Dynamic interaction between the dual specificity phosphatase MKP7 and the JNK3 scaffold protein beta-arrestin 2.

JNK scaffold proteins bind JNK and upstream kinases to activate subsets of JNK and localize activated JNK to specific subcellular sites. We previously demonstrated that the dual specificity phosphatases (DSPs) MKP7 and M3/6 bind the scaffold JNK-interacting protein-1 (JIP-1) and inactivate the bound subset of JNK (1). The G protein-coupled receptor (GPCR) adaptor beta-arrestin 2 is also a JNK3 scaffold. It binds the upstream kinases ASK1 and MKK4 and couples stimulation of the angiotensin II receptor AT1aR to activation of a cytoplasmic pool of JNK3. Here we report that MKP7 also binds beta-arrestin 2 via amino acids 394-443 of MKP7, the same region that interacts with JIP-1. This region of MKP7 interacts with beta-arrestin 2 at a central region near the JNK binding domain. MKP7 dephosphorylates JNK3 bound to beta-arrestin 2, either following activation by ASK1 overexpression or following AT1aR stimulation. Initial AT1aR stimulation causes a rapid (within 5 min) dissociation of MKP7 from beta-arrestin 2. MKP7 then reassociates with beta-arrestin 2 on endocytic vesicles 30-60 min after initial receptor stimulation. This dynamic interaction between phosphatase and scaffold permits signal transduction through a module that binds both positive and negative regulators.

The response to particular stimuli and the subsequent cellular localization of active MAPKs is thought to be maintained by scaffold proteins (16). For example, the JIP family of scaffold proteins are kinesin cargo proteins that recruit MLKs, MKK7, and JNK (17,18,25) and also the DSPs, MKP7, and M3/6 (1). Such binding of both kinases and phosphatases has also been observed in the family of protein kinase A-anchoring proteins (AKAPs) (19); yotiao/AKAP450 binds both PKA and phosphatase PP-1 to control N-methyl-D-aspartate receptor signaling (20).
Cell Culture-293T and COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. 293T and COS-7 cells were transfected using Lipofectamine (Invitrogen) according to the manufacturer's instructions.
Pull-downs, Immunoprecipitations, and Immunoblots-293T cells were washed once with ice-cold buffered saline then lysed in buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% IGEPAL CA-630, 5 mM EDTA, and a protease inhibitor mixture (Roche Applied Science). After 10 min of incubation on ice, the extracts were centrifuged at 14000 ϫ g for 15 min. Supernatants were subject to pull-down or im-* This work was supported by the Association for International Cancer Research. 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 U.S.C. Section 1734 solely to indicate this fact.
FIG. 1. Selective binding of the dual specificity phosphatase MKP7 to MAPK scaffold ␤-arrestin 2. A, top panel, a plasmid expressing Xp-tagged MKP7 (0.75 g) was introduced into 293T cells together with GST, GST-tagged ␤-arrestin 2, and GST-tagged JIP-1 (0.2 g). GST-containing complexes were isolated with glutathione-Sepharose beads (GST pull-down) from 5 ϫ 10 6 cells, and the presence of MKP7 in the precipitates was examined by immunoblot using the anti-Xpress tag antibody. Densitometry analysis indicates 4.5 and 1.2% of total expressed MKP7 was present in the GST-JIP-1 and GST-␤-arrestin 2 precipitates, respectively. The relative expression levels of GST, GST-␤-arrestin 2, GST-JIP-1, and MKP7 are also shown in the whole cell extract. Bottom panel, a construct expressing M3/6 (0.75 g) was introduced into 293T cells Dynamic Interaction between MKP7 and ␤-Arrestin 2 25652 munoprecipitation, as described below. Samples were separated by 10% SDS-PAGE and electrotransferred to nitrocellulose membranes (Hybond ECL, Amersham Biosciences). Membranes were subjected to immunoblotting with anti-Xpress (Invitrogen), anti-M3/6 polyclonal (also used to detect GST), anti-FLAG M2 monoclonal (Sigma), or anti-HA High Affinity (Roche) antibodies and horseradish peroxidase-conjugated secondary antibodies (Dako). Blots were developed using ECL reagents (Amersham Biosciences). For immunoprecipitation or pulldown experiments, extracts were made as described above and incubated with either glutathione-Sepharose 4B (Amersham Biosciences) for GST-containing constructs or with anti-FLAG M2 antibody with protein G-Sepharose beads (Sigma) for 3 h. Beads were then washed three times in lysis buffer and resuspended in an appropriate amount of gel loading buffer (final concentration 50 mM Tris, pH 6.8, 2% SDS, 100 mM dithiothreitol, 4% glycerol). For experiments to determine JNK phosphorylation, COS-7 cells were lysed directly into gel-loading buffer and blots probed with anti-phospho-JNK (Promega), anti-HA High Affinity, anti-FLAG M2, or anti-Xpress antibodies.
Immunostaining-COS-7 cells were transfected on glass coverslips. The cells were incubated with 1 M angiotensin II (Sigma) for the indicated times. The cells were then washed with phosphate-buffered saline solution, fixed with 4% paraformaldehyde (Sigma), and lysed with 0.2% Triton X-100/phosphate-buffered saline. Indirect immunofluorescence was performed by incubation with the MKP7 polyclonal antiserum in 1% bovine serum albumin/0.5% goat serum albumin/phosphate-buffered saline for 1 h at 37°C. Texas Red-conjugated secondary antibody (Jackson Immunoresearch) was used and images prepared on a Zeiss confocal microscope using Bio-Rad Lasersharp software.

Selective
Binding of MKP7 to ␤-Arrestin 2-We recently reported that the DSPs, MKP7, and M3/6 bind the JNK scaffold proteins JIP-1 and JIP-2 (1). As ␤-arrestin 2 has been identified as a scaffold protein for JNK3, the neuronal isoform of JNK (23,24), we asked whether any DSPs could bind ␤-arrestin 2. We co-expressed GST-tagged ␤-arrestin 2 with Xpress-tagged members of the DSP family and examined ␤-arrestin 2 precipitates for presence of the phosphatases using the anti-Xpress tag antibody. Under these conditions, MKP7 bound ␤-arrestin 2 ( We next performed co-precipitation analysis of MKP7 deletion mutants with ␤-arrestin 2 ( Fig. 1, C and D). This identified a deletion mutant containing amino acids 1-394 that did not bind ␤-arrestin 2, whereas a mutant containing amino acids 1-443 did (Fig. 1D, lanes 6 and 8). The sequence between amino acids 394 and 443 is also critical for MKP7 binding to JIP-1. The equivalent region of M3/6 is very similar to that of MKP7, which suggests either that ␤-arrestin 2 (unlike JIP-1) discriminates between the two sequences or that M3/6 binding is not detected within the sensitivity of our assay. The MKP7 C-terminal fragment residues 360 -665 were sufficient for binding both JIP-1 and ␤-arrestin 2, whereas residues 1-394 of MKP7 did not bind (Fig. 1E, compare lanes 5, 6, 8, and 9). As shown in Fig. 1F, MKP7-(1-394), which cannot bind ␤-arrestin 2, can still bind JNK3 (lane 3). Thus MKP7 binding to ␤-arrestin 2 is dependent on residues within its C-terminal region. This region is sufficient for the interaction, demonstrating that the interaction is independent of JNK3 binding to MKP7.
Angiotensin II ligation to the AT1aR stimulates JNK3 phosphorylation in the presence of ␤-arrestin 2 (23). We co-expressed JNK3, ␤-arrestin 2, and AT1aR in COS-7 cells and showed that angiotensin II stimulated phosphorylation of JNK3 at 15 and 30 min, with the signal disappearing after 60 min (Fig. 3C, lanes 3 and 4), similar to previously published data (3). In the absence of ␤-arrestin 2 and addition of MKP7, stimulation of JNK3 phosphorylation remains the same as above (Fig. 3C, lanes 8 and 9). However, when MKP7 was expressed along with ␤-arrestin 2, JNK3 phosphorylation was equivalent at 15 min after angiotensin II stimulation but completely abrogated 30 min after stimulation (Fig. 3C, lanes 3 and  4; and 13 and 14). This data indicates MKP7 utilizes the interaction with ␤-arrestin 2 to specifically dephosphorylate JNK3, thus allowing a more transient JNK3 phosphorylation under AT1aR activation.
MKP7 Interacts with ␤-Arrestin 2 in Resting Cells but Is Released after AT1aR Stimulation-To explain the initial equivalent JNK3 activation seen in the presence of MKP7, followed by its rapid inactivation, we postulated that MKP7 interaction with ␤-arrestin 2 might be regulated by AT1aR stimulation. To detect MKP7 location, we used a previously described polyclonal antiserum (1) that detects a cytoplasmic protein in transfected COS-7 cells (Fig. 4A). We used leptomycin B to inhibit the nuclear export of MKP7 (10, 11) and demonstrated resulting nuclear fluorescence accumulation to confirm the specificity of this staining (data not shown). We then co-expressed MKP7 with components of the AT1aR signaling system and ␤-arrestin 2. Fig. 4A shows overexpressed GFP-␤arrestin 2 (green) and MKP7 (red) were both present in the cytoplasm of resting cells. Angiotensin II ligation to AT1aR has been reported to cause internalization of JNK3 bound to ␤-arrestin 2 on endocytic vesicles (23). Fig. 4A shows that ␤-arrestin 2 (green) moved to the plasma membrane 5 min after agonist activation of the AT1aR, whereas MKP7 remains cytoplasmic. After 15 min, green vesicles containing ␤-arrestin 2 (but not MKP7) had formed, but after 30 min, the vesicles appeared yellow, indicating the recruitment of MKP7. After 60 min, yellow endosomal structures had begun clustering at the perinuclear region of the cell. These experiments suggest that MKP7 is released from ␤-arrestin 2 after activation of the AT1aR and is then re-recruited after ϳ30 min to ␤-arrestin 2 present on endocytic vesicles.
To confirm this dynamic interaction between ␤-arrestin 2 and MKP7, we examined MKP7 binding to ␤-arrestin 2 after AT1aR stimulation. For these experiments, we used 293T cells to express sufficient protein for detection after immunoprecipitation rather than the COS-7 cells in which we measured subcellular localization and JNK3 dephosphorylation. In the absence of JNK3, the binding of MKP7 to ␤-arrestin 2 changed relatively little after angiotensin II stimulation (Fig. 4B, lanes  3-7). However with the addition of JNK3, MKP7 binding to ␤-arrestin was lost rapidly after angiotensin II stimulation, returning to the initial level after 60 min (Fig. 4B, lanes 9 -13). The level of JNK3 binding to ␤-arrestin 2 remained the same throughout the time course, as previously described (23) (data not shown). Together, these data demonstrate that activation of the AT1aR causes rapid dissociation of MKP7 from ␤-arrestin 2. This dissociation is dependent on the presence of JNK3 and is reversed 30 -60 min after AT1aR stimulation. Because JNK3 is required for MKP7 dissociation from ␤-arrestin 2 triggered by angiotensin II, we asked whether JNK3 activation per se is sufficient to cause MKP7 dissociation. To do this, we used overexpressed ASK1 as a JNK3 activator. Fig. 4C shows that ASK1 causes a significant decrease in the amount of MKP7 bound to ␤-arrestin 2, suggesting that it is the stimulation of JNK3 phosphorylation that leads to MKP7 dissociation from ␤-arrestin 2.

DISCUSSION
␤-arrestins were first described as adaptor proteins involved in the desensitization and internalization of G protein-coupled receptors. After GPCR activation, the cytoplasmic tail of the receptor becomes phosphorylated by G protein-coupled receptor kinases, leading to ␤-arrestin recruitment. ␤-arrestin then blocks further G protein interaction and bind proteins required for the internalization of GPCRs, such as clathrin (26,27), AP-2 (28), N-ethylmaleimide-sensitive factor (29), Ral-guanine nucleotide dissociation stimulator (30), and ADP ribosylation factor 6 (31). In response to agonist binding, ␤-arrestin 2 also undergoes rapid ubiquitination by Mdm2, which is essential for receptor internalization (32).
In addition to their roles in receptor inhibition, ␤-arrestins are also signal transducers and have been shown to interact with intracellular signaling components, including both tyrosine and MAP kinases. For example, ␤-arrestins have been shown to recruit the Src family kinases (33) to endothelin ETa receptors (34), neurokinin 1 receptors (35), and CXCR1 receptors (36), resulting in the activation of the MAPK ERK1/2 (37). ␤-arrestin 2 has also been shown to directly recruit components of the ERK MAPK signaling module to the AT1aR (38). Although G protein-dependent ERK signaling is responsible for the activation of transcription factor Egr-1 after AT1aR stimulation, the ␤-arrestin 2-bound pool of ERK is retained in the cytoplasm, targeting activated ERK to other cytoplasmic subexpressed with Xp-tagged MKP7 (0.75 g) in 293T cells. GST complexes were isolated using glutathione-Sepharose beads (GST pull-down) from 5 ϫ 10 6 cells, and the presence of MKP7 was examined by immunoblot using the anti-Xpress tag antibody. In (i), densitometry analysis indicates 0.6 and 5.8% of total expressed MKP7 was present in GST-␤-arrestin 2 and GST-(165-410) precipitates, respectively. In (ii), 0.5 and 2.3% of total expressed MKP7 was present in GST-␤-arrestin 2 and GST-(1-164) precipitates, respectively. Relative expression levels are also shown.  strates (39,40). ␤-arrestin 2 was also recently reported to bind the neuronal JNK isoform, JNK3, and upstream components ASK1, and indirectly MKK4, and to stimulate JNK3 activation through AT1aR stimulation (23). Similar to ERK, JNK may bind ␤-arrestin 2 to maintain an extra nuclear location close to specific cytosolic substrates (41). As well as GPCRs, numerous receptors have been found that utilize ␤-arrestin-type scaffold proteins to coordinate both the kinases and phosphatases required for receptor function. These include the ␤-adrenergic receptors (42, 43), AT1R (44), and mGluRs (45).
We have shown that the dual specificity phosphatase MKP7 binds ␤-arrestin 2, using residues 394 -443 of MKP7 previously FIG. 4. MKP7 transiently dissociates from ␤-arrestin 2 after AT1aR stimulation and JNK3 activation. A, plasmids expressing MKP7 (0.2 g), GFP-tagged ␤-arrestin 2 (0.2 g), HA-tagged JNK3 (0.2 g), and HA-tagged AT1aR (0.2 g) were co-expressed in COS-7 cells. After 48 h, the cells were stimulated with 1 M angiotensin II (Ang II) for the indicated time periods. MKP7 was detected by immunofluorescence using an anti-MKP7 polyclonal antiserum with a Texas Red-conjugated donkey anti-rat secondary antibody. The images show the separate staining of MKP7 (red) and ␤-arrestin 2 (green), as well as a merge of the two; yellow indicates co-localization. Scale bars are included to indicate size. B, GST or GST-␤-arrestin 2 (0.2 g) were expressed together with HA-tagged AT1aR (0.4 g) and either Xp-tagged MKP7 only (0.75 g) or Xp-tagged MKP7 and HA-tagged JNK3 (0.75 g). 293T cells were stimulated with 1 M angiotensin II for the indicated times. GST-containing complexes were isolated using glutathione-Sepharose beads (GST pull-down), and the presence of MKP7 and JNK3 (data not shown) was examined by immunoblot. Relative expression levels are also shown. The graph represents the level of MKP7 binding to ␤-arrestin 2 at each time point calculated by densitometry and showing levels as arbitrary units. C, GST-␤-arrestin 2 (0.2 g) was expressed along with Xp-tagged MKP7 (0.75 g) and HA-tagged JNK3 (0.75 g) and/or HA-ASK1 (0.75-0.5 g) in 293T cells. GST-containing complexes were obtained using glutathione-Sepharose beads (GST pull-down), and the presence of MKP7 was examined by immunoblot using the anti-Xpress tag antibody. Relative expression levels of all plasmids used are also shown. The graph represents the level of MKP7 binding to ␤-arrestin 2 in the presence of different concentrations of ASK1 calculated by densitometry and showing levels as arbitrary units.
identified as the JIP binding domain (1). The interaction of this region of MKP7 with these two diverse scaffold proteins suggests that it may function as a general scaffold-binding domain. The interaction between MKP7 and ␤-arrestin 2 is dynamic, meaning they dissociate after ␤-arrestin 2 is recruited to the AT1aR by angiotensin stimulation. Recent data have indicated ␤-arrestin 2, upon activation by binding to the phosphorylated C terminus of GPCRs, undergoes a conformational change (46). This conformational change could be involved in the triggering of JNK3 activation. We have shown that the presence of JNK3 on the scaffold is necessary for MKP7 dissociation and also that ASK1 overexpression leads to MKP7 dissociation. We therefore propose that activation of JNK3 triggers MKP7 dissociation from ␤-arrestin 2. After 30 -60 min, depending on the cell system, MKP7 reassociates with ␤-arrestin 2 to specifically dephosphorylate JNK3. This cycle of MKP7 dissociation from and reassociation with ␤-arrestin 2 explains how a signal can be transmitted through a scaffold protein that apparently binds both activating and inhibitory components. By binding into this complex, under these circumstances, MKP7 may play a role in regulating JNK in the cytoplasm.
These studies have been performed in cells overexpressing the signaling components; however, endogenous interaction between ␤-arrestin 2 and JNK3 has been seen in brain lysates (23), where the widely expressed MKP7 is also present (10,11). Data has shown seven-transmembrane receptors Frizzled (47) and Smoothened (48,49), as well as non-GPCRs, such as TGF␤ (50), are coupled to ␤-arrestin scaffold proteins. It is therefore possible that MKP7 and other DSPs can regulate MAPK signaling in a number of receptor systems.