Claudin-11/OSP-based Tight Junctions of Myelin Sheaths in Brain and Sertoli Cells in Testis

Rat anti–mouse occludin mAb (MOC37) was raised and characterized as described previously (Saitou et al., 1997). Mouse anti-neurofilament 200-kD mAb and rat anti-FLAG mAb were purchased from Boehringer Mannheim Biochemicals and Eastman Kodak Co., respectively. Mouse L cells and their transfectants were grown in DME supplemented with 10% fetal calf serum.

Kidney total RNA was isolated according the method described by Chomczynski and Sacchi (1987). Poly(A)⁺ was prepared from the total RNA using oligo-dT cellulose (New England Biolabs, Inc.). First strand cDNA was prepared from this poly(A)⁺ RNA with Superscript™ II reverse transcriptase (GIBCO BRL) and used for PCR. Claudin-11/OSP cDNA was amplified from mouse kidney first strand DNA by PCR using the previously reported sequence of mouse OSP (Bronstein et al., 1996). The amplified cDNA was subcloned into pGEM-T Easy Vector (Promega). The lack of sequence errors was confirmed by DNA sequence analysis using a Dye Terminator Cycle Sequence Kit (Applied Biosystems, Inc.).

Mouse claudin-11/OSP was tagged with FLAG-peptide at its COOH terminus. To construct FLAG-tagged claudin-11/OSP-expression vectors, EcoRI site was introduced at the 3′-end of claudin-11/OSP cDNA by PCR, and amplified fragments were subcloned into pBluescript SK(−)- Flag-tag. The insert was excised by SalI-XbaI digestion followed by blunting with T4-polymerase, and then introduced into pCAGGSneodelEcoRI (Niwa et al., 1991), which was provided by Dr. J. Miyazaki (Osaka University). Mouse L cells were used for transfection as described previously (Furuse et al., 1998b), and the clones stably expressing tagged protein were screened by fluorescence microscopy with anti-FLAG mAb.

The expression of claudin-11/OSP in various mouse tissues was examined by Northern blotting using a Mouse Multiple Tissue Northern Blot (Clontech). The DNA fragment of mouse claudin-11/OSP (entire ORF) was radiolabeled with [³²P]dCTP and used as a probe for Northern blotting. Hybridization was performed in ExpressHyb™ Hybridization Solution (Clontech) at 68°C for 12 h. The membranes were washed with 2× SSC containing 0.1% SDS at room temperature for 30 min and then with 0.1× SSC containing 0.1% SDS at 50°C for 30 min. The membranes were exposed to imaging plates for 12 h, and the signals were visualized using a BAS2000 Bio-Imaging Analyzer (Fuji Photo Film Co. Ltd.).

A polypeptide, CNRFYYSSGSSSPTHAKSAHV, corresponding to the COOH-terminal cytoplasmic domain of mouse claudin-11/OSP was synthesized, and coupled via cysteine to keyhole limpet hemocyanin. This conjugated peptide was used as an antigen to generate polyclonal antibodies (pAb) in rabbits. The antiserum was affinity-purified with glutathione- S-transferase (GST) fusion protein with the COOH-terminal cytoplasmic domain of claudin-11/OSP.

Lysates of E. coli expressing GST/claudin fusion proteins were subjected to one-dimensional SDS-PAGE (12.5%), according to the method of Laemmli (1970), and gels were stained with Coomassie brilliant blue R-250. For immunoblotting, proteins were electrophoretically transferred from gels onto nitrocellulose membranes, which were then incubated with the first antibody. Bound antibodies were detected with biotinylated second antibodies and streptavidin-conjugated alkaline phosphatase (Amersham Corp.). Nitroblue tetrazolium and bromochloroindolyl phosphate were used as substrates for detection of alkaline phosphatase.

L transfectants plated on glass coverslips were fixed with 1% formaldehyde in PBS for 10 min at room temperature. Mouse brain and testis were frozen using liquid nitrogen, and frozen sections ∼10–20 μm thick were cut on a cryostat, mounted on glass slides, air-dried, and fixed in 95% ethanol at 4°C for 30 min followed by 100% acetone at room temperature for 1 min. These samples were processed for immunofluorescence microscopy as described previously (Furuse et al., 1998b; Morita et al., 1999). Specimens were observed using a fluorescence microscope, Zeiss Axiophot photomicroscope (Carl Zeiss, Inc.), or a Bio-Rad MRC 1024 confocal fluorescence microscope (Bio-Rad Laboratories) equipped with a Zeiss Axiophot photomicroscope. For each stereoscopic image (see Fig. 5), 30 optical sections (0.3–0.4-μm interval) were accumulated in the computer.

For conventional freeze-fracture analysis, tissues or cultured L fibroblasts were fixed in 2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.3) for 3 h at room temperature, washed with 0.1 M sodium cacodylate buffer three times, immersed in 30% glycerol in 0.1 M sodium cacodylate buffer for 2 h, and then frozen in liquid nitrogen. Frozen samples were fractured at −100°C and platinum-shadowed unidirectionally at an angle of 45° in Balzers Freeze Etching System (BAF060; Bal-Tec). The samples were then immersed in household bleach, and replicas floating off the samples were washed with distilled water. Replicas were picked up on formvar-filmed grids, and examined with a JEOL 1200EX electron microscope (JEOL) at an acceleration voltage of 100 kV.

The immunoelectron microscopic technique for examining freeze-fracture replicas was described in detail previously (Fujimoto, 1995; Moroi et al., 1998), except that samples were frozen in a high-pressure freezer (Baltec HPM010; Bal-Tec). Immunoelectron microscopy using ultrathin cryo-sections was performed essentially according to the method developed by Tokuyasu (Tokuyasu, 1980; Fujimoto et al., 1992). Samples were examined with a JEOL 1200EX electron microscope (JEOL) at an acceleration voltage of 80 kV.

Using the previously reported nucleotide sequence of mouse OSP (Bronstein et al., 1996), we amplified a full-length cDNA encoding mouse OSP by PCR, and confirmed that its open reading frame encoded a protein of 207 amino acids with a calculated molecular mass of 22.1 kD. OSP showed rather weak sequence similarity to claudins: it was almost equidistantly related to previously identified members of the claudin family (claudin-1 to -8; ∼30% identity at the amino acid sequence level to each member). As shown in Fig. 1, comparison between OSP and claudin-1 revealed that identical amino acids were almost evenly distributed throughout these molecules.

Next, we introduced cDNA encoding OSP with a FLAG- sequence at its COOH terminus into cultured L fibroblasts which lacked TJs or the expression of claudins (Furuse et al., 1998b). Immunofluorescence microscopy of the stable transfectants with anti-FLAG mAb showed that expressed FLAG-OSP was concentrated at cell–cell borders as planes or on thin cellular protrusions (Fig. 2, a–d). This mAb gave no signal from parent L fibroblasts. Then, these stable L transfectants expressing FLAG-OSP were fixed with glutaraldehyde and examined by conventional freeze-fracture electron microscopy (Fig. 2 e). In these cells, TJ strand/groove-like structures were frequently observed to be arranged in a parallel manner, whereas in parent L cells these structures were not detected. These strands were associated with the P-face, and were mostly discontinuous with intervening spaces of various widths (Fig. 2 e, inset). On the E-face, complementary continuous grooves were identified, containing scattered particles (Fig. 2 e, inset). The OSP-induced strands did not bifurcate frequently and showed a tendency to run parallel to each other.

These characteristics of OSP, i.e., sequence similarity to claudins and ability to induce TJ strand/groove-like structures in L fibroblasts, led us to regard OSP as a member of the claudin family, and we tentatively designated it as claudin-11 after consulting the human gene nomenclature committee (http://www.gene.ucl.ac.uk/nomenclature/). (In the updated human gene database, at least 15 members of the claudin family were found, all of which were already assigned to claudin-1 to -15.)

We next examined the expression of claudin-11/OSP in various tissues by Northern blotting. As shown in Fig. 3 A, claudin-11/OSP mRNA was detected as a 2.3-kb band in large amounts in the brain and testis and in only a trace amount in the kidney, suggesting that this type of claudin is involved in the TJ formation specifically in the brain and testis. Then, to examine the subcellular distribution of claudin-11/OSP, we generated polyclonal antibodies (pAbs) in rabbits using a synthesized peptide corresponding to COOH-terminal 20 amino acids of claudin-11/ OSP as an antigen. By immunoblotting, one affinity-purified pAb (pAb CL11-2) specifically recognized the GST fusion protein with COOH-terminal cytoplasmic domain of claudin-11/OSP, but not that of claudin-1 to -8, which were produced in E. coli (Fig. 3 B). Using this pAb, we first examined the subcellular distribution of claudin-11/OSP in the brain. When frozen sections of mouse brain were immunofluorescently stained with anti–claudin-11/OSP pAb, a large number of intensely stained linear structures were seen scattered in random directions in the cortex (Fig. 4 a). In deeper regions, these linear structures were occasionally arranged in a parallel manner to form thick bundles (Fig. 4 b). In the brain, TJs are known to be developed in vascular endothelial cells. Since these TJs were stained positively with anti-occludin antibody, we then stained frozen sections of brain doubly with anti–claudin-11/OSP pAb and anti-occludin mAb. However, as shown in Fig. 4, c and d, the claudin-11–positive linear structures did not overlap with occludin-positive endothelial TJs.

Then, frozen sections of the brain cortex were doubly stained with anti–claudin-11/OSP pAb and anti-neurofilament mAb, examined by confocal microscopy, and stereoscopic images were generated by computer (Fig. 5). At higher magnification, each claudin-11/OSP-positive linear structure was seen to run in a gentle spiral around a neurofilament-positive axon. These images suggested that these claudin-11/OSP signals were derived from myelin sheaths surrounding individual axons.

As reported previously (Dermietzel et al., 1974, 1980; Reale et al., 1975; Schnapp and Mugnaini, 1978; Tabira et al., 1978), conventional freeze-fracture electron microscopy of glutaraldehyde-fixed mouse brain revealed that so-called interlamellar strands of oligodendrocytes were arranged in a gentle spiral around axons (Fig. 6 a). Similarly to TJ strands in L transfectants expressing claudin-11/OSP (see Fig. 2), these interlamellar strands were associated with the P-face, and were mostly discontinuous with intervening spaces of various widths (Fig. 6 a, inset). When freeze-fracture replicas from unfixed brains or optic nerves were immunolabeled with anti–claudin-11/OSP pAb, these interlamellar strands were specifically labeled (Fig. 6, b and c). Furthermore, as shown in Fig. 7, when ultrathin cryo-sections of the brain were incubated with the same pAb, transverse sections of myelin sheaths were labeled radially, which may correspond to the so-called radial component of myelin (Peters, 1961, 1964; Dermietzel, 1974).

Finally, frozen sections of mouse testis were doubly stained with anti–claudin-11/OSP pAb and anti-occludin mAb. As shown in Fig. 8, a and b, both claudin-11/OSP and occludin were concentrated and precisely colocalized in a linear fashion at the most basal region of lateral membranes of adjacent Sertoli cells where TJs were reported to be well developed. Endothelial cells of microvessels were occludin-positive, but claudin-11/OSP-negative (Fig. 8, a and b). Immunofreeze-fracture electron microscopy revealed that claudin-11/OSP was exclusively localized on well-developed TJ strands of Sertoli cells (Fig. 8 c).

Interlamellar strands (radial components) of myelin sheaths are of great interest in terms of the physiological functions of oligodendrocytes (see Schnapp and Mugnaini, 1978). Although their appearance on freeze-fracture images is similar to those of epithelial and/or endothelial TJs, TJ-specific proteins, especially occludin, have not been shown to be localized at these interlamellar strands. On the other hand, we recently identified novel gene family members, claudin-1 to -8, which are major constituents of TJ strands, and found that single or several claudins are copolymerized to form TJ strands in various tissues (Furuse et al., 1998a,b; Morita et al., 1999). However, Northern blotting revealed that they were not expressed in large amounts in the brain as compared with non-neuronal tissues (Furuse et al., 1998a; Morita et al., 1999). In this study, we found that OSP (oligodendrocyte-specific protein) shows a sequence similarity to claudins, although the degree of identity was at most ∼30% (see Fig. 1). OSP was first identified as a protein specifically expressing in oligodendrocytes, but it has not been characterized in detail (Bronstein et al., 1996). We have demonstrated that OSP has the ability to form TJ strand-like structures in L fibroblasts, and that it is exclusively localized at interlamellar strands. Based on these observations, we concluded that interlamellar strands can be regarded as a variant of TJ strands composed of a new claudin, claudin-11/OSP.

As previously reported, when claudin-1 or -2 was singly expressed in L fibroblasts, TJ strands were induced, and through their ramification a huge network of strands was formed (Furuse et al., 1998b). In contrast, the claudin-11/ OSP-induced strands scarcely branched in L fibroblasts, and showed a tendency to run parallel to each other. Interestingly, interlamellar strands (and also TJ strands in Sertoli cells [Gilula et al., 1976]) run parallel without branching (see Figs. 6 and 8), while typical TJ strands in epithelial cells frequently ramify to form networks (Staehelin, 1973, 1974). These findings favored the notion that claudin-11/ OSP is a major constituent of interlamellar strands (and also TJ strands in Sertoli cells) and that the frequency of ramification of strands depends on the intrinsic nature of each claudin species.

It is intriguing that TJ strands in oligodendrocytes (interlamellar strands) share a specific claudin species, claudin-11/OSP, with those in Sertoli cells. In the testis, well-developed TJ strands in Sertoli cells (so-called Sertoli junctions) closely protect spermatogenic cells from the external environment, which is known as the blood-testis barrier (Dym and Fawcett, 1970; Russell and Peterson, 1985). Similarly to the brain, Northern blotting analysis indicated that the testis does not express claudin-1 to -8 in large amounts (Furuse et al., 1998a; Morita et al., 1999). Therefore, it is likely that in the testis claudin-11/OSP is a major component of TJ strands, which function as a tight barrier, and that in the brain interlamellar strands of oligodendrocytes also function as a barrier to closely isolate the extracellular compartment within myelin sheaths. In the central nervous system, neurons are also closely protected from the external environment, but this blood-brain barrier is known to be established by TJ strands in vascular endothelial cells in vertebrates (Rubin, 1991). However, in invertebrates such as insects, TJs of the perineurial sheath in the central nervous system (Lane et al., 1994) and/or pleated septate junctions (possible invertebrate equivalent of TJs) of glia cells were reported to be directly involved in the blood-nerve barrier (Lane et al., 1977; Juang and Carlson, 1992; Baumgartner et al., 1996). It is still premature to discuss this issue further, but it would be interesting to examine the phylogenetical relationship between claudin-11/ OSP and the other claudins in future studies.

Sertoli junctions were initially called “the junctional specializations of Sertoli cells” (Flickinger and Fawcett, 1967), and thought to be different from typical TJs in other epithelial cells (Gilula et al., 1976). These Sertoli junctions were characterized by (a) discrete bundles of filaments running parallel to the junction membranes, (b) cisternae of the endoplasmic reticulum located deeper to the layer of filaments, and (c) the parallelly arranged strands that did not anastomose extensively. However, recent findings on the localization of ZO-1 (Byers et al., 1991; Balda and Anderson, 1993), symplekin (Keon et al., 1996), and occludin (Moroi et al., 1998) in Sertoli junctions indicated that this specialization is a variant of TJs found in many other epithelia. This study, i.e., the occurrence of claudin-11/ OSP in Sertoli junctions, conclusively supported this notion.

Another issue that we should discuss here is the sequence differences in the cytoplasmic tail between claudin-11/OSP and the other claudins. All of claudin-1 to -8 end in -Y-V at their COOH termini (Furuse et al., 1998a; Morita et al., 1999), whereas only claudin-11/OSP ends in -H-V. In this sense, claudin-11/OSP is rather distantly related to the other claudins. The COOH-terminal -E-S/T-D-V motif in the Shaker K⁺ channel (-ETDV)/NMDA R2 subunit (-ESDV) and -E-Y-Y-V motif in neurexins were shown to be responsible for their binding to PDZ domains of Dlg/PSD-95 family proteins and LIN-2/CASK, respectively (Kim et al., 1995; Komau et al., 1995; Hata et al., 1996; Niethammer et al., 1996). Considering that three PDZ-containing MAGUK family proteins, ZO-1, ZO-2, and ZO-3, are exclusively concentrated at the cytoplasmic surface of TJ strands in epithelial and endothelial cells (Stevenson et al., 1986; Gumbiner et al., 1991; Balda et al., 1993; Itoh et al., 1993; Willott et al., 1993; Jesaitis and Goodenough, 1994; Haskins et al., 1998), it is expected that the COOH-terminal -Y-V in claudin-1 to -8 binds to some of the PDZ domains of the MAGUK family proteins to recruit them to TJ strands. So far, occludin has been regarded as a membrane binding partner for ZO-1 and ZO-2 (Furuse et al., 1994; Itoh et al., 1999), but it was shown that even in occludin-deficient TJ strands ZO-1 (and also ZO-2; unpublished data) was still recruited (Saitou et al., 1998). These findings are consistent with the above expectation, and led to the further speculation that the COOH-terminal -H-V of claudin-11/OSP shows distinct affinity to ZO-1, ZO-2, and/or ZO-3 from the other claudins. Actually, ZO-1 and ZO-2 were not concentrated at the interlamellar strands of oligodendrocytes (data not shown; see Saitou et al., 1997).

In this study, we found that although the amino acid sequence of claudin-11/OSP is fairly diversified from the eight previously identified claudins, it has the ability to induce TJ strands in L fibroblasts and constitutes the interlamellar strands in oligodendrocytes as well as TJ strands of Sertoli cells in the testis. Molecular manipulation of claudin-11/OSP including targeted gene disruption will more directly unravel not only the physiological functions of this unique claudin family member in the brain and testis, but also the etiologies of some myelin-related as well as spermatogenesis-related disorders.

We thank all the members of our laboratory (Department of Cell Biology, Faculty of Medicine, Kyoto University) for helpful discussions. Our thanks are also due to Ms. K. Furuse for her excellent technical assistance.

This study was supported in part by a Grant-in-Aid for Cancer Research and a Grant-in-Aid for Scientific Research (A) from the Ministry of Education, Science and Culture of Japan to Sh. Tsukita.

E-face

extracellular face

GST

glutathione-S-transferase

OSP

oligodendrocyte-specific protein

P-face

protoplasmic face

pAb

polyclonal antibody

TJ

tight junction