top, Exon-intron structures of NLC1-A, NLC1-B, and NLC1-C. Boxes indicate exons, with unblackened boxes indicating untranslated regions and blackened boxes indicating coding regions. Predicted gene regions and 1 kb of upstream region were screened for sequence variations. Fourteen additional polymorphisms, including two nonsynonymous substitutions, were detected and were examined for possible associations, but no polymorphisms showed stronger association than D21S0012m, rs13048981, and rs13046884. Bottom, P values for individual SNPs.

Figure 3.

LD block structure. A, Pairwise LD between polymorphisms, estimated with the parameter | D′| in cases (upward triangle) and controls (downward triangle). B, Pairwise LD between polymorphisms, estimated with the parameter r² in cases (upward triangle) and controls (downward triangle). The figures are adjusted for physical distance. Regions of high and low degrees of LD are shown in red and blue, respectively. For SNPs, significant P values of association analyses with individual genotyping of 190 case and 190 control samples are shown.

Figure 4.

Expression analysis for NLC1-A, NCL1-B, and NCL1-C, with the use of RT-PCR. A, Schematic drawing of the specific primers for RT-PCR. B, Expected size of RT-PCR products from cDNA or genomic DNA. On the basis of the UCSC Genome Browser, products with the expected size were amplified from cDNA for NLC1-A and NLC1-C in samples of whole brain, hypothalamus, and several other organs, but, for NLC1-B, only the products from genomic DNA were observed (C). Amplified products were confirmed by direct sequencing. Lane 1, Heart; lane 2, liver; lane 3, spleen; lane 4, pancreas; lane 5, lung; lane 6, whole brain; lane 7, hypothalamus; lane 8, kidney; lane 9, skeletal muscle; lane 10, sperm. NC = negative control. M = 100-bp ladder size marker.

Figure 5.

Effects of the microsatellite marker D21S0012m in the promoter region and of SNP rs13046884 in the intron 1 of NLC1-A on transcriptional activity. Reporter-gene constructs contained the sequences from IVS1+31 to IVS1+327 for rs13046884 or 80–987 nt upstream of the transcription initiation site for D21S0012m. The chart shows luciferase expression from each reporter in transfected HeLa cells or NB-1 cells, relative to empty vector. Data are means of at least three independent experiments. Error bars represent SDs.