Exome sequencing study in patients with multiple sclerosis reveals variants associated with disease course

The discovery cohort comprised 20 MS patients classified according to their disease course into benign and aggressive phenotypes. Patients with benign phenotypes (n = 10) were defined as having an Expanded Disability Status Scale (EDSS) equal or lower than 3.0 after 15 or more years from disease onset [8] and never received MS therapies. Patients with aggressive disease courses (n = 10) reached an EDSS score equal or higher than 6.0 within the first 5 years after disease onset, regardless of treatment [9]. All patients included in the discovery cohort were recruited at the Centre d’Esclerosi Múltiple de Catalunya (Cemcat). Additional file 1: Table S1 summarizes demographic and main clinical characteristics of the discovery cohort.

Genomic DNA was extracted from peripheral blood using standard methods. An exome sequencing approach was applied to the discovery cohort in order to identify genes associated with benign and aggressive disease courses. Exome sequencing was based on an Illumina HiSeq2000 sequencing platform and an Agilent’s SureSelect Target Enrichment System for 51 Mb. Sequencing was done with a 50× of coverage and reads were aligned against the human reference genome (GRCh37/hg19 assembly) using the Burrows-Wheeler Alignment tool (BWA) [10]. After reads mapping, low-quality reads and sequences flagged as PCR duplicates were removed from the BAM file using the Sequence Alignment/Map (SAM) [10] and Picard Tools. Unmasked variants were annotated considering all possible transcripts for each target gene, and in some cases variants located within a coding sequence when considering one isoform could be positioned within a non-coding region when considering another isoform, thus resulting also in the identification of intronic variants. Exome sequencing was performed in Sistemas Genómicos (Valencia, Spain).

For the variant calling process, different algorithms were applied, including VarScan [11] and the Genome Analysis Toolkit (GATK) [12]. Python scripts were developed to combine variants. Variants annotation was based on Ensembl and NCBI databases. For the selection of significant variants, a Fisher exact test was applied to the benign and aggressive phenotypes. For prioritization and selection of the most promising variants, the following criteria were applied: (i) presence of two or more statistically significant variants per gene; (ii) odds ratio difference of the prevalence for the variant between disease phenotypes equal or higher than 2; (iii) absence of the variant in one disease phenotype and presence of the variant in ≥ 50% of patients belonging to the counterpart phenotype; (iv) missense variants, splice region variants, and variants reported as possible deleterious mutations; and (v) biological and functional relevance of the target genes to MS, as reported in the literature. A total of 16 independent variants satisfying 2 or more of the aforementioned criteria were selected for validation.

Two independent cohorts with benign and aggressive disease courses were included in the study in order to validate the selected variants from the exome sequencing approach.

The first validation cohort included 194 MS patients from 7 MS centers [Bilbao (n = 56); UCSF (n = 55); Madrid—Hospital Clínico (n = 32); Barcelona—Hospital Clinic (n = 23); Madrid—Ramón y Cajal (n = 16); Madrid—Puerta de Hierro (n = 9); Girona (n = 3)]. Of these, 107 MS patients had benign phenotypes and 87 aggressive disease courses.

The second validation cohort consisted of 257 MS patients from Canada, 224 patients with benign phenotypes and 33 with aggressive disease courses. MS patients were ascertained through the Canadian Collaborative Project on the Genetic Susceptibility to Multiple Sclerosis (CCPGSMS) [13].

Clinical criteria to classify patients into benign and aggressive disease courses were the same as those applied to the discovery cohort, except for the second validation cohort in which treatment information on patients with benign disease course was not available. Similar to the discovery cohort, patients with benign phenotypes from the first validation cohort never received MS therapies. A summary of demographic and clinical characteristics of the first and second validation cohorts is shown in Additional file 1: Table S1.

The study was approved by the corresponding local ethics committees, and all participants provided informed consent.

Genotyping of selected variants in the first validation cohort was performed using an OpenArray technology (Thermo Fisher Scientific, Massachusetts, USA) and following the manufacturer’s instructions. Briefly, DNA samples were loaded into custom designed arrays using an OpenArray® AccuFill System (Thermo Fisher Scientific). QuantStudio™ 12K Flex system (Thermo Fisher Scientific) was used for sample amplification and fluorescent data collection. Hapmap samples with known genotype were included as internal controls of the process. Genotype was assigned using Taqman Genotyper Software (Thermo Fisher Scientific). Genotyping was performed by the Human Genotyping laboratory of the Spanish National Cancer Research Centre (CNIO).

In the second validation cohort, selected variants were genotyped using a MassArray iPLEX platform (Sequenom, San Diego, CA, USA) as previously described [14].

Gene expression levels for CPMX2, NLRP9, and IGSF9B were determined by real-time PCR in peripheral blood mononuclear cells (PBMC) available from a subgroup of untreated MS patients from the first validation cohort. In order to avoid a confounding effect of disease course in the expression levels for these genes, analysis was restricted to the group of patients with aggressive disease course (n = 8 for CPXM2; n = 9 for NLRP9; n = 7 for IGSF9B). Briefly, PBMC were isolated by Ficoll-Isopaque density gradient centrifugation (Gibco BRL, Life Technologies LTD, Paisley, UK) and stored in liquid nitrogen until used. Total RNA was extracted from PBMC using TRIzol® reagent (Invitrogen, Carlsbad, CA) and cDNA synthesized using the High Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA, USA). Messenger RNA expression levels for CPMX2, NLRP9, and IGSF9B were determined by real-time PCR using TaqMan® probes specific for each gene (Applied Biosystems, Foster City, CA, USA). The housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an endogenous control (Applied Biosystems). Assays were run on the ABI PRISM® 7900HT system (Applied Biosystems) and data were analyzed using the 2^−∆∆CT method [15]. Results were expressed as fold change in gene expression in MS patients carrying the risk allele relative to non-carrier patients.

Paraffin-embedded brain samples of chronic active lesions from four MS patients were provided by the UK Multiple Sclerosis Tissue Bank and stained with hematoxylin and eosin (HE) and Klüver-Barrera (KB) for inflammation and demyelination assessment. Four-micrometer-thick, paraffin-embedded serial sections were deparaffined in xylene and rehydrated in alcohol. Endogenous peroxidase activity was blocked with hydrogen peroxide (2%), methanol (70%), and PBS for 20 min. Antigen retrieval was performed in TE buffer (1 M TrismaBase and 1 mM EDTA) (pH = 9) in the microwave. Non-specific protein binding was blocked with 0.2% of bovine albumin (BSA) in PBS. Sections were incubated overnight at 4 °C with the following primary antibodies: rabbit anti-CPXM2 (Biorbyt), rabbit anti-NLRP9 (Abcam), and rabbit anti-IGSG9B (Abcam). Samples were incubated for 1 h at room temperature with goat-anti rabbit HRP secondary antibody (Dakocytomation) and stainings were visualized with 3,3′diaminobenzidine (Sigma, St Louis, MO, USA) as a chromogenic substrate.

A flowchart summarizing the main steps of the study design and analysis is depicted in Fig. 1. In order to identify genetic variants associated with disease course, we first performed exome sequencing in an initial cohort of 20 MS patients classified according to their disease course into benign (n = 10) and aggressive (n = 10). A total of 915 single-nucleotide polymorphisms (SNPs) were differentially distributed between MS patients with benign and aggressive disease courses (uncorrected p values < 0.05; data not shown). From the list of differentially distributed SNPs and after applying the selection criteria described in the “Methods” section, 16 SNPs were chosen for validation in two validation cohorts. Table 1 shows a description of the 16 selected SNPs from the discovery cohort, and Table 2 provides the results of the exome sequencing analysis for these 16 SNPs.

Selected SNPs were genotyped in two independent validation cohorts composed of MS patients classified according to their disease course. A total of 194 MS patients were included in the first validation cohort and 257 MS patients in the second validation cohort. Meta-analysis of genotyping results in both cohorts identified two SNPs, rs28469012 and rs10894768, significantly associated with MS disease course. As shown in Fig. 2, the minor allele for rs28469012, an intronic SNP positioned in CPXM2 (carboxypeptidase X, M14 family, member 2), was nominally significantly associated with more aggressive disease course [p-value; odds ratio (95% confidence interval): p = 0.04; 1.81 (1.03–3.18)] (Fig. 2). In this case, a modest degree of heterogeneity was found between both validation cohorts.

The minor allele for rs10894768, which corresponds to a synonymous SNP located in IGSF9B (immunoglobulin superfamily member 9B), was nominally significantly associated with benign disease course [p = 0.04; 0.70 (0.49–0.99)]. Meta-analysis of this variant supported the originally identified association with homogeneous outcomes for both cohorts (I² = 0).

A trend for significance was observed in a third SNP, rs10423927, an intronic variant located in NLRP9 (NLR family pyrin domain containing 9), with the minor allele associated with benign disease course [p = 0.09; 0.58 (0.31–1.09)]. Similar to the IGSF9B variant, meta-analysis for rs10423927 also supported the association in the discovery cohort and resulted in similar outcomes in both validation cohorts (I² = 0).

It should be noted that rs2374639, a synonymous variant positioned in VPS50 (VPS50, EARP/GARPII complex subunit), showed a significant association in the second validation cohort with the same protective effect observed in the discovery cohort [p = 0.04; 0.47 (0.23–0.95)]. However, the heterogeneity found between both validation cohorts did not allow to perform a joint analysis (Table 2; I² = 80%).

Genotyping frequencies and results of statistical analysis in the two validation cohorts for all selected variants are provided in Table 2.

In order to explore the functional consequences of the polymorphisms associated with MS disease course, we first investigated whether mRNA expression levels for IGSF9B, CPXM2, and NLRP9 in PBMC differed between MS patients carrying the minor allele associated with disease course and non-carrier patients. CPXM2 and NLRP9 expression was not detected in PBMC from MS patients (data not shown). IGSF9B expression was detected in PBMC from MS patients, although no differences were observed between carriers and non-carriers of the minor allele associated with benign disease course (Additional file 1: Figure S1).

In order to determine whether the selected genetic variants associated with disease course could be modifying the expression of nearby genes in the region, we analyzed cis-expression quantitative trait loci (eQTLs) in 48 tissues from the Genotype-Tissue Expression (GTEx) project [16], lymphoblastoid cell lines in 465 individuals from the Geuvadis project [17], and trans-eQTLs from expression data of 9196 tumor samples in 33 cancer types from the PancanQTL study [18]. We observed correlation of the rs10894768 genotypes with IGSF9B expression in pancreatic tissue (p = 3.6 × 10^− 6) from 171 individuals and thyroid tissue (p = 5.6 × 10^− 11) from 323 individuals from the GTEx study [16]. The minor allele for this polymorphism, which was more represented in patients with benign disease course, was associated with lower IGSF9B expression (Additional file 1: Figure S2). Of note, IGSF9B was highly expressed in cerebellar hemisphere, cerebellum and also in cortex, hypothalamus, and spinal cord; however, no eQTLs were described for these tissues most likely due to the lower number of samples analyzed in brain tissue (105 samples from cerebellar hemisphere and 125 from cerebellum).

Based on the negative results observed for IGSF9B, CPXM2, and NLRP9 expression in peripheral blood, we wondered whether selected genes associated with MS disease course could play more specific roles at the CNS level by investigating their expression in brain tissue from patients. Figure 3 depicts the results of IGSF9B, CPXM2, and NLRP9 immunohistochemistries in brain tissue sections from chronic active lesions of MS patients. IGSF9B showed a diffuse neutrophil staining in both the gray and the white matter. The most intense IGSF9B staining was detected in the cytoplasm of microglia/macrophages and astrocytes located at the margins of MS lesions, where inflammatory activity is highest. IGSF9B immunostaining was also observed in less active areas of the lesions, albeit to a lesser degree compared to more active areas. CPXM2 and NLRP9 exhibited a similar and very discrete immunostaining. Their expression was restricted to the cytoplasm of few microglia/macrophages located at the margins of MS lesions showing the highest inflammatory activity (Fig. 3).