Involvement of a Heterotrimeric G Protein α Subunit in Tight Junction Biogenesis

The tight junction (TJ) of polarized epithelial cells is critical for maintaining an impermeant barrier and epithelial cell polarity. The signaling events important for TJ assembly require regulated calcium stores and protein kinase C (PKC), but the earliest signaling events in the cascade have not been well defined. We now show that Gαi2 in Madin Darby canine kidney (MDCK) cells localizes to a region overlapping with the TJ. To further analyze the localization of Gα subunits in epithelial cells, rat Gαo, Q205Lαo (Gαo “activated” by point mutation) and plasmid without insert (PC) were transfected into MDCK cells and localized by immunofluorescence and confocal microscopy. Similar to endogenous Gαi2, Gαo-MDCK cells localize Gαo, (84% similar to Gαi2) in the subapical region overlapping with ZO-1 (zona occludens-I), a key component of the TJ. PC-MDCK cells have no detectable Gαo. In Gαo-MDCK cells, a physical association of Gαo with components of the TJ was detectable by immunoprecipitation of ZO-1. Immunoprecipitates of ZO-1 from Gαo-MDCK cells consistently coprecipitated Gαo. Constitutively active Q205LGαo localized to the subapical lateral membrane similar to wild-type Gαo. To determine if constitutively activated Gα subunits can affect TJ biogenesis, the formation of tight junctions in PC, Gαo, and Q205Lαo-MDCK cells was followed by measurement of transepithelial resistance (TER) during the Ca2+ switch, a model widely used to study mechanisms of junctional assembly. Baseline and post Ca2+ switch TER values did not differ among the cell lines. However, constitutively activated Q205Lαo-MDCK cells developed TER significantly faster than PC and Gαo cells in the early phase (0-4 h) (54 ± 4 versus 23 ± 3 (PC); 12 ± 1 (Gαo) Ω·m2/h) and late phase (4-h peak) (117 ± 10 versus 45 ± 5 (PC); 66 ± 7 (Gαo) Ω·cm2/h) after Ca2+ switch. Peak TER values were significantly higher in Q205Lαo-MDCK cells (1168 ± 107 versus 437 ± 37 (PC); 548 ± 54 (Gαo) Ω·cm2). These results indicate that Gαo and Q205Lαo expressed in MDCK cells are localized near the junctional complex, associate with at least one TJ protein, and that activated Gαo accelerates TJ biogenesis without significantly affecting the maintenance of the TJ. Together, these results suggest an important role for heterotrimeric G proteins in TJ assembly.

The tight junction (TJ) 1 serves not only a barrier function in epithelial tissue, but may also play an essential role in establishing and maintaining lipid and protein polarity (reviewed in Refs. 1 and 2). In developing tissues as well as cultured cell models, mechanisms involved in regulating the bioassembly of tight junctions appear significantly different from those necessary for their maintenance. MDCK cells in culture provide an excellent and widely utilized model for the assembly of tight junctions. MDCK cells maintained in low calcium media (LC media) lack cell-cell contact, intercellular junctions, and apicalbasolateral polarization of lipids and protein (3, 4 -7). Upon "switching" to normal calcium media (NC media), the cells rapidly develop characteristics of a polarized, tight, transporting epithelium. In many respects, the MDCK cell calcium switch model recapitulates in vitro key events in epithelial morphogenesis (8,9). The presence of a single cell type undergoing intercellular junction formation in a synchronous manner allows the process to be followed by immunocytochemical and biochemical, as well as physiological means.
We have previously shown that, as MDCK cells form junctions and polarize, there are impressive changes in intracellular calcium concentration (3,5). Most interestingly, at sites of cell-cell contact, where tight junctional assembly occurs, we observed localized increases in intracellular calcium (5). When these increases were buffered with cell-permeant calcium chelators or when endoplasmic reticulum calcium stores were depleted with thapsigargin (3,7), the assembly of the TJ was perturbed, as determined by immunocytochemistry and transepithelial electrical resistance (TER, a functional measure of TJ integrity). Furthermore, the association of TJ proteins (e.g. ZO-1) with the cytoskeletal fraction was disrupted, as determined by their solubility profile in detergent (Triton X-100) extracts. These results indicate that internal calcium stores, which are known to substantially overlap with the IP 3 -sensitive stores in the endoplasmic reticulum, are critical for TJ formation during polarized epithelial biogenesis. The results also suggest that TJ molecules are sorted in a calcium-dependent fashion to the lateral surface of the plasma membrane where they associate with the actin cytoskeleton.
Because regulated calcium stores are critical for TJ formation and because these stores are released in parallel with activation of protein kinase C (PKC), the role of this kinase in TJ formation has been studied. When PKC was inactivated by independent mechanisms (calphostin C or low concentrations of H7), both the sorting and assembly of TJ proteins was perturbed, as determined by immunocytochemistry and TER (10,11). Consistent with this, diacylglycerol analogs accelerate TJ assembly (12 cytosol to the membrane fraction, and at least one PKC isoform was translocated directly to the vicinity of the TJ (11). During TJ assembly, several key TJ proteins, including ZO-1 and ZO-2, appear to be phosphorylated in a PKC-dependent fashion (11).
A key question remains to be answered regarding the early signaling events that are critical for TJ biogenesis. If regulated calcium stores and PKC are necessary for TJ assembly, it is likely that a heterotrimeric G protein(s) plays an essential role early in the process. Reagents that modulate G protein function such as cholera toxin, pertussis toxin, and AlF 3 have been shown to affect TJ biogenesis during the Ca 2ϩ switch in MDCK cells (4) and G␣ i2 has been localized in the vicinity of the tight junction overlapping with ZO-1 in LLC-PK 1 cells (13). In the course of studying the localization of G␣ subunits in MDCK cells, we have expressed G␣ o and activated G␣ o (Q205L) in MDCK cells. The highly conserved Gln-205 residue in G␣ o is important during the transition state of GTP hydrolysis. The Gln 3 Leu mutation significantly slows GTP hydrolysis and results in a constitutively "activated" G␣ subunit (14,15). We now show that transfection of G␣ o and Q205L␣ o into MDCK cells leads to their localization near the junctional complex, association with at least one TJ protein (ZO-1), and that expression of activated G␣ o (Q205L) accelerates TJ biogenesis.

EXPERIMENTAL PROCEDURES
Plasmid Construction and Cell Culture-Rat G␣ o cDNA was cloned into Bluescript (Stratagene) as described previously (16) and recloned into the EcoRI/ApaI sites of pcDNA3 (Invitrogen). pMN-Q205L␣ o was provided by Dr. R. Iyengar, and the cDNA was amplified by PCR using primers to the 5Ј-end and to the opposite strand 3Ј-end. The resulting cDNA was cloned into pcDNA3 and sequenced by T7-double-stranded DNA sequencing (U. S. Biochemical Corp). MDCK cells were maintained in Dulbecco's modified Eagle's Medium, supplemented with antibiotics plus 5% fetal calf serum. Transfected cell lines were maintained in G418 (500 g/ml; Life Technologies, Inc.).
Transfections-Subconfluent MDCK cells (ATCC, Rockville MD) were transfected with 10 g of linearized (PvuI), G␣ o , Q205L␣ o , or pcDNA3 by the calcium phosphate precipitation method. G418-resistant colonies were analyzed for expression by Western blot analysis using R4 polyclonal antibody provided by Dr. E. Neer as described previously (17). Two clones expressing Q205L␣ o and four G␣ o clones were further analyzed. There were no significant differences between clones.
Immunolocalization-Untransfected, PC-, G␣ o -, and Q205L␣ o -MDCK cells were plated at confluence on coverslips or Transwell filters (12 mm) (Costar), washed with PBS, and fixed with methanol (100%, Ϫ70°C) for 10 min and stained as previously described (11). Samples were incubated with one or two of several antibodies; rabbit polyclonal G␣ o (R4) at 1:100 dilution, rat monoclonal to ZO-1 (undiluted; hybridoma courtesy of D. Goodenough), mouse monoclonal to E-cadherin (courtesy of B. Gumbiner) (undiluted), mouse monoclonal to gp135 (apical marker; courtesy of G. Ajakian). The G␣ i2 antibody (AS7; Du-Pont NEN) was used at 1:50 dilution. One or two of the following secondary antibodies were used: goat anti-rabbit Texas Red or fluorescein, goat anti-rat Texas Red or fluorescein, or goat anti-mouse Texas Red or fluorescein (all from Jackson ImmunoResearch, West Grove, Pa). Coverslips or filters were visualized on a Nikon Labphot-2 microscope equipped with epifluorescence or a Bio-Rad MRC600 confocal microscope equipped with 2 filter blocks for dual label image acquisition.
Immunoprecipitation-Confluent G␣ o -and PC-MDCK cells were scraped in cold PBS, resuspended in 1 ml of buffer A (50 mM Tris, pH 7.6, 1% Triton X-100, 0.5% sodium deoxycholate, 0.2% SDS, 100 mM NaCl, 2 mM EDTA, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and soy and lima bean trypsin inhibitors), frozen and thawed three times, and triturated through a 27-gauge needle. Samples were centrifuged at 10,000 rpm for 10 min, and the supernatant was precleared with 25 l of goat anti-rat Sepharose beads (Cappel). The supernatant was incubated with ZO-1 hybridoma supernatant (200 l) or an equivalent amount of rat serum or control rat hybridoma and 50 l of goat anti-rat Sepharose beads overnight at 4°C with rocking. These conditions quantitatively immunoprecipitate ZO-1 and several other proteins of the tight junction (11). Beads were washed and samples were analyzed by SDS-PAGE and Western with the R4 G␣ o antibody as described previously (17). The nitrocellulose was de-veloped using the Supersignal TM CL-HRP substrate system (Pierce) with horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch).
Ca 2ϩ Switch and TER-MDCK cells were plated at ϳ3 ϫ 10 5 cells on 12-mm polycarbonate filters (Transwell; Costar) and allowed to establish confluence over 24 h in medium without antibiotics. Medium was changed to LC medium (minimal essential media with 5% dialyzed fetal calf serum, without Ca ϩ ; previously shown to be 1-4 M (3)) for 16 -24 h. At the start of each experiment, cells were switched to NC medium, containing 1.8 mM Ca 2ϩ , TER was measured using a Millipore ERS electrical resistance system as described (3), and the results were expressed in ⍀⅐cm 2 .

RESULTS AND DISCUSSION
Epithelial cells are known to constitutively express a variety of G␣ subunits, including G␣ i2 , G␣ i3 , G␣ s , and G␣ q . In rat enterocytes, G␣ i2 and G␣ i3 are localized in the basolateral membrane while G␣ s appears evenly distributed in both apical and basolateral membranes (18). Along the renal tubule, G␣ subunits can be found in either the apical, basolateral, or both domains depending upon the cell type (19,20). In cultured renal epithelial cells such as LLC-PK 1 cells (porcine renal cell line with proximal tubule characteristics), G␣ i2 overlaps with the tight junction (13), and G␣ i3 is found predominantly in the Golgi (21,22). Fig. 1 shows that G␣ i2 in wild-type MDCK cells has a distribution similar to that reported for LLC-PK 1 cells. G␣ i2 partially overlaps with the tight junction protein, ZO-1 when identical fields are visualized by standard immunofluorescent microscopy. Fig. 1A shows linear staining of both G␣ i2 and ZO-1 that is consistent with basolateral localization. In addition, there is detectable intracellular staining of G␣ i2 that FIG. 1. Immunolocalization of G␣ i2 in MDCK cells. A, wild-type MDCK cells were grown to confluence on coverslips and double-stained with rabbit G␣ i2 antibody and ZO-1 hybridoma as described under "Experimental Procedures." The coverslips were visualized on a Nikon Labphot-2 microscope, and the identical field was photographed in the same focal plane with Tmax 3200 film (Kodak) for 3-6 s using the FITC filter for ZO-1 and Texas Red filter for G␣ i2 . Magnification ϭ ϫ 1030. B, coverslips of wild-type MDCK cells were prepared as described in A, and confocal images of two cells (marked 1 and 2) were obtained on Bio-Rad MRC 600 confocal microscope with dual filters that permits simultaneous detection of fluorescein and Texas Red. The same cells (1 and 2) labeled with G␣ i2 and ZO-1 antibodies are displayed in the XZ orientation with the apical (Ap) membrane marked with a white line. Digital images were labeled and merged with images from A using Adobe Photoshop Mountain View, CA. Magnification ϭ ϫ 1050.

Role of G Proteins in Tight Junction Biogenesis 25751
is more prominent than is seen for ZO-1. To refine the localization of G␣ i2 and ZO-1 in MDCK cells, confocal images in the XZ orientation were obtained. Fig. 1B shows two cells doublelabeled with G␣ i2 and ZO-1 antibodies. The distribution of G␣ i2 and ZO-1 along the lateral membrane is similar but not identical. There is overlap of the two proteins although it appears that G␣ i2 has a slightly broader distribution along the lateral membrane and some intracellular staining. In our studies aimed at identifying elements involved in the targeting of G proteins to specific locales in epithelial cells, we began with a G protein not normally expressed in epithelial cells; G␣ o the major G protein in neuronal tissues. We used MDCK cells, which have been widely employed to study epithelial junctions and polarity. When G␣ o was transfected into MDCK cells, the protein was found to be highly expressed and localized predominantly laterally by standard immunofluorescence (not shown). More detailed examination of the cellular localization using confocal microscopy revealed that the G␣ o subunit had been efficiently targeted to the lateral-subapical region of the transfected cells in the vicinity of the apical junctional complex (Fig. 2A). Subsequent double-labeling studies and confocal microscopy with antibodies against the adherens junctional protein, E-cadherin, and the TJ protein, ZO-1, revealed that the transfected G␣ o subunit localized to a region between the TJ and adherens junctions, at least partially overlapping both ( Fig. 2A). Thus, the distribution of G␣ o in G␣ o -MDCK cells is remarkably similar to the distribution of G␣ i2 in wild-type MDCK cells (Fig. 1). Fig. 2B, left panel, shows that there is no detectable staining of PC-MDCK cells with G␣ o antibody.
Given the close proximity of G␣ o to the tight junction, we sought to determine whether the transfected G␣ o subunit associated with proteins of the TJ. A number of laboratories have previously shown that antiserum against ZO-1 coimmunoprecipitates at least three TJ proteins: ZO-1, ZO-2, and p130 (11,12). When transfected MDCK cells were immunoprecipitated with antibodies against ZO-1, G␣ o could be consistently detected in G␣ o -MDCK cells but not in PC-MDCK cells (Fig. 3). The conditions for immunoprecipitation have previously been determined to quantitatively immunoprecipitate almost all of the ZO-1 (11). This result indicates that a fraction of G␣ o in transfected cells associates into a complex containing at least one TJ protein and raises the possibility of a functional interaction between G␣ o and the TJ. In order to further examine this possibility, we transfected a constitutively activated G␣ o mutant (Q205L) into MDCK cells. This point mutation is constitutively "activated" and maintained in a receptor-activated conformation.
Standard and confocal microscopy revealed that, like wildtype G␣ o , Q205L␣ o is also targeted to the lateral membrane in the vicinity of tight junction. Fig. 2B, right panel, demonstrates linear lateral surface staining of Q205L␣ o when imaged by confocal microscopy. To address the role of constitutively activated G␣ o in the TJ, transepithelial electrical resistance (TER), a reliable measure of TJ integrity, was monitored both in confluent monolayers and also during the MDCK cell "Ca 2ϩ switch." The MDCK cell Ca 2ϩ switch is a model widely employed to study the bioassembly of the TJ and has been particularly useful in elucidating signaling events (3, 4, 5, 10 -12). The time course for TER development at various times are shown in Fig. 4A, and the rates of TER development for the early phase (0 -4 h) and later phase (4 h-peak) are shown in Fig. 4B. In confluent monolayers, there was no significant difference between the cell lines in baseline TER values ( (548 Ϯ 54 ⍀⅐cm 2 ) and PC-MDCK (437 Ϯ 37) cells (Fig. 4A). These data suggest an important role for signaling initiated by a heterotrimeric G protein in TJ assembly and distinct from a role in the maintenance of the formed TJ. Consistent with this notion, TER measurements 28 h after cell contact was established were similar to those in confluent monolayers with no significant difference between transfected cells and mocktransfected controls (Fig. 4A). Thus, the localization of an activated G␣ subunit into the TJ accelerates TJ biogenesis. Together, these results suggest a critical role for a heterotrimeric G protein in TJ bioassembly, presumably through activation of calcium signaling and PKC, although a direct link remains to be established. Of course, a limitation of these experiments is their heterologous nature since the key G protein in TJ assembly in MDCK cells must be something other than G␣ o . This could be G␣ i2 , which is found in the vicinity of the TJ or another unidentified G protein associated with the TJ.
The association of at least a fraction of G␣ o with ZO-1 is interesting not just because it suggests an intimate physical association of a G protein with a critical component of the TJ, but also because of similarities with neuronal cells. G␣ o is concentrated in the distal tip of neuronal processes and is regulated by GAP-43, a protein essential for accurate pathway development of growing neurites (23,24). Expression of consti-tutively activated G␣ o increases the number of neurites per cell possibly by affecting protein kinase C and intracellular calcium release (25,26). ZO-1 is a member of the membrane-associated guanylate kinase (MAGUK) family of proteins, and, apart from ZO-2, its greatest reported homology is with an essential protein of the postsynaptic density, PSD-95 (27). G␣ o is also enriched in the post-synaptic density (28), although it is unknown if G␣ o associates with PSD-95. G␣ o also regulates N-type calcium channels in the heart (29) as well as several other signaling pathways (reviewed in Ref. 30). Thus, the effect of G␣ o in the Ca 2ϩ switch of MDCK cells could conceivably result from modulation of intracellular and/or extracellular calcium. This intriguing parallel raises the possibility that these proteins (ZO-1, ZO-2, and PSD-95), and perhaps other MAGUK family members, may possess structural motifs that confer an ability to associate with G proteins. ZO-1 and PSD-95 share 55% amino acid sequence similarity, and each of these proteins is about 40% similar to GAP-43. However, there are no obvious sequence motifs in the primary sequences of these proteins that suggest a site for interaction with G␣ subunits. Important questions in TJ biogenesis remain regarding events upstream and downstream to the activation of G proteins and precisely which G protein(s) are involved. Nevertheless, the results of this study provide evidence for an important role of G␣ subunits in the early events of tight junction biogenesis.