MEKK1 is required for inducible urokinase-type plasminogen activator expression.

Urokinase-type plasminogen activator (uPA) regulates the remodeling of extracellular matrix and controls reparative processes such as wound healing and liver regeneration. Here we show inducible uPA expression is controlled by MEKK1, a MAPK kinase kinase that regulates the ERK1/2 and JNK pathways. MEKK1 is activated in response to growth factors and cytoskeletal changes. We have found MEKK1 to be necessary for uPA up-regulation in response to treatment with phorbol 12-myristate 13-acetate or basic fibroblast growth factor. We demonstrate that growth factor-treated MEKK1-deficient fibroblasts display greatly reduced uPA expression and activity compared with control fibroblasts. Further, we show that growth factor-induced uPA expression requires MEKK1-dependent MKK1 and JNK activity and that transfection of MEKK1 into knockout cells restores inducible uPA expression and activity. Importantly, disrupted expression of MEKK2, a related MAPK kinase kinase, had no effect on uPA activity. Therefore, we conclude that MEKK1 expression is required for PMA- or FGF-2-induced signals to control uPA expression and function.

Tissue remodeling in response to stress is a critical homeostatic function in multiple organ systems. One physiological process involving tissue remodeling is that of wound healing. Wound healing is a complex process involving multiple cell types and regulated proteolytic degradation of extracellular matrix components (1). The loss of activity of some key proteases results in an impaired ability to repair wounds (2)(3)(4). Among the important enzymes expressed and secreted during the wound healing response is the serine protease uPA 1 (2,5,6). Known uPA substrates are diverse, and include the zymogen plasminogen (7). uPA activity promotes a proteolytic cascade by converting plasminogen to its active form, plasmin, which can then cleave and activate matrix metalloproteinases (8), thereby enhancing tissue remodeling processes by degrading extracellular matrix components. Beyond wound healing, uPA is required for liver regeneration (9), whereas disregulation of uPA activity plays a major role in human diseases including cancer metastasis (7,10). As uPA can have a profound impact on homeostasis, both activity and expression are subject to tight regulation. uPA activity is controlled posttranslationally through cleavage to produce an activated form (7) as well as through localization by binding to its cognate receptor (uPAR) (7). uPA is also regulated at the transcriptional level. AP-1 transcription factor duplexes play a major role in control of inducible uPA expression through binding of enhancer regions in the promoter of the uPA gene (11). AP-1 complex components and assembly are, in turn, subject to regulation by MAPK activity. Transcription factor and AP-1 component c-Jun may be phosphorylated and consequently activated by the MAPK c-Jun NH 2 -terminal kinase (JNK) (12). JNK may also phosphorylate the CREB (cAMP-responsive element-binding protein) family member ATF-2. ATF-2/c-Jun heterodimers have been shown to be a predominant form of AP-1 that promotes uPA expression (11), and therefore JNK activity is an integral part of uPA regulation. Another MAPK, extracellular signal-regulated kinase (ERK1/2), is also key to AP-1 regulation, as its activation leads to the induction of c-Fos, which may join with c-Jun to form a second AP-1 heterodimer complex for the regulation of uPA expression (12).
MEKK1 is a 196-kDa serine-threonine kinase that functions as an MAPKKK in the JNK pathway and can modulate the ERK1/2 pathway (13). Although regulation of these MAPK pathways has been attributed to other MAPKKKs, genetic studies have begun to define a specific role for MEKK1 in tissue homeostasis. Targeted disruption of the MEKK1 gene in mice results in a failure to activate JNK in response to aortic banding, and MEKK1Ϫ/Ϫ mice display reduced tolerance to hypertensive cardiac insult (14). Indeed, MEKK1-deficient mice show an increased propensity to cardiac rupture and congestive heart failure consistent with defective tissue remodeling capability. 2 We have previously shown MEKK1 to be activated in response to an array of stimuli, including epidermal growth factor receptor ligation (15) and cytoskeletal alteration by treatment with nocodazole and taxol (13,16). Further, we have found serum and lysophosphatidic acid-induced JNK activation to be reduced in MEKK1-deficient embryonic stem cells (13). MEKK1 is thus a MAPKKK that regulates MAPK signaling in response to cytoskeletal changes and a diverse group of growth factors.
In this work, we demonstrate that MEKK1 regulates uPA expression in response to growth factor receptor ligation and by phorbol ester stimulation. Our data are consistent with MEKK1 being a junction that integrates different signals to common pathways required for uPA expression. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-The anti-phospho JNK and anti-phospho-ERK monoclonal antibodies and the rabbit polyclonal antibody against ERK2 were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The rabbit polyclonal antibody against uPA was a generous gift of the Finsen Laboratory (Copenhagen, Denmark). HRP-sheep antimouse IgG was purchased from Amersham Biosciences. Protein A-HRP conjugate was purchased from Zymed Laboratories (San Francisco, CA). The constructs for full-length uPA and the uPA promoter were purchased from American Type Culture Collection (Manassas, VA).
Cell Lines-Both primary and immortalized mouse embryo fibroblasts (MEFs) were obtained as described previously (17).
Northern blot Analysis-Stimulated cells were lysed, and RNA was extracted with RNeasy (Qiagen, Valencia, CA) according to manufacturer's recommendations. RNA analysis by Northern hybridization was performed as described previously (33). The membrane was hybridized with a 460-bp uPA cDNA fragment (EcoRI digest).
Transient Transfection-Adherent cells were transfected by lipofection. Briefly, 4 ϫ 10 5 cells were seeded on 100-mm cell culture plates and incubated in complete medium overnight. cDNA expression constructs were incubated in serum-free medium with LipofectAMINE (Invitrogen) at room temperature for 30 min and then diluted with serum-free medium and incubated with cells at 37°C for 2 h, after which time the LipofectAMINE mixture was replaced with complete medium and the cells returned to 37°C for 24 h. Complete medium was then removed, the cells were rinsed, and incubation was continued with serum-free medium for an additional 24 h. For production of stable transfectants, cells were subjected to selection and maintained in the presence of G418.
Immunoblotting-Proteins were separated by SDS-PAGE and then transferred to nitrocellulose (Schleicher & Schuell). Membranes were blocked in 5% milk (diluted in Tris-buffered saline and 0.1% Tween 20) and incubated with the appropriate antibody at 4°C overnight. HRPprotein A or HRP-sheep anti-mouse IgG was used as secondary reagent. After extensive washing, the targeted proteins were detected by enhanced chemiluminescence (ECL). Where indicated, blots were stripped by treatment with 2% SDS and 100 mM 2-mercaptoethanol in Trisbuffered saline and then reprobed with the desired antibodies and detected by ECL.
Luciferase Assay-The Luciferase assay kit was purchased from Promega (Madison, WI), and assays were carried out according to manufacturer's recommendations.
uPA Colorimetric Assay-Urokinase-specific chromogen S2444 was purchased from DiaPharma Group Inc. (West Chester, OH). Tissue culture medium was removed from the stimulated MEFs, diluted with S2444 in Tris buffer (50 mM Tris and 38 mM NaCl), and incubated at 37°C for 24 h. uPA activity cleaves S2444 to produce p-nitroaniline, which was assessed by color change at 405 nm.
Fluorescent Microscopy-Immortalized, serum-starved MEFs were attached to coverslips and stimulated with 100 nM PMA for the indicated times. Cells were fixed in 3% paraformaldehyde, 3% sucrose pH 7.4, washed, permeabilized with 0.2% Triton, and blocked in 4% bovine serum albumin. Cells were bound by anti-uPA and Cy3-conjugated anti-rabbit.
Zymogram-To analyze uPA activity in MEF cell culture supernatants 1.2 ϫ 10 6 MEFs were plated into a 10-cm tissue culture dish and allowed to adhere overnight. Medium was then removed, and the cells were rinsed with 5 ml of serum-free IMDM to remove residual serum. Cells were incubated for 24 h in 4 ml of serum-free IMDM with either vehicle or stimuli. Cell-free supernatant (150 l) samples were mixed with 50 l of nondenaturing sample buffer (40% glycerol, 8% SDS, 250 mM Tris-HCl, pH 6.8, 0.05% bromphenol blue), and proteins were separated within 10% SDS-PAGE co-polymerized with 0.025 unit of plasminogen (Calbiochem) and 2 g/ml ␣-casein (Sigma). The gel was washed for 1 h in 300 ml of 2.5% Triton X-100. The gel was rinsed three times with water to remove the Triton X-100 and rocked for an addi-tional 2 h in water. For casein proteolysis, the gel was immersed in zymography enzyme buffer (25 mM Tris, pH 7.4, 100 mM NaCl, and 10 mM CaCl 2 ) and incubated for 20 -24 h at 37°C. Following this incubation the gel was stained for 2 h in Coomassie Blue stain (40% methanol, 10% acetic acid and 2.5 g/liter R250 Coomassie Blue dye (Sigma)) and destained with multiple changes of destain solution (30% methanol, 10% acetic acid) until light bands appeared on a blue background.

MEKK1
Regulates PMA-induced uPA Expression-Wound healing requires coordinated proteolysis of various wound matrix components (1). We observed that MEKK1-deficient mice developed wounds, through over-grooming or fighting, that were persistent and healed poorly (not shown). Importantly, these persistent wounds were not observed in mice deficient for MEKK2, a related MAPKKK. This apparent healing defect is very similar to one reported for mice deficient in the protease uPA. Carmeliet and colleagues (18) observed facial wound healing defects in uPA-deficient mice. uPA activity regulates the activation of other proteases necessary for the healing process and has been linked to activation of immunoregulatory cytokine TGF-␤ (19). MEKK1 signaling through ERK1/2 and JNK pathways controls AP-1 activity (12), and uPA expression is transcriptionally regulated through the binding of AP-1-PEA3 complexes to sequences in the uPA promoter (11), which suggested to us that MEKK1 might regulate uPA expression. Thus, we investigated the role of MEKK1 in the regulation of uPA expression. We tested fibroblasts derived from both parental strain and MEKK1Ϫ/Ϫ embryos (MEFs) for uPA expression. PMA induces a dynamic reorganization of the actin cytoskeleton and is a well characterized, potent stimulator of uPA expression (20,21). As we have demonstrated previously that MEKK1 regulates both ERK1/2 and JNK signaling (13), we predicted that MEKK1-deficient cells would show impaired PMA-induced uPA induction. We observed a dramatic reduction in PMA-induced uPA mRNA in MEKK1-deficient fibroblasts as measured by Northern blot analysis (Fig. 1A). This loss of uPA induction observed in Northern blot analysis was mirrored in reporter assays using luciferase-based uPA promoter constructs (Fig. 1, B-D), which indicated that MEKK1 is required for PMA to stimulate uPA promoter activity. Activation of the uPA promoter requires the kinase activity of MEKK1, as luciferase activity of MEKK1Ϫ/Ϫ fibroblasts transfected with a kinase-inactive MEKK1 mutant mimicked that of control cells (Fig 1C). Loss of uPA induction in response to PMA is a function of MEKK1 deficiency because the stable expression of MEKK1 in knockout cells restored regulation of the uPA promoter (Fig. 1, C and D). Thus, our data indicate that MEKK1 is essential for PMA-induced uPA expression.
MEKK1-deficient Fibroblasts Lose uPA Activity-Because our data indicated that MEKK1 plays a role in regulation of uPA expression, we wanted to determine whether the apparent loss of uPA message observed in MEKK1Ϫ/Ϫ cells would result in a corresponding decrease in uPA activity. Consistent with decreased uPA message in MEKK1Ϫ/Ϫ cells, a colorimetric assay for measuring uPA activity confirmed that MEKK1 deficiency is also associated with a reduction in secreted urokinase activity (Fig. 2A). In addition, zymographic analysis demonstrated the requirement of MEKK1 for regulation of uPA activity. We co-polymerized plasminogen and casein in SDS-PAGE and then used this gel to separate proteins of conditioned medium from wild-type and MEKK1Ϫ/Ϫ fibroblast cultures. In a subsequent incubation in reaction buffer, in-gel uPA activated plasminogen to form active plasmin, which then cleaved casein. Upon staining of the gel with Coomassie Blue, areas of casein proteolysis could then be observed as clear bands. Our results demonstrate that PMA stimulates uPA protein expression and secretion from wild type fibroblasts ( Fig. 2B). In contrast, uPA activity is markedly inhibited in MEKK1Ϫ/Ϫ fibroblasts in response to PMA. Immunoblotting confirmed that induction of secreted uPA protein expression in both primary and immortalized fibroblasts is MEKK1-dependent (Fig. 2B). We further verified the requirement of MEKK1 in uPA expression by immunofluorescence analysis. Stimulation of wild-type fibroblasts with PMA resulted in an increased accumulation of uPA in the Golgi complex (Fig. 2C). Although MEKK1Ϫ/Ϫ fibroblasts exhibit a basal level of uPA expression, the presence of Golgi-associated uPA was not observed in PMAstimulated MEKK1Ϫ/Ϫ fibroblasts (Fig. 2C). To confirm the specificity of MEKK1 in regulating fibroblast uPA expression, we repeated the zymography assay with MEKK2Ϫ/Ϫ fibroblasts. Our zymography assays clearly showed that PMA-induced uPA activity in MEKK2Ϫ/Ϫ fibroblasts was not reduced when compared with activity observed in wild-type fibroblasts (Fig. 2D). Altogether, these findings demonstrate that MEKK1 is an MAPKKK that controls uPA expression in fibroblasts.
FGF-2-induced uPA Expression Requires MEKK1-Because MEKK1 is required for PMA-induced uPA expression, we wanted to determine whether MEKK1-deficient fibroblasts were similarly impaired for uPA induction in response to growth factors that activate ERK1/2 and JNK signaling. When the FGF receptor tyrosine kinase is bound by its cognate ligand fibroblast growth factor 2 (FGF-2), it initiates a potent ERK1/2 signal (22). Further, FGF-2 stimulation induces uPA expression in NIH3T3 fibroblasts (22). Northern blot analysis revealed that FGF-2 was capable of up-regulating uPA expression in wild-type fibroblasts (Fig. 3A). However, as with PMA stimulation, MEKK1-deficient fibroblasts were unable to upregulate uPA expression to FGF-2 (Fig. 3A). Importantly, ligand-induced uPA induction was restored in MEKK1-deficient cells that had been transfected with MEKK1 (Fig. 3A, add  back), demonstrating that FGF-2-induced uPA expression is MEKK1-dependent and not caused by epigenetic differences between cell lines. This loss of uPA expression resulted in a functional loss of uPA activity (Fig. 3B). Consistent with the Northern blot analysis, FGF-2-inducible secreted uPA activity is reduced to background levels in MEKK1-deficient cells (Fig.  3B). Furthermore, expression of wild-type MEKK1 in MEKK1Ϫ/Ϫ fibroblasts, either through transient transfection (Fig. 3C) or in stably transfected cell lines (Fig. 3B, add back), restored uPA activity. Furthermore, this restoration of uPA activity absolutely requires MEKK1 activity, as expression of a kinase-inactive MEKK1 mutant did not rescue uPA activity (Fig. 3C). Our results indicate that MEKK1 is absolutely required for FGF-2-induced uPA up-regulation in fibroblasts.

MEKK1 Regulates MAPK Signaling Downstream of FGF-2 Receptor
Ligation-Control of uPA expression is largely dependent on the binding of the consensus AP-1-PEA3 sequence within the uPA promoter (21,23). Both ERK1/2 and JNK signaling impact AP-1 activity (12), and we have previously shown that MEKK1 activity plays a role in both pathways (13). To determine whether MEKK1 expression is necessary for FGF-2-induced ERK1/2/JNK signaling, we stimulated either wild-type or MEKK1-deficient fibroblasts and assessed MAPK activity by immunoblot with antibodies specific for activated ERK or JNK (Fig. 4A). We clearly show that ligand-induced ERK and JNK phosphorylation is dramatically reduced in the MEKK1-deficient samples (Fig. 4A). Importantly, other FGF-2-induced pathways, such as phosphatidylinositol 3-kinase activation and consequent AKT phosphorylation, remain intact in MEKK1-deficient cells (Fig. 4B). This finding demonstrates that disruption of MEKK1 expression specifically impacts MAPK signaling.
We wanted to determine whether ERK1/2, JNK, or both of these signaling pathways were responsible for FGF-2-induced uPA activity. We therefore repeated the zymography assay using wild-type fibroblasts treated with the MEK inhibitor UO126 or the JNK inhibitor SP600125. Analysis of uPA activity revealed it to be highly sensitive to MEK inhibition, as U0126 treatment blocked FGF-2-induced uPA expression (Fig.  5). Interestingly, treatment with p38 inhibitor SB203586 enhanced FGF-2-induced uPA activity (Fig. 5). We conclude that regulation of both the ERK1/2 and JNK signaling pathways, a property of MEKK1 signaling, is necessary for FGF-2-induced uPA expression. Altogether, our data are consistent with JNK and ERK1/2 activities, modulated by MEKK1, functioning in concert to regulate uPA expression. DISCUSSION Protease-dependent tissue remodeling plays an important role in normal homeostasis and wound healing and is a factor in significant human pathologies including myocardial infarction (24) and cancer metastasis (7). Further, uPA has been linked to arterial neointima formation and vascular wound FIG. 1. MEKK1 is required for uPA expression. PMA-induced uPA expression was assessed by Northern blot (A) and luciferase reporter assays (B-D). A, wildtype and MEKK1Ϫ/Ϫ fibroblasts were treated with 100 nM PMA for 18 h. RNA (5 g) was probed for uPA and ␤-actin. B, wild-type and MEKK1Ϫ/Ϫ immortalized fibroblasts were transfected with a uPApromoter/luciferase (uPA/Luc) construct and treated with 100 nM PMA for 18 h. Fold increase in uPA induction was assessed by luciferase assay. C, MEKK1Ϫ/Ϫ fibroblasts were transiently transfected with empty vector, wild-type MEKK1, or kinase-inactive MEKK1 K1253M together with uPA/Luc and treated with PMA. D, MEKK1Ϫ/Ϫ fibroblasts clones that stably express wild-type or kinase-inactive MEKK1 were treated with PMA. Both C and D were assayed for uPA induction as described in B. The results shown are the mean Ϯ S.E. of at least three independent experiments. healing in mouse models (25,26). The protease plasmin, a vital component in these processes, itself is capable of degrading multiple matrix components. Further, plasmin regulates the activity of matrix metalloproteinases (27) by cleaving and consequently activating inactive precursors. As unrestricted proteolysis would be deleterious to the organism, plasmin activity is tightly regulated. Proteins upstream of plasmin that activate this protease cascade are therefore an integral part of the remodeling process. uPA can activate plasmin by cleaving plasminogen, the precursor zymogen form of plasmin. Understanding the regulation of uPA expression is therefore critical in defining the molecular basis of normal reparative processes and the pathophysiology of multiple diseases. Indeed, the capacity to remodel the mix of components that comprise the extracellular matrix is of paramount importance for maintaining homeostasis.
As uPA activates plasmin, the loss of uPA expression observed in MEKK1Ϫ/Ϫ fibroblasts would be predicted to result in a reduced capacity to remodel matrix components. This defect could be manifested by an impaired ability to heal wounds, and in fact, this very defect has been observed in uPA-deficient mice (18). Our observation that facial wounds persisted in our MEKK1-deficient mouse colony suggests that MEKK1 is involved in the wound healing process. Mice lacking MEKK1Ϫ/Ϫ have an inhibited wound healing response that mimics uPA deficiency and that is not observed in MEKK2 knockout mice. What properties of MEKK1 make it an important MAPKKK for tissue remodeling? First, MEKK1 is activated in response to changes in cytoskeletal structure (13,16), cell shape, and adherence (13,16) and in response to epidermal growth factor and lysophosphatidic acid (13,15). In contrast, other MAPKKKs, like MEKK2, appear to be involved primarily in responses to specific growth factor and cytokine receptors but not in changes to the cytoskeleton, adherence, or cell shape (28). Second, MEKK1 interacts with Rac and Cdc42, important GTPases in the control of the cytoskeleton, cell shape, and migration. Finally, MEKK1 regulates the JNK and ERK1/2 pathways, MAPKs that have been shown clearly to be involved in the regulation of uPA expression. We have demonstrated that MEKK1-deficient fibroblasts do not have the ability to up-regulate uPA in response to FGF-2 and PMA. Further, we provide evidence that FGF-2-induced uPA expression specifically requires MEKK1-dependent JNK and ERK1/2 activities. It is intriguing that, whereas other MAPKKKs are known to regulate ERK1/2 or JNK in response to specific stimuli, MEKK1 remains indispensable for uPA up-regulation in response to these growth factors. Growth factor-dependent uPA expression is blocked in MEKK1-deficient fibroblasts, demon- strating that other MAPKKKs such as Raf (which regulates ERK1/2 signaling) and MEKK2 (which controls JNK) are unable to compensate for the loss of MEKK1 activity. One reason for this requirement may be that, by its ability to coordinately regulate both ERK1/2 and JNK, MEKK1 is uniquely suited to control uPA induction. Alternately, MEKK1 may be needed for effective MAPK signaling in response to these ligands. MEKK1 has been found to associate with cytoskeletal (29) and focal adhesion proteins, 2 suggesting that MEKK1 localization may figure prominently in the organization of signaling complexes.
As uPA protein was detectable by immunoblot in all of the basal fibroblasts tested, we conclude that MEKK1 expression is not absolutely required for basal uPA expression. Thus, other regulatory pathways independent of MEKK1 regulate uPA expression. In contrast to inducible uPA expression, basal uPA transcription has been shown to be driven by SP1 binding to the proximal promoter (30). SP1-dependent transcription is regulated differently than that reliant upon AP-1 (12,31) and may in part be regulated through ERK1/2 signaling (32). Indeed, our results showing basal uPA activity to be sensitive to MEK but not JNK inhibition is consistent with SP1-dependent constitutive transcription being controlled by the ERK1/2 pathway, independently of JNK activity (Fig. 4). There are additional MAPKKKs capable of activating the JNK and ERK1/2 pathways that are regulated by upstream stimuli different from those that activate MEKK1. The MEKK1 and MEKK2 knockouts have defined their selective roles in regulating different physiological functions. The pathological consequences observed with MEKK1 or MEKK2 deletion are different, even though the downstream MAPK signaling pathways remain intact. For this reason, MAPKKK knockouts, like those of MEKK1 and MEKK2, can be more subtle than total loss of a signaling pathway, such as found for the JNKs, but can be much more telling in regards to physiological regulation in adult animals. The MEKK1 knockout defines MEKK1 as a critical MAPKKK controlling the expression of uPA from multiple stimuli. MEKK1 expression is required for uPA induction downstream of both receptor tyrosine kinase (FGF receptor) and protein kinase C activation (PMA), and thus these pathways converge at or upstream of MEKK1 (Fig. 6). Our data place MEKK1 at the nexus of the signaling pathways that control uPA expression and, as such, reveal the utility of modulating MEKK1 as a means of regulating uPA-dependent cell functions. The control of proteolytic degradation of extracellular matrix required for tumor metastasis is just one example of such a function. The importance of MEKK1 in these processes validates the use of MAPKKKs as targets for inhibition by small molecule inhibitors in specific diseases including cancer metastasis.