Wortmannin-sensitive Trafficking Pathways in Chinese Hamster Ovary Cells DIFFERENTIAL EFFECTS ON ENDOCYTOSIS AND LYSOSOMAL SORTING*

Phosphatidylinositol (PI) 3 (cid:42) -kinases are a family of lipid kinases implicated in the regulation of cell growth by oncogene products and tyrosine kinase growth factor receptors. The catalytic subunit of the p85/p110 PI 3 (cid:42) kinase is homologous to VPS-34, a phosphatidylinositol- specific lipid kinase involved in the sorting of newly synthesized hydrolases to the yeast vacuole. This sug- gests that PI 3 (cid:42) -kinases may play analogous roles in mammalian cells. We have measured a number of secre- tory and endocytic trafficking events in Chinese hamster ovary cells in the presence of wortmannin, a potent inhibitor of PI 3 (cid:42) -kinase. Wortmannin caused a 40–50% down-regulation of surface transferrin receptors, with a dose dependence identical to that required for maximal inhibition of the p85/p110 PI 3 (cid:42) -kinase in intact cells. The redistribution of transferrin receptors reflected a 60% increase in the internalization rate and a 35% decrease in the recycling rate. Experiments with fluorescent transferrin showed that entry of transferrin receptors into the recycling compartment and efflux of receptors out of the compartment were slowed by wortmannin. Wortmannin altered the morphology of the recycling compartment, which was more vesiculated than in untreated cells. Using Semliki Forest virus as a probe, we also found that delivery of the endocytosed virus its lysosomal site of degradation slowed by wortman- nin, whereas endosomal acidification was unaffected. In contrast to these effects on endocytosis and recycling, wortmannin did not affect intracellular processing of newly synthesized viral spike proteins. Wortmannin did induce missorting of the lysosomal enzyme cathepsin to the secretory pathway, but only at a dose 20-fold greater than that required to inhibit p85/p110 PI 3 kinase activity or to redistribute transferrin receptors. Our data demonstrate the presence of wortmannin-sen-sitive enzymes at three distinct steps of the endocytic cycle in Chinese hamster ovary cells: internalization, transit from early endosomes to the recycling and deg- radative compartments, and transit from the recycling compartment back to the cell surface. The wortmannin- sensitive enzymes critical for endocytosis and recycling are distinct from those involved in sorting newly syn- thesized lysosomal enzymes. boxymethylated incubation m S-Labeled cathepsin D immunoprecipitated with anti-cathepsin D antibody (a generous gift of Dr. Gregory Conner, University of Miami School of Medicine) and protein A-Sepharose (Pharmacia Biotech Inc.), eluted, and analyzed by SDS-polyacrylamide electrophoresis (10%

Phosphatidylinositol (PI) 1 3Ј-kinases are a family of lipid kinases that contain homologous catalytic domains, but distinct regulatory elements. The first known PI 3Ј-kinase was a heterodimeric lipid kinase composed of an 85-kDa regulatory subunit and a 110-kDa catalytic subunit (1)(2)(3)(4)(5). This enzyme was discovered in transformed cells, but is also stimulated by mitogens including platelet-derived growth factor and insulin (6). The p85/p110 PI 3Ј-kinase is activated when the two SH2 domains in p85 bind to phosphorylated YXXM motifs in tyrosine kinase receptors or their substrates (7,8); activation by interactions with p21 ras , Rho, and the SH3 domains of Src family kinases has also been described (9 -11). Activation of the p85/p110 PI 3Ј-kinase increases the intracellular concentrations of PI(3,4)P 2 and PI(3,4,5)P 3 (12). Although these lipid products activate calcium-independent protein kinase C isoforms in vitro, their function in intact cells is not known (13,14). p110 also contains a serine/threonine protein kinase activity with an apparently limited range of substrates (15). In addition to multiple isoforms of both the 85-and 110-kDa subunits, the recent description of several novel PI 3Ј-kinase isoforms with varying substrate specificities and regulatory mechanisms has added to the complexity of 3-phosphoinositide signaling (16 -18).
The p110 catalytic subunit is 55% homologous to the Saccharomyces cerevisiae protein VPS-34 (2,19). Disruption of the VPS-34 gene causes missorting of vacuolar hydrolases to the secretory pathway and also abolishes PI 3Ј-kinase activity in yeast (20). Although VPS-34 is a PI 3Ј-kinase, its substrate specificity is limited to phosphatidylinositol, unlike the p85/ p110 mammalian enzyme that also utilizes PI(4)P and PI(4,5)P 2 (21). Moreover, mammalian p110 is lethal when expressed in yeast (22). However, a mammalian homologue to VPS-34 has been cloned, which, like VPS-34, is a lipid kinase activity specific for phosphatidylinositol (16). Several experiments suggest that PI 3Ј-kinases play a role in vesicular trafficking in mammalian cells. Mutagenesis of the two PI 3Јkinase-binding sites in the platelet-derived growth factor receptor disrupts the post-endocytic sorting of this receptor (23). However, these same sites bind to SH2-containing mole-cules other than PI 3Ј-kinase (24), which complicates the interpretation of this finding. PI kinase inhibitors such as wortmannin and LY294002 inhibit histamine secretion in RBL-2H3 cells, exocytosis in neutrophils stimulated with chemotactic agents, and translocation of the GLUT4 glucose transporter in insulin-stimulated adipocytes and skeletal muscle (25)(26)(27)(28)(29)(30)(31). These data have been interpreted as demonstrating a role for the p85/p110 PI 3Ј-kinase in these processes. However, a number of PI kinase isoforms are sensitive to wortmannin, including both PI 3-kinases and PI 4-kinases (16,24,32). Despite questions as to the specificity of these findings, there is a growing body of evidence suggesting a role for 3-phosphoinositides in intracellular trafficking.
In this study, we examined a number of endocytic and biosynthetic trafficking pathways in CHO cells using wortmannin, which irreversibly inhibits several PI kinase isoforms. Treatment of cells with 100 nM wortmannin causes a 50% reduction of cell-surface transferrin receptors due to an increase in the internalization rate and a decrease in the recycling rate. Consistent with these changes, the morphology of the recycling compartment is altered by wortmannin, and movement of transferrin into and out of the recycling compartment is slowed. We also find that wortmannin slows the delivery of endocytosed Semliki Forest virus to its lysosomal site of degradation. However, we were unable to detect any changes in the transit of newly synthesized viral glycoproteins through the Golgi apparatus and to the plasma membrane, and we detected alterations in the sorting of newly synthesized lysosomal enzymes only at micromolar concentrations of wortmannin. Our data demonstrate the presence of wortmannin-sensitive enzymes that regulate receptor-mediated endocytosis and recycling. These enzymes are pharmacologically distinct from those involved in the sorting of newly synthesized lysosomal proteases.
Inhibition of PI 3Ј-Kinase by Wortmannin-Wortmannin (Sigma) was stored in dimethyl sulfoxide at Ϫ70°C and diluted into protein-free medium just prior to use. Inhibition of PI 3Ј-kinase in intact cells was measured by incubating CHO/IR cells with the indicated concentrations of wortmannin or dimethyl sulfoxide carrier (0.1%, v/v) for 30 min at 37°C. After an additional incubation in the absence or presence of 100 nM insulin, the cells were lysed, and PI 3Ј-kinase activity in anti-p85 immunoprecipitates was determined as described previously (7).
Transferrin Uptake and Recycling-TRVb-1 cells cultured in 6-well dishes were incubated in the absence or presence of wortmannin (100 nM unless otherwise indicated, in a final concentration of 0.1% dimethyl sulfoxide) for 45 min at 37°C; control experiments showed that 0.1% dimethyl sulfoxide does not affect trafficking parameters in TRVb-1 cells. Transferrin endocytosis was measured, and endocytic rate constants were calculated using the In/Sur method as described previously (33,36). In experiments examining both internalization and surface binding at 37°C, parallel sets of untreated or wortmannin-treated cells were incubated with 125 I-labeled transferrin for 0 -10 min and then washed with neutral buffer, to determine surface-bound radioactivity, or with acidic buffer, to determine intracellular radioactivity. Uptake of 55 Fe-labeled transferrin was measured by incubating untreated or wortmannin-treated cells (100 nM; 45 min) in the presence of 10 g/ml 55 Fe-labeled transferrin as described (33). To measure recycling, TRVb-1 cells were incubated with 125 I-labeled transferrin for 2 h at 37°C, followed by an additional 45 min with 125 I-labeled transferrin in the absence or presence of 100 nM wortmannin. The cells were washed with 50 mM MES (pH 5.0) and 150 mM NaCl to remove surface-bound transferrin and incubated in 600 g/ml unlabeled transferrin and 100 g/ml desferoxamine in the continued absence or presence of wortmannin. At various times, the medium and one wash were collected, and the cells were solubilized to determine recycled and cell-associated radioactivity.
Insulin Endocytosis-Cells were pretreated with 100 nM wortmannin or carrier (0.1% (v/v) dimethyl sulfoxide) for 30 min at 37°C. Insulin internalization was measured, and internalization rate constants were calculated as described previously (37).
Analysis of Transferrin Trafficking by Immunofluorescence-In pulse-chase experiments, TRVb-1 cells grown on coverslips were incubated in 100 nM wortmannin or carrier for 30 min at 37°C. Cells were incubated with Cy3-transferrin (10 g/ml) for 3 min, washed, and chased in the continued absence or presence of wortmannin. At various times, the cells were washed and fixed in 3.7% formaldehyde as described (38) and examined using a Bio-Rad MRC 600 laser scanning confocal microscope with Nikon ϫ60 NA 1.4 planapo optics. Serial images were collected at 0.6-m steps over the depth of the cells and projected using the maximum pixel method and were printed using a Sony Mavigraph color video printer. In the sorting experiments, cells were incubated in Cy3-transferrin for 15 min, washed, and then chased with medium containing FITC-transferrin (10 g/ml) in the absence or presence of 100 nM wortmannin. Cells were washed three times after the chase period, fixed in 3.7% formaldehyde, and examined using a Zeiss Axiovert microscope at ϫ63 or ϫ100. Images were photographed directly from the microscope using identical exposure times.
Uptake and Degradation of 35 S-Labeled Semliki Forest Virus-[ 35 S]Methionine-labeled Semliki Forest virus was prepared and purified as described (39). CHO/IR or TRVb-1 cells grown in 35-mm plates were pretreated in the absence or presence of 100 nM wortmannin for 30 min at 37°C. The cells were then incubated for 1.5 h at 4°C with 35 S-labeled Semliki Forest virus in the continued absence or presence of wortmannin. After a rinse in cold medium to remove unbound virus, internalization and degradation of surface-bound SFV were determined as described (40).
Measurement of Endosomal Acidification with Conformation-specific Antibodies-The effect of wortmannin on endosomal acidification was determined by following the acid-induced conformational change in the Semliki Forest virus E1 spike protein as described previously (41). Briefly, TRVb-1 cells were incubated for 30 min at 37°C in the presence of 100 nM wortmannin or carrier and incubated with 35 S-labeled SFV at 4°C as described above. The cells were warmed rapidly to 37°C to initiate endocytosis and lysed at various times. Lysates were incubated with a rabbit polyclonal antibody against the SFV spike protein or with monoclonal antibody E1a-1, which specifically recognizes the SFV E1 glycoprotein only after it has undergone an acid-dependent conformational change (40). Immunoprecipitated proteins were analyzed as described (40).
Transit of Newly Synthesized Vesicular Stomatitis Virus Glycoproteins-Biosynthetic trafficking of the vesicular stomatitis virus spike protein (VSV-G) was analyzed as described previously (42). Briefly, CHO cells cultured in 35-mm dishes were infected with vesicular stomatitis virus (20 plaque-forming units/cell). The cells were treated without or with 100 nM wortmannin for 30 min at 37°C in methioninefree medium, labeled with [ 35 S]methionine (50 Ci/well) for 5 min, and chased for various periods of time in medium containing 0.15 mg/ml methionine. Total radiolabeled VSV-G was precipitated with trichloroacetic acid (10% final concentration) and analyzed by SDS-polyacrylamide gel electrophoresis (43). Plasma membrane VSV-G was labeled by biotinylation with N-hydroxysuccinimidobiotin (Pierce), collected on streptavidin-agarose beads, and analyzed as described (44). Total and plasma membrane 35 S-labeled VSV-G was quantitated using a Phos-phorImager (Molecular Dynamics, Inc.).
Processing of Newly Synthesized Semliki Forest Virus Spike Proteins-Processing of the SFV E1 and E2 spike proteins in wortmannintreated cells was analyzed as described previously (40). Briefly, CHO cells grown in 35-mm dishes were infected with SFV (100 plaqueforming units/cell). Cells were incubated for 30 min in the absence or presence of 100 nM wortmannin in methionine-free medium, labeled with [ 35 S]methionine (50 Ci/plate) for 30 min, and chased in medium containing excess methionine in the continued absence or presence of wortmannin. At various times, 35 S-labeled spike proteins were immunoprecipitated from solubilized cells and analyzed as described (40). Viral budding was determined by immunoprecipitation of 35 S-labeled proteins from solubilized aliquots of the medium.
Secretion of Cathepsin D-Analysis of cathepsin D synthesis was performed as described by Richo and Conner (45). TRVb-1 or CHO/IR cells, grown in 35-mm dishes, were incubated for 30 min in wortmannin as indicated in methionine-free medium, labeled with [ 35 S]methionine for 30 min, and chased in medium containing excess methionine in the continued absence or presence of wortmannin. At various times, the medium was removed, SDS was added (0.4% final concentration), and the samples were boiled. The cells were solubilized in 0.5 ml of boiling lysis buffer as described (45). Both the medium and cells were carboxymethylated by a 1-h incubation in 50 mM iodoacetamide. 35 S-Labeled cathepsin D was immunoprecipitated with anti-cathepsin D antibody (a generous gift of Dr. Gregory Conner, University of Miami School of Medicine) and protein A-Sepharose (Pharmacia Biotech Inc.), eluted, and analyzed by reducing SDS-polyacrylamide gel electrophoresis (10% resolving gel).

Inhibition of PI 3Ј-Kinase by Wortmannin-
We determined the concentration of wortmannin required to inhibit the p85/ p110 PI 3Ј-kinase in intact cells. A 30-min incubation with 100 nM wortmannin reduced both basal and insulin-stimulated p85/ p110 activity by 70 -80% (data not shown). Basal PI 3Ј-kinase activity showed half-maximal inhibition at 30 nM wortmannin and maximal inhibition at 100 nM, with no additional decrease in activity at 1 M wortmannin (Fig. 1). Subsequent experiments used 100 nM wortmannin unless otherwise indicated.
Effect of Wortmannin on the Endocytic Pathway-Previous studies suggest a role for wortmannin-sensitive enzymes in the endocytic pathway (23). We examined the effect of wortmannin on transferrin receptor trafficking using TRVb-1 cells, which overexpress the human transferrin receptor. Treatment of TRVb-1 cells with wortmannin resulted in a depletion of cellsurface transferrin receptors ( Fig. 2A). Half-maximal depletion was observed at 30 nM wortmannin, and a maximal 40 -50% depletion was achieved at 100 nM wortmannin. No further decrease in cell-surface receptors was observed at concentrations as high as 1 M wortmannin (data not shown). The effect of wortmannin on cell-surface transferrin receptors corresponded with the dose response for inhibition of the p85/p110 PI 3Ј-kinase.
We then measured the uptake of 125 I-labeled transferrin during a 10-min incubation at 37°C in TRVb-1 cells. Pretreatment with wortmannin for 30 min decreased the accumulation of surface-bound and intracellular transferrin (Fig. 2B). The change in cell-surface transferrin binding reflected a redistribution of cell-surface transferrin receptors into an intracellular compartment, as total cellular transferrin binding was unaffected by wortmannin (data not shown). To determine the mechanism for this redistribution, we measured the rates of internalization and recycling in wortmannin-treated cells. The internalization rate constant for 125 I-labeled transferrin was calculated by the In/Sur method (36); a representative experiment is shown in Fig. 2C. In data pooled from nine experiments, the rate constant for transferrin receptor endocytosis was increased 55% by wortmannin (Table I). Similarly, the rate of 55 Fe uptake was increased 60% by treatment of cells with wortmannin (Table I). Although the calculated rate constants differed using the two methods, in both cases, treatment of cells with 100 nM wortmannin caused a similar increase in the internalization rate (Table I). Although there was a decrease in the absolute amount of internalized 125 I-labeled transferrin in wortmannin-treated cells (Fig. 2B), the internalization rate constants are calculated on a per receptor basis and are in fact increased by wortmannin.
In addition to the effects on internalization, the rate of transferrin recycling was reduced by treatment of cells with wortmannin. Cells were preloaded with 125 I-labeled transferrin in the absence of wortmannin (to prevent any differences in the concentration of intracellular ligand) and then incubated in 100 M desferoxamine in the absence or presence of wortmannin (a representative experiment is shown in Fig. 2D). In data pooled from seven experiments, we observed a 33% decrease in the rate of 125 I-labeled transferrin efflux from the wortmannintreated cells (Table I).
The steady-state distribution of transferrin receptors between the interior and surface of cells is determined by both the rates of endocytosis (k e ) and recycling (k r ). Changes in either parameter will be reflected in a new steady-state distribution. The increase in k e and the decrease in k r observed in wortmannin-treated cells predict a redistribution of transferrin receptors from the cell surface to the cell interior, resulting in a decrease in surface binding. The magnitude of the predicted change in surface binding, based on the ratio of the changes in k e and k r , was similar to the observed decrease in cell-surface 125 I-labeled transferrin binding (42% as opposed to 45%) ( Table  I).
We also measured the internalization of 125 I-labeled insulin in CHO/IR cells, which overexpress the human insulin receptor. Unlike transferrin receptors, which internalize constitutively, insulin receptors enter coated pits in a ligand-stimulated manner (46,47). Moreover, changes in the insulin receptor membrane-spanning domain that increase the rate of receptor lateral diffusion on the cell surface increase the rate of receptor endocytosis (48), suggesting that events prior to entry into coated pits are rate-limiting for insulin receptor endocytosis. We found that the rate constant for 125 I-labeled insulin uptake in CHO/IR cells was unaffected by wortmannin (k e ϭ 0.116 Ϯ 0.006 versus 0.125 Ϯ 0.01 in the absence and presence of wortmannin) (data not shown). Thus, the ligand-stimulated events that precede entry of insulin receptors into coated pits appear to be independent of wortmannin-sensitive enzymes.
Morphology of the Recycling Compartment-We used Cy3transferrin to visualize the intracellular itinerary of internalized transferrin receptors. In control cells incubated with Cy3transferrin for 3 min followed by a 2-min chase in the absence of transferrin, receptor-bound transferrin is present in early endosomes and sorting endosomes (Fig. 3A), compartments that have been previously shown to colabel with endocytosed low density lipoprotein (38). After an additional 10 min of chase, the receptor-bound transferrin is sorted away from the degradation pathway to the pericentriolar recycling compartment, located near the Golgi apparatus (Fig. 3B). The receptor and its ligand then return to the plasma membrane for another round of internalization. Exit from the pericentriolar compartment is the rate-limiting step in the recycling of transferrin and results in a concentration of the transferrin receptor in the pericentriolar region of the cell (Fig. 3B) (38).
In contrast, pretreatment of cells with wortmannin delayed the sorting of internalized transferrin to the recycling compartment. When wortmannin-treated cells were labeled with Cy3-transferrin and chased in wortmannin for 2 min, we saw an accumulation of ligand in numerous peripheral punctate structures similar to those seen in control cells (Fig. 3C). The intensity of labeling was decreased in wortmannin-treated cells relative to control cells, reflecting the decrease in cell-surface transferrin receptors caused by preincubation with wortmannin (Fig. 2B). After 12 min of chase, the distribution of labeled transferrin in wortmannin-treated cells was significantly different than in control cells, with a persistence of label in peripheral punctate structures (Fig. 3D). After 16 min of chase, pericentriolar concentration could be seen in some wortmannin-treated cells (data not shown). However, the labeled structures in these cells were more vesiculated and less uniform in appearance that those in control cells (data not shown; but see Fig. 4, A and B).
These data suggest that wortmannin reduces the rate of delivery of internalized transferrin from endosomes to the recycling compartment. Alternatively, the altered morphology of the labeled compartment in wortmannin-treated cells could reflect a change in the fidelity of sorting, with delivery of internalized transferrin to a different compartment. To differentiate these possibilities, we examined the degree to which transferrin internalized prior to wortmannin treatment mixed with transferrin internalized in the presence of wortmannin. Cells were preincubated to equilibrium by a 45-min incubation with Cy3-transferrin in the absence of wortmannin; under these conditions, label accumulates in the pericentriolar recycling compartment. The cells were then washed and chased in medium containing FITC-transferrin in the absence or presence of 100 nM wortmannin. During this incubation, Cy3-transferrin effluxes from the cells, whereas FITC-transferrin accumulates.
In untreated cells labeled with Cy3-transferrin and then chased with FITC-transferrin for 15 min, residual Cy3 labeling of the recycling compartment was easily observable (Fig. 4A). Moreover, FITC-transferrin could be seen to accumulate in the same intracellular compartment that contains Cy3-transferrin (Fig. 4C). In cells chased with FITC-transferrin in the presence of wortmannin for 15 min, the residual Cy3 labeling of the recycling compartment is much brighter than in control cells (Fig. 4B), indicating that wortmannin slowed the recycling of transferrin receptors back to the cell surface. Moreover, the vesiculation of the pericentriolar compartment in wortmannintreated cells is evident (Fig. 4B). However, the altered compartment is still clearly labeled with FITC-transferrin (Fig. 4D), indicating that this compartment is in communication with early endosomal structures containing newly internalized transferrin.
The wortmannin-induced slowing of efflux from the recycling compartment is even more striking after a 45-min chase. In control cells chased for 45 min, Cy3-transferrin in the pericen-  (36) or based on the accumulation of 55 Fe after incubation with 55 Fe-labeled transferrin (33). The rate of recycling was calculated as the inverse slope of a semilog plot of intracellular transferrin versus time (33). Percent change induced by wortmannin (100 nM) is given in parentheses. The predicted percent change in cell-surface transferrin binding is calculated based on the assumption that at equilibrium, the cell-surface and intracellular concentrations of transferrin receptors are determined by the ratio of their inward (endocytic) and outward (exocytic) rate constants. Thus, the predicted change in surface binding is determined by the percent change in the inward and outward rate constants. The calculation was performed using the k e derived from the In/Sur method. The observed percent change in surface binding induced by a 45-min treatment with 100 nM wortmannin is taken from experiments described in Fig. 2 triolar structures has been substantially washed out (Fig. 4E), replaced by FITC-transferrin that entered during the chase (Fig. 4G). In contrast, the punctate pericentriolar structures in wortmannin-treated cells are still brightly labeled with Cy3transferrin after 45 min (Fig. 4F). Again, Cy3-transferrin colocalizes with newly internalized FITC-transferrin (Fig. 4H), suggesting that the punctate pericentriolar structures in wortmannin-treated cells are in communication with early endosomal structures. The increased labeling of the recycling compartment during the FITC-transferrin chase in wortmannintreated cells (Fig. 4, compare C with D and G with H) reflects the wortmannin-induced redistribution of transferrin-receptor complexes into the cell interior ( Fig. 2A). Taken together, the data in Figs. 3 and 4 show that wortmannin has a pronounced inhibitory effect on both entry into and efflux from the recycling compartment.
Effects of Wortmannin on Lysosomal Delivery and Endosomal Acidification-The endocytic system internalizes both ligands and receptors by a common pathway, subsequently sorting receptors for recycling and ligands for lysosomal degradation. To determine whether wortmannin affects the sorting of ligands into a degradative pathway, we measured the degradation of endocytosed 35 S-labeled SFV, which is sorted from endosomes to the lysosome (49). In control cells, we observed a characteristic 10 -15-min lag period before the appearance of viral degradation products in the medium, which reached a maximum at ϳ90 min (Fig. 5). In contrast, the time lag before the detection of degraded SFV in the medium was longer (20 -30 min) in wortmannin-treated cells, and degraded SFV in the medium of wortmannin-treated cells was half that of control cells by 90 min. The differences in the rates of SFV degradation could not be explained by differences in the rates of internalization of surface-bound SFV, which were not affected by wortmannin (data not shown).
Although these data suggest that wortmannin had significant effects on the kinetics of late endosomal sorting, the altered kinetics of SFV degradation could be due to effects on the acidification of the endosomal lumen. We therefore examined the endocytic uptake of 35 S-labeled SFV using a conformationspecific antibody to the SFV E1 spike protein. The E1 spike protein undergoes a conformational change when exposed to the acidic conditions of the early endosome (pH 6.2 or lower), exposing an epitope that is recognized by monoclonal antibody E1a-1 (40). Untreated (Fig. 6A) or wortmannin-treated (Fig. 6B) cells were allowed to bind 35 S-labeled SFV at 4°C and then warmed rapidly to 37°C. At various times, the cells were lysed, and proteins were precipitated with either anti-spike antibody or conformation-specific antibody E1a-1. No difference in the total amount of cell-associated virus was detected in control versus wortmannin-treated cells (Fig. 6, A and B, lanes a). In both treated and untreated cells, a decrease in cell-associated virus due to lysosomal degradation was apparent by 15-30 min (Fig.  6, A and B, lanes d and e). Similarly, in both untreated and wortmannin-treated cells, the E1 spike protein was first recognized by antibody E1a-1 after 5 min at 37°C (Fig. 6, A and B,  lanes j), with levels of E1a-1-precipitable protein reaching maximal levels by 15 min (lanes l). In control experiments, treatment of cells with monensin completely blocked the conversion of E1 to the acidic conformation (Fig. 6, A and B, lanes o), whereas brief treatment of bound virus with acidic buffer (pH 5.5) caused maximal conversion of the E1 protein to its acidic conformation (lanes p). We conclude that endosomal acidification is not grossly affected by wortmannin.
Effect of Wortmannin on Biosynthetic Trafficking-We examined the transit and processing of VSV-G and the SFV E2 glycoprotein to determine the potential effects of wortmannin on biosynthetic vesicular trafficking. To measure the net rate of delivery of newly synthesized membrane glycoprotein to the plasma membrane in the absence or presence of wortmannin, we infected CHO cells with vesicular stomatitis virus, pulselabeled the cells with [ 35 S]methionine, and specifically labeled plasma membrane VSV-G by a cell-surface biotinylation protocol at 4°C. The kinetics of VSV-G arrival at the plasma membrane was consistent with previous studies (50) and was unaffected by 100 nM wortmannin (data not shown). We also examined the processing of the SFV E1 and E2 spike proteins, which occurs in a post-Golgi compartment (40,51). SFV-infected cells were incubated in the absence or presence of 100 nM wortmannin and pulse-labeled with [ 35 S]methionine. Processing of the SFV spike proteins was then analyzed by immunoprecipitation and SDS-polyacrylamide gel electrophoresis. Wortmannin had no effect on the cleavage of the E2 glycoprotein from its 62-kDa precursor form to its mature form (data not shown). Secretion of assembled virions was also unaffected by wortmannin (data not shown). Taken together, the experiments with VSV and SFV indicate that wortmannin does not affect transit of newly synthesized glycoproteins from the endoplasmic reticulum, through the Golgi apparatus, and to the plasma membrane.
Effect of Wortmannin on Lysosomal Sorting-Lysosomal hydrolases are sorted to the lysosome in CHO cells by virtue of the binding of mannose 6-phosphate residues to the mannose 6-phosphate receptor in the Golgi apparatus (52). A small proportion of lysosomal hydrolases are also secreted, and this alternative pathway is augmented during transformation of a number of cell types (53). In yeast, mutation of VPS-34 causes the missorting of vacuolar hydrolases to the secretory pathway (20). Moreover, two recent studies have suggested that wortmannin diverts newly synthesized cathepsin D into the secretory pathway (54,55). To determine if treatment of CHO cells with wortmannin caused a similar perturbation of lysosomal enzyme sorting, we measured the secretion of cathepsin D into the medium of cells treated with 100 nM wortmannin. Fig. 7A shows total intracellular cathepsin D and secreted cathepsin D from control or wortmannin-treated cells; the two gels were intentionally overexposed for identical lengths of time to permit visualization of secreted cathepsin D. 100 nM wortmannin had no effect on either the synthesis or stability of intracellular cathepsin D (Fig. 7A, upper panel), and secretion of cathepsin D was extremely low in both untreated and wortmannin-treated cells (lower panel). In five separate experiments, we detected no significant increases in cathepsin D secretion in cells treated with 100 nM wortmannin.
In contrast to the data obtained with 100 nM wortmannin, we saw a significant effect on the secretion of cathepsin D after treatment of cells with 2 M wortmannin. This concentration of drug is 20-fold higher than that required for maximal effects on p85/p110 PI 3Ј-kinase activity or transferrin receptor redistribution. Fig. 7B shows identical exposures of intracellular and secreted cathepsin D from control cells, cells treated with 2 M wortmannin, or cells treated with 50 M chloroquine. Treatment of cells with 2 M wortmannin had no significant effect on the synthesis or stability of intracellular cathepsin D, whereas treatment of cells with chloroquine increased the intracellular half-life of newly synthesized cathepsin D, presumably by inhibiting lysosomal degradation (Fig. 7B, upper panel). Notably, both 2 M wortmannin and 50 M chloroquine had pronounced effects on the secretion of newly synthesized cathepsin D. Secreted cathepsin D was undetectable in control cells (Fig. 7B, lower panel) using a fluorographic exposure time one-third of that used in Fig. 7A. In contrast, secreted cathepsin D was easily seen after 3 h in cells treated with 2 M wortmannin (Fig.  7B, lower panel). 2 M wortmannin induced a level of cathepsin D secretion similar to that seen in cells treated with chloroquine, a drug that is known to divert lysosomal enzymes into the secretory pathway (56).

Morphology of the Recycling Compartment in the Presence of Micromolar
Concentrations of Wortmannin-These data suggest that high doses of wortmannin induce changes in cellular physiology quite different than those observed at 100 nM wortmannin. We therefore examined the morphology of the recycling compartment loaded with Cy3-transferrin for 45 min in the presence of 1 M wortmannin (Fig. 8). A high dose of wortmannin induced the formation of a tubular network that extended throughout the cytoplasm. Moreover, there was a much more marked dispersal of the pericentriolar transferrinlabeled compartment and an increase in large punctate structures in the cell periphery. Thus, micromolar concentrations of TRVb-1 cells pretreated with carrier (A) or 100 nM wortmannin (B) were allowed to bind 35 S-labeled SFV on ice for 90 min. After washes to remove unbound virus, the cells were warmed to 37°C in the continued absence or presence of wortmannin. At various times, the cells were lysed, and 35 S-labeled viral proteins were immunoprecipitated with anti-spike antibody (A and B, lanes a-h) or antibody E1a-1 (lanes i-p), which recognizes the E1 spike protein after it has undergone an aciddependent conformational change. Lanes g and o show immunoprecipitates from incubations that included monensin (10 M) to block acidification. Lanes h and p show cells treated with pH 5.5 buffer for 1 min prior to lysis.

FIG. 7. Secretion of cathepsin D in wortmannin-treated cells.
A, TRVb-1 cells were incubated in 100 nM wortmannin or carrier for 30 min in methionine-free medium plus an additional 30 min with [ 35 S]methionine, followed by a chase in the continued absence or presence of wortmannin. At various times, the medium was harvested, the cells were lysed, and intracellular or secreted 35 S-labeled cathepsin D was immunoprecipitated and analyzed by SDS-polyacrylamide gel electrophoresis. B, cells were incubated in the absence or presence of 2 M wortmannin or 50 M chloroquine for 30 min and then processed as described above. The fluorographic exposure for B was one-third that used in A, which was intentionally overexposed to visualize secreted cathepsin D.
wortmannin induce changes distinct from those observed at concentrations that maximally inhibit the p85/p110 PI 3Ј-kinase. DISCUSSION Studies from several different laboratories suggest that PI 3Ј-kinase plays a role in endocytic or other trafficking events in mammalian cells. In this paper, we examined trafficking events in CHO cells treated with 100 nM wortmannin (27,57,58), which maximally inhibits the p85/p110 PI 3Ј-kinase in these cells. We find that 100 nM wortmannin causes a significant redistribution of the transferrin receptor due to changes in both the endocytic and recycling rate constants. Moreover, wortmannin induces a vesicularization of the recycling compartment labeled with fluorescent transferrin and reduces the rate at which transferrin enters and exits this compartment. Transferrin receptor trafficking and p85/p110 PI 3Ј-kinase activity show similar sensitivities to wortmannin, consistent with the involvement of p85/p110 or a related enzyme in the regulation of endocytosis and recycling.
The effects of 100 nM wortmannin appear to be specific for the endocytic system, as we could not detect any changes in biosynthetic trafficking of SFV or VSV spike proteins. We did detect an enhancement of the secretion of cathepsin D, a phenotype similar to that observed in yeast containing mutant VPS-34 alleles, but only at micromolar concentrations of wortmannin. These data are consistent with those of Balch and co-workers (54), who reported missorting of cathepsin D at 1-2 M wortmannin in CHO and normal rat kidney cells, but differ from those of Davidson (55), who observed effects on cathepsin D sorting at 100 nM wortmannin in K-532 cells. Presumably, the discordance reflects different sensitivities of the lysosomal enzyme-sorting apparatus in different cell types. However, our CHO cell data strongly suggest that the wortmannin-sensitive enzymes involved in lysosomal sorting are distinct from those involved in transferrin recycling and are unlikely to be the p85/p110 PI 3Ј-kinase. Furthermore, treatment of cells with 1 M wortmannin causes an extensive tubulation and vesiculation of the transferrin-labeled compartment, which is barely visible at 100 nM wortmannin. The disruption of transferrin receptor cycling in CHO cells by 100 nM wortmannin therefore defines a sorting event(s) that is pharmacologically distinct from the wortmannin-sensitive steps in lysosomal enzyme delivery.
The identity of the wortmannin-sensitive enzymes involved in transferrin endocytosis and recycling is not clear, as the specificity of wortmannin with regard to PI 3Ј-kinase is extremely problematic. Wortmannin was initially described as an inhibitor of the myosin light chain kinase, but it irreversibly inhibits PI 3Ј-kinase at 100-fold lower concentrations (26,27,57,58). Both ␣and ␤isoforms of the p110 PI 3Ј-kinase are half-maximally inhibited by low nanomolar concentrations of wortmannin, as is the recently cloned PI-specific PI 3Ј-kinase ; the sensitivity of the ␤␥-stimulated PI 3Ј-kinase is not yet clear (16,18,59). In addition, a PI 4-kinase important in the regulation of intracellular PI(4)P and PI(4,5)P 2 levels is half-maximally inhibited by 100 nM wortmannin and completely inhibited by 1 M wortmannin (32). This finding is particularly important; PI(4,5)P 2 has been implicated in the regulation of AP-2 adaptin complexes and cytoskeletal components such as profillin and plays a role in regulated secretion (60 -62). Moreover, PIP 2 has been shown to bind to pleckstrin homology domains, suggesting that disruption of PIP 2 synthesis could affect numerous cellular functions (63). The existence of multiple wortmannin-inhibited enzymes, which are regulated by different mechanisms and may exist in distinct intracellular locations, compromises the use of wortmannin as a putatively specific inhibitor of PI 3Ј-kinase. Studies currently in progress, involving the microinjection of inhibitory anti-p110 antibodies into TRVb-1 cells, will address the question of whether the wortmannin-sensitive enzyme involved in transferrin trafficking is the p85/p110 PI 3Ј-kinase.
The kinetic analysis of transferrin receptor trafficking and Semliki Forest virus degradation in wortmannin-treated cells demonstrates that three distinct steps in endocytic trafficking are affected by the drug. First, we see an increase in the rates of 125 I-labeled transferrin endocytosis and 55 Fe accumulation, both of which measure the rate of constitutive coated pitmediated internalization. It is therefore possible that a wortmannin-sensitive enzyme is important in regulating the early rate-limiting steps governing coated pits. We did not see changes in the rates of internalization of either 125 I-labeled insulin or surface-bound 35 S-labeled SFV. However, both insulin and SFV initially bind to receptors located in the microvillous regions of the cell membrane and then move to coated pits (47,49). Furthermore, the rate of insulin receptor endocytosis appears to be limited by the rate of receptor diffusion through the plasma membrane (48) and therefore is probably not a reflection of the absolute rate of coated pit endocytosis.
A second site of wortmannin action may be the sorting endosome. In CHO cells, recycled ligands are sorted from material destined for degradation by an iterative fractionation process in which the sorting endosome matures into a late endosome (38). Consequently, if wortmannin affects the maturation of the sorting endosome, then both transport of the recycling membrane to the pericentriolar recycling compartment and delivery of material to lysosomes could be affected. Consistent with this possibility, wortmannin slows the movement of internalized receptor-bound transferrin from endosomes to the pericentriolar recycling compartment as well as the degradation of internalized SFV, which requires delivery to the lysosome. These changes do not appear to be due to defects in luminal acidification, which is known to affect intracellular trafficking of the transferrin receptor (64,65), as the acid-dependent conformational changes in the SFV E1 spike protein are normal in wortmannin-treated cells. Although our data could be explained by a common wortmannin-sensitive step in the sorting compartment, we cannot rule out the possibility of independent wortmannin-sensitive steps involved in transit of ligands between the sorting endosome and the recycling compartment or the lysosome. These steps could involve Rab5 or another Rab isoform, as has been recently suggested by Li et al. (66).
A third potential site of action is the recycling compartment, defined as the pericentriolar structure that is labeled by fluo- rescent transferrin, but not by ligands like low density lipoprotein that are destined for lysosomal degradation (38). Transport from the pericentriolar recycling compartment is the ratelimiting step in transferrin recycling (38). The morphology of the recycling compartment is significantly altered by 100 nM wortmannin, and efflux of transferrin out of the compartment is slowed in wortmannin-treated cells. Although the mechanism by which wortmannin affects the recycling compartment is not yet clear, there is precedent for the regulation of transit from the recycling compartment. Thus, transferrin receptor recycling from the compartment to the plasma membrane is slowed by either alkalinization of the lumen or oligomerization of transferrin receptors in the presence of multivalent transferrin (64,67).
The redistribution of transferrin receptors caused by wortmannin is similar to that seen in a mutant liver line (Trf1) that expresses only State 1 receptors (68,69), defined by Weigel (68) as receptors resistant to down-regulation by metabolic and cytoskeletal inhibitors. However, residual surface receptors in Trf1 cells show unchanged endocytosis and accelerated recycling (69). Receptor redistributions in wortmannin-treated cells therefore involve different mechanisms than those operative in Trf1 cells.
For the most part, the studies presented here examine trafficking under basal conditions of PI 3Ј-kinase activity. Some activation of PI 3Ј-kinase may occur during insulin-stimulated endocytosis in CHO/IR cells, although the studies were performed at insulin concentrations well below the half-maximal concentrations required for activation of PI 3Ј-kinase (70,71). Moreover, we have previously shown that mutant insulin receptors that do not activate PI 3Ј-kinase still undergo normal ligand-stimulated endocytosis (72,73). Siddle and co-workers (74) recently reported that wortmannin blocks insulin-stimulated up-regulation of the transferrin receptor in 3T3-L1 cells, and Joly et al. (23) have shown that treatment of cells with wortmannin or mutation of PI 3Ј-kinase-binding sites alters the post-endocytic routing of internalized platelet-derived growth factor receptors. Both of these studies examined cells under conditions of mitogen-induced PI 3Ј-kinase activation and may not be strictly comparable to the studies presented here.
In summary, we have evaluated the effects of wortmannin on vesicular trafficking in CHO cells. We find that concentrations of wortmannin that maximally inhibit the p85/p110 PI 3Јkinase have pronounced effects on three distinct steps in the endocytic cycle: internalization, transit from early endosomes to the recycling and degradative compartments, and transit from the recycling compartment back to the cell surface. A substantial change in the morphology of the sorting compartment is also seen under these conditions. However, we find no effects on the biosynthetic trafficking of viral glycoproteins or lysosomal enzymes at these concentrations. In contrast, wortmannin-induced secretion of cathepsin D was observed at significantly higher doses of wortmannin. We have therefore identified two thresholds of wortmannin-sensitive events in vesicular sorting, suggesting that the effects on endocytic and lysosomal trafficking are mechanistically distinct.