Inhibition of Calcium-independent Mannose 6-Phosphate Receptor Incorporation into trans-Golgi Network-derived Clathrin-coated Vesicles by Wortmannin*

The transport of pro-cathepsin D from thetrans-Golgi network (TGN) to the endosomal pathway is dependent on binding to the calcium-independent mannose 6-phosphate receptor (ci-M6PR), which is incorporated into TGN-derived clathrin-coated transport vesicles (CCVs). Inhibition of this transport step by wortmannin has led to the proposal that it is dependent upon a phosphoinositide 3-kinase activity necessary for ci-M6PR recruitment into TGN-derived CCVs or in the formation of those vesicles (Brown, W. J., DeWald, D. B., Emr, S. D., Plutner, H., and Balch, W. E. (1995) J. Cell Biol. 130, 781–796; Davidson, H. W. (1995) J. Cell Biol. 130, 797–806). In this study we have addressed the effect of wortmannin on the TGN step of the ci-M6PR cycle. CCVs from K562 cells, pretreated or not with 250 nm wortmannin, were purified on equilibrium density gradients. Quantification of TGN-derived CCVs, assessed by γ-adaptin content in purified vesicle fractions, showed that the formation of the vesicles was only marginally decreased after 20 min of treatment with the drug, while for the same wortmannin treatment, the amount of ci-M6PR recruited into those vesicles was decreased by 70% compared with control. At a later time point (2 h), a reduction in the amount of γ-adaptin in CCV fractions was also observed. These findings demonstrate that inhibition of ci-M6PR recruitment into CCVs but not of vesicle formation is the primary reason for the observed defect in cathepsin D transport following wortmannin treatment.

For many years, intracellular membrane traffic has been considered to comprise protein-regulated events governing vesicle formation and fusion, while a passive role has been generally assumed for phospholipids. During the same period, phosphoinositides have been widely studied for their participation in transducing inputs received by cell surface receptors. For example, activation of G-protein-coupled receptors may result in stimulation of phospholipase C hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP 2 ), 1 leading to the production of the second messengers inositol 1,4,5-trisphosphate and diacylglycerol. Phosphoinositide 3-kinases (PI3Ks) have also been extensively studied for their role in cellular proliferation and differentiation resulting from their interaction with growth factor receptors (1). More recently, different types of PI3K have been isolated and cloned, which may play other roles in the cell, including the regulation of membrane traffic (2). Together with separate observations linking the small GTPase ADP-ribosylation factor, which is involved in budding of coated vesicles, to activation of phospholipase D, these studies have promoted widespread interest in the role of lipids in governing aspects of membrane traffic (3).
The use of wortmannin, a specific inhibitor of PI3K in the low nanomolar range (4,5), has allowed the requirement for PI3K activity in many cellular processes to be tested. Many aspects of membrane traffic are influenced by application of the drug, including fluid phase endocytosis (6 -8), early endosome fusion (7,9,10), transferrin receptor recycling (10,11), platelet-derived growth factor (PDGF) receptor down-regulation (12,13), and delivery of pro-cathepsin D from the trans-Golgi network (TGN) to the endocytic pathway (14,15).
Two major PI3K classes have now been characterized. The first class includes the wortmannin-sensitive heterodimeric mammalian PI3Ks represented by the p85/p110 kinases, which link to receptors with tyrosine kinase activity (1), and the p101/p117 or p101/p120 PI3Ks, which are activated by heterotrimeric G-protein ␤␥-subunits (16). These heterodimeric PI3Ks exhibit an in vivo substrate preference for PIP 2 (17). The second class, specific for phosphatidylinositol (PtdIns), is exemplified by the wortmannin-resistant Vps34p from Saccharomyces cerevisiae (18) and a recently identified wortmannin-sensitive mammalian PtdIns 3-kinase (19).
Among membrane traffic pathways involving the phosphoinositide machinery, the transport of vacuolar enzymes from the TGN to the yeast vacuole was the first in which a PI3K activity was implicated. This result was based on the observation of abnormal secretion of carboxypeptidase Y in a S. cerevisiae strain mutated for a gene encoding Vps34p, subsequently identified as the yeast PtdIns 3-kinase (18,20,21). In the wild type strain, the enzyme is routed to the vacuolar compartment. On the basis of wortmannin sensitivity, a role for a PI3K was reported in mammalian cells for the calcium-independent mannose 6-phosphate receptor (ci-M6PR)-mediated transport of lysosomal protease cathepsin D (14,15). This pathway is analogous to the TGN-vacuole pathway in yeast. In these studies the pro-cathepsin D was observed to be secreted into the extracel-lular medium following wortmannin treatment, instead of being delivered to the early endosomal compartment and then processed during its transport toward the lysosome. The authors proposed that wortmannin was inhibiting a PI3K activity necessary for a transport step in the ci-M6PR cycle, most likely involved at the exit from the TGN (14). However, no direct evidence was obtained to determine if the PI3K activity was required in the sorting of ci-M6PR into TGN-derived clathrincoated vesicles (CCVs) or, alternatively, if the formation of the vesicles was itself dependent on PI3K activity. The answer to this question may have a wider significance as receptor sorting events in other subcellular compartments have also been shown to be wortmannin-sensitive. These include transferrin receptor recycling (10,11), trafficking of the GLUT4 transporter (22), and the traffic of PDGF receptor through endosomal compartments (12,13).
In the present study we have developed a method to investigate the role of the PI3K activity in the ci-M6PR-mediated transport of lysosomal enzymes from the TGN to the early endosome in K562 cells. The method is based on an established protocol of CCV purification, which takes advantage of their characteristic density due to the presence of their protein coat (23). Moreover, the coat of TGN-derived CCVs contains a specific adaptor protein, ␥-adaptin, allowing us to distinguish them from co-isolated plasma membrane-derived CCVs, which contain ␣-adaptin rather than ␥-adaptin. Purification of CCVs from K562 cells, pretreated or not with wortmannin, and quantification of the relative amounts of ␥-adaptin and ci-M6PR in the vesicle fractions indicated that wortmannin was not primarily affecting CCV formation at the TGN, but that the vesicles produced were depleted of ci-M6PR. We conclude that a wortmannin-sensitive PI3K is implicated in ci-M6PR sorting into TGN-derived CCVs but not in the vesicle formation step itself.
Isolation of CCVs from K562 Cells-CCVs were isolated from K562 cells by a modification of the method of Woodman and Warren (24). 1.6 ϫ 10 8 cells were washed twice (700 ϫ g for 6 min) with 37°C phosphate-buffered saline (PBS) and incubated for 3 h in 20 ml of MEM supplemented with Pen/Strep, 2% (v/v) 1 M Hepes, pH 7.3, 5% (v/v) FBS, and 0.5 mCi of Tran 35 S-Label. Approximately 1.3 ϫ 10 9 nonlabeled cells were then combined with 35 S-labeled cells, washed with 37°C PBS, and resuspended at 2.5 ϫ 10 6 cells/ml in 37°C RPMI 1640 medium supplemented with Pen/Strep, 1% (v/v) FBS, and 2% (v/v) 1 M Hepes, pH 7.3. Cells were finally split into three flasks and incubated at 37°C with or without 250 nM wortmannin for 20 and 120 min. The following steps were performed at 4°C. Cells were washed twice with vesicle buffer (10 mM MES, 75 mM KOAc, 2 mM EGTA, 1 mM MgCl 2 , 140 mM sucrose, 1 mM dithiothreitol, 1 g/ml pepstatin A, 2 g/ml aprotinin, 2 g/ml leupeptin A, pH 6.6, supplemented with 250 nM wortmannin when used for wortmannin-treated cells) and resuspended into 5 ml of vesicle buffer. Cells were broken open using a stainless steel homogenizer (EMBL cell cracker) by passing 15 times through a 8.02-mm bore containing a 8.002-mm diameter ball. Post-nuclear supernatants were prepared by a 10-min centrifugation at 1,000 ϫ g, treated for 30 min with 50 g/ml ribonuclease A, and centrifuged for 40 min at 7,000 ϫ g to prepare post-mitochondrial supernatants. For each condition, postmitochondrial supernatant was applied to the top of two 10-ml continuous deuterium oxide gradients of 9 -90% (w/v) 2 H 2 O in vesicle buffer and centrifuged for 35 min at 45,000 ϫ g. Ten 1-ml fractions were collected from the top of each gradient. Fractions 2-7 were combined, diluted up to 24 ml with vesicle buffer, and centrifuged for 40 min at 100,000 ϫ g. The pellets were resuspended in vesicle buffer, pooled for each condition, applied to the top of one 10-ml continuous gradient of 2% (w/v) Ficoll 400/9% (w/v) 2 H 2 O ϩ 20% (w/v) Ficoll 400/90% (w/v) 2 H 2 O in vesicle buffer, and centrifuged for 12 h at 76,000 ϫ g. Ten 1-ml fractions were collected from each gradient, precipitated with trichloroacetic acid, resuspended in reducing SDS-PAGE sample buffer, and incubated for 10 min at 90°C. Finally, fractions were run on 10% SDS-PAGE gels, which were either Coomassie-stained and dried before quantification of 35 S label using a PhosphorImager (Molecular Dynamics) or transferred to nitrocellulose membrane. Immunoblots of ␥-adaptin and ci-M6PR proteins were quantified by phosphorimaging (Molecular Dynamics) after 125 I-protein A affinity binding (67 nCi/ml). Alternatively immunoblots of ␣-adaptin, ␥-adaptin, and ci-M6PR were detected using a SuperSignal CL-HRP substrate system (Pierce & Warriner, Chester, UK) following incubation with HRP-coupled secondary antibody and quantified with IPLab Gel (Signal Analytics) from the Hyperfilm-ECL scan (ScanMaker plug-in for Adobe Photoshop, Microtek Lab).
Analysis of 2 H 2 O Rate Sedimentation Gradients-Post-mitochondrial supernatants were prepared as above and loaded on a continuous rate sedimentation gradient of 9 -90% 2 H 2 O in vesicle buffer. After 35 min of centrifugation at 45,000 ϫ g, 10 1-ml fractions were collected from the top of the gradient, diluted 1.5-fold with vesicle buffer, and centrifuged for 45 min at 100,000 ϫ g. Pellets were resuspended in 100 l of reducing SDS-PAGE sample buffer and run on a 10% SDS-PAGE gel, and proteins were transferred to nitrocellulose membrane. Immunoblots for ␥-adaptin protein were detected using a SuperSignal CL-HRP substrate system following incubation with HRP-coupled secondary antibody and quantified as above.

RESULTS
Wortmannin Inhibits ci-M6PR-mediated Transport of Procathepsin D from TGN to Endosomal Compartment-Pro-cathepsin D is synthesized in human cells as a 53-kDa proenzyme, which is modified by the addition of M6P residues that provide the recognition signal for ci-M6PR-mediated sorting at the TGN into CCVs (25). Pro-cathepsin D is then delivered to endosomal compartments, where it is cleaved to form the 47-kDa intermediate form and then the 31-kDa mature form (26). As shown previously for rat clone 9 hepatocytes, NRK, CHO, MDBK (14), and K562 cells (15), wortmannin treatment triggers a mistargeting of pro-cathepsin D. Following a 10-min pulse radiolabeling with Tran 35 S-Label, pro-cathepsin D remained associated with cells kept at 4°C for 120 min and was not processed (Fig. 1, no chase). When cells were incubated for 120 min at 37°C in the presence of extracellular M6P and in the absence of wortmannin most of the pro-cathepsin D was normally matured and remained associated with the cell, while only a small amount was secreted in the medium (Fig. 1, ctrl). Incubation with 250 nM wortmannin promoted pro-enzyme secretion into the medium, while very few mature or intermediate forms of cathepsin D were found associated with the cell or in the medium (Fig. 1, wrt). A minor intracellular band corresponding to the expected molecular weight of the intermediate form is apparent in control and wortmannin-treated cells at similar levels. The significance of this band is unclear, but it is important to note that no processed form is secreted into the medium, indicating that the route taken by the majority of the cathepsin D molecules does not dispose them to processing. Several studies have shown that endocytic trafficking and processing of externally applied cathepsin D is not inhibited by wortmannin (14,15,27), so it is highly unlikely that cathepsin D is secreted via an endosomal compartment.
Analysis of Rate Sedimentation and Equilibrium Gradients-The presence of proteins coating the vesicle lipid bilayers confers upon vesicles a high density, which can be exploited for the preparation of highly enriched fractions of CCVs on Ficoll/ 2 H 2 O equilibrium density gradients. Post-mitochondrial supernatants containing ϳ35% of total radioactivity incorporated in cells (data not shown) were first applied to 10-ml 9 -90% 2 H 2 O velocity gradients (see "Experimental Procedures"). Immunoblot analysis of gradient fractions for ␥-adaptin (Fig. 2a) showed that membrane associated ␥-adaptin was concentrated in fractions 2-7 (1 ml, fraction 1 ϭ top). This is in agreement with the ATP-dependent immunoprecipitation profile of 125 Itransferrin observed previously by Woodman and Warren (24) in the same gradient. Moreover, quantification of ␥-adaptin immunoblots showed no significant difference in ␥-adaptin distribution regardless of wortmannin treatment (Fig. 2b). Fractions 2-6, representing ϳ6% of total radioactivity in cells (data not shown) were pooled for further purification. Fig. 3a demonstrates typical radioactivity profiles after fractions 2-6 were applied on 10-ml Ficoll/ 2 H 2 O equilibrium density gradients. Similar radioactivity profiles were obtained through the gradi-ent regardless of wortmannin treatment. Two peaks were observed accounting for contaminating membranes (lightest peak, fractions 1-4) and for CCVs (heaviest peak, fractions 6 -8) as demonstrated previously on the basis of their fusion activity (23). CCV peak in control cells represented ϳ1% of total radioactivity associated with cells as estimated previously by Pearse (28). Fig. 3b shows a typical autoradiography of fractions collected from the Ficoll/ 2 H 2 O gradient and analyzed by SDS-PAGE. Fractions 6 -8 contained a major band at 180 kDa and other bands at ϳ100 kDa and 45-50 kDa consistent with the presence of clathrin heavy chain and the adaptor complexes, respectively (28).
Wortmannin Causes a Decrease in ci-M6PR Content of CCVs Formed at the TGN-To determine if wortmannin treatment was inhibiting CCV formation at the TGN or ci-M6PR recruitment in those vesicles, we analyzed ␥-adaptin and ci-M6PR contents in fractions collected from the Ficoll/ 2 H 2 O gradients. The level of ␥-adaptin is the critical indicator of TGN-derived clathrin-coated vesicles and the ratio of ␥-adaptin over total Tran 35 S-Label radioactivity (␥-adaptin/total 35 S) in the gradient is the appropriately normalized ratio for judging their relative abundance. The ratio of ci-M6PR/␥-adaptin is the critical indicator of receptor content in TGN-derived CCVs. Clathrin heavy chain, accounting for both plasma membrane-and TGN-derived CCVs and ␣-adaptin (only plasma membranederived CCVs), has also been analyzed in those fractions to provide accessory information regarding the effect of wortmannin on CCV formation in K562 cells (Fig. 5). Fig. 4 shows results obtained from the analysis of a typical experiment. ␥-Adaptin (Fig. 4a) and ci-M6PR (Fig. 4b) in fractions 5 to 8 were detected from immunoblots using 125 I-protein A affinity binding and subsequent phosphorimaging. ␣-Adaptin (Fig. 4c) was detected from immunoblots using a SuperSignal CL-HRP substrate system. Clathrin heavy chain content was directly determined by PhosphorImager analysis of the 35 S signal (Fig.  4d). Intensity measurements of immunoblots (Fig. 4, a-c) or 35 S label (Fig. 4d) were normalized by ratioing over total Tran 35 S-Label radioactivity measured in fractions 1-10 of the Ficoll/ 2 H 2 O gradients (Fig. 3a). Normalized results obtained The ␥-adaptin/total 35 S ratio, which gives a measure for the amount of CCVs formed at the TGN due to the specific localization of ␥-adaptin in that compartment, was slightly decreased to 81% of control after 20 min of treatment with wortmannin but significantly decreased to 46% after 2 h of incubation with the drug (Fig. 5). Quantification of ci-M6PR immunoblots revealed a much larger decrease in ci-M6PR/total 35 S ratio than in ␥-adaptin/total 35 S; normalized ratios were 28% and 7% compared with control cells after 20 and 120 min of treatment, respectively. Moreover, ci-M6PR/␥-adaptin ratios were calculated as an indicator for the level of ci-M6PR in TGN-derived CCVs. Taking this ratio as 100% in control cells, ci-M6PR/␥-adaptin decreased to 34% and 15% after 20 min and 2 h of incubation with wortmannin, respectively, indicating a significant decrease in ci-M6PR incorporation into TGN-derived CCVs.
Densitometry measurement of immunoblots for ␣-adaptin in CCV fractions showed that ␣-adaptin/total 35 S ratio was increased to 140% of control following 20 min of treatment with wortmannin, in accord with previous observations showing a 1.4-fold increase in transferrin receptor internalization into K562 cells following wortmannin treatment (10). After 120 min of wortmannin treatment, this elevation of endocytic vesicles is no longer apparent and there is a decrease in levels of clathrin heavy chain in line with a reduction in both types of CCVs. Results presented in Fig. 5 are derived from a single preparation obtained from cells incubated under the three conditions and for which all but ␣-adaptin levels were quantified by phosphorimaging. These results were confirmed by two other experiments in which ␥-adaptin and ci-M6PR levels in CCV fractions were quantified from immunoblots by densitometry following development with a SuperSignal CL-HRP substrate system. DISCUSSION Previous studies (14,15) have shown that the ci-M6PRmediated transport of lysosomal pro-cathepsin D from the TGN to the endosomal compartment is inhibited in cells treated with the PI3K specific inhibitor wortmannin. In wortmannintreated cells, the pro-enzyme is abnormally secreted into the extracellular medium instead of being delivered to endosomal compartments, where it would be processed (see also Fig. 1). It has also been shown that wortmannin does not affect the addition of the M6P recognition signal to the pro-cathepsin D, the binding of M6P-modified pro-cathepsin D to the ci-M6PR in the Golgi compartment (15), or the transport of the ci-M6PR through the endocytic pathway (14,15). It was therefore concluded that wortmannin inhibits a PI3K involved in ci-M6PR dependent traffic from the prelysosomal compartment (PLC) to the early endosome via the TGN. However, it remained unresolved, precisely which function is affected by wortmannin. Four models can be proposed (Fig. 6, i-iv). PI3K activity could be required (i) for sorting of the ci-M6PR into TGN-derived CCVs, (ii) directly in CCV formation at the TGN, (iii) for fusion of TGN-derived CCVs with endosomes, or (iv) for recycling of the ci-M6PR from the PLC to the TGN.
In this study we have first repeated an experiment of Davidson (15) and obtained similar results in terms of pro-cathepsin D secretion into the extracellular medium following wortmannin treatment of K562 cells, indicating that in our hands the cells are responding similarly (Fig. 1). We have then per- formed experiments to discriminate the four models proposed above by preparing CCVs from wortmannin-treated and untreated cells and then quantifying their ␥-adaptin and ci-M6PR contents. By ratioing the quantified protein values for each condition, we are able to obtain internally consistent values independent of variation in the load applied to the final gradient, much as ratiometric fluorescence measurements reflect ion concentrations independent of fluorophore loading.
Model ii was originally proposed by Emr and co-workers (20), on the grounds that phosphorylation of lipid headgroups would increase the average headgroup area relative to the acyl chain area and promote membrane curvature necessary for budding according to the bilayer couple hypothesis of Sheetz and Singer (29). This model could also be proposed from another point of view, following the recent observation that binding of Golgi coat proteins to lipid membranes is sensitive to the lipid composition of those membranes (3). We find that, following a 20-min incubation with wortmannin, the amount of ␥-adaptin in purified CCV fractions was only slightly decreased compared with control cells (␥-adaptin/total 35 S), indicating that CCV formation at the TGN was largely unaltered by the drug. These observations are incompatible with model ii.
Model iii could be proposed, particularly on the basis that PI3K is known to regulate the homotypic fusion properties of early endosomes in vitro (7,9,10) and by extension might also regulate their heterotypic fusion with transport vesicles. The fact that the absolute quantity of TGN-derived vesicles was not increased by wortmannin treatment argues against model iii, which predicts an accumulation of the vesicles. Moreover, following a 20-min treatment with wortmannin, the recruitment of ci-M6PR into CCVs was greatly diminished compared with control cells (Fig. 5). This demonstrates that the original population of vesicles has been replaced by a population depleted in ci-M6PR, which is inconsistent with a simultaneous shutdown in vesicle formation and vesicle consumption.
Finally, the fact that the depletion of ci-M6PR from CCVs occurred after only 20 min of treatment with wortmannin argues against an inhibition of ci-M6PR recycling from the PLC to the TGN (model iv) because the half-time for recycling has been estimated between 1 and 2 h in K562 cells (30). In most cell lines, the steady state distribution of ci-M6PR shows accumulation in the TGN, so exit from the TGN should have a relatively long half-time. Most importantly, a recent study by Nakajima and Pfeffer (27) has shown that wortmannin fails to inhibit the recycling of ci-M6PR in K562 cells under conditions where pro-cathepsin D processing is inhibited. In addition, the drug had no effect on an in vitro assay that reconstitutes the transport step between late endosomes and the trans-Golgi network (27). Brown et al. have reported that the PLC compartment was depleted in ci-M6PR after long exposure to wortmannin while ci-M6PR was still present in the TGN (14). They have also shown that the block to cathepsin D transport from the TGN occurs shortly after application of wortmannin. We have found that the distribution of ci-M6PR on a 0.7-1.3 M sucrose equilibrium gradient was unchanged by wortmannin treatment and co-distributed with a TGN46 (31) peak (data not shown). Taking all these observations together, we consider model iv unlikely at least for K562 cells.
Our data on K562 cells are therefore most consistent with model i, in which the wortmannin treatment inhibits the recruitment of the ci-M6PR into CCVs derived from the TGN. This result is in agreement with the observations by Shpetner et al. (32), who have shown by immunofluorescence that ␥-adaptin distribution in HepG2 cells was not affected by 10 min of wortmannin treatment. We have also made the same observation with HeLa cells after more prolonged treatment (data not shown).
The PI3K involved in ci-M6PR sorting is still unidentified. Involvement of a PI3K in protein sorting at the TGN was originally reported in a yeast strain mutated in the VPS34 gene encoding for the yeast, wortmannin-insensitive PtdIns 3-kinase, Vps34p (18). A defect in that gene was accompanied by an abnormal secretion of carboxypeptidase Y in the extracellular medium instead of being delivered, through receptor-mediated transport, to the yeast lysosome-like vacuole (33). It has also been shown that the active form of this kinase forms a complex with Vps15p, a serine/threonine kinase, which participates in the recruitment of Vps34p to the TGN membrane and to its activation (20,34). Vps34p appears to be the only PI3K present in yeast, and can only use PtdIns as substrate. Waterfield and co-workers have recently cloned a mammalian PtdIns 3-kinase (19,35), which represents the best candidate for a role in ci-M6PR sorting. This kinase is 1) a PtdIns 3-kinase sharing relatively high homology with Vps34p, 2) associated with and FIG. 5. CCVs formed at the TGN are depleted in ci-M6PR following wortmannin treatment. Densitometry measurements of SDS-PAGE autoradiography (clathrin heavy chain) or immunoblots detected with 125 I-protein A (␥-adaptin, ci-M6PR) or using a SuperSignal CL-HRP substrate system (␣-adaptin) were normalized over total 35 S counted in fractions 1-10 of equilibrium density gradients. Ratios obtained for wortmannin-treated cells were then expressed as a percentage of those obtained for control cells. ci-M6PR content in ␥-adaptin containing vesicles was also normalized to control cells by ratioing ci-M6PR/␥-adaptin for wortmannin-treated cells over ci-M6PR/␥-adaptin obtained for control cells. Bars represent ratios for ␥-adaptin/total 35 S (a), ci-M6PR/total 35 S (b), ci-M6PR/␥-adaptin (c), ␣-adaptin/total 35 S (d), and clathrin heavy chain/total 35 S (e). Conditions were: no wortmannin (ctrl), 20 min (20 min), or 120 min (120 min) with 250 nM wortmannin. Results presented in this figure were confirmed by two other independent experiments, where quantification of ␥-adaptin and ci-M6PR were done from immunoblots detected using a SuperSignal CL-HRP substrate system instead of 125 I-protein A.
FIG. 6. Models proposed for the wortmannin inhibition of ci-M6PR pathway. Wortmannin could inhibit: i, the sorting of the ci-M6PR (filled ovals) at the TGN; ii, TGN-derived CCV formation; iii, fusion of those vesicles with the early endosomal compartment (EE); or iv, recycling of the ci-M6PR from the prelysosomal compartment (PLC) to the TGN. activated by another protein (p150) with homology to Vps15p, 3) only able to use PtdIns as a substrate, and 4) wortmanninsensitive. The enzyme is distinct from other mammalian PI3Ks, which couple to activated growth factor receptors or G-protein-coupled receptors and prefer PIP 2 as substrate in vivo.
Phosphorylation of the ci-M6PR is tightly associated with its exit from the TGN via the CCV route (36). It is tempting to speculate that the serine/threonine kinase activity of the p150 adaptor subunit of the PtdIns 3-kinase plays a role in this process, although this activity would not be expected to be inhibited directly by wortmannin. 3-Phosphorylated phosphoinositides may themselves regulate this or another crucial serine/threonine kinase, just as phosphatidylinositol 3,4-bisphosphate has recently been shown to specifically activate the c-Akt kinase (37). The fact that CCV formation continues as normal, despite severe depletion of ci-M6PR in the vesicles themselves, suggests that they can form independently of ci-M6PR incorporation, and that other proteins (e.g. LAMP-1; Ref. 38) interacting with ␥-adaptin might traverse the TGN to endosome route irrespective of wortmannin treatment. This would be consistent with the observation that alkaline phosphatase in a Vps34 mutant strain of yeast reaches the vacuole as normal (20). Specific sorting of receptors coupled to PI3Ks may be a general feature shared by ci-M6PR and the plasma membrane receptors for PDGF (13) and colony-stimulating factor (39). For each case it remains to be resolved precisely how PI3K activity is translated into a sorting signal.