Interaction of Activator of G-protein Signaling 3 (AGS3) with LKB1, a Serine/Threonine Kinase Involved in Cell Polarity and Cell Cycle Progression

Activator of G-protein signaling 3 (AGS3) has a modular domain structure consisting of seven tetratricopeptide repeats (TPRs) and four G-protein regulatory (GPR) motifs. Each GPR motif binds to the α subunit of Gi/Go (Giα > Goα) stabilizing the GDP-bound conformation of Gα and apparently competing with Gβγ for GαGDP binding. As an initial approach to identify regulatory mechanisms for AGS3-G-protein interactions, a yeast two-hybrid screen was initiated using the TPR and linker region of AGS3 as bait. This screen identified the serine/threonine kinase LKB1, which is involved in the regulation of cell cycle progression and polarity. Protein interaction assays in mammalian systems using transfected cells or brain lysate indicated the regulated formation of a protein complex consisting of LKB1, AGS3, and G-proteins. The interaction between AGS3 and LKB1 was also observed with orthologous proteins in Drosophila where both proteins are involved in cell polarity. LKB1 immunoprecipitates from COS7 cells transfected with LKB1 phosphorylated the GPR domains of AGS3 and the related protein LGN but not the AGS3-TPR domain. GPR domain phosphorylation was completely blocked by a consensus GPR motif peptide, and placement of a phosphate moiety within a consensus GPR motif reduced the ability of the peptide to interact with G-proteins. These data suggest that phosphorylation of GPR domains may be a general mechanism regulating the interaction of GPR-containing proteins with G-proteins. Such a mechanism may be of particular note in regard to localized signal processing in the plasma membrane involving G-protein subunits and/or intracellular functions regulated by heterotrimeric G-proteins that occur independently of a typical G-protein-coupled receptor.

Activator of G-protein signaling 3 (AGS3) has a modular domain structure consisting of seven tetratricopeptide repeats (TPRs) and four G-protein regulatory (GPR) motifs. Each GPR motif binds to the ␣ subunit of G i /G o (G i ␣ > G o ␣) stabilizing the GDP-bound conformation of G␣ and apparently competing with G␤␥ for G␣ GDP binding. As an initial approach to identify regulatory mechanisms for AGS3-G-protein interactions, a yeast two-hybrid screen was initiated using the TPR and linker region of AGS3 as bait. This screen identified the serine/ threonine kinase LKB1, which is involved in the regulation of cell cycle progression and polarity. Protein interaction assays in mammalian systems using transfected cells or brain lysate indicated the regulated formation of a protein complex consisting of LKB1, AGS3, and Gproteins. The interaction between AGS3 and LKB1 was also observed with orthologous proteins in Drosophila where both proteins are involved in cell polarity. LKB1 immunoprecipitates from COS7 cells transfected with LKB1 phosphorylated the GPR domains of AGS3 and the related protein LGN but not the AGS3-TPR domain. GPR domain phosphorylation was completely blocked by a consensus GPR motif peptide, and placement of a phosphate moiety within a consensus GPR motif reduced the ability of the peptide to interact with G-proteins. These data suggest that phosphorylation of GPR domains may be a general mechanism regulating the interaction of GPR-containing proteins with G-proteins. Such a mechanism may be of particular note in regard to localized signal processing in the plasma membrane involving G-protein subunits and/or intracellular functions regulated by heterotrimeric G-proteins that occur independently of a typical G-protein-coupled receptor.
AGS3 1 was identified in a functional screen for receptorindependent activators of G-protein signaling (1,2). Surprisingly the activation of G-protein signaling in the functional screen was independent of nucleotide exchange on the G␣ subunit suggesting unexpected mechanisms for regulating the activation state of heterotrimeric G-proteins. AGS3 interacts with G-proteins (G i /G o ) via its four G-protein regulatory (GPR) or GoLoco motifs, each of which interacts with G␣ (G i Ͼ G o ) and stabilizes the GDP-bound conformation of G␣ (3). The GPR motif is also found in other proteins including LGN, RGS12, RGS14, Rap1GAPII, Pcp2, and G18.1b (1,4). AGS3 also contains seven tetratricopeptide repeats (TPRs) in the first half of the protein, and these domains may serve as a regulatory domain for the GPR-G-protein interaction, or they may target the protein to different microdomains within the cell (5). A similar motif structure is found in the AGS3-related protein LGN in mammals (6,7) as well as in the AGS3/LGN ortholog Pins in Drosophila melanogaster, which is a key determinant of cell polarity (8 -12). A role for GPR-containing proteins and G-proteins in cell polarity is also suggested by studies in Caenorhabditis elegans (13,14).
AGS3 and other accessory proteins (proteins distinct from receptors, G-proteins, and effectors) may influence receptormediated signaling events and/or mediate signal input to Gproteins independently of a G-protein coupled receptor. Such proteins may also serve as alternative binding partners for G-protein subunits independently of heterotrimer formation (1,2,15), and the existence of these accessory proteins suggests unexpected functional roles for G-proteins within the cell. As an initial approach to define the cellular control mechanisms for AGS3-G-protein interactions, we sought to identify binding partners for the TPR domains of AGS3.
We isolated several candidate AGS3-TPR-interacting proteins in a yeast two-hybrid screen, one of which corresponded to the carboxyl-terminal 107 amino acids of LKB1, also known as serine/threonine kinase 11 (STK11) (16). Loss of LKB1 is implicated in Peutz-Jeghers syndrome, a rare inherited intestinal polyposis syndrome (16 -18), and LKB1 is actually the mammalian counterpart of the C. elegans gene par-4, which is a member of a group of polarity-determining genes during embryogenesis in both C. elegans and Drosophila (19 -21). Significantly LKB1 phosphorylates AGS3 in its GPR domain, and this was completely blocked by a consensus GPR motif peptide. Placement of a phosphate moiety within a consensus GPR motif markedly reduced the ability of the peptide to interact with G-proteins suggesting that phosphorylation of GPR motifs may be a general mechanism regulating the interaction of GPR-containing proteins with G-proteins. Such a mechanism may be of particular note in regard to the localized signal processing in the plasma membrane involving G-protein subunits and/or intracellular functions regulated by heterotrimeric G-proteins that occur independently of a typical G-protein coupled receptor.

EXPERIMENTAL PROCEDURES
Materials-Yeast strains pretransformed with prey libraries, c-Myc monoclonal antiserum, and KC-8 chemically competent cells were obtained from Clontech (Palo Alto, CA). Bait vector pGBKT7 and yeast strains Y187 and AH109 were kindly provided by Dr. Tim McQuinn (Medical University of South Carolina). [␥-32 P]ATP and [ 32 P]orthophosphate were obtained from PerkinElmer Life Sciences. Sodium orthovanadate and RNase A were obtained from Sigma. Okadaic acid was obtained from Calbiochem. pMAL-c2x and amylose-agarose beads were obtained from New England Biolabs (Boston, MA). Other materials were obtained as described elsewhere (3,7).
Yeast Two-hybrid Screening-AGS3-TPR (Met 1 -Ile 462 ) was generated by PCR. Restriction enzyme-digested PCR products were subcloned into pGBKT7 to generate the TPR bait construct. TPR and empty pGBKT7 vector were transformed into AH109 by the lithium acetate method. Expression of bait fusion proteins was confirmed by immunoblotting with anti-c-Myc. Basal activity of bait strains was assayed by nutritional selection. AH109 yeast strains expressing TPR as bait were mated with Y187 yeast strains expressing an 11-day-old mouse embryo cDNA library by following the manufacturer's protocol. The mated yeast culture was plated onto 120 quadruple dropout (Trp Ϫ Leu Ϫ His Ϫ Ade Ϫ ) plates that were then incubated at 30°C for 7 days. ␤-Galactosidase activity was screened using the colony-lift filter assay according to the manufacturer's directions using diploid p53/ SV40 large T antigen interaction (diploid strain PJ69-2A[pVA3-1] ϫ Y187[pTD1-1]) as a positive control as supplied by the manufacturer. Yeast plasmid DNA was isolated and used to transform competent KC-8 Escherichia coli cells. Transformants containing the prey vector were selected by plating onto M9 Leu Ϫ plates. Plasmids isolated from KC-8 transformants were transformed into XL1-Blue E. coli cells for further processing and retransformation of yeast strains.
Immunoprecipitation and Cell Labeling-100-mm dishes of confluent COS7 cells were transfected with either 10 g of empty vector (pcDNA3), 5 g of pcDNA3::AGS3 ϩ 5 g of empty vector, or 5 g of pcDNA3::AGS3 ϩ 5 g of pcDNA3::LKB1 (mouse). After 24 h, cells were lysed in Nonidet P-40 lysis buffer and incubated on ice for 1 h. The lysate was centrifuged at 100,000 ϫ g for 30 min at 4°C and precleared with Gamma-Bind Sepharose (Amersham Biosciences). The precleared lysates (1 mg of protein) were incubated with 5 g of anti-LKB1 (Upstate Biotechnology, Inc., Lake Placid, NY) for 12-18 h at 4°C. Gamma-Bind Sepharose was added, and incubation continued for 30 min. The resin was pelleted and used for kinase assays. Samples were washed three times with Nonidet P-40 lysis buffer, resuspended in 5ϫ protein sample buffer, and placed in a boiling water bath for 3 min followed by SDS-PAGE and immunoblotting with AGS3-specific (PEP32) (3) and LKB-specific (P6) (22) antisera. PC12 cells were labeled with [ 32 P]orthophosphate (8500 -9120 Ci/mmol) according to Kang et al. (23) except that phosphate-free Dulbecco's modified Eagle's medium was used.
Kinase Assay-LKB1 immunoprecipitates were washed three times with Nonidet P-40 lysis buffer and three times with kinase buffer A (50 mM Tris, pH 7.5, 0.1% ␤-mercaptoethanol, 0.1 mM EGTA, 10 mM MnCl 2 , 0.5 M okadaic acid) (17). The Gamma-Bind Sepharose was then resuspended in kinase buffer A containing 10 M [␥-32 P]ATP (1000 cpm/pmol) and 1 M purified GST fusion protein. Reactions were incubated at 30°C for 1 h. The Gamma-Bind Sepharose beads were pelleted, and the supernatant was removed and incubated with glutathione-Sepharose (Amersham Biosciences) for 30 min at 24°C to isolate the GST fusion proteins. The glutathione-Sepharose was pelleted and washed three times with kinase buffer A, resuspended in 5ϫ protein sample buffer, and placed in a boiling water bath for 3 min followed by SDS-PAGE and autoradiography.

RESULTS AND DISCUSSION
Although AGS3 clearly interacts with G-proteins and the GPR motifs in AGS3 and other GPR-containing proteins stabilize the GDP-bound conformation of G␣, only a subpopulation of AGS3 and G-proteins are associated with each other in brain lysates (3), and the two proteins exhibit minimal overlap in terms of their subcellular distribution (5). These data suggest that the interaction between AGS3 and G-proteins is a regulated event. As part of a broader strategy to address this issue and define the role of AGS3-G-protein interactions in cellular function, we used a protein interaction screen to identify binding partners for the TPR and linker region of AGS3. A yeast two-hybrid screen of a mouse 11-day-old embryonic cDNA library using AGS3-TPR (Met 1 -Ile 462 ) as bait yielded several candidate AGS3 binding partners. The screen was run with high stringency by directly using quadruple dropout selection (Trp Ϫ Leu Ϫ His Ϫ Ade Ϫ ) followed by a secondary selection for colonies that exhibited strong ␤-galactosidase activity within 30 min. Sixteen cDNAs were isolated, five of which encoded DNAbinding proteins or proteins involved in regulation of transcription or translation. Of the remaining 11, one cDNA clone encoded an extracellular protein, and five encoded previously unidentified proteins or proteins of unknown function. The remaining cDNA clones encoded portions of murine robo-1, an axonal guidance receptor during central nervous system development (24); microtubule/actin cross-linking factor (MACF, also known as ACF7), a member of the plakin family implicated in epithelial and neuronal polarity (25); MARCKS (myristoylated alanine-rich C kinase substrate)-like protein, a regulator of actin dynamics, migration, and neuronal development (26); and LKB1/STK11 (27). LKB1 and the Drosophila AGS3/ LGN ortholog Pins are both involved in various aspects of cell polarity and development. As a tertiary screen to select for proteins of potential interest, we asked whether GST fusion proteins of each of these cDNA clones interacted with fulllength AGS3 in brain lysates. Only GST-MACF 2 and GST-LKB1 effectively pulled down AGS3 from rat brain lysates. We first focused our effort on LKB1 as the LKB1 ortholog in C. elegans was previously identified as a PAR (partitioning defective) gene (19) involved in asymmetric division of C. elegans embryos. AGS3/LGN orthologs or proteins containing GPR motifs are also involved in similar events in Drosophila and C. elegans.
The cDNA clone encoding LKB1 contained the last 107 amino acids of the coding region of LKB1. The interaction of AGS3-TPR (Met 1 -Ile 462 ) with LKB1 in the yeast two-hybrid screen required amino acids Asp 338 -Ile 462 in the AGS3 coding region that connects the TPR and GPR domains. 2 Additional regions of AGS3 and LKB1 may also interact with each other in the context of the full-length proteins. 3,4 We then asked whether the interaction between the carboxyl terminus of LKB1 and AGS3 was observed in a mammalian system using a GST fusion protein of the LKB1 fragment isolated in the yeast two-hybrid screen. LKB1-CT (Asp 330 -Gln 436 ) effectively interacted with endogenous, full-length AGS3 in rat brain lysates (Fig. 1). Interestingly this complex also contained G i ␣ subunits, which is likely due to an interaction of G-proteins with the GPR motifs of AGS3 (3). The presence of G-proteins in this complex was nucleotidedependent in that it was not observed in the presence of the nonhydrolyzable GTP analog GTP␥S, which is consistent with the demonstrated preference of GPR motifs for the GDP-bound conformation of G␣ (1,3,28). The interaction of AGS3 itself with LKB1 was not influenced by guanine nucleotides.
The LKB1-AGS3 interaction was further addressed with the full-length proteins in the intact cell. Co-transfection of cDNAs encoding full-length AGS3 and LKB1 in COS7 cells and subsequent immunoprecipitation with LKB1 antiserum resulted in co-immunoprecipitation of AGS3 (Fig. 1B). This interaction was specific for LKB1 as immunoprecipitation with LKB1 antiserum from cells transfected with AGS3 alone did not coimmunoprecipitate AGS3 (Fig. 1B). In contrast to the results obtained with the GST-LKB1-CT (Asp 330 -Gln 436 ) fusion protein in which G i ␣ was brought down with the LKB1-CT⅐AGS3 complex from brain lysates, G i ␣ was not found in the co-immunoprecipitation complex of the full-length proteins suggesting that LKB1 may process incoming signals to regulate the interaction between AGS3 and G-proteins or target the protein to a microdomain where G-proteins are inaccessible. 3 To provide further evidence for a functional interaction between LKB1 and AGS3, we asked whether the interaction was evolutionarily conserved in Drosophila. Drosophila LKB1 (DmLKB1) co-immunoprecipitated with the AGS3 ortholog Pins, and conversely Pins co-immunoprecipitated with DmLKB1 in Drosophila embryo lysates, indicating an interac- Similar results were obtained in three separate experiments. The Input lane represents 1 ⁄10 of the total volume in each interaction assay. B, GTP␥ 35 S (500 nM) binding to G i ␣ (100 nM) was measured after incubation with increasing concentrations of control GPR consensus peptide and a phosphorylated Ser 16 (PhosphoS16) GPR peptide as described under "Experimental Procedures." Protein interaction assays and GTP␥S binding assays were performed as described previously (28). Data are expressed as the percentage of specific binding (ϳ0.5 pmol) observed in the absence of added peptide and are expressed as the mean Ϯ S.E. of two experiments with duplicate determinations. tion of the full-length Drosophila proteins (Fig. 1C). In both the COS7 transfectants and the Drosophila embryos only a subpopulation of AGS3 (or Pins) was actually complexed with LKB1 following immunoprecipitation. This may reflect the affinity of the interaction, stoichiometric considerations, and/or regulation of the interaction by an as yet undefined signal(s). The AGS3 ortholog in Drosophila plays a critical role in cell polarity that apparently also involves heterotrimeric G-proteins (8 -12). The LKB1 ortholog in Drosophila was also recently identified in a genetic screen for defects in oocyte and epithelial cell polarity (21). The demonstration of an interaction between LKB1 and AGS3/LGN orthologs in Drosophila provides additional evidence for functionality of this interaction and suggests a role for this interaction in the regulation of cell polarity and cell division.
The 180-amino acid kinase domain (Lys 44 -Pro 314 ) of LKB1 is upstream of the region interacting with AGS3 (Fig. 1). To determine whether AGS3 is phosphorylated by LKB1, we performed in vitro kinase assays using LKB1 immunoprecipitated from LKB1-transfected COS7 cells and purified GST-AGS3 fusion proteins as substrate (Fig. 2A). The region of AGS3 containing GPR motifs was specifically phosphorylated by LKB1, whereas the TPR-linker domain of AGS3 was not ( Fig.  2A). The specific phosphorylation of the region of AGS3 containing the GPR motifs is of particular interest as this region of the protein serves as a potential scaffold for G-protein ␣ subunits (3). There are 24 serines/threonines in the GPR domain of AGS3, 16 of which are found in the GPR motifs themselves. All known GPR motifs contain 1-3 serines/threonines. LKB1 also phosphorylated the GPR domain found in the AGS3-related protein LGN, and this is of particular interest as both LKB1 and LGN regulate the progression of the cell cycle (7,29,30). 3 [ 32 P]Orthophosphate labeling experiments in PC12 cells, which express endogenous AGS3, followed by immunoprecipitation with AGS3-specific antiserum indicate that AGS3 is indeed phosphorylated (Fig. 2C).
Although we do not as yet know the precise site of phosphorylation of AGS3-GPR by LKB1, the phosphorylation of the GPR domain was completely blocked by a consensus GPR peptide (Fig. 2B). The action of the GPR peptide is specific as a scrambled GPR peptide containing the same residues in a different order was ineffective (Fig. 2B). These data suggest that the GPR motif is either itself the site of phosphorylation and docks within the active site of LKB1 or it is an additional anchor for protein interaction. Peptidomimetics derived from the GPR motif may actually be a path for the development of LKB1 inhibitors.
As the GPR peptide effectively blocked AGS3-GPR phosphorylation by LKB1 immunoprecipitates, we asked whether phosphorylation within the GPR motif could potentially influence the interaction of AGS3 with G-proteins. We initially addressed this possibility by placing a phosphate moiety on a serine found in the core of the GPR motif. A 28-amino acid peptide encompassing the core consensus GPR motif effectively inhibits the interaction of GPR-containing proteins with G i ␣ and also mimics the action of GPR-containing proteins by inhibiting the binding of GTP␥S to G-protein (28). This action of the GPR peptide involves discrete residues within the GPR motif (4, 31), mutation of which leads to a loss of activity as indicated for the Q22A peptide (28,31) (Fig. 3A). Phosphorylation of a serine residue within the GPR motif immediately downstream of the invariant residue Gln 15 markedly decreased the ability of the GPR motif to interact with G i ␣ and to inhibit GTP␥S binding to G i ␣ (Fig. 3, A and B).
LKB1 and other serine/threonine kinases may exert a regulatory influence on the interaction of G-protein ␣ subunits with GPR motif-containing proteins by phosphorylation of residues within the GPR motif. GPR motifs are found in several proteins involved in signal propagation including LGN, RGS12, RGS14, Rap1GAP, Pcp2, and G18.1b (1,4). A recent report also suggested that phosphorylation near the GPR motif of RGS14 may influence GPR-G i ␣ interaction (32). Such a mechanism may allow the discrete regulation of G-protein signaling in specific subcellular compartments that occurs independently of a Gprotein-coupled receptor. The localized regulation of such events is a signature mechanism for the determination of cell polarity and asymmetric cell division observed in stem cells during tissue development.