Implication of genetic variants in primary microRNA processing sites in the risk of multiple sclerosis

Abstract

Background: Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system with a well-established genetic contribution to susceptibility. Over 200 genetic regions have been linked to the inherited risk of developing MS, but the disease-causing variants and their functional effects at the molecular level are still largely unresolved. We hypothesised that MS-associated single-nucleotide polymorphisms (SNPs) affect the recognition and enzymatic cleavage of primary microRNAs (pri-miRNAs).

Methods: Our study focused on 11 pri-miRNAs (9 primate-specific) that are encoded in genetic risk loci for MS. The levels of mature miRNAs and potential isoforms (isomiRs) produced from those pri-miRNAs were measured in B cells obtained from the peripheral blood of 63 MS patients and 28 healthy controls. We tested for associations between SNP genotypes and miRNA expression in cis using quantitative trait locus (cis-miR-eQTL) analyses. Genetic effects on miRNA stem-loop processing efficiency were verified using luciferase reporter assays. Potential direct miRNA target genes were identified by transcriptome profiling and computational binding site assessment.

Findings: Mature miRNAs and isomiRs from hsa-mir-26a-2, hsa-mir-199a-1, hsa-mir-4304, hsa-mir-4423, hsa-mir-4464 and hsa-mir-4492 could be detected in all B-cell samples. When MS patient subgroups were compared with healthy controls, a significant differential expression was observed for miRNAs from the 5' and 3' strands of hsa-mir-26a-2 and hsa-mir-199a-1. The cis-miR-eQTL analyses and reporter assays pointed to a slightly more efficient Drosha-mediated processing of hsa-mir-199a-1 when the MS risk allele T of SNP rs1005039 is present. On the other hand, the MS risk allele A of SNP rs817478, which substitutes the first C in a CNNC sequence motif, was found to cause a markedly lower efficiency in the processing of hsa-mir-4423. Overexpression of hsa-mir-199a-1 inhibited the expression of 60 protein-coding genes, including IRAK2, MIF, TNFRSF12A and TRAF1. The only target gene identified for hsa-mir-4423 was TMEM47.

Interpretation: We found that MS-associated SNPs in sequence determinants of pri-miRNA processing can affect the expression of mature miRNAs. Our findings complement the existing literature on the dysregulation of miRNAs in MS. Further studies on the maturation and function of miRNAs in different cell types and tissues may help to gain a more detailed functional understanding of the genetic basis of MS.

Funding: This study was funded by the Rostock University Medical Center (FORUN program, grant: 889002), Sanofi Genzyme (grant: GZ-2016-11560) and Merck Serono GmbH (Darmstadt, Germany, an affiliate of Merck KGaA, CrossRef Funder ID: 10.13039/100009945, grant: 4501860307). NB was supported by the Stiftung der Deutschen Wirtschaft (sdw) and the FAZIT foundation. EP was supported by the Landesgraduiertenförderung Mecklenburg-Vorpommern.

Keywords: B cells; Expression quantitative trait loci; Genetic risk; Immune reconstitution therapy; MicroRNA target genes; MicroRNAs; Multiple sclerosis; Primary microRNA processing; Single-nucleotide polymorphisms.

Figure 1

Overview of the study. This study comprised three parts: First, primary miRNAs encoded in genetic risk loci for MS were identified by a database-driven approach. Second, DNA, B cells and B-cell RNA were collected from MS patients as well as from healthy controls. Flow cytometry was used to phenotype B-cell subpopulations. The expression levels of mature miRNAs and isomiRs were measured using specific qPCR assays. The subjects were genotyped with respect to MS-associated SNPs within the miRNA-coding regions. Third, the effect of the SNPs on the enzymatic cleavage of the primary miRNAs by the Microprocessor complex was investigated using luciferase reporter assays. Transcriptome profiling was used to identify miRNA target genes. MIR=microRNA, MS=multiple sclerosis, qPCR=quantitative polymerase chain reaction, SNP=single-nucleotide polymorphism.

Figure 2

Differential expression of microRNAs in B cells from MS patients and healthy controls. The expression of 10 mature miRNAs and isomiRs is visualised for the six study groups. These miRNAs could be detected in >50% of the samples and are derived from 6 primary miRNA transcripts from MS-associated genetic regions. For this plot, we always selected the 5p and 3p miRNA forms (canonical miRNA or isomiR) that had the lowest P-value in the group comparisons (Table 3). The expression was quantified relative to the reference miRNA hsa-miR-191-5p. Higher data points indicate higher expression levels. The y-axis on the left displays ΔC_T values in an inverted manner and the y-axis on the right displays the data converted in linear scale (2^−ΔCT×1000). Black horizontal lines indicate the means per group. Significance values <0.05 from pairwise Tukey post hoc tests are shown above the brackets. IRT=immune reconstitution therapy, miRNA=microRNA, MS=multiple sclerosis, n=number, PPMS=primary progressive multiple sclerosis, RRMS=relapsing-remitting multiple sclerosis.

Figure 3

Levels of hsa-miR-199a-5p and -3p as well as hsa-miR-4423-5p and -3p per genotype. The B-cell expression levels of 4 miRBase-annotated mature miRNAs are visualised. (a and b) A non-significant tendency towards higher expression of hsa-miR-199a-5p and -3p was noted in carriers of the MS risk allele (RA) of SNP rs1005039. (c and d) The RA of SNP rs817478 was associated with a lower expression of hsa-miR-4423-5p in the cis-miR-eQTL analysis (Table 4). Accordingly, the respective SNPs were suspected to affect the miRNA stem-loop processing by Drosha. The number of samples from subjects carrying 0, 1 or 2 RA is indicated below each beeswarm/violin plot. The left and right y-axes refer to the raw qPCR data (ΔC_T values displayed in an inverted manner) and the converted data (2^−ΔCT×1,000), respectively. The means per genotype group are shown as black lines. The Tukey test P-value reaching the significance level of α=0.05 is indicated above the bracket. eQTL=expression quantitative trait locus, miRNA=microRNA, MS=multiple sclerosis, n=number, qPCR=quantitative polymerase chain reaction, SNP=single-nucleotide polymorphism.

Figure 4

Allele-specific processing of microRNA precursor sequences in the reporter assay. (a) The predicted secondary structures of the miRNA stem-loops were visualised using the web-based tool forna. Both SNPs are located in a CNNC motif in the 3’ flanking region. The MS risk allele (RA) is marked in red. The Drosha cleavage sites are indicated according to the miRBase annotation. (b and c) The processing of hsa-mir-199a-1 and hsa-mir-4423 was analysed using a luciferase-based assay. For this purpose, the miRNA precursor sequences were cloned into the 3’ UTR of the GLuc gene within a reporter vector. GLuc/SEAP ratios were then measured after transient transfection of HeLa cells with plasmids carrying either the MS risk allele or the alternative allele. The decrease in relative luminescence was calculated by subtracting from 1 the quotient of the GLuc/SEAP ratio and the respective ratio that was obtained for the negative control. Higher bars thus indicate higher processing efficiencies. The bars and error bars show the means and standard deviations of 3 biological replicates each. Welch t-test P-values are given above the bars. The MS RA of the SNP rs1005039 conferred a higher hsa-mir-199a-1 cleavage efficiency (b), whereas the RA of SNP rs817478 was associated with a significantly diminished processing of hsa-mir-4423, independently of the amount of plasmid DNA used for transfection and the readout time point (c). ANOVA=analysis of variance (3-way additive model), GLuc=Gaussia luciferase, miRNA=microRNA, MS=multiple sclerosis, SEAP=secreted alkaline phosphatase, SNP=single-nucleotide polymorphism, UTR=untranslated region.