The 12q13.3–12q14.1 region has been associated to several autoimmune diseases8
but no causal variant has been found so far for any of them exception made for a rare variant in the coding region of CYP27B1
gene associated with MS.17
This variant was not detected in our sample set.
In the present work, by integrating several strategies, we have uncovered the variant most strongly associated with MS in the region. This variant also affects the activity of an enhancer in an allele-dependent and orientation-dependent manner and correlates with the expression of five genes of the locus. Interestingly, the major allele of this SNP (rs10877013) correlates with high luciferase activity and at the same time with high expression of TSFM
and low expression of FAM119B, CYP27B1
Although it could seem contradictory that one enhancer may differentially affect the expression of different genes, parallel situations have been described extensively in two scenarios: first, the same transcription factor is capable of activating or repressing depending on the context;32
second, a single DNA sequence can be bound by different factors with opposite activities.34
On the other hand, we cannot forget the important role of H3K4me preventing aberrant gene expression or modulating transcriptional response. H3K4me turns to mark for both positive and negative regulation. The role of H3K4me2 in histones deacetylation could serve as a fine-tuning regulatory mechanism.35
It has been described that interacting promotes are coexpressed in a tissue-specific manner. Based on the Chromatin Interaction Analysis with Paired-End-Tag sequencing results mapping long-range chromatin interactions associated with RNAPII and H3K4me2,36
we observed that the genes in the analysed region are engaged through promoter–promoter, promoter–enhancer interactions forming a multigene complex that could provide a structural framework for cotranscription (). In this sense, the effect of a polymorphism on one enhancer embedded in this multigene complex could have diverging effects depending on the interaction of this enhancer with the different promoters. Additionally, the interactions could be tissue-type or cell-type specific, as has been shown by Li et al.36
Therefore, it is not evident which of the genes in the region may be responsible for the genetic association with MS, since they seem to be coordinately regulated, and the enhancer containing the MS-associated variant may be affecting either of them; in fact, it acts as an eQTL for all of them. We observed that the Spearman's correlation coefficients (r) between the expression levels of the different genes in this cluster and the rs6581155-tagged SNPs were different in the different assays performed. For instance, in monocytes the best correlation is observed with FAM119B
(r=−0.52), however in data obtained from LCL by RNA-seq a medium-high correlation coefficient is found for both CYP27B1
(r=−0.472) and FAM119B
(r=−0.474), and with microarrays the correlation coefficient is very similar for TSFM
. Based on data from LCL cells, TSFM, FAM119B
would be candidates, while FAM119B
appears to emerge from monocytes. In fact, Gandhi et al20
reported that FAM119B
expression level was much lower in MS patients carrying the susceptibility haplotype. We observed that this correlation occurs also in samples from healthy donors as they are the CEU HapMap samples. However, we do not know the precise conditions (cell type, stage of disease, etc) under which expression of these genes could be determinant for the development of MS. For the same reason the variant-specific enhancer activity of fragment III, reported in this work, could vary in different cell-types. Here we have used in the activity assays the same cell-type used in the eQTL analysis to reproduce the effect observed in the expression of these genes. The Raji cells are lymphoblast-like cells derived from B-lymphocyte as the LCL from HapMap. The effect of the enhancer on the gene expression of proximal tubule cells of the kidney and the disease-activated macrophages, which are the major source of CYP27B1,38
could shed light on the involvement of this variant in MS.
Figure 3 Datasets on Long-Range Chromatin Interaction with Paired-End-Tag sequencing (ChIA-PET) in the studied region. (A) Coordinates of chr12 in B37 and known genes by University of California Santa Cruz (UCSC) browser. (B) H3K4me2 ChIA-PET data obtained from (more ...)
Relying on published data from other groups on the concentration of precursors of vitamin D in serum, two independent groups reported a correlation between the major alleles of rs703842, tagged by rs6581155 in the present work, with high 25-hydroxyvitamin D (25(OH)D) concentrations in MS patients.39
On the other hand, Lange et al41
reported a correlation of the major allele of rs10877012, in total LD with rs6581155, and low 1,25(OH)2
D serum levels. The major allele of the MS-associated SNPs, in total LD with the previously mentioned SNPs, correlated with low expression of the CYP27B1
gene. This enzyme catalyses the conversion of 25(OH)D to 1,25(OH)2
D, thus a low expression of this enzyme, could reconcile the above seemingly conflicting data, via accumulation of 25(OH)D and a low 1,25(OH)2
D product. These data support the CYP27B1
gene as a very plausible candidate gene for association of the region with MS.
Vitamin D, besides having well-known control functions of calcium and phosphorus metabolism, bone formation and mineralisation, also has a complex role in the maintenance of immune homeostasis. The administration of vitamin D in animal models leads to improvement of immune-mediated symptoms. The correlation between the MS- associated variants and circulating levels of vitamin D supports the important role of vitamin D in susceptibility to MS. Thus, if this is the case in human MS, both the pharmacogenomics and the expression analysis of all the genes affecting the causing factor can reveal the kind of intervention necessary for neutralising a pathogenic situation.42
In this work we localise a group of variants in almost total LD that explain the association of the 12q13.3–12q14.1 region with MS. One of these variants (rs10877013) strongly affects the activity of one enhancer in the region, in an allele-dependent and orientation-dependent way, that could be the cause of the alteration of TSFM, TSPAN31, FAM119B, CYP27B1 and AVIL gene expression, due to the promoter multigene interactions observed in the region. Some of the tagged variants have been described to be associated with the vitamin D serum concentration. Our results point to one of these as the causal gene for MS association in the locus.