Nonsteroidal anti-inflammatory drugs (NSAIDs) such as sulindac and indomethacin are a major cause of gastric erosions and ulcers. Induction of apoptosis by NSAIDs is an important mechanism involved. Understanding how NSAIDs affect genes that regulate apoptosis is useful for designing therapeutic or preventive strategies and for evaluating the efficacy of safer drugs being developed. We investigated whether growth arrest and DNA damage-inducible 45α (GADD45α), a stress signal response gene involved in regulation of DNA repair and induction of apoptosis, plays a part in NSAID-induced gastric mucosal injury and apoptosis in vivo in mice and in vitro in cultured human AGS and rat RGM-1 gastric epithelial cells. Intraperitoneal administration of sulindac and indomethacin both resulted in up-regulation of GADD45α expression and induction of significant injury and apoptosis in gastric mucosa of wild-type mice. GADD45α(−/−) mice were markedly more resistant to both sulindac- and indomethacin-induced gastric mucosal injury and apoptosis than wild-type mice. Sulindac sulfide and indomethacin treatments also concentration-dependently increased GADD45α expression and apoptosis in AGS and RGM-1 cells. Antisense suppression of GADD45α expression significantly reduced sulindac and indomethacin-induced activation of caspase-9 and apoptosis in AGS cells. Pretreatments with exogenous prostaglandins and small interfering RNA suppression of cyclooxygenase (COX)-1 and -2 did not affect up-regulation of GADD45α by sulindac sulfide and indomethacin in AGS cells. These findings indicate that GADD45α up-regulation is a COX-independent mechanism that is required for induction of severe gastric mucosal apoptosis and injury by NSAIDs, probably via a capase-9-dependent pathway of programmed cell death.
NSAIDs such as indomethacin, sulindac, and others are commonly used drugs worldwide. However, they frequently cause gastric erosions and ulcers. NSAIDs can cause gastric injury through a variety of mechanisms, many of which culminate in induction of apoptosis. A well known mechanism is inhibition of prostaglandin synthesis by inhibiting cyclooxygenase (COX)-1 and COX-2 enzymes. Inhibition of both COX-1 and COX-2 seems crucial for effective induction of gastric injury, because nonselective NSAIDs (i.e., indomethacin, sulindac) cause severe gastric erosions, whereas COX-2-selective NSAIDs (e.g., celecoxib) produce significantly less injury (Wallace et al., 2000; Tanaka et al., 2001). Although selective NSAIDs are less damaging to gastric mucosa, their adverse cardiovascular effects have prompted their recent withdrawal from the market.
Several strategies are currently being developed to overcome gastrointestinal toxicities associated with NSAIDs while minimizing the adverse cardiovascular effects. One strategy involves attachment of small chemical components to nonselective NSAIDs such that release of these moieties upon administration of the drug would produce gastroprotective effects [e.g., nitric oxide-releasing NSAIDs (Singh et al., 2009) and hydrogen sulfide-donating NSAIDs (Fiorucci et al., 2006)]. Another strategy being considered is selective inhibition of microsomal prostaglandin E2 synthase-1 (mPGEs-1)-derived prostaglandin E2 (PGE2) formation as an alternative to general inhibition of physiologically relevant prostanoid synthesis by NSAIDs (Koeberle and Werz, 2009). Excessive PGE2 formation from COX-2 and mPGEs-1 activities has been linked to inflammation, pain, fever, atherosclerosis, and tumorigenesis. Inhibition of mPGEs-1 is expected to reduce inflammation, fever, and pain while allowing continued biosynthesis of gastroprotective prostanoids.
Mounting evidence indicates that NSAIDs also exert COX/prostaglandin-independent effects, such as inhibition of epidermal growth factor signaling, regulation of cell cycle proteins such as p21 and cyclins, inhibition of the mitogen-activated protein-kinase pathway (Ishikawa et al., 1998; Tegeder et al., 2001), and increased ubiquitination of proteins such as hypoxia-inducible factor-1α and β-catenin in T cells for degradation by the ubiquitin-proteasome (Jones et al., 2002; Dihlmann et al., 2003). NSAIDs also directly activate proapoptotic factors such as caspases and Bax (Zhou et al., 2001), and they down-regulate levels of survivin, an antiapoptosis protein, both in vivo in human and rat gastric mucosa and in vitro in rat gastric epithelial RGM-1 cells (Chiou et al., 2005). Our recent studies showed that exogenous PGE2 treatments do not alter survivin levels in the rat gastric mucosa or gastric epithelial cells, suggesting that regulation of survivin by NSAIDs occurs through COX/prostaglandin-independent pathways (Chiou et al., 2005). Further understanding of how nonselective NSAIDs affect genes involved in the regulation of apoptosis remains crucial for designing new therapeutic or preventive strategies against induction of gastric mucosal injury and is also useful for evaluating the efficacy of new drugs being developed.
GADD45α is a stress signal response gene involved in regulation of DNA repair and apoptosis (Hollander et al., 1999; Hildesheim et al., 2002). GADD45α-null mice exhibit severe genomic instability characterized by aneuploidy, centrosome amplification, aberrant mitosis, and cytokinesis (Hollander et al., 1999). More strikingly, mice lacking GADD45α are susceptible to carcinogenesis induced by ionizing radiation, UV irradiation, and dimethylbenzanthracene (Hollander et al., 2001). In addition, GADD45α is important in cellular response to a variety of stress stimuli and genotoxic agents that are known to induce apoptosis, such as hypoxia, growth factor withdrawal, ionizing radiation, hydroxyurea, and methanesulfonate (Fornace et al., 1992). UV radiation-induced apoptosis is deficient in GADD45α-null mice (Hollander et al., 1999). Rebamipide, a gastroprotective agent that inhibits indomethacin-induced apoptosis, also down-regulates GADD45α expression (Naito et al., 2005), suggesting that NSAIDs may activate GADD45α expression as a mechanism of gastric cell apoptosis and gastric damage. The role of GADD45α in NSAID-induced gastric mucosal injury and apoptosis has not been explored. In this study, we investigated whether 1) GADD45α expression is up-regulated by sulindac and indomethacin treatments both in vitro in human gastric AGS and normal rat gastric RGM-1cells and in vivo in mouse gastric mucosa; 2) antisense suppression of GADD45α expression in AGS cells confers resistance to NSAIDs-induced apoptosis and whether GADD45α-null mice are resistant to NSAID-induced gastric injury; 3) exogenous PGE2 treatment and siRNA suppression of COX-1 and -2 could inhibit up-regulation of GADD45α by sulindac and indomethacin in AGS cells; and 4) NSAIDs induce growth inhibition and caspase-9 activation in gastric epithelial cells, and whether they are associated with GADD45α up-regulation.
Materials and Methods
Animals, Treatments, and Tissues.
These studies were approved by the Subcommittee for Animal Studies of the Long Beach Department of Veterans Affairs Medical Center. GADD45α(−/−) B6129F1 mice (Hollander et al., 1999; Mouse Models of Human Cancers Consortium Repository, Frederick MD), their wild-type (WT) age-matched siblings, and WT age-matched C57BL6 mice (The Jackson Laboratory, Bar Harbor ME), weighing 20–24 g) were fasted for 16 h. Groups of 12 mice each were then administered the intraperitoneal injections of the following: 1) vehicle (0.2 M Na2CO3 + NaH2PO4 in 4:1 ratio), 2) 15 mg/kg indomethacin (Sigma-Aldrich, St. Louis, MO), or 3) 30 mg/kg sulindac (Sigma-Aldrich). The indomethacin concentration used has been used previously to effectively induce experimental gastric mucosal injury in rodents within a few hours. The sulindac concentration was chosen because the daily recommended dose in humans for sulindac is double that of indomethacin. Four, 8, and 24 h after administration of drugs, the mice were euthanized with 200 mg/kg pentobarbital, and laparotomy was immediately performed to excise the stomachs. Gastric samples for biochemical studies were prepared by scraping the mucosa and freezing it immediately in liquid nitrogen. Gastric samples for immunohistochemical studies were frozen and sectioned.
Quantitation of Macroscopic Injury in Gastric Mucosa.
The stomachs were opened along the greater curvature and photographed using a Canon Powershot A70 digital camera (Canon USA Inc., Lake Success, NY). The area of visible mucosal erosions was measured using Photoshop CS3 (Adobe Systems, Mountain View, CA). Mucosal damage was expressed as percentage of damage of the entire gastric mucosa and calculated as follows: (total lesion area/total area of the gastric mucosa) × 100.
Analysis of Apoptosis in the Gastric Mucosa (Caspase-3 Activity Assay).
Frozen gastric mucosal sections from 4 h post-NSAIDs administration were incubated with anti-caspase-3 or anti-active caspase-3 rabbit polyclonal antibodies (BioVision, Mountain View, CA), followed by Texas Red-conjugated goat anti-rabbit secondary antibodies (Vector Laboratories, Burlingame, CA). Staining was visualized using a Optiphot epifluorescence microscope (Nikon, Melville, NY) with Omega filter fluorescein isothiocyanate/Texas Red. Percentage of apoptotic cells within and around the periphery of each gastric mucosal erosion was determined as follows: number of active caspase-3-positive (red) cells/500 gastric mucosal cells in and around erosion cross sections (blue Hoechst counterstained) × 100.
Sections from five different gastric mucosal erosions were analyzed per treatment condition. Five randomly selected intact gastric mucosal areas were similarly analyzed per treatment condition.
Quantitation of fold gastric mucosal caspase-3 activity was performed using gastric mucosal scrapings and the colorimetric CasPASE apoptosis assay kit (Geno Technologies, Inc, Maryland Heights, MO), following manufacturer's instructions and as described previously (Chiou et al., 2005). Gastric mucosal scrapings from six mice were used per condition and treatment.
Cell Lines and Treatments.
Human gastric mucosal AGS cells were cultured in RPMI 1640 medium (Sigma-Aldrich) supplemented with 10% FBS and 1% antibiotics. Normal rat gastric epithelial RGM-1 cells were cultured in Dulbecco's modified Eagle's medium:F-12 medium (Sigma-Aldrich) supplemented with 20% FBS. For quantification of GADD45α mRNA expression, cells were grown in six-well culture plates until they were approximately 70% confluent, incubated in serum-free medium overnight, and then treated with vehicle only (dimethyl sulfoxide); 0.2, 0.1, or 0.05 mM indomethacin; or 0.1, 0.05, or 0.01 mM sulindac sulfide for 3 h. Then, total RNA was purified for real-time reverse transcription (RT)-PCR analysis. We utilized the sulindac metabolite because sulindac is a prodrug that is insoluble in culture medium.
Sulindac sulfide concentrations used were lower than those of indomethacin because this active sulindac metabolite was very potent in inducing apoptosis. Once cells were dead, they were no longer useful for subsequent biochemical assays.
For determination of GADD45α protein up-regulation, cells were grown and serum-starved as described above and then treated with vehicle only, 0.2 mM indomethacin, or 0.1 mM sulindac sulfide for 6 h. Then, total protein was extracted for Western blot analysis.
To examine effects of exogenous prostaglandins on GADD45α up-regulation by NSAIDs, cells were treated with 1 or 10 μM PGE2 (Cayman Chemical, Ann Arbor, MI) for 6 h or pretreated with the same concentrations of PGE2 for 30 min and then treated with 0.2 mM indomethacin or 0.1 mM sulindac sulfide. Vehicle only, 0.2 mM indomethacin, and 0.1 mM sulindac sulfide treatments were used as controls. At 6 h after NSAID treatment, before onset of significant apparent cellular damage, cells were collected for Western blot analysis.
To examine the effects of siRNA suppression of COX-1 and COX-2 on GADD45α up-regulation by NSAIDs, 10 nM siRNA targeting COX-1 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), COX-2 (Santa Cruz Biotechnology, Inc.), or both were transfected into AGS cells at 50% confluence using RNAiMax transfection reagent (Invitrogen, Carlsbad CA), following the manufacturer's instructions. Mock transfection using only the transfection reagent but no siRNA, and transfection using a control RNA (Santa Cruz Biotechnology, Inc.) that does not silence mammalian mRNAs were included as controls. The control RNA is conjugated to a green fluorescent dye for monitoring transfection efficiency with fluorescence microscopy (percentage of fluorescent cells/500 total cells counted). Forty-eight hours after transfection, cells were treated with vehicle only, sulindac sulfide (0.1 mM), and indomethacin (0.2 mM) for 4 h. Then, total RNA was isolated for RT-PCR analysis of COX-1, COX-2, GADD45α, and β-actin expressions, using gene-specific primers following the manufacturer's instructions (Santa Cruz Biotechnology, Inc.).
Total RNA was isolated using the RNeasy mini kit (QIAGEN, Valencia, CA), according to the manufacturer's instructions. We used 0.3 μg of total RNA from each sample in a random primed RT reaction with the Moloney murine leukemia virus reverse transcriptase (Applied Biosystems, Foster City, CA) to synthesize first-strand cDNA. Quantitation of mRNA expression (first-strand cDNA) was performed by real-time qPCR using custom-made, gene-specific primers (Invitrogen). The human GADD45α primer sequences were described previously (Chiou and Hoa, 2009). Survivin primers also were described previously (Chiou et al., 2003). Real-time qPCR data were analyzed using a method described by Pfaffl (2001). This method accounts for variations in amplification efficiency in the following formula: ratio = (Etarget)δCt target (control-treated)/(Eref)δCt ref (control-treated), where target refers to GADD45α or survivin, and β-actin is used as reference to which all GADD45α and survivin PCR products were normalized. E refers the amplification efficiency based on the slope of the standard dilution curve. Real-time qPCR analysis was performed on the iCycler (Bio-Rad Laboratories, Hercules, CA) using IQ SYBR Green Supermix (Bio-Rad Laboratories). Each PCR was performed in triplicate, and results are presented as average fold expression ± S.D.
Western Blot Analysis.
Cells were lysed with protein lysis buffer [20 mM Tris-HCl, pH 7.9, 1.5 mM MgCl2, 550 mM NaCl, 0.2 mM EDTA, 2 mM dithiothreitol, and 20% (v/v) glycerol]. We separated 100 μg of protein per sample by 10% SDS-polyacrylamide gel electrophoresis and transferred it onto nitrocellulose membranes. The membranes were blocked in skim milk and incubated with rabbit-polyclonal anti-GADD45α, rabbit polyclonal anti-survivin, rabbit polyclonal anti-procaspase-9, and rabbit polyclonal active caspase-9 p10 antibodies (Santa Cruz Biotechnology, Inc.) at 4°C overnight. The membrane was then washed and incubated for 1 h with peroxidase-conjugated goat anti-rabbit secondary antibody (Sigma-Aldrich) for 1 h. The protein signals were visualized using ECL chemoluminescence reagent (GE Healthcare, Little Chalfont, Buckinghamshire, UK) and by exposure to X-Omat film (Eastman Kodak, Pittsburgh, PA). The membranes were stripped and reincubated with monoclonal anti-β-actin antibodies (Sigma-Aldrich) and then with peroxidase-conjugated anti-mouse antibodies (BD Transduction Laboratories, Lexington, KY). Protein signal densities were quantified using a MetaMorph Imaging System, version 3.0 (Molecular Devices, Sunnyvale, CA). Protein signal densities were subtracted from background densities and normalized to the corresponding β-actin signal densities.
Antisense Suppression of GADD45α.
Morpholino antisense oligonucleotides were designed to block expression of human GADD45α mRNA (Gene Tools, Philomath, OR). The oligonucleotide sequence used was described previously (Chiou and Hoa, 2009). AGS cells were seeded in six-well tissue culture plates to obtain 70% confluence the next day. Endoporter transfection reagent (Gene Tools) was used alone (mock transfection) or to deliver either negative control (inverse sequence to the antisense) or the antisense oligonucleotide at 5 μM concentration into cells. Forty-eight hours later, Western blot analysis was used to determine the extent of suppression of GADD45α protein. At this time, cells were serum-starved overnight for treatment with vehicle only, 0.2 or 0.1 mM indomethacin, or 0.1 or 0.05 mM sulindac sulfide for 24 h followed by apoptosis assay and Western blot analysis for caspase-9 activation.
Determination of Growth Inhibition and Apoptosis in Response to Indomethacin and Sulindac Sulfide Treatments in Cell Culture Studies.
For growth inhibition assays, 104 AGS and RGM-1 cells per well were seeded and grown in reduced serum medium (0.5% FBS) overnight, and then triplicate wells were treated with vehicle only; 0.002, 0.01, 0.02, and 0.2 mM indomethacin; or 0.001, 0.005, 0.01, and 0.1 mM sulindac sulfide. Proliferation assays were performed using the CyQUANT cell proliferation assay kit (Invitrogen), according to the manufacturer's instructions, at 0, 24, 48, and 72 h, with renewal of culture medium containing NSAIDs every 24 h. Fluorescence intensity data were obtained using NOVAStar (BMG Labtech, Durham, NC) and converted to number of cells according to the manufacturer's instructions.
For apoptosis assays, AGS and RGM-1 cells were grown and serum-starved as described above. Then, they were treated with vehicle only; 0.2, 0.1, or 0.05 mM indomethacin; or 0.1, 0.05, or 0.01 mM sulindac sulfide. Twenty-four hours later, the number of cells undergoing apoptosis was measured using BD ApoAlert apoptosis detection kit (BD Biosciences, San Jose, CA), following the manufacturer's instructions, and flow cytometry analysis as described previously (Ge et al., 2009). In brief, induced cells were trypsinized, gently washed, and labeled with annexin V per the manufacturer's protocol. Ten thousand cells were analyzed by flow cytometry using an FACSCalibur cytometer (BD Biosciences), with single laser emitting excitation wavelength at 488 nm.
Student's two-tailed t test was used to compare data between two groups. One-way analysis of variance and Bonferroni's correction were used to compare data between three or more groups. A p value ≤0.05 was considered statistically significant.
Gastric Mucosal Injury and Apoptosis in NSAID-Treated WT Versus GADD45α(−/−) Mice.
Vehicle-only treatment resulted in low basal amounts of visible gastric mucosal injury in WT mice at all the time points examined (Fig. 1A). Sulindac and indomethacin treatments caused severe visible gastric lesions in WT mice at all the time points examined, compared with vehicle-only treatment (Fig. 1A). The extent of gastric mucosal injury was similar at 4 and 8 h and slightly less at 24 h after NSAIDs were administered (Fig. 1A). Vehicle-only treatment also resulted in low basal amounts of visible gastric mucosal injury in GADD45α(−/−) mice at all the time points examined (Fig. 1B). Compared with vehicle-only treatment, treatment with sulindac did not cause significantly greater amounts of visible gastric mucosal injury at all the time points examined in GADD45α(−/−) mice (Fig. 1B). Indomethacin treatment caused slightly greater amounts of visible gastric mucosal injury in GADD45α(−/−) mice but the increase was not statistically significant (Fig. 1B).
Immunofluorescence staining of gastric mucosal sections showed that caspase-3 is expressed almost ubiquitously in both the WT and GADD45α(−/−) mouse gastric mucosa treated with vehicle only, sulindac, or indomethacin (Fig. 2, top). The white dotted lines demarcate the borders of gastric mucosal lesions found in sulindac- and indomethacin-treated mice (Fig. 2, top) and showed that caspase-3 is also detectable in areas surrounding the lesions. This result is consistent with the finding that caspase-3 expression is widely detected in normal human gastric mucosa (Kania et al., 2003), and ti enabled us to measure caspase-3 activity in the mouse gastric mucosa as a method to determine the extent of apoptosis induction by NSAIDs. Basal caspase-3 activity was detected in the gastric mucosa of vehicle-only-treated WT mice (Fig. 2, top left graph). Sulindac and indomethacin treatments increased caspase-3 activity 3.1 ± 0.2- and 4.2 ± 0.2-fold, respectively, in gastric mucosa of WT mice (Fig. 2, top left graph). Basal caspase-3 activity in gastric mucosa of GADD45α(−/−) mice treated with vehicle only was slightly lower than that in gastric mucosa of vehicle-only-treated WT mice (0.79 ± 0.1-fold; Fig. 2, top right graph). Sulindac and indomethacin treatments increased caspase-3 activity 1.2 ± 0.2- and 1.4 ± 0.3-fold, respectively, in gastric mucosa of GADD45α(−/−) mice (Fig. 2, top right graph). The increases in caspase-3 activity due to indomethacin and sulindac treatments in the GADD45α(−/−) gastric mucosa were significant but drastically less than in WT gastric mucosa.
To determine the percentage of apoptotic cells in intact gastric mucosa and erosions induced by sulindac and indomethacin, cross-sections of intact mucosa and erosions were stained with active caspase-3 antibodies, and the percentage of positive cells was determined. Basal amounts of apoptotic cells were detected in WT intact gastric mucosa with vehicle-only treatment (8.1 ± 0.7%; Fig. 2, bottom left graph). Compared with vehicle treatment, sulindac and indomethacin treatments slightly but significantly increased the percentage of apoptotic cells in the intact areas of the gastric mucosa in WT mice (12.3 ± 0.8 and 11.5 ± 0.4, respectively; Fig. 2, bottom left graph). Both sulindac- and indomethacin-induced erosions contained significantly greater proportions of apoptotic cells (32.1 ± 4.3 and 38.5 ± 3.3%, respectively; Fig. 2, bottom left graph). Basal amounts of apoptotic cells were also detected in GADD45α(−/−) intact gastric mucosa with vehicle-only treatment (5.7 ± 1.2%; Fig. 2, bottom right graph). Sulindac and indomethacin treatments also slightly increased the percentage of apoptotic cells in the intact areas of the gastric mucosa in GADD45α(−/−) mice (8.1 ± 0.7 and 7.6 ± 0.5, respectively; Fig. 2, bottom right graph). Both sulindac- and indomethacin-induced erosions contained significantly higher proportions of apoptotic cells (15.1 ± 1.3 and 18.0 ± 0.9%, respectively; Fig. 2, bottom right graph). Although the percentage of apoptotic cells was significantly increased with NSAIDs treatments in gastric mucosa and erosions in GADD45α(−/−) mice, they were markedly lower than that in WT mice.
Up-Regulation of GADD45α Expression in Gastric Mucosa of WT Mice.
Gastric mucosal scrapings from WT mice treated with sulindac and indomethacin were analyzed for GADD45α protein expression compared with those treated with vehicle only. Both sulindac and indomethacin treatments significantly increased GADD45α protein levels in the gastric mucosa compared with vehicle-only treatments (2.5 ± 0.2- and 3.8 ± 0.9-fold increase, respectively; Fig. 3, see graphs). Western blots showing representative GADD45α protein levels found in gastric mucosal scrapings from vehicle-only-, sulindac-, and indomethacin-treated mice are shown Fig. 3 (top).
Up-Regulation of GADD45α mRNA and Protein Expression by Sulindac and Indomethacin Treatments in Human Gastric Mucosal AGS and Normal Rat Gastric Epithelial RGM-1 Cells in Culture.
AGS and RGM-1 cells were treated with vehicle only and with increasing concentrations of sulindac sulfide and indomethacin. GADD45α mRNA expression levels were then measured by quantitative real-time PCR. Both sulindac sulfide and indomethacin treatments significantly increased GADD45α mRNA levels in both cell lines, and the greater the NSAID concentration used, the greater the fold increase (Fig. 4, A and B). Our treatments with sulindac sulfide down-regulated survivin mRNA expression in both cell lines, and the fold down-regulation increased with increasing NSAID concentrations (Fig. 4A). Survivin mRNA down-regulation was measured as independent confirmation of the effectiveness of our sulindac sulfide treatments, because others have shown that sulindac treatment down-regulates survivin via transcriptional control in cultured cells (Zhang et al., 2004). Survivin mRNA was not measured in indomethacin-treated cells, because we have shown previously that indomethacin treatment does not down-regulate survivin at the mRNA level but that it enhances survivin protein degradation (Chiou and Mandayam, 2007).
Treatments with sulindac sulfide and indomethacin both up-regulated GADD45α protein levels compared with vehicle-only treatment in both cell lines (Fig. 4C). Treatments with sulindac sulfide and indomethacin both down-regulated survivin protein levels compared with vehicle-only treatment in both cell lines (Fig. 4 C), verifying that our sulindac sulfide and indomethacin treatments were effective in these cells.
Induction of Apoptosis in AGS and RGM-1 Cells by Sulindac Sulfide and Indomethacin Treatments.
Treatments of both AGS and RGM-1 cells with indomethacin and sulindac sulfide induced significant amounts of apoptosis compared with vehicle-only treatment, as shown by annexin V binding assay and flow cytometry (Fig. 5). The amount of apoptosis induction increased with increasing indomethacin and sulindac sulfide concentrations used (Fig. 5, C and D, respectively). Thus, there is a positive association between the extent of apoptosis induction and the extent of GADD45α up-regulation by indomethacin and sulindac sulfide treatments in both cell lines.
Inhibition of Sulindac Sulfide- and Indomethacin-Induced Apoptosis, and Caspase-9 Activation, in AGS Cells by Antisense Suppression of GADD45α Expression.
To suppress GADD45α expression in AGS cells, we transfected antisense oligonucleotides into the cells, and then we monitored GADD45α protein levels by Western blot analysis. Compared with mock transfection and transfection with the negative control oligonucleotide, transfection with antisense GADD45α oligonucleotides resulted in approximately 60% reduction of GADD45α protein levels (Fig. 6A).
Flow cytometric quantitation of annexin V-bound apoptotic cells revealed that both low and high concentrations of sulindac sulfide and indomethacin induced significant amounts of apoptosis in mock-transfected cells and that the fold apoptosis induction increased with increasing NSAID concentrations used (Fig. 6, B and C, respectively). Significant apoptosis was also induced by all concentrations of sulindac sulfide and indomethacin in control oligonucleotide-transfected cells, at levels comparable with or slightly lower than that of mock-transfected cells (Fig. 6, B and C, respectively). In contrast, cells transfected with antisense GADD45α oligonucleotides were resistant to sulindac sulfide- and indomethacin-induced apoptosis. Only the high concentrations of these NSAIDs used were able to induce significant amounts of apoptosis, and the fold inductions were dramatically lower than that in mock and control oligonucleotide-transfected cells (Fig. 6, B and C).
We investigated whether antisense GADD45α suppression affects the caspase-9 pathway of apoptosis. Western blot analysis showed that low levels of procaspase-9 were expressed in the cells, and treatments with sulindac sulfide and indomethacin did not alter procaspase-9 levels compared with vehicle treatment (Fig. 7A). Basal levels of active caspase-9 p10 subunit were detected in mock and control oligonucleotide-transfected cells treated with vehicle only, and anti-GADD45α transfection reduced the basal levels of active caspase-9 (Fig. 7A). Sulindac sulfide and indomethacin treatments dramatically increased active caspase-9 p10 subunit levels in mock and control oligonucleotide-transfected cells but not in anti-GADD45α oligonucleotide-transfected cells, compared with vehicle-only treatment (Fig. 7A).
Analysis of Antiproliferative Effects of Sulindac Sulfide and Indomethacin Treatments in AGS and RGM-1 Cells.
Increased GADD45α expression is associated with cell growth inhibition, and deletion of GADD45α frequently results in uncontrolled proliferation (Siafakas and Richardson, 2009). Therefore, we examined the effects of sulindac sulfide and indomethacin on AGS and RGM-1 cell proliferation and its relationship to GADD45α up-regulation. Vehicle-only-treated AGS and RGM-1 cells multiplied at a rate of 2 × 103 and 8 × 103 cells/day, respectively (Fig. 7B). Indomethacin treatment at lower than 0.02 mM did not have any detectable effect on proliferation of either cell line (negative data; data not shown). At 0.02 mM, indomethacin treatment reduced proliferation of both AGS and RGM-1 cells to 0.5 × 103 and 4 × 103 cells/day, respectively (Fig. 7B). This concentration of indomethacin was insufficient to induce apoptosis and did not up-regulate GADD45α mRNA expression in either cell line (negative data; data not shown). Indomethacin treatment at 0.2 mM was included in this assay as positive control for up-regulation of GADD45α and loss of cell numbers due to induction of cell death (Figs. 4 and 7B). Sulindac sulfide treatment at concentrations lower than 0.01 mM did not up-regulate GADD45α expression, induce apoptosis, or have any detectable effect on AGS and RGM-1 proliferation (negative data; data not shown). Sulindac sulfide treatment at 0.01 mM reduced proliferation of AGS and RGM-1 cells and induced apoptosis (Figs. 4 and 7B). Sulindac sulfide at 0.1 mM was included as positive control for up-regulation of GADD45α expression and loss of cell numbers due to induction of cell death (Figs. 4 and 7B). Thus, indomethacin but not sulindac sulfide treatment at concentrations below what is necessary to induce apoptosis had growth inhibitory effects on gastric epithelial cells but did not up-regulate GADD45α expression.
Exogenous PGE2 Treatment and siRNA Suppression of COX-1 and -2 Did Not Inhibit Up-Regulation of GADD45α Expression by Sulindac Sulfide and Indomethacin.
To determine whether up-regulation of GADD45α by sulindac sulfide and indomethacin occurred through prostaglandin/COX-dependent pathways, we examined whether pretreatment with exogenous PGE2, and siRNA suppression of COX-1 and -2 expression, would inhibit up-regulation of GADD45α expression by sulindac sulfide and indomethacin treatments in AGS cells. As expected, sulindac sulfide treatment dramatically up-regulated GADD45α protein levels compared with vehicle-only treatment (Fig. 8A). Treatments with PGE2 alone did not affect GADD45α protein levels compared with vehicle-only treatment, and pretreatment with PGE2 also did not affect up-regulation of GADD45α protein levels by sulindac sulfide (Fig. 8A). Likewise, indomethacin treatment significantly up-regulated GADD45α protein levels compared with vehicle-only treatment (Fig. 8B). Pretreatment with PGE2 also did not affect up-regulation of GADD45α protein levels by indomethacin (Fig. 8B). Thus, up-regulation of GADD45α expression by sulindac sulfide and indomethacin treatments did not occur through prostaglandin/COX-dependent pathways.
Compared with mock transfections with no siRNAs, transfections with control RNA did not affect the expression of COX-1, COX-2, or GADD45α in AGS cells (Fig. 8C, left). Transfections with siRNAs targeting COX-1 and COX-2 effectively reduced COX-1 and COX-2 expression, respectively, to almost undetectable levels (Fig. 8C, left, top and middle). Cotransfection of COX-1 and COX-2 siRNA effectively reduced both COX-1 and -2 to undetectable levels, as visualized by using both COX-1 and COX-2 gene-specific primers in the same PCR (Fig. 8C, bottom left). Effective suppression of COX-1, COX-2, or both by their respective siRNAs did not affect GADD45α levels (Fig. 8C, left). As expected, sulindac sulfide and indomethacin treatments markedly increased GADD45α mRNA levels compared with vehicle-only treatment in control RNA-transfected AGS cells (Fig. 8C, top right). Transfections with siRNA targeting COX-1, COX-2, or both did not inhibit up-regulation of GADD45α mRNA expression by sulindac sulfide and indomethacin (Fig. 8C, right, middle and bottom).
In this study, we showed that 1) GADD45α expression was up-regulated by sulindac and indomethacin treatments both in vivo in mouse gastric mucosa and in vitro in AGS and RGM-1 cells. 2) Apoptotic cells constitute a significant percentage of cells in and around NSAID-induced gastric mucosal erosions in vivo. 3) Up-regulation of GADD45α expression by NSAID treatments in AGS and RGM-1 cells was accompanied by apoptosis. 4) GADD45α null mice were resistant to NSAID-induced gastric injury and apoptosis, and antisense suppression of GADD45α expression in AGS cells inhibited NSAIDs-induced caspase-9 activation and apoptosis. 5) Pretreatments with PGE2 and siRNA suppression of COX-1 and -2 did not inhibit up-regulation of GADD45α by sulindac sulfide and indomethacin in AGS cells. These data suggest that GADD45α up-regulation is a COX-independent mechanism that is necessary for induction of severe gastric injury and caspase-9-dependent apoptosis pathway by sulindac and indomethacin.
In WT mice, indomethacin treatment up-regulated GADD45α expression more effectively (3.8 ± 0.9-fold) than sulindac treatment (2.5 ± 0.2-fold) and induced greater gastric mucosal injury (17.7 ± 14.6% at 4 h) and apoptosis (4.2 ± 0.2-fold versus vehicle only) than sulindac treatment (14.1 ± 8.4% injury at 4 h and 3.1 ± 0.2-fold versus vehicle only). This positive association of the magnitude of GADD45α up-regulation with extent of gastric mucosal injury and apoptosis suggests a causative effect between GADD45α up-regulation and induction of injury and apoptosis in gastric mucosa. However, GADD45α up-regulation is not the sole mechanism responsible, because in gastric mucosa of GADD45α(−/−) mice, treatments with sulindac and indomethacin could still induce caspase-3 activation to significantly higher levels than treatment with vehicle only, and in a moderate amount of apoptotic cells in and around gastric mucosal erosions. These findings indicated that both sulindac and indomethacin treatments induced apoptosis pathways that did not require the presence of GADD45α gene expression.
In agreement with our previous experience that acute mucosal damage by NSAIDs is rapidly produced via systemic effects, in this study, sulindac and indomethacin treatments produced severe gastric mucosal damage in WT mice as early as 4 h after intraperitoneal injection. The extent of injury was sustained for at least 8 h after NSAID administration. The decrease in gastric mucosal damage observed at 24 h after NSAIDs were administered was probably due to healing of the gastric mucosa in the absence of repeat NSAID treatment after the initial injection. Sulindac and indomethacin treatments did not induce significant gastric mucosal damage at all time points up to 24 h in GADD45α-null mice, suggesting that inhibition of NSAID-induced gastric mucosal injury in GADD45α-null mice was not due to a delay in injury induction. Furthermore, the effective resistance of GADD45α null mice to NSAID-induced gastric mucosal damage suggested that deletion of GADD45α may also inhibit other nonapoptotic cell deaths that also contribute to gastric mucosal injuries, because our data showed that only approximately a third of gastric mucosal cells in and around NSAID-induced erosions were apoptotic cells.
We used the human gastric carcinoma AGS cells as our in vitro human model because at present there is no normal human gastric mucosal cell model available. Normal human gastric epithelial cells have poor viability in culture, making it difficult to establish a cell line for culture studies. We used an immortalized non-neoplastic rat gastric epithelial cell line, RGM-1, as a normal cell model.
As in the in vivo mouse model, in both AGS and RGM-1 cells the extent of apoptosis increased with greater GADD45α up-regulation by higher concentrations of both sulindac sulfide and indomethacin treatments. In RGM-1 cells, indomethacin treatment up-regulated GADD45α and induced apoptosis to greater extent than sulindac sulfide treatment. In AGS cells, however, sulindac sulfide treatments up-regulated GADD45α expression to comparable levels as indomethacin treatments but induced markedly greater apoptosis than did indomethacin treatment. An explanation for this is that, in addition to the GADD45α pathway, sulindac sulfide induced more numbers of and/or more potent apoptosis pathways than did indomethacin in AGS cells. An alternative explanation is that GADD45α may play a greater role in sulindac sulfide-induced than indomethacin-induced apoptosis pathway(s) in AGS cells. Our data support the latter supposition, because consistent levels of GADD45α suppression by antisense oligonucleotides in both sulindac sulfide- and indomethacin-treated cells resulted in greater inhibition of sulindac sulfide-induced apoptosis (approximately 2.7-fold decrease relative to mock-transfected) than indomethacin-induced apoptosis (approximately 1.6-fold decrease relative to mock-transfected).
Results from our AGS studies were inconsistent with our in vivo data showing that indomethacin is a more potent drug than sulindac in up-regulation of GADD45α expression and induction of gastric mucosal injury and gastric cell apoptosis. These differences may be due in part to the fact that we used sulindac, a prodrug that requires further breakdown in vivo to produce active metabolites, in our animal studies, whereas we used sulindac sulfide, the active metabolite of sulindac, directly in our cell culture studies. After the breakdown of sulindac in vivo, the resulting concentration of the active sulfide metabolite may be greatly diminished relative to the concentration of sulindac we administered. Another explanation is that the gastric carcinoma AGS cells responded to each drug differently than non-neoplastic gastric epithelial cells. Our data showing that indomethacin treatment is more potent than sulindac sulfide treatment in up-regulating GADD45α expression and inducing apoptosis in RGM-1 cells support this supposition. A third possibility remains that human gastric epithelial cells responded differently to each drug than rodent gastric epithelial cells.
This study is the first to demonstrate that GADD45α expression is required for induction of severe gastric mucosal injury and apoptosis by NSAIDs. They are consistent with our previous findings and those of others showing that GADD45α expression is up-regulated in various cancer cells in response to NSAIDs and that GADD45α is involved in injury of other tissues. We have shown previously that indomethacin and sulindac treatments up-regulate GADD45α expression in colon cancer cells in culture (Chiou et al., 2009). GADD45α up-regulation has also been observed in various prostate, breast, and stomach cancer cells in culture in association with induction of interleukin-24 by sulindac, aspirin, ibuprofen, acetaminophen, and naproxen, resulting in cancer cell apoptosis and/or growth arrest (Zerbini et al., 2006). GADD45α up-regulation is also observed in adult primary sensory and motor neurons after peripheral axonal injury (Befort et al., 2003), and GADD45α has been shown to participate in lipopolysaccharide- and ventilator-induced inflammatory lung injury (Meyer et al., 2009).
NSAIDs are known to have antiproliferative effects on a variety of cell types. GADD45α overexpression has been shown to lead to growth arrest (Hollander and Fornace, 2002). Therefore, the possibility arises that up-regulation of GADD45α expression may mediate NSAID-induced growth inhibition of gastric epithelial cells. The relationship between GADD45α up-regulation and growth inhibition of gastric epithelial cells by NSAIDs is unknown. This study showed that low concentrations of indomethacin, but not sulindac sulfide, that was insufficient to induce cell death had growth inhibitory effects on AGS and RGM-1 cells, but did not up-regulate GADD45α expression. This indicated that in the gastric epithelial cells we used, GADD45α is not involved in inhibition of cell proliferation in response to NSAID treatments.
Our current study provided evidence that GADD45α up-regulation by NSAIDs triggers a caspase-9-dependent pathway of apoptosis. Caspase-9 is the initiator caspase involved in the intrinsic or mitochondrial pathway of apoptosis (Li et al., 1997). Once activated, it cleaves and activates downstream effector caspases such as caspase-3 and/or -7 to induce apoptosis (Taylor et al., 2008). Our finding does not preclude the possibility that other caspase pathways are triggered by GADD45α up-regulation. The mechanism of how GADD45α up-regulation leads to apoptosis in gastric mucosa, and what caspase pathways are involved, require further investigation.
Our current data support the idea that, in addition to inhibition of COX enzymes, COX-independent mechanisms of NSAID action are also important contributors to induction of gastric mucosal damage and apoptosis. Disruption of COX-independent pathways such as genetic ablation of GADD45α expression in vivo and antisense suppression of GADD45α expression in cultured cells could largely inhibit NSAID-induced gastric mucosal injury and gastric cell apoptosis. Our current data also suggest that effective strategies may be developed for minimizing NSAID-induced gastric injury by utilizing methods that result in inhibition of GADD45α expression, such as siRNA or antisense oligonucleotide treatment.
We thank Drs. M. C. Hollander (National Cancer Institute, Bethesda, MD) and J. D. Ashwell (National Cancer Institute) for providing the GADD45α null mice and the extra breeding pairs for this study.
This work is supported by the Department of Veterans Affairs Merit Review [Grant GAST-011-08S (to S.-K.C.)].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- nonsteroidal anti-inflammatory drug
- microsomal prostaglandin E2 synthase-1
- prostaglandin E2
- growth arrest and DNA damage-inducible 45α
- small interfering RNA
- wild type
- fetal bovine serum
- reverse transcription-polymerase chain reaction
- quantitative polymerase chain reaction.
- Received March 16, 2010.
- Accepted May 21, 2010.
- U.S. Government work not protected by U.S. copyright