|Home | About | Journals | Submit | Contact Us | Français|
A Genome-wide association study (GWAS) identified significant association between variants in MEIS1, BTBD9, and MAP2K5/SKOR1 and restless legs syndrome (RLS). However, many independent replication studies are needed to unequivocally establish a valid genotype-phenotype association across various populations. To further validate the GWAS findings, we investigated three variants, rs2300478 in MEIS1, rs9357271 in BTBD9 and rs1026732 in MAP2K5/SKOR1 in 38 RLS families and 189 RLS patients/560 controls from the U.S. for their association with RLS.
Both family-based and population-based case-control association studies were carried out.
The family-based study showed that SNP rs1026732 in MAP2K5/SKOR1 was significantly associated with RLS (P=0.01). Case-control association studies showed significant association between all three variants and RLS (P=0.0001/OR=1.65, P=0.0021/OR=1.59 and P=0.0011/OR=1.55 for rs2300478, rs9357271 and rs1026732, respectively).
Variants in MEIS1, BTBD9 and MAP2K5/SKOR1 confer a significant risk of RLS in a U.S. population.
Restless legs syndrome (RLS, MIM 102300) is a common neurological disorder which is characterized by: 1) an urge to move the legs, usually due to uncomfortable sensations primarily in the legs; 2) transient improvement with movement; 3) worsening of symptoms with inactivity; 4) worsening of symptoms in the evening or night . Periodic limb movements of sleep (PLMS) are seen in most cases. RLS can also be secondary to renal failure, iron deficiency, pregnancy, neuropathy, myelinopathy, multiple sclerosis and possibly Parkinson disease [2-4]. RLS disturbs general health, mental health, and quality of life regardless of whether it is idiopathic or secondary to other causes. Pathologic studies showed reduced brain and cerebrospinal fluid iron and iron associated proteins [5,6]. Dopamine agonists are the best-studied treatment for RLS .
RLS has a particularly high prevalence rate in Caucasian populations, and is present in approximately 10% of the U.S. population . Genetic factors have been shown to be involved in the pathogenesis of RLS [9,10]. Published linkage analysis in families have identified seven genetic loci for RLS, RLS1 on chromosome 12q12-q21, RLS2 on 14q13-21, RLS3 on 9p24-p22, RLS4 on 2q33, RLS5 on 20p13, RLS6 on 19p13 and RLS7 on 16p12.1 [10-16]. A recent genome-wide association study (GWAS) using 500,000 single nucleotide polymorphisms (SNPs) identified four additional genetic loci associated with RLS, which are represented by SNPs on chromosome 2p14 (MEIS1), 6p21.2 (BTBD9), 15q23 (MAP2K5/SKOR1 or MAP2K5/LBXCOR1), and 9p24.1-p23 (PTPRD) [17,18]. Interestingly, GWAS also identified a significant association between a variant in BTBD9 and PLMS, which is a clinical feature in a majority of RLS patients . In general, GWAS results provide powerful evidence to support the association between SNPs and disease. However, rigorous replication studies in independent populations by independent research groups are important to establish an unequivocal association. We have recently shown that SNP rs1975197 in the PTPRD gene within the RLS3 locus reported by our group conferred a significant risk of RLS in the U.S. population . Vilariño-Güell et al. reported that variants in MEIS1 and BTBD9 were associated with RLS in a U.S. population, but four SNPs in MAP2K5/SKOR1 did not show any significant association with RLS in the same population . Kemlink et al. performed an independent replication study for SNPs in MEIS1, BTBD9 and MAP2K5/SKOR1 in three European populations . The difference between these two replication studies with regard to MEIS1, BTBD9 and MAP2K5/SKOR1 SNPs suggests that more replication studies are needed to further validate the GWAS findings.
In this study, we performed association studies in 38 RLS families and a case-control population of 189 RLS patients and in 560 controls in a U.S. population to assess the association between SNP rs2300478 in MEIS1, rs9357271 in BTBD9 and rs 1026732 in MAP2K5/SKOR1 and RLS.
This study was approved by local institutional review boards on human subject research and written consent was obtained from the participants. For the family-based association study, a total of 38 RLS families, including 15 families used in the PTPRD replication research , were studied. There were 611 subjects and 186 affected individuals in the 38 families. The summary statistics of the 38 families are shown in Table 1. We have DNA samples for 150 RLS patients and 85 unaffected family members in 38 RLS families and they were genotyped and analyzed in this study.
For the population-based association study, a total of 189 RLS patients and 560 Caucasian controls were studied as previously described . The 189 RLS patients included 38 probands from the families used for the family-based association study, and were all Caucasians in the U.S.
The study subjects were evaluated for the clinical diagnosis of RLS by two expert neurologists (W.G.O. and N.F.S.) following the diagnostic criteria for RLS described in the 1995 IRLSSG and 2003 NIH reports [1,23,24]. The control subjects for the case-control study were general population controls who were not specifically screened for the presence or absence of RLS.
We selected SNP rs2300478 in MEIS1, rs9357271 in BTBD9 and rs1026732 in MAP2K5/SKOR1 for the replication study. SNP rs2300478 was the only variant in MEIS1 identified by the GWAS . For the BTBD9 locus, GWAS identified two SNPs with similar P values and we selected rs9357271 with a higher minor allele frequency. For the MAP2K5/SKOR1 locus, GWAS identified 6 SNPs and we selected rs1026732 with the best P value.
SNPs were genotyped using the TaqMan allelic discrimination genotyping assay with a protocol recommended by the manufacturer (Applied Biosystems). The assay kit contained the forward target-specific PCR primer, the reverse primer, and the TaqMan MGB probes labeled with two special dyes for two alleles: FAM and VIC. Genotyping was performed in a 5-μl PCR reaction which contained 25 ng/μl DNA, 2.5 μl of TaqMan Universal PCR Master Mix and 0.25 μl of TaqMan SNP genotyping assay. The PCR program was 95°C for 10 min, 60 cycles of 92°C for 15 seconds and 60°C for 1 min and, finally, 4°C for storage . Genotyping data were collected using an ABI PRISM 7900HT Sequence Detection System. Genotypes were called using Software SDS Version 2.1 with automatic allele calling.
The missing genotype rates were 1.6% for cases and 2% for controls for rs2300478, 1.6% for cases and 3.4% for controls for rs9357271 and 1.1% for cases and 3.4% for controls for rs1026732. For the family-based study, All 235 DNA samples were well genotyped for rs2300478, rs9357271 and rs1026732 respectively.
The genotyping data were analyzed for Hardy-Weinberg equilibrium using a Chi-square test (http://www.oege.org/software/hardy-weinberg.html) . Power calculation for the case-control cohort was carried out by the nQuery Advisor 7.0 program using minor allele frequencies (MAF) and odds ratios (ORs) from the previous GWAS report.
For the family-based association study, Sib-TDT was carried out to test the association between a SNP and RLS in the 38 families using the TDT/STDT program 1.1 [27, 28]. The Sib-TDT analyzes whether the risk allele is transmitted preferentially to affected offspring [25,28]. The risk allele is considered to be linked and associated with RLS if it is transmitted more frequently to affected offspring. Sib-TDT was used in this study because RLS is mostly a late-onset disease and parental data were not complete for all parents . There were 112 affected-affected sib pairs, 77 affected-unaffected sib pairs, and 90 unaffected-unaffected sib pairs in the 38 RLS families.
For the population-based case-control association study, we used a Pearson 2×2 contingency table Chi-square test for allelic association and 2×3 contingency tables for genotypic association assuming three different inheritance models, i.e., an additive, dominant or recessive model (SAS version 9.0). ORs and 95% confidence intervals (CI) were estimated using SAS version 9.0. P values were adjusted for multiple testing using the Bonferroni method, thus a P value of 0.05/3=0.017 was considered to be significant.
A family-based association study was carried out in 38 U.S. RLS families with 150 RLS patients and 84 unaffected members using Sib-TDT analysis. As shown in Table 2, SNP rs1026732 in MAP2K5/SKOR1 showed significant association with RLS (P=0.01). However, we did not observe significant association of RLS with SNP rs2300478 in MEIS1 and rs9357271 in BTBD9. The results suggest that SNP rs1026732 in MAP2K5/SKOR1 is a more important risk factor in familial RLS than SNP rs2300478 in MEIS1 and rs9357271 in BTBD9.
A population-based association study was performed with a group of 189 RLS patients and a group of 560 controls in the U.S. The three SNPs, rs2300478 in MEIS1, rs9357271 in BTBD9 and rs1026732 in MAP2K5/SKOR1, were all in Hardy-Weinberg equilibrium in both patient and control groups (p>0.05). Power analysis with reported MAF and ORs from GWAS studies suggests that our case-control population has a sufficient power of 99%, 84% and 77% for detecting the association between RLS and SNP rs2300478, rs9357271 and rs1026732, respectively. All three SNPs showed significant allelic association with RLS with P values of 0.0001 (OR=1.65), 0.0021 (OR=1.59) and 0.0011 (OR=1.55) for rs2300478, rs9357271 and rs1026732, respectively (Table 3). These results suggest that SNP rs2300478, rs9357271 and rs1026732 confer a significant risk of RLS in the U.S. population.
Genotypic association analysis was also carried out to further investigate the association between SNPs rs2300478, rs9357271 and rs1026732 and RLS. Highly significant genotypic association was identified between SNP rs2300478 in MEIS1 and RLS with all three inheritance models, although the association was relatively more significant with an additive model (P=5.0×10−6, 8.6×10−5, 2.8×10−5 assuming an additive, recessive or dominant model, respectively, Table 4). Significant genotypic association was also identified for SNP rs1026732 in MAP2K5/SKOR1 with an additive or recessive model (P=0.0066, 0.0032, respectively, Table 4), but not with a dominant model after Bonferroni correction for multiple testing (Table 4). No significant genotypic association was identified for SNP rs9357271 in BTBD9.
We previously mapped the third RLS locus on chromosome 9p24-22 (RLS3) by family-based linkage analysis . A later GWAS identified the same RLS locus by showing the significant association between SNPs in the PTPRD gene at the RLS3 locus and RLS . Recently, we used a family-based association study and a population-based association study to independently replicate the GWAS finding on SNP rs 1975197 in PTPRD . The same case-control cohort used in the PTPRD replication study was used in this study to assess the three other RLS loci identified by GWAS. Our population-based association studies showed that SNP rs2300478 in MEIS1, rs9357271 in BTBD9 and rs1026732 in MAP2K5/SKOR1 were all significantly associated with RLS (Table 3).
Vilariño-Güell et al. also studied 11 SNPs in the three GWAS RLS loci in a U.S. population and found that SNP rs12469063 in MEIS1 and rs4714156, one of 5 SNPs genotyped in BTBD9, were significantly associated with RLS . In contrast, 4 SNPs in MAP2K5/SKOR1 (rs11635424, rs884202, rs3784709 and rs6494696) studied by Vilariño-Güell et al did not show any significant association with RLS . Vilariño-Güell et al. did not study SNP rs1026732 in MAP2K5/SKOR1, which showed highly significant association with both familial and sporadic RLS in our study. Moreover, consistent with our results, in a replication study in combined European samples involving Czech, Austrian and Finnish RLS patients, all three loci were confirmed to be associated with RLS . Together, available evidence suggests that SNP rs2300478 in MEIS1, rs9357271 in BTBD9 and rs1026732 in MAP2K5/SKOR1 confer significant risk of RLS.
We also used family-based association studies in 38 families to show that SNP rs1026732 in MAP2K5/SKOR1, but not rs2300478 in MEIS1 and rs9357271 in BTBD9, was associated with RLS. This further supports that MAP2K5/SKOR1 is associated with RLS in the U.S. population and that rs1026732 in MAP2K5/SKOR1 may play a role in susceptibility to familial RLS. Together, the results in this study may suggest that SNP rs1026732 in MAP2K5/SKOR1 contributes to a risk in both familial and sporadic RLS, whereas rs2300478 in MEIS1 and rs9357271 in BTBD9 may be associated with sporadic RLS.
Despite the strong genetic evidence showing the significant association between SNP rs2300478 in MEIS1, rs9357271 in BTBD9 and rs1026732 in MAP2K5/SKOR1 and RLS, there is lack of direct evidence to establish MEIS1, BTBD9 and MAP2K5/SKOR1 as pathogenic genes causing RLS. The MEIS1 gene encodes a Meis homeobox protein with expression high in neuroblasma cell lines [29-31]. Xiong L et al did not find any coding or exon-intron boundary mutation after sequencing the MEIS1 gene in 285 RLS probands. However, they reported an intronic variant of MEIS1 associated with RLS, and the expression level of MEIS1 was significantly decreased in brain tissue from homozygous variant carriers . Vilariño-Güell et al. also performed sequencing analysis of MEIS1 in 71 RLS probands. They identified a variant p.R272H that was co-segregated with RLS in a family with 6 members, but the variant was detected in one of 853 controls . These studies provide some evidence to functionally link MEIS1 to RLS.
The BTBD9 gene encodes a BTB/POZ domain-containing protein . Quantitative trait loci (QTL) studies showed that the iron content in the ventral midbrain was related to the BTBD9 gene in the mouse . Stefansson et al. also reported that human serum ferritin levels were significantly decreased with the SNP rs3923809 allele in BTBD9 . Some RLS patients showed an iron deficiency in the central nervous system . Vilariño-Güell et al. did not identify any novel variant in the BTBD9 coding region or exon-intron boundaries in 71 RLS probands . Future studies are needed to determine whether the decreased ferritin levels associated with rs9357271 is causative for development of RLS.
The MAP2K5 gene encodes mitogen-activated protein kinase 5, which interacts with and activates MAPK7/ERK5 [36-38]. Pharmacological evidence suggests that the mitogen-activated protein kinase signaling pathways involving extracellular signal-regulated kinases (ERKs) play important roles in neuroprotection of dopaminergic neurons , which may somehow be linked to positive response to dopaminergic drugs by some RLS patients [40-42]. SKOR1 (LBXCOR1) is a SKI family transcriptional co-repressor 1. The homologous gene Corl1 in mice was highly selectively expressed in the central nervous system (CNS), interacted with Lbx1 and acted as a transcriptional co-repressor for Lbx1 in regulating cell fate determination in the dorsal spinal cord . No mutation screening has been reported for MAP2K5 and SKOR1 in RLS patients. In our study, SNP rs1026732, which is associated with RLS, is located between MAP2K5 and SKOR1, thus it is unknown which of the two genes is the causative gene for RLS. Future studies are needed to clarify the issue, and mutation sequencing of MAP2K5 and SKOR1 in RLS probands may find functional SNPs that are associated with RLS.
In conclusion, our population-based case control association studies have established that SNP rs2300478 in MEIS1, rs9357271 in BTBD9, and rs1026732 in MAP2K5/SKOR1 are associated with risk of RLS in the U.S. population. Interestingly, SNP rs1026732 in MAP2K5/SKOR1, but not SNP rs2300478 in MEIS1 and rs9357271 in BTBD9, appeared to be a risk locus for familial RLS in a family-based study. Future studies are needed to identify the true causative SNPs at these three RLS loci and to determine whether or not MEIS1, BTBD9, and MAP2K5/SKOR1 are causative genes for RLS.
Study funding: This study was supported in part by grants from the NIH (R01 HL094498), the China Scholarship Council, a Key Program of Hubei Natural Science Funds (2008CDA047), China National 863 Scientific Program (2006AA02Z476), the China National Basic Research Program (973 program 2007CB512002), and a Wuhan City Academic Leadership award (200951830560).
Financial disclosures and potential conflict of interest: Dr. W. G. Ondo is a speaker and consultant for Glaxo Smith Kline, Allergan, Ipsen, TEVA, Lundbeck, and Novartis. Dr. N. Foldvary receives research support from Glaxo Smith Kline. Other authors have nothing to report.
Author roles: Qinbo Yang was involved in the conception and execution of research plans, genotyping, writing of the first draft and help with statistical analysis. Lin Li was involved in statistical analysis. Qiuyun Chen was involved in conception and supervision of the research project, and review and critique of the manuscript. Nancy Foldvary-Schaefer was involved in ascertainment of study subjects and clinical phenotyping of study subjects. William G. Ondo was involved in conception of the research projects, review and critique of the manuscript, ascertainment of study subjects and clinical phenotyping of study subjects and supervision of the clinical aspects of the project. Qing Kenneth Wang was involved in overall supervision of the project, conception, design and organization of the project, obtaining funding and critical revision of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.