MAP Kinase Cascades Are Activated in Astrocytes and Preadipocytes by 15-Deoxy-Δ12–14-prostaglandin J2 and the Thiazolidinedione Ciglitazone through Peroxisome Proliferator Activator Receptor γ-independent Mechanisms Involving Reactive Oxygenated Species*

15-Deoxy-Δ12–14-prostaglandin J2 (dPGJ2) and thiazolidinediones are known as ligands for the peroxisome proliferator activator receptor γ (PPARγ) a member of the nuclear receptor superfamily. Herein, we show that dPGJ2 activates, in cultured primary astrocytes, Erk, Jnk, p38 MAP kinase, and ASK1, a MAP kinase kinase kinase, which can be involved in the activation of Jnk and p38 MAP kinase. The activation kinetic is similar for the three MAP kinase. The activation of the MAP kinases is detectable around 0.5 h. The activation increases with dPGJ2 in a dose dependent manner (0–15 μm). A scavenger of reactive oxygenated species (ROS), N-acetylcysteine (NAC) at 20 mm, completely suppresses the activation of MAP kinases and ASK1, suggesting a role for oxidative stress in the activation mechanism. Other prostaglandin cyclopentenones than dPGJ2, A2, and to a lesser degree, A1 also stimulate the MAP kinases, although they do not bind to PPARγ. Ciglitazone (20 μm), a thiazolidinedione that mimics several effects of dPGJ2 in different cell types, also activates the three MAP kinase families and ASK1 in cultured astrocytes. However the activation is more rapid (it is detectable at 0.25 h) and more sustained (it is still strong after 4 h). NAC prevents the activation of the three MAP kinase families by ciglitazone. Another thiazolidinedione that binds to PPARγ, rosiglitazone, does not activate MAP kinases, indicating that the effect of ciglitazone on MAP kinases is independent of PPAR γ. Ciglitazone and less strongly dPGJ2 activate Erk in undifferentiated cells of the adipocyte cell line 1B8. Ciglitazone also activates Jnk and p38 MAP kinase in these preadipocytes. Our findings suggest that a part of the biological effects of dPGJ2 and ciglitazone involve the activation of the three MAP kinase families probably through PPARγ-independent mechanisms involving ROS.

15-Deoxy-⌬ 12-14 -prostaglandinJ2 (dPGJ2) 1 has been shown to bind and to activate peroxisome proliferator-activated recep-tor ␥ (PPAR␥). This receptor, a member of the ligand-regulated nuclear receptor family, is a molecular target of the thiazolidinedione class of antidiabetic drugs (1,2), including ciglitazone, pioglitazone, and rosiglitazone. This receptor has a critical role in adipogenesis, glucose metabolism, placental function, and macrophage functions (3)(4)(5). It has been suggested that much of the effects of dPGJ2 and thiazolidinediones may be mediated by this receptor. However, Rossi et al. (6) have shown that dPGJ2 also can directly inhibit the IKK␤ subunit of IKK, the protein kinase responsible for the activation of NFB by proinflammatory stimuli. This effect of dPGJ2 can be mimicked by other prostaglandin cyclopentenones and is clearly independent of PPAR␥. dPGJ2 also activates Erk cascade in mesangial cells in a manner probably independent of PPAR␥ as ciglitazone does not mimic this effect (7).
In the cellular model studied by our group, rat cultured astrocytes, which express PPAR␥, we observed (this paper) 2 that dPGJ2 promotes apoptosis and prevents the induction of thyroid hormone deiodinases, enzymes strongly induced by signals activating protein kinase A or protein kinase C (8,9). To understand the mechanisms of action of dPGJ2, we examined the signaling pathways activated by this prostaglandin, and we focused on the mitogen-activated protein kinase (MAP kinase) cascades. Three MAP kinase families have been identified in mammalian cells: the extracellular regulated kinases (Erks), the c-Jun N-terminal kinases (Jnks), and the p38 MAPKs (for review, see Ref. 10). MAP kinases transduce information into the different compartments of the cell. The three MAP kinase families have different downstream targets that they phosphorylate on serine and threonine residues located adjacent to proline residues. The three MAP kinase families are activated by upstream dual-specificity kinases (MAP kinase kinases) phosphorylating threonine and tyrosine residues in a threonine-X-tyrosine site specific for each MAP kinase family. MAP kinase kinases are activated by specific MAP kinase kinase kinases. The three MAP kinase families are activated in response to various stimuli. The Erk family is activated preferentially by growth factors usually acting through a Ras-Raf-dependent cascade, while Jnk and p38 MAPK families are preferentially activated by stress and cytokines of the tumor necrosis factor ␣ family. However, the specificity of activating stimuli is relative, depending on the cell type. For example, in astrocytes Erk is strongly activated by a number of stress factors including oxidative stress (11). 3 We found that dPGJ2 and the thiazolidinedione ciglitazone activate the MAP kinase cascades in astrocytes and in preadipocytes. We demonstrated that (i) ROS generation is required for this effect, (ii) the effect of dPGJ2 on MAP kinases is mimed by other prostaglandin cyclopentenones that do not bind to PPAR␥, and (iii) the effect of ciglitazone on MAP kinases was not mimed by an other thiazolidinedione ligand of PPAR␥, rosiglitazone.

Materials
Sprague-Dawley rats were from our own breeding colony. Fetal calf serum (FCS), Dulbecco's modified Eagle's medium (DMEM), and Ham's F-12 culture mediums were from Invitrogen. dPGJ2 was from Biomol and Cayman. Ciglitazone, rosiglitazone, and prostaglandins A 1 and A 2 were from Cayman. N-Acetylcysteine (NAC) and 4Ј,6-diamidino-2-phenylindole (DAPI) were from Sigma. Antibodies against active forms (phosphorylated forms) of Erk, Jnk, and p38 MAPK were purchased from Promega. Antibody against ASK1 was from Santa Cruz Biotechnology. Horseradish peroxidase-conjugated secondary antibody was from Vector Laboratories. PVDF membranes were furnished by PerkinElmer Life Sciences. Chemiluminescence systems were purchased from Amersham Biosciences and PerkinElmer Life Sciences. All others products were of quality grade.

Methods
Cell Cultures-Astrocytes were prepared as described previously (12). Briefly, astroglial cells were obtained from the cerebral hemispheres of 2-day-old Sprague-Dawley rats. They were plated in 10-cm 2 dishes and grown at 37°C, 5% CO 2 , in DMEM, supplemented with: 6 g/liter glucose, 2.4 g/liter sodium bicarbonate, antibiotics (100 units/ml penicillin, 100 g/liter streptomycin, and 0.25 g/liter amphotericin B), and 10% FCS. The medium was changed every 2-3 days until the cells reached confluence at about 10 days. The cells were further cultured for a week, with daily changes, in a 1:1 mixture of DMEM and Ham's F-12 medium supplemented with 5.2 g/liter glucose, 1.8 g/liter sodium bicarbonate and the antibiotics listed above. Ninety-five percent of these cells contained, as we have described previously (13), the immunoreactive glial fibrillary acidic protein, a specific marker of astrocytes.
The immortalized mouse adipocyte cell line, termed 1B8, was grown in standard adipocyte culture medium supplemented with 10% FCS (14) until confluence (preadipocytes). After cells reached confluence, FCS content was reduced to 5% and supplemented with 1 nM triiodothyronine and 20 nM insulin until complete differentiation into adipocytes about 1 week later (15). The serum concentration was reduced to 0.5% for 24 h in the culture medium of adipocytes and preadipocytes before treatment by dPGJ2 and ciglitazone.
Protein Assays-The protein content of the samples in Laemmli buffer (16) was determined by the method of Mc Knight (17), using bovine serum albumin as standard.
Immunoblotting-Cultures were treated with different agents, and at the end of the incubation period, cells were scraped off in Laemmli buffer. Identical amounts of proteins from each sample were separated by SDS-PAGE and then transferred onto PVDF membranes by semidry transfer. The membranes were blocked with 3% bovine serum albumin in PBS, 0.1% Tween.
Immunodetections were performed by incubating the PVDF membrane with antibodies in PBS, 0.1% Tween and 1% bovine serum albumin. Three antibodies were directed against the phosphorylated forms (active forms) of MAP kinases: anti-active Erk (P-Erk) (1/5000), antiactive Jnk (P-Jnk) (1/5000), and anti-active p38 MAPK (P-p38) (1/2000). Another antibody was directed against and ASK1 (1/1000). Detection of antigen-antibodies complexes was performed with horseradish peroxidase-conjugated secondary antibody and the revelation was made by chemiluminescence reaction. All the experiments were repeat at least three times.
Apoptosis-Apoptosis was evaluated by nuclear staining with DAPI. Cells were treated with the different agents for 24 h, fixed in 4% paraformaldehyde, and permeabilized with 0.1% Triton X-100. Cell nuclei were stained with DAPI (0.1 g/ml). Nuclei were visualized by fluorescence microscopy. Nuclei of apoptotic cells showed characteristic morphological changes (condensation and fragmentation). Apoptotic cells and nonapoptotic cells were counted. The results are expressed as percent of apoptotic cells RESULTS MAP Kinase Activation by dPGJ2-Activation of Erk, Jnk, and p38 MAP kinase was observed after addition of dPGJ2 to cultured astrocytes. The activation kinetic obtained with 15 M dPGJ2 was similar for the three MAP kinases (Fig. 1A). This activation was delayed about half an hour and reached its maximum after 2 h. Then, the activation declined and returned to the basal level after 6 h. The vehicle, 0.1% Me 2 SO did not stimulate the MAP kinases. The MAP kinase activation was concentration dependent. Fig. 1B shows the activation of Erk, Jnk, and p38 MAP kinase after addition of increasing concentrations of dPGJ2 (0 -15 M) to cultured astrocytes for 2 h. For each MAP kinase family, the maximum activation was obtained at 15 M dPGJ2.
For a number of stimuli (18 -24), activation of MAP kinases requires the generation of reactive oxygenated species (ROS). Accumulation of ROS can be prevented by NAC, a scavenger of ROS. Cultured astrocytes were pretreated 15 min with 20 mM NAC before the addition of dPGJ2 (15 M) to the culture medium. Fig. 1C shows that NAC completely prevents the activation of MAP kinases by dPGJ2, suggesting that the formation of ROS is involved. NAC does not affect the phosphorylation of MAP kinases in unstimulated cells (not shown).
Other Prostaglandin Cyclopentenones Activate MAP Kinase Cascades-dPGJ2 contains an ␣,␤-unsaturated carbonyl group in the cyclopentane ring structure like prostaglandins A and other prostaglandins J. These prostaglandins are named prostaglandin cyclopentenones, and most of them share various biological effects (25), although several of them, like prostaglandins A, do not bind PPAR␥. We tested whether prostaglandins A also activate MAP kinase cascades. Fig. 2 shows that both prostaglandins A 2 and A 1 are able to activate MAP kinases. However, they are weaker activators than dPGJ2. NAC also inhibits MAP kinase activation by prostaglandins A 1 and A 2 .
MAP Kinase Activation by Ciglitazone-As reported for other cells, the thiazolidinedione ciglitazone promoted similar effects on astrocytes than dPGJ2. By example, ciglitazone induced apoptosis like dPGJ2, although its effect was slower than that of dPGJ2 (Table I). These observations prompted us to examine the effect of ciglitazone on MAP kinase cascades.
Activation of Erk, Jnk, and p38 MAP kinase was observed after addition of 20 M ciglitazone to cultured astrocytes (Fig.  3A). The kinetic was similar for the three families of MAP kinases. Activation was detectable after 15 min and was sustained for several hours. The vehicle, 0.1% Me 2 SO, did not stimulate the MAP kinases as shown in Fig 1A. In this particular experiment, two forms of activated p38 MAP kinase were seen clearly, although in most of the experiments only one form was easily detectable. Also, in some experiments, an additional minor band was visible. We have not established the reason of these variations. The MAP kinase activation by ciglitazone is concentration-dependent. Fig. 3B shows the activation of Erk, Jnk, and p38 MAP kinase after addition of increasing concentrations of ciglitazone (0 -20 M) to cultured astrocytes for 2 h. For each MAP kinase family, the maximum activation was obtained with 20 M ciglitazone.
Cultured astrocytes were pretreated for 15 min with 20 mM NAC before the addition of ciglitazone (20 M) to the culture medium. Fig. 3C shows that NAC reduces the activation of MAP kinases by ciglitazone.
ASK1 Is Activated by dPGJ2 and Ciglitazone-As we have shown previously (24), the MAP kinase kinase kinase ASK1, which can promote the activation of Jnk and p38 MAP kinase, is expressed in cultured astrocytes. The rate of migration on SDS-PAGE of ASK1 from stressed cells is decreased, which probably means that it is activated. Fig. 4 shows that ASK1 from cultured astrocytes treated with ciglitazone or dPGJ2 also shifts on SDS-PAGE. In astrocytes treated with dPGJ2, all the molecules of ASK1 clearly shifted. This shift promoted by dPGJ2 is prevented by NAC. In astrocytes treated by ciglitazone, only a fraction of ASK1 molecules shifted, and this was not clearly prevented by NAC. These observations show that ASK1 is probably activated in astrocytes by ciglitazone or dPGJ2, but that the mechanisms of activation may be different.
Rosiglitazone, Another Ligand of PPAR␥, Does Not Activate MAP Kinase Cascades-The effect of rosiglitazone on MAP kinase cascades was studied and compared with that of ciglitazone. Astrocytes were incubated with various concentrations of rosiglitazone (2-20 M) for 2 h. Astrocytes were also incubated with 20 M rosiglitazone for various times (0.5-6 h). In all cases, no activation of MAP kinases was detected, although ciglitazone in the same cultures promoted a strong activation of MAP kinases. Fig. 5 illustrates the difference of the effects of ciglitazone and rosiglitazone on MAP kinase cascades.
Ciglitazone, and to a Lesser Degree dPGJ2, Activate MAP Kinases in Preadipocytes-Observations made with astrocytes prompted us to examine whether ciglitazone and dPGJ2 also activate the MAP kinase cascades in preadipocytes known to differentiate into adipocytes after treatment with ciglitazone or dPGJ2. Preadipocytes of the cell line 1B8, maintained in an undifferenciated state (see "Methods"), were incubated with ciglitazone (20 M) or dPGJ2 (15 M) for 2 h, conditions that promoted a strong activation of MAP kinases in cultured astrocytes. The activation of MAP kinases was monitored by Western blot. Fig. 6A shows that ciglitazone strongly activated Erk but also p38 MAP kinase and Jnk. dPGJ2 less strongly activated Erk and did not stimulate Jnk and p38 MAP kinase. These observations suggest that MAP kinase activation, particularly Erk activation, could play a role in the action of ciglitazone and dPGJ2. Note that neither ciglitazone nor dPGJ2 activated MAP kinases in differentiated adipocytes (Fig. 6B). DISCUSSION We show that dPGJ2 and ciglitazone both promote the activation of MAP kinases (Erk, Jnk, and p38 MAP kinase) in  15.9 Ϯ 5.14 cultured astrocytes and preadipocytes. Ciglitazone and dPGJ2 are known as ligands of PPAR␥, and their effects on MAP kinases might be mediated by this nuclear receptor. However, some works suggest that dPGJ2 and thiazolidinediones also may act by a nongenomic mechanism. By example Chawla et al. (26) have obtained PPAR␥ deficient macrophages by using embryonic stem cells from mice deficient of PPAR␥, and PPAR␥ ligands still exert anti-inflammatory effects in these macrophages. Our results indicate that the activation of MAP kinase cascades by dPGJ2 and ciglitazone in astrocytes probably involves PPAR␥-independent mechanisms, because a classical ligand of PPAR␥ as rosiglitazone does not activate MAP kinases. Today, from the examination of the structures of these molecules (ciglitazone and roziglitazone), we are not able to explain their different effect on MAP kinase activation. The observation that prostaglandins A 2 and A 1 , which do not bind PPAR␥ but are prostaglandin cyclopentenones like dPGJ2, also activate MAP kinase cascades is also in favor of PPAR␥-inde-  pendent mechanisms. Our results implicates that, to promote their biological effects, some thiazolidinediones act only through PPAR␥-dependent mechanisms, while other thiazolidinediones act through PPAR␥-dependent and -independent mechanisms. MAP kinases might be elements of the PPAR␥independent mechanisms.
Examination of the activation kinetic of MAP kinases shows that the effect of ciglitazone and dPGJ2 is rather rapid. Activation is easily detectable within 30 min after addition of dPGJ2 to the culture medium and after 15 min for ciglitazone. Maximum is reached after 2 h. Activation is more sustained with ciglitazone than with dPGJ2. This difference may be due to a different stability of these molecules in the culture medium, but other explanations are possible.
In previous work, we have shown that ASK1, a MAP kinase kinase kinase able to activate Jnk and p38 MAP kinase cascades, is present in astrocytes and is activated by oxidative stress (24). In the present work, we show that ASK1 is activated by ciglitazone and dPGJ2, but that the activation of ASK1 by ciglitazone is not complete and is not blocked by NAC. However, the activation of Jnk and p38 MAP kinase by ciglitazone and dPGJ2, like the activation of Erk, requires ROS, because the activation of all these cascades is prevented in the presence of the ROS scavenger, N-acetylcysteine. To explain that, we propose (i) activation of ASK1 by ciglitazone must involve a mechanism insensitive to NAC, and (ii) ciglitazone must activate another MAP kinase kinase kinase able to activate Jnk and p38 MAP kinase by a mechanism sensitive to NAC. The activation of Erk cascade involves MAP kinase kinase kinases different from those able to activate Jnk and p38 MAP kinase cascades (10). Note that the activation of MAP kinases by a number of stressors, including drugs (18 -24), appears to require ROS, suggesting that the activation of MAP kinases by ROS might be a rather general response to stress factors. ROS (H 2 O 2 ) is required also for MAP kinase activation by growth factors in some cellular models as vascular smooth muscle cells (27). However, in other cases, the activation of MAP kinases by various stimuli does not require ROS (28,29). 3 The concentrations of ciglitazone and dPGJ2 efficient to activate MAP kinases in astrocytes and preadipocytes are similar to those that promote adipogenesis (1, 2), modifications of glucose metabolism, and placental and macrophage functions (3)(4)(5). A role for MAP kinases in the actions of ciglitazone and dPGJ2 can be proposed. Erk is generally considered as an inhibitor of the adipogenesis (30). This may be due to the inhibition of the transcriptional activity of PPAR␥ phosphorylated on a consensus site for Erk (31). In contrast, it has been shown that p38 MAP kinase is required for the first steps of adipogenesis (32). The activation of p38 MAP kinase by ciglitazone in preadipocytes, as we have observed, can play a role in initiating their differentiation into adipocytes. Activation of Erk (by dPGJ2 and ciglitazone) can potentiate the proliferation of preadipocytes before the initiation of their differentiation.
In astrocytes, preadipocytes, and likely in other cell types, the activation of the three families of MAP kinases by dPGJ2 and ciglitazone must affect the expression of a number of genes involved in various cellular functions, because each MAP kinase cascade is known to control the expression of several genes. dPGJ2 also can act by inhibiting the NFB pathway (6), considered by many authors as a survival pathway. Consequently, depending on the cell types and environmental conditions, these drugs can affect multiplication, cell death, and differentiated functions. In the case of astrocytes cultured in chemically defined medium (DMEM/F-12), we observed that dPGJ2 and more slowly ciglitazone promoted cell death.
An article by Takeda et al. (33), published while this manuscript was in preparation, shows that Erk cascade is activated in vascular smooth muscle cells by dPGJ2, pioglitazone, and troglitazone. This work and our work, taken together, show that the activation of MAP kinases by dPGJ2 and some thiazolidinediones can be obtained in various types of cells. The activation of the three MAP kinase families by dPGJ2 and some thiazolidinediones suggests a new concept of the molecular mechanism of action of these drugs. This may drive to use thiazolidinediones in therapeutic treatments for their ability to activate MAP kinases cascades or, contrary to this, to look for a reduction of undesirable side effects.