DNA methylation changes in CD4+ T cells isolated from multiple sclerosis patients on dimethyl fumarate.

BACKGROUND: Dimethyl fumarate is an oral treatment for multiple sclerosis, whose mechanism of action is not fully understood.
OBJECTIVE: To investigate the effects of dimethyl fumarate on DNA methylation in the CD4+ T cells of multiple sclerosis patients.
METHODS: We performed Illumina EPIC arrays to investigate the DNA methylation profiles of CD4+ T cells derived from multiple sclerosis patients before and after dimethyl fumarate treatment.
RESULTS: Treatment with dimethyl fumarate resulted in 97% of differentially methylated positions showing hypermethylation. Four genes, SNORD1A, SHTN1, MZB1 and TNF had a differentially methylated region located within the transcriptional start site.
CONCLUSION: This study investigates the effect of dimethyl fumarate on DNA methylation in multiple sclerosis patients.

Introduction

Although increasing numbers of treatments are available for multiple sclerosis (MS), the exact mechanism of action of these therapies is often unclear. Patients are frequently required to trial several treatments to identify which is most suitable for their disease activity. Dimethyl fumarate (DMF; Tecfidera, Biogen Idec, Cambridge MA, USA) is approved in Europe and Australia as a first-line oral drug for the treatment of relapsing–remitting multiple sclerosis, and its use is associated with a reduction in disease activity and a variable effect on progression.1,2

Although the exact mode of action is not fully elucidated, DMF has been shown to have both anti-inflammatory and anti-oxidative properties. Decreased absolute lymphocyte counts and a shift in T lymphocyte polarisation from T helper (Th)1 and Th17 (pro-inflammatory) to Th2 phenotype (anti-inflammatory) has been reported after DMF treatment in MS patients.3 DMF also promotes translocation of nuclear factor erythroid 2-related factor 2 into the nucleus, which upregulates the transcription of anti-oxidative enzymes.3

DNA methylation refers to the epigenetic modification whereby the addition/removal of methyl groups to CpG dinucleotides regulates gene transcription. We, and others, have assessed global methylation profiles in CD4+ and CD8+ T cells from MS patients compared to healthy controls.4–6 Our studies have demonstrated altered methylation profiles in the CD4+ T cells of treatment-naive patients or in the absence of treatment. However, the effect of disease-modifying therapies (DMTs) on methylation profiles remains unclear. Neither group found significant changes in CD8+ T cells.5,6

Here we performed a longitudinal study of the genome-wide methylation profiles of CD4+ T cells in MS patients before and after DMF treatment.

Methods

We recruited seven MS patients (three men and four women) who were either treatment naive or had been off DMT for at least 3 months and were planning to start DMF therapy (Table 1). The majority of patients had not had steroid use for at least 2 months prior to entry into this study (Table 1). Samples were collected and processed as previously described.7 Blood was collected prior to the first dose of DMF and 6 months following treatment initation. At 6 months, all patients remained on therapy and had no change in Expanded Disability Status Score (EDSS). Two patients had evidence of disease activity as assessed by the appearance of new lesions on magnetic resonance imaging (MRI). However, both of these patients showed no new disease activity at their next MRI and remain on therapy.

Table 1.

MS cohort demographic/clinical features at baseline (prior to DMF treatment).

Sex	Age (years)	Prior DMT	Prior steroids (days prior to collection)	EDSS at baseline	EDSS at 6 months
M	23	Naive	312	1	1
M	40	Naive	7	3	3
M	35	Naive	54	1.5	1.5
F	32	Naive	3400	2.5	2.5
F	42	(Fingolimod)	1224	3	3
F	31	(Glatiramer acetate)	72	1.5	1.5
F	43	(Peginterferon beta-1a,interferon beta-1a)	95	2	2

MS: multiple sclerosis; DMF: dimethyl fumarate; DMT: disease-modifying therapy; EDSS: Expanded Disability Scale Status.

CD4+ T cells were extracted using magnetic isolation kits (Stem Cell Technologies, Canada) and purity (mimimum threshold ≥90%) was assessed using the FACS CantoII (BD Biosciences) system. Purified DNA was bisulphite converted and hybridised to Illumina EPIC arrays. Raw fluorescence data were processed using a combination of R/Bioconductor and custom scripts. Differences in mean methylation before and after the 6-month treatment period were tested using a paired samples t-test for each CpG. A CpG was considered a differentially methylated position (DMP) if the P value was less than 0.0005 and the absolute difference in mean methylation between groups was greater than 5%. A differentially methylated region (DMR) was defined as two or more contiguous DMPs located within 500 bp of each other, whose methylation changes were in the same direction. If a DMP was located outside of the 500 bp region but was less than 500 bp from the last DMP it was also included in the DMR.

Results and discussion

In total, 945 DMPs were identified when comparing the 6-month time point to baseline, the majority of which were hypermethylated after treatment (912; 97%) (see Supplementary Table 1). The most altered DMP between baseline and treatment was 17.5% hypermethylated (cg14048158); however, this site maps to an area with no known gene association. To identify sites of potential functional consequence, we filtered DMPs to include only those with a DMR, gene name and position annotation. Table 2 shows the DMPs with the largest percentage change for each of the resulting 64 genes.

Table 2.

DMRs with gene name and annotation.

Chr.	CpG ID	Position	Gene name	Element	Mean methylation		% Change	T stat.	P value
					Baseline	6 Months
1	cg16144718	23115066	EPHB2	Body	0.50	0.62	11.47	7.06	4.05 × 10–4
1	cg06808725	32264502	SPOCD1	Body	0.44	0.56	11.74	6.64	5.64 × 10–4
1	cg24533227	42145514	HIVEP3	5′UTR	0.65	0.76	10.63	6.71	5.32 × 10–4
1	cg02410801	55046065	ACOT11	Body	0.56	0.66	9.02	8.87	1.15 × 10–4
1	cg25130912	201982886	ELF3	Body	0.66	0.76	9.67	7.11	3.89 × 10–4
2	cg05333614	1168186	SNTG2	Body	0.72	0.77	5.36	7.77	2.40 × 10–4
2	cg03771015	15831147	LOC101926966	Body	0.62	0.70	7.78	6.35	7.16 × 10–4
2	cg14501323	31279457	GALNT14	Body	0.81	0.87	6.37	6.96	4.38 × 10–4
2	cg10796691	65135899	LOC400958	Body	0.60	0.69	8.36	9.16	9.53 × 10–5
2	cg16603943	121614683	GLI2	Body	0.51	0.63	11.66	6.35	7.13 × 10–4
2	cg20772458	158983130	UPP2	Body	0.74	0.80	5.89	7.26	3.46 × 10–4
2	cg18707238	218688237	TNS1	Body	0.73	0.78	5.52	6.47	6.49 × 10–4
3	cg15756415	14932169	FGD5	Body	0.42	0.54	12.05	6.48	6.40 × 10–4
3	cg02790932	23373256	UBE2E2	Body	0.65	0.72	7.55	8.65	1.31 × 10–4
3	cg00049674	123058535	ADCY5	Body	0.61	0.68	7.51	7.01	4.20 × 10–4
5	cg27073488	14262157	TRIO	Body	0.70	0.75	5.39	8.81	1.19 × 10–4
5	cg16375820	55289001	IL6ST	5′UTR	0.31	0.23	–7.30	–9.25	9.03 × 10–5
5	cg27346756	90431802	ADGRV1	Body	0.58	0.64	6.52	7.38	3.18 × 10–4
5	cg16558774	132579360	FSTL4	Body	0.68	0.75	6.66	9.33	8.58 × 10–5
5	cg11988321	138725622	MZB1	TSS200	0.42	0.54	12.21	16.30	3.39 × 10–6
6	cg04095776	31106941	PSORS1C1	Body	0.66	0.72	6.25	7.34	3.28 × 10–4
6	cg19978379	31542671	TNF	TSS1500	0.54	0.67	13.00	7.09	3.95 × 10–4
6	cg15496866	40491590	LRFN2	5′UTR	0.61	0.72	11.04	7.47	2.97 × 10–4
6	cg01473948	148823785	SASH1	Body	0.59	0.66	7.47	6.91	4.54 × 10–4
7	cg13800949	47343103	TNS3	Body	0.79	0.85	5.90	8.14	1.85 × 10–4
7	cg14797899	69882555	AUTS2	Body	0.68	0.78	9.85	7.70	2.51 × 10–4
7	cg02170577	104939331	SRPK2	Body	0.72	0.77	5.03	6.94	4.44 × 10–4
7	cg05476934	133859100	LRGUK	Body	0.52	0.60	8.64	6.74	5.18 × 10–4
7	cg09891341	138619424	KIAA1549	Body	0.78	0.84	6.24	6.81	4.90 × 10–4
7	cg06679384	158049077	PTPRN2	Body	0.60	0.66	6.23	7.57	2.76 × 10–4
9	cg08290373	8633541	PTPRD	Body	0.68	0.78	10.02	6.66	5.52 × 10–4
9	cg17557530	90193634	DAPK1	Body	0.61	0.73	12.22	6.62	5.74 × 10–4
9	cg06749278	97662692	C9orf3	Body	0.75	0.83	7.50	6.91	4.55 × 10–4
10	cg16203213	45398814	TMEM72-AS1	Body	0.66	0.74	7.24	9.01	1.05 × 10–4
10	cg26754789	49857879	ARHGAP22	Body	0.74	0.79	5.36	9.19	9.34 × 10–5
10	cg13312268	50019744	WDFY4	ExonBnd	0.72	0.78	6.84	6.68	5.44 × 10–4
10	cg12552633	71573337	COL13A1	Body	0.45	0.55	10.40	6.72	5.29 × 10–4
10	cg24587741	79313774	KCNMA1	Body	0.68	0.75	7.07	6.89	4.62 × 10–4
10	cg17753789	81026766	ZMIZ1	Body	0.69	0.76	7.74	6.52	6.21 × 10–4
10	cg16035098	118886914	SHTN1	TSS1500	0.46	0.55	9.16	9.66	7.04 × 10–5
10	cg01613414	126693304	CTBP2	Body	0.52	0.62	10.31	7.40	3.13 × 10–4
11	cg09731767	503628	RNH1	5′UTR	0.53	0.61	7.79	9.85	6.32 × 10–5
11	cg11922498	4936427	OR51G2	1stExon	0.65	0.71	6.09	9.45	7.99 × 10–5
11	cg00842359	10686144	MRVI1	5′UTR	0.67	0.77	9.37	9.28	8.85 × 10–5
11	cg14595291	35993855	LDLRAD3	5′UTR	0.50	0.65	14.89	6.42	6.76 × 10–4
11	cg00964019	117593395	DSCAML1	Body	0.76	0.82	5.81	7.08	3.97 × 10–4
12	cg11439695	2561024	CACNA1C	Body	0.46	0.56	9.93	8.25	1.71 × 10–4
12	cg17451712	122293122	HPD	Body	0.68	0.75	7.13	6.57	5.95 × 10–4
14	cg03725784	61992305	PRKCH	Body	0.41	0.54	12.42	7.14	3.81 × 10–4
14	cg11198334	75040680	LTBP2	Body	0.63	0.74	10.81	6.90	4.56 × 10–4
14	cg07399096	91050031	TTC7B	Body	0.69	0.75	6.38	6.58	5.91 × 10–4
14	cg15325186	102562217	HSP90AA1	Body	0.51	0.60	8.52	7.79	2.36 × 10–4
15	cg25814224	51572976	CYP19A1	5′UTR	0.65	0.71	6.39	10.93	3.49 × 10–5
16	cg02260059	78262124	WWOX	Body	0.72	0.78	5.78	6.62	5.74 × 10–4
17	cg04456720	54250143	ANKFN1	Body	0.67	0.77	10.60	8.93	1.10 × 10–4
17	cg19439071	74557625	SNORD1A	TSS200	0.56	0.67	10.87	8.68	1.29 × 10–4
17	cg11476241	78866235	RPTOR	Body	0.56	0.66	10.10	6.45	6.58 × 10–4
18	cg13297582	13288627	LDLRAD4	5′UTR	0.51	0.62	10.73	7.79	2.36 × 10–4
18	cg03385871	46311648	CTIF	Body	0.45	0.56	11.75	6.56	6.00 × 10–4
19	cg07345937	1175444	SBNO2	TSS1500	0.53	0.63	9.74	10.58	4.19 × 10–5
20	cg10453816	37499530	PPP1R16B	Body	0.51	0.64	13.41	6.85	4.74 × 10–4
20	cg04991444	50057438	NFATC2	Body	0.41	0.53	12.25	9.96	5.94 × 10–5
21	cg10919441	44143035	PDE9A	5′UTR	0.67	0.73	6.59	7.18	3.68 × 10–4

DMT: disease-modifying therapy; Chr.: chromosome.

Four genes had at least two adjacent DMPs located in the transcriptional start site (TSS) (Table 3). SNORD1A (small nucleolar RNA, C/D box 1A) encodes for an uncharacterised small nucleolar RNA. SHTN1 encodes shootin1, a protein involved in neuronal polarisation of axons.8 MZB1 (marginal zone B and B1 cell-specific protein) codes for an endoplasmic reticulum calcium regulator. While it has not previously been linked to MS, a study by Belkaya et al. (2013) found that overexpression of miR-185 resulted in a nearly five-fold decrease of MZB1 in mice.9 This decrease corresponded with lymphopenia and a reduced proliferative response in CD4+ T cells.9 The observed increase in DNA methylation identified in the MZB1 TSS in our dataset may result in a similar decrease in MZB1 transcription. A resulting decrease in CD4+ T cells would be consistent with the known anti-inflammatory action of DMF.

Table 3.

Genes with DMRs in the transcriptional start site.

Chr.	CpG ID	Position	Gene name	Element	Mean methylation		% Change	T stat.	P value
					Baseline	6 months
5	cg11988321	138725622	MZB1	TSS200	0.421426	0.543513	12.20872	16.29927	3.39 × 10–6
5	cg04359635	138725975	MZB1	TSS1500	0.513568	0.633291	11.97221	6.619687	5.72 × 10–4
6	cg19978379	31542671	TNF	TSS1500	0.537903	0.667883	12.99794	7.088953	3.95 × 10–4
6	cg24452282	31542740	TNF	TSS1500	0.470376	0.584539	11.41626	6.396426	6.88 × 10–4
10	cg16035098	118886914	SHTN1	TSS1500	0.458233	0.549798	9.156483	9.664102	7.04 × 10–5
10	cg23251794	118886883	SHTN1	TSS1500	0.641729	0.707866	6.613676	9.553839	7.51 × 10–5
17	cg19439071	74557625	SNORD1A	TSS200	0.562719	0.671405	10.86864	8.67833	1.29 × 10–4
17	cg07180212	74557703	SNORD1A	TSS200	0.62678	0.726408	9.962796	6.422776	6.73 × 10–4
17	cg13664588	74557494	SNORD1A	TSS1500	0.527718	0.624507	9.678917	6.578915	5.92 × 10–4

DMR: differentially methylated region; Chr.: chromosome.

The fourth DMR identified is located at the TSS of tumour necrosis factor (TNF). TNF is a pro-inflammatory cytokine that is produced by many cell types, including lymphocytes (reviewed in Wajant et al).10 TNF binding to its receptor activates the nuclear factor kappa B (NF-κB) pathway, which activates the transcription of genes involved in cell survival and proliferation, inflammatory response and anti-apoptotic factors. Hypermethylation at the TNF TSS may result in decreased TNF production, and a decrease in activation of the NF-κB pathway. One known mechanism of action for DMF is preventing translocation of NF-κB to the nucleus, resulting in a decrease of pro-inflammatory cytokines and an increase of anti-inflammatory cytokines (reviewed in Pistono et al.).3 It is possible that altered DNA methylation profiles at the TNF TSS may contribute to this mechanism.

DMF has previously been linked to other epigenetic mechanisms in a study by Kalinin et al. (2013), in which they reported that DMF increased expression of histone deacetylases in cultured rat astrocytes.11 Both DNA methylation and histone deacetylation are associated with gene repression.12 Taken together there is now evidence that DMF may act as an epigenetic modifier with the function of shutting down transcription associated with pro-inflammatory activity.

One limitation of this study is that we only assessed patients who started DMF treatment. Also, athough the majority of patients were stable at the time of baseline collection, two patients had recently had a relapse, only one of whom was treated with steroids. We are therefore unable to determine for certain if the changes in methylation profiles are due to treatment effects or stabilisation of disease. Future studies comparing changes following different therapies and different disease severities are required. A further limitation is the small sample size and lack of transcriptional data. Future studies characterising treatment responses in larger populations that also investigate the functional changes at the transcriptional level are warranted.

This is the first longitudinal study to investigate the effect of DMF on the DNA methylation of CD4+ T cells of MS patients. Of the most interest, the DMRs identified at TNF and MZB1 provide a potential novel mechanism of action for DMF. Treatment with DMF resulted in overall hypermethylation suggesting that DMF may act to promote DNA methylation. Larger studies are warranted to elucidate further the functional link between DMF and epigenetic mechanisms.

Click here for additional data file.

Supplemental material for DNA methylation changes in CD4+ T cells isolated from multiple sclerosis patients on dimethyl fumarate by Vicki E Maltby, Rodney A Lea, Karen A Ribbons, Katherine A Sanders, Daniel Kennedy, Myintzu Min, Rodney J Scott and Jeannette Lechner-Scott in Multiple Sclerosis Journal – Experimental, Translational and Clinical