Phosphorylation of Pax2 by the c-Jun N-terminal kinase and enhanced Pax2-dependent transcription activation.

The Pax gene family encodes DNA-binding proteins that can both activate and repress transcription of specific target genes during embryonic development. Pax proteins are required for pattern formation and cell differentiation in a broad spectrum of developing tissues. Consistent with its expression in the intermediate mesoderm, the optic cup and stalk, and the otic vesicle, Pax2, a member of the Pax2/5/8 subfamily, is essential for the development of the renal epithelia, the optic cup, and the inner ear. In addition to a DNA binding domain, the Pax2 protein contains a carboxyl-terminal transactivation domain rich in serine, threonine, and tyrosine. In this report, we demonstrate that the Pax2 transactivation domain is phosphorylated by the c-Jun N-terminal kinase, but not the ERK1/2 or p38 MAP kinases and that phosphorylation is coincident with increased transactivation of a Pax2-dependent reporter gene. Activation of JNK by either upstream kinase MEKK1 or DLK or by expression of Wnt signaling proteins significantly enhances Pax2 phosphorylation in cells. In vitro kinase assays using immunoprecipitated JNK or constitutively active, recombinant JNK show phosphorylation of GST-Pax2 fusion proteins. In transfected cells, phosphorylation of Pax2 correlates with increased transactivation of a Pax2-dependent reporter gene, suggesting that serine/threonine phosphorylation of the transactivation domain is important for Pax2 activity. Pax2 can form a complex with the JNK scaffolding protein JIP1, and this interaction is enhanced by activation of the JNK signaling module with the upstream kinase DLK. The data demonstrate that Pax2 is a new target for the JNK signaling module and point to a novel mechanism for mediating Pax-dependent transcription regulation.

, and a partial-paired type homeobox. Pax2/5/8 proteins are transcription regulators that bind DNA via the amino-terminal-paired domain, whereas the carboxyl-terminal region is required for transactivation of target genes (2,3).
In mouse and man, the Pax2 gene is essential for the development of the kidneys (4), optic cup, and inner ear (5). Pax2 encodes at least two alternatively spliced messages that produce proteins of 392 and 415 amino acids, differing only by a 23 amino acid insertion (6). Genes known to be up-regulated by Pax2 include WT1 (7) and gdnf (8) in the developing kidney and engrailed-2 (9) in the developing hindbrain. In transfected cells, transcription activation requires the Pax2 carboxyl-terminal domain that is rich in serine and threonine residues, which may be potential sites for phosphorylation. In the zebrafish Pax6 protein, the carboxyl-terminal transactivation domain is phosphorylated at multiple serine residues by the mitogen-activated protein kinases (MAPK) 1 p38 MAPK and ERK1/2 to increase the transactivation potential (10). Of the Pax2/5/8 subfamily, only Pax8 has been studied with respect to phosphorylation, though it is not clear which kinases are involved (11). MAP kinase cascades are involved in transmitting signals generated at the cell surface into the cytosol and nucleus and consist of three sequentially acting enzymes: a MAP kinase, an upstream MAP kinase kinase (MEK), and a MEK kinase (MEKK) (12,13). The extracellular signal-regulated kinase 1/2 (ERK 1/2), the c-Jun N-terminal kinase/stress-activated protein kinases (JNK/SAPK), or the p38MAP kinases, can translocate to the nucleus and subsequentely phosphorylate a variety of transcription factors. MEKK1 phosphorylates MKK4/MKK7 to activate JNK but can also activate ERK and p38 MAPK in transfected cells.
This report addresses the phosphorylation state of the Pax2 protein and its ability to activate transcription. We show that the carboxyl-terminal activation domain is phosphorylated at serine and threonine residues and that Pax2 phosphorylation is coincident with the enhanced ability to transactivate a reporter gene. Pax2 is a substrate for the c-Jun N-terminal kinase (JNK) but, unlike Pax6, is not phosphorylated by ERK or p38 MAPK. Activation of JNK by either the upstream kinases MEKK1 or DLK or by expression of Wnt signaling proteins increases Pax2 phosphorylation and enhances the Pax2 transactivation potential. The data point to an important role for JNK in modifying the Pax2 transactivation domain and stimulating Pax2-dependent gene expression.

MATERIALS AND METHODS
DNA Constructs-The CMV-Pax-2bHA and carboxyl-terminal deletions and the PRS4-CAT reported plasmid, containing five tandem * This work was supported by National Institutes of Health Grants DK54740 and DK54723 (to G. R. D.). 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  copies of the pax2 binding site PRS4 cloned upstream of the herpes simplex virus thymidine kinase promoter, were as described (3). The CMV-MEKK1 wild-type, CMV-MEKK1 K432M, HA-DLK, Myc-Erk, FLAG-JNK, Myc-p38 MAPK, were provided by L. Holzman. The Myc-JIP1 was a gift of B. Margolis. The Wnt4 (gift of A. McMahon), Wnt11, and Wnt7a (IMAGE Consortium) coding sequences were cloned into CMV-CB6 and sequenced for verification. The Pax2-GST fusion protein contained amino acids 197-415 and was purified by affinity to glutathione-agarose. GST-Elk (Cell Signaling Technology, cat. no. 9184S) and GST-ATF-2 (Cell Signaling Technology, cat. no. 9224S) were obtained commercially. GST-c-Jun (1-223) was as described (14). Constitutively active, recombinant JNK was purchased from Upstate Biotechnology (Lake Placid, NY).
Western Blot-Western blots were performed with equal amounts of protein obtained by lysis of transiently transfected cells in PK lysis buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EGTA, 1 mM Na 3 VO 4 , 50 mM NaF, 1% Triton X-100, 10% glycerol, and a mixture of protease inhibitors (Roche Molecular Biochemicals, cat no. 1836170). The lysates were separated by SDS-PAGE and immunoblotted with antibodies indicated in the legends. Horseradish peroxidase-conjugated secondary antibodies were used to detect antigen-antibody complexes by ECL detection system (Amersham Biosciences).
Immunolinked Kinase Assays-Cells were lysed in PK lysis buffer after 40 h of transient transfection. After incubation with antibodies against the epitope of the protein, the immunocomplex was adsorbed to protein-G-Sepharose and washed with kinase buffer (25 mM HEPES, pH 7.4, 20 mM MgCl 2 , 0.5 mM EGTA, 12.5 mM ␤-glycerophosphate, 0.1 mM orthovanadate, 0.5 mM NaF). The complexes were incubated for 30 min at 30°C in 50 l of kinase buffer containing 20 M ATP, 5 Ci of [␥-32 P]ATP (3000 Ci/mmol, Amersham Biosciences), 2 g of GST-Pax2-(197-415), and 2 g of either GST-Elk-1 (ERK assay, Cell Signaling), GST-c-Jun (JNK assay) or GST-ATF-2 (p38 MAPK assay, Cell Signaling). The reactions were terminated by the addition of 12 l of 6ϫ SDS loading buffer, 10 mM EGTA and were subjected to SDS-PAGE followed by transfer to polyvinylidene difluoride membranes and autoradiography. The amount of kinase that was immunoprecipitated was assessed by re-immunoblotting of the nitrocellulose. For the in vitro kinase assay, 50 milliunits of recombinant JNK were used to phosphorylate equal amounts (2 g) of GST-Pax2 or GST-c-Jun in kinase buffer under the same conditions as above or in the presence of excess cold ATP for Western blotting.
In Vivo Labeling-293 cells were transfected with expression plasmids as indicated in the legends. The cells were cultured in phosphatefree Dulbecco's modified Eagle's medium for 2 h after 40 h of transient transfection, and then labeled with 250 Ci of [ 32 P]orthophosphate for 4 h. After washing with cold phosphate-buffered saline, lysates were prepared using lysis buffer and then incubated with anti-HA antibodies at 4°C for 2 h. The immunocomplexes were adsorbed to protein G-Sepharose and were subjected to SDS-PAGE followed by blotting onto nitrocellulose and autoradiography. The amount of Pax2 that was immunoprecipitated was assessed by re-immunoblotting of the nitrocellulose. For amino acid analysis, immunoprecipitated Pax2 was blotted to polyvinylidene difluoride membranes, excised, soaked in methanol, and subsequently hydrolyzed in 6 N HCl for 2 h at 110°C. Samples were dried, washed repeatedly with water, and spotted onto thin layer chromatography cellulose plates with fluorescent markers for Ser(P), Thr(P), and Tyr(P). The plates were electrophoresed for 1 h at 1000 V, air-dried, sprayed with acetone to visualize markers, and exposed to film.
Immunoprecipitation of JNK Module-293 cells were transfected with the expression plasmids as indicated in the legends. After 40 h, lysates were prepared in IP buffer (20 mM Tris-HCl pH 8.0, 100 mM NaCl, 0.5% Triton X-100, 10 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmeth-ylsulfonyl fluoride, and protease inhibitor mixture). Lysates were incubated with anti-Myc M5 monoclonal antibodies for 2 h at 4°C. Antibodies were captured with protein G-Sepharose for 1 h and washed three times with IP-wash (same as IP buffer except 0.1% Triton X-100). Proteins were eluted from protein G-Sepharose by boiling in 2ϫ SDS-PAGE sample buffer.
Transient Transactivation Assays-For 5 ϫ 10 5 of 293 cells, 0.5 g of CAT reporter plasmid was cotransfected with various doses of effector plasmid and 0.25 g of p-CH110, a ␤-galactosidase reporter plasmid used to monitor transfection efficiency. 40 h after transfection, cells were harvested and extracts were prepared and normalized for transfection efficiency. CAT analysis was performed essentially as described (15). Percent acetylation was determined by scintillation counting, and a value of 1.0 was arbitrarily assigned to control parental plasmids. Each experiment was repeated three times and the results presented as averages Ϯ 1 S.D. from the mean.

Pax2 Is a Substrate for the c-Jun N-terminal Kinase-
The activation domain of Pax2 contains multiple serine, threonine, and tyrosine residues that could serve as potential phosphorylation sites. To determine the status of Pax2 phosphorylation, cells were transfected with epitope-tagged Pax2 and labeled with [ 32 P]orthophosphate. Immunoprecipitation of Pax2 revealed one major form of phosphorylated protein (Fig. 1A) and at least one minor form of slightly higher molecular mass. The phospho-Pax2 was excised from the gel and subjected to acid hydrolysis (Fig. 1B). Amino acids were separated by thin layer chromatography using phosphotyrosine, -serine, and -threonine markers and subjected to autoradiography. The immunoprecipitated Pax2 protein was phosphorylated at both serine and threonine residues, but not at tyrosine residues. Given the serine/threonine phosphorylation pattern and previously published reports on Pax6 phosphorylation, we examined the ability of the three major MAP kinases to phosphorylate Pax2 (Fig.  2). Cells were transfected with epitope-tagged forms of ERK1, JNK, and p38 MAP kinases and the upstream activator  4) and analyzed on SDS-PAGE gels. Note major phosphoprotein migrating at ϳ48 kDa and slightly higher molecular mass species at about 50 kDa (arrows). B, amino acid analysis of 32 P-labeled Pax2b immunoprecipitated with anti-HA (1) or anti-FLAG (2). Markers for phosphoserine, threonine, and tyrosine were traced from the TLC plates (dotted lines).
MEKK1. The kinases were immunoprecipitated and used for immunocomplex kinase assays with GST-Pax2 and control substrates. Transfected MEKK1 can activate all three pathways in 293 cells, as demonstrated by increased phosphorylation of ERK, JNK, and p38 in total cell lysates ( Fig. 2A). However of the immunoprecipitated kinases, only JNK was able to efficiently phosphorylate GST-Pax2 (Fig. 2B). The ability of activated JNK to phosphorylate Pax2 was confirmed using a recombinant, constitutively active form of JNK1 (Fig. 2C). The in vitro kinase assay using recombinant JNK and the GST fusion proteins as substrates indicated that GST-Pax2 was as efficient a substrate as GST-c-Jun (Fig. 2C). In vitro phosphorylation of GST-Pax2 was also done with an excess of cold ATP, and the proteins were detected by Western blotting (Fig. 2D). Using anti-Pax2 antibodies, a slower migrating form of the GST-Pax2 fusion protein is observed upon phosphorylation with recombinant JNK. Thus, the antibodies can distinguish between the unphosphorylated and phosphorylated forms of the GST-Pax2 fusion protein.
The phosphorylation of Pax2 was also examined directly in cells co-transfected with MEKK1. To map the domain of phosphorylation more precisely, carboxyl-terminal deletion constructs of Pax2b were utilized (Fig. 3). Specific antibodies recognize the 392-amino acid Pax2b protein, which migrates at ϳ48 kDa. Upon co-transfection with MEKK1, a slower migrating Pax2b species was observed at ϳ50 kDa in SDS-PAGE gels (Fig. 3B). Preincubation of the lysates with either calf intestinal phosphatase or potato acid phosphatase eliminates this slower migrating Pax2b form, indicating that the shift in mobility is due to phosphorylation, similar to what was observed in the in vitro assay (Fig. 2D). The MEKK1-dependent, phosphorylated Pax2b form is seen with carboxyl-terminal deletion constructs 1-373, 1-333, but not with deletions that contain only amino acids 1-279. Thus, Pax2b phosphorylation is primarily within the carboxyl-terminal transactivation domain spanning amino acids 279 -415. However, phosphorylation at additional sites between amino acids 1-279 that might not affect the mobility of Pax2 in SDS-PAGE gels cannot be ruled out. Strikingly, this region contains 19 serine and 11 threonine residues, making definitive assignment of the phosphorylation sites difficult.
Activation of Pax2 Phosphorylation by Wnt Signals-During kidney and neural development, Wnt signaling is known to mediate inductive interactions. While the canonical Wnt signaling pathway activates the TCF/LEF family of transcription factors by inhibition of the GSK3 kinase and translocation of ␤-catenin to the nucleus, alternate Wnt signaling pathways can work through the activation of JNK (16,17). Particularly during the specification of planar cell polarity (18), activation of JNK is known to require Disheveled and is uncoupled from the TCF/LEF pathway (19 -21). Given the relationship between Wnt signaling, kidney development, and JNK activation, we examined the ability of Wnt proteins to activate JNK and stimulate Pax2 phosphorylation. Cells were co-transfected with expression plasmids for Wnt4, Wnt11, and Wnt7a and examined for JNK activation and Pax2 phosphorylation (Fig. 4A). Analysis of total lysates from 293 cells co-transfected with Wnt expression vectors and Pax2b indicated co-incident phosphorylation of Pax2, as evidenced by the slower migrating Pax2b isoform (Fig. 4A). Activation of the Wnt pathway results in the slower migration of the disheveled protein, through phosphorylation at multiple sites, as reported previously (22,23). Cells transfected with each of the three Wnt constructs also exhibited co-incident activation of JNK (Fig. 4A). The phosphorylation of Pax2 was confirmed by in vivo labeling with Pax2 Phosphorylation by JNK [ 32 P]orthophosphate (Fig. 4B). MEKK1 co-transfected cells showed high levels of phospho-Pax2, whereas Wnt11 co-transfected cells showed less phospho-Pax2 but significantly more than cells expressing Pax2 alone. The data indicate that activation of JNK by either Wnt proteins or MEKK1 results in increased phosphorylation of Pax2.
Pax2 Interacts with the JIP1/JNK Signaling Module-Activation of the JNK pathway can also occur with via the mixed lineage kinases (MLK or DLK) in association with the scaffolding proteins of the JIP family (14,24,25). To determine whether Pax2 interacts directly with this JNK signaling module, cells were co-transfected with Pax2, JIP1, DLK, and JNK (Fig. 5). Expression of DLK alone slightly increased the amount of phospho-Pax2 present in total lysates. At low doses of DLK, Pax2 phosphorylation was further enhanced by co-expression with JIP1 or JNK. Immunoprecipitation of JIP1 also pulled down Pax2 and JNK, indicating that at least some Pax2 was associated with the JNK signaling module. Strikingly, Pax2 could be co-immunoprecipitated with JIP1 in the absence of DLK, but the amount of JIP1-associated Pax2 increased significantly upon activation of the JNK module with DLK. More phosphorylated Pax2 was observed by expression of DLK, together with exogenous JNK, and this resulted in even more JIP1-associated Pax2. In the absence of exogenous JIP1, cotransfection of DLK and JNK produced even more phosphorylated Pax2. At first this may seem counterintuitive, but transfection of exogenous JIP1 provides more scaffolding such that individual proteins may not all assemble on the same scaffold and thus effectively titrate out some of the activities (14). The data suggest that DLK-dependent phosphorylation of Pax2 by JNK is mediated by a direct association with the JNK signaling module, perhaps through interactions with JIP1.
Transactivation of Pax2 Reporter Genes-To assess the effects of Pax2 phosphorylation on the transactivation potential, we utilized a reporter plasmid containing five copies of the PRS4 Pax2b binding sequence upstream of a minimal promoter and the chloramphenicol acetyltransferase (CAT) gene. Using a fixed amount of Pax2b plasmid (125 ng) and increasing amounts of MEKK1, Pax2-dependent reporter activation increased dramatically with as little as 50 ng of MEKK1 expression plasmid (Fig. 6A). Pax2 transfection can stimulate the reporter plasmid activity by 8 -10-fold. This increased to more than 50-fold by co-transfection of 200 ng of MEKK1 expression plasmid. In the absence of Pax2, MEKK1 had no effect on the expression of the reporter. Furthermore, a kinase-deficient MEKK1 construct showed only a slight increase at the highest doses. Pax2-dependent transactivation was also enhanced in the presence of Wnt expression plasmids (Fig. 6B). Expression of Wnt4, Wnt11, and Wnt7a all generated a 2-3-fold increase in Pax2-dependent activation of the reporter gene. Whereas Wnt gene expression alone had no effect on CAT activity in the absence of Pax2 (data not shown). As reported previously (3), deletion of the transactivation domain reduced Pax2-dependent activation (Fig. 6C). A carboxyl-terminal deletion of amino acids 373-415 still showed a robust response to MEKK1 although the basal level of Pax2-dependent activation was down about 2-fold. However, a carboxyl-terminal deletion of amino acids 333-415 exhibited a further reduction in basal activation and a loss of responsiveness to MEKK1. Western blotting indicated that there was some degree of Pax2-(1-333) phosphorylation in response to MEKK1 (Fig. 3B). However, it is unlikely that these remaining phosphorylation sites are necessary for the increased activation potential in response to MEKK1. These data suggest that the critical phosphorylated residues in Pax2 that mediate the MEKK1-induced transcriptional activation are located between amino acids 333 and 415. Similarly, activation of JNK via DLK and JIP increased the transactivation potential of Pax2 (Fig. 6D). DLK and Pax2 alone increased reporter activity only slightly. However, co-transfection of DLK with JIP1 increased the Pax2-dependent transactivation potential by 3-4-fold. Thus in all cases examined, the factors that increase JNK activity and result in increased amounts of Pax2 phosphorylation also significantly enhance the Pax2 dependent transactivation potential of the PRS4 reporter vector. DISCUSSION In this report, we demonstrate that Pax2, a member of the Pax family of transcription regulators, is phosphorylated by JNK. Also known as SAPK, JNK phosphorylates a variety of nuclear transcription factors, including c-Jun, Elk1, and ATF2.
In particular, phosphorylation of Ser-63 and Ser-73 of c-Jun results in increased activity of the AP-1 transcription factor, which consists of a c-Fos/c-Jun heterodimer (13). Indeed, the sequence around Ser-63 of c-Jun (LLTSP) is very similar to a conserved serine at position 393 of the Pax2 protein (LLSSP). However, phosphorylation of Pax2 may occur at multiple sites. Deletion of the serine/threonine-rich, carboxyl-terminal transactivation domain significantly reduces phosphorylation such that a shift in Pax2b mobility, upon JNK activation, is no longer observed. The phosphorylated residues that mediate the JNK-dependent increase in transactivation reside between amino acids 333 and 415 of the activation domain. Within this region there are 11 serine and 7 threonine residues, of which 6 and 1 respectively are absolutely conserved within the Pax2/5/8 family. We have begun making point mutations of these conserved residues. However, preliminary data indicate that multiple serine substitution must be made before the MEKK1-dependent increase in transactivation is abrogated (data not shown).
By definition, the transactivation domain is required for maximal activation potential of the Pax2 reporter gene, PRS4-CAT, as defined previously in the absence of exogenous JNK activators. A basal level of phospho-Pax2 is present that most probably accounts for the 8 -10-fold transactivation potential observed without JNK activation. This basal levels accounts for the metabolically labeled phospho-Pax2 observed without JNK activation and the small amount of high molecular weight Pax2b form observed upon long exposures of Western blots. Basal levels of activated JNK and phospho-c-Jun are inevitably observed in proliferating cells stimulated with serum. Because there are presently no practical JNK inhibtors, we were unable to test whether JNK activation was absolutely required for Pax2-mediated transactivation. However, several lines of evidence indicate that JNK is the kinase responsible for Pax2 phosphorylation, which correlates with increased transactivation. Of the major MAP kinases, only JNK is able to phosphorylate Pax2 in the immunolinked kinase assay. Furthermore, recombinant constitutively active JNK phosphorylates Pax2 in vitro as efficiently as it phosphorylates c-Jun. Increased JNK activity leads to increased Pax2 phosphorylation and increased Pax2-dependent transactivation, regardless of whether JNK is activated by MEKK1, Wnt signaling, or DLK. Pax2 can associate with the JNK signaling module as demonstrated by the co-immunoprecipitation of Pax2 with JIP1. Strikingly, activation of the signaling module by the DLK results in increased JIP1-associated Pax2. A specific inhibitor of MEK1 (PD98059), which suppresses phosphorylation of the ERK1/2 MAPK, does not significantly reduce Pax2 phosphorylation nor does it reduce Pax2-dependent transactivation (data not shown). While JNK is likely to be the kinase that mediates phosphorylation of Pax2, the effect on Pax2 activity is less clear. It remains to be determined whether Pax2 phosphorylation by JNK is merely co-incident with enhanced activation, whether it is a prerequisite for further modification by other kinases, or whether there is indeed a causative link to enhanced Pax2 activation potential.
In the developing kidney and central nervous system, Wnt signaling and Pax genes play critical roles in establishing patterns and specifying cell types. During kidney development, Pax2 is expressed in the metanephric mesenchyme prior to induction by the ureteric bud epithelium (8). Upon induction, Pax2-expressing cells proliferate, aggregate around the tips of the ureteric bud, and undergo conversion to an epithelial phenotype (6,26). This response to inductive signals is Pax2-dependent (4, 27) and also requires Wnt4 signaling (28). That Wnt4 can activate JNK and stimulate Pax2-dependent gene FIG. 6. Transactivation potential of Pax2. A, MEKK1 increases the transactivation of a Pax2b reporter gene. Cells were transfected with Pax2 alone or with increasing amounts of MEKK1, and with the reporter plasmid PRS4-CAT. The control sample has reporter only and the MEKK1 sample contains the reporter and MEKK1. B, Wnt signaling increases Pax2-dependent transactivation. Cells were transfected with Pax2b, PRS4-CAT, and increasing amounts of Wnt expression plasmids. Wnt expression alone has no effect on the levels of PRS4-CAT (data not shown). C, deletion analysis within the Pax2 transactivation domain. Cells were transfected with full-length Pax2-(1-415) or with deletion constructs containing the indicated amino acids and with (ϩ) or without (Ϫ) MEKK1. Note that Pax2b-(1-333) shows no increase in activity caused by MEKK1. D, activation of Pax2-dependent reporter gene by the DLK/JIP1 signaling module. Note that co-transfection of JIP1 (100 ng) and DLK (100 ng) increases Pax2-dependent transactivation 3-4-fold. DLK or JIP1 alone have no effect on reporter gene expression. For all panels, the mean values of three independent transfections for CAT activity were shown relative to control expression plasmid, which was assigned a value of 1.0. activation suggests that phosphorylation of Pax2 is an essential component of the induction response. Thus, Pax2 may be an immediate downstream target for inductive signals that can translate these signals by initiating the epithelial-specific genetic program.
Pax2 and the related genes Pax5 and Pax8 are also active during neural development, particularly in the mid-brain/hindbrain junction of the central nervous system (29 -31). Mice carrying targeted mutations for both JNK1 and JNK2 show regional defects in the developing hindbrain, including decreased apoptosis prior to neural tube closure in and around the Pax2/5/8 expression domain (32). Thus, modulation of Pax activity by JNK may also be required for the regulation of neuronal cell number through selective activation of apoptosis. In any event, more direct studies in normal and mutant mice are warranted to more precisely determine the phosphorylation state of Pax proteins during specific stages of development.