Expression of auxilin or AP180 inhibits endocytosis by mislocalizing clathrin: evidence for formation of nascent pits containing AP1 or AP2 but not clathrin

During receptor-mediated endocytosis, clathrin triskelions polymerize and form clathrin-coated pits on the plasma membrane that then invaginate into the cell to form clathrin-coated vesicles (Keen, 1990; Pearse and Robinson, 1990). Similar clathrin-coated pits also form on the trans-Golgi network. In addition to clathrin and receptors, these pits contain assembly proteins (APs) that catalyze the polymerization of clathrin triskelions and in some cases bind the receptors localized in the clathrin-coated pits. A number of different APs have been described. AP1, AP2, AP3 and AP4 are multimeric subunit complexes of about 270 kDa (Keen, 1990; Robinson and Kreis, 1992; Simpson et al., 1997; Dell’Angelica et al., 1999; Hirst et al., 1999); AP1 occurs on the trans-Golgi network, AP2 on the plasma membrane, AP3 on both the trans-Golgi membrane and endosomes, and AP4 on perinuclear structures. AP180 (Ungewickell and Oestergaard, 1989) and auxilin (Ahle and Ungewickell, 1990) are neuronal specific APs that consist of single subunits of 92 kDa and 100 kDa, respectively; CALM and GAK, which have recently been described, are the non-neuronal homologs of AP180 (Dreyling et al., 1996) and auxilin (Kanaoka et al., 1997; Greener et al., 2000), respectively. Finally β-arrestin is specifically involved in recruiting β-adrenergic receptors to clathrin-coated pits but unlike APs it does not induce clathrin polymerization (Goodman et al., 1996; Goodman et al., 1997).

In addition to APs a number of other proteins have been discovered that are involved in the formation, invagination, and pinching off of clathrin coated vesicles including dynamin, amphiphysin, epsin, eps15, endophilin, syndapin I and the small GTPase protein, Rab5-GDI (Van der Bliek et al., 1993; Takei et al., 1995; David et al., 1996; Chen et al., 1998; Tebar et al., 1996; McLauchlan et al., 1998; Ringstad et al., 1999; Schmidt et al., 1999; Qualmann et al., 1999). In addition, rho, rac, phospholipids, and actin are involved in clathrin coat assembly and receptor recruitment (Rapoport et al., 1997; Takei et al., 1998; Lamaze et al., 1996; Lamaze et al., 1997; Munn et al., 1995; Gaidarov et al., 1999) as is dephosphorylation of many of the proteins involved in formation of clathrin-coated pits (Wilde and Brodsky, 1996; Slepnev et al., 1998). Finally, after they pinch off, the clathrin-coated vesicles are uncoated in an ATP dependent process by Hsc70 and its partner proteins, auxilin or cyclin G-associated kinase (GAK). These partner proteins not only assemble clathrin but also have J-domains that enable them to interact with Hsc70 (Prasad et al., 1993; Ungewickell et al., 1995; Jiang et al., 1997; Greener et al., 2000). Recently, Cremona et al. (Cremona et al., 1999) have shown that hydrolysis of PIP2 by synaptojanin is also important for uncoating in vivo.

An important question regarding regulation of uncoating by Hsc70 is why premature uncoating of clathrin-coated pits does not occur before they pinch-off to form clathrin-coated vesicles. Since the J-domain proteins auxilin and GAK are critical for uncoating it seemed possible that the level of auxilin or GAK present in the cell would not only affect the rate and extent of uncoating of clathrin-coated vesicles but also whether or not clathrin-coated pits were uncoated by Hsc70; if excess auxilin or GAK caused premature uncoating of clathrin-coated pits before they pinched off to form clathrin-coated vesicles, it might markedly inhibit clathrin-mediated endocytosis. On the other hand, quantitative western blot analysis showed that the level of auxilin present in neuronal cells is almost 10 times the level of GAK present in non-neuronal cells (Greener et al., 2000), probably because recycling of synaptic vesicles requires clathrin-mediated endocytosis to occur much more rapidly in neuronal cells than in non-neuronal cells. Therefore, markedly increasing the level of auxilin or GAK present in cultured cells might actually increase the rate of uncoating of clathrin-coated vesicles and thereby increase the rate of clathrin-mediated endocytosis rather than inhibit it.

In the present study, we found that expression of auxilin or GAK markedly decreased clathrin-mediated endocytosis in HeLa and Cos cells and, at the same time, in many of the cells led to the formation of clathrin-Hsc70-auxilin granules in the cytosol and a decrease in clathrin associated with clathrin-coated pits on the plasma membrane and the trans-Golgi network. However, clathrin-mediated endocytosis was also inhibited by auxilin with its J-domain mutated so that it no longer supported uncoating by Hsc70 in vitro although in this case the clathrin in the cytosol did not form granules but appeared to become aggregated in the cytosol. A similar effect occurred when AP180 or its clathrin-binding domain was expressed. Surprisingly, however, in none of these cases was localization of AP1 or AP2 affected despite the mislocalization of clathrin suggesting first, that expression of clathrin-binding proteins inhibits endocytosis by causing mislocalization of clathrin away from nascent pits, and second, that the binding of APs to these pits occurs independently of clathrin.

HeLa and Cos cells were purchased form ATCC. Mouse neuroblastoma N2A cells were a gift from Dr Y. Peng Loh (NICHD, NIH). Cells were maintained in DMEM supplemented with 10% fetal bovine serum, 2 mM glutamine, penicillin (100 unit/ml), and streptomycin (100 unit/ml) in a humidified incubator with 5% CO₂ at 37°C. All media and supplements were obtained from Biofluids, Inc. (Rockville, MD, USA). For the purpose of immunofluorescence studies calcium phosphate precipitation was used to transfect cell. In order to get higher transfection efficiency in biochemical assays, SuperFect (QIAGEN, Valencia, CA, USA) was used as transfection reagent.

Monoclonal antibody M5 against Flag was obtained from Kodak Scientific Imaging System (Rochester, NY USA). A rabbit antiserum to Flag was from ZYMED Laboratories, Inc. (San Francisco, CA, USA). Monoclonal anti-HA antibody (HA. 11) was purchased from Berkeley Antibody Co. (Richmond, CA, USA). Rabbit antibody against Golgi β coatomer (β-COP), monoclonal anti-clathrin heavy chain (X22), monoclonal anti-α-adaptin (AP.6) are from Affinity BioReagents, Inc. (Golden, CO, USA). Monoclonal antibody against γ-adaptin of AP1 (100/3) was obtained from Sigma (St Louis, MO, USA). Monoclonal mouse anti-human transferrin receptor antibody was purchased from Biomeda Corp. (Foster City, CA, USA). A rabbit antiserum to human transferrin was obtained from Boehringer Mannhein (Indianapolis, IN, USA). Monoclonal and polyclonal antibodies against Hsc70 were purchased from Stressgen Biotechnologies Corp. (Victoria, BC, Canada). Fluorescence-conjugated secondary antibodies were from Jackson ImmunoResearch laboratories, Inc. (West Grove, PA, USA). ¹²⁵I-sheep anti-mouse antiserum was from Amersham Pharmacia Biotech (Piscataway, NJ, USA).

Auxilin and AP180 and their truncated mutants were prepared as Flag fusion proteins (Fig. 1A-B) using a pFlag-CMV-2 expression vector from Kodak Scientific Imaging System (Rochester, NY, USA). Auxilin and AP180 cDNA were subcloned to give pTG176 and pTG135 expressing wild-type AP180 and wild-type auxilin, respectively. This wild-type construct of auxilin was later used to make pTG177 expressing mutated auxilin contains a non-active J-domain, where the HPDK conserved motif was changed to AAAK. N- and C-terminal fragments of auxilin were made as follows. The first 1,224 base pairs of auxilin cDNA were subcloned to give pTG168 expressing the 45 kDa N-terminal fragment of auxilin which contains the tensin domain. The last 1,521 base pairs of auxilin cDNA were subcloned to give pTG197 expressing the 56 kDa C-terminal fragment of auxilin which contains the clathrin binding and J-domains. N- and C-terminal fragments of AP180 were made as follows. The first 1,674 base pairs of AP180 cDNA were subcloned to give pTG190 expressing the 61 kDa N-terminal domain of AP180 containing the domain which interacts with phospholipase D. The last 1,791 base pairs of AP180 cDNA were subcloned to give pTG192 expressing the 65 kDa C-terminal domain of AP180 containing its clathrin binding domain (Lee et al., 1997). GAK was subcloned into GFP vector (Kioka et al., 1999) with the epitope tag at the N-terminal of GAK.

For immunofluorescence microscopy studies, cells were grown on glass coverslips and transfected 24-48 hours before the assay. Cells were washed three times with PBS and incubated with DMEM containing 0. 5% BSA for 30 minutes at 37°C. Human transferrin was then added to the media at the final concentration of 30 μg/ml. Incubation was continued at 37°C for 5 to 10 minutes. To measure bulk fluid-phase endocytosis, 1 mg/ml lysine-fixable FITC-dextran (70,000) from Molecular Probes was added to the cells in DMEM containing 0.5% BSA for 1 hour at 37°C. Cells were washed quickly with PBS, fixed in 2% formaldehyde and processed for immunofluorescence microscopy.

To measured transferrin internalization biochemically, a modified biotinylated transferrin uptake assay (Smythe et al., 1992) was used. Briefly, Cos cells grown in 6-well plates were first depleted of endogenous transferrin. Biotinylated transferrin was then added and incubated with cells for minutes. After removing free biotinylated transferrin by washing, avidin was added to the plates to mask surface-bound biotinylated transferrin. Internalized biotin activity in cell lysates were then assayed quantitatively using streptavidin-horseradish peroxidase in an ELISA plate coated with a rabbit antibody against transferrin. Experiments were performed in triplicate.

Cells were treated as indicated in each experiment, fixed in 2% formaldehyde at room temperature for 15 minutes. After washing the cells three time with PBS containing 10% FBS, cells were incubated with primary antibodies for 1 hour at room temperature. Cells were washed again three times and incubated with fluorescence-conjugated secondary antibodies.

Rat adipose cells from male rats were prepared and GLUT4 endocytosis assays were conducted as previously described (Al-Hasani et al., 1998). Briefly, cells were transfected by electroporation with HA-GLUT4 alone or cotransfected with HA-GLUT4 and Flag-AP180, auxilin or their mutants as indicated. Cell surface GLUT4 in absence or presence of insulin (1×10⁴ microunits/ml) were measured by the binding of monoclonal anti-HA antibody to cell surface HA-GLUT4 followed by the addition of ¹²⁵I-sheep anti-mouse antibody. Cell surface associated radioactivity was counted in a γ-counter. Unless stated otherwise, the values obtained from transfected cells were subtracted from all other values to correct nonspecific antibody binding. Antibody binding assays were routinely performed in duplicate, but occasionally were done in quadruplicate.

To determine the effect of expression of auxilin in cultured cells, we first expressed auxilin with a N-terminal Flag epitope in Cos and HeLa cells. Fig. 2A shows the fluorescence obtained when Cos cells were stained with an anti-Flag antibody to detect the expressed auxilin. Transferrin uptake in the same cells was also imaged by immunofluorescence microscopy as shown in Fig. 2B. Comparison of Fig. 2A and B shows that, in the non-transfected cells, the transferrin was mainly localized to the recycling endosome, whereas the cells expressing auxilin showed marked inhibition of transferrin uptake. Likewise, in HeLa cells, expression of auxilin markedly inhibited transferrin uptake (Fig. 2C,D). Quantification of this effect (Table 1) showed that 10% of the control cells had little or no transferrin uptake compared to 65% of the transfected cells. Interestingly, while in both transfected HeLa and Cos cells the expressed auxilin was cytosolic, its distribution did not appear to be uniform. Although the expressed auxilin varied from a grainy appearance to obvious speckles which ranged in size from tiny particles to large granules, its cellular appearance did not seem to be related to the inhibition of transferrin uptake. In contrast, the uptake of FITC-dextran, a marker for bulk fluid-phase endocytosis, was comparable in auxilin transfected and control cells (Fig. 2E,F). These results establish that expression of auxilin specifically affects clathrin-dependent receptor mediated endocytosis in transfected cells, but not clathrin-independent fluid phase uptake.

To further verify that auxilin inhibits transferrin uptake, we biochemically compared the uptake of biotinylated transferrin in mock and auxilin transfected Cos cells. Using Cos cells in which about 30% of the population was transfected with auxilin, we found that these cells took up about 30% less transferrin than the mock-transfected cells in good agreement with the 30% transfection efficiency (data not shown). Therefore, both biochemical and fluorescence microscopy studies showed that transient expression of auxilin markedly inhibits clathrin-mediated endocytosis.

Since auxilin contains three domains, we next investigated which of these domains is required for inhibition of endocytosis. We expressed either the N-terminal tensin domain, the C-terminal portion of auxilin lacking the tensin domain but containing the clathrin-binding domain and the J-domain, or full-length auxilin with the critical residues, HPDK, of the J-domain (Sell et al., 1990; Tsai and Douglas, 1996) mutated to AAAK, so that, in vitro, we found that the mutated auxilin no longer supported uncoating (data not shown). As shown in Table 1, expression of the N-terminal tensin domain of auxilin in both HeLa and Cos cells had no significant effect on transferrin uptake, whereas, the C-terminal portion of auxilin, which acts like intact auxilin in vitro in uncoating clathrin coated vesicles, also acted like intact auxilin in vivo, inhibiting endocytosis and forming auxilin granules in the cytosol (Fig. 2G,H). We next tested whether the expressed auxilin had to interact with Hsc70 in order to inhibit endocytosis by expressing auxilin with its J-domain mutated so that it could no longer interact with Hsc70 in vitro. Unexpectedly, we found that expression of this mutated auxilin inhibited transferrin uptake just like expression of intact auxilin (Fig. 2I,J; Table 1), raising the possibility that inhibition of endocytosis by expression of auxilin is not related to its involvement in uncoating of clathrin but rather to its activity as an assembly protein. Interestingly, however, in contrast to what is observed with expression of intact auxilin or the C-terminal portion of auxilin with an intact J-domain, there was a morphological difference in that none of the cells expressing auxilin with a mutated J-domain showed formation of auxilin granules.

The observation that expression of auxilin inhibits endocytosis is intriguing because expression of numerous other intact proteins involved in endocytosis such as Eps15, dynamin, amphiphysin, rho, rac, and β-arrestin do not inhibit endocytosis (Benmerah et al., 1998; Damke et al., 1994; Lamaze et al., 1996; Goodman et al., 1996; Wigge et al., 1997); endocytosis is inhibited only by expression of domains or mutants of these proteins that interfere with the function of the parent proteins (Damke et al., 1994; Wigge et al, 1997; Benmerah et al., 1998; Benmerah et al., 1999; Owen et al., 1999; Nesterov et al., 1999). It therefore seems possible that expression of auxilin inhibits endocytosis by overwhelming the regulatory mechanisms in place to prevent inappropriate polymerization of clathrin in the cytosol. If so, inhibition of endocytosis by expression of auxilin may be a general phenomenon that not only occurs with auxilin but with other clathrin-binding proteins as well. To investigate this question we determined whether endocytosis is inhibited by expression of the nerve-specific AP180, which, like auxilin, is monomeric.

As we observed with expression of auxilin, expression of AP180 markedly inhibited transferrin uptake in both Cos (Fig. 3A,B) and HeLa cells (Fig. 3C,D), although like auxilin with a mutated J-domain, the expressed AP180 did not form granules. Quantification of the inhibition of transferrin uptake (Table 1) showed that expression of AP180 reduced transferrin uptake in about 95% of the transfected population of Cos and HeLa cells, which shows that regardless of the level of expression of AP180, it causes a marked reduction in clathrin mediated endocytosis. This is a greater inhibition of transferrin uptake than we observed with auxilin and, in agreement with this observation, we found that in the transfected Cos cells much of the transferrin was localized on the plasma membrane (Fig. 3A,B), an effect that we did not observe with expression of auxilin. Similarly, expression of AP180 in HeLa cells may also have caused much of the transferrin to accumulate on the plasma membrane as shown by the diffuse staining of the transferrin in the transfected cells (Fig. 3C,D). By acid washing the cells briefly in 0.5% acetic acid/0.5 M NaCl, pH 2.4, to remove cell surface-bound transferrin, the transferrin associated with the AP180 transfected HeLa cells was completely removed (data not shown). This establishes that the transferrin is associated with the plasma membrane. Similar to the results obtained in cells expressing auxilin, the expression of AP180 was specific to clathrin-mediated endocytosis since fluid phase uptake, as measured by uptake of FITC-dextran, was unaffected by expression of AP180 (Fig. 3E,F).

The association of transferrin with the plasma membrane of the AP180 transfected cells predicts that there should be an increase in transferrin receptor on the plasma membrane in transfected cells expressing AP180. Fig. 3G,H show that the transferrin receptors in non-transfected HeLa cells are localized to coated pits and endosomal compartments, while in the AP180 transfected cells much of the receptor appears to be localized diffusely on the plasma membrane. Therefore, our data strongly suggest that expression of AP180 is inhibiting internalization of the transferrin receptor rather than a later step in endocytosis.

Since neither Cos nor HeLa cells normally express AP180, we carried out a similar experiment using mouse neuro-2A cells after first demonstrating by western blot analysis, using an antibody specific for AP180, that these cells indeed express AP180 (data not shown). As we observed for Cos and HeLa cells, the transfected neuro-2A cells expressing AP180 showed markedly reduced transferrin uptake (Fig. 3I,J). Western blot experiments showed that the transfected neuro-2A cells produced, after correcting for transfection efficiency, about 20-fold more AP180 than normal neuro-2A cells (data not shown). Therefore, even in cells that normally express AP180, over-expression of this protein markedly inhibits transferrin uptake.

We next examined whether it is, in fact, the clathrin binding domain of AP180 that is causing inhibition of transferrin uptake or whether this inhibition is due to the ability of AP180 to inhibit phospholipase D activity (Lee et al., 1997). The latter activity is localized to the N-terminal fragment of AP180 (Lee et al., 1997), while the C-terminal fragment has the clathrin assembly activity (Ye and Lafer, 1995). Like expression of intact AP180, expression of the C-terminal fragment of AP180 inhibited transferrin uptake in HeLa cells, while expression of the N-terminal fragment of AP180 had no effect. Table 1 quantifies the effect of the C- and N-terminal fragments of AP180 on transferrin uptake in a large population of transfected HeLa and Cos cells. These results clearly show that it is the clathrin-assembly activity of AP180 that is responsible for inhibiting transferrin uptake, not its ability to inhibit phospholipase D activity.

If expression of AP180 and auxilin or their clathrin binding domains indeed inhibits transferrin uptake by inhibiting endocytosis non-specifically, uptake of proteins other than transferrin receptor should also be affected by this expression. To test this prediction we investigated the effect of expression of AP180, auxilin and their clathrin-binding domains on the level of GLUT4 glucose transporter present on the plasma membrane of adipocytes. In its cycle between the plasma membrane and an intracellular compartment, the GLUT4 glucose transporter is thought to be internalized by clathrin-mediated endocytosis (Robinson et al., 1992; Chakrabarti et al., 1994; Volchuk et al., 1998). Therefore if expression of AP180 and auxilin inhibits internalization of GLUT4 in a primary culture of rat adipocytes, transfected cells should display a higher level of GLUT4 on the cell surface in the absence of insulin. Fig. 4 shows that this is indeed the case. As we observed for internalization of transferrin, expression of AP180 had a somewhat greater effect than expression of auxilin. In fact, expression of the clathrin binding portion of AP180 brought the basal level of GLUT4 glucose transporters on the plasma membrane up to the level observed in the presence of insulin while expression of the N-terminal fragments of AP180 and auxilin had almost no effect. These data confirm that, for uptake of GLUT4 transporter as well as transferrin receptor, expression of the APs auxilin or AP180 interferes with clathrin-mediated endocytosis.

We next investigated whether expression of AP180 affects the distribution of clathrin in HeLa cells since our data strongly suggested that it is the clathrin-binding ability of AP180 that is required for inhibition of clathrin-mediated endocytosis. Fig. 5 shows the localization of clathrin in HeLa cells expressing either intact AP180 (Fig. 5A,B) or the C-terminal fragment of AP180 (Fig. 5C,D). Normally clathrin is associated with both the plasma membrane and the trans-Golgi network but in the transfected cells, there seemed to be a loss of clathrin from the trans-Golgi network and an appearance of aggregated clathrin in the cytosol. In agreement with the observed loss of clathrin from the trans-Golgi network, using a chimeric protein, the IL-2 receptor α chain (Tac) containing a signal localization sequence to the lysosome (Marks et al., 1995), we found that over-expression of AP180 or its C-terminal fragment increased plasma membrane association of the chimeric Tac, indicating inhibition of transport of this fusion protein from the trans-Golgi network to the lysosome (data not shown). On the other hand, expression of the N-terminal fragment of AP180 had no effect on the distribution of clathrin (Fig. 5E,F).

The observation that expression of AP180 or its clathrin binding domain removed clathrin from the trans-Golgi network allowed us to investigate whether this affected the distribution of AP1 on the trans-Golgi network. Strikingly, despite the decrease in clathrin associated with the trans-Golgi network in the cells expressing AP180, there was no change in the distribution of the γ chain of AP1; it remained bound to the trans-Golgi network even in the absence of clathrin (Fig. 5G,H). As expected, the Golgi coat protein β-coatomer also appeared normal in AP180 transfected cells (Fig. 5I,J). We also investigated whether AP180 affects the localization of clathrin and AP2 on the plasma membrane. The presence of aggregated clathrin in the cytosol partially obscured the amount of clathrin associated with the plasma membrane, but confocal microscopy suggested that there was indeed less clathrin associated with the plasma membrane of HeLa cells expressing AP180 than with the plasma membrane of untransfected cells (Fig. 6A-C). Furthermore, in agreement with our observation that AP1 localization on the trans-Golgi network is unaffected by expression of AP180, we observed no significant difference in the amount of AP2 associated with the plasma membranes of the transfected and untransfected cells (Fig. 6D-G).

This lack of effect of AP180 on AP2 localization was further confirmed by comparing the fluorescence intensity per unit area in control and AP180 expressing cells. Using the Metamorph imaging computer program, we found that the fluorescence intensity was 18.01±4.90 and 18.68±4.20 in control and AP180 expressing cells, respectively. These results suggest that the AP2 pit density was not significantly affected due to expression of AP180. Therefore, in a result that has important implications for the mechanism of clathrin-coated pit formation, the sequestration of clathrin does not significantly affect the localization of the key APs involved in receptor recruitment and clathrin polymerization at the plasma membrane and trans-Golgi network suggesting that these APs bind to nascent pits independently of clathrin.

To investigate further whether expression of clathrin APs affect clathrin distribution without affecting the localization of AP2 and AP1, we investigated the effect of auxilin on the localization of clathrin and the APs. Strikingly, we found that, in both Cos cells (data not shown) and HeLa cells, auxilin granules always contained clathrin (Fig. 7A,B) and Hsc70 (Fig. 7C,D). Using colocalization with a Lamp-1 antibody, we determined that these proteins are not localized in the lysosomes (data not shown). The association of clathrin with the auxilin granules was accompanied by a marked decrease of clathrin in the cytosol, which, in turn, made it easier to discern than in the cells expressing AP180, that there was a marked decrease in clathrin associated with the clathrin-coated pits on the plasma membrane as well as on the trans-Golgi network. On the other hand, there was no apparent association of either AP2 or AP1 with the granules. And consistent with this observation, we did not observe significant redistribution of either AP2 (Fig. 7E-H) or AP1 (Fig. 7I,J) in these cells confirming that, as in cells expressing AP180, nascent pits containing APs form on the plasma membrane and the trans-Golgi network of these cells in the absence of clathrin.

Further support that nascent pits containing APs can form on the plasma membrane and the trans-Golgi network of cells independent of clathrin binding comes from studies with the auxilin homolog, GAK, which, in contrast to auxilin, is an endogenous protein in HeLa cells. Cells transfected with GFP-GAK showed decreased transferrin uptake, an effect that was particularly dramatic in the cells showing formation of GAK granules (Fig. 8A,B). Furthermore, as with the auxilin granules, clathrin was associated with the GAK granules (Fig. 8C,D), and in cells with GAK granules, there was a marked decrease in clathrin associated with the trans-Golgi network and clathrin-coated pits on the plasma membrane. In fact, in some cases, almost all of the clathrin in the cell was associated with the GAK granules making it particularly clear that, compared to the dramatic changes in clathrin distribution, there was no significant change in the distribution of AP1 and AP2. Specifically, the AP1 was still associated with the trans-Golgi network (Fig. 8I,J), while the AP2 retained its punctate appearance on the plasma membrane although some cytosolic AP2 appeared to be associated with the GAK-clathrin granules in the cytosol (Fig. 8G,H). Interestingly, when the cells transfected with either auxilin or GAK were stained for Hsc70, we found that the granules that contained clathrin and auxilin or GAK also contained Hsc70 (Fig. 7C,D and Fig. 8E,F), which explains why we did not observe these granules in cells expressing auxilin with a mutated J-domain that could not interact with Hsc70. Therefore, in the cells expressing GAK as well as auxilin, nascent pits containing APs form on the plasma membrane and the trans-Golgi network even though clathrin is not associated with these pits.

There is strong evidence that Hsc70, acting with the J-domain proteins auxilin or GAK, plays a major role in uncoating clathrin-coated vesicles both in vitro and in vivo, but does not uncoat clathrin-coated pits (Heuser and Steer, 1989). We were, therefore, interested in whether expression of auxilin or GAK in cultured cells increased or decreased clathrin-mediated endocytosis. In the present study we found that expression of either auxilin or over-expression of GAK inhibited clathrin-mediated endocytosis in Cos and HeLa cells. However, even expression of auxilin with a mutated J-domain inhibited clathrin-mediated endocytosis. Furthermore, expression of the clathrin assembly protein AP180 also inhibited endocytosis and here too it was the clathrin-binding domain of AP180 that was responsible for this inhibition. Since this work was completed, Tebar et al. (Tebar et al., 1999) showed that CALM, a homolog of AP180 expressed in non-neuronal cells, also inhibited clathrin-mediated endocytosis when it was expressed in Cos cells where it is normally present, again supporting the view that expression of proteins or domains of proteins that act as clathrin assembly proteins inhibit clathrin-mediated endocytosis.

Studies on localization of clathrin provided an explanation for the inhibition of clathrin-mediated endocytosis by over-expression of clathrin-binding proteins. In cells expressing auxilin, GAK, AP180, or their clathrin-binding domains, clathrin was either aggregated in the cytosol or, in the case of GAK or auxilin with an intact J-domain, in GAK- or auxilin-clathrin-Hsc70 granules. At the same time there was a decrease in the level of clathrin associated with the trans-Golgi network and the plasma membrane. This led to an opportunity to determine whether the absence of clathrin from clathrin-coated pits affected the distribution of the clathrin assembly proteins AP1 and AP2. Strikingly, we did not observe a decrease in the level of AP1 associated with the trans-Golgi network, nor did we observe a change in the distribution of AP2 on the plasma membrane. Interestingly, when Tebar et al. (Tebar et al., 1999) expressed CALM, they observed a similar depletion of clathrin from the trans-Golgi network with no effect on AP1 distribution, but did not observe a decrease in clathrin at the plasma membrane. Therefore, our results show for the first time that, even in the absence of clathrin binding, AP2 apparently forms nascent pits on the plasma membrane.

Both APs and clathrin are present in the cytosol as well as on cellular membranes and therefore, when clathrin-coated pits form, both APs and clathrin must be recruited to the membrane. There has been speculation that formation of clathrin-coated pits involves co-assembly of clathrin, AP2 and receptors on the plasma membrane (Pearse and Crowther, 1987) but our data suggest that AP recruitment is independent of clathrin recruitment. In this regard, there is strong evidence that AP1 is recruited to the trans-Golgi network by the binding of ARF1, which then dissociates once the AP1 and clathrin are bound (Zhu et al., 1998), but it is not yet understood what causes recruitment of AP2 to the plasma membrane. In any case our results strongly suggest that nascent pits containing APs can form in the absence of clathrin binding. These data are consistent with the observation that when clathrin is removed from existing coated pits by potassium depletion or treatment with hypertonic solution, AP2 remains behind (Hansen et al., 1993; Brown et al., 1999). They are also consistent with the finding of Hannan et al. (Hanna et al., 1998) who found that clathrin and AP2 are independently uncoated from clathrin-coated vesicles. Finally, they are consistent with the results of Liu et al. (Liu et al., 1998) who found that, when they over-expressed clathrin hubs, not only was endocytosis inhibited but, in addition, there was increased clathrin heavy chain associated with the plasma membrane. Yet despite this increase, there was no change in the distribution of AP2.

Since many of the proteins involved in endocytosis shuttle between cytosolic and membrane-bound pools, a key question in the regulation of endocytosis is what keeps these proteins from polymerizing in the cytosol. There is evidence that phosphorylation may regulate clathrin polymerization in the cell (Wilde and Brodsky, 1996) and there is also evidence that Hsc70 acting as a chaperone may form a complex with clathrin triskelions and APs that prevent them from polymerizing in the cytosol (Eisenberg and Greene, 1998; Black et al., 1991). In this regard, our observation that GAK- or auxilin-clathrin-Hsc70 granules form in cells expressing auxilin or over-expressing GAK provides the first direct evidence that clathrin, Hsc70, and auxilin indeed form a complex in vivo as well as in vitro (Jiang et al., 2000), although we could only demonstrate complex formation in cells over-expressing auxilin or GAK. The data presented in this paper also show that the mechanisms that prevent clathrin from polymerizing in the cytosol can be overwhelmed by increasing the levels of APs in the cell suggesting that polymerization of clathrin in clathrin-coated pits rather than in the cytosol depends on multiple regulatory factors including the concentration of APs present in the cell.

The observation that expression of AP180 or auxilin inhibits clathrin-mediated endocytosis provides a simple method of testing whether a given process in the cell involves clathrin-mediated endocytosis. Using this method we confirmed that the GLUT4 glucose transporters are internalized from the plasma membrane by clathrin-mediated endocytosis and, at the same time, demonstrated that expression of AP180 and auxilin not only inhibited endocytosis in immortalized cells but also in primary tissue culture cells. Our studies on the GLUT4 transporter show that the strongest inhibition of clathrin-mediated endocytosis occurred with the clathrin-binding domain of AP180. In fact, expression of this domain inhibited clathrin-mediated endocytosis so strongly that the basal level of the GLUT4 transporter on the plasma membrane of transfected adipocytes almost reached the same level as in cells treated with insulin. On the other hand, although it has been reported that the GLUT4 glucose transporter interacts with AP1 and AP3 (Gillingham et al., 1999), over-expression of AP180 or its clathrin binding domain had no effect on the transport of GLUT4 glucose transporters to the plasma membrane in the presence of insulin suggesting that clathrin is not involved in this transport. Therefore, in future studies, expression of the clathrin-binding domain of AP180 should provide a simple method of determining whether clathrin-mediated endocytosis is involved in various processes in the cell, a method that will compliment the use of clathrin hubs to inhibit endocytosis (Liu et al., 1998). Since one method decreases the level of clathrin heavy chain associated with the plasma membrane while the other increases it, agreement between the effects of these two methods will strengthen the conclusion that clathrin-mediated endocytosis is required for a particular process.

We thank Drs Julie Donaldson and Harish Radhakrishna for their many helpful discussions, Dr Kenneth Yamada for the GFP-vector, Dr Ivan Bonifacino for the Tac construct, and Dr Xufeng Wu for her valuable help with the confocal microscopy work.