Many animal organs are composed of epithelial tubes, each with a lumen. The epithelial tissue maintains a distinct environment of luminal and interstitial compartments by regulating solute transport and functioning as a barrier to the diffusion of solutes between these two compartments. The functions of the tissue largely depend on the polarized nature of the cells. Their plasma membrane is functionally divided into apical domain facing the lumen and basolateral domain contacting adjacent cells and the underlying tissue, which membrane domains differ in the compositions of proteins and lipids. In vertebrates, these two membrane domains of epithelial cells are separated by tight junctions (TJs), which appear in ultrathin section electron photomicrographs as a series of focal contacts between adjacent cells, and in freeze-fracture electron photomicrographs, as a set of continuous, circumferential networks of intramembranous particle strands. TJs not only restrict the diffusion of molecules between the apical and basolateral membrane domains but also function as the major barrier regulating paracellular movement of water and solutes across epithelial sheets, thereby playing a critical role in the physiological functions of tubular organs (Anderson, 2001; Tsukita et al., 2001).
In several tubular organs of mammals, such as the salivary and mammary glands, lumens and TJs are established de novo in the epithelium during development (Redman, 1987; Hieda and Nakanishi, 1997; Nanba et al., 2001; Hogan and Kolodziej, 2002). In the submandibular salivary gland (SMG) of mice the epithelial tissue arises at embryonic day (E) 11 to make solid cell masses with no lumens. These epithelial cells lack TJs and do not have defined apical and basolateral membrane domains. As the epithelium grows and branches into lobules and stalks, lumens appear first in the stalk at E14, at which time the luminal cells have formed TJs, resulting in the ductal structure of the stalk epithelium. Subsequently, the lobules develop a lumen covered with TJ-bearing cells by E16 to form terminal tubules or acini, leading to the establishment of the complete tubular structure in the entire epithelium (Redman, 1987; Hieda et al., 1996; Hieda and Nakanishi, 1997). Little is known, however, about the molecular aspect of TJ assembly and its roles in the lumen formation of these developing organs.
The claudin transmembrane proteins, with molecular masses of approximately 23 kDa, have been identified recently as components of TJ strands. Claudins comprise a large family consisting of more than 20 members in mammals, and their expression pattern varies considerably among tissues (Furuse et al., 1998a; Morita et al., 1999a; Rahner et al., 2001; Tsukita et al., 2001; Kiuchi-Saishin et al., 2002). Several lines of evidence have demonstrated the direct involvement of claudins in the formation of TJ strands as well as their barrier function (Tsukita et al., 2001). For instance, claudins expressed singly in fibroblasts lacking TJs reconstitute well-developed networks of TJ strands between the cells (Furuse et al., 1998b; Morita et al., 1999b, c, 2002). In contrast, these strands in MDCK epithelial cells, which express primarily claudin-1 and -4, are disorganized upon the specific down-regulation of claudin-4 induced by treatment with a Clostridium perfringens enterotoxin fragment (Sonoda et al., 1999). Notably, the toxin-treated cells exhibit a significant increase in TJ permeability. Also, the renal absorption of Mg2+ ions through the paracellular pathway depends on claudin-16, whose mutation is associated with the human disease known as hereditary hypomagnesemia (Simon et al., 1999).
In the present study, we examined the expression patterns of claudins (claudin-1 to -12) in the developing mouse SMG. Claudin-3 to -8, -10, and -11 were found to be localized at TJs of luminal cells in the developing epithelium and to exhibit the spatiotemporally regulated patterns of expression. Claudin-5 was also expressed in mesenchymal cells, probably endothelial cells. Our data provide a clue to uncovering not only the roles of claudin-based TJs in epithelial tubulogenesis but also the mechanisms regulating these physiological functions of the junctions.
RESULTS AND DISCUSSION
Expression of Claudin Members in Developing SMG
The expression of mRNAs for claudin-1 to -12 in the developing mouse SMG was examined by reverse transcriptase-polymerase chain reaction (RT-PCR) using primer sequences specific for these molecules. PCR products with the expected size were hardly detectable for claudin-2 and -9 but were clearly detected for claudin-1, -3 to -8, and -10 to -12 (Fig. 1A). The identity of the PCR products was verified by sequencing. Next, the protein expression of claudin-1 to -8, -10, and -11 was analyzed by Western blotting using specific antibodies characterized previously (Furuse et al., 1999; Kubota et al., 1999; Morita et al., 1999a– c; Kiuchi-Saishin et al., 2002). Claudin-9 and -12 were not examined because specific antibodies for them were not available. As shown in Figure 1B, antibodies against claudin-1, -3 to -8, and -10 recognized proteins with molecular mass of approximately 23 kDa in the detergent-insoluble fraction of E16 SMG rudiments, although the claudin-6 signal was rather weak. In addition to the 23-kDa protein, a protein of approximately 19 kDa was detected with anti–claudin-10 antibody, which may represent a claudin-10 isoform of 194 amino acids (DDBJ/GenBank/EMBL accession no. BC021770). Claudin-2 and -11 were almost negative under our conditions, although claudin-11 mRNA was detected as described above. These data indicate the expression of multiple claudins in the developing submandibular gland.
Differential Expression Patterns of Claudins in Developing SMG Epithelium
Immunofluorescence microscopy was performed to examine the expression patterns of claudins, except claudin-9 and -12 in the developing SMG. Tissue sections of the glands taken from E14, E16, and newborn mice were double-stained for a given claudin and ZO-1, a cytoplasmic protein localized at tight junctions. At all of the developmental stages examined, staining for claudin-1 was weak and obscure and that for claudin-2 was hardly detectable (data not shown). It turned out, however, that antibodies against the eight other claudins, including claudin-3 to -8, -10, and -11, gave clear staining at TJs in the developing SMG, displaying the differential expression patterns as described below in detail. A few claudins were present at the basolateral membrane as well as TJs, as reported previously (Gregory et al., 2001; Rahner et al., 2001; Coyne et al., 2003).
Claudin-3, -5, and -7.
These claudins were expressed in all luminal cells of the epithelium at all of the developmental stages examined. At E14, claudin-3 showed a thin line of distribution in colocalization with ZO-1 along the lumen appearing in the ducts but was not detectable in lobules composed of cell masses (Fig. 2A–C). By E16, the distal, lobular epithelium also developed lumens to form terminal tubules, and claudin-3 was precisely colocalized with ZO-1 along these lumens (Fig. 2D–F). This claudin was also localized at ZO-1–positive TJs in the entire submandibular gland epithelium of newborn mice (Fig. 2G–I). The colocalization with ZO-1 was also observed for claudin-5 (Fig. 2J–L) and claudin-7 (Fig. 2M–O) in almost all of the luminal cells at all of the developmental stages examined, although the concentration of claudin-7 at TJs was less obvious due to its intense and uniform staining along the basolateral surface of the cells.
We noted that, in addition to its presence in epithelial cells, claudin-5 was also expressed and colocalized with ZO-1 in certain mesenchymal cells (Fig. 2J–L). No staining for the other claudins examined was detectable in mesenchyme (see below). Also, occludin, another TJ transmembrane protein (Furuse et al., 1993; Saitou et al., 1997), was undetectable in mesenchyme; however, it was precisely colocalized with ZO-1 in epithelium (data not shown). The expression profile of TJ molecules suggests the mesenchymal cells positive for claudin-5 to be endothelial cells (Morita et al., 1999c; Tsukita et al., 2001).
Claudin-4, -6, and -8.
In addition to claudin-3 to -5, claudin-4, -6, and -8 were expressed and localized at TJs in the ducts of the E14 SMG (data not shown). Examination of SMGs at later developmental stages, however, revealed the spatially and temporally regulated expression patterns of these claudins. Claudin-4 in the E16 SMG was expressed and colocalized with ZO-1 in the ducts, with its staining extending to the vicinity of the terminal tubules (Fig. 3A,B). In the postnatal SMG, the expression of this claudin was restricted to the ducts (Fig. 3C,D). Claudin-6 was expressed and concentrated at TJs only in the ducts at E16 (Fig. 3E,F), whereas after birth it was almost completely absent (Fig. 3G,H). This finding is consistent with its embryo-specific expression in various organs (Morita et al., 1999a, 2002). The obvious expression and concentration at TJs of claudin-8 were detected mainly in the ducts at E16 (Fig. 3I,J) but were found in both the ducts and terminal tubules of the postnatal SMG (Fig. 3K,L).
Claudin-10 and -11.
Unlike that of claudin-3 to -8, the expression of claudin-10 and -11 was undetectable in the E14 SMG (data not shown), but it became evident in E16 and postnatal SMGs. Notably, staining for claudin-10 was hardly recognizable in the ducts but was obvious in the terminal tubules, where this claudin molecule was colocalized with ZO-1 at the apical side of cells, although it was also present at the basolateral surfaces (Fig. 4A–D). Similarly, staining for claudin-11 was detected mainly in terminal tubules of both E16 and postnatal SMGs; however, this claudin molecule appeared to be expressed also in the ductal region adjacent to the terminal tubules (Fig. 4E–H). In addition, compared with claudin-10, claudin-11 showed weaker staining intensity and was precisely colocalized with ZO-1, which may explain the failure to detect claudin-11 on Western blots.
To date, more than 20 members of the claudin TJ proteins have been identified; however, little information is available about the expression patterns of these claudin members in developing organs. We showed in the present study that, of the 12 members of the claudin family investigated (claduin-1 to -12), claudin-3 to -8, -10, and -11 were expressed in epithelium of the developing mouse SMG, with claudin-5 also expressed in mesenchyme, probably in endothelial cells. Notably, immunofluorescence microscopy demonstrated that the expression pattern of these claudins was spatiotemporally regulated in the developing epithelium. In the SMG of newborn mice, claudin-3, -5, -7, and -8 were expressed in both the ducts and terminal tubules; claudin-4 in the ducts; and claudin-10 and -11 in the terminal tubules. The region-specific combinations of claudins in the SMG epithelium may contribute to the differences in the transepithelial electrical resistance and permeability to molecules between the ducts and the terminal tubules or acini (Hand, 1987), because variations in the tightness of TJ strands are determined by the combinations and mixing ratios of claudin species (Simon et al., 1999; Tsukita and Furuse, 2000; Furuse et al., 2001; Tsukita et al., 2001). It is of note that salivary glands exhibit increased permeability of TJs after adrenergic stimulation (Hand, 1987; Segawa, 1994). This change might involve the transient regulation of claudin expression and interaction between claudin members coexpressed. We have not determined the distribution of claudin-12 in the SMG, and it would also be possible that some of the claudins that were not examined here are also expressed in the gland. Yet, the SMG may serve as a model to study not only roles of claudin-based TJs in the development of tubular organs but also the mechanism(s) regulating the physiological functions of these junctions.
SMG rudiments were dissected from ddY strain mice (Nihon SLC, Hamamatsu, Japan) at E14, E16, and postnatal day 1 in Hanks' balanced salt solution. The discovery of the vaginal plug was designated as E0.
Guinea pig anti–claudin-1 antibody, rabbit antibodies against claudin-2, -3, -4, -6, -8, -10, and -11, mouse anti–ZO-1 antibody, and rat anti-occludin antibody were characterized as described previously (Itoh et al., 1993; Saitou et al., 1997; Furuse et al., 1999; Kubota et al., 1999; Morita et al., 1999a– c; Kiuchi-Saishin et al., 2002). Rabbit antibodies against claudin-1, -5, and -7 were purchased from Zymed Laboratories (South San Francisco, CA).
RNA preparation from SMG rudiments, reverse transcription to cDNA, and PCR were performed as described previously (Umeda et al. 2001). PCR primers for claudin-1 to -4 and claudin-6 to -11 and those for claudin-12 were designed according to Turksen and Troy (2002) and Niimi et al. (2001), respectively; and those for claudin-5 were 5′-ATGGGGTCTGCAGCGTTGGA-3′ (forward) and 5′-TGTCGTAATCGCCATTGGCC-3′ (reverse). The reaction mixtures were heated at 94°C for 5 min and then subjected to 35 cycles for amplification. Cycle parameters for amplification were as follow: for claudin-1, -3, -5, and -8, 94°C for 1 min, 65°C for 1 min, and 72°C for 1 min; for claudin-2, -4, - 6, -7, and -9 to -11, 94°C for 1 min, 60°C for 1 min, and 72°C for 1 min; and for claudin-12, 94°C for 1 min, 62°C for 1 min, and 72°C for 1 min. Aliquots of PCR products were subjected to agarose gel electrophoresis and visualized by ethidium bromide staining. The identity of the PCR products was verified by nucleotide sequencing.
SMG rudiments were homogenized in lysis buffer (20 mM Tris-HCl, 150 mM NaCl, 2 mM ethylenediaminetetraacetic acid (EDTA), 1 mM EGTA, 1% Triton X-100, 0.1 mM phenylmethanesulfonyl fluoride, 1 μg/ml leupeptin; pH 7.4), and incubated on ice for 1 hr. Homogenates were centrifuged at 15,000 × g for 30 min at 4°C, and the pellets were then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (12%). Proteins were electrophoretically transferred from the gel to the Hybond-P polyvinylidene difluoride membranes (Amersham Biosciences, Piscataway, NJ), which were then incubated with primary antibodies. Bound antibodies were detected with biotinylated secondary antibody and a VECTASTAIN ABC kit (Vector Laboratories, Burlingame, CA).
SMG rudiments were embedded in OCT compound and frozen in liquid nitrogen. Tissue sections of 6-μm thickness were cut in a cryostat, air-dried, and fixed in methanol at −20°C for 10 min. The sections were treated with 0.5% Triton X-100 in phosphate-buffered saline (PBS) for 10 min and subsequently with 1% bovine serum albumin. They were then incubated with primary antibodies for 1 hr, washed in PBS, and incubated for 1 hr with fluorescein isothiocyanate- or rhodamine- conjugated secondary antibody (Chemicon, Temecula, CA; Cappel, Durham, NC). After several washes with PBS, the sections were mounted with 50% glycerol/PBS containing 0.1% p-phenylenediamine and were observed with a fluorescence microscope (BX-50 epifluorescence microscope, Olympus, Tokyo, Japan) equipped with a charge-coupled device camera.