Involvement of Gαi2 in the Maintenance and Biogenesis of Epithelial Cell Tight Junctions*

Polarized epithelial cells have highly developed tight junctions (TJ) to maintain an impermeant barrier and segregate plasma membrane functions, but the mechanisms that promote TJ formation and maintain its integrity are only partially defined. Treatment of confluent monolayers of Madin-Darby canine kidney (MDCK) cells with AlF4 − (activator of heterotrimeric G protein α subunits) results in a 3–4-fold increase in transepithelial resistances (TER), a reliable indicator of TJ integrity. MOCK cells transfected with activated Gα0 (Q205L) have acclerated TJ formation (Denker, B. M., Saha, C., Khawaja, S., and Nigam, S. J. (1996) J. Biol. Chem. 271, 25750–25753). Gαi2 has been localized within the tight junction, and a role for Gαi2in the formation and/or maintenance of the tight junction was studied by transfection of MDCK cells with vector without insert (PC), wild type Gαi2, or a GTPase-deficient mutant (constitutively activated), Q205Lαi2. Tryptic conformational analysis confirmed expression of a constitutively active Gαi2 in Q205Lαi2-MDCK cells, and confocal microscopy showed a similar pattern of Gαi2 localization in the three cell lines. Q205Lαi2-MDCK cells had significantly higher base-line TER values than wild type Gαi2- or PC-MDCK cells (1187 ± 150 versus 576 ± 89 (Gαi2); 377 ± 52 Ω·cm2 (PC)), and both Gαi2- and Q205Lαi2-transfected cell lines more rapidly develop TER in the Ca2+ switch, a model widely used to study the mechanisms of junctional assembly. Treatment of cells with AlF4 − during the Ca2+ switch had little effect on the kinetics of TER development in Gαi2- or Q205Lαi2-MDCK cells, but PC cells reached half-maximal TER significantly sooner in the presence of AlF4 − (similar times to Gαi2-transfected cells). Base-line TER values obtained after the switch were significantly higher for all three cell lines in the presence of AlF4 −. These findings indicate that Gαi2 is important for both the maintenance and development of the TJ, although additional Gα subunits are likely to play a role.

Polarized epithelia have developed highly specialized membrane functions enabling vectorial transport across the cellular layer. The junctional complex of epithelial cells includes gap junctions, adherens junctions, and tight junctions. The tight junction (TJ) 1 is the most apical component of the junctional complex and provides two essential functions: (i) the permeability barrier to paracellular fluxes and (ii) the "fence" separating the apical and basolateral membrane domains. In developing tissues as well as cell culture models, the critical signaling events important to junction formation appear to be quite different from mechanisms that maintain junctional integrity. The TJ is composed of a complex of proteins that includes occludin, the only transmembrane protein identified so far (1). There are several peripherally attached membrane proteins found in the TJ including the zona occludens family (ZO-1, -2, and -3) (2)(3)(4). ZO proteins are members of the MAGUK (membrane associated guanylate kinase) superfamily that are often found at sites of cell-cell contact and may function to couple extracellular signaling pathways with the cytoskeleton. Other proteins found in or near the TJ include cingulin, 7H6, symplekin, unidentified phosphoproteins, and a series of signal transduction molecules (reviewed in Ref. 5).
MDCK cells are a cultured epithelial cell line that has been extensively utilized for studies of epithelial polarity, targeting of proteins, and the study of intercellular junctions (6). The Ca 2ϩ switch model of TJ formation in MDCK cells has been widely utilized to gain insights into the function of polarized epithelial cells (7)(8)(9)(10)(11) and recapitulates many of the critical molecular events of epithelial morphogenesis. MDCK cells cultured in low calcium (M) lack cell-cell contact, polarity, and junctions. "Switching" to normal calcium medium (NC) triggers a series of molecular events that leads to establishment of the polarized phenotype with characteristics of a tight transporting epithelium. Tight junction development can be followed by measuring the transepithelial resistance (TER), a rapid and reproducible assessment of tight junction integrity. Because MDCK cells are clonal and TJ development can be synchronized in the Ca 2ϩ switch, the role of specific proteins on TJ biogenesis can be studied in this system by cDNA transfections.
The critical role of calcium in the formation of intercellular junctions is well established. Extracellular calcium is required for homotypic interactions of E-cadherin and is likely to be the initial event of junctional complex formation (12). Regulated intracellular calcium stores are also important for tight junction biogenesis. There are local increases in intracellular calcium concentration at the points of cell-cell contact (9), and chelation of intracellular calcium perturbs TER development (13). Thapsigargin depletes intracellular endoplasmic reticulum stores of calcium, and thapsigargin treatment of MDCK cells prior to initiation of cell-cell contact prevents TER development and the sorting of ZO-1 to the TJ (7). The signaling events important for TJ biogenesis are complex and utilize a variety of pathways. Phosphorylation events are important as several proteins become phosphorylated in the TJ, and protein kinase C (PKC) isoforms translocate to the TJ during biogenesis. PKC inhibitors markedly inhibit the development of TER in the calcium switch, and PKC agonists stimulate ZO-1 translocation to the membrane. The importance of PKC in tight junction biogenesis, as well as regulated calcium stores, suggests important roles for heterotrimeric G proteins. The proximity of several G proteins to the TJ also suggests they may have potential roles in regulating the development and/or maintenance of the TJ. PKC and PKC␣, have also been localized in the vicinity of the TJ (14 -17). We previously demonstrated that expressing a constitutively activated G␣ o (Q205L) in MDCK cells significantly accelerated TJ biogenesis without affecting base-line resistances. Although G␣ o is a member of the G protein family inhibited by pertussis toxin (ϳ80% similar to G␣ i1-3 ), its receptors and effectors are distinct, and furthermore, G␣ o is not detected in renal epithelia or MDCK cells (16,18,19). Several G␣ i family members are expressed in epithelial cells, and G␣ i2 has been shown to overlap with the tight junction in epithelial cell lines (16,17). Taken together, these observations raise the possibility that G␣ i2 may be an important regulator of tight junctions. To test this hypothesis, we initially looked for effects of AlF 4 Ϫ (activator of G␣ subunits) on tight junctions in control cell lines and then established MDCK cell lines overexpressing wild type G␣ i2 and a constitutively activated G␣ i2 (GTPase-deficient, Q205L␣ i2 ). We find that AlF 4 Ϫ significantly increases TER in control cells and accelerates TER development during the Ca 2ϩ switch. The effects of AlF 4 Ϫ can be reproduced in MDCK cells expressing activated G␣ i2 , indicating that this G␣ subunit is critical to the development and maintenance of tight junctions.

EXPERIMENTAL PROCEDURES
Plasmid Construction and Cell Culture-Rat G␣ i2 cDNA was cloned into Bluescript (Stratagene) as described previously (20) and recloned into the EcoRI and ApaI sites of pcDNA3 (Invitrogen). Q205L␣ i2 was provided by Dr. Gary Johnson and cloned into Bluescript using HindIII sites and then into pcDNA3 using XhoI and XbaI sites. 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.) Transfection-Subconfluent MDCK cells (ATCC, Manassas, VA) were transfected with 10 g of linearized plasmid by calcium phosphate precipitation method as described previously (16). G418-resistant colonies were analyzed for increased G␣ i2 expression by Western blot using a rabbit polyclonal antibody directed toward the C terminus of G␣ i2 (AS7; NEN Life Science Products). Control cells were obtained by transfecting pcDNA3 without insert, and all cell lines were established in parallel.
Tryptic Analysis of Transfected Clones-Confluent PC-, G␣ i2 -, or Q205L␣ i2 -MDCK cells were washed twice with PBS and then scraped into buffer A (50 mM Tris-HCl, pH 7.5, 6 mM MgCl 2 , 75 mM sucrose, 1 mM dithiothreitol, 1 mM EDTA). Cells were frozen and thawed three times and triturated ten times through a 27 gauge needle. All samples were incubated at 30°C with no added nucleotide or 100 M GTP␥S. Samples were immediately placed on ice, and trypsin was added (20 pmol of L-1-(tosylamido)-2-phenylethyl chloromethyl ketone-treated trypsin (Sigma). All samples were incubated at 30°C for 20 min, and digestion was terminated by the addition of SDS-polyacrylamide gel electrophoresis sample buffer followed by boiling for 5 min. Samples were then analyzed by SDS-polyacrylamide gel electrophoresis and Western blot using AS7 anti-G␣ i2 rabbit polyclonal antibody (1:1,000) and ECL (Pierce) with goat anti-rabbit horseradish peroxidase (1:10,000).
Immunohistochemistry-PC-, G␣ i2 -, or Q205L␣ i2 -MDCK cells were grown on coverslips or Transwell filters (12 mm) (Costar), rinsed with PBS, and fixed with methanol (100%, Ϫ70°C) for 10 min. Cells were then washed with PBS and blocked as described previously (16). Samples were incubated with rabbit polyclonal G␣ i2 (AS7, from NEN Life Science Products) at several dilutions and rat monoclonal to ZO-1 (undiluted supernatant; courtesy of D. Goodenough) for 1 h. Cells were washed with PBS three times at 5-min intervals and incubated with secondary antibodies (fluorescein-or Texas Red-conjugated goat antirabbit or anti-rat IgG; Jackson Immuno Research, West Grove, PA) at 1:100 with for 1 h. Coverslips were visualized on a Nikon Labphot-2 immunofluorescence microscope or a Bio-Rad 1024 confocal microscope using the 63ϫ oil immersion objective. Images were processed in Adobe Photoshop (Adobe, CA) and figure compiled in Adobe Illustrator (Adobe, CA).
Ca 2ϩ Switch and Measurement of TER-MDCK cells were plated on 12-mm transwell filter (Costar) at confluence (ϳ3 ϫ 10 5 cells) and allowed to attach for 24 -36 h to form a tight monolayer in normal Ca 2ϩ containing medium (NC). Cells were placed in low Ca 2ϩ (1-4 M) medium (low calcium) for 1 h followed by switch to NC medium. TER was measured using a Millipore (Bedford, MA) electrical resistance system, and the results are expressed in ⍀⅐cm 2 . TER was measured in stable monolayers and during Ca 2ϩ switch in the presence and absence of aluminum fluoride (AlF 4 Ϫ ; 3 mm NaF ϩ 50 M AlCl 3 , Sigma).

RESULTS AND DISCUSSION
Several lines of evidence have suggested the involvement of heterotrimeric G proteins in tight junction formation. Early studies with G protein modulators such as pertussis toxin, cholera toxin, AlF 4 Ϫ , and a variety of other agents showed variable effects on TJ biogenesis (10). Several confocal studies have localized G␣ i2 , G␣ i3 , and G␣ 12 in the vicinity of the tight junction (16,17,19,21). Recently, we demonstrated that G␣ o (a member of the G␣ family inhibited by pertussis toxin) expressed in MDCK cells localizes to the subapical lateral membrane overlapping with ZO-1 in the tight junction (16), and this was subsequently confirmed in another study (19). A constitutively active mutant of G␣ o (Q205L) also localizes in this region, and cells expressing G␣ o showed no differences in baseline junctional properties as determined by transepithelial resistance. However, in the Ca 2ϩ switch, MDCK cells expressing activated G␣ o (Q205L␣ o -MDCK) developed tight junctions at twice the rate and reached significantly higher peak TER values than either G␣ o -MDCK or PC-MDCK cells. Although G␣ o is not normally expressed in epithelia, this observation raises the possibility that one of the G␣ subunits normally found in this location could have a fundamental role in regulating the development and/or maintenance of the TJ.
To further examine the role of G protein ␣ subunits affecting the TJ, we studied the effects of the G protein activator AlF 4 Ϫ on TJ formation in wild type (not transfected; WT-MDCK) and vector (pcDNA3) transfected MDCK cells (PC-MDCK). AlF 4 Ϫ has no known effects on small GTP binding proteins but activates heterotrimeric G␣ subunits. Crystal structures of G␣ i1 obtained with GDP complexed with AlF 4 Ϫ reveal that the position of the ␥-phosphate is occupied by AlF 4 Ϫ . AlF 4 Ϫ in this position prevents catalysis by immobilizing Gln 204 and Arg 178 (22). Fig. 1 shows that wild type MDCK cells and PC transfected cells cultured on filters develop significantly higher TER in the presence of AlF 4 Ϫ . Untreated steady state TER values were similar between the cell lines, and AlF 4 Ϫ reproducibly increased TER values 3-4-fold. This finding is consistent with activation of one or more endogenous G␣ subunits that results in enhanced steady state resistances.
Because AlF 4 Ϫ activates all G␣ subunits in MDCK cells and several studies have placed G␣ i2 in close proximity to the TJ, we tested the hypothesis that G␣ i2 was important to this process by stably expressing G␣ i2 and a constitutively activated G␣ i2 (Q205L␣ i2 ) in MDCK cells. The amount of transfected G␣ i2 was determined relative to the levels of endogenous G␣ i2 in PC-MDCK cells. Western blots of the three cell lines using identical amounts of total protein were analyzed (not shown) using NIH image (Wayne Rasband, NIH). Relative to PC-MDCK cells, the level of G␣ i2 in G␣ i2 -MDCK cells was 3.9 Ϯ 0.4-fold (n ϭ 7) increased, and for Q205L␣ i2 -MDCK the level was of G␣ i2 was 1.8 Ϯ 0.2-fold (n ϭ 7) above PC-MDCK cells. To confirm that constitutively activated G␣ i2 was expressed in these transfected MDCK cells, we utilized a tryptic cleavage analysis of G␣ i2 (Fig. 2). This technique has been widely utilized as an indicator of G␣ subunit conformation (23,24) and is based on the observation that G␣ subunits have a different cleavage pattern depending on whether they are folded into an active or inactive conformation. In the active conformation (GTP-liganded), there is only a single tryptic site accessible near the N terminus (approximately Arg 21 ) resulting in a slightly truncated protein (39 kDa instead of 41 kDa for G␣ i2 ). In the inactive (GDP-liganded) conformation, an additional site becomes accessible in the ␣2 helix or switch region (near Arg 209 ) resulting in peptides of approximately 25 and 17 kDa. Fig. 2 demonstrates the tryptic cleavage patterns of cell homogenates from each of the transfected cell lines (PC-, G␣ i2 -, and Q205L␣ i2 -MDCK cells). In PC-and G␣ i2 -transfected cells, untreated G␣ i2 migrates at 41 kDa (first lane of each set) and is stabilized in the active conformation (39 kDa) if preincubated with the nonhydrolyzable GTP analogue, GTP␥S (last lane of each set). However, in the absence of added nucleotide (middle lane of each set) the G␣ subunits should be GDP bound from endogenous GTP/GDP in the cell, and G␣ i2 is cleaved into 25and 17-kDa fragments with no detectable 39-kDa peptide. The 25-kDa fragment (derived from the N terminus) is not detectable with the AS7 antibody, and the 17-kDa fragment is more labile (25) and consequently is not well visualized on these blots (not shown). Tryptic digestion of Q205L␣ i2 -transfected cells shows that in the absence of added nucleotide (middle lane), there is a fraction of G␣ i2 that is tryptic-resistant, indicating persistence of G␣ i2 in an "active" conformation. The amount of G␣ i2 in the active conformation of Q205L␣ i2 -MDCK cells is a small fraction of the starting material. This is due, in part, to the observation that G␣ subunits with mutations that result in an active conformation are more sensitive to proteolysis than wild type G␣ subunits activated with GTP␥S (20). In addition, it is necessary to do these studies on whole cell lysates in the absence of protease inhibitors. The Western blots were deliberately overexposed to look for bands migrating at 39 kDa. This analysis confirms expression of constitutively activated G␣ i2 in Q205L␣ i2 -MDCK cells.
We have previously demonstrated that WT-MDCK cells express some G␣ i2 in the subapical lateral membrane overlapping with ZO-1 (16). To eliminate the possibility that the transfection process affects G␣ i2 localization, PC-MDCK cells were characterized by confocal microscopy. Fig. 3A shows a confocal image of PC-MDCK cells costained with antibodies to ZO-1 and G␣ i2 (used at 1:25 dilution). The confocal images reveal that G␣ i2 partially colocalizes with ZO-1 at the level of the tight junction. There is significant intracellular staining that was also seen in WT-MDCK cells (16). To determine whether trans- fected G␣ i2 subunits were localized in a similar manner to the endogenous G␣ i2 , the G␣ i2 antibody (AS7) was diluted to a point where the endogenous G␣ i2 was barely detectable. Fig. 3B shows a confocal analysis of PC-, G␣ i2 -, and Q205L␣ i2 -MDCK stained and analyzed under identical conditions using a 1:100 dilution of the G␣ i2 antibody. In panel a, PC-MDCK cells only demonstrate faint intracellular staining, but in panels b and c, transfected G␣ i2 and Q205L␣ i2 can be visualized in the subapical lateral membrane overlapping with the TJ marker, ZO-1. Again, there is intracellular staining that is similar to the endogenous G␣ i2 (Fig. 3A). Overall the pattern of transfected G␣ i2 and Q205L␣ i2 is very similar to that seen with the endogenous G␣ i2 subunits. These results confirm that transfected G␣ i2 and Q205L␣ i2 partition between the lateral membrane overlapping with the TJ and intracellular compartments. This finding is similar to our prior findings with G␣ o -transfected MDCK cells (16).
Because transfected G␣ i2 and Q205L␣ i2 were localized in a manner similar to that of the endogenous G␣ i2 , we next determined whether G␣ i2 localization in the TJ had any functional consequences for the tight junction. PC-, G␣ i2 -, and Q205L␣ i2 -MDCK cells were simultaneously analyzed under steady state conditions and also by using the Ca 2ϩ switch. TJ integrity was followed by measurement of transepithelial resistance.  (16), and several clones were analyzed with no significant differences seen among the clones. To gain insight into the mechanism of higher TER values observed in Q205L␣ i2 -MDCK cells, all three cell lines were simultaneously analyzed in the Ca 2ϩ switch. The elevated TER in Q205L␣ i2 -MDCK cells could be achieved by differences in the kinetics of TER development. Nonlinear regression analysis of the TER data between 0 -12 h for all of the cell lines (Fig. 4A) indicates an asymptotic approach to peak TER. Although the data do not precisely fit standard kinetic models, the kinetics of TER development in these cells is similar to what has been reported in other studies (15,26). The time to half-maximal TER is a useful value for discussing the effects of G␣ i2 expression on TER biogenesis, and these values were calculated for each cell line in the presence and absence of AlF 4 Ϫ (Table I). PC-, G␣ i2 -, and Q205L␣ i2 -MDCK cells were plated at confluent density on Transwell filters, and base-line TER values were obtained 36 -40 h later. Cells were then placed in low calcium medium and TJ biogenesis followed over time using the Ca 2ϩ switch as described under "Experimental Procedures." Blanks were subtracted for each experiment. A, base line (BL) and time course of TER development. TER values were significantly higher for Q205L␣ i2 -MDCK cells (1187 Ϯ 150 ⍀⅐cm 2 ; p Ͻ 0.001) than for G␣ i2 -MDCK (576 Ϯ 89 ⍀⅐cm 2 ) or PC-MDCK (377 Ϯ 52 ⍀⅐cm 2 ). The difference between G␣ i2and PC-MDCK was not significantly different. TER values were obtained every 2 h after switching from low calcium to NC medium at time 0. Post-Ca 2ϩ switch base-line values are shown at 26 h. There was a decrease in base-line TER values for each of the three cell lines, but Q205L␣ i2 remained significantly higher than the other two cell types. Results are expressed as the means Ϯ S.E. of 12 independent experiments with n ϭ 4 -6 for each cell line in each experiment. Graphs were generated and statistical analyses were performed on data using Graphpad Prism 2.0 (Graphpad Software, Inc.) B, post-Ca 2ϩ switch base-line TER values obtained at 26 h for the six independent experiments Ϯ AlF 4 Ϫ . The differences Ϯ AlF 4 Ϫ were significant for each of the cell lines (p ϭ 0.008 for PC and p ϭ 0.002 for both G␣ i2 and Q205L␣ i2 -MDCK cells). Ϫ (n ϭ 6; added in the low calcium medium and continued after the switch). TER Max is the line defined by the maximal TER obtained between 6 -12 h and was held constant for calculation of the time to one-half maximal TER (T 50 ). The differences in T 50 between PC-MDCK cells and G␣ i2 -and Q205L␣ i2 -MDCK cells were significant (p Ͻ 0.02), and the difference with or without AlF 4 Ϫ was significant only for PC cells (p Ͻ 0.05). Ϫ . Similar effects of AlF 4 Ϫ were seen with the three cell lines cultured in the steady state (not shown). This raises the possibility that AlF 4 Ϫ activates additional G␣ subunits in the steady state that enhances transepithelial resistance.
Taken together, these studies offer direct evidence that G␣ i2 is a critical regulator of tight junction biogenesis and affects base-line characteristics of the tight junction. The protein composition of the TJ is complex with one integral membrane protein identified so far (occludin), several peripherally attached proteins with partially defined functions (including ZO-1, -2, and -3) and a variety of signal transduction molecules including PKC isoforms, G␣ subunits, and tyrosine kinases (see Ref. 5 for review). How these diverse proteins function to maintain and regulate the development of tight junctions is not well understood. G proteins could be activated within the TJ through a classical seven-transmembrane receptor (although none yet identified in the TJ) or alternatively through a modulatory protein that promotes GDP release or slows GTP hydrolysis. Additional transmembrane proteins must exist within the TJ (27), and there are multiple examples of modulatory proteins that affect G protein function. GTPase activating proteins (RGS proteins; regulators of G protein signaling; reviewed in Ref. 28) interact with G␣ subunits, and nucleotide exchange factors that promote GDP release have been described for many small G proteins such as Ras (29). Although analogous proteins for G␣ subunits have not yet been identified, such proteins may exist and could provide mechanisms for activation of G␣ subunits in the TJ or within intracellular compartments (30). Our findings that G␣ i2 is important for both the maintenance and development of the TJ does not exclude roles for other G␣ subunits, and in fact the effects of AlF 4 Ϫ on the steady state TER suggests that other G␣ subunits are likely to enhance this barrier.