We have developed a model system for studying integrin regulation of mammalian epithelial tubule formation. Application of collagen gel overlays to Madin-Darby canine kidney (MDCK) cells induced coordinated disassembly of junctional complexes that was accompanied by lamellipodia formation and cell rearrangement (termed epithelial remodeling). In this study, we present evidence that the Rho signal transduction pathway regulates epithelial remodeling and tubule formation. Incubation of MDCK cells with collagen gel overlays facilitated formation of migrating lamellipodia with membrane-associated actin. Inhibitors of myosin II and actin prevented lamellipodia formation, which suggests that actomyosin function was involved in regulation of epithelial remodeling. To determine this, changes in myosin II distribution, function, and phosphorylation were studied during epithelial tubule biogenesis. Myosin II colocalized with actin at the leading edge of lamellipodia thereby providing evidence that myosin is important in epithelial remodeling. This possibility is supported by observations that inhibition of Rho kinase, a regulator of myosin II function, alters formation of lamellipodia and results in attenuated epithelial tubule development. These data and those demonstrating myosin regulatory light-chain phosphorylation at the leading edge of lamellipodia strongly suggest that Rho kinase and myosin II are important modulators of epithelial remodeling. They support a hypothesis that the Rho signal transduction pathway plays a significant role in regulation of epithelial tubule formation.
- signaling pathway
epithelia are physiological barriers to movement of solutes and fluids in multicellular organisms. To provide these barriers, epithelia are organized into sheets or tubules with individual cells held together by intercellular junctions (15, 19). Morphological and biochemical studies have demonstrated that epithelial cells are polarized and have structurally and functionally distinct apical and basolateral membranes (62). Epithelial biogenesis is currently an area of intense study in mammalian cells at both the subcellular and cellular levels (39, 62). Formation and migration of epithelial sheets in embryos are critical events during development (19, 31, 39) and work toward identification of the factors involved has benefited from genetic analysis of invertebrate model systems (32).
An important approach to studying epithelial development has been observation of cellular dynamics and epithelial tubule development in collagen gels. This takes advantage of observations that extracellular matrix (ECM) plays an important role in epithelial biogenesis (7, 39). The Madin-Darby canine kidney (MDCK) cell line has been used in model systems to study ECM and regulation of epithelial tubule formation. Epithelial cysts formed by culturing MDCK cells in suspension have polarized distributions of membrane proteins (59). Resuspension of these cysts in collagen gel induces extensive membrane remodeling and subsequent epithelial polarity reversal (60). Another model utilizes clonal growth of MDCK cells in collagen gel to produce polarized cysts with lumenal microvilli (60). Treatment of these cysts with hepatocyte growth factor (HGF) induces cell migration and formation of polarized tubular extensions that contain adherens and tight junctions (39, 44, 63).
Our laboratory used the collagen gel-overlay model developed by Bissell and colleagues (21) for the study of epithelial tubule biogenesis. When MDCK cells are overlaid with collagen gel, the cells undergo extensive remodeling that leads to formation of polarized epithelial tubules (21, 41, 50, 53, 64). This model system has the advantage that the collagen gel overlay can be removed to allow physiological and biochemical analyses of the remodeled epithelium that remains attached to the ECM substrate (41, 50, 64). Based on studies in our laboratory and others, there is good evidence that epithelial tubule formation requires the binding of integrin to collagen gel (see discussion).
Integrins are a family of ECM receptors that regulate epithelial cell attachment to the basal lamina and modulate an extensive signal transduction pathway (13, 18, 24, 49). We and others have used the collagen gel-overlay model to demonstrate that epithelial tubule formation is regulated by integrin binding to ECM (41, 43, 50, 64). It has been determined that α2β1-integrin is the critical receptor regulating MDCK epithelial tubule formation in collagen (43, 47, 53). More recently, we demonstrated that epithelial tubule formation was accompanied by extensive cell rearrangement, disruption of the junctional complex, and formation of lamellipodia (41). Furthermore, increased tight-junction permeability and cell rearrangement were completely blocked by an inhibitory β1-integrin monoclonal antibody, which provides evidence that integrin signal transduction pathways are involved in regulation of epithelial tubule formation in this model system (41, 50).
Integrin signal transduction plays an important role in regulation of cell adhesion and migration (11, 24), and there is evidence that cell binding to ECM activates the Rho signaling pathway (13, 18, 49). Rho family members include Rho, Rac, and Cdc42. These small GTPases are involved in organization of the actin cytoskeleton, formation of cell junctions, and regulation of cell motility (17, 20). The Rho signaling pathway regulates the activity of Rho kinase and myosin light-chain kinase (MLCK), which are downstream effectors involved in modulation of myosin II function (3). Along with myosin phosphatase, these kinases modulate a reciprocal cycle of myosin light-chain (MLC) phosphorylation and dephosphorylation regulating actomyosin interactions that are involved in stress-fiber formation, cell motility, and cell-to-cell interactions (10, 11, 23). Furthermore, Rho family GTPases are important modulators of epithelial cell-to-cell interactions in keratinocytes and MDCK cells (8, 9, 25, 57). Epithelial junctional complexes are associated with actin microfilaments, and there is considerable evidence that adherens and tight-junction structure and function are regulated by the cytoskeleton (1, 4, 26, 35, 57). In this article, we present data on the function of the Rho signaling pathway in integrin-regulated epithelial tubule formation using the collagen gel-overlay model system.
MATERIALS AND METHODS
Cell culture. MDCK II cells were cultured in DMEM that contained 10% FBS at 37°C in an atmosphere of 5% CO2-95% air as previously described (41, 42). For experiments, cells were plated at 2.5 × 105 cells/ml on the following type I collagen-coated substrates: coverglasses for immunofluorescence and confocal microscopy, micropore filters (0.45-μm pores; Millipore) for transepithelial electrical resistance (TER) measurements, and 35-mm-diameter tissue-culture wells (Falcon) or Transwell filters (0.4-μm pores; Costar) for SDS-PAGE and immunoblotting. Once plated, cells were cultured for 2 days to produce confluent monolayers. For integrin-regulated epithelial tubule formation, cell monolayers were incubated with type I collagen gel overlays as previously described (21, 40, 50) for the times indicated. At different times the gels were removed by aspiration and the cells were prepared for microscopy or biochemistry. For experiments utilizing inhibitors, the drugs were included in collagen gels at their final concentrations.
Antibodies and reagents. The antibodies used in this study included mouse monoclonal antibody (MAb) 3F2 against MDCK apical membrane protein gp135 (produced by our laboratory; Ref. 42), MAb against β-catenin (purchased from BD Transduction Laboratories), rabbit antibody against nonmuscle myosin II (Biomedical Technologies), and rabbit antibodies against phosphorylated MLC-Ser19 and MLC-Thr18/Ser19 (from Cell Signaling Technologies). Rabbit antibodies against myosin IIA and IIB (coupled to Fluor-X) were provided by Dr. Anne Bresnick (Albert Einstein College of Medicine). Drugs were obtained from the following sources: latrunculin A was obtained from Sigma Chemical, phalloidin-Texas red was from Molecular Probes, blebbistatin and ML-7 were from Calbiochem, and Y-27632 was a gift from the Mitsubishi Pharma (Yokohama, Japan) and was also purchased from Calbiochem. Goat anti-rabbit IgG-FITC and goat anti-mouse IgG-Texas red secondary antibodies were purchased from Jackson Laboratories.
Immunofluorescence and confocal microscopy. MDCK cells on coverglasses were fixed with 4% formaldehyde in PBS for 30 min at 4°C and then permeabilized with 0.05% Triton X-100 (TX-100) for 10 min at 4°C. In some experiments, cells were permeabilized with 0.05% TX-100 before formaldehyde fixation. This procedure was used to enhance actin and myosin II staining (see Fig. 2) as described previously (37). For gp135 and β-catenin localization, cells were fixed with methanol for 5 min at –20°C. Fixed cells were washed with PBS and blocked in 3% BSA-1% goat serum-PBS (BSA-GS). Primary antibodies, secondary antibodies, and phalloidin-Texas red were diluted in BSA-GS as follows: myosin IIA-Fluor-X (1:25 dilution), myosin IIA (1:25 dilution), phosphorylated MLC-Ser19 and MLC-Thr18/Ser19 (1:20 dilution), gp135 (1:5 dilution), β-catenin (1:100 dilution), phalloidin-Texas red (1:75 dilution), and goat anti-rabbit IgG-FITC and goat anti-mouse IgG-Texas red secondary antibodies (1:100 dilution). Fixed cells were incubated with antibodies or phalloidin for 1 h each and mounted in 10% glycerol-PBS with 12% triethyldiamine (Sigma Chemical) to prevent bleaching. Immunofluorescence microscopy was done on a Zeiss microscope equipped with epifluorescence optics. Images were taken on Kodak Elite film. Confocal microscopy was done on a Bio-Rad Radiance 2000 laser scanning confocal microscope. Z-series images were captured at 1-μm intervals starting at a level just above the apical cell surface. Immunofluorescence and confocal images were assembled into figures using Adobe Photoshop and Canvas software on a G4 Power Macintosh computer.
TER measurements. The permeability of tight junctions was determined using TER measurements as described previously (41). Briefly, MDCK cells were grown to confluency on collagen-coated micropore filters attached to Lexan plastic cylinders. Cells were incubated without or with collagen gel overlays and inhibitor for 2 or 6 h at 37°C. The Rho-kinase inhibitor Y-27632 was included in both the collagen gel and basolateral compartment below the filter to ensure that no changes in drug concentration occurred during increased tight-junction permeability. TER (in Ω·cm2) was determined using 3 M KCl-agarose salt bridges and Hg-HgCl electrodes to pass 10 μA of current and measure voltage changes. Initial TER measurements were taken before the start of an experiment to determine the integrity of individual monolayers (average TER readings, 150–300 Ω·cm2) and were compared with the final reading at the conclusion of an experiment to determine percent change. Statistical evaluation of TER and biochemical data was done using a two-tailed Student's t-test.
SDS-PAGE and immunoblotting. To determine the association of myosin II and β-catenin with the cytoskeleton, we used a differential detergent-extraction procedure described previously (41). Briefly, cells were incubated without or with collagen gel overlays that contained or lacked inhibitor for the indicated times. Cells were extracted with CSK buffer that contained 0.5% TX-100 for 10 min at 4°C to produce the soluble fraction (22, 41). After removal of the soluble fraction, the remaining TX-100-insoluble residue was extracted with RIPA buffer that contained 1% TX-100, 1% sodium deoxycholate, and 0.1% SDS for 10 min at 4°C to produce the cytoskeletal fraction. Both CSK and RIPA buffers contained the appropriate protease and phosphatase inhibitors (41). Proteins from the soluble and cytoskeletal fractions were separated by SDS-PAGE, transferred to nitrocellulose sheets, and detected via immunoblotting using primary antibodies against either myosin II (1:250 dilution) or β-catenin (1:250 dilution) followed by either goat anti-rabbit IgG or goat anti-mouse IgG coupled to alkaline phosphatase (1:2,000 dilution) and enhanced chemiluminescence (Amersham). Quantitation of protein levels was done using either scanning densitometry or NIH Image software, and the data are presented as soluble/cytoskeletal ratios (22, 41).
Actin and myosin II are involved in epithelial remodeling. In previous studies, we observed that incubation of MDCK monolayers with collagen gel overlays induced formation of lamellipodia and extensive epithelial remodeling (41, 50). Because lamellipodial migration in other epithelial developmental systems is due to actin polymerization (6, 12, 32, 56) and myosin function (10, 23), experiments were done to establish whether similar mechanisms were utilized for MDCK lamellipodia formation. Phalloidin staining demonstrated that actin was localized primarily to the lateral membrane (Fig. 1a) as described previously (25, 42). Incubation with collagen gel overlays induced epithelial remodeling including the formation of lamellipodia that contained membrane-associated actin filaments (Fig. 1c). The ultrastructure of these lamellipodia was presented in a previous study (41). To determine whether myosin II was involved in epithelial tubule formation, MDCK cells were incubated with collagen gel overlays that contained blebbistatin, a recently characterized myosin II inhibitor (51). Blebbistatin (10 μM) alone had no effect on the actin paracortical staining pattern (Fig. 1b). However, inclusion of blebbistatin in collagen gel overlays prevented lamellipodia formation and cell rearrangement (Fig. 1d), which strongly suggests that actomyosin function plays an important role in regulation of epithelial remodeling. Furthermore, latrunculin A (0.25 μM), which is an inhibitor of actin polymerization, disrupted the actin staining pattern, and inclusion of this inhibitor in collagen gel overlays prevented lamellipodia formation (R. Eisen and G. Ojakian, unpublished data).
Actomyosin interactions during lamellipodia formation. Numerous studies have shown that interactions between actin and myosin II are important in regulation of cell-to-ECM binding, cell-to-cell interactions, and motility in nonmuscle cells (3, 10, 11, 23). Therefore, we focused this study on myosin II modulation of epithelial remodeling and tubule formation. Cells incubated for 6 h without or with collagen gel overlays were fixed and double labeled for actin and myosin II. For this we used an antibody against myosin IIA, which is one of the two predominant myosin isoforms expressed in nonmuscle cells (10). Confocal microscopy demonstrated that myosin IIA on lateral membranes colocalized with actin (Fig. 2, a–c). Using available antibodies, we were unable to detect myosin IIB. Z-series confocal scans through control monolayers determined that myosin IIA was colocalized to the upper 50% of the lateral membranes, whereas actin staining was detected over the entire length. These results are consistent with previous observations of MDCK and other epithelial cells (33, 37, 51). After incubation with collagen gel overlays, confocal analysis demonstrated actin and myosin IIA colocalization at the leading edge of lamellipodia (Fig. 2, d and e), which suggests that actomyosin interactions were occurring.
The redistribution of actin and myosin II during lamellipodia formation and cell migration suggested that dynamic changes in actomyosin interactions were occurring during epithelial remodeling and tubule formation. To study this possibility in more detail, we utilized a detergent-extraction procedure that has been used to demonstrate cytoskeletal membrane protein interactions in MDCK cells (1, 22, 35, 41). Differential detergent extraction and immunoblot analysis were performed on cells incubated without or with collagen gel overlays for 6 h. Quantitative immunoblotting demonstrated that ∼40% of myosin II was in the soluble fraction whereas ∼60% was associated with the cytoskeletal fraction (Fig. 3). After 6 h of incubation with collagen gel overlays, a dramatic rearrangement had occurred: ∼70% of the myosin II was in the soluble fraction, while ∼30% was associated with the cytoskeleton (Fig. 3). These results suggest that myosin II was in a dynamic state during epithelial tubule formation.
Rho-kinase regulation of cell motility and MLC phosphorylation. Actomyosin function plays an important role in the regulation of mammalian cell migration (1, 10, 11, 23) and cell polarity development (29). There is considerable evidence that the Rho signaling pathway regulates myosin function through its downstream effector Rho kinase in nonmuscle cells (10, 11, 23). To determine whether the Rho signaling pathway was involved in regulation of epithelial tubule formation, we utilized Y-27632, a specific inhibitor of Rho kinase (14, 55). This inhibitor has been used to modulate the Rho signal transduction pathway in cultured cells (2, 29, 38, 61, 63). MDCK cells were incubated without or with collagen gel overlays in the absence or presence of Y-27632. Based on previous work, β-catenin was localized to characterize the initial stages of lamellipodia formation (41). Incubation with Y-27632 alone for 6 h or collagen gel overlays for 2 h had no effect on β-catenin lateral membrane staining (Fig. 4, b and c). However, after 2 h of incubation with collagen gel overlays that contained Y-27632, short lamellipodia with membrane-associated β-catenin had formed (Fig. 4d). At later stages of epithelial tubule formation (6 h), lamellipodia in collagen gel-treated monolayers had a broad leading edge (Fig. 4e), whereas those in collagen gel that contained Y-27632 were not elongated (Fig. 4f); this suggests that epithelial cell migration was altered. These data and those presented below provided evidence that Rho kinase was involved in modulation of epithelial remodeling.
Our laboratory has used TER measurements to quantitate epithelial remodeling in MDCK monolayers (41). Because Y-27632 appeared to be affecting lamellipodia formation, TER measurements were done in the presence and absence of Y-27632 to determine the epithelial integrity. This concentration of inhibitor was previously shown to produce small changes in intestinal epithelial cell tight-junction permeability (58). Consistent with these observations, Y-27632 induced a significant reduction in TER (∼25% below control) that was similar to incubation with collagen gel alone (Fig. 5). Importantly, collagen gel that contained Y-27632 produced a large reduction (∼50%) in TER that was additive compared with Y-27632 and collagen alone (Fig. 5).
In addition to tight junctions, we have demonstrated that a coordinated disassembly of adherens junctions also occurs during collagen-mediated epithelial remodeling (41). To study the possibility that Rho kinase was involved in regulation of adherens junction function, quantitative measurements on β-catenin association with the cytoskeleton were done after incubation of MDCK cells with collagen gel overlays that contained or lacked Y-27632. These data demonstrated that incubation with Y-27632 alone had a small effect on dissociation of β-catenin from the cytoskeleton (Fig. 6). Incubation with collagen gel alone induced an approximately twofold dissociation of β-catenin, whereas incubation with collagen that contained Y-27632 produced an additive β-catenin dissociation (Fig. 6). Importantly, these changes were comparable to those observed for TER (approximately fourfold) under identical conditions (Fig. 5), which provides additional evidence for the coordinated disassembly of tight and adherens junctions during epithelial remodeling.
Previously we demonstrated that incubation of MDCK cells with collagen gel overlays for 24 h induced formation of multicellular polarized epithelial tubules (40, 50). This was determined by staining the apical lumens with MAb 3F2 against the apical membrane protein gp135 (42). Here we use this procedure to determine the role of Rho kinase in regulation of tubule formation. Incubation with collagen gel overlays for 24 h induced formation of large branching tubular structures (Fig. 7, a and b). These tubules were identical to those observed previously by using light and electron microscopy (41, 50). Inclusion of Y-27632 in collagen gel overlays prevented formation of large tubules, and smaller tubular structures were observed on the lateral membranes between adjacent cells (Fig. 7, c and d). Interestingly, these small tubules were similar to those observed in subconfluent MDCK cultures treated with collagen gel overlays (40). Occludin staining demonstrated that tight junctions were associated with both the small and large tubular structures (Fig. 7). The punctate cytoplasmic occludin staining may represent vesicles carrying components to developing tight junctions. Taken together, these data (see Figs. 4, 5, 6, 7) strongly suggest that Rho kinase played a significant role in regulation of epithelial remodeling and tubule formation in our model system.
Phosphorylation of myosin regulatory light chain. The Rho signal transduction pathway modulates myosin II activity and cell motility by regulating MLC-Ser19 phosphorylation (10, 23, 33, 52). To determine whether MLC function was important in regulation of epithelial tubule formation, we used an antibody that specifically recognizes phosphorylated MLC-Ser19 (P-MLC-Ser19). P-MLC-Ser19 staining was not detected in control cells (R. Eisen and G. Ojakian, unpublished observations), which suggests that little or no MLC was phosphorylated in quiescent cell monolayers. However, after incubation with collagen gel overlays, confocal microscopy demonstrated a significant level of P-MLC-Ser19 staining (Fig. 8), which suggests that myosin activity had increased during epithelial remodeling. Importantly, P-MLC-Ser19 was found at the leading edge of migrating lamellipodia colocalized with actin (Fig. 8c). Similar observations were made using an antibody against P-MLC-Thr18/Ser19 (Fig. 8f). These data were consistent with those that suggested that antibodies against P-MLC-Ser19 also recognized P-MLC-Thr18/Ser19 (33). Inclusion of Y-27632 in collagen gel overlays inhibited phosphorylation of MLC-Ser19 (Fig. 9), which provides supporting evidence that Rho kinase was involved in regulation of MLC phosphorylation during epithelial remodeling.
There is evidence that phosphorylation of MLC can be regulated by either of two possible mechanisms. One is through reciprocal phosphorylation regulated by MLCK and myosin phosphatase (23), and the other utilizes direct phosphorylation of MLC by Rho kinase (3, 54). To make this determination in our model system, we used the MLCK inhibitor ML-7 in collagen gel-overlay experiments. We determined that ML-7 had no effect on epithelial remodeling (Fig. 10), thereby providing further evidence that MLC could serve as a substrate for Rho kinase. These data are supported by observations that ML-7 did not inhibit MLC-Ser19 phosphorylation on lamellipodia during epithelial remodeling in collagen gel (R. Eisen and G. Ojakian, unpublished observations).
Our laboratory and others have demonstrated that incubation of MDCK monolayers with collagen gel overlays offers an excellent model system for studying regulation of epithelial tubule formation (40, 41, 50, 53, 64). Here we present evidence that the Rho family signal transduction pathway plays an important role in cytoskeletal and integrin regulation of this developmental process. These data support a hypothesis that Rho-kinase activation is a critical component in epithelial tubule development and that myosin II is an important downstream effector in our model system of tubulogenesis. We propose that integrin interaction with collagen activates the Rho signal transduction pathway to increase MLC phosphorylation levels and regulate activation of myosin II during epithelial remodeling. This activation allows interaction of myosin II with lamellipodial actin thereby increasing actomyosin tension in the leading edge and promoting epithelial cell migration, remodeling, and tubule formation. Based on our observations that Y-27632 inhibited phosphorylation of MLC and formation of large multicellular epithelial tubules, it appeared likely that downstream Rho signaling was modulated through Rho kinase and not MLCK. These data were supported by observations that the MLCK inhibitor ML-7 did not inhibit epithelial remodeling or MLC-Ser19 phosphorylation. They also suggest that MLC serves as a substrate for Rho kinase during epithelial tubule formation, a model consistent with that previously demonstrated for MDCK and other cells (3, 52, 54).
During completion of the studies presented here, another laboratory published a paper demonstrating the effects of Y-27632 on MDCK cysts in collagen gel and HGF-induced epithelial tubule formation (63). Based on the evidence presented in this article, we confirm and extend their findings. There are similarities and differences between the HGF model system and the collagen gel-overlay model used in our study. Although both models utilize formation of nonpolarized cell extensions, considerable differences in lamellipodia behavior were observed during epithelial tubule formation. Incubation of cysts with Y-27632 produced short cell extensions, whereas this Rho-kinase inhibitor did not affect the morphology of monolayers used in our study. Cysts incubated with combined HGF and Y-27632 exhibited pronounced increases in the number and length of cell extensions and extensive cell migration from the cyst epithelium (63). In contrast, incubation of MDCK monolayers with collagen gel overlays that contained Y-27632 did not induce increases in lamellipodia length or number. Quite the contrary, Rho-kinase inhibition prevented cell migration and resulted in attenuated formation of epithelial tubules. These studies demonstrate that the capacity to form apical lumens in the collagen-overlay model is not affected by Rho-kinase inhibition. They suggest that epithelial tubule biogenesis utilizes two distinct mechanisms that are not physiologically coupled: 1) myosin II-regulated cell migration, and 2) cellular targeting mechanisms required for polarized apical lumen formation. Interestingly, despite the apparent differences between the HGF and collagen gel-overlay models in response to Rho-kinase inhibition, increased MLC phosphorylation was observed in both during stimulation of epithelial tubule formation. This suggests that both model systems utilize common signal transduction pathways that are subject to additional modulation by growth factors.
Myosin II interactions with actin also play an important role in regulation of nonmuscle cell motility (10, 11, 31). Alterations in myosin II activity can have profound effects on formation of stress fibers and lamellipodia, cell-to-ECM binding, cell-to-cell interactions, and motility depending on the cell type (16, 20, 23, 30, 38). In our study, we obtained evidence that myosin II binding to the cytoskeleton was reduced during epithelial remodeling, thereby potentially allowing myosin II redistribution from actin filaments to lamellipodial membranes. Furthermore, myosin IIA and P-MLC-Ser19 colocalized with actin at the leading edge of MDCK lamellipodia during epithelial tubule formation with a distribution similar to that observed in migrating cells including MDCK cells (16, 28, 33, 34). We propose that signaling initiated through the α2β1-integrin activated the Rho family signal transduction pathway to increase MLC phosphorylation, promote actomyosin interactions at the leading edge of lamellipodia, and modulate MDCK cell motility during epithelial tubule biogenesis.
Although inclusion of Y-27632 in collagen gel overlays accelerated disassembly of tight and adherens junctions, MDCK cell migration appeared to be inhibited. These observations are supported by studies that demonstrate that Y-27632 inhibits leukocyte migration (2, 61). Therefore, it appears likely that increased cell migration due to decreases in cell-to-cell adhesion was not sufficient to form large epithelial tubules in collagen. Rather, our observations that Rho-kinase inhibition reduced the size of tubule lumens provides strong evidence that myosin II activity is required to modulate the orderly assembly of epithelial cells into large multicellular tubules. This hypothesis is supported by studies on the mechanism of dorsal closure in Caenorhabditis elegans and Drosophilia embryos. During this developmental process, actin- and myosin-containing lamellipodia at the leading edge of migrating epithelial sheets zipper together to form an intact epithelium (45). Importantly, inhibition of either the Rho signaling pathway or myosin II function altered dorsal closure (27, 32) in a manner analogous to that observed during Rho-kinase inhibition of MDCK epithelial tubule formation.
The initial steps in migration of MDCK cells from monolayers may require cortical tension produced by actomyosin interactions as a driving force. In principal, this suggested mechanism is similar to the purse-string contraction hypothesis proposed for epithelial wound closure (5, 27). Our confocal analysis demonstrates that myosin II is localized to the upper 50% of the MDCK lateral membrane adjacent to tight junctions. Increased cortical tension due to integrin signaling through the Rho pathway could regulate cell migration from the monolayer utilizing mechanisms similar to those described for both epithelial wound healing and apoptosis. During epithelial wound closure, there is redistribution of actin and myosin II including P-MLC-Ser19 to lateral membrane sites adjacent to extruded cells (5, 27, 33), which suggests that cell contraction is involved. Taken together, these observations indicate that a common mechanism is used for a variety of epithelial cell functions including extrusion of apoptotic cells from epithelial layers due to actomyosin contraction (36, 46).
The Rho signal transduction pathway also plays a prominent role in the regulation of epithelial permeability and cell-to-cell adhesion. Expression of constitutively active and dominant-negative forms of Rac and Rho GTPases in MDCK cells and keratinocytes affected both tight and adherens junction structure and function (9, 25, 26). There is also evidence to demonstrate that Y-27632 inhibited tight and adherens junction formation in keratinocytes and intestinal and MDCK cells (8, 48, 58). However, studies exist to support the possibility that distinct signaling pathways regulate adherens and tight junction function. C3 transferase, a specific Rho inhibitor, disrupted adherens junctions (48) whereas Y-27632 did not (48, 58). Furthermore, although both Y-27632 and collagen gel overlays affected adherens and tight junction function to the same extent in our studies, Y-27632 in combination with collagen produced additive effects. This is consistent with observations that Rho and Rho kinase have numerous downstream effectors (3) and supports a hypothesis that Rho signaling during integrin-regulated epithelial tubule formation is mediated through diverse pathways.
We thank Dr. Anne Bresnick for expert advice and for generously providing anti-myosin II antibodies. We also thank Drs. Bill Chirico, John Condeelis, Eva Cramer, John Lewis, Anne Müsch, and Tim Sutton for helpful discussion; the Mitsubishi Pharma for providing Y-27632; Shereaf Walid for critically reading the manuscript; Vincent Garofalo for graphic arts assistance; Una Yearwood for typing the manuscript; and Dr. Jeremy Weedon (Scientific Computing Center) for assistance with statistical analysis.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-60570 and a grant from the American Heart Association, Heritage Affiliate (to G. K. Ojakian).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society