EBioMedicine 10 (2016) 227–235

Contents lists available at ScienceDirect

EBioMedicine

journal homepage: www.ebiomedicine.com

Research Paper

A Minimally-invasive Blood-derived Biomarker of Oligodendrocyte

Cell-loss in Multiple Sclerosis
John A. Olsen a, Lauren A. Kenna a, Regine C. Tipon a, Michael G. Spelios a, Mark M. Stecker b, Eitan M. Akirav a,c,⁎
a
Research Institute, Islet Biology, Winthrop-University Hospital, Mineola, NY, USA
b
Department of Neuroscience, Winthrop-University Hospital Mineola, NY, USA
c
Stony Brook University School of Medicine, Stony Brook, NY, USA

a r t i c l e i n f o a b s t r a c t

Article history: Multiple sclerosis (MS) is a neurodegenerative disease of the central nervous system (CNS). Minimally invasive
Received 31 March 2016 biomarkers of MS are required for disease diagnosis and treatment. Differentially methylated circulating-free
Received in revised form 15 June 2016 DNA (cfDNA) is a useful biomarker for disease diagnosis and prognosis, and may offer to be a viable approach
Accepted 23 June 2016
for understanding MS. Here, methylation-specific primers and quantitative real-time PCR were used to study
Available online 27 June 2016
methylation patterns of the myelin oligodendrocyte glycoprotein (MOG) gene, which is expressed primarily in
Keywords:
myelin-producing oligodendrocytes (ODCs). MOG-DNA was demethylated in O4+ ODCs in mice and in DNA
Relapsing-remitting multiple sclerosis from human oligodendrocyte precursor cells (OPCs) when compared with other cell types. In the cuprizone-
Oligodendrocyte fed mouse model of demyelination, ODC derived demethylated MOG cfDNA was increased in serum and was as-
Cuprizone sociated with tissue-wide demyelination, demonstrating the utility of demethylated MOG cfDNA as a biomarker
Human patients of ODC death. Collected sera from patients with active (symptomatic) relapsing-remitting MS (RRMS) demon-
Circulating free DNA strated a higher signature of demethylated MOG cfDNA when compared with patients with inactive disease
Biomarker discovery and healthy controls. Taken together, these results offer a minimally invasive approach to measuring ODC
death in the blood of MS patients that may be used to monitor disease progression.
© 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction DNA methylation is a mechanism involved in the control of tissue-

specific gene expression (Jones, 2012). DNA hypermethylation of CpG
Multiple sclerosis (MS) is a neurodegenerative disease characterized dinucleotides acts as an inhibitor of transcription while hypomethyla-
by demyelination of axons in the central nervous system (CNS). The tion is associated with increased gene expression. Differences in DNA
cells that produce myelin in the CNS, oligodendrocytes (ODCs) (Bunge methylation patterns between different cell types can be used to deter-
et al., 1962; Bunge, 1968), die from an autoimmune response that mine the origin of the DNA. Accordingly, methylated circulating-free
results in demyelinated lesions in the brain and spinal cord DNA (cfDNA) is used as a biomarker of tumor progression (Nawroz et
(Noseworthy et al., 2000; Olsen and Akirav, 2015). Axonal demyelin- al., 1996; Chen et al., 1996; Crowley et al., 2013). Our group has success-
ation reduces the signal strength of nerve impulses (Waxman, 1977; fully adapted this approach for the detection of β-cell loss in patients
Felts et al., 1997) and may leave the axon exposed to degeneration with autoimmune type 1 diabetes (Akirav et al., 2011), where differen-
(Ferguson et al., 1997). Demyelinated lesions can be visualized by mag- tially methylated β-cell derived insulin cfDNA was used to detect β-cell
netic resonance imaging (MRI); however, this procedure is costly and loss. A recent report by Lehmann-Werman et al. showed the utility of
cannot detect diffused or mild myelin degeneration or measure the using cfDNA to detect cell loss in MS, by measuring demethylated
active loss of ODCs. Although new innovations in MS therapy are cur- cfDNA of MBP and WM1 genes, showing an increase in the levels of
rently being explored (Olsen and Akirav, 2015), there are currently no demethylated DNA in patients with MS (Lehmann-Werman et al.,
clinically approved molecular biomarkers of ODC death in MS. This 2016).
unmet need impairs MS diagnosis, prognosis, and assessment of clinical In contrast to MBP, which is found in both ODCs and Schwann cells,
intervention. and WM1, whose tissue expression patterns are largely unknown, my-
elin oligodendrocyte glycoprotein (MOG) is a CNS-specific protein
expressed solely by ODCs as an integral part of the myelin sheath
⁎ Corresponding author at: 101 Mineola Blvd., Rm. 04-39, Mineola, NY 11501, USA. (Linnington et al., 1984; Lebar et al., 1986; Gardinier et al., 1992; Johns
E-mail address: [email protected] (E.M. Akirav). and Bernard, 1999). The specificity of MOG in ODCs suggests that the

http://dx.doi.org/10.1016/j.ebiom.2016.06.031
2352-3964/© 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
228 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235

Table 1
Primer sequences and PCR protocols for mouse MOG analysis.

PCR type Primer designation Primer sequence 5′ → 3′ Product length PCR protocol

First-step PCR Forward GAGTGATAGGATTAGGGTATTTTATT 169 bp 50 cycles, annealing temperature 57 °C

Reverse TCTACATCTTAATCCTTACCATTTC
Methylation-specific nested qPCR Common forward GAGTGATAGGATTAGGGTATTTTATT 87 bp 40 cycles, annealing temperature 64 °C
Hypermeth-specific reverse TAACGTTCTTCCCAAAAAATATACG
Hypometh-specific reverse TAACATTCTTCCCAAAAAATATACA
Native sequence first-step PCR Forward GAGTGATAGGACCAGGGTATCCCATC 169 bp Cycles no. variable, annealing temperature 57 °C
Reverse CTGCATCTTGGTCCTTGCCATTTC

MOG gene may present with unique methylation patterns in ODCs, 2.4. Enrichment of O4+ Cells
which can be used to detect ODC-derived DNA in the blood and assess
disease activity in patients with MS. Here we identify demethylated O4+ cells were purified from the brains of five C57BL/6 mice
CpG dinucleotides in ODCs from mouse and human origin that are ab- (Jackson Laboratory, Bar Harbor, ME) and treated with the Neural Disso-
sent in other cell types in the brain and periphery. Methylation-specific ciation Kit (P) and Anti-O4 MicroBeads, followed by magnetic purifica-
primers for demethylated MOG-DNA were able to detect ODC-specific tion on the autoMACS Pro Separator (Miltenyi Biotec Inc., Auburn, CA).
DNA in purified primary mouse ODCs and primary human oligodendro-
cyte precursor cells (OPCs). When used to detect ODC MOG-DNA in the 2.5. FACS Staining
sera, these primers showed an increase in ODC-derived cfDNA in mice
treated with cuprizone. Testing of human primers successfully detected Evaluation of O4+ fraction purity was performed by fluorescence-
ODC MOG cfDNA in sera of patients with relapsing-remitting MS activated cell sorting (FACS) analysis. Magnetically purified ODCs
(RRMS) and showed a correlation with disease activity. were washed twice and suspended in FACS buffer containing anti-O4 la-
beled antibody or isotype control and analyzed using an Accuri FACS an-
alyzer (BD Biosciences, San Diego, CA). FACS data was analyzed using
2. Methods
FlowJo software (FlowJo, Ashland, OR).
2.1. Mice
2.6. Plasmid Injections
C57BL/6 female mice purchased from Jackson Laboratory (Bar
Harbor, ME) were started on a diet of rodent chow with 0.2% cuprizone Four C57BL/6 mice (Jackson Laboratory, Bar Harbor, ME) were each
(Research Diets Inc., New Brunswick, NJ) at 8 weeks of age (Day 0). Mice administered an intravenous injection of TOPO plasmid containing the
were divided into 2 groups of 6 (total n = 12) with blood sampling by demethylated mouse MOG sequence. Five minutes following injection,
cheek pouch bleed on Days 0, 7, 21, 35, and 49 for group 1 and Days 0, blood was obtained by heart puncture.
14, 28, and 42 for group 2. On Day 56 blood was obtained from all
mice via heart puncture. Brain tissues were collected for histological anal- 2.7. Histology
ysis of myelination. All animal care was approved by the Winthrop-
University Hospital Institutional Animal Care and Use Committee. Brain tissues from cuprizone or sham treated mice were collected
immediately following euthanasia, embedded in O.C.T. compound
(Fisher, Waltham, MA) and snap frozen in liquid nitrogen. Tissues
2.2. Cell and Cell Lines
were sectioned at 7 μm thickness and mounted on poly-D-lysine coated
slides, followed by formalin fixation. Myelin staining was done using a
SW10 immortalized neuronal murine Schwann cells were pur-
NovaUltra Luxol Fast Blue Staining Kit according to the manufacturer's
chased from American Type Culture Collection (Manassas, VA) and cul-
instructions (IHCworld, Woodstock, MD). Stained slides were imaged
tured using the provided protocols. Cells were pelleted for DNA
using a bright field on a Nikon Eclipse Ti confocal microscope (Nikon,
extraction. Human OPC genomic DNA extracts (ScienCell Research Lab-
Melville, NY).
oratories, Carlsbad, CA) were used as a positive control sample for
demethylated MOG-DNA. This genomic DNA sample was obtained
from ScienCell Research Laboratories from early passage human OPCs 2.8. Human Subject Samples
using the AllPrep DNA/RNA Mini Kit (Qiagen N.V., Valencia, CA).
Serum samples were obtained from BioServe Biotechnologies, Ltd.
(Beltsville, MD). Samples were designated as “Active” if having a relapse
2.3. Mouse Tissues at the time of sampling, “Inactive” if in remission at the time of sam-
pling, or “Healthy Control” based on donor information. Disease activity
Mouse liver, kidney, and brain tissue were obtained from a C57BL/6 was determined by specialized neurologists at the time of sampling and
mouse (Jackson Laboratory, Bar Harbor, ME) and homogenized for DNA included a combination of disease activity by patient admission, EDSS
extraction. scoring and MRI imagining where applicable. In some cases, MS activity

Table 2
Primer sequences and PCR protocols for human MOG analysis.

PCR type Primer designation Primer sequence 5′ → 3′ Product length PCR protocol

First-step PCR Forward GGGTAGTTTAGAGTGATAGGATTAAGATAT 152 bp 50 cycles, annealing temperature 57 °C

Reverse TAAAAATAAACCACCCTAAAAAAAA
Methylation-specific nested qPCR Common forward GGGTAGTTTAGAGTGATAGGATTAAGATAT 97 bp 40 cycles, annealing temperature 64 °C
Hypermeth-specific reverse TAACGTTTTTCTCAAAAAATATACG
Hypometh-specific reverse TAACATTCTTCCCAAAAAATATACA
 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235 229

Table 3 2.10. Bisulfite Sequence Prediction and Primer Design

Demographic and disease history for RRMS and control subjects.

Parameters Groups Prediction of methylated CpG in specific regions of the human and
Ctrl RRMS
mouse MOG gene was done using the MethPrimer program (Li and
Dahiya, 2002). Sequence output was used to design methylation insen-
Inactive Active
sitive and sensitive primers.
N (all females) 20 20 20
Ethnicity (W/AA) 20/0 19/1 20/0
Age (years) 41.3 ± 1.5 40.20 ± 1.75 41.3 ± 1.5 2.11. First-step PCR and Gel Extraction
Age at dx (years) – 32.5 ± 2.1 32.8 ± 1.6
Disease duration (years) – 7.7 ± 1.3 8.6 ± 1.1 A first-step PCR using non-methylation-specific primers and EpiTaq
a
Duration from last epi (years) – 1.5 ± 0.5 –
b
HS Kit (Clonetech Laboratories Inc., Mountain View, CA) was run to in-
Time between epi (years) – 1.9 ± 0.4 1.6 ± 0.6
crease template availability. First-step products were run on a 2% aga-
a
18/20 patients reporting. rose gel and isolated using the QIAquick Gel Extraction Kit (Qiagen
b
29/40 patients reporting.
N.V., Valencia, CA). No template controls in the first-step reaction
were free of products. Mouse and human primer sequences are de-
scribed in Tables 1 and 2, respectively.
was determined primarily by physician assessment and patient reports.
Demographic information, disease activity, disease duration, duration 2.12. Cloning of MOG-DNA
from the last episode, and time between episodes are summarized in
Table 3. Bisulfite treated first-step PCR products purified from agarose gel
were used as templates for a TOPO TA cloning reaction using the pCR
2.1TOPO vector (Invitrogen, Carlsbad, CA). ODC + and liver first-step
2.9. DNA Purification and Bisulfite Conversion MOG products were used as the mouse demethylated and methylated
inserts, respectively. Human demethylated and methylated DNA
DNA extraction from homogenized tissue, sera, and cell pellets was (Zymo Research, Irvine, CA) first-step MOG products were used for
performed using the DNEasy Blood and Tissue Kit (Qiagen N.V., Valencia, the human demethylated and methylated inserts, respectively. NEB 5-
CA). DNA concentration was measured with the Quant-iT PicoGreen alpha competent Escherichia coli (New England Biolabs, Ipswich, MA)
dsDNA Assay Kit (Life Technologies, Carlsbad, CA). DNA extracts were were used for transformation with the TOPO MOG products and incu-
bisulfite treated with the EZ DNA Methylation-Direct Kit (Zymo bated at 37 °C overnight on ImMedia Kan Agar plates (Invitrogen, Carls-
Research, Irvine, CA). bad, CA). Individual colonies were added to a culture of LB broth/

Fig. 1. Illustrative representation of biomarker assay used to detect oligodendrocyte death in multiple sclerosis. Myelin-producing oligodendrocytes die and release genomic DNA into
circulation. Blood is collected from the patient, DNA is purified, and bisulfite converted. Post-bisulfite conversion, samples are run on first-step PCR using methylation-unspecific
primers and loaded onto an agarose gel. First-step PCR product is extracted and used as a template for qPCR utilizing methylation-specific primers.
230 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235

Kanamycin (Gibco, Carlsbad, CA) and shaken at 37 °C overnight. The Zyppy-96 Plasmid Miniprep (Zymo Research, Irvine, CA) prior to se-
QIAprep Spin Miniprep Kit (Qiagen N.V., Valencia, CA) was used to pu- quencing. Purified clones were sequenced by the Keck Biotechnology
rify the TOPO plasmid from the cultures. Research Laboratory (New Haven, CT).

2.13. McrBC Restriction Enzyme Reaction 2.16. Methylation-sensitive Real Time PCR

Purified DNA from human liver, brain, and spinal cord fraction was Gel-extracted first-step PCR products served as a template for quan-
treated with the McrBC methylation-specific restriction enzyme (New titative real-time PCR (qPCR) using methylation-sensitive primers
England Biolabs Inc., Ipswich, MA). After treatment, 70 ng of treated (Sigma-Aldrich Corp., St. Louis, MO) designed for bisulfite treated
and untreated liver, spinal cord or brain DNA, as well as a TOPO plasmid MOG-DNA. All reactions were run using LightCycler 480 SYBR Green I
containing the appropriate native MOG insert, were run on PCR using Master (Roche Diagnostics GmbH, Germany) PCR mix on a CFX96
native MOG primers. Samples were run on PCR and were removed at Real-Time System (Bio-Rad Laboratories Inc., Hercules, CA). Relative
cycle 32 or 35. PCR products were run on a 2% agarose gel and imaged quantification of demethylated DNA was calculated using the demeth-
using a 4000R Image Station (Eastman Kodak Co., Rochester, NY). ylation index (DMI) = 2(methylated cycle number) − (demethylated cycle number).

2.14. Sequencing of MOG-DNA 2.17. Statistical Analysis

MOG first-step PCR gel extracts and MOG Plasmid DNA were se- Results are reported as mean ± SEM. Statistical significance (p b 0.05)
quenced at the Keck Biotechnology Research Laboratory (New Haven, of differences between means was calculated by one-way ANOVA and
CT). Tukey's post hoc test using Prism 5 (GraphPad software). In some
cases, t-test was used to compare two individual treatment groups.
2.15. Analysis of MOG Methylation in Primary Murine O4+ and O4− ROC analysis of Active vs. Inactive RRMS was performed with Prism 5.
Fractions
3. Results
First-step PCR product of bisulfite treated DNA from mouse O4+ and
−
O4 fractions, SW10 Schwann cells, and liver were cloned using pCR 3.1. MOG Gene DNA Shows Unique Methylation in the CNS
2.1TOPO vector (Invitrogen, Carlsbad, CA). NEB5-alpha competent E.
coli (New England Biolabs, Ipswich, MA) were used for the transforma- The concept of using DNA methylation as a biomarker of ODC loss is
tion reaction and individual colonies were treated with the Zymo described in Fig. 1. Previous reports show the expression of MOG in the

Fig. 2. Murine brain and spinal cord show differential methylation in the MOG gene, due to their O4+ cell population. A. Sanger sequencing results of bisulfite treated DNA from murine
tissues. Arrows point toward CpG sites where cytosines (C) are preserved in methylated samples (Liver, Kidney), or converted to thymines (T) in samples containing demethylated CpGs,
leading to a mixed population of C′s and T's (Brain, Spinal Cord). B. Methylation-sensitive DNA digestion was performed on the magnetically spinal cord of liver derived-DNA using the
McrBC enzyme. Digested DNA was subjected to semi-quantitative PCR (cycles 32 and 35) and run on agarose gel. UnTx—untreated DNA receiving all reaction components except
McrBC enzyme. Pos—positive control using cloned native MOG DNA. NTC—no template control. C and D. O4+ and O4− cells were separated from four separate murine brains by
magnetic beads. FACS analysis showed N92.6 ± 3.9% enrichment of O4+ cells among four independent preparations when compared to O4− fractions. E. DNA from murine O4+ cells
is differentially methylated in the MOG gene compared to DNA from O4− cells, the SW10 Schwann cell line, and liver. Sequence analysis was performed on first-step PCR product of
each sample, 10 clones from each are shown (○ represent demethylated cytosines; ●, methylated cytosines). Locations in relation to the MOG transcription start site are listed,
methylation-specific murine primers incorporate the CpG site at bp +2752.
 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235 231

brain, however, the role of epigenetics in controlling MOG expression MOG demethylation is unique to ODCs (Fig. 2E and Supplemental
remains largely unknown. To examine whether the MOG gene is Fig. 1). Taken together these data show demethylation patterns in
demethylated in the CNS, DNA from murine liver, kidney, brain and spi- the MOG-DNA from primary ODCs when compared with other CNS
nal cord was isolated and subjected to bisulfite sequencing. Bisulfite cell types, Schwann cells, and peripheral tissues, suggesting that
treatment of DNA leads to the conversion of cytosine (C) to thymine these differences in methylation can be used as a biomarker of ODC
(T) in demethylated (DeMeth) CpGs (C → T), while methylated loss.
(Meth) mCpGs remain intact (C → C). Sequence analysis of DNA from
murine brain and spinal cord tissues revealed an increase in C → T con-
version when compared with DNA from liver and kidney, demonstrat- 3.2. Methylation Specific Primers Detect DeMeth MOG-DNA in Murine CNS
ing the presence of DeMeth CpGs in CNS tissue (Fig. 2A). Analysis of Tissues and Purified ODCs
MOG DNA methylation in native DNA by the methylation sensitive
restriction enzyme, McrBC, confirmed the presence of demethylated The identification of demethylated CpG dinucleotides in primary
MOG DNA in spinal cord DNA when compared with liver DNA ODCs and CNS tissues prompted us to design methylation-specific
(Fig. 2B). Similar results were obtained using brain DNA (data not primers capable of discerning between Meth and DeMeth CpGs in posi-
shown). ODCs are considered the only source of MOG protein in tion + 2752 of the murine MOG gene (Table 1 and Fig. 3A). Plasmids
the brain. Therefore, in order to confirm that MOG-DNA was indeed containing the Meth and DeMeth DNA sequence of the MOG gene
demethylated in ODCs and not in other cell types, brains from nor- were used as controls for primer sensitivity and specificity. Serial dilu-
mal mice were digested and cells labeled with anti-O4 magnetic tion of plasmids over a 6 log dilution range showed the ability of
beads, followed by magnetic separation. FACS analysis of enriched DeMeth specific primers to detect DeMeth DNA even when diluted at
O4+ and O4− primary cell fractions were 92.6 ± 3.9% pure in four in- 1:1000 in Meth-DNA (Fig. 3B, R2 = 0.987, p b 0.0001). When tested
dependent O4+ preparations when compared with the O4− fraction on bisulfite-treated DNA from murine tissues, methylation-specific
(Fig. 2C and D). Purified O4+ and O4− cells were lysed and genomic primers showed a strong signal in both brain and spinal cord tissues
DNA purified and bisulfite converted. Analysis of MOG-DNA methyl- when compared with liver and kidney DNA (Fig. 3C. ANOVA
ation state in both fractions revealed unique methylation patterns in p b 0.0001. Liver vs. Brain or SC p b 0.001. Kidney vs. Brain or SC
O4+ cells but not in O4− cells (Fig. 2E). Schwann cells mediate the p b 0.001). Comparison of purified O4+ ODCs with both O4− and
myelination of peripheral nerves and are MOG negative. Analysis of SCW10 showed similar results, with O4+ cells producing a statistically
the DNA methylation patterns in DNA from murine Schwann cell higher signal when compared with either cell (Fig. 3D. ANOVA p =
line and liver (negative control) showed a complete methylation of 0.0008, O4+ vs. O4− p b 0.01. O4+ vs. Schwann cells p b 0.01). Taken to-
four CpG dinucleotides in MOG gene, demonstrating the fact that gether, these results show the ability of methylation-specific primers to

Fig. 3. Methylation-specific primers display high specificity and sensitivity and can detect demethylated MOG-DNA in murine brain, spinal cord, and O4+ cells. A. A depiction of MOG gene
region utilized for mouse methylation-specific qPCR analysis; cytosine at bp +2752 from MOG transcription start site incorporated into reverse primer sequence. B. Methylation-specific
primers tested using plasmids containing methylated and demethylated murine MOG-DNA inserts over a wide range of serial dilutions (R2 = 0.987, p b 0.0001). C. Methylation-specific
primers were used in qPCR with bisulfite treated DNA from murine liver, kidney, brain, and spinal cord. Three independent analyses used to compute DMI averages; Liver vs. Brain/Spinal
Cord p b 0.001, Kidney vs. Brain/Spinal Cord p b 0.001. D. Methylation-specific primers were used in qPCR with murine O4+ cells, O4− cells, and SW10 Schwann cells. Three independent
analyses used to compute DMI averages; ANOVA p = 0.0008, O4+ vs. O4− p b 0.01, O4+ vs. SW10 Schwann cells p b 0.01.
232 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235

identify DeMeth MOG-DNA in ODCs, suggesting that these primers may 3.4. Human MOG-DNA is Demethylated in the Brain and Can Be Detected
be used to detect ODC-derived cfDNA in the serum. by Methylation-specific Primers in Primary Oligodendrocyte Precursor Cells

3.3. Circulating Free DeMeth MOG-DNA Is Detected in Sera of Mice with The detection of DeMeth MOG-DNA in mouse CNS and primary
Widespread Demyelination ODCs suggested that MOG may be demethylated in CNS and ODCs of
human origin. The demethylated CpG pairs detected in the mouse
To examine whether DeMeth-specific primers can be used to detect assay are evolutionarily conserved between both human and mouse
DeMeth MOG in the blood, healthy C57BL/6 mice were injected with a (Fig. 5A). Analysis of bisulfite-treated DNA from human brain revealed
plasmid containing an insert of DeMeth DNA of the MOG gene. Blood the presence of demethylated CpG dinucleotides while these CpG
was collected 5 min post injection by heart puncture and circulating pairs were fully methylated in DNA from human liver (Fig. 5B). This
free serum DNA was purified. The level of DeMeth DNA in serum DNA analysis was supported by cloning of purified DNA from both brain
was expressed as DMI. Extracted DNA from plasmid treated mice and liver which showed the presence of unique demethylation patterns
showed a 24-fold increase in DeMeth MOG-DNA when compared on single DNA strands in the brain. These patterns were absent when
with untreated controls (Fig. 4A, Tx vs. Ctrl p b 0.012), demonstrating liver DNA was cloned and sequenced Fig. 5C). The difference in MOG-
the ability of DeMeth-specific primers to detect DeMeth MOG-DNA in DNA methylation between the brain and liver allowed for the design
the serum. Next, we used the cuprizone mouse model of MS. In this of methylation-specific primers for the human MOG gene (Table 2).
model, ODC injury is induced by treatment with the copper chelating Similar to the mouse assay, cloned plasmids containing the Meth and
agent, cuprizone. Cheek pouch bleeding was performed on a weekly DeMeth DNA sequences of the human MOG gene were synthesized,
basis following cuprizone administration to measure the levels of and mixed at varied ratios over 6 logarithmic concentrations followed
ODC-derived DeMeth MOG cfDNA in the blood. Baseline DMI (Fig. 4B, by analysis using methylation-specific qPCR. A positive correlation
solid line) was established using normal untreated mice (n = 9). Aver- was observed between DMI level and MOG DeMeth plasmid (Fig. 5D,
age MOG DMIs (n = 12, 6 mice per time point) reached their highest R2 = 0.992, p b 0.0001). Next, we compared DMI values of DNA from ol-
peak over baseline at Day 14, with increased DMI levels continuing igodendrocyte precursor cells (OPCs) and liver tissues. Methylation-
through Days 21 and 28 before returning to baseline levels on Day 35 specific primers yielded DMI values of OPC DNA that were 136-fold
(Fig. 4B). Luxol fast-blue staining of brain tissue sections showed a higher than those of liver DNA, showing a sensitivity and specificity of
gross demyelination in cuprizone-treated mice but not untreated the methylation specific primers (Fig. 5E). These differences were fur-
mice, confirming the dramatic effect on myelin in the brain (Fig. 4C). ther validated by comparing DMI values of plasmids containing Meth
These data demonstrate the ability of methylation-specific primers to or DeMeth MOG-DNA (data not shown). Taken together, these results
detect DeMeth MOG-DNA in the blood of mice treated with the ODC- show the presence of differentially methylated CpG dinucleotides in
specific toxin, supporting the use of DeMeth MOG-DNA as a biomarker the human brain and OPCs. Methylation-specific primers can success-
of ODC cell loss in MS. fully detect these DeMeth CpGs in DNA purified from the OPC,

Fig. 4. Methylation-specific primers can detect DeMeth MOG-DNA in the serum of plasmid injected mice and in cuprizone-treated mice which can be correlated with demyelination in
neural tissue. A. Bisulfite-treated DNA from the serum of mice injected with DeMeth MOG plasmid and non-injected mice was run on qPCR with methylation-specific primers; Tx vs.
Ctrl p b 0.012. B. Bisulfite-treated DNA from the serum of cuprizone-fed mice (n = 12) was run on qPCR with methylation-specific primers; MOG DMIs peak at Day 14 and remain
elevated over baseline until Day 35. C. Brain sections from cuprizone-fed and control mice stained for myelin using Luxol fast-blue; arrows indicate the region of native myelination.
 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235 233

suggesting that this approach may be used to detect ODC death in pa- 4. Discussion
tients with MS.
Differentially methylated cfDNA can be used as a biomarker of can-
3.5. Circulating Free DeMeth MOG-DNA Levels Are Increased in Patients cer and autoimmune diabetes. Here we report the presence of differen-
with Active Relapsing-remitting MS tially methylated CpG dinucleotides in the coding region of the MOG
gene in ODCs. These DeMeth CpGs are detected by methylation-specific
The ability of methylation-specific primers to successfully detect primers and measure the presence of ODC-derived MOG cfDNA in the
DeMeth MOG-DNA from OPCs suggests that this approach may be blood of mice with widespread demyelination. Moreover, methyla-
used to detect ODC-derived MOG cfDNA in patients with relapsing- tion-specific primers for human MOG-DNA can detect increased ODC
remitting MS (RRMS). Serum samples from RRMS patients with active loss in the sera of patients with active RRMS when compared with inac-
or inactive disease at the time of blood sampling were compared to tive disease and healthy controls. Our results describe a new assay for
sera of age and gender-matched healthy individuals for DMI values. measuring differentially methylated MOG cfDNA as a minimally inva-
MS patients with active and inactive disease showed similar age sive biomarker of ODC loss in RRMS. A biomarker capable of measuring
medians and disease duration (Fig. 6A: Age—Inactive = 40.2 ± 1.75, ODC loss may advance our ability to better diagnose MS progression in
Active = 41.30 ± 1.5; Fig. 6B: Disease duration—Inactive = 7.7 ± 1.3, conjunction with current imaging techniques, allow for better disease
Active = 8.6 ± 1.1). DNA was purified from 300 μl of sera, bisulfite-con- prognosis and modeling of ODC cell loss, and for measuring clinical effi-
verted, amplified by a first-step PCR reaction, and analyzed by qPCR cacy of novel therapeutic agents.
using methylation-specific primers. DMI values were calculated based MOG expression is found predominantly in ODCs (Coffey and
on three independent PCR reactions. DMI values of patients with active McDermott, 1997; Solly et al., 1996). Recent reports suggest that meth-
disease were 3.6 folds higher than inactive disease showing statistical ylation may play an active role in the transcription of MOG as the ex-
significance (Fig. 6C, DMI—Healthy controls = 9.23 × 10− 2 ± pression of methyl-CpG binding protein 2 can bind to Meth CpGs and
2.16 × 10−2, Inactive MS = 4.89 × 10− 2 ± 2.49 × 10− 2, Active = inhibit the expression of several myelin components, including MOG
18.36 × 10− 2 ± 5.18 × 10− 2. ANOVA p b 0.029; Inactive vs. Active (Solly et al., 1996). Despite these findings, little is known about the
p b 0.05). ROC analysis of samples showed an AUC of 0.7475 with 95% state of MOG-DNA methylation and the role methylation plays in gene
confidence interval of 0.59–0.9. This analysis reached statistical signifi- regulation. The data presented above show the presence of several
cance (Fig. 6D, p b 0.007). Taken together, our data show, for the first demethylated CpGs in DNA from primary mouse ODCs and primary
time, the utility of MOG cfDNA to detect active MS in patients with human OPCs when compared with other tissues. These demethylated
RRMS. dinucleotides were absent in other cell types and tissues and are

Fig. 5. Methylation-specific primers display high specificity and sensitivity and can detect demethylated MOG-DNA in human brain and liver. A. A depiction of MOG gene region utilized for
human methylation-specific qPCR analysis; cytosines at bps +2410 and +2430 from MOG transcription start site incorporated into reverse primer sequence. B. Sanger sequencing results
of bisulfite treated DNA from human tissues. Arrows point toward CpG sites where cytosines (C) are preserved in the methylated sample (Liver), or converted to thymines (T) in a sample
containing demethylated CpGs, leading to a mixed population of C′s and T's (Brain). Red arrows indicate CpGs incorporated into reverse primers. C. DNA from the brain is differentially
methylated in the MOG gene compared to DNA from the liver. Sequence analysis was performed on first-step PCR product of each sample, 13 clones of liver and 23 clones of brain
DNA are shown (○ represent demethylated cytosines; ●, methylated cytosines). Locations in relation to the MOG transcription start site are listed; methylation-specific human
primers incorporate the CpG sites at bp +2410 and +2430. D. Methylation-specific primers tested using plasmids containing methylated and demethylated human MOG-DNA inserts
over a wide range of serial dilutions (R2 = 0.992, p b 0.0001). E. Methylation-specific primers were used in qPCR with bisulfite treated DNA from human liver and oligodendrocyte
precursor cells. Three independent analyses used to compute DMI averages.
234 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235

evolutionarily conserved in the MOG coding region. While our data does reemergence of ODCs in the brain. Our overall findings suggest that
not address the role of methylation in the regulation of MOG gene ex- ODC loss can be detected during the acute or toxic phase of cuprizone
pression, our findings provide a strong support for it in both murine administration while in later time points the ablation of ODCs resulted
and human cells. The presence of DeMeth CpG pairs in MOG-DNA in reduced ODC-derived DNA in the blood.
from ODCs provided an opportunity to design methylation-specific The detection of ODC-derived DeMeth MOG-DNA in the cuprizone
primers for the detection of ODC-derived DNA. This approach was pre- mouse model lead us to examine the methylation state of the MOG
viously used to detect the loss of insulin producing β-cells in patients gene in human DNA. Analysis of an evolutionarily conserved region of
and animal models of type 1 diabetes (Akirav et al., 2011; Olsen et al., MOG-DNA in human brains and liver showed an identical demethyla-
2016) and for identifying cell loss in MS (Lehmann-Werman et al., tion pattern in CpG dinucleotides also present in mouse MOG. The fact
2016). Methylation specific primers showed a high degree of specificity that sequencing analysis of brain DNA revealed only a partial conversion
for Meth or DeMeth MOG-DNA when tested on both artificially synthe- of C → T suggests that non-ODC cells contribute to the Meth form of
sized DNA and DNA from ODC and brain origin, suggesting that these MOG-DNA while ODCs are the sole source of DeMeth MOG-DNA as
primers may detect as little as 1 copy of DeMeth DNA in 1000 copies seen in the O4+ and O4− cell fractions in the mouse. Methylation-
of Meth MOG-DNA. specific primers designed to distinguish between the Meth and DeMeth
The fact that these primers were able to detect DeMeth MOG-DNA in form of CpGs + 2410 and + 2430 showed a high degree of specificity
mice injected with artificial DNA confirmed the ability of the assay to and sensitivity similar to that of the mouse assay, and were able to de-
detect ODC-like DNA in vivo. To further test the ability of the primers tect DeMeth MOG in brain DNA at DMI levels nearly 10,000 folds higher
to detect ODC cell loss, we used the cuprizone model of ODC-toxicity than those observed in the liver. When tested, these primers showed a
and widespread demyelination in mice. In previous reports, ODC gene high DMI signal in OPCs when compared with liver, showing the ability
expression in the brain, used as an indicator of ODC function and of the assay to detect ODC-like DNA.
mass, was reduced early during treatment, with the reemergence of The high degree of sensitivity and specificity of the human methyla-
MAG-expressing cells in the brains of cuprizone-treated mice from tion-specific primers allowed us to analyze blood samples from 40 pa-
week 6 onwards post treatment (Lindner et al., 2008; Praet et al., tients with RRMS, with active (n = 20) or inactive (n = 20) disease.
2014). Additional studies report complete remyelination in the Healthy controls (n = 20) were also used to establish the baseline back-
cuprizone mice upon removal of the cuprizone diet. The remyelination ground DMI levels of normal subjects. Disease activity in RRMS patients
was complemented with the detection of several myelin proteins, was determined by clinical episodes and, in some cases, by the forma-
excluding MOG, whose expression was delayed for several weeks tion of new brain lesions in the CNS. While both RRMS patients with in-
(Lindner et al., 2008; Praet et al., 2014). Indeed, Methylation-specific active and active disease were similar in age, gender, and duration of
primers detected ODC-cell loss as early as 3 weeks post cuprizone treat- disease, patients with active RRMS had significant elevation in DMI
ment, with the highest increase detected at Day 14 post treatment. values. ROC analysis of active and inactive RRMS revealed average
Moreover, our results showed relatively low levels of ODC-derived AUC and preliminary cutoff values of specificity and sensitivity for
DeMeth MOG-DNA in the blood of mice from week 5 until week 8 assay performance. Additional studies consisting of larger patient co-
post-treatment, corresponding to previous reports showing the horts would be needed to validate these values and allow for their use

Fig. 6. Methylation-specific primers can detect elevated levels of demethylated MOG cfDNA in patients with RRMS. A. Age at diagnosis of relapsing-remitting multiple sclerosis for both
active and inactive disease groups. B. Duration of disease of relapsing-remitting multiple sclerosis for both active and inactive groups. C. Methylation-specific primers were used in qPCR
with bisulfite treated DNA extracted from sera from Healthy Controls, Inactive and Active RRMS patients; ANOVA p b 0.029, Inactive vs. Active p b 0.05. D. ROC analysis of samples showed
an AUC of 0.7475 with 95% confidence interval of 0.59–0.9. This analysis reached statistical significance (p b 0.007).
 J.A. Olsen et al. / EBioMedicine 10 (2016) 227–235 235

in the clinic. To our knowledge, this is the first report demonstrating the Akirav, E.M., Lebastchi, J., Galvan, E.M., et al., 2011. Detection of beta cell death in diabetes
using differentially methylated circulating DNA. Proc. Natl. Acad. Sci. U. S. A. 108 (47),
utility of differentially methylated ODC-derived cfDNA as a biomarker of 19018–19023.
cell loss in RRMS where DMI levels reflect disease activity. Bunge, R.P., 1968. Glial cells and the central myelin sheath. Physiol. Rev. 48 (1), 197–251.
This report describes the differential methylation of the MOG gene in Bunge, M.B., Bunge, R.P., Pappas, G.D., 1962. Electron microscopic demonstration of con-
nections between glia and myelin sheaths in the developing mammalian central ner-
ODCs and brain tissues of both mouse and human origin. The unique vous system. J. Cell Biol. 12, 448–453.
DeMeth patterns in ODCs provide a new biomarker for the detection Chen, X.Q., Stroun, M., Magnenat, J.L., et al., 1996. Microsatellite alterations in plasma DNA
of ODC-cell loss in MS. Currently, diagnosis of MS is based on several of small cell lung cancer patients. Nat. Med. 2 (9), 1033–1035.
Coffey, J.C., McDermott, K.W., 1997. The regional distribution of myelin oligodendrocyte
criteria (Polman et al., 2011), including identification of lesions by MRI glycoprotein (MOG) in the developing rat CNS: an in vivo immunohistochemical
and positive identification of biomarkers in cerebral spinal fluid (CSF). study. J. Neurocytol. 26 (3), 149–161.
MS is a financially burdening disease (Adelman et al., 2013; Kobelt et Crowley, E., Di Nicolantonio, F., Loupakis, F., Bardelli, A., 2013. Liquid biopsy: monitoring
cancer-genetics in the blood. Nat. Rev. Clin. Oncol. 10 (8), 472–484.
al., 2006), making MRIs cost-prohibitive for many underinsured pa-
Felts, P.A., Baker, T.A., Smith, K.J., 1997. Conduction in segmentally demyelinated mam-
tients. Despite the ongoing identification of useful biomarkers in CSF malian central axons. J. Neurosci. 17 (19), 7267–7277.
testing (Teunissen et al., 2015; Fitzner et al., 2015), lumbar punctures Ferguson, B., Matyszak, M.K., Esiri, M.M., Perry, V.H., 1997. Axonal damage in acute mul-
are an invasive procedure which prevents collecting samples from tiple sclerosis lesions. Brain J. Neurol. 120 (Pt 3), 393–399.
Fitzner, B., Hecker, M., Zettl, U.K., 2015. Molecular biomarkers in cerebrospinal fluid of
healthy individuals. The collection of cfDNA is a cost effective, minimally multiple sclerosis patients. Autoimmun. Rev. 14 (10), 903–913.
invasive procedure that can be implemented for diagnosis, prognosis, Gardinier, M.V., Amiguet, P., Linington, C., Matthieu, J.M., 1992. Myelin/oligodendrocyte
and monitoring of treatment efficacy. The ability to measure ODC loss glycoprotein is a unique member of the immunoglobulin superfamily. J. Neurosci.
Res. 33 (1), 177–187.
may prove beneficial not only to MS but also to other disease afflicting Johns, T.G., Bernard, C.C., 1999. The structure and function of myelin oligodendrocyte gly-
ODC survival such as progressive multifocal leukoencephalopathy coprotein. J. Neurochem. 72 (1), 1–9.
(PML). Jones, P.A., 2012. Functions of DNA methylation: islands, start sites, gene bodies and be-
yond. Nat. Rev. Genet. 13 (7), 484–492.
Supplementary data to this article can be found online at http://dx. Kobelt, G., Berg, J., Lindgren, P., Fredrikson, S., Jonsson, B., 2006. Costs and quality of life of
doi.org/10.1016/j.ebiom.2016.06.031. patients with multiple sclerosis in Europe. J. Neurol. Neurosurg. Psychiatry 77 (8),
918–926.
Lebar, R., Lubetzki, C., Vincent, C., Lombrail, P., Boutry, J.M., 1986. The M2 autoantigen of
Funding central nervous system myelin, a glycoprotein present in oligodendrocyte mem-
brane. Clin. Exp. Immunol. 66 (2), 423–434.
This work was supported by the National Multiple Sclerosis Society Lehmann-Werman, R., Neiman, D., Zemmour, H., et al., 2016. Identification of tissue-spe-
cific cell death using methylation patterns of circulating DNA. Proc. Natl. Acad. Sci. U.
(Grant No. PP2130), the Marilyn Hilton Award for Innovation in MS S. A. 113 (13), E1826–E1834.
Research, and Winthrop University Hospital Pilot Award. Granting Li, L.C., Dahiya, R., 2002. MethPrimer: designing primers for methylation PCRs. Bioinfor-
agencies did not play a role in the design, data collection, or analysis matics 18 (11), 1427–1431.
Lindner, M., Heine, S., Haastert, K., et al., 2008. Sequential myelin protein expression dur-
of data. ing remyelination reveals fast and efficient repair after central nervous system demy-
elination. Neuropathol. Appl. Neurobiol. 34 (1), 105–114.
The Duality of Interest Linnington, C., Webb, M., Woodhams, P.L., 1984. A novel myelin-associated glycoprotein
defined by a mouse monoclonal antibody. J. Neuroimmunol. 6 (6), 387–396.
Nawroz, H., Koch, W., Anker, P., Stroun, M., Sidransky, D., 1996. Microsatellite alterations
The authors declare that there is no duality of interest associated in serum DNA of head and neck cancer patients. Nat. Med. 2 (9), 1035–1037.
with this manuscript. Noseworthy, J.H., Lucchinetti, C., Rodriguez, M., Weinshenker, B.G., 2000. Multiple sclero-
sis. N. Engl. J. Med. 343 (13), 938–952.
Olsen, J.A., Akirav, E.M., 2015. Remyelination in multiple sclerosis: cellular mechanisms
Author Contributions and novel therapeutic approaches. J. Neurosci. Res. 93 (5), 687–696.
Olsen, J.A., Kenna, L.A., Spelios, M.G., Hessner, M.J., Akirav, E.M., 2016. Circulating differen-
tially methylated amylin DNA as a biomarker of beta-cell loss in type 1 diabetes. PLoS
EMA and JAO conceived and designed the experiments. JAO, LAK, One 11 (4), e0152662.
RCT and EMA performed the experiments. JAO and EMA analyzed the Polman, C.H., Reingold, S.C., Banwell, B., et al., 2011. Diagnostic criteria for multiple scle-
data. JAO, LAK, MGS, MMS and EMA wrote the paper. rosis: 2010 revisions to the McDonald criteria. Ann. Neurol. 69 (2), 292–302.
Praet, J., Guglielmetti, C., Berneman, Z., Van der Linden, A., Ponsaerts, P., 2014. Cellular and
molecular neuropathology of the cuprizone mouse model: clinical relevance for mul-
Acknowledgments tiple sclerosis. Neurosci. Biobehav. Rev. 47, 485–505.
Solly, S.K., Thomas, J.L., Monge, M., et al., 1996. Myelin/oligodendrocyte glycoprotein
(MOG) expression is associated with myelin deposition. Glia 18 (1), 39–48.
The authors thank the patients that participated in the studies Teunissen, C.E., Malekzadeh, A., Leurs, C., Bridel, C., Killestein, J., 2015. Body fluid bio-
markers for multiple sclerosis-the long road to clinical application. Nat. Rev. Neurol.
11 (10), 585–596.
References Waxman, S.G., 1977. Conduction in myelinated, unmyelinated, and demyelinated fibers.
Arch. Neurol. 34 (10), 585–589.
Adelman, G., Rane, S.G., Villa, K.F., 2013. The cost burden of multiple sclerosis in the Unit-
ed States: a systematic review of the literature. J. Med. Econ. 16 (5), 639–647.