1

Ottoboni et al.

Supplemental Data

Clinical relevance and functional consequences of the TNFRSF1A multiple sclerosis

locus.

Linda Ottoboni PhD,*, § Irene Y. Frohlich MPH,*, § Michelle Lee BSc,*, § Brian C. Healy PhD,¶ Brendan T. Keenan MS,*, § Zongqi Xia MD,*, § Tanuja Chitnis MD,¶ Charles R. Guttmann MD,† Samia J. Khoury MD,¶ Howard L. Weiner MD,¶ David A. Hafler MD#, ‡ and Philip L. De Jager MD*, §, ¶, ‡.

Index

§  Supplementary Material and Methods (p. 2)

§  Supplemental Figures (p. 11)

§  Supplemental References (p. 21)


Material and Methods

Subjects with MS

The MS subjects used in this study are of European ancestry and are part of the Comprehensive Longitudinal Investigation of MS at the Brigham & Women’s Hospital (CLIMB). All subjects were older than 18 years of age and met criteria of either MS using the revised McDonald diagnostic criteria e-1 or clinically isolated syndrome. Untreated subjects are defined as having no disease-modifying treatment in the preceding 6 weeks for glatiramer acetate (GA) or interferon–β (IFNβ), the preceding 4 weeks for steroids, or the preceding one year for cytotoxic, monoclonal or investigational agents. Clinical data is collected prospectively at six months intervals, and the patients undergo magnetic resonance imaging (MRI) yearly. Details of the CLIMB study design have been published previously 13. Demographic details of these subjects are presented in supplemental Table e-1. The expanded disability status scale (EDSS) at the time of blood sampling and the year of first symptoms were used to calculate the MS severity score (MSSS) for individual subjects e-2. If the disease duration was <3 years, an MSSS was not calculated. Prospectively collected serum samples from 215 of these MS subjects were analyzed in this study. A small number of healthy controls (n=47, spouses or friends of subjects with MS) serum samples were collected in the same time frame as the MS subjects. All samples were stored frozen at -80°C, and each aliquot had been thawed no more than twice before use.

The BWH PhenoGenetic Project

This collection of healthy subjects was designed to provide a pool of subjects with which to facilitate the functional dissection of genetic variants; it currently has over 1700 subjects. Subjects are genotyped for the variant of interest, and individuals are recalled based on their genotype. In this study, only subjects of self-reported European ancestry were used. A bank of frozen serum samples (maintained at -80°C) was accessed for the serum characterization of healthy subjects. For the monocyte experiments detailed below, a coordinator recalls the selected subjects, and the laboratory member performing the experiment remains blinded to an individual’s genotype. On every experimental day, subjects of each genotype category are sampled and processed.

Standard protocol approvals, registrations, and patient consent.

All subjects were consented using a protocol approved by the Institutional Review Board of Partners Healthcare.

Genotyping

SNP genotype data were available for the MS subjects from existing Affymetrix Genechip 6.0 data 5 (Affymetrix, Santa Clara, CA). Genotype data for healthy control subjects from the BWH PhenoGenetic Project were generated using the Sequenom MassArray platform (Sequenom, San Diego, CA) at the Broad Institute’s Genotyping Analysis Platform. We applied standard quality-control criteria on subjects and on SNPs (Hardy–Weinberg equilibrium p > 1 × 10−3; minor allele frequency > 0.01, genotype call rate > 0.95).

Cell line

HEK293T cells were obtained from American Type Culture Collection (ATCC, Manassas, VA) and cultured as recommended.

Minigene construct

Total RNA was extracted from PBMC of a donor homozygous for the susceptible allele rs1800693C using the RNeasy Mini kit (Qiagen, Hilden, Germany). cDNA synthesis was carried out using SuperscriptIII-RT and oligo(dT) (Invitrogen, Grand Island, NY). The forward 5′ TAGCTGTCTGGCATGGGCCTC 3′ and reverse 5′ CGCAGCCTCATCTGAGAAGACTG 3’ primers were used to amplify the FL and Δ6 isoforms, using Phusion® High-Fidelity (New England Biolabs, Ipswich, MA). The PCR products have been gel-purified using QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany), amplified into modified pBluescript, ligated into pcDNA3.1(+) (Invitrogen, Grand Island, NY) using EcoRV (New England Biolabs, Ipswich, MA) and transformed chemically into One Shot TOP10 (Invitrogen, Grand Island, NY) competent Escherichia coli cells. Inserts in the vectors have been confirmed by sequencing.

In Vitro Transient Transfection Assays, Flow cytometry staining, Western Blot

The constructs with the full length and D6 isoforms subcloned into the pcDNA3.1(+) vector were transiently transfected into HEK293T using a calcium phosphate approach (along with a parallel GFP construct to test for transfection efficiency). Cells were harvested after 48 hours and stained for surface and intracellular expression of TNFRSF1A. Phycoerythrin (PE) conjugated anti-human TNFRSF1A mouse monoclonal antibody (Clone 16803, R&D Systems, Minneapolis, MN) or an IgG1 isotype control antibody (BioLegend, San Diego, CA) were used. For western blot on cell lysates and supernatant of transfected cells we have used anti-TNFR1 clone H5 from Santa Cruz Biotechnology (Dallas, Texas) and rabbit-anti-mouse IgG secondary antibody.

ELISA: soluble TNFRSF1A measurements

The laboratory member performing the ELISA was blinded to each subject’s genotype during the generation and analysis of the data. ELISA measurement of TNFRSF1A on serum samples of healthy controls and MS subjects were performed using the DuosetÒ ELISA Development System (DY225) according to the manufacturer’s recommendations (R&D Systems, Minneapolis, MN). Briefly, 100 μl of standard or serum sample diluted 1:10 using PBS supplemented with 1% BSA were added to each well of a 96-well microplate that was pre-coated with specific capture antibodies and incubated overnight. at 4° C. Wells were then washed with the appropriate buffer and incubated for 1h at RT with detection antibody. Afterwards plates have been incubated with biotinylated antibody reagent. After 8 washes, Streptavidin-HRP Reagent was added and incubated for 30 min at room temperature. Finally, the SureBlue TMB substrate (KPL, Gaithersburg, MD) was added. The enzyme-substrate reaction produced a luminescent signal that was immediately detected using a Bio-Rad Benchmark microplate reader (Bio-Rad Laboratories, Philadelphia, PA). All assays were performed in triplicate and outliers were excluded (CV > 5%) from the analysis. All samples, randomized by genotype, storage time and status of the samples (MS samples vs. healthy controls), were plated in total of 4 batches. Two samples have been used as internal control and measured in each plate across all plates of the 4 different batches. Those samples have been used for intra-plates normalization within and across batches. The sensitivity of our ELISA was 25 pg/ml.

Purification of peripheral mononuclear cells, isolation of monocytes and culturing condition

Peripheral blood mononuclear cells (PBMC) were obtained from fresh blood by density-gradient centrifugation with Ficoll (GE Healthcare, Princeton NJ). CD14+ monocytes were purified from mononuclear cells with anti-CD14 magnetic beads as recommended by the manufacturer (Miltenyi Biotec, Auburn, CA). The purity of monocytes was over 90% as tested by flow cytometry. Monocytes were kept in culture for 48h, in RPMI 1640 medium (Lonza-Biowhittaker, Walkersville, MD) supplemented with fetal bovine serum and human macrophage colony–stimulating factor (M-CSF; 10 ng/ml) (216-MC, R&D Systems, Minneapolis, MN) in the presence or absence of recombinant human TNFα (10 ng/ml) (210-TA, R&D Systems, Minneapolis, MN). Both supernatants and RNA cell lysates were collected 48h after the addition of TNFα.

ELISA: CXCL10 measurements

Supernatant were collected from human monocytes 48hrs after TNFα stimulation. Unstimulated and stimulated samples were collected and kept at -80° C until tested. The paired capture/detection antibody set is from BD Biosciences (San Jose, CA) (human CXCL10, 555046 and 555048, both used at 1 mg/ml). Recombinant human CXCL10 (266-IP, R&D Systems, Minneapolis, MN) was used to generate a standard curve. The ELISA developing protocol was as described above for soluble TNFRSF1A. The sensitivity of our ELISA was 45 pg/ml.

Antibodies and Detection of cell surface TNFRSF1A expression (CD120a) by flow cytometry

Cell surface expression of TNFRSF1A in the transfected cell population and in whole blood was assayed by flow cytometry after immunofluorescent staining. The anti-human monoclonal antibodies used for whole blood cell surface immunostaining were phycoerythrin (PE) conjugated anti-human TNFRSF1A mouse monoclonal antibody (Clone 16803, R&D Systems, Minneapolis, MN), Alexa Fluor 647–conjugated anti-CD16 (Clone 3G8, BioLegend, San Diego, CA), Fluorescein isothiocyanate (FITC)–conjugated anti-CD3 (Clone UCHT1, Biolegend, San Diego, CA) and Pacific Blue–conjugated anti-CD14 (clone M5E2, BioLegend, San Diego, CA). The isotype control antibodies used were PE-conjugated mouse IgG1 (BioLegend, San Diego, CA). To minimize potential variation due to antibody batch differences, all antibodies were obtained prior to the start of each experiment (discovery and replication). Blood staining occurred within 5 h of blood being drawn and kept on ice. After blocking in the presence of FcR receptor blocking at twice as much the concentration recommended by the manufacturer (Miltenyi Biotec, Auburn, CA), each sample was stained for 40 min on ice with either isotype control antibody and the antibody cocktail above mentioned and then lysed with BD FACS Lysing Solution (BD Biosciences, San Jose, CA). After erythrocyte lysis, the samples were stored at 4 °C until analysis by flow cytometry which was performed within 5 hours after staining. Immunostained samples were analyzed using a BD LSRII Flow Cytometer with BD FACSDiVa Software (BD Biosciences, San Jose, CA) and cytometry data were analyzed using FlowJo (Tree Star, Inc., Ashland, OR). MFI Expression level were calculated on the difference of (MFI – isotype staining) for each individual sample.

Shedding experiment

Purified CD14+ cells have been stimulated for 25 min with 0, 1, 5, 10 and 20 ng/ml of phorbol-12-myristate-13-acetate (PMA, Sigma-Aldrich, St. Louis, MO) at 37 C in polypropylene tubes at a concentration of 1.5x106 cells/ml. After incubation, cells have been washed twice with RPMI culture medium and subsequently stained on the surface for TNFRSF1A expression and isotype.

Quantitative real-time PCR for full-length and Δ6 isoform quantitation

RNA was extracted from pre-genotyped banked frozen PBMC with the Qiagen RNeasy Micro kit (Qiagen, Hilden, Germany). 300ng of total RNA were converted to cDNA using Superscript III from Invitrogen (Grand Island, NY). qPCR was performed with TaqMan reagent on an AB 7500 HT instrument using 20 ng total RNA in each assay (Applied Biosystems, Grand Island, NY). Full-length TNFRSF1A was measured using the commercial probe Hs00533565_g1 (Applied Biosystems, Grand Island, NY) specific for the exon 6/7 boundary while TNFRSF1A-Δ6 isoform was amplified by PCR using the following primers: 5′-CACCTGCCATGCAGGTTTCT-3′ (forward) and 5′-AGGGATAAAAGGCAAAGACCAAA-3′ (reverse) and fluorescent reporter 5′-TCTCCTGTAGTAAGCACCACA-3′. Relative fold changes of mRNA expression were evaluated with the 2-(dCt) normalizing to b2Microglobulin.

NanoString nCounter assay

We followed detailed protocol for the nCounter Analysis System (NanoString Technologies, Seattle, WA) as reported by Geiss et al. e-3. In brief, 2 sequence-specific probes were constructed for each gene of interest (Interferon genes, see below). One probe was covalently linked to an oligonucleotide containing biotin (the capture probe), and the other was linked to a color-coded molecular tag that provided the signal (the reporter probe). The nCounter CodeSet for these studies contained probe pairs for 47 genes (46 genes of interest and 1 housekeeping). Each sample was individually hybridized using 4 ml of total RNA lysate, without prior cDNA production. All genes and controls were assayed simultaneously in a multiplexed reaction. To account for differences in hybridization, the raw data were normalized to the standard curve generated via the nCounter system spike-in controls present in all reactions and using the most stable housekeeping gene among the 8 that were tested in the panel (SDHA). Specifically, a first normalization step calculated based on the geometric mean of the positive controls present in each reaction was applied. A second normalization step was performed using the housekeeping gene, and, ultimately, background level subtraction (average of negative control probes plus 2 times the standard deviation) was performed.

Statistical analysis

Linear models adjusting for age, gender, and symptom duration (except for MSSS) were used to compare neurological clinical outcomes among genotypes, using an additive model that considers the effect of each additional rs1800693G allele. T2 hyperintense lesion volume (T2LV) measures have been log transformed. Baseline outcome and last outcomes measures defined respectively as first and final non-missing value during follow up were used to test the associations between the TNFRSF1A rs1800693G susceptible allele and MS clinical outcomes. The genetic power calculator (http://pngu.mgh.harvard.edu/~purcell/gpc) was used to perform the power calculations using the reported sample sizes and an a = 0.05. Among MS individuals under treatment (GA or IFNb), Cox Proportional Hazards Models adjusted for age at first symptom, gender and symptom duration were used to test the association between the rs1800693G allele and time to first inflammatory event, defined as an MRI event (new T2 hyperintense lesion or new gadolinium-enhancing lesion), clinical relapse or EDSS increase in the first 6-25 months while the subjects were on treatment.

The COMBAT algorithm (as implemented in Genepattern toolkit e-4) has been applied to remove batch effect when combining the ELISA and expression data of the sample set1 and 2. In the analysis of ELISA data the difference between the stimulated and non stimulated conditions has been used to control for variability from one sample to another.

For analysis of serum ELISA data, a 4 parameters logistic regression model (4PL) was applied to the raw data, followed by a normalization approach across plates. Briefly, the average of the geometric mean of repeated measures on two samples present in all plates have been calculated. The ratio between the average geometric mean and the geometric mean in each individual plate was used as a coefficient of correction for each plate. Data have been analyzed with a linear model to test for associations between genotype and soluble level of TNFRSF1A, assuming a genotypic additive model and adjusting for batch effects, age at serum sampling, gender and treatment (when needed). Analysis of variance (ANOVA) and linear model (adjusting for batch effects, age at serum sampling, gender) were used to assess differences in soluble TNFRSF1A among the group of healthy subjects and the 3 treatment groups (GA, IFNβ and untreated). Linear additive model as implemented in MatrixEQTL (R Bioconductor e-5) was used to test for associations between rs1800693 genotype and soluble CXCL10, TNFRSF1A transcript levels and gene transcripts. Two tails Welch–T-test for samples with unequal variance was used to compare genotype rs1800693GG and rs1800693AA with TNFRSF1A cell surface expression in human monocyte cell subsets. Age and sex were not included as covariates in the analysis because no significant association was found.