PAR-6 regulates aPKC activity in a novel way and mediates cell-cell contact-induced formation of the epithelial junctional complex


* E-mail:


Background PAR-6, aPKC and PAR-3 are polarity proteins that co-operate in the establishment of cell polarity in Caenorhabditis elegans and Drosophila embryos. We have recently shown that mammalian aPKC is required for the formation of the epithelia-specific cell-cell junctional structure. We have also revealed that a mammalian PAR-6 forms a ternary complex with aPKC and ASIP/PAR-3, and localizes at the most apical end of the junctional complex in epithelial cells.

Results The ternary complex formation and junctional co-localization of PAR-6 with aPKC and ASIP/PAR-3 are observed during the early stage of epithelial cell polarization. In addition, over-expression of the PAR-6 mutant with CRIB/PDZ domain in MDCK cells disturbs the cell-cell contact-induced junctional localization of tight junction proteins, as well as inhibiting TER development. Furthermore, the binding of Cdc42:GTP to the CRIB/PDZ domain of PAR-6 enhances the kinase activity of PAR-6-bound aPKC. Detailed analyses suggest that the binding of PAR-6 to aPKC has the intrinsic potential to activate aPKC, which is only released when Cdc42:GTP binds to the CRIB/PDZ domain.

Conclusion The results indicate the involvement of PAR-6 in the aPKC function which is required for the cell-cell adhesion-induced formation of epithelial junctional structures, possibly through the cooperative regulation of aPKC activity with Cdc42.


Studies of asymmetric cell division during embryogenesis in Caenorhabditis elegans and Drosophila have provided evidence that a transient asymmetric (polarized) distribution of proteins at the cell periphery is required prior to asymmetric cell division (Knoblich 1997). In the C. elegans one-cell embryo, the six par (partitioning defective) genes are required for the unequal distribution of most cell-fate determinants thus far identified (Guo & Kemphues 1996). Previously, we found that a mammalian homologue of the par-3 gene product, ASIP, binds specifically to atypical PKC (aPKC) through its evolutionarily conserved domain (Izumi et al. 1998). Based on this finding, we have shown that a C. elegans atypical PKC, PKC-3, binds to PAR-3 and plays an indispensable role in this cell polarization event with PAR proteins (Tabuse et al. 1998). Subsequent genetic analyses revealed that PKC-3, PAR-3 and PAR-6 co-operate interdependently to establish cell polarity during the early phase of C. elegans embryogenesis (Hung & Kemphues 1999; Tabuse et al. 1998; Watts et al. 1996). Recent experiments on Drosophila embryos have further demonstrated that Drosophila homologues of aPKC, PAR-3 (Bazooka) and PAR-6 are also required for the establishment of cell polarity of the embryonic epidermis as well as neuroblasts (Kuchinke et al. 1998; Muller & Wieschaus 1996; Petronczki & Knoblich 2001; Schober et al. 1999; Wodarz et al. 1999; Yu et al. 2000), further supporting the presence of an evolutionarily conserved cell-polarization machinery consisting of aPKC, PAR-3 and PAR-6.

In mammals, epithelial tissues and cells have provided an experimental system in which the essential features of cell asymmetry have been revealed. Fully polarized epithelial cells contain apicobasal polarity that provides the basis for directed transport across the epithelium. These cells have an asymmetric junctional complex that includes adherens junctions and tight junctions that localize to the most apical end (Balda & Matter 1998). Studies on cultured epithelial cells, such as MDCK cells, have revealed that these junctional structures play essential roles in the establishment of epithelial cell polarity (Drubin & Nelson 1996; Eaton & Simons 1995; Gumbiner 1996). Recently, we have shown that aPKC, ASIP/PAR-3 and a mammalian homologue of PAR-6 form a ternary complex through aPKC as a linker molecule, and co-localize to the apical end of the junctional complex in polarized epithelial cells (Suzuki et al. 2001). We have further demonstrated that the introduction of a kinase defective aPKC mutant (aPKC-kn) into MDCK cells blocks the cell-cell contact-induced formation of junctional structures and thus inhibits the development of cell surface polarity (Suzuki et al. 2001). Considering the close functional interaction of these proteins in the polarization of the C. elegans one-cell embryo, these results suggest that the PAR-6-aPKC-ASIP/PAR-3 ternary complex, as well as the kinase activity of aPKC, play critical roles in the regulation of mammalian epithelial cell polarity. Importantly, the dominant negative effect of aPKC-kn was observed only when it is expressed in cells without cell polarity, and has no significant effect on polarized cells (Suzuki et al. 2001). This raises the intriguing possibility that the role of the PAR-6-aPKC-ASIP/PAR-3 ternary complex is not the maintenance but rather the establishment of cell polarity.

Recent studies have also demonstrated an unexpected molecular interaction between the CRIB/PDZ domain of PAR-6 and Cdc42/Rac1 (Joberty et al. 2000; Johansson et al. 2000; Lin et al. 2000; Noda et al. 2001; Qiu et al. 2000). In addition, recombinant Cdc42:GTP can pull down endogenous PAR-6, aPKC and ASIP/PAR-3 from MDCK lysates (Joberty et al. 2000). Interestingly, in an over-expression system in COS cells, immunoprecipitated aPKCζ showed increased kinase activity when co-transfected with PAR-6, and this increase was suppressed by the additional co-transfection of dominant negative Cdc42/Rac1 (Qiu et al. 2000). In addition, the over-expression of PAR-6, but not its mutant lacking the CRIB motif, enhances Rac1-dependent cell transformation in NIH3T3 cells (Qiu et al. 2000). Taken together with the critical importance of aPKC kinase activity during the early phase of epithelial cell polarization (Suzuki et al. 2001) and the role of Cdc42 in cell polarization in yeast (Johnson 1999), these results raise the possibility that aPKC activity in protein complexes containing PAR-6, aPKC and ASIP/PAR-3 is regulated by Cdc42, and is involved in the early phase of epithelial cell polarization. However, the exact role of PAR-6 in the epithelial cell polarization, as well as the biochemical consequence of the interaction between the aPKC-PAR complex and Cdc42, remains to be clarified.

The present study was thus undertaken to clarify the nature and functional role of the ternary complex during the early phase of cell polarization in MDCK epithelial cells. We present evidence that the pre-existing ternary complex is targeted to the cell-cell contact region at an early stage of cell-cell contact. Results of experiments using adenovirus-mediated gene transfer also provide functional evidence supporting PAR-6 as being involved in the aPKC signalling that is required for cell-cell contact induced epithelial cell polarization. In addition, we also provide the first evidence supporting the in vivo interaction between Cdc42 and the ternary complex comprising PAR-6, aPKC and ASIP/PAR-3. We further reveal a biochemical consequence of the interaction between aPKC and PAR-6, that is, PAR-6 regulates aPKC activity through its interaction with the regulatory region of the kinase, and Cdc42 modifies this regulation by binding to the CRIB/PDZ domain of PAR-6.


Concerted targeting of PAR-6, aPKCλ and ASIP to the epithelial junctional complex in a cell-cell adhesion-dependent manner

As a first step to evaluate the functional importance of the molecular interaction among PAR-6, aPKC and ASIP/PAR-3, we focused on the cellular distribution of the three proteins during the early phase of cell polarization. Previous experiments have shown that calcium depletion from the culture medium of MDCK cells results in a disappearance of cell-cell contacts and a loss of tight junctions and adherens junctions. Furthermore, adding calcium to the culture medium (calcium switch) causes the formation of cell-cell contacts with functional junctional complexes. (Gonzalez-Mariscal et al. 1985; Rajasekaran et al. 1996; Yamamoto et al. 1997). Immunofluorescent analysis of aPKC, ASIP/PAR-3 and PAR-6 in polarized MDCK cells has revealed that all co-localize in the cell-cell contact region with a tight junction (TJ) marker, ZO-1 (Izumi et al. 1998; Suzuki et al. 2001). As shown in Fig. 1A, calcium depletion results in a disappearance of the specific staining for the three proteins at the cell surface (calcium switch; time = 0). However, after calcium is restored, the stainings in the cell-cell contact region start to appear within 20 min and are complete within 3 h. Although we could not perform a double labelling analysis of these proteins, their complete co-localization with ZO-1 supports the idea that they co-localize at cell-cell junctional regions from the very early stage of junctional formation. Interestingly, the confocal z-sectional analysis indicates that staining starts to appear on the basal side within 20 min and reaches the most apical tip within 60 min after cell-cell contact (Fig. 1B; data not shown).

Figure 1.

PAR-6, aPKCλ and ASIP are targeted to the cell-cell contact region in a cell adhesion-dependent manner. (A) Gradual targeting of PAR-6, aPKCλ and ASIP to the cell-cell contact region after calcium switch. Confluent MDCK cells were incubated in a low calcium medium for 20 h, and subjected to calcium switch by replacing the medium with normal calcium medium. The cells were fixed with 2% paraformaldehyde at 0, 20, 60 and 180 min after calcium switch. Cells were doubly stained with antibodies for PAR-6, aPKCλ (λ1), or ASIP (red), together with that for ZO-1 (green). Merged views for all double stainings are shown. Bar = 25 µm (B) Confocal z-sectional view of the MDCK cells shown in (A) doubly stained with anti-PAR-6 (red) and anti-ZO-1 (green). Merged views are shown. The arrowheads indicate the position of the basal membrane. Bar = 10 µm (C) Identification of the PAR-6-aPKCλ-ASIP complex during the early stages of cell polarization. Confluent MDCK cells (10 cm dish × 4) were grown in normal growth medium for 20 h (normal calcium), or in low calcium medium for 20 h, followed by incubation in normal calcium medium (calcium switch) for 0, 20, 60 and 180 min. The cell lysates (Sup; total 7 mg protein) were processed for immunoprecipitation (IP) with 3 µg of GW2AP antibody or control IgG as indicated at the top. Co-immunoprecipitated proteins were analysed using the anti-aPKCλ (ι) mAb or anti-ASIP pAb.

Taken together with our previous demonstration of the presence of a PAR-6-aPKC-ASIP/PAR-3 ternary complex in polarized MDCK cells (Suzuki et al. 2001), the results suggest the intriguing possibility that the ternary complex is formed in response to cell-cell contact. To test this possibility, we examined the components in PAR-6 immunoprecipitates during the early stage of cell polarization. Supernatant fractions taken from cells at the times indicated were used as starting materials from which PAR-6 and associated proteins, ASIP/PAR-3 and aPKCλ were immunoprecipitated. As shown in Fig. 1C, the amounts of co-precipitated ASIP and aPKCλ did not change significantly through the time when cell-cell junctional structures were gradually formed (0–180 min). Notably, even in the absence of calcium, almost the same amounts of aPKCλ and ASIP bind PAR-6. These results suggest that PAR-6 forms a stable protein complex through the disappearance and reformation of cell-cell contact induced by calcium switch, and that the cell-cell contact recruits the preformed protein complex to the cell-cell contact region.

Over-expression of PAR-6 mutant affects the cell-cell contact-induced targeting of TJ proteins, ZO-1, Claudin-1 and Occludin and inhibits TER development

We recently found that the adenovirus-mediated introduction of a dominant negative mutant of aPKC (aPKC kn) suppresses the cell-cell contact-induced polarization of MDCK cells (Suzuki et al. 2001). To our knowledge, this adenovirus-mediated gene transfer technique is the best way to express ectopic proteins transiently in confluent MDCK cells without any artificial effects on the cell-cell junctional structures; other methods, such as lipofection, make it difficult to evaluate beyond the background cell damage, especially without statistical analysis. In addition, these methods are difficult to apply to confluent monolayers of epithelial cells. Thus, to examine whether PAR-6 is also involved in the maintenance and/or establishment of epithelial cell polarity, we constructed an adenovirus vector encoding T7-tagged PAR-6 or its mutant, introduced them into confluent MDCK cells, and evaluated the effects on junctional structures. In this study, the LacZ expression vector was used as a negative control. As shown in Fig. 2B, more than 80% of the cells are positive for anti-T7 antibody. In this condition, we firstly examined the effects of the over-expression of PAR-6 or its mutant on restoration of TJ barrier function after calcium switch by measurement of TER. As shown Fig. 2A, over-expression of PAR-6 ΔaPKCBD suppresses the TER development after calcium switch, whereas only a small effect was observed for PAR-6 wt. We next examined the localization of ZO-1, a TJ maker, in PAR-6 or its mutant over-expressing cells. When introduced and expressed in polarized MDCK cells, PAR-6 wt, as well as PAR-6 ΔaPKCBD, does not affect the localization of ZO-1 (Fig. 2B,D). However, as observed for an aPKC kn (Suzuki et al. 2001), cells that over-express PAR-6 ΔaPKCBD but not PAR-6 wt frequently (in about 13% of cells) show a mislocalization of ZO-1 if the cell-cell adhesion is subjected to the dissociation–reassociation process by calcium-switch (Fig. 2C,D). Only a small effect was observed when the LacZ expression vector was employed (Fig. 2D), indicating that this effect of ectopic PAR-6 mutant is not due to artificial effects arising from the adenovirus infection. These results are consistent with the above data on TER measurement (Fig. 2A), indicating that over-expressed PAR-6 ΔaPKCBD interferes with the TJ development process rather than its maintenance. In PAR-6 ΔaPKCBD over-expressing cells, most cells that lost proper ZO-1 staining also lost staining for TJ membrane proteins, claudin-1 and occludin, at the cell-cell contact (Fig. 2E,F). On the other hand, E-cadherin accumulation at the cell-cell contact region is less affected (Fig. 2E,F). Taken together, these results indicate that over-expression of PAR-6 ΔaPKCBD affects the localization of the TJ proteins without causing a complete disruption of cell-cell adhesion.

Figure 2.

Over-expression of a C-terminal mutant lacking the aPKC-BD of PAR-6 affects the junctional localization of TJ proteins and TER development. (A–C) MDCK cells grown on filters were infected with an adenovirus vector encoding T7-PAR-6 wt, T7-PAR-6 ΔaPKCBD or LacZ, and then cells were cultured with or without calcium switch. (A) The effects of ectopic expression of indicated proteins on the TER development after calcium switch. (B and C) Immunostaining of MDCK cells over-expressing T7-PAR-6 wt or ΔaPKCBD with antibodies against T7 (Omni)-tag (green) and ZO-1 (red) (B; –calcium switch, C; +calcium switch). (D) The percentage of cells that lost proper ZO-1 staining was quantified (total 1000 cells). (E and F) MDCK cells grown on glass coverslips were infected with an adenovirus vector encoding T7-PAR-6 ΔaPKCBD, and then cells were subjected to calcium switch. (E) Comparison of the localization of junctional proteins (indicated in the right panel) with that of ZO-1 (left panel). (E) The percentage of cells that lost proper staining of the indicated proteins in cells showing aberrant ZO-1 staining was quantified. Values shown in (A), (D) and (F) are means (± SD) of three independent experiments. Bars (B, C and E) = 20 µm.

Although somewhat weaker, the effects caused by over-expression of PAR-6 ΔaPKCBD are quite similar to the effects caused by aPKC kn over-expression (Suzuki et al. 2001). In addition, the junctional localization of aPKC and ASIP/PAR-3 is also disrupted in PAR-6 mutant over-expressing cells (Fig. 2E,F). Taken together with our previous results—that aPKC-kn over-expression disrupts the junctional localization of PAR-6 and ASIP/PAR-3—these results show an evolutionarily conserved coincidence with the C. elegans aPKC-PAR system, in which aPKC, PAR-3 and PAR-6 show interdependence on their asymmetric localizations to the anterior periphery. Considering the physical interaction and concerted targeting to the cell-cell junction of PAR-6 and aPKC in an early stage of cell polarization, these results suggest that PAR-6 is involved in a aPKC function on TJ development.

PAR-6 and Cdc42 act in cooperation to regulate aPKC activity

The fact that the PAR-6 mutant which is unable to interact with aPKC and ASIP exerts a similar effect with a dominant negative mutant of aPKC on TJ biogenesis, further suggests the possibility that over-expression of the C-terminal region of PAR-6 interferes with the aPKC signalling pathway required for epithelial junctional formation. This region of PAR-6 contains the well-conserved CRIB/PDZ domain that interacts directly with the GTP-bound form of Cdc42/Rac1 (Joberty et al. 2000; Johansson et al. 2000; Lin et al. 2000; Noda et al. 2001; Qiu et al. 2000). Figure 3 shows that PAR-6 immunoprecipitates from MDCK cells that stably over-express tag-PAR-6 contain endogenous Cdc42 and Rac1, as well as aPKCλ and ASIP. This is the first in vivo evidence for a molecular interaction between the PAR-6-aPKC-ASIP/PAR-3 ternary complex and Cdc42/Rac1. Taken together, these results suggest the possibility that PAR-6 functions as a linker molecule between the activated form of Cdc42/Rac1 and the ternary complex, which mediates the aPKC signalling to epithelial cell polarization. To test this possibility, we examined Cdc42 activating aPKC through PAR-6 as a linker molecule. For this purpose, we over-expressed aPKCλ together with T7-tagged wild-type PAR-6 or mutant in COS1 cells, and monitored the kinase activity that co-precipitated with PAR-6 in the presence or absence of the activated form of recombinant Cdc42. As shown in Fig. 4A, the kinase activity of the PAR-6 immunoprecipitate was specifically activated (≈2.3-fold) by the addition of GTP-γS-pre-incubated GST-Cdc42 (lane 3). Neither GST (pre-incubated with GDP or GTP-γS) nor GDP-pre-incubated GST-Cdc42 showed such effects (lanes 1, 2 and 4). Since the kinase activity of the PAR-6 precipitate was significantly reduced, and its dependence on Cdc42 disappeared when a kinase deficient mutant of aPKCλ (aPKCλ kn) was used instead of wild-type aPKCλ (lanes 5–8), we conclude that the above results reflect the activation of aPKCλ by an activated form of Cdc42. Next, we co-expressed and precipitated a PAR-6 mutant (PAR-6 ΔCRIB/PDZ) lacking the CRIB/PDZ domains that are indispensable for binding with Cdc42/Rac1 (Joberty et al. 2000), instead of wild-type of PAR-6. In this case, no Cdc42 dependence of aPKCλ activity was detected (lanes 9–12), indicating that the Cdc42-dependent activation of aPKCλ is mediated by the interaction between Cdc42 and PAR-6. This is further supported by the results shown in Fig. 5 that the kinase activity in the aPKCλ immunoprecipitate was not affected by the presence of activated Cdc42 if PAR-6 was not co-expressed (Fig. 5, lanes 1–4). Taken together, these results strongly suggest that Cdc42 can activate aPKC kinase activity by binding with the CRIB/PDZ domains of PAR-6.

Figure 3.

In vivo interaction of Cdc42/Rac1 with over-expressed PAR-6. An MDCK stable clone that inducibly over-expresses Flag-PAR-6 (wt) was grown in culture medium in the presence (induction–) or absence (induction+) of 20 ng/mL of Doxycycline for 3 days to form a confluent monolayer. The cell lysates (Sup) were processed for immunoprecipitation (IP) with anti-Flag mAb, and examined for the co-precipitation of endogenous Cdc42, Rac1, aPKCλ and ASIP.

Figure 4.

Cooperative regulation of aPKC activity by PAR-6 and Cdc42. (A) COS cells were co-transfected with expression vectors encoding T7-PAR-6 (wt or ΔCRIB/PDZ) and aPKCλ (wt or kn). Then T7-PAR-6 was immunoprecipitated with anti-T7 mAb. After incubation for 20 min on ice with GST or GST-Cdc42 preloaded with GTP-γS or GDP, the kinase activity of the immunoprecipitates was assayed by adding MBP and [γ-32P] ATP. γ-32P incorporation into MBP was examined by SDS-PAGE followed by autoradiography. Values shown are the means (± SD) of MBP phosphorylation, normalized to the amount of aPKC, obtained in three (lanes 1–8) or two (lanes 9–16) independent experiments. The value (lane 2) was taken as 1. (B) A hypothetical model for the regulation of aPKC activity by Cdc42 through PAR-6 (see text for details).

Figure 5.

The effects of PAR-6 or its mutant on aPKC activity. COS cells were co-transfected with expression vectors encoding T7-aPKCλ wt and Flag-PAR-6 (wt or ΔCRIB/PDZ). Then T7-aPKCλ was immunoprecipitated with anti-T7 mAb, and the effects of Cdc42 on the kinase activity of the immunoprecipitates was evaluated as described in the legend to Fig. 4A. Values shown are the means (± SD) of MBP phosphorylation, normalized to the amount of aPKC, obtained in three independent experiments. The value (lane 2) was taken as 1.

Unexpectedly, the kinase activity of the PAR-6 ΔCRIB/PDZ immunoprecipitate is significantly higher than that of the PAR-6 wild-type. Even if normalized for the amount of precipitated aPKCλ, the PAR-6 ΔCRIB/PDZ immunoprecipitate shows a greater than 3.5-fold higher kinase activity than that in the PAR-6 wt-immunoprecipitate (Fig. 4A, lanes 9–12). This increase is not due to the co-precipitation of unknown kinases with PAR-6 ΔCRIB/PDZ because the enhancement was not observed when aPKCλ kn was used (lanes 13–16). Therefore, these results suggest that the N-terminal region of PAR-6 has an intrinsic ability to activate aPKCλ kinase activity, which is usually suppressed by the presence of CRIB/PDZ domains. One role of the binding of the activated form of Cdc42 to the CRIB/PDZ domain of PAR-6 may be to relieve this suppression (Fig. 4B). The aPKCλ activity in the PAR-6 ΔCRIB/PDZ immunoprecipitate is even higher than the Cdc42-activated aPKCλ activity observed in the PAR-6 immunoprecipitate (about 1.5-fold; compare lanes 3 and 11). However, this may be due to the fact that the recombinant Cdc42 used in this study is a lipid-unmodified protein. Generally, lipid-unmodified small G protein shows a weaker interaction with its target molecule than a lipid-modified one.

In Fig. 5, we performed similar experiments to those shown in Fig. 4 to evaluate the effects of PAR-6 and Cdc42 compared with the aPKCλ basal activity. Here, T7-tagged aPKCλ and Flag-tagged PAR-6 were co-expressed in COS1 cells, and the over-expressed T7-tagged aPKCλ was immunoprecipitated instead of PAR-6. In this assay, the precipitated aPKCλ does not necessarily form a complex with PAR-6, but if appropriate conditions are chosen (cDNA amount for transfection), the effects of PAR-6 on aPKC can be evaluated. The results reveal that the co-expression of PAR-6 with aPKCλ reduces the kinase activity by ≈ 40%, while the addition of activated Cdc42 restores the activity to levels observed with aPKCλ alone (lanes 5–8). On the other hand, co-expression of PAR-6 ΔCRIB/PDZ increases the aPKCλ activity by about 1.2-fold, independently of the addition of activated Cdc42 (lanes 9–12), although the amount of PAR-6 ΔCRIB/PDZ that co-precipitates with aPKCλ is similar to that of PAR-6 wt (data not shown). These results are consistent with those in Fig. 4A, and suggest that aPKC kinase activity may be suppressed by the interaction with PAR-6. The association between the activated form of Cdc42 and the CRIB/PDZ domain of PAR-6 might release this suppression, resulting in the restoration rather than the potentiation of aPKC activity.


In the present study, we provide evidence supporting the functional importance of the PAR-6-aPKC-ASIP/PAR-3 ternary complex during the early stages of polarization in epithelial MDCK cells. Firstly, this ternary complex exists even in cells without cell-cell contacts and thus without cell polarity, and it is recruited to the cell-cell contact region following the calcium-induced formation of cell-cell contacts that triggers cell-polarization (Fig. 1). Secondly, the over-expression of PAR-6 mutant by adenovirus-mediated gene transfer results in the mislocalizations of the TJ associated proteins ZO-1, claudin-1 and occludin, as well as aPKC and PAR-3/ASIP and suppression of TJ barrier function, in a cell-cell adhesion-dependent manner (Fig. 2). Taken together with our previous demonstration of the dominant negative effect of aPKC-kn on the process of epithelial cell polarization but not on its maintenance (Suzuki et al. 2001), the present results suggest that the early stage of cell polarization induced by the establishment of cell-cell contact is the major target of the PAR-6-aPKC-ASIP/PAR-3 ternary complex.

Our present results also suggest that PAR-6 cooperates together with Cdc42/Rac1 to regulate aPKC activity in the polarization of epithelial cells. Firstly, the C-terminal region with CRIB/PDZ domains of PAR-6 affects the establishment of the junctional structures of MDCK cells (Fig. 2). Secondly, endogenous Cdc42/Rac1 interacts with the PAR-6-aPKC-ASIP/PAR-3 ternary complex in MDCK cells (Fig. 3). Thirdly, Cdc42 activates aPKC kinase activity through its protein–protein interaction with aPKC by way of PAR-6 (Fig. 4). The present in vitro kinase assays using recombinant Cdc42 exclude the possibility that Cdc42 merely activates aPKC by recruiting it to the membrane where a direct aPKC activator is present. As for the activation mechanism of aPKC, the direct binding of lipids such as phosphatidyl serine or phosphatidyl inositol phosphates, as well as direct phosphorylation, probably by PDK1, have been suggested (Chou et al. 1998; Le Good et al. 1998; Nakanishi et al. 1993). Based on the finding that PAR-6 ΔCRIB/PDZ can potentiate aPKC activity by about 3.5-fold compared with full-length PAR-6 (Fig. 4A), we conclude that the conserved N-terminal region of PAR-6, which interacts with the N-terminal region of aPKC, has an intrinsic ability to activate aPKC. Thus, we propose a model in which the presence of the CRIB/PDZ domain in the PAR-6 molecule inhibits this ability of the N-terminal region of PAR-6 due to an intramolecular interaction, and the GTP-dependent binding of Cdc42 to the CRIB/PDZ domain may release this inhibition (Fig. 4B).

Here, we also suggest the possibility that the binding of full-length PAR-6 to aPKC suppresses aPKC activity (Fig. 5). This suppression of aPKC by PAR-6 might be a mechanism by which aPKC activity is suppressed in the PAR-6-aPKC-ASIP/PAR-3 complex in the cytosol of cells that have lost cell polarity. Then, cell-cell adhesions might lead to the activation of aPKC by the interaction of the PAR-6-aPKC-ASIP/PAR-3 ternary complex with activated Cdc42. In support of this, Cdc42 has been shown to be activated by E-cadherin-mediated cell-cell adhesions in epithelial cells (Kim et al. 2000).

More recently, the depletion of Cdc42 by RNAi has been shown to result in defects in asymmetric cell division in the C. elegans early embryo (Gotta et al. 2001; Kay & Hunter 2001; Jantsch-Plunger et al. 2000). Moreover, Cdc42 interacts with PAR-6 and regulates the localization of PAR proteins, including PAR-6 and PAR-3 (Gotta et al. 2001; Kay & Hunter 2001). These observations indicate the involvement of Cdc42 in aPKC-PAR signalling in C. elegans, and support our present results obtained in mammalian epithelial cells. In the PAR-6-aPKC-ASIP/PAR-3 ternary complex in epithelial cells, PAR-6 may mediate the upstream signal from Cdc42/Rac1 to aPKC and regulate aPKC activity, whereas ASIP/PAR-3, with three PDZ domains, may function as an adapter molecule targeting the ternary complex to the cell-cell contact region. In fact, we recently found that PDZ1 in ASIP can bind to the C-terminal region of JAM, a TJ membrane protein belonging to IgG superfamily (Ebnet et al. 2001), suggesting that JAM is a membrane protein that anchors this signalling complex to TJ.

Recently, Qui et al. showed that PAR-6 links Rac1/Cdc42 to aPKC in the signalling pathway leading to cell transformation in NIH-3T3 fibroblasts (Qiu et al. 2000), and Noda et al. (2001) showed that PAR-6 mediates the localization of aPKC to the Rac-induced membrane ruffles in HeLa cells. In addition, we recently confirmed that the PAR-6-aPKC-ASIP/PAR-3 ternary complex also exists in these cells (our unpublished results). Therefore, the evolutionarily conserved signalling complex, the PAR-6-aPKC-ASIP/PAR-3 ternary complex, may act more widely than expected.

Experimental procedures

Expression vectors

Human PAR-6 expression vectors encoding PAR-6 wt, a full-length PAR-6 (1–346 amino acids), PAR-6 ΔaPKCBD, lacking amino acids 1–125, corresponding to the aPKC-binding domain, and PAR-6 ΔCRIB/PDZ, lacking amino acids 126–258 corresponding to the CRIB and PDZ domains, has been previously described (Suzuki et al. 2001). Expression vectors encoding aPKCλ wt and aPKCλ kn, a kinase-deficient mutant, have been previously described (Akimoto et al. 1994, 1996; Izumi et al. 1997; Ohno et al. 1994). Cdc42/pGEX for GST-Cdc42 was from Drs T. Sasaki and Y. Takai (Osaka University, Osaka, Japan). The adenovirus expression vectors were generated using cosmid vector, pAxCAwt, as previously described (Miyake et al. 1996).


Anti-human PAR-6 pAb (GW2AP), anti-ASIP pAb (C2–3AP), and anti-aPKCλ (λ1) pAb have been previously described (Akimoto et al. 1994; Izumi et al. 1998; Suzuki et al. 2001). Anti-aPKCλ (ι) mAb, anti-E-cadherin mAb, anti-Cdc42 mAb and anti-Rac1 mAb (Transduction Laboratories); anti-ZO-1 mAb/pAb, anti-Claudin-1 pAb and anti-Occludin pAb (ZYMED Laboratories Inc); anti-T7 mAb (Novagen); anti-T7 (Omni) pAb (Santa Cruz); anti-Flag mAb (M2) (Sigma)

Cell culture and adenovirus infection

COS and MDCKΙΙ cells were grown in DMEM containing 10% FCS, penicillin and streptomycin under an air−5% CO2 atmosphere at constant humidity. Calcium switch assay and adenovirus infection were performed as previously described (Suzuki et al. 2001). Briefly, MDCKΙΙ cells were seeded on 10 mm diameter Transwell filters (Corning Coaster Corp.) or 12 mm round glass coverslips and grown in normal calcium medium to form a confluent monolayer. For adenovirus infection, the cells were incubated for 2 h with 200 µL of the appropriate virus vector diluted to 3 × 108 pfu/mL. When the calcium switch was performed, the cells were incubated for 20 h with low calcium medium containing 3 µm calcium as previously described (Stuart et al. 1994), and then transferred to normal calcium medium. When the calcium switch was omitted, the cells were incubated in normal calcium medium after virus infection. Transepithelial electrical resistance (TER) was measured as previously described (Suzuki et al. 2001). MDCK Tet-Off stable cell lines inducibly expressing Flag-PAR-6 were produced as described (Jou & Nelson 1998) using a Tet-off Gene Expression System (Clontech Laboratories, Inc.).

Immunoprecipitation and Western blot analysis

MDCK II cells cultured in 10 cm dishes were suspended in 250 µL of lysis buffer containing 20 mm Hepes, pH 7.5, 150 mm NaCl, 1 mm EDTA, 50 mm NaF, 1 mm Na3VO4, 10 µg/mL leupeptin, 1 mm PMSF, 1.8 µg/mL aprotinin and 1% Triton X-100. After 30 min incubation on ice, the lysates were clarified by centrifugation at 14 000 r.p.m. for 30 min and incubated with antibodies preabsorbed on Protein G-Sepharose (Amersham Pharmacia Biotech) for 1 h at 4 °C. After washing four times with lysis buffer, the immunocomplexes were eluted with SDS sample buffer for SDS-PAGE (Laemmli 1970). Western blot analysis was performed as previously described (Suzuki et al. 2001). Antibodies were detected by chemiluminescence ECL (Amersham Pharmacia Biotech).

In vitro protein kinase assays

COS cells in 10 cm dishes were transfected with expression vectors encoding T7-tagged PAR-6 or aPKCλ by electroporation (Gene Pulser; Bio-Rad Laboratories). The cells were then suspended in 200 µL of lysis buffer containing 20 mm Tris-HCl, pH 7.5, 0.25 m Sucrose, 1.2 mm EDTA, 20 mmβ-ME, 150 mm NaCl, 1 mm NaF, 1 mm Na3VO4, 1 mm sodium pyrophosphate, 1% Triton X-100, 0.5% NP40, 10 µg/mL leupeptin, 1 mm PMSF and 1.8 µg/mL aprotinin, and then T7-tagged PAR-6 or aPKCλ was immunoprecipitated with anti-T7 antibody. The immunoprecipitates in 40 µL of phosphorylation buffer (20 mm Tris-HCl, pH 7.5, 5 mm MgCl2, 1 mm EGTA) were pre-incubated with 0.6 µm GST or GST-Cdc42 preloaded with GTP-γS or GDP for 20 min on ice. Then the phosphorlation reaction was started by adding 10 µL of substrate mix containing 1 µm Myelin Basic Protein (MBP) (Sigma) and 40 µm ATP (2 µCi/50 µL [γ-32P] ATP). After 20 min at 30 °C, the reaction was stopped by the addition of SDS sample buffer and the products were subjected to SDS-PAGE, autoradiography and Western blotting. For the preparation of preloaded GST or GST-Cdc42, 10 µm GST or GST-Cdc42 was incubated in loading buffer (20 mm Tris-HCl, pH 7.5, 10 mm EDTA, 1 mm DTT, 5 mm MgCl 2, 40 µm GTP-γS or GDP) for 30 min at 30 °C, after which MgCl2 was added to a final concentration of 15 mm.

Immunofluorescence microscopy

MDCK II cells plated on glass coverslips or filters were fixed with 2% paraformaldehyde, and stained as previously described (Suzuki et al. 2001). The samples were then mounted with Vectashield (Vector Laboratories, Inc.) and viewed with a Confocal microscope system (Microradiance; Bio-Rad laboratories).


We thank Drs Takuya Sasaki and Yoshimi Takai (Osaka University) for the construction of GST-Cdc42, and Drs Mutsuki Amano and Kozo Kaibuchi (Nara Institute of Science and Technology) for technical advice. This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (S. Ohno) and the Japanese Society for the Promotion of Science (S. Ohno and T. Yamanaka).


  1. Communicated by: Yoshimi Takai