Modulation of angiogenesis by dithiolethione-modified NSAIDs and valproic acid

Article abstract of DOI:10.1038/sj.bjp.0707198

Background and purpose: Angiogenesis involves multiple signaling pathways that must be considered when developing agents to modulate pathological angiogenesis. Because both cyclooxygenase inhibitors and dithioles have demonstrated anti-angiogenic properti

Full text of DOI:10.1038/sj.bjp.0707198

British Journal of Pharmacology (2007) 151, 142–151
2007 Nature Publishing Group All rights reserved 0007–1188/07 30.00
&
$
www.brjpharmacol.org
RESEARCH PAPER
Modulation of angiogenesis by dithiolethione-
modified NSAIDs and valproic acid
JS Isenberg1, Y Jia1, L Field1, LA Ridnour2, A Sparatore3, P Del Soldato4, AL Sowers2, GC Yeh5,
TW Moody6, DA Wink2, R Ramchandran1 and DD Roberts1
2
1Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; Radiation Biology Branch,
National Cancer Institute, National Institutes of Health, Bethesda, MD, USA; 3Istituto di Chimica Farmaceutica, University of Milan,
Milan, Italy; 4CTG Pharma, Milan, Italy; 5Laboratory of Metabolism, NCI-Frederick, Frederick, MD, USA and 6Center for Cancer
Research, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA
Background and purpose: Angiogenesis involves multiple signaling pathways that must be considered when developing
agents to modulate pathological angiogenesis. Because both cyclooxygenase inhibitors and dithioles have demonstrated anti-
angiogenic properties, we investigated the activities of a new class of anti-inflammatory drugs containing dithiolethione
moieties (S-NSAIDs) and S-valproate.
Experimental approach: Anti-angiogenic activities of S-NSAIDS, S-valproate, and the respective parent compounds were
assessed using umbilical vein endothelial cells, muscle and tumor tissue explant angiogenesis assays, and developmental
angiogenesis in Fli:EGFP transgenic zebrafish embryos.
Key results: Dithiolethione derivatives of diclofenac, valproate, and sulindac inhibited endothelial cell proliferation and
induced Ser78 phosphorylation of hsp27, a known molecular target of anti-angiogenic signaling. The parent drugs lacked this
activity, but dithiolethiones were active at comparable concentrations. Although dithiolethiones can potentially release
hydrogen sulphide, NaSH did not reproduce some activities of the S-NSAIDs, indicating that the dithioles regulate
angiogenesis through mechanisms other than release of H2S. In contrast to the parent drugs, S-NSAIDs, S-valproate, NaSH,
and dithiolethiones were potent inhibitors of angiogenic responses in muscle and HT29 tumor explants assessed by
3-dimensional collagen matrix assays. Dithiolethiones and valproic acid were also potent inhibitors of developmental
angiogenesis in zebrafish embryos, but the S-NSAIDs, remarkably, lacked this activity.
Conclusions and implication: S-NSAIDs and S-valproate have potent anti-angiogenic activities mediated by their dithiole
moieties. The novel properties of S-NSAIDs and S-valproate to inhibit pathological versus developmental angiogenesis suggest
that these agents may have a role in cancer treatment.
British Journal of Pharmacology (2007) 151, 142–151. doi:10.1038/sj.bjp.0707198; published online 12 March 2007
Keywords: angiogenesis inhibitors; hsp27 phosphorylation; dithioles; nonsteroidal anti-inflammatory drugs
Abbreviations: ADT, 5-[p-methoxyphenyl]-3H-1,2-dithiole-3-thione; D3T, 1,2-dithiole 3-thione; hpf, hours postfertilization;
EGFP, enhanced green fluorescent protein; FCS, fetal calf serum; Hsp27, heat-shock protein 27; HUVEC,
human umbilical vein endothelial cells; NSAID, nonsteroidal anti-inflammatory drug; Fli, Friend leukemia
inhibitor
Introduction
Angiogenesis is the process of new vessel formation from
an established vascular plexus (Folkman, 2006). Under
nonpathological conditions, it is a tightly regulated process
representing a balance between stimulating and inhibiting
factors (Ferrara and Kerbel, 2005). A number of diseases
involve dysregulation of angiogenesis (Carmeliet, 2005).
Lack of an angiogenic response in the face of tissue ischemia
underlies myocardial infarction and peripheral vascular
disease. Inappropriate angiogenesis is required for tumor
growth and metastasis (Cao, 2005). Thus, selective control of
angiogenic responses would have clinical utility in several
major diseases (Verheul and Pinedo, 2005).
Dithiolethiones are sulfur-based compounds widely dis-
tributed in the human diet with a range of known activities,
including inhibition of lipid peroxidation, protective proper-
ties against radiation injury and redox cytotoxicity, and
anticancer effects (Kensler et al., 2000). Some dithioles have
Correspondence: Dr DD Roberts, Laboratory of Pathology, National Cancer
Institute, National Institutes of Health, Building 10, Room 2A33, 10 Center
Drive, Bethesda, MD 20892-1500, USA.
E-mail: droberts@helix.nih.gov
Received 30 August 2006; revised 29 November 2006; accepted 24 January
2007; published online 12 March 2007

Dithiole-modified NSAIDs inhibit angiogenesis
JS Isenberg et al
143
known effects on endothelial cells or antiangiogenic activ-
ities (Ben-Mahdi et al., 2000; Ruggeri et al., 2002). These
compounds also protect against tetrachloride and acetami-
nophen hepatotoxicity, possibly owing to their antioxidant
properties modulating cellular glutathione, which provides
cytoprotection against oxidative stress (Kensler et al., 1992;
Kwak et al., 2001).
water. Handling and care of animals was in compliance with
the guidelines established by the Animal Care and Use
Committees of the National Cancer Institute and the
National Institutes of Health. Zebrafish were grown and
maintained at 28.51C under the National Cancer Institute
guidelines.
A number of molecular pathways are induced by dithio-
lethiones. Dithiolethiones induce redox coactivator Ref-1,
which enhances AP-1-dependent transcription of c-fos and
c-jun (Yao and O’Dwyer, 2003). Dithiolethiones also induce
antioxidant-response element (ARE)-regulated genes via
dissociation of nuclear transcription factor erythroid 2p45
(NF-E2)-related factor 2 (Nrf2) from Kelch-like ECH-asso-
ciated protein 1 (Keap1), which sequesters Nrf2 in the
cytoplasm (Lee and Surh, 2005). The activation of these
pathways is dependent on oxidation of key thiols in the
molecular targets.
Cells
Human umbilical vein endothelial cells (HUVEC; Cambrex,
Walkersville, MD, USA) were maintained in endothelial cell
growth medium (Cambrex) and 2% fetal calf serum (FCS) in
5% CO2 at 371C. Cells were utilized between passages 4 and
8. Purity of cultures was monitored by immunochemical
staining with monoclonal human anti-CD31 antibody
(Sigma, St Louis, MO, USA).
The above observations led us to investigate the effects of
a novel class of dithiolethione-modified anti-inflammatory
drugs (Li et al., 2006) on angiogenic responses in vascular
cells. Unlike the unmodified drugs, we found that dithiol-
ethione-modified S-nonsteroidal anti-inflammatory drug
(NSAIDs) and S-valproate demonstrated significant anti-
angiogenic activities, inhibiting endothelial cell proliferation
and vascular cell outgrowth and invasion of extracellular
matrix under wound healing as well as tumor-driven
conditions. Likewise, simple dithiolethiones potently inhib-
ited these angiogenic responses, suggesting that the anti-
angiogenic properties of the S-NSAIDs and S-valproate reside
in their dithiolethione moiety.
Synthesis of dithiolethione derivatives S-diclofenac, S-sulindac
and S-valproate
The 5-(4-hydroxyphenyl)-3H-1,2-dithiole-3-thione (ADT-
OH) was prepared by demethylation of ADT as described
previously (Bottcher et al., 1951). The compounds
S-diclofenac and S-sulindac were prepared adding to the
respective free acids, ADT-OH, dicyclohexylcarbodiimide
and a catalytic amount of 4-dimethylaminopyridine in
dichloromethane (Figure 1) (Li et al., 2006). After removal
of dicyclohexylurea, the compounds were purified by
column chromatography. S-valproate was prepared by
refluxing 2-propylpentanoyl chloride with ADT-OH and
N,N-diisopropylethylamine in anhydrous tetrahydrofurane.
After evaporation of the solvent, the residue was purified
by chromatography and crystallized with ethyl ether. All
the compounds had chromatographic purity 498%. The
compounds were characterized by melting point (m.p.)
(Buchi apparatus), 1-H-NMR spectra (Varian Mercuri 300
VX spectrometer in CDCl3 or DMSO-d6), high-resolution
mass spectra (APEX II ICR-FTMS Bruker Daltonics mass
Methods
Animals
C57B16 mice and Cr:(NCr)-athymic nu fBR mice (NCI,
Frederick, MD, USA) were housed in
a pathogen-free
environment and allowed ad libitum access to food and
Figure 1 Synthesis and chemical structures of S-NSAIDs and S-valproate. Synthetic pathways and chemical structures of the ADT-OH
derivatives of diclofenac, sulindac and valproate.
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
144
JS Isenberg et al
spectrometer) and by elemental analysis (Carlo Erba EA-
1100CHNS-O instrument).
USA). Briefly, 5 Â 103 cells suspended in 100 ml of culture
medium were added to each well of a 96-well culture plate
(Nunc, Denmark) and incubated for 72 h. Appropriate zero-
time controls were run for all assays and the optical density
readings obtained then subtracted from those obtained at
72 h. Results represent the mean7s.e. of at least three
separate experiments.
(1) 2-[(2,6-dichlorophenyl)amino]benzeneacetic 4-(5-thioxo-
5H-1,2-dithiol-3-yl)phenyl ester (S-diclofenac): m.p. 151–
1531C; 1H NMR (CDCl3): d 7.75–7.60 (m, 2H), 7.45–7.15
(m, 7H), 7.10–6.90 (m, 2H), 6.65–6.55 (m, 2H), 4.10 (s, 2H);
HRMS (ESI) calculated for C23H15Cl2NO2S3Na (M þ Na)þ
:
525.95342; found: 525.95438. Elemental analysis calculated
for C23H15Cl2NO2S3: C 54.76, H 3.00, N 2.78, S 19.07;
found: C 54.64, H 3.27, N 2.64, S 19.18.
Explant cytospin analysis
Type I collagen matrix from explant constructs was digested
with type 2 collagenase in growth medium and collected
by centrifugation. The cell pellet was resuspended in growth
medium at a concentration of 0.5 Â 106 cells mlÀ1. Standard
cytospin cuvettes were loaded with 200 ml of cell suspension
and spun at 800 r.p.m. for 3 min and allowed to air dry. Slides
were then fixed with acetone, stained with anti-CD31 (clone
WM-59; Sigma) and anti-a smooth muscle actin (clone 1A4,
Sigma), developed with the Vectastain Elite ABC kit (Vector
Labs, Burlington, CA, USA) following the manufacturer’s
instructions and the cell counts obtained.
(2) (Z)-5-Fluoro-2-methyl-1-[[4-(methylsulfinyl)phenyl]-
methylene]-1H-indene-3-acetic acid 4-(thioxo-5H-[1,2]
dithiol-3-yl)-phenyl ester. (S-sulindac): m.p. 155–1561C;
1H NMR (DMSO-d6): d 8.00 (d, 2H), 7.65–7.80 (m, 5H),
7.40 (s, 1H), 7.30 (d, 2H), 7.10–7.25 (m, 2H), 6.70 (t, 1H),
4.00 (s, 2H), 2.80 (s, 3H), 2.20 (s, 3H). HRMS (ESI)
calculated for C29H21FO3S4Na (M þ Na)þ : 587.02498;
found: 587.02397.
(3) 2-Propylpentanoic acid 4-(5-thioxo-5H-1,2-dithiol-3-yl)-
phenyl ester (S-valproate): mp 69.5–70.51C; 1H NMR
(DMSO): d 7.95 (d, 2H), 7.80 (s, 1H), 7.25 (d, 2H), 2.70–
2.55 (m, 1H), 1.42–1.70 (m, 4H), 1.25–1.40 (m, 4H),
0.85–1.00 (m, 6H); HRMS (ESI) calculated for C17H21O2S3
(M þ H)þ : 353.06982; found: 353.06956. Elemental ana-
lysis calculated for C17H20O2S3: C 57.92; H 5.72, S 27.29;
found: C 58.05, H 5.69, S 27.34.
Cyclooxygenase assay
HT-29 cells were incubated with the S-NSAID compounds at
1 mM for 30 min, and then 20 mM arachidonic acid was added
for 5 min. The supernatant was then removed and assayed
for prostaglandin E2 (PGE2) using an enzyme-linked
immunoassay.
Explant invasion assay
Biopsies (1 mm3) from the pectoralis major muscle of 8-week-
old wild-type C57B16 mice were harvested and explanted
into type I collagen gel in 96-well tissue culture plates as
described by Isenberg et al. (2005). Following gelation,
explants were incubated in endothelial growth medium
(EGM) þ 2% FCS in the presence or absence of a dose range of
the indicated treatment agents or the comparable concen-
tration of vehicle at 371C and 5% CO2. The selection of
treatment agent concentrations used in explant and other
in vitro cell assays was based upon data obtained from in vivo
experiments (Li et al., 2006). Maximum cell migration
through the matrix was measured following 7 days of
incubation. In other experiments, C3H athymic nude mice
were injected subcutaneously with 106 HT29 adenocarcino-
ma tumor cells to the lateral thigh. Animals were killed when
tumors reached 1 cm. Tumor biopsies of 1 mm3 were
harvested immediately following death and explanted into
type I collagen matrix as described above and incubated in
EGM with 1% FCS. Following 7 days of incubation in the
presence of treatment agents, maximum vascular cell out-
growth was measured. The treatment doses employed in vitro
and in vivo were based on previously published data utilizing
these agents (Li et al., 2006). Vascular cell outgrowth of
collagen matrices was quantified as the distance of farthest
cell invasion from the explant border in each of the four
quadrants. Results represent the mean7s.e. of at least three
separate experiments.
Western blot analysis
HUVEC at 60% confluence were starved overnight in
endothelial basal medium þ 1% FCS and then treated with
the indicated dosages of compounds for 60 min. Cells were
collected into 1 Â sodium dodecyl sulfate sample buffer
containing 1 Â protease inhibitors. Cell lysates were sepa-
rated by electrophoresis and blotted onto polyvinylidene
fluoride membranes. Membranes were blocked with 5%
nonfat dry milk at RT for 1 h, and incubated with anti-Hsp27
Ser78 phosphorylation Ab (Stressgen, Ann Arbor, MI, USA)
overnight at 41C, then with goat-anti-mouse-horseradish
peroxidase (Jackson ImmunoResearch, West Grove, PA, USA)
at RT for 1 h. After three washes with phosphate-buffered
saline, protein was determined by enhanced chemilumines-
cence. Membranes were reprobed with antiactin antibody for
sample loading control.
Embryonic Zebrafish toxicology studies
Zebrafish were maintained as described in The Zebrafish
Book (Westerfield, 2000). Transgenic Tg(fli1:EGFP)y10 em-
bryos were dechorinated at 10 h postfertilization (hpf) with
2 mg mlÀ1 pronase (Sigma, Cat. No. P-8811) stock in 2–3 ml
Blue Water (E3 solution: 5 mM NaCl, 0.17 mM KCl, 0.33 mM
CaCl2, 0.33 mM MgSO4, 10À5% Methylene Blue). Embryos
were then added to 2 ml of Blue Water containing desired
concentration of drug compound or dimethylsulfoxide
(DMSO) control at either 10 hpf or in some experiments at
18 hpf. All drug compounds were dissolved in DMSO to 1 mM
Cell proliferation
Proliferation of HUVEC was measured using a nonradioac-
tive colorimetric assay (CellTiter 96, Promega, Madison, WI,
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
JS Isenberg et al
145
before being diluted in Blue Water. Embryos were incubated
at 281C until 48 hpf and images were captured with a Zeiss
AxioCam HRC and Carl Zeiss SV11 Epi-Fluorescence stereo
microscope (Thornwood, NY, USA). The number of animals
or embryos employed ranged from 30 to 100/drug concen-
tration per experiment.
vascular cell outgrowth. The dithiolethiones ADT and D3T
also significantly inhibited wound-driven vascular cell out-
growth (Figure 2b). In contrast, the corresponding unmodi-
fied valproate, sulindac and diclofenac had either no effect
on wound-driven angiogenesis or tended to stimulate the
response (data not shown), indicating that the dithiole
moieties of the S-NSAIDs and S-valproate mediated their
antiangiogenic activities. Vehicle alone did not inhibit the
muscle explant vascular cell response (data not shown).
Statistics
Results represent the mean7s.e. of at least three separate
experiments. Significance was determined with the Student’s
t-test, or one- or two-way analysis of variance with Tukey’s
post hoc analysis and using a Po0.05. A standard software
package (Origin version 7, OriginLab Corp., Northampton,
MA, USA) was utilized throughout.
S-NSAIDs and S-valproate inhibit tumor-driven vascular
outgrowth in 3D collagen explants
Utilizing a novel ex vivo model of tumor-driven angiogenesis,
we found that HT-29 tumor-driven angiogenic responses
were also very sensitive to the inhibitory effects of S-NSAIDs
and S-valproate (Figure 2c and d). Cytospin analysis of cell
outgrowth from tumor explants demonstrated a preponder-
ance of vascular cells composed of endothelial cells
(3575%), vascular smooth muscle cells (3371%) and
other unidentified cells (3377%) (Table 1). S-NSAID and
Materials
3H-1,2-dithiole-3-thione (D3T) was obtained from LKT
Laboratories (St Paul, MN, USA). Sodium hydrosulfide and
valproic acid were purchased from Sigma (St Louis, MO,
USA). Agents with limited solubility in water were initially
prepared as stock solutions in DMSO and serial dilutions
in the indicated cell-culture medium performed with final
concentrations of DMSO never exceeding 0.1%.
S-valproate treatment of HT-29 explants resulted in
a
decrease in vascular migration through extracellular matrix
(Figure 2c) as and in a decrease in the number of cells in the
outgrowth (Figure 2d). As in the muscle explants, nonmo-
dified parent compounds were much less active and did not
significantly inhibit tumor-driven angiogenesis except at
4100 mM (Figure 2e). Both ADT and D3T potently inhibited
tumor-driven angiogenesis in the explant assay (Figure 2f).
Vehicle alone did not inhibit the tumor explant vascular cell
response (data not shown).
Results
Chemical characterization and solubility of S-NSAIDs
The ADT-OH derivatives of diclofenac, sulindac and valpro-
ate were synthesized as outlined in Figure 1 and according to
Li et al.(2006). Nuclear magnetic resonance, mass spectro-
metry and elemental analysis were all consistent with the
predicted structures. The compounds S-diclofenac, S-sulin-
dac and S-valproate were found to have lower solubility as
compared with the parent compound and were all practi-
cally insoluble in water (water solubility o0.01%) and very
lipophilic. The calculated log P (ACD/Labs software V9.02,
Toronto, Canada) at 251C, for example, of compound
S-diclofenac is 7.18 vs a calculated log P of the parent
compound of 4.06, for compound S-valproate, the calculated
log P is 5.88 and where as that of the parent compound is 2.72.
S-NSAIDs and S-valproate inhibit endothelial cell proliferation
HUVEC proliferation was inhibited in a dose-dependent
manner by S-NSAIDs and S-valproate (Figure 3a). Consistent
with the explant data, the nonmodified parent compounds
had little or no effect on cell proliferation at equivalent
dosages (Figure 3b). As in vascular outgrowth, ADT inhibited
HUVEC proliferation. D3T demonstrated a slight, but not
statistically
(Figure 3c). In contrast, H2S stimulated endothelial cell
proliferation in dose-dependent manner (Figure 3d).
significant
inhibition
on
proliferation
a
Vehicle alone did not inhibit the vascular cell response (data
not shown).
COX inhibition by S-NSAIDs
To confirm that the S-NSAIDs retain their COX inhibitory
activities, they were assessed for inhibiting conversion of
arachidonic acid to PGE2 in HT-29 cells. Compared to basal
levels of 90721 pg mlÀ1, S-diclofenac and S-sulindac at 1 mM
reduced PGE2 production to 1679 and 58713, respectively
(mean7s.e., n ¼ 4).
Dithioles but not S-NSAIDs disrupt embryonic angiogenesis
To investigate the effect of the drugs on developmental
angiogenesis, we utilized the zebrafish angiogenesis model
system. In zebrafish, the dorsal aorta and posterior cardinal
vein, which traverse along the axis of the embryo, represent
vasculogenic vessels, whereas sprouts emerging from the
axial vessels represent the angiogenic intersomitic vessels.
The effect of drugs were studied in transgenic fish
Tg(fli1:EGFP) that carry a vascular specific 15 kb fli (Friend
leukemia inhibitor) promoter driving enhanced green fluor-
escent protein (Lawson and Weinstein, 2002) in angioblasts
and mature vasculature. S-NSAIDs and S-valproate demon-
strated minimal toxicity and had little effect upon inter-
S-NSAIDs and S-valproate inhibit wound-driven angiogenesis
Muscle biopsies from C57Bl6 mice were explanted into 3D
type I collagen gels in a 96-well plate format and incubated
in the presence of EGM and the indicated concentrations
of S-NSAIDs, S-valproate or H2S (Figure 2a). The S-NSAIDs,
S-valproate and H2S were dose-dependent inhibitors of
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
146
JS Isenberg et al
Figure 2 S-NSAIDs, S-valproate and dithiole thiones inhibited vascular outgrowth from muscle and tumor explants. Pectoralis major muscle
(a and b) or HT29 tumour biopsies (c, e and f) were explanted in type I collagen matrices and incubated in growth medium with the indicated
treatments for 7 days and vascular outgrowth quantified. Control and treated-HT29 adenocarcinoma explants were incubated under similar
conditions and, following matrix digestion, total cell counts determined using cytospin preparations (d). In (d), S-valproate, S-diclofenac and
S-sulindac are shown as ACS2, ACS15 and ACS18, respectively. S-NSAIDs and S-valproate caused significant inhibition (Po0.05) at all
concentrations (a, b, d and f) and from 10 to 100 mM (c), compared with control. Parent compounds were effective (Po0.05) at 100–1000 mM
(e) compared with control.
Table 1 Immunohistochemical analysis of HT29 xenograft explants
somitic vessels formation at the dose ranges employed in
zebrafish embryos (Figure 4a). However, there were moderate
numbers of gross developmental abnormalities noted among
embryos treated with these agents (Figure 4a). Importantly,
solubility limits for these compounds were encountered at
25 mM, above which precipitation occurred, probably owing
to the reactivity with Blue Water components. In contrast,
treatment of embryos with valproate (Figure 4b–b00) and
higher doses of sulindac (Figure 4c–c00), the parent com-
pounds of S-valproate and S-sulindac, respectively, were
associated with a significant increase in intersomitic vessel
defects. However, structural defects were also observed in
25 mM sulindac-treated embryos. D3T and ADT treatment
Cell type
%
Endothelial cells
Vascular smooth muscle cells
Other cell types
34.675a
32.771
32.777
HT29 adenocarcinoma xenograft explants were cultured in 3D matrices for
7 days. Vascular outgrowths were collected after digesting the collagen gels
using collagenase type 2. Slides were prepared by cytospin and stained using
CD31 (PECAM-1) or a-smooth muscle actin antibodies. CD31-positive
endothelial cells and a-smooth muscle actin-positive perivascular cells were
counted in 10 randomly selected fields. Results are the means of at least three
independent experiments.
aValues are the means7s.e.m.; n ¼ 3.
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
JS Isenberg et al
147
resulted in both vessel defects and gross developmental
defects (Figure 5a and d). In the case of D3T, we noticed
defects in the notochord at concentrations as low as 1 mM
that progressed in severity as the concentration increased to
5 mM (Figure 5c, asterisk) and beyond. To investigate whether
vessel defects were indeed primary, we added the drug at
18 hpf postnotochord development (Figures 5b, e and f).
We noticed intersomitic vessel defects in these embryos,
although at much higher concentrations of 25 mM
(Figure 5e). Interestingly, ADT treatment at 10 hpf was
associated with significant delay in gastrulation (Figure 5d),
and the drug uptake was evident by the yellow color seen
in the yolk photographs (Figure 5d). When ADT was added
to embryos at 18 hpf, the vessel defects were apparent
(Figure 5f), although the concentrations required to observe
this effect was higher (25 mM) than when treated at 10 hpf
(7.5 mM). In most of the drug-treated embryos, we did not
notice any defect in vasculogenesis, as both the dorsal aorta
and posterior cardinal vein remained unaffected.
S-NSAIDs, S-valproate, H2S and dithioles induce Hsp27
phosphorylation
Phosphorylation of heat shock protein 27 (Hsp27) is
a convergent target of signaling in endothelial cells exposed
to the angiogenesis inhibitors thrombospondin-1, fumagillin,
endostatin and TNP-470 (Keezer et al., 2003). Similar
regulation of Hsp27 phosphorylation was subsequently
reported for endorepellin, an antiangiogenic fragment of
the heparan sulfate proteoglycan perlecan (Bix et al.,
2004). To determine whether Hsp27 phosphorylation is also
regulated by S-NSAIDs, endothelial cells were exposed to
various concentrations of S-NSAIDs, H2S and D3T (Figures 6
and 7). After 60 min, S-sulindac stimulated Hsp27 Ser78
phosphorylation at 10–100 mM (Figure 6a), but sulindac
lacked activity at the same concentrations (Figure 6b). The
dithiolethione D3T showed an inverse dose response in the
same range, with strongest induction of Hsp27 phosphoryla-
tion at 1 mM (Figure 6a). Further dilution of the D3T showed
maximal phosphorylation at 0.3 mM (Figure 5c). ADT was
much less active and stimulated only weak Hsp27 phosphor-
ylation at 10 mM (Figure 6b). H2S was also much less active,
inducing maximal Hsp27 phosphorylation at 100 mM
(Figure 6a).
H2S in some cases parallels NO in cellular responses, and
NO, via activation of cGMP-dependent protein kinase, is
known to induce phosphorylation of Hsp27 at Thr143 (Butt
et al., 2001). Interestingly, spermine-NONOate also increased
Hsp27 Ser78 phosphorylation in endothelial cells (Figure 6a).
The maximal response at 100 mM suggests that this is
mediated by reactive nitrogen intermediates that form at
Figure
3 S-NSAIDs and S-valproate inhibited endothelial cell
proliferation. HUVEC (5 Â 103 cells wellÀ1) were plated on 96-well
plates and incubated for 72 h in EGM þ 2% FCS with the indicated
concentrations of S-NSAIDs and S-valproate (a), parent compounds
(b), dithioles alone (c) and hydrogen sulfide (d). Cell proliferation
was assayed via the colorimetric change obtained after incubation
with MTS [3-(4,5 dimethylthiazol-2-yl)-5-(3-carboxymethoxyphe-
nyl-2-(4-sulfophenyl)-2H-tetrazolium] reagent using a microplate
reader at 490 nm and is presented as percent of untreated control
wells and corrected for the initial cell signal. Treatments were
effective (Po0.05) at concentrations from 10 to 1000 mM (a), 0.1–
100 mM (c) and 1–1000 mM (d) compared with control.
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
148
JS Isenberg et al
Figure 4 S-NSAIDs and S-valproate did not cause significant defects in vessels in developing zebrafish embryos. Dechorinated zebrafish
embryos were incubated in the presence of the indicated concentrations of S-NSAIDs and S-valproate or nonderivatized parent compound of
S-sulindac. Vascular defects in intersegmental vessels, gross defects or embryo death were quantified (a) (n ¼ 40–60 embryos/experimental
point). Representative embryos treated as indicated were photographed 48 hpf with light (b–d) or fluorescent illumination (b0–d0).
Enlargements of the fluorescent images (b00–d00) show trunk intersegmental vessels branching from the dorsal aorta. Data are expressed as %
defective.
stress levels of NO rather than by NO/cGMP signaling that is
typically observed at o1 mM spermine-NONOate (Thomas
et al., 2004). Oxidizing intermediates formed from reactive
oxygen species induce Hsp27 phosphorylation in endothe-
lial cells through the p38/MAPKAP2 pathway (Nguyen et al.,
2004). Using 100 mM H2O2, we observed moderate phosphor-
ylation of Ser78 (Figure 6b) but much less than in response to
D3T, suggesting that the latter induces Hsp27 phosphoryla-
tion through a different mechanism.
To examine further the potential role of p38/MAPKAP2
signaling in dithiolethione-induced Hsp27 phosphorylation,
we examined the effects of these on mitogen-activated
protein (MAP) kinase activation (Figure 7b). Neither p38,
JNK nor Erk showed increased phosphorylation after treat-
ment for 60 min with ADT, D3T or S-valproate. S-diclofenac,
S-sulindac and sulindac modestly increased p38 phosphor-
ylation. Therefore, increased p38 activation does not
account for the dithiolethione-induced phosphorylation
of Hsp27.
Treatment with S-diclofenac and S-valproate also induced
Hsp27 phosphorylation, with maximal responses at 100 mM
and similar to S-sulindac (Figure 7a). However, the corre-
sponding parent compounds diclofenac and valproate also
moderately increased Hsp27 phosphorylation (Figure 7a).
These responses were maximal at 10 and/or 100 mM. The
Hsp27 response to valproate is consistent with its reported
antiangiogenic activity (Michaelis et al., 2004) and the
intersomitic vessel defects in valproate-treated zebrafish
embryos.
Discussion and conclusions
NSAIDs such as diclofenac are well studied for their anti-
inflammatory properties via COX inhibition (Ulrich et al.,
2006). Here, we show that S-NSAIDs and S-valproate
significantly inhibit pro-angiogenic responses of vascular
cells under both wound healing and tumor growth condi-
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
JS Isenberg et al
149
Figure 5 Dithiole moieties caused vascular and developmental defects in zebrafish embryos. Zebrafish embryos were incubated in the
presence of the indicated concentrations of dithiole moieties and counted for developmental and vascular abnormalities (a) (n ¼ 40–100
embryos/experimental point) or added at 18 hpF (b, e, f) (n ¼ 30 embryos/experimental point). Representative embryos were photographed
48 hpf with light (c–f) or fluorescent illumination (c0–f0). Data are expressed as % defective. High magnification images showing intersomitic
vessels are shown in c00, d0, e00 and f00.
tions. Vascular cell migration through extracellular matrix
was significantly blocked by S-NSAIDs and S-valproate.
Additionally, total cell numbers obtained from digested
extracellular matrix in tumor explants were dramatically
decreased by S-NSAIDs, S-valproate and dithioles, consistent
with the activities of these agents in inhibiting endothelial
cell proliferation. In contrast, the parent drugs had no
inhibitory effects on vascular cell outgrowth in either muscle
or tumor explants. Treatment with the dithioles ADT and
D3T mimicked the inhibitory effects of the S-NSAIDs and
S-valproate in ex vivo wound and tumor-driven explants,
suggesting that the antiangiogenic activity of the S-NSAIDs
and S-valproate resides in their dithiole moiety. Although
the two S-NSAIDs tested are COX inhibitors (Li et al., 2006),
their antiangiogenic activities appear to be independent of
COX inhibition.
increased cell proliferation and migration. Consistent with
their activity for blocking angiogenic responses under
wound healing and tumor-driven conditions, we found that
S-NSAIDs, S-valproate and dithioles inhibited endothelial
cell proliferation in vitro. In contrast, hydrogen sulfide
stimulated endothelial cell proliferation in a dose-dependent
manner. In contrast to some other activities of S-NSAIDs and
dithioles, their antiangiogenic activity is probably indepen-
dent of H2S release (Distrutti et al., 2006; Li et al., 2006).
Because cyclooxygenases are also known to play a role in
angiogenesis (Gately and Li, 2004), the lack of effects of
diclofenac and sulindac on endothelial cell proliferation and
their weak inhibition of explant angiogenesis were unex-
pected. The most active metabolite of sulindac for cyclo-
oxygenase inhibition is the sulfide, but previous studies have
indicated that other sulindac metabolites can also influence
angiogenic responses independent of cyclooxygenase inhibi-
tion (Gourzoulidou et al., 2005). Sulindac sulfide induces
Angiogenesis requires a number of specific responses in
both endothelial and vascular smooth muscle cells including
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
150
JS Isenberg et al
antangiogenic in zebrafish embryos but not for endothelial
cell proliferation suggests that this activity of sulindac is
context-dependent.
As a heat-shock protein, Hsp27 supports cell survival and
recovery from thermal and oxidative stress (Arrigo et al.,
2005). However, it also plays important roles in normal cell
signaling. Mediated by MAPKAP-2 downstream from p38,
phosphorylation of Hsp27 stabilizes the actin cytoskeleton
and decreases cell migration (Landry and Huot, 1999; Keezer
et al., 2003). Consistent with the effects of other angiogen-
esis inhibitors to stimulate Hsp27 phosphorylation (Keezer
et al., 2003; Bix et al., 2004) we found that phosphorylation
of Hsp27 was induced by S-NSAIDs, S-valproate, D3T and
H2S, suggesting that these agents exert some of their
antiangiogenic effects by regulating signaling pathways
downstream of Hsp27. It is known that activation of the
p38 pathway stimulates Hsp27 phosphorylation at all three
serine residues (S15, S78 and S82). However, S-NSAID-
induced Hsp27 phosphorylation is mainly at S78. In
addition, the phosphorylation of Hsp27-S78 is not sensitive
to the p38-specific inhibitor SB203580 (data not shown).
Together, these results suggest these compounds might
induce Hsp27-S78 phosphorylation through other signaling
pathways.
Figure 6 S-sulindac and D3T but not sulindac increased phosphor-
ylation of Hsp27. HUVEC were serum starved overnight and then
incubated for 60 min in the presence of S-sulindac, D3T, spermine-
NO (1–100 mM) or hydrogen sulfide (10–1000 mM) (a) and with
H2O2, ADT or sulindac (1–100 mM) (b), cells lysed and Western blot
assays performed using an antibody specific for Hsp27 phosphory-
lated at Ser78. In other experiments (c), an expanded concentration
range of D3T was employed.
The potent effects of S-NSAIDs, S-valproate and simple
dithioles on wound healing and tumor-driven angiogenesis
in an explant model correlated with their inhibitory effects
on endothelial cell proliferation in vitro. Although direct
toxicity studies were not carried out in murine models,
others have employed H2S-releasing derivatives of diclofenac
in vivo in rat models with success, suggesting that these
agents may be useful in therapeutic situations (Bhatia et al.,
2005; Li et al., 2006; Ong et al., 2006). However, in a zebrafish
model the S-NSAIDs and S-valproate demonstrated relatively
modest effects upon intersomitic vessel formation. This
selective activity of S-NSAIDs and S-valproate may reflect
the known mechanistic differences between developmental
angiogenesis and pathological angiogenesis in the adult
(Carmeliet, 2004). From this standpoint, our zebrafish
screening approach could be useful to differentiate drugs
that inhibit embryonic angiogenesis as opposed to tumor-
associated angiogenesis in the adult. Alternatively or con-
comitantly, hydrolytic cleavage of the S-NSAIDs and
S-valproate maybe required to release the antiangiogenic
dithiole moiety. It is possible that hydrolysis does not occur
in the zebrafish developmental angiogenesis model owing
to the absence of a secreted esterase.
Figure 7 S-valproate and S-diclofenac and their parent drugs
altered Hsp27 phosphorylation. HUVEC were treated with
S-valproate, S-diclofenac and their representative parent compounds
at the indicated concentrations for 60 min, and Western blot analysis
performed for Hsp27 (Ser78P) (a). Cell lysates from HUVEC treated
for 60 min with the indicated agents were analyzed for phospho-
p38, phospho-JNK or phospho-ERK (b).
Activity of valproate in the zebrafish model is consistent
with earlier published reports of its inhibition of vessel
development in the chicken chorioallantoic membrane and
in mice (Michaelis et al., 2004; Zgouras et al., 2004), and
thereby validates the zebrafish model as a screening tool for
angiogenesis inhibitors. The endothelial inhibition results
and the tumor explant studies argue that both S-NSAIDs
and S-valproate are angiogenesis inhibitors, with the primary
antiangiogenic effects most likely residing in the
dithiolthione moieties. These effects are separate from any
antiangiogenic or anti-inflammatory effects of the parent
compounds. S-NSAIDs and S-valproate, by combining the
favorable activities of conventional anti-inflammatory drugs
endothelial cell apoptosis via caspase-3 activation (Flis et al.,
2006). However, earlier studies showed that sulindac sulf-
oxide, the sulfone and sulfide all inhibit angiogenesis in
chick chorioallantoic membrane angiogenesis assays, and
the sulfone was in fact 10-fold more active than the sulfide,
which in turn was twice as active as sulindac itself (Sharma
et al., 2001; Elwich-Flis et al., 2003). Our data that sulindac is
British Journal of Pharmacology (2007) 151 142–151

Dithiole-modified NSAIDs inhibit angiogenesis
JS Isenberg et al
151
and dithioles, and the decreased gastrointestinal side effects
in the case of S-diclofenac (Li et al., 2006), may increase
their effectiveness as drugs. The antiangiogenic activities of
S-NSAIDs may be useful in cancer therapy.
Folkman J (2006). Angiogenesis. Annu Rev Med 57: 1–18.
Gately S, Li WW (2004). Multiple roles of COX-2 in tumor angiogenesis:
a target for antiangiogenic therapy. Semin Oncol 31: 2–11.
Gourzoulidou E, Carpintero M, Baumhof P, Giannis A, Waldmann H
(2005). Inhibition of angiogenesis-relevant receptor tyrosine
kinases by sulindac analogues. Chembiochem 6: 527–531.
Isenberg JS, Ridnour LA, Perruccio EM, Espey MG, Wink DA, Roberts
DD (2005). Thrombospondin-1 inhibits endothelial cell responses
to nitric oxide in a cGMP-dependent manner. Proc Natl Acad Sci
USA 102: 13141–13146.
Acknowledgements
This research was supported in part by the Intramural
Research Program of the National Institutes of Health,
National Cancer Institute, Center for Cancer Research.
Keezer SM, Ivie SE, Krutzsch HC, Tandle A, Libutti SK, Roberts DD
(2003). Angiogenesis inhibitors target the endothelial cell cyto-
skeleton through altered regulation of heat shock protein 27 and
cofilin. Cancer Res 63: 6405–6412.
Kensler T, Styczynski P, Groopman J, Helzlsouer K, Curphey T,
Maxuitenko Y et al. (1992). Mechanisms of chemoprotection by
oltipraz. J Cell Biochem Suppl 16I: 167–172.
Kensler TW, Curphey TJ, Maxiutenko Y, Roebuck BD (2000).
Chemoprotection by organosulfur inducers of phase 2 enzymes:
dithiolethiones and dithiins. Drug Metabol Drug Interact 17: 3–22.
Kwak MK, Egner PA, Dolan PM, Ramos-Gomez M, Groopman JD,
Conflict of interest
Dr Piero Del Soldato is a shareholder of CTG Pharma, Milan,
Italy. This company has patents on reagents used in this
study. Dr Anna Sparatore received a grant from CTG Pharma.
Itoh
K et al. (2001). Role of phase 2 enzyme induction in
chemoprotection by dithiolethiones. Mutat Res 480–481: 305–315.
Landry J, Huot J (1999). Regulation of actin dynamics by stress-
activated protein kinase 2 (SAPK2)-dependent phosphorylation of
heat-shock protein of 27 kDa (Hsp27). Biochem Soc Symp 64: 79–89.
Lawson ND, Weinstein BM (2002). In vivo imaging of embryonic
vascular development using transgenic zebrafish. Dev Bio 248:
307–318.
References
Arrigo AP, Virot S, Chaufour S, Firdaus W, Kretz-Remy C, Diaz-Latoud
C (2005). Hsp27 consolidates intracellular redox homeostasis by
upholding glutathione in its reduced form and by decreasing iron
intracellular levels. Antioxid Redox Signal 7: 414–422.
Ben-Mahdi MH, Gozin A, Driss F, Andrieu V, Christen MO, Pasquier C
(2000). Anethole dithiolethione regulates oxidant-induced tyro-
sine kinase activation in endothelial cells. Antioxid Redox Signal 2:
789–799.
Lee JS, Surh YJ (2005). Nrf2 as
a novel molecular target for
chemoprevention. Cancer Lett 224: 171–184.
Li L, Rossoni G, Sparatore A, Lee L, Del Soldato P, Moore P (2006).
Anti-inflammatory and gatrointestinal sparing activity of a novel
H2S-releasing diclofenac agent: new insights into the biological
roles of H2S. Free Radic Biol Med 42: 706–719.
Michaelis M, Michaelis UR, Fleming I, Suhan T, Cinatl J, Blaheta RA
et al. (2004). Valproic acid inhibits angiogenesis in vitro and in vivo.
Mol Pharmacol 65: 520–527.
Nguyen A, Chen P, Cai H (2004). Role of CaMKII in hydrogen
peroxide activation of ERK1/2, p38 MAPK, HSP27 and actin
reorganization in endothelial cells. FEBS Lett 572: 307–313.
Ong MM, Sparatore A, del Soldato P, Moore PK, Treinen-Moslen M,
Boelsterli UA (2006). Differential covalent adduct biodistribution
of diclofenac and a diclofenac sulfurated ester in rat liver and
small intestine. The Toxicologist S2136: 438.
Bhatia M, Sidhapuriwala JAS, Moore PK (2005). Treatment with H2S-
releasing derivative of diclofenac reduces inflammation in carra-
geenan-induced hindpaw oedema. Inflammation Res (Suppl. a):
S185.
Bix G, Fu J, Gonzalez EM, Macro L, Barker A, Campbell S et al. (2004).
Endorepellin causes endothelial cell disassembly of actin cytoske-
leton and focal adhesions through alpha2beta1 integrin. J Cell Biol
166: 97–109.
Bottcher B, Bauer F, Trithiones V (1951). Some new trithiones.
Chemische Berichte 84: 458–463.
Ruggeri BA, Robinson C, Angeles T, Wilkinson J, Clapper ML (2002).
The chemopreventive agent oltipraz possesses potent antiangio-
genic activity in vitro, ex vivo, and in vivo and inhibits tumor
xenograft growth. Clin Cancer Res 8: 267–274.
Sharma S, Ghoddoussi M, Gao P, Kelloff GJ, Steele VE, Kopelovich L
(2001). A quantitative angiogenesis model for efficacy testing of
chemopreventive agents. Anticancer Res 21: 3829–3837.
Thomas DD, Espey MG, Ridnour LA, Hofseth LJ, Mancardi D, Harris
CC et al. (2004). Hypoxic inducible factor 1alpha, extracellular
signal-regulated kinase, and p53 are regulated by distinct thresh-
old concentrations of nitric oxide. Proc Natl Acad Sci USA 101:
8894–8899.
Ulrich CM, Bigler J, Potter JD (2006). Non-steroidal anti-inflamma-
tory drugs for cancer prevention: promise, perils and pharmaco-
genetics. Nat Rev Cancer 6: 130–140.
Verheul HM, Pinedo HM (2005). Angiogenesis inhibitors: what is the
clinical future? Prog Drug Res 63: 67–91.
Butt E, Immler D, Meyer HE, Kotlyarov A, Laass K, Gaestel M (2001).
Heat shock protein 27 is a substrate of cGMP-dependent protein
kinase in intact human platelets: phosphorylation-induced
actin polymerization caused by HSP27 mutants. J Biol Chem 276:
7108–7113.
Cao Y (2005). Tumor angiogenesis and therapy. Biomed Pharmacother
59: S340–S343.
Carmeliet P (2004). Manipulating angiogenesis in medicine. J Intern
Med 255: 538–561.
Carmeliet P (2005). Angiogenesis in life, disease and medicine.
Nature 438: 932–936.
Distrutti E, Sediari L, Mencarelli A, Renga B, Orlandi S, Russo G et al.
(2006). 5-Amino-2-hydroxybenzoic acid 4-(5-thioxo-5H-[1,2]dithiol-
3yl)-phenyl ester (ATB-429), a hydrogen sulfide-releasing derivative
of mesalamine, exerts antinociceptive effects in a model of pos-
tinflammatory hypersensitivity. J Pharmacol Exp Ther 319: 447–458.
Elwich-Flis S, Soltysiak-Pawluczuk D, Splawinski
J (2003). Anti-
Westerfield M (2000). The Zebrafish Book. University of Oregon Press,
Eugene, Oregon, p363.
Yao KS, O’Dwyer PJ (2003). Role of the AP-1 element and redox
factor-1 (Ref-1) in mediating transcriptional induction of DT-
diaphorase gene expression by oltipraz: a target for chemopreven-
tion. Biochem Pharmacol 66: 15–23.
Zgouras D, Becker U, Loitsch S, Stein J (2004). Modulation of
angiogenesis-related protein synthesis by valproic acid. Biochem
Biophys Res Commun 316: 693–697.
angiogenic and apoptotic effects of metabolites of sulindac on
chick embryo chorioallantoic membrane. Hybrid Hybridomics 22:
55–60.
Ferrara N, Kerbel RS (2005). Angiogenesis as a therapeutic target.
Nature 438: 967–974.
Flis S, Soltysiak-Pawluczuk D, Jedrych A, Jastrzebski Z, Remiszewska
M, Splawinski J (2006). Antiangiogenic effect of sulindac sulfide
could be secondary to induction of apoptosis and cell cycle arrest.
Anticancer Res 26: 3033–3041.
British Journal of Pharmacology (2007) 151 142–151