Wortmannin, an inhibitor of phosphoinositide 3-kinase, inhibits transcytosis in polarized epithelial cells.

Wortmannin, an inhibitor of phosphoinositide 3-kinase, inhibits both basolateral to apical and apical to basolateral transcytosis of ricin in Fisher rat thyroid (FRT) cells by 50% at 100 nM in a continuous transcytosis assay. In MDCK cells, a similar effect of wortmannin on basolateral to apical transcytosis of ricin was found, whereas apical to basolateral transcytosis was inhibited to a lesser degree. Transcytosis of dimeric IgA in MDCK cells expressing the polymeric immunoglobulin receptor was also reduced to 50% of controls, suggesting that wortmannin inhibits membrane translocation rather than sorting of specific proteins in the transcytotic pathway. This effect of wortmannin is selective, however, in that endocytosis at the basolateral domain and recycling at both the basolateral and apical membrane domains are unaffected, and apical endocytosis and apical secretion are only moderately reduced. We have shown previously that cAMP stimulates a late stage in basolateral to apical transcytosis in MDCK cells through activation of protein kinase A (Hansen, S. H., and Casanova, J.E. (1994) J. Cell Biol. 126, 677-687). Elevation of cellular cAMP still induced a 100% increase in transcytosis in wortmannin-treated cells, but transcytosis was no longer increased when compared to cells which received no drugs. In contrast, in experiments using a 17 degrees C block to accumulate ricin internalized from the basolateral surface in the apical compartment of MDCK cells, wortmannin had little effect on the stimulation of transcytosis by activators of protein kinase A observed under these conditions. The data thus suggest the existence of a wortmannin-sensitive step in the transcytotic pathway, positioned after endocytosis but prior to translocation into the protein kinase A-sensitive apical compartment, implying a role for phosphoinositide 3-kinase in an intermediate step in transcytosis in polarized epithelial cells.

ated transport processes are required to allow selective traffic to and from the apical and basolateral membrane domains and between these domains (1,2). The latter process, transcytosis, i.e. vesicle-mediated transepithelial transport, has been shown to be a brefeldin A-sensitive process (3,4) which in the case of basolateral to apical transcytosis is inhibited by calmodulin antagonists (5,6) and requires intact microtubules (7)(8)(9). However, little is known about how constitutive transcytosis is actually regulated.
Mammalian PI3K 1 (p85/110), a heterodimer consisting of an 85-kDa and a 110-kDa subunit, has been implicated in a wide array of processes involving signaling through transmembrane receptors such as mitogenesis, cell transformation, cellular differentiation, apoptosis, membrane ruffling, and histamine secretion (for reviews, see Refs. 10 -12). The 85-kDa regulatory subunit contains one SH3 domain and two SH2 domains which mediate interactions of PI3K with tyrosine-phosphorylated proteins (10 -12). The 110-kDa catalytic subunit phosphorylates PtdIns, PtdIns-4-P, and PtdIns-4,5-P 2 in the 3-position (13) and has also been shown to posses protein kinase activity in that it phosphorylates the 85-kDa subunit and the insulin receptor substrate 1 on serine (14 -16). Phosphorylation of the 85-kDa subunit inhibits the lipid kinase activity of the 110-kDa subunit of PI3K, a possible mechanism for regulating PI3K activity (14,15). Both the phosphoinositide and the protein kinase activities of PI3K are inhibited by nanomolar concentrations of the microbial product wortmannin (16 -18), which has proven to be a useful tool in defining the role of PI3K in biological processes (10,11).
The PI3K has also been implicated in vesicular trafficking. Mutagenesis studies have demonstrated that, upon ligand binding, the autophosphorylated PDGF receptor tail recruits PI3K to the membrane through binding of the SH2 domains in the 85-kDa subunit of PI3K to phosphotyrosine residues 740 and 751 of the PDGF receptor (19). It has been shown that the these two tyrosine residues are required for efficient lysosomal transport of activated PDGF receptors (20) and that this process is inhibited by 25-50 nM wortmannin (21). A similar mechanism may be operating in the postendocytic sorting of the receptor for colony-stimulating factor (22,23). Furthermore, insulin signaling through PI3K has been shown to recruit glucose transporters to the cell surface (24), which is also a vesiclemediated process (25,26).
The most compelling evidence for the significance of phosphoinositide derivatives in membrane traffic has come from studies by Emr and co-workers (27) on vacuolar protein sorting in the yeast, Saccharomyces cerevisiae. These studies have shown that the Vps34 gene encodes a phosphatidylinositolspecific 3-kinase (PtdIns 3-kinase) with a catalytic domain homologous to that of the 110-kDa subunit of mammalian PI3K (27). Through interaction with Vps15p, a serine-threonine kinase, Vps34p mediates sorting of soluble hydrolytic enzymes to the vacuole (28). Extensive analysis of Vps15 and Vps34 mutants have linked efficient sorting of soluble hydrolytic enzymes to the actual generation of PtdIns 3-phosphate (29). In addition to its substrate specificity, this PtdIns 3-kinase differs from PI3K further in that it is less sensitive to wortmannin, requiring low micromolar concentrations for inhibition (30). Furthermore, yeast deficient in Vps34 fail to complement End12, a mutant defective in postendocytic sorting of internalized ␣-factor to the vacuole, suggesting a role for Vps34p in this process (31). Very recently, the sorting and transport of lysosomal enzymes in mammalian cells has also been shown to require PI3K activity (32,33).
The importance and diversity of PI3K activity in cell biology, including its significance in the previously described aspects of vesicular trafficking, led us to investigate a possible role for the PI3K in transcytosis in polarized epithelial cells, using wortmannin as a tool. As demonstrated in this work, we found that treatment of FRT cells with 100 nM wortmannin resulted in a 50% reduction in basolateral to apical as well as apical to basolateral transcytosis of ricin, a toxic lectin which binds to terminal galactose residues on glycoproteins and glycolipids and thus serves as an effective marker of bulk membrane transport (4,34). Wortmannin had no effect on endocytosis from the basolateral membrane domain of FRT cells, while some reduction in apical endocytosis was detected. In MDCK cells expressing the polymeric immunoglobulin receptor (pIgR), basolateral to apical transcytosis of both ricin and dimeric IgA (dIgA) was reduced to 50% of controls in the presence of 100 nM wortmannin, showing that bulk membrane and receptor-mediated transcytosis are affected similarly. It has previously been shown that transcytosis in MDCK cells is stimulated by activators of protein kinase A (PKA) (35,36), and that the stimulation occurs at a late stage in the basolateral to apical transcytotic pathway (36). Using a 17°C block to accumulate ricin internalized from the basolateral domain in the apical compartment of MDCK cells, we provide evidence that wortmannin exerts its inhibitory effect on postendocytic transport earlier in the transcytotic pathway than the late stage where PKA stimulates. These data thus implicate a role for PI3K at an intermediate step in transcytosis in polarized epithelial cells.

MATERIALS AND METHODS
Reagents-Wortmannin, obtained from Sigma, was dissolved in dimethyl sulfoxide at a concentration of 1 mM and distributed in 50-l aliquots which were stored at Ϫ20°C. An aliquot of this stock was thawed and diluted immediately prior to each experiment to minimize degradation of wortmannin, and the remainder was discarded. The concentration of dimethyl sulfoxide did not exceed 0.04% in any experiment in this work. All other reagents used in this work were obtained, prepared, and used as described previously (36).
Cell Culture-Fisher rat thyroid (FRT) cells were kindly provided by Dr. Michael P. Lisanti (Whitehead Institute of Biomedical Research) and grown in F12 Coon's modified medium, obtained from Biofluids, Inc. (Rockville, MD), supplemented with 5% fetal calf serum, 2 mM L-glutamine, and antibiotics. MDCKII cells expressing the pIgR were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics. Both cell lines were maintained in a humidified atmosphere of 5% CO 2 , 20% O 2 , and 75% N 2 and split 1:20 three times in 2 weeks with one change of medium between each split. 5 d before experimentation, confluent 100-mm plates with FRT or MDCK cells were trypsinized, the cells were resuspended in 12 ml of medium per plate, and 0.5/1.5 ml of cell suspension was seeded per 12/24-mm Transwell polycarbonate filter with a pore size of 0.4 m (Costar Corp., Cambridge, MA).
Transcytosis Experiments with 125 I-Ricin-A 12-well plate with 12-mm filters was processed in each experiment measuring transcytosis of ricin in FRT or MDCK cells, and these experiments were performed exactly as described previously for MDCK cells (36). Briefly, in a 37°C continuous basolateral to apical transcytosis experiment, the cells were rinsed twice with Hanks' buffered saline containing 0.6% BSA (HBSA) and incubated with vehicle only or with drugs (always in both chambers simultaneously) as detailed in the figure legends. Next, 125 I-ricin (250 ng/ml, 1 ϫ 10 4 cpm/ng), iodinated with chloramine T, was added from the basolateral side, and the filters were incubated for 15 or 60 min. Transport was arrested by transferring the plate to slush ice, the basolateral side was rinsed 6 times, and the apical medium containing released transcytosed counts was sampled. The filters were then incubated for 5 min at 37°C in the presence of 0.1 M lactose in both chambers to release membrane-associated 125 I-ricin, the basal and apical media were sampled, the filters were excised, and media and filters were counted. Total cell-associated counts were calculated as counts released from the basolateral side with 0.1 M lactose plus counts remaining in the cells. Intracellular ricin was expressed as the ratio of counts remaining in the cells to total cell-associated counts. Transcytosed ricin was expressed as the ratio of counts released into the apical medium plus counts released from apical membrane with 0.1 M lactose to total cell-associated counts. Apical to basolateral transcytosis experiments were performed identically except for the reversed polarity. More than 95% of transcytosed counts were precipitable by trichloroacetic acid, and this fraction was not affected by treatment with wortmannin or other drugs. Furthermore, Ն96% of ricin bound to the surface of either FRT or MDCK cells in 1 h at 4°C could be removed with lactose irrespective of whether the cells had been treated with wortmannin. In experiments utilizing a 17°C temperature block (8), MDCK cells were incubated with 400 ng of 125 I-ricin for 4 h at 17°C to allow translocation of the marker into the apical cytoplasm. Next, the basolateral side of the filter was rinsed 6 times, and the cells were incubated further at 17°C for 30 min with or without drugs as described in the legend to Fig. 4. The cells were then warmed to 37°C for 15 min in the presence of 0.1 M lactose on both sides with or without drugs, and finally the apical and basal media as well as the filters were sampled as above. Transcytosis of ricin in 17°C experiments was expressed as in 37°C continuous transcytosis experiments.
Transcytosis Experiments with 125 I-dIgA-Human dimeric IgA was prepared as described previously (37) from serum generously provided by Dr. Per Brandtzaeg (Rikshospitalet, Oslo, Norway) and iodinated to a specific activity of 2 ϫ 10 4 cpm/ng using chloramine T. Transcytosis assays were performed essentially as described previously (38). Briefly, MDCK cells stably expressing the rabbit polymeric immunoglobulin receptor (39), cultured on 12-mm Transwells were pretreated for 15 min with or without wortmannin (100 nM), then allowed to internalize 125 I-dIgA (400 ng/ml) from the basolateral surface for 10 min at 37°C in the continuous presence or absence of the drug. Filters were then washed 4 times rapidly with HBSA, placed in fresh HBSA with or without wortmannin, and chased for the times indicated in Fig. 2. At each time point, apical and basolateral media were harvested and replaced with fresh media. After the final time point, filters were cut from their holders and placed in Microfuge tubes. All samples were precipitated on ice for 15 min by addition of an equal volume of 20% trichloroacetic acid. Trichloroacetic acid-soluble counts did not exceed 4% of the total in any experiment. Apical recycling experiments were performed essentially as described (7) except that cells were treated with or without wortmannin as described above.
Determination of Cellular cAMP-Cellular cAMP levels were determined essentially as described by Shimizu et al. (40). Briefly, cells cultured on 12-mm Transwells were labeled overnight with 2 Ci/ml [ 3 H]adenine. Triplicate filters were then treated with the appropriate agonist for 30 min at 37°C. In some cases, cells were treated with 100 nM wortmannin for 15 min before or after addition of agonist. Filters were then cut from their holders and placed in 1.0 ml of stop solution containing 5% trichloroacetic acid, 1 mM ATP, and 1 mM cAMP. After extraction for 30 min on ice, the supernatants were subjected to sequential chromatography on Dowex and alumina (1 ml bed volume). The Dowex column was washed with H 2 0, and the ATP was recovered in the first 4 ml. The cAMP was collected in the subsequent 10 ml which were applied to the alumina column and eluted with 10 ml of 10 mM imidazole, 26 mM HCl. Data are expressed as the ratio of cAMP to ATP.
Secretion Experiments-MDCK cells on 24-mm filters were rinsed twice with phosphate-buffered saline containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ , incubated for 20 min at 37°C in Dulbecco's modified Eagle's medium minus cysteine and methionine containing 40 mM Hepes, pH 7.4, and pulse-labeled with 1 mCi/ml EXPRE 35 S 35 S (Dupont NEN) for 15 min at 37°C. The cells were then chased for 2 h at 20°C to accumulate labeled proteins in the TGN (41) and during the last 30 min at 20°C treated with drugs as stated in the legend to Fig. 5. Next, the filters were incubated for 15-60 min at 37°C in the continued presence or absence of drugs, and the apical media (400 l) containing the labeled secreted proteins were sampled. Aliquots thereof were mixed 1:2 with sample buffer containing 5% ␤-mercaptoethanol, boiled, and subjected to SDS-polyacrylamide gel electrophoresis and fluorography. Densitometry was performed using a Hewlett-Packard Scanjet IICX scanner and IPLab Gel densitometry software (Signal Analytics Corp., Vienna, VA).

Wortmannin Inhibits Bidirectional Transcytosis of Ricin in
FRT Cells-To investigate a possible role for PI3K in membrane traffic in polarized epithelial cells, we tested the effect of wortmannin on transcytosis of ricin in FRT cells in a 1-h continuous transcytosis assay. As shown in Fig. 1, wortmannin inhibited both basolateral to apical and apical to basolateral transcytosis of ricin in FRT cells by 20 -25% at 10 nM and by 50% at concentrations Ն50 nM. The dose-response of wortmannin on transcytosis with significant effects at nanomolar concentrations is highly suggestive of a role for PI3K in transcytosis and similar to other processes in which PI3K has been implicated (18,21,(42)(43)(44)(45)(46). The effect of wortmannin on basolateral to apical transcytosis of ricin reached a plateau at 50 -100 nM, whereas apical to basolateral transcytosis continued to decrease at higher wortmannin concentrations (Fig. 1, A  and B; data not shown). Wortmannin did not affect internalization of ricin from the basolateral domain, whereas endocytosis from the apical surface was diminished in the presence of wortmannin (Fig. 1, C-F). However, as seen by comparing B and D of Fig. 1, the reduction in apical endocytosis following treatment with wortmannin is too small to account for the effect on apical to basolateral transcytosis in FRT cells. In all subsequent experiments, we used 100 nM, a concentration at which wortmannin almost completely inhibits PI3K activity without any known effect on other lipid or protein kinases. It

FIG. 1. Wortmannin inhibits transcytosis of ricin in FRT cells. Monolayers of FRT cells grown 5 days on
Transwell filters were preincubated with 0 -250 nM wortmannin in both basal and apical chambers for 15 min at 37°C before addition of 125 I-ricin (250 ng/ml; 1 ϫ 10 4 cpm/ng) to the basolateral or apical side of the cells and further incubated for 60 min at 37°C. The medium containing soluble transcytosed 125 I-ricin was harvested directly, while membrane-associated 125 I-ricin was recovered by incubation with 0.1 M lactose for 5 min at 37°C. Finally, the filters containing the cells with intracellular 125 I-ricin were excised, and all samples were counted. In a basolateral to apical transcytosis experiment, total cell-associated counts were calculated as counts removed with 0.1 M lactose from the basolateral side of the filter plus counts remaining in the cells. Intracellular ricin was expressed as the ratio of counts in the cells to total cell-associated counts. Transcytosed ricin was expressed as the ratio of soluble transcytosed counts plus transcytosed counts recovered with 0.1 M lactose to total cell-associated counts. Data were expressed similarly in apical to basolateral transcytosis experiments except for the reversed polarity. should furthermore be noted that 100 nM wortmannin had no effect on the transepithelial resistance even after prolonged (Ն2 h) incubations with the drug (data not shown).
Wortmannin Inhibits Transcytosis of Ricin and dIgA in MDCK Cells-100 nM wortmannin also reduced basolateral to apical transcytosis in MDCK cells to nearly 50% of control, whereas transport in the opposite direction was inhibited by only 30% (data not shown). To further characterize the putative role for PI3K in transcytosis in polarized epithelial cells, we determined the effect of wortmannin on receptor-mediated transport of dIgA in MDCK cells expressing the pIgR. As for ricin, basolateral to apical transcytosis of dIgA was reduced to 50% of the control level in the presence of 100 nM wortmannin ( Fig. 2A). A small stimulation in apical to basolateral transcytosis of dIgA was observed following treatment with wortmannin, but the transcytosed amounts under these conditions are so low that it is difficult to derive any conclusion based on this result (Fig. 2B). Neither endocytosis (data not shown) nor recycling from either surface (Fig. 2, A and B) were significantly affected by wortmannin. This is consistent with reports that PI3K (21) and Vps34p/End12 (31) play no significant roles in internalization of PDGF receptors in mammalian cells or ␣-factor in yeast, respectively. Thus, the main effect of wortmannin on vesicular transport of dIgA is a selective inhibition of basolateral to apical transcytosis. Since bulk-membrane (ricin) and receptor-mediated (dIgA) transport are similarly affected by wortmannin, the data implicate a role for PI3K in postendo-cytic membrane translocation, rather than sorting of specific proteins, in the transcytotic pathway.
Activators of PKA Stimulate Transcytosis Bidirectionally in FRT Cells-We have previously found that activators of PKA stimulate both receptor-mediated and bulk membrane transcytosis in MDCK cells, and that this stimulation is more pronounced for basolateral to apical transcytosis than transport in the opposite direction (36). The same polarity was found for the inhibition of transcytosis by wortmannin in MDCK cells in the present study. To examine whether activators of PKA affected transcytosis in FRT cells with the same (lack of) polarity as wortmannin, we performed many of the experiments previously carried out with MDCK cells. In FRT cells, forskolin (FSK) and 8-Br-cAMP in the presence of the phosphodiesterase inhibitor Ro 20 -1724, strongly stimulate both basolateral to apical and apical to basolateral transcytosis of ricin in a 1-h continuous transcytosis assay (Fig. 3). We also found that cholera toxin, which constitutively activates G s ␣ through ADPribosylation, stimulates transcytosis of ricin bidirectionally in FRT cells (data not shown). In further agreement with the data obtained with MDCK cells, the stimulation in transport in both directions by FSK was significantly reduced in the presence of H-89, a selective inhibitor of PKA, and completely eliminated when FSK was substituted with dideoxyforskolin, an analog which does not activate adenylate cyclase (Fig. 3). The stimulation in transcytosis correlated with a 10-fold elevation of cellular levels of cAMP (data not shown). Thus, in two polarized epithelial cell lines with distinct differences in protein sorting, wortmannin inhibits the same vesicular transport processes that are stimulated by activators of PKA.

Wortmannin Inhibits an Intermediate
Step in Transcytosis-The data presented above implicate a role for both PI3K and PKA in bidirectional transcytosis in FRT cells as well as basolateral to apical transport in MDCK cells. This similarity made it interesting to study the relationship between the effects of wortmannin and activators of PKA on transcytosis. Furthermore, since PKA exerts it stimulatory role at a late stage in the basolateral to apical transcytotic pathway in MDCK cells, after internalized ligand has been translocated into the apical compartment, it was possible to use activators of PKA to position the inhibitory effect of wortmannin in the transcytotic pathway.
Wortmannin did not affect the generation of cAMP by Ro 20 -1724 and FSK (data not shown). In the presence of 100 nM wortmannin, the combination of forskolin and Ro 20 -1724 still increased transcytosis by 100% above the level of wortmannintreated cells in a continuous 1-h transcytosis assay in both FRT and MDCK cells (Fig. 4, A-C). However, transcytosis was no longer increased when compared to cells which received no drugs (Fig. 4, A-C). In contrast, in pulse-chase experiments using a 17°C block to accumulate ricin internalized from the basolateral surface in the apical compartment of MDCK cells, the presence of wortmannin had little effect on the stimulation in transcytosis by forskolin and Ro 20 -1724 observed under these conditions (Fig. 4D). These data suggest that the effects of wortmannin and agents which increase cellular levels of cAMP are separable and implicate a role for PI3K in an intermediate step in transcytosis after internalization but before translocation to the PKA-sensitive compartment. apical media sampled from metabolically labeled cells (47,48). In one set of experiments, the cells were pulse-labeled for 15 min with a mixture of [ 35 S]methionine and [ 35 S]cysteine and finally chased for 60 -120 min at 37°C. The presence of 100 nM wortmannin throughout these experiments had no effect on the amount of gp 80 secreted into the apical or the basolateral medium (data not shown). In another set of experiments, following pulse-labeling, the cells were incubated for 2 h at 19.5°C to accumulate labeled proteins in the TGN (41). During the last 30 min at 19.5°C, the cells were treated with or without 100 nM wortmannin and, in some experiments, were incubated further in the presence or absence of Ro 20 -1724 and FSK for the last 15 min. Next, the cells were warmed to 37°C for 15, 30, or 60 min with or without drugs. Under these conditions, treatment with 100 nM wortmannin reduced apical delivery of gp 80 by 25% after 15 min at 37°C, whereas no reduction was observed at later time points (Fig. 5A). Furthermore, wortmannin did not reduce the previously demonstrated strong stimulation in gp 80 secretion by Ro 20 -1724 and FSK after 15 min (Fig. 5B) or at later time points (data not shown). Treatment with wortmannin thus causes a slight but reproducible reduction in the rate of constitutive delivery of gp 80 from the TGN to the apical surface. Similar results were found for apical secretion of soluble endogenous proteins from FRT cells (data not shown).

DISCUSSION
In an attempt to identify mechanisms regulating vesicular transport in polarized epithelial cells, we have investigated a possible role for PI3K in transcytosis, using the fungal metabolite wortmannin as a tool. At nanomolar concentrations, wortmannin is an irreversible, noncompetitive, and specific inhibitor of this kinase (17,18). Indeed, we found that as little as 10 nM wortmannin gave a significant (25%) reduction in bidirectional transcytosis of ricin in FRT cells, while, at a concentration of 100 nM, transcytosis was reduced to 50% of control values. In MDCK cells, both receptor-mediated (dIgA) and bulk-membrane (ricin) transcytosis was inhibited by 50% in the basolateral to apical direction, whereas apical to basolateral transport of ricin was inhibited by 25%. Recycling of dIgA was unimpaired at either the basolateral or apical surface. These data suggest a role for PI3K in transcytosis, a postendocytic vesicular trafficking pathway that is unique to epithelial cells.
As mentioned in the introduction, a role for PI3K in postendocytic sorting of activated PDGF receptors has been described. The observation that mutant receptors lacking PI3K binding sites are efficiently internalized but not degraded has led to the assumption that PI3K is involved in the sorting of the internalized receptors into the lysosomal pathway (20). This hypothesis has been supported by the demonstration that 25-50 nM wortmannin also inhibits degradation of receptors in the presence of PDGF (21). Our results, however, using ricin and dIgA as markers of bulk membrane and receptor-mediated transcytosis, respectively, suggest that wortmannin at similar concentrations significantly inhibits the constitutive flow of membrane into the transcytotic pathway rather than sorting of individual proteins.
Another constitutive membrane transport process that is known to require PI3K activity is the sorting of soluble hydrolytic enzymes to the yeast vacuole. Vps34p, the PtdIns 3-kinase which is essential to this process, has an IC 50 of 4 M and requires 10 M wortmannin for complete inhibition, a 100-fold higher concentration than mammalian p110 (30). Vps34p appears to be required for a specific sorting event since trafficking of vacuolar membrane proteins remains unaffected following inactivation of Vps34p (29). Yet, the possibility remains that soluble and membrane vacuolar proteins are cargoed in two separate vesicle populations, of which only the first requires Vps34p for a budding or other membrane translocation event. The organelle in the biosynthetic pathway where the action of Vps34p is required is probably the TGN, although Munn and before addition of 125 I-ricin to the basal or apical side of the filters. The cells were then incubated for 60 min at 37°C, and counts representing transcytosed, intracellular, and bound 125 I-ricin were obtained as described in Fig. 1. A, basolateral to apical transcytosis. B, apical to basolateral transcytosis. The data were first expressed as the ratio of transcytosed to total cell-associated (bound ϩ internalized) counts and then normalized to controls (100%) from the same experiments. The data are representative of at least two separate experiments, each performed in triplicate, with a coefficient of variation less than 15% in the data shown. C, summary of the protocol for these experiments.
Riezman (31) have recently determined that Vps34p is also required for transport of internalized ␣-factor to the vacuole, suggesting that it may function at the exit from late endosomes as well. Very recently, Brown et al. (33) have presented evidence in favor of a similar role for a PI3K in transport of newly synthesized lysosomal enzyme (cathepsin D) from the TGN to late endosomes/prelysosomes in the mammalian cell lines NRK and CHO. As for Vps34p, high wortmannin concentrations (1-3 M) were required for maximal inhibition of cathepsin D processing in late endosomes. Furthermore, this mammalian enzyme appears to function in a M6PR-dependent sorting event in the TGN, since M6PRs were detected at normal levels in the TGN but were depleted from the late endosomes in wortmannin-treated cells (33). These authors suggest that the most likely candidate is a PtdIns 3-kinase characterized by Stephens et al. (49), which, like Vps34p, has a high IC 50 for wortmannin. Davidson (32), however, has found that in K562, NRK, and Cos 1 cells the sorting of procathepsin D is inhibited by low nanomolar concentrations of wortmannin. Davidson suggests that the transport of lysosomal enzymes in these cells is regulated by a mammalian homolog of Vps34p, which very recently has been cloned by Volinia et al. (50) and shown to interact with a mammalian homolog of Vps15p. This PtdIns 3-kinase is indeed inhibited by concentrations of wortmannin in the low nanomolar range (50). Thus, it is possible that homologs of Vps34p function in transport of soluble hydrolytic enzymes to the lysosome in mammalian cells.
The effect of wortmannin on transcytosis was further characterized through experiments that included activators of PKA. These experiments were prompted by the finding that wortmannin appeared to inhibit vesicular transport processes that are stimulated by PKA activators. In 37°C continuous transcytosis experiments, the inhibition by wortmannin in percent was the same in cells treated with PKA activators as in controls. In contrast, in experiments with MDCK cells employing a 17°C temperature block where marker internalized from the basolateral domain is accumulated in the apical compartment (8), the presence of wortmannin had little effect on the stimulation by PKA activators on transport of ricin to the apical membrane. These results show that wortmannin exerts its inhibitory effect on transcytosis prior to the late stage in the transcytotic pathway where activators of PKA stimulate. In addition, it was found that wortmannin has a small inhibitory effect on apical secretion of gp 80 which also is stimulated strongly by activa- tors of PKA. A simple hypothesis compatible with these data is that wortmannin inhibits vesicle budding from a postendocytic compartment whereas activators of PKA exert their effect by stimulating fusion of transcytosing and secretory vesicles with the apical membrane in MDCK cells (and possibly with both apical and basolateral membranes in FRT cells).
Very recently, an isoform of the 110-kDa subunit of PI3K, p110␥, has been cloned which has no binding site for the 85-kDa regulatory subunit and is activated by G protein subunits (51). Although it is tempting to speculate that this isoform is involved in the regulation of transcytosis, there are several discrepancies which are difficult to reconcile with this possibility. First, as discussed above, the data obtained in this work implicate PKA (activated by G s ␣, adenylyl cyclase, and cAMP) and PI3K at different stages in the transcytotic pathway. Second, the p110␥ is less sensitive to wortmannin requiring higher concentrations for complete inhibition than those significantly inhibiting transcytosis. Third, in addition to ␤␥ subunits of G proteins (51,52), the activity of p110␥ appears to be stimulated strongly by the G i ␣ group of G␣ subunits (51), whereas pertussis toxin, an inhibitor of this group, has no effect on transcytosis (36).
Serine/threonine kinases clearly serve important roles in regulating transcytosis. In addition to PKA, it has been shown that transcytosis of the pIgR is also stimulated by phorbol esters which activate protein kinase C (53). It remains unknown, however, if this effect of phorbol esters is specific for the pIgR or, like PKA, affects bulk membrane transport as well.
Furthermore, it has been demonstrated that sorting of the pIgR into the transcytotic pathway is stimulated by phosphorylation of the receptor itself by an as yet unidentified kinase (38). The catalytic subunit of PI3K is also a protein kinase (see introduction), and it cannot be excluded that it is the protein and not the lipid kinase activity of PI3K which is involved in the regulation of constitutive transcytosis, as both activities are inhibited by wortmannin. However, there is to our knowledge no precedence for a role for the serine kinase activity of PI3K in membrane traffic.
How the PI3K functions in membrane traffic is not known in detail, but it has been suggested that the negative charge of the phosphoinositide products of the kinase may be important in membrane invagination and/or recruitment of adaptor proteins (12). The negative charge of phospholipid products have also been implicated in the possible function of other lipid kinases in membrane traffic (54), like phosphatidylinositol 4-kinase and phosphatidylinositol 4-phosphate 5-kinase, which may operate through a mechanism that involves ADP-ribosylation factor (55,56). These suggestions fit our data well in that the effect of wortmannin does not appear to distinguish between receptor-mediated and bulk membrane transport, arguing against a role for PI3K in sorting but in favor of a possible function in formation of vesicles that mediate transcytosis. This could also explain the differential effect of wortmannin on the transcytotic pathways in MDCK cells, since the apical and basolateral membranes of these cells have been shown to exhibit distinct differences in lipid composition (57). It seems likely that either the extent to which the charged phospholipids are generated and/or their effect on vesicle formation would differ depending on whether the membrane derives from the apical or basolateral domain of MDCK cells. FRT cells, in contrast to most other polarized epithelial cells studied (including MDCK), have been shown to sort most of their glycosylphosphatidylinositol-anchored proteins to the basolateral domain (58). This suggests that the lipid compositions of the apical and basolateral membrane domains in FRT cells differ substantially from those of MDCK cells and could provide an explanation for the lack of polarity in the effect of wortmannin on transcytosis in FRT cells. However, whether this is actually the case remains speculative until more is known about the role of lipids and their kinases in membrane traffic.