Both microtubules and actin filaments are required for efficient postendocytotic traffic of the polymeric immunoglobulin receptor in polarized Madin-Darby canine kidney cells.

It has been postulated that membrane traffic in polarized epithelial cells requires both actin filaments and microtubules. We have tested this hypothesis by analyzing the effect of cytochalasin D (cytoD; an actin-disrupting agent), by itself or in combination with nocodazole (a microtubule depolymerizing agent), on postendocytic traffic in Madin-Darby canine kidney cells. CytoD treatment inhibited basolateral to apical transcytosis of IgA in polymeric immunoglobulin receptor-expressing cells by approximately 45%, but had little effect on basolateral recycling of transferrin. Apical recycling of IgA was also inhibited by approximately 20%. Like nocodazole, cytoD acted at an early step in transcytosis, and inhibited translocation of IgA between the basolateral early endosomes and the apical recycling endosome. There was little inhibition of the subsequent release of IgA from the apical recycling endosome of cytoD- or nocodazole-treated cells. Order-of-addition experiments suggest that the cytoD-sensitive step preceded the nocodazole-sensitive step. Treatment with both cytoD and nocodazole inhibited transcytosis 95%. These results suggest that in addition to microtubules, efficient postendocytic traffic in polarized epithelial cells also requires actin filaments.

Epithelial cells also have an actin cytoskeleton that is associated with a variety of cell structures including the microvilli, the terminal web, the tight and adherent junctions, and stress fibers (1, 16 -19). The exact role of actin filaments in membrane trafficking is unknown. In some systems, vesicle fusion with the plasma membrane is accompanied by local actin depolymerization (20 -22). In these examples actin may act as a molecular fence, preventing fusion until localized depolymerization allows vesicles access to the plasma membrane. In other systems an actin-myosin motor complex may be required for pinching off forming endocytic pits (23)(24)(25).
Additional evidence suggests that actin in conjunction with unconventional myosin motors (reviewed in Ref. 26) may be responsible for directing trans-Golgi network-derived vesicles to the apical plasma membrane of intestinal epithelial cells (27)(28)(29). Vesicles derived from this organelle contain both brush-border myosin I and the microtubule motor protein dynein. In light of this finding, it has been proposed that fusion with the apical plasma membrane requires the sequential transport of apical transport vesicles along both the microtubule and actin filament cytoskeleton (27)(28)(29). Presumably, initial vesicle transport would occur along microtubules, and, because microtubules rarely extend past the terminal web (16,30), a shift to an actin-based motor mechanism such as myosin I would be utilized to navigate vesicles through the terminal web. Similarly, traffic to or through the basolateral actin cortex may also require these unconventional myosins attached to transport vesicles (29).
In MDCK 1 cells, basolaterally internalized macromolecules transit between the basolateral early endosomes, the apical recycling endosome (ARE), the late endosomes, and the apical plasma membrane (31,32). A prediction of the above described sequential transport hypothesis is that many of these membrane trafficking steps may require both actin filaments and microtubules (27,29). We have tested this prediction by analyzing the effect of cytoD, by itself or in combination with nocodazole, on postendocytic traffic in polarized MDCK cells. The cytochalasins are actin-disrupting agents that have been used widely to study the role of actin filaments in different biological systems (33)(34)(35). We find the following: 1) cytoD inhibits basolateral to apical transcytosis by approximately 45%, but has little effect on the basolateral recycling pathway; 2) cytoD also inhibits the apical recycling pathway by approximately 20%; 3) like nocodazole, cytoD inhibition is early in the transcytotic pathway and inhibits transit between the basolateral early endosomes and the ARE; 4) nocodazole inhibits transcytosis by 60 -70%, and order-of-addition experiments suggest that the cytoD step precedes the nocodazole-sensitive step; 5) treatment with both cytoD and nocodazole inhibits transcytosis Ͼ95%. Our results suggest that the actin cytoskeleton has a previously unappreciated role in postendocytic traffic of polarized epithelial cells, and that postendocytic traffic requires both an intact actin filament and microtubule cytoskeleton.

EXPERIMENTAL PROCEDURES
Antibodies, Proteins, and Other Reagents-Purified human polymeric IgA was kindly provided by J.-P. Vaerman (Louvain Catholic University, Louvain, Belgium) and was used at a concentration of 50 g/ml. Rat monoclonal antibody ascites to ZO-1, a protein associated with tight junctions, was obtained from Chemicon (Temecula, CA) and was used at 1:100 dilution. FITC-phalloidin (Molecular Probes) was dissolved in methanol at a final concentration of 200 units/ml and stored at Ϫ20°C. Prior to its use, an aliquot was dried and brought up in the appropriate buffer at a final concentration of 10 units/ml. Canine apotransferrin (Sigma) was loaded with iron as described (36) and used at 25 g/ml. Rabbit polyclonal antibodies against this protein have been described (31). The mouse monoclonal antibody DM1␣, specific for ␣-tubulin, was obtained from Sigma and used at 1:1000 dilution. A mouse monoclonal anti-actin antibody (clone AC-40) was obtained from Sigma and used at 1:100 dilution. FITC-or Cy3-conjugated goat anti-human IgA, goat anti-rabbit IgG, goat anti-mouse IgG, or goat anti-rat IgG were obtained from Jackson ImmunoResearch Laboratories (West Grove, PA) and were used at 5-10 g/ml. Goat anti-mouse horseradish peroxidase was obtained from Jackson ImmunoResearch Laboratories and used at a 1:25,000 dilution.
Cell Culture-MDCK strain II cells expressing the wild-type rabbit pIgR have been described (37). Although the results presented in this paper are from one clone of MDCK strain II cells transfected with the pIgR cDNA, we have obtained very similar results using two other independent clones of pIgR-expressing MDCK cells (one of which was kindly provided by K. Singer, University of California, San Francisco, CA). Cells were maintained in minimal essential medium (Cellgro; Fisher) supplemented with 5% fetal bovine serum (HyClone, Logan, UT), 100 units/ml penicillin, and 100 g/ml streptomycin in 5% CO 2 , 95% air. In order to maintain a high level of receptor expression, new cells were thawed every 2-3 weeks, and were split 1:10 and passaged once weekly. Cells were cultured on 12-or 24-mm diameter, 0.4-m pore size Transwells (Costar, Cambridge, MA) as described (37). The cells were fed after the 3rd day and used 3-4 days post-culture.
Internalization of Ligands-Ligands were internalized from the apical and/or basolateral surface of filter-grown MDCK cells. Prior to Tf internalization, cells were incubated for 60 min at 37°C in MEM/BSA (MEM, Hank's balanced salt solution, 0.6% w/v BSA, 20 mM HEPES, pH 7.4) to deplete intracellular stores of Tf. All incubations in MEM/ BSA were performed in a circulating water bath. For basolateral uptake of ligands, the cells (cultured on 12-mm Transwells) were rinsed with MEM/BSA at 37°C and the edge of the filter on the side opposite the cells was carefully blotted to remove excess medium. The Transwell unit was placed on a 25-l drop of MEM/BSA containing the ligand. For apical uptake, the cells (cultured on 12-mm Transwells) were rinsed with MEM/BSA, excess fluid was aspirated from the cell side of the Transwell and 150 l of ligand, diluted in MEM/BSA, was added. All incubations were performed in a humid chamber.
Nocodazole Treatment and Extraction of Tubulin-Nocodazole (Calbiochem) was dissolved in Me 2 SO at 33 mM and stored at Ϫ20°C. In all experiments in which this drug was used, cells were pretreated 60 min at 4°C in the presence of 33 M nocodazole. The drug was included in subsequent incubations. The effectiveness of this treatment was assessed by extracting monomeric tubulin from filter-grown MDCK cells as described previously (7). Immunoblotting was performed as follows. Equal volumes of extracts I (monomeric tubulin) and II (polymeric tubulin) were resolved by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (10% gels). The gels were then equilibrated in 10 mM CAPS-NaOH, pH 11.0 (CAPS buffer), for 10 min at room temperature before proteins were transferred to Immobilon-P membrane (Millipore, Bedford, MA), for 75 min at 375-mA constant current in CAPS buffer.
The Immobilon-P membrane was blocked overnight in 5% BSA dissolved in Dulbecco's phosphate-buffered saline (PBS), and then incubated with a 1:1000 dilution of DM1␣ ascites (specific for ␣-tubulin) in 1% dry nonfat milk/PBS for 120 min at room temperature. Unbound primary antibody was removed by three 15-min washes in TBST (25 mM Tris, pH 8.0, 500 mM NaCl, 25 mM KCl, 0.05% [w/v] Tween 20), and three washes with PBS, and the Immobilon-P membrane incubated 60 min at room temperature with goat anti-mouse horseradish peroxidase diluted 1:25,000 in 1% dry nonfat milk/PBS. Following three 15-min washes in TBST, tubulin was detected using the SuperSignal chemiluminescent detection system (Pierce) following the protocol described by the manufacturer. The resulting bands were quantified by densitometry.
The above analysis was performed on cells incubated in the presence or absence of nocodazole or in the presence of 25 g/ml cytoD (see below) for 2.5 h at 37°C. In the presence of nocodazole, all of the tubulin was recoverable in the monomeric form (no polymeric tubulin was detected by Western blotting). In the absence of nocodazole, 23% of the tubulin was in the monomeric form and 76% was in the polymeric form. In the presence of cytoD, 15% of the tubulin was in the monomeric form and 85% was in the polymeric form. These observations confirm the effectiveness of nocodazole in depolymerizing microtubules, and that cytoD did not increase the amount of monomeric tubulin present in the cell.
CytoD Treatment and Quantification of F-actin Content-CytoD (Sigma) was dissolved in Me 2 SO at 25 mg/ml and stored at Ϫ20°C. In many experiments cells were pretreated for 15 min at 37°C with this drug. Extending this pretreatment period to 60 min had no additional effect. In experiments in which both cytoD and nocodazole were simultaneously added to the cells, the cells were first pretreated with nocodazole for 60 min at 4°C, and then cytoD and nocodazole included in subsequent reactions. The final concentration of Me 2 SO in the medium was Յ0.2%. At this concentration Me 2 SO alone (in the absence of drug) had no effect on the assays described.
A fluorometric assay was used to measure the F-actin (polymerized actin) content of MDCK cells (cultured on 24-mm Transwells) as described previously (38). The protocol was performed as detailed; however, following the methanol extraction the samples were read directly in a Perkin-Elmer spectrofluorimeter (excitation of 539 nm and an emission of 570 nm) using pure methanol as a blank. Filters, after extraction, were boiled 10 min in 100 mM triethanolamine, pH 8.6, 5 mM EDTA, 0.5% SDS, vortex-shaken for 15 min at 4°C, and protein content assessed using the BCA protein assay kit as described by the manufacturer (Pierce). BSA was used as the standard.
The above analysis was performed on cells incubated in the presence or absence of nocodazole or in the presence of 25 g/ml cytoD (see below) for 2.5 h at 37°C. Values (reported as the emission per microgram of protein) were as follows (n ϭ 4): control, 9.4 Ϯ 0.7; nocodazole, 8.7 Ϯ 0.3; cytoD, 8.1 Ϯ 0.7. As described previously cytoD had only a small effect on the total amount of F-actin present in the cell (33,39), and the F-actin content was not significantly affected by nocodazole treatment.
Fixation and Fluorescent Labeling of Cells-Samples were fixed with paraformaldehyde using a pH-shift protocol or with 0.25% (v/v) glutaraldehyde, quenched, blocked, stained, mounted, and stored as described previously (31). When staining with FITC-phalloidin, cells were fixed with glutaraldehyde and quenched with NaBH 4 as described (31), and then stained in PBS alone for 30 min at 37°C. The cells were then washed three times for 5 min with PBS, postfixed, mounted, and stored as described (31).
Scanning Laser Confocal Analysis of Fluorescently Labeled Cells-The samples were analyzed using an argon laser coupled to a Molecular Dynamics (Mountain View, CA) Multiprobe 2001 confocal, attached to a Diaphot microscope (Nikon, Melville, NY) with a Plan Apo 60 ϫ 1.4 numerical aperture objective lens (Nikon). The samples were scanned using the appropriate filter combinations. Collection parameters were as follows: laser output ϭ 25 milliwatts, PMT-1 set to 750 mV, laser attenuation at 3%, 50 m slit. The images (512 ϫ 512 pixels, 0.8 m pixel size) were acquired using ImageSpace software. The images were converted to tag-information-file-format (TIFF) and the contrast levels of the images adjusted in the Photoshop program (Adobe Co., Mountain View, CA) on a Power PC Macintosh 9500 (Apple, Cupertino, CA). Every attempt was made to collect and process images in an identical manner. The contrast-corrected images were imported into Pagemaker 6.0 (Adobe Co.) and printed from an Agfa 9800 imagesetter at 2400 dots per inch.
Assessment of Monolayer Permeability-Monolayer permeability was assessed by measuring the transepithelial resistance of control and drug-treated cells using a Millipore (Bedford, MA) ERS device. The value from a blank filter (with no cells, but otherwise treated identically) was subtracted from the resistance values of filters on which cells were plated. In addition, leakage of 125 I-BSA (1.0 -2.0 ϫ 10 7 cpm/g) across the monolayer was measured. Approximately 150,000 cpm of 125 I-BSA was added to the apical side of control or drug-treated monolayers for 2.5 h at 37°C. At the end of the incubation period, the amount of 125 I-BSA in the apical medium, basolateral medium, or cells was quantified with a ␥ counter (Packard, Palo Alto, CA).
Analysis of 125 I-IgA Postendocytotic Fate and Endocytosis-125 I-IgA was iodinated using the ICl method to a specific activity of 1.0 -2.0 ϫ 10 7 cpm/g (37). The postendocytotic fate of a preinternalized cohort of 125 I-IgA (at 5-10 g/ml) was analyzed as described (40). In brief, filtergrown MDCK cells expressing the wild-type pIgR were treated with or without drug, and 125 I-IgA internalized from the basolateral cell surface of the cells for 10 min at 37°C. Where appropriate, drug was included in all subsequent steps. The basal surface of the cells were rapidly washed three times, the apical and basolateral medium aspirated, and replaced with fresh medium. The cells were then incubated for 3 min at 37°C. This wash procedure takes 5 min at 37°C. Fresh medium was added to the cells, and they were chased for up to 2 h at 37°C. At the designated time points, the apical and basolateral media (0.5 ml) were collected and replaced with fresh media. After the final time point, filters were cut out of the insert and the amount of 125 I-IgA quantified with a ␥ counter. The media samples were precipitated with 10% trichloroacetic acid for 30 min on ice, and then centrifuged in a microcentrifuge for 15 min at 4°C. The amount of 125 I-IgA in the trichloro-acetic acid-soluble (degraded) and insoluble fractions (intact) was quantified with a ␥ counter. An equal number of MDCK cells (not expressing the pIgR) were treated identically, and these values were subtracted from those of MDCK cells expressing the pIgR. The postendocytic fate of apically internalized 125 I-IgA was essentially as described above. However, following ligand internalization the apical surface of the cells were treated three times for 10 min with 25 g/ml trypsin at 4°C to remove cell surface bound ligand. The cells were then treated with 125 g/ml soybean trypsin inhibitor for 10 min at 4°C. The postendocytic fate was determined as described above.
Endocytosis of 125 I-IgA was measured as described (40). At the conclusion of the experiment, 125 I-IgA was stripped from the cell surface by incubating the cells for 60 min at 4°C with 750 mM glycine, pH 2.5, diluted 1:5 with PBS ϩ (Dulbecco's phosphate-buffered saline with 0.5 mM MgCl 2 and 0.9 mM CaCl 2 ). The Transwells were rinsed with PBS ϩ , and the filters were cut out of their holders. The total 125 I-IgA initially bound to the cells included ligand recycled to the basolateral surface, ligand stripped from the cell surface with acid, and cell-associated ligand not sensitive to glycine stripping (endocytosed), and was quantified in a ␥ counter. Values for endocytosed ligand from filters that were never warmed up to 37°C (typically less than 10% of the total bound counts) were subtracted from the endocytosis values of cells that were allowed to internalize ligand at 37°C. In addition, an equal number of MDCK cells (not expressing the pIgR) were treated identi- cally, and these values were subtracted from those of MDCK cells expressing the pIgR.
Analysis of 125 I-Tf Recycling-Iron-saturated Tf was iodinated to a specific activity of 5.0 -9.0 ϫ 10 6 cpm/g using ICl as described (40). The cells were depleted of endogenous Tf by incubating for 60 min in MEM/BSA. When appropriate cytoD was included during the last 15 min of this starvation period and included in all subsequent steps. 125 I-Tf (5 g/ml) was internalized from the basolateral surface of the cells for 45 min at 37°C. The cells were washed three times for 5 min each with ice-cold MEM/BSA, and then warmed up to 37°C for 2.5 min to allow for receptor internalization as described previously (36). The medium was aspirated, fresh medium was added, and the postendocytic fate of 125 I-Tf assessed as described above. 125 I-Tf uptake was inhibited Ͼ95% when the radioactive ligand was internalized in the presence of a 100-fold excess of cold ligand.

CytoD Rapidly Disrupts the Actin Cytoskeleton, but Has Little Effect on Monolayer
Integrity-Our goal was to determine the effect of the actin-disrupting agent cytoD on postendocytic traffic in MDCK cells. CytoD is one member of a large group of cytochalasins (Ͼ20 members), which has been shown to cap the barbed or rapidly growing ends of actin filaments, nucleate actin polymerization, and shorten actin filaments (33)(34)(35). The overall effect of cytoD is to disrupt the actin cytoskeleton by creating short filaments and aggregates that are either nonfunctional or functional in a reduced capacity (33). Unlike nocodazole, which depolymerizes microtubules, cytoD does not dramatically alter the total F-actin (polymerized actin) content of the cell (33,39), a result we have confirmed in MDCK cells (see "Experimental Procedures").
First, we assessed how rapidly cytoD altered the actin filament architecture of the MDCK cells, and what effect this treatment had on the integrity of the cell monolayer. The distribution of filamentous actin in untreated cells, as assessed by FITC-phalloidin staining, is shown in Fig. 1 (A-D). Phalloidin is a small molecule that binds and stabilizes actin filaments (33). These series of panels are optical sections obtained with a scanning laser confocal microscope. A fine punctate staining, characteristic of the actin-based core of the microvilli, was observed at the apex of the cell (Fig. 1A). Filamentous actin was also observed at the margins of each cell both at the apical pole of the cell above the nucleus (Fig. 1B) and along the lateral membranes (Fig. 1C). At the base of the cell were a series of parallel bundles of actin stress fibers (Fig. 1D). These results are consistent with previous characterizations of the actin cytoskeleton in MDCK cells (41).
The distribution of actin was markedly altered when cells were treated with cytoD. Panels E-H of Fig. 1 are optical sections through a portion of the monolayer treated 5 min at 37°C with 25 g/ml cytoD. The cells in this series of sections had different heights (a characteristic of MDCK monolayers) so that the cells in the upper right of panels E-G were taller than those in the lower left of each of these panels. As a result, microvillar staining was observed in the cells in the lower left of panel E, and focal accumulations of actin were seen in the cells in the upper right of this panel. These focal accumulations of actin were more apparent at the apical pole of the cell (Fig.  1F), and along the lateral surfaces of the cell (Fig. 1G). At the base of the cell, the stress fibers had the appearance of having contracted along tracts. This may reflect the loss of membrane attachments. Focal accumulations of actin in cytoD-treated cells have been described previously (33,34,41). Similar changes were observed following cytoD treatment for 60 min at 37°C (Fig. 1, I-L). Focal accumulations of actin were observed apically (panels I and J) and laterally (panel K) and were more pronounced than observed after 5 min of cytoD treatment. In addition, the foci at the base of the cell appeared more ran-  Filter-grown MDCK cells were rapidly cooled to 4°C, and left untreated (control, 1 and 25 g/ml cytoD) or treated with nocodazole (Noc and NocϩcytoD) for 60 min at 4°C. The cells were then incubated at 37°C for up to 3 h in the absence of drugs (control), in the presence of 1 or 25 g/ml of cytoD (cytoD), in the presence of nocodazole and 25 g/ml cyto D (NocϩcytoD), or in the presence of nocodazole alone (Noc). At the indicated times trans-epithelial resistance was measured. The values are mean Ϯ S.D. (n ϭ 3) and are representative of three separate experiments. domly distributed than after the 5-min treatment (panel L).
There were no additional changes in the actin cytoskeleton when cytoD treatment was extended to 2 h. The general archi-tecture of the monolayer (cell height, shape, size, and cell-cell adhesion) as well as tight junctional morphology (see below) was unaffected by cytoD treatment. Finally, similar observations were made when an anti-actin antibody (which recognizes both filamentous and monomeric actin) was used instead of FITC-phalloidin (data not shown). One side effect of cytoD treatment is that transepithelial resistance across the monolayer can be altered (41,42). However, this effect was concentration dependent. After a 3-h treatment with 1 g/ml cytoD, transepithelial resistance across the monolayer was reduced by approximately 50%, while at higher concentrations (25 g/ml) there was little effect on transepithelial resistance (Table I). These observations are similar to those reported previously (41). To measure protein permeability across the monolayer, 125 I-BSA was added to the apical chamber of control or cytoD-treated cells and the amount of 125 I-BSA that "leaked" across the cell (via paracellular diffusion or apical to basal transcytosis) was assessed. In either untreated cells and those treated with 25 g/ml cytoD, Յ0.2% of the 125 I-BSA was found in the basal chamber during a 2.5-h incubation at 37°C. In cells treated with 1 g/ml cytoD, approximately 0.5% of the 125 I-BSA leaked into the basal chamber. These data suggest that the protein permeability of monolayers treated with 25 g/ml cytoD was largely intact and unaffected by drug treatment. In the following experiments, we have chosen to use the higher concentration of cytoD (25 g/ml) as it rapidly altered the actin cytoskeleton without altering the integrity of the MDCK cell monolayer, and this concentration of drug has previously been used in other studies where the effect of this drug has been studied in MDCK cells (23,25).
Like Nocodazole, CytoD Alters the Postendocytotic Traffic of IgA in pIgR-expressing MDCK Cells-We next determined the effect of cytoD pretreatment on the postendocytic traffic of 125 I-IgA in polarized polymeric immunoglobulin receptor (pIgR)-expressing MDCK cells. Newly synthesized pIgR is delivered directly to the basolateral cell surface (43), where it binds its ligand. The receptor-ligand complex is then endocytosed and delivered to peripheral basolateral early endosomes. From this compartment, the pIgR-ligand complex is translocated in a microtubule-dependent step to an apical endosomal compartment, which has been termed the apical recycling endosome (ARE) (31,44). This tubular vesicular compartment receives proteins recycling from both plasma membrane domains as well as transcytosing proteins (31,32,44). Upon exit from the ARE, the pIgR is cleaved by a proteinase at, or in transit to, the apical plasma membrane, releasing the pIgR's large extracellular domain (with bound ligand) into the apical medium. This cleaved form of the receptor is known as the secretory component.
In these experiments, 125 I-IgA is internalized basolaterally for 10 min at 37°C and the cells washed over a 5-min period of time to allow nonspecifically bound ligand to dissociate from the filter and cell surface, and for receptor bound ligand to be internalized. Under these internalization conditions, the majority of ligand resides in the ARE (31,45). Following this internalization and wash protocol, less than 5% of total cell-  associated 125 I-IgA was present at the cell surface (as determined by acid-stripping at 4°C) of control or cytoD-pretreated cells. In a subsequent 2-h chase, the amount of pre-internalized ligand released apically (transcytosed), released basolaterally (recycled), degraded, or remaining cell associated was quantified. Under control conditions approximately 80% of the 125 I-IgA was transcytosed ( Fig. 2A). However, in cells pretreated with cytoD there was a marked inhibition of transcytosis. A 2-2.5-fold decrease in transcytosis was observed at the earliest time points, and at the end of the assay there was approximately a 45% decrease in transcytosis ( Fig. 2A). The fraction of ligand released basolaterally (recycled) in these assays is shown in Fig. 2B, and demonstrates that cytoD caused a 2-fold stimulation of recycling over that observed in control cells. A surprising effect of cytoD treatment was that the amount of ligand degraded and released extracellularly (trichloroacetic acid-soluble fraction) increased over time (Fig.  2C). In the first 30 min of the assay, during which time the majority of transcytosis was observed, there was no increase in degradation. However, by 2 h, the total amount of ligand degraded increased approximately 4-fold over control values. We do not know whether this represents IgA being misdirected to lysosomes, or if it reflects missorting of degradative enzymes to the endosomal system. The former possibility may be less likely, as it has been shown recently that degradation of ␣2 macroglobulin in BWTG3 cells (a mouse hepatoma cell line) is inhibited by cytoD (46). Alternatively, if IgA remains in the early endosomes for extended periods of time, it may be more susceptible to proteinases that are present in this compartment.
The effects of cytoD on transcytosis and recycling were remarkably similar to those observed in cells treated with nocodazole (decreased transcytosis coupled with increased recycling) (6, 15). However, there was no difference in the microtubule distribution between control and cytoD-treated cells (as assessed by immunofluorescence; data not shown), and cytoD did not increase the amount of monomeric tubulin in the cell (see "Experimental Procedures"). These results suggest that the effect of cytoD was not the result of alterations in the microtubular network. In addition, the cytoD-mediated inhibition of transcytosis was not due to cytotoxicity, as the effect of a 2-h treatment with cytoD was reversible following a 2-h wash out period (Fig. 2D). The reversibility of cytoD is consistent with past observations (24,41,47). In summary, cytoD had profound effects on IgA postendocytic fate: decreasing basolateral to apical transcytosis, and increasing IgA recycling and degradation. The effects of cytoD were rapid, specific for the actin cytoskeleton, and reversible.
Like Nocodazole, CytoD Treatment Partially Inhibits Apical Recycling, but Has No Effect on Basolateral Recycling-We have also analyzed the effect of cytoD pretreatment on the basolateral and apical recycling pathways. CytoD had essentially no effect on the kinetics of Tf recycling from the basolateral cell surface (Fig. 3A). There was a relatively small increase in degradation of 125 I-Tf in cytoD-treated cells (6.8% in cytoDtreated cells versus 3.7% in control cells). Since there is no known endogenous marker of the apical recycling pathway, we have measured the apical recycling of IgA internalized from the apical pole of the cells as described previously (31,40). A small fraction of the pIgR escapes cleavage at the apical cell surface and is capable of binding and endocytosing ligand (40). Because apical endocytosis is impaired in cytoD-treated cells (including that of IgA, data not shown) (23), we removed cell surface ligand by proteinase treatment to ensure we only measured the fate of a preinternalized cohort of ligand. As shown in Fig. 3B, the majority of ligand in control cells was recycled apically and little was transcytosed in the apical-basolateral direction. Cy-toD inhibited apical recycling by about 20%. In addition, there was a small increase in apical to basolateral transcytosis. These effects were very similar to those observed in cells treated with nocodazole (6,15).
CytoD, like Nocodazole, Inhibits Transit between the Basolateral Early Endosomes and the ARE-Next, experiments were designed to determine at which step cytoD inhibited transcytosis. As described above, the first step in transcytosis is endocytosis of pIgR-ligand complexes, followed by delivery to peripheral basolateral endosomes. CytoD treatment had only a small effect on basolateral endocytosis (Table II). After 1 min there was approximately a 25% inhibition, but by 5 min the inhibition was reduced to 10%. This observation is consistent with previous reports that basolateral endocytosis is largely unaffected by cytoD treatment (23,24). In addition, the observed inhibition was not a result of decreased binding sites as cytoD-treated cells internalized more ligand than control cells (data not shown); this is consistent with the increased recycling and decreased transcytosis of pIgR ligand observed in Fig. 2.
The final step of transcytosis is the exit of pIgR-IgA complexes from the ARE into the apical secretions (31,44). To determine if exit of IgA from the ARE was affected by actin disruption, the following experiment was performed. 125 I-IgA was internalized under the standard conditions (10-min pulse followed by a 5-min wash). In untreated control cells, the majority of ligand (approximately 80%) is present in the ARE following this internalization protocol (45). CytoD was then included during the subsequent 2-h chase, and 125 I-IgA release into the apical medium was assessed. CytoD inhibited 125 I-IgA transcytosis by 30% at the earliest time point (7.5 min); however, the overall effect on ligand transcytosis was minimal (Fig. 4). There was a small increase in recycling, but little effect on degradation, or the amount of cell-associated ligand remaining at the end of the experiment. These data suggested that cytoD may have been acting on a step prior to exit of ligand from the ARE. We have previously published a density-shift assay to quantify delivery of IgA and pIgR from the basolateral early endosomes to the ARE (the step that precedes ARE exit) (31,45). However, we have yet to define reproducible assay conditions that allow for us to use this assay, as it requires the apical internalization of a horseradish peroxidase-conjugated marker of the apical recycling pathway. As described above, apical endocytosis is significantly impaired in cytoD-treated cells (23). Instead, we have performed a morphological analysis of IgA transport from the basolateral early endosomes to ARE in control and cytoD-pretreated cells (Fig. 5, A-F). Following a 10-min pulse and 5-min wash, basolaterally internalized IgA was predominantly found at the apical pole of control cells in a fine vesicular compartment previously characterized as the ARE (Fig. 5A) (31). The thin bright lines surrounding each cell is simultaneous staining for the tight junction associated pro-tein ZO-1. There was faint staining of IgA along the lateral (Fig. 5B) and basal (Fig. 5C) surface of the cell in small punctate vesicles, which have previously been characterized as basolateral early endosomes (31).
In contrast, when cells were pretreated with cytoD, there was less accumulation of IgA in the ARE (Fig. 5D), and a corresponding increase in lateral (Fig. 5E) and basal (Fig. 5F) staining. The basolateral staining observed was not a result of cytoD-mediated ARE redistribution, as IgA internalized into the ARE and subsequently treated with cytoD remained at the apical pole of the cell (in the ARE) and was not found basolaterally (data not shown). The tight junction staining of ZO-1 was unaltered by cytoD treatment, consistent with our data demonstrating that 25 g/ml cytoD has little effect on monolayer permeability. There were patches of cytoD-treated cells with a greater accumulation of apical ligand, but these cells also had a corresponding increase in basolateral staining. When control cells were chased for 60 min at 37°C in ligand-free medium (following the ligand pulse and wash), there was little IgA left in the cell. However, in cytoDtreated cells IgA staining in the basolateral endosomes persisted (data not shown). The above observations suggested that the action of cytoD was to prevent the efficient translocation of IgA between basolateral early endosomes and the ARE. In this respect, the effect of cytoD was similar to nocodazole, which also prevents transit of macromolecules between the basolateral early endosomes and the ARE (31,45). However, the effect of nocodazole is more pronounced as much less IgA accumulates in the ARE of nocodazole-treated cells.
The transport of Tf (a marker of the basolateral recycling pathway) from the basolateral early endosomes to the ARE was similarly inhibited by cytoD treatment (data not shown). We note that recycling of Tf does not require transit through the ARE. In nocodazole-treated cells (in which delivery to the ARE is prevented) (22), Tf recycling occurs with almost identical kinetics to untreated cells (data not shown). The above experiments indicate that a major effect of cytoD treatment was to prevent the efficient translocation of membrane proteins from the basolateral early endosomes to the ARE.
CytoD Arrests Traffic at a Step prior to the Nocodazole-sensitive Step-Our results suggested that cytoD, like nocodazole, inhibited traffic between the basolateral early endosomes and the ARE. To determine whether cytoD and nocodazole arrested IgA transcytosis at discrete steps early in the transcytotic pathway, we have performed order-of-addition experiments. In these experiments, ligand was internalized in the presence of either cytoD or nocodazole to accumulate ligand at the basolateral pole of the cell. The cells were then washed, and the ability of the other drug to inhibit transcytosis in a subsequent reaction was assessed. First, IgA was internalized in cytoD-pretreated cells using the standard internalization protocol. Under these conditions, some of the IgA had escaped the basolateral endosomes and was already in the ARE as shown in Fig. 5. This fraction of ligand is expected to be nocodazole-insensitive (as described above; see also Fig. 8). During a subsequent 2-h chase at 37°C, the cells were either left untreated, incubated in the continued presence of cytoD, or treated with nocodazole. As shown in Fig.  6A, the effect of cytoD was largely irreversible, and similar to cells incubated in the continued presence of cytoD. (In Fig. 2D, we have demonstrated that the effect of cytoD was reversible after a 2-h chase in drug-free media; however, in this experimental protocol, there was presumably not enough time for the actin-cytoskeleton to recover from cytoD treatment). Compared to the levels of transcytosis observed in cells incubated in cytoD during the 2-h chase, nocodazole inhibited subsequent transcytosis by approximately 30% (Fig. 6A). The result of these treatments on ligand recycling are shown in Fig. 6B. The above observations suggest that at least a fraction of ligand in cytoDtreated cells was blocked at a step that preceded the nocodazole-sensitive step.
The reverse experiment was also performed. 125 I-IgA was internalized in nocodazole-treated cells, the cells were washed, and then treated with nocodazole, or nocodazole and cytoD for an additional 15 min at 37°C. This step was included to insure that the cytoD would have time to act. The subsequent release of ligand was measured in a 2-h chase in the absence of any drug, in the presence of cytoD alone, or in the continued presence of nocodazole. The effect of nocodazole was reversible (Fig.  6C, no addition) when compared to reactions in which nocodazole was included in the 2-h chase. In contrast to the effect seen in Fig. 6A, cytoD had little inhibitory effect on ligand transcy-tosis compared to that observed when nocodazole was included in the 2-h chase. The effects of these treatments on basolateral recycling are shown in Fig. 6D. The above observations are consistent with the cytoD-sensitive step preceding the microtubule dependent step.
Both Microtubules and Actin Filaments Are Required for Efficient Transcytosis-Our data suggested that in addition to microtubules, efficient transcytosis also requires actin filaments. It was also plausible that these two systems may be functionally redundant so that in the absence of one the other could function in its place, or at a reduced level of efficiency. This might have explained why neither nocodazole, nor cytoD completely inhibited transcytosis (Fig. 7A). In fact, in cells treated with both nocodazole and cytoD prior to ligand internalization, 125 I-IgA transcytosis was almost completely inhibited (Fig. 7A), while recycling was correspondingly increased (Fig. 7B). This inhibition was not the result of increased monolayer permeability or cytotoxicity as trans-epithelial resistance (Table I) cell morphology were unaffected by this treatment (data not shown). In addition, the effect of this combination of drugs on transcytosis was reversible, had a relatively modest effect on Tf recycling, and did not inhibit apical recycling over the inhibition observed with cytoD or nocodazole alone (data not shown). As expected, IgA accumulated in the basolateral endosomes in cells treated with this drug combination (data not shown).
Once ligand has accumulated into the ARE, its subsequent release is largely insensitive to the action of either cytoD and nocodazole (Fig. 8). However, in cells in which the combination of cytoD and nocodazole was added after 125 I-IgA had been preinternalized into the ARE, there was a 50% reduction in exit of ligand from this compartment into apical secretions (Fig. 8). As described above, one possible explanation is that in the absence of either microtubules or actin filaments the other system is functionally redundant; however, when both systems are impaired, exit of ligand from the IgA to the apical plasma membrane is inhibited. That inhibition was not complete may reflect that some vesicle transport was past the step where cytoskeletal interaction was necessary.

DISCUSSION
Although the actin cytoskeleton forms many specialized structures in epithelial cells, its role in postendocytic traffic of these cells is largely unexplored. Actin cytoskeletal rearrangements have been noted in gastric parietal cells in response to secretion (48). However, the relationship between these two phenomenon is unknown. Apical endocytosis is selectively inhibited by cytoD treatment of MDCK cells and Caco-2 cells, whereas basal endocytosis is largely unaltered (23)(24)(25). In addition, uptake of ricin and fluid-phase markers by Vero cells is inhibited by cytoD treatment, while Tf internalization is unaffected (47). In non-polarized BWTG3 mouse hepatoma cells, cytoD inhibits ␣-2 macroglobulin degradation, and recycling of Tf from the recycling endosome to the plasma membrane of these cells (46).
In the present analysis, we have found that cytoD, a drug that selectively disrupts the actin cytoskeleton (33), has profound effects on postendocytic sorting in MDCK cells. The most notable effect is the inhibition (approximately 45% after 2 h) of basolateral to apical transcytosis with a corresponding increase in ligand recycling. The increased recycling may be a result of an increased residence time of IgA in basolateral early endosomes. In fact, a morphological analysis suggests that transcytosing IgA accumulates in the basolateral early endosomes, and less is delivered to the ARE of cytoD-treated cells. Once delivered to the ARE, there is little effect on subsequent release of ligand in cells treated with cytoD. The effect of cytoD on transcytosis is not due to cytotoxicity or a nonspecific affect on basolateral endosome function as the effects were reversible, and the basolateral recycling pathway, marked by Tf, is unaffected by cytoD treatment.
The cytoD inhibition of transcytosis is not absolute and may reflect the fact that even in cytoD-treated cells polymerized actin (present as short filaments and aggregates) is still be found in the cell (33). These short filaments and aggregates may still retain some functionality. An alternative, but not mutually exclusive possibility is that the unaffected microtubule network remains intact and may functionally substitute for the actin cytoskeleton to some degree. In support of this hypothesis is the observation that neither nocodazole (60 -70% inhibition) nor cytoD (45% inhibition) completely inhibit transcytosis (this work and Refs. 6 and 15), while the combination of both drugs acts to inhibit transcytosis almost completely (Ͼ95% inhibition). A cooperative interaction between microtubules and the actin cytoskeleton is also suggested in yeast cells with a mutation in their MYO2 gene (which encodes an unconventional myosin). These cells do not form buds, have a disrupted actin cytoskeleton, and exhibit a build-up of cytoplasmic vesicles (49). Interestingly, overexpression of the kinesin-like microtubule motor, Smy1p, suppresses this defect (50). These Values for control and cytoD are from Fig. 2 and 4, respectively, and are included for the purpose of making comparisons. Recycling values were: control, 9.5 Ϯ 1.9%; cytoD, 13.9 Ϯ 4.1%; nocodazole, 8.6 Ϯ 0.7%; nocodazole ϩ cytoD, 16.5 Ϯ 0.2%. Values for degradation were: control, 5.8 Ϯ 0.7%; cytoD, 7.7 Ϯ 0.6%; nocodazole, 6.9 Ϯ 2.0%; nocodazole ϩ cytoD, 3.5 Ϯ 0.8%. The remainder of the counts were intracellular. Values are mean Ϯ S.D. (n ϭ 3) and are representative of two separate experiments.
observations again suggest that one cytoskeletal system can operate in place of the other.
While the exact role of actin in basolateral to apical transport along the endocytic pathway of MDCK cells is unknown, our data suggest that the actin cytoskeleton, presumably in conjunction with a myosin motor complex, may be responsible for both generating transport vesicles from the basolateral early endosomes and/or delivering them through the cortical actin network. In fact, recent evidence suggests that the endosomes of MDCK cells are surrounded by actin and contain the unconventional myosin motor myr4. 2 Moreover, membrane interactions between vesicles derived from apical or basolateral early endosomes and transferrin-containing endosomes are facilitated by actin filaments and depend on the unconventional myosin myr4 and calmodulin. 2 Actin, in conjunction with an unconventional myosin motor, may help to form pIgR-containing transport vesicles via a contractile ring (similar to that seen during cytokinesis), or by exerting tension between actin filaments and the endosomal membrane to generate long tubular processes (as proposed in Refs. 46 and 51). These tubules are commonly seen emanating from the vesicular portion of an endosome (52) and may be capable of or required for transport.
An alternative but not mutually exclusive possibility is that delivery of endosomal vesicles may require actin-activated mechanochemical motors, in addition to microtubule motors, as proposed for trans-Golgi network-derived vesicles (28,29). In this case, an unconventional myosin motor (possibly myr4) would be responsible for transporting the endosomal vesicles through the cortical actin cytoskeleton underlying the basolateral plasma membrane. Once past this step, transport vesicles would require a microtubule motor protein such as cytoplasmic dynein, which would allow them to interact with the microtubular network. The order-of-addition experiments described in this manuscript are consistent with a model requiring the sequential transport of ligand-occupied pIgR between actin and microtubule cytoskeleton. The vesicles, thus attached, would be directed toward the apical pole of the cell where they would fuse with the ARE. Transcytotic vesicles forming from the ARE may also contain actin-and microtubule-based motor proteins. As demonstrated, exit of IgA from the ARE into apical secretions was inhibited when both microtubules and the actin cytoskeleton were disrupted. Additional work will be required to assess if the tubular/vesicular elements of the ARE contain motor proteins.
Finally, recent evidence suggests that IgA transport is a regulated process (reviewed in Ref. 53). In response to IgA binding to the pIgR, intracellular calcium levels rise, inositol triphosphate turnover is stimulated, and protein kinase C becomes activated (54). It is also known that wortmannin, a phosphatidylinositol 3-kinase inhibitor, blocks IgA transcytosis at an early step in the transcytotic pathway (55). Interestingly, many actin-binding proteins, including members of the myosin I family, are themselves regulated by changes in calcium, phospholipids, and phosphorylation (2,26), and, as such, the actin cytoskeleton may be an important player in signal transduction (2). We have found that while transcytosis of ligand occupied pIgR (i.e. pIgR with bound IgA) is selectively inhibited by cytoD, the kinetics of transcytosis of ligand unoccupied pIgR may actually be stimulated by cytoD. 3 By stimulating secondary messenger pathways, IgA is apparently modulating its own vesicle's interaction with the cytoskeleton, and the ultimate rate of its own transcytosis.