Casein-kinase-II-dependent phosphorylation of PPARγ provokes CRM1-mediated shuttling of PPARγ from the nucleus to the cytosol

Subcellular compartmentalization of transcription factors is an important mechanism to regulate their activity. Often, transcription factors localize to the cytoplasm in their basal, unstimulated state, to allow rapid access to receptors at the plasma membrane (Xu and Massague, 2004). For transcription factors such as signal transducer and activators of transcription (STAT) proteins, which demand receptor-dependent phosphorylation to become activated, cytosolic localization is a prerequisite for their activation (O'Shea and Murray, 2008). Upon activation, they translocate to the nucleus, initiate expression of target genes and are finally exported from the nucleus for their recycling in successive rounds of activation and deactivation.

Other transcription factors require ligand binding to become activated (Weigel and Moore, 2007). Some of these ligand-dependent transcription factors are localized in the cytoplasm and translocate to the nucleus upon ligand binding. Others, such as members of the nuclear hormone receptor family are located mainly in the nucleus, even in the basal, unstimulated state (Glass, 2006). Peroxisome proliferator-activated receptors (PPARs) are one such group of receptors, and they have been described to affect pro-inflammatory gene expression in a manner that is independent of DNA binding (Daynes and Jones, 2002). PPARγ is a ligand-dependent nuclear hormone receptor that attenuates pro-inflammatory signaling by direct protein-protein interaction and thus scavenges co-factors or transcription factors such as NFAT, STAT or NF-κB from binding to the DNA, and concomitantly inhibits transactivation of their pro-inflammatory target genes (Tontonoz and Spiegelman, 2008). Moreover, PPARγ inhibits MAP kinase signaling or degrades co-repressors, known to block NF-κB-dependent transactivation (Straus and Glass, 2007).

These mechanisms require nuclear PPARγ localization, but recent data provide evidence that PPARγ also localizes to the cytoplasm (von Knethen et al., 2007). Little is known about the mechanisms that provoke nucleocytoplasmic shuttling of PPARγ. Burgermeister and colleagues showed that PPARγ is shuttled from the nucleus by binding via its AF2 domain to MEK1 (Burgermeister et al., 2007). This complex is actively exported from the nucleus to the cytosol by CRM1 in response to phorbol ester stimulation, thereby causing PPARγ downregulation in the nucleus and inhibiting PPARγ-dependent transactivation. Moreover, Kelly and co-workers demonstrated that in response to anaerobic gut bacteria PPARγ binds to RelA, a member of the NF-κB transcription factor family, in the nucleus (Kelly et al., 2004). This complex is shuttled to the cytosol by an unknown CRM1-independent pathway, attenuating NF-κB-dependent inflammation. These data suggest that different export mechanisms cause cytosolic PPARγ localization.

Considering the requirement to precisely regulate the pathway whereby PPARγ attenuates hypo-inflammatory states as well as chronic inflammation (Daynes and Jones, 2002; Gerry and Pascual, 2008; Rigamonti et al., 2008; Zingarelli and Cook, 2005), characterization of mechanisms regulating this confined process is necessary. Thus, our study defines signaling pathways that lead to ligand-independent cytoplasmic PPARγ localization in untreated cells. CK-II-dependent phosphorylation at Ser16 and Ser21 of the PPARγ AF1 domain is a premise of an active export, requiring the nuclear export protein CRM1. Moreover, this process demands RanGTP and phosphorylation of RanBP3, to efficiently shuttle PPARγ to the cytoplasm, where it functions in a cytosolic signaling pathway in response to its activation.

Taking our previous work showing that the PPARγ deletion constructs PPARγ1 Δ32-198 and PPARγ1 Δ51-406 localize to the cytoplasm (von Knethen et al., 2007) into consideration, we focused on the two remaining parts of PPARγ: residues 1-31 of the AF1 domain and residues 406-475 of the AF2 domain. As the AF2 domain is important for phorbol-ester-initiated PPARγ export from the nucleus (Burgermeister et al., 2007), we first analyzed whether residues 406-475 of the PPARγ AF2 domain account for cytoplasmic PPARγ localization in untreated control cells. We deleted residues 407-475 using the DsRed-Monomer-C1-PPARγ1 wild-type construct (von Knethen et al., 2007) to obtain the DsRed-Monomer-C1-PPARγ1 Δ407-475 deletion mutant (supplementary material Fig. S1). Transiently transfected HEK293 cells demonstrate that deletion of amino acids 407-475 does not alter cytosolic PPARγ localization (Fig. 1A) in our system. However, the localization pattern of this construct indicates that PPARγ localizes to the cytoplasm independently of agonist binding, because this domain includes the amino acids conferring ligand binding, as well as ligand-dependent transactivation. To define the domain involved in cytosolic PPARγ localization, we deleted residues 1-31 using the DsRed-Monomer-C1-PPARγ1 wild-type construct to obtain DsRed-Monomer-C1-PPARγ1 Δ1-31 (supplementary material Fig. S1). In HEK293 cells transiently transfected with this construct, we only noticed nuclear PPARγ localization (Fig. 1B). We conclude that one or more of the amino acids within the first 31 residues is responsible for cytosolic PPARγ localization.

To clarify mechanisms for cytosolic PPARγ localization, we transiently transfected HEK293 cells with a full-length PPARγ construct linked to DsRed (DsRed-Monomer-C1-PPARγ1 wild type). Twenty-four hours after transfection, cells were treated for 1 hour with the ERK1/2 inhibitor PD98059 (25 μM) (Fig. 2B), the p38 MAPK inhibitor SB203850 (10 μM) (Fig. 2C), a classical PKC inhibitor Gö6976 (1 μM) (Fig. 2D), the PI3 kinase inhibitor LY294002 (30 μM) (Fig. 2E), or the casein kinase II (CK-II) inhibitor DRB (100 μM) (Fig. 2F). Nuclei were counterstained using DAPI. Interestingly, PD98059 and DRB both blocked cytosolic PPARγ localization, suggesting a mechanism involving ERK1/2 and/or CK-II. Effectiveness of inhibitors was confirmed (data not shown).

In silico analysis of residues 1-31 of PPARγ revealed Ser16 and Ser21 as potential phosphorylation sites for CK-II. Moreover, CK-II is a kinase that is known to be constitutively active in untreated cells (Litchfield, 2003). To check whether Ser16 and Ser21 are putative phosphorylation sites, we mutated Ser16 and Ser21 into alanine to block their phosphorylation. As shown in Fig. 3A, in transiently transfected HEK293 cells the DsRed-Monomer-C1-PPARγ1 S16A/S21A mutant (supplementary material Fig. S1) does not localize in the cytosol. To further prove the important phosphorylation of these residues, we mutated both serines into glutamic acid (S16E/S21E), to mimic a constitutive phosphorylation. As expected, DsRed-Monomer-C1-PPARγ1 S16E/S21E (supplementary material Fig. S1) mainly localized to the cytosol in transiently transfected HEK293 cells (Fig. 3B). From these results, we conclude that CK-II-dependent phosphorylation of PPARγ at Ser16 and Ser21 provokes its cytosolic localization. To verify these immunofluorescence data and to determine the cellular localization of PPARγ, we examined the impact of RXRα, which is a heterodimerization partner of PPARγ. We transiently transfected HEK293 cells with DsRed-Monomer-C1-PPARγ1 wild type, DsRed-Monomer-C1-PPARγ1 S16E/S21 or DsRed-Monomer-C1-PPARγ S16A/S21A in combination with an RXRα-EGFP construct. Nuclear and cytosolic fractions were isolated and analyzed for PPARγ and RXRα localization with or without 1 μM rosiglitazone treatment. As shown in Fig. 3C, DsRed-Monomer-C1-PPARγ1 wild type mainly localized to the nucleus (Fig. 3C, lane 1), although a small, but significant amount was also found in the cytosol. By contrast, the DsRed-Monomer-C1-PPARγ S16E/S21E mutant completely localized to the cytosolic fraction (Fig. 3C, lane 3), whereas the DsRed-Monomer C1-PPARγ S16A/S21A mutant, in line to our immunofluorescence data, showed only nuclear localization (Fig. 3C, lane 5). Interestingly, stimulation of the transiently transfected cells with 1 μM rosiglitazone altered neither exclusive nuclear RXRα localization nor PPARγ localization (Fig. 3C, lanes 2, 4 and 6). Moreover, reporter analysis showed no PPARγ-dependent transactivion using cells expressing the S16E/S21E mutant compared with cells transiently transfected with the PPARγ wild-type construct, further supporting our data on cytosolic PPARγ localization. Strikingly, PPARγ-dependent transactivation was enhanced in cells transiently transfected with the PPARγ S16A/S21A mutant, which allows only nuclear localization (supplementary material Fig. S2).

Taking into consideration that induction of PPARγ target genes requires nuclear PPARγ localization, CK-II inhibition should enhance expression of PPARγ target genes, such as CD36, in response to PPARγ-specific agonists. Therefore, we treated RAW 264.7 macrophages with 1 μM rosiglitazone to induce PPARγ-dependent gene induction and determined how increasing concentrations of DRB alter PPARγ-dependent CD36 expression. Having established CD36 as a PPARγ target gene in our experimental system (supplementary material Fig. S3), we found that 100 nM DRB significantly increased rosiglitazone-dependent CD36 expression in RAW 264.7 macrophages (Fig. 4A, black columns). To verify that CK-II inhibition also alters PPARγ-dependent CD36 expression in primary cells, we treated peritoneal macrophages derived from wild-type PPARγ mice (Mac-PPARγ wild type) with 1 μM rosiglitazone and with increasing concentrations of DRB (from 10 nM to 100 μM). Expression of CD36 mRNA in primary cells was enhanced in response to DRB in a concentration-dependent manner, and was most effective with 10 μM DRB. This suggests that inhibition of CK-II retains PPARγ in the nucleus, which consequently amplifies PPARγ target gene expression; this was further supported by only nuclear PPARγ localization in peritoneal macrophages derived from PPARγ wild-type mice in response to DRB treatment (supplementary material Fig. S4). Final evidence for the role of CK-II comes from CK-II-knockdown experiments. As shown in Fig. 4B, stable knockdown of CK-II in HEK293 cells (supplementary material Fig. S5) blocks cytosolic PPARγ localization (Fig. 4B,C). From these data we conclude that CK-II is important for cytosolic PPARγ localization under resting conditions.

Because phosphorylation of PPARγ by CK-II does not explain the inhibitory role of PD98059 (Fig. 2B) on cytosolic PPARγ localization, we focused on proteins, such as CRM1, that are known to be involved in nuclear-cytosolic shuttling (Hutten and Kehlenbach, 2007). CRM1 exports proteins from the nucleus to the cytoplasm, requiring in most, but not all instances, the GTPase Ran and a co-factor, such as RanBP3, for effective cargo export (Askjaer et al., 1999; Engelsma et al., 2004). Because CRM1-dependent shuttling is blocked in response to leptomycin B, we analyzed the effect of leptomycin B on PPARγ localization. HEK293 cells, transiently transfected with either a full-length PPARγ1 construct (DsRed-Monomer-C1 PPARγ1 wild type) or the PPARγ1 mutant DsRed-Monomer-C1 PPARγ1 S16E/S21E, which localized in untreated cells primarily in the cytoplasm (Fig. 3B), were treated with 50 nM leptomycin B for 1 hour. As shown in Fig. 5, leptomycin B blocked cytosolic PPARγ localization of the full-length PPARγ construct (Fig. 5A) and it even abolished cytosolic localization of PPARγ when Ser16 and Ser21 were mutated to glutamic acid (Fig. 5B). Therefore, mimicking phosphorylation of these two amino acids is not sufficient for cytosolic PPARγ localization, thus pointing to CRM1 as an additional factor that is required for nuclear export.

Based on these data, we went on to determine whether Ran and RanBP3 are involved in CRM1-dependent PPARγ shuttling. Therefore, we transiently transfected HEK293 cells with a wild-type Ran construct or a Ran Q69L construct, which functions as a dominant-negative mutant, not allowing GTP binding to Ran and thus, possibly inhibiting CRM1-dependent shuttling (Gil et al., 2006). HEK293 cells were co-transfected with a DsRed-Monomer-C1 PPARγ1 wild-type construct to investigate PPARγ localization. To guarantee that all cells express not only the full-length PPARγ construct, but also express the particular Ran construct, cells were transfected with a vector ratio of 1:3 (DsRed-Monomer-C1 PPARγ1 wild type to Ran construct). As we assumed, overexpression of wild-type Ran enhanced cytosolic PPARγ localization (Fig. 6A). Interestingly, the dominant-negative Ran Q69L mutant completely retained PPARγ in the nucleus (Fig. 6C), indicating the requirement of a functional Ran to export PPARγ from the nucleus. PD98059 prevents cytosolic PPARγ localization, even in cells overexpressing wild-type Ran (Fig. 6B), suggesting that a different factor is involved in causing cytosolic PPARγ localization. In cells transfected with Ran Q69L, PD98059 did not alter nuclear PPARγ localization (Fig. 6D).

Therefore, we considered RanBP3, which recently has been shown to be phosphorylated at Ser58 by RSK1/2, which was activated by ERK1/2 to enhance cargo binding to CRM1 and concomitant nuclear export (Yoon et al., 2008). We analyzed RanBP3 phosphorylation in RAW 264.7 macrophages in untreated controls, in cells incubated for 1 hour with DRB, PD98059 or leptomycin B. Western blotting, using an antibody specific for RanBP3-S58 phosphorylation, showed that phosphorylation of RanBP3 was not affected by DRB or leptomycin B (Fig. 7, lower panel, second and fourth lane), but was significantly reduced by the MEK1/2 inhibitor PD98059, which concomitantly abolished ERK1/2 activity (Fig. 7, lower panel, third lane). A densitometric analysis of these results using actin as control is provided in the upper panel of Fig. 7. These data suggest that ERK1/2-mediated phosphorylation of RanBP3 is a prerequisite for the export of PPARγ from the nucleus.

To prove the role of RanBP3 in PPARγ export, we transfected a FLAG-tagged vector encoding RanBP3 into HEK293 cells. After 24 hours, cells were treated for 1 hour with 25 μM PD98059 or 100 μM DRB. Total protein lysates were prepared and immunoprecipitated for FLAG-RanBP3. Immunoprecipitates were analyzed for co-immunoprecipitation of PPARγ and CRM1. As shown in Fig. 8A, PD98059 abolished RanBP3-dependent co-immunoprecipitation of PPARγ and CRM1 compared with control conditions. As expected, DRB inhibited co-immunoprecipitation of PPARγ with RanBP3, but left RanBP3 co-immunoprecipitation of CRM1 unaltered. We conclude that co-immunoprecipitation of PPARγ with CRM1 requires CK-II activity, probably because of Ser16 and Ser21 phosphorylation. However, CRM1-dependent export demands Ran GTPase activity and RanBP3 phosphorylation, presumably by ERK1/2. In a final set of experiments HEK293 cells were treated for 1 hour with 25 μM PD98059 or 100 μM DRB. Total cell lysates were prepared and used for immunoprecipitation of CRM1. As shown in Fig. 8B, PD98059 did not alter the amount of co-immunoprecipitated PPARγ, whereas DRB almost completely blocked PPARγ co-immunoprecipitation. Based on these findings we were interested to clarify whether PPARγ directly binds to CRM1. Therefore, we used bacterially expressed proteins (data not shown) in a cell-free system to analyze whether PPARγ can be co-immunoprecipitated by CRM1 immunoprecipitation. As shown in Fig. 8C, PPARγ was not co-immunoprecipitated with CRM1 in combination with wild-type Ran. However, in line with our assumption, PPARγ was not co-immunoprecipitated with CRM1 in combination with Ran Q69L, although CRM1 co-immunoprecipitated both wild-type Ran and Ran Q69L. These data support the hypothesis that CRM1 does not directly bind to PPARγ.

To further elucidate the role of ERK1/2 in cytosolic PPARγ localization, we stably knocked down ERK1 and ERK2, using a lentiviral shRNA approach in HEK293 cells. Knockdown of ERK1 and ERK2 was differently effective in various clones (supplementary material Fig. S6). For further analysis, we used clone 1 with ERK1 knockdown and clone 3 with ERK2 knockdown. Whereas ERK2 knockdown did not alter PPARγ localization (data not shown), ERK1 knockdown completely retained DsRed-Monomer C1 PPARγ1 wild type in the nucleus (Fig. 9E, right panel). In line with our assumption of a mechanism requiring RanBP3 phosphorylation, knockdown of ERK1 inhibited basal RanBP3 phosphorylation (Fig. 9A, upper panel, right lane compared with left lane). To verify a role of ERK1 in cytosolic PPARγ localization, we transiently transfected HEK293 cells with a vector encoding the constitutively active protein ERK2-MEK1-LA, which has been shown to act as a super activator of the MAP kinase cascade (Robinson et al., 1998). This fusion construct is characterized by the mutation of four MEK1 leucine residues, which are crucial for nuclear export, into alanine (Robinson et al., 1998). This region is also known to control MEK1 activity (Mansour et al., 1996; Mansour et al., 1994). In line with this, we found expression of ERK2-MEK1-LA exclusively in the nuclear fraction (Fig. 10, cytosolic vs nuclear fraction). To quantify the effect of ERK1 knockdown and ERK2-MEK1-LA overexpression on phosphorylation of ERK1, ERK2 and RanBP3, a densitometric quantification is provided in Fig. 9B,C,D. Having established its functional expression in ERK1-knockdown cells (Fig. 9A), we determined whether ERK2-MEK1-LA restores cytosolic PPARγ localization in ERK1-knockdown cells. As shown in Fig. 9E, the ERK2-MEK1-LA protein indeed provoked cytosolic PPARγ localization in ERK1-knockdown cells, which was even stronger than in control cells (Fig. 9E, first panel vs second panel).

To test whether the results obtained with the DsRed-Monomer C1 PPARγ constructs are valid for endogenous PPARγ, we performed a set of experiments using HEK293 control and ERK1-knockdown cells. Cytosolic and nuclear fractions were separated by SDS gel electrophoresis and expression of PPARγ was analyzed by western blotting. As shown in Fig. 10A (first vs second panel), only a small amount of PPARγ was located in the cytosol in control cells. Densitometric quantification demonstrated that approximately 20% of the total PPARγ is located in the cytosol (Fig. 10B). In ERK1-knockdown cells, less than 5% of total PPARγ localizes in the cytosol (Fig. 10A, third vs fourth panel, and Fig. 10B), clearly indicating a role of ERK1 in cytosolic PPARγ localization. This assumption is supported by the use of control and ERK1-knockdown cells transiently transfected with the ERK2-MEK1-LA constitutively active kinase. Expression of this construct provoked an almost complete cytosolic PPARγ localization, although in control HEK293 cells this effect was more intense than in ERK1-knockdown cells (Fig. 10A,B). From these data, we conclude that ERK1 provokes cytosolic PPARγ localization under resting conditions via RanBP3 phosphorylation. Fig. 11 proposes a model of hierarchical regulation, based on our study, to explain the cytosolic localization of PPARγ.

We provide evidence for a previously unrecognized export mechanism provoking cytosolic PPARγ localization in control cells. This is of interest because cytosolic PPARγ has been shown to affect pro-inflammatory signaling independently of DNA binding. In the nucleus, this is achieved mainly by protein-protein interactions to scavenge the bound proteins from their signaling pathways, thereby inhibiting signal transduction and concomitant gene induction (Daynes and Jones, 2002; Glass, 2006; Tontonoz and Spiegelman, 2008). It is likely that PPARγ also affects signaling cascades in the cytoplasm, similarly to its role in the nucleus, i.e. by direct protein-protein interaction (von Knethen et al., 2007). Thus, an export mechanism provoking cytosolic PPARγ localization in unstimulated control cells appears to be a prerequisite.

Recent data support this assumption, demonstrating cytosolic PPARγ localization in response to phorbol ester activation of cells, decrease in nuclear PPARγ and consequent inhibition of PPARγ-dependent transactivation (Burgermeister et al., 2007). In this report, PPARγ directly interacts with MEK1, and this complex is exported from the nucleus in a CRM1-dependent fashion. PPARγ binding to MEK1 is mediated by the AF2 domain of PPARγ, particularly the last 16 amino acids of the C-terminus. Because in our system these amino acids are not involved in cytosolic PPARγ localization under control conditions, this implies a different export mechanism is responsible for the distribution of PPARγ in untreated cells.

Our data indicate the involvement of CK-II, because the CK-II-specific inhibitor DRB, as well as CK-II knockdown completely blocked cytosolic PPARγ localization. In line with this, CK-II is active under control conditions and furthermore, has been found to facilitate phosphorylation of nuclear proteins (Litchfield, 2003). Moreover, CK-II-dependent phosphorylation has been described to mediate the export of nuclear proteins such as ribosomal S6 kinase1 II (S6K1 II). Here, CK-II phosphorylation of Ser17 causes export of S6K1 II, as shown by transfecting a S17E mutant that mimicks constitutive phosphorylation in association with cytosolic localization (Panasyuk et al., 2006). Comparison of the first 100 amino acids of PPARα, PPARβ, PPARγ, ER1 and ER2 demonstrates that several putative CK-II phosphorylation sites exist in these proteins (supplementary material Fig. S7). However, whether CK-II phosphorylation of these amino acids also mediates cytosolic localization of these proteins has to be clarified. In analogy to these observations, the PPARγ S16E/S21E mutant in our experiments particularly localizes to the cytosol as well. In addition to the role of CK-II in direct phosphorylation of PPARγ, a different component seems to be involved in PPARγ export, which is blocked in response to PD98059, the classical ERK1/2 inhibitor.

Therefore, we focused on established nuclear export proteins such as CRM1 (Hutten and Kehlenbach, 2007). In line with our assumption, the CRM1-specific inhibitor leptomycin B (LMB) abolished cytosolic PPARγ localization. The molecular export inhibition mechanism involves the direct binding of LMB to Cys529 of CRM1, thereby blocking CRM1 binding to the nuclear export signal (NES) of cargo proteins (Kudo et al., 1999), concomitantly freezing cargo export. Therefore, our data demonstrating only nuclear localization of PPARγ in response to LMB support the assumption of a CRM1-dependent export mechanism in our system. However, owing to the lack of a classical leucine-rich nuclear export signal, PPARγ cannot directly bind to CRM1. Therefore, an unknown adapter protein has to connect PPARγ and CRM1. The characterization of this protein is currently under investigation.

Further evidence for a CRM1-dependent mechanism came from experiments in which we overexpressed Ran in HEK293 cells, which causes enhanced cytosolic PPARγ localization. This might be explained by the fact that RCC, a GTP exchange factor of Ran, is restricted to the nucleus (Xu and Massague, 2004). This causes a RanGTP-RanGDP gradient, where RanGTP is more abundant in the nucleus and RanGDP is more abundant in the cytoplasm; this is even more enhanced in cells overexpressing Ran. However, based on our experiments showing that Ran overexpression provokes an increase in cytosolic PPARγ localization, Ran seems to be a rate-limiting enzyme for PPARγ export. This was further emphasized by the dominant-negative mutant Ran Q69L, which cannot bind GTP, and thus abrogates CRM1-dependent shuttling, as established for PTEN (Gil et al., 2006). Overexpressing the dominant-negative Ran Q69L mutant inhibits cytosolic PPARγ localization under resting conditions. Interestingly, PD98059 still blocked cytosolic PPARγ localization, even in cells overexpressing Ran. From these data we conclude that another mechanism or factor is required for PPARγ export.

Taking a recent manuscript into consideration, which demonstrated ERK1/2-dependent phosphorylation of the CRM1 adaptor protein RanBP3, thereby enhancing the binding of CRM1 to its cargo (Yoon et al., 2008), we assumed that RanBP3 might also be involved in PPARγ export in untreated cells. In line with our expectation, we found that PD98059 inhibits RanBP3 phosphorylation, indicating a possible involvement of this adaptor protein in PPARγ export from the nucleus. Although overexpression of RanBP3 does not enhance cytosolic PPARγ localization in HEK293 cells, transfection of a plasmid encoding an ERK2-MEK1-LA constitutively active kinase enhances RanBP3 phosphorylation and amplifies cytosolic PPARγ localization. These data suggest that RanBP3 expression is not rate limiting for PPARγ export, but its phosphorylation. In line with this, ERK1 knockdown, which provokes a decreased basal RanBP3 phosphorylation, inhibits cytosolic PPARγ localization. Further support came from our data in HEK293 cells determining PPARγ localization in cytosolic and nuclear fractions. Here, ERK2-MEK1-LA compensates for knockdown of ERK1. It remains elusive whether this is a cell- type-specific phenomenon. However, co-immunoprecipitation assays in RAW 264.7 macrophages verify the role of CK-II and RanBP3 phosphorylation in the CRM1-dependent nuclear export of PPARγ. This is in line with our previous data showing cytosolic PPARγ localization in RAW 264.7 macrophages under resting conditions (von Knethen et al., 2007).

If there are several parallel mechanisms of PPARγ export, they might operate under different physiological contexts or be subject to different forms of regulation. It will be challenging to characterize their different outputs in an adoptive gene transfer model in vivo, with vectors encoding PPARγ mutants that allow either nuclear or cytosolic localization. Moreover, characterization of the impact of cell and PPARγ activation upon cellular compartmentalization of PPARγ will be essential to modulate PPARγ-dependent signaling, which might be important to inhibit hyper-inflammatory or chronic inflammatory diseases.

We cultured RAW 264.7 and primary murine peritoneal macrophages in RPMI 1640 (PAA Laboratories, Linz, Austria) and HEK293 cells in DMEM high glucose (PAA Laboratories). Both media were supplemented with 100 U/ml penicillin (PAA Laboratories), 100 μg/ml streptomycin (PAA Laboratories) and 10% heat-inactivated fetal calf serum (PAA Laboratories).

C57BL/6 LysM-CRE mice (LysM-CRE^+/+) were bred with C57BL/6 PPARγ^fl/fl mice as described (Akiyama et al., 2002; Kuhn et al., 1995). Genotyping of LysM-CRE^+/+ PPARγ^fl/fl mice was performed as described (Jennewein et al., 2008). C57BL/6 LysM-CRE^+/+ PPARγ^fl/fl (Mac-PPARγ-KO) and PPARγ^fl/fl (Mac-PPARγ wild type) mice (25-35 g) were used for the experiments. Animals had free access to water and pellet food. Animals were entrained to a regular 12:12 hour light:dark cycle. For the experiments, mice were sacrificed and peritoneal macrophages were isolated by peritoneal lavage with 5 ml PBS. All procedures performed on these mice followed the guidelines of the Hessian animal care and use committee (authorization no. V54-19c20/15-F144/01 and 02).

To determine intracellular PPARγ localization in response to DRB treatment, we seeded peritoneal macrophages derived by peritoneal lavage from wild-type PPARγ mice directly on a slide. After 24 hours, cells were treated as indicated and fixed on the slides by overnight incubation in 4% paraformaldehyde at 4°C. Thereafter, cells were permeabilized in PBS containing 0.5% Triton X-100 for 15 minutes. After a washing step in PBS, cells were incubated for 2 hours with a 1:100 dilution of rabbit α-PPARγ antibody (Santa Cruz, Biotechnology, Heidelberg, Germany). After three washing steps with PBS, cells were incubated with a secondary goat α-rabbit antibody (1:250) labeled with Alexa Fluor 546 (Invitrogen, Karlsruhe, Germany) for 2 hours at 4°C. After three 5 minute washing steps with PBS, cells were counterstained with DAPI (1 μg/ml in PBS for 15 minutes). After a final 5 minute wash in PBS, cells were covered with Vectashield mounting medium (Linaris, Wertheim, Germany) and a coverslip. PPARγ localization was determined using an AxioScope fluorescence microscope with the ApoTome upgrade (Carl Zeiss, Frankfurt, Germany, lens 63× 0.6 NA; ocular 10×), documented by a CCD camera (Carl Zeiss) and the AxioVision Software (Carl Zeiss).

Site-directed mutagenesis to generate double amino acid exchanges (S16A/S21E, S16E/S21E) and deletion of amino acids 1-31 and 407-475 were performed using the QuikChange XLII kit (Stratagene, Karlsruhe, Germany). The following primers were used: Δ1-31, 5′-GCTCCCAGTCCGGACTCAGATCTCGAATCAAGCCCTTCACTACTGTTGACTTCTCC-3′ and 5′-GGAGAAGTCAACAGTAGTGAAGGGCTTGATTCGAGATCTGAGTCCGGACTGGGAGC-3′; Δ407-475, 5′-CCTGCTACAAGCCCTGGAGCTCCAGCTGTAGGATCCACCGGATCTAGATAACTGATC-3′ and 5′-GATCAGTTATCTAGATCCGGTGGATCCTACAGCTGGAGCTCCAGGGCTTGTAGCAGG-3′; S16A/S21A, 5′-GGGATCGCCTCCGTGGATCTCGCCGTAATGGAAGACACTCCC-3′ and 5′-GGGAGTGTCTTCCATTACGGCGAGATCCACGGAGGCGATCCC-3′; S16E/S21E, 5′-GGGATCGAGTCCGTGGATCTCGAGGTAATGGAAGACACTCCC-3′ and 5′-GGGAGTGTCTTCCATTACCTCGAGATCCACGGACTCGATCCC-3′. Bold indicates changed nucleotides. The pDsRed-Monomer-C1-PPARγ1 wild-type vector (von Knethen et al., 2007) was used as a template. An initial denaturation step was performed at 95°C for 1 minute and followed by 18 cycles at 95°C for 50 seconds, annealing at 60°C for 50 seconds, and extension at 68°C for 7 minutes. A final extension phase was performed at 68°C for 7 minutes. Correct orientation and sequence of the generated vectors was verified by restriction analyses and/or sequencing.

To elucidate mechanisms of PPARγ localization requiring efficient transfection or transduction of expression or knockdown vectors, HEK293 cells were used. HEK293 cells are easy to transfect and transduce and moreover, HEK293 cells also express endogenous PPARγ. To determine macrophage-specific responses, transiently transfected RAW 264.7 macrophages were used. Additionally, RAW cells stably transduced with a dominant-negative PPARγ mutant [RAW 264.7 PPARγ d/n (Johann et al., 2006)] were used.

Therefore, to follow PPARγ distribution, HEK293 cells were seeded directly onto a slide, and transiently transfected by CaPO₄ precipitation with pDsRed-Monomer-C1 PPARγ1 wild type, pDsRed-Monomer-C1 PPARγ1 Δ1-31, pDsRed-Monomer-C1 PPARγ1 Δ407-475, pDsRed-Monomer-C1 PPARγ1 S16A/S21A, or pDsRed-Monomer-C1 PPARγ1 S16E/S21E alone, with combinations of pDsRed-Monomer-C1 PPARγ1 wild type with pcDNA3-FLAG-Ran wild type or pcDNA3-FLAG-Ran Q69L, or pDsRed-Monomer-C1 PPARγ1 wild type with pcDNA3-FLAG-RanBP3, or with combinations of pDsRed-Monomer-C1-PPARγ1 wild type with RXRα-EGFP, or pDsRed-Monomer-C1-PPARγ S16E/S21E with RXRα-EGFP, or pDsRed-Monomer-C1-PPARγ S16A/S21A with RXRα-EGFP. 24 hours after transfection, cells were used for experiments. Cells were treated for 1 hour with 25 μM PD98059, 10 μM SB203580, 1 μM Gö6976, 30 μM LY294002, 100 μM DRB or with 50 nM leptomycin B. Distribution of different DsRed-PPARγ constructs was analyzed using an AxioScope fluorescence microscope with the ApoTome upgrade (Carl Zeiss, lens 63× 0.6 NA; ocular 10×), documented by a CCD camera (Carl Zeiss) and the AxioVision Software (Carl Zeiss).

RNA from RAW 264.7 macrophages and primary peritoneal macrophages was extracted using peqGold RNAPure (Peqlab Biotechnologie, Erlangen, Germany). Total RNA (1 μg) was transcribed using iScript cDNA Synthesis Kit (Bio-Rad). Quantitative real-time-PCR (qPCR) was performed using MyiQ real-time PCR system (Bio-Rad) and Absolute Blue QPCR SYBR Green Fluorescein Mix (Thermo Scientific, Karlsruhe, Germany). For CD36 and actin expression, validated QuantiTect Primer Assays were purchased from Qiagen (Hilden, Germany). Real-time PCR results were quantified using Gene Expression Macro (version 1.1) from Bio-Rad with actin expression as the internal control.

Cells were co-transfected with the Renilla luciferase control vector pRL-CMV (Promega, Mannheim, Germany) and the PPRE containing p(AOX)₃-TK-luc construct using jetPEI cationic polymer transfection reagent (Biomol, Hamburg, Germany) following the instructions of the manufacturer. A total of 5×10⁵ cells were seeded in 24-well plates for 7 hours. Following transfection, incubations continued for 24 hours, followed by individual stimulation. Firefly luciferase activity normalized to Renilla luciferase activity was determined after 24 hours of treatment with 1 μM rosiglitazone.

Transfected HEK293 cells were stimulated or left as controls, as indicated. Subsequently cells were harvested and lysed in lysis buffer (50 mM Tris-HCl, 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet-40, 1 mM PMSF, pH 8.0). To ensure cell lysis, cells were sheared 10 times with a 16-gauge needle followed by brief sonication (Branson sonifier; 10 seconds; duty cycle 100%, output control 60%). Cell debris was removed by centrifugation (10,000 g, 5 minutes) and 1 mg protein was used for immunoprecipitation. Sample volume was adjusted with lysis buffer to 1 ml. Immunoprecipitation was performed as indicated using 5 μg monoclonal anti-FLAG antibody (Sigma) or 5 μg anti-exportin-1 antibody (Bethyl-Laboratories, Biomol, Hamburg, Germany) and incubated at 4°C overnight. Thereafter 100 μl μMACS protein-G microbeads (Miltenyi Biotech, Bergisch-Gladbach, Germany) were added when the monoclonal antibody was used and 100 μl μMACS protein-A beads were added to the polyclonal antibodies and incubated overnight. Lysate was applied onto an equilibrated μ column, already placed in the magnetic field of a μMACS separator. The flowthrough was collected and saved for further analysis. The column was rinsed four times with 200 μl lysis buffer followed by two washes with low ionic buffer (20 mM Tris-HCl, pH 7.5). Afterwards the remaining proteins were eluted using 50 μl boiling SDS sample buffer.

For SDS-PAGE, cell lysis was achieved with lysis buffer (50 mM Tris-HCl, 5 mM EDTA, 150 mM NaCl, 0.5% Nonidet-40, 1 mM PMSF, pH 8.0) and sonication (Branson sonifier; 20 seconds, duty cycle 100%, output control 60%). Whole-cell lysates were cleared by centrifugation (10,000 g, 5 minutes), and the protein concentration was determined with the Lowry method. Protein (80 μg) was resolved on 10% polyacrylamide gels and blotted onto nitrocellulose sheets basically following standard methodology. Equal loading and correct protein transfer to nitrocellulose was routinely quantified by Ponceau S staining. Filters were incubated with the anti-exportin-I antibody (1:500, Transduction Laboratories, BD Biosciences, Heidelberg, Germany), anti RanBP3-pS58 antibody (1:500, Invitrogen) anti-PPARγ antibody (1:500, Santa Cruz Biotechnology), anti-RFP antibody (1:1000, MBL, Biozol, Eching, Germany), anti-lamin antibody (1:1000, Santa Cruz Biotechnology), anti-EGFP antibody (1:1000, Clontech), anti-CK-II (1:1000, Abcam) or anti-actin antibody (1:2000, Amersham Biosciences, Freiburg, Germany) overnight at 4°C. IRDye700 or IRDye800 polyclonal antibodies (1:5000, Licor, Bad Homburg, Germany) and the The Odyssey Infrared Imaging System were used for direct infrared fluorescence detection and quantification.

His-tagged bacterial expression vectors for CRM1 (pz-tagged-CRM1), kindly provided by Maarten Fornerod (Department of Tumor Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands), Ran wild type (pQE-Ran) and Ran Q69L (pQE-Ran Q69L, both kindly provided Maarten Fornerod with the permission of Dirk Görlich (Department for Cellular Logistics, MPI für Biophysikalische Chemie, Göttingen, Germany) and a GST-tagged PPARγ1 construct (pGEX4-T2 PPARγ) were transformed into BL21 bacteria. IPTG-dependent induction was performed based on standard procedures. Enrichment of expressed His-tagged proteins was performed using the BugBuster Ni-NTA His Bind Purification Kit from Novagen. Ran wild type and Ran Q69L loading with GTP was done according to personal communication of Maarten Fornerod. PPARγ-GST purification was performed using Glutathione-Sepharose 4 Fast Flow (GE Healthcare) according to the instructions of the distributor. 5 μg purified CRM1, Ran wild type or Ran Q69L and PPARγ1 were incubated for 15 minutes in binding buffer. Then, 3 μg of an anti-CRM1 antibody was added for 1 hour at 4°C and 50 μl μMACS protein-A microbeads (Miltenyi Biotech, Bergisch-Gladbach, Germany) were added and incubated for 1 hour at 4°C. Lysate was applied onto an equilibrated μ column, already placed in the magnetic field of a μMACS separator. The flowthrough was collected and saved for further analysis. The column was rinsed four times with 200 μl lysis buffer followed by two washes with low ionic buffer (20 mM Tris-HCl, pH 7.5). Then the remaining proteins were eluted using 50 μl boiling SDS sample buffer.

For efficient gene knockdown, a lentiviral shRNA approach was used. Validated MISSION shRNA plasmids for human CK-IIα (NM_177559; MISSION TRC shRNA Target Set TRCN0000010672, TRCN0000010673, TRCN0000001985, TRCN0000001986, TRCN0000001987), human ERK1 (NM_002746; MISSION TRC shRNA Target Set TRCN0000006150, TRCN0000006151, TRCN0000006152, TRCN0000010997, TRCN0000010998) and human ERK2 (NM_138957; MISSION TRC shRNA Target Set TRCN0000010039, TRCN0000010040, TRCN0000010041, TRCN0000010050, TRCN0000195517) as well as Mission Non-Target shRNA control vector as a negative control were obtained from Sigma. For generation of lentiviral particles, 3×10⁶ HEK293T cells were seeded in 5 ml complete DMEM medium. The following day, cells were transfected using jetPEI (Biomol, Hamburg, Germany) according to the distributor's instructions. Briefly, 3 μg of the shRNA plasmid in combination with 26 μl of the packaging mix (Sigma) and 6 μl jetPEI (Biomol, Hamburg, Germany) were used. Afterwards, cells were cultured for 2 days. Then, medium was changed and after a further 2 days, the supernatant containing the infectious lentiviral particles was taken. To transduce target cells, lentivirus containing culture supernatants was filtered through a 0.45 μm filter (Millipore, Schwalbach, Germany) and thereafter directly spun twice onto HEK293 cells (90 min, 500 g, RT). Transduced HEK293 cells were grown for 1 week in complete medium before puromycin was added for antibiotic selection of positive clones. Following 2 weeks of puromycin selection, knockdown efficiency was analyzed at the protein level by western blot analysis.

To overexpress a constitutively active ERK2-MEK1-LA protein, we transiently transfected HEK293 cells with a vector encoding ERK2-MEK1-LA kindly provided by Melanie H. Cobb (Robinson et al., 1998) using jetPEI (Biomol, Hamburg, Germany) according to the distributor's instructions. Expression was verified by western blot analysis.

To isolate cytosolic and nuclear fractions, 1×10⁷ HEK293 cells were lysed, basically as described (Nelson et al., 2006). Briefly, following cell activation for the times indicated, 4×10⁶ cells were washed in 1 ml ice-cold PBS, centrifuged at 1000 g for 5 minutes, resuspended in 200 μl ice-cold cell lysis buffer [150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 5 mM EDTA, NP-40 (0.5% v/v), Triton X-100 (1.0% v/v), 0.5 mM PMSF, 1 μg/ml leupeptin, 10 mM NaF], lysed by pipetting up and down several times, and centrifuged at 12,000 g for 1 minute. Aliquots of the supernatant containing cytosolic proteins were frozen in liquid nitrogen and stored at −70°C. Sedimented nuclei were resuspended in 50 μl ice-cold saline buffer (50 mM HEPES-KOH, 50 mM KCl, 300 mM NaCl, 0.1 mM EDTA, 10% glycerol, 1 mM DTT, 0.5 mM PMSF, pH 7.9, 1 μg/ml leupeptin, 10 mM NaF), left on ice for 20 minutes, vortexed, and centrifuged at 15,000 g for 5 minutes at 4°C. Aliquots of the supernatant, containing nuclear proteins, were frozen in liquid nitrogen and stored at −70°C. Protein was determined using a Bio-Rad II kit.

Each experiment was performed at least three times. Statistical analysis was performed using the paired t-test. We considered P-values ≤0.05 as significant. Otherwise representative data are shown.

This work was supported by a grant from the Deutsche Forschungsgemeinschaft (KN493/9-1 and SFB TP3 and TP8) and Deutsche Krebshilfe (10-2008). We thank Mien-Chie Hung (Department of Molecular and Cellular Oncology, Anderson Cancer Center, Houston, TX) and Rafael Pulido (Centro de Investogación Príncipe Felipe, Valencia, Spain) for providing the Ran wild type and Ran Q69L vectors; John Blenis (Department of Cell Biology, Harvard Medical School, Boston, MA) for sending us the FLAG-tagged RanBP3 construct; Maarten Fornerod (Department of Tumor Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands) for providing the z-tagged CRM1 vector, Ran wild type and Ran Q69L constructs for bacterial expression; and Dirk Görlich (Department for Cellular Logistics, MPI für Biophysikalische Chemie, Göttingen, Germany) for the permission to use the latter two vectors. We thank Nadja Wallner for expert technical assistance.