Structure of a Cell Polarity Regulator, a Complex between Atypical ...
THE JOURNAL OF BIOLOGICAL CHEMISTRY © 2005 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 280, No. 10, Issue of March 11, pp. 9653–9661, 2005 Printed in U.S.A.
Structure of a Cell Polarity Regulator, a Complex between Atypical PKC and Par6 PB1 Domains* Received for publication, August 26, 2004, and in revised form, October 27, 2004 Published, JBC Papers in Press, December 7, 2004, DOI 10.1074/jbc.M409823200
Yoshinori Hirano‡, Sosuke Yoshinaga‡, Ryu Takeya§, Nobuo N. Suzuki‡, Masataka Horiuchi‡, Motoyuki Kohjima§, Hideki Sumimoto§¶, and Fuyuhiko Inagaki‡储 From the ‡Department of Structural Biology, Graduate School of Pharmaceutical Sciences, Hokkaido University, N-12, W-6, Kita-ku, Sapporo 060-0812, Japan, §Medical Institute of Bioregulation, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan, and ¶CREST, Japan Science and Technology, Kawaguchi 332-0012, Japan
A complex of atypical PKC and Par6 is a common regulator for cell polarity-related processes, which is an essential clue to evolutionary conserved cell polarity regulation. Here, we determined the crystal structure of the complex of PKC and Par6␣ PB1 domains to a resolution of 1.5 Å. Both PB1 domains adopt a ubiquitin fold. PKC PB1 presents an OPR, PC, and AID (OPCA) motif, 28 amino acid residues with acidic and hydrophobic residues, which interacts with the conserved lysine residue of Par6␣ PB1 in a front and back manner. On the interface, several salt bridges are formed including the conserved acidic residues on the OPCA motif of PKC PB1 and the conserved lysine residue on the Par6␣ PB1. Structural comparison of the PKC and Par6␣ PB1 complex with the p40phox and p67phox PB1 domain complex, subunits of neutrophil NADPH oxidase, reveals that the specific interaction is achieved by tilting the interface so that the insertion or extension in the sequence is engaged in the specificity determinant. The PB1 domain develops the interaction surface on the ubiquitin fold to increase the versatility of molecular interaction.
Cell polarity is a fundamental property related to many cellular processes. Recently, a complex of atypical protein kinase C (aPKC)1 and Par6 (aPKC-Par6) has received much attention as a common cell polarity regulator conserved from nematodes to mammals. Par6 was initially identified as one of the products of par genes (par-1 to par-6) encoding essential components of cell division in the early embryo found by lethal genetic screening of Caenorhabditis elegans (1). In addition to Par proteins, a defect in Pkc3 (a homologue of aPKC in C. elegans) also disrupts embryonic polarity (2). aPKC-Par6 is a conserved signaling complex from C. elegans to Homo sapiens (3) and is now regarded as a common regulator for asymmetric
* This work is supported by the National Project on Protein Structural and Functional Analyses from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. 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. The atomic coordinates and structure factors (code 1WMH) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/). 储 To whom correspondence should be addressed: Graduate School of Pharmaceutical Sciences, Hokkaido University, N-12, W-6, Kita-ku, Sapporo 060-0812, Japan. Tel.: 81-11-706-3976; Fax: 81-11-706-4979; E-mail: [email protected]. 1 The abbreviations used are: aPKC, atypical protein kinase C; TJ, tight junction; SeMet, selenomethionine; MDCK, Madin-Darby canine kidney; HA, hemagglutinin; wt, wild type; OPCA, OPR, PC, and AID. This paper is available on line at http://www.jbc.org
cell division (4, 5), directed cell migration (6), axon specification (7), and tight junction (TJ) formation in epithelial cells (8) in various species. Thus, aPKC-Par6 is an essential clue to evolutionary conserved cell polarity regulation (9). aPKC-Par6 is characteristically located at a specific region of cells, such as the anterior cortex of the nematode embryo (1, 2), apical cortex of fly neuroblasts (10, 11), leading edge of migrating cells (6), and TJs in epithelial cells (8), implying that the complex formation is required for aPKC specific localization. The regulation of cellular localization as well as kinase activity of aPKC is essential for phosphorylation of its intrinsic target. Thus, the scaffold proteins that bind to PKCs play a crucial role in the control of the cellular localization of PKCs (12, 13). In particular, TJ formation is a prototype of cell polarity events. TJs encircle cells at the most apical region of the lateral membrane and function as both paracellular and intramembrane diffusion barriers. Moreover TJ formation triggers cell proliferation and differentiation (14, 15). aPKC-Par6 is localized at the TJs and together with Par3 is involved in TJ assembly. Although recruitment of the ternary complex to the TJs and regulation of TJ formation are still elusive, phosphorylation by aPKC seems to be a clue to signal transduction because the kinase activity deficient aPKC abrogates TJ formation (16). Notably, Par6 can regulate aPKC kinase activity in a Cdc42dependent manner (17). The complex formation of aPKC and the scaffold protein Par6 is established through the interaction between PB1 domains existing at the N terminus of both proteins. In TJ formation, the disruption of PB1-PB1 interaction in aPKC-Par6 causes mislocalization of aPKC or Par6 (18). Therefore, PB1PB1 interaction between aPKC and Par6 is related to their localization and has an indispensable role in cell polarity-related processes. Until now, the PB1 domain has been found in more than 200 signaling proteins including p40phox, p67phox, cytosolic factors of NADPH oxidase, Bem1p, Cdc24p, MEK5 and p62/ZIP on the SMART data base (19). Recent studies have demonstrated that PB1 domain functions as a protein binding module through PB1-mediated heterodimerization or homo-oligomerization (20 –23). The PB1 domain is further classified into two types, type I and type II PB1 domains. The PB1-mediated complex is formed through the interaction between type I and type II PB1 domains. Type I PB1 adopts the OPCA motif as a binding site, a remarkable consensus sequence in the PB1 domain with conserved acidic and hydrophobic residues (12, 24, 25). On the other hand, type II PB1 requires a conserved lysine residue in 1 (Fig. 1). Binding assays for PB1-PB1 interaction suggest that the electrostatic interaction between conserved acidic residues on the OPCA motif in type I-PB1 and a conserved basic residue in type II PB1 is responsible for PB1-PB1 interaction.
9653
9654
Complex Structure of Cell Polarity Regulator
FIG. 1. Structure-based sequential alignment of the PB1 domain. The conserved hydrophobic residues are highlighted in red, and hydrophilic residues are highlighted in blue. Secondary structure components of type I, type I and type II, or type II PB1 are derived from the tertiary structure of Cdc24p, PKC, or Par6␣, respectively. The OPCA motif is enclosed in green. The two conserved acidic regions of type I, A1 and A2, are boxed, and the two conserved basic region of type II, B1 and B2/B2⬘, are highlighted in orange and boxed. Residues involved in the interaction between PKC PB1 and Par6␣ are mapped on the sequence. Stars represent residues with side chains involved in an intermolecular salt bridge and/or hydrogen bond. Asterisks represent residues involved in an intermolecular hydrophobic interaction. Filled circles represent residues with the main chain involved in an intermolecular hydrogen bond.
Despite its electrostatic nature, the PB1-PB1 interaction is highly specific and exclusive (20). Interestingly, aPKC binds to several proteins such as Par6, ZIP/p62, and MEK5 through its PB1 domain (18, 26) and mediates distinct signals in a binding protein-dependent manner. To date, whereas structure determination of the PB1 domain has been performed (21, 23, 27, 28), the general and specific properties of the PB1-PB1 interaction mode are yet to be elucidated. Here, we determined the crystal structure of the PB1 domain complex of PKC and Par6␣. Although the electrostatic interaction was assumed to be responsible for complex formation, the structure reveals salt bridge formation in the PB1-PB1 complex. Furthermore, it provides insights into the general and specific properties of PB1-PB1 interaction. We also identified the role of PB1-PB1 interaction in TJ formation in aPKC-Par6. EXPERIMENTAL PROCEDURES
Plasmid Construction—Complementary DNA fragments encoding various lengths of human Par6␣ and PKC were prepared as described previously (18, 29). Mutations leading to the indicated amino acid substitutions were introduced by PCR-mediated site-directed mutagenesis, and the mutated fragments were cloned into the indicated vectors. All the constructs were sequenced to confirm their identities. Sample Preparation for Crystallization—The PKC (16 –99) gene was cloned in pGEX-6P-1 (Amersham Biosciences), and the Par6␣ (14 –95) gene was cloned in pET HP, a modified version of pET-28a (Novagen), converting the thrombin protease recognition site into the PreScission protease recognition site. Both constructs were co-expressed in Escherichia coli strain BL21 (DE3). Because PKC-(16 –99) and Par6␣-(14 – 95) form a complex in E. coli cells, the complex is acquired via purification from bacterial lysate by affinity chromatography using glutathione-Sepharose 4B resin (Amersham Biosciences). After proteins were treated with PreScission protease (Amersham Biosciences), purification using anion exchange chromatography and gel filtration chromatography were performed. On the process of purification, PKC(16 –99) and Par6␣-(14 –95) retained a stable complex. Finally, the complex of PKC PB1 and Par6␣ PB1 was concentrated up to 12–15 mg/ml in a buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, and 5 mM dithiothreitol, and this solution was used for crystallization. Selenomethionine (SeMet)-substituted protein was expressed in E. coli strain B834 (DE3) with LeMaster medium (30). Purification was performed in the same manner as native protein preparation, and SeMet substitution for Met was confirmed using matrix-assisted laser
desorption ionization time-of-flight mass spectrometry. After concentration to 12–15 mg/ml, SeMet derivatives were used as a crystallization sample. Crystallization—Crystallization screening was performed using a sitting drop vapor diffusion method at 293 K. 100 l of volume of reservoir was applied to wells on crystal clear strips no indent (Hampton Research). Each drop consisted of 0.5 l of protein solution and an equal amount of reservoir solution. Approximately 12–15 mg/ml of the complex of PKC PB1 and Par6␣ PB1 in 50 mM Tris (pH 8.0), 150 mM NaCl, and 1 mM dithiothreitol was used for crystallization. A single crystal was obtained in 3.6 M sodium formate and 0.1 M Tris (pH 8.0) at 293 K by using a microseeding technique. Data Collection and Structure Determination—Prior to x-ray diffraction experiments, crystals were soaked in a cryoprotectant solution (3.6 M sodium formate, 0.1 M Tris (pH 8.0), 20% glycerol) for several seconds and were immediately flash-cooled to 100 K in a nitrogen gas stream. Native diffraction data were collected on an MAR CCD detector using synchrotron radiation at the BL41XU beam line, SPring-8. The camera to crystal distance was set to 150 mm, and a total of 180 images were collected at 1° oscillation angle per image. The diffraction data were processed using the HKL2000 program (31). The crystal belongs to space group P21212 with unit cell parameters a ⫽ 94.150 Å, b ⫽ 38.363 Å, c ⫽ 45.478 Å and a cell volume of 164,261 Å3. The molecular mass of a complex of PKC PB1 and Par6␣ PB1 was ⬃19,427 Da, thus, the Matthews coefficient (VM) was 2.1 Å3/Da and the solvent content was 42%, assuming one molecule exist in an asymmetric unit. Multiwavelength anomalous dispersion data sets were collected using a single crystal of SeMet-substituted protein at three wavelengths at the BL41XU beam line, SPring-8 as well as native data sets. The x-ray diffraction statistics are summarized in Table I. Determination of the selenium site in SeMet-substituted proteins, phasing, and refinement were performed by crystallography and NMR system software (32). Modeling was performed by Turbo-Frodo (33). Finally, refinement was performed up to 1.5 Å, and the statistics are summarized in Table I. The coordinates and structure factor of a PB1 domain complex of PKC and Par6␣ have been deposited in the Protein Data Bank (accession code 1WMH). The molecular surface model was generated by the GRASP program (34). All figures of the structural model except the model of electrostatic surface potential were created by Pymol software. Yeast Two-hybrid Experiments—Yeast HF7c cells containing a HIS3 reporter gene were co-transformed with a pair of pGBT9 (Clontech) encoding the wild-type or mutant PKC and pGADGH (Clontech) encoding the wild-type or mutant Par6␣ as described previously (35). Following the selection for Trp⫹ and Leu⫹ phenotype, the transfor-
Complex Structure of Cell Polarity Regulator
9655
TABLE I Crystallographic analysis statistics Data collection
Space group Unit cell (Å)
Wavelength (Å) Resolution range (Å) Observed/unique reflections Completeness (%) Rmergea 具I/典 Refinement statistics Resolution range (Å) Rwork (%) Rfree (%)b r.m.s.d. from ideal valuesc Bonds (Å) Angles (deg) Model quality (Ramachandran plot) (%) Residues in most favored regions Residues in additionally allowed regions Residues in generously allowed regions Residues in disallowed regions
SeMet (Peak)
a ⫽ 94.663 b ⫽ 38.407 c ⫽ 45.570 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9793 50.0–1.9 97,678/26,471 99.3 0.044 13.5
SeMet (Edge)
a ⫽ 94.648 b ⫽ 38.393 c ⫽ 45.546 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9795 50.0–1.9 97,467/26,435 99.2 0.044 13.8
P21212
SeMet (Remote)
Native
a ⫽ 94.574 b ⫽ 38.382 c ⫽ 45.537 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9843 50.0–1.9 96,296/26,133 98.7 0.045 13.4
a ⫽ 94.150 b ⫽ 38.363 c ⫽ 45.478 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9791 50.0–1.5 17,9821/50,330 98.5 0.054 16.5
50.0–1.5 21.6 22.4 0.004 1.20 93.8 5.5 0.0 0.7
Rmerge ⫽ ⌺hkl⌺i兩I(hkl)i ⫺ 具I(hkl)典兩/⌺hklI(hkl), where I(hkl)i indicates the observed intensity, and 具I(hkl)典 is the mean intensity. Rfree was calculated on a random 9.9% reflections of the data. c r.m.s.d., root mean square deviation. a b
mants were tested for their ability to grow on plates lacking histidine supplemented with 2 mM 3-aminotriazole to suppress background growth according to the manufacturer’s recommendation (Clontech). Formation of Epithelial Tight Junction—To investigate the effect of PKC or Par6␣ on formation of TJs, we performed a calcium switch method using canine kidney epithelial MDCK cells. Briefly, MDCK cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum (36) and transfected with pEF-BOS-hemagglutinin, a vector for expression as a HA-tagged protein, that encodes the indicated form of PKC or Par6␣ (36). After incubation for 24 h in Dulbecco’s modified Eagle’s medium with 10% fetal calf serum, the transfected cells were seeded on 12-mm round glass coverslips and grown in normal calcium medium to form confluent monolayers. The monolayer cells were cultured for 60 min in the calcium-free medium S-MEM (Invitrogen) and then incubated for 2 h in the calcium-containing medium Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum. After three washes with phosphate-buffered saline, cells were fixed for 15 min with 3.7% formaldehyde and then permeabilized for 15 min in phosphate-buffered saline containing 0.5% Triton X-100. After blocking for 60 min with phosphate-buffered saline containing 3% bovine serum albumin, the samples were initially incubated with anti-HA (Y11) rabbit polyclonal antibodies (Santa Cruz Biotechnology) and anti-ZO-1 mouse monoclonal antibody (36) and subsequently probed with Alexa FluorTM 594-labeled goat anti-rabbit IgG antibodies (Molecular Probes) and Alexa FluorTM 488-labeled goat anti-mouse IgG antibodies (Molecular Probes). Images were visualized with an LSM 510 confocal laser scanning microscope (Carl Zeiss). RESULTS AND DISCUSSION
Overall Structure of the PB1 Complex of PKC and Par6␣— The crystal structure of the PB1 complex of PKC and Par6␣ has been determined to a resolution of 1.5 Å by a multiwavelength anomalous dispersion technique using an SeMet-substituted protein as described under “Experimental Procedures.” The crystal belongs to the space group P21212, and one complex exists per asymmetric unit. Of a total of 175 residues, 5 Nterminal residues of PKC PB1 and 4 N-terminal residues of Par6␣ PB1 that are derived from cloning artifacts and Cys99 of PKC PB1, a C-terminal residue of the construct, are disordered. In Par6␣ PB1, Pro87 adopts a cis conformation. A summary of the collected data and the structural statistics is given in Table I. The overall structure of the complex is shown in Fig. 2A. Both PB1 domains comprise two ␣ helices and a mixed -sheet which consists of five  strands that belong to the ubiquitin
superfamily. The OPCA motif region of PKC (Leu56-Leu83) has a ␣ fold, and the  turn is categorized as a type I ⫹ G1 -bulge, characteristic of type I PB1 domains (23). The PB1 domains of PKC and Par6␣ are remarkably similar as shown in Fig. 2B (root mean square deviation ⫽ 0.86 Å for backbone atoms). Because the solution structure of PKC (free state) was determined by NMR (28), the free and the complex states of PKC PB1 can be compared (Fig. 2C). Both of the overall structures are well superimposed (root mean square deviation ⫽ 0.96 Å for backbone atoms). We performed superimposition between free and complex state structures, using not only over all structure but also more limited region such as OPCA motif. As the result, free and complex state structures are uniformly superimposed, and notable conformational exchange is not observed except for the loop region that is flexible in free structure. Interaction Surface of the PKC and Par6␣ PB1 Complex— Despite the overall structural similarity between the PKC and Par6␣ PB1 domains, they formed an asymmetric heterodimer using a different surface from each protein (Fig. 2A). The interaction surface of PKC PB1 is the OPCA motif corresponding to 3, 4, and ␣2, whereas that of Par6␣ PB1 is composed of 1, 2, the C-terminal region of ␣1, and the N-terminal region of 5. Thus, PKC serves as a type I PB1 domain and Par6␣ serves as a type II PB1 domain, respectively, as was suggested by the previous mutational analyses (18, 27). In PKC PB1, the OPCA motif is responsible for the interaction with Par6␣ PB1, but no other regions are involved in the interaction. The interaction surface of PKC PB1 presents two acidic regions, A1 (Asp63, Glu65, and Asp67) and A2 (Glu76), whereas that of Par6␣ is comprised of two basic regions, B1 (Lys19, Arg28 and Arg89) and B2⬘ (Arg27) (Fig. 3A). In Par6␣ PB1, Arg27 and Arg28, two consecutive basic residues on 2, direct their side chains to the opposite side so that Par6␣ PB1 presents two distinct basic regions, B1 and B2⬘, at both sides of 2 (Fig. 3A, right). The complementarity of the surface potential between A1-B1 and A2-B2⬘ supports the notion that electrostatic interaction plays a major role in PB1 heterodimer formation. In addition to these interactions, Arg82 (PKC)-Glu17 (Par6␣) and a hydrophobic interaction, including Leu75 of PKC and Phe29,
9656
Complex Structure of Cell Polarity Regulator
FIG. 2. Structural properties of a PB1 complex of PKC-Par6␣. A, a ribbon representation of a PB1 complex of PKC-Par6␣. PKC PB1 is indicated in pink, and Par6␣ PB1 is indicated in green. The direct interaction surfaces of PKC PB1 and Par6␣ PB1 are indicated in red and blue, respectively. B, a superposition of PKC PB1 and Par6␣ PB1 in the PB1 complex. The color of each PB1 domain is the same as in A. C, a superposition of the free state (solution structure) and complex state (crystal structure) of PKC PB1. The free state is indicated in cyan, and the complex state is indicated in pink.
Leu45, and Val49 of Par6␣, also contribute to the intermolecular interaction (Fig. 3A). Description of the Interface in the PKC-Par6␣ PB1 Domain Complex—A detailed inspection of the interface gives the characteristic interaction mode in the PKC-Par6␣ complex (Fig. 3, A and B). In PKC PB1, A1 is located on the  turn in the OPCA motif formed by the conserved acidic residues, Asp63, Glu65, and Asp67. As the  turn adopts a type I ⫹ G1  bulge, these acidic residues orient to the same side to form an acidic surface. B1 of Par6␣ consists of Lys19 in 1, Arg28 in 2, and Arg89 in 5. The complementarily charged side chains of PKC and Par6␣ form salt bridges as shown in Fig. 2B. Notably, Lys19 of Par6␣ PB1 forms salt bridges with all of the three conserved acidic residues. The interaction mediated by Lys19 is consistent with its strict conservation in type II PB1. Arg28 and Arg89 of Par6␣ interact with Glu65 and Asp63 of PKC, which fix these carboxyl groups to assist salt bridge formation with Lys19 as well as to compensate for charges on the salt bridge network. Thus, A1-B1 is a major interaction site in the complex formation (Fig. 3B). In the complex of p40phox-p67phox PB1 domains, the corresponding acidic residues on the OPCA motif in p40phox (Asp289, Glu291, and Asp293) interact with the conserved lysine residue in p67phox (Lys355) in a manner similar to that in the PKC-Par6␣ complex (27). The importance of this interaction is evident considering that the type I PB1 domain contains an arginine residue instead of a lysine residue. To form tight salt bridges with the three acidic residues, the lysine residue is essential in this position (Fig. 1). In addition to the A1-B1 interaction, two other interactions, including A2-B2⬘ and Arg82-Glu17, are to be noted (Fig. 3B). A2 is located on ␣2 of PKC PB1 to interact with B2⬘ of Par6␣ PB1. Glu76 located on ␣2 of PKC PB1 forms a salt bridge with Arg27 on 2 of Par6␣ PB1, which also forms a hydrogen bond with Ser72 of PKC. Glu17 of Par6␣ and Arg82 of PKC are located close to each other to form a salt bridge. Furthermore, intramolecular salt bridge formation between Glu79 and Arg82 of PKC and between Glu17 and Arg28 of Par6␣ is observed (Fig. 3B). Besides these salt bridges, there are hydrophobic interactions
involving Leu75 on ␣2 of PKC and Leu45 and Val49 on 3 and Phe29 on 2 of Par6␣ (Fig. 3A). In the interface, not only intermolecular salt bridges but also a number of intramolecular salt bridges and hydrogen bonds are formed. These interactions are observed throughout the interface and form a network (Fig. 3C). Although the intramolecular salt bridges do not directly contribute to the binding, the salt bridge network may affect their side chain coordination to assist in the formation of a stable complex. There are a relatively small number of hydrophobic interactions and watermediated hydrogen bonds in the interface. The surface area of the PKC-Par6␣ complex is ⬃1141 Å2, which is appreciably smaller than those of other protein complexes (37). Tight interaction between PB1 domains is mainly caused by the salt bridge and hydrogen bond formations that amounted to 10 in the complex and much less to hydrophobic interactions. Therefore, the interface of the PKC-Par6␣ PB1 complex is very compact but sufficient to form a stable protein-protein complex. Mutational Analyses of the Residues Responsible for Complex Formation—The mutational analyses using the yeast two-hybrid method were performed to confirm the molecular interaction. The conserved acidic residue (Asp63) in PKC PB1 and the conserved lysine residue in Par6␣ (Lys19) were mutated to alanine, which completely abrogated the binding with each counterpart. The K20A mutation in PKC PB1 and the D63A mutation in Par6␣ showed similar binding affinity to the wild type (Fig. 4). These results are consistent with the previous mutational experiments using a pull-down binding assay (18, 27) and confirm that PKC PB1 serves as type I and Par6␣ PB1 as type II (Fig. 1). Furthermore, we designed several mutants based on the present structure of the PKC-Par6␣ PB1 complex. First, we made mutations at residues R28A, R89A, and R28A/R89A in Par6␣ to investigate the contribution of these residues to the A1-B1 interaction. Although R28A and R89A slightly diminished the interaction, the R28A/R89A double mutant completely impaired the interaction with PKC PB1, consistent with the present structure (Fig. 3, B and C). Then we further investigated the two interaction sites, A2-B2⬘ (Ser72/
Complex Structure of Cell Polarity Regulator
9657
FIG. 3. Detailed molecular interaction between PKC PB1 and Par6␣ PB1. A, ribbon and stick diagrams of a PB1 complex of PKC-Par6␣ represented in an open book style. The two acidic regions of PKC PB1 and the two basic regions of Par6␣ PB1 in the interaction surface are encircled in red and blue, respectively. B, PKC-Par6␣ interface with the corresponding final 2兩Fo兩 ⫺ 兩Fc兩 electron density at 2.0 is shown. A1, A2, B1, and B2⬘ indicate the same as in A. Orange dotted lines indicate intermolecular hydrogen bonds, and blue dotted lines indicate intramolecular hydrogen bonds. C, a stick model of manifold salt bridge network in the interaction surface. Blue spheres represent water molecules. Orange dotted lines indicate intermolecular hydrogen bonds, and blue dotted lines indicate intramolecular hydrogen bonds or water molecule-mediated hydrogen bonds.
Glu76-Arg27) and Arg82-Glu17. Although E76A or R82A mutations in PKC had a slight effect on binding to Par6␣, the E76A/R82A double mutant completely abolished binding to Par6␣ PB1. Thus, the Glu76 and Arg82 of PKC PB1 were confirmed to be involved in the complex formation. The mutational studies by the yeast two-hybrid method (Fig. 4) support the interaction mode revealed herein in the structure of the PKC-Par6␣ PB1 domains (Fig. 3). General Properties of PB1-PB1 Interaction—The PB1 domains were classified into two types, the type I and the type II PB1 domains. Type I uses the OPCA motif, whereas type II uses the conserved lysine residue located on the opposite surface to the OPCA motif for heterodimerization. There is another type of PB1 domain, classified as type I and type II, which contains both the OPCA motif and the conserved lysine residue (23). For example, aPKC PB1 binds to Par6 PB1 as a type I PB1 domain and to MEK5 PB1 as a type II PB1 domain. Now, we aligned the PB1 domains on the structural basis, taking the previous studies into consideration (Fig. 1) (21, 23, 27, 28). We show the interfaces of both type I and type II PB1 domains in an open book style (Fig. 5). The interface in the type I PB1
domain is strictly located on the OPCA motif. Two acidic regions, A1 and A2, are presented on the OPCA motif of the type I PB1 domains that interact with the type II PB1 domains in an electrostatic manner. In contrast to the type I PB1 domain, the general properties of the type II PB1 domain are less understood because of low sequence homology except for the strictly conserved lysine residue involved in B1 that interacts with A1 of the type I PB1 domain. Our present study reveals the generality of the PB1-PB1 interaction especially in the type II PB1 domain. In the PB1 complex of PKC and Par6␣, A1 of PKC interacts with B1, including Lys19, Arg28, and Arg89 of Par6␣. Notably, as the Arg89 of Par6␣ is positioned in 5 as well as Lys19, the conserved lysine residue in 1 interacts with A1. Arg89 corresponds to a highly conserved basic residue in 5 of the type II PB1 domain (Fig. 1). Thus, the conserved basic residue in 5 of the type II PB1 domain is suggested to generally interact with A1 of the type I PB1 domain. In contrast to the A1-B1 interaction, the A2-mediated interaction is quite complicated. In the structure of the p40phoxp67phox PB1 heterodimer, Lys382 in ␣1 of p67phox PB1 interacts
9658
Complex Structure of Cell Polarity Regulator
FIG. 4. PB1 domain-mediated interaction between PKC and Par6␣. A, two-hybrid interaction of Par6␣ with the wild-type or mutant PB1 domain of PKC. Yeast HF7c cells were co-transformed with a pair of pGBT9 encoding wild-type PKC PB1 (wt) or PKC PB1 mutants, E76A, R82A, E76A/R82A, K20A, or D63A, and pGADGH encoding the full-length wild-type Par6␣-F (wt). Following the selection for the Trp⫹ and Leu⫹ phenotype, the histidine-dependent (right) and -independent (left) growth of the transformant was tested as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results. B, two-hybrid interaction of PKC with the wild-type or mutant PB1 domain of Par6␣. HF7c cells were co-transformed with a pair of pGBT9 encoding the full-length wild-type PKC-F (wt) and pGADGH encoding wild-type Par6␣ PB1 (wt) or mutants, R28A, R89A, R28A/R89A, K19A, or D63A. Following the selection for the Trp⫹ and Leu⫹ phenotype, the histidine-dependent (right) and -independent (left) growth of the transformant was tested as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results.
with A2 of p40phox PB1 (27). This basic residue in ␣1 (referred to basic region 2, B2) is also conserved in Bem1p PB1 (corresponding to Arg510), and its contribution to the formation of the PB1 domain complex is supported by the tertiary structure and mutational experiments (21, 23). However, in other type II PB1 domains, the residue involved in B2 is not conserved on the sequential alignment (Fig. 1). Accordingly, the A2-B2 interaction appears to be specific for p40phox-p67phox and Cdc24pBem1p PB1. However, surprisingly, Arg27, a basic residue in 2 of Par6␣, interacts with Glu76 in A2 of PKC. Arg27 presents its side chain to a similar position as Lys382 in p67phox PB1 in the tertiary structure regardless of their different position in the sequence (Fig. 1). So Arg27 of Par6␣ and Lys382 of p40phox give similar electrostatic potential at an equivalent position and form a conserved basic surface (Fig. 5). This basic residue in 2 (basic region 2⬘, B2⬘) is consistently conserved in type II PB1 domains except for p67phox and Bem1p. Thus, the A2-B2/B2⬘ interaction is considered to be conserved as a general property of the PB1-PB1 interaction. Specificity Determinants for PKC-Par6␣ and p40phoxp67phox Complex Formation—Although the formation of salt bridges between A1-B1 and A2-B2/B2⬘ has been identified as a general property of PB1-PB1 interaction, the PB1-PB1 interaction is known to be highly specific and exclusive. As a unique feature of the p40phox-p67phox PB1 heterodimer, the C-terminal non-conserved region of p40phox PB1 was shown by mutational analyses to be responsible for stable complex formation with p67phox PB1 (25). Therefore, in addition to the typical A1-B1 and A2-B2/B2⬘ interactions, the C-terminal extension of p40phox located outside of the canonical PB1 domain boundary provides a specific interface essential for the formation of the p40phox-p67phox PB1 complex. However, the PB1 domain of Cdc24p exists at the C terminus of the protein, and therefore involvement of the C-terminal extension outside of the canonical PB1 domain is excluded from the complex formation with
FIG. 5. General interaction surface of the PB1 domain. Electrostatic surface potentials of the interaction surfaces of type I and type II PB1 domains are presented. The conserved acidic regions of the type I PB1 domain, A1 and A2, are encircled in orange, and the conserved basic regions of the type II PB1 domain, B1and B2/B2⬘, are encircled in blue.
Bem1p PB1 (Fig. 1). Additionally, PKC does not require an extension to bind to ZIP/p62 (28). According to these observations, the interaction mode between the PKC and Par6␣ PB1 domains seems to be distinct from that between p40phox and p67phox PB1 domains. Now, because we know the atomic detail of the two PB1 complexes, PKC-Par6␣ and p40phox-p67phox, we can discuss the molecular interactions that give specificity to each complex. Superposition of the type II PB1 domain of each complex revealed that although the general interaction sites, A1-B1 and A2-B2/B2⬘, are conserved, the interaction surfaces were tilted downward for PKC-Par6␣ and upward for p40phox-p67phox, resulting in the formation of a distinct interface in the PB1 domain complexes. In the PKC-Par6␣ PB1 domains, ␣2 of PKC PB1 is protruded to extensively interact with the residues in 1 and 2 of Par6␣, whereas in the complex of p40phoxp67phoxPB1 domains, the C-terminal extension of p40phox PB1 is engaged in the interaction with p67phox PB1 (Fig. 6A). The molecular surfaces for each complex are highlighted in Fig. 6B. In the PKC-Par6␣ PB1 complex, Leu75 of PKC, located in the N-terminal region of ␣2, interacts with a hydrophobic patch comprised of Phe29, Leu45, and Val49 of Par6␣ (Fig. 6B). In addition, Arg82 of PKC at the C-terminal region of ␣2 forms a salt bridge with Glu17 of Par6␣. Thus, ␣2 of PKC PB1 interacts extensively with Par6␣ PB1.
Complex Structure of Cell Polarity Regulator
9659
FIG. 6. Distinct interaction mode between PKC-Par6␣ and p40phoxp67phox. A, overlay of the PB1 complexes of PKC-Par6␣ and p40phox-p67phox, where the type II PB1 domains of the PB1 complexes (Par6␣ and p67phox, respectively) were overlapped to the maximum. Each of the PB1 domains of PKC, Par6␣, p40phox, and p67phox is indicated in pink, green, orange, or blue, respectively. B, interaction surfaces of PB1 heterodimers of PKC-Par6␣ and p40phox-p67phox are presented in an open book style. The regions involved in ␣2 of type I PB1-mediated interaction are highlighted in red, and the regions involved in the C-terminal extension mediated interaction are highlighted in blue. Other interaction sites are shown in yellow.
In the complex of p40phox-p67phox PB1 domains, the C-terminal extension of p40phox PB1 is engaged in the interaction with the 1/2 loop, 3/4 loop, and the C-terminal region of 5 in p67phox PB1 mainly through hydrophobic interaction and hydrogen bond formation (27). However, in contrast to the complex of PKC-Par6␣ PB1 domains, ␣2 of p40phox PB1 only slightly interacts with p67phox PB1 because of the upward tilt (Fig. 6B). Thus, the change in the tilt angle between the two PB1 complexes is sufficient to endow specificity in the complex of the PB1 domains. Such specific interaction caused by the tilt in the interface was shown in the complex of Ras binding domain and Ras or Rap1A. Ras or Rap1A interacts with a specific effector such as Raf, RalGDS, or phosphatidylinositol 3-kinase ␥ by adjusting the relative orientation of each domain (38). Beside the case of Ras or Rap1A-Ras binding domain interaction, ubiquitin-mediated interactions with Npl4 zinc finger (39) and ubiquitin E2 variant domain (40) is a good exam-
ple. Ubiquitin uses a hydrophobic patch on a  sheet as a common interface with them, whereas ubiquitin confers a specific interface with each domain by tilting its orientation. The specific recognition in domain-domain interaction achieved by tilting the orientation may be a general strategy developed in the process of evolution. Role of the PB1-PB1 Interaction between PKC and Par6␣ in Tight Junction Formation—MDCK cells normally form TJs when aPKC (wt) is overexpressed, whereas TJ assembly in MDCK cells is disturbed when a kinase-deficient mutant of aPKC is overexpressed (16). Thus, the kinase activity of aPKC is known to be important for epithelial polarity. However, the importance of the PB1-PB1 interaction between aPKC and Par6 in TJ assembly is still unclear. Accordingly, we investigated the role of PB1-PB1 interaction in the control of TJ assembly. In accordance with previous reports, MDCK cells normally formed TJs when PKC (wt) was overexpressed (Fig.
9660
Complex Structure of Cell Polarity Regulator
FIG. 7. Role of the PB1 domain-mediated interaction between PKC and Par6␣ in TJ formation of MDCK cells. A–G, MDCK cells were transfected with pEF-BOS-HA encoding PKC (wt), PKC (E76A/R82A), or PKC (D63A). The transfected cells were subjected to the calcium switch method. The sample was then fixed and doubly stained with the anti-HA (A–C) and anti-ZO-1 (D–F) antibodies, and the images were visualized as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results. G, the indicated percentage of MDCK cells ectopically expressing the wild-type or mutant PKC showed incomplete peripheral ZO-1 staining. Values shown in G are means (⫾S.D.) of three independent experiments. H–N, MDCK cells were transfected with pEF-BOS-HA encoding Par6␣ (wt), Par6␣ (K19A), or Par6␣ (R28A/R89A). The transfected cells were subjected to the calcium switch method. The sample was then fixed and doubly stained with the anti-HA (H–J) and anti-ZO-1 (K–M) antibodies, and the images were visualized as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results. N, the indicated percentage of MDCK cells ectopically expressing the wild-type or mutant Par6␣ showed incomplete peripheral ZO-1 staining. Values shown in N are means (⫾S.D.) of three independent experiments.
7). In contrast, overexpression of PKC mutants (PKC D63A and PKC E76A/R82A), which do not bind with Par6, disturbed normal localization of ZO-1, a tight junction marker, implying that TJ assembly was inhibited (Fig. 7). Thus, aPKC requires the interaction with Par6 via the PB1 domain as well as kinase activity for the control of TJ assembly. On the other hand, overexpression of Par6 is known to inhibit TJ assembly (41). Par6 contains a CRIB motif and a PDZ domain besides the PB1 domain. Par6 binds with the activated form of Cdc42 via the CRIB and PDZ (42, 43). Par6 also binds with Par3 (42, 43), Lgl (5, 44, 45), and Pals1 (46) via the PDZ domain, suggesting that Par6 is a scaffold of a multiprotein complex for controlling cell polarity. The overexpression of Par6 may upset the balance of the multiprotein complex to bring about abnormal epithelial cell polarity. We evaluated the effect of Par6␣ mutants that do not bind with aPKC to clarify the role of the Par6␣ PB1 domain. When Par6␣ (wt) was overexpressed in MDCK cells, TJ assembly was inhibited, which is consistent with the results of the previous study (Fig. 7) (41). However, when Par6␣ K19A and R27A/R28A mutants that do not interact with aPKC were overexpressed in MDCK cells, TJ assembly was little inhibited (Fig. 7). Thus, the role of Par6 in the TJ assembly is to interact with aPKC via the PB1 domain. These results strongly suggest that the PB1-PB1 interaction between aPKC and Par6 plays an important role in the regulation of epithelial cell polarity. CONCLUSION
The crystal structure of the PB1 heterodimer of PKC and Par6␣ revealed that the salt bridge formation (A1-B1) between the conserved acidic residues in the OPCA motif of PKC PB1 and the conserved lysine residue in Par6␣ PB1 is mainly re-
sponsible for complex formation. Interestingly, the ammonium group of the conserved lysine residue is tightly held with the three conserved acidic residues in the OPCA motif of PKC PB1. Such salt bridge formation is also observed in the p40phoxp67phox PB1 complex and is regarded as characteristic in PB1 heterodimer formation. In addition, the formation of another salt bridge (A2-B2/B2⬘) was observed. Although the electrostatic surface potential is quite similar in the Par6␣ and p67phoxPB1 domains, the residue responsible for the formation of basic surface is not aligned in the sequence but makes up the equivalent basic surface in the tertiary structure. Thus, the A1-B1 and A2-B2/B2⬘ interactions are regarded as general properties in the PB1-PB1 interaction. The PB1-PB1 interaction is known to be highly specific and exclusive. The comparison of the PB1 complexes of PKC-Par6␣ and p40phox-p67phox revealed differences in the relative orientation between PB1 domains. The C-terminal extension outside of the PB1 domain boundary of p40phox PB1 is involved in the interaction with p67phox PB1, whereas ␣2 in aPKC PB1 extensively interacts with Par6. Therefore, not only diversity in amino acid sequence but also the relative orientation of the PB1 domains endow specificity in PB1-PB1 interaction. Although the PB1 domain adopts a ubiquitin fold, it develops insertions and extensions in the sequence as interaction sites. For example, the PB1 domain develops an insertion of the OPCA motif as canonical interaction sites. On the other hand, the p40phox PB1 domain develops a C-terminal extension as a specific interaction site with the p67phox PB1 domain. The above consideration leads to the hypothesis that the ubiquitin scaffold presents several surfaces as interaction sites for cognate partners, thus expanding the versatility of the protein interaction modes.
Complex Structure of Cell Polarity Regulator Acknowledgments—We thank Nobuyuki Muroya, Kenjiro Hara, and Dr. Satoru Yuzawa for their technical advice and support. We also thank Dr. Masahide Kawamoto, Dr. Hisanobu Sakai, and all the staff of the BL41XU beamline, SPring-8 for assistance with x-ray data collection. REFERENCES 1. Watts, J. L., Etemad-Moghadam, B., Guo, S., Boyd, L., Draper, B. W., Mello, C. C., Priess, J. R., and Kemphues, K. J. (1996) Development 122, 3133–3140 2. Tabuse, Y., Izumi, Y., Piano, F., Kemphues, K. J., Miwa, J., and Ohno, S. (1998) Development 125, 3607–3614 3. Ohno, S. (2001) Curr. Opin. Cell Biol. 13, 641– 648 4. Betschinger, J., Mechtler, K., and Knoblich, J. A. (2003) Nature 422, 326 –330 5. Cai, Y., Yu, F., Lin, S., Chia, W., and Yang, X. (2003) Cell 112, 51– 62 6. Etienne-Manneville, S., and Hall, A. (2003) Nature 421, 753–756 7. Shi, S. H., Jan, L. Y., and Jan, Y. N. (2003) Cell 112, 63–75 8. Izumi, Y., Hirose, T., Tamai, Y., Hirai, S., Nagashima, Y., Fujimoto, T., Tabuse, Y., Kemphues, K. J., and Ohno, S. (1998) J. Cell Biol. 143, 95–106 9. Macara, I. G. (2004) Nat. Rev. Mol. Cell. Biol. 3, 220 –231 10. Petronczki, M., and Knoblich, J.A. (2000) Nat. Cell Biol. 3, 43– 49 11. Wodarz, A., Ramrath, A., Grimm, A., and Knust, E. (2000) J. Cell Biol. 150, 1361–1374 12. Moscat, J., and Diaz-Meco, M. T. (2000) EMBO Rep. 1, 399 – 403 13. Newton, A. C. (2001) Chem. Rev. 101, 2353–2364 14. Tsukita, S., Furuse, M., and Itoh, M. (2001) Nat. Rev. Mol. Cell. Biol. 2, 285–293 15. Matter, K., and Balda, M. S. (2003) Nat. Rev. Mol. Cell. Biol. 4, 225–236 16. Suzuki, A., Yamanaka, T., Hirose, T., Manabe, N., Mizuno, K., Shimizu, M., Akimoto, K., Izumi, Y., Ohnishi, T., and Ohno, S. (2001) J. Cell Biol. 152, 1183–1196 17. Yamanaka, T., Horikoshi, Y., Suzuki, A., Sugiyama, Y., Kitamura, K., Maniwa, R., Nagai, Y., Yamashita, A., Hirose, T., Ishikawa, H., and Ohno, S. (2001) Genes Cells 6, 721–731 18. Noda, Y., Kohjima, M., Izaki, T., Ota, K., Yoshinaga, S., Inagaki, F., Ito, T., and Sumimoto, H. (2003) J. Biol. Chem. 278, 43516 – 43524 19. Schultz, J., Milpetz, F., Bork, P., and Ponting, C. P. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 5857–5864 20. Ito, T., Matsui, Y., Ago, T., Ota, K., and Sumimoto, H. (2001) EMBO J. 20, 3938 –3946 21. Terasawa, H., Noda, Y., Ito, T., Hatanaka, H., Ichikawa, S., Ogura, K., Sumimoto, H., and Inagaki, F. (2001) EMBO J. 20, 3947–3956 22. Ponting, C. P., Ito, T., Moscat, J., Diaz-Meco, M. T., Inagaki, F., and Sumimoto, H. (2002) Trends Biochem. Sci. 27, 10 23. Yoshinaga, S., Kohjima, M., Ogura, K., Yokochi, M., Takeya, R., Ito, T.,
9661
Sumimoto, H., and Inagaki, F. (2003) EMBO J. 22, 4888 – 4897 24. Ponting, C. P. (1998) Protein Sci. 5, 2353–2357 25. Nakamura, R., Sumimoto, H., Mizuki, K., Hata, K., Ago, T., Kitajima, S., Takeshige, K., Sakaki, Y., and Ito, T. (1998) Eur. J. Biochem. 251, 583–589 26. Lamark, T., Perander, M., Outzen, H., Kristiansen, K., Øvervatn, A., Michaelsen, E., Bjørkøy, G., and Johansen, T. (2003) J. Biol. Chem. 278, 34568 –34581 27. Wilson, M. I., Gill, D. J., Perisic, O., Quinn, M. T., and Williams, R. L. (2003) Mol. Cell 12, 39 –50 28. Hirano, Y., Yoshinaga, S., Ogura, K., Yokochi, M., Noda, Y., Sumimoto, H., and Inagaki, F. (2004) J. Biol. Chem. 279, 31883–31890 29. Noda, Y., Takeya, R., Ohno, S., Naito, S., Ito, T., and Sumimoto, H. (2001) Genes Cells 6, 107–119 30. LeMaster, D. M., and Richards, F. M. (1985) Biochemistry 24, 7263–7268 31. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307–326 32. Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., GrosseKunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta Crystallogr. Sect. D Biol. Crystallogr. 54, 905–921 33. Cambillau, C., and Roussel, A. (1997) Turbo-Frodo, version Open GL 1, Universite´ Aix-Marseile II, Marseille, France 34. Nicholls, A., Sharp, K. A., and Honig, B. (1991) Proteins 11, 281–296 35. Koga, H., Terasawa, H., Nunoi, H., Takeshige, K., Inagaki, F., and Sumimoto, H. (1999) J. Biol. Chem. 274, 25051–25060 36. Kohjima, M., Noda, Y., Takeya, R., Saito, N., Takeuchi, K., and Sumimoto, H. (2002) Biochem. Biophys. Res. Commun. 299, 641– 646 37. Lo Conte, L., Chothia, C., and Janin, J. (1999) J. Mol. Biol. 285, 2177–2198 38. Pacold, M. E., Suire, S., Perisic, O., Lara-Gonzalez, S., Davis, C. T., Walker, E. H., Hawkins, P. T., Stephens, L., Eccleston, J. F., and Williams, R. L. (2000) Cell 103, 931–943 39. Alam, S. L., Sun, J., Payne, M., Welch, B. D., Blake, B. K., Davis, D. R., Meyer, H. H., Emr, S. D., Sundquist, W. I. EMBO J. (2004) 23, 1411–1421 40. Sundquist, W. I., Schubert, H. L., Kelly, B. N., Hill, G. C., Holton, J. M., Hill, C. P. (2004) Mol. Cell 13, 783–789 41. Gao, L., Joberty, G., and Macara, I. G. (2002) Curr. Biol. 12, 221–225 42. Joberty, G., Petersen, C., Gao, L., and Macara, I. G. (2000) Nat. Cell Biol. 2, 531–539 43. Lin, D., Edwards, A. S., Fawcett, J. P., Mbamalu, G., Scott, J. D., and Pawson, T. (2000) Nat. Cell Biol. 2, 540 –547 44. Plant, P. J., Fawcett, J. P., Lin, D. C., Holdorf, A. D., Binns, K., Kulkarni, S., and Pawson, T. (2003) Nat. Cell Biol. 5, 301–308 45. Yamanaka, T., Horikoshi, Y., Sugiyama, Y., Ishiyama, C., Suzuki, A., Hirose, T., Iwamatsu, A., Shinohara, A., and Ohno, S. (2003) Curr. Biol. 13, 734 –743 46. Hurd, T. W., Gao, L., Roh, M. H., Macara, I. G., and Margolis, B. (2003) Nat. Cell Biol. 5, 137–142
Vol. 280, No. 10, Issue of March 11, pp. 9653–9661, 2005 Printed in U.S.A.
Structure of a Cell Polarity Regulator, a Complex between Atypical PKC and Par6 PB1 Domains* Received for publication, August 26, 2004, and in revised form, October 27, 2004 Published, JBC Papers in Press, December 7, 2004, DOI 10.1074/jbc.M409823200
Yoshinori Hirano‡, Sosuke Yoshinaga‡, Ryu Takeya§, Nobuo N. Suzuki‡, Masataka Horiuchi‡, Motoyuki Kohjima§, Hideki Sumimoto§¶, and Fuyuhiko Inagaki‡储 From the ‡Department of Structural Biology, Graduate School of Pharmaceutical Sciences, Hokkaido University, N-12, W-6, Kita-ku, Sapporo 060-0812, Japan, §Medical Institute of Bioregulation, Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582, Japan, and ¶CREST, Japan Science and Technology, Kawaguchi 332-0012, Japan
A complex of atypical PKC and Par6 is a common regulator for cell polarity-related processes, which is an essential clue to evolutionary conserved cell polarity regulation. Here, we determined the crystal structure of the complex of PKC and Par6␣ PB1 domains to a resolution of 1.5 Å. Both PB1 domains adopt a ubiquitin fold. PKC PB1 presents an OPR, PC, and AID (OPCA) motif, 28 amino acid residues with acidic and hydrophobic residues, which interacts with the conserved lysine residue of Par6␣ PB1 in a front and back manner. On the interface, several salt bridges are formed including the conserved acidic residues on the OPCA motif of PKC PB1 and the conserved lysine residue on the Par6␣ PB1. Structural comparison of the PKC and Par6␣ PB1 complex with the p40phox and p67phox PB1 domain complex, subunits of neutrophil NADPH oxidase, reveals that the specific interaction is achieved by tilting the interface so that the insertion or extension in the sequence is engaged in the specificity determinant. The PB1 domain develops the interaction surface on the ubiquitin fold to increase the versatility of molecular interaction.
Cell polarity is a fundamental property related to many cellular processes. Recently, a complex of atypical protein kinase C (aPKC)1 and Par6 (aPKC-Par6) has received much attention as a common cell polarity regulator conserved from nematodes to mammals. Par6 was initially identified as one of the products of par genes (par-1 to par-6) encoding essential components of cell division in the early embryo found by lethal genetic screening of Caenorhabditis elegans (1). In addition to Par proteins, a defect in Pkc3 (a homologue of aPKC in C. elegans) also disrupts embryonic polarity (2). aPKC-Par6 is a conserved signaling complex from C. elegans to Homo sapiens (3) and is now regarded as a common regulator for asymmetric
* This work is supported by the National Project on Protein Structural and Functional Analyses from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. 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. The atomic coordinates and structure factors (code 1WMH) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/). 储 To whom correspondence should be addressed: Graduate School of Pharmaceutical Sciences, Hokkaido University, N-12, W-6, Kita-ku, Sapporo 060-0812, Japan. Tel.: 81-11-706-3976; Fax: 81-11-706-4979; E-mail: [email protected]. 1 The abbreviations used are: aPKC, atypical protein kinase C; TJ, tight junction; SeMet, selenomethionine; MDCK, Madin-Darby canine kidney; HA, hemagglutinin; wt, wild type; OPCA, OPR, PC, and AID. This paper is available on line at http://www.jbc.org
cell division (4, 5), directed cell migration (6), axon specification (7), and tight junction (TJ) formation in epithelial cells (8) in various species. Thus, aPKC-Par6 is an essential clue to evolutionary conserved cell polarity regulation (9). aPKC-Par6 is characteristically located at a specific region of cells, such as the anterior cortex of the nematode embryo (1, 2), apical cortex of fly neuroblasts (10, 11), leading edge of migrating cells (6), and TJs in epithelial cells (8), implying that the complex formation is required for aPKC specific localization. The regulation of cellular localization as well as kinase activity of aPKC is essential for phosphorylation of its intrinsic target. Thus, the scaffold proteins that bind to PKCs play a crucial role in the control of the cellular localization of PKCs (12, 13). In particular, TJ formation is a prototype of cell polarity events. TJs encircle cells at the most apical region of the lateral membrane and function as both paracellular and intramembrane diffusion barriers. Moreover TJ formation triggers cell proliferation and differentiation (14, 15). aPKC-Par6 is localized at the TJs and together with Par3 is involved in TJ assembly. Although recruitment of the ternary complex to the TJs and regulation of TJ formation are still elusive, phosphorylation by aPKC seems to be a clue to signal transduction because the kinase activity deficient aPKC abrogates TJ formation (16). Notably, Par6 can regulate aPKC kinase activity in a Cdc42dependent manner (17). The complex formation of aPKC and the scaffold protein Par6 is established through the interaction between PB1 domains existing at the N terminus of both proteins. In TJ formation, the disruption of PB1-PB1 interaction in aPKC-Par6 causes mislocalization of aPKC or Par6 (18). Therefore, PB1PB1 interaction between aPKC and Par6 is related to their localization and has an indispensable role in cell polarity-related processes. Until now, the PB1 domain has been found in more than 200 signaling proteins including p40phox, p67phox, cytosolic factors of NADPH oxidase, Bem1p, Cdc24p, MEK5 and p62/ZIP on the SMART data base (19). Recent studies have demonstrated that PB1 domain functions as a protein binding module through PB1-mediated heterodimerization or homo-oligomerization (20 –23). The PB1 domain is further classified into two types, type I and type II PB1 domains. The PB1-mediated complex is formed through the interaction between type I and type II PB1 domains. Type I PB1 adopts the OPCA motif as a binding site, a remarkable consensus sequence in the PB1 domain with conserved acidic and hydrophobic residues (12, 24, 25). On the other hand, type II PB1 requires a conserved lysine residue in 1 (Fig. 1). Binding assays for PB1-PB1 interaction suggest that the electrostatic interaction between conserved acidic residues on the OPCA motif in type I-PB1 and a conserved basic residue in type II PB1 is responsible for PB1-PB1 interaction.
9653
9654
Complex Structure of Cell Polarity Regulator
FIG. 1. Structure-based sequential alignment of the PB1 domain. The conserved hydrophobic residues are highlighted in red, and hydrophilic residues are highlighted in blue. Secondary structure components of type I, type I and type II, or type II PB1 are derived from the tertiary structure of Cdc24p, PKC, or Par6␣, respectively. The OPCA motif is enclosed in green. The two conserved acidic regions of type I, A1 and A2, are boxed, and the two conserved basic region of type II, B1 and B2/B2⬘, are highlighted in orange and boxed. Residues involved in the interaction between PKC PB1 and Par6␣ are mapped on the sequence. Stars represent residues with side chains involved in an intermolecular salt bridge and/or hydrogen bond. Asterisks represent residues involved in an intermolecular hydrophobic interaction. Filled circles represent residues with the main chain involved in an intermolecular hydrogen bond.
Despite its electrostatic nature, the PB1-PB1 interaction is highly specific and exclusive (20). Interestingly, aPKC binds to several proteins such as Par6, ZIP/p62, and MEK5 through its PB1 domain (18, 26) and mediates distinct signals in a binding protein-dependent manner. To date, whereas structure determination of the PB1 domain has been performed (21, 23, 27, 28), the general and specific properties of the PB1-PB1 interaction mode are yet to be elucidated. Here, we determined the crystal structure of the PB1 domain complex of PKC and Par6␣. Although the electrostatic interaction was assumed to be responsible for complex formation, the structure reveals salt bridge formation in the PB1-PB1 complex. Furthermore, it provides insights into the general and specific properties of PB1-PB1 interaction. We also identified the role of PB1-PB1 interaction in TJ formation in aPKC-Par6. EXPERIMENTAL PROCEDURES
Plasmid Construction—Complementary DNA fragments encoding various lengths of human Par6␣ and PKC were prepared as described previously (18, 29). Mutations leading to the indicated amino acid substitutions were introduced by PCR-mediated site-directed mutagenesis, and the mutated fragments were cloned into the indicated vectors. All the constructs were sequenced to confirm their identities. Sample Preparation for Crystallization—The PKC (16 –99) gene was cloned in pGEX-6P-1 (Amersham Biosciences), and the Par6␣ (14 –95) gene was cloned in pET HP, a modified version of pET-28a (Novagen), converting the thrombin protease recognition site into the PreScission protease recognition site. Both constructs were co-expressed in Escherichia coli strain BL21 (DE3). Because PKC-(16 –99) and Par6␣-(14 – 95) form a complex in E. coli cells, the complex is acquired via purification from bacterial lysate by affinity chromatography using glutathione-Sepharose 4B resin (Amersham Biosciences). After proteins were treated with PreScission protease (Amersham Biosciences), purification using anion exchange chromatography and gel filtration chromatography were performed. On the process of purification, PKC(16 –99) and Par6␣-(14 –95) retained a stable complex. Finally, the complex of PKC PB1 and Par6␣ PB1 was concentrated up to 12–15 mg/ml in a buffer containing 50 mM Tris (pH 8.0), 150 mM NaCl, and 5 mM dithiothreitol, and this solution was used for crystallization. Selenomethionine (SeMet)-substituted protein was expressed in E. coli strain B834 (DE3) with LeMaster medium (30). Purification was performed in the same manner as native protein preparation, and SeMet substitution for Met was confirmed using matrix-assisted laser
desorption ionization time-of-flight mass spectrometry. After concentration to 12–15 mg/ml, SeMet derivatives were used as a crystallization sample. Crystallization—Crystallization screening was performed using a sitting drop vapor diffusion method at 293 K. 100 l of volume of reservoir was applied to wells on crystal clear strips no indent (Hampton Research). Each drop consisted of 0.5 l of protein solution and an equal amount of reservoir solution. Approximately 12–15 mg/ml of the complex of PKC PB1 and Par6␣ PB1 in 50 mM Tris (pH 8.0), 150 mM NaCl, and 1 mM dithiothreitol was used for crystallization. A single crystal was obtained in 3.6 M sodium formate and 0.1 M Tris (pH 8.0) at 293 K by using a microseeding technique. Data Collection and Structure Determination—Prior to x-ray diffraction experiments, crystals were soaked in a cryoprotectant solution (3.6 M sodium formate, 0.1 M Tris (pH 8.0), 20% glycerol) for several seconds and were immediately flash-cooled to 100 K in a nitrogen gas stream. Native diffraction data were collected on an MAR CCD detector using synchrotron radiation at the BL41XU beam line, SPring-8. The camera to crystal distance was set to 150 mm, and a total of 180 images were collected at 1° oscillation angle per image. The diffraction data were processed using the HKL2000 program (31). The crystal belongs to space group P21212 with unit cell parameters a ⫽ 94.150 Å, b ⫽ 38.363 Å, c ⫽ 45.478 Å and a cell volume of 164,261 Å3. The molecular mass of a complex of PKC PB1 and Par6␣ PB1 was ⬃19,427 Da, thus, the Matthews coefficient (VM) was 2.1 Å3/Da and the solvent content was 42%, assuming one molecule exist in an asymmetric unit. Multiwavelength anomalous dispersion data sets were collected using a single crystal of SeMet-substituted protein at three wavelengths at the BL41XU beam line, SPring-8 as well as native data sets. The x-ray diffraction statistics are summarized in Table I. Determination of the selenium site in SeMet-substituted proteins, phasing, and refinement were performed by crystallography and NMR system software (32). Modeling was performed by Turbo-Frodo (33). Finally, refinement was performed up to 1.5 Å, and the statistics are summarized in Table I. The coordinates and structure factor of a PB1 domain complex of PKC and Par6␣ have been deposited in the Protein Data Bank (accession code 1WMH). The molecular surface model was generated by the GRASP program (34). All figures of the structural model except the model of electrostatic surface potential were created by Pymol software. Yeast Two-hybrid Experiments—Yeast HF7c cells containing a HIS3 reporter gene were co-transformed with a pair of pGBT9 (Clontech) encoding the wild-type or mutant PKC and pGADGH (Clontech) encoding the wild-type or mutant Par6␣ as described previously (35). Following the selection for Trp⫹ and Leu⫹ phenotype, the transfor-
Complex Structure of Cell Polarity Regulator
9655
TABLE I Crystallographic analysis statistics Data collection
Space group Unit cell (Å)
Wavelength (Å) Resolution range (Å) Observed/unique reflections Completeness (%) Rmergea 具I/典 Refinement statistics Resolution range (Å) Rwork (%) Rfree (%)b r.m.s.d. from ideal valuesc Bonds (Å) Angles (deg) Model quality (Ramachandran plot) (%) Residues in most favored regions Residues in additionally allowed regions Residues in generously allowed regions Residues in disallowed regions
SeMet (Peak)
a ⫽ 94.663 b ⫽ 38.407 c ⫽ 45.570 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9793 50.0–1.9 97,678/26,471 99.3 0.044 13.5
SeMet (Edge)
a ⫽ 94.648 b ⫽ 38.393 c ⫽ 45.546 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9795 50.0–1.9 97,467/26,435 99.2 0.044 13.8
P21212
SeMet (Remote)
Native
a ⫽ 94.574 b ⫽ 38.382 c ⫽ 45.537 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9843 50.0–1.9 96,296/26,133 98.7 0.045 13.4
a ⫽ 94.150 b ⫽ 38.363 c ⫽ 45.478 ␣ ⫽  ⫽ ␥ ⫽ 90° 0.9791 50.0–1.5 17,9821/50,330 98.5 0.054 16.5
50.0–1.5 21.6 22.4 0.004 1.20 93.8 5.5 0.0 0.7
Rmerge ⫽ ⌺hkl⌺i兩I(hkl)i ⫺ 具I(hkl)典兩/⌺hklI(hkl), where I(hkl)i indicates the observed intensity, and 具I(hkl)典 is the mean intensity. Rfree was calculated on a random 9.9% reflections of the data. c r.m.s.d., root mean square deviation. a b
mants were tested for their ability to grow on plates lacking histidine supplemented with 2 mM 3-aminotriazole to suppress background growth according to the manufacturer’s recommendation (Clontech). Formation of Epithelial Tight Junction—To investigate the effect of PKC or Par6␣ on formation of TJs, we performed a calcium switch method using canine kidney epithelial MDCK cells. Briefly, MDCK cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum (36) and transfected with pEF-BOS-hemagglutinin, a vector for expression as a HA-tagged protein, that encodes the indicated form of PKC or Par6␣ (36). After incubation for 24 h in Dulbecco’s modified Eagle’s medium with 10% fetal calf serum, the transfected cells were seeded on 12-mm round glass coverslips and grown in normal calcium medium to form confluent monolayers. The monolayer cells were cultured for 60 min in the calcium-free medium S-MEM (Invitrogen) and then incubated for 2 h in the calcium-containing medium Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum. After three washes with phosphate-buffered saline, cells were fixed for 15 min with 3.7% formaldehyde and then permeabilized for 15 min in phosphate-buffered saline containing 0.5% Triton X-100. After blocking for 60 min with phosphate-buffered saline containing 3% bovine serum albumin, the samples were initially incubated with anti-HA (Y11) rabbit polyclonal antibodies (Santa Cruz Biotechnology) and anti-ZO-1 mouse monoclonal antibody (36) and subsequently probed with Alexa FluorTM 594-labeled goat anti-rabbit IgG antibodies (Molecular Probes) and Alexa FluorTM 488-labeled goat anti-mouse IgG antibodies (Molecular Probes). Images were visualized with an LSM 510 confocal laser scanning microscope (Carl Zeiss). RESULTS AND DISCUSSION
Overall Structure of the PB1 Complex of PKC and Par6␣— The crystal structure of the PB1 complex of PKC and Par6␣ has been determined to a resolution of 1.5 Å by a multiwavelength anomalous dispersion technique using an SeMet-substituted protein as described under “Experimental Procedures.” The crystal belongs to the space group P21212, and one complex exists per asymmetric unit. Of a total of 175 residues, 5 Nterminal residues of PKC PB1 and 4 N-terminal residues of Par6␣ PB1 that are derived from cloning artifacts and Cys99 of PKC PB1, a C-terminal residue of the construct, are disordered. In Par6␣ PB1, Pro87 adopts a cis conformation. A summary of the collected data and the structural statistics is given in Table I. The overall structure of the complex is shown in Fig. 2A. Both PB1 domains comprise two ␣ helices and a mixed -sheet which consists of five  strands that belong to the ubiquitin
superfamily. The OPCA motif region of PKC (Leu56-Leu83) has a ␣ fold, and the  turn is categorized as a type I ⫹ G1 -bulge, characteristic of type I PB1 domains (23). The PB1 domains of PKC and Par6␣ are remarkably similar as shown in Fig. 2B (root mean square deviation ⫽ 0.86 Å for backbone atoms). Because the solution structure of PKC (free state) was determined by NMR (28), the free and the complex states of PKC PB1 can be compared (Fig. 2C). Both of the overall structures are well superimposed (root mean square deviation ⫽ 0.96 Å for backbone atoms). We performed superimposition between free and complex state structures, using not only over all structure but also more limited region such as OPCA motif. As the result, free and complex state structures are uniformly superimposed, and notable conformational exchange is not observed except for the loop region that is flexible in free structure. Interaction Surface of the PKC and Par6␣ PB1 Complex— Despite the overall structural similarity between the PKC and Par6␣ PB1 domains, they formed an asymmetric heterodimer using a different surface from each protein (Fig. 2A). The interaction surface of PKC PB1 is the OPCA motif corresponding to 3, 4, and ␣2, whereas that of Par6␣ PB1 is composed of 1, 2, the C-terminal region of ␣1, and the N-terminal region of 5. Thus, PKC serves as a type I PB1 domain and Par6␣ serves as a type II PB1 domain, respectively, as was suggested by the previous mutational analyses (18, 27). In PKC PB1, the OPCA motif is responsible for the interaction with Par6␣ PB1, but no other regions are involved in the interaction. The interaction surface of PKC PB1 presents two acidic regions, A1 (Asp63, Glu65, and Asp67) and A2 (Glu76), whereas that of Par6␣ is comprised of two basic regions, B1 (Lys19, Arg28 and Arg89) and B2⬘ (Arg27) (Fig. 3A). In Par6␣ PB1, Arg27 and Arg28, two consecutive basic residues on 2, direct their side chains to the opposite side so that Par6␣ PB1 presents two distinct basic regions, B1 and B2⬘, at both sides of 2 (Fig. 3A, right). The complementarity of the surface potential between A1-B1 and A2-B2⬘ supports the notion that electrostatic interaction plays a major role in PB1 heterodimer formation. In addition to these interactions, Arg82 (PKC)-Glu17 (Par6␣) and a hydrophobic interaction, including Leu75 of PKC and Phe29,
9656
Complex Structure of Cell Polarity Regulator
FIG. 2. Structural properties of a PB1 complex of PKC-Par6␣. A, a ribbon representation of a PB1 complex of PKC-Par6␣. PKC PB1 is indicated in pink, and Par6␣ PB1 is indicated in green. The direct interaction surfaces of PKC PB1 and Par6␣ PB1 are indicated in red and blue, respectively. B, a superposition of PKC PB1 and Par6␣ PB1 in the PB1 complex. The color of each PB1 domain is the same as in A. C, a superposition of the free state (solution structure) and complex state (crystal structure) of PKC PB1. The free state is indicated in cyan, and the complex state is indicated in pink.
Leu45, and Val49 of Par6␣, also contribute to the intermolecular interaction (Fig. 3A). Description of the Interface in the PKC-Par6␣ PB1 Domain Complex—A detailed inspection of the interface gives the characteristic interaction mode in the PKC-Par6␣ complex (Fig. 3, A and B). In PKC PB1, A1 is located on the  turn in the OPCA motif formed by the conserved acidic residues, Asp63, Glu65, and Asp67. As the  turn adopts a type I ⫹ G1  bulge, these acidic residues orient to the same side to form an acidic surface. B1 of Par6␣ consists of Lys19 in 1, Arg28 in 2, and Arg89 in 5. The complementarily charged side chains of PKC and Par6␣ form salt bridges as shown in Fig. 2B. Notably, Lys19 of Par6␣ PB1 forms salt bridges with all of the three conserved acidic residues. The interaction mediated by Lys19 is consistent with its strict conservation in type II PB1. Arg28 and Arg89 of Par6␣ interact with Glu65 and Asp63 of PKC, which fix these carboxyl groups to assist salt bridge formation with Lys19 as well as to compensate for charges on the salt bridge network. Thus, A1-B1 is a major interaction site in the complex formation (Fig. 3B). In the complex of p40phox-p67phox PB1 domains, the corresponding acidic residues on the OPCA motif in p40phox (Asp289, Glu291, and Asp293) interact with the conserved lysine residue in p67phox (Lys355) in a manner similar to that in the PKC-Par6␣ complex (27). The importance of this interaction is evident considering that the type I PB1 domain contains an arginine residue instead of a lysine residue. To form tight salt bridges with the three acidic residues, the lysine residue is essential in this position (Fig. 1). In addition to the A1-B1 interaction, two other interactions, including A2-B2⬘ and Arg82-Glu17, are to be noted (Fig. 3B). A2 is located on ␣2 of PKC PB1 to interact with B2⬘ of Par6␣ PB1. Glu76 located on ␣2 of PKC PB1 forms a salt bridge with Arg27 on 2 of Par6␣ PB1, which also forms a hydrogen bond with Ser72 of PKC. Glu17 of Par6␣ and Arg82 of PKC are located close to each other to form a salt bridge. Furthermore, intramolecular salt bridge formation between Glu79 and Arg82 of PKC and between Glu17 and Arg28 of Par6␣ is observed (Fig. 3B). Besides these salt bridges, there are hydrophobic interactions
involving Leu75 on ␣2 of PKC and Leu45 and Val49 on 3 and Phe29 on 2 of Par6␣ (Fig. 3A). In the interface, not only intermolecular salt bridges but also a number of intramolecular salt bridges and hydrogen bonds are formed. These interactions are observed throughout the interface and form a network (Fig. 3C). Although the intramolecular salt bridges do not directly contribute to the binding, the salt bridge network may affect their side chain coordination to assist in the formation of a stable complex. There are a relatively small number of hydrophobic interactions and watermediated hydrogen bonds in the interface. The surface area of the PKC-Par6␣ complex is ⬃1141 Å2, which is appreciably smaller than those of other protein complexes (37). Tight interaction between PB1 domains is mainly caused by the salt bridge and hydrogen bond formations that amounted to 10 in the complex and much less to hydrophobic interactions. Therefore, the interface of the PKC-Par6␣ PB1 complex is very compact but sufficient to form a stable protein-protein complex. Mutational Analyses of the Residues Responsible for Complex Formation—The mutational analyses using the yeast two-hybrid method were performed to confirm the molecular interaction. The conserved acidic residue (Asp63) in PKC PB1 and the conserved lysine residue in Par6␣ (Lys19) were mutated to alanine, which completely abrogated the binding with each counterpart. The K20A mutation in PKC PB1 and the D63A mutation in Par6␣ showed similar binding affinity to the wild type (Fig. 4). These results are consistent with the previous mutational experiments using a pull-down binding assay (18, 27) and confirm that PKC PB1 serves as type I and Par6␣ PB1 as type II (Fig. 1). Furthermore, we designed several mutants based on the present structure of the PKC-Par6␣ PB1 complex. First, we made mutations at residues R28A, R89A, and R28A/R89A in Par6␣ to investigate the contribution of these residues to the A1-B1 interaction. Although R28A and R89A slightly diminished the interaction, the R28A/R89A double mutant completely impaired the interaction with PKC PB1, consistent with the present structure (Fig. 3, B and C). Then we further investigated the two interaction sites, A2-B2⬘ (Ser72/
Complex Structure of Cell Polarity Regulator
9657
FIG. 3. Detailed molecular interaction between PKC PB1 and Par6␣ PB1. A, ribbon and stick diagrams of a PB1 complex of PKC-Par6␣ represented in an open book style. The two acidic regions of PKC PB1 and the two basic regions of Par6␣ PB1 in the interaction surface are encircled in red and blue, respectively. B, PKC-Par6␣ interface with the corresponding final 2兩Fo兩 ⫺ 兩Fc兩 electron density at 2.0 is shown. A1, A2, B1, and B2⬘ indicate the same as in A. Orange dotted lines indicate intermolecular hydrogen bonds, and blue dotted lines indicate intramolecular hydrogen bonds. C, a stick model of manifold salt bridge network in the interaction surface. Blue spheres represent water molecules. Orange dotted lines indicate intermolecular hydrogen bonds, and blue dotted lines indicate intramolecular hydrogen bonds or water molecule-mediated hydrogen bonds.
Glu76-Arg27) and Arg82-Glu17. Although E76A or R82A mutations in PKC had a slight effect on binding to Par6␣, the E76A/R82A double mutant completely abolished binding to Par6␣ PB1. Thus, the Glu76 and Arg82 of PKC PB1 were confirmed to be involved in the complex formation. The mutational studies by the yeast two-hybrid method (Fig. 4) support the interaction mode revealed herein in the structure of the PKC-Par6␣ PB1 domains (Fig. 3). General Properties of PB1-PB1 Interaction—The PB1 domains were classified into two types, the type I and the type II PB1 domains. Type I uses the OPCA motif, whereas type II uses the conserved lysine residue located on the opposite surface to the OPCA motif for heterodimerization. There is another type of PB1 domain, classified as type I and type II, which contains both the OPCA motif and the conserved lysine residue (23). For example, aPKC PB1 binds to Par6 PB1 as a type I PB1 domain and to MEK5 PB1 as a type II PB1 domain. Now, we aligned the PB1 domains on the structural basis, taking the previous studies into consideration (Fig. 1) (21, 23, 27, 28). We show the interfaces of both type I and type II PB1 domains in an open book style (Fig. 5). The interface in the type I PB1
domain is strictly located on the OPCA motif. Two acidic regions, A1 and A2, are presented on the OPCA motif of the type I PB1 domains that interact with the type II PB1 domains in an electrostatic manner. In contrast to the type I PB1 domain, the general properties of the type II PB1 domain are less understood because of low sequence homology except for the strictly conserved lysine residue involved in B1 that interacts with A1 of the type I PB1 domain. Our present study reveals the generality of the PB1-PB1 interaction especially in the type II PB1 domain. In the PB1 complex of PKC and Par6␣, A1 of PKC interacts with B1, including Lys19, Arg28, and Arg89 of Par6␣. Notably, as the Arg89 of Par6␣ is positioned in 5 as well as Lys19, the conserved lysine residue in 1 interacts with A1. Arg89 corresponds to a highly conserved basic residue in 5 of the type II PB1 domain (Fig. 1). Thus, the conserved basic residue in 5 of the type II PB1 domain is suggested to generally interact with A1 of the type I PB1 domain. In contrast to the A1-B1 interaction, the A2-mediated interaction is quite complicated. In the structure of the p40phoxp67phox PB1 heterodimer, Lys382 in ␣1 of p67phox PB1 interacts
9658
Complex Structure of Cell Polarity Regulator
FIG. 4. PB1 domain-mediated interaction between PKC and Par6␣. A, two-hybrid interaction of Par6␣ with the wild-type or mutant PB1 domain of PKC. Yeast HF7c cells were co-transformed with a pair of pGBT9 encoding wild-type PKC PB1 (wt) or PKC PB1 mutants, E76A, R82A, E76A/R82A, K20A, or D63A, and pGADGH encoding the full-length wild-type Par6␣-F (wt). Following the selection for the Trp⫹ and Leu⫹ phenotype, the histidine-dependent (right) and -independent (left) growth of the transformant was tested as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results. B, two-hybrid interaction of PKC with the wild-type or mutant PB1 domain of Par6␣. HF7c cells were co-transformed with a pair of pGBT9 encoding the full-length wild-type PKC-F (wt) and pGADGH encoding wild-type Par6␣ PB1 (wt) or mutants, R28A, R89A, R28A/R89A, K19A, or D63A. Following the selection for the Trp⫹ and Leu⫹ phenotype, the histidine-dependent (right) and -independent (left) growth of the transformant was tested as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results.
with A2 of p40phox PB1 (27). This basic residue in ␣1 (referred to basic region 2, B2) is also conserved in Bem1p PB1 (corresponding to Arg510), and its contribution to the formation of the PB1 domain complex is supported by the tertiary structure and mutational experiments (21, 23). However, in other type II PB1 domains, the residue involved in B2 is not conserved on the sequential alignment (Fig. 1). Accordingly, the A2-B2 interaction appears to be specific for p40phox-p67phox and Cdc24pBem1p PB1. However, surprisingly, Arg27, a basic residue in 2 of Par6␣, interacts with Glu76 in A2 of PKC. Arg27 presents its side chain to a similar position as Lys382 in p67phox PB1 in the tertiary structure regardless of their different position in the sequence (Fig. 1). So Arg27 of Par6␣ and Lys382 of p40phox give similar electrostatic potential at an equivalent position and form a conserved basic surface (Fig. 5). This basic residue in 2 (basic region 2⬘, B2⬘) is consistently conserved in type II PB1 domains except for p67phox and Bem1p. Thus, the A2-B2/B2⬘ interaction is considered to be conserved as a general property of the PB1-PB1 interaction. Specificity Determinants for PKC-Par6␣ and p40phoxp67phox Complex Formation—Although the formation of salt bridges between A1-B1 and A2-B2/B2⬘ has been identified as a general property of PB1-PB1 interaction, the PB1-PB1 interaction is known to be highly specific and exclusive. As a unique feature of the p40phox-p67phox PB1 heterodimer, the C-terminal non-conserved region of p40phox PB1 was shown by mutational analyses to be responsible for stable complex formation with p67phox PB1 (25). Therefore, in addition to the typical A1-B1 and A2-B2/B2⬘ interactions, the C-terminal extension of p40phox located outside of the canonical PB1 domain boundary provides a specific interface essential for the formation of the p40phox-p67phox PB1 complex. However, the PB1 domain of Cdc24p exists at the C terminus of the protein, and therefore involvement of the C-terminal extension outside of the canonical PB1 domain is excluded from the complex formation with
FIG. 5. General interaction surface of the PB1 domain. Electrostatic surface potentials of the interaction surfaces of type I and type II PB1 domains are presented. The conserved acidic regions of the type I PB1 domain, A1 and A2, are encircled in orange, and the conserved basic regions of the type II PB1 domain, B1and B2/B2⬘, are encircled in blue.
Bem1p PB1 (Fig. 1). Additionally, PKC does not require an extension to bind to ZIP/p62 (28). According to these observations, the interaction mode between the PKC and Par6␣ PB1 domains seems to be distinct from that between p40phox and p67phox PB1 domains. Now, because we know the atomic detail of the two PB1 complexes, PKC-Par6␣ and p40phox-p67phox, we can discuss the molecular interactions that give specificity to each complex. Superposition of the type II PB1 domain of each complex revealed that although the general interaction sites, A1-B1 and A2-B2/B2⬘, are conserved, the interaction surfaces were tilted downward for PKC-Par6␣ and upward for p40phox-p67phox, resulting in the formation of a distinct interface in the PB1 domain complexes. In the PKC-Par6␣ PB1 domains, ␣2 of PKC PB1 is protruded to extensively interact with the residues in 1 and 2 of Par6␣, whereas in the complex of p40phoxp67phoxPB1 domains, the C-terminal extension of p40phox PB1 is engaged in the interaction with p67phox PB1 (Fig. 6A). The molecular surfaces for each complex are highlighted in Fig. 6B. In the PKC-Par6␣ PB1 complex, Leu75 of PKC, located in the N-terminal region of ␣2, interacts with a hydrophobic patch comprised of Phe29, Leu45, and Val49 of Par6␣ (Fig. 6B). In addition, Arg82 of PKC at the C-terminal region of ␣2 forms a salt bridge with Glu17 of Par6␣. Thus, ␣2 of PKC PB1 interacts extensively with Par6␣ PB1.
Complex Structure of Cell Polarity Regulator
9659
FIG. 6. Distinct interaction mode between PKC-Par6␣ and p40phoxp67phox. A, overlay of the PB1 complexes of PKC-Par6␣ and p40phox-p67phox, where the type II PB1 domains of the PB1 complexes (Par6␣ and p67phox, respectively) were overlapped to the maximum. Each of the PB1 domains of PKC, Par6␣, p40phox, and p67phox is indicated in pink, green, orange, or blue, respectively. B, interaction surfaces of PB1 heterodimers of PKC-Par6␣ and p40phox-p67phox are presented in an open book style. The regions involved in ␣2 of type I PB1-mediated interaction are highlighted in red, and the regions involved in the C-terminal extension mediated interaction are highlighted in blue. Other interaction sites are shown in yellow.
In the complex of p40phox-p67phox PB1 domains, the C-terminal extension of p40phox PB1 is engaged in the interaction with the 1/2 loop, 3/4 loop, and the C-terminal region of 5 in p67phox PB1 mainly through hydrophobic interaction and hydrogen bond formation (27). However, in contrast to the complex of PKC-Par6␣ PB1 domains, ␣2 of p40phox PB1 only slightly interacts with p67phox PB1 because of the upward tilt (Fig. 6B). Thus, the change in the tilt angle between the two PB1 complexes is sufficient to endow specificity in the complex of the PB1 domains. Such specific interaction caused by the tilt in the interface was shown in the complex of Ras binding domain and Ras or Rap1A. Ras or Rap1A interacts with a specific effector such as Raf, RalGDS, or phosphatidylinositol 3-kinase ␥ by adjusting the relative orientation of each domain (38). Beside the case of Ras or Rap1A-Ras binding domain interaction, ubiquitin-mediated interactions with Npl4 zinc finger (39) and ubiquitin E2 variant domain (40) is a good exam-
ple. Ubiquitin uses a hydrophobic patch on a  sheet as a common interface with them, whereas ubiquitin confers a specific interface with each domain by tilting its orientation. The specific recognition in domain-domain interaction achieved by tilting the orientation may be a general strategy developed in the process of evolution. Role of the PB1-PB1 Interaction between PKC and Par6␣ in Tight Junction Formation—MDCK cells normally form TJs when aPKC (wt) is overexpressed, whereas TJ assembly in MDCK cells is disturbed when a kinase-deficient mutant of aPKC is overexpressed (16). Thus, the kinase activity of aPKC is known to be important for epithelial polarity. However, the importance of the PB1-PB1 interaction between aPKC and Par6 in TJ assembly is still unclear. Accordingly, we investigated the role of PB1-PB1 interaction in the control of TJ assembly. In accordance with previous reports, MDCK cells normally formed TJs when PKC (wt) was overexpressed (Fig.
9660
Complex Structure of Cell Polarity Regulator
FIG. 7. Role of the PB1 domain-mediated interaction between PKC and Par6␣ in TJ formation of MDCK cells. A–G, MDCK cells were transfected with pEF-BOS-HA encoding PKC (wt), PKC (E76A/R82A), or PKC (D63A). The transfected cells were subjected to the calcium switch method. The sample was then fixed and doubly stained with the anti-HA (A–C) and anti-ZO-1 (D–F) antibodies, and the images were visualized as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results. G, the indicated percentage of MDCK cells ectopically expressing the wild-type or mutant PKC showed incomplete peripheral ZO-1 staining. Values shown in G are means (⫾S.D.) of three independent experiments. H–N, MDCK cells were transfected with pEF-BOS-HA encoding Par6␣ (wt), Par6␣ (K19A), or Par6␣ (R28A/R89A). The transfected cells were subjected to the calcium switch method. The sample was then fixed and doubly stained with the anti-HA (H–J) and anti-ZO-1 (K–M) antibodies, and the images were visualized as described under “Experimental Procedures.” The experiments were repeated more than four times with similar results. N, the indicated percentage of MDCK cells ectopically expressing the wild-type or mutant Par6␣ showed incomplete peripheral ZO-1 staining. Values shown in N are means (⫾S.D.) of three independent experiments.
7). In contrast, overexpression of PKC mutants (PKC D63A and PKC E76A/R82A), which do not bind with Par6, disturbed normal localization of ZO-1, a tight junction marker, implying that TJ assembly was inhibited (Fig. 7). Thus, aPKC requires the interaction with Par6 via the PB1 domain as well as kinase activity for the control of TJ assembly. On the other hand, overexpression of Par6 is known to inhibit TJ assembly (41). Par6 contains a CRIB motif and a PDZ domain besides the PB1 domain. Par6 binds with the activated form of Cdc42 via the CRIB and PDZ (42, 43). Par6 also binds with Par3 (42, 43), Lgl (5, 44, 45), and Pals1 (46) via the PDZ domain, suggesting that Par6 is a scaffold of a multiprotein complex for controlling cell polarity. The overexpression of Par6 may upset the balance of the multiprotein complex to bring about abnormal epithelial cell polarity. We evaluated the effect of Par6␣ mutants that do not bind with aPKC to clarify the role of the Par6␣ PB1 domain. When Par6␣ (wt) was overexpressed in MDCK cells, TJ assembly was inhibited, which is consistent with the results of the previous study (Fig. 7) (41). However, when Par6␣ K19A and R27A/R28A mutants that do not interact with aPKC were overexpressed in MDCK cells, TJ assembly was little inhibited (Fig. 7). Thus, the role of Par6 in the TJ assembly is to interact with aPKC via the PB1 domain. These results strongly suggest that the PB1-PB1 interaction between aPKC and Par6 plays an important role in the regulation of epithelial cell polarity. CONCLUSION
The crystal structure of the PB1 heterodimer of PKC and Par6␣ revealed that the salt bridge formation (A1-B1) between the conserved acidic residues in the OPCA motif of PKC PB1 and the conserved lysine residue in Par6␣ PB1 is mainly re-
sponsible for complex formation. Interestingly, the ammonium group of the conserved lysine residue is tightly held with the three conserved acidic residues in the OPCA motif of PKC PB1. Such salt bridge formation is also observed in the p40phoxp67phox PB1 complex and is regarded as characteristic in PB1 heterodimer formation. In addition, the formation of another salt bridge (A2-B2/B2⬘) was observed. Although the electrostatic surface potential is quite similar in the Par6␣ and p67phoxPB1 domains, the residue responsible for the formation of basic surface is not aligned in the sequence but makes up the equivalent basic surface in the tertiary structure. Thus, the A1-B1 and A2-B2/B2⬘ interactions are regarded as general properties in the PB1-PB1 interaction. The PB1-PB1 interaction is known to be highly specific and exclusive. The comparison of the PB1 complexes of PKC-Par6␣ and p40phox-p67phox revealed differences in the relative orientation between PB1 domains. The C-terminal extension outside of the PB1 domain boundary of p40phox PB1 is involved in the interaction with p67phox PB1, whereas ␣2 in aPKC PB1 extensively interacts with Par6. Therefore, not only diversity in amino acid sequence but also the relative orientation of the PB1 domains endow specificity in PB1-PB1 interaction. Although the PB1 domain adopts a ubiquitin fold, it develops insertions and extensions in the sequence as interaction sites. For example, the PB1 domain develops an insertion of the OPCA motif as canonical interaction sites. On the other hand, the p40phox PB1 domain develops a C-terminal extension as a specific interaction site with the p67phox PB1 domain. The above consideration leads to the hypothesis that the ubiquitin scaffold presents several surfaces as interaction sites for cognate partners, thus expanding the versatility of the protein interaction modes.
Complex Structure of Cell Polarity Regulator Acknowledgments—We thank Nobuyuki Muroya, Kenjiro Hara, and Dr. Satoru Yuzawa for their technical advice and support. We also thank Dr. Masahide Kawamoto, Dr. Hisanobu Sakai, and all the staff of the BL41XU beamline, SPring-8 for assistance with x-ray data collection. REFERENCES 1. Watts, J. L., Etemad-Moghadam, B., Guo, S., Boyd, L., Draper, B. W., Mello, C. C., Priess, J. R., and Kemphues, K. J. (1996) Development 122, 3133–3140 2. Tabuse, Y., Izumi, Y., Piano, F., Kemphues, K. J., Miwa, J., and Ohno, S. (1998) Development 125, 3607–3614 3. Ohno, S. (2001) Curr. Opin. Cell Biol. 13, 641– 648 4. Betschinger, J., Mechtler, K., and Knoblich, J. A. (2003) Nature 422, 326 –330 5. Cai, Y., Yu, F., Lin, S., Chia, W., and Yang, X. (2003) Cell 112, 51– 62 6. Etienne-Manneville, S., and Hall, A. (2003) Nature 421, 753–756 7. Shi, S. H., Jan, L. Y., and Jan, Y. N. (2003) Cell 112, 63–75 8. Izumi, Y., Hirose, T., Tamai, Y., Hirai, S., Nagashima, Y., Fujimoto, T., Tabuse, Y., Kemphues, K. J., and Ohno, S. (1998) J. Cell Biol. 143, 95–106 9. Macara, I. G. (2004) Nat. Rev. Mol. Cell. Biol. 3, 220 –231 10. Petronczki, M., and Knoblich, J.A. (2000) Nat. Cell Biol. 3, 43– 49 11. Wodarz, A., Ramrath, A., Grimm, A., and Knust, E. (2000) J. Cell Biol. 150, 1361–1374 12. Moscat, J., and Diaz-Meco, M. T. (2000) EMBO Rep. 1, 399 – 403 13. Newton, A. C. (2001) Chem. Rev. 101, 2353–2364 14. Tsukita, S., Furuse, M., and Itoh, M. (2001) Nat. Rev. Mol. Cell. Biol. 2, 285–293 15. Matter, K., and Balda, M. S. (2003) Nat. Rev. Mol. Cell. Biol. 4, 225–236 16. Suzuki, A., Yamanaka, T., Hirose, T., Manabe, N., Mizuno, K., Shimizu, M., Akimoto, K., Izumi, Y., Ohnishi, T., and Ohno, S. (2001) J. Cell Biol. 152, 1183–1196 17. Yamanaka, T., Horikoshi, Y., Suzuki, A., Sugiyama, Y., Kitamura, K., Maniwa, R., Nagai, Y., Yamashita, A., Hirose, T., Ishikawa, H., and Ohno, S. (2001) Genes Cells 6, 721–731 18. Noda, Y., Kohjima, M., Izaki, T., Ota, K., Yoshinaga, S., Inagaki, F., Ito, T., and Sumimoto, H. (2003) J. Biol. Chem. 278, 43516 – 43524 19. Schultz, J., Milpetz, F., Bork, P., and Ponting, C. P. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 5857–5864 20. Ito, T., Matsui, Y., Ago, T., Ota, K., and Sumimoto, H. (2001) EMBO J. 20, 3938 –3946 21. Terasawa, H., Noda, Y., Ito, T., Hatanaka, H., Ichikawa, S., Ogura, K., Sumimoto, H., and Inagaki, F. (2001) EMBO J. 20, 3947–3956 22. Ponting, C. P., Ito, T., Moscat, J., Diaz-Meco, M. T., Inagaki, F., and Sumimoto, H. (2002) Trends Biochem. Sci. 27, 10 23. Yoshinaga, S., Kohjima, M., Ogura, K., Yokochi, M., Takeya, R., Ito, T.,
9661
Sumimoto, H., and Inagaki, F. (2003) EMBO J. 22, 4888 – 4897 24. Ponting, C. P. (1998) Protein Sci. 5, 2353–2357 25. Nakamura, R., Sumimoto, H., Mizuki, K., Hata, K., Ago, T., Kitajima, S., Takeshige, K., Sakaki, Y., and Ito, T. (1998) Eur. J. Biochem. 251, 583–589 26. Lamark, T., Perander, M., Outzen, H., Kristiansen, K., Øvervatn, A., Michaelsen, E., Bjørkøy, G., and Johansen, T. (2003) J. Biol. Chem. 278, 34568 –34581 27. Wilson, M. I., Gill, D. J., Perisic, O., Quinn, M. T., and Williams, R. L. (2003) Mol. Cell 12, 39 –50 28. Hirano, Y., Yoshinaga, S., Ogura, K., Yokochi, M., Noda, Y., Sumimoto, H., and Inagaki, F. (2004) J. Biol. Chem. 279, 31883–31890 29. Noda, Y., Takeya, R., Ohno, S., Naito, S., Ito, T., and Sumimoto, H. (2001) Genes Cells 6, 107–119 30. LeMaster, D. M., and Richards, F. M. (1985) Biochemistry 24, 7263–7268 31. Otwinowski, Z., and Minor, W. (1997) Methods Enzymol. 276, 307–326 32. Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., GrosseKunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T., and Warren, G. L. (1998) Acta Crystallogr. Sect. D Biol. Crystallogr. 54, 905–921 33. Cambillau, C., and Roussel, A. (1997) Turbo-Frodo, version Open GL 1, Universite´ Aix-Marseile II, Marseille, France 34. Nicholls, A., Sharp, K. A., and Honig, B. (1991) Proteins 11, 281–296 35. Koga, H., Terasawa, H., Nunoi, H., Takeshige, K., Inagaki, F., and Sumimoto, H. (1999) J. Biol. Chem. 274, 25051–25060 36. Kohjima, M., Noda, Y., Takeya, R., Saito, N., Takeuchi, K., and Sumimoto, H. (2002) Biochem. Biophys. Res. Commun. 299, 641– 646 37. Lo Conte, L., Chothia, C., and Janin, J. (1999) J. Mol. Biol. 285, 2177–2198 38. Pacold, M. E., Suire, S., Perisic, O., Lara-Gonzalez, S., Davis, C. T., Walker, E. H., Hawkins, P. T., Stephens, L., Eccleston, J. F., and Williams, R. L. (2000) Cell 103, 931–943 39. Alam, S. L., Sun, J., Payne, M., Welch, B. D., Blake, B. K., Davis, D. R., Meyer, H. H., Emr, S. D., Sundquist, W. I. EMBO J. (2004) 23, 1411–1421 40. Sundquist, W. I., Schubert, H. L., Kelly, B. N., Hill, G. C., Holton, J. M., Hill, C. P. (2004) Mol. Cell 13, 783–789 41. Gao, L., Joberty, G., and Macara, I. G. (2002) Curr. Biol. 12, 221–225 42. Joberty, G., Petersen, C., Gao, L., and Macara, I. G. (2000) Nat. Cell Biol. 2, 531–539 43. Lin, D., Edwards, A. S., Fawcett, J. P., Mbamalu, G., Scott, J. D., and Pawson, T. (2000) Nat. Cell Biol. 2, 540 –547 44. Plant, P. J., Fawcett, J. P., Lin, D. C., Holdorf, A. D., Binns, K., Kulkarni, S., and Pawson, T. (2003) Nat. Cell Biol. 5, 301–308 45. Yamanaka, T., Horikoshi, Y., Sugiyama, Y., Ishiyama, C., Suzuki, A., Hirose, T., Iwamatsu, A., Shinohara, A., and Ohno, S. (2003) Curr. Biol. 13, 734 –743 46. Hurd, T. W., Gao, L., Roh, M. H., Macara, I. G., and Margolis, B. (2003) Nat. Cell Biol. 5, 137–142
