Screening, Synthesis, and in Vitro Evaluation of Vinyl Sulfones as Inhibitors of Complement-Dependent Cytotoxicity in Neuromyelitis Optica

Article abstract of DOI:10.1002/cmdc.201500546

Neuromyelitis optica (NMO) is a demyelinating autoimmune disease of the optic nerve and spinal cord triggered by binding of NMO-specific immunoglobulin G (NMO-IgG) auto-antibodies to the water channel aquaporin-4 (AQP4) in astrocytes. To find potential NM

Full text of DOI:10.1002/cmdc.201500546

DOI: 10.1002/cmdc.201500546
Communications
Screening, Synthesis, and In Vitro Evaluation of Vinyl
Sulfones as Inhibitors of Complement-Dependent
Cytotoxicity in Neuromyelitis Optica
Eun Ji Ju+,[a, b] Seul Ki Yeon+,[a, b] Jong-Hyun Park+,[a] So Young Cheon,[c] Ji Won Choi,[a, b]
Taehwan Ha,[a] Bo Ko Jang,[a] Siwon Kim,[a, d] Yong Gu Kang,[a] Hayoung Hwang,[e]
Sung Jin Cho,[e] Eunji Cheong,[b] Yong Sun Bahn,[b] Ae Nim Pae,[a, d] Sung Min Kim,*[c] and
Ki Duk Park*[a, d]
Neuromyelitis optica (NMO) is a demyelinating autoimmune
disease of the optic nerve and spinal cord triggered by binding
of NMO-specific immunoglobulin G (NMO-IgG) auto-antibodies
to the water channel aquaporin-4 (AQP4) in astrocytes. To find
potential NMO therapeutics, a screening system was estab-
lished and used to identify inhibitors of NMO-IgG-mediated
complement-dependent cytotoxicity (CDC). The screening of
approximately 400 compounds yielded potent hit compounds
with inhibitory effects against CDC in U87-MG cells expressing
human AQP4. Derivatives of the hit compounds were synthe-
sized and evaluated for their inhibition of CDC. Of the small
molecules synthesized, (E)-1-(2-((4-methoxyphenyl)sulfonyl)vin-
Neuromyelitis optica (NMO) is an autoimmune disorder of the
optic nerve and spinal cord of the central nervous system
(CNS).[1] The symptoms of NMO include optic neuritis (inflam-
mation of the optic nerve with a sudden decrease of vision)
and acute myelitis (inflammation of the spinal cord). NMO was
previously thought to be a variant of multiple sclerosis (MS)
but in 2004, Lennon et al.[1b] reported that NMO immuno-
globulin G (NMO-IgG) was present in patients with NMO but
not in those with MS. Because NMO-IgG is detected with
a high sensitivity and specificity in clinically defined NMO, this
marker is now used as a major diagnostic criterion.[2] The etiol-
ogy of NMO is elusive but recent studies have suggested that
it may involve the binding of pathogenic NMO-IgG auto-anti-
body to the water channel aquaporin-4 (AQP-4), which is ex-
pressed by almost all CNS astrocytes, but it is particularly en-
riched in the spinal cord gray matter, the posterior optic nerve,
and the floor of the fourth ventricle.[3] Several magnetic reso-
nance imaging (MRI) results of patients with NMO showed that
the associated brain lesions were extensively localized in high
AQP4 expression sites.[4]
yl)-[4-[(3-trifluoromethyl)phenyl]
methoxy]benzene
(5c)
showed the most potent activity in both stably transfected
U87-MG cells and mice-derived astrocytes. The results of this
study suggest that 5c, which targets NMO-IgG-specific CDC,
may be useful as a research tool and a potential candidate for
therapeutic development for the treatment of NMO.
An in vitro assay showed that NMO-IgG was bound to the
extracellular domain of AQP4, and in the presence of active
complement, this binding leads to strong complement activa-
tion and rapid complement-dependent cytotoxicity (CDC).[5]
Moreover, a recent study reported that passive transfer (intra-
thecal injection) of NMO-IgG and human activated comple-
ment into an animal model triggers symptoms similar to
NMO.[6]
[a] E. J. Ju,+ S. K. Yeon,+ Dr. J.-H. Park,+ J. W. Choi, Dr. T. Ha, B. K. Jang, S. Kim,
Y. G. Kang, Dr. A. N. Pae, Dr. K. D. Park
Center for Neuro-Medicine, Korea Institute of Science and Technology
Seoul, 02792 (Republic of Korea)
E-mail: kdpark@kist.re.kr
[b] E. J. Ju,+ S. K. Yeon,+ J. W. Choi, Prof. E. Cheong, Prof. Y. S. Bahn
Department of Biotechnology, Yonsei University
Seoul, 03722 (Republic of Korea)
Considering that strong humoral responses are a central fea-
ture of NMO, common therapies include general immunosup-
pressants and plasma exchange to achieve a sustained deple-
tion of NMO-IgG and complement. However, these therapies
are associated with severe side effects. Several therapeutic
strategies for perturbing complement proteins or interleukin
(IL)-6 receptor, and depleting neutrophils, eosinophils, or
B cells (CD19) are under clinical evaluation for the treatment of
NMO.[7] Other therapeutic approaches have been developed to
block the binding of NMO-IgG to AQP4 and decrease CDC.
Among them, aquaporumab, a nonpathogenic human mono-
clonal antibody, competitively displaces NMO-IgG in the serum
of patients with NMO. This direct blocker greatly decreased
NMO-IgG-dependent cytotoxicity and NMO pathology in both
in vivo and in vitro models of NMO.[7] In an alternative ap-
[c] Dr. S. Y. Cheon, Prof. S. M. Kim
Department of Neurology, College of Medicine, Seoul National University
Seoul, 03080 (Republic of Korea)
E-mail: sueh916@gmail.com
[d] S. Kim, Dr. A. N. Pae, Dr. K. D. Park
Department of Biological Chemistry, University of Science and Technology
Daejeon, 34132 (Republic of Korea)
[e] Dr. H. Hwang, Dr. S. J. Cho
New Drug Development Center, Daegu-Gyeongbuk Medical Innovation
Foundation
Daegu 41061 (Republic of Korea)
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500546. The Supporting Information
contains the complete Experimental Section for the work described, in-
cluding synthetic methods, characterization data and biological assay
protocols.
ChemMedChem 2016, 11, 377 – 381
377
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
proach, small molecules were identified as possible blockers
that decrease the binding of NMO-IgG to AQP4 by competi-
tively binding to the extracellular surface of AQP4.[8] Com-
pounds that specifically block NMO-IgG binding to AQP4 could
provide an approach to inhibiting CDC and further inflamma-
tory cascades. The small molecules identified so far exhibit rel-
atively weak inhibitory activity against CDC and unfavorable
pharmacological properties. Here, we developed a cell-based
screening system and identified potent inhibitors of CDC
caused by NMO-IgG binding to AQP4.
Our cell-based assay used stably transfected U87-MG cells
expressing high levels of the M23 isoform of the human AQP4
channel. Research studies have revealed that the binding of
NMO-IgG from NMO patients occurs more with transfected
cells expressing the M23 isoform than those with M1 isoform.[9]
AQP4 overexpression in the stable cell line was confirmed
using reverse transcription-polymerase chain reaction (RT-PCR)
and Western blot analysis (Figure S1 in the Supporting Infor-
mation).
To establish a screening system for inhibitors of CDC caused
by a human NMO-IgG autoantibody from patients with NMO,
we measured lactate dehydrogenase (LDH) release from cells
with damaged membranes. The binding of NMO-IgG to AQP4
recruited complement (C1q) and formed a membrane attack
complex (MAC), which disrupted the plasma membrane fol-
lowed by LDH release (Figure 1a). To validate the efficacy of
our CDC assay system, NMO-IgG from patients with NMO or
control-IgG from healthy individuals was treated with increas-
ing concentrations of human complement in either non-trans-
fected or AQP4-overexpressing cells. NMO-IgG incubation with
AQP4 overexpressing cells significantly increased LDH release
dose-dependently whereas the control groups did not exhibit
this effect (Figure S2 in the Supporting Information).
Figure 1. Screening system for the identification of inhibitors of neuromyeli-
tis optica-specific immunoglobulin G autoantibodies (NMO-IgG)/comple-
ment-mediated cytotoxicity. A) Schematic of the screening assay used for
identifying potent inhibitors of (a) NMO-IgG binding to aquaporin-4 (AQP4),
(b) complement binding to NMO-IgG, and (c) membrane attack complex
(MAC) formation by measuring lactate dehydrogenase (LDH) release from
AQP4-expressing U87-MG cells incubated with NMO-IgG and 5% comple-
ment. B) Chemical structure of hit compound 5a.
and selected a hit compound (5a) that decreased the LDH re-
lease by >30% (Figure 1b).
NMO-IgG acquired from the sera of patients with NMO was
provided by Seoul National Hospital, Korea. Pathogenic waste
sera from patients with severe relapsing NMO who had under-
gone plasmapheresis was stored at À808C and bio-banked for
future reference. The use of NMO-IgG derived from patient
sera is valuable for developing in vitro, in vivo, and ex vivo dis-
ease models of NMO.[10] This study was approved by the Insti-
tutional Review Board of the Seoul National University Hospital
(IRB number: H-1012-023-317). All patients provided written in-
formed consent. NMO-IgG was purified from sera, using the
method outlined in the Supporting Information, because
serum contains other IgGs, complement, cytokines, and plasma
protein-bound toxins. The severity of disease progression of
each patient differs, as does the concentration of IgG, and
therefore, the purified NMO-IgG was subjected to activity test-
ing prior to the screening experiments. The NMO-IgG binding
affinity and CDC pathogenicity of the NMO-IgGs from different
patients were tested using the same concentrations, and sam-
ples with optimal activity were selected (concentration of each
NMO-IgG that induced 50% CDC).
The analysis of the screening results revealed that the vinyl
sulfone group is a potentially favorable motif for inhibitory ac-
tivity against CDC. Accordingly, we synthesized five vinyl sul-
fone derivatives by introducing H or CF3 instead of F (5b and
5c) or by placing the OMe or PhOMe groups in different posi-
tions on the aryl group (5d and 5e). Vinyl sulfone derivatives
5a–e were prepared in three steps (Scheme 1).[11] First, substi-
tuted benzenethiols (1) were coupled with (diethoxyphosphor-
yl)methyl-4-methylbenzenesulfonate in the presence of a base
to obtain the sulfides (2). In the next step, the sulfides (2) were
oxidized with meta-chloroperoxybenzoic acid (mCPBA) at 08C
to obtain the sulfones (3). The final compounds (5a–e) were
obtained by Horner–Emmons olefination reactions with the de-
sired substituted benzaldehydes (4; for synthetic methods, see
the Supporting Information).
Using a CDC assay, we found that compound 5c exhibited
the most potent inhibitory activity (43.4% decrease in LDH re-
lease at 25 mm). The substitution of the trifluoromethyl group
(CF3) instead of the fluorine group (F) on the ring C significant-
ly increased the inhibitory activity in the CDC assay. When the
electron-withdrawing group (F or CF3) was replaced with hy-
drogen (5b), the inhibitory activity was decreased (Figure 2).
We placed a methoxy group at the 2-position on ring A in-
stead of the 4-position and observed that the 2-methoxy deriv-
To identify potent small-molecule inhibitors of NMO-IgG-de-
pendent CDC, we screened ~400 synthetic small molecules
from our central nervous system (CNS)-focused chemical library
ChemMedChem 2016, 11, 377 – 381
www.chemmedchem.org
378
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Scheme 1. Synthesis of vinyl sulfone derivatives 5a–e. Reagents and conditions: a) (diethoxyphosphoryl)methyl 4-methylbenzenesulfonate, CsCO2, DMF, RT,
overnight, 69–73%; b) meta-chloroperoxybenzoic acid, CH2Cl2, RT, 4 h, 68%; c) [(substituted phenyl)methoxy]benzaldehydes (4), nBuLi, THF, À788C, 2 h, 73–
90%.
Figure 2. Potency of synthetic compounds (5a–e) against neuromyelitis
Figure 3. Analysis of inhibition of neuromyelitis optica-specific immuno-
globulin G autoantibodies (NMO-IgG) binding to AQP4 by compound 5c
using immunofluorescence in aquaporin-4 (AQP4)-expressing U87-MG cells,
which expressed DsRed were treated with compound 5c (50 mm) before in-
cubation with neuromyelitis optica-specific immunoglobulin-G autoantibod-
ies (NMO-IgG, 6 mgmLÀ1). Binding of NMO-IgG or control-IgG to AQP4 was
detected using fluorescein isothiocyanate (FITC)-conjugated antihuman sec-
ondary antibody. Binding affinities were determined using Harmony soft-
ware programmed to automatically calculate nonlinear regression of back-
ground-subtracted green/red fluorescence intensity ratios. Values shown are
the relative binding affinity (%); data are the meanÆSEM (n=5).
optica-specific immunoglobulin G autoantibodies (NMO-IgG) and human
complement-dependent cytotoxicity (CDC) in cell cultures Five vinyl sulfone
derivatives (5a–e) were used at a final concentration of 25 mm before incu-
bation with 100 mgmLÀ1 NMO-IgG and 5% pooled human complement in
aquaporin-4 (AQP4)-expressing U87-MG cells. After 40 min, lactate dehydro-
genase (LDH) release was measured. Arbidol was used as a positive con-
trol.[12] Black bars show LDH release (%) [data are the meanÆSEM n=4] and
grey bars show CDC inhibition (%) compared with dimethyl sulfoxide
(DMSO)-treated group (no inhibition).
ative (5d) had a similar inhibitory activity with the 4-methoxy
derivative (5a). Next, we changed a methylphenyl group at-
tached to the para position on ring B to the ortho position
(5e) but did not observe a clear change in activity.
The highest potency compound (5c) was further evaluated
for its ability to decrease CDC in primary astrocytes highly ex-
pressing AQP4. It has been suggested that NMO-IgG binding
to AQP4 is accompanied by complement-dependent astrocyte
cytotoxicity, which produces NMO lesions in the disease patho-
genesis.[13] CDC in astrocytes was tested using a two-color
staining method in which live and dead cells were stained
blue and green, respectively. Mouse primary astrocytes were
incubated with NMO-IgG in the presence of human comple-
ment with or without 5c. Compound 5c decreased cytotoxici-
ty dose-dependently in the NMO-IgG/complement-treated
AQP-4-expressing astrocytes (Figure 4). We found that 5c de-
creased cell death of primary astrocytes by 46.4% at 10 mm,
which indicated a higher CDC inhibition than was shown in
U87-MG cells (43.4% decrease at 25 mm). This result may be
Next, we evaluated the ability of 5c to block the binding of
NMO-IgG to AQP4 using immunofluorescence. An NMO-IgG
polyclonal antibody against the three-dimensional epitopes on
the extracellular surface of AQP4 was incubated with the
AQP4-expressing U87-MG cells (DsRed positive) in the absence
or presence of 5c for 30 min. Then, the cells were washed with
phosphate-buffered saline (PBS), fixed, and then incubated
with a fluorescein isothiocyanate (FITC)-conjugated antihuman
secondary antibody. The NMO-IgG but not the control-IgG effi-
ciently bound to the AQP4-overexpressing U87-MG cells. The
immunofluorescence micrographs showed a substantial de-
crease in NMO-IgG binding to AQP4 following 5c treatment
(Figure 3). This result suggests that 5c physically interferes
with the binding of polyclonal NMO-IgG to AQP4 (Figure 1a).
ChemMedChem 2016, 11, 377 – 381
www.chemmedchem.org
379
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Figure 4. Inhibitory effect of compound 5c against neuromyelitis optica-specific immunoglobulin G autoantibodies (NMO-IgG)-dependent complement-de-
pendent cytotoxicity (CDC) in primary mouse astrocytes. Primary astrocytes from cortices of wildtype mice were induced with CDC with NMO-IgG in the pres-
ence of compound 5c at the indicated concentration. A) Cell staining (dead cells: green; total cells: blue) and B) relative % cell death (Sytox green staining
index/Hoechst staining index). Compound 5c blocked cell death in a concentration-dependent manner. Black bars show Sytox green index (%) [data are the
meanÆSEM n=5] and grey bars show CDC inhibition (%) compared with dimethyl sulfoxide (DMSO)-treated group (no inhibition).
due to overexpression of the M23 isoform of AQP4 in U87-MG
cells, which probably enhances the binding efficiency of NMO-
IgG to AQP4 compared with endogenous AQP4 in astrocytes.
Compound 5c was identified as a potent inhibitor of CDC
induced by NMO-IgG binding to AQP4. For a compound to be
used as a drug candidate, a number of pharmacokinetic re-
quirements must be met. To assess the metabolic stability of
5c, we determined its degree of degradation using liver micro-
somes from four different animal species (human, dog, rat, and
mouse) and plasma from two different animal species (human
and rat). Compound 5c exhibited favorable liver microsomal
stabilizing effects in the human, rat, and mouse with 43.9, 69,
and 62.5%, respectively, remaining after 30 min incubation
with nicotinamide adenine dinucleotide phosphate (NADPH).
In addition, it exhibited complete resistance to metabolism in
dog liver microsomes (~100%; Table 1). Compound 5c was
also stable in human and rat plasma during 120 min incuba-
tion. The favorable stability of 5c against the human enzymes
suggests that the compound may have a high bioavailability
when administered to humans.
in Table 1, for the subtypes evaluated, the IC50 values of com-
pound 5c were all over 10 mm except for 2C19 (slightly below
10 mm), indicating that 5c is not likely to cause drug-drug in-
teraction-related side effects.
In conclusion, we report on a novel compound 5c, which
has inhibitory effects against NMO-IgG/CDC and may be a po-
tential therapeutic candidate for NMO treatment. Further eval-
uations in ex vivo and in vivo animal models and drug optimi-
zation are necessary to evaluate its desired drug properties.
Acknowledgements
Funding for this study was provided by supporting grants from
the Korea Health Technology R&D Project, the Korean Ministry of
Health and Welfare (HI12C1022), and the Korea Institute of Sci-
ence and Technology (KIST, 2E25240).
Keywords: aquaporin-4
cytotoxicity neuromyelitis
immunoglobulin G autoantibodies · vinyl sulfones
·
complement-dependent
·
optica NMO-specific
·
Metabolism-related drug–drug interactions in vivo might
cause adverse reactions or severe side effects, and this was
evaluated using a cytochrome 450 (CYP) inhibition test, partic-
ularly in subtypes 1A2, 2C9, 2C19, 2D6, and 3A4.[14] As shown
[1] a) C. F. Lucchinetti, R. N. Mandler, D. McGavern, W. Bruck, G. Gleich, R. M.
Ransohoff, C. Trebst, B. Weinshenker, D. Wingerchuk, J. E. Parisi, H. Lass-
mann, Brain 2002, 125, 1450–1461; b) V. A. Lennon, D. M. Wingerchuk,
T. J. Kryzer, S. J. Pittock, C. F. Lucchinetti, K. Fujihara, I. Nakashima, G. G.
Weinshenker, Lancet 2004, 364, 2106–2112; c) G. J. Hengstman, P. Wes-
seling, C. W. Frenken, P. J. Jongen, Mult. Scler. 2007, 13, 679–682.
[2] D. M. Wingerchuk, V. A. Lennon, S. J. Pittock, C. F. Lucchinetti, B. G.
Weinshenker, Neurology 2006, 66, 1485–1489.
Table 1. In vitro profile of compound 5c determined using metabolic sta-
bility and cytochrome P450 (CYP) inhibition assays.
[3] a) V. A. Lennon, T. J. Kryzer, S. J. Pittock, A. S. Verkman, S. R. Hinson, J.
Exp. Med. 2005, 202, 473–477; b) S. R. Hinson, S. J. Pittock, C. F. Lucchi-
netti, S. F. Roemer, J. P. Fryer, T. J. Kryzer, V. A. Lennon, Neurology 2007,
69, 2221–2231; c) T. Misu, K. Fujihara, A. Kakita, H. Konno, M. Nakamura,
S. Watanabe, T. Takahashi, I. Nakashima, H. Takahashi, Y. Itoyama, Brain
2007, 130, 1224–1234.
Species[a]
Stability[b]
Microsomes
CYP Inhibition
Subtype % Inhibition
Plasma
at 10 mm
Human
Dog
Rat
43.9
~100
69.0
~100
–
86.7
–
1A2
2C9
2C19
2D6
3A4
<5
<5
43.4
<5
<5
[4] D. M. Wingerchuk, V. A. Lennon, C. F. Lucchinetti, S. J. Pittock, B. G.
Weinshenker, Lancet Neurol. 2007, 6, 805–815.
Mouse
62.5
[5] a) P. Waters, S. Jarius, E. Littleton, M. I. Leite, S. Jacob, B. Gray, R. Ger-
aldes, T. Vale, A. Jacob, J. Palace, S. Maxwell, D. Beeson, A. Vincent, Arch.
Neurol. 2008, 65, 913–919; b) S. Jarius, F. Aboul-Enein, P. Waters, B.
Kuenz, A. Hauser, T. Berger, W. Lang, M. Reindl, A. Vncent, W. Kristofer-
itsch, Brain 2008, 131, 3072–3080; c) S. Jarius, F. Paul, D. Franciotta, P.
[a] All microsomes were purchased from BD Gentest (Woburn, MA, USA);
[b] % Remaining was determined after 30 min (microsomes) or 120 min
(plasma).
ChemMedChem 2016, 11, 377 – 381
www.chemmedchem.org
380
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Waters, F. Zipp, R. Hohlfeld, A. Vincent, B. Wildemann, Nat. Clin. Pract.
Neurol. 2008, 4, 202–214; d) T. Vincent, P. Saikali, R. Cayrol, A. D. Roth,
A. Bar-Or, A. Prat, J. P. Antel, J. Immunol. 2008, 181, 5730–5737.
[6] S. Saadoun, P. Waters, B. A. Bell, A. Vincent, A. S. Verkman, M. C. Papado-
poulos, Brain 2010, 133, 349–361.
[7] a) M. C. Papadopoulos, J. L. Bennett, A. S. Verkman, Nat. Rev. Neurol.
2014, 10, 493; b) L. Tradtrantip, N. Asavapanumas, P. W. Phuan, A. S.
Verkman, PLoS One 2014, 9, e106824.
[11] S. Y. Woo, J. H. Kim, M. K. Moon, S.-H. Han, S. K. Yeon, J. W. Choi, B. K.
Jang, H. J. Song, Y. G. Kang, J. W. Kim, J. Lee, D. J. Kim, O. Hwang, K. D.
Park, J. Med. Chem. 2014, 57, 1473–1487.
[12] L. Tradtrantip, H. Zhang, M. O. Anderson, S. Saadoun, P. W. Phuan, M. C.
Papadopoulos, J. L. Bennett, A. S. Verkman, FASEB J. 2012, 26, 2197–
2208.
[13] A. S. Verkman, P. W. Phuan, N. Asavapanumas, L. Tradtrantip, Brain
Pathol. 2013, 23, 684–695.
[8] P. W. Phuan, M. O. Anderson, L. Tradtrantip, H. Zhang, J. Tan, C. Lam,
J. L. Bennett, A. S. Verkman, J. Biol. Chem. 2012, 287, 36837–36844.
[9] a) J. M. Crane, C. Lam, A. Rossi, T. Gupta, J. L. Bennett, A. S. Verkman, J.
Biol. Chem. 2011, 286, 16516–16524; b) A. Rossi, J. M. Crane, A. S. Verk-
man, Glia 2011, 59, 1056–1063.
[14] L. C. Wienkers, T. G. Heath, Nat. Rev. Drug Discovery 2005, 4, 825–833.
Received: November 22, 2015
[10] H. Zhang, J. L. Bennett, A. S. Verkman, Ann. Neurol. 2011, 70, 943–954.
Published online on January 25, 2016
ChemMedChem 2016, 11, 377 – 381
www.chemmedchem.org
381
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim