Identification and temporal stability of conformational epitopes of autoantibodies against Myelin Oligodendrocyte Glycoprotein recognized by patients with different inflammatory central nervous system diseases

Myelin Oligodendrocyte Glycoprotein (MOG) is one of the few
proteins known to be localized on the outermost sheath of central
nervous system (CNS) myelin. Due to this localization, MOG is
accessible to antibodies. Anti-MOG antibodies are demyelinating and
enhance clinical symptoms in a number of animal models of CNS
inflammation. Autoantibodies recognizing conformationally intact
MOG are found in different inflammatory diseases of the CNS, but
their antigenic epitopes had not been mapped. In this work, 9
variants of MOG with an intracellular enhanced green fluorescent
protein (EGFP) tag were expressed on the cell surface of human HeLa
cells and used to analyze sera from 111 patients (104 children, 7
adults), who had antibodies recognizing cell-bound human MOG. These
patients had different diseases, namely acute disseminated
encephalomyelitis (ADEM), one episode of transverse myelitis or
optic neuritis, multiple sclerosis (MS), anti-aquaporin-4
(AQP4)–negative neuromyelitis optica (NMO), and chronic relapsing
inflammatory optic neuritis (CRION). The expression levels of the
mutants were comparable and cells with a defined expression level
(fluorescence intensity in the EGFP channel of 102-103) were gated.
Each MOG-mutant was recognized by at least one MOG-specific mAb.
This allowed the comparison of binding to the different mutants. In
order to assess the reproducibility of the system, binding of the
111 sera to the mutants was analyzed up to three times in
independent experiments, yielding a very good reproducibility of
the binding percentage with an absolute SD of 7.8% in the case of
low recognition of a mutant and a relative SD of 20% in the case of
high recognition of a mutant. The applied variants of MOG gave
insight into epitope recognition of 98 patients. All epitopes
identified in this work were located at loops connecting the
ß-strands of MOG. The immunodominant epitope of human anti-MOG
antibodies was at the membrane-proximal CC’-loop containing aa42,
which is not present in rodent MOG. This loop was recognized by
about half of all patients. Overall, seven epitope patterns were
distinguished, including the one mainly recognized by mouse mAbs at
the FG-loop around aa104. Evidence from mouse models of CNS
inflammation shows that anti-MOG antibodies recognizing different
epitopes can be demyelinating and thus pathogenic. This suggests
that not only those antibodies recognizing the same epitope of MOG
as the pathogenic mAbs (i.e. the FG-loop), but also the ones
recognizing the CC'-loop are pathogenic in humans, as both epitopes
allow for the recognition of cell-bound MOG. In half of the
patients, the anti-MOG response was directed to a single epitope.
To analyze the effect of glycosylation on the recognition of MOG by
human autoantibodies, a “non-glycosylation mutant” N31D was made.
Digestion with PNGaseF and Western blot analysis confirmed that N31
was the only used N-glycosylation site of the MOG constructs in
HeLa cells. Glycosylation of MOG was not needed for antibody
binding, but 8% of the patients recognized deglycosylated MOG at
least two-fold better. The epitope specificity was not linked to
certain disease entities. The individual epitope recognition
patterns stayed constant in 11 analyzed patients over an
observation period of up to 5 years without evidence for
intramolecular epitope spreading. Some patients with acute
syndromes had anti-MOG IgG at disease onset, but rapidly lost their
anti-MOG IgG reactivity. These patients were able to generate a
long-lasting IgG response to measles and rubella virus vaccine
indicating that the loss of anti-MOG reactivity was not reflective
of a lack of capacity for longstanding IgG responses. Human
anti-MOG antibodies are mainly of the IgG1 isotype, which can
activate complement and antibody dependent cellular cytotoxicity.
Upon binding to MOG in the CNS, human anti-MOG antibodies are hence
expected to cause demyelination. Transfer experiments with purified
human anti-MOG antibodies have not been performed yet. The fact
that the majority of human anti-MOG antibodies did not recognize
rodent MOG has implications for animal studies. Using the described
assay will help to identify patient samples appropriate for these
transfer experiments and finally lead to the formal proof of the
pathogenicity of human anti-MOG antibodies. This work also gives
important information for future detection of potential mimotopes
and the development of anti-MOG antibody detection assays and might
pave the way to antigen-specific depletion.