A Role for HSP70 in Protecting against Indomethacin-induced Gastric Lesions*

A major clinical problem encountered with the use of nonsteroidal anti-inflammatory drugs (NSAIDs), such as indomethacin, is gastrointestinal complications. Both NSAID-dependent cyclooxygenase inhibition and gastric mucosal apoptosis are involved in NSAID-produced gastric lesions, and this apoptosis is mediated by the endoplasmic reticulum stress response and resulting activation of Bax. Heat shock proteins (HSPs) have been suggested to protect gastric mucosa from NSAID-induced lesions; here we have tested this idea genetically. The severity of gastric lesions produced by indomethacin was worse in mice lacking heat shock factor 1 (HSF1), a transcription factor for hsp genes, than in control mice. Indomethacin administration up-regulated the expression of gastric mucosal HSP70. Indomethacin-induced gastric lesions were ameliorated in transgenic mice expressing HSP70. After indomethacin administration, fewer apoptotic cells were observed in the gastric mucosa of transgenic mice expressing HSP70 than in wild-type mice, whereas the gastric levels of prostaglandin E2 for the two were indistinguishable. This suggests that expression of HSP70 ameliorates indomethacin-induced gastric lesions by affecting mucosal apoptosis. Suppression of HSP70 expression in vitro stimulated indomethacin-induced apoptosis and activation of Bax but not the endoplasmic reticulum stress response. Geranylgeranylacetone induced HSP70 at gastric mucosa in an HSF1-dependent manner and suppressed the formation of indomethacin-induced gastric lesions in wild-type mice but not in HSF1-null mice. The results of this study provide direct genetic evidence that expression of HSP70 confers gastric protection against indomethacin-induced lesions by inhibiting the activation of Bax. The HSP inducing activity of geranylgeranylacetone seems to contribute to its gastroprotective activity against indomethacin.

Nonsteroidal anti-inflammatory drugs (NSAIDs), 2 such as indomethacin, are a useful family of therapeutics (1). An inhib-itory effect of NSAIDs on cyclooxygenase (COX) activity is responsible for their anti-inflammatory actions because COX is an enzyme essential for the synthesis of prostaglandins (PGs), such as PGE 2 , which have a strong capacity to induce inflammation. On the other hand, NSAID use is associated with gastrointestinal complications (2)(3)(4).
Although PGE 2 has a strong protective effect on gastrointestinal mucosa, the inhibition of COX by NSAIDs is not the sole explanation for the gastrointestinal side effects of NSAIDs (5). We have recently demonstrated that NSAIDs induce apoptosis in cultured gastric cells and at gastric mucosa in a manner independent of COX inhibition (6 -10). As for the molecular mechanism governing this apoptosis, we have proposed the following pathway. Permeabilization of cytoplasmic membranes by NSAIDs stimulates Ca 2ϩ influx and increases intracellular Ca 2ϩ levels, which in turn induces the endoplasmic reticulum (ER) stress response (6,11,12). In the ER stress response, an apoptosis-inducing transcription factor, CCAAT enhancerbinding protein homologous transcription factor (CHOP), is induced, as we have previously shown, CHOP is essential for NSAID-induced apoptosis (7). CHOP induces expression of p53 up-regulated modulator of apoptosis (PUMA) and the resulting translocation and activation of Bax. We have already shown that both PUMA and Bax play an important role in NSAID-induced mitochondrial dysfunction, activation of caspases and apoptosis (13). Furthermore, we have suggested that both COX inhibition (measured as a decrease in the gastric PGE 2 level) and gastric mucosal apoptosis are required for the formation of NSAID-induced gastric lesions in vivo (10,14). Therefore, maintenance of gastric PGE 2 levels or protection against gastric mucosal apoptosis is important for protection of gastric mucosa from NSAID-induced lesions.
Heat shock proteins (HSPs) are induced by various stressors, including NSAIDs, and induction of HSPs, especially HSP70, provides cellular resistance to NSAIDs (15,16). Interestingly, geranylgeranylacetone (GGA), a leading anti-ulcer drug on the Japanese market, has been reported to be a nontoxic HSP inducer (17). We have previously reported that treatment of * This work was supported by grants-in-aid of scientific research from the Ministry of Health, Labour, and Welfare of Japan, grants-in-aid for scientific research from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and grants-in-aid of the Japan Science and Technology Agency. 1 To whom correspondence should be addressed. Tel. and Fax: 81-96-371-4323; E-mail: mizu@gpo.kumamoto-u.ac.jp. 2  cultured gastric mucosal cells with GGA protects cells from NSAID-induced cell death and is accompanied by induction of HSP70 (18). These previous results suggest that HSP70 protects gastric mucosa from NSAID-induced gastric lesions; however, no direct evidence exists. The results also suggest that the protective effect of GGA against NSAID-produced gastric lesions is due to its HSP inducing activity. However, because GGA mediates various other gastroprotective mechanisms, such as an increase in gastric mucosal blood flow, stimulation of surface mucus production, and direct protection of gastric mucosal cell membranes (19 -21), it remains unclear whether up-regulation of HSPs represents major mode of gastroprotective activity of GGA against NSAIDs. Up-regulation of HSP expression by stressors is achieved at the level of transcription through a consensus cis-element (heat shock element) and a transcription factor (heat shock factor 1 (HSF1)), which specifically binds to a heat shock element located upstream of the hsp genes (22). Disruption to the activity of HSF1 leads to the loss of stressor-induced HSP up-regulation (22,23). We recently used HSF1-null mice to obtain genetic evidence for a protective role for HSPs against production of gastric lesions (24). However, NSAID-produced gastric lesions, which involve not only irritant-induced mucosal damage but also a decrease in the PGE 2 level, were not tested. Furthermore, although transgenic mice expressing HSP27 display a phenotype of resistance to irritant-induced gastric lesions (25), the effect of artificial up-regulation of HSP70 on the production of gastric lesions has not been tested. In this study, we used HSF1-null mice and transgenic mice expressing HSP70 to examine the role of HSP70 in the pathogenesis of NSAID-produced gastric lesions. The results suggest that HSP70 is protective against the production of indomethacin-induced gastric lesions. Furthermore, the results suggest that HSP70 achieves this protective effect through inhibiting mucosal apoptosis rather than affecting the gastric level of PGE 2 .

EXPERIMENTAL PROCEDURES
Chemicals and Animals-RPMI 1640 was obtained from Nissui Pharmaceutical Co. (Osaka, Japan). Paraformaldehyde, fetal bovine serum, and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide were obtained from Sigma. Indomethacin was obtained from Wako Co. (Osaka, Japan). A PGE 2 enzyme-linked immunosorbent assay kit was purchased from Cayman Chemical (Ann Arbor, MI). Antibodies against HSP70 were purchased from StressGen (San Diego, CA), and actin and the N-terminal region of Bax (Bax N20) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Terminal deoxynucleotidyl transferase (TdT) was obtained from TOYOBO (Osaka, Japan). Biotin 14-ATP, Alexa Fluor 488 goat anti-rabbit (or anti-mice) immunoglobulin G, and Alexa Fluor 488 conjugated with streptavidin were purchased from Invitrogen. Mounting medium for immunohistochemical analysis (VECTASHIELD) was from Vector Laboratories (Burlingame, CA). 4Ј,6-Diamidino-2-phenylindole dihydrochloride (DAPI) was from Dojindo (Kumamoto, Japan). The RNeasy kit and HiPerFect were obtained from Qiagen, the first strand cDNA synthesis kit was from Takara (Kyoto, Japan), and SYBR GREEN PCR Master Mix was from ABI (Foster City, CA). HSF1-null mice and their wild-type counterparts (ICR) (10 -12 weeks of age and 30 -35 g) or transgenic mice expressing HSP70 (gifts from Drs. Angelidis and Pagoulatos (University of Ioannina, Greece)) and their wild-type counterparts (C57/BL6) (6 -8 weeks of age and 20 -25 g) were prepared as described previously (27). Homozygotic transgenic mice were used in experiments. The experiments and procedures described here were carried out in accordance with the Guide for the Care and Use of Laboratory Animals as adopted and promulgated by the National Institutes of Health (Bethesda, MD) and were approved by the Animal Care Committee of Kumamoto University.
Gastric Damage Assay-The gastric ulcerogenic response was examined as described previously (14), with some modifications. Mice fasted for 18 h were orally administered with indomethacin. Eight hours later, the animals were sacrificed, after which their stomachs were removed, and the areas of gastric mucosal lesions were measured by an observer unaware of the treatment they had received. Calculation of the scores involved measuring the area of all the lesions in square millimeters and summing the values to give an overall gastric lesion index. The gastric PGE 2 level was determined by enzymelinked immunosorbent assay according to the manufacturer's instructions.
Cell Culture, Measurement of Caspase Activity and K ϩ Efflux, and siRNA Targeting-Human gastric adenocarcinoma (AGS) cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere of 95% air with 5% CO 2 at 37°C. The cells were exposed to indomethacin by changing the medium. The cells were cultured for 24 h and then used in experiments. Cell viability was determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide method as previously described (28). The measurement of caspase-3-like activity was as previously described (9,11). K ϩ efflux from cells was monitored as previously described (21). The transfection with pcDNA3.1 containing the hsp70 gene (29) was carried out using Lipofectamine (TM2000) according to the manufacturer's instructions. The cells were used for experiments after a 20-h recovery period.
Fluorescence-activated Cell Sorting (FACS) Analysis-Apoptosis was monitored using PI analysis by FACS as described previously (11). The cells were collected by centrifugation, and the pellets were fixed with 70% ethanol and recentrifuged. The pellets were resuspended in phosphate-citrate buffer (0.2 M Na 2 HPO 4 and 4 mM citric acid) and incubated for 20 min at room temperature. After centrifugation, the pellets were resuspended in DNA staining solution (50 mg/ml PI and 10 g/ml RNase A) and incubated for 20 min at room temperature. Samples were scanned with a FACSCalibur (Becton Dickinson) cell sorter. For measuring only PI-DNA mediated fluorescence spe-cifically, we scanned cells without PI staining and only PI solution (without cells) to determine the signal threshold. For excluding debris and cell fragments, thresholds were set for eliminating the low channels (from 1 to 15), in other words, cells with low PI fluorescence (the G 1 peak was set as channel 400 of total 1024 channels) or for eliminating small cells by adjusting FSC and SSC light scatter parameters (according to the manufacturer's protocols). Apoptotic cells appeared as a hypodiploid peak (sub-G 1 peak) because of nuclear fragmentation and loss of DNA, and we calculated number of cells included this peak and determined the ratio of apoptotic cells to total cells (10, 000). For statistical analysis, we measured three different samples in the same experiment.
Real Time RT-PCR Analysis-Total RNA was extracted from gastric tissues or AGS cells using the RNeasy kit according to the manufacturer's protocol. Samples (2.5 g of RNA) were reverse-transcribed using the first strand cDNA synthesis kit according to the manufacturer's instructions. Synthesized cDNA was used in real time RT-PCR (Bio-Rad Chromo 4 system) experiments using iQ SYBR Green Supermix and analyzed with Opticon Monitor software according to the manufacturer's instructions. The real time PCR cycle conditions were 95°C for 3 min, followed by 45 cycles at 95°C for 10 s and at 60°C for 60 s. Specificity was confirmed by electrophoretic analysis of the reaction products and by inclusion of template-or reverse transcriptase-free controls. To normalize the amount of total RNA present in each reaction, glyceraldehyde-3-phosphate dehydrogenase or actin cDNA was used as an internal standard. The primers were designed using the Primer3 website (Table 1).
Immunoblotting Analysis-Total protein was extracted as described previously (30). The protein concentration of each sample was determined by the Bradford method (31). Samples were applied to 8% (HSP70) or 10% (actin, Bax, and cytochrome c) polyacrylamide SDS gels and subjected to electrophoresis, after which the proteins were immunoblotted with appropriate antibodies.
Histological, Immunohistochemical, and TdT-mediated Biotinylated UTP Nick End Labeling (TUNEL) Analyses-Gastric tissue samples were fixed in 4% buffered paraformaldehyde and embedded in paraffin before being cut into 4-m sections.
For histological examination (hematoxylin and eosin staining), the sections were stained first with Mayer's hematoxylin and then with 1% eosin alcohol solution. The samples were mounted with Malinol and inspected with the aid of an Olympus BX51 microscope.
For immunohistochemical analysis, the sections were blocked with 2.5% goat serum for 10 min, incubated for 12 h with antibody against HSP70 (1:250 dilution) in the presence of 2.5% bovine serum albumin and finally incubated for 1 h with Alexa Fluor 488 goat anti-mouse immunoglobulin G in the presence of DAPI (5 g/ml). The samples were mounted with VECTASHIELD and inspected using fluorescence microscopy (Olympus BX51).
For TUNEL assay, the sections were incubated first with proteinase K (20 g/ml) for 15 min at 37°C, then with TdT and biotin 14-ATP for 1 h at 37°C, and finally with Alexa Fluor 488 conjugated with streptavidin for 1 h. The samples were mounted with VECTASHIELD and inspected using fluorescence microscopy (Olympus BX51).
Staining of Cells-The cells were cultured on four-well Lab-Tek II glass slides (Nunc). After fixation with 4% formaldehyde for 20 min and permeabilization with 0.5% Triton X-100 for 5 min, nonspecific binding sites were blocked with 3% bovine serum albumin for 30 min. Immunostaining to detect the active form of Bax was performed with a polyclonal antibody (Bax N20) and Alexa Fluor 488 goat anti-rabbit immunoglobulin G. The cells were simultaneously stained with DAPI (5 g/ml). The cells were mounted with VECTASHIELD and inspected using fluorescence microscopy (Olympus BX51).
Statistical Analysis-All of the values are expressed as the means Ϯ S.E. The Tukey test or the Student's t test for unpaired results was used to evaluate differences between more than three groups or between two groups, respectively. Differences were considered to be significant for values of p Ͻ 0.05.

RESULTS
Enhanced Gastric Ulcerogenic Response Induced by Indomethacin in HSF1-null Mice-The development of gastric lesions following oral administration of indomethacin was compared between wild-type and HSF1-null mice. Administration of indomethacin produced gastric lesions in a dose-dependent manner, and this lesion production was significantly worse in HSF1-null mice relative to wild-type controls (Fig. 1A). This result shows that HSF1 plays an important role in protecting gastric mucosa from indomethacin-induced lesions.
As mentioned above, both a decrease in PGE 2 and mucosal apoptosis play important roles in NSAID-produced gastric lesions, and we have examined these processes in HSF1-null mice. As shown in Fig. 1B, there was no significant difference in the gastric level of PGE 2 between the HSF1-null mice and wildtype mice in either the presence or absence of indomethacin treatment. The level of gastric mucosal apoptosis was determined by TUNEL assay. An increase in the number of TUNELpositive (apoptotic) cells was observed after the indomethacin HSP70 and NSAID-induced Gastric Lesions JULY 17, 2009 • VOLUME 284 • NUMBER 29 administration, and this increase was more apparent in HSF1null mice than wild-type mice (Fig. 1, C and D). These results suggest that HSF1 protects the gastric mucosa from lesion formation by inhibiting indomethacin-induced apoptosis rather than by affecting the gastric PGE 2 level. We monitored the expression of hsp mRNAs in gastric tissues by real time RT-PCR. Indomethacin administration upregulated the expression of hsp70 mRNA, with this expression significantly lower in indomethacin-treated HSF1-null mice than in the wild-type mice ( Fig. 2A). Lack of the hsf1 gene did not affect the background level of expression of hsp70 mRNA ( Fig. 2A). In contrast, the expression of hsp25, hsp60, hsp90␣, and hsp90␤ mRNAs was not affected by either indomethacin administration or lack of the hsf1 gene ( Fig. 2A). Based on these results, we subsequently focused on HSP70.
Immunoblotting and immunohistochemical analyses demonstrated that indomethacin administration increased the level of HSP70 in gastric mucosa in wild-type mice but not in HSF1null mice (Fig. 2, B-D). Based on the results in Fig. 2, we proposed that the inability of HSF1-null mice to induce expression of HSP70 is responsible for their phenotypic sensitivity to indomethacin-induced gastric lesions and mucosal apoptosis.
Reduced Gastric Ulcerogenic Response Induced by Indomethacin in Transgenic Mice Expressing HSP70-To test the idea described above, the level of indomethacin-induced gastric lesions was compared between transgenic mice expressing HSP70 and wild-type mice. Relative to control mice, formation of indomethacin-induced gastric lesions was significantly suppressed in transgenic mice expressing HSP70 (Fig. 3A). By immunoblotting and immunohistochemical analyses, we confirmed that HSP70 expression was much higher in the gastric tissues of the transgenic mice than in those of the wild-type mice, regardless of whether or not they were treated with indomethacin (Fig. 3, B-D). These results suggest that HSP70 expression somehow suppresses the formation of indomethacin-induced gastric lesions.
As shown in Fig. 3E, there was no significant difference in the gastric level of PGE 2 between transgenic mice expressing HSP70 and wild-type mice. On the other hand, the level of indomethacin-induced gastric mucosal apoptosis was lower in transgenic mice expressing HSP70 than in wild-type mice (Fig.  3, F and G).
Mechanism for the Role of HSP70 in Protecting against Indomethacin-induced Apoptosis-To understand the molecular mechanism for HSP70-conferred protection against indomethacin-induced cell death in vitro, we first examined the effect of siRNA specific for HSP70 on indomethacin-induced cell death in AGS cells. Transfection of cells with siRNA for HSP70 inhibited the expression of HSP70 both in the presence and absence of indomethacin (Fig. 4A). As shown in Fig. 4B, treatment of cells with indomethacin induced cell death in a dose-dependent manner. Transfection of cells with siRNA for HSP70 stimulated this cell death but did not affect cell viability in the absence of indomethacin (Fig. 4B). We also found that overexpression of HSP70 slightly inhibited indomethacin-induced cell death (Fig. 4, C and D). Based on results reported in one of our previous papers (8), we speculated that the cell death evidenced in Fig. 4 (B and D) is mediated by apoptosis. To confirm this idea, we monitored indomethacin-induced apoptosis, using FACS analysis and measurement of caspase-3-like activity. Both analyses showed that indomethacin-induced apoptosis is stimulated by transfection of cells with siRNA for HSP70 (Fig. 4, E and F). The results in Fig. 4 suggest that knocking down of HSP70 stimulated indomethacin-induced apoptosis in AGS cells.
As described above, indomethacin-induced apoptosis is mediated by sequential induction of various cellular phenomena. We next examined which step of this apoptotic pathway is affected by expression of HSP70.
We have previously monitored NSAID-dependent permeabilization of cytoplasmic membranes by monitoring K ϩ efflux from cells (6,21). The K ϩ concentration in the culture medium increased in the presence of indomethacin, showing that K ϩ efflux from AGS cells was stimulated (Fig. 5A). Down-regulation of expression of HSP70 by siRNA did not affect the K ϩ efflux (Fig. 5B). These results suggest that expression of  ϫ200). The experiments shown in Fig. 2 were repeated at least three times, and basically similar results were obtained.
HSP70 did not affect indomethacin-dependent membrane permeabilization.
Next we examined the effect of expression of HSP70 on the ER stress response, using real time RT-PCR. As shown in Fig.  5C, treatment of cells with indomethacin up-regulated the mRNA expression of not only chop and puma but also atf4, which is an ER stress response-related transcription factor that regulates these genes (13), showing that treatment of AGS cells with indomethacin induces the ER stress response. Down-regulation of expression of HSP70 by siRNA did not affect the indomethacindependent up-regulation of expression of these genes (Fig.  5C), suggesting that expression of HSP70 did not affect the ER stress response.
We previously reported that in the pathway for NSAID-induced apoptosis, activation of Bax through conformational change and its resulting translocation from the cytosol to mitochondria occur after induction of ER stress response (13). The effect of expression of HSP70 on activation of Bax was tested by immunostaining analysis using an antibody that specifically recognizes the active form of Bax. This antibody can recognize only the active form in the immunostaining assay but can recognize all forms of Bax in the immunoblotting assay because of the denaturing of proteins in the latter assay (32). As shown in Fig.  5D, the active form of Bax was observed in indomethacin-treated cells but not in control cells, showing that indomethacin induces the conformational change in Bax. Furthermore, the levels of the active form of Bax observed with indomethacin treatment increased in cells transfected with siRNA for HSP70 (Fig. 5D). As shown in Fig.  5E, the number of Bax (active form)positive cells increased in the presence of indomethacin, and the transfection with siRNA for HSP70 enhanced this increase, suggesting that expression of HSP70 suppresses the indomethacin-dependent activation of Bax. As shown in Fig. 5 (F and G), the amount of Bax or cytochrome c in the cytosol fractions decreased or increased, respectively, in the presence of indomethacin, and these alterations were further enhanced by transfection with siRNA for HSP70. This sug-  F, magnification, ϫ200). The values are the means Ϯ S.E. (n ϭ 3-6). **, p Ͻ 0.01; *, p Ͻ 0.05; n.s., not significant. The experiments shown in Fig. 3 were repeated at least three times and basically similar results were obtained. H&E, hematoxylin and eosin.  Fig. 4 were repeated at least two times, and basically similar results were obtained. JULY 17, 2009 • VOLUME 284 • NUMBER 29

JOURNAL OF BIOLOGICAL CHEMISTRY 19711
gested that expression of HSP70 suppresses indomethacin-dependent translocation of Bax and mitochondrial outer membrane permeabilization (mitochondrial dysfunction).
Anti-ulcer and HSP-inducing Activities of GGA in HSF1-null Mice-To evaluate the contribution of the HSP inducing activity of GGA to its protective effect against NSAID-produced gastric lesions, we investigated the activities of GGA in HSF1-null mice. First, we examined the effect of GGA and/or indomethacin on gastric expression of HSP70. An induction of HSP70 expression by indomethacin or GGA alone was observed in the gastric mucosa of wild-type mice (Fig. 6A). Preadministration of GGA enhanced the indomethacin-induced expression of HSP70 (Fig. 6A). On the other hand, HSP70 expression was not observed in HSF1-null mice under  A and B, the level of K ϩ efflux was measured using a K ϩ ion-selective electrode. Melittin (10 M) was used to establish the 100% level of K ϩ efflux. C, the relative expression of each gene was monitored by real time RT-PCR using a specific primer for each gene. The values normalized to the actin gene are expressed relative to the control sample. D, immunostaining with the antibody against the N-terminal region of Bax (Bax N20) and DAPI staining were performed as described under "Experimental Procedures" (magnification, ϫ400). E, ϳ400 -600 cells were randomly counted for staining with Bax N20. F, after subcellular fractionation, cytosolic fractions were analyzed by immunoblotting with an antibody against Bax (Bax N20), cytochrome c, or actin. G, the band intensity of BAX and cytochrome c in three independent experiments (one of them is shown in F) was determined and shown. A-C, E, and G, values are the means Ϯ S.E. (n ϭ 3-4). **, p Ͻ 0.01; *, p Ͻ 0.05; n.s., not significant. The experiments shown in Fig. 5 were repeated at least two times, and basically similar results were obtained. FIGURE 6. Effect of indomethacin and/or GGA on expression of HSP70 and production of gastric lesions. HSF1-null mice (ϫ/ϫ) and wild-type mice (ICR) (ϩ/ϩ) were orally administered 50 mg/kg GGA (10 ml/kg as an emulsion with 5% gum arabic), 1 h after which they were orally administered 10 mg/kg (A-D) or the indicated doses (E) of indomethacin, and the stomach was removed after 8 h. A-D, immunohistochemical and immunoblotting analyses were performed as described in the legend of Fig. 2 (A and B, magnification, ϫ200). D, experiments shown in C were repeated for six mice/condition, and the results (band intensity of HSP70) were analyzed statistically. The values are the means Ϯ S.E. (n ϭ 6). **, p Ͻ 0.01. E, gastric mucosal lesions were measured as described in the legend of Fig. 1. The values are the means Ϯ S.E. (n ϭ 3-4). **, p Ͻ 0.01; n.s., not significant. The experiments shown in Fig. 6 were repeated at least two times, and basically similar results were obtained. H&E, hematoxylin and eosin.
any conditions (Fig. 6B). Similar results were observed in immunoblotting assay (Fig. 6, C and D). Fig. 6E shows the effect of preadministration of GGA on indomethacin-produced gastric lesions in wild-type and HSF1-null mice. Preadministration of GGA significantly suppressed the formation of indomethacin-induced gastric lesions in wild-type mice but not in HSF1null mice (Fig. 6E). This result shows that HSF1 is required for the anti-ulcer activity of GGA against indomethacin. GGA was administrated prior to the administration of indomethacin in Fig. 6E. It seems that preinduction of HSP70 is required to protect against the indomethacin-induced gastric lesions, because indomethacin alone can induce HSP70 to an extent similar to that of GGA alone (Fig. 6D). Overall, the results in Fig. 6 suggest that the loss of the protective effect of GGA against indomethacin in HSF1-null mice is due to the lack of expression of HSP70. In other words, the HSP inducing activity of GGA contributes to its protective effect against the formation of NSAID-produced gastric lesions.

DISCUSSION
A number of previous observations have suggested that HSPs, especially HSP70, play an important role in protecting gastric mucosa against the development of lesions (17,18,(33)(34)(35)(36)(37)(38)(39). However, prior to the current study, little direct evidence, genetic or otherwise, existed to support this idea. We recently addressed this issue by using mice that lack the ability to induce HSPs; HSF1-null mice are more susceptible to ethanol-or hydrochloric acid-produced gastric lesions (24). However, the effect of up-regulation of expression of HSP70 on irritant-induced production of gastric lesions had not been tested. Furthermore, NSAIDs, which are clinically more important as a cause of gastric lesions, were not tested in our previous study. In the present study, we have shown that HSF1-null mice are more susceptible to the formation of indomethacin-induced gastric lesions than controls. Furthermore, we have shown that transgenic mice expressing HSP70 are more resistant to the formation of indomethacin-induced gastric lesions. These results offer direct genetic evidence that expression of HSP70 protects the gastric mucosa against the formation of NSAID-induced gastric lesions. HSP70 normally has a reasonable level of expression in cells, and this may contribute to protection of gastric mucosa.
HSP70 seems to protect gastric mucosa from NSAID-induced lesions through inhibiting apoptosis rather than affecting gastric PGE 2 levels; indomethacin-induced gastric mucosal apoptosis, but not the induced decrease in the gastric level of PGE 2 , was ameliorated in transgenic mice expressing HSP70. We also showed that up-regulation or down-regulation of expression of HSP70 in cultured gastric cells make cells resistant or sensitive, respectively, to indomethacin-induced apoptosis. Of the various steps in the NSAID-induced apoptosis pathway (permeabilization of cytoplasmic membranes, stimulation of Ca 2ϩ influx across cytoplasmic membranes, increase in the intracellular Ca 2ϩ level, induction of the ER stress response (up-regulation of expression of CHOP and PUMA), translocation and activation of Bax, mitochondrial dysfunction, and activation of caspases), the translocation and activation of Bax seems to be the target of HSP70 for conferral of its inhibitory effect on NSAID-induced apoptosis. This may be concluded because indomethacin-dependent translocation and activation of Bax but not up-regulation of expression of CHOP and PUMA was enhanced by down-regulation of HSP70 expression. Although the mechanism by which HSP70 suppresses the activation of BAX is unclear at present, the direct interaction between HSP70 and Bax may be responsible, as previously reported (40,41). HSP70 binds to Apaf-1, thereby preventing activation of caspases, or HSP70 suppresses the apoptotic pathway downstream of ycaspase-3 activation and apoptosis-inducing factor-induced chromatin condensation (42)(43)(44)(45). Thus, these effects of HSP70 may also be involved in HSP70-dependent suppression of indomethacin-induced apoptosis. The effect of siRNA for HSP70 on apoptosis was more clear in Fig. 4E (apoptosis monitored by FACS) than Fig. 4F (apoptosis monitored by caspase-3-like activity), and we consider that this is because HSP70 suppresses the apoptotic pathway downstream of caspase-3 and caspase-3-independent pathway, such as apoptosis-inducing factor-induced chromatin condensation.
Of the various HSPs tested, oral administration of indomethacin up-regulated expression of HSP70 only. Although it is already known that NSAIDs induce expression of HSP70 in cultured cells (10,15,16), this is the first demonstration that administration of NSAIDs induces expression of HSP70 at the gastric mucosa in animals. Diclofenac (another NSAID) was recently reported to induce expression of HSP70 at the gastric mucosa in humans (46). Combining these results, it seems that induction of HSP70 by NSAIDs at the gastric mucosa plays an important role in protection against the formation of NSAIDproduced gastric lesions in humans. Similar specific up-regulation of expression of HSP70 (not a general induction of HSPs) has also been observed for ethanol-induced gastric lesions (24), although various HSPs are induced by these stressors (NSAIDs and ethanol) in cultured cells (15). Thus, expression of HSP70 seems to be specifically regulated at the gastric mucosa in response to stressors, although the mechanism is unknown at present.
In addition to the cytoprotective effect of HSP70, an antiinflammatory effect of HSP70 was recently revealed. Up-regulation of HSP70 expression by heat shock was found to inhibit the inflammatory stimuli-dependent activation of nuclear factor B (NF-B), which is responsible for induction of production of various pro-inflammatory cytokines (such as tumor necrosis factor-␣, IL-1␤, and IL-6) (47,48). We recently reported that lipopolysaccharide-induced production of tumor necrosis factor-␣, IL-1␤, and IL-6 was inhibited in peritoneal macrophages prepared from transgenic mice expressing HSP70, compared with those from wild-type controls (27). Given that it is well known that pro-inflammatory cytokines stimulate the production of NSAID-induced gastric lesions (49), it is possible that expression of HSP70 suppresses the development of NSAID-induced gastric lesions through its anti-inflammatory effect.
GGA has attracted considerable attention as an HSP inducer, largely because of its clinical value as an anti-ulcer drug and because it can induce HSPs without affecting cell viability (17). We have previously reported that GGA makes cultured gastric cells resistant to indomethacin and simultaneously up-regu-lates expression of HSP70 (18). In the current study we have shown that artificial expression of HSP70 makes cells resistant to indomethacin. Furthermore, we have revealed that preadministration of GGA suppresses the formation of indomethacin-induced gastric lesions in wild-type mice but not in HSF1null mice in which up-regulation of gastric mucosal expression of HSP70 by GGA alone and stimulation of the indomethacininduced expression of HSP70 by GGA were not observed. These results strongly suggest that the HSP inducing activity of GGA contributes to its gastroprotective activity against NSAIDs. These findings should be clinically relevant because it was recently reported that orally administered GGA (at clinical doses) up-regulated HSP70 expression at the gastric mucosa in humans while simultaneously suppressing diclofenac-induced gastric damage (46). Thus, we propose that nontoxic HSP70 inducers are therapeutically beneficial for NSAID-induced gastric lesions.