Fig 1.
A low-frequency SNP within MERTK is significantly associated with MS susceptibility.
124 SNPs on chromosome 2 were directly genotyped in 3268 MS cases and 3579 controls. (A) The negative log of the unadjusted p-value of each SNP is plotted against the relative position on chromosome 2, with a schematic of the gene structures shown underneath (B). Two SNPs, both within MERTK, reach significance using the Sidak adjustment for multiple testing at a nominal α = 0.05. Association p-values were determined using a Chi-square test. (C) Schematic of the pattern of LD across the whole region, with a large single block of high LD (D'>0.99), including the whole of the MERTK gene, highlighted with a red triangle.
Table 1.
Top 10 SNPs within MERTK associated with Multiple Sclerosis susceptibility.
Table 2.
Gene model association of rs13414207 with Multiple Sclerosis.
Fig 2.
SNPs within MERTK define both risk and protective haplotypes associated with MS susceptibility.
28 SNPs within the MERTK gene form a single block of very high LD (D'>0.99, LOD≥2). The five most frequent haplotypes (population frequency >1%) are shown in this schematic, along with the p-value of association of each haplotype with MS susceptibility as determined using a Chi-square test. Arrowhead indicates the haplotype-tagging allele of rs13414207 in haplotype 5. The alleles presented for rs17174870 are inferred from data obtained from sequencing the opposite strand but are presented as CT to maintain consistency with the remainder of the study.
Table 3.
Individuals used for resequencing of MERTK.
Fig 3.
Heat map of variants found in MERTK grouped according to haplotype.
The sequence of each group following resequencing was compared with the reference genome (GRCh37/hg19). Coloured lines indicate a base that is variant compared with the reference genome. Mapping of the groups shows that haplotype groups 3 and 4 were most closely related to the reference sequence, showing many invariant nucleotides. Conversely haplotype groups 2 and 5 showed the greatest differences to the reference sequence and an apparently close relationship, sharing many variants.
Table 4.
Summary of variants identified in MERTK.
Fig 4.
The haplotype structure of variants identified in MERTK.
The 52 variants in MERTK genotyped for association with MS susceptibility fall into 4 separate blocks. Haplotype blocks are connected with thick lines if connections are observed in >10% samples and thin lines if connections are observed in >1% samples. A schematic of the MERTK gene is shown underneath indicating the relationship of the blocks to the physical structure of MERTK. The haplotypes coloured in red are significantly associated with MS susceptibility (p<0.05) and the haplotype coloured in blue is associated with protection (p<0.05). The p-value of association was determined using a Chi-square test. #This variant represents a tri-nucleotide in-del (T/- = TGG; -)
Table 5.
Results of the association testing for variants identified in MERTK.
Fig 5.
Discordant effect of rs7422195 in the presence or absence of HLA-DRB1*15:01.
All samples (n = 3000) were first stratified according to the number of DR15 alleles then by genotype at rs7422195. (A) The frequency of the A-allele of rs7422195 was calculated for cases and controls within each DR15 genotype group, showing a clear decrease in the frequency of the A-allele with increasing copies of DR15 within MS cases, and the opposite effect in healthy controls. The total number of samples within each DR15 genotype group is included below the group name on the x-axis, with the number of individuals used to calculate each point represented in brackets on the graph. (B) Disease frequency for each group was calculated as the number of MS cases divided by the total number of cases and healthy controls for each genotype. The minor allele at rs7422195 shows an increase in the disease risk in the absence of DR15, but a clear decrease in the disease frequency amongst individuals carrying two copies of DR15.
Table 6.
Association of MERTK SNPs in populations stratified by DR15 status.
Fig 6.
Expression of MERTK in monocytes is genotype dependent.
Immune cell subsets were purified from peripheral blood mononuclear cells obtained from MS cases and healthy controls using magnetic cell sorting. (A) Gene expression was measured using the Affymetrix Human ST1.0 array. The genotype for each participant was determined using Immunochip (Illumina) and expression data stratified by genotype at rs17174870. Expression of the MERTK gene in monocytes was significantly associated with genotype (p = 2.215 x 10−5, padj = 0.006) but not with phenotype (p>0.05). We then stratified individuals carrying various combinations of the haplotype blocks identified in association testing and stratified expression for individuals homozygous for those in block 2 (B) and block 3 (C), with the tag-SNP and relevant allele shown for each haplotype. (B) Individuals homozygous for the haplotype tagged by rs17174870(T) showed significantly lower MERTK expression compared with haplotypes A or B (p<0.0001). (C) Individuals homozygous for the haplotype tagged by rs7422195(A) showed significantly increased expression of MERTK compared with the haplotype tagged by the alternative (G) allele. (D) When samples are stratified by rs7422195 expression of MERTK is increased with increased copies of the minor (A) allele [Mean expression±SD: 8.654±0.369 (GG), 9.012±0.5514 (AG) p<0.001 vs GG, 9.186±0.3434 (AA) p<0.0001 vs GG] (E) When samples are stratified by rs56361454 expression of MERTK is increased with increased copies of the minor (T) allele [Mean expression±SD: 8.655±0.4489 (CC), 9.041±0.5529 (CT) p<0.0001 vs CC, 9.139±0.3603 (TT) p<0.01 vs CC]. (F) Schematic of the genomic region surrounding rs7422195 and rs56361454 showing transcription factor binding sites and DNase sensitive regions. MERTK expression on the surface of monocytes was determined using flow cytometric analysis of whole blood and correlated with MERTK gene expression. MERTK surface expression was significantly correlated with gene expression in (G) CD14Hi (r2 = 0.3434, p = 0.0066) and (H) CD14Lo (r2 = 0.3624, p = 0.005) monocytes. (I) The proportion of CD14HiMERTKHi monocytes is increased with increasing dose of the minor (A) allele of rs7422195 (p<0.01 GG vs AA). All grouped expression data are plotted as Tukey box and whiskers.
Table 7.
rs17174870 genotype dependent expression of MERTK in immune cell subtypes.
Fig 7.
The MERTK gene contains a novel alternative final exon but usage is not genotype-dependent.
(A) Exon junctions in RNA sequencing data were visualised using IGV_2.3.35. Representative junction maps for a low (rs7422195 GG) and high (rs7422195 AA) expressing sample are shown. Each shows the presence of a junction from exon 18 to a putative novel alternative final exon excluding exon 19. (B) RT-PCR analysis of a high and low expressing sample show the presence of this exon only in combination with exon 18 and not exon 19. (C) Exon usage analysis shows overall increased expression of each exon in samples homozygous for the rs7422195 minor (A) allele, but no difference in exon usage following normalisation for overall expression level (D).
Fig 8.
Disease course is altered in the presence of MERTK susceptibility-associated variants.
Individuals initially presenting with a relapsing-remitting course of MS were stratified by both DR15 status and genotype at rs7422195. (A) In the presence of the minor (A) allele of rs7422195, DR15 negative individuals (n = 370) showed a strong trend towards increased probability of progression (p = 0.07) (B) In the presence of the major (G) allele of rs7422195 DR15 homozygous individuals (n = 68) showed a strong trend towards increased probability of progression (p = 0.081)