Full text of DOI:10.1038/sj.bjp.0704497

British Journal of Pharmacology (2002) 135, 407 ± 416
ã
2002 Nature Publishing Group All rights reserved 0007 ± 1188/02 $25.00
www.nature.com/bjp
Disparate e€ects of non-steroidal anti-in¯ammatory drugs on
apoptosis in guinea-pig gastric mucous cells: inhibition of basal
apoptosis by diclofenac
1Miranda Ashton & *,1Peter J. Hanson
1Pharmaceutical Sciences Institute, Aston University, Aston Triangle, Birmingham B4 7ET
1
Non-steroidal anti-in¯ammatory drugs (NSAIDs) induce apoptosis in gastrointestinal cancer cell
lines. Similar actions on normal gastric epithelial cells could contribute to NSAID gastropathy. The
present work therefore compared the actions of diclofenac, ibuprofen, indomethacin, and the cyclo-
oxygenase-2 selective inhibitor, NS-398, on a primary culture of guinea-pig gastric mucous epithelial
cells.
2
Cell number was assessed by staining with crystal violet. Apoptotic activity was determined by
condensation and fragmentation of nuclei and by assay of caspase-3-like activity. Necrosis was
evaluated from release of cellular enzymes.
3
Ibuprofen (250 mM for 24 h) promoted cell loss, and apoptosis, under both basal conditions and
when apoptosis was increased by 25 mM N-Hexanoyl-D-sphingosine (C6-ceramide).
4 Diclofenac (250 mM for 24 h) reduced the proportion of apoptotic nuclei from 5.2 to 2.1%, and
caused inhibition of caspase-3-like activity, without causing necrosis under basal conditions. No such
reduction in apoptotic activity was evident in the presence of 25 mM C6-ceramide.
5
The inhibitory e€ect of diclofenac on basal caspase-3-like activity was also exhibited by the
structurally similar mefenamic and ¯ufenamic acids (1 ± 250 mM), but not by ni¯umic acid.
Inhibition of superoxide production by the cells increased caspase-3-like activity, but the
6
inhibitory action of diclofenac on caspase activity remained. Diclofenac did not a€ect superoxide
production.
7
Diclofenac inhibited caspase-3-like activity in cell homogenates and also inhibited human
recombinant caspase-3.
In conclusion, NSAIDs vary in their e€ect on apoptotic activity in a primary culture of guinea-
8
pig gastric mucous epithelial cells, and the inhibitory e€ect of diclofenac on basal apoptosis could
involve an action on caspase activity.
British Journal of Pharmacology (2002) 135, 407 ± 416
Keywords: Non-steroidal anti-in¯ammatory drug; apoptosis; caspase; gastric epithelial cell; stomach
Abbreviations: NO, nitric oxide; NSAIDs, non-steroidal anti-in¯ammatory drugs; TNF-a, tumour necrosis factor alpha; N-
Hexanoyl-D-sphingosine, C6-ceramide
Introduction
In the gastrointestinal tract the size of the mucosal cell
population is determined by the balance between cell gain
from proliferation and cell loss from apoptosis (programmed
cell death) and necrosis (cell lysis). Apoptosis occurs close to
the proliferative zone and at the surface of the epithelium
(Hall et al., 1994). Too much apoptosis can lead to tissue
atrophy or potentially to the induction of a compensatory
hyper-proliferative response.
Non-steroidal anti-in¯ammatory drugs (NSAIDs) damage
the gastric mucosa (Wallace, 1997; Hawkey, 2000). Inhibition
of prostaglandin biosynthesis by NSAIDs, with associated
reductions in mucus and bicarbonate secretion, and of blood
¯ow, is very probably of major importance in the acute
action of NSAIDs. In animal models a tumour necrosis
factor a (TNF-a) mediated recruitment of neutrophils to the
capillary endothelium (Santucci et al., 1995) is an important
mediator of NSAID damage. Neutrophils may release
damaging reactive oxygen intermediates, and obstruct blood
¯ow. Direct topical actions of NSAIDs on gastric surface
epithelial cells have also been suggested (see Wallace, 1997).
Administration of NSAIDs to rats as a bolus causes
apoptosis by a mechanism that involves TNF-a (Fiorucci et
al., 1999). Furthermore apoptosis is increased in the stomach
of human patients taking NSAIDs (Zhu et al., 1998).
Increased apoptosis could contribute to either the gastric
atrophy or the hyperplasia seen with NSAIDs (Taha et al.,
1992), the ®nal result depending on whether or not there is a
compensatory proliferative response. Experiments with
transformed cell lines (Lu et al., 1995; Chan et al., 1998;
Simmons et al., 1999; Kusuhara et al., 1999; Zhu et al., 1999)
have indicated that NSAIDs can induce apoptosis by a direct
action on cells. Furthermore, ¯urbiprofen increased apoptosis
in a primary culture of guinea-pig gastric epithelial cells
*Author for correspondence; E-mail: P.J.Hanson@aston.ac.uk

408
M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
(Johal & Hanson, 2000) showing that such a target for NSAIDs
could be present in non-transformed cells. The guinea-pig cell
preparation consists of at least 90% of gastric mucous epithelial
cells (Byrne & Hanson, 1998; Teshima et al., 1998), and exhibits
spontaneous apoptosis that is dependent on the presence of
serum and protein synthesis, and which may represent the
normal process of turnover of gastric surface mucous cells (Johal
& Hanson, 2000; Tsutsumi et al., 2000). Nuclear fragmentation
and condensation and increased activity of the e€ector
proteolytic enzyme, caspase 3 (Song & Steller, 2000), can be
used as indicators of apoptosis, while necrosis can be estimated
from the early release of enzymes such as lactate dehydrogenase
and acid phosphatase.
penicillin, 100 mg ml71 streptomycin and 2.5 mg ml71 of
amphotericin B) and cultured on six- or twelve-well culture
plates. The culture medium was renewed after 24 h, and
again at 48 h, at which time some experiments were initiated
by the inclusion of agents in the culture medium. For
experiments involving shorter treatment periods a further
change of medium was performed on the day on which cells
were harvested after a total 72 h of culture. Agents were
dissolved in dimethylsulphoxide, the concentration of which
was never greater than 0.25% (v v71), and which was always
the same in control and experimental wells. NS-398 was not
used at concentrations higher than 100 mM because of its
limited solubility in culture medium.
The aim of the present work was to examine three
commonly used NSAIDs, which do not exhibit major
selectivity as inhibitors of cyclo-oxygenases-1 and -2 (Warner
et al., 1999) and the selective cyclo-oxygenase-2 inhibitor NS-
398 (Warner et al., 1999) for direct e€ects on apoptosis in
primary cultures of guinea-pig gastric epithelial cells. Cyclo-
oxygenase-2 selective agents exhibit considerably less gastro-
toxicity than conventional non-selective NSAIDs (Whittle,
2000), but have nevertheless been shown to increase apoptosis
in colon cancer cell lines (Elder et al., 2000). Contrary to the
®ndings obtained with transformed cell lines (e.g. Simmons et
al., 1999), we ®nd that there are major di€erences between
the NSAIDs in their e€ect on apoptotic activity in the
primary culture of gastric epithelial cells.
Crystal violet assay
Cells were grown in 12-well plates. The culture medium was
removed and 0.5 ml of crystal violet (0.4 g 100 ml71) in 30%
(v v71) methanol was added to each well for 20 min at room
temperature. After removal of the stain, wells were washed
three times with 1 ml of water and then allowed to dry. Stain
was released from cells by incubation with agitation for
30 min at room temperature with 400 ml of sodium dodecyl
sulphate (1 g 100 ml71) in each well. Absorbance of 150 ml of
the extracts was read at 570 nm on a plate reader.
Attached and detached cell protein
Medium was removed from the plates and centrifuged at
12,0006g for 20 s. The pellet of detached cells was washed
with 1 ml of 138 mM NaCl, 2.7 mM KCl, 10 mM phosphate
pH 7.4 (PBS), and the pellet ®nally resuspended in 0.15 ml of
PBS containing 1% (v v71) Triton X-100. The wells were
washed with 1 ml of PBS, and attached cells were scraped o€
into 0.4 ml of PBS containing 1% (v v71) Triton X-100.
Protein was measured by using bicinchoninic acid reagent
with bovine serum albumin as standard (Redinbaugh &
Turley, 1986).
Methods
Animals
Male Dunkin-Hartley guinea-pigs of 200 ± 300 g body weight
were obtained from Charles River, Margate, Kent U.K. and
were fed on SDS Economy guinea-pig diet supplied by
Lillico, Betchworth, Surrey, U.K.
Materials
Detection of apoptosis by staining with Hoechst 33258
RPMI 1640 medium, Hanks-bu€ered salt solution without
phenol red, foetal calf serum, antibiotics and amphotericin B
were from Life Technologies, Paisley U.K. Pronase E (70,000
PU units per g) was purchased from Merck, Lutterworth,
U.K. [3H]-thymidine was from Amersham Pharmacia, Little
Chalfont, U.K. Recombinant murine tumour necrosis factor-
a (TNF-a, 66107 units mg71) was from Roche, Lewes, U.K.
Other reagents, including activated human recombinant
caspase-3 with a C-terminal histidine tag on the 13.5 kDa
subunit, were from Sigma, Poole, U.K.
Cells were removed from the culture plate by exposure to
trypsin/EDTA, and were subjected to centrifugation for 20 s
at 12,0006g. Cells were resuspended in 4% (w v71
)
paraformaldehyde in PBS for 10 min at room temperature,
subjected to centrifugation at 12,0006g for 10 s, resuspended
in ethanol: water (4 : 1) and stored at 48C. Cells were
transferred to glass slides by use of a Shandon cytocentrifuge,
and nuclei were stained with 8 mg ml71 of Hoechst 33258.
Slides were coded, and the proportion of apoptotic nuclei was
determined by counting using a ¯uorescent microscope with
apoptosis de®ned as the presence of two or more condensed
bodies per nucleus.
Isolation and culture of gastric mucous cells
The method has been described in detail previously (Byrne &
Hanson, 1998). Gastric mucosa was minced with ®ne scissors,
incubated with 45 ml of RPMI 1640 containing 2 g l71
bovine serum albumin (isolation medium) and 0.5 mg ml71
pronase for 20 min at 378C, and then, after centrifugation,
incubated with 45 ml of isolation medium containing
0.4 mg ml71 collagenase for 20 min. Cells were ®ltered
through 150 mm nylon mesh, washed in culture medium
(RPMI 1640 containing 10% foetal calf serum, 100 u ml71
Assay of caspase activity
Plates were placed on ice, the culture medium was removed
and 1 ml of PBS was added to each well. The cells were
removed from the plate with a cell-scraper, transferred to
microfuge tubes, and were subjected to centrifugation for 20 s
at 12,0006g. Pellets were rinsed with 1.0 ml of ice cold PBS
and then were resuspended, by pipetting up and down 10
British Journal of Pharmacology vol 135 (2)

M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
409
times, in 110 ml of homogenization bu€er (in mM) HEPES
100, NaCl 140, EDTA 1, (pH 7.4), containing phenylmethyl-
sulphonyl ¯uoride 0.5, aprotinin 5 mg ml71
,
pepstatin
5 mg ml71 and leupeptin 10 mg ml71. Extracts were then
frozen in liquid nitrogen and stored at 7708C. Before assay
homogenates were thawed and refrozen twice, before a ®nal
thaw and centrifugation at 12,0006g for 20 min at 48C.
Twenty-®ve ml of supernatant was incubated at 378C with
175 ml of assay bu€er consisting of 100 mM HEPES (pH 7.4)
and 20% (v v71) glycerol containing protease inhibitors at the
same concentration as above, and the caspase substrate Ac-
Asp-Glu-Val-Asp-7-amido-4-methylcoumarin at a ®nal con-
centration of 15 mM. Formation of ¯uorescent product was
measured on a Victor multilable counter. Five mM dithio-
threitol was added to the assay bu€er to stabilize the activity
of recombinant caspase-3, but dithiothreitol was not included
in the assay of tissue homogenates since concentrations up to
20 mM had no e€ect on activity.
Lactate dehydrogenase activity
Samples were prepared as described for the protein assay,
except that culture medium from wells with and without cells
was retained after centrifugation and Triton X-100 added to
give a ®nal concentration of 1% (v v71). Lactate dehydro-
genase activity was assayed immediately by mixing samples
with 0.15 ml of 50 mM potassium phosphate bu€er contain-
ing 4.4 mM pyruvate and 0.4 mM NADH and measuring the
rate of change of absorbance at 340 nm on a plate reader.
Results were corrected for lactate dehydrogenase present in
the serum component of culture medium, and the activity in
the medium was then expressed as a percentage of that in
medium, detached cells and attached cells (total activity).
Acid phosphatase activity
Preparation of samples and analysis of results was as for
lactate dehydrogenase. Enzyme activity was determined by
the breakdown of p-nitrophenyl phosphate at pH 4.8 as
described by Wong & Tepperman (1994).
Incorporation of [3H]-thymidine
Two mCi of [3H]-thymidine (25 Ci mmol71) was added to
each well of a 12-well plate. After 24 h the medium was
removed and the wells washed with 1 ml of ice-cold PBS.
Cells were scraped into 0.75 ml of trichloroacetic acid (10 g
100 ml71 (TCA)), and left for 20 min on ice. The contents of
the wells, and 0.5 ml of TCA used to rinse the wells, were
transferred to microfuge tubes, centrifuged at 12,0006g for
1 min, the pellet washed with TCA, and ®nally dissolved in
0.2 ml NaOH (0.3 M) containing 0.1 g 100 ml71 sodium
dodecylsulphate at 378C for 4 h. Incorporation of [3H]-
normalized to the attached cell protein in control wells, which was
77+7 mg per well. In (c) cell protein in the culture medium has been
normalized to that in control incubations which was 12.4+1.9 mg per
well. Data are means+s.e.mean from respectively four and six
cultures for the crystal violet and protein measurements. Data were
analysed before normalization by ANOVAR and Dunnett's test.
*P50.05 and **P50.01 for di€erence from control.
Figure 1 E€ect of incubation with non-steroidal anti-in¯ammatory
drugs (NSAIDs, 250 mM except for NS-398 at 100 mM) for 24 h on
estimates of cell number. Staining of attached cells with crystal violet
(a) has been normalized to the absorbance at 570 nm obtained with
control cells, which was 1.162+0.154. In (b) data have been
British Journal of Pharmacology vol 135 (2)

410
M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
thymidine was determined by scintillation counting and was
expressed as a function of cellular protein.
Production of superoxide
Wells were washed three times with 1 ml of PBS at 378C. Hanks-
bu€ered salt solution (0.75 ml) without phenol red and
containing 80 mM cytochrome c from horse heart (Sigma
C7752) was added to each well and the plate incubated at 378C
for 1 h. Superoxide dismutase from bovine erythrocytes (Sigma
S2515) was included at 100 units per ml where appropriate.
Reduction of cytochrome c was measured at 550 nm and an
extinction coecient of 216103 ml mmol71 cm71 used in the
calculation.
Statistical analysis
Figure 2 E€ect of incubation with non-steroidal anti-in¯ammatory
drugs (NSAIDs, 250 mM except for NS-398 at 100 mM) on release into
the culture medium of lactate dehydrogenase (LDH) after 4 h (open
bars, n=4 cell batches) and 24 h (®lled bars, n=5 cell batches), and
on release of acid phosphatase after 24 h (hatched bars, n=3 cell
Results are presented as means+s.e.mean, with n equal to the
number of separate cell cultures, and unless stated otherwise
were subjected to analysis of variance, to remove variation
between cell cultures, followed by either a Newman-Keuls or
a Dunnett's multiple comparison test. Data on the percentage
of apoptotic cells were arcsine-transformed before analysis.
batches) Results are means+s.e.mean and are expressed as
a
percentage of total enzyme activity. Data were analysed by
ANOVAR and Dunnett's test. *P50.05 and **P50.01 for di€erence
from appropriate control.
Figure 3 E€ect of incubation with non-steroidal anti-in¯ammatory drugs (NSAIDs, 250 mM except for NS-398 at 100 mM) for 24 h
on the proportion of apoptotic nuclei (a,c) and on caspase-3-like activity (b,d) in attached cells. Results are means+s.e.mean for
basal conditions in (a) and (b), n=6 and n=5 respectively, and in the presence of 25 mM C6-ceramide in (c) and (d), n=4 in both
cases. The control caspase activity used for normalization in (b) was 199+25, and in (d), 498+147 pmol min71mg protein71. Data
were analysed by ANOVAR and Dunnett's test. *P50.05; **P50.01; ***P50.001 for di€erence from control or from 25 mM C6-
ceramide alone.
British Journal of Pharmacology vol 135 (2)

M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
411
Results
appropriate direction, of caspase-3-like activity (Figure 3b).
Caspase-3-like activity is elevated in cells which have been
detached from the culture plate with the majority undergoing
apoptosis (Johal & Hanson, 2000). E€ects of NSAIDs on
caspase activity (pmol min71 mg protein71) in detached cells
were (control result versus that with NSAID, n=number of
preparations, P value for paired t-test for e€ect of NSAID):
250 mM indomethacin: 1733+466 vs 1439+445, n=4, P50.05;
250 mM diclofenac: 1614+411 vs 1154+359, n=5 P50.05;
250 mM ibuprofen: 1159+307 vs 2048+301, n=6, P50.05;
100 mM NS-398: 2045+530 vs 821+275, n=4, P50.05.
Effect of incubation with NSAIDs for 24 h on the number
of cells
Both ibuprofen (250 mM) and NS-398 (100 mM) reduced the
number of viable cells attached to the culture plate as
estimated by staining with crystal violet, while 250 mM
indomethacin or diclofenac were without e€ect (Figure 1a).
A similar pattern of results was obtained if attached cell
protein was used as an indicator of cell number (Figure 1b).
Reductions in attached cells were not re¯ected in signi®cantly
increased numbers of cells in suspension as indicated by
measurement of detached cell protein (Figure 1c). Indeed the
Effect of incubation with NSAIDs for 24 h on stimulated
apoptotic activity
only di€erence from control was
measurement in response to diclofenac.
a reduction in this
Since basal apoptotic activity in attached cells was low (Figure
3a), we investigated the e€ect of TNF-a, which increases
apoptosis in suspensions of guinea-pig chief cells (Fiorucci et
al., 1996). The proportion of apoptotic nuclei in cells treated for
24 h with 20 ng ml71 TNF-a was 3.1+0.22 compared with
3.4+0.68 in control cells (no e€ect by paired t-test, n=3).
However 25 mM C6-ceramide, which is produced by TNF-a
(Hannun & Luberto, 2000), and which is elevated in gastric
cells infected with Helicobacter pylori (Masamune et al., 1999),
increased the percentage of apoptotic nuclei to 9.7+1.5 from
6.7+1.0 in controls (P50.05 by paired t-test, n=4), and
elevated caspase-3-like activity to 497+147 pmol min71 mg
protein71 from 277+89 pmol min71 mg protein71 in controls
(P50.05). In the presence of 25 mM C6-ceramide no NSAID
reduced the percentage of apoptotic cells although ibuprofen
was the only one to produce a signi®cant stimulation above that
with ceramide alone (Figure 3c). All NSAIDs stimulated
caspase activity above that found with ceramide alone except
diclofenac which had no e€ect (Figure 3d). Results for cell
Effect of incubation with NSAIDs on release of enzyme
activity from cells
None of the NSAIDs a€ected the release of lactate
dehydrogenase after 4 h of incubation. Ibuprofen (250 mM)
increased lactate dehydrogenase release after 24 h and NS-
398 (100 mM) increased release of both lactate dehydrogenase
and acid phosphatase after this time interval (Figure 2).
Effect of incubation with NSAIDs for 24 h on basal
apoptotic activity
The proportion of apoptotic nuclei in cells attached to the
culture plate was increased slightly, but signi®cantly by
ibuprofen (250 mM) and decreased in response to diclofenac
(250 mM) and NS-398 (100 mM) (Figure 3a). These changes in
apoptotic activity were accompanied by changes, in the
Figure 4 E€ect of the concentration of non-steroidal anti-in¯ammatory drug (NSAID) on caspase-3-like activity in attached cells
after 24 h of exposure. Results are means+s.e.mean. The signi®cance of any e€ect of NSAID concentration on caspase activity was
assessed by analysis of variance with the following results (number of cell cultures): ibuprofen, P50.001 (5); diclofenac, P50.01 (4);
indomethacin, P40.05 not signi®cant; NS-398, P50.05 (3).
British Journal of Pharmacology vol 135 (2)

412
M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
Figure 5 E€ect of the concentration of non-steroidal anti-in¯ammatory drugs (NSAIDs) structurally related to diclofenac on
caspase-3-like activity in attached cells after 24 h of exposure. Results are means+s.e.mean, and data for diclofenac are taken from
Figure 4. The signi®cance of any e€ect of NSAID concentration on caspase activity was assessed by analysis of variance with the
following results (three cell cultures in each case): ni¯umic acid, P40.05 not signi®cant; mefenamic acid, P50.05; ¯ufenamic acid,
P50.01.
Figure 6 E€ect of the concentration of ibuprofen (®lled squares)
and diclofenac (open circles) on incorporation of [3H]-thymidine into
cellular DNA during 24 h of exposure. Results are means+s.e.mean,
for four cultures for diclofenac and three for ibuprofen. The control
[3H]-thymidine incorporation used for normalization was
1761+377 c.p.m. per mg protein. The signi®cance of any e€ect of
NSAID concentration was assessed by analysis of variance followed
Figure 7 E€ect of the concentration of diclofenac on caspase-3-like
activity in the presence (®lled bars) and absence (clear bars) of NG-
monomethyl-L-arginine (2 mM). Results are means+s.e.mean of
respectively four and ®ve separate cell cultures and have been
normalized to the activities with or without NG-monomethyl-L-
arginine which were: 181+35 and 203+56 pmol min71 mg protein71
respectively. *P50.05, **P50.01 for comparison with the appro-
priate control by ANOVAR and Dunnett's test.
by
a Dunnett's test for comparison with control: *P50.05,
**P50.01.
number, as assessed by the crystal violet assay, and expressed as
a percentage of basal conditions were (n=4): ceramide,
75+1.3; ceramide+250 mM indomethacin, 65.4+3.3; cerami-
de+250 mM diclofenac, 64+4.7; ceramide+250 mM ibuprofen,
55.8+1.6; ceramide+100 mM NS-398, 61+3.36. All results,
except that for indomethacin+ceramide were signi®cantly
di€erent (P50.05) from ceramide alone by ANOVAR and
Dunnett's test.
Effect of exposure to different concentrations of NSAIDs
for 24 h on basal caspase-3-like activity
The e€ects of ibuprofen, diclofenac and NS-398 were dose-
related, and the di€erential actions of the NSAIDs on
caspase activity were exhibited over a range of concentration
(Figure 4). Exposure of cells for 24 h to NSAIDs structurally
British Journal of Pharmacology vol 135 (2)

M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
413
related to diclofenac showed that mefenamic acid and
¯ufenamic acid also decreased caspase-3-like activity but that
ni¯umic acid did not (Figure 5).
to the assay system to give
a
®nal concentration of
diclofenac in the assay of 300 mM, then signi®cant
inhibition relative to control was found after 1, 2 and
3 h of incubation (Figure 8a). If diclofenac was added to
active recombinant human caspase-3 to give a concentra-
tion of 2.5 mM and the mixture added to the assay system
Effect of time on inhibition of basal caspase-3-like activity
by 250 mM diclofenac
to give
a concentration of 250 mM, then in both the
Cells were incubated with and without diclofenac for varying
periods of time and caspase activity after incubation with
diclofenac expressed as a percentage of that in control wells
with the following results: 15 min, 98+3% (4); 1 h, 67+3%
(4), P50.01 by paired t-test; 3 h, 60+6% P50.01 by paired
t-test; 6 h, 60+8% (4) P50.05 by paired t-test; and 24 h,
46+7% P50.01 by paired t-test.
presence and absence of NSAID activity was linearly
related to time but that with diclofenac was signi®cantly
lower (Figure 8b,c). Preincubation for 1 h with diclofenac
before assay did not increase the inhibitory e€ect (paired
t-test, Figure 8c). With diclofenac at concentrations (mM)
in the assay of 0.1, 1, 10, 100 and 250 the activity of
caspase-3 (per cent control, n=4) was : 98+6, 98+4,
96+4, 92+9 and 84+6 respectively.
Effect of ibuprofen and diclofenac on proliferative activity
of the gastric mucous cells
Discussion
Two hundred and ®fty and 500 mM diclofenac substantially
reduced incorporation of [3H]-thymidine into DNA over a
24 h period (P50.05 and 50.01 respectively by ANOVAR
and Dunnett's test), while using the same analysis procedure,
no signi®cant e€ect of 250 or 500 mM ibuprofen was detected
(Figure 6).
The main ®ndings of this investigation are ®rstly that under
basal conditions in primary cultures of guinea-pig gastric
epithelial cells there are major di€erences between the actions
of NSAIDs on apoptotic activity in contrast with the uniform
stimulatory e€ects seen in transformed cell lines (Lu et al.,
1995; Chan et al., 1998; Simmons et al., 1999; Zhu et al.,
1999), and secondly that one NSAID, diclofenac, inhibits
basal apoptosis with an e€ect on caspase-3 possibly
contributing to this action.
Lack of involvement of nitrogen and oxygen radicals in
the inhibitory action of diclofenac
Nitric oxide (NO) inhibits apoptosis in guinea-pig gastric
mucous cells (Potter & Hanson, 2000). However, inclusion of
the non-selective inhibitor of NO synthase, NG-monomethyl-L-
arginine, did not prevent the inhibitory e€ect of diclofenac on
caspase-3-like activity (Figure 7). Guinea-pig gastric mucous
cells produce superoxide in substantial amounts relative to
activated neutrophils (Teshima et al., 1998). Reduction of
exogenous cytochrome c by the cultured cells was decreased by
87+4% (6) by added superoxide dismutase, and superoxide
production was therefore taken as cytochrome c reduction
inhibitable by superoxide dismutase. Values obtained were
(nmol h71 mg protein71, n=5): control 25+7, 250 mM
diclofenac, 27+6; 10 mM diphenyleneiodonium (inhibitor
targeting the ¯avoprotein component of NADPH oxidase),
5.9+1.9, and 250 mM diclofenac plus 10 mM diphenyleneiodo-
nium, 4.8+1.7. Thus diclofenac did not inhibit superoxide
production, which, however, was reduced by diphenyleneiodo-
nium (P50.05 by ANOVAR and Dunnett's test).
In animal models intragastric administration of NSAIDs
causes release of TNF-a probably from macrophages
(Fiorucci et al., 1999). E€ects of TNF-a on the primary
culture system were therefore investigated. No change in
apoptotic activity was detected. This result contrasts with
that obtained using a similar concentration of TNF-a in
freshly isolated guinea-pig chief cells. It is possible that the
guinea-pig gastric epithelial cells had lost receptors for TNF-
a during culture, but it should be remembered that many
`normal' cells e.g. hepatocytes do not show loss of viability to
TNF-a (Leist et al., 1994; Wallach, 1997), and that the
trauma of isolation, or the lack of attachment to
a
substratum, could have sensitized the chief cells. However
C6-ceramide, which is produced in cells stimulated with TNF-
a (Hannun & Luberto, 2000), and which is elevated in gastric
cells incubated with Helicobacter pylori (Masamune et al.,
1999), elevated apoptotic activity and was used to test for
e€ects of NSAIDs in cells in which apoptosis was stimulated
above basal.
The NSAIDs could have a€ected the number of cells
attached to the culture plate by causing apoptosis, necrosis,
inducing detachment, or promoting proliferation. The
relationship between numbers of cells on the culture plate
and those in suspension is likely to be complex. Thus a
reduction in cells on the plate over a period of 24 h because
of apoptosis or necrosis will not necessarily be re¯ected in an
increased number of cells in suspension as the dead cells may
have been lost from the system. In two cases under basal
conditions the number of cells on the plate was reduced. With
ibuprofen the primary cause of cell loss was probably
increased apoptotic activity as evidenced by an enhanced
caspase-3-like activity and an increased percentage of
attached cells showing condensed and fragmented nuclei.
There was no evidence for ibuprofen causing major
Caspase-3-like activity (pmol min71mg protein71) in cells
incubated for 3 h with 10 mM diphenyleneiodonium was
245+39, and with superoxide dismutase (100 u ml71) and
catalase (100 u ml71) was 303+91, (P50.001 after ANO-
VAR and Newman Keuls test versus control caspase-3-like
activity of 135+41 (all n=4)). However, the inhibitory e€ect
of 250 mM diclofenac was not signi®cantly altered being
respectively: 66+5.6%, 67+9.7% and 58+7.7% under
control conditions, with diphenyleneiodonium, and with
catalase and superoxide dismutase (all n=4).
Effect of diclofenac on caspase-3-like activity in
homogenates, and on recombinant caspase-3
If homogenates prepared from control cells were incubated
with diclofenac at 2.5 mM and then the homogenate added
British Journal of Pharmacology vol 135 (2)

414
M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
Figure 8 E€ect of diclofenac (®lled bars and symbols) on caspase-3-like activity in cell homogenates (a) and on recombinant
caspase-3 (b,c). In (a) results are means+s.e. for ®ve separate homogenates, with the concentration of diclofenac at 2.5 mM during
preincubation at 208C and 300 mM during the assay. Control homogenate activity use for normalization was 167+31 pmol
min71 mg protein71. In (b,c) the concentration of diclofenac was 2.5 mM during preincubation and 250 mM in the assay and the
results are means+s.e.mean for ®ve repetitions in the absence (clear bars/open symbols) or presence of diclofenac (®lled bars/
symbols) with 50 ng of enzyme in each assay.
alterations in proliferative activity or primary cell detach-
ment. The time course of lactate dehydrogenase release,
elevated after 24 h, but not after 4 h, is most easily
interpreted as indicating `secondary-necrosis' i.e. enzyme
release following on from primary apoptotic cell death.
Ibuprofen, and the structurally similar ¯urbiprofen (Johal &
Hanson, 2000) are the only NSAIDs in this primary culture
which showed enhanced apoptosis under basal conditions in
a manner similar to previous results with cell lines (Lu et al.,
1995; Chan et al., 1998; Simmons et al., 1999; Kusuhara et
al., 1999; Zhu et al., 1999). Ibuprofen also further enhanced
apoptotic activity above that obtained with 25 mM C6-
ceramide.
The other NSAID, which caused cell loss under basal
conditions, was the cyclo-oxygenase-2 selective inhibitor NS-
398. However, in this instance both indices of apoptotic
activity were reduced by comparison with control, while
release of both acid phosphatase and lactate dehydrogenase
were elevated after 24 h compared with control. This pro®le
would imply some form of slow necrotic cell death. It may be
relevant that in L929 ®brosarcoma cells inhibition of caspase
activity promoted necrotic cell death in response to TNF-a
(Vercammen et al., 1998). In colorectal adenoma and
carcinoma cell lines incubated with a similar range of
concentrations of NS-398, no increase in apoptosis of
attached cells was found, but by contrast with the present
British Journal of Pharmacology vol 135 (2)

M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
415
results NS-398 stimulated detachment and consequently
apoptosis (Elder et al., 2000). Interpretation of the action
of NS-398 is further complicated by the stimulatory e€ect of
NS-398 on apoptotic activity in the gastric cells in the
presence of C6-ceramide. Cyclo-oxygenase-2 selective inhibi-
tors, including NS-398, show reduced gastric toxicity relative
to non-selective NSAIDs (Futaki et al., 1993; Whittle, 2000).
It is unclear what relevance the present data with NS-398
have to the role of cyclo-oxygenase-2 selective inhibitors as
antiin¯ammatory agents. Thus the concentration of NS-398
in the stomach after a dose appropriate to inhibit cyclo-
oxygenase-2 is unknown. It is also uncertain whether NS-398,
which has not been developed for use in humans, is
representative of all cyclo-oxygenase-2 selective inhibitors,
but investigation of this question is beyond the scope of the
present investigation.
Under basal conditions, indomethacin did not show any
major e€ects on cell number, indices of apoptosis or cell
detachment of the guinea-pig gastric epithelial cells which is in
marked contrast to the stimulation of apoptosis found in AGS
gastric cancer cells (Zhu et al., 1999) and in colon cancer cells
(Chan et al., 1998). However, it did stimulate caspase-3-like
activity above that obtained with C6-ceramide alone.
Diclofenac inhibited basal apoptotic activity, and by
contrast with NS-398, this inhibition was not associated with
any increase in enzyme release above control. Diclofenac also
inhibited caspase-3-like activity when this was enhanced by
detachment of cells under basal conditions, or in attached
cells by inhibition of superoxide production. However, when
caspase and apoptotic activity was stimulated by C6-ceramide
the inhibitory e€ect of diclofenac was lost. Although
apoptosis was decreased under basal conditions there was
no increase in cell number, which can be explained by the
inhibition of proliferative activity by diclofenac. The smaller
loss of cells into the medium (Figure 1c) may re¯ect the
reduced cell `turnover' in the preparation. Inhibition of
proliferation by NSAIDs of colonic cell lines in well-
established (Rigas & Shi€, 1999). Because the result for
diclofenac exhibited the greatest contrast with data for cells
lines, where stimulation of apoptosis is found (e.g. Simmons
et al., 1999, J774.2 macrophages; Kusuhara et al., 1999, AGS
cells), the possible mechanism of the inhibition of caspase-3-
like activity was investigated further. Previous work has
indicated that changes in caspase-3-like activity correlate well
with the proportion of apoptotic cells attached to the culture
plate (Johal & Hanson, 2000; Potter & Hanson, 2000).
The structure-activity relationship was probed by using
fenemates. Mefenamic acid and ¯ufenamic acid have respec-
tively 2,3-dimethylphenyl amino and 3-tri¯uorophenyl amino
substituents somewhat di€erent from the equivalent region of
diclofenac, yet were still active. By contrast, changing the ring
with the carboxyl group from benzoic acid, to nicotinic acid as
in ni¯umic acid, caused a loss of inhibitory activity. All four
compounds are inhibitors of cyclo-oxygenases-1 and -2
(Warner et al., 1999) so inhibition of either form of cyclo-
oxygenase is unlikely to be involved in the inhibitory action of
diclofenac. Flufenamic acid is an activator of peroxysome
proliferator-activated receptor g (PPARg), but this receptor is
unlikely to be the target in the present work because
indomethacin, which was essentially inactive, and ibuprofen,
which stimulates apoptosis, also activate this receptor
(Lehmann et al., 1997).
An interesting feature of guinea-pig gastric epithelial cells,
possibly distinctive relative to gastrointestinal cell lines, is
their production of superoxide (Teshima et al., 1998), which
has been reported to inhibit apoptosis in these cells
(Rokutan et al., 1999). However, diclofenac did not a€ect
superoxide production, and although inhibition of super-
oxide production, or its removal by superoxide dismutase
and catalase, over a 3 h period increased caspase-3-like
activity the inhibitory e€ect of diclofenac was unaltered.
Thus, superoxide production was neither a€ected by
diclofenac nor required for its inhibitory action. Similarly,
although NO inhibits apoptosis in guinea-pig gastric
epithelial cells (Johal & Hanson, 2000; Potter & Hanson,
2000), inhibition of NO synthase did not prevent the action
of diclofenac.
Johal & Hanson (2000) found that the stimulatory e€ect of
¯urbiprofen on caspase-3-like activity required an incubation
period longer than 6 h to become apparent, but the inhibitory
e€ect of diclofenac was apparent within 1 h. This suggested that
a direct interaction of diclofenac, or a metabolite (Tang et al.,
1999), on caspase activity might be possible. Indeed incubation
of tissue homogenates, or recombinant human caspase-3, with
diclofenac produced an inhibition of enzyme activity, which
remained constant during the assay period. Whether an
adequate concentration of diclofenac in the vicinity of
caspase-3 was achieved within intact cells, or whether the action
of diclofenac is somehow enhanced in the cellular environment is
unclear. The loss of the inhibition of caspase-3-like activity in
cells incubated with C6-ceramide suggests that incubation with
ceramide may modify some cellular component required for the
inhibitory e€ect of diclofenac.
The stimulation of basal apoptosis with ibuprofen and
inhibition with diclofenac were obtained with concentrations
of 100 mM and above. Plasma concentrations in humans, on a
moderate dosage of ibuprofen and diclofenac, can reach
200 mM (Davies, 1998) and 10 mM (Davies & Anderson, 1997)
respectively. The results with diclofenac may therefore be
more relevant to potential topical e€ects of NSAIDs, but
estimation of the appropriate concentration is complicated by
the likelihood of binding of NSAIDs to protein (Lin et al.,
1987), and the possible accumulation of NSAIDs with
carboxyl groups in surface epithelial cells, in the presence
of acid in vivo, by `ion trapping' (Hawkey, 2000). In
conclusion, the potential of NSAIDs to contribute to
gastropathy by increasing apoptosis through a direct e€ect
on surface epithelial cells may not be uniform.
The growth inhibitory and pro-apoptotic activities of
NSAIDs on cancer cells, which are all found within the
concentration ranges used in this investigation, are of consider-
able interest because of the linkage between usage of NSAIDs
and a reduction in the risk of colon and stomach cancer (Boland
et al., 2000). The target for the pro-apoptotic activity of NSAIDs
remains to be clearly de®ned (Boland et al., 2000). The lack of a
comparable uniform e€ect of NSAIDs on apoptosis in the
present work raises questions as to whether the same target is
present in primary cultures of gastric mucous epithelial cells, or
alternatively implies that some NSAIDs may have multiple sites
of action on the apoptotic process.
Klairi Kavvada cultured the cells for the experiments with TNF-a.
British Journal of Pharmacology vol 135 (2)

416
M. Ashton & P.J. Hanson
NSAIDs and apoptosis in gastric cells
References
BOLAND, C.R., SINICROPE, F.A., BRENNER, D.E. & CARETHERS,
J.M. (2000). Colorectal cancer prevention and treatment.
Gastroenterology, 118, S115 ± S128.
BYRNE, C.R. & HANSON, P.J. (1998). Induction of heat shock protein
72 by a nitric oxide donor in guinea-pig gastric mucosal cells. Eur.
J. Pharmacol., 353, 117 ± 122.
CHAN, T.A., MORIN, P.J., VOGELSTEIN, B. & KINZLER, K.W. (1998).
Mechanisms underlying nonsteroidal antiin¯ammatory drug-
mediated apoptosis. Proc. Natl. Acad. Sci. U.S.A., 95, 681 ± 686.
DAVIES, N.M. (1998). Clinical pharmacokinetics of ibuprofen ± The
®rst 30 years. Clin. Pharmacokinet., 34, 101 ± 154.
DAVIES, N.M. & ANDERSON, K.E. (1997). Clinical pharmacokinetics
of diclofenac. Therapeutic insights and pitfalls. Clin. Pharmaco-
kinet., 33, 184 ± 213.
ELDER, D.J.E., HALTON, D.E., CREW, T.E. & PARASKEVA, C. (2000).
Apoptosis induction and cyclooxygenase-2 regulation in human
colorectal adenoma and carcinoma cell lines by the cycloox-
ygenase-2-selective non-steroidal anti-in¯ammatory drug NS-
398. Int. J. Cancer, 86, 553 ± 560.
FIORUCCI, S., ANTONELLI, E., SANTUCCI, L., MORELLI, O.,
MIGLIETTI, M., FEDERICI, B., MANNUCCI, R., DEL SOLDATO,
P. & MORELLI, A. (1999). Gastrointestinal safety of nitric oxide-
derived aspirin is related to inhibition of ICE-like cysteine
proteases in rats. Gastroenterology, 116, 1089 ± 1106.
POTTER, C.L. & HANSON, P.J. (2000). Exogenous nitric oxide inhibits
apoptosis in guinea pig gastric mucous cells. Gut, 46, 156 ± 163.
REDINBAUGH, M.G. & TURLEY, R.B. (1986). Adaptation of the
bicinchoninic acid protein assay for use with microtitre plates
and sucrose gradient fractions. Anal. Biochem., 153, 267 ± 271.
RIGAS, B. & SHIFF, S.J. (1999). Non-steroidal anti-in¯ammatory
drugs (NSAIDs), cyclooxygenases, and the cell cycle. Adv. Exp.
Med. Biol., 470, 119 ± 126.
ROKUTAN, K., TESHIMA, S., KAWAI, T. & NIKAWA, T. (1999).
Superoxide anion produced by gastric pit cells plays a crucial role
in autocrine/paracrine regulation of their apoptosis. Gastro-
enterology, 116, A640.
SANTUCCI, L., FIORUCCI, S., DI MATTEO, F.M. & MORELLI, A.
(1995). Role of tumour necrosis factor a release and leukocyte
margination in indomethacin-induced gastric injury in rats.
Gastroenterology, 108, 393 ± 401.
SIMMONS, D.L., BOTTING, R.M., ROBERTSON, P.M., MADSEN, M.L.
VANE, J.R. (1999). Induction of acetaminophen-sensitive
&
cyclooxygenase with reduced sensitivity to nonsteroid antiin-
¯ammatory drugs. Proc. Natl. Acad. Sci. U.S.A., 96, 3275 ± 3280.
SONG, Z. & STELLER, H. (2000). Death by design: mechanism and
control of apoptosis. Trends Biochem. Sci., 24, M49 ± M52.
TAHA, A.S., NAKSHABENDI, I., LEE, F.D., STURROCK, R.D.
&
RUSSELL, R.I. (1992). Chemical gastritis and Helicobacter pylori
related gastritis in patients receiving non-steroidal anti-in¯am-
matory drugs: Comparison and correlation with peptic ulcera-
tion. J. Clin. Pathol., 45, 135 ± 139.
FIORUCCI, S., SANTUCCI, L., MIGLIORATI, G., RICCARDI, C.,
AMOROSI, A., MANCINI, A., ROBERTI, R.
& MORELLI, A.
(1996). Isolated guinea-pig gastric chief cells express tumour
necrosis factor receptors coupled with the sphingomyelin path-
way. Gut, 38, 182 ± 189.
TANG, W., STEARNS, R.A., WANG, R.W., CHIU, S.-H.L. & BAILLIE,
T.A. (1999). Roles of human hepatic cytochrome P450s 2C9 and
3A4 in the metabolic activation of diclofenac. Chem. Res.
Toxicol., 12, 192 ± 199.
TESHIMA, S., ROKUTAN, K., NIKAWA, T. & KISHI, K. (1998). Guinea
pig gastric mucosal cells produce abundant superoxide anion
through an NADPH oxidase-like system. Gastroenterology, 115,
1186 ± 1196.
TSUTSUMI, S., ROKUTAN, K., TSUCHIYA, T. & MIZUSHIMA, T.
(2000). Spontaneous apoptotic DNA fragmentation in cultured
guinea pig gastric mucosal cells. Dig. Dis. Sci., 45, 291 ± 297.
VERCAMMEN, D., BEYAERT, R., DENEKER, G., GOOSSENS, V., VAN
LOO, G., DECLERCQ, W., GROOTEN, J., FIERS, W. & VANDENA-
BEELE, P. (1998). Inhibition of caspases increases the sensitivity
of L929 cells to necrosis mediated by tumour necrosis factor. J.
Exp. Med., 187, 1477 ± 1485.
WALLACE, J.L. (1997). Nonsteroidal anti-in¯ammatory drugs and
gastroenteropathy: The second hundred years. Gastroenterology,
112, 1000 ± 1016.
WALLACH, D. (1997). Cell death induction by TNF: a matter of self
control. Trends Biochem. Sci., 22, 107 ± 109.
WARNER, T.D., GIULIANO, F., VOJNOVIC, I., BUKASA, A., MITCH-
ELL, J.A. & VANE, J.R. (1999). Nonsteroid drug selectivities for
cyclo-oxygenase-1 rather than cyclo-oxygenase-2 are associated
with human gastrointestinal toxicity: a full in vitro analysis. Proc.
Natl. Acad. Sci. U.S.A., 96, 7563 ± 7568.
FUTAKI, N., YOSHIKAWA, K., HAMASAKA, Y., ARAI, I., HIGUCHI,
S., IISUKA, H. & OTOMO, S. (1993). NS-398, a novel non-steroidal
anti-in¯ammatory drug with potent analgesic and antipyretic
e€ects, which causes minimal stomach lesions. Gen. Pharmacol.,
24, 105 ± 110.
HALL, P.A., COATES, P.J., ANSARI, B. & HOPWOOD, D. (1994).
Regulation of cell number in the mammalian gastrointestinal
tract: the importance of apoptosis. J. Cell. Sci., 107, 3569 ± 3577.
HANNUN, Y.A. & LUBERTO, C. (2000). Ceramide in the eukaryotic
stress response. Trends Cell Biol., 10, 73 ± 80.
HAWKEY, C.J. (2000). Nonsteroidal anti-in¯ammatory drug gastro-
pathy. Gastroenterology, 119, 521 ± 535.
JOHAL, K. & HANSON, P.J. (2000). Opposite e€ects of ¯urbiprofen
and the nitroxybutyl ester of ¯urbiprofen on apoptosis in
cultured guinea-pig gastric mucous cells. Br. J. Pharmacol.,
130, 811 ± 818.
KUSUHARA, H., KOMATSU, H., SUMICHIKA, H. & SUGAHARA, K.
(1999). Reactive oxygen species are involved in the apoptosis
induced by non-steroidal antiin¯ammatory drugs in cultured
gastric cells. Eur. J. Pharmacol., 383, 331 ± 337.
LEHMANN, J.M., LENHARD, J.M., OLIVER, B.B., RINGOLD, G.M. &
KLIEWER, S.A. (1997). Peroxisome proliferator-activated recep-
tors a and g are activated by indomethacin and other non-
steroidal anti-in¯ammatory drugs. J. Biol Chem., 272, 3406 ±
3410.
LEIST, M., GANTNER, F., BOHLINGER, I., GERMANN, P.G., TIEGS,
G. & WENDEL, A. (1994). Murine hepatocyte apoptosis induced
in vitro and in vivo by TNF-a requires transcriptional arrest. J.
Immunol., 153, 1778 ± 1788.
WHITTLE, B.J.R. (2000). COX-1 and COX-2 products in the gut:
therapeutic impact of COX-2 inhibitors. Gut, 47, 320 ± 325.
WONG, H.M.
& TEPPERMAN, B.L. (1994). Reduced glutathione
modulates Ca2+- mediated damage to rabbit isolated gastric
mucosal cells. Am. J. Physiol., 267, G1 ± G9.
LIN, J.H., COCCHETTO, D.M.
&
DUGGAN, D.E. (1987). Protein
ZHU, G.H., WONG, B.C.Y., EGGO, M.C., CHING, C.K., YUEN, S.T.,
CHAN, E.Y.T., LAI, K.C. & LAM, S.K. (1999). Non-steroidal anti-
in¯ammatory drug-induced apoptosis in gastric cancer cells is
blocked by protein kinase C activation through inhibition of c-
myc. Brit. J. Cancer, 79, 393 ± 400.
ZHU, G.H., YANG, X.L., LAI, K.C., CHING, C.K., WONG, B.C.Y.,
YUEN, S.T., HO, J. & LAM, S.K. (1998). Nonsteroidal antiin-
¯ammatory drugs could reverse Helicobacter pylori-induced
apoptosis and proliferation in gastric epithelial cells. Dig. Dis.
Sci., 43, 1957 ± 1963.
binding as a primary determinant of the clinical pharmacokinetic
properties of non-steroidal anti-in¯ammatory drugs. Clin.
Pharmacokin., 12, 402 ± 432.
LU, X., XIE, W., REED, D., BRADSHAW, W.S. & SIMMONS, D.L.
(1995). Non-steroidal antiin¯ammatory drugs cause apoptosis
and induce cyclooxygenases in chicken embryo ®broblasts. Proc.
Natl. Acad. Sci. U.S.A., 92, 7961 ± 7965.
MASAMUNE, A., SHIMOSEGAWA, T., MASAMUNE, O., MUKAIDA,
N., KOIZUMI, M. & TOYOTA, T. (1999). Helicobacter pylori-
dependent ceramide production may mediate increased inter-
leukin 8 expression in human gastric cancer cell lines. Gastro-
enterology, 116, 1330 ± 1341.
(Received February 10, 2001
Revised October 23, 2001
Accepted November 12, 2001)
British Journal of Pharmacology vol 135 (2)