ERK and the F-box protein β TRCP target STAT1 for degradation.

The transcription factor STAT1 has roles in development, homeostasis, cellular differentiation, and apoptosis and has been postulated to function as a tumor suppressor. STAT1 is activated by tyrosine or serine phosphorylation in response to specific cytokines or following a variety of stress-induced stimuli. STAT1 activity is carefully regulated to prevent sustained STAT1-mediated transcription, although the molecular mechanisms involved in the modulation of STAT1 stability are poorly understood. Here we show that activated STAT1 is degraded at the proteasome by a mechanism involving the F-box E3 ligase, SCF(betaTRCP). Active p42/p44 MAPK-ERK phosphorylates STAT1 on serine 727 and targets it for proteasomal degradation. SCF(betaTRCP) binds wild-type STAT1 but not the nonphosphorylatable mutant STAT1(S727A). Moreover, silencing betaTRCP expression or pharmacological inhibition of ERK activity stabilized STAT1 expression. These data suggest that constitutively active ERK may inappropriately degrade STAT1, with loss of its pro-apoptotic and tumor suppressor functions.

The transcription factor STAT1 has roles in development, homeostasis, cellular differentiation, and apoptosis and has been postulated to function as a tumor suppressor. STAT1 is activated by tyrosine or serine phosphorylation in response to specific cytokines or following a variety of stress-induced stimuli. STAT1 activity is carefully regulated to prevent sustained STAT1-mediated transcription, although the molecular mechanisms involved in the modulation of STAT1 stability are poorly understood. Here we show that activated STAT1 is degraded at the proteasome by a mechanism involving the F-box E3 ligase, SCF ␤TRCP . Active p42/p44 MAPK-ERK phosphorylates STAT1 on serine 727 and targets it for proteasomal degradation. SCF ␤TRCP binds wild-type STAT1 but not the nonphosphorylatable mutant STAT1 S727A . Moreover, silencing ␤TRCP expression or pharmacological inhibition of ERK activity stabilized STAT1 expression. These data suggest that constitutively active ERK may inappropriately degrade STAT1, with loss of its pro-apoptotic and tumor suppressor functions.
The signal transducers and activators of transcription (STATs) 2 are a family of latent cytoplasmic transcription factors. After ligation of cytokine receptors, STATs become phosphorylated by receptor kinases and dimerize and translocate to the nucleus where they modulate expression of STAT-responsive genes (1,2). STAT1 is the classical mediator of the effects of interferon-␥ (IFN␥), and binding of IFN␥ to its receptor results in Janus kinase-mediated phosphorylation of a specific tyrosine (Tyr 701 ) residue found in the C-terminal transcriptional domain of STAT1 (2,3). STAT1 is also phosphorylated, at least in part by the p38 MAPK (4) and ERK pathway (5), on a serine (Ser 727 ) residue, again located in the transcriptional domain, and phosphorylation of both residues appears to be required for maximal transcriptional activity (6).
The function of STAT1 is regulated by various post-translational modifications that act in a highly coordinated manner to integrate signals from secreted cytokines. IFN␥ stimulation induces Tyr 701 phosphorylation and DNA binding of STAT1, but the transcriptional response is also regulated by phosphorylation of Ser 727 in the transcriptional domain. Phosphorylation of Tyr 701 and Ser 727 are independent events, but they act in concert, as tyrosine phosphorylation and presumably dimer formation of STAT1 enhance the serine phosphorylation (5). Moreover, STAT1 Ser 727 phosphorylation is generally considered to have differential effects on STAT1-induced target genes (5).
STAT1 is also known to function as a co-activator for the tumor suppressors p53 and p73 and to regulate their pro-apoptotic properties (7,8). STAT1 is also required for modulating cell cycle checkpoints following DNA damage (9). Thus, cells lacking STAT1 have defects in both their intra-S and G 2 /M checkpoints in response to DNA damage (9). These results imply that STAT1 plays pivotal roles in regulating the apoptotic as well as cell cycle checkpoint responses following DNA damage and is evidence that STAT1 may function as a tumor suppressor. Indeed, STAT1/p53-deficient mice have an increased incidence of spontaneous tumors compared with single p53deficient mice (10), strongly suggesting that STAT1 and p53 cooperate to promote tumorigenesis.
The ubiquitin-proteasome pathway plays a central role in regulating many cellular processes such as cell cycle progression by targeting phosphorylated regulatory proteins for degradation via distinct F-box-associated E3 ligases (11). Overactivity in the degradation of tumor suppressor proteins is an underlying mechanism that promotes tumorigenesis (11). Several reports have identified E3 ligases that specifically target STAT1 for degradation. This has been particularly well investigated in the context of viral infection. For example, the simian viral protein SV5 enables the virus to circumvent the interferon-mediated host cell anti-viral response by degrading STAT1 (12,13). The SV5 protein interacts with the damaged DNAbinding protein-1 (DDB1) and Cul4A to form an active E3 ligase. More recently, the STAT-interacting LIM protein has also been shown to form an E3 ligase that targets STAT1 and STAT4 proteolysis via the ubiquitin-proteasome pathway (14). However, E3 ligases that specifically target phosphorylated forms of STAT1 for proteolysis in mammalian cells have not been characterized. In this study, we provide evidence that the F-box E3 ligase ␤TRCP interacts with and promotes STAT1 proteasomal degradation in an ERK-dependent manner.
Cell Culture, Cell Lines, Agents, and siRNA Transfection-STAT1-deficient MEF cells (STAT1 Ϫ/Ϫ MEF) and their wildtype counterparts were maintained in supplemented Dulbecco's modified Eagle's medium containing 10% v/v fetal bovine serum and nonessential amino acids and were a gift from David E. Levy. The leukemic cell lines were purchased from American Type Culture Collection (ATCC). Urocortin and MG132 proteasomal inhibitor were purchased from Sigma, and MEK1 inhibitor U0126, the p38 inhibitor SB203580, and lactacystin were purchased New England Biolabs Inc., Beverly, MA.
In Vitro Kinase Assays-Kinase reactions were carried out in 40 l of kinase buffer (25 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 1 mM CaCl 2, 1 mM dithiothreitol, 1.25 g of phosphatidylserine) containing 10 Ci of [␥-32 P]ATP (3000 Ci mmol Ϫ1 ; PerkinElmer Life Sciences) together with recombinant active ERK (1 unit) (New England Biolabs) and either myelin basic protein (Invitrogen) or recombinant wild-type STAT1␣ or STAT1␤, mutant STAT1 Y701F , or mutant STAT1 S727A (kindly provided by Uwe Vinkemeier) proteins incubated at 25°C for 20 min. Reactions were terminated by addition of SDS sample buffer, and the 32 P-incorporated proteins were fractionated by SDS-PAGE and visualized by autoradiography. Membranes were subsequently blotted with the anti-STAT1 (E-23) antibody to reveal STAT1 quantities.
Confocal Microscopy-STAT1 Ϫ/Ϫ MEF cells were seeded on glass coverslips at a density of 10 3 cells per 13-mm coverslip and transfected the following day using FuGENE 6 and incubated for another 24 h. Cells were fixed in paraformaldehyde, permeabilized in paraformaldehyde, 0.1% v/v Triton X-100, and stained using antigen specific antibodies as outlined (8). Alexa-Fluor 488 donkey anti-mouse IgG, Texas Red goat anti-rat conjugates were used as secondary antibodies (10 g/ml) and were purchased from Molecular Probes (Invitrogen). Coverslips were mounted in Vecta Shield mounting mix containing 4Ј,6diamidino-2-phenylindole (Vecta Laboratories, UK), visualized at room temperature using a Leica TCS SP2 laser scanning confocal microscope with 63ϫ objectives, and analyzed using LCS Lite (Leica) and Adobe Photoshop 6.0 software.

STAT1 Levels Are Regulated by the Ubiquitin Pathway-
During ongoing studies on STAT1 activation, we had observed that levels of unphosphorylated or phosphorylated STAT1 are reduced over time following various stressful stimuli, suggesting that STAT1 is modulated during cellular stimulation. To understand the mechanism for this rapid removal of activated STAT1, we examined the effects of H 2 O 2 -mediated oxidative stress in mouse embryonic fibroblasts (MEFs). H 2 O 2 resulted in phosphorylation of STAT1 Tyr 701 and STAT1 Ser 727 , and this was accompanied by a reduction in the levels of both total STAT1 and phospho-STAT1 (Fig. 1A). Similar observations were also observed following exposure to the DNA-damaging agent cisplatin (data not shown). Treatment with the proteasomal inhibitor, lactacystin (Fig. 1A), reduced STAT1, phospho-STAT1 Tyr-701 , and STAT1 Ser-727 degradation (Fig. 1A). These results imply that STAT1 and phosphorylated STAT1 levels are regulated by the proteasomal pathway. We also confirmed that STAT1 becomes ubiquitylated, and this is enhanced following H 2 O 2 treatment in MEF cells as assessed by co-immunoprecipitation assays (Fig. 1B). We also show that ubiquitylated STAT1 co-localizes with the 20 S proteasome following H 2 O 2 (Fig. 1C). These finding demonstrate that STAT1 levels are modulated by the ubiquitin pathway.
ERK Phosphorylates and Modulates STAT1 Levels-Phosphorylation of regulatory proteins can act as a trigger for their degradation by the ubiquitin-proteasomal pathway (11). To determine whether proteasomal degradation of STAT1 also required its prior phosphorylation, and which kinase was involved, we studied the effects of various kinase inhibitors on STAT1 degradation following oxidative stress. The MEK-ERK inhibitor U0126, but not the p38 kinase inhibitor SB203580, reduced STAT1 degradation, suggesting that ERK-mediated STAT1 phosphorylation was important for its proteasomal breakdown following oxidative stress. The effectiveness of these MAPK kinase inhibitors was confirmed by reduced levels of activated phospho-ERK (pERK) and phospho-p38 respectively ( Fig. 2A).
Next we investigated the mechanism of ERK-mediated STAT1 proteolysis and examined whether activated pERK is able to interact with and directly phosphorylate STAT1␣. Coimmunoprecipitation assays using a specific anti-STAT1 antibody confirmed a weak STAT1 and pERK interaction following oxidative stress (Fig. 2B) and not under normal conditions (data not shown). Moreover, co-localization of pERK and STAT1 was confirmed in vivo by confocal microscopy following oxidative stress (Fig. 2C). A conserved STAT1 serine at position 727 has a  possible ERK consensus phosphorylation motif. Recombinant active ERK readily phosphorylated recombinant wild-type STAT1 (Fig. 3A) in an in vitro kinase assay. In contrast, recombinant active pERK was not able to phosphorylate a STAT1 S727A mutant but was still able to phosphorylate a STAT1 Y701F mutant (Fig. 3A). In addition, recombinant active pERK was unable to phosphorylate STAT1␤ (Fig. 3A), a C-terminal truncated STAT1 that lacks STAT1 serine 727 but retains tyrosine 701.
Next, to further examine the involvement of the MAPK-ERK cascade in STAT1 degradation, we assessed the ability of constitutively active MEK1 to promote STAT1 degradation. Cells expressing constitutively active MEK1 and exposed to oxidative stress indeed showed enhanced STAT1 degradation, suggesting that STAT1 levels are inversely correlated with enhanced pERK activity (Fig. 3B). These experiments suggest that STAT1 is efficiently phosphorylated by ERK and confirms that STAT1 Ser-727 is the major ERK substrate.
␤TRCP Interacts Specifically with Phospho-STAT1 Serine 727-The targeting of specific proteins for proteasomal degradation is governed by the F-box protein subunit of the SCF (SKP1-CUL-F-box) complex that directly recruits specific substrates (11). HA-tagged F-box proteins were transfected into MEF cells for 48 h. The transfected cells were then treated with H 2 O 2 for 4 h to activate ERK. The cells were lysed and the lysates incubated with GST-STAT1. Fig. 4A shows that the GST-STAT1 (which will be serine-phosphorylated by the active endogenous ERK in the lysate) interacts with Fbw2 and Fbw1/␤TRCP but not Fbw4. No interactions were seen in lysates treated with the ERK inhibitor U0126 (data not shown).
Because Fbw1/␤TRCP has been well documented to play a role in targeting key checkpoint cell cycle regulators, and we have also implicated STAT1 in checkpoint regulation (11), we focused the rest of our studies on the role of the STAT1/␤TRCP association. To investigate whether endogenous STAT1 and ␤TRCP interact, we immunoprecipitated ␤TRCP from the same cell lysates used in the experiments shown in Fig. 4A (IP-␤TRCP) and blotted with an anti-total STAT1 antibody (Fig.  4B). The data show that IP with the anti-␤TRCP antibody (but not the control preimmune antibody) results in pull down of STAT1. ␤TRCP and STAT1 were also co-localized, and at least a portion of the co-localized proteins was associated with a proteasomal marker (Fig. 4C).
Next, we examined whether the STAT1/␤TRCP interaction is dependent on STAT1 phosphorylation. Wild-type STAT1 or mutant STAT1 Y701F immunoprecipitated efficiently with ␤TRCP, but STAT1␤ or a STAT1 S727A mutant did not (Fig.  4D). This indicates that serine 727 is essential for this interaction and suggests that serine 727 may be phosphorylated in vivo and mediate the binding of STAT1 with ␤TRCP.
To confirm that ␤TRCP inhibition stabilizes STAT1, we next tested whether direct inhibition of ␤TRCP expression by siRNA also inhibited degradation of STAT1. As shown in Fig.  5A, human fibrosarcoma cells (3fTGH) transfected with ␤TRCP siRNA but not a control siRNA showed more abundant STAT1 protein expression following treatment with cycloheximide and H 2 O 2 . Similarly in MEF cells, knock down of ERK1 or ERK2 by siRNA also resulted in increased STAT1 but not STAT3 protein expression (Fig. 5B). Moreover, ␤TRCP siRNA knockdown increased expression of its known Cdc25A substrate (Fig. 5B). Overexpression of ␤TRCP also reduced the expression of serine 727-phosphorylated STAT1 following oxidative stress, and this was abolished by proteasome inhibition (Fig. 5C). These results further support the suggestion that STAT1 degradation is mediated by ␤TRCP.
To further investigate the importance of serine 727 in STAT1 degradation, we re-expressed STAT1 wild-type or STAT1 S727A mutant forms into STAT1 Ϫ/Ϫ MEF cells. Compared with wild-type STAT1, the half-life of the STAT1 S727A mutant was increased following treatment with cycloheximide and H 2 O 2 (Fig. 6A). Similar results were also obtained in human STAT1-deficient U3A fibrosarcoma cells stably expressing wild-type STAT1 or the STAT1 S727A mutant (data not shown). These results confirm that the phosphorylation state of serine 727 is an important determinant of STAT1 stability by triggering its proteasomal degradation.
STAT1 and ERK Kinase Activity Inversely Correlate in Leukemic Cell Lines-The finding that ERK is frequently activated in a variety of cancers, including acute myeloid leukemias (16), prompted us to test whether STAT1 levels and ERK activity are inversely correlated in leukemic cell lines. We have preliminary data showing that levels of constitutive phospho-STAT1 serine 727 varied significantly (compared with phospho-STAT3 serine 727) in a panel of leukemic cell lines (data not shown). We next examined whether the activity of the ERK pathway influences STAT1 levels in two of the cell lines, Ramos, which was a low STAT1 expressor, and RL, a high STAT1 expressor. Constitutive ERK activity was significantly higher in Ramos compared with the RL cell line (Fig. 6B). Treating the cells with the MEK1-ERK pharmacological inhibitor U0126 not only reduced pERK activity and decreased constitutive STAT1 serine 727 phosphorylation but also enhanced STAT1 levels, suggesting that pERK activity is associated with reduced levels of STAT1 (Fig. 6B).
Furthermore, MEF cells subjected to serum deprivation followed by addition of serum (which contains numerous growth factors) enhanced phospho-ERK, and this was also associated with reduced expression of STAT1 (Fig. 6C). Importantly, this effect was abrogated with the MEK1-ERK pharmacological inhibitor U0126. In addition, the cytokine urocortin, which we have studied extensively in the past (16), is also able to enhance phospho-ERK and reduce STAT1 expression (Fig. 6C). This effect of urocortin on STAT1 is also reversed by the MEK-ERK inhibitor. Thus, this is further evidence that activation of the ERK pathway is linked to STAT1 turnover.

DISCUSSION
STAT1 plays important roles in the interferon-␥ response and also following various stressful stimuli that induce apoptotic or cell cycle checkpoint responses (2,3,5,7,9). However, the molecular mechanisms that govern the stability of activated/phosphorylated STAT1 have not been fully elucidated. Here we demonstrate that STAT1 phosphorylated on Ser 727 is proteasomally degraded by the RING finger domain E3 ligase, ␤TRCP.
Although the E3 ligase SLIM has been previously shown to target unphosphorylated STAT1 (14), our data indicate that ␤TRCP only recognizes serine-phosphorylated STAT1 and suggests that different E3 ligases target STAT1 for proteasomal breakdown depending on its phosphorylation status. Many ␤TRCP substrates, such as IB (17), have a DSGXXS degron consensus motif that allows binding following serine phosphorylation. However, other ␤TRCP substrates, including Wee1 (18), have been found to lack such a consensus site. The interaction of Wee1 with ␤TRCP requires phosphorylation of two serine residues, Ser 53 and Ser 123 by polo-like kinase 1 (Plk1) and CDK, respectively (18). Although the sequence surrounding phosphorylated Ser 53 (DpSAFQE) is similar to the conserved ␤TRCP-binding motif, the role of Ser 123 phosphorylation (EEGFGSSpSPVK) in ␤TRCP binding to Wee1 is unclear.
In this study we demonstrate that the STAT1 Ser 727 site, which also lacks the DSGXXS degron motif, is required for binding to ␤TRCP, because the STAT1 S727A mutant failed to bind the ligase (Fig. 3D). However, because a region less than 20 residues downstream of serine 727 ( 744 DpSMMN 748 ) in STAT1 is comparable with a ␤TRCP consensus sequence, serine 727 phosphorylation may play a similar role in targeting ␤TRCP to this downstream sequence in STAT1 as serine 123 phosphorylation in Wee1. In addition to Wee1, Cdc25A, a checkpoint cell cycle regulator, and Emi1, a negative regulator of the E3 ligase APC, are also degraded by the ␤TRCP pathway (19,20) suggesting that ␤TRCP-mediated degradation may be important for mitotic progression by regulating protein levels of several key cell cycle regulators. It is known that serine 727 phosphorylation of STAT1 can be mediated by at least two kinase cascades, p38 and ERK (4,5,21). The same serine residues in other proteins, such as residue 422 in eIF4B and Ser 123 in Cdc25A, have also been shown to be phosphorylated by more than one kinase (22,23). In STAT1, mutation of serine 727 to alanine results in loss of roughly 80% of IFN␥ responsiveness (21) but has much less effect on the response to type I interferon. These differential effects may also depend on how serine 727 phosphorylation affects recruitment of co-activators/co-repressors, because phosphoserine 727 is required for the interaction between STAT1 and MCM5 (24), thus coupling the transcriptional activity of STAT1 to cell cycle regulation. Moreover, recruitment of the co-activator CBP is strongly reduced at the interferon-responsive GBP promoter in cells expressing mutant STAT1 S727A following IFN␥ treatment (25). STAT1 is expressed as two isoforms STAT1␣ and STAT1␤, which lack the last 38 amino acids of the C-terminal domain, including the serine 727 residue. Therefore, the biological outcome of STAT1␣ serine phosphorylation by different kinases may depend on the cell and stimulus context. However, because STAT1␤ still contains the tyrosine 701 residue, we cannot rule out that this site may also be regulated by post-translational processing.
We have previously demonstrated that the cardioprotective agent urocortin mediates its protective effects against ischemia/reperfusion-induced apoptotic cell death via activation of the MAPK-ERK pathway. Thus, pretreatment with a pharmacological phospho-ERK inhibitor inhibited the survival-promoting effect of urocortin (16). Moreover, apoptosis in cardiac myocytes exposed to ischemia/ reperfusion injury is, at least in part, STAT1-dependent (26). Thus, these data, together with this study, suggest that the cardioprotective effects of urocortin may be mediated via ERK-STAT1 turnover and reduction in pro-apoptotic STAT1 levels.
Here we propose that phosphorylation of serine 727 in STAT1 by p38 and ERK leads to opposite effects. p38-mediated phosphorylation has been reported to maximize STAT1 tran- FIGURE 5. ERK or ␤TRCP and STAT1 serine 7272 is required for STAT1 degradation. A, ␤TRCP silencing reduces STAT1 degradation following oxidative stress. MEF cells were transfected with ␤TRCP siRNA or control luciferase siRNA and exposed to H 2 O 2 at the indicated times. Western blotting was carried out with the indicated antibodies anti-STAT1␣ antibody (C-24) (ST1) or anti-␤TRCP (␤TRCP). Similar results were observed in three independent experiments. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. B, ERK or ␤TRCP silencing results in STAT1 protein up-regulation. MEF cells were transfected with either ␤TRCP-(TP), ERK1-(E1), ERK2-(E2), or luciferase (Lu) (control)specific siRNA oligonucleotides and exposed to H 2 O 2 (200 M). Western blotting was performed with the indicated antibodies. Similar results were observed in three independent experiments. C, exogenous expression of ␤TRCP reduces phospho-STAT1 serine 727 following oxidative stress. Wild-type MEF cells were transfected with HA-tagged ␤TRCP and exposed to H 2 O 2 (200 M) for 4 h alone or in the presence of the proteasome inhibitor lactacystin (LC). Western blotting was performed with the indicated antibodies, anti-phosphoserine STAT1 (pST1) or anti-HA (HA). Similar results were observed in three independent experiments. FIGURE 6. Stability of STAT1 is modulated by the activity of ERK Kinase. A, mutant STAT1 S727A degradation is significantly reduced following oxidative stress. STAT1-deficient MEF cell were transfected with wild-type STAT1 or mutant STAT1 S727A or a green fluorescent protein expression vector and exposed to cycloheximide (CHX) plus H 2 O 2 (200 M) for the indicated times. Western blotting was carried out with the indicated antibodies. In the right panel, the results shown in A were subjected to densitometric analysis and results were normalized against green fluorescent protein (GFP) expression levels, showing wild-type STAT1␣ (ST1-WT, triangles) or mutant STAT1 S727A (ST1-727, squares) levels. Similar results were observed in three independent experiments. B, pharmacologic inhibition of ERK activity enhances STAT1 levels. Leukemic cell lines Ramos and RL were treated with the MEK1 inhibitor U0126 (U) (1 M) for 6 h and analyzed by Western blotting with the indicated antibodies, anti-STAT1␣ antibody (C-24) (ST1) or phospho-ERK (pERK). Similar results were observed in three independent experiments. Lane C, control. C, wild-type MEF cells were arrested by serum starvation for 24 h (0) and then allowed to cycle again following addition of 10% serum (10) or 10% serum plus addition of the MEK1 inhibitor U0126 (1 M) (10 ϩ). FCS, fetal calf serum. After 6 h cell lysates were analyzed by Western blotting with the indicated antibodies, anti-STAT1␣ antibody (C-24) (ST1) or phospho-ERK (pERK). Similar results were observed in three independent experiments (left panel). Wild-type MEF cells were treated with 1 ϫ 10 6 M urocortin (Uc) or urocortin plus the MEK1 inhibitor U0126 (1 M)(Uc ϩ). After 6 h cell lysates were analyzed by Western blotting with the indicated antibodies, anti-STAT1␣ antibody (C-24) (ST1) or phospho-ERK (pERK). Similar results were observed in three independent experiments (right panel). scriptional activity and may mediate the cell cycle arrest induced by IFN␥ through enhancement of the transcription of cell cycle arrest and apoptotic genes. In contrast, proliferative growth factors and other cytokines, which are known to activate ERKs, would target STAT1 for proteasomal breakdown, compromising its transcriptional activity and thus promoting cell cycle progression. This may be reflected in the molecular pathology of some tumors, where constitutively active ERKs lead to STAT1 degradation, and thus allow unregulated cell proliferation.
Our present findings describe a novel pathway for the regulation of STAT1 stability mediated via its phosphorylation and proteasomal degradation in an ERK-␤TRCP-dependent manner. Serine 727 phosphorylation by ERKs, although not by p38 MAPK, is necessary for ␤TRCP binding, although it does not form part of a classic ␤TRCP degron, because the S727A mutant no longer binds ␤TRCP. The high levels of active ERK induced by proliferative growth factors, and which are constitutive in certain tumors, may also have a role in promoting STAT1 degradation and reduce its functional activity.