Evidence for a Functional Interaction between Cingulin and ZO-1 in Cultured Cells*

Cingulin, a protein component of the submembrane plaque of tight junctions (TJ), contains globular and coiled-coil domains and interacts in vitro with several TJ and cytoskeletal proteins, including the PDZ protein ZO-1. Overexpression of Xenopus cingulin in transfected Xenopus A6 cells resulted in the disruption of endogenous ZO-1 localization, suggesting that cingulin functionally interacts with ZO-1. GlutathioneS-transferase pull-down experiments showed that a conserved ZO-1 interaction motif (ZIM) at the NH2 terminus of cingulin is required for cingulin-ZO-1 interaction in vitro. An NH2-terminal region of cingulin, containing the ZIM, was sufficient, when fused to coiled-coil sequences, to target transfected cingulin to junctions. However, deletion of the ZIM did not abolish junctional localization of transfected cingulin in A6 cells, suggesting that cingulin can be recruited to TJ through multiple protein interactions. Interestingly, the ZIM was required for cingulin recruitment into ZO-1-containing adherens junctions of Rat-1 fibroblasts, indicating that cingulin junctional recruitment does not require the molecular context of TJ. Cingulin coiled-coil sequences enhanced the junctional accumulation of expressed cingulin head region in A6 cells, but purified recombinant cingulin did not form filaments under physiological conditions in vitro, suggesting that the cingulin coiled-coil domain acts primarily by promoting dimerization.

The ability of vertebrate organisms to develop a complex body organization depends critically on the creation and maintenance of separate body, organ, and tissue compartments. This, in turn, depends on the ability of epithelia and endothelia lining different body surfaces to form semipermeable barriers that prevent the free diffusion of molecules and cells between extracellular compartments. The structure that allows epithelia and endothelia to form a semipermeable barrier is the tight junction (TJ), 1 a gasket-like seal surrounding the apicolateral region of polarized cells. TJ also define the separation between apical and basolateral domains of the plasma membrane that show different composition and function. Thus, TJ are essential for separation of tissue compartments and maintenance of epithelial cell polarity. The structure and function of TJ vary depending on type of tissue and physiological and pathological states, suggesting that changes in the expression and/or function of different TJ components underlie such modulation.
Much progress has been made in recent decades in elucidating the molecular organization of TJ. TJ comprise membrane and submembrane (cytoplasmic plaque) domains. The membrane domain of TJ contains transmembrane Ig-like proteins and four-membrane pass proteins important for cell-cell adhesion and selective paracellular permeability (see Refs. 1 and 2 for reviews). The cytoplasmic plaque of TJ contains a large number of proteins, including PDZ-containing, signaling, and cytoskeletal proteins (2,3). TJ plaque proteins probably play an important role in epithelial tissue formation, as suggested by the observation that mice lacking the plaque protein AF-6 die early during development, with epithelial cells losing their polarity (4,5). TJ plaque proteins may recruit and regulate membrane proteins and function as targets and effectors of signaling pathways involved in control of cell differentiation and growth (reviewed in Refs. 2 and 3). Recent experiments highlighted the role of the evolutionarily conserved Par-3-Par-6-atypical protein kinase C complex in establishing epithelial polarity and TJ assembly in mammalian systems (6 -9). However, the molecular steps leading to the formation of TJ and the role of most TJ plaque proteins in TJ function are not known.
Cingulin is a non-PDZ TJ plaque protein predicted to be a parallel dimer of two polypeptides, each consisting of a globular "head" region, a coiled-coil "rod" region, and a small globular tail (10 -12). Cingulin forms a complex in vivo with the PDZ proteins ZO-1 (13)(14)(15) and ZO-2 (16,17), and an NH 2 -terminal head fragment of Xenopus cingulin interacts in vitro with high affinity with ZO-1 (12). In addition, cingulin interacts in vitro with ZO-2, ZO-3, myosin, AF-6, actin, and cingulin itself (12,18). However, it is not clear whether cingulin interaction with ZO-1 or any other protein plays a significant role in the assembly and maintenance of the cytoplasmic plaque of TJ in vivo. Furthermore, the precise cingulin sequence(s) and protein interactions required for its junctional assembly are not known. Previously, we showed that full-length Xenopus cingulin, but not NH 2 -terminal and COOH-terminal fragments, is targeted to junctions in transfected Madin-Darby canine kidney cells, indicating that sequences within both head and rod domains are required for TJ localization in this model system (12). In this paper, we analyzed in more detail the cingulin sequences important for junctional targeting in vivo, by transient transfection of Xenopus cingulin mutants into cultured Xenopus epithelial (A6) cells. In addition, we used GST pull-down assays to map cingulin sequences involved in its interaction with ZO-1 and ZO-2. We noted that cingulin overexpression affected ZO-1 localization in A6 cells, and we identified a conserved NH 2 -terminal sequence of cingulin, which is important in junctional recruitment and in vitro interaction with ZO-1.
The cDNAs coding for canine ZO-2 (a gift of Dr. D. Goodenough, Harvard University) and human ZO-1 (a gift of Dr. J. Anderson, Yale University) were used for in vitro transcription and translation as described (12). Baculovirus stock for the production of full-length mouse ZO-1 and mouse ZO-2 were gifts of Dr. Shoichiro Tsukita (University of Kyoto) and Dr. Bruce Stevenson (University of Alberta), respectively. An expressed sequence tag clone containing a fragment of mouse cingulin cDNA was obtained from the Primary Data base of the German Human Genome Project (RZPD, Berlin, Germany) and completely sequenced. 2 Baculovirus-mediated Expression of Cingulin and ZO-2 in Insect Cells-Constructs for insect cell expression of untagged or His-tagged full-length Xenopus cingulin were generated in pFASTBAC1 or pFAST-BACHTa (Invitrogen). Bacmid isolation, viral stock production, and preparation of lysates was as described in Ref. 19. His-tagged Xenopus cingulin and mouse ZO-2 were purified from lysates of insect cells extracted in 300 mM NaCl, 50 mM NaH 2 PO 4 , 10 mM imidazole, pH 8.0. Lysates were clarified by centrifugation at 100,000 ϫ g and loaded onto Ni 2ϩ -nitrilotriacetic acid superflow resin (Qiagen). Purified protein was eluted with 250 mM imidazole.
GST Pull-down Assays and Measurement of K d Values-Constructs for expression of GST fusion proteins in bacteria were generated by subcloning fragments amplified by PCR into pGEX-4T (Amersham Biosciences). Bacterial expression of GST fusion proteins and GST pulldown assays were as described (12). Fusion proteins were quantitated by reference to a bovine serum albumin standard curve, and each assay contained the same amount of fusion protein. To measure dissociation constants (K d values), ϳ0.75 g of cingulin GST-head fusion protein (containing residues 1-378 of Xenopus cingulin) or 2.0 g of cingulin GST-rod fusion protein (containing residues 851-1368 of Xenopus cingulin) bound to the glutathione-Sepharose resin were incubated with increasing amounts of target protein (1-15 g) present in insect cell lysate (Xenopus cingulin) or purified by affinity chromatography (mouse ZO-2). As a control, 2.0 g of GST was conjugated to beads and incubated with the same target proteins. After washing, proteins bound to the beads were analyzed by gel electrophoresis and quantitated by scanning Coomassie-stained bands in gels containing BSA standard (12).
Cell Culture and Transfections-Xenopus A6 renal epithelial cells were cultured in modified L-15 medium containing 10% fetal bovine serum, and antibiotics (20). Rat-1 fibroblasts were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Constructs for transient transfections were generated in the vector pcDNA3.1Myc/His (Invitrogen) either by using convenient restriction sites or by cloning restriction fragments or PCR products in frame with the Myc tag. The sequences of regions amplified by PCR were controlled by dideoxy sequencing. Construct ⌬(1-40, 50 -1038) was generated by ligating oligonucleotides coding for residues 41-55 (containing XbaI-BamHI sites at their 5Ј-and 3Ј-ends, respectively) with the DNA generated by digestion of full-length, Myc-tagged Xenopus cingulin with XbaI and BamHI. A6 cells and Rat-1 fibroblasts were transiently transfected at 50 -75% density using the TransIT transfection reagent (Myrus Corp.), and the protocol recommended by the manufacturers. After 1-2 days in culture, cells were fixed in phosphate-buffered saline containing 3% paraformaldehyde and 0.3% Triton and immunofluorescently stained with anti-Myc and anti-ZO-1 (followed by fluorescein isothiocyanate anti-mouse and TRITC anti-rabbit). At least 200 cells for each construct (from a minimum of 2 different experiments) were scored for localization of the expressed protein. Cells were scored as having "junctional" localization when the Myc labeling was precisely co-localized with the ZO-1 labeling along the cell junction(s), and no significant cytoplasmic labeling was detected. These cells in most cases expressed relatively low levels of the transfected protein. Cells were scored as displaying "junctional and cytoplasmic" labeling when Myc labeling was detected not only accumulated at the cell periphery in association with ZO-1 but also in the cytoplasm. The relative intensities and amounts of ZO-1-associated and cytoplasmic Myc labeling were variable depending on the type of construct and level of expression. Higher levels of expression of the transfected protein typically correlated with stronger cytoplasmic Myc staining. "Cytoplasmic" labeling indicated that Myc staining was only detectable in the cytoplasm and in some cases in the nucleus but never in association with ZO-1 in those cells that expressed the transfected protein and also displayed ZO-1 junctional labeling. Cells were scored as displaying "ZO-1 redistribution" when high levels of expression of the transfected protein correlated with a loss of the discrete linear junctional staining pattern of ZO-1, and the appearance of diffuse, less intense (yet clearly above background) subjunctional ZO-1 staining. These cells were always a subset of the "junctional and cytoplasmic" or "cytoplasmic" group.
In Vitro Assembly of Purified Cingulin and Myosin-Purified recombinant Xenopus cingulin (18) was dialyzed for 1 h at 4°C against 50 mM sodium acetate, pH 5.0, 5 mM DTT. After dialysis, samples were applied onto carbon collodion-coated grids, washed with 0.24 M ammonium formate, stained with 1% uranyl acetate, and observed with a Phillips EM 410 electron microscope.
To study the effect of ionic strength and pH on cingulin assembly, purified protein was dialyzed for 1 h at 4°C against 5 mM MgCl 2 , 0.5 mM EGTA, 0.1 mM DTT, 10 mM imidazole at different pH values (6.5, 7.5, and 8.5) and NaCl concentrations (5, 50, 150, or 300 mM). After dialysis, the turbidity measurements at 340 nm were performed as described (21). The data obtained for pH values of 6.5, and 8.5 were similar to those obtained at pH 7.5 at all ionic strengths tested.  (Fig. 1A, inset). Anti-ZO-1 antibodies cross-react with one polypeptide of apparent M r ϳ220,000 in A6 cell lysates (Fig. 1B, inset), suggesting that amphibian and mammalian ZO-1 share not only junctional localization and epitopes but also molecular size.

Overexpression of Full-length Xenopus Cingulin in Xenopus
Myc-tagged Xenopus cingulin was transfected into A6 cells, and cells were stained with anti-Myc and anti-ZO-1 antibodies to detect expressed cingulin and endogenous ZO-1, respectively. When cingulin was expressed at low levels, it was colocalized with ZO-1 in a thin linear pattern along junctional regions ( Fig. 1, C and CЈ, arrowhead). In these cells, ZO-1 localization was identical to that of nonexpressing cells. When the exogenous cingulin was expressed at high levels, it accumulated in a broad subjunctional area extending from the cell border into the peripheral cytoplasm ( Fig. 1, C and CЈ, double arrows). These cells did not show abnormal morphology and appeared to maintain contact with adjacent, low expressing cells. However, the cell border staining of ZO-1 was disrupted and appeared as a diffuse staining that was colocalized with the cingulin anti-Myc staining (Fig. 1, C and CЈ, double arrows). Thus, cingulin overexpression disrupts the localization of ZO-1 in A6 cells.
Shared Regions in Xenopus and Human Cingulin Are Involved in Interaction with Mammalian ZO-1 and ZO-2 in Vitro-Cingulin forms a complex in vivo with ZO-1 and ZO-2 (12). Thus, we mapped the sequences important for the in vitro interaction of Xenopus cingulin with full-length ZO-1 and ZO-2, with the perspective of testing their role in junctional recruitment in vivo. Since Xenopus cDNAs for ZO-1 and ZO-2 are not available, we used mammalian ZO-1 and ZO-2, expressed in reticulocyte or baculovirus systems. To verify the significance of the in vitro mapping data, we used both Xenopus and mammalian cingulin in GST pull-down experiments with mammalian ZO-1 and ZO-2.
Deletion mutants of the Xenopus and human cingulin head domains were expressed in bacteria as GST fusion proteins and used as baits to pull down human ZO-1 from reticulocyte lysates and mouse ZO-1 from insect cell lysates. As shown in Fig.  2A, fusion proteins lacking residues 41-55 of Xenopus cingulin did not interact with ZO-1, indicating that this sequence is required for cingulin-ZO-1 interaction in vitro. Similarly, deletion of the first 62 residues but not the first 47 residues of human cingulin head resulted in loss of interaction ( Fig. 2A and data not shown), indicating that residues 48 -61 of human cingulin are required for ZO-1 interaction in vitro. Identical results were obtained with human ZO-1 (not shown).
Alignment of Xenopus, mouse, and human cingulin sequences in this region reveals a remarkable conservation (Fig.  2B). We denoted as "ZIM" a novel ZO-1 interaction motif, the sequence whose deletion resulted in loss of cingulin-ZO-1 interaction in vitro. Although the ZIM was required, it was not sufficient for ZO-1 interaction, since proteins containing only residues 41-55 of Xenopus cingulin fused to GST failed to interact with ZO-1 (not shown). Additional residues in the head are probably necessary in order to allow the ZIM to fold into an appropriate conformation for ZO-1 binding.
A similar battery of deletion mutants was used to identify Xenopus and human cingulin sequences involved in ZO-2 interaction in vitro. Fusion proteins of Xenopus and human cingulin that lacked the ZIM still interacted with ZO-2 (Fig. 2C).
In addition, fusion proteins comprising the first part of cingulin head (residues 1-149 in Xenopus, 1-107 in human cingulin) did not interact with ZO-2. This indicated that sequences within residues 150 -295 of Xenopus cingulin and residues 108 -406 of human cingulin are required for ZO-2 interaction. A GST fusion protein encoding residues 1-378 of Xenopus cingulin head interacted with ZO-2 in a saturable, specific manner, with a calculated K d of ϳ15 nM (Fig. 2D).
No interaction was detected between any cingulin coiled-coil rod fusion proteins and either ZO-1 or ZO-2 (not shown). Thus, sequences at the very NH 2 terminus of the cingulin head are required for in vitro interaction with ZO-1, and sequences in the central part of the head are required for ZO-2 interaction (Fig. 2E).

Identification of Xenopus Cingulin Sequences Involved in Its Junctional Recruitment in A6 Epithelial Cells: Role of ZIM and
Coiled-coil Sequences-To identify cingulin sequences important for its junctional recruitment in vivo, a battery of mutant constructs of Xenopus cingulin containing a COOH-terminal Myc tag were transiently transfected into A6 cells. Mutants were designed to obtain progressive deletions from the NH 2 or COOH terminus of the cingulin molecule or internal deletions. The distribution of the expressed proteins was quantitatively evaluated by scoring at least 200 expressing cells. The distributions were scored as "junctional," "junctional and cytoplasmic," or "cytoplasmic," depending on the co-localization of transfected protein with endogenous ZO-1 (Fig. 3, see "Experimental Procedures"). Fig. 4 shows a selection of representative examples of A6 cells expressing different Xenopus cingulin constructs.
Many of the expressed proteins showed junctional localization. Among these, a first group were deleted for 100 or fewer residues from the NH 2 terminus of the head but contained all of the coiled-coil rod and tail sequence (1-1368, 25-1368, 41-1368, 56 -1368, 73-1368, and 101-1368) (Figs. 3 and 4 (panels A and AЈ and panels B and BЈ). Further deletion of NH 2terminal head sequence (residues 193-1368) resulted in a protein that was never detected exclusively in a junctional localization, even when expressed at low levels, but only in junctional and cytoplasmic (Fig. 4, C-CЈ) and cytoplasmic localizations (Figs. 3 and 4 (C and CЈ)). When the first 294 residues of the head were deleted (295-1368), Ͼ90% of cells showed only cytoplasmic localization (Figs. 3 and 4 (D and DЈ). The junctional localization of proteins lacking the ZIM (residues 56 -1368, 73-1368, 101-1368, and 193-1368) suggested that ZO-1 interaction is not absolutely required for junctional targeting and that cingulin recruitment to TJ can occur through multiple interactions.
A second group of junctionally localized proteins contained either the first 378 residues of the head (residues 1-378) alone or the whole head fused to a stretch of coiled-coil sequence (residues 1-480, 1-551, 1-847, or ⌬445-1038) (Figs. 3 and 4  (panels E and EЈ, F and FЈ, and G and GЈ)). The observation that head sequences alone can be localized at junctions indicated that the coiled-coil region is not absolutely required for junctional targeting. However, when no sequence or only short stretches of coiled-coil sequence were fused to the head, the percentage of cells showing only junctional localization was low (Ͻ10%), and most of the labeling was detected in the cytoplasm and nucleus rather than at junctions (Figs. 3 and 4 (panels E and EЈ and panels G and GЈ)). Thus, cingulin coiled-coil sequences appeared to stabilize junctional recruitment of head sequences.
A third type of junctionally localized protein contained an internal deletion, resulting in the fusion of the first 99 residues of the head to the last 330 residues of the rod/tail sequence (⌬100 -1038) (Figs. 3 and 4 (H and HЈ)). This indicated that although the ZIM-containing NH 2 -terminal region of cingulin is not absolutely required, it is sufficient for junctional localization when fused to coiled-coil sequences.
When the ZIM sequence alone was fused to the coiled-coil sequence, the expressed protein was localized exclusively in the cytoplasm (⌬1-40, 56 -1038) ( Fig. 3 and data not shown). Similarly, proteins containing the COOH-terminal part of the rod/ tail region (residues 1039 -1368) and proteins containing the last ϳ60 residues of the head fused to all of the rod/tail sequence (residues 377-1368) showed only cytoplasmic localization (Figs. 3 and 4 (panels I and IЈ and panels J and JЈ)).
Redistribution of ZO-1 in a significant percentage of cells (ϳ20%) was observed when cells expressed the full-length Myctagged Xenopus cingulin molecule and mutants with short NH 2 -terminal deletions of the cingulin head (Fig. 3D). Deletion of the ZIM was correlated with a decrease in the percentage of cells showing ZO-1 redistribution. However, expression of pro-teins containing ZIM but lacking additional head and/or rod sequences did not cause significant ZO-1 redistribution, even in the presence of junctional localization (for example, ⌬100 -1038). This indicates that additional head sequences, besides the ZIM, are necessary in order for cingulin to disrupt the localization of ZO-1.

ZIM Is Required for Junctional Recruitment of Cingulin in Rat1
Fibroblasts-The observation that cingulin overexpression induces ZO-1 redistribution and that the ZIM-containing protein (⌬100 -1038) targets to junctions in A6 cells suggests that ZO-1 interaction is of primary importance in cingulin junctional recruitment. To test whether cingulin junctional recruitment requires the molecular context of epithelial TJ, Rat-1 fibroblasts were transfected with full-length Xenopus cingulin. Rat-1 fibroblasts lack TJ and do not express specific cingulin, occludin, or claudin immunoreactive polypeptides (data not shown). However, these cells express ZO-1, which is localized along cell-cell adherens junctions (Fig. 5AЈ). When Rat-1 fibroblasts were transfected with ZIM-containing Xenopus or human cingulin, the expressed protein was colocalized with ZO-1 at sites of cell-cell adhesion (Fig. 5, panels A and AЈ  and panels C and CЈ). However, if the expressed proteins lacked the ZIM, they were localized in the cytoplasm (Fig. 5, panels B  and BЈ and panels D and DЈ). Thus, in Rat-1 fibroblasts, exog-enous Xenopus and human cingulins are recruited into cadherin-based cell-cell adhesion sites in a ZIM-dependent and TJ-independent fashion.

The Coiled-coil Region of Cingulin Mediates Self-interaction but Does Not Mediate Filament Formation in Vitro under Physiological Conditions-Transfection experiments indicated that
the coiled-coil region of cingulin alone does not target to junctions, but its presence appears to promote junctional recruitment of head sequences (Fig. 3). In vitro experiments suggest that the coiled-coil region of cingulin mediates self-interaction (12), suggesting that stabilization of junctional recruitment may occur through coiled-coil-mediated cingulin self-assembly. To study cingulin self-assembly in further detail, we analyzed recombinant, full-length Xenopus cingulin by GST pull-down assays, electron microscopy, and turbidity assays.
A GST fusion protein of a fragment of Xenopus cingulin rod (residues 851-1368) interacted with full-length Xenopus cingulin in insect cell lysates in a saturable manner, with a K d of ϳ30 nM (Fig. 6A). Thus, the C-terminal part of the cingulin rod is sufficient for a high affinity interaction with full-length cingulin.
We next purified His-tagged full-length Xenopus cingulin and dialyzed it against a low ionic strength solution, at pH 5.0. We have previously shown that under these conditions purified chicken cingulin rod forms aggregates (12). Electron microscopic analysis of negatively stained samples revealed discrete aggregates (mean length 268 Ϯ 20 nm, mean width 35 Ϯ 7 nm, n ϭ 30) similar to the "twisted tangles" observed previously (Fig. 6B). The aggregates appeared as bundles of 3-5 subfilaments (Fig. 6BЈ, arrow). Thus, the presence of the head region of cingulin does not prevent rod-mediated self-assembly under these conditions.
To test whether full-length cingulin can form filaments under physiological conditions as does myosin, purified Xenopus cingulin and purified myosin were dialyzed separately against solutions of increasing ionic strength. After dialysis, the turbidity of each solution was determined by measurement of absorbance at 340 nm (Fig. 6C). The turbidity of myosin solutions was consistently higher than those of cingulin at all ionic strengths tested, and at physiological NaCl concentration (ϳ150 mM) the turbidity of cingulin solutions was negligible. No filaments were detected in dialyzed cingulin samples by electron microscopy (data not shown). Thus, unlike myosin, purified cingulin does not assemble into filaments under physiological conditions in vitro.

DISCUSSION
In this study, we provide evidence that a conserved NH 2terminal region of cingulin interacts in vitro with ZO-1 and is necessary and/or sufficient for junctional recruitment of cingulin coiled-coil sequences in cultured cells. This result and the observation that overexpression of cingulin disrupted the localization of endogenous ZO-1 in A6 cells suggest that cingulin and ZO-1 are functionally linked in the organization of the TJ plaque.
The redistribution of ZO-1 in cells overexpressing cingulin indicates a functional interaction between these proteins. Some previous studies have addressed the effect of overexpression of TJ proteins in TJ organization and assembly in vivo. Overexpression of NH 2 -terminal fragments of ZO-1 in corneal cells results in disruption of endogenous ZO-1, ZO-2, occludin, and cadherin (23). In Madin-Darby canine kidney cells, expression of truncation mutants of ZO-1 that fail to localize to junctions results in dramatic changes in cell morphology (24). An NH 2terminal fragment of ZO-3, but not wild-type ZO-3, delayed junctional localization of ZO-1 and other TJ proteins during TJ assembly by calcium switch in Madin-Darby canine kidney The numbers above and below the diagrams indicate amino acid residues of Xenopus cingulin included in each construct. The numbers preceded by ⌬ indicate amino acid residues deleted in the construct (for example ⌬1-40, 56 -1038 contains residues 41-55 fused to residues 1039 -1368, fused to Myc tag). All constructs were made in pcDNA3.1A and contained a C-terminal Myc tag that was used for immunofluorescence localization of the expressed protein.
cells (25). Similarly, mutants of atypical protein kinase C, and wild-type and mutant forms of Par-6 proteins prevented or delayed TJ formation during the calcium switch but had no effect on pre-existing TJ (6 -9). In all of these cases, the mechanism by which protein redistribution occurs or assembly is prevented or delayed is not known. We also do not know whether ZO-1 redistribution in cells that overexpress cingulin is a direct effect of a cingulin-ZO-1 interaction or an indirect consequence of an effect of cingulin on other junctional or cytoskeletal proteins. However, we observed that cells overexpressing full-length cingulin did not display an altered overall morphology and that Myc-cingulin labeling was distributed in a submembrane area rather than throughout the cytoplasm. This suggests that disruption of ZO-1 localization is not the consequence of a dramatic reorganization of cellular architecture or a toxic effect but rather that cingulin and ZO-1 remain associated in the subjunctional area. The objective of future studies will be to develop stable cell lines in order to determine whether overexpression of cingulin affects the functional properties of TJ and the organization of additional TJ proteins. Here, we asked whether an interaction between cingulin and ZO-1 is involved in cingulin junctional recruitment by analyzing the role of specific sequences in junctional targeting and ZO-1 interaction.
Analysis of the localization of cingulin mutant proteins in epithelial and nonepithelial cells suggested that the structural requirements for cingulin junctional recruitment depend on cell type and thus, presumably, on junction composition. Rat-1 fibroblasts lack TJ, occludin, claudin, and cingulin but contain adherens junctions and ZO-1. In these cells, cingulin junctional recruitment occurred in a ZIM-dependent manner, suggesting that cingulin can associate with the submembrane plaque of adherens-type junctions through its interaction with ZO-1. However, in A6 epithelial cells, deletion of the ZIM was not correlated with loss of junctional localization, indicating that these cells contain additional proteins or mechanisms that help recruit cingulin. Since deletion of the first ϳ300 residues of Xenopus cingulin essentially abolished junctional localization (Figs. 3 and 4), such additional mechanisms/interactions are probably mediated by this region of the head. However, their role in the physiological recruitment of cingulin into junctions remains unclear. On the other hand, the observation that the first 99 residues of cingulin fused to coiled-coil sequence were sufficient for junctional targeting suggests that cingulin interaction with ZO-1 plays a primary role in its recruitment into epithelial junctions. In addition, the ZIM sequence is highly conserved between amphibian and mammalian species, suggesting a role in cingulin function. We cannot exclude the possibility that the ZIM mediates cingulin interaction with other proteins in addition to ZO-1. However, it seems unlikely that such a small region could be involved in interaction with more than one protein at any given time.
The observation that cingulin can be recruited into fibroblast adherens junctions indicates that cingulin does not require the molecular context of TJ in order to be associated with sites of cell-cell contact. This observation may be relevant to understanding the role of cingulin and its interaction with ZO-1 in TJ biogenesis. In preimplantation mouse embryos, accumulation of ZO-1 (␣ Ϫ isoform) at cadherin-based junctions precedes the assembly of cingulin, occludin, and the ZO-1 ␣ ϩ isoform into new junctional sites (26,27). In addition, studies on cultured cells suggest that TJ formation occurs when cadherins are sorted out of and occludin is sorted into ZO-1-containing complexes (28,29). Thus, ZO-1 appears to be first associated with cadherin-based adherens-type junctions and then is redistributed into TJ. Evidence from studies on early Xenopus embryos shows that during the first cleavage, cingulin is recruited from the apical submembrane cortex directly into new apical junc-tional sites (22,30). On the other hand, ZO-1-containing vesicles are targeted to the new basolateral membrane and then accumulate into the cingulin-containing sites (30). This sug- FIG. 6. Self-assembly of cingulin in vitro. A, quantitative analysis of cingulin rod-mediated self-interaction in vitro. The plot shows the amount of full-length Xenopus cingulin interacting with bacterially expressed GST-cingulin rod fusion protein (GST fused to residues 851-1368) as a function of free cingulin. Inset, Coomassie-stained gel with bound fractions and Scatchard plot analysis of the data. The estimated K d value for the interaction between the rod fragment of cingulin and full-length cingulin was ϳ30 nM. B, in vitro assembly of purified recombinant Xenopus cingulin after dialysis against 50 mM sodium acetate, pH 5.0, negative staining and electron microscopy (see "Experimental Procedures"). Bars, 100 nm. Inset BЈ shows a higher magnification image of twisted tangle bundles, with an arrow indicating subfilaments within bundles. C, histogram showing turbidity of purified Xenopus cingulin (white bars) and purified smooth muscle myosin (black bars) samples after dialysis against solutions containing the indicated concentrations of NaCl (5, 50, 150, and 300 mM) plus 5 mM MgCl 2 , 0.5 mM EGTA, 0.1 mM DTT, 10 mM imidazole, pH 7.5. The data are from one representative experiment of three. gests that TJ assembly could result from the recruitment of a "basolateral complex," containing proteins such as ZO-1, into a preformed "apical complex," containing cingulin. In early Xenopus embryos, it appears that cingulin is not associated with basolateral cadherin-based adhesion sites. Indeed, junctional assembly of cingulin and ZO-1 occurs even under conditions where cadherin adhesion is disrupted (22,30). Taken together, these observations raise a number of hypothetical scenarios for the role of cingulin-ZO-1 interaction in TJ biogenesis. For example, the high affinity cingulin-ZO-1 interaction could play a role in sorting ZO-1 and associated proteins out of cadherinassociated complexes and into TJ complexes. Alternatively, or in addition, the cingulin-ZO-1 interaction may favor the redistribution of cingulin from a cytoskeleton-associated submembrane pool into the TJ plaque, possibly through a cadherinassociated intermediate complex. To test definitively these hypotheses and dissect the molecular steps in TJ assembly, it would be useful to obtain cells lacking cingulin or ZO-1.
The coiled-coil rod domain of cingulin appeared to stabilize the recruitment of cingulin head sequences into junctions, since head sequences alone or followed by short stretches of coiledcoil were rarely localized exclusively at junctions in transiently transfected cells (Fig. 3). On the other hand, proteins containing all or part of the rod did not localize at junctions in transfected A6 cells, confirming earlier observations using Xenopus blastomeres (12). Thus, although cingulin rod fragments interact with full-length cingulin with high affinity in vitro (Fig.  6A), such interaction does not appear sufficient to support junctional recruitment of transfected cingulin by endogenous cingulin (Fig. 4, J and JЈ). What then is the role of the coiledcoil region of cingulin? By analogy with conventional myosin II, we speculated that the cingulin rod may allow the molecule to form filaments. However, although formation of aggregates was detected at pH 5, no assembly of purified recombinant cingulin was observed under physiological conditions, suggesting that cingulin cannot form filaments (Fig. 6C). This is in agreement with the observation that the periodicities of distributions of acidic and basic residues in the Xenopus cingulin rod sequence do not have matching peaks and that the peaks themselves are of lower intensities than those of myosin (31). Thus, the present evidence suggests that the main role of the coiled-coil domain of cingulin is to allow two cingulin subunits to form a parallel dimer. However, the possibility of oligomer formation or intramolecular association cannot be excluded.
In summary, we dissected the structural requirements for cingulin targeting to junctions of A6 cells and Rat-1 fibroblasts, and we identified an NH 2 -terminal region of cingulin that is required for in vitro ZO-1 interaction and is sufficient and/or necessary for cingulin junctional recruitment. Furthermore, we showed that cingulin overexpression results in ZO-1 redistri-bution, suggesting that cingulin and its interaction with ZO-1 are important in the organization of the TJ plaque.