Suppression of Apoptosis by All-trans-Retinoic Acid

Retinoic acid induces apoptosis of various cells, whereas little is known about its anti-apoptotic potential. In this report, we describe an anti-apoptotic property of all-trans-retinoic acid (t-RA) in mammalian cells. Mesangial cells exposed to hydrogen peroxide (H2O2) exhibited shrinkage of the cytoplasm, membrane blebbing, condensation of nuclei, and DNA fragmentation. Pretreatment with t-RA attenuated the morphologic and biochemical hallmarks of apoptosis. t-RA also inhibited apoptosis of mesangial cells triggered by pyrrolidine dithiocarbamate, whereas it did not prevent tumor necrosis factor-α-induced apoptosis. The anti-apoptotic effect against H2O2 was similarly observed in NRK49F fibroblasts, but not in Madin-Darby canine kidney epithelial cells and ECV304 endothelial cells. Mesangial cells exposed to H2O2 undergo apoptosis via the activator protein 1 (AP-1)-dependent pathway. We found that t-RA abrogated the H2O2-induced expression of c-fos/c-jun and activation of AP-1. Furthermore, t-RA inhibited H2O2-triggered activation of c-Jun N-terminal kinase (JNK), and dominant-negative inhibition of JNK attenuated the H2O2-induced apoptosis. These data disclosed the novel potential of retinoic acid as an inhibitor of apoptosis. The anti-apoptotic action of t-RA was ascribed, at least in part, to dual suppression of the cell death pathway mediated by JNK and AP-1.

Apoptosis of glomerular cells is observed in several types of glomerulonephritis (1)(2)(3)(4). The molecular mechanisms involved in the apoptotic process have not been identified yet, but several possibilities are postulated. During initiation and progression of inflammation, toxic substances elaborated by leukocytes may induce apoptosis of glomerular cells. Putative triggers include cytokines, nitric oxide, and reactive oxygen intermediates (ROI) 1 (5)(6)(7)(8). ROI play crucial roles in the generation of a broad array of human and experimental glomerular diseases (9). Using hydrogen peroxide (H 2 O 2 ) as a trigger, recent studies have shown that ROI induces apoptosis of glomerular mesangial cells (8,10,11).
Multiple signaling cascades may be involved in the H 2 O 2initiated apoptosis of glomerular cells. Pathways mediated by activator protein 1 (AP-1) are possible candidates. AP-1 is generally regarded as a redox-sensitive transcription factor (12). AP-1, mainly composed of either homodimers of c-Jun or heterodimers of c-Jun and c-Fos, binds to the particular cis element, 12-O-tetradecanoylphorbol-13-acetate response element (TRE), and initiates transcription of target genes (13). Several reports have shown the importance of c-Jun N-terminal kinase (JNK) and its substrate c-Jun in the signaling pathways to apoptosis. For example, exposure of cells to apoptotic stimuli including ultraviolet light, ␥-irradiation, tumor necrosis factor-␣ (TNF-␣), and ceramide triggers JNK activity (14 -17). Dominant-negative inactivation of SEK1 (JNK kinase), JNK, or c-Jun prevents certain apoptotic processes (14,15,(17)(18)(19). Furthermore, constitutive activation of the JNK-AP-1 pathway results in apoptotic cell death (19 -21).
In apoptosis of mesangial cells exposed to H 2 O 2 , activation of AP-1 also plays a crucial role. We previously reported that H 2 O2 induces expression of c-jun and activation of AP-1 (10). Down-regulation of c-Jun/AP-1 using either a dominant-negative mutant of c-jun, an antisense c-jun, or a pharmacological inhibitor of c-jun attenuated the H 2 O 2 -initiated apoptosis (10). Furthermore, suppression of c-jun expression and AP-1 activation by flavonoid quercetin and heparin was closely associated with attenuation of H 2 O 2 -induced apoptosis in mesangial cells (11,22).
Retinoic acid (RA) is an active metabolite of vitamin A and regulates a wide range of biological processes including cell proliferation, differentiation, and morphogenesis (23). The action of retinoids, including RA, is mediated by specific nuclear receptors, namely, retinoic acid receptors (RAR-␣, -␤, -␥) and retinoid X receptors (RXR-␣, -␤, -␥). RXRs form homodimers and heterodimers with RARs or other nuclear hormone receptors and function as transcriptional regulators. All-trans-RA (t-RA), for example, activates RAR-RXR heterodimers and exerts its biological actions via binding to particular cis response elements, retinoic acid response elements (24). In certain cell types, RA functions as a potent inhibitor of AP-1 (25). A previous study showed that t-RA inhibited serum-induced activation of AP-1 in mesangial cells (26).
RA is known to induce apoptosis in various cell types includ-ing tumor cells and embryonic cells. In contrast, little is known about its anti-apoptotic potential in mammalian cells. In the present report, we investigate whether and how t-RA modulates apoptosis mediated by AP-1. This study highlights especially the effect of t-RA on the H 2 O 2 -triggered apoptosis of mesangial cells in which the AP-1 pathway plays a crucial role.

Cells and Transfectants
Mesangial cells (SM43) were established from isolated glomeruli of a male Sprague-Dawley rat and identified as being of the mesangial cell phenotype as described before (27). The fibroblast cell line NRK49F, the epithelial cell line MDCK, and the endothelial cell line ECV304 were purchased from American Type Culture Collection (ATCC, Manassas, VA). All cells were maintained in Dulbecco's modified Eagle's medium/ Ham's F-12 (Life Technologies, Inc., Gaithersburg, MD) supplemented with 100 units/ml penicillin G, 100 g/ml streptomycin, 0.25 g/ml amphotericin B, and 10% fetal calf serum (FCS). Medium containing 1% FCS was generally used for experiments.
Nuclear factor-B (NF-B)-inactive mesangial cells were created as follows. SM43 mesangial cells were exposed to diluted retrovirus that introduces a super-repressor mutant of IB␣ (IB␣M) and a neomycin phosphotransferase gene (28). This retroviral vector was generated by transfection of the helper-free ecotropic packaging line ⍀E (29) with pLIB␣MSN (28). Stable infectants were selected in the presence of G418 (750 g/ml), and SM/IB␣M cells were established. SM/IB␣M cells exhibit blunted activation of NF-B in response to interleukin-1␤ and TNF-␣, when examined by electrophoretic mobility shift assay (30).

Assessment of Apoptosis
Microscopic Analyses-Morphologic examination was performed using a phase-contrast microscope. For fluorescence microscopy, cells were fixed with 4% formaldehyde in phosphate-buffered saline for 10 min and stained by Hoechst 33258 (10 g/ml; Sigma) for 1 h. Apoptosis was identified using morphological criteria including shrinkage of the cytoplasm, membrane blebbing, and nuclear condensation and/or fragmentation. In contrast to other cell types, MDCK cells undergoing apoptosis easily detach from the substratum. For this cell type, Hoechst analysis was performed using floating cells. To confirm that the major mechanism of cell death induced by H 2 O 2 , PDTC, and TNF-␣ is apoptosis, cells were stained with acridine orange (50 g/ml) and ethidium bromide (50 g/ml) for 10 min without fixation. The percentages of apoptosis (condensed/fragmented green nuclei) against total cell death (condensed/fragmented green nuclei ϩ orange nuclei) were evaluated by fluorescence microscopy. Assays were performed in quadruplicate.
Ladder Detection Assay-After the induction of apoptosis, both attached and detached cells (5 ϫ 10 5 cells/sample) were harvested and subjected to ladder detection assay, as described previously (11).
Trypan Blue Analysis-The final step of apoptosis, secondary necrosis (33), was evaluated by trypan blue exclusion. After the induction of apoptosis, both attached and detached cells were gently trypsinized and mixed with the same volume of 0.4% trypan blue solution (Sigma). Percentages of viable cells were evaluated by light microscopy. Assays were performed in quadruplicate.

Reporter Assay
The effect of t-RA on the activity of AP-1 was evaluated by a transient transfection assay as described before (10,11). In brief, using the calcium phosphate coprecipitation method, mesangial cells cultured in 24-well plates (1 ϫ 10 5 cells/well) were transiently transfected with an AP-1 reporter plasmid pTRE-LacZ (0.33 g/well) (34) or a control plasmid pCI-␤Gal (0.33 g/well; a gift from Promega, Madison, WI). pTRE-LacZ introduces a ␤-galactosidase gene (lacZ) under the control of tandemly repeated TREs. pCI-␤Gal introduces lacZ under the control of the immediate-early enhancer/promoter of human cytomegalovirus. After transfection, cells were incubated for 48 h in 10% FCS in the presence or absence of t-RA (5 M) and subjected to 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) assay to evaluate AP-1 activity. To examine the effect of t-RA on the H 2 O 2 -induced activation of AP-1, transfected cells were incubated in 0.5% FCS for 24 h, pretreated with t-RA in 2.5% FCS for 2 h, and then stimulated by H 2 O 2 (100 M) for 40 h. Assays were performed in quadruplicate.
X-Gal assay was performed as described before (35). The number of X-gal-positive cells transfected with pTRE-LacZ was counted and normalized by the number of X-gal-positive cells transfected with pCI-␤Gal. The mean value of 4 wells was used to compare data in different groups.

Transient Transfection with a Dominant-negative Mutant of JNK
Mesangial cells cultured in 24-well plates were co-transfected with pcDNA3-DN-JNK1 (a gift of Dr. Roger Davis, University of Massachusetts Medical School) or an empty plasmid pcDNA3 (Invitrogen, San Diego, CA) (0.5 g/well, respectively) together with pCI-␤Gal (0.17 g/well). pcDNA3-DN-JNK1 encodes a dominant-negative mutant of JNK1 (36). After incubation overnight, medium was replaced with 1% FCS/Dulbecco's modified Eagle's medium/Ham's F-12. After 24 h, cells were treated with H 2 O 2 (100 -200 M, 8 -12 h) and subjected to X-gal assay. Percentage of shrunk/rounded blue cells against the total number of blue cells was calculated for each well, and the mean value of four wells was used to compare data in different groups (37). Assays were performed in quadruplicate.

Northern Blot Analysis
Expression of c-fos and c-jun was examined by Northern blot analysis (38). In brief, confluent mesangial cells cultured in the presence of 1% FCS were pretreated with t-RA (5 M) for 2 h and stimulated by 75-100 M H 2 O 2 for 30 min and 2 h. Total RNA was extracted by the single-step method (39) and subjected to analysis. cDNAs for c-Fos (40), c-Jun (41), and glyceraldehyde-3-phosphate dehydrogenase (42) were used as probes.

JNK Assay
Confluent mesangial cells cultured in 6-well plates in the presence of 1% FCS for 24 h were pretreated with t-RA (5 M) for 2 h and exposed to 100 M H 2 O 2 for 1 h. Activity of JNK was evaluated by phosphorylation of c-Jun, using the SAPK/JNK Assay Kit (New England Biolabs, Herts, United Kingdom) following the protocol provided by the manufacturer.

Statistical Analysis
All experiments were repeated at least twice. Data were expressed as mean Ϯ S.E. Statistical analysis was performed using the non-parametric Mann-Whitney U test to compare data in different groups. p Value of Ͻ 0.05 was used to indicate a statistically significant difference.

Suppression of H 2 O 2 -induced Apoptosis of Mesangial Cells by t-RA-Rat mesangial cells cultured in the presence of 1% FCS
were pretreated with t-RA (5 M) for 2 h and stimulated by H 2 O 2 (100 M). In the absence of t-RA, mesangial cells exposed to H 2 O 2 showed shrinkage of the cytoplasm, membrane blebbing, and condensation of nuclei that are typical of apoptosis. The morphological alteration occurred within several hours. Acridine orange-ethidium bromide staining confirmed that the major mechanism of cell death (75.3 Ϯ 3.7%, after 3 h) was apoptosis. Pretreatment with t-RA substantially inhibited these morphologic changes (Fig. 1A). Staining of the cells with Hoechst 33258 exhibited condensation and fragmentation of nuclei in H 2 O 2 -exposed cells, whereas it was dramatically sup-pressed by the treatment with t-RA (Fig. 1B). Consistently, agarose gel electrophoresis detected DNA ladder formation in H 2 O 2 -exposed cells, which was markedly attenuated by treatment with t-RA (Fig. 1C).
The apoptotic process is divided into three phases. In the first and second phases, function of cellular membranes is retained intact, but in the third phase, cell membranes are progressively degenerated (33). The final step of apoptosis was, therefore, evaluated by trypan blue exclusion. Confluent mesangial cells were pretreated with or without t-RA and stimulated by H 2 O 2 for 16 h. After the induction of apoptosis, both attached and detached cells were gently trypsinized and used for the analysis. When exposed to H 2 O 2 , the percentage of viable cells was reduced from 89.5 Ϯ 0.6% to 17.0 Ϯ 3.4% (mean Ϯ S.E.) (Fig. 1D). Pretreatment with t-RA significantly improved the cell survival to 69.0 Ϯ 2.1% (p Ͻ 0.05). The cytoprotective effect of t-RA was dose-dependent. Obvious improvement in cell survival was observed at concentrations higher than 1 M, and a maximum effect was achieved by 5 M t-RA (Fig. 1E).
Effect of t-RA on Apoptosis of Mesangial Cells Triggered by Other Stimuli-The anti-apoptotic potential of t-RA was investigated using different stimuli. PDTC is known to induce apoptosis in certain cell types (43)(44)(45). The pro-apoptotic action of PDTC is supposed to be via activation of AP-1 and/or inactivation of NF-B (31,32). Mesangial cells were pretreated with t-RA and stimulated by PDTC (10 -20 M) in the presence of 1% FCS. Mesangial cells exposed to PDTC showed shrinkage of the cytoplasm. Acridine orange-ethidium bromide staining confirmed that the major mechanism of cell death (82.0 Ϯ 4.4%

FIG. 1. Suppression of hydrogen peroxide (H 2 O 2 )-induced apoptosis of mesangial cells by t-RA.
A, phase-contrast microscopy. Confluent rat mesangial cells (SM43) were pretreated with (ϩ) or without (Ϫ) t-RA (5 M) for 2 h in the presence of 1% FCS, exposed to H 2 O 2 (100 M) for 5 h, and subjected to analysis. B, Hoechst staining. After the induction of apoptosis, cells were stained by Hoechst 33258 and examined by fluorescence microscopy. C, ladder detection assay. Mesangial cells were pretreated with or without t-RA for 2 h, exposed to H 2 O 2 (75 or 100 M) for 24 h, and subjected to agarose gel electrophoresis. D, trypan blue analysis. Confluent mesangial cells were pretreated with or without t-RA and exposed to H 2 O 2 (100 M). After 16 h, both attached and detached cells were gently trypsinized and used for trypan blue analysis. Data are expressed as mean Ϯ S.E. Asterisks indicate statistically significant differences (p Ͻ 0.05). Assays were performed in quadruplicate. E, dose-dependent effect of t-RA on mesangial cell survival. Cells were pretreated with 0.1, 0.5, 1, 5, or 7.5 M t-RA for 2 h, exposed to H 2 O 2 (100 M) for 24 h, and subjected to trypan blue analysis. after 16 h) was apoptosis. Pretreatment with t-RA reversed the morphologic change ( Fig. 2A). Staining of the cells with Hoechst 33258 exhibited condensation and fragmentation of nuclei in PDTC-treated cells. It was suppressed by treatment with t-RA (Fig. 2B, left panel). The relative percentages of apoptotic cells were significantly reduced from 16.8 Ϯ 1.6% (PDTC alone) to 1.0 Ϯ 0.3% (t-RA ϩ PDTC) (versus untreated control, 0.8 Ϯ 0.3%) (Fig. 2B, right panel). Consistently, agarose gel electrophoresis detected DNA fragmentation in PDTCexposed cells, and it was attenuated by treatment with t-RA (Fig. 2C).
We further tested the effect of t-RA on apoptosis triggered by another apoptosis inducer, TNF-␣ (46). Like other cell types, cultured mesangial cells are resistant to TNF-␣-induced apoptosis. It is due to induction of anti-apoptotic proteins by TNF-␣ via NF-B-dependent mechanisms (30,47). To sensitize mesangial cells to TNF-␣-induced apoptosis, we created NF-B-inactive mesangial cells, SM/IB␣M, by expression of a superrepressor mutant of IB␣, IB␣M. The established SM/IB␣M cells exhibited substantial susceptibility to TNF-␣-induced cellular injury (30). Acridine orange-ethidium bromide staining confirmed that the major mechanism of cell death (75.8 Ϯ 2.5% after 16 h) was apoptosis. Using the established cells, the effect of t-RA was tested. Microscopic analysis showed that, in contrast to H 2 O 2 -and PDTC-initiated apoptosis, t-RA did not affect morphological changes (shrinkage and round-up of the cells) induced by TNF-␣ (250 units/ml) (Fig. 2D). Like in wildtype mesangial cells, H 2 O 2 -induced damage was attenuated by t-RA in SM/IB␣M cells. Hoechst staining and agarose gel electrophoresis exhibited consistent results. That is, (i) condensation and fragmentation of nuclei induced by TNF-␣ was not attenuated by t-RA (percentages of apoptotic cells: 28.2 Ϯ 2.1% by TNF-␣ alone, and 25.9 Ϯ 2.9% by t-RA ϩ TNF-␣, not statistically different) (Fig. 2E) and (ii) DNA fragmentation in-

FIG. 2. Effect of t-RA on apoptosis of mesangial cells triggered by other stimuli.
A, phase-contrast microscopy. Mesangial cells were pretreated with (ϩ) or without (Ϫ) t-RA (5 M) for 2 h in the presence of 1% FCS and exposed to pyrrolidine dithiocarbamate (PDTC; 20 M) for 24 h. B, Hoechst staining. After the induction of apoptosis (10 M PDTC), cells were stained by Hoechst 33258. Percentages of condensed and/or fragmented nuclei are shown on the right, mean Ϯ S.E. An asterisk indicates a statistically significant difference (p Ͻ 0.05). C, ladder detection assay. Mesangial cells were pretreated with or without t-RA for 2 h, exposed to PDTC (10 M) for 24 h, and subjected to agarose gel electrophoresis. D, phase-contrast microscopy. Nuclear factor-B-inactive mesangial cells, SM/IB␣M, were pretreated with (ϩ) or without (Ϫ) t-RA (5 M) for 2 h in the presence of 1% FCS and exposed to TNF-␣ (250 units/ml) or H 2 O 2 (100 M) for 24 h. E, Hoechst staining. After the induction of apoptosis, SM/IB␣M cells were stained by Hoechst 33258 and examined by fluorescence microscopy. F, ladder detection assay. SM/IB␣M cells were pretreated with or without t-RA for 2 h, exposed to TNF-␣ or H 2 O 2 for 24 h, and subjected to agarose gel electrophoresis. duced by TNF-␣ was unaffected by the pretreatment with t-RA (Fig. 2F). In contrast, DNA laddering induced by H 2 O 2 was inhibited by t-RA in SM/IB␣M cells.

Effect of t-RA on Apoptosis in Other Cell Types Triggered by H 2 O 2 -
To examine whether the anti-apoptotic effect of t-RA against H 2 O 2 is specific to mesangial cells, NRK49F fibroblasts, MDCK epithelial cells, and ECV304 endothelial cells were tested. Dose-dependent effects of H 2 O 2 on individual cell type was initially examined to determine minimum concentrations required for cellular damage. Compared with mesangial cells, NRK49F, MDCK, and ECV304 cells were found to be relatively resistant to H 2 O 2 -induced injury. The minimum concentrations required were 150 -200 M for NRK49F cells, 400 M for MDCK cells, and 200 -400 M for ECV304 cells (data not shown). Using these concentrations, effects of t-RA on H 2 O 2induced apoptosis were examined. NRK49F fibroblasts exposed to H 2 O 2 exhibited shrinkage of the cytoplasm, membrane blebbing, and condensation of nuclei. Pretreatment with t-RA substantially inhibited these morphologic changes (Fig. 3A). Hoechst staining showed condensation and fragmentation of nuclei in H 2 O 2 -exposed cells, whereas it was suppressed by treatment with t-RA (Fig. 3B, left panel). The percentages of apoptotic cells were significantly reduced from 35.0 Ϯ 5.9% (H 2 O 2 alone) to 15.0 Ϯ 1.7% (t-RA ϩ H 2 O 2 ) (versus untreated control, 3.7 Ϯ 1.1%) (Fig. 3B, right panel). Consistently, agarose gel electrophoresis detected DNA ladder formation in H 2 O 2 -exposed NRK49F cells, and it was attenuated by treatment with t-RA (Fig. 3C).
In contrast to mesangial cells and NRK49F fibroblasts, t-RA did not diminish H 2 O 2 -induced apoptosis of MDCK cells. Morphological analysis, Hoechst staining, and agarose gel electrophoresis showed typical features of apoptosis in H 2 O 2 -exposed MDCK cells, and the apoptotic process was not affected by the pretreatment with t-RA (Fig. 3, D-F). The percentages of apoptotic cells were 17.0 Ϯ 0.8% in H 2 O 2 alone and 14.7 Ϯ 1.0% in t-RA ϩ H 2 O 2 (Fig. 3E, right panel, not statistically different). Similar unresponsiveness to t-RA was observed in ECV304 endothelial cells (data not shown).
Effect of t-RA on the JNK-AP-1 Pathway-Activation of AP-1 is a crucial signaling event that mediates H 2 O 2 -induced apoptosis in mesangial cells (10,11). We examined the effect of t-RA on the activity of AP-1, especially focusing on expression of AP-1 components and activation of JNK. In the presence of serum (10%), mesangial cells exhibit constitutive AP-1 activity. Reporter assays showed that t-RA (5 M) significantly suppressed the basal activity of AP-1 (Fig. 4A). Compared with the untreated control (100 Ϯ 9.5%), the activity of AP-1 was decreased to 48.2 Ϯ 5.4% by the treatment with t-RA.
The effect of t-RA on the oxidant-induced activation of AP-1 was further examined by reporter assays. In response to H 2 O 2 , mesangial cells exhibited up-regulation of AP-1 activity (196.4 Ϯ 21.7%). Pretreatment with t-RA abrogated the H 2 O 2induced activation of AP-1 (106.0 Ϯ 7.2%) (Fig. 4B).
To identify molecular mechanisms involved in the suppressive action of t-RA on AP-1, its effect on the expression of c-fos and c-jun was examined. Mesangial cells were pretreated with or without t-RA for 2 h and stimulated by H 2 O 2 for 0.5 and 2 h. Northern blot analysis detected substantial induction of c-fos and c-jun mRNAs in response to H 2 O 2 . Pretreatment with t-RA completely abolished the oxidant-induced expression of c-fos and c-jun (Fig. 4C).
The activity of AP-1 is regulated by phosphorylation-dependent activation by JNK. We therefore examined whether or not t-RA affects the activity of JNK. Mesangial cells were pretreated with or without t-RA, stimulated by H 2 O 2 for 1 h and subjected to the JNK assay. After stimulation with H 2 O 2 , substantial induction of JNK activity was observed. Pretreatment with t-RA markedly diminished the activation of JNK in response to H 2 O 2 (Fig. 4D).
To examine whether JNK is required for the H 2 O 2 -induced apoptosis, mesangial cells were transiently co-transfected with an empty plasmid or an expression plasmid encoding a dominant-negative mutant of JNK1 together with pCI-␤Gal that introduces a ␤-galactosidase gene. After incubation for 24 h in the presence of 1% FCS, cells were treated with H 2 O 2 for 12 h and subjected to X-gal assay. Percentages of shrunk/rounded blue cells (apoptotic cells) against total numbers of blue cells were evaluated. As shown in Fig. 4E, treatment with H 2 O 2 significantly increased round cells in mock-transfected cells (2.4 Ϯ 0.2 fold versus untreated control, p Ͻ 0.05). In contrast, in the cells transfected with the JNK mutant, significant increase of apoptotic cells was not observed after exposure to H 2 O 2 (1.1 Ϯ 0.1-fold versus untreated control). DISCUSSION RA has been considered as a potential therapeutic agent for malignant diseases, especially for the treatment of leukemia (25). It is based on the pharmacological potential of RA to induce growth arrest, cellular differentiation, and apoptosis (48). RA triggers apoptosis of a variety of cell types including embryonic cells and tumor cells. In contrast, little is known about anti-apoptotic action of RA. Previous studies have shown that RA may inhibit apoptosis of T cells, leukemic cells, and hematopoietic cells (49 -52). Currently, it is unknown whether RA inhibits apoptosis of non-leukocyte lineage. Using H 2 O 2 as a trigger, the present report provides novel evidence for the anti-apoptotic potential of RA. Our data showed that t-RA attenuates H 2 O 2 -induced apoptosis in mesangial cells and fibroblasts. The molecular mechanisms involved in its anti-apo-ptotic action was not fully elucidated, but the current results suggested that the JNK-AP-1 pathway is one of its potential targets. The fact that t-RA inhibited apoptosis triggered by RA has been generally regarded as an inhibitor of AP-1 (23). However, previous studies indicated that the manner of which RA affects the AP-1 pathway varies from cell type to cell type. For example, RA inhibits expression of c-fos and c-jun in synovial fibroblasts (53). In human bronchial epithelial cells, growth factor-induced activation of JNK is also inhibited by RA (54). However, in vascular smooth muscle cells, RA inhibits AP-1 activity without suppressing expression of c-fos and c-jun. (55). In human skin, RA inhibits ultraviolet-triggered accumulation of c-Jun via a post-transcriptional mechanism (56). RA suppresses endothelin-triggered activation of extracellular signal-regulated kinase, but not JNK in aortic smooth muscle cells (57). Furthermore, RA does not inhibit c-jun and c-fos expression and activity of AP-1 in activated myofibroblasts and monocytes (58,59). RA may rather up-regulate expression of c-fos/ c-jun and activity of AP-1 in certain cell types (60 -63). The effect of RA on the JNK-AP-1 pathway is thus different, depending on cell types and triggers. In the present report, we have shown that RA inhibits activation of JNK, expression of c-fos/c-jun, and up-regulation of AP-1 activity in H 2 O 2 -exposed mesangial cells. As we previously reported, activation of AP-1 plays a crucial role in the oxidant-induced apoptosis in mesangial cells (10,11). The current data suggested the dual intervention by t-RA in the AP-1-mediated, apoptotic pathway.
Biological actions of RA are mediated by RARs and RXRs. Yang et al. (64) showed that binding of both RARs and RXRs is required for efficient inhibition of T cell apoptosis by RA. Currently, it is unknown whether suppression of apoptosis by RA requires transactivation of retinoic acid response elements. Inhibition of AP-1 by RA may occur independently of retinoic acid response elements (65). For example, RA can inhibit activation of AP-1 via physical interaction of RAR⅐RXR complexes with c-Jun (53). In addition to its suppressive effects on JNK and c-fos/c-jun, sequestration of AP-1 proteins by RAR-RXR heterodimers (66) may be involved in the anti-apoptotic action of t-RA observed here.
Retinoids possess antioxidant activity (48). For example, RA inhibits lipid peroxidation by scavenging lipid peroxyl radicals (67,68). The cytoprotective action of RA against H 2 O 2 might be simply via scavenging cytotoxic ROI. Alternatively, RA may induce endogenous antioxidant enzymes. Recently, we showed that t-RA up-regulates catalase activity and the reduced form of glutathione (GSH) content via transcriptional up-regulation of catalase and ␥-glutamil-cysteine synthetase, the limiting enzyme of GSH synthesis (69). However, the following evidence seems to exclude these possibilities. (i) Pretreatment is required for the anti-apoptotic action of t-RA, but preincubation for short periods (less than 15 min) is sufficient to suppress the cytotoxic effect of H 2 O 2 . 2 (ii) t-RA did not inhibit H 2 O 2 -induced apoptosis in some cell types including MDCK epithelial cells and ECV304 endothelial cells. (iii) t-RA also inhibited apoptosis induced by another trigger, antioxidant PDTC.
RA might inhibit H 2 O 2 -induced apoptosis in other ways. It has been shown that, in particular cell types, RA activates NF-B, a potent anti-apoptotic molecule (70). t-RA could inhibit apoptosis of mesangial cells via up-regulation of NF-B. However, our data using NF-B-inactive mesangial cells and PDTC excluded this possibility. That is, the anti-apoptotic effect of t-RA was also observed in H 2 O 2 -stimulated SM/IB␣M cells similarly to that in H 2 O 2 -triggered wild-type mesangial cells. Furthermore, mesangial cell apoptosis induced by the NF-B inhibitor PDTC was significantly attenuated by t-RA.
A recent report has shown that mannose 6-phosphate/insulin-like growth factor-II receptor is a receptor for RA and that the binding of RA to the mannose 6-phosphate/insulin-like growth factor-II receptor enhances the primary functions of this receptor (71). Insulin-like growth factor-II is a prominent survival factor of mesangial cells exposed to cycloheximide, etoposide, or serum deprivation (72). The anti-apoptotic action of t-RA observed here might be due to activation of the survival pathway via the insulin-like growth factor-II receptor.
Interestingly, we found that TNF-␣-induced apoptosis was not inhibited by t-RA. This result may lead to some confusion, because (i) TNF-␣ induces apoptosis via generation of ROI (73)(74)(75), and (ii) t-RA suppresses ROI-induced apoptosis, as shown in this report. A possible explanation for this is that ROI other than H 2 O 2 , e.g. superoxide anion (O 2 . ), may be involved in the TNF-␣-induced apoptosis and that t-RA selectively inhibits the action of H 2 O 2 , but not other ROI. Our recent data support this possibility. That is, we found that scavengers of O 2 . , but not scavengers of H 2 O 2 , inhibited TNF-␣-induced apoptosis in mesangial cells. 3 Furthermore, in contrast to H 2 O 2 -triggered apoptosis, apoptosis induced by O 2 . releasing agents was not inhibited by t-RA. 2 Taken together, these data support the idea that t-RA suppresses the action of particular ROI including H 2 O 2 .
The reason for the lack of effects of t-RA on MDCK cells and ECV304 cells is unknown. As described above, the anti-AP-1 action of t-RA is different from cell type to cell type. The different responses to t-RA may be due to different effects of t-RA on the JNK-AP-1 pathway in individual cell types. Alternatively, different expression levels of RARs and RXRs might have caused the different responsiveness to t-RA. Further investigation is required to examine these possibilities.