Regulation of Vesicle Trafficking in Madin-Darby Canine Kidney Cells by Rab11a and Rab25*

Polarized epithelial cells maintain the polarized distribution of basolateral and apical membrane proteins through a process of receptor-mediated endocytosis, sorting, and then recycling to the appropriate membrane domain. We have previously shown that the small GTP-binding proteins, Rab11a and Rab25, are associated with the apical recycling system of Madin-Darby canine kidney cells. Here we have utilized inducible expression of wild-type, dominant negative, and constitutively active mutants to directly compare the functions of Rab25 and Rab11a in postendocytic vesicular transport. We found that a Rab11a mutant deficient in GTP binding, Rab11aS25N, potently inhibited both transcytosis and apical recycling yet failed to inhibit transferrin recycling. Similarly, expression of either wild type Rab25 or the active mutant Rab25S21V inhibited both apical recycling and transcytosis of IgA by greater than 50% but had no effect on basolateral recycling of transferrin. Interestingly, the GTPase-deficient mutant Rab11aS20V inhibited basolateral to apical transcytosis of IgA, but had no effect on either apical or basolateral recycling. These results indicate that neither Rab11a nor Rab25 function in the basolateral recycling of transferrin in polarized

Recycling receptors (e.g. transferrin receptor) subsequently enter a tubular and pericentriolar recycling endosome, where they are repackaged and transferred back to the plasma membrane (2).
The endocytic pathway of polarized epithelial cells exhibits an additional level of complexity because these cells maintain distinct apical and basolateral membrane domains (3)(4)(5). Studies in polarized MDCK 1 cells have shown that fluid-phase markers internalized from the apical or basolateral surface of the cell accumulate in distinct apical or basolateral early endosomes before accumulating in common late endosomes (6,7). However, electron microscopy studies indicate that transferrin internalized from the basolateral side of the Caco-2 cell is found in an apical endosomal compartment that is also accessible to apical endocytic markers (8,9). In polarized MDCK cells, polymeric IgA receptor (pIgR) internalized from the basolateral surface is first delivered to basolateral sorting endosomes and then transported to apical tubular endosomal compartments before being delivered to the apical plasma membrane, and these endosomes are also accessible to apically internalized IgA (10 -13). These results suggest the existence of apical recycling endosomes that receive cargoes from both apical and basolateral early endosomes. It is believed that recycling receptors and transcytosing markers are sorted in these apical recycling endosomes, but the exact nature of these compartments and the way(s) in which they interact with other partners in the endocytic pathway remain obscure.
Interestingly, the apical recycling endosome in polarized cells shares similar features with the recycling endosome in nonpolarized cells: 1) it has a tubulovesicular morphology; 2) it is located close to the centrosome; 3) it is dependent on the integrity of microtubules; and 4) it is accessible to membranebound but not fluid-phase endocytic markers (10,14). These similarities have suggested that the apical recycling endosome is the counterpart of the recycling endosome of nonpolarized cells. It is possible that polarized cells would not require additional endosomal compartments for the polarized sorting of membrane constituents. Instead, the recycling endosome could be modified to function in polarized cells through the expression of polarized cell-specific regulatory proteins (15).
Rab proteins are Ras-like small GTPases, which are involved in regulating various aspects of the membrane trafficking process. More than 50 Rabs are present in distinct membrane compartments in mammalian cells. Cycles of GTP binding and hydrolysis of these proteins are linked to the recruitment of * This work was supported by Grants DK48370 and DK43405 from NIDDK, National Institutes of Health (to J. R. G.), a Veterans Administration Merit Award, a National Research Service Award postdoctoral fellowship (to X. W.), Grants AI32991 and DK33506 from the National Institutes of Health, and the Good Samaritan Foundation (to J. E. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ These authors contributed equally to this work. specific effectors on cellular membranes. The active GTP-bound form is then thought to regulate membrane vesicle docking with target membrane surfaces (for review see Refs. 16 -18). A subset of Rab proteins participates in the endocytic pathway. For example, Rab5 regulates transport from plasma membrane to early endosome and homotypic fusion between early endosomes (19); Rab7 functions in transport from early endosome to late endosome and lysosome (20,21); and Rab11a modulates transport through recycling endosomes (22). Many Rab proteins in the endocytic pathway are expressed ubiquitously. However, Rab17 and Rab25 are only found in polarized epithelial cells. Rab17 was identified in the epithelial cells of kidney and lung during cell polarization (23). In polarized Eph4 cells, Rab17 was localized to the apical endosomal membranes, and Rab17 mutants that were defective in either GTP binding or GTP hydrolysis stimulated basolateral to apical transcytosis and apical recycling (24). In an MDCK cell line expressing the transcytotic pIgR, Rab17 was found in an apical tubular endosome structure that was accessible to dimeric IgA internalized from the apical or basolateral cell surface. In contrast with the results of Zacchi, et al. (24), overexpression of Rab17 in this cell line inhibited basolateral to apical transcytosis (25). Although its precise function is unclear, these results imply that Rab17 regulates membrane trafficking through the apical recycling endosome in polarized epithelial cells. Rab25 was cloned from a gastric parietal cell cDNA library and is expressed only in epithelial tissues such as the gastrointestinal mucosae, kidney, and lung (26). Recently, we have studied Rab25 in an MDCK cell line that stably expressed the pIgR (14). We found that Rab25 was located in an apical pericentriolar endosomal compartment that was dependent on intact microtubules for its integrity. Moreover, this compartment was accessible to IgA internalized from either the apical or basolateral side of the cell surface. Interestingly, a closely related Rab, Rab11a, colocalized with Rab25 in this subapical compartment. Because Rab11a is known to function in the recycling endosome to regulate membrane trafficking in nonpolarized cells, it is possible that the ubiquitous role of Rab11a in plasma membrane recycling applies to polarized MDCK cells. Introduction of Rab25 in polarized cells such as MDCK may provide an additional check point in the trafficking pathway. Our observation that overexpression of Rab25 inhibits IgA transcytosis and apical recycling supports this view (14).
Whereas our previous studies suggested that Rab11a and Rab25 both were associated with the apical recycling system in MDCK cells, the relative contributions of Rab11a and Rab25 to plasma membrane recycling in polarized MDCK cells have not been studied. In this study we inducibly expressed Rab11a and Rab25 along with their corresponding constitutively active and dominant negative mutants and examined their effects on basolateral recycling of transferrin as well as transcytosis and apical recycling of polymeric IgA. We demonstrate that mutants of Rab11a deficient in GTP binding fail to inhibit basolateral recycling of transferrin, yet have pronounced effects on apical recycling and transcytosis of polymeric IgA. Somewhat surprisingly, the GTPase-deficient form of Rab11a also inhibited transcytosis but had no effect on apical or basolateral recycling. Wild type Rab25 and GTPase-deficient Rab25 behaved similarly, with both potently inhibiting apical recycling and transcytosis, without affecting transferrin recycling. These studies demonstrate that both Rab11a and Rab25 function at a late stage in the transcytotic and apical recycling pathways, beyond the point at which transferrin has been sorted away from IgA. This pattern of sorting is clearly different from the situation in nonpolarized cells in which Rab11a is present on transferrin-containing compartments and regulates trans-ferrin recycling.

EXPERIMENTAL PROCEDURES
Materials-Monoclonal antibodies against Rab11a (8H10) and Rab25 (12C3) were prepared as described previously (27,28). Rabbit polyclonal antibodies against Rab11a were purchased from Zymed Laboratories Inc. Secondary antibodies conjugated to Cy2, Cy3, and Cy5 were purchased from Jackson Immunochemicals. Prolong antifade mounting medium was obtained from Molecular Probes. All culture media were purchased from Life Technologies, Inc. G418 and hygromycin were purchased from Calbiochem. Human dimeric IgA was purified from the serum of a patient with excessive IgA production, as described previously (29). Oligonulceotides were synthesized and purified by the IMMAG Molecular Biology Core Facility.
Vector Construction-The rabbit Rab11a wild type sequence in pET19b was mutated to either Rab11aS25N or Rab11aS20V, and Rab25 wild type in pGEX-2T was mutated to Rab25S21V with single nucleotide site-directed mutations using a two complementary oligonucleotide method with Pfu polymerase (Stratagene). Rabbit Rab11a constructs (wild type, S25N, and S20V) and rabbit Rab25S21V were then cloned into the tetracycline regulatable eukaryotic expression vector pTRE (CLONTECH). All vector sequences were confirmed by automated DNA sequencing in the IMMAG Molecular Biology Core Facility. The adenovirus containing a tetracycline repressible expression vector for Rab11aN124I was a gift of Dr. Hsiao-Ping Moore (Univ. California, Berkeley) and has been detailed previously (30). Construction and propagation of adenoviral vectors for Rab25 wild type and Rab25T26N have been detailed previously (14).

Analysis of Recombinant Rab11a and Rab25
Mutants-Histidinetagged Rab11a and Rab11aS20V and glutathione S-transferase-tagged Rab25 and Rab25S21V were expressed and purified as described previously (14,26,27). GTP binding was assessed as described previously (14). Intrinsic GTPase and GTPase activity in the presence of cytosol as a donor of GTPase-activating protein activity were assayed as previously published (14).
Establishment of Tetracycline Repressible Cell Lines-T23 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Stable transfections of T23 cells were performed with the LipofectAMINE PLUS reagent (Life Technologies, Inc.) in the presence of 20 ng/ml doxycycline. 4 g of supercoiled Rab11a or Rab25 plasmid and 0.2 g of pCB7 plasmid (hygromycin-resistant) were used for each co-transfection. Clones were selected with 300 g/ml hygromycin and in the presence of 20 ng/ml doxycycline. Hygromycinresistant clones were further selected for protein expression 24 h after the removal of doxycycline. Positive clones were maintained in culture medium with 300 g/ml hygromycin and 20 ng/ml doxycycline.
Assessment of Protein Expression in Transfected Cells-To induce recombinant protein expression in stable cell lines, cells were incubated with fresh phosphate-buffered saline (5 min each for 3 times) and then placed in Dulbecco's modified Eagle's medium containing 0.2 ng/ml doxycycline. Experiments were performed 24 h after the induction. For viral infections, cells were infected at a ratio of 10 plaque-forming units/cell in the presence of 0.2 ng/ml doxycycline, and experiments with viral infection were performed 16 h after the infection.
Immunocytochemistry-Cells fixed on permeable filters were stained in the culture well inserts. Cells were permeabilized and blocked with 5% goat serum, 0.3% Triton X-100 in phosphate-buffered saline for 30 min at 4°C and then incubated with primary antibodies as appropriate for 2 h at room temperature. Primary antibodies were utilized at the following dilutions: murine monoclonal Rab25 ascites (12C3), 1:50; rabbit polyclonal anti-Rab11a, 1:200; rat monoclonal anti-ZO-1, 1:300; murine monoclonal anti-Rab11a ascites (8H10), 1:100. Following washing in phosphate-buffered saline, cells were then incubated with fluorochrome-conjugated secondary antibodies for 30 min. In cells loaded with dimeric IgA, primary anti-IgA antibodies directly conjugated with fluorescein isothiocyanate were used at 1:100 dilution. For double and triple labeling studies, all primary and secondary antibodies were incubated together with the cells. Following final washing, cells were mounted in Prolong antifade solution (Molecular Probes) and examined with scanning confocal fluorescence microscopy in the Core Imaging Facility of the Institute of Molecular Medicine (Molecular Dynamics, Sunnyvale, CA). For triple labeling studies, section series (thirty 0.3-m optical sections) were performed twice with dual imaging with 488/647 nm excitation laser lines to visualize Cy2/fluorescein isothiocyanate and Cy5 fluorochromes followed by reimaging with 568 nm of laser excitation to visualize Cy3. Section series were rendered in look through projections using Imagespace software (Molecular Dynamics) on a Silicon Graphics Indy workstation. For localization of IgA, cells were incubated with purified human dimeric IgA (10 g/ml) in either the apical or basolateral medium for 30 min at 37°C in Hanks' buffered saline containing calcium, magnesium, and 0.2% bovine serum albumin (Hanks' bovine serum albumin), washed three times quickly in the same medium, and fixed for confocal microscopy as described above.

RESULTS
GTPase-deficient Mutants of Rab11a and Rab25-Most previous studies of Rab protein function, including those involving Rab11a, have utilized WDTAGQE to WDTAGLE mutations in the P3 GTP binding motif to generate putative dominant active proteins (31,32). However, we have previously shown that Rab11aQ70L exhibits normal GAP-stimulated GTPase activity (14). Additionally, Rab25, which has WDTAGLE as its wild type sequence, is also an active GTPase (26). We therefore sought to construct other mutations in Rab11a and Rab25, which would inhibit their GTPase activities. Fig. 1 demonstrates that Rab11aS20V and Rab25S21V both display the characteristics of dominant active proteins. Both display specific binding of GTP, albeit at a somewhat reduced capacity (Fig. 1, A and B). However, neither mutant shows any significant intrinsic GTPase activity and the mutations abolish GAPstimulated GTPase activity (as assayed in the presence of gastric cytosol, Fig. 1, C and D).
Expression of Rab11a and Rab25 Mutants in MDCK Cells-To compare the roles of Rab11a and Rab25 in MDCK cells, we prepared stably transfected cell lines overexpressing Rab11a wild type, the dominant negative Rab11aS25N mutant, Rab11aS20V, and Rab25S21V under the control of a tetracycline repressible promoter. To facilitate the analysis of the transcytotic and apical recycling pathways, all lines were derived from MDCK cells stably expressing the pIgR (T23 cells) (33). We have previously reported the construction of recombinant adenovirus directing the expression of Rab25 wild type and Rab25T26N dominant negative (14). As we have previously reported for cells infected with the Rab25 adenovirus constructs (14), the four stable cells lines did not express the Rab11a and Rab25 constructs in the presence of tetracycline ( Fig. 2). However, lowering of tetracycline to 0.2 ng/ml elicited a prominent expression of the Rab proteins (Fig. 2). All six recombinant proteins were expressed at similar levels.
Transferrin Trafficking in MDCK Cells-To investigate the effects of Rab protein expression on basolateral recycling, we studied the trafficking of 125 I-canine transferrin in the four stable cell lines as well as in pIgR-transfected MDCK cells infected with adenovirus coding for wild type Rab25 or Rab25T26N. Cells cultured on permeable filters supports were allowed to internalize 125 I-transferrin from the basolateral medium in polarized cells for 30 min. Release of 125 I-transferrin into the apical and basal media was then followed over a 60-min time course to assess transcytosis and basolateral recycling, respectively. As shown in Fig. 3 GTPase-deficient Rab11aS20V had any effect on either basolateral recycling or transcytosis of transferrin (Fig. 3A). This result was somewhat surprising in that previous observations in nonpolarized cells (22, 31) indicated that Rab11aS25N potently inhibited transferrin recycling. To confirm these findings we also infected polarized MDCK cells with a tet-regulated adenovirus encoding a second GTP binding-deficient mutant, Rab11aN124I (30). In agreement with the findings described above, no effect on transferrin basolateral recycling was observed in cells expressing Rab11aN124I (data not shown). It should be noted, however, that unattenuated expression of either Rab11aS25N or Rab11aN124I did cause a decrease of tight junction integrity leading to some leakage of transferrin through the paracellular space (data not shown). Under standard assay conditions (0.2 ng/ml doxycycline) tight junction integrity typically remained intact as determined by measurements of transepithelial resistance. As occasional leakiness was observed even in the presence of doxycycline, only monolayers that retained transepithelial resistance similar to that of controls were used. Similarly, none of the Rab25 constructs had any effect on trafficking of transferrin (Fig. 3). Similar results were observed using a shorter (10 min) pulse of transferrin (data not shown). These data suggest that, in polarized cells, neither Rab11a nor Rab25 participate in basolateral recycling of transferrin.
Effects of Rab11a and Rab25 Mutants on Apical IgA Recy-  cling-To investigate apical membrane recycling and basolateral to apical transcytosis, we studied the transport of polymeric IgA in the transfected cell lines. To assess apical recycling, 125 I-IgA was internalized from the apical medium for 30 min, and then the appearance of 125 I-IgA in the apical and basolateral media was determined over a 30-min time course. As previously noted (14), little apical to basolateral transcytosis was observed in MDCK cells, and none of the transfected constructs altered this pattern (Fig. 4). Whereas neither Rab11a nor Rab11aS20V altered apical recycling, the dominant negative Rab11aS25N did decrease recycling by 25%. In contrast, wild type Rab25 and Rab25S21V strongly inhibited apical recycling by 35 and 50%, respectively (Fig. 4). As previously reported (14), the GTP binding mutant Rab25T26N did not alter apical recycling or apical-to-basolateral transcytosis. These data suggest that both Rab11a and Rab25 are involved in the process of apical recycling.
Effects of Rab11a and Rab25 Mutants on IgA Transcytosis-To assess basolateral to apical transcytosis, 125 I-IgA was internalized from the basal medium for 10 min, and then the appearance of 125 I-IgA in the apical and basolateral media was determined over a 60-min time course. As previously noted (14), we observed little basolateral recycling of IgA in these pIgR-transfected MDCK cells. In these assays, wild type Rab11a had no effect on transcytosis (Fig. 5). However, Rab11aS20V and Rab11aS25N both inhibited transcytosis by 25 Ϯ 1% and 29 Ϯ 2%, respectively. Expression of Rab11aN124I in virally transfected cells elicited an even more pronounced inhibition of transcytosis (data not shown). As with apical recycling, wild type Rab25 and Rab25S20V both inhibited transcytosis by 55 Ϯ 1% and 57 Ϯ 1%, respectively. As previously noted, Rab25T26N had no effect on transcytosis. Similar amounts of ligand were internalized under each condition, indicating that the endocytic rate was unaffected by any of the expressed constructs. These data suggest that both Rab11a and Rab25 are involved in the regulation of basolateral to FIG. 4. Effects of Rab11a and Rab25 constructs on apical recycling of polymeric IgA. MDCK monolayers stably expressing the pIgR and tetracycline-regulated transactivator along with constructs for wild type Rab11a, Rab11aS25N, Rab11aS20V, wild type Rab25, Rab25T26N, or Rab25S21V were cultured 24 h in the absence (circles) or presence (squares) of 20 ng/ml doxycycline. To assess apical recycling, 125 I-dimeric IgA was internalized from the apical surface for 30 min at 37°C. Cells were then cooled to 4°C and washed, and surface-bound ligand removed by incubation with trypsin (10 g/ml) for 1 h at 4°C. Trypsinization was stopped by washing with cold medium containing soybean trypsin inhibitor (50 g/ml). Filters were then placed in warm medium, and ligand was recovered from the apical (recycled, open symbols) or basolateral (transcytosed, closed symbols) medium at the times indicated. Data are the average of two determinations and are expressed as a percentage of total internalized IgA. All data represent trichloroacetic acid precipitable counts (typically 97% of internalized ligand was trichloroacetic acid precipitable) and are representative of three separate experiments.

FIG. 5. Effects of Rab11a and Rab25 constructs on transcytosis of polymeric IgA.
MDCK monolayers stably expressing the pIgR and tetracycline-regulated transactivator along with constructs for wild type Rab11a, Rab11aS25N, Rab11aS20V, wild type Rab25, Rab25T26N, or Rab25S21V were grown in the absence (circles) or presence (squares) of 20 ng/ml doxycycline. To assess transcytosis, cells were loaded with 125 I-dimeric IgA from the basolateral medium for 10 min at 37°C. Filters were then washed; fresh medium was added to apical and basolateral chambers and incubated for the time points indicated. Data are the average of two determinations and are expressed as a percentage of total internalized cpm. Apical medium (containing transcytosed ligand, filled symbols) and basolateral medium (recycled ligand, open symbols) were harvested at each time point. Data are the average of two determinations and are expressed as a percentage of total internalized IgA. All data represent trichloroacetic acid precipitable counts (typically 97% of internalized ligand was trichloroacetic acid precipitable) and are representative of three separate experiments. Effects of Rab11a and Rab25 Mutants on IgA Trafficking Visualized through Confocal Fluorescence Microscopy-To elucidate further the effects of the Rab mutations on trafficking, we studied the morphological localization of polymeric IgA in comparison with overexpressed Rab proteins in cells exposed to polymeric IgA in their basal medium. To visualize the transfected Rab11a proteins, we utilized a monoclonal antibody raised against rabbit Rab11a that does not detect endogenous canine Rab11a in 4% paraformaldehyde-fixed tissues (14,27). Fig. 6 demonstrates that there was considerable overlap of IgA staining with Rab11a in cells overexpressing wild type Rab11a protein. In contrast, in cells overexpressing Rab11S25N, whereas the transfected protein did appear to be associated with vesicles in the pericentrosomal region, the co-localization with IgA was less apparent (Fig. 5, c and d). In Rab11S20Vtransfected cells, the Rab11aS20V immunoreactivity was distributed in a more diffuse punctate pattern throughout the subapical region of the MDCK cells (Fig. 6g). Still, we did observe considerable overlap of the Rab11aS20V immunostaining with IgA (Fig. 6h).
As we have reported previously, in cells overexpressing Rab25, Rab25 immunoreactivity showed extensive overlap with both endogenous Rab11a and IgA. However, whereas Rab25T26N immunoreactivity was observed dispersed throughout the cell cytoplasm, endogenous Rab11a and IgA were distributed together in a normal pattern of pericentriolar vesicles. In cells overexpressing Rab25S21V, Rab25 immunoreactivity was observed dispersed through the subapical region and in the perijunctional region (Fig. 7g). Endogenous Rab11a staining was also dispersed in the upper portion of the cell, as was the staining for polymeric IgA. Although there were areas of overlap, it was clear that much of the Rab25S21V staining was not colocalized with endogenous Rab11a. DISCUSSION Many studies of Rab protein function have been facilitated by the use of dominant negative and dominant active mutants, which are deficient in either GTP binding or GTPase activity, respectively. Earlier investigations in nonpolarized cells have examined the effects of Rab11aQ70L as a putative dominant active Rab11a mutant (22, 31). However, we have recently found that Rab11aQ70L possesses a normal GTPase activity (14). In addition, Rab25, which contains a WDTAGLE sequence in its wild type amino acid sequence, is also an active GTPase. In contrast, Rab11aS20V and Rab25S21V demonstrate the dominant active phenotype with loss of both intrinsic and GAPstimulated GTPase. These results underline the importance of establishing the GTP binding and GTPase characteristics of Rab25T26N (d-f) or MDCK cells stably expressing both the pIgR and Rab25S21V (g-i) were cultured on permeable supports in the presence of 2 ng/ml doxycycline. Cells were loaded with dimeric IgA for 30 min at 37°C from the basal medium. Cells were triple labeled with monoclonal anti -Rab25 (a, d, and g), fluorescein isothiocyanate-conjugated anti-IgA (b, e, and h), and polyclonal anti-Rab11a (c, f, and i). Look through projections were constructed as in Fig. 6. Arrowheads indicate the positions utilized to render the XZ projections presented under each XY projection panel. In wild type Rab25-transfected cells, IgA trafficked into a vesicle compartment that was immunoreactive for both Rab11a and Rab25. In Rab25T26N-transfected cells, Rab25 immunoreactivity was observed in a perinuclear position and throughout the cytoplasm. IgA immunoreactivity codistributed with endogenous Rab11a. In Rab25S20V-transfected cells, Rab25 immunoreactivity was observed in vesicles distributed in the junctional regions of the cell. Both IgA and endogenous Rab11a were also distributed in a junctional distribution. Bar, 5 m.
Rab protein mutants as a prerequisite to studies in intact cells.
A thorough analysis of the function of some Rab proteins has been problematic because of the toxic effects of mutant Rabs in eukaryotic cells. It is therefore not surprising that we have been unable to produce stable cell lines overexpressing the GTP binding mutants for Rab11a and Rab25, Rab11aS25N and Rab25T26N, respectively. Previously, as well as in the present investigation, we have utilized an adenoviral transfection system in which Rab25T26N expression was controlled by a tetracycline repressible promoter system (14). In the present study, we have used a similar tetracycline repressible system to construct stable cell lines that inducibly express Rab11aS25N as well as the wild type Rab11a and the GTPase-deficient mutant Rab11aS20V. In previous studies, Rab11aS25N transiently transfected in nonpolarized cells (22) demonstrated a vesicular rather than a cytosolic distribution that at least partially colocalized with Golgi markers. We have found that Rab11aS25N expressed in polarized MDCK cells also partially localized in a vesicle pattern, but we did not observed significant overlap with Golgi markers (data not shown). Another GTP binding-deficient Rab11a mutant, Rab11aN124I, demonstrated cytosolic distribution without prominent vesicular localization. 2 Both dominant negative mutants potently inhibited IgA trafficking but had little effect on basolateral recycling of transferrin. Forte and colleagues (30) have recently reported that Rab11aN124I inhibited regulated translocation of the parietal cell H/K-ATPase to the apical canalicular membrane. Because Rab11a is a localized to H/K-ATPase-containing parietal cell tubulovesicles (28,34), these data are consistent with a role for Rab11a in apical recycling.
The interactions of pathways regulating and coordinating trafficking to the apical and basolateral membranes in polarized cells remain controversial. Cargoes endocytosed from the basolateral membrane are sorted in the early endocytic process to separate cargoes to be directed to the lysosomes through the late endosomal pathway (e.g. low density lipoprotein) or cargoes undergoing recycling to either the basolateral or apical membranes (e.g. transferrin receptor and the polymeric IgA receptor, respectively). Recycling cargoes appear, at least in part, to traverse into a tubulovesicular sorting compartment. Thus, in MDCK cells, Hopkins and colleagues (13) have found that basolaterally endocytosed transferrin is transported into a common tubular sorting system also containing IgA receptor internalized from either the apical or basolateral plasma membrane. These data indicate that ligands destined for basolateral recycling are mixed with those undergoing transcytosis or apical recycling. The pericentrosomal position of this tubular sorting system suggested that it was analogous to the plasma membrane recycling system in nonpolarized cells (13,35). Because the recycling endosomes in nonpolarized cells are labeled by antibodies to Rab11a, we had previously hypothesized that by analogy Rab11a would be a marker of this common tubular sorting compartment in polarized cells. However, recent data have questioned this supposition. Dunn and colleagues (12) have studied the trafficking of fluorescently labeled transferrin and IgA in polarized MDCK cells. Whereas their results support the segregation of transferrin and IgA away from the low density lipoprotein, they found that transferrin separates from IgA prior to entry into a Rab11a positive compartment. Thus, in polarized cells transferrin was not observed in endosomes labeled with antibodies to Rab11a.
In addition, we have recently found that myosin Vb is a downstream target of Rab11a (36). Expression of the carboxyl tail of myosin Vb lacking a motor domain acts as a dominant negative regulator of transport through the apical recycling system leading to accumulation of IgA but not transferrin (36). These results suggest that there are at least three critical sorting stages after endocytosis: 1) the sorting of recycling cargoes from lysosomally directed ones, 2) the trafficking of basolaterally recycling cargoes out of a common tubular sorting endosome, and 3) the movement of transcytosing and apically recycling cargoes into a more apical endosomal compartment, which contains Rab11a and Rab25 (Fig. 8).
The inability of the Rab11aS25N and Rab11aN124I mutants to inhibit transferrin recycling supports the work of Dunn and colleagues (12). We further observed that dominant active Rab11aS20V overexpression had no effect on either basolateral recycling of transferrin or apical recycling of polymeric IgA, but Rab11aS20V did significantly inhibit transcytosis of IgA. Several possibilities may explain this observation. Cycling of Rab11a may be required for maintenance of the transcytotic pathway, whereas apical recycling does not have an absolute requirement for Rab11a cycling. Morphological studies do suggest that Rab11aS20V altered the localization of the apical recycling endosomes, to a more diffuse distribution in the subapical region of the cell. Previous studies with Rab17 have suggested that cycling of Rab17 between its GDP-and GTPbound forms is required for normal transcytotic rates (25). In any case, these studies all indicate that Rab11a may associate with different effectors at different points in the membrane recycling pathways. Some of these effectors may require cycling of the Rab11a between its GDP-and GTP-bound forms. Alternatively, it is possible that apically recycling and transcytosing cargoes are maintained in separate or segregated membranes. This separation could be accomplished either through different vesicle pools or through regional segregation within a tubulovesicular structure. Indeed, van Ijzendoorn and Hoekstra (37) have recently provided evidence that distinct sorting pathways may exist in the apical recycling endosome for different sphingolipids. Similarly, Rab11a and Rab25 may be associated with differentiable pathways through the recycling system.
As we have previously reported, overexpression of wild type Rab25 dramatically inhibited both apical recycling and transcytosis. In this study we found that the GTPase-deficient mutant Rab25S21V was even more potent than wild type Rab25 in inhibiting transcytosis and apical recycling. As described above for the Rab11 mutants, neither wild type Rab25 nor Rab25S21V had any effect on basolateral recycling of transferrin. As reported previously, the GTP binding mutant Rab25T26N did not affect any of the three membrane-recycling parameters. It is possible that this lack of transport inhibition accrues from a release of the inhibitory function of Rab25 on apically directed transport.
All of these studies are indicative of the complexity of the apical recycling system in polarized epithelial cells. The diagram in Fig. 8 depicts two hypotheses that may explain the existing data on the roles of Rab11a and Rab25. Both hypotheses are built on the general construct that Rab11a is not present in the tubular common recycling endosomal sorting system (RE) where IgA and transferrin would still be colocalized. The model in Fig. 8A proposes that Rab11a regulates the transport of apically directed cargoes out of the common RE to the apical recycling endosome (ARE). Hopkins and co-workers (35) have proposed a mechanism for sorting within the RE in which basolaterally recycling proteins are selectively removed by clathrin-mediated budding, whereas the remaining endosomal membranes are enriched in transcytosing (or, presumably, apically recycling) cargo and constitute what has come to be known as the ARE. In such a model, the Rab-dependent fusion machinery may not be required for formation of the ARE, rather, Rab11a, through its association with myosin Vb, may facilitate the movement of these endosomal membranes through the actin-rich terminal web. A similar transport role has been proposed for Rab6, through its association with its effector, rabkinesin-6 (38). An inability of the constitutively active mutant Rab11aS20V to release its myosin effector may underlie the observed inhibitory effect of this mutant on transport. Alternatively, Rab11a may function in a more traditional role and mediate the fusion of carrier vesicles with the ARE.
Our observation that both wild-type and constitutively active Rab25 inhibit apically directed transport suggests that Rab25 is a negative regulator of this process. In classical terms, Rab25 may regulate endosome-endosome fusion within the ARE (Fig.  8A). In this regard, it is notable that overexpression of Rab5 induces the formation of enlarged sorting endosomes and inhibits recycling presumably by stimulating endosome-endosome fusion (22). Indeed, we do observe an enlargement of ARE membranes in Rab25-expressing cells (14). It is therefore pos-sible that endogenous Rab25 does function physiologically in a positive manner to regulate transport through the ARE. An alternative hypothesis shown in Fig. 8B suggests that Rab11a and Rab25 regulate distinct pathways through the ARE. In this model, it is more likely that these proteins would compete for common exchange factors or effectors, thus accounting for the effects of Rab25 on trafficking.
In summary, using inducible expression of Rab proteins and their mutants, we have examined the function of both Rab11a and Rab25 on vesicle trafficking in polarized MDCK cells. These studies establish quantitative evidence that Rab11a functions differently in nonpolarized and polarized cells. Whereas Rab11a appears to regulate transferrin trafficking in nonpolarized cells, in polarized MDCK cells the predominance of both quantitative and morphological data supports the interpretation that Rab11a does not regulate basolateral recycling. Rather, both Rab11a and Rab25 appear to control aspects of trafficking into or out of a discrete apical endosomal system, separate from tubular sorting endosomes. These studies sup- In both models, transferrin receptor (TfR) endocytosed from the basolateral membrane and pIgGR endocytosed from either the basolateral or apical membranes traffic through sorting endosomes (S.E.) into a common RE. Transferrin receptor is then recycled to the basolateral membrane out of the RE, and apically directed cargoes are trafficked into the ARE system. A, Rab11a is involved in the process of trafficking to and through the ARE, whereas Rab25 acts as a negative regulator retarding exit from the ARE. B, Rab11a and Rab25 represent separable pathways through the ARE that are potentially competitive. port a multilevel organization for the sorting and recycling of cargoes in polarized epithelial cells.