Body Mass Index in Multiple Sclerosis modulates ceramide-induced DNA methylation and disease course

Abstract

Background: Multiple Sclerosis (MS) results from genetic predisposition and environmental variables, including elevated Body Mass Index (BMI) in early life. This study addresses the effect of BMI on the epigenome of monocytes and disease course in MS.

Methods: Fifty-four therapy-naive Relapsing Remitting (RR) MS patients with high and normal BMI received clinical and MRI evaluation. Blood samples were immunophenotyped, and processed for unbiased plasma lipidomic profiling and genome-wide DNA methylation analysis of circulating monocytes. The main findings at baseline were validated in an independent cohort of 91 therapy-naïve RRMS patients. Disease course was evaluated by a two-year longitudinal follow up and mechanistic hypotheses tested in human cell cultures and in animal models of MS.

Findings: Higher monocytic counts and plasma ceramides, and hypermethylation of genes involved in negative regulation of cell proliferation were detected in the high BMI group of MS patients compared to normal BMI. Ceramide treatment of monocytic cell cultures increased proliferation in a dose-dependent manner and was prevented by DNA methylation inhibitors. The high BMI group of MS patients showed a negative correlation between monocytic counts and brain volume. Those subjects at a two-year follow-up showed increased T1 lesion load, increased disease activity, and worsened clinical disability. Lastly, the relationship between body weight, monocytic infiltration, DNA methylation and disease course was validated in mouse models of MS.

Interpretation: High BMI negatively impacts disease course in Multiple Sclerosis by modulating monocyte cell number through ceramide-induced DNA methylation of anti-proliferative genes. FUND: This work was supported by funds from the Friedman Brain Institute, NIH, and Multiple Sclerosis Society.

Keywords: Epigenetics; Immunity; Lipids; Neurodegeneration; Obesity.

Copyright © 2019 The Authors. Published by Elsevier B.V. All rights reserved.

Figures

Fig. 1

Immune profiling reveals a BMI-dependent increase in blood monocyte counts in MS patients but not healthy individuals. Immune cell populations from fresh whole blood were quantified using flow cytometry. (a) Differences in T cell counts due to BMI were assessed in a primary cohort of MS patients (n = 52). (b) Differences in monocyte counts due to BMI were assessed in a primary cohort of MS patients (n = 52), a validation cohort of MS patients (n = 91), and a cohort of healthy individuals (n = 50). One-way MANCOVA was used to adjust for age and sex, followed by pair-wise comparisons with Bonferroni correction to determine statistically significant differences in the two BMI groups (normal BMI = white dots, high BMI = gray dots) (*p < .05).

Fig. 2

Lipidomic profiling reveals increased abundance of ceramide species in MS patients with high BMI, but not in healthy individuals. Unbiased lipidomic analysis was performed on plasma samples from MS patients (n = 48) and healthy individuals (n = 40) using an MS/MSALL platform. Lipids that differed in abundance due to BMI were assessed using multiple t-tests with 5% FDR correction, q < 0.05 were considered significant (*q < 0.05, **q < 0.01). (a) Pie chart representation of numbers and percentage of lipids in MS patients by lipid family and class. Red = increased abundance, blue = decreased abundance, white = no change in patients with high BMI. (b) Relative abundance of ceramide species with 18:1 backbone that were significantly different in abundance in MS patients with high BMI (gray dots) compared to those with normal BMI (white). (c) The most highly significantly abundant ceramide species in MS patients were not statistically different in healthy individuals with high BMI (gray dots) compared to those with normal BMI (white). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Exposure to ceramides results in dose-dependent increase in proliferation and DNA methylation. (a) Cultured monocytes (i.e. THP1 cells) were treated with fluorescently labeled Ceramide C16 (NBD-Ceramide C16) or untreated for 24 h. 3D reconstruction of the cells was conducted on Imaris to visualize localization of NBD-Ceramide (green). (b) Cultures either untreated or treated with Ceramide C16 (Cer C16) for 1, 12, or 24 h were fixed at the same time. After immunocytochemistry for Ki67 to detect proliferating cells or5-methylcytosine as marker for global DNA methylation the percentage of cells relative to the total DAPI+ nuclei was calculated and plotted in the graphs. Statistical differences in the percentage of Ki67+ cells and percentage of 5mc + cells relative to total DAPI+ nuclei were assessed by Student's t-test. (c) Cultures were treated for 1 (left panels) or 24 (right panels) hours with concentrations of a mixture of ceramides C16, C22, and C24:1, mimicking those detected in the plasma of MS patients with normal (Low) or high BMI (High) in the absence or presence of the DNA methylation inhibitor, 5-aza-2′-deoxycytidine (5-aza). After fixation, cells were stained for Ki67 (green) as proliferation marker, 5-methylcytosine (5mC, red), as marker for global DNA methylation and DAPI (blue), to identify nuclei. Confocal images were acquired at 20× magnification using the Zeiss LSM800 confocal (scale bar = 25um). High and High+5-aza groups were compared to Low Cer at 24 h using Dunnett's multiple comparisons test (*p p < .01) (n = 3 independent biological replicates, each conducted in triplicate wells and analyzing 3 images/well). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Genome-wide distribution of DNA methylation marks in monocytes of high BMI MS patients. DNA methylation analysis was conducted on DNA from monocytes of MS patients using the Infinium HumanMethylation450 BeadChip. DNA samples from the monocytes of MS patients with high BMI (n = 23) were compared to those from patients with normal BMI (n = 25). Differentially methylated regions (DMRs) were identified using a 1 kb sliding window and significantly different CpGs were determined by evaluating the combined p values of CpG sites within each region, using the Stouffer's method with a 1% FDR correction. (a) Methylation values of individual CpGs within DMRs were measured as beta values, or the signal intensity from the methylated probe in relation to the sum of intensities from both methylated and unmethylated probes. Beta values of CpGs in MS samples with high BMI were subtracted from those in with normal BMI and a frequency histogram of the beta value differences was plotted to visualize the frequency of beta value differences in a given range. Positive beta value differences, indicating DNA hypermethylation in high BMI MS samples, are plotted in red, whereas negative differences, indicating DNA hypomethylation, are in blue. (b) Genomic distribution of the hypermethylated DMR in patients with high BMI (black bars), compared to the distribution of all the CpGs in the array (white bar). The relative proportions of CpGs in islands, shores (2 kb flanking the islands), shelves (2 kb flanking the shores) and sea (regions outside the previous three categories) are indicated. (c) Distribution of the hypermethylated CpGs distributed within in MS patients with high BMI (black bars) within distinct gene features, compared to the distribution of all the CpGs in the array (white bar). The relative proportion of differentially methylated CpGs in high BMI MS patients is represented relative to RefSeq gene promoters (2 kb from TSS), 5’UTR, 1st exon, gene bodies, 3’UTR and intergenic regions. Note that the CpGs hypermethylated in MS patients with high BMI were enriched at promoters and intergenic region and depleted at gene bodies, consistent with a repressive pattern of DNA methylation. Enrichment defined as statistically significantly different from the proportion of total array CpGs in a given location, which was calculated by chi- squared test, *p < .05, **p < .01. (d) Pathway enrichment analysis was performed on hypermethylated CpGs using the KEGG database on Enrichr. Enrichment score was calculated as the -log (adjusted p value). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5

Chromosomal distribution and identification of genes with most significantly hypermethylated regions in MS patients with high BMI identifies negative regulators of cell proliferation. (a) Circos plot representing the chromosomal distribution of DMRs and CpGs within DMRs. Chromosomal numbers are indicated in the innermost circle. The inner red and blue circle depict the beta value differences for each CpG within a DMR. The more hypermethylated the CpG (bright red) the farther it is from the center of the circos plot, and the more hypomethylated the CpG (blue) the closer it is to the center of the plot. Dark red lines in the outer circle denote the genomic location of the DMRs and dark red dots indicate the mean beta value difference for each DMR. Genes overlapping DMRs are indicated in the outermost circle and genes with the greatest methylation differences are highlighted in red. (b) Manhattan plot highlighting the overall methylation difference, or mean beta change, for each DMR on each chromosome. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6

Exposure to ceramides results in increased DNA methylation and reduced transcription of negative regulators of cell proliferation. (a) Cultured human monocytes (i.e. THP1 cells) were either untreated (blue) or treated with Ceramide C16 (red) for 24 h. DNA was harvested and processed for Sequenom Mass Array EpiTyper, in order to quantify differential methylation of CpGs (Y axis) at the indicated genomic locations (X axis). The gene name is shown on top of each graph. (b) samples from the same cells described in (a) were used for RNA isolation and quantification of the corresponding gene transcripts by using real time-PCR and normalized to reference gene transcripts. (c) Cultures were treated for 24 h with concentrations of a mixture of ceramides C16, C22, and C24:1, mimicking those detected in the plasma of MS patients with normal (Low) or high BMI (High) in the absence or presence of the DNA methylation inhibitor, 5-aza-2′-deoxycytidine (5-aza). DNA was harvested and processed for Sequenom Mass Array EpiTyper, in order to quantify differential methylation of CpGs (Y axis) at the indicated genomic locations (X axis). The gene name is shown on top of each graph. Overall methylation differences for the entire region of interest for each gene were assessed using two-way ANOVA (*p < .05, ***p < .0001) (n = 3/group, n = 2 replicates/group). (d) Transcript levels of NRXN1, FZD7, and TP63 were assessed using real time-PCR and normalized to the levels of thehousekeeping gene, RPLP0. Cer C16 groups were compared to Untreated groups using Student's t-test, while the High and High +5-aza groups were compared to the Low group, using Dunnett's multiple comparisons test (*p < .05, **p < .01, ***p < .001) (n = 3 independent experiments, n = 3 biological replicates/experiment, n = 3 technical replicates/biological replicate). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 7

Monocytes uniquely correlate with reduced brain volume in overweight/obese MS patients, who exhibit worsened disease course at the two-year follow up. (a) Brain volumetric analysis was performed on T1-weighted images from MRI scans and correlated with CD14+ monocyte counts from fresh whole blood, which had been obtained by flow cytometry, in a primary MS cohort (n = 41), validation MS cohort (n = 80) and a healthy cohort (n = 48) using linear regression, p < .05 considered significant. (b) Differences in the proportion of patients who showed changes in T1 lesion load (ΔT1) <1 ml, between 0 and 1 ml, and >1 ml during a 2 year period, between MS patients with normal or high BMI were assessed by chi-squared test (*p < .05). (c) Differences in the proportion of patients who had disease activity, defined by clinical relapse or new lesions, during a 2 year period, between MS patients with normal or high BMI were assessed by chi-squared test (*p < .05). (d) Change in EDSS in a 2 year period was categorized as better (<−0.5), same (−0.5 to 0.5), or worse (>0.5) and correlated with BMI using ordinal regression, p < .05 considered significant.

Fig. 8

High-fat diet fed mice show increased monocytic infiltration and greater neurological deficits in an experimental model of MS. (a) Female C57/bl6 mice were fed either a low-fat (LFD) or high-fat diet (HFD) for 5 weeks, and then immunized with 200 ng MOG35–55 using a protocol to induce EAE. (b) Ascending paralysis was graded on a scale of 0 to 5 with high scores indicating increased clinical disability (black = LFD-EAE mice, gray = HFD-EAE mice). Differences in EAE scores were evaluated between LFD-EAE and HFD-EAE mice at each time point using Mann-Whitney U test, respectively (*p < .05, **p < .01) (n = 15/group). (c) Weight change from baseline (i.e. day of immunization) was calculated for each time point and differences between LFD-EAE and HFD-EAE mice were evaluated by Sidak's multiple comparisons test (**p < .01). (d) Area under the curve of the EAE disease course and maximum EAE score were also used to evaluate differences in disease course and statistical analysis was performed using Mann-Whitney U test (*p < .05). (e-g) Mice were sacrificed at the peak of disease and spinal cord tissue sectioned for histological analysis. Spinal cord sections were stained with Fluoromyelin (FM)to stain myelin and lesion areas and Neurofilament H (NFH) to stain axons. Images were acquired using the Zeiss LSM 800 confocal microscope and the 20× objective (scale bar = 50um). Myelin and axonal content were quantified as the area of FM or NFH positivity, respectively, in relation to the whole spinal cord area. Spinal cord sections were also stained with Iba1 and CD45 to mark peripheral monocytes/activated microglia (Iba1+ CD45+) and leukocytes (CD45+) and counterstained with DAPI, to identify nuclei. Confocal images were acquired using the Zeiss LSM 800 and the 20× objective (scale bar = 25um). Iba1 + CD45+ area and CD45+ area was quantified within the lesion to enhance representation of peripheral monocytes and peripheral leukocytes, respectively, and was calculated as a percentage in relation to the total spinal cord area. Spinal cord sections were stained with 5mc and Iba1 to assess DNA methylation levels in monocytes. Images were acquired using the Zeiss LSM 800 at 63× (scale bar = 10um). 5mc intensity was normalized to DAPI intensity in Iba1+ cells within the lesion. Statistical differences between LFD-EAE (white) and HFD-EAE (gray) groups were assessed by Student's t-test (*p < .05) (n = 3 mice/group, n = 2 sections/mouse, n = 3 images/section).