Overexpression of the Xenopus Tight-Junction Protein Claudin Causes Randomization of the Left–Right Body Axis

Developmental Biology 230, 217–229 (2001) doi:10.1006/dbio.2000.0116, available online at http://www.idealibrary.com on Overexpression of the Xenopus Tight-Junction Protein Claudin Causes Randomization of the Left–Right Body Axis

Brenda J. Brizuela,1 Oliver Wessely, and E. M. De Robertis2 Howard Hughes Medical Institute and Department of Biological Chemistry, University of California at Los Angeles, Los Angeles, California 90095-1662

This study presents Xenopus claudin (Xcla), a tight-junction protein that is abundantly expressed in eggs and neuroecto- dermal precursors during early development. It was isolated via a differential screen for mRNAs enriched in microsomes in the Xenopus blastula. The Xcla protein contains four transmembrane domains and a carboxy-terminal cytoplasmic region with a putative PDZ-binding site. We show that this PDZ-binding site of Xcla is critical for its correct localization on the cell membrane and that a truncated form leads to delocalization of the tight-junction protein ZO-1. Overexpression of Xcla causes changes in the cell adhesion properties of blastomeres and leads to visceral situs randomization. The results suggest that left–right axial patterning is very sensitive to changes in regulation of cell–cell interactions and implicate a tight-junction protein in the determination of left–right asymmetry. © 2001 Academic Press Key Words: Xenopus; left–right asymmetry; Claudin; ZO-1; tight junctions; cell adhesion; cell–cell interactions.

INTRODUCTION tight-junction proteins (Kubota et al., 1999; Fanning et al., 1999). Tight junctions provide an apical seal, also known as the Tight junctions contain integral membrane proteins that zonula occludens, in polarized epithelial and endothelial bridge adjacent cells and peripheral cytoplasmic proteins cells. Tight junctions are found associated with adherens that interact with the cytoskeleton (reviewed in Citi and junctions and gap junctions in most epithelial cell types and Cordenosi, 1998). In avian and mammalian epithelia the are linked intracellularly to the actin cytoskeleton (Kojima membrane-spanning component of the tight-junction com- et al., 1999; Fanning et al., 1998). These junctions play a plex includes the four-transmembrane proteins occludin major role in specific cell–cell recognition, sorting, and and family members of the more recently identified clau- communication and in the maintenance of the cohesive dins (Furuse et al., 1993, 1998a). The cytoplasmic compo- sheet of cells required for epithelial integrity and morpho- nent of the tight-junction complex contains several periph- genesis. Two important functions previously described for eral proteins, in particular zonula occludens-1 (ZO-1) tight junctions are a barrier function that allows epithelia to (Stevenson et al., 1986), ZO-2 (Gumbiner et al., 1991; Itoh separate fluid compartments (e.g., blood/brain) by control- et al., 1999a), and ZO-3 (Haskins et al., 1998). ZO-1 is ling the permeability of the paracellular pathway to ions thought to be responsible for clustering transmembrane and larger molecules and a fence function that restricts the proteins or receptors in their correct positions by binding to mobility of membrane proteins and lipids across the bound- the COOH-terminus of these proteins via 90-amino-acid ary between apical and basolateral membrane domains, establishing the polarity of epithelial cells (Fanning et al., repeats called PDZ domains, so designated because of their 1998; Citi, 1993; Mu¨ller and Hausen, 1995). More recently conservation in the PSD-95 (postsynaptic density-95), a third function in cell adhesion has been proposed for Dlg-A (discs-large), and ZO-1 proteins (Fanning and Ander- son, 1999). Tight junctions play an important role during

1 early embryonic development. For example, functional in- Present address: Zoology Department, North Carolina State University, Raleigh, NC 27695. tercellular tight junctions maintain epithelial integrity dur- 2 To whom correspondence should be addressed. E-mail: ing the formation of the blastocoele cavity in Xenopus [email protected]. (Cardellini et al., 1996; Merzdorf et al., 1998). The regula-

membrane-bound or secreted proteins translated at these earliest stages of embryogenesis, we purified mRNAs associated with the endoplasmic reticulum at the blastula stage and carried out a differential screen of microsomal versus nonmi- crosomal mRNAs. In the present study we characterize one of the clones isolated, Xcla. Although the function of this protein was unknown at the time the studies were initiated, rapid progress in this active area of research revealed that Xcla was a member of the Claudin gene family of tight-junction proteins. Xcla mRNA is expressed maternally in the oocyte and blastula stages and has a dynamic zygotic expression pattern in epithelial cells at the gastrula, neurula, and tail-bud stages. We show that, like members of the claudin gene family in other systems, Xcla is an indispensable component of the tight-junction complex in Xenopus embryos. Overexpressed Xcla localizes at cell–cell contact points and increases cell adhesion, while a mutant construct lacking the C-terminal PDZ-binding domain changes ZO-1 localization. Perturbation of cell adhesion and cell migration have previously been correlated with randomization of left– right asymmetries in Xenopus embryos. Elimination of fibronectin fibrils from the blastocoele roof over which cardiac and visceral primordia move during development results in global randomization of left–right asymmetries (Yost, 1992). In addition, communication via gap junctions has been shown to play an essential role in left– right asymmetry in Xenopus (Levin and Mercola, 1998) and has been implicated in heterotaxia in humans (Britz- Cunningham et al., 1995). Furthermore, N-cadherin, a component of adherens junctions, is required for left– right asymmetry in the chick (Garcia-Castro et al., 2000). Interestingly, injection of synthetic mRNA encoding Xcla into Xenopus embryos causes bilateral expression of Xnr-1 in lateral plate mesoderm and left–right randomization of the visceral organs (heart, gut, and gall bladder). This phenotype is not accompanied by a concomitant FIG. 1. (A) Alignment of the Xcla amino acid sequence with decrease in gap-junction communication. The results human Claudins 4, 3, and 5. (B) The putative Xenopus Claudin suggest that left–right axial patterning is very sensitive to protein has four transmembrane domains, two extracellular loops, changes in cell junction function. one intracellular loop, and a cytoplasmic carboxyl-terminus. (C, D) Immunostaining of Xcla-HA-injected animal caps in (C) nonper- meabilized and (D) permeabilized cell membranes; this demon- MATERIALS AND METHODS strates experimentally that the predicted structure shown in B is the correct topology and confirms that the protein is located in the cell membrane and is concentrated at regions of cell contact (inset Isolation of Microsomal RNA Fractions from in D). Xenopus Embryos To screen for membrane or secreted proteins that might function in early development, a method for microsome isolation was developed for Xenopus embryos from the methods of tion of tight-junction complex assembly is dependent on Adelman et al. (1973) and of Kopiczynki et al. (1998). Xenopus the localization of tight-junction transmembrane proteins embryos were allowed to develop to blastula stage 9 (8.5 h after within the cell (Fleming et al., 1993) and has been impli- fertilization), and 500 healthy embryos were selected, homoge- nized in a glass homogenizer with a loose pestle in 6 ml STKMH cated in the timing of blastocoele formation (Sheth et al., homogenization buffer (0.5 M sucrose, 50 mM Tris–HCl, pH 7.5, 2000). 25 mM KCl, 5 mM MgCl, 0.5 mg/ml heparin), and centrifuged In Xenopus, early development takes place in the absence of (45 s, 5220g) in an Eppendorf 5402 microcentrifuge to remove the RNA synthesis until the midblastula transition and com- yolk. The postyolk supernatant (PYS) was decanted from the pletely relies on maternally inherited proteins and mRNAs yolk pellet and adjusted to a final concentration of 150 mM KCl, (Newport and Kirschner, 1982). In an effort to identify 2 M sucrose, 0.5 mg/ml heparin. High Kϩ concentration is used

FIG. 2. Xenopus claudin expression pattern studied by in situ hybridization. (A) Maternal expression of Xcla in oocyte stages I–IV. (B) RT-PCR of RNA isolated from 16-cell stage embryos; Xcla expression is enriched in the animal hemisphere. Vg1 was used as a vegetal control and histone H4 as a loading control. (C–F) Xcla mRNA in the animal pole of 2-, 4-, 8-, and 128-cell embryos. (G–J) Zygotic expression in the late gastrula, neurula, and neural tube stages. Note that the expression is repressed in the neural plate midline and in a transient band that appears to coincide with the midbrain/hindbrain boundary (arrowhead in I). (K and L) Tail-bud stage embryos. Xcla neural expression can be detected in the developing neuroepithelium at the tip of the tail, in the otic vesicles, and in the branchial arch endoderm (arrowheads in K), as well as in nasal placode and pronephros (arrowheads in L).

to destabilize weak protein–protein interactions and help disso- (Tel-Test, Inc.) and tested by RT-PCR for the presence of chordin ciate free polysomes from membrane-bound polysomes (rough and Frzb-1 mRNA, which encode known secreted molecules. microsomes). The high sucrose density allows the PYS to form the bottom of a step gradient, producing a better separation of Isolation of Xcla cDNA the ﬂoating microsome fraction. The addition of heparin inhibits ribonuclease activity. The PYS was then placed in the bottom of The fractions that tested positive for microsomes marked by a sucrose density step gradient overlaid by 1.8 and 1.3 M chordin and Frzb-1 mRNA were transcribed into cDNA by the Cap STKMH. After ultracentrifugation (13 h, 212,000g) in a Beckman Finder method (Clontech). cDNAs with inserts between 600 and SW 50.1 rotor, fractions were collected using a peristaltic pump. 1200 bp, prepared by sizing columns, were cloned into the pCR2.1 RNA was then puriﬁed from the fractions with RNA-Stat 50LS vector (Invitrogen), and 77 individual clones were sequenced using

an ABI sequencer and analyzed using the BlastN and BlastX search dorsoanterior index (DAI ϭ 5) were scored. Xnr-1 expression was programs (nonredundant GenBank database). After analyzing these determined by in situ hybridization at stage 22 after injection of sequences we determined that 19 independent genes were repre- Xcla or Xcla-⌬C (600 pg total RNA) into the animal cap region of sented. These isolated clones consisted of partial cDNAs of 13 each blastomere at the four-cell stage. Gap-junction transfer novel genes, 5 cDNAs with varying degrees of homology to assays were performed as described in Levin and Mercola (1998), occludin/ELL, spermidine N1-acetlytransferase, phosphatidylino- after two injections of Xcla or Xcla-⌬C (600 pg total RNA) before sitol glycan-B, ILV5, and human ␤-catenin, in addition to Xcla first cleavage. (described here). Xcla was isolated independently seven times, and its full-length cDNA was then isolated from a dorsal lip Xenopus gastrula library (Blumberg et al., 1991) and sequenced on both Immunohistochemistry and Confocal Microscopy strands (GenBank Accession No. AF224712). Related Xenopus ␮ clones appear in the GenBank EST database, suggesting that other L-cell fibroblasts were transfected with 1 g of the HA-tagged ⌬ claudin family members may be present in Xenopus later in forms of Xcla or Xcla- C in a CMV promoter-driven expression development. vector (pCS2), on eight-chamber slides (Nunc/Lab-Tek), and cultured for 48 h in MEM with 10% fetal calf serum. Cultured cells were fixed and stained as stated elsewhere (Furuse et al., RT-PCR and in Vitro Transcription/Translation 1998b). Injected animal caps were dissected at stage 10.5, followed by fixation and staining as stated elsewhere (Miller and RT-PCR to test microsomal fractions were performed using Moon, 1997). Counterstain with BODIPY (Molecular Probes; 30 primers for chordin, Frzb-1, and Vg1; histone H4 was used as the min at room temperature at 2 units/ml) was used to emphasize nonsecreted control, and ornithine decarboxylase and elongation the cell membrane in some samples prior to mounting in ␣ factor 1- were used as loading controls. All primers and conditions Vectashield (Vector Laboratories). Cell culture specimens and are described in http://www.hhmi.ucla.edu/derobertis/protocol- animal cap explants were observed using a fluorescence Zeiss Ј page/protocol.html. RT-PCR primers for Xcla were (forward) 5 - Axiophot photomicroscope, and confocal microscopy was per- Ј Ј GGA ACA ACA TAG TGG TGG C-3 and (reverse) 5 -GAC AGT formed on an MRC 1024 scanner (Bio-Rad) attached to an Ј AAG GGC ACG GGC A-3 (annealing temperature 52°C, 20 Eclipse TE 300 Nikon microscope. The ZO-1 antibody used here, cycles). Oocytes and embryo stages were determined as in Nieu- which is cross-reactive with Xenopus, was antiserum 10153 wkoop and Faber (1994). In situ hybridization was performed as raised in rabbit against a human ZO-1/GST fusion protein described (Belo et al., 1997). (Merzdorf et al., 1998) and kindly provided to us by Dr. Daniel Goodenough. Plasmid Construction and RNA Injections The pCS2-Xcla-HA C-terminal HA tag construct was gener- RESULTS ated by PCR addition of the amino acids YPYDVPDYA after Val214 and subcloned into the EcoRI–XhoI sites of the pCS2ϩ Cloning and Characterization of cDNA Encoding vector. The pCS2-Xcla-⌬C-HA construct was made by PCR Xenopus Claudin mutagenesis to remove the C-terminal amino acids 192–214, adding an HA-tag after amino acid 191. pCS2-Xcla-⌬C-HA Patterning of the early frog embryo requires the expres- expresses a truncated form of the protein, which lacks the sion of secreted morphogens and depends on the commu- putative PDZ binding site. To generate synthetic mRNAs nication between cells for the correct interpretation of pCS2-Xcla, pCS2-Xcla-HA, or pCS2-Xcla-⌬C-HA was digested these signals. For this reason, we decided to use microsome with NotI and transcribed with SP6 RNA polymerase (Kim et al., fractionation to screen for molecules which are either 1998). mRNAs for green fluorescent protein (GFP), prolactin, present at cell membranes or secreted into the extracellular and lacZ were described (Kim et al., 1998; Zernicka-Goetz et al., space. This method is based on the observation that those 1996). proteins targeted to the cell membrane or secretory path- To determine the cellular localization of the Xenopus Claudin way have their mRNA transcripts bound to the endoplas- protein, four injections of 300 pg Xcla mRNA containing an HA-tag were introduced into the animal pole at the 4- to 8-cell mic reticulum (Adelman et al., 1973; Kopczynski et al., stage, followed by immunohistological staining (see below). 1998; Diehn et al., 2000). A maternal mRNA isolated Dissociation and reaggregation experiments were done as de- multiple times (seven independent isolates) was chosen for scribed in Kim et al. (1998) following injection of 300 pg Xcla or further analysis and a full-length cDNA was isolated. Xcla Xcla-⌬C mRNA into each blastomere at the 4-cell stage. The cell encoded a novel member of the Claudin family of tight- adhesion phenotype was determined after one injection of 400 or junction membrane proteins (Fig. 1A). The Xcla protein has 800 pg Xcla or Xcla-⌬C mRNA into the A4 blastomere at the 62% amino acid identity to human Claudin-4 (Morita et al., 32-cell stage. Xcla and Xcla-⌬C were co-injected with 250 pg 1999a), 59% identity to human Claudin-3 (Morita et al., GFP and 250 pg lacZ mRNA as lineage tracers. 1999a), and 52% identity to human Claudin-5 (Morita et al., Effects on left–right asymmetry were analyzed after two 1999a,b; Sirotkin et al., 1997; Chen et al., 1998). Xcla shows dorsal, two ventral, two left, or two right injections of Xcla or Xcla-⌬C (400 or 600 pg mRNA total per embryo) at the four- to lower sequence similarity to Claudin-6, Claudin-1, eight-cell stage. Defects in left–right asymmetry were deter- Claudin-2, and other members of this recently identified mined by scoring the situs of the heart, gut, and gall bladder family of tight-junction proteins. Since Xcla is the first under a dissecting microscope in embryos anesthetized with member of the claudin family isolated in Xenopus, and 0.1% tricaine (Sigma) at stage 45. Only embryos with a normal none of the other vertebrate Claudins have been analyzed

Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. Xenopus Claudin and Left–Right Asymmetry 221 for its expression pattern during early embryogenesis, with tures in L-cells, a fibroblast cell line which does not form current knowledge it is not possible to say whether Xcla is tight junctions and does not endogenously express claudin a homologue to a previously isolated claudin. The level of (Furuse et al., 1998a,b). Similar results could be obtained amino acid identities, however, would suggest that Xcla using epitope-tagged full-length Xcla. After transfection may be a novel Claudin family member. The Xcla cDNA into L-cell and staining with an anti-HA antibody the encodes a protein of approximately 25 kDa as judged by in protein was found at cell contact points between the L-cells vitro coupled transcription/translation (not shown). Like (Figs. 3B and 3D). Interestingly, a construct lacking the last other members of this family, Xcla ends in the sequence 23 amino acids from the COOH-terminus of Xcla, which KNYV, a potential PDZ-binding motif (Fig. 1A), which is deleted the potential PDZ-binding domain (Xcla-⌬C, Fig. thought to be recognized by ZO-1 (Morita et al., 1999a; 3A), failed to concentrate at cell contacts and appeared to be Songyang et al., 1997; Itoh et al., 1999b). Hydropathy localized mostly in intercellular organelles (Fig. 3E). There- analyses of the amino acid sequence of Xcla predicted four fore, the COOH-terminus of Xcla was required for the transmembrane regions (Fig. 1B). To clarify the in vivo proper localization of the protein to cell contact points. topology of the protein, we used an immunohistochemical Interaction of Claudin with ZO-1 via the PDZ domain approach to determine the localization of the COOH- (Itoh et al., 1999b) may position the Claudin/ZO-1 complex terminus. Synthetic mRNA of a COOH-terminal HA- in the apical region of the epithelial cell, where tight tagged Xcla was injected into the four blastomeres of junctions are formed. Therefore, we wanted to test whether eight-cell embryos, the animal caps were dissected at stage Xcla-⌬C affected not only its own localization, but also the 10.5 and stained with an anti-HA antibody in the presence formation of endogenous cell junctions marked by ZO-1. To or absence of Triton X-100 detergent. Staining at the cell this end, we injected synthetic mRNA encoding the HA- surface could be detected only after permeabilization of the tagged versions of either the mutant or the wild-type membrane with detergent, demonstrating that the HA- protein into the animal pole of four-cell stage embryos, tagged C-terminus of the protein is intracellular (compare dissected animal cap explants at stage 10.5, and analyzed Figs. 1C and 1D). them for the localization of ZO-1 and the injected protein construct using immunohistochemistry. As expected for Expression of Xenopus Claudin tight-junction-like structures, full-length Xcla-HA protein colocalized with endogenous ZO-1 in the cell membrane of In order to study the expression of Xcla during develop- animal cap explants (Figs. 4A–4C). However, in embryos ment we performed whole-mount in situ hybridization (Fig. expressing Xcla-⌬C-HA not only did the mutant protein not 2). Xcla is strongly expressed maternally in the animal pole localize at the membrane, but in addition endogenous ZO-1 of oocytes and cleaving embryos after fertilization (Figs. 2A lost its localization at the surface membrane, was found and 2C–2F). This animal pole enrichment of the maternal predominantly in intracellular organelles, and was not Xcla mRNA could be confirmed using embryos dissected colocalized with Xcla-⌬C-HA in the cytoplasm (Figs. 4D into animal cap and vegetal mass at the 16-cell stage and and 4E). These results were reproducible and the effects of analyzed by RT-PCR (Fig. 2B). After the midblastula tran- both Xcla and Xcla-⌬C could be obtained in four indepen- sition and the onset of zygotic transcription, Xcla expres- dent experiments (n ϭ 12 for each experimental condition) sion could be detected in neural plate epithelium (Fig. 2G) involving 48 caps total. which subsequently narrowed as the neural folds close to form the neural tube (Figs. 2H–2J). Expression is downregu- lated in the midline and in a transient band that appears to Xcla and Xcla-⌬C Affect the Adhesive Properties of coincide with the midbrain/hindbrain boundary (Fig. 2I). In Epithelial Cells tail-bud stage embryos, Xcla mRNA is found in the tip of To test whether the effects of Xcla and Xcla-⌬C mRNA the tail, nasal placode, and otic vesicles, as well as in on tight junctions might influence the adhesive properties branchial arch endoderm, pronephros, and pronephric duct of cells in the embryo, we co-injected mRNAs encoding (Figs. 2K and 2L). It is interesting to note that all of the either form of the protein together with lacZ lineage tracer. tissues which show expression of Xcla are of epithelial Embryos injected into a single blastomere (see Materials character. Northern blot analysis showed that Xcla is and Methods) were fixed at stage 11 and stained for expressed as a single transcript of approximately 1.2 kb ␤-galactosidase activity. Injection of control prolactin from oocyte through all embryonic stages (not shown). We mRNA showed a gradual dispersion of the injected cells in conclude that Xcla mRNA is stored maternally in the a smooth pattern as shown in Fig. 5A (94%, n ϭ 77). animal region of the egg and expressed zygotically in Injection of full-length (nonepitope tagged) Xcla mRNA developing epithelia. caused cells to adhere more tightly and thereby prevented cell dispersion (62%, n ϭ 103, Fig. 5B). Unexpectedly, The COOH-Terminus of Xcla Is Required for ectopic expression of the mutant construct Xcla-⌬C re- Localization within the Membrane sulted in a broad dispersion of large groups or clumps of It has recently been shown that ectopically expressed cells, with some clumps localizing far away from the point claudin cDNA can reconstitute tight-junction-like struc- of injection (88%, n ϭ 107, Fig. 5C). The Xcla-⌬C pheno-

type could be rescued to a degree of cell dispersion resem- bling that of the wild type by co-injection of the full-length Xcla and Xcla-⌬C mRNA (76%, n ϭ 107, Fig. 5D). This result suggests that ectopically expressed wild-type Xcla may form oligomers with Xcla-⌬C to rescue the phenotype. Finally, to further document changes in adhesive behavior, ectodermal explants of embryos injected animally with Xcla or Xcla-⌬C mRNA, together with mRNA for the lineage tracer GFP, were analyzed for their cell adhesion behavior with uninjected cells (Fig. 5E). Interestingly, in the case of full-length Xcla the effect of overexpression of a tight-junction protein could already be detected during the dissociation procedure. The Xcla-injected cells became much more tightly associated and could not be easily dissociated in calcium/magnesium-free solution. Xcla-⌬C- injected cells, however, were readily dissociated. Dissociated microinjected animal cap cells were mixed together with cells of control embryos injected with rhodamine-dextran as a lineage tracer, reaggregated by addition of Ca2ϩ, and analyzed by fluorescence microscopy after 12 h (Kim et al., 1998). In two independent experiments, cells injected with Xcla and GFP sorted out from wild-type cells (Fig. 5F, n ϭ 13), as did cells injected with Xcla-⌬C and GFP (Fig. 5G, n ϭ 9). Cells injected with control prolactin and GFP mRNAs dispersed more uni- formly among uninjected or rhodamine-dextran-injected wild-type cells (Fig. 5H, n ϭ 13). Taken together, these results suggest that overexpression of Xcla leads to changes in cell adhesion. A mutant version lacking the C-terminal PDZ binding domain, Xcla-⌬C, also does this, although to a lesser degree, causing changes in behavior of the injected cells. The ability of Xcla-⌬C to change cell adhesion may be due to its interaction with the endogenous claudin present in the cell or the sequestering of other tight-junction components. The effects of Xcla-⌬C on cell adhesion may also be due to interactions, via other domains still present in the protein, with occludin or the junctional adhesion molecule (JAM). JAM is a member of the Ig superfamily which localizes to the tight junctions of both epithelial and endothelial cells and has been shown to both mediate homotypic adhesion and influence monocyte transmigra- FIG. 3. The PDZ binding domain is required for Xcla localization tion (Martin-Padura et al., 1998). at cell contact points in L-cells. (A) Schematic representation of the epitope-tagged constructs used; see text for details. Representative examples from two independent experiments, in which many Randomization of Left–Right Asymmetry microscope fields involving hundreds of cells were examined. (C, D) Xcla-HA and (E) Xcla-⌬C-HA transfected into L-cells and To determine whether interfering with tight-junction analyzed by immunofluorescence using an anti-HA antibody. Note assembly had consequences for normal development, we that Xcla-HA is localized at contact points between cells, whereas injected synthetic mRNA of Xcla-⌬C at the four-cell stage Xcla-⌬C was never found in intracellular locations (compare ar- (see Materials and Methods) and examined the resulting rows in C–E). (B) Phase contrast; (C) fluorescence image of cells shown in (B). (D and E) Confocal microscopy. FIG. 4. Wild-type Xcla, but not Xcla-⌬C, colocalizes with ZO-1 in the cell surface membrane. Confocal microscopy of Xenopus animal cap explants injected with synthetic mRNA encoding Xcla-HA Xcla-HA and ZO-1 are colocalized in spots along the membrane in (A–C) or Xcla-⌬C-HA (D and E) and uninjected control (F). The Xcla-HA-injected caps (C), but located mostly in intracellular explants were stained with anti-HA monoclonal antibody to visu- organelles from the membrane in Xcla-⌬C-HA-injected caps (E). alize the exogenous Xcla protein (red) and with anti-ZO-1 poly- Endogenous ZO-1 protein is localized in the membrane of unin- clonal antibody to show the endogenous ZO-1 protein (green). jected caps (F).

FIG. 5. Xcla overexpression affects cell adhesion. (A–D) Embryos co-injected once into the A4 blastomere at the 32-cell stage with lacZ and prolactin, Xcla, Xcla-⌬C, or Xcla plus Xcla-⌬C mRNAs, followed by staining for ␤-galactosidase activity at stage 11. Overexpression of wild-type Xcla (B) causes cells to be more adherent than in prolactin mRNA-injected control embryos (A), whereas overexpression of mutated Xcla-⌬C (C) also causes cells to adhere, but they disperse as clumps of cells. This altered cell adhesion can be rescued to normal by co-injection of Xcla plus Xcla-⌬C (D). (E–H) Dissociated animal caps injected with Xcla/GFP (F), Xcla-⌬C/GFP (G), or control prolactin/GFP (H) were mixed with dissociated animal caps injected with rhodamine-dextran (red cells in F–H) and allowed to reaggregate by the addition of Ca2ϩ. Both Xcla and Xcla-⌬C mRNA lead to cell sorting from rhodamine-dextran injected cells, while prolactin-injected control cells disperse evenly throughout the aggregate.

embryos at stage 45. Microinjection produced a very spe- (42% bilateral Xnr-1, n ϭ 133). Interestingly, we observed ciﬁc and reproducible phenotype, heterotaxia, in animals very few cases of right-sided expression or complete loss of that appeared phenotypically normal at stage 45. In the case Xnr-1 (Table 1). The higher frequency of bilateral Xnr-1 of Xcla-⌬C, 17% of the scored embryos (n ϭ 255) had expression compared to the heterotaxia phenotype in the randomization of the heart, gut, and/or gall bladder (Figs. visceral organs (34% versus 28% for Xcla mRNA and 42% 6C and 6E). In contrast, uninjected or prolactin mRNA- versus 17% for Xcla-⌬C mRNA) can be explained by the injected control embryos displayed only 1% heterotaxia fact that left-sided Xnr-1 is a prerequisite for the establish- (n ϭ 438) (Figs. 6B and 6D). This suggested that functional ment of the left–right axis. In bilaterally expressing em- tight junctions are required for the determination of the bryos the orientation of each organ would be established left–right axis. Interestingly, this observation was not re- stricted to the use of the mutant construct, which disrupts tight junctions. Heterotaxia could also be observed when overexpressing wild-type non-epitope-tagged Xcla mRNA TABLE 1 ϭ or Xcla-HA mRNA (29% heterotaxia, n 175). Overexpression of Xcla or Xcla-⌬C Causes Bilateral Xnr-1 In Xenopus the earliest marker for left–right asymmetry Expression in Lateral Plate Mesoderm is the expression of Xenopus Nodal-related 1 (Xnr-1)inthe left lateral plate (Lowe et al., 1996) (Fig. 6F). We therefore Controls analyzed the effects of Xcla and Xcla-⌬C on Xnr-1 expres- Xnr-1 prolactina Uninjected Xclaa Xcla-⌬Ca sion. Embryos were injected into each blastomere at the expression n ϭ 35 n ϭ 530 n ϭ 118 n ϭ 133 four-cell stage with Xcla, Xcla-⌬C, or control prolactin mRNA, or left uninjected, and then ﬁxed at stage 22 and Bilateral 6% 4% 34% 42% analyzed by whole-mount in situ hybridization (Figs. 6F– Absent 3% 0.6% 0% 0% 6H). In the case of the prolactin mRNA-injected (n ϭ 35) Right side 0% 0.4% 1% 5% and uninjected (n ϭ 530) embryos, 91 and 95% of the Left side 91% 95% 65% 53% embryos showed normal left-sided expression of Xnr-1, Note. Lateral plate mesoderm expression of Xnr-1 at stage 22 as respectively (Table 1). This was drastically altered upon analyzed by in situ hybridization. overexpression of Xcla (n ϭ 118), which caused 34% of the a Four injections into the animal region of each blastomere at the embryos to express Xnr-1 bilaterally. A similar result was 4-cell stage with Xcla, Xcla-⌬C, or prolactin mRNAs (600 pg total also obtained in embryos injected with Xcla-⌬C mRNA mRNA per embryo).

randomly and thus can also result in a completely normal situs. Pioneering work by Levin and Mercola (1998) has shown that disruption of gap-junction communication (GJC) also leads to situs inversus. To address the question of whether the phenotypic effects observed were caused by a secondary effect on gap-junction communication, we used the approach of Levin and Mercola (1998). Fertilized albino eggs were injected prior to the first cleavage with either Xcla or Xcla-⌬C mRNA. Then at the 16-cell stage one random animal blastomere was injected with a mixture of Lucifer yellow and rhodamine-lysinated dextran dye. Embryos were maintained in culture for an additional 20 min before fixation and analysis by fluorescence microscopy. During the time between injection and fixation the low-molecular- weight compound Lucifer yellow (522 Da) could cross from the injected cell into a neighboring cell via gap junctions, while the Ͼ10 kDa rhodamine-lysinated dextran remained within the injected cell. Thus, the number of embryos with differential transport of Lucifer yellow is an indicator for GJC (Fig. 7). Using this approach we found that 24% of the embryos (n ϭ 49) injected with the mixture of Lucifer yellow and rhodamine-lysinated dextran showed transfer. This number was reduced to 11% (n ϭ 27) when the embryos were incubated in heptanol, a compound that inhibits GJC in Xenopus (Levin and Mercola, 1998). Injec- tion of Xcla-⌬C (n ϭ 37) had no effect on GJC, whereas Xcla mRNA (n ϭ 24) led to a slight increase (Figs. 7A–7G). The 24% transfer via GJC in control embryos was less than the 41% reported for dorsal blastomeres by Levin and Mercola (1998) and might be explained by the fact that we injected albino embryos at random animal blastomeres, whereas previous workers have reported higher GJC in dorsal blastomeres and lower GJC in ventral blastomeres (Levin and Mercola, 1998). We preferred albino embryos for this initial set of experiments as we had noted that the ventral pigment can quench the fluorescent signal in some batches of embryos. To confirm that the failure of Xcla-⌬C to affect GJC (Fig. 7G, 26% compared to the 24% for controls) was not due to the low basal level of Lucifer yellow transfer, we repeated the experiment on dorsal blastomeres of embryos with strong dorsal–ventral polarity (Klein, 1987). Under these conditions, a higher basal rate of GJC was obtained. Indeed, after allowing 10 min for transfer after co-injection of rhodamine-dextran and Lucifer yellow, dye transfer was observed in 63% (n ϭ 32) of the control FIG. 6. Overexpression of Xcla mRNA causes bilateral Xnr-1 embryos and in 65% (n ϭ 69) of the Xcla-⌬C mRNA- expression and left–right randomization. (A–H) Embryos were injected embryos (Fig. 7H). Thus, the phenotype of hetero- injected at the four-cell stage with Xcla or Xcla-⌬C mRNA and taxia caused by Xcla-⌬C cannot be caused by a decrease in scored for inversions at stage 45 (B–E) or analyzed by Xnr-1 in GJC. For these and other reasons discussed below, we favor situ hybridization at stage 22 (F–H). (B, D, and F) Uninjected the interpretation that interfering with tight-junction as- control embryos. (C) Inverted heart and gall bladder (compare sembly can affect left–right asymmetry independent of arrowheads in B and C) and (E) inverted gut coiling (compare with D) in Xcla-⌬C-injected embryos. (F) Normal left-sided changes in gap-junction communication, although we can- Xnr-1 expression in control embryo from dorsal view. Bilateral not rule out the possibility that the slight increase in GJC Xnr-1 expression from dorsal view in (G) Xcla and (H) Xcla-⌬C (particularly in the case of Xcla) could be in part responsible mRNA-injected embryos. for the left–right phenotype.

FIG. 7. Gap junction communication (GJC) is not inhibited by overexpression of Xcla or Xcla-⌬C. Albino embryos injected with either Xcla (A and B) or Xcla-⌬C (C and D) prior to the ﬁrst cleavage, followed by one injection of a mixture of Lucifer yellow (LY) and rhodamine-dextran (RLD) at the 16-cell stage, are shown under ﬂuorescence. The presence of a cell with LY only (arrows) adjacent to a cell which was injected with both LY ϩ RLD (arrowheads) indicates dye transfer through gap junctions. Heptanol is an inhibitor of GJC; the heptanol-treated embryo shown in E and F is an example of inhibition of GJC. The heptanol treatment was less effective in our case (11% versus 5% for Levin and Mercola, 1998), perhaps due to subtle technical differences. (G and H) Percentage of embryos showing LY transfer after heptanol treatment (n ϭ 27), no treatment (n ϭ 49 albino; n ϭ 32 pigmented), Xcla injection (n ϭ 24) or

DISCUSSION the cells until they achieve barrier formation. These ﬁnd- ings correlate with a molecular model in which maturation Xcla and Cell Junctions in the Early Embryo of the tight junction occurs as ZO-1 sequentially interacts with transmembrane components of the tight-junction In the course of a screen designed to identify membrane complex (Rajasekaran et al., 1996). or secreted molecules translated during early Xenopus development using microsome fractionation, we found that the Xcla transcript is an abundant microsome-bound Xcla mRNA and Left–Right Asymmetry mRNA in the Xenopus blastula. Using microinjected ani- The left–right polarity phenotype of overexpressing Xcla mal cap explants, we have shown that the COOH-tail of in Xenopus embryos opens exciting developmental ques- Xcla is located intracellularly and that it is required for the tions. Xcla mRNA causes a randomization of the situs correct localization of Xcla to the cell surface membrane as preceded by bilateral expression of Xnr-1. This left-sided well as for its colocalization with ZO-1. Our experiments expression of Xnr-1 in the lateral plate mesoderm is the indicate that Xcla, like other members of the claudin earliest marker so far characterized for left–right asymme- family, is able to concentrate at cell contact points in try in Xenopus (Capdevila et al., 2000). Hyatt and Yost L-cells which form tight-junction-like structures and that (1998) have proposed that two subsequent steps cause this the ability of Xcla to form these structures requires the asymmetric expression. Early in development, a left–right putative PDZ-binding region. coordinator interacts with the Spemann organizer and to- It has been previously reported that Claudin-1 and gether both establish left–right asymmetry. Although not Claudin-2 can recruit occludin into the tight-junction com- yet proven deﬁnitively, the TGF␤ molecule Vg1 appears to plex (Furuse et al., 1998b). Our present results show that be a key player in this process. Later, the establishment of the COOH-terminus of Xcla is required for interaction with a midline barrier is thought to enforce asymmetries (Meno ZO-1, allowing both proteins to be relocated from the et al., 1998). The perturbation of tight-junction assembly by cytoplasm to the membrane, such that ZO-1 can then overexpression of Xcla or Xcla-⌬C might interfere with the function to correctly position the proteins within the apical later event in left–right patterning since, in contrast to Vg1 region of the membrane, a prerequisite for mature tight- (Hyatt and Yost, 1998), it does not lead to right-sided but junction formation (Merzdorf et al., 1998; Itoh et al., rather to bilateral expression of Xnr-1. 1999b). As shown in Fig. 8, both occludin and claudins have It has been proposed that Lefty-1 may provide a repressive four transmembrane domains, two extracellular loops, one signal that prevents left-sided determinants from crossing intracellular loop, and a cytoplasmic tail (Furuse et al., the midline to the right side of the embryo (Meno et al., 1993, 1998a; this work). This protein structure and topology 1998). Mutants that lack this protein show isomerism and is shared by another family of junction proteins, the con- have bilateral expression of the left-side determinants nexin gap-junction proteins (Goodenough et al., 1996). Nodal and lefty-2 (Meno et al., 1998). Recently, another The hypothetical model shown in Fig. 8 indicates how molecule, Caronte, has been implicated as an inducer of Xcla could function in the localization of multiple proteins left-sided Nodal expression in chicken (Rodriguez Esteban into a complex in the tight junction. Oligomers of Occludin et al., 1999; Yokouchi et al., 1999). SHH and Pitx2 are might bind to ZO-1 (Furuse et al., 1994; Chen et al., 1997), positive factors inducing the left-sided determination, which may then be recruited to the surface membrane by while cSnR and FGF8 have been implicated in the right- interaction with the COOH-terminus of Claudin. Several sided pathway in chicken (Ryan et al., 1998; Patel et al., models have been described for the formation of mature and 1999; Isaac et al., 1997; Boettger et al, 1999). While in functional tight junctions (Merzdorf et al., 1998; Rajaseka- Xenopus no asymmetric gene expression before Xnr-1 has ran et al., 1996; Citi and Cordenosi, 1998). Tight-junction been reported to date, it is, however, possible that some formation in Xenopus cleavage stage embryo, as demon- members of the hedgehog family and cerberus-related genes strated by Merzdorf et al. (1998), starts deep in the intercel- may also be involved in this process. Interestingly, cells lular spaces, leaving exposed ion-permeable membrane in from desert hedgehog (dhhϪ/Ϫ) mutant mice have abnormal the cleavage furrows, and as these junctions mature, they tight junctions that allow the passage of proteins and other move toward a progressively more apical position between macromolecules through the intercellular space, suggesting

Xcla-⌬C injection (n ϭ 37 albino; n ϭ 69 pigmented). Ectopic expression of Xcla somewhat increased the percentage of embryos with GJC (G). This observation might be explained if the tighter adhesion of cells injected with Xcla mRNA led to an increase in gap-junction communication. (H) Dorsal injections into pigmented embryos showing a higher frequency of GJC that is, nevertheless, not affected by Xcla-⌬C mRNA microinjection. FIG. 8. A model for tight-junction assembly. Oligomers of occludin may form a complex with ZO-1. Claudin binding to ZO-1 may then facilitate the assembly of the mature apical tight junction. Transmembrane proteins located at the tight junction associate with cytoplasmic proteins ZO-2 and ZO-3, which also bind to ZO-1 and actin.

Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. Xenopus Claudin and Left–Right Asymmetry 227 that hedgehog may not only have an instructive role, but in the brain, heart, and gut (Essner et al., 2000; Campione et may also actively maintain cell polarity (Parmantier et al., al., 1999). Again, this phenotype is thought to be caused by 1999). Mice mutant for claudin-11 also have abnormal tight effects on cell behavior. Overexpression of the different junctions, and dhh and claudin-11 are expressed in similar Pitx2 isoforms leads to increased cellular adhesion during tissues of the embryo (Morita et al., 1999c; Gow et al., early development and it is therefore tempting to speculate 1999). that Pitx2 regulates asymmetric organ development in a Experiments by Levin and Mercola have demonstrated similar manner (Essner et al., 2000). Other researchers have that interfering with gap junctions causes heterotaxia in implicated gap junctions in left–right determination based Xenopus (Levin and Mercola, 1998). We now report that a on studies involving connexins in human mutations and in heterotaxia phenotype can be obtained by overexpression of Xenopus and chick embryo experiments (Britz- the tight-junction claudin constructs Xcla-⌬C and Xcla. Cunningham et al., 1995; Levin and Mercola, 1998, 1999). However, overexpression of Xcla constructs did not prevent The results reported here implicate the tight-junction Xcla gap-junction communication between cells in this assay. protein in left–right determination (Figs. 6F–6H). Taken Experiments with Lucifer yellow showed dye transfer together, these findings suggest that tight, adherens, and through gap junctions in embryos injected with Xcla-⌬C or gap junction cell molecules play important roles in left– wild-type Xcla. In our assay for Xnr-1 expression, we right asymmetry and it will be an important issue for the observed a striking increase in bilateral Xnr-1 expression future to understand how the functions of the different after overexpression of Xcla or Xcla-⌬C mRNA. In the case junctions are related to each other. of gap junctions, the predominant phenotype seen by over- In conclusion, interfering with the function of any of the expression of mutant or wild-type connexins was the loss of three types of cell junctions present in epithelial cells can Xnr-1 expression, although some bilateral expression was lead to changes in left–right situs: gap junctions (Levin and also observed, but to a lesser extent in the case of a Mercola, 1998), adherens junctions (Garcia Castro et al., dominant-negative connexin mRNA (Levin and Mercola, 2000), and tight junctions. The results presented here for 1998). Since claudins cause ectopic Xnr-1 expression on the Xcla, a tight-junction protein that can affect cell adhesion, right side (resulting in bilateral expression), whereas con- highlight the importance of correct cell–cell interactions nexins cause mostly loss of normal Xnr-1 expression on the for left–right axis determination during vertebrate develop- left side, we suggest that changes in both types of epithelial ment and introduce a novel molecule involved in this cell junction might lead to left–right asymmetries. The process. molecular mechanism in both cases would be upstream of the Nodal signaling pathway since both show alterations of Xnr-1 expression. Because Xcla (but not Xcla-⌬C) mRNA ACKNOWLEDGMENTS caused an increase in GJC, the possibility that the effect of Xcla on left–right asymmetry might be mediated by Thanks are due to Professor Daniel A. Goodenough for gener- changes in GJC cannot be eliminated. While we cannot ously providing the ZO-1 antibody. We also thank D. Geissert, C. conclude that multiple pathways exist, the data indicate Kemp, and A. Cuellar for technical assistance and the members of that tight-junction proteins (this work), gap-junction pro- the De Robertis lab for helpful comments on the manuscript. This teins (Levin and Mercola, 1998), and zonula adherens pro- work was supported by the NIH Grant R37 HD 21503-15. B.J.B. is teins (Garcia-Castro et al., 2000) may be all involved in supported by a supplemental grant from the NICHD and O.W. by left–right determination. The simplest interpretation is an HFSPO postdoctoral fellowship. E.D.R. is an Investigator of the that changes in cell–cell interactions can lead to changes in Howard Hughes Medical Institute. left–right symmetry. The ability of Xcla to affect cell adhesion and migration suggests that cell contacts during early Xenopus develop- REFERENCES ment influence left–right asymmetries. This has recently also been underscored by studies in both chicken and Adelman, M. R., Blobel, G., and Sabatini, D. D. (1973). An im- zebrafish. Garcia-Castro et al. (2000) report that molecules proved cell fractionation procedure for the preparation of rat liver governing cell adhesion or migration direct asymmetry in membrane-bound ribosomes. J. Cell Biol. 56, 191–205. avians. Their studies show that the adherens junction Belo, J. A., Bouwmeester, T., Leyns, L., Kertesz, N., Gallo, M., protein N-cadherin has both the spatial and the temporal Follettie, M., and De Robertis, E. M. (1997). Cerberus-like is a asymmetrical expression appropriate to be involved in secreted factor with neutralizing activity expressed in the ante- left–right asymmetry. They further show that inhibiting rior primitive endoderm of the mouse gastrula. Mech. Dev. 68, 45–57. N-cadherin with specific antibodies causes randomization Boettger, T., Wittler, L., and Kessel, M. (1999). FGF8 functions in of heart looping and that blocking N-cadherin function the specification of the right body side of the chick. Curr. Biol. 9, altered the expression of Snail and Pitx2 but not that of 277–280. Nodal (Garcia-Castro et al., 2000). In zebrafish, differen- Blumberg, B., Wright, C. V., De Robertis, E. M., and Cho, K. W. tially regulated isoforms of the transcription factor Pitx2 (1991). Organizer-specific homeobox genes in Xenopus laevis are implicated in the development of left–right asymmetry embryos. Science 253, 194–196.

Briel, M. M., and Miesfeld, R. L. (1991). Isolation and characteriza- localizing at tight junctions with no sequence similarity to tion of transcripts induced by androgen withdrawal and apoptotic occludin. J. Cell Biol. 141, 1539–1550. cell death in rat ventral prostate. Mol. Endocrinol. 5, 1381–1388. Furuse, M., Sasaki, H., Fujimoto, K., and Tsukita, S. (1998b). A Britz-Cunningham, S. H., Shah, M. M., Zuppan, C. W., and single gene product, claudin-1 or -2, reconstitutes tight junction Fletcher, W. H. (1995). Mutations of the Connexin43 gap- strands and recruits occludin in fibroblasts. J. Cell Biol. 143, junction gene in patients with heart malformations and defects 391–401. of laterality. N. Engl. J. Med. 332, 1323–1329. Garcia-Castro, M. I., Vielmetter, E., and Bronner-Fraser M. (2000). Campione, M., Steinbeisser, H., Schweickert, A., Deissler, K., van N-cadherin, a cell adhesion molecule involved in establishment Bebber, F., Lowe, L. A., Nowotschin, S., Viebahn, C., Haffter, P., of embryonic left–right asymmetry. Science 288, 1047–1051. Kuehn, M. R., and Blum, M. (1999). The homeobox gene Pitx2: Goodenough, D. A., Goliger, J. A., and Paul, D. L. (1996). Connex- Mediator of asymmetric left–right signaling in vertebrate heart ins, connexons, and intercellular communication. Annu. Rev. and gut looping. Development 126, 1225–1234. Biochem. 65, 475–502. Capdevila, J., Vogan, K. J., Tabin, C. J., and Izpisuá Belmonte, J. C. Goliger, J. A., and Paul, D. L. (1996). Connexins, connexons, and (2000). Mechanisms of left–right determination in vertebrates. intercellular communication. Annu. Rev. Biochem. 65, 475–502. Cell 101, 9–21. Gow, A., Southwood, C. M., Li, J. S., Pariali, M., Riordan, G. P., Cardellini, P., Davanzo, G., and Citi, S. (1996). Tight junctions in Brodie, S. E., Danias, J., Bronstein, J. M., Kachar, B., and Lazza- early amphibian development: Detection of junctional cingulin rini, R. A. (1999). CNS myelin and Sertoli cell tight junction from the 2-cell stage and its localization at the boundary of strands are absent in Osp/claudin-11 null mice. Cell 99, 649– distinct membrane domains in dividing blastomeres in low 659. calcium. Dev. Dyn. 207, 104–113. Gumbiner, B., Lowenkopf, T., and Apatira, D. (1991). Identification Chen, Y., Merzorf, C., Paul, D. L., and Goodenough, D. A. (1997). of a 160-kDa polypeptide that binds to the tight junction protein COOH terminus of occludin is required for tight junction barrier ZO-1. Proc. Natl. Acad. Sci. USA 88, 3460–3464. function in early Xenopus embryos. J. Cell Biol. 138, 891–899. Haskins, J., Gu, L., Wittchen, E. S., Hibbard, J., and Stevenson, B. R Chen, Z., Zandonatti, M., Jakubowski, D., and Fox, H. S. (1998). (1998). ZO-3, a novel member of the MAGUK protein family Brain capillary endothelial cells express MBEC1, a protein that is found at the tight junction, interacts with ZO-1 and occludin. related to the Clostridium perfringens enterotoxin receptors. J. Cell Biol. 141, 199–208. Lab. Invest. 78, 353–363. Hyatt, B. A., and Yost, H. J. (1998). The left–right coordinator: The Citi, S. (1993). The molecular organization of tight junctions. J. Cell role of Vg1 in organizing left–right axis formation. Cell 93, Biol. 121, 485–489. 37–46. Citi, S., and Cordenosi, M. (1998). Tight junction proteins. Bio- Isaac, A., Sargent, M. G., and Cooke, J. (1997). Control of vertebrate chim. Biophys. Acta 1448, 1–11. left–right asymmetry by a snail-related zinc finger gene. Science Diehn, M., Eisen, M. B., Botstein, D., and Brown, P. O. (2000). 275, 1301–1304. Large-scale identification of secreted and membrane-associated Itoh, M., Morita, K., and Tsukita, S. (1999a). Characterization of gene products using DNA microarrays. Nat. Genet. 25, 58–62. ZO-2 as a MAGUK family member associated with tight as well Essner, J. J., Branford, W. W., Zhang, J., and Yost, H. J. (2000). as adherens junctions with a binding affinity to occludin and Mesendoderm and left–right brain, heart and gut development alpha catenin. J. Biol. Chem. 274, 5981–5986. are differentially regulated by pitx2 isoforms. Development 127, Itoh, M., Furuse, M., Morita, K., Kubota, K., Saitou, M., and 1081–1093. Fanning, A. S., Jameson, B. J., Jesaitis, L. A., and Anderson, J. M. Tsukita, S. (1999b). Direct binding of three tight junction- (1998). The tight junction protein ZO-1 establishes a link be- associated MAGUKs, ZO-1, ZO-2, and ZO-3, with the COOH tween the transmembrane protein occludin and the actin cy- termini of claudins. J. Cell Biol. 147, 1351–1363. toskeleton. J. Biol. Chem. 273, 29745–29753. Katahira, J., Sugiyama, H., Inoue, N., Horiguchi, Y., Matsuda, M., Fanning, A. S., and Anderson, J. M. (1999). PDZ domains: Funda- and Sugimoto, N. (1997). Clostridium perfringens enterotoxin mental building blocks in the organization of protein complexes utilizes two structurally related membrane proteins as func- at the plasma membrane. J. Clin. Invest. 103, 767–772. tional receptors in vivo. J. Biol. Chem. 272, 26652–26658. Fanning, A. S., Mitic, L. L., and Anderson, J. M. (1999). Transmem- Kim, S.-H., Yamamoto, A., Bouwmeester, T., Agius, A., and De brane proteins in the tight junction barrier. J. Am. Soc. Nephrol. Robertis, E. M. (1998). The role of paraxial protocadherin in 10, 1337–1345. selective adhesion and cell movements of the mesoderm during Fleming, T. P., Hay, M., Javed, Q., and Citi, S. (1993). Localization Xenopus gastrulation. Development 125, 4681–4691. of tight junction protein cingulin is temporally and spatially Klein, S. L. (1987). The first cleavage furrow demarcates the regulated during early mouse development. Development 117, dorsal–ventral axis in Xenopus embryos. Dev. Biol. 120, 299– 1135–1144. 304. Furuse, M., Hirase, T., Itoh, M., Nagafuchi, A., Yonemura, S., Kojima, T., Sawada, N., Chiba, H., Kokai, Y., Yamamoto, M., Tsukita, S., and Tsukita, S. (1993). Occludin: A novel integral Urban, M., Lee, G. H., Hertzberg, E. L., Mochizuki, Y., and Spray, membrane protein localizing at tight junctions. J. Cell Biol. 123, D. C. (1999). Induction of tight junctions in human connexin 32 1777–1788. (hCx32)-transfected mouse hepatocytes: Connexin 32 interacts Furuse, M., Itoh, M., Hirase, T., Nagafuchi, A., Yonemura, S., with occludin. Biochem. Biophys. Res. Commun. 266, 222–229. Tsukita, S., and Tsukita, S. (1994). Direct association of occludin Kopczynski, C. C., Noordermeer, J. N., Serano, T. L., Chen, W.-Y., with ZO-1 and its possible involvement in the localization of Pendleton, J. D., Lewis, S., Goodman, C. S., and Rubin, G. M. occludin at tight junctions. J. Cell Biol. 127, 1617–1626. (1998). A high throughput screen to identify secreted and trans- Furuse, M., Fujita, K., Hiiragi, T., Fujimoto, K., and Tsukita, S. membrane proteins involved in Drosophila embryogenesis. Proc. (1998a). Claudin-1 and -2: Novel integral membrane proteins Natl. Acad. Sci. USA 95, 9973–9978.

Kubota, K., Furuse, M., Sasaki, H., Sonoda, N., Fujita, K., Nagafu- Rajasekaran, A. K., Hojo, M., Huima, T., and Rodriguez-Boulan, E. chi, A., and Tsukita, S. (1999). Ca(2ϩ)-independent cell-adhesion (1996). Catenins and zonula ccludens-1 form a complex during activity of claudins, a family of integral membrane proteins early stages in the assembly of tight junctions. J. Cell Biol. 132, localized at tight junctions. Curr. Biol. 9, 1035–1038. 451–463. Levin, M., and Mercola, M. (1998). Gap junctions are involved in Rodriguez Esteban, C., Capdevila, J., Economides, A. N., Pascual, J., the early generation of left–right asymmetry. Dev. Biol. 203, Ortiz, A., and Izpisua Belmonte, J. C. (1999). The novel Cer-like 90–105. protein Caronte mediates the establishment of embryonic left– Levin, M., and Mercola, M. (1999). Gap junction-mediated transfer right asymmetry. Nature 401, 243–251. of left–right patterning signals in the early chick blastoderm is Ryan, A. K., Blumberg, B., Rodriguez-Esteban, C., Yonei-Tamura, upstream of Shh asymmetry in the node. Development 126, S., Tamura, K., Tsukui, T., de la Pena, J., Sabbagh, W., Green- 4703–4714. wald, J., Choe, S., Norris, D. P., Robertson, E. J., Evans, R. M., Lohr, J. L., Danos, M. C., and Yost, H. J. (1997). Left–right Rosenfeld, M. G., and Izpisua Belmonte, J. C. (1998). Pitx2 asymmetry of a Nodal-related gene is regulated by dorsoanterior determines left–right asymmetry of internal organs in verte- midline structures during Xenopus development. Development brates. Nature 394, 545–551. 124, 1465–1472. Sheth, B., Fesenko, I., Collins, J. E., Moran, B., Wild, A. E., Lowe, L. A., Supp, D. M., Sampath, K., Yokoyama, T., Wright, Anderson, J. M., and Fleming, T. P. (1997). Tight junction C. V., Potter, S. S., Overbeek, P., and Kuehn, M. R. (1996). assembly during mouse blastocyst formation is regulated by late Conserved left–right asymmetry of Nodal expression and alter- expression of ZO-1 alphaϩ isoform. Development 124, 2027– Nature ations in murine situs inversus. 381, 158–161. 2037. Martin-Padura, I., Lostaglio, S., Schneemann, M., Williams, L., Sheth, B., Moran, B., Anderson, J. M., and Fleming, T. P. (2000). Romano, M., Fruscella, P., Panzeri, C., Stoppacciaro, A., Ruco, L., Post-translational control of occludin membrane assembly in Villa, A., Simmons, D., and Dejana, E. (1998). Junctional adhe- mouse trophectoderm: A mechanism to regulate timing of tight sion molecule, a novel member of the immunoglobulin super- junction biogenesis and blastocyst formation. Development 127, family that distributes at intercellular junctions and modulates 831–840. monocyte transmigration. J. Cell Biol. 142, 117–127. Sirotkin, H., Morrow, B., Saint-Jore, B., Puech, A., Das Gupta, R., Merzdorf, C. S., Chen, Y. H., and Goodenough D. A. (1998). Patanjali, S. R., Skoultchi, A., Weissman, S. M., and Kucher- Formation of functional tight junctions in Xenopus embryos. lapati, R. (1997). Identification, characterization, and precise Dev. Biol. 195, 187–208. mapping of a human gene encoding a novel membrane-spanning Miller, J. R., and Moon, R. (1997). Analysis of the signaling activities of localization mutants of beta-catenin during axis protein from the 22q11 region deleted in velo-cardio-facial syn- specification in Xenopus. J. Cell Biol. 139, 229–243. drome. Genomics 42, 245–251. Morita, K., Furuse, M., Fujimoto, K., and Tsukita, S. (1999a). Songyang, Z., Fanning, A. S., Fu, C., Xu, J., Marfatia, S. M., Chishti, Claudin multi-gene family encoding four-transmembrane do- A. H., Crompton, A., Chan, A. C., Anderson, J. M., and Cantley, main protein components of tight junction strands. Proc. Natl. L. C. (1997). Recognition of unique carboxyl-terminal motifs by Acad. Sci. USA 96, 511–516. distinct PDZ domains. Science 275, 73–77. Morita, K., Sasaki, H., Furuse, M., and Tsukita, S. (1999b). Endo- Stevenson, B. R., Siliciano, J. D., Mooseker, M. S., and Goodenough, thelial claudin: Claudin-5/TMVCF constitutes tight junctions D. A. (1986). Identification of ZO-1: A high molecular weight strands in endothelial cells. J. Cell Biol. 147, 185–194. polypeptide associated with the tight junction (zonula occludens) Morita, K., Sasaki, H., Fujimoto, K., Furuse, M., and Tsukita, S. in a variety of epithelia. J. Cell Biol. 103, 755–766. (1999c). Claudin-11/OSP-based tight junctions of myelin Toyofuku, T., Yabuki, M., Otsu, K., Kuzuya, T., Hori, M., and Tada, sheaths in brain and Sertoli cells in testis. J. Cell Biol. 145, M. (1998). Direct association of the gap junction protein 579–588. connexin-43 with ZO-1 in cardiac myocytes. J. Biol. Chem. 273, Mu¨ller, H., and Hausen, P. (1995). Epithelial cell polarity in early 12725–12731. Xenopus development. Dev. Dyn. 202, 405–420. Yokouchi, Y., Vogan, K. J., Pearse, R. V., 2nd, and Tabin, C. J. Newport, J., and Kirscher, M. (1982). A major developmental (1999). Antagonistic signaling by Caronte, a novel Cerberus- transition in early Xenopus embryos. II. Control of the onset of related gene, establishes left–right asymmetric gene expression. transcription. Cell 30, 687–696. Cell 98, 573–583. Nieuwkoop, P., and Faber, J. (1994). “Normal Table of Xenopus Yost, H. J. (1992). Regulation of left–right asymmetries by extra- laevis, Daudin,” 2nd ed. North Holland, Amsterdam. [Reprinted cellular matrix. Nature 357, 158–161. by Garland Publishing, New York] Zernicka-Goetz, M., Pines, J., Ryan, K., Siemering, K. R., Haseloff, Parmantier, E., Lynn, B., Lawson, D., Turmaine, M., Namini, J., Evans, M. J., and Gurdon, J. B. (1996). An indelible lineage S. S., Chakrabarti, L., McMahon, A. P., Jessen, K. R., and marker for Xenopus using a mutated green fluorescent protein. Mirsky, R. (1999). Schwann cell-derived Desert hedgehog Development 122, 3719–3724. controls the development of peripheral nerve sheaths. Neuron 23, 713–724. Received for publication May 19, 2000 Patel, K., Isaac, A., and Cooke, J. (1999). Nodal signalling and the Revised November 9, 2000 roles of the transcription factors SnR and Pitx2 in vertebrate Accepted November 9, 2000 left–right asymmetry. Curr. Biol. 9, 609–612. Published online January 19, 2001