INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The plasma membrane of epithelial cells displays apical and basolateral domains with distinct composition and properties. The maintenance of the cell surface asymmetry requires junctional complexes such as, tight junctions (TJ), adherens junctions, and desmosomes. TJ act as a selective barrier restricting the diffusion of ions and solutes across the paracellular space (gate function). They also form a "fence" preventing lateral diffusion of plasma membrane proteins and lipids. Freeze-fracture electron microscopy (EM) revealed that the TJ was composed of linear rows that were tightly connected to similar rows on adjacent cell membranes. Much work has been devoted to understanding the molecular architecture of TJ. Occludin and claudins, two transmembrane proteins associated with TJ strands, are thought to seal the paracellular space and to generate a series of regulated channels within TJ membranes for the passage of ions and small molecules (Tsukita and Furuse, 1999). Exactly, how TJ assemble is still a matter of debate. Several peripheral membrane proteins such ZO-1, ZO-2, and ZO-3 are thought to link occludin and claudins to the underlying actin cytoskeleton. ZO proteins contain 3 PDZ and 1 SH3 domains that may recruit and cluster proteins to TJ (Cordenonsi et al., 1999; Itoh et al., 1999; Wittchen et al., 1999; Balda and Matter, 2000). However, it remains to be determined how these protein complexes may control both the dynamics of TJ assembly and the TJ barrier functions. Adherens junctions are positioned below TJ and consist of adhesion and signaling molecules. E-cadherin, a basolateral transmembrane protein, plays a key role in the formation and maintenance of junctional complexes (Gumbiner et al., 1988). However, expression of a dominant negative mutant of E-cadherin in MDCK cells did not affect tight junctions (Troxell et al., 2000). The mechanisms that regulate the assembly and maintenance of TJ and adherens junctions are not well understood.

Previous studies suggest that junctional complexes may constitute a specific targeting site on the lateral plasma membrane for delivery of specialized cargo vesicles (Louvard, 1980). This proposal is supported by the finding that many proteins involved in membrane trafficking are concentrated in cell-cell junctions (Zahraoui et al., 2000). In MDCK cells, antibodies against Sec8, an exocyst subunit, inhibit vesicles delivery to the basolateral membrane (Grindstaff et al., 1998). Three small GTPases, Rab3b, Rab8, and Rab13, are localized in cell-cell contacts in epithelial cells (Huber et al., 1993; Weber et al., 1994; Zahraoui et al. 1994). Rab proteins are involved in the regulation of different steps of exocytic and endocytic pathways (Schimmoller et al., 1998; Chavrier and Goud, 1999; Zerial and McBride, 2001). Although the exact role of Rab proteins is still unclear, they may control the assembly of protein complexes involved in various cellular processes including vesicle movement and fusion or actin organization (Stenmark et al., 1995; Kato et al., 1996; Peränen et al., 1996; Ren et al., 1996; Echard et al., 1998; Nielsen et al., 1999; White et al., 1999). Recently, we have found that Rab13 is recruited to cell-cell contacts at an early stage of assembly of junctional complexes (Sheth et al., 2000), suggesting that Rab13 may be a good candidate for regulating the assembly cell-cell junctions. In this study, we investigate the role of Rab13 in TJ functions. The expression of the constitutively active form of GFP-Rab13 (Q67L mutant) affected TER and increased the paracellular flux of small tracers. It also induced a disorganization of TJ strand networks and a delay in the localization of the TJ transmembrane protein, claudin1. In contrast, the inactive GFP-Rab13T22N mutant reduced paracellular permeability and had no effect on TJ strand organization and claudin1 localization. Our results provide a new insight into the role of Rab13 protein in regulating TJ assembly.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Antibodies

The mouse mAb anti-GFP was purchased from Roche (Roche Diagnostics GmbH, Mannheim, Germany). The monoclonal antigp58 and gp-114 antibodies were kindly provided by K. Simons (Max Planck Institute, Dresden, Germany). The rat monoclonal raised against ZO-1 (R40.76) was used (Anderson et al., 1989). The polyclonal rabbit antioccludin and anticlaudin-1 were purchased from Zymed (Zymed Laboratories, Inc., South San Francisco, CA). The affinity-purified goat anti-mouse, or anti-rat or anti-rabbit IgG conjugated to either TRITC or Cy5 were purchased from Jackson ImmunoResearch Laboratories, Inc.(West Grove, PA).

Mutagenesis and Transfection

Dominant negative (T22N) and the constitutively active (Q67L) Rab13 mutants were generated from Rab13 cDNA using site-directed mutagenesis kit (Stratagene, La Jolla, CA), and 5' TCGGGGGTGGGCAAGAATTGTCTGATCATTCGCTT-3' and 5'-GGGACACGGCTGGCCTAGAGCGGTTCAAGACAATA-3' oligonucleotides respectively. An EcoRI site was introduced by PCR upstream of the ATG codon of Rab13. EcoRI-BamHI inserts encoding Rab13WT, Rab13T22N, and Rab13Q67L were fused to the C terminus of the enhanced green fluorescent protein (GFP) and cloned into pGFP-C3 vector (Clontech Inc., Palo Alto, CA). All constructions were verified by sequencing. Stable MDCK cell lines expressing GFP-Rab13WT, GFP-Rab13T22N, GFP-Rab13Q67L, and the mock MDCK expressing GFP were generated by transfection using the effectene transfection reagent according to the manufacturer instructions (Qiagen GmbH, Hilden, Germany). Positive clones were selected and cloned in the same medium supplemented with 1 mg/ml G418 (Life Technologies, Eggenstein, Germany). The expression of GFP, GFP-Rab13 WT, T22N, and Q67L mutants was checked by immunobloting and by immunofluorescence using a monoclonal anti-GFP antibody. Two independent clones of GFP-Rab13T22N and GFP-Rab13Q67L were grown up and subsequently analyzed.

Cell Culture and Growth

MDCK cells (strain II) were grown in DMEM supplemented with 10% fetal calf serum, 2 mM glutamine, 100 U/ml penicillin, and 10 mg/ml streptomycin. The cultures were incubated at 37°C under a 10% CO₂ atmosphere.

In all our experiments, 600,000 cells/cm² were plated onto polycarbonate filters (0.4-µm pore size and 12-mm diameter; Transwell; Costar Corp., Cambridge, MA) as instant confluent monolayers and grown for 3 d. The expression of the ectopic proteins was induced by treatment with 10 mM sodium butyrate for 15 h before analysis. Parental MDCK and MDCK expressing GFP alone were also treated with 10 mM sodium butyrate. Stable transfected clones were maintained under selection in 500 µg/ml G418. To remove extracellular calcium, confluent MDCK cells were rinsed three times in low calcium medium (S-MEM) and incubated with S-MEM for 15 h. Cells were then rinsed with normal DMEM, incubated in DMEM for 0, 2, 4, and 15 h, and analyzed by immunofluorescence.

Immunofluorescence Microscopy

Parental or transfected MDCK cells were washed with PBS containing 1 mM CaCl₂, 0.5 mM MgCl₂ and fixed with 2% paraformaldehyde in PBS for 15 min. The free aldehyde groups were quenched for 15 min with 50 mM NH₄Cl in PBS. Cells were then permeabilized with 0.5% Triton X-100 in PBS for 15 min and then blocked in PBS buffer containing 0.5% Triton X-100 and 0.2% BSA. All subsequent incubations with antibodies and washes were performed with this buffer. Cells were incubated overnight at 4°C with the primary antibodies, rinsed three times for 10 min with the blocking buffer, and then incubated with affinity-purified secondary antibodies raised in goat and conjugated to TRITC or Cy5. After 45 min the filters were washed three times more with the same buffer and three times with PBS. The samples were mounted in 50% glycerol-PBS and analyzed with a Leica SP2 confocal laser scanning microscope (Leica Microscopy and Systems GmbH, Mannheim, Germany).

Measurement of Transepithelial Electrical Resistance

MDCK cells were plated on filters as instant confluent monolayers. TER of filters (12-mm diameter) were determined by applying an AC square wave current of 20 µA at 12.5 Hz and measuring the voltage deflection with a Ag/AgCl electrode using an Epithelial VoltOhmMeter (EVOM; World Precision Instruments, Sarasota, FL). TER values were determined by subtracting the contribution of filter and medium.

Paracellular Flux Assay

Paracellular permeability was measured using three different tracers: [³H]mannitol (182 Da), 4 kDa FITC-Dextran, and 40 kDa FITC-Dextran. Cells were grown on filters to confluency for 3 d, and the monolayers were treated overnight with sodium butyrate. The stock solution of FITC-Dextran (20 mg/ml; Sigma-Aldrich Chemie GmbH, Deisenhofen, Germany) was dialyzed against P buffer (10 mM HEPES, pH 7.4, 1 mM sodium pyruvate, 10 mM glucose, 3 mM CaCl₂, 145 mM NaCl) and diluted to 2 mg/ml in P buffer before the assay. We measured paracellular diffusion from the apical to the basolateral domain. The assay was started by replacing the basolateral medium with 500 µl of P buffer and the apical culture medium with 250 µl of solution containing 2 mg/ml of 4K FITC-Dextran or 40K FITC-Dextran. Monolayers were incubated at 37°C for 3 h, and the basal chamber media was collected. FITC-Dextran was measured with a fluorometer (excitation: 392 nm; emission: 520 nm; Perkin Elmer-Cetus Applied Biosystems, Inc., Überlingen, Germany). Paracellular flux of [³H]mannitol was measured as described (Balda et al., 1993). Briefly, we added to the upper chamber 500 µl of tissue culture medium containing 1 mM mannitol and 4 µCi/ml [³H]mannitol. After 1 h at 37°C, aliquots were removed, and radioactivity was counted in a liquid scintillation counter (Wallac Oy, Turku, Finland).

Freeze-fracture Electron Microscopy and Immunolabeling

MDCK cells were plated in 10-cm diameter tissue culture plates and grown to confluency for 3 d. For conventional freeze-fracture, cells were fixed in 2% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4, for 30 min at room temperature. Cells were scraped from the substrate with a plastic cell scraper and infiltrated with 30% glycerol for 2 h at 4°C. Freeze-fracture was performed at 130°C in a Balzers freeze-fracture 301 or 400 unit (Balzers, Liechtenstein; Fujimoto, 1997; Dunia et al., 2001). Replicas were examined using a Philips CM12 electron microscope (Eindhoven, The Netherlands).

Diffusion of BODIPYR6G-Sphingosylphosphorylcholin

BodipyR6G-sphingosylphosphorylcholin was prepared by reacting BodipyR6G SE (the succinimidyl ester of the 4,4-difluoro-5-phenyl-4-bora-3a,4a-diaza-s-indacene-3-propionic acid purchased from Molecular Probes Inc., Eugene, OR) with sphingosylphosphorylcholin (Sigma-Aldrich Chimie GmbH) as follows: a suspension of sphingosylphosphorylcholin (10.6 mg, 0.0228 mmol) in anhydrous dimethylformamide (5 ml) was placed, under inert atmosphere, in a 10-ml round-bottom flask fitted with a magnetic bar. BodipyR6G SE (10 mg, 0.0228 mmol) was added, in one portion, to the mixture, and the stirring was continued for 48 additional hours at room temperature. Removal of the solvent at 50°C under reduced pressure provided N-(4,4-difluoro-5-phenyl)-4-bora-3a,4a-diaza-s-indacene-3-propionyl)sphingosylphosphorylcholin (16.7 mg, 93% yield) as a dark powder: Mass Spectrum-FAB: m/z 787 (M + H)⁺, 809 (M + Na)⁺. This fluorescent lipid was thus obtained satisfactorily pure for subsequent utilization without further purification as judged by TLC on silica gel (eluent: dichloromethane/methanol/triethylamine: 50/49/1).

BODIPYR6G-sphingosylphosphorylcholin/BSA complexes were obtained by adding 400 µl of BODIPYR6G-sphingosylphosphorylcholin stock solution (1 mM in DMSO) to 10 ml of BSA solution (0.8 mg/ml defatted BSA in 10 mM HEPES, pH 7.4, 145 mM NaCl). To label the cells, monolayers were washed twice with cold P buffer, then 250 µl of fluorescent lipid/BSA was added to the apical chamber, and the cells were incubated for 10 min on ice. Cells were first washed four times with P buffer and then either left on ice for 1 h or directly mounted in P buffer. For mounting, double-sided tape on each side of the microscope slide was used to support the coverslip and to avoid pressure on the sample. The lateral diffusion of the fluorescent lipid was analyzed within the first 10 min by confocal microscopy.

Immunoblotting

MDCK cells expressing GFP, GFP-Rab13T22N, and GFP-Rab13Q67L were grown on 10-cm-diameter culture plates for 3 d and treated with 10 mM sodium butyrate for 15 h. They were extracted in 1 ml of 0.5% Triton X-100 (vol/vol), 10 mM Tris-HCl, pH 7.6, 120 mM NaCl, 25 mM KCl, 1.8 mM CaCl₂, 1 mM sodium vanadate, 50 mM NaF, 1 mM pefabloc, and a mixture of protease inhibitors (1 µg/ml pepstatin, 10 µg/ml leupeptin, and aprotinin) on a rocker platform for 20 min at 4°C. Solubilized materials (Triton X-100 soluble) were recovered by pelleting at 18,500 × g for 15 min at 4°C. Pellets (Triton X-100 insoluble) were dissolved in 250 µl of the same buffer containing 1% SDS at 100°C. Supernatants were adjusted to 1% SDS and protein concentration was determined using the Bio-Rad protein assay kit (Bio-Rad laboratories, Hercules, CA). Equal amounts of Triton X-100 soluble (S) and insoluble (P) pools were separated by SDS-PAGE and transferred electrophoretically to nitrocellulose filters. The upper and lower parts of the filters were probed with antioccludin and anticlaudin1 antibodies respectively. After blotting, soluble and insoluble occludin and claudin1 protein bands were quantified using the Scion image program (NIH image, Rockville, MD) and Microsoft Excel software (Redmond, WA). The insoluble (P)/soluble (S) ratio was calculated (mean + SEM).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Expression of GFP-Rab13 in MDCK Cells

Rab proteins cycle between GDP-bound inactive and GTP-bound active states. To investigate the role of Rab13 in TJ function, we generated two mutants, Rab13T22N and Rab13Q67L. In Ras and other Rab proteins, the equivalent of Q67L mutant is deficient in GTP hydrolysis, so it cannot undergo catalytic cycling to GDP-bound form and remains bound to its effectors on target membranes; therefore, this mutant is considered as an activated form of the GTPase. The Rab13T22N mutant is expected to be inactive. The corresponding mutation in Ras and Rab has a lower affinity for GTP than for GDP. It also appears to sequester the exchange factor thereby inhibiting the function of the endogenous protein. Rab13 wild-type (WT), Rab13T22N, and Rab13Q67L mutants fused to the C terminus of the enhanced green fluorescent protein (GFP) were expressed in stably transfected MDCK cells. Unfortunately, our anti-Rab13 antibodies did not detect the endogenous protein, we could not compare the level of expression, nor the localization of the exogenic GFP-Rab13 to that of the endogenous protein. Immunoblot analysis using anti-GFP antibody revealed that the expression of the ectopic GFP-Rab13 proteins was very weak, in particular for the GFP-Rab13T22N protein (Figure 1A). This prompted us to treat the cells with 10 mM sodium butyrate for 15 h to stimulate the expression of the transfected cDNAs. Under these conditions, the sodium butyrate treatment did not cause changes in TJ structure or functions (Balda et al., 1996; this work). Compared with nontreated cells, the sodium butyrate treatment led to a 1.5-fold increase in the amount of GFP, GFP-Rab13 WT, and GFP-Rab13Q67L proteins and a 6-fold increase in the expression of GFP-Rab13T22N protein (Figure 1A). Except for the TER experiments, the expression of the ectopic proteins was induced for 15 h with 10 mM sodium butyrate before analysis. Parental MDCK and MDCK expressing GFP alone were also treated with 10 mM sodium butyrate. We checked that this treatment had no effect on the targeting of GFP-Rab13.

View larger version (112K): [in this window] [in a new window]   Figure 1.   Expression of GFP-Rab13 proteins in stably transfected MDCK cells in presence (+) or absence () of butyric acid. (A) Proteins in equal amounts of cell extracts prepared from cells expressing GFP (lanes 1 and 5), GFP-Rab13T22N (lanes 2 and 6), GFP-Rab13Q67L (lanes 3 and 7), and GFP-Rab13WT (lanes 4 and 8) were separated by SDS-PAGE and transferred to nitrocellulose filters. The filter was probed with monoclonal anti-GFP antibody. (B) Immunofluorescence localization of GFP-Rab13 in stably transfected MDCK cells. Cells expressing GFP, GFP-Rab13T22N, or GFP-Rab13Q67L were grown for 1 d to form small islets of differentiating MDCK cells. After sodium butyrate treatment, cells were processed for immunofluorescence. Arrowheads indicate the periphery of cells devoid of junctional complexes. Cells were viewed with a Leica microscope and photographed. GFP-Rab13 WT and Q67L proteins are found in the perinuclear region and in regions of cell that are in contact with neighboring cells (arrows). Note the absence of GFP-Rab13 WT and Q67L staining from the edges of cells devoid of cell-cell junctions (arrowheads). (C) Cells expressing GFP, GFP-Rab13wt, GFP-Rab13T22N, or GFP-Rab13Q67L were grown on Transwell filters for 3 d. After sodium butyrate treatment and fixation, cells were processed for confocal laser scanning microscopy. Shown are one (x-y) focal plane taken at the tight junction level and a z-section. GFP-Rab13 wild-type and GFP-Rab13Q67L constructs were targeted to the lateral membrane including junction's area. Bar, 10 µm

In isolated islets of differentiating MDCK cells, GFP-Rab13 WT and Q67L were found in the perinuclear region and in cell-cell contacts where junctional complexes developed with neighboring cells. GFP-Rab13 was absent from the edges of cells devoid of junctional complexes. The GFP-Rab13T22N mutant was observed on cytoplasmic structures distributed throughout the cells. GFP alone was found in the nucleus and as a diffuse staining in the cytoplasm (Figure 1B). These results indicated that the recruitment of GFP-Rab13 to the plasma membrane required cell-cell contact formation. In confluent cells, GFP-Rab13WT and Rab13Q67L mutant were distributed along the lateral plasma membrane including tight junctions as well as on cytoplasmic structures. No labeling either due to binding to apical or basal plasma membranes could be detected. GFP-Rab13T22N was observed on cytoplasmic structures distributed throughout the cells (Figure 1C). To ascertain whether the presence of GFP tag may have disturbed the localization of Rab13, we transiently expressed a carboxy terminus VSV-G-tagged Rab13 T22N and Q67L mutants in MDCK cells. Similarly to GFP-Rab13 constructs, we found that VSV-Rab13Q67L localized to the plasma membrane and VSV-Rab13T22N to cytoplasmic structures (supplementary material). Moreover, we also examined the distribution of an other GFP-tagged protein Rab8, which shares high amino acid identity with Rab13 (Zahraoui et al., 1994) and is also associated with tight junctions in MDCK cells (Huber et al., 1993). GFP-Rab8Q67L localized to perinuclear region and to the plasma membrane (supplementary material) was consistent with the localization of endogenous Rab8 in MDCK cells (Huber et al., 1993). These data suggest that the GFP tag did not affect the distribution of Rab proteins in MDCK cells.

Expression of GFP-Rab13Q67L Delays the Formation of Electrically Tight Epithelial Monolayers

Zonula occludens tightly seal the apex of adjacent cells to form an epithelial sheet. The transepithelial electrical resistance (TER) can be used to monitor the tightness of the seal between neighboring cells. Because the recruitment of Rab13 to the apicolateral junctional complex occurred at an early stage in the assembly of TJ (Zahraoui et al., 1994; Sheth et al., 2000), we postulated that it might be involved in TJ sealing. Therefore, we measured TER over a 120-h time period of parental MDCK, GFP, GFP-Rab13WT, GFP-Rab13T22N, and GFP-Rab13Q67L cells grown on filters (Figure 2). Because the same number of cells was plated per filter, the TER from different stable cell lines could be compared. Parental MDCK and cells expressing GFP showed a rapid increase in TER that reached 300  × cm² by 24 h. In cells expressing GFP-Rab13T22N, TER reached 400  × cm² by 24 h. In contrast, cells expressing the active GFP-Rab13Q67L exhibited a significant delayed TER development. TER developed slowly and reached a maximal value of 200  × cm² only by 48 h. Similar alteration of TER recovery was seen in GFP-Rab13WT cells; although the effect was less pronounced than in GFP-Rab13Q67L cells, TER value reached 200  × cm² by 24 h. This effect was expected because the overexpression of many RabWT proteins has been shown to have an effect that is similar to GTP-bound forms of Rab proteins (Bucci et al., 1992; Peränen et al., 1996; White et al., 1999). After the peak, all cell lines recovered a similar steady state TER of 100-150  × cm² by 72-96 h. The TER of three independent Q67L clones was measured. These clones exhibited similar TER development. This excluded the possibility that the TER lag found in MDCK expressing GFP-Rab13Q67L mutant was due to clonal variation. These results indicate that the TER delay observed in GFP-Rab13Q67L-expressing cells was specific and that the expression of the constitutively active GFP-Rab13Q67L mutant delayed the formation of electrically tight epithelial monolayers. They also indicate that the effect of GFP-Rab13 occurred during TJ assembly.

View larger version (32K): [in this window] [in a new window]   Figure 2.   TER development in MDCK cells and MDCK expressing GFP, GFP-Rab13WT, GFP-Rab13T22N, or GFP-Rab13Q67L. Cells were plated on filters as instant confluent monolayers (0 min). TER was measured at different times >120 h, and background resistance of empty filters was subtracted. Average values are given ± SEM of each cell clone as derived from four independent determinations and performed in duplicate. In these experiments, cells were not treated with sodium butyrate. At the end of each experiments, the integrity of the monolayer was checked by ZO-1 staining. The results of two independent clones of GFP-Rab13T22N and GFP-Rab13Q67L cells are shown. We measured the TER of a third independent GFP-Rab13Q67L clone, which exhibited similar TER development than the two presented here.

Active Rab13 Disrupts TJ Gate Function for Nonionic Molecules

TJ also act as a selective diffusion barrier for nonionic molecules. We examined whether Rab13 mutants could affect the diffusion of three nonionic tracers through the paracellular space: [³H] mannitol (182Da), 4 kDa FITC-Dextran, and 40 kDa FITC-Dextran (Figure 3). In confluent MDCK and in cells expressing GFP, a small amount of these tracers diffused across the monolayer of cells. When cells expressing GFP-Rab13Q67L were analyzed, they showed a higher amount of paracellular diffusion of the three tracers. Compared with control cells, the expression of GFP-Rab13Q67L induced an 8-fold increase in the paracellular permeability of [³H]mannitol and a 10- to 12-fold increase in the diffusion of 4- and 40-kDa FITC-Dextran (Figure 3). In monolayers expressing the inactive GFP-Rab13T22N, the diffusion of the three tracers was reduced compared with that of control cells. We conclude that the expression of the GFP-Rab13Q67L mutant induced the leakage of small nonionic tracers from the apical domain.

View larger version (17K): [in this window] [in a new window]   Figure 3.   Paracellular flux of nonionic tracers in MDCK and in GFP-Rab13-transfected cells. Parental MDCK cells and cells expressing GFP, GFP-Rab13WT, GFP-Rab13T22N, or GFP-Rab13Q67L were plated as instant confluent monolayers and grown for 3 d. Cells were treated with sodium butyrate for 15 h. Paracellular flux of [3H]mannitol, 4 kDa FITC-Dextran (FD4), and 40 kDa FITC-Dextran (FD40) was measured in the apical to basolateral direction. The amount of tracer diffusion is normalized to parental MDCK cells. The data of three independent experiments performed in duplicate are reported as the mean ± SEM

Involvement of Rab13 in Tight Junction Fence Function

In addition to its gate function, TJ provide a physical fence for free diffusion of lipids and proteins in the outer leaflet of the plasma membrane between apical and basolateral domains. To examine whether the expression of active Rab13 affected the fence function of TJ, BODIPY-sphingosylphosphorylcholin/BSA was applied for 10 min at 4°C to the apical surface of butyrate-treated cell monolayers grown on filters. After washing, the diffusion of BODIPY-sphingosylphosphorylcholin/BSA was either immediately analyzed by confocal microscopy or analyzed after 60 min of incubation. The confocal x-z sections (Figure 4) show the distribution of membrane-labeled BODIPY-sphingosylphosphorylcholin/BSA in control cells and in Rab13 mutants. In cells expressing GFP or GFP-Rab13T22N mutant and in MDCK cells, the fluorescent lipid was confined to the apical domain at both 0 and 60 min. No lateral staining was detected. However, in MDCK expressing GFP-Rab13Q67L, BODIPY-sphingosylphosphorylcholin/BSA was not restricted after 60 min to the apical membrane but diffused across the TJ and stained the lateral membrane. When these cells were analyzed immediately after labeling with the lipid, BODIPY-sphingosylphosphorylcholin/BSA labeled only the apical membrane, indicating that BODIPY-sphingosylphosphorylcholin/BSA leakage after 60 min could not have resulted from an increase in the paracellular pathway.

View larger version (83K): [in this window] [in a new window]   Figure 4.   Diffusion of BODIPY-sphingosylphosphorylcholin/BSA complex from the apical to the lateral membrane in MDCK cells and in cells expressing GFP, GFP-Rab13T22N, and GFP-Rab13Q67L. Cells were grown on Transwell filters for 3 d and treated with sodium butyrate for 15 h. Fluorescent lipid/BSA was loaded to the apical surface for 10 min at 0°C, washed four times in P buffer (see MATERIAL AND METHODS), and either immediately processed for confocal microscopy (0 min) or left for an additional 60 min on ice (60 min). The lateral diffusion of the fluorescent lipid was then determined by z-sectioning on a confocal microscope. Bar, 10 µm.

We next investigated whether the expression of Rab13 mutants interfered with the polarized distribution of gp114 and gp58, two membrane proteins that are targeted and restricted to the apical and basolateral domains, respectively. Figure 5 shows the x-z confocal sections of gp114 and gp58. In control cells (MDCK and MDCK-expressing GFP) as well as in cells expressing GFP-Rab13T22N mutant, gp114 and gp58 proteins were correctly localized. In GFP-Rab13Q67L cells, gp114 and gp58 staining was not specifically restricted to apical and basolateral membrane domains, respectively, although the cell morphology was altered. Both gp114 and gp58 staining was still detected in the cytoplasm and also on the lateral membrane for gp114. Taken together, the results indicate that the expression of the activated form of GFP-Rab13 altered the distribution of apical and basolateral membrane proteins.

View larger version (54K): [in this window] [in a new window]   Figure 5.   Distribution of the apical (A) gp114 and the basolateral (B) gp58 membrane proteins. Parental MDCK cells and cells expressing GFP, GFP-Rab13T22N, or GFP-Rab13Q67L were grown on Transwell filters for 3 d. Monolayers were fixed, permeabilized, and stained either with mouse monoclonal anti-gp114 or anti-gp58 antibodies. The primary antibodies were visualized with goat anti-mouse Cy5. Cells were then analyzed by confocal microscopy. The z-sections show the localization of gp58 and gp114 in transfected and untransfected MDCK cells. Bar, 10 µm.

Active Rab13 Induces Dramatic Changes in TJ Strand Organization

Given the effect of the activated form of Rab13 on TJ gate and fence function, we next asked whether this Rab13 mutant could affect the TJ structure. We examined by freeze-fracture EM the TJ strand organization, which provides a measure of TJ integrity. Freeze-fracture analysis of the surface of the native MDCK cell line showed that the apical cell border is characterized by finger-like projections (microvilli) and close contact between the juxtaposed cells. Fracture faces of the apical plasma membranes displayed the characteristic organization of TJ:linear strands interconnected to form a junctional complex of anastomosing ridges on the protoplasmic face (PF) and linear depressions or grooves on the exoplasmic face (EF). Two to three parallel strands formed by linear rows encircle, without interruption, the apical cell periphery (Figure 6A). A similar type of TJ linear strands was detected by freeze-fracture in MDCK cells expressing GFP-Rab13T22N (Figure 6B). Conversely, the MDCK cells expressing GFP-Rab13Q67L were characterized by increased tortuosity and overlapping of both lateral and apical cell borders with rather large interspaces between the adjoining plasma membranes. In many areas of the freeze-fractured plasma membranes the TJ strands were randomly assembled and appeared as pleomorphic clusters of anastomosing filaments. Surprisingly, the tight junctional structures running along the apical cell periphery did not form a continuous belt but rather a discontinuous series of stranded clusters (Figure 6C). Furthermore, in some areas, TJ strand clusters appeared as a loose meshwork that was not restricted to the apical region of the lateral membrane but extended along the lateral domain of cells expressing GFP-Rab13Q67L (Figure 6D).

View larger version (158K): [in this window] [in a new window]   Figure 6.   Freeze-fracture replica images of TJ of (A) MDCK, (B) MDCK expressing GFP-Rab13T22N and (C and D) GFP-Rab13Q67L cells grown for 3 d. The apical cell border is characterized by finger-like projections. (A and B) TJ linear and continuous strands occupy the apical zone of the lateral membrane. PF, protoplasmic fracture face; EF, exoplasmic fracture face. Bar, 100 nm. (C) The TJ domain comprises clusters of furrows on EF (arrows), interconnected by one single particulate strand running along the intermembrane space (arrowhead). (D) Note that individual strands are fragmented, and a subunit pattern is visible (arrowheads). The arrow points to a cluster of particulate entities segregated from the strands. Bar, 150 nm.

Rab13Q67L Mutant Affects Claudin1 Localization

We next examined whether the alteration of TJ strand organization induced by Rab13Q67L mutant correlated with changes in the localization of occludin and claudin1, two TJ transmembrane proteins, and ZO-1, a peripheral TJ protein. For this purpose, MDCK clones were plated at high identical density, grown on filters for 3 d, and then processed for confocal microscopy. The confocal x-y sections showed prominent sharp ring-like structures of occludin and ZO-1 on the lateral membrane between neighboring cells. The distribution of these proteins was very similar in parental as well as in MDCK expressing GFP, GFP-Rab13T22N (Figure 7). We found that occludin is also associated with the lateral membrane and is not restricted to TJ. Overexpression of GFP-Rab13Q67L affected slightly the localization of ZO-1 but did not apparently affect that of occludin. In control cells (MDCK, MDCK-GFP), claudin1 hexagonal staining characteristic of junctional complexes was detected. Some claudin1-positive vesicular structures were observed. Claudin1 distribution in GFP-Rab13T22N-expressing cells appeared similar to those observed in control cells. In GFP-Rab13Q67L-expressing cells, claudin1 was weakly detected at cell-cell junctions on the lateral plasma membrane; most of claudin1 immunoreactivity was observed in the cytoplasm (Figure 7), suggesting that the activated GFP-Rab13Q67L altered the distribution of claudin1.

View larger version (160K): [in this window] [in a new window]   Figure 7.   Confocal immunofluorescence localization of ZO-1, occludin, and claudin1 in MDCK cells, MDCK expressing GFP, GFP-Rab13T22N, and GFPRab13Q67L. Cells were grown on filters for 3 d and processed for immunofluorescence with either anti-occludin, anti-ZO-1, and anti-claudin1 antibodies as described in MATERIALS AND METHODS. Images of one focal section across the tight junction and X-Z views showed ZO-1, occludin, and claudin1 immunolabeling. Bar, 10 µm.

Junctional complexes are dynamic structures that are continuously assembled/disassembled in vivo. To analyze whether GFP-Rab13Q67L interferes with the recruitment to the lateral membrane of claudin1, confluent MDCK clones were incubated in low calcium medium for 15 h to dissociate cell-cell junctions. Subsequent restoration of physiological level of calcium results in the synchronous de novo assembly of adherens and tight junctions. The localization of claudin1, ZO-1, and E-cadherin was analyzed by immunofluorescence at 0, 2, 4, and 15 h after induction of cell-cell contacts. In control cells as well as cells expressing GFP-Rab13 T22N and Q67L mutants, removal of calcium from the medium resulted in the dissociation of cell-cell junctions and redistribution of E-cadherin, ZO-1, and claudin1 (Figure 8). Membrane dissociation of these proteins was reversible. Within 2 h after readdition of calcium to the culture medium, the intracellular E-cadherin staining decreased, and the plasma membrane labeling increased at cell-cell contacts. Although the kinetics of membrane localization of this protein appeared similar in control cells and in MDCK expressing GFP-Rab13 T22N and Q67L mutants (Figure 8A), we cannot rule out that GFP-Rab13Q67L might be involved in the targeting of E-cadherin. Identical pattern of recovery of claudin1 and ZO-1 was observed in MDCK expressing GFP and GFP-Rab13T22N (Figure 8, B and C). Expression of GFP-Rab13Q67L slightly delayed the recruitment of ZO-1 to the cell-cell junctions. Interestingly, plasma membrane recruitment of claudin1 in MDCK expressing GFP-Rab13Q67L was delayed. Within 4 h after readdition of calcium, most of the claudin1 staining was not detected at the plasma membrane and was still observed in the cytoplasm. Fifteen hours after calcium add-back, claudin1 was detected at the lateral membrane (Figure 8C). The recruitment of GFP-Rab13Q67L protein to the plasma membrane occurred earlier (2 h) during cell-cell contact reassembly (supplementary material). These data confirm that the expression of GFP-Rab13Q67L delayed claudin1 recruitment and to a lesser extent that of ZO-1 to the lateral membrane during junctional complex assembly and did not efficiently impair the localization to the plasma membrane of E-cadherin.

View larger version (124K): [in this window] [in a new window]   Figure 8.   Expression of GFP6Rab13Q67L delays the recruitment of claudin1 to the plasma membrane during TJ assembly after calcium switch. Confluent MDCK expressing GFP, GFP-Rab13T22N, and GFP-Rab13Q67L were incubated in low calcium medium for 15 h and then reincubated in normal medium for 0, 2, and 4 h and stained with (A) anti-E-cadherin, (B) anti-ZO-1, and (C) anti-claudin1 antibodies. Within 2 h after readdition of calcium to the culture medium, E-cadherin accumulated in the lateral membrane in control cells and in MDCK expressing GFP-Rab13 T22N and Q67L mutants. A similar pattern of plasma membrane recruitment of ZO-1 and claudin1 was observed in MDCK expressing GFP and GFP-Rab13T22N. Plasma membrane recruitment of claudin1 in MDCK expressing GFP-Rab13Q67L was delayed but recruitment of ZO-1 is only slightly affected. Bar, 10 µm.

Detergent Solubility of Tight Junction Proteins

In epithelial cells, TJ membrane proteins partitioned in two distinct pools: a Triton X-100-insoluble fraction that could represent a specific pool associated with TJ membrane microdomain (Nusrat et al., 2000) and a TX-100 soluble pool that may represent the non-TJ-associated pool. Because these two pools are likely functionally distinct, we analyzed whether there were any differences in the Triton X-100 extractability of the occludin and claudin1 from cells expressing GFP or GFP-Rab13 mutants. Figure 9 shows the distribution of TJ proteins occludin and claudin1 in Triton X-100-soluble and -insoluble fractions. The Triton X-100 insolubility of occludin in cells expressing GFP, GFP-Rab13T22N, and GFP-Rab13Q67L was similar. We did not detect any change in the Triton X-100 extractability of claudin1 in MDCK expressing GFP or GFP-Rab13T22N. However, expression of GFP-Rab13Q67L coincided with a decrease (40%) in claudin1 TX-100-insoluble pool. As shown in Figure 9B, the insoluble/soluble ratio of claudin1 in GFP-Rab13Q67L cells was lower than those in GFP or GFP-Rab13T22N cells. This suggests that the protein was poorly associated with TJ membranes.

View larger version (28K): [in this window] [in a new window]   Figure 9.   Triton X-100 extractability of TJ proteins occludin and claudin1 in MDCK cells expressing GFP, GFP-Rab13T22N, and GFP-ab13Q67L. Cells were grown for 3 d, extracted in buffer containing 0.5% Triton X-100 for 20 min at 0°C, and centrifuged. Proteins in equal amounts of Triton X-100-soluble (S) and -insoluble (P) fractions were separated by SDS-PAGE, and transferred to nitrocellulose filters. The upper and lower parts of the filters were probed with anti-occludin and anti-claudin1 antibodies, respectively. Results shown are representative of three experiments. (B) Quantification of three independent experiments performed as in A. After blotting, soluble and insoluble occludin and claudin1 protein bands were quantified using the Scion image program (NIH image, Rockville, MD) and Microsoft Excel software. The insoluble (P)/soluble (S) ratio was calculated (mean ± SEM). Note that the claudin1 P/S ratio from MDCK expressing GFP-Rab13Q67L was lower than those from the other cells.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Our results demonstrate that expression of GFP-Rab13Q67L but not the inactive GFP-Rab13T22N mutant alters the distribution of the tight junction membrane protein, claudin1

As a consequence, TJ formed in MDCK cells expressing GFP-Rab13GTP are structurally disorganized and functionally leaky for small tracers despite being an effective barrier to ion flow

To investigate the role of Rab13 in epithelial TJ organization, we generated two Rab13 mutants, T22N and Q67L. Rab13 mutants were expressed in MDCK cells as GFP-Rab13 fusion proteins. Our finding that GFP-Rab13 WT and Q67L are localized on the lateral membrane including TJ in MDCK cells is different from data on Caco-2 cells reported earlier by us. Immunofluorescence labeling of polarized intestinal Caco-2 cells showed that Rab13 colocalized with ZO-1 (Zahraoui et al., 1994). Unfortunately, our attempts (including this work) to localize Rab13 by EM have failed. The precise localization of Rab13 might be different in polarized MDCK and Caco-2 cells. We note that in MDCK cells, as in Caco-2 cells, the formation of cell-cell contact is required for the recruitment of Rab13 to the plasma membrane. It is possible that the antipeptide we used earlier (Zahraoui et al., 1994) recognized only the TJ associated pool of Rab13. Interestingly, depending on the antibodies used, claudins are not restricted to TJ but can be found on the lateral membrane (Rahner et al., 2001). Recently, Yeaman et al. (2001) used a panel of monoclonal antibodies to Sec6 and Sec8 and found that some antibodies stained Sec6/s8 at cell-cell contact sites and others antibodies stained Sec6/s8 in a perinuclear compartment. The implication of this lateral distribution of GFP-Rab13 in MDCK cells is unclear. Perhaps, Rab13 has an additional function independent of TJ in MDCK cells. The presence of Rab13 in epithelial and fibroblast cells strengthens the notion that Rab13 may have additional role, independent of TJ, in cell surface polarization (Zahraoui et al., 1994). Alternatively, the lateral distribution of GFP-Rab13 might represent a pool that can be recruited into TJ to acutely alter the localization of claudin1 and disturb the paracellular diffusion properties of TJ in response to physiological signals.

TJ Gate and Fence Functions Are Impaired by the Activated GFP-Rab13

Measurement of TER and tracer flux across the TJ was used to monitor the formation of the TJ barrier during TJ assembly. Although GFP-Rab13 WT and T22N are weakly expressed, they affected TER recovery. We suggest that a weak expression of the two proteins is likely sufficient to affect TER development in these cells. We found that neither inactive Rab13T22N, nor active Rab13Q67L mutants could significantly affect steady state TER. However, when TER is measured during course of TJ formation, it indicated that expression of GFP-Rab13Q67L caused a significant delay in TER development. The TER lag induced by active GFP-Rab13Q67L and the fact that its expression did not affect TER steady state strongly suggested that Rab13 acts specifically during the assembly of TJ. Analyses of tracer (mannitol, 4 and 40 kDa dextran) movement through TJ revealed that expression of inactive GFP-Rab13T22N reduced the diffusion of the small tracers. In contrast, expression of GFP-Rab13Q67L resulted in a significant increase in paracellular flux. These data indicated that constitutively active Rab13 reduced the tightness of the seal between neighboring cells. Claude (1978) suggested the existence of series of compartmentalized aqueous channels that fluctuate between closed and open states within the TJ strand network. According to this model, flux measurements of tracers reflect the sum of fluxes through TJ pores over a period of time, whereas TER is an instantaneous indicator of permeability. Consequently, the relationship between TER and paracellular flux could be nonlinear (Balda et al., 1996, Madara, 1998; McCarthy et al., 2000). This could explain the functional dissociation of paracellular permeability from TER in cells expressing GFP-Rab13 mutants. In addition to disrupting TJ gate function, we found that the expression of GFP-Rab13Q67L altered the localization of the apical, gp114, and the basolateral, gp58, membrane proteins as well as the lateral diffusion of a lipid, BODIPY-sphingosylphosphorylcholin/BSA. In contrast, the expression of the inactive mutant (Rab13T22N) did not alter the localization of gp58 and gp114 proteins or the diffusion of fluorescent lipid. Thus, the structural organization of TJ responsible for the intramembrane fence and the paracellular permeability seal appear to be altered in GFP-Rab13Q67L cells. It is possible that the partial mislocalization of gp114 and gp58 in GFP-Rab13Q67L cells might not be due to the disruption of TJ fence per se but to a mechanism involving GFP-Rab13GTP in the maintenance of apico-basolateral polarity in epithelial cells.

One of the most striking observations made in this study is the formation of aberrant strands along the lateral surface of cells expressing GFP-Rab13Q67L but not GFP-Rab13T22N. The TJ of cells expressing the GFP-Rab13Q67L mutant appeared as a fragmented and clustered assembly of anastomotic filaments. Our data suggest that TJ strand organization in GFP-Rab13Q67L cells is sufficiently disrupted to permit lipids to diffuse through the TJ barrier and to alter the organization of apical and basolateral membrane proteins. Although we observed a reduction in tracer diffusion through the TJ, we did not observe any change in TJ strand organization in GFP-Rab13T22N cells, suggesting that an efficient functional paracellular barrier may not require several intramembrane strands.

Rab13 Is Required for TJ Assembly

TJ strands appear to be formed by homo- or heteropolymers of claudins and occludin, which pair laterally with those in apposing membranes to seal the paracellular space between adjacent cells. These integral membrane proteins are linked to the underlying actin cytoskeleton via cytosolic proteins such as ZO-1 (Tsukita and Furuse, 1999, 2000). At present it is not clear how TJ protein organization might affect TJ barrier functions. Although occludin is a key component of TJ strands (Furuse et al., 1993; Balda et al., 1996; Wong and Gumbiner, 1997), recent work suggests that it is not sufficient to generate TJ strands. For example, visceral endoderm cells in which the occludin gene is knocked out still develop a normal TJ strand network (Saitou et al., 1998). Several reports indicate that claudins are required for the generation of TJ strands and for the establishment of the paracellular diffusion barrier (Furuse et al., 1998a; Tsukita and Furuse, 2000). Expression of claudin1 and 2 in fibroblasts lacking TJ induces the formation of strands that are morphologically similar to the epithelial TJ strands (Furuse et al., 1998b). Analysis of OSP/claudin11-deficient mice reveals the absence of TJ strands in myelin sheets of oligodendrocytes and sertoli cells and paracellin1/claudin16 mutations cause renal Mg²⁺ wasting (Gow et al., 1999; Simon et al., 1999). Our data indicate that expression of GFP-Rab13Q67L dramatically altered the distribution of claudin1 but not significantly that of occludin. In these cells, ZO-1 exhibited only a slight distribution change and still formed a continuous junctional ring, suggesting that its localization is not determined only by Rab13. We suggest that alteration of TJ barrier functions in cell expressing GFP-Rab13Q67L is due to the abnormal organization of claudin1, and to a lesser extent to that of ZO-1, in the TJ area. Moreover, our data revealed that the claudin1 Triton X-100-insoluble pool that might represent the TJ-associated fraction was dramatically reduced in MDCK expressing GFP-Rab13Q67L. Although the mechanisms involved are not known, we speculate that alteration of the dynamics of claudin1 association to/dissociation from TJ membrane by GFP-Rab13Q67L may perturb the structural and functional properties of the intramembrane strands.

Interestingly, calcium switch experiments indicated that the expression of GFP-Rab13Q67L mutant delayed the recruitment of claudin1 to the plasma membrane, strengthening the notion that Rab13 may play an important role in coordinating the recruitment of claudin1/ZO-1 to specific domains on the lateral membrane. Unfortunately, we did not detect any interaction between Rab13 and claudin1/ZO-1, suggesting that the effect of Rab13 on both proteins is likely indirect. Probably because of the lack of classical ectodomains, we failed to biotinylate claudin1 and were therefore unable to show whether GFP-Rab13Q67L inhibited the transport of claudin1 to the plasma membrane. It has been suggested that E-cadherin mediates the translocation of ZO-1 that recruits other TJ proteins required for functional assembly of TJ (Rajasekaran et al., 1996; Troxell et al., 2000). The effect of GFP-Rab13Q67L on TJ does not appear to be mediated by E-cadherin, because overexpression of active GFP-Rab13 does not significantly impair the recruitment of E-cadherin to the lateral membrane. Rab proteins act in conjunction with a variety of protein effectors to regulate different steps of exocytic and endocytic pathways (Schimmoller et al., 1998; Chavrier and Goud, 1999; Zerial and McBride, 2001). Accordingly, it is tempting to speculate that Rab13, in its GTP-bound and membrane bound form, recruits an effector (s) that inhibits the recruitment of claudin1 to the cell surface. Conversely, Rab13T22N (inactive form) unable to recruit the effector to the lateral membrane may favor the establishment of TJ gate and fence functions. Alternatively, it is also possible that in cells expressing the GTPase-deficient mutant, GFP-Rab13Q67L, claudin1 is transported to the cell surface but is reinternalized because it is not incorporated into TJ. The opposite effects of the GFP-Rab13 Q67L and T22N mutants on TJ suggest that Rab13GTP is required before claudin1 recruitment, whereas GTP hydrolysis by Rab13 occurs after claudin1 recruitment, inactivating Rab13 and allowing claudin1 incorporation into TJ. So, our data indicate that Rab13 may play an important role in the assembly of TJ and thus in the establishment of polarity in epithelial cells.