Gα12 Directly Interacts with PP2A

The Gα12/13 family of heterotrimeric G proteins modulate multiple cellular processes including regulation of the actin cytoskeleton. Gα12/13 interact with several cytoskeletal/scaffolding proteins, and in a yeast two-hybrid screen with Gα12, we detected an interaction with the scaffolding subunit (Aα) of the Ser/Thr phosphatase, protein phosphatase 2A (PP2A). PP2A dephosphorylates multiple substrates including tau, a microtubule-associated protein that is hyperphosphorylated in neurofibrillary tangles. The interaction of Aα and Gα12 was confirmed by coimmunoprecipitation studies in transfected COS cells and by glutathione S-transferase (GST)-Gα12 pull-downs from cell lysates of primary neurons. The interaction was specific for Aα and Gα12 and was independent of Gα12 conformation. Endogenous Aα and Gα12 colocalized by immunofluorescent microscopy in Caco-2 cells and in neurons. In vitro reconstitution of GST-Gα12 or recombinant Gα12 with PP2A core enzyme resulted in ∼300% stimulation of PP2A activity that was not detected with other Gα subunits and was similar with GTPγS- and GDP-liganded Gα12. When tau and active kinase (Cdk5 and p25) were cotransfected in to COS cells, there was robust tau phosphorylation. Co-expression of wild type or QLα12 with tau and the active kinase resulted in 60 ± 15% reductions in tau phosphorylation. In primary cortical neurons stimulated with lysophosphatitic acid, a 50% decrease in tau phosphorylation was observed. The Gα12 effect on tau phosphorylation was inhibited by the PP2A inhibitor, okadaic acid (50 nm), in COS cells and neurons. Taken together, these findings reveal novel, direct regulation of PP2A activity by Gα12 and potential in vivo modulation of PP2A target proteins including tau.

Heterotrimeric G proteins 1 regulate many fundamental cellular responses, and the canonical receptor-G protein-effector paradigm has become significantly more complex in recent years. G protein signaling is modulated through interactions with families of regulatory proteins that affect nucleotide binding and hydrolysis. These include the RGS (regulators of G protein signaling) protein family that stimulate G␣ GTPase activity (reviewed in Ref. 1), and the GPR (G protein regulatory) protein family that share a conserved motif that inhibits GDP release from the G␣ i /␣ o families of G␣ subunits (2)(3)(4). G␣ 12/13 have multiple cellular functions including regulation of the actin cytoskeleton (5) and many functions are shared by both G␣ 12 and G␣ 13 subunits. G␣ 12/13 regulation of the actin cytoskeleton and stress fiber formation occurs through direct interactions with Rho regulatory proteins (6,7), and other aspects of G␣ 12 /␣ 13 signaling are regulated through specific interactions with membrane or scaffolding proteins. For example, the binding of G␣ 12 and G␣ 13 to the C terminus of E-cadherin displaces ␤-catenin permitting it to signal (8,9). We identified direct binding of both wild type and constitutively active (QL)␣ 12 to ZO-1 and regulation of paracellular permeability of Madin-Darby canine kidney cells (10,11). In addition, G␣ 12 interacts with AKAP-lbc, a scaffolding molecule that organizes components of cAMP signaling and Rho signaling machinery (12). Here, we have identified the A␣ subunit of Ser/ Thr phosphatase PP2A as another "scaffolding" protein that selectively interacts with G␣ 12 .
Tau is a microtubule-associated protein that is predominantly expressed in neurons and functions to stabilize the cytoskeleton. Tau phosphorylation is necessary for microtubule binding and other protein interactions, but hyperphosphorylated tau is the predominant component of neurofibrillary tangles, a pathologic hallmark finding found in several neurodegenerative disorders including Alzheimer disease and FTDP-17 (frontotemporal dementia and Parkinson's disease linked to chromosome 17; reviewed in Ref. 13). Several kinases phosphorylate tau including mitogen-activated protein kinase (MAP), glycogen synthase 3␤ (GSK-3␤), tau-tubulin kinase, and cyclindependent kinases 2 and 5 (Cdk) (reviewed in Ref. 13). Likewise, several protein phosphatases (PP) of the 1, 2A, and 2B families can dephosphorylate tau proteins in vitro, and PP2A directly binds to tau and microtubules (14). To determine the relevance of the yeast two-hybrid interaction between G␣ 12 and the A␣ subunit of PP2A, we demonstrate direct and specific G␣ 12 -dependent stimulation of PP2A activity in vitro. Furthermore, we show that G␣ 12 -mediated signaling in COS cells and primary cultured neurons stimulates PP2A-mediated dephosphorylation of the microtubule-binding protein, tau.
Baculovirus purified G␣ i1 was provided by Stephen Graber (West Virginia University), and G␣ 12 and G␣ s were generously provided by Patrick Casey (Duke University). Polyclonal rabbit anti-G␣ 12 and G␣ 13 and goat-anti A␣ antibodies were from Santa Cruz Biotechnology, monoclonal anti-Myc mouse antibodies from Invitrogen. Antibodies to tau were from Upstate Biotechnology (Lake Placid, NY), and GFP was from Abcam Inc. (Cambridge, MA). Secondary antibodies were all obtained from Molecular Probes. Polyclonal rabbit anti-A␣ was provided by D. Virshup, and tau phosphospecific antibody CP9 was generously provided by Peter Davies (Albert Einstein College of Medicine). Lysophosphatitic acid (LPA) was from Avanti Polar Lipids (Alabaster, AL).
Yeast Two-hybrid Screening-The Matchmaker TM (Clontech, Palo Alto, CA) was utilized as described previously (17). Mouse G␣ 12 cDNA was cloned into the bait vector pAS-2 and a human fetal kidney library, in pACT-2, was screened using standard selection methodology as described previously (17) except that co-transformed yeast were incubated at room temperature for 6 -10 days. Clones growing on selective plates at room temperature were regrown and screened for ␤-galactosidase activity.
PP2A Phosphatase Activity-Purified PP2A core enzyme (A␣ and catalytic unit) was obtained from Upstate Biotechnology and kinetic analysis of phosphatase activity determined by Malachite Green phosphatase assay (Upstate Biotechnology). The phospho-peptide (K-R-pT-I-R-R) was used as substrate to determine PP2A activity alone and in combination with GST, and GST-G␣ 12 , GST-G␣ 13 , and GST-G␣ s at 1:5 molar ratio). A similar comparison was done with recombinant G␣ 12 , G␣ s , and G␣ i1 at 1:1 molar ratio. G proteins were incubated with GDP or GTP␥S (50 M) or AlF 4 Ϫ (3 mm NaF ϩ 50 M AlCl 3 ) for 15 min at 30°C followed by incubation with PP2A for 30 min at 23°C. The reaction was initiated by addition of substrate at t ϭ 0, and after 30 min the reaction was terminated, absorbance measured at 624 nm, and enzyme activity determined (nmol of phosphate/min/units).
Primary Neuronal Cultures and Agonist Stimulation-The cortex of E18 Sprague-Dawley rats was separated in 2.5% trypsin (Invitrogen) and 0.5-1 ml 0.16% (w/v) DNase I (Sigma) dissolved in HBSS. Cells were plated at ϳ1 ϫ 10 5 cells/cm 2 on poly-L-lysine-coated plates or coverslips. HBSS was replaced with Neurobasal media (Invitrogen) supplemented with glutamine and B-27 and penicillin/streptomycin antibiotic solution. Experiments were done after 10 -14 days in culture by adding vehicle, isoproterenol (1 M), or LPA (10 M) for 30 -60 min followed by washing, scraping in modified RIPA buffer, and Western analysis on identical amounts of total protein using CP-9 antibody. Blots were stripped and reprobed with tau antibodies.
Statistics and Quantization-Western blots with exposures in the linear range were scanned using a desktop scanner and the images analyzed in NIH Image 1.63 (Wayne Rasband). Statistics were compiled with GraphPad Prism (GraphPad Software, San Diego, CA). Results are expressed as the mean Ϯ standard deviation. Statistical significance was determined using two-tailed t test.

RESULTS AND DISCUSSION
During our search for novel binding partners for G␣ 12 , we identified an interaction with the A␣ subunit of PP2A. This interaction was confirmed in several assays and we found direct G␣ 12 -dependent stimulation of PP2A activity in vitro. Furthermore, studies in COS cells and primary neurons reveal G␣ 12 -dependent stimulation of PP2A activity resulting in reduced phosphorylation of a target protein, tau.
Yeast two-hybrid screening of a human embryonic kidney library with mouse wild type G␣ 12 subunit resulted in five confirmed positive clones. Two of these were identical and encoded a 1.1-kb fragment of the regulatory subunit A (PR 65) of PP2A␣ (GenBank TM accession number NM_014225). The coding sequence in this fragment included amino acids 225 to the C terminus (590) encompassing repeats 7-15. PP2A is one of eight classes of serine/threonine phosphatases, is ubiquitously expressed, and is a major regulator of many fundamental cellular processes (reviewed in Ref. 19). PP2A is composed of a catalytic subunit (C), a scaffolding subunit (A), and a regulatory subunit (B). The catalytic and scaffolding subunits bind tightly to form a core dimer (AC) that is a functional unit. The regulatory subunits are numerous and diverse and provide specificity and localization for the many functions of PP2A. Many proteins interact with PP2A including signaling proteins and transcription factors, membrane receptors and transporters, protein kinases, cytoskeletal proteins, and others (reviewed in Ref. 20).
To confirm this interaction, we cotransfected myc-tagged PP2A A␣ and wild type G␣ 12 or constitutively active Q229L␣ 12 in COS cells. Endogenous G␣ 12 protein is not detectable nor can any be immunoprecipitated (last lane in Fig. 1A). Immunoprecipitation of myc-tagged A␣ subunits precipitated the co-expressed G␣ 12 subunits (wild type and QL, Fig. 1A). Control immunoprecipitations with G␣ 12 antibody and myc antibody in vector-transfected cells did not detect any G␣ 12 . Stripping and reprobing the blot with A␣ rabbit polyclonal antibody revealed the myc-tagged A␣ subunit migrating at ϳ80 kDa. Fig. 1B shows that immunoprecipitating the endogenous A␣ subunit coprecipitated a fraction of the transfected wild type and QL␣ 12 subunits. Controls with serum and beads alone were negative. Parallel experiments with G␣ 13 , G␣ i2 , and G␣ s failed to detect any interaction with A␣ (results not shown). To document this interaction in non-transfected cells, GST pull-downs from primary cortical neurons were performed. Fig. 1C shows interaction of endogenous A␣ from neurons with GST-G␣ 12 but not GST, GST-G␣ 13 , or GST-G␣ s . These findings suggested that the interaction of G␣ 12 with A␣ subunits did not depend upon G␣ 12 conformation. To confirm this, wild type G␣ 12 and myc-A␣ were cotransfected into COS cells and divided into three equal fractions. Lysates were incubated at 23°C for 20 min in the presence of GDP, GTP␥S, or AlF 4 Ϫ . Following immunoprecipitation of myc-A␣ subunits and analysis by G␣ 12 Western blot, there were no significant differences in the amount precipitated (Fig.  1D). It was previously shown in separate studies that G␣ 12 and PP2A are localized in epithelial cell tight junctions (21,22). Endogenous G␣ 12 and A␣ were colocalized by immunofluorescent microscopy using antibodies to G␣ 12 and A␣ in Caco-2 cells and primary cultured neurons (Fig. 1E). The two proteins colocalize in the lateral membrane of Caco-2 cells (Fig. 1E) and in the cell body and processes of primary neurons (Fig. 1E).
Although G␣ 12 and A␣ interact, there is no known regulation of PP2A by G proteins. To determine whether G␣ 12 affects PP2A activity, phosphatase activity of purified core enzyme (A␣ and catalytic subunit) was measured in the presence and absence of several G␣ subunits. Fig. 2A shows that GST-G␣ 12 significantly stimulates PP2A phosphatase activity by nearly 300% above the activity of the enzyme alone. Parallel assays done with GST alone or GST-G␣ 13 failed to demonstrate any significant effect on PP2A activity. Consistent with binding results in Fig. 1, preincubation of GST-␣ subunits with GDP or GTP␥S did not significantly affect the G␣ 12 -mediated stimulation of PP2A activity. AlF 4 Ϫ modestly decreased base-line PP2A activity (by ϳ25%), but there was still nearly 3-fold stimulation of PP2A activity (data not shown). These results suggest that the effect on PP2A activity was specific for G␣ 12 and independent of bound nucleotide. However, GST-G␣ proteins may not appropriately exchange guanine nucleotides. To address this, PP2A phosphatase activity was measured in the presence and absence of purified G␣ subunits. Fig. 2B shows that recombinant G␣ 12 stimulates PP2A phosphatase activity over 300%, whereas recombinant G␣ s and G␣ i1 had no demonstrable effect. Similar to the results with GST-G␣ 12 , and consistent with the binding data (Fig. 1D), the G␣ 12 -stimulated phosphatase activity was similar for GDP-and GTP␥S-liganded G␣ 12 .
Based upon these findings, we hypothesized that G␣ 12 -stimulated PP2A activity would result in less phosphorylation of a target substrate. To test this in cells, we utilized the microtubule associated protein, tau (GFP-tagged) and determined the effect of G␣ 12 expression on its phosphorylation in COS cells. We have previously demonstrated that the function of GFP-tau is indistinguishable from wild type tau (16). Fig. 3A shows a Western blot of COS cells transfected with GFP-tau (4R), Cdk5ϩp25 (the active kinase), and either wild type or QL␣ 12 . As expected for COS cells, expression of tau (4R, GFP-tagged) alone reveals no detectable Thr-231 phosphorylation (as detected with CP9 antibody; Fig. 3A, lane 1). When GFP-tau is coexpressed with the active kinase (4Rϩkinase) a robust signal is apparent. Cotransfection of either wild type or QL␣ 12 along with tau and the active kinase results in ϳ60% reduction in Thr-231 phosphorylation. The total tau in each lane was similar, and the bar graph (Fig. 3A) summarizes the relative effect of G␣ 12 expression on Thr-231 phosphorylation. Consistent with the effects of G␣ 12 occurring through regulation of PP2A, the PP2A phosphatase inhibitor, okadaic acid (50 nM), nearly completely inhibited the G␣ 12 mediated stimulation of PP2A activity (ϩO.A., Fig. 3). As expected from the lack of PP2A stimulation by other G␣ subunits (Fig. 2), parallel experiments with cotransfected wild type and QL␣ 13 in COS cells showed no change in tau phosphorylation (results not shown). Finally, we determined the extent of tau phosphorylation in primary cortical neurons stimulated with the G␣ 12/13 agonist, LPA. Fig. 3B shows a 50% reduction in endogenous tau phosphorylation when the cells were stimulated with LPA for 60 min. There was no effect of isoproterenol on tau phosphorylation, and the effect of LPA was inhibited by preincubation with okadaic acid (Fig. 3B).
Taken together, these results provide direct evidence for stimulation of PP2A phosphatase activity by G␣ 12 but not the related G␣ 13 subunit. In addition, G␣ 12 -stimulated PP2A ac-

FIG. 1. G␣ 12 and A␣ coimmunoprecipitate in COS cells and colocalize in Caco-2 cells and primary cortical neurons. A,
Western blot of COS cells cotransfected with myc-tagged A␣ and wild type or activated (QL) G␣ 12 . Cotransfections and immunoprecpitations (IP) were done as described under "Experimental Procedures." Plasmids used for cotransfections are listed below and immunoprecipitation with G␣ 12 rabbit polyclonal antibody or mouse anti-myc monoclonal antibody is shown above. The blot was probed with G␣ 12 antibody (top lane) and then reprobed with rabbit A␣ antibody (lower lane). B, G␣ 12 Western blot of COS cells transfected with wild type or QL␣ 12 and immunoprecipitated with endogenous A␣ subunit antibody, serum, or beads alone as described under "Experimental Procedures." The arrow denotes G␣ 12 . C, GST pull-downs from primary neuron lysate (500 g). Neuron lysate (20 g) is shown in first lane and represents 4% of input. GST, GST-G␣ 12 , and GST-G␣ 13 (1 g) were incubated with neuronal lysates, washed, and eluted. Samples were analyzed by SDS-PAGE followed by Western blot analysis with A␣ antibody. D, Western blots of myc-A␣ and wild type G␣ 12transfected COS cells incubated with guanine nucleotides. COS cells were transfected in a single plate and divided in to 3 equivalent aliquots followed by incubation with the nucleotide. A␣ was immunoprecipitated using myc antibody and samples analyzed by Western for G␣ 12   tivity may be an important regulatory pathway in cells that could affect the phosphorylation state of proteins such as tau. Strategies to decrease tau phosphorylation have received intense interest, since some features of Alzheimer and other neurodegenerative diseases are associated with hyperphosphorylated tau in neurofibrillary tangles. Furthermore, there is complex regulation of tau phosphorylation by multiple kinases and phosphatases (reviewed in Ref. 13). Several lines of evidence support the notion that G␣ 12 may have an important role in regulating tau phosphorylation in the central nervous system. The expression pattern of G␣ 12 in the central nervous system overlaps with the locations in which neurofibrillary tangles are known to develop including the cortex and hippocampus (23). Ligands for many G␣ 12/13 -coupled receptors have been implicated in Alzheimer and other neuropathologic processes. For example, endothelin and thrombin can couple to G␣ 12/13 , and activation of these receptors may contribute to neuronal cell death and amyloid plaque development in Alzheimer disease (24 -26). In addition, GSK-3 is one of several kinases that phosphorylates tau, and recent studies reveal regulation of GSK-3 activity by G␣ 12/13 (27,28).
These results plus recent studies suggest that the G␣ 12/13 family of G proteins may regulate serine/threonine phosphatases through multiple mechanisms. An interaction of QL␣ 12 with the Ser/Thr protein phosphatase type 5 (PP-5) was recently identified. Both QL␣ 12 and QL␣ 13 interacted with the N-terminal tetratricopeptide repeat unique to the PP-5 family of Ser/Thr phosphatases and stimulated phosphatase activity (29,30). In contrast, our studies reveal similar binding of both wild type and QL␣ 12 to PP2A, and we do not detect any interactions with G␣ 13 . Taken together, this suggests that protein phosphatase activities of several families may be regulated by G␣ 12/13 and raises the possibility that G protein regulation of serine/threonine phosphatases is a more general phenomenon.
The stimulation of PP2A phosphatase activity by G␣ 12 independent of nucleotide bound suggests a novel mechanism of regulation. The G␣ 12 -stimulated PP2A phosphatase activity in vitro reveals that core enzyme (A␣ and catalytic subunit) and G␣ 12 are sufficient for the stimulation. As a result, G␤␥ and PP2A regulatory B subunits are not necessary for this effect, but future studies are needed to determine any potential regulatory role of G␤␥ and/or B subunits on this activity. Our findings are consistent with a direct interaction of A␣ and G␣ 12 , although we cannot exclude the possibility that G␣ 12 also interacts with the catalytic subunit of PP2A. Additional studies are needed to address this question and whether the mechanism of stimulation by G␣ 12 is direct or through a secondary conformational change in A␣. The observation that G␣ 12 binds to the scaffolding A␣ subunit of PP2A and regulates phosphatase activity is another example of signaling modulation through multiprotein complexes. This novel pathway of G␣ 12 -stimulated PP2A phosphatase activity was not seen with other related G␣ subunits and raises the possibility for specific modulation of some PP2A target proteins in vivo. This may have major implications for treating specific diseases such as neurodegenerative processes mediated by tau hyperphosphorylation.