Identification of a PDZ protein, PIST, as a binding partner for Rho effector Rhotekin: biochemical and cell-biological characterization of Rhotekin–PIST interaction

Abstract

Among various effector proteins for the small GTPase Rho, the function(s) of Rhotekin is (are) almost unknown. We have identified PIST [PDZ (PSD-95, Discs-large and ZO-1) domain protein interacting specifically with TC10 (a Rho-family small GTPase)] as a binding partner for Rhotekin, using yeast two-hybrid screening. Rhotekin was found to associate with PIST in vitro and in both polarized and non-polarized MDCK (Madin–Darby canine kidney) cells. The C-terminal SPV (Ser-Pro-Val) motif of Rhotekin exhibited binding to the PDZ domain of PIST. The binding was markedly inhibited by an activated version of Rho and partially by that of Rac or Cdc42 in COS7 cells. In contrast, TC10 had no effects on the binding. Immunofluorescence analyses revealed the co-localization of PIST and Rhotekin at the Golgi apparatus in non-polarized fibroblast-like MDCK cells and AJs (adherens junctions) in the fully polarized cells. PIST and Rhotekin are recruited from the cytosol to AJs as the cell becomes polarized. Expression of constitutively active Rho or prevention of Rhotekin–PIST interaction induced diffuse cytoplasmic distribution of Rhotekin in polarized MDCK cells. These results suggest that there is (1) Rho-dependent regulation of Rhotekin-PIST interaction, (2) involvement of PIST in the recruitment of Rhotekin to AJs and (3) a possible role(s) for these two proteins in cell-polarity development and/or maintenance.

INTRODUCTION

Among the Rho family of small GTPases, which includes Rho, Rac, Cdc42 and TC10 (a Rho-family small GTPase), Rho regulates various fundamental cellular processes such as the reorganization of actin cytoskeleton, the formation of focal adhesions, cell movement, cytokinesis, transcription and cell proliferation [1,2]. A variety of downstream effectors of Rho have been identified by intensive studies and shown to play pivotal roles in the Rho-dependent cellular events. Rhotekin is one of the Rho effectors containing a PH (pleckstrin homology) domain and two proline-rich motifs towards the C-terminus (Figure 1A) [3]. It is notable that both human and mouse Rhotekins exhibit, at their C-termini, the sequence QSPV that matches the X(^S/_T)XV consensus known for proteins recognizing PDZ domains. The PDZ domain, the name of which corresponds to the first letters of PSD-95, Discs-large and ZO-1, is known to be present in a rapidly increasing number of proteins exhibiting diverse functions [4].

Figure 1. Identification of PIST as a Rhotekin-binding protein.

(A, B) Structures of Rhotekin (A), PIST (B) and their truncation mutants. The structural domain labelled ‘Pro’ is a proline-rich motif. The regions used for the bait (Rhotekin-C, amino acids 513–551) in yeast two-hybrid screening and Rhotekin-ΔRBD are indicated in (A). Positions of the original PIST fragment identified in the screening (PIST-C), PIST-PDZ, and PIST-Coil are indicated in (B). Numbers refer to amino acid positions. (C) Interaction of Rhotekin-C with the PDZ domain of PIST in yeast two-hybrid assays. Y190 cells co-transformed with pYTH9-Rhotekin-C and pACT2 harbouring a PIST mutant lacking PDZ domain were analysed for growth on medium lacking histidine, tryptophan and leucine, but with 3-aminotriazole. The plus sign (+) represents the growth of the transformed yeast colonies in 3 days. The minus sign (−) represents failure of growth of the transformed yeast colonies in 7 days. (D) Co-sedimentation of PIST with GST–Rhotekin-C. The COS7-cell lysate expressing HA–PIST was incubated with GST or GST-Rhotekin-C, followed by pull-down with glutathione–agarose beads. The resultant samples and the original lysate (‘Lysate’) were separated by SDS/10%-PAGE, followed by Western blotting with anti-HA 12CA5 and anti-GST antibodies to detect PIST and GST proteins respectively.

Rhotekin has recently been reported to control gene-transcriptional events [5,6]. For example, Rhotekin and its binding partner with a PDZ domain, Tax-interacting protein-1 (TIP-1), were shown to co-ordinately regulate Rho-dependent activation of SRE (serum-responsive element) [5]. Rhotekin was also reported to interact with, and disrupt, cytoskeletal septin filaments [7]. However, our knowledge about the physiological significance of Rhotekin is very limited and fragmentary when compared with what we know about several other Rho effectors [1,2].

The presence of functional domains and motifs in Rhotekin (Figure 1A) strongly suggests physiological roles for this protein through protein–protein interactions, but molecules interacting with Rhotekin are almost unknown. To elucidate the function of Rhotekin, we performed screening of its binding partners using the yeast two-hybrid method. Here we identified the Golgi-associated PDZ protein PIST (PDZ domain protein interacting specifically with TC10); also called FIG (fused in glioblastoma), GOPC (Golgi-associated PDZ and coiled-coil motif-containing protein) or CAL [CFTR (cystic fibrosis transmembrane conductance regulator)-associated ligand] [8–11] as an interaction partner for Rhotekin. PIST is a putative effector protein for a TC10 family of Rho proteins [8], and shown to be associated with Golgi-apparatus via binding to the Q-SNARE (Q-soluble NSF attachment protein receptor) protein syntaxin-6 [9]. PIST was also reported to bind to the cell-surface proteins frizzled 5 and frizzled 8, CFTR, chloride channel ClC-3B CALEB/NGC (chicken acidic leucine-rich epidermal growth factor-like domain-containing brain protein)/neuroglycan C) and to regulate the expression of these proteins at the plasma membrane [10–13]. Recent work has shown that a PIST isoform, namely neuronal PIST, is involved in autophagy and neurodegeneration [14] and AMPA (α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid) receptor trafficking in synapses [15].

In the present study we found that PIST and Rhotekin concentrated at the Golgi apparatus in non-polarized MDCK cells and at AJs (adherens junctions) as the cell polarity develops. Co-localization of Rhotekin and PIST, and their interaction in MDCK cells, seems to be regulated by Rho activity. These two proteins are recruited, with slightly different time courses, from the Golgi apparatus to AJs as cell–cell adhesion becomes established. The results obtained suggest that PIST may play some role in the anchoring of Rhotekin at AJs and that the Rhotekin–PIST interaction may be involved in epithelial cell polarity formation and/or maintenance.

MATERIALS AND METHODS

Plasmid construction

Expression plasmids of mouse PIST with three HA (haemagglutinin) tags at the N-terminus (HA–PIST) and constitutively active and dominant negative TC10 (L75- and N31-TC10) were kindly provided by Dr I. G. Macara (Center for Cell Signaling, Department of Microbiology, University of Virginia, Charlottesville, VA, U.S.A.). Rhotekin was kindly given by Dr S. Narumiya (Department of Pharmacology and Horizontal Medical Research Organization, Kyoto University Faculty of Medicine, Kyoto, Japan). Rhotekin-ΔSPV lacking three C-terminal amino acids (Ser-Pro-Val), Rhotekin-C (amino acids 513–551), Rhotekin-ΔRBD lacking the active Rho-binding domain (amino acids 89–551), the PDZ domain of PIST (PIST-PDZ, amino acids 235–462), the coiled-coil region of PIST (PIST-Coil; amino acids 1–201) and PIST C-terminal fragment (PIST-C; amino acids 133–462) were produced by PCR (Figures 1A and 1B) and subcloned into various vectors, including pYTH9, pGEX-4T3, pEGFP, pRK5-Myc and pRK5-FLAG. All constructs were verified by DNA sequencing.

Yeast two-hybrid analyses

pYTH9-Rhotekin-C was used as a bait (Figure 1A) in the two-hybrid screen with human heart cDNA library fused to pACT2 (BD Biosciences/Clontech), the Matchmaker Two-hybrid System Protocol being followed. Subsequent two-hybrid interaction analyses were carried out as described in [16].

In vitro co-sedimentation assays

GST (glutathione S-transferase) and GST–Rhotekin-C were ex-pressed in Escherichia coli and affinity-purified on glutathione–Sepharose beads (Amersham Pharmacia Biotech) according to the manufacturer's instruction. COS7 cell lysate expressing HA–PIST was incubated with GST or GST–Rhotekin-C bound to glutathione–Sepharose beads for 90 min at 4 °C in 25 mM Tris/HCl, pH 7.5, containing 0.1% Triton X-100, 0.1% Nonidet P40, 25 mM NaCl, 1 mM PMSF, 10 μg/ml leupeptin and 10 μg/ml aprotinin. The beads were then washed four times in the buffer, separated by SDS/10%-(w/v)-PAGE and immunoblotted with a monoclonal anti-HA 12CA5 or a polyclonal anti-GST antibody as described in [17]. Immunoreactive bands were visualized using Western Lightning Chemiluminescence Reagent (PerkinElmer).

Preparation and characterization of antibodies

Using a PIST-C fragment expressed in E. coli as an antigen, a rabbit polyclonal antibody specific for PIST was generated and affinity-purified. Anti-Rhotekin antibody was also produced and affinity-purified as described in [7]. A Western-blot analysis was carried out, and immunoreactive bands were visualized as described in [17]. To confirm the specificity of the purified antibody, it was preabsorbed with the antigen.

Cell culture, transfection and immunofluorescence

A variety of cell lines were cultured essentially as described in [18,19]. For preparing the cell lysates used in Western-blot analyses, cells were treated with 10% (w/v) trichloroacetic acid for 15 min on ice, rinsed with ice-cold PBS, and suspended in SDS sample-loading buffer. Transient transfection was carried out using the Nucleofector system (Amaxa Biosystems, Gaitherburg, MD, U.S.A.). Calcium-switch experiments using MDCK cells were performed as described in [20]. Immunofluorescence analysis was done as described in [17]. To detect PIST or Rhotekin, affinity-purified anti-PIST or anti-Rhotekin antibody was used as the primary antibody. Monoclonal anti-ZO-1 (Chemicon), anti-β-catenin (BD Biosciences) and anti-(Golgi 58K protein) (Sigma–Aldrich) antibodies were used as TJ (tight junction), AJ and Golgi-apparatus markers respectively. Alexa Fluor 488-labelled IgG or FluoroLink Cy3-linked IgG (Molecular Probes) was used as a secondary antibody. Fluorescent images were obtained using a FluoView confocal microscope (Olympus).

Immunoprecipitation

Immunoprecipitation was done as previously described [16]. Briefly, COS7 cells expressing tagged proteins, or MDCK cells, were homogenized with lysis buffer containing 40 mM Tris/HCl, pH 7.5, 50 mM NaCl, 5 mM NaF, 100 μM Na₃VO₄, 0.5% Noni-det P40, 10 μg/ml aprotinin and 10 μg/ml leupeptin. Insoluble material was removed by centrifugation at 4 °C for 60 min at 10000 g, and the lysate was used. As for FLAG–Rhotekin or the mutant, immunoprecipitation was done using 1 μg of anti-FLAG M2 antibody (Sigma–Aldrich). After washing the precipitates three times with the lysis buffer, they were subjected to SDS/10%-PAGE. Western blotting was then carried out with polyclonal anti-HA (Santa Cruz Biotechnology) or anti-FLAG (Sigma–Aldrich). Immunoprecipitation of endogenous PIST with Rhotekin from the lysates of fibroblast-like non-polarized (∼47 mg of protein) or polarized MDCK cells (∼45 mg of protein) was carried out with 4 μg of anti-PIST-C or rabbit IgG, and Western blotting was done as described in [21]. The relative level of protein in the immunoprecipitate compared with the total amount in the lysate was calculated using NIH Image software based on densitometry and expressed as a percentage.

RESULTS

Identification of PIST as a Rhotekin-binding partner by yeast two-hybrid screening

In order to search for candidate proteins involved in the Rho/Rhotekin signalling, we attempted to identify binding partners for Rhotekin. We focused on the C-terminus of the protein, since it contains a consensus binding motif for the class I PDZ domain. Many proteins binding to this domain are thought to play important roles in cell polarity, cell adhesion and cell signalling [4]. We performed a yeast two-hybrid screen with Rhotekin-C as a bait and a cDNA library from human heart (Figure 1A). After screening of 2.0×10⁵ clones, we obtained 83 positive ones. Partial cDNA sequence analyses revealed that two clones showing a strong positive interaction with Rhotekin-C corresponded to the C-terminal region (starting at amino acid 133) of PIST (Figure 1B). PIST contains two putative coiled-coil domains and a PDZ domain. Whereas the second coiled-coil domain has been reported to be important for the localization of PIST to the Golgi apparatus [9], a putative leucine zipper present within the domain is also essential for the interaction with activated TC10 [8].

Since the C-terminal amino acids of Rhotekin specify a consensus sequence recognizing PDZ domains, it is most likely that it is through its C-terminus that Rhotekin binds to the PDZ domain of PIST. We thus tested, using yeast two-hybrid analysis, whether Rhotekin-C interacts with a mutant of the PIST fragment lacking most of the PDZ domain. As shown in Figure 1(C), the deletion mutant did not bind with Rhotekin-C, indicating that the PDZ domain is essential for the interaction with Rhotekin. To further analyse the interaction between C-terminus of Rhotekin with PIST, we performed pull-down analyses. A purified GST-Rhotekin-C or GST was incubated with COS7 cell lysates expressing HA–PIST, and the resulting complexes were separated on SDS/polyacrylamide gels, followed by Western blotting with 12CA5. As shown in Figure 1(D), GST-Rhotekin-C was capable of binding HA–PIST under the conditions used.

Interaction of Rhotekin with PIST in cells

The data obtained by two-hybrid analyses and pull-down assays strongly suggest that Rhotekin and PIST form a complex in cells. To test this possibility, immunoprecipitation analyses were carried out using the COS7-cell transient-expression method. As shown in Figures 2(A) and 2(B), PIST was co-immunoprecipitated with Rhotekin. The association between PIST and Rhotekin was lost when Rhotekin-ΔSPV, a Rhotekin mutant lacking three C-terminal amino acids, was used instead of the wild-type (Figures 2A and 2B). These results confirmed that Rhotekin interacts with PIST in cells and that the three C-terminal amino acids of Rhotekin are essential for association with the PDZ domain of PIST.

Figure 2. Association of Rhotekin with PIST in cells.

(A) Co-immunoprecipitation analyses of HA–PIST with FLAG–Rhotekin, FLAG–Rhotekin-ΔSPV or FLAG-tag. Lysates from transfected COS7 cells were immunoprecipitated with M2 as described in the Materials and methods section. A proportion (20%) of the precipitated materials was subjected to SDS/10%-PAGE, followed by Western blotting with a polyclonal anti-HA antibody. The relative levels of HA–PIST in the immunopellets were measured by densitometry and expressed as percentages of the total amount in the lysate used in each assay. (B) Expression of each protein in (A) was confirmed by Western blotting of cell lysates (3% of total volume) with 12CA5 (upper panel) or M2 (lower panel). (C) Effects of activated versions of Rho, Rac, Cdc42, TC10 and dominant-negative Rho on the interaction between Rhotekin and PIST. COS7 cells were transfected with HA–PIST, FLAG–Rhotekin, FLAG–L63RhoA, FLAG–L61Rac, FLAG–L61Cdc42, HA-L75TC10 and Myc-N19Rho in various combinations. Lysates from transfected cells were immunoprecipitated with M2 as in (A). Western blotting was carried out with a polyclonal anti-HA antibody as in (A) (upper panel). The blotting membrane was re-probed with polyclonal anti-FLAG antibody to confirm the precipitated FLAG-Rhotekin (lower panel). The relative levels of HA–PIST in the immunopellets were measured as in (A). (D) Expression of each protein in (C) was confirmed by Western blotting of cell lysates (3% of total volume) with 12CA5, M2 and 9E10.

Since Rhotekin and PIST are target proteins for Rho and TC10 respectively, it is possible that the functions and/or properties of Rhotekin and PIST may be modified by interaction with Rho and/or TC10. We thus questioned whether the interaction between Rhotekin and PIST is influenced by the presence of a mutated version of Rho or TC10. As shown in Figures 2(C) and 2(D), an activated mutant of Rho, L63Rho (i.e. [Leu⁶³]Rho), inhibited the Rhotekin–PIST interaction. Constitutively activated versions of Rac (L61Rac) and Cdc42 (L61Cdc42) also inhibited the Rhotekin–PIST interaction to some extent (Figures 2C and 2D). The observed partial effects may be due to activation of endogenous Rho by these mutants, as has been described for fibroblastic cells [22,23]. On the other hand, dominant-negative forms of Rho (N19Rho), Rac (N19Rac) and Cdc42 (N17Cdc42) had little effect on the Rhotekin–PIST interaction (Figures 2C and 2D, and results not shown). These results suggest that the Rhotekin–PIST interaction is inhibited by activated Cdc42–Rac–Rho cascade in the COS7-cell expression system and that inhibition of this signalling cascade has no effect on Rhotekin–PIST interaction, although the precise molecular mechanism underlying the observed findings remains to be elucidated. It is noteworthy that expression of constitutively activated L75TC10 and dominant negative N31TC10 had no effect on the binding of Rhotekin with PIST (Figures 2C and 2D, and results not shown).

Detection of PIST in various cell lines and of the Rhotekin–PIST complex in MDCK cells, both before and after polarization

To further explore the interaction between PIST and Rhotekin, we developed a rabbit polyclonal antibody (anti-PIST-C) against the bacterially synthesized PIST-C. It was affinity-purified on a column to which the antigen had been conjugated. Specificity of the antibody was confirmed by Western-blot analyses using lysate from COS7 cells expressing HA–PIST. As shown in the upper-left panel of Figure 3(A), we detected HA–PIST with an apparent molecular mass of 68 kDa in the lysate by Western blotting. It should be noted that HA–PIST contains three HA tags, so that the apparent molecular mass is ∼3.5 kDa higher than that of endogenous PIST. Western-blot analyses revealed two PIST protein bands of 60 and 35 kDa in both HA–PIST-expressing and control cells. We assume that there are splicing variants detected by anti-PIST-C, although further analyses are required to clarify this possibility. Preincubation of the antibody with the antigen inhibited the immunoreactivity (Figure 3A, upper right panel). We next did Western-blot analyses to detect endogenous PIST in lysates from a variety of mammalian cell lines and detected proteins with apparent molecular masses of 60, 46 and 35 kDa (Figure 3B). The 60 kDa protein was detected in all cell lines tested, whereas the expression of 46 and 35 kDa proteins was cell-type-specific. The 60 and 46 kDa isoforms had previously been detected in some tissues [9]. It is not clear whether the 35 kDa protein is a splicing variant or a degradation product.

Figure 3. Detection of endogenous PIST and immunocomplex formation of endogenous PIST and Rhotekin.

(A) Lysates from COS7 cells (30 μg of each) expressing HA–PIST (left lane of each panel) or HA-tag (−, right lane of each panel) were immunoblotted with anti-PIST-C (upper left panel) or the antibody preabsorbed by the antigen (upper right panel). The blot membranes were re-probed with an anti-tubulin monoclonal antibody (Sigma–Aldrich) for loading control (lower panels). (B) Detection of endogenous PIST proteins. Lysates (30 μg of each) of HeLa, T24, MDCK, G361, Saos2, SK-LC-2, CASKY, C6D6 and U251 cells were subjected to SDS/10%-PAGE, and Western blotting was done to detect endogenous PIST. Molecular size markers are on the extreme left. (C) Immunocomplex formation of endogenous PIST and Rhotekin in MDCK cells. PIST was immunoprecipitated with anti-PIST-C from polarized (upper panels) or non-polarized cells (lower panels) as described in the Materials and methods section. In control experiments, rabbit IgG was used for immunoprecipitation. A proportion (10%) of the precipitated material was subjected to SDS/10%-PAGE, followed by Western blotting with anti-Rhotekin (left panels) or anti-PIST-C (right panels). The cell extract (input in each panel was 0.1% of the total amount) was also used as a control. The relative levels of Rhotekin and PIST in the immunopellets compared with the total amount of protein used in the immunoprecipitation assays were calculated as described in the Materials and methods section and expressed as percentages.

Immunocomplex formation of Rhotekin and PIST expressed in COS7 cells strongly suggests their physiological interaction in tissues. We thus tested whether endogenous Rhotekin and PIST associate with each other. We first confirmed that anti-PIST-C immunoprecipitates HA–PIST with FLAG–Rhotekin expressed in COS7 cells (results not shown). We then immunoprecipitated endogenous PIST from extracts of polarized or non-polarized MDCK cells with anti-PIST-C. As shown in Figure 3(C), the amount of Rhotekin in the immunoprecipitate from non-polarized cells was comparable with that from polarized cells on the basis of Western-blot analyses. Although these data suggest that Rhotekin forms a complex with PIST both before and after polarization, it is possible that Rhotekin interacts not only with PIST, but also with yet-unidentified other proteins under polarized and/or non-polarized conditions, since (1) many other PDZ proteins are likely to be enriched at cell–cell contact sites, and (2) we could immunoprecipitate only ∼10% of PIST from the lysates used in the analyses.

Subcellular localization of PIST and Rhotekin in MDCK cells

Although PIST is known to be a Golgi-resident protein in several cell types, neuronal PIST is localized partly in the synapse of distal dendrites of hippocampal neurons [15], suggesting (a) polarity-related function(s) of PIST. Since the distribution of PIST during the polarity-forming process remains generally unknown, we examined the subcellular localization of PIST in non-polarized fibroblast-like and fully polarized epithelial MDCK cells, using a confocal microscope. In addition, since the intracellular distribution of Rhotekin has not been observed at all, we also determined the localization of Rhotekin in non-polarized and polarized MDCK cells. We then compared the localization patterns of PIST with those of Rhotekin to examine whether these proteins are co-localized in the cells. We first double-stained endogenous PIST or Rhotekin with a Golgi marker, Golgi 58K protein, to clarify the relationship of these proteins to the Golgi apparatus. As shown in Figure 4(A), panels (a) and (b), PIST was significantly co-localized with the Golgi 58K protein in non-polarized MDCK cells, as observed in other cell lines [9–13], and also localized at peripheral cell regions where cells contact with neighbouring ones (arrowheads in Figure 4A, panel a). Distribution of PIST at peripheral cell areas, as well as in the Golgi apparatus, suggests a possible function of the protein in the trafficking of various molecules from the juxtanuclear region to the secretory pathway towards the plasma membrane, as has been proposed recently [13,15]. As in the case of PIST, the staining pattern of Rhotekin was reminiscent of that of the Golgi apparatus (Figure 4A, panels c and d). It is, however, noteworthy that Rhotekin was not observed at peripheral cell regions of fibroblast-like MDCK cells, suggesting that Rhotekin does not have a physiological role in these non-polarized cells.

Figure 4. Subcellular distribution of PIST and Rhotekin in non-polarized fibroblast-like and fully polarized MDCK cells.

(A) Fibroblast-like non-polarized cells were double-stained using anti-PIST-C (a) or anti-Rhotekin (c) with anti-Golgi 58K (b and d). Arrowheads show PIST localization at cell–cell contact sites. (B) Fully polarized cells were double-stained as in (A). The horizontal bars represent 20 μm.

On the other hand, in the fully polarized cells, endogenous PIST was highly concentrated at cell–cell contact areas and weakly localized in the cytoplasm (Figure 4B, panels a and b). Rhotekin is also distributed at regions of cell–cell contact and the cytoplasm (Figure 4B, panels c and d). These results suggest that PIST and Rhotekin play some role in the formation and/or maintenance of cell–cell contacts, possibly in a concerted manner. It also should be noted that the staining pattern of Rhotekin in the cytoplasm of polarized MDCK cells became apparently denser and thicker, and overlapped with the compacted dot-like staining of the Golgi apparatus. These results imply that the Golgi-related function of Rhotekin is still retained in polarized MDCK cells.

Subcellular distribution patterns of Rhotekin and PIST in non-polarized MDCK cells were different from those observed in the polarized cells (Figure 4). We thus further examined their localization during the establishment of cell polarity. As shown in Figure 5(A), panels (a) and (e), after 1 day of plating the cells showed fibroblastic morphology, although a TJ marker, ZO-1, was already localized at cell/cell boundaries. In such cells, PIST was distributed not only in the Golgi apparatus, but also at sites of cell–cell contact. After 3 days of plating, PIST was distributed mainly at cell–cell contact areas (Figure 5A, panels b and f). Finally, PIST became highly concentrated at cell–cell contact sites and only faintly present in the Golgi apparatus by day 7 after plating (Figure 5A, panels d and h). On the other hand, Rhotekin is concentrated only in the Golgi apparatus at 1 day after plating (Figure 5B, panels a and e). The Golgi-enriched localization pattern was still obvious at 3 days after plating, although the protein distributed partially in cytoplasm at this time point (Figure 5B, panels b and f). The compacted dot-like staining pattern of Rhotekin became obvious in the Golgi apparatus by day 5 (Figure 5B, panels c and g). Localization of Rhotekin at cell–cell contact areas was observed weakly after 5 days of plating and became clear at day 7 (Figure 5B, panels d and h). These results demonstrated that PIST became localized at cell–cell contact areas earlier than did Rhotekin during MDCK-cell polarization, and that these two proteins may not interact with each other throughout the cell-polarity-establishment process. It is also likely that the subcellular distribution of PIST and Rhotekin is regulated spatiotemporally by different molecular machineries, in spite of their co-localization in the Golgi apparatus and at cell–cell contact areas in non-polarized and polarized MDCK cells respectively.

Figure 5. Changes in distribution patterns of PIST and Rhotekin during the establishment of MDCK-cell polarity.

(A, B) Cells after plating at low density in normal medium were fixed at indicated time points and double-stained with anti-PIST-C (A, panels a–d) and anti-ZO-1 (A, panels e–h) or anti-Rhotekin (B, panels a–d) and anti-ZO-1 (B, panels e–h). (C, D) Polarized cells were cultured in low-calcium medium for 12 h, then shifted to medium containing normal calcium concentrations (0 h) and incubated for an additional 2, 6 or 24 h. At indicated time points, cells were fixed and double-stained for PIST (C, panels a–d) and ZO-1 (C, panels e–h), or Rhotekin (D, panels a–d) and ZO-1 (D, panels e–h). (E) Localization of PIST (panels a–f) and Rhotekin (panels g–l) at the vertical section in fully polarized cells. Cells were double-stained for PIST with β-catenin (panels a–c) or ZO-1 (panels d–f), and for Rhotekin with β-catenin (panels g–i) or ZO-1 (panels j–l). The merged images are also shown (panels c, f, i and l). The horizontal bars represent 20 μm.

To further understand the role of Rhotekin and PIST in the polarity-establishment process, we analysed redistribution patterns of these proteins at early stages in the development of cell–cell contacts using the calcium-switch assay [20]. In this assay, MDCK monolayers treated with low-calcium medium for 12 h, which induced rapid dissolution of the AJ complex (as monitored by ZO-1 localization), were transferred to medium containing normal calcium levels. As shown in Figures 5(C), panel (e), and (D), panel (e), culturing confluent MDCK cells in low-calcium medium induced the loss of cell–cell contacts, and ZO-1 was hardly detected. At this time point (0 h), both PIST and Rhotekin redistributed in cytoplasm and perinuclear regions (Figures 5C and 5D, panels a). When the cells were transferred back to the normal-calcium medium, the cell–cell contact sites were gradually formed again. After 2 h of transfer of the cells to the normal-calcium medium, both PIST and Rhotekin were largely retained in the Golgi apparatus, whereas cell–cell contact sites were partially formed (Figures 5C and 5D, panels b and f). In 6 h, there was a considerable appearance of PIST at the cell–cell contact sites, whereas Rhotekin remained distributed largely in the cytoplasm and only partly at the cell–cell contact sites (Figures 5C and 5D, panels c and g). In 24 h, both PIST and Rhotekin became enriched at the sites of cell–cell contacts in the polarized MDCK cells (Figure 5C and 5D, panels d and h). These results demonstrated that polarized distribution of Rhotekin and PIST at the cell–cell contact sites was dependent on the maintenance of calcium-dependent cell–cell adhesion.

The temporal change in the subcellular localization of PIST occurs earlier than does that of Rhotekin during the establishment of cell polarity, both in the normal cell culture process and in the calcium-switching experiment. It is therefore likely that PIST is first translocated from the Golgi apparatus to lateral plasma membranes and interacts with them through its C-terminal coiled-coil region, as described in [9]. Rhotekin is then recruited to the Golgi and interacts through its C-terminal amino acids with the PDZ domain of PIST. Since interaction between PIST and Rhotekin is inhibited by co-expression of activated Rho GTPases, it is tempting to speculate that Rho signals regulate the establishment and/or maintenance of cell polarity by controlling Rhotekin–PIST interaction.

From the results obtained here, PIST and Rhotekin were found to be enriched at cell–cell contact sites in polarized MDCK cells. We thus examined the precise localization of endogenous PIST and Rhotekin in fully polarized MDCK cells by confocal microscopy. When cells were double-stained with anti-PIST-C and anti-β-catenin, the PIST signal coincided well with that of β-catenin, mainly along lateral membranes, on the basis of the vertical sectional image (Figure 5E, panels a–c). By contrast, ZO-1 was concentrated at the most apical regions of the cell–cell adhesion sites, and vertical sectional images showed that PIST was only slightly co-localized with ZO-1 (Figure 5e, panels d–f). On the other hand, when cells were double-stained for Rhotekin and β-catenin, the Rhotekin signal partially coincided with that of β-catenin at cell–cell adhesion areas (Figure 5E, panels g–i). Co-staining of Rhotekin with ZO-1 demonstrated that Rhotekin is not particularly co-localized with ZO-1 (Figure 5E, panels j–l). Vertical sectional images revealed that neither PIST nor Rhotekin are distributed in the most apical cytoplasmic regions. These findings indicate that PIST and Rhotekin are partly, but clearly, co-localized at AJs, although their functions there are not clear. PIST and Rhotekin were also observed in the cytoplasm, where their localization patterns appear to be slightly different (Figures 5A–5D, panels d).

Possible regulation of Rhotekin localization by Rho and PIST

From the finding in the present study that PIST is recruited to AJs earlier during MDCK cell polarization than is Rhotekin, PIST is likely to be responsible for recruiting Rhotekin to the AJ. To test this possibility, we tried to decrease the PIST protein level in MDCK cells using an RNA-interference method. However, we could not knock down PIST using several small-interfering-RNA duplexes whose sequences are from mouse or rat PIST, perhaps as a result of sequence mismatch (since the cDNA sequence of canine PIST is not available). Instead, to gain some insight into the physiological significance of PIST, we tried to down-regulate PIST by introducing its fragments into the cell. For this purpose, we constructed two PIST fragments, GFP (green fluorescent protein)–PIST-PDZ and GFP–PIST-Coil, as described in the Materials and methods section, and tested to see whether these fragments inhibited PIST–Rhotekin interactions in MDCK cells. One or both of the fragments was expected to inhibit as-yet-unidentified PIST function(s) by preventing interactions with PIST-binding partners. As shown in Figure 6(A), immunocomplex formation of FLAG–Rhotekin with HA–PIST was prevented in the presence of GFP–PIST-PDZ. In this case, GFP–PIST-PDZ was found to form a complex preferentially with Rhotekin, suggesting a higher affinity of this fragment for Rhotekin. By contrast, GFP–PIST-Coil had no effect on PIST–Rhotekin interaction in the immunoprecipitation assay. From these biochemical observations it is strongly suggested that the PIST-PDZ fragment alters intracellular localization of Rhotekin by preventing interaction between Rhotekin and PIST. We next carried out cell-biological analyses and expressed PIST-PDZ in MDCK cells. As shown in Figure 6(B), PIST-PDZ distributed in the cytoplasm and the staining pattern of endogenous Rhotekin was dramatically altered: the immunoreactivity of Rhotekin increased greatly in the cytoplasm, although the dissociation of Rhotekin from AJs was not clear because of the very high immunoreactivity of Rhotekin. By contrast, PIST-Coil had little effect on the localization of endogenous Rhotekin (Figure 6C). These results suggest a possible important role of PIST, especially the PDZ domain, in Rhotekin localization. It should be noted that localization of β-catenin at cell–cell contact sites was not affected by expression of PIST-PDZ or PIST-Coil fragments (results not shown).

Figure 6. Rho signal is crucial for localization of Rhotekin, but not for PIST, in MDCK cells.

(A) Effects of the PIST fragments PIST-PDZ and PIST-Coil on the interaction between Rhotekin and PIST. COS7 cells were transfected with HA–PIST, FLAG–Rhotekin, GFP–PIST-PDZ and GFP–PIST-Coil in various combinations. Expression of each protein was confirmed by Western blotting (WB) of the cell lysates (3% of total volume) with a mixture of 12CA5 and monoclonal anti-GFP antibody (Santa Cruz Biotechnology) (left panel). Lysates from transfected cells were immunoprecipitated with M2 as described in the Materials and methods section. A proportion (20%) of the precipitated material was subjected to SDS/10%-PAGE, followed by Western blotting with anti-PIST-C (upper right panel). Precipitated FLAG–Rhotekin was confirmed by re-probing the blotting membrane with M2 (lower right panel). (B) Effects of PIST-PDZ on Rhotekin distribution in polarized MDCK cells. Cells expressing PIST-PDZ fragment were stained for Rhotekin (panel a). Distribution of PIST-PDZ is also shown (panel b). (C) Effects of PIST-Coil on Rhotekin distribution in polarized MDCK cells. Cells expressing PIST-Coil fragment were stained for Rhotekin (panel a). Distribution of PIST-Coil was also shown (panel b). (D) Expression of Myc-L63Rho (panels a–d) or Myc-Rhotekin-ΔRBD (panels e and f) in polarized MDCK cells. Endogenous Rhotekin (panel a) or PIST (panels c and e) was double-stained with 9E10 (panels b, d and f). (E) Expression of Myc-L63Rho in non-polarized MDCK cells. Endogenous Rhotekin (panel a) was double-stained with 9E10 (panel b). Note that cells expressing L63Rho show round-up shapes. The horizontal bars represent 20 μm.

Since the interaction of Rhotekin with PIST was inhibited by activated Rho (Figure 2), it is most likely that constitutively active Rho disrupts co-localization of Rhotekin and PIST in MDCK cells. To test this possibility, we expressed L63Rho in the cells and analysed the intracellular localization of endogenous PIST and Rhotekin. As shown in Figure 6(D), panels (a) and (b), Rhotekin became localized in the cytoplasm with a higher intensity on expression of L63Rho, as is the case for PIST-PDZ expression. We assume that L63Rho binds with the RBD domain of Rhotekin, inhibits its interaction with PIST and consequently causes dissociation of PIST from AJs of polarized MDCK cells.

On the basis of the results shown in Figure 5, during MDCK cell polarization PIST translocates to AJs earlier than does Rhotekin. However, the possibility cannot be excluded that the Rho signal controls the recruitment of PIST to AJs through as-yet-unidentified pathways. We thus tested the effects of L63Rho on the localization of endogenous PIST. Consequently, expression of L63Rho slightly increased the cytoplasmic localization of PIST, but PIST was mainly enriched at AJs, as is the case in non-transformed cells (Figure 6D, panels c and d). We also questioned whether Rhotekin per se affects the distribution of PIST. As shown in Figure 6(D), panels (e) and (f), Rhotekin-ΔRBD, which does not trap endogenous active Rho, also had little effect on the endogenous PIST localization. Taken together, it is supposed that (1) dissociation of Rhotekin from PIST and its subsequent cytoplasmic localization may be due to the Rho-mediated conformational change of Rhotekin, although other possibilities cannot be ruled out, (2) Rhotekin localization at AJs is dependent on its interaction with PIST, and (3) Rho/Rhotekin signalling does not play a role in PIST localization.

The present findings suggest that PIST is responsible for recruiting Rhotekin to AJs during MDCK-cell polarization. However, the mechanism by which PIST dissociates from Rhotekin at the Golgi apparatus at the initiation of cell polarization is not clear. It is also possible that the Rho signal plays a pivotal role in PIST–Rhotekin interaction and/or their dissociation at Golgi apparatus in non-polarized MDCK cells. If this is the case, Rho signalling through Rhotekin is likely to regulate the initial step of recruitment of PIST to AJs. We thus expressed L63 Rho in non-polarized MDCK cells in order to analyse the effects of Rho signals on the localization of PIST in the cells. Consequently, cells expressing L63Rho showed a contracted round-up shape, but L63Rho had little effect on PIST localization in non-polarized MDCK cells (Figure 6E). When Rhotekin-ΔRBD was expressed in non-polarized MDCK cells, PIST localization at the Golgi apparatus did not change (results not shown). Therefore we assume that Rho/Rhotekin signals do not play a central role in intracellular PIST localization in non-polarized MDCK cells, as is the case in polarized ones.

DISCUSSION

Cadherin-mediated AJs are cell–cell adhesion structures that have been extensively studied. Classical cadherins, including E-cadherin, N-cadherin and VE-cadherin, are single-span transmembrane proteins that establish calcium-dependent cell–cell contacts [24]. Homophilic interactions between the extracellular domains of cadherins on neighbouring cells are integral to the establishment of AJs. The intracellular regions of cadherins interact with various partners, most notably the catenins, p120 (p120catenin) and β-catenin, which then interact with α-catenin, linking the cadherin complex to the actin cytoskeleton and strengthening the adhesion.

Several signalling pathways, including Rho-family-mediated ones, regulate cadherin adhesion, and in turn, the adhesive complexes influence a number of pathways. The Rho family of small GTPases has been reported to be important in the regulation of epithelial structure, function and assembly [25–30]. These small GTPases regulate epithelial intercellular junctions via distinct biochemical mechanisms, and imbalance in active/resting GTPase levels is thought to induce perturbations in barrier functions. Rac-driven lamellipodia and Cdc42-driven filopodia are reported to be important for establishment of cadherin adhesion, maybe through the initiation of cell–cell contacts [31,32]. An effector protein for Cdc42 and Rac, IQGAP1 (IQ motif containing GTPase activating protein 1), is one of the regulatory molecules of AJs. IQGAP1 localizes to sites of cell–cell contacts and affects E-cadherin-mediated adhesion differently, depending on its binding partners [33].

In spite of its importance, the precise molecular mechanism governing Rho-dependent cell–cell contact regulation is almost unknown when compared with that of Rac and Cdc42. Rho has been shown to negatively regulate cadherins through the actin cytoskeleton or other interactions independent of IQGAP and β-catenin [34]. Two effectors of Rho, Dia and ROCK (Rho kinase), have been reported to participate in AJ formation and show opposing effects on AJs [35]. During AJ formation, a low level of Rho activity is required for Dia-mediated AJ stabilization, but once AJs are established, Rho is inhibited to prevent AJ disruption, which is mediated by ROCK [35].

In the present study we demonstrated that Rhotekin interacts with PIST in vitro and in cells. This interaction raised the possibility of a novel function of Rhotekin other than the control of gene transcription and septin filament organization so far described [5–7]. PIST has been reported to be involved in vesicle transport between Golgi and plasma membranes [8–12,14]. Since Rhotekin and PIST are co-localized and interact with each other in the Golgi apparatus in non-polarized fibroblast-like MDCK cells, we suggest that Rhotekin, in concert with PIST, functions in Golgi-related cellular processes, such as vesicle trafficking in cells, although additional studies are needed to clarify this possibility. On the other hand, concentration of these proteins at the AJ in polarized MDCK cells suggests their novel roles in cell-polarity regulation. Rhotekin and PIST are downstream effectors for Rho and TC10 respectively, and it is therefore possible for them to be regulated by these GTPases. Indeed, the interaction of PIST with Rhotekin was significantly inhibited by activated Rho. Rho activity is reported to be dramatically suppressed in polarized MDCK cells when compared to non-polarized ones; calcium-switch experiments using MDCK cells revealed that Rho activity started to decrease after 1 h of restoration of normal calcium levels, so that, by 8 h, it became very low [36]. From these observations and our results showing that the activated version of Rho inhibited Rhotekin–PIST interaction, Rhotekin and PIST are likely to form stable complexes during AJ formation and maturation.

The Rho-dependent molecular mechanism regulating PIST–Rhotekin interaction is not obvious. It is possible that Rho-mediated conformational change in Rhotekin is responsible for its dissociation from PIST. In the case of mDia1 (a mammalian homologue of the Drosophila diaphanous protein), activated Rho has been reported to activate mDia1 by disrupting its intramolecular interaction, and consequently mDia1 is thought to interact with its binding partners through its domains exposed outside by the conformational change [37]. It is, however, not clear whether a similar mechanism regulates Rhotekin conformation and its function(s), since intramolecular interaction of the N-terminal half fragment with the C-terminal half has not yet been observed both in vitro and in COS7 cells (results not shown). The role of TC10 in Rhotekin–PIST interaction also remains to be elucidated. A recent report suggests that TC10 regulates CFTR expression by interacting with PIST [38].

On the basis of the present findings we propose a hypothesis whereby Rhotekin and PIST co-ordinately function in the establishment and/or maintenance of cell polarity in a Rho-dependent manner. Although Rhotekin interacts with PIST through the PDZ domain, the precise molecular mechanism, and physiological significance of their interaction, has not been elucidated. One report demonstrated a physiological role of the interaction between Rhotekin and TIP-1, a PDZ protein. Co-expression of TIP-1 with activated Rho and Rhotekin caused SRE activation, this effect of TIP-1 being lost when the Rhotekin–TIP-1 interaction was prevented [5]. In the case of PIST, extensive analyses will be required to clarify the molecular mechanism of Rhotekin–PIST complex formation, as the function of these proteins during cell polarization remains almost unknown.