Restless legs syndrome-associated intronic common variant in Meis1 alters enhancer function in the developing telencephalon

Abstract

Genome-wide association studies (GWAS) identified the MEIS1 locus for Restless Legs Syndrome (RLS), but causal single nucleotide polymorphisms (SNPs) and their functional relevance remain unknown. This locus contains a large number of highly conserved noncoding regions (HCNRs) potentially functioning as cis-regulatory modules. We analyzed these HCNRs for allele-dependent enhancer activity in zebrafish and mice and found that the risk allele of the lead SNP rs12469063 reduces enhancer activity in the Meis1 expression domain of the murine embryonic ganglionic eminences (GE). CREB1 binds this enhancer and rs12469063 affects its binding in vitro. In addition, MEIS1 target genes suggest a role in the specification of neuronal progenitors in the GE, and heterozygous Meis1-deficient mice exhibit hyperactivity, resembling the RLS phenotype. Thus, in vivo and in vitro analysis of a common SNP with small effect size showed allele-dependent function in the prospective basal ganglia representing the first neurodevelopmental region implicated in RLS.

Figure 1.

Association signals, sequence conservation, and epigenetic signatures in embryonic and adult stages of the MEIS1/Meis1 locus (A) RLS-associated locus on 2p14 with the lead SNP rs12469063 (violet diamond) and its genetic environment. (B) The x-axis represents the genomic position, referring to the hg18 genome annotation. The right-hand y-axis represents the recombination rate, and the negative log10 of the nominal P-values of all SNPs genotyped are given on the left-hand y-axis. The color-coded LD between SNPs is based on the lead SNP rs12469063 (violet diamond) as a reference. rs2300478, which lies in a nonconserved region, is depicted with a red circle. Recombination rate and r² values were estimated with the CEU population from HapMap II (release 22). The LD block below is based on HapMap data, measured in r² and visualized with Haploview. Highlighting refers to the magnitude of pairwise LD, with a white-to-black shading indicating lower to higher LD values. The lower part of Figure 1B uses mouse genome annotation mm9 and lists from top to bottom: (1) conservation between the mouse genome and the genome of the rat, human, frog, zebrafish, and Fugu; (2) corresponding localization of the human RLS-SNPs rs12469063 (violet diamond) and rs2300478 (red circle); (3) seven analyzed HCNRs in turquoise boxes; (4) the conservation in mammals (dark blue); (5) DNA binding of EP300 (ChIP-seq) in E11.5 forebrain (blue) and midbrain (gray); and (6) DNase I hypersensitivity in whole brain of E14.5 embryos and adult mice (red and orange peaks, respectively).

Figure 2.

In vivo allele-specific enhancer function of HCNR 617 in zebrafish. (A) Results of the zebrafish enhancer screen with the expression domains for the protective and risk allele in orange and purple, respectively. (B) Representative embryos of the F1 generation with the protective ([A] adenine) and the risk allele ([G] = guanine) construct of HCNR 617 showed an allele-specific difference in reporter EGFP expression in the neural tube. The protective allele of HCNR 617 drove EGFP expression in the retina, fore-, mid-, and hindbrain and spinal cord. The risk allele significantly reduced the expression almost to background levels (bar represents 250 µm). (C) Boxplot (left) The risk allele reduced the fluorescence intensity to 70% compared to the protective allele (100%) (Wilcoxon rank sum test, P = 0.08; EGFP-brain signal measured, n_protective = 5, n_risk = 5). The bar chart (right) represents the percentage of all EGFP-positive founders in the respective area (y-axis) with respect to the protective (orange) and risk (purple) allele of HCNR 617. The risk allele leads either to a significant reduction or a complete loss of the EGFP signal.

Figure 3.

In vivo and in vitro identification of allele-specific HCNR 617 enhancer function in the ganglionic eminences of the mouse. (A) Representative transgenic mouse embryos after pronucleus injection of HCNR 617 protective (left) and risk allele constructs (right) at stage E12.5. Blue color indicates regions expressing the reporter gene beta-galactosidase (lacZ). Arrowhead indicates the reproducible telencephalic signal (bar represents 2.5 mm). (B) Analysis of the signal intensity in Theiler-staged embryos (stages 19 and 20; n_protective = 4, n_risk = 4) showed a reduction down to 35% (Wilcoxon rank sum test, P = 0.029). Stereological volume estimation according to Cavalieri revealed a significant volume reduction down to 24% for the risk allele (Wilcoxon rank sum test, P = 0.029). (C) Electrophoretic mobility shift assay using oligonucleotides encompassing rs12469063 showed allele-specific differences of DNA-protein complex formation using E12.5 forebrain nuclear extract (lanes 3 and 4, arrow and arrowhead). Specificity is proven by competition with unlabeled risk and protective oligonucleotide in excess (lanes 6–18). CTRL, control (no protein). (D) Frontal sections through the forebrain reveal lacZ reporter activity in the mantle zone (mz) of the ganglionic eminences, in the same region as Meis1 transcript and MEIS1 protein (arrowheads; bar represents 500 µm); no transcripts are visible in the ventricular (vz) and subventricular (svz) zones. (E) Transcripts of all additional GWAS-RLS risk loci map to the embryonic ganglionic eminences. mRNA of Ptprd, Btbd9, and Map2k5 was detected in the mantle zone (mz) of the LGE and MGE. Tox3 was expressed in the adjacent ventricular (vz) and subventricular (svz) zone (arrowheads; bar represents 500 µm).

Figure 4.

rs12469063 alters a CREB1 transcription factor binding site. (A) Principle work flow for the identification of upstream binding factors by affinity chromatography and mass spectrometry. (B) CREB consensus sequence compared to the human sequence spanning rs12469063. (C,D) Creb/CREB and Meis1/MEIS1 were both detected in the mantle zone of the ganglionic eminences and show colocalization (bar in C represents 500 µm; bar in D represents 250 µm). (D) Magnification of rectangle in C. (E) Supershift EMSA assay with a specific antibody against CREB showed significant reduction of DNA-protein complex formation compared to unspecific IgG antibody (control) for overexpressed CREB1 in 293T nuclear cell lysate (lanes 1–6) and E12.5 forebrain nuclear extract (lanes 7–12, arrowhead). Complete abolishment of specific binding with an oligonucleotide, in which the entire CREB motif is deleted (lane 13). CREB consensus oligonucleotide shows specific binding, being absent with the mutated CREB consensus oligonucleotide (lanes 14,15). Arrow indicates additional allele-specific band.

Figure 5.

MEIS1 transcriptional target genes in the E12.5 ganglionic eminences. Summarized heat map from unpaired two class analysis (SAM) of genes regulated between heterozygous Meis1^tm1Mtor and wild-type forebrain tissues at E12.5. Expression profiles of male and female embryos clearly separate. Similar gene profiles are grouped together (rows) and color code gives the mean fold changes of the respective genes for each mutant embryo. Yellow represents up-regulation and blue represents down-regulation in comparison to the respective wild-type control group. The number of genes that are included in each group (A–G) with similar patterns is given. Individual genes are listed in Supplemental Table S4.

Figure 6.

Motor restlessness/hyperactivity in adult heterozygous Meis1^tm1Mtor mice. Heterozygous Meis1^tm1Mtor (het) traveled a higher total distance (P = 0.003, two-tailed t-tests, n = 29) (A) and moved with a higher speed on average (P = 0.004, two-tailed t-tests, n = 30) (B) in a 20-min open field test for spontaneous locomotor activity in a novel environment. (C) Oxygen consumption (mL/h) plotted versus body mass (g). Scatterplot with regression lines split by sex and genotype to adjust for body mass variation. The shift in regression lines indicates higher energy turnover in heterozygous Meis1^tm1Mtor mice (linear regression model, P = 0.0005). (D) Mean distance traveled (cm/20 min) monitored by infrared light beams during the indirect calorimetry trial. Heterozygous Meis1^tm1Mtor mice tended to show increased locomotor activity under home cage conditions (two-way ANOVA, P = 0.06).