Aquaporin 4-Specific T Cells in Neuromyelitis Optica Exhibit a Th17 Bias and Recognize Clostridium ABC Transporter

Abstract

Aquaporin 4 (AQP4)-specific autoantibodies in neuromyelitis optica (NMO) are immunoglobulin (Ig)G1, a T cell-dependent Ig subclass, indicating that AQP4-specific T cells participate in NMO pathogenesis. Our goal was to identify and characterize AQP4-specific T cells in NMO patients and healthy controls (HC).
Peripheral blood T cells from NMO patients and HC were examined for recognition of AQP4 and production of proinflammatory cytokines. Monocytes were evaluated for production of T cell-polarizing cytokines and expression of costimulatory molecules.
T cells from NMO patients and HC proliferated to intact AQP4 or AQP4 peptides (p11-30, p21-40, p61-80, p131-150, p156-170, p211-230, and p261-280). T cells from NMO patients demonstrated greater proliferation to AQP4 than those from HC, and responded most vigorously to p61-80, a naturally processed immunodominant determinant of intact AQP4. T cells were CD4(+), and corresponding to association of NMO with human leukocyte antigen (HLA)-DRB1*0301 and DRB3, AQP4 p61-80-specific T cells were HLA-DR restricted. The T-cell epitope within AQP4 p61-80 was mapped to 63-76, which contains 10 residues with 90% homology to a sequence within Clostridium perfringens adenosine triphosphate-binding cassette (ABC) transporter permease. T cells from NMO patients proliferated to this homologous bacterial sequence, and cross-reactivity between it and self-AQP4 was observed, supporting molecular mimicry. In NMO, AQP4 p61-80-specific T cells exhibited Th17 polarization, and furthermore, monocytes produced more interleukin 6, a Th17-polarizing cytokine, and expressed elevated CD40 and CD80 costimulatory molecules, suggesting innate immunologic dysfunction.
AQP4-specific T-cell responses are amplified in NMO, exhibit a Th17 bias, and display cross-reactivity to a protein of an indigenous intestinal bacterium, providing new perspectives for investigating NMO pathogenesis.

Full text

RAPID COMMUNICATION
Aquaporin 4-Specific T Cells in
Neuromyelitis Optica Exhibit a Th17
Bias and Recognize Clostridium
ABC Transporter
Michel Varrin-Doyer, PhD,
1
Collin M. Spencer, BS,
1
Ulf Schulze-Topphoff, PhD,
1
Patricia A. Nelson, PhD,
1
Robert M. Stroud, PhD,
2
Bruce A. C. Cree, MD, PhD,
1
and Scott S. Zamvil, MD, PhD
1
Objective: Aquaporin 4 (AQP4)-specific autoantibodies in neuromyelitis optica (NMO) are immunoglobulin (Ig)G1, a
T cell-dependent Ig subclass, indicating that AQP4-specific T cells participate in NMO pathogenesis. Our goal was
to identify and characterize AQP4-specific T cells in NMO patients and healthy controls (HC).
Methods: Peripheral blood T cells from NMO patients and HC were examined for recognition of AQP4 and
production of proinflammatory cytokines. Monocytes were evaluated for production of T cell-polarizing cytokines
and expression of costimulatory molecules.
Results: T cells from NMO patients and HC proliferated to intact AQP4 or AQP4 peptides (p11–30, p21–40, p61–80,
p131–150, p156–170, p211–230, and p261–280). T cells from NMO patients demonstrated greater proliferation to
AQP4 than those from HC, and responded most vigorously to p61–80, a naturally processed immunodominant
determinant of intact AQP4. T cells were CD4
þ
, and corresponding to association of NMO with human leukocyte
antigen (HLA)-DRB1*0301 and DRB3, AQP4 p61–80-specific T cells were HLA-DR restricted. The T-cell epitope
within AQP4 p61–80 was mapped to 63–76, which contains 10 residues with 90% homology to a sequence within
Clostridium perfringens adenosine triphosphate-binding cassette (ABC) transporter permease. T cells from NMO
patients proliferated to this homologous bacterial sequence, and cross-reactivity between it and self-AQP4 was
observed, supporting molecular mimicry. In NMO, AQP4 p61–80-specific T cells exhibited Th17 polarization, and
furthermore, monocytes produced more interleukin 6, a Th17-polarizing cytokine, and expressed elevated CD40 and
CD80 costimulatory molecules, suggesting innate immunologic dysfunction.
Interpretation: AQP4-specific T-cell responses are amplified in NMO, exhibit a Th17 bias, and display cross-reactivity
to a protein of an indigenous intestinal bacterium, providing new perspectives for investigating NMO pathogenesis.
ANN NEUROL 2012;72:53–64
Neuromyelitis optica (NMO) is a rare, disabling,
sometimes fatal, central nervous system (CNS)
demyelinating disease characterized by severe attacks of
optic neuritis and transverse myelitis.
1
NMO is consid-
ered to be primarily a humoral autoimmune disease, as a
majority of NMO patients develop autoantibodies
(NMO immunoglobulin [Ig]G) against aquaporin 4
(AQP4),
2
the predominant CNS water channel, which is
abundantly expressed on astrocytes. AQP4-specific anti-
bodies in NMO serum are IgG1, a subclass of mature
IgG that requires help from T cells,
3
indicating that
AQP4-specific CD4
þ
T cells participate in the genesis of
this adaptive humoral response. Passive transfer of
AQP4-specific antibodies alone did not produce CNS
pathology, but did promote development of NMO-like
lesions in recipient animals when CNS inflammation was
View this article online at wileyonlinelibrary.com. DOI: 10.1002/ana.23651
Received Apr 30, 2012, and in revised form May 8, 2012. Accepted for publication May 21, 2012.
Address correspondence to Dr Zamvil, Department of Neurology, University of California at San Francisco, 675 Nelson Rising Lane, NS-215A, San
Francisco, CA 94143. E-mail: zamvil@ucsf.neuroimmunol.org
From the Departments of
1
Neurology and
2
Biochemistry, University of California at San Francisco, San Francisco, CA.
Additional supporting information can be found in the online version of this article.
V
C2012 American Neurological Association 53

induced by myelin-specific T cells.
4,5
T cells are detected
within active NMO lesions.
6
Further, NMO lesions are
characterized by an abundance of eosinophils and neutro-
phils, and elevated levels of IL-17 have been associated
with NMO,
7
suggesting involvement of Th17 cells.
However, as no previous studies have identified or char-
acterized proliferative AQP4-specific T cells in NMO
patients, their potential role in NMO pathogenesis is
largely unknown.
In this report, we first identified peripheral blood T
cells from NMO patients and healthy controls (HC) that
proliferated in response to discrete AQP4 peptides or
intact AQP4. T cells from NMO patients demonstrated
greater proliferation to this autoantigen than those from
HC, and responded most frequently to p61–80. After
defining the p61–80 core T-cell determinant, residues
63–76, we conducted a homology search with known
microbes. We discovered that AQP4 p63–76 contains
strong homology to aa 204–217 of an adenosine triphos-
phate-binding cassette (ABC) transporter permease of
Clostridium perfringens, a bacterial species that contains
both commensal and pathogenic strains for humans. T
cells from NMO patients responded to the homologous
ABC transporter peptide and exhibited cross-reactivity
between this foreign antigen and AQP4 p63–76, findings
that support molecular mimicry. When compared to
HC, AQP4 p61–80-specific T cells from NMO patients
exhibited Th17 polarization. Monocytes from NMO
patients produced significantly higher levels of the Th17-
polarizing cytokine interleukin (IL)-6, suggesting that
immunologic dysfunction in NMO may also include the
innate immune compartment. Collectively, our findings
establish that AQP4-specific proliferative T cells exist,
and support a Th17 bias in the adaptive immune
response in NMO. Our demonstration of T-cell molecu-
lar mimicry may stimulate further evaluation of the
potential role of the Clostridium species in NMO
pathogenesis.
Patients and Methods
Patients
Fifteen NMO patients (12 females and 3 males, aged 44.3 6
13.8 years) fulfilling Mayo Clinic diagnostic criteria
8
and 9 HC
(5 females and 4 males, aged 40.8 610.7 years) were recruited
from the University of California at San Francisco (UCSF)
Multiple Sclerosis Center. A majority of NMO patients had
been treated with rituximab,
9
and none had been treated with
azathioprine, mycophenolate mofetil, cyclophosphamide, or
other immunosuppressive medications. None of the patients
had received steroids within 2 months preceding blood draws.
Blood was collected by venipuncture. This study was approved
by the UCSF Committee on Human Research (Protocol # 10-
00650), and written informed consent was obtained from sub-
jects prior to enrollment.
T-Cell Proliferation Assays
Peripheral blood mononuclear cells (PBMC) were isolated by
density gradient centrifugation over Ficoll (Ficoll-Paque PLUS;
GE Healthcare, Milwaukee, WI) according to the manufac-
turer’s instructions. T-cell proliferation was evaluated by
[
3
H]thymidine incorporation or 5,6-carboxylfluorescein diace-
tate succinimidyl ester (CFSE) dilution assays. In thymidine
incorporation assays, PBMC were cultured with antigens in 96-
well plates at either 1 10
5
cells (AQP4 pools, in at least 10
wells) or 5 10
5
cells (individual peptides, in duplicate) per
well for 6 days. Cultures were then pulsed with [
3
H]thymidine
and harvested 18 hours later. Positive wells were defined as hav-
ing counts per minute (cpm) values greater than control cpm
average values þ3 standard deviations or stimulation index
(SI) >2. Alternatively, PBMC were stained with 0.5lM CFSE
(Invitrogen, Carlsbad, CA), according to the manufacturer’s
instructions. Cells were cultured in the presence of antigens for
10 days. T-cell proliferation was assessed by flow cytometric
evaluation of CFSE dilution. Proliferation was expressed as the
cell division index (defined as the number of CFSE
low
T cells
cultured with antigen/number of CFSE
low
T cells without anti-
gen). In all cases, culture medium consisted of X-VIVO 15
(Lonza, Walkersville, MD) supplemented with penicillin
(100U/ml) and streptomycin (0.1mg/ml).
Antigens
Peptides were synthesized by Genemed Synthesis (San Antonio,
TX) with purity >95% by high-performance liquid chromatog-
raphy analysis. Overlapping AQP4 20-mer peptides were offset
by 10 amino acids (Supplementary Table). Peptides correspond-
ing to certain hydrophobic AQP4 sequences were synthesized
in overlapping 15-mer peptide pairs. Truncated peptides within
the 61–80 region (p61–78 GTEKPLPVDMVLISLCFG; p61–
76 GTEKPLPVDMVLISLC; p61–74 GTEKPLPVDMVLIS;
p61–72 GTEKPLPVDMVL; p63–80 EKPLPVDMVLISLC
FGLS; p65–80 PLPVDMVLISLCFGLS; p67–80 PVDMVL
ISLCFGLS; p69–80 DMVLISLCFGLS), AQP4 p63–76
(EKPLPVDMVLISLC), and bacterial peptide ABC-transporter
permease (TP) p204–217 (FIILPVSMVLISLV) were as quoted.
Full-length recombinant human (rh) AQP4 (1–323) was
expressed in Pichia pastoris and purified as described.
10
Tetanus
toxoid was obtained from List Biological Laboratories (Camp-
bell, CA).
Flow Cytometry Analysis
Single-cell suspensions were incubated with human serum to
prevent nonspecific antibody binding, then stained with anti-
bodies against CD3, CD4, CD8, CD25, major histocompati-
bility complex (MHC) class II, CD40, CD80, and CD86
(eBioscience, San Diego, CA and BD Biosciences, Mississauga,
ON, Canada). Intracellular cytokine production by CD4
þ
T
cells and antigen-presenting cells (APC) was analyzed by moni-
toring the expression of interferon (IFN)-c, IL-17, IL-6, IL-1b,
ANNALS of Neurology
54 Volume 72, No. 1

and IL-10 (1:100) (eBioscience). Foxp3 staining was performed
according to the manufacturer’s protocol (eBioscience). For in-
tracellular cytokine staining, T cells were stimulated with phor-
bol 12-myristate 13-acetate (50ng/ml) plus ionomycin (500ng/
ml) in the presence of GolgiStop (1ll/ml) (BD Biosciences).
CD14
þ
cells were stimulated with lipopolysaccharide (LPS;
1lg/ml; Sigma-Aldrich, St Louis, MO) for 4 or 20 hours in
the presence of GolgiStop. Cells were analyzed by flow cytome-
try on a FACS Canto flow cytometer (BD Biosciences).
Blocking of HLA Alleles with Antibodies
Inhibition of the proliferation of PBMC to AQP4 p61–80 and
rhAQP4 was studied by using mouse monoclonal anti–HLA-
DR (clone G46-6; BD Biosciences), anti–HLA-DQ (clone
HG-38; Abcam, Cambridge, MA), anti–HLA-DP (clone B7/
21; Abcam), and isotype control (clone G155-178; BD Bio-
sciences). Antibodies (1lg/ml) were added to CFSE-stained
PBMC cultures 1 hour before addition of antigens.
Antigen Recall Experiments
PBMC were initially stimulated with antigens. After 10 days,
cells were restimulated with rhAQP4 (5lg/ml), AQP4 peptides,
or bacterial peptide (10lg/ml), in the presence of irradiated
autologous APC. Following 3 days of stimulation, cultures were
pulsed with [
3
H]thymidine and harvested 18 hours later. SI was
calculated by dividing cpm in wells with antigen by cpm in
control wells with no antigen for each assay test group.
Analyses for Protein Sequence Homology and
MHC Core Binding Motifs
Sequence similarities between AQP4 and other proteins were
addressed using the protein–protein Basic Local Alignment
Search Tool from the National Center for Biotechnology Infor-
mation (NCBI). The prediction of the core binding motif
within the AQP4 61–80 sequence for HLA-DRB1*0301 and
HLA-DRB3*0202 was performed with netMHCII-1.1
11
and
net MHCII-2.2,
12
programs that utilize relative affinities of
identified determinants from the Immune Epitope Database.
HLA Typing
High-resolution HLA typing was performed by the UCSF
Immunogenetics and Transplantation Laboratory (UCSF
Department of Surgery). The following HLA loci were analyzed
using sequence-based typing: DRB1, DRB3/4/5, DQA1,
DQB1, DPA1, and DPB1. Sequence ambiguities outside exon
2 were resolved.
Statistics
Statistical analysis was performed using either Prism (GraphPad
Software, La Jolla, CA) or STATA (StataCorp, College Station,
TX) software. The nonparametric Mann–Whitney Utest was
used to compare data. Paired ttests were performed to compare
cpm values with antigens to control values with no antigens
presented in Figure 3E. A value of p0.05 was considered
significant.
Results
T Cells from NMO Patients Recognize Discrete
AQP4 Determinants and Are Restricted by HLA-
DR Molecules
In general, antigen-specific T cells recognize linear pep-
tide fragments of 10–15 amino acids in association with
MHC (HLA) proteins expressed on APC.
13
To identify
AQP4-specific T cells in NMO patients, we initially
tested proliferation of PBMC to a library of 32 synthetic
overlapping 15-mer and 20-mer peptides encompassing
the 323-amino acid sequence of full-length human
AQP4 (M1 isoform). Here, we studied separate pools
containing 5 overlapping AQP4 peptides. By [
3
H]thymi-
dine incorporation, we detected more frequent prolifera-
tive responses in primary cultures to AQP4 pools 1–55,
46–100, 126–170, 201–250, and 241–300 (Fig 1A).
Lymphocytes from HC also proliferated to some of these
pools, and exhibited comparable responses to tetanus
toxoid.
Having identified candidate regions of AQP4 con-
taining T-cell determinants, we then tested proliferative
responses of NMO patients to individual AQP4 peptides.
T-cell determinants were identified within p21–40, p61–
80, p131–150, p156–170, and p211–230, which corre-
sponded to intracellular, extracellular, and transmembrane
sequences of AQP4 (see Fig 1). Interestingly, 3 of these
AQP4 determinants, p61–80, p131–150, and p211–230
are respectively located in extracellular A, C, and E loops,
AQP4 domains targeted by NMO-IgG.
14
The fluores-
cent dye 5,6-carboxylfluorescein diacetate succinimidyl
ester (CFSE) dilution assay is considered a more power-
ful and sensitive method for detecting proliferation of
rare autoantigen-specific human T cells than the tradi-
tional [
3
H]thymidine incorporation.
15
Using this
approach, we examined responses to individual AQP4
peptides identified in our initial screening, and also to
AQP4 T-cell determinants common to mouse strains
with distinct MHC haplotypes.
16,17
We detected a robust
proliferative T-cell response to p61–80, which is located
within the extracellular A loop, in all NMO patients
tested. T-cell responses were observed to AQP4 p21–40,
p156–170, p11–30, and p261–280, although we did not
detect substantial proliferation to the latter 2 peptides in
our initial [
3
H]thymidine incorporation assays. T cells
from HC also recognized these AQP4 peptides, but
again, the proliferative responses were both lower and
less frequent than in NMO patients. Proliferating AQP4-
specific T cells were predominantly CD4
þ
, and the pro-
portion of CD4
þ
T cells that responded to AQP4 p61–
80 was higher in NMO patients than HC.
Presentation of native protein antigens by APC
generally requires proteolytic processing.
18–21
Therefore,
Varrin-Doyer et al: AQP4-Specific T Cells in NMO
July 2012 55

FIGURE 1: T cells from neuromyelitis optica (NMO) patients recognize discrete determinants of aquaporin 4 (AQP4). Periph-
eral blood mononuclear cells (PBMC) were tested for proliferation to (A) pools of AQP4 peptides (n 58 NMO and n 53
healthy controls [HC]) and to (B) individual AQP4 peptides identified from those pools. In A and B, PBMC were cultured for 6
days in the presence of AQP4 pools (10lg/ml) or AQP4 peptides (10lg/ml), respectively, then pulsed with [
3
H]thymidine and
harvested 18 hours later. In A, positive wells were defined as values >control counts per minute average values 13 standard
deviations. (C) AQP4 determinants are represented within a human AQP4 topological diagram using TOPO2 transmembrane
protein display software (http://www.sacs.ucsf.edu/TOPO2/).
50
(D, E) PBMC were examined by 5,6-carboxylfluorescein diace-
tate succinimidyl ester (CFSE) dilution for proliferation to individual AQP4 peptides (10lg/ml), recombinant human (rh) AQP4
(5lg/ml), or in E, tetanus toxoid (TT; 1lg/ml) after 10 days of culture. CFSE was measured in CD3
1
, CD4
1
, and CD8
1
T cells
by flow-activated cell sorting and quantified by cell division index (CDI). CDI >2(broken lines) was considered positive. (F)
Recall T-cell proliferation ([
3
H]thymidine incorporation) to individual AQP4 peptides (10lg/ml) or rhAQP4 (5lg/ml) was
detected after initial stimulation with rhAQP4 (5lg/ml) for 10 days. In A and E, error bars indicate standard error of the mean;
in B, D, and F, horizontal lines indicate mean values. *p<0.05 Mann–Whitney Utest. Ag 5antigens.
ANNALS of Neurology
56 Volume 72, No. 1

we examined whether the AQP4 peptides we identified
contained natural T-cell determinants of intact AQP4.
When T cells initially stimulated with rhAQP4 were
tested for recall responses to individual AQP4 peptides,
we observed proliferation to AQP4 p21–40 and p61–80,
indicating that these are naturally processed determinants
of AQP4 (see Fig 1F). Among peptides that we exam-
ined, AQP4 p61–80 was clearly immunodominant. Sev-
eral studies have identified over-representation of HLA-
DPB1*0501, HLA-DRB1*0301, or HLA-DRB3 in
NMO patients,
22–24
suggesting that these MHC II alleles
could serve as restriction elements for CD4
þ
T cells in
NMO. We also identified a high representation of these
HLA alleles in our patient cohort; in particular, we noted
an over-representation of HLA-DRB3*0202 among
NMO subjects (Table). Using MHC II-blocking antibod-
ies, we observed that T-cell proliferative responses to
AQP4 p61–80 were inhibited by anti–HLA-DR, but
were not statistically inhibited by anti–HLA-DQ or anti–
HLA-DP, demonstrating that HLA-DR molecules serve
as restriction elements for T cells that recognize this de-
terminant (Fig 2). Proliferation of AQP4 p61–80-specific
T cells from HLA-matched HC (see Table) was also
inhibited by anti–HLA-DR antibodies. Furthermore, a
similar MHC II-restriction profile was observed after
stimulating T cells from NMO patients with rhAQP4,
suggesting that other AQP4 determinants may also be re-
stricted by HLA-DR molecules.
AQP4 p63–76-Specific T Cells Cross-React with
C. perfringens ABC-TP p204–217
To characterize the fine specificity of AQP4 p61–80-spe-
cific T cells, we examined proliferation to truncated pep-
tides corresponding to sequences within this region.
AQP4 p61–80-specific T cells proliferated in response to
p61–78 and p61–76, but not to p61–74 or p61–72 (see
Fig 3). Shorter AQP4 peptides truncated from the N-ter-
minal sequence, p65–80, p67–80, and p69–80, also
stimulated proliferation of p61–80-specific T cells,
although less efficiently than p63–80. Collectively, these
findings indicated that p63–76 contained the core deter-
minant of AQP4 61–80. In this regard, we observed that
p61–80-specific T cells responded nearly as efficiently to
p63–76 as to p61–80. Interestingly, AQP4 63–76 con-
tains the predicted binding motif for HLA-DRB1*0301
and HLA-DRB3*0202.
Immune responses to pathogens may elicit cross-
reactivity to self-antigens that share structural or sequence
homology.
25,26
This process, known as molecular mimicry,
is considered an important potential mechanism in auto-
immunity. Having found that p63–76 contains an im-
munodominant AQP4 T-cell epitope, we addressed
whether this sequence might share homology with other
proteins. We identified 90% homology between AQP4
66–75 and the 10-amino acid sequence 207–216 within
conserved ABC-TP proteins from several strains of the
bacterium C. perfringens (NCBI protein reference sequen-
ces ZP_02952885.1, ZP_02638213.1, ZP_02634520.1,
ZP_02630305.1; 90% positives, 90% identities, 0%
gaps; see Fig 3). T cells from NMO patients proliferated
significantly to ABC-TP p204–217, although less
intensely than to AQP4 p61–80 and AQP4 p63–76. To
directly test for cross-reactivity, T cells initially stimulated
with AQP4 p63–76 or ABC-TP were tested for recall
responses in a reciprocal manner. Importantly, AQP4-
primed T cells proliferated to ABC-TP p204–217 and
vice versa, supporting molecular mimicry between this
bacterial transmembrane protein and AQP4. Confirming
specificity of those recall responses, we did not observe
proliferation to AQP4 p156–170.
AQP4 p61–80-Specific T Cells from NMO
Patients Exhibit Proinflammatory Th17
Polarization
Although indirect, some clinical and histologic data sug-
gest that Th17 cells may participate in NMO pathogene-
sis.
6,27
Thus, we examined proinflammatory cytokine
production in proliferating AQP4-specific T cells. In
comparison to HC, we observed significantly higher fre-
quencies of IL-17
þ
single- and IL-17
þ
IFN-c
þ
double-
positive cells that recognized p61–80 in NMO patients
(Fig 4). An increased frequency of Th17 cells from
NMO patients was observed after stimulation with
TABLE: Human Leukocyte Antigen Haplotypes of NMO Patients and Healthy Controls
Subjects DRB1*1501
a
DQB1*0602
a
DRB1*0301
b
DRB3*0202
b
DPB1*0501
b
NMO patients 3/15, 20% 2/15, 13% 7/15, 47% 11/15, 73% 7/15, 47%
Healthy controls 2/8, 25% 1/8, 12.5% 3/8, 37.5% 3/8, 37.5% 3/8, 37.5%
a
Alleles associated with multiple sclerosis susceptibility.
b
Alleles associated with NMO susceptibility.
NMO ¼neuromyelitis optica.
Varrin-Doyer et al: AQP4-Specific T Cells in NMO
July 2012 57

rhAQP4, but was not significant. No Th17 bias was
detected in response to AQP4 p156–170, suggesting that
the Th17 polarization may be epitope specific. In con-
trast, IFN-cproduction by AQP4-specific T cells
appeared unchanged between the 2 groups. Thus, the
Th17/Th1 ratio was elevated in NMO patients in
response to the immunodominant determinant AQP4
p61–80, but not to the other antigens tested. Interest-
ingly, we did not detect a difference in the frequency of
peripheral blood regulatory T cells (Treg) from NMO
patients and HC. By contrast, the examination of AQP4-
specific T cells revealed a significantly reduced frequency
of Treg in NMO patients in response to rhAQP4, but
not to p61–80.
Monocytes from NMO Patients Exhibit
Proinflammatory Polarization
APC, including monocytes and other myeloid cells,
express costimulatory molecules and secrete specific cyto-
kines that participate in activation and promote lineage
commitment of antigen-specific T cells. In this regard,
IL-6 is critical for Th17 differentiation.
28
Previous stud-
ies have indicated that serum IL-6 levels are elevated in
NMO patients.
7
As we observed that AQP4 p61–80-spe-
cific T cells from NMO patients exhibited Th17 polar-
ization, we questioned whether there were alterations in
expression of costimulatory molecules or increased pro-
duction of IL-6 by myeloid APC. In comparison to HC,
there was no evident change in frequency of peripheral
FIGURE 2: Human leukocyte antigen (HLA)-DR serves as a restriction element for aquaporin 4 (AQP4)-specific T cells. (A, B)
5,6-carboxylfluorescein diacetate succinimidyl ester (CFSE)-labeled peripheral blood mononuclear cells (PBMC) from neuromye-
litis optica (NMO) patients were cultured for 10 days with antigens (Ag) alone or in combination with antibodies against HLA-
DR, HLA-DQ, or HLA-DP or isotype control antibodies. T-cell proliferation was evaluated by flow-activated cell sorting analysis
of CFSE dilution. Inhibitory effects of blocking antibodies were examined on proliferating CD4
1
T cells (n 57 NMO in A and n
54 NMO in B). T-cell proliferation is expressed as cell division index (CDI). (C) PBMC from healthy controls (HC) were similarly
examined after stimulation with AQP4 p61–80 (n 52). Error bars represent standard error of the mean. *p<0.05, Mann–Whit-
ney Utest.
ANNALS of Neurology
58 Volume 72, No. 1

FIGURE 3: Cross-reactivity between aquaporin 4 (AQP4) p63–76 and Clostridium perfringens adenosine triphosphate-binding
cassette (ABC) transporter permease (TP) p204–217. (A) The T-cell epitope within AQP4 p61–80 was mapped by testing recall
proliferation of AQP4 p61–80-reactive T cells from neuromyelitis optica (NMO) patients to truncated AQP4 peptides (10lg/ml)
in the presence of irradiated autologous antigen-presenting cells (APC). (B) AQP4 p63–76 appeared to contain p61–80 core de-
terminant. Proliferation was measured by [
3
H]thymidine incorporation after 3 days. Data are representative of 3 independent
experiments. (C) Sequence homology between AQP4 p63–76 and C. perfringens ABC-TP p204–217 was identified using the pro-
tein–protein Basic Local Alignment Search Tool from National Center for Biotechnology Information. Top bracket represents the
predicted core binding motif for human leukocyte antigen (HLA)-DRB1*0301 and HLA-DRB3*0202 within AQP4 p63–76 (netMH-
CII-1.1 and netMHCII-2.2 programs). (D) 5,6-Carboxylfluorescein diacetate succinimidyl ester-labeled peripheral blood mononu-
clear cells (PBMC) from 3 NMO patients were stimulated with antigens (10lg/ml) and cultured for 10 days before evaluating
proliferation by flow-activated cell sorting. (E) PBMC from 4 NMO patients were initially stimulated for 10 days with AQP4 p63–
76 or ABC-TP p204–217 at 10lg/ml. Recall responses to peptides in the presence of irradiated autologous APC were evaluated
by [
3
H]thymidine incorporation after 3 days. Paired ttests were performed to compare counts per minute (cpm) values of each
antigen to cpm values of no-antigen controls, *p<0.05, **p<0.01. In A, B, and E, data are presented as means of duplicate or
triplicate wells; error bars throughout indicate standard error of the mean. CDI 5cell division index.
Varrin-Doyer et al: AQP4-Specific T Cells in NMO
July 2012 59

FIGURE 4: Aquaporin 4 (AQP4) p61–80-specific T cells exhibit a proinflammatory bias. Peripheral blood mononuclear cells
(PBMC) were stained with 5,6-Carboxylfluorescein diacetate succinimidyl ester (CFSE) and cultured for 10 days with AQP4 pep-
tides (10lg/ml) or recombinant human (rh) AQP4 (5lg/ml). (A) CD4
1
CFSE
low
proliferating T cells were analyzed for interleukin
(IL)-17 and interferon (IFN)-cproduction by intracellular staining after stimulation with phorbol 12-myristate 13-acetate/Ionomy-
cin for 5 hours. (B) Frequencies of IL17
1
IFN-c
2
, IL17
1
IFN-c
1
, and IL17
2
IFN-c
1
were examined among proliferating p61–80-spe-
cific CD4
1
T cells (n 58 NMO and n 55 healthy controls [HC]), p156–170-specific CD4
1
T cells (n 56 NMO and n 53 HC),
and rhAQP4-specific CD4
1
T cells (n 56 NMO and n 55 HC). Frequencies of IL-17 and IFN-csingle positive T cells were used
to calculate Th17/Th1 ratio. (C) PBMC were examined by fluorescence-activated cell sorting (FACS) for expression of regulatory
T cells (Treg)markers including CD4, CD127, and CD25. (D) CFSE-labeled PBMC were cultured for 10 days with AQP4 p61–80
(10lg/ml) or rhAQP4 (5lg/ml). Proliferating CD4
1
T cells (cell division index >2) were examined by FACS for expression of
CD25
high
, defined as the top half of CD25
1
cells, and Foxp3 (n 58 NMO p61–80, n 56 HC p61–80, n 57 NMO rhAQP4, and
n55 HC rhAQP4). Box and whisker plots include the median, distribution, and range. **p<0.01 Mann–Whitney Utest.
ANNALS of Neurology
60 Volume 72, No. 1

blood monocytes (Supplementary Fig). However, we
observed increased expression of CD40 and CD80 (Fig
5A), costimulatory molecules that can be associated with
proinflammatory T-cell polarization.
29,30
The frequency
of IL-6–producing monocytes was similar in NMO
patients and HC. Nevertheless, there were both relative
and absolute increases of intracellular IL-6 production af-
ter LPS stimulation in monocytes from NMO patients
(see Fig 5B, C). No such differences were observed in
expression of IL-1band IL-10. These results indicate
that in addition to the known involvement of adaptive
immunity, phenotypic changes of cells within the innate
immune system may also contribute to NMO
pathogenesis.
Discussion
In this report, we have demonstrated for the first time
that peripheral blood T cells from NMO patients and
HC proliferate in response to intact AQP4 and AQP4
peptides. However, the frequency and magnitude of T-
cell responses to AQP4 determinants was greater in
NMO patients. Expansion of those autoreactive T cells
provides further evidence that AQP4 is the autoantigen
in NMO. It is notable that 3 of the AQP4 T-cell deter-
minants, p61–80, p131–150, and p211–230 are respec-
tively located in extracellular A, C, and E loops, AQP4
domains targeted by NMO-IgG.
14
Whereas antibodies
that target membrane proteins frequently bind conforma-
tional determinants exposed on their extracellular
FIGURE 5: CD14
1
monocytes from neuromyelitis optica (NMO) patients exhibit increased expression of certain costimulatory
molecules and production of interleukin (IL)-6. (A) Peripheral blood mononuclear cells (PBMC) were rested for 4 hours at 378C.
Expression of costimulatory (CD80, CD86, and CD40) and major histocompatibility complex class II molecules was analyzed by
flow-activated cell sorting gating on the CD14
1
population (n 58 NMO and n 58 healthy controls [HC]). (B, C) PBMC were
stimulated with LPS lipopolysaccharide (LPS; 1lg/ml) for 4 hours. Expression of IL-6 in CD14
1
monocytes was analyzed by in-
tracellular cytokine staining, before and after LPS stimulation. In C, horizontal lines indicate mean values; in A and B, error
bars represent standard error of the mean. *p<0.05, **p<0.01 Mann–Whitney Utest. MFI 5mean fluorescent intensity.
Varrin-Doyer et al: AQP4-Specific T Cells in NMO
July 2012 61

surfaces,
31
CD4
þ
T cells, which are restricted by HLA-D
molecules, recognize linear processed peptides that can
originate from extracellular, transmembrane, or intracel-
lular domains.
21
Thus, T and B cells may recognize dis-
tinct epitopes of the same autoantigen.
32,33
Nevertheless,
with regard to the B–T collaboration required for IgG
production, it is intriguing that p61–80, a naturally
processed immunodominant AQP4 T-cell determinant,
also represents a target for pathogenic AQP4-specific
antibodies.
14
The frequency of AQP4 p61–80-specific Th17 cells
was significantly elevated in NMO patients, a finding
that suggests that these autoantigen-specific T cells are a
source of IL-17 that drives immunopathogenesis in
NMO. In this regard, it was recently demonstrated that
Th17 cells more efficiently drive naive B cells to secrete
Ig than Th1 cells.
34
Although our study relates to the pe-
ripheral immune response in NMO, our finding that
AQP4-specific T cells exhibit a Th17 bias may also be
relevant to development of CNS inflammation in NMO.
Although neutrophils and eosinophils comprise the pre-
dominant cell types within NMO lesions, T cells are also
detected.
6,35
As it is recognized that NMO IgG alone
does not induce CNS inflammation and Th17 cells can
promote tissue accumulation of neutrophils, AQP4-spe-
cific T cells may be sentinel adaptive immune cells
directing CNS inflammation in NMO. Via IL-17 secre-
tion, Th17 cells may compromise integrity of the
BBB,
36,37
promote endothelial activation, and stimulate
transendothelial migration of neutrophils.
38
Thus,
AQP4-specific Th17 cells may participate in multiple
steps of NMO pathogenesis.
Many genetic and environmental factors may con-
tribute to the development of autoimmunity. There is
increasing evidence that commensal and pathogenic gut
microbiota alter susceptibility to multiple sclerosis (MS),
rheumatoid arthritis, type I diabetes, and systemic lupus
erythematosus.
39,40
In this report, we observed a striking
sequence homology between the AQP4 T-cell epitope
p63–76, which contains predicted binding motifs for 2
NMO-associated HLA-DR molecules,
41
and p204–217
of a C. perfringens ABC-TP. C. perfringens is a ubiquitous
gram-positive spore-forming bacterium found in human
commensal gut flora and also includes specific strains fre-
quently associated with enterotoxin-mediated food poi-
soning.
42
We observed T-cell reactivity to ABC-TP
p204–217 in peripheral blood of NMO patients, as well
as cross-reactivity between it and AQP4 p63–76. Inter-
estingly, we detected 60 to 70% homology between p63–
76 and the expressed or predicted ABC-TP in other Clos-
tridium species, including the commensal bacteria C.
scindens and C. hylemonae as well as the pathogenic strain
C. sporogenes. Thus, molecular mimicry could account
for T-cell reactivity to AQP4. Besides molecular mimicry,
microbes can also exploit innate mechanisms that stimu-
late proinflammatory or anti-inflammatory immune
responses. Recently, it was observed that commensal Clos-
tridium-related species alter the balance between Th17
and Treg in mice, and can influence development of
autoimmunity.
43,44
We hypothesize that a Clostridium
species may have dual functions in NMO pathogenesis,
(1) exposing a determinant that cross-reacts with self-
antigen; and (2) serving as its own proinflammatory
adjuvant, promoting Th17 polarization. This demonstra-
tion of molecular mimicry may stimulate further investi-
gation of the potential role of Clostridium species in
NMO pathogenesis.
Based upon the presumption that the AQP4-spe-
cific antibodies of NMO IgG are pathogenic,
approaches that reduce humoral immunity, including
plasmapheresis, intravenous IgG, and CD20 B-cell
depletion, are commonly used in NMO therapy.
Although favorable responses to those treatments have
been reported, they are often incomplete.
45
Recognition
that AQP4-specific antibodies are T cell-dependent, and
alone are not pathogenic in the absence of CNS inflam-
mation, suggests that therapies directed against the cel-
lular arm of NMO pathogenesis could be beneficial.
Interestingly, IFN-b, an approved MS therapy that
alters cellular immune responses and may influence
proinflammatory Th17 activity,
46
has provoked NMO
exacerbations.
47–49
Our observation that T cells specific
for the immunodominant AQP4 epitope exhibit Th17
polarization support testing of agents that target the IL-
17 axis in NMO.
Collectively, our data provide a foundation to
address the potential role of AQP4-specific T cells in
driving adaptive humoral and cellular immune responses
in NMO pathogenesis. Our observations provide a possi-
ble connection between gastrointestinal microbiota, Th17
polarization, and molecular mimicry in the development
of CNS autoimmunity.
Acknowledgment
M.V.-D., C.M.S., P.A.N., S.S.Z., and R.M.S. were sup-
ported by a research grant from the Guthy Jackson Char-
itable Foundation. M.V.-D. and U.S.-T. are postdoctoral
fellows of the National Multiple Sclerosis Society
(NMSS). U.S.-T. is a fellow of the Deutsche For-
schungsgemeinschaft (SCHU 2587/1-1). S.S.Z. also
received grant support from the NIH (RO1 AI073737,
RO1 AI059709, and RO1 NS063008), the NMSS
(RG4124), and the Maisin Foundation. R.M.S. was
ANNALS of Neurology
62 Volume 72, No. 1

supported by NIH 2P50 GM073210, GM24485, and
U54GM094625.
Potential Conflicts of Interest
C.M.S.: grants/grants pending, NIH, NMSS; travel, Teva
Neuroscience. P.A.N.: consultancy, Teva Neurosciences;
travel, Guthy Jackson Charitable Foundation. R.M.S.:
grants/grants pending, NIGMS/NIH, Guthy-Jackson
Foundation. S.S.Z.: grants/grants pending, NIH, NMSS;
travel, Teva Neuroscience.
References
1. Graber DJ, Levy M, Kerr D, Wade WF. Neuromyelitis optica
pathogenesis and aquaporin 4. J Neuroinflammation 2008;5:22.
2. Lennon VA, Kryzer TJ, Pittock SJ, et al. IgG marker of optic-spinal
multiple sclerosis binds to the aquaporin-4 water channel. J Exp
Med 2005;202:473–477.
3. Nurieva RI, Chung Y. Understanding the development and func-
tion of T follicular helper cells. Cell Mol Immunol 2010;7:190–197.
4. Bennett JL, Lam C, Kalluri SR, et al. Intrathecal pathogenic anti-
aquaporin-4 antibodies in early neuromyelitis optica. Ann Neurol
2009;66:617–629.
5. Bradl M, Misu T, Takahashi T, et al. Neuromyelitis optica: pathogenic-
ity of patient immunoglobulin in vivo. Ann Neurol 2009;66:630–643.
6. Lucchinetti CF, Mandler RN, McGavern D, et al. A role for hu-
moral mechanisms in the pathogenesis of Devic’s neuromyelitis
optica. Brain 2002;125(pt 7):1450–1461.
7. Uzawa A, Mori M, Arai K, et al. Cytokine and chemokine profiles
in neuromyelitis optica: significance of interleukin-6. Mult Scler
2010;16:1443–1452.
8. Wingerchuk DM, Lennon VA, Pittock SJ, et al. Revised diagnostic
criteria for neuromyelitis optica. Neurology 2006;66:1485–1489.
9. Jacob A, Weinshenker BG, Violich I, et al. Treatment of neuro-
myelitis optica with rituximab: retrospective analysis of 25
patients. Arch Neurol 2008;65:1443–1448.
10. Ho JD, Yeh R, Sandstrom A, et al. Crystal structure of human
aquaporin 4 at 1.8 A and its mechanism of conductance. Proc
Natl Acad Sci U S A 2009;106:7437–7442.
11. Nielsen M, Lundegaard C, Lund O. Prediction of MHC class II
binding affinity using SMM-align, a novel stabilization matrix align-
ment method. BMC Bioinformatics 2007;8:238.
12. Nielsen M, Lund O. NN-align. An artificial neural network-based
alignment algorithm for MHC class II peptide binding prediction.
BMC Bioinformatics 2009;10:296.
13. Zamvil SS, Mitchell DJ, Moore AC, et al. T-cell epitope of the
autoantigen myelin basic protein that induces encephalomyelitis.
Nature 1986;324:258–260.
14. Owens G, Ritchie A, Lam C, Bennett J. Mutagenesis of aquaporin-
4 extracellular domains defines binding patterns of neuromyelitis
optica intrathecal IgG. Mult Scler 2011;17(10 suppl):S291–S292.
15. Mannering SI, Morris JS, Jensen KP, et al. A sensitive method for
detecting proliferation of rare autoantigen-specific human T cells.
J Immunol Methods 2003;283:173–183.
16. Nelson PA, Khodadoust M, Prodhomme T, et al. Immunodomi-
nant T cell determinants of aquaporin-4, the autoantigen associ-
ated with neuromyelitis optica. PLoS One 2010;5:e15050.
17. Kalluri SR, Rothhammer V, Staszewski O, et al. Functional charac-
terization of aquaporin-4 specific T cells: towards a model for neu-
romyelitis optica. PLoS One 2011;6:e16083.
18. Slavin AJ, Soos JM, Stuve O, et al. Requirement for endocytic
antigen processing and influence of invariant chain and H-2M defi-
ciencies in CNS autoimmunity. J Clin Invest 2001;108:1133–1139.
19. Soos JM, Morrow J, Ashley TA, et al. Astrocytes express elements
of the class II endocytic pathway and process central nervous sys-
tem autoantigen for presentation to encephalitogenic T cells. J
Immunol 1998;161:5959–5966.
20. Zamvil S, Nelson P, Trotter J, et al. T-cell clones specific for mye-
lin basic protein induce chronic relapsing paralysis and demyelin-
ation. Nature 1985;317:355–358.
21. Vyas JM, Van der Veen AG, Ploegh HL. The known unknowns of anti-
gen processing and presentation. Nat Rev Immunol 2008;8:607–618.
22. Matsushita T, Matsuoka T, Isobe N, et al. Association of the HLA-
DPB1*0501 allele with anti-aquaporin-4 antibody positivity in Jap-
anese patients with idiopathic central nervous system demyelinat-
ing disorders. Tissue Antigens 2009;73:171–176.
23. Brum DG, Barreira AA, dos Santos AC, et al. HLA-DRB association
in neuromyelitis optica is different from that observed in multiple
sclerosis. Mult Scler 2010;16:21–29.
24. Deschamps R, Paturel L, Jeannin S, et al. Different HLA class II
(DRB1 and DQB1) alleles determine either susceptibility or resist-
ance to NMO and multiple sclerosis among the French Afro-Ca-
ribbean population. Mult Scler 2011;17:24–31.
25. Fujinami RS, Oldstone MB. Amino acid homology between the
encephalitogenic site of myelin basic protein and virus: mecha-
nism for autoimmunity. Science 1985;230:1043–1045.
26. Wucherpfennig KW, Strominger JL. Molecular mimicry in T cell-
mediated autoimmunity: viral peptides activate human T cell
clones specific for myelin basic protein. Cell 1995;80:695–705.
27. Warabi Y, Yagi K, Hayashi H, Matsumoto Y. Characterization of
the T cell receptor repertoire in the Japanese neuromyelitis
optica: T cell activity is up-regulated compared to multiple sclero-
sis. J Neurol Sci 2006;249:145–152.
28. Acosta-Rodriguez EV, Napolitani G, Lanzavecchia A, Sallusto F.
Interleukins 1beta and 6 but not transforming growth factor-beta
are essential for the differentiation of interleukin 17-producing
human T helper cells. Nat Immunol 2007;8:942–949.
29. Katzman SD, Gallo E, Hoyer KK, Abbas AK. Differential require-
ments for Th1 and Th17 responses to a systemic self-antigen.
J Immunol 2011;186:4668–4673.
30. Kuchroo VK, Das MP, Brown JA, et al. B7-1 and B7-2 costimula-
tory molecules activate differentially the Th1/Th2 developmental
pathways: application to autoimmune disease therapy. Cell 1995;
80:707–718.
31. Benjamin DC, Berzofsky JA, East IJ, et al. The antigenic structure
of proteins: a reappraisal. Annu Rev Immunol 1984;2:67–101.
32. Weber MS, Prod’homme T, Patarroyo JC, et al. B-cell activation
influences T-cell polarization and outcome of anti-CD20 B-cell
depletion in central nervous system autoimmunity. Ann Neurol
2010;68:369–383.
33. Kaushansky N, Zhong MC, Kerlero de Rosbo N, et al. Epitope
specificity of autoreactive T and B cells associated with experi-
mental autoimmune encephalomyelitis and optic neuritis induced
by oligodendrocyte-specific protein in SJL/J mice. J Immunol
2006;177:7364–7376.
34. Morita R, Schmitt N, Bentebibel SE, et al. Human blood
CXCR5(þ)CD4(þ) T cells are counterparts of T follicular cells and
contain specific subsets that differentially support antibody secre-
tion. Immunity 2011;34:108–121.
35. Popescu BF, Lennon VA, Parisi JE, et al. Neuromyelitis optica
unique area postrema lesions: nausea, vomiting, and pathogenic
implications. Neurology 2011;76:1229–1237.
36. Kebir H, Kreymborg K, Ifergan I, et al. Human TH17 lymphocytes
promote blood-brain barrier disruption and central nervous sys-
tem inflammation. Nat Med 2007;13:1173–1175.
Varrin-Doyer et al: AQP4-Specific T Cells in NMO
July 2012 63

37. Huppert J, Closhen D, Croxford A, et al. Cellular mechanisms of
IL-17-induced blood-brain barrier disruption. FASEB J 2010;24:
1023–1034.
38. Roussel L, Houle F, Chan C, et al. IL-17 promotes p38 MAPK-de-
pendent endothelial activation enhancing neutrophil recruitment
to sites of inflammation. J Immunol 2010;184:4531–4537.
39. Berer K, Mues M, Koutrolos M, et al. Commensal microbiota and
myelin autoantigen cooperate to trigger autoimmune demyelin-
ation. Nature 2011;479:538–541.
40. Ochoa-Reparaz J, Mielcarz DW, Begum-Haque S, Kasper LH. Gut,
bugs, and brain: role of commensal bacteria in the control of cen-
tral nervous system disease. Ann Neurol 2011;69:240–247.
41. Nielsen M, Justesen S, Lund O, et al. NetMHCIIpan-2.0—
improved pan-specific HLA-DR predictions using a novel concur-
rent alignment and weight optimization training procedure. Immu-
nome Res 2010;6:9.
42. Lindstrom M, Heikinheimo A, Lahti P, Korkeala H. Novel insights
into the epidemiology of Clostridium perfringens type A food poi-
soning. Food Microbiol 2011;28:192–198.
43. Ivanov, II, Atarashi K, Manel N, et al. Induction of intestinal Th17
cells by segmented filamentous bacteria. Cell 2009;139:485–498.
44. Atarashi K, Tanoue T, Shima T, et al. Induction of colonic regula-
tory T cells by indigenous Clostridium species. Science 2011;331:
337–341.
45. Lindsey JW, Meulmester KM, Brod SA, et al. Variable results after
rituximab in neuromyelitis optica. J Neurol Sci 2012;317:103–105.
46. Axtell RC, Raman C, Steinman L. Interferon-beta exacerbates
Th17-mediated inflammatory disease. Trends Immunol 2011;32:
272–277.
47. Warabi Y, Matsumoto Y, Hayashi H. Interferon beta-1b exacer-
bates multiple sclerosis with severe optic nerve and spinal cord
demyelination. J Neurol Sci 2007;252:57–61.
48. Shimizu J, Hatanaka Y, Hasegawa M, et al. IFNbeta-1b may
severely exacerbate Japanese optic-spinal MS in neuromyelitis
optica spectrum. Neurology 2010;75:1423–1427.
49. Palace J, Leite MI, Nairne A, Vincent A. Interferon Beta treatment
in neuromyelitis optica: increase in relapses and aquaporin 4 anti-
body titers. Arch Neurol 2010;67:1016–1017.
50. Crane JM, Tajima M, Verkman AS. Live-cell imaging of aquaporin-
4 diffusion and interactions in orthogonal arrays of particles. Neu-
roscience 2010;168:892–902.
ANNALS of Neurology
64 Volume 72, No. 1