15d-PGJ 2 and Rosiglitazone Suppress Janus Kinase-STAT Inflammatory Signaling through Induction of Suppressor of Cytokine Signaling 1 (SOCS1) and SOCS3 in Glia*

Peroxisome proliferator-activated receptor (PPAR)- (cid:1) agonists are now emerging as therapeutic drugs for various inflammatory diseases. However, their molecular mechanism of action remains to be elucidated. Here we report a novel mechanism that underlies the PPAR- (cid:1) agonist-mediated suppression of brain inflammation. We show that 15-deoxy- (cid:2) 12,14-prostaglandin J 2 (15d- PGJ 2 ) and rosiglitazone reduce the phosphorylation of STAT1 and STAT3 as well as Janus kinase 1 (JAK1) and JAK2 in activated astrocytes and microglia. The PPAR- (cid:1) agonist-mediated reduction in phosphorylation leads to the suppression of JAK-STAT-dependent inflammatory responses. The effects of 15d-PGJ 2 and rosiglitazone are not mediated by activation of PPAR- (cid:1) . 15d-PGJ 2 and rosiglitazone rapidly induce the transcription of suppressor of cytokine signaling (SOCS) 1 and 3, which in turn inhibit JAK activity in activated glial cells. In addition, Src homology 2 domain-containing protein phos-phatase 2 (SHP2), another negative regulator of JAK activity, is also involved in their anti-inflammatory action. Our data suggest that 15d-PGJ 2 and rosiglitazone suppress the initiation of JAK-STAT inflammatory signaling independently of PPAR-

Astrocytes and microglia are cells with immune functions in the central nervous system. They are rapidly activated in response to pathological stimuli and lead to various inflammatory processes. Upon activation, astrocytes and microglia change their immunophenotype as well as the expression pattern of inflammatory mediators including cytokines, chemokines, and neurotoxic substances at the transcriptional level (1,2). Activated glial cells can exert both beneficial and harmful actions on brain. Although transient activation is beneficial for defense processes, either chronic activation or overactivation may lead to or exacerbate neuronal diseases such as Alzheimer's disease, ischemia, and human immunodeficiency virus encephalitis (3). Inflammatory mediators are involved in many effects of activated glial cells. Whereas cytokines such as interleukin-10 reduce glial cytotoxicity (4), other cytokines such as interferon-␥ (IFN-␥) 1 cause activation of glial cells (5), eventually leading to neuronal injury via inflammatory cascade.
The JAK-STAT (Janus kinase-signal transducers and activators of transcription) cascade is an essential inflammatory signaling pathway that mediates immune responses. Specific subtypes of JAK and STAT are activated by different signals and transduce signals from cell surface receptors to different subsets of target genes, many of which are involved in immune responses (6,7). To prevent detrimental effects, the intensity and duration of JAK-STAT activation are tightly regulated. Recently, a family of proteins called suppressors of cytokine signaling (SOCS) has been isolated (8 -10). Generally, SOCS are present in cells at very low levels but are rapidly transcribed after exposure of cells to stimulus. SOCS can negatively regulate the response of immune cells either by inhibiting JAK activity or by competing with signaling molecules for binding to the phosphorylated receptor (11). For example, SOCS1 and SOCS3 bind to the JAKs and inhibit their tyrosine kinase activity (12). The ability of SOCS to suppress cytokine signaling in vivo has been confirmed by animal models including knock-out mice (13,14). SOCS family is now considered as important regulators of normal immune physiology and immune disease (15).
Peroxisome proliferator-activated receptor (PPAR)-␥ is a ligand-dependent nuclear receptor whose ligands include several prostanoids including 15-deoxy-⌬12,14-prostaglandin J 2 (15d-PGJ 2 ) (16) and antidiabetic thiazolidinediones such as rosiglitazone (17). Although PPAR-␥ was originally shown to play a key role in adipocyte differentiation and lipid metabolism (18), some PPAR-␥ ligands have recently been reported to exert anti-inflammatory actions by reducing inflammation-associated molecules such as cytokines and nitric oxides (19 -24). However, the precise mechanisms of action underlying the anti-inflammatory effects of PPAR-␥ agonists are poorly understood. Here we provide new insights into the anti-inflammatory actions of 15d-PGJ 2 and rosiglitazone based on their effects on JAK-STAT inflammatory signaling. We found that there is a PPAR-␥-independent link between the anti-inflammatory action of 15d-PGJ 2 and rosiglitazone and the induction of SOCS1 and SOCS3. Our findings suggest not only a novel molecular explanation for the therapeutic efficacy of PPAR-␥ agonists but also a new potential therapeutic intervention for inflammatory diseases.
Cell Culture-Primary astrocytes and microglia were cultured from the cerebral cortices of 1-day-old Sprague-Dawley rats. The cortices were triturated into single cells in minimal essential medium (Invitrogen) containing 10% fetal bovine serum and incubated for 2-3 weeks. Microglia were detached from the flasks, and the cells remaining in the flask were harvested with 0.1% trypsin to prepare pure astrocytes. BV2 murine microglia cells were obtained from Dr. E. J. Choi and maintained in Dulbecco's modified Eagle's medium (Invitrogen) with 5% fetal bovine serum.
Real-time RT-PCR-For transcript quantification purposes, the LightCycler TM system (Roche Molecular Biochemicals) was used. The mRNA levels were determined using LightCycler-RNA Master SYBR Green1 or LightCycler-DNA Master SYBR Green1. Glucose-6-phosphate dehydrogenase housekeeping gene set (Roche Molecular Biochemicals) was used as external standards, and the variability in the initial quantities of cDNA was normalized to the internal control, glyceraldehyde-3-phosphate dehydrogenase. Negative control was included in each set of experiments. Melting curve analysis was performed to enhance specificity of amplification reaction, and the LightCycler software, version 3.5, was used to compare amplification in the experimental samples during the log-linear phase to the standard curve.
Western Blot Analysis-Cell lysates were separated by SDS-PAGE and transferred to nitrocellulose membrane. The membrane was incubated with primary antibodies and peroxidase-conjugated secondary antibodies (Vector Laboratories, Burlingame, CA) and then visualized using an enhanced chemiluminescence system (Promega). For Western blot analysis and RT-PCR, cells were serum-starved at least 24 h before drug treatment.
Luciferase Assay-Transient transfections were performed in duplicate on 35-mm dishes using LipofectAMINE Plus reagents (Invitrogen). To normalize the variations in cell number and transfection efficiency, all of the cells were co-transfected with pCMV-␤-galactisodase.

RESULTS
To better define the mechanism of action by which PPAR-␥ agonists suppress inflammation, we examined the effects of 15d-PGJ 2 and rosiglitazone on JAK-STAT inflammatory signaling in rat brain. Pretreatment of 15d-PGJ 2 or rosiglitazone for 2 h markedly suppressed the IFN-␥-stimulated phosphorylation of STAT1 and STAT3 in rat primary astrocytes (Fig. 1a). At all of the time points tested from 2 min to 2 h after IFN-␥ treatment, phosphorylation of STAT1 and STAT3 was strongly reduced by treatment with either of the PPAR-␥ agonists (data FIG. 1. PPAR-␥ agonists reduce the phosphorylation of JAKs and STATs. a, primary astrocytes were pretreated with 15d-PGJ 2 (10 M) or rosiglitazone (20 M) for 2 h and stimulated with IFN-␥ (10 units ml Ϫ1 ) for 5 min. Western blots were probed with pSTAT1 (Tyr-701) and pSTAT3 (Tyr-705) antibody. The membrane was then stripped and analyzed with STAT1 antibody to determine loading. b, the phosphorylated level of JAK1 and JAK2 was determined by phospho-JAK1 or JAK2 antibody. c, rat primary astrocytes were stimulated with lipopolysaccharide (100 ng ml Ϫ1 ) for 2 h in the presence of 15d-PGJ 2 (10 M) or rosiglitazone (20 M). LPS, lipopolysaccharide; 15d, 15d-pGJ 2 ; Ro, rosiglitazone.

FIG. 2. PPAR-␥ agonists also suppress the events downstream of JAK-STAT activation.
a, primary astrocytes were transfected with 8-GAS luciferase reporter plasmid and pCMV-␤-galactosidase, and the cells were treated with combinations of 15d-PGJ 2 (10 M), rosiglitazone (20 M), and/or IFN-␥ (10 units ml Ϫ1 ) for 18 h. The cell extract was assayed for luciferase activity and ␤-galactosidase activity. b and c, primary astrocytes were pretreated with 15d-PGJ 2 (10 M), and rosiglitazone (20 M) for 2 h and then stimulated with IFN-␥ (10 units ml Ϫ1 ) for 2 h. Total RNA was analyzed for the messenger levels of indicated genes. d, primary astrocytes were pretreated with 15d-PGJ 2 (10 M) and rosiglitazone (20 M) for 2 h and stimulated with IFN-␥ (10 units ml Ϫ1 ) for 1 h. Western blot analysis was performed with interferon regulatory factor-1 and actin antibody. MCP, monocyte chemoattractant protein-1; IRF, interferon regulatory factor-1; Con, control; IL, interleukin; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; 15d, 15d-pGJ 2 ; Ro, rosiglitazone. not shown). We then investigated whether PPAR-␥ agonists affected the activation of JAK1 and JAK2, because phosphorylation of STATs mainly depends on the activation of JAKs (26). The phosphorylation of JAK1 and JAK2 by IFN-␥ were reduced in the presence of PPAR-␥ agonists (Fig. 1b). We obtained similar results with another activator of brain inflammation, lipopolysaccharide. Stimulation with lipopolysaccharide resulted in phosphorylation of STAT1 and STAT3, but this was significantly inhibited by pretreatment with PPAR-␥ agonists (Fig. 1c). Similar effects were observed in primary microglia and C6 astroglioma cells (data not shown). These results show that 15d-PGJ 2 and rosiglitazone can inhibit the activation of JAK-STAT signaling by reducing their phosphorylation.
We further investigated the events downstream of STAT activation to evaluate the functional importance of the reduction in phosphorylation by 15d-PGJ 2 and rosiglitazone. First, we examined the ␥-interferon-activated sequence (GAS) promoter activity in primary astrocytes, because activated STAT dimers bind to GAS elements. Transient transfection analysis showed that PPAR-␥ agonists significantly reduced the IFN-␥ responsiveness of the GAS promoter (Fig. 2a). We then examined the mRNA level of genes whose promoters have binding sites for STATs and act as mediators of inflammation, namely monocyte chemoattractant protein-1 (27) and interferon-inducible protein-10 (28). IFN-␥ rapidly induced the transcription of both genes, but this induction was inhibited by pretreatment with 15d-PGJ 2 or rosiglitazone (Fig. 2b). Similar patterns of transcriptional repression were observed for pro-inflammatory cytokines such as tumor necrosis factor-␣ and interleukin-1␤ (Fig. 2c). We further tested the effects of PPAR-␥ agonists at the level of protein expression. As expected, PPAR-␥ agonists suppressed the IFN-␥-stimulated induction of interferon regulatory factor-1 protein, which is a regulator of host defense and has binding sites for STAT in its promoter (Fig. 2d) (29). These observations convincingly demonstrate that PPAR-␥ agonists can exert inhibitory actions on JAK-STAT inflammatory signaling, thus regulating brain inflammation.
An important question raised by the above findings is how PPAR-␥ agonists influence JAK-STAT signaling. To address this question, we examined whether PPAR-␥ mediates the inhibitory action of its agonists on JAK-STAT signaling. First, we determined the existence of PPAR-␥ in rat primary astrocytes and microglia. Even in unstimulated cells, PPAR-␥ transcripts could be detected (Fig. 3a and data not shown). We then compared the inhibitory effects of PPAR-␥ agonists on JAK-STAT signaling between cells transiently transfected with PPAR-␥ and vector. Interestingly, we did not observe any differences of GAS-luciferase activity in cells with PPAR-␥ compared with cells with vector (Fig. 3b). Different concentrations either of PPAR-␥ constructs or of drugs also did not affect the GAS activity (data not shown). In contrast, a significant activation of PPAR-␥-responsive element (PPRE)-luciferase was observed in the same experimental conditions (Fig. 3b). Thus, overexpression of PPAR-␥ did not lessen GAS activity but enhanced PPRE activity, indicating that PPAR-␥ agonists act independently of PPAR-␥ activation in the JAK-STAT signaling pathway. Overexpression of PPAR-␥ consistently failed to improve the inhibitory effects of PPAR-␥ agonists on JAK-STAT phosphorylation. In cells transiently transfected with vector, PPAR-␥, or PPAR-␥, S112A (a PPAR-␥ active mutant) (30), PPAR-␥ agonists exerted equivalent effects on the phosphorylation of Western blots were probed with pSTAT1, pSTAT3, or STAT1 antibody. Ro, rosiglitazone. b and c, the transcript level of indicated molecules in BV2 cells was analyzed by RT-PCR in the same conditions of Fig. 2, b  and c. 15d, 15d-pGJ 2 . LPS, lipopolysaccharide; Con, control; ICAM, intercellular adhesion molecule 1.
FIG. 5. 15d-PGJ 2 promptly induces the mRNA expression of SOCS1 and SOCS3 in primary astrocytes. The mRNA levels of SOCS1 and SOCS3 were determined by quantitative real-time PCR analysis with external standards (glucose-6-phosphate dehydrogenase) and internal control (glyceraldehyde-3-phosphate dehydrogenase) as described under "Experimental Procedures." The graph represents relative RNA levels of SOCS1 (ࡗ) and SOCS3 (q) from LightCycler RT-PCR results using LightCycler software, version 3.5. STATs (Fig. 3c). To further confirm this effect, we examined the effects of PPAR-␥ agonists in murine BV2 microglial cells that are reported to lack PPAR-␥ or express it at very low levels (31). We could not detect the PPAR-␥ transcripts in BV2 cells, but we did observe an inhibitory effect of PPAR-␥ agonists on STAT phosphorylation (Fig. 4a). Furthermore, PPAR-␥ agonists reduced the transcription of not only IFN-␥-responsive genes but also inflammation-associated genes in BV2 cells (Fig. 4, b and  c). Thus, we conclude that the inhibitory actions of 15d-PGJ 2 and rosiglitazone on JAK-STAT signaling occur independently of PPAR-␥ activation in brain.
We next focused our attention on how JAK activity might be inhibited independently of PPAR-␥ in cells pretreated with 15d-PGJ 2 and rosiglitazone. We considered the possibility that SOCS1 and SOCS3 could be involved in these events because they have been identified as negative regulators of JAK activation and as key regulators in the immune system (15). Interestingly, we found that 15d-PGJ 2 and rosiglitazone rapidly induced the transcription of SOCS1 as well as SOCS3 in primary astrocytes and BV2 cells. Within 1-2 h after treatment of either 15d-PGJ 2 or rosiglitazone, the transcript levels of SOCS1 and SOCS3 were significantly elevated in a dose-dependent fashion (Figs. 5 and 6). In agreement with JAK-STAT signaling events, overexpression of PPAR-␥S112A failed to increase the ability of PPAR-␥ agonists to induce SOCS1 and SOCS3 (Fig. 6d), indicating that elevation of SOCS transcripts occurs independently of PPAR-␥. To confirm the functional role of SOCS in brain inflammatory responses, we tested whether overexpression of SOCS1 or SOCS3 attenuated the phosphorylation of STATs in primary astrocytes. Levels of phosphorylated STAT1 and STAT3 were considerably attenuated in primary astrocytes transiently transfected with either SOCS1 or SOCS3 compared with vector (Fig. 6e). These results suggest that SOCS1 and SOCS3 may mediate the anti-inflammatory action of 15d-PGJ 2 and rosiglitazone.
In subsequent experiments, we unexpectedly found that the phosphorylations of JAKs and STATs were also suppressed at 5 min after simultaneous exposure of cells to IFN-␥ and either 15d-PGJ 2 or rosiglitazone, implying that 15d-PGJ 2 or rosiglitazone does not act only via the induction of SOCS1 and SOCS3 on JAK-STAT inflammatory signaling. Thus, we further examined the mechanism explaining the rapid inhibition of JAK-STAT activation following treatment with 15d-PGJ2 or rosiglitazone. Because SHP has been known to be an important negative regulator of JAK-STAT signaling, we investigated the involvement of SHP in anti-inflammatory action of 15d-PGJ 2 and rosiglitazone. Interestingly, we observed that SHP2 was phosphorylated within 5 min after treatment of 15d-PGJ 2 (Fig.  7). The activity of SHP2 has been reported to correlate with its phosphorylation, and the phosphorylated SHP2 can catalyze the tyrosine phosphorylation of JAKs, receptor, or other cellular proteins (32). In this regard, the activation of SHP2 may be part of a mechanism involved in inhibitory action of 15d-PGJ 2 and rosiglitazone. DISCUSSION Some PPAR-␥ agonists have been reported to effectively suppress inflammation and are thus promising potential therapeutic agents for chronic inflammatory diseases such as Alzheimer's disease (33). Recent in vivo studies provide evidence for the therapeutic potential of PPAR-␥ agonists in various inflammatory diseases (22,23,34). However, the mechanisms underlying the role of PPAR-␥ agonists in ameliorating inflammation are poorly understood. Most of the information on the mechanism of action of PPAR-␥ agonists is based on studies of NFB activity. PPAR-␥ agonists have been shown to inhibit multiple steps in the NFB signaling pathway through covalent modi- FIG. 6. Effects of 15d-PGJ 2 and rosiglitazone on expression of SOCS1 and SOCS3. a-c, primary astrocytes (a and c) and BV2 (b) cells were treated with IFN-␥ (10 units ml Ϫ1 ), 15d-PGJ 2 (15d) (10 M), or rosiglitazone (20 M) as indicated. RT-PCR analysis was performed using SOCS1 primers (a and b) and SOCS3 primers (c), and the products were analyzed on ethidium bromide-stained agarose gel. d, primary astrocytes were transiently transfected with pSV or PPAR-␥S112A, and the cells were treated with IFN-␥ or 15d-PGJ 2 for 2 h. The transcript level of SOCS1 was detected by RT-PCR. e, primary astrocytes were transiently transfected with pEF, pEF-SOCS1, or pEF-SOCS3, and the cells were treated with IFN-␥ (10 units ml Ϫ1 ) for 5 min. Western blots were probed with pSTAT1, pSTAT3, or STAT1 antibody. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. fications of IB kinase (31,35,36) and trans-repression of NFB, resulting in suppression of inflammation (19,21). In this paper, we suggest a novel pathway underlying the antiinflammatory action of PPAR-␥ agonists in activated glial cells based on the JAK-STAT signaling in inflammatory response.
The possible involvement of STAT in the anti-inflammatory action of PPAR-␥ agonists had been previously proposed (19), but evidence to directly support this hypothesis was lacking.
Here, we show that 15d-PGJ 2 and rosiglitazone can act at two points: first by rapidly attenuating JAK phosphorylation followed by later modulation of JAK-STAT inflammatory signaling in brain (Figs. 1 and 2). Recent reports indicate that JAKs and STATs play important roles in many diverse cellular events in brain (37)(38)(39), especially in neuroinflammatory disorders including Alzheimer's disease (25) and experimental autoimmune encephalomyelitis (40). In this regard, it is interesting to note that PPAR-␥ agonists can inhibit the activity of JAK and STAT, which may provide an explanation for previous reports of the beneficial actions of PPAR-␥ agonists in the treatment of neuronal diseases (33).
Because of the anti-inflammatory action of PPAR-␥ agonists, PPAR-␥ has received much attention as a target for the development of anti-inflammatory drugs (24). However, whether the anti-inflammatory effects of PPAR-␥ agonists are PPAR-␥-dependent or independent is controversial (19,21,31). Thus, we investigated the involvement of PPAR-␥ in the inhibition of JAK-STAT phosphorylation by its agonists. In all of the signaling events that we tested, PPAR-␥ agonists appeared to act independently of PPAR-␥ activation. Previously, other researchers reported that PPAR-␥ agonists could reduce inflammation independently of PPAR-␥ via direct modulation of NFB activity (36,37). Consistent with these works, our present data indicate that PPAR-␥ agonists also act in JAK-STAT inflammatory signaling independently of PPAR-␥ These observations support the recent proposal by Chawla et al. (21) that PPAR-␥ is required for lipid metabolism but not for anti-inflammatory effects.
How then might PPAR-␥ agonists act independently of PPAR-␥? Interestingly, we found that PPAR-␥ agonists rapidly induced the transcription of SOCS1 and SOCS3, key physiological regulators of immune responses and therapeutic targets for human immune diseases (15). One consequence of SOCS induction is the inhibition of JAK activation, thus reducing inflammatory signaling. We also found that overexpression of either SOCS1 or SOCS3 led to suppression of STAT phosphorylation in primary astrocytes (Fig. 6e). Recently, Wesemann et al. (41) report that overexpression of SOCS1 inhibits CD40 expression by blocking STAT1 activation in macrophage, resulting in the suppression of IFN-␥-induced tumor necrosis factor-␣ secretion and subsequent activation of NFB. Thus, it is possible that SOCS induction could suppress NFB activity in addition to JAK-STAT activity. There have been several reports regarding the beneficial actions of SOCS in the treatment of immune diseases such as hematological malignancies (42), inflammatory arthritis (14), and infectious disease (43). The results we report here further support the therapeutic role of SOCS. It is interesting that 15d-PGJ 2 and rosiglitazone are potent inducers of SOCS expression, because the only SOCSinducing agents known previously were cytokines and growth factors (15).
These observations lead to the question of how expression of SOCS1 and SOCS3 might be induced by 15d-PGJ 2 and rosiglitazone. Although SOCS are negative feedback regulators that are rapidly transcribed in a STAT-dependent manner (15), we could not observe the phosphorylation of JAKs and STATs by treatment of PPAR-␥ agonists alone for 5 min to 2 h (Fig. 8 and data not shown), indicating that other pathways are responsible for their induction. Preliminary data from our laboratory suggest a role for protein kinases A and C in the induction of SOCS1 and SOCS3 by PPAR-␥ agonists. 2 It has been previously reported that the promoter of SOCS1 and SOCS3 contain cAMP binding sites and that either cAMP (44) or phorbol 12-myristate 13-acetate (45) can stimulate the transcription of SOCS3. Thus, it is conceivable that certain protein kinases may mediate the induction of SOCS, but additional studies are needed to evaluate this possibility.
Here we suggest the interesting note that there is a PPAR-␥-independent link between anti-inflammatory action of PPAR-␥ agonists and induction of SOCS. In addition to SOCS1 and SOCS3, SHP2, another negative regulator of JAK-STAT signaling, seems to be also involved in anti-inflammatory action of 15d-PGJ 2 and rosiglitazone. As shown in Fig. 7, SHP2 was phosphorylated upon treatment of 15d-PGJ 2 . The phosphatase activity of SHPs has been reported to depend primarily on their phosphorylation and interaction with tyrosine-phosphorylated proteins such as JAKs, receptor, and other cellular proteins (32). This observation is in agreement with our recent data obtained from other experiments. We recently revealed that the activation of SHPs may be closely related with attenuation of brain inflammation in activated microglia, 3 which supports that SHP2 as well as SOCS1 and SOCS3 may be involved in the inflammatory action of 15d-PGJ 2 and rosiglitazone.
Although work on the contribution of SOCS to human immune diseases is still in its infancy, some studies suggest that the enforced expression of SOCS might be beneficial for the treatment of some immune diseases. Thus, our results provide a strong foundation for the future development of therapeutic drugs for the treatment of inflammatory diseases.