Receptor-mediated endocytosis is sensitive to antibodies against the uncoating ATPase (hsc70)

Clathrin-coated vesicles (CVs) participate in a variety of transport processes, e.g. the selective and efficient uptake of molecules into eukaryotic cells via receptor-mediated endocytosis (Goldstein et al., 1979; Mellman et al., 1987). Early events of this process comprise ligand-receptor interaction, aggregation of these complexes in coated pits and formation of CVs that serve as short-lived vehicles destined to deliver their cargo to the endosomal compartment (Pearse, 1980). Fusion of coated vesicles with endosomes can only occur after the clathrin-containing coat is released from CVs (Altstiel and Branton, 1983). This uncoating reaction requires the hydrolysis of ATP. Several groups contributed to the identification of a putative ‘CV-uncoating ATPase’ (Patzer et al., 1982; Schlossman et al., 1984), which was finally purified from mammalian brain and shown to strip clathrin triskelia from clathrin baskets and isolated CVs, in a complex process involving several ATP-dependent steps (Schmid and Rothman, 1985a-c; Greene and Eisenberg, 1990; Paddenberg et al., 1990). A functional role for UA in the uncoating process in living cells has been postulated (DeLuca-Flaherty et al., 1990), but not demonstrated so far.

More recently, the uncoating ATPase (UA) has been identified as a member of the 70 kDa heat shock (hsp70) family. By two-dimensional gel electrophoresis, immunological crossre-activity and sequence homology (Ungewickell, 1985; Chappell et al., 1986), UA is identical with the 70,800 Da cognate heat shock protein (hsc70), which is constitutively expressed and only slightly heat-inducible. All members of the hsp70 family exhibit extensive sequence homologies, particularly within a high-affinity ATP-binding site at their N terminus (Lindquist and Craig, 1988). Many of the constitutively expressed 70 kDa proteins have been shown to serve as ‘molecular chaperones’, mediating correct folding, unfolding and refolding of proteins in various translocation processes that involve crossing of intracellular membranes (Ellis, 1987; Welch, 1991). hsc70, the CV-uncoating ATPase, also exhibits this chaperoning activity in several cellular processes: it apparently facilitates protein transport into the nucleus (Shi and Thomas, 1992; Imamoto et al. 1992) and into lysosomes (Terlecky et al., 1992). Its activity on coated vesicles is presumably also associated with a chaperoning function: the enzyme probably protects clathrin from misfolding during the uncoating process.

In stress situations, such as temperature shock, the hsp70 proteins may generally act as protein stabilizers, since a large fraction of them is bound to polypeptides under such conditions (Beckman et al.,1992; Baler et al., 1992). Upon temperature shift from 37°C to 42°C, hsc70 migrates from the cytoplasmic to the nuclear compartment (Ungewickell, 1985; Riabowol et al., 1989), where it presumably interacts with a specific heat shock transcription factor (Baler et al., 1992).

Recently, it was shown that polyclonal antibodies against hsc70, when microinjected into human hepatoma cells, rapidly block the import of nucleus-specific proteins into the nuclear compartment (Imamoto et al., 1992). Here, we show that microinjection of a well-characterized monoclonal antibody against UA/hsc70 also interferes rapidly with uptake and transportation of ligand-receptor complexes in two different cell systems. These events were correlated with the accumulation of coated structures in the periphery of the injected cells. Thus, our results strongly support the hypothesis that UA/hsc70 is indeed involved in an essential step during receptor-mediated endocytosis.

BHK-21 cells were obtained from the American Type Culture Collection (Rockville, MD/USA). The cell clone 2/14, derived from BHK-21 cells transfected with human FcRII cDNA, has been described previously (Engelhardt et al., 1990). BHK-21 cells, clone 2/14 cells and human fibroblasts from healthy donors (primary cultures) were grown in DMEM (Gibco, Eggenstein/FRG), supplemented with 10% fetal calf serum (FCS, Seromed Biochrom, Berin/FRG) and 2 mM L-glutamine, at 5% CO₂.

For microinjection experiments and all light microscopic analyses, cells were grown on glass coverslips. Viability and integrity of microinjected cells were tested by analysis of cell proliferation and Trypan Blue exclusion 24 hours after injection, using standard protocols.

CVs were purified from fresh porcine brain, according to standard protocols (Altstiel and Branton, 1983). After several centrifugation steps, they were finally subjected to gel filtration on Sephacryl S-1000 (Pharmacia, Freiburg/FRG) (Campbell et al., 1984). Purity of the preparations was monitored by SDS-PAGE and electron microscopy.

UA/hsc70 was purified according to Schlossman et al. (1984)with only slight modifications. The first high-speed supernatant of the CV preparation served as starting material. It was dialyzed for 48 hours against 0.025 M Tris-HCl, pH 7.0, and subsequently centrifuged for 60 minutes at 35,000 g. The supernatant was subjected to ion exchange chromatography (DEAE-52, Whatman, Maidstone/UK) as described. Further purification steps included hydroxyapatite chromatography and affinity chromatography on ATP-agarose (Sigma, Deisenhofen/FRG) according to Schlossman et al. (1984).

Discrete fragments of purified UA/hsc70 were obtained by limited proteolysis using the following enzymes: Staphylococcus aureusV8 protease (molar ratio of UA/hsc70 to protease 1:20); α-chymotrypsin (1:13); papain (1:11); and trypsin (1:6). All proteases were purchased from Sigma. Digestion protocols and termination of the reaction were as described (Westmeyer et al., 1990).

For sequencing, UA/hsc70 (150 μg) was digested with V8 protease, subjected to SDS-PAGE and electroblotted onto Fluorotrans-membrane (Pall, Dreieich/FRG). The blotted polypeptide profile was visualized by Ponceau Red staining. A discrete 26 kDa fragment present in all digests was selected for amino acid sequencing in a microseqencer (Knauer, model 816; Bad Homburg/FRG). The N-terminal 10 amino acids of the V8-generated 26 kDa fragment were thus identified.

A mouse monoclonal antibody to UA/hsc70 was generated according to standard protocols. This antibody (3C5), of the IgG 1 isotype, was used after dialysis against phosphate buffered saline (PBS), pH 7.2. For localization of UA/hsc70 after heat shock experiments, a commercially available mouse monoclonal antibody (BRM 22; Sigma, Deisenhofen) was employed. Clathrin-coated vesicles were labeled with a monoclonal mouse antibody against brain clathrin heavy chain (IgM; Dianova, Heidelberg). Mouse IgG (Sigma) and preimmune serum from the 3C5 producing animal, diluted to 8-15 mg/ml in PBS, were used as controls. In fluorescence studies, all mouse antibodies were detected with Texas Red-labeled goat anti-mouse F(ab)₂, and human IgG was labeled with FITC/goat anti-human F(ab)₂(both from Dianova, Hamburg/FRG). For immunoblots (see below), horseradish peroxidase(HRP)-coupled rabbit anti-mouse IgG (Sigma) was used.

BHK-21 cells and human fibroblasts grown in culture dishes were trypsinized, concentrated by centrifugation and solubilized in sample buffer (2% SDS, 8% β-mercapthoethanol; 4% glycerol; 5 mM EGTA, 0.01% Bromophenol Blue in 0.06 M Tris-HCl, pH 6.8). Purified UA/hsc70, proteolytic fragments and cell extracts (all in sample buffer) were separated on 10% SDS-polyacrylamide gels according to Laemmli (1970). For two-dimensional gel electrophoresis, the iso-electric focussing gels contained ampholytes (Pharmacia, Freiburg/FRG, and Serva, Heidelberg/FRG), yielding a pH gradient of 3 to 10. Separation in the second dimension was achieved by SDS-PAGE on 10% slab gels (Rodemann, 1990). Polypeptides were blotted onto nitrocellulose and incubated with the first and HRP-coupled second antibodies followed by chloronaphthol and H₂O₂, according to standard protocols. For the quantification of UA/hsc70 in BHK cells, the proteins in total extracts and purified UA were separated on 10% SDS-gels, blotted onto nitrocellulose and reacted with 3C5 followed by HRP-coupled second antibodies and reagents for enhanced chemoluminescence (ECL; Amersham-Buchler, Braunschweig). Densitometry of the ECL-X-ray films was used to calculate the amount of UA in BHK cells, in comparison with a calibration curve obtained with purified UA.

Heat-aggregated human IgG (ahIgG) was used as a homologous ligand for the FcRII expressed on 2/14 cells. Monomeric human IgG (Dianova) was heated to 63°C for 25 minutes, cooled rapidly to 0°C and centrifuged at 30,000 gfor 30 minutes at 4°C. The supernatant, containing primarily small aggregates, was divided into aliquots and kept frozen at −70°C until use. Low density lipoprotein (LDL, density: 1.019-1.063 g/l) was isolated from pooled human plasma as described (Havel et al., 1955). For gold-labeled LDL, monodisperse gold sols of 15 nm diameter gold particles were freshly prepared by reduction of chloroauric acid with sodium citrate as described (Frens, 1973). Conjugation of LDL to the gold particles was performed as described previously (Robenek et al., 1982).

The assay involved quantification of the clathrin heavy chain released from isolated CVs (Greene and Eisenberg, 1990). Purified CVs (80 μg protein) were mixed with 10 μg UA/hsc70 in a total volume of 100 μl buffer (20 mM HEPES-NaOH, pH 7.0, 25 mM KCl, 10 mM (NH₄)₂SO₄, 2 mM magnesium acetate, 0.8 mM dithiothreitol) with or without 5 mM ATP. After 20 minutes at 37°C, the reaction mixture was centrifuged in an Airfuge (Beckman Instruments, Munich/FRG) at 100,000 gfor 10 minutes. The supernatants were subjected to SDS-PAGE, and the amount of clathrin was monitored by scanning the Coomassie Blue-stained 180 kDa bands, using a Chromoscan 3 (Joyce Loebl, Gatesway/UK). Some aliquots were incubated with the antibody 3C5 at equimolar ratio with respect to UA/hsc70, or a mouse control serum at twice this protein concentration.

For microinjection, clone 2/14 cells and human fibroblasts were grown on coverslips marked with a diamond pen. All cells within a marked area were injected according to standard protocols (Westmeyer et al., 1990), using glass capillaries and a commercial micromanipulation and microinjection system (Eppendorf-Nethler-Hinz GmbH, Hamburg/FRG). Cells were injected with either 3C5 or mouse IgG (at 4 mg protein/ml), or mouse serum (at 8-15 mg protein/ml), within 10-15 minutes. After microinjection, the cells were reincubated at 37°C/5% CO₂for one or two hours prior to further experimental handling or analysis. Before fixation, all cells outside the marked area were removed with a rubber policeman.

One hour after microinjection, BHK-21 cells, injected with either 3C5 or serum, and non-injected controls were subjected to a temperature shift from 37°C to 42°C for 2 hours. Subsequently, the cells were fixed and processed for immunofluorescence (see below). The intracellular location of endogenous UA/hsc70 was revealed either with a second incubation of 3C5 or with the antibody BRM 22 from Sigma, followed by goat anti-mouse IgG coupled to Texas Red (see Antibodies, above). For both antibodies, identical fluorescence images were obtained.

The CV-mediated endocytosis of ahIgG in FcRII-transfected BHK-21 cells (clone 2/14) has been described elsewhere in detail (Höning et al., 1991). One or two hours after microinjection, the cells were chilled to 4°C and incubated with ahIgG (2 μg/coverslip) for 45 minutes. Binding was terminated by repeated washings with ice-cold medium. Subsequently, the cells were warmed to 37°C for different time periods, and endocytosis was stopped by washing with ice-cold medium. The cells were then fixed and processed for either light or electron microscopic analysis.

To stimulate enhanced LDL uptake under experimental conditions, human fibroblasts were precultured in DMEM without FCS for 24 hours. Prior to incubation with the gold-conjugated ligands, microinjected cells or non-injected controls were washed three times with ice-cold PBS. Binding of LDL-gold (40 μg/ml) was performed below 4°C for 1 hour. Subsequently, the cells were washed extensively and either fixed directly for surface replication, or triggered to endocytose the ligand by replacing ice-cold buffer with 37°C medium. After 5 or 10 minutes at this temperature, these cells were also washed with ice-cold PBS and rapidly fixed to terminate endocytosis.

For light microscopic analysis, the cells were fixed in 3.7% formaldehyde in PBS for 20 minutes and permeabilized with 0.2% Triton X-100 for 10 minutes. Double labeling for injected antibodies and ahIgG were subsequently carried out by simultaneous incubation with the respective second antibodies (see above) for 30 minutes. Finally, the cells were embedded in Mowiol 4.88 (Hoechst, Frankfurt/M./FRG) and examined in a conventional epifluorescence microscope (Axiophot, Zeiss, Oberkochen/FRG), equipped with FITC and Texas Red detection filter systems. Morphology, adhesion, integritity of intracellular organelles and microfilament organization of antibody-injected cells were examined in rhodamine-phalloidin (Sigma, Deisenhofen, FRG) and rhodamine-wheat germ agglutinin (Sigma)-stained specimens, using phase-contrast, reflection contrast and fluorescence microscopy. Photographs were taken on Kodak Tri-X Pan.

Alternatively, the cells were analysed with a confocal laser scanning microscope (Lasersharp MRC 500, Bio-Rad, Munich/FRG), attached to an inverted microscope (Zeiss). Two independent confocal detection channels were used for simultaneous detection of FITC and Texas Red fluorescence. During image capturing either a Kalman (xysections) or an accumulative filter (xzsections) was used to reduce photomultiplier noise. For edge enhancement, images were further processed by applying crispening convolutions (SOM software; BioRad). The step size in xzsections was kept constant at 0.27 μm during all sections. Photographs were taken with a 35 mm camera from a flat screen video monitor system using a Kodak T-Max film.

For surface replication, human fibroblasts were fixed in ice-cold 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2, for 2 hours, dehydrated in an ethanol series and subjected to critical-point or air drying. The bottom of the culture dish with the attached cells was cut into 1 cm²pieces with a soldering iron. Platinum/carbon surface replicas of the cell monolayers were made in a Balzers BA 300 apparatus (Balzers, Lichtenstein), equipped with an electron gun evaporator and a quartz crystal thickness monitor. Replicas were obtained by shadowing the cell surface with platinum/carbon at an angle of 38°, followed by carbon at 90°. The replicas were cleaned overnight in household bleach and washed in distilled water.

For thin sectioning, the cells grown on glass coverslips and injected with either 3C5 or control serum were fixed directly in the culture dish and dehydrated as described above. After embedding in Epon 812, the glass coverslip was removed with fluoric acid. Only the area containing microinjected cells, identified by diamond pen marking, was cut out to obtain thin sections.

All specimens were picked up on 200-mesh copper grids and examined in a transmission electron microscope (model 410, Philips, Eindhoven/NL) at 60 kV.

In vitro release of clathrin from CVs was recorded in four independent experiments. In the microinjection analysis, 58-108 cells were injected on at least three different coverslips. The exact numbers for each experiment are given in the corresponding figure legends. For quantification at the light microscopic level, every injected cell was counted. BHK cells that were fixed and processed after a 20 minute endocytotic period could be easily allocated to one of two categories according to the distribution of fluorescently labeled ahIgG: ‘normal endocytosis’ was defined by bright staining of large vesicles distributed throughout the cell and concentrated in the perinuclear area; ‘subnormal endocytosis’ was defined by weak, dotted staining and lack of large vesicles and perinuclear label.

Coated structures (clathrin-decorated pits and vesicles) were quantified on thin sections of clone 2/14 cells that had been injected with either 3C5 or control serum. On each section, the entire peripheral region (plasma membrane and the underlying cortical cytoplasm, approximately 1-2 μm in depth) of 10-16 injected cells was analysed for coated structures.

For quantification of surface replicas obtained from gold-LDL-labeled human fibroblasts, a minimum of 12 cells per time point was examined, and each cell was photographed. Label was counted on standard squares (61 μm2), which represented approximately 6% of the dorsal plasma membrane.

The relative standard deviations (mean values divided by standard deviation) are given for all data.

To study the role of UA/hsc70 in ligand-receptor uptake, we wanted to reduce the amount of functional UA/hsc70 in living cells. To this end, we characterized in detail a monoclonal antibody, 3C5, raised against purified pig brain UA/hsc70. The antibody bound selectively to a 70 kDa polypeptide in one-dimensional gel profiles derived from human fibroblasts (Fig. 1A), and identified one spot at a position expected for UA/hsc70 in two-dimensional gels prepared from BHK-21 extracts (Fig. 1B). In proteolytic digests of the brain protein, 3C5 recognized two fragments of 60 and 26 kDa, respectively, but not a prominent 44 kDa fragment (Fig. 1A). Micro-sequencing of the 26 kDa immunoreactive fragment, in combination with information on the UA/hsc70 domain structure available in the literature (Chappel et al., 1987; Flaherty et al., 1991), revealed that the 3C5 epitope is located in the C-terminal domain, which contains the clathrin-binding site (Fig. 1D). Thus, 3C5 reacts with the same domain that is involved in interaction with CVs, and it binds specifically and selectively to the putative UA/hsc70 in extracts of BHK-21 and human fibroblasts.

In an in vitro assay, we found that 3C5 interferes with the ATP-dependent uncoating activity of the purified protein on isolated CVs. UA/hsc70, as purified from brain, effectively removes clathrin from isolated coated vesicles, in an ATP-dependent reaction (cf. Paddenberg et al., 1990). In the presence of equimolar amounts of 3C5, this activity was reduced approximately to the level seen without exogenous ATP and UA, while control serum proteins, added at twice this concentration, had very little effect (Fig. 1C). Thus, in this assay, the complexes formed between 3C5 and its antigen are of sufficient affinity to interfere with the interaction between the enzyme and CV.

The formation and stability of a complex between 3C5 and rodent UA/hsc70 were tested in vivo, by analysing the effect of microinjected 3C5 on intracellular mobility of the enzyme. Based on a calibration with purified brain UA/hsc70, the amount of UA/hsc70 present in BHK cells was determined to 0.2% of the total protein (not shown). When BHK cells were subjected to heat shock (42°C, 2 hours), a large fraction of this amount was translocated from the cytoplasm into the nucleus (not shown), as had been described for other fibroblastic cells (Ungewickell, 1985; Riabowol et al., 1989). This translocation was not hampered in cells that had been injected with either an unrelated mouse IgG (4 mg/ml) or preimmune serum taken from the 3C5-producing mouse (8-15 mg/ml). In contrast, in cells that had been injected with 4 mg/ml of 3C5, the enzyme was retained in the cytoplasm (Fig. 2). In these experiments, UA/3C5 was visualized by staining the injected cells either with another, commercially available antibody against UA (see Materials and Methods) or by adding 3C5 again after fixation. No difference was found between these two protocols: in the control antibody- or serum-injected cells, the nuclei were intensely labeled (Fig. 2, top), while the 3C5-injected cells showed bright label of the cytoplasm and dark nuclei (Fig. 2), identical to that in cells not subjected to heat stress. Fig. 3gives the corresponding quantitative data demonstrating that heat-induced translocation of UA/hsc70 was inhibited in more than 90% of 3C5-injected cells. Thus, this antibody binds with high efficiency to UA/hsc70 in living BHK cells, forming heat-resistent complexes.

In the next set of experiments, 3C5 and control serum were microinjected into cells actively engaged in CV-mediated ligand-receptor uptake. By double labeling, the intracellular distributions of injected antibodies, of the ligand and of clathrin-coated structures were assessed.

BHK-21 cells (clone 2/14) transfected with the receptor for aggregated human immunoglobulin G (FcRII) rapidly and efficiently internalize this ligand (ahIgG) by receptor-mediated endocytosis. Qualitative and quantitative aspects of this system have been described in detail elsewhere (Höning et al., 1991). In immunofluorescence, the ligand, as labeled with fluorescent F(ab)₂against human IgG, was seen distributed in small dots in the cortical area of the cell early (2-5 minutes) after onset of endocytosis. Later (10-20 minute time points), the ligand was primarily detectable in larger vesicles of nonhomogeneous size that were frequently concentrated in the perinuclear area, and these organelles were found to be part of the endosomal/lysomal system (Höning et al., 1991).

Identical patterns were observed in the present study with cells injected with unrelated mouse IgG or preimmune serum (4 mg/ml and 8-15 mg/ml, respectively), and allowed to endocytose for 20 minutes, as seen in Fig. 4C,C′ and D,D′. In contrast, when the antibody 3C5 was injected (at 4 mg/ml) most cells showed a markedly altered labeling pattern at that time point. The cytoplasmic vesicles were greatly reduced in size, and not preferentially seen in the perinuclear area (Fig. 4A,A′,B,B′). Only a few of the 3C5-injected cells showed the large, presumably endocytotic vesicles that were prominent in controls, and those had obviously received little 3C5, as was seen by the weak labeling for this antibody (cells marked with short arrows in Fig. 4A,A′ and B,B′). Such en face images as were obtained from conventional fluorescence analyses permitted allocation of the cells to two categories defined as ‘normal’ and ‘subnormal endocytosis’ (for details of definition, see Materials and Methods). The quantitative data thus obtained revealed that approximately 86% of non-injected or control serum-injected BHK-21 (clone 2/14) cells showed the large cytoplasmic vesicles associated with efficient endocytosis of ahIgG, while this value was decreased to 22% in 3C5-injected cells (Fig. 5).

Previously, we had analysed the correlation between ahlgG and clathrin labeling patterns in non-injected BHK-21 clone 2/14 cells. The small dots containing ahlgG at 2 to 5 minutes after triggering endocytosis were also labeled for clathrin, while the endosomal/lysosomal vesicles filled with the ligand were not or only partially positive for clathrin in immunofluorescence (Höning et al., 1991). In the present study, we examined the clathrin and ligand labeling patterns in control serum- and 3C5-injected cells. The images of double-labeled cells, as seen by confocal microscopy with clathrin labeled in red and the ligand in green, are depicted in Fig. 6. At 5 minutes after the onset of endocytosis in the control-injected cells, most of the ligand was confined to small dots that were also clathrin-positive, as indicated by their yellow colour (Fig. 6C), but some had apparently left the clathrin-bound compartment and had been transported to clathrin-free vesicles (green dots in C). These images were identical with those of untreated cells (cf. Höning et al., 1991). In contrast, in 3C5-treated cells, all ligand-containing vesicles were also clathrin-labeled (only yellow dots are seen in Fig. 6A). This difference became much more pronounced at 20 minutes after triggering endocytosis: only the control-injected cells showed the large endocytotic vesicles concentrated in the perinuclear area, free of clathrin label (green colour, Fig. 6D). In the 3C5-injected cells, the ligand persisted in clathrin-positive structures distributed more peripherally (Fig. 6B, yellow colour). The ligand was, however, not simply bound to the cell surface but internalized, as concluded from the observation that these patterns proved resistant to washing in buffers of high salt and low pH (2.0). Thus, these images suggested that ahlgG uptake by CV was specifically arrested by 3C5 at an early step of endocytosis, when the ligand was still associated with clathrin-coated pits or vesicles.

We then examined the effect of microinjected 3C5 on clathrin-coated structures in this system at the ultrastructural level. All cells within a diamond mark drawn on the coverslip of the cell culture were injected either with control serum or with 3C5. After removal of all cells outside this area, the injected cells were embedded and sectioned. Screening the cortical regions of all cells seen in each section revealed only a few coated structures per cell section, as expected. No differences in morphology were seen between coated structures of 3C5-injected and control cells (not shown). However, counting clathrin-coated pits and vesicles on 10-16 cells that had been fixed at 10 minutes after triggering endocytosis revealed a statistically significant difference between 3C5-injected and control cells, indicating that 3C5 injection led to an increase in CVs (clathrin-coated circular structures) of approximately 50% and in coated pits of 25% (Fig. 7). The coated vesicles may of course also be deeply invaginated coated pits, but these data show that clathrin-coated structures accumulate in the cellular periphery as a consequence of 3C5 injection.

A detailed study of the kinetics of ahIgG uptake was performed by confocal laser scanning microscopy on single cells. Optical sections in the xz-axis allowed for analysis of the internalization process. As seen in Fig. 8, the ligand, added to non-injected and control serum-injected cells, is associated with the dorsal membrane before uptake. Within 20 minutes after triggering endocytosis, it progressively filled the depth of the cytoplasm towards the ventral membrane (Fig. 8A). Conversely, in 3C5-injected cells, the initial strong fluorescence seen at 0°C, indicative of normal ahIgG-binding to the dorsal membrane (0 minutes), did not spread throughout the cytoplasm, but gave rise to only a few distinct streaks at 5 minutes and smaller dots at 20 minutes after onset of endocytosis (Fig. 8B). These images, obtained from optical sections through single cells, as well as the en face (xy) images (Fig. 8, upper panels), correspond well to the images shown in Figs 4and 6, and are consistent with the view that ligand is taken up but is arrested at an early stage of coated vesicle-mediated endocytosis.

The system used in the experiments described above employs transfected cells that grossly overexpress an exogenous receptor molecule. Therefore, it seemed advisable to test the effect of 3C5 injection in a situation of natural endocytosis. In this context, we examined the influence of 3C5 on the uptake of LDL by human fibroblasts. As in the ultrastructural analyses described above, all cells grown within a diamond mark on the coverslip were injected, and all cells outside this area were removed. LDL-gold was added to controls and injected cells at 4°C. The cells were then processed for surface replicas. Fig. 9shows the distribution of bound LDL on the dorsal plasma membrane in a surface replica of a control cell. As previously described (cf. Robenek et al., 1982), the gold particles appeared in aggregates of variable size even at low temperature, consistent with previous data showing clustering of the LDL receptor prior to ligand internalization (Goldstein et al., 1985). At the resolution of such replicas, shadows associated with the larger aggregates were easily detectable, identifying their location in small depressions that may correspond to submembranous coated pits (cf. large aggregates marked by arrows with smaller, shadowless aggregates marked by arrowheads in Fig. 9). Subsequently, cells were induced to endocytose the gold-labeled LDL by warming them to 37°C. At various time points, replicas were prepared, and the decrease in label was quantified, using standard squares of 61 μm²on individual cells. With non-injected controls, it was found that approximately 75% of the label is internalized within 5 minutes of endocytosis in this system (Fig. 10). After that time point, the residual gold particles remain almost at the same level, and no further uptake of large aggregates can be seen (not shown). For control serum-injected cells, identical values were found. In contrast, 3C5-injected fibroblasts demonstrated only a slight reduction of such aggregates at 5 minutes after onset of endocytosis: 70% remained at the cell surface. Thus, endocytosis of gold-LDL was reduced to approximately 30%. At 10 minutes, there was no further decrease in surface-associated particles (Fig. 10). Thus, we observed that LDL endocytosis into fibroblasts was severely affected by 3C5 injection. This might be the result of either reduced uptake or rapid return of intact ligand-receptor complexes to the cell surface, by reversing the normal route.

Antibodies have been used in several studies to analyse discrete steps in CV-mediated, receptor-dependent endocytosis of ligands. The role of CV assembly in this process has been investigated by antibodies against clathrin. While the microin-jection of a polyclonal antibody against brain clathrin had no effect on the uptake of α₂-macroglobulin in fibroblasts (Wehland et al., 1981), monoclonal antibodies that interfered with clathrin assembly in vitro also affected CV-mediated endocytosis: after fusion with anti-clathrin-loaded erythrocytes, CV1 cells showed a 50% decrease in uptake of Semliki Forest virus particles (Doxsey et al., 1987). The importance of the functional state of the receptor molecules has been emphasized in a study of EGF uptake in various cell lines transfected with EGF-receptor constructs: microinjection of monoclonal antibodies against phosphotyrosine interfered with endocytosis of ligand-receptor complexes, probably by impairing the tyrosine phosphorylation-dependent activity of the receptor kinase (Glenney et al., 1988). In the present study, we microinjected antibodies against hsc70, the CV uncoating ATPase, to study the role of this enzyme in uptake and processing of ligand-receptor complexes.

The usefulness of antibodies for such studies depends, of course, crucially on the quality of the antibodies employed. Therefore, a large part of this work was focussed on the characterization of 3C5, a monoclonal antibody against pig brain UA/hsc70. The antibody binds specifically to the presumptive UA/hsc70 in hamster and human fibroblasts, it is directed to the very UA/hsc70 domain that contains the clathrin-binding site, and it inhibits quantitatively the in vitro uncoating of CVs. The affinity of 3C5 for its target in living cells was characterized in heat-shock experiments. Antigen-antibody complexes were apparently formed readily in the cytoplasm and were sufficiently stable to resist a possible dissociation at 42°C. Similar results had previously been described by Riabowol et al. (1988) after injecting rat embryo fibroblasts with a mixture of four monoclonal antibodies, and by Imamoto et al. (1992), who used polyclonal antibodies against UA/hsc70. With the properties described, 3C5 seemed a promising tool with which to analyse the effect of decreasing UA/hsc70 levels on receptormediated endocytosis.

For microinjection, we used a concentration of 4 mg/ml, which supposedly resulted in an approximately equimolar concentration of 3C5 to UA/hsc70, based on the following calculation. A tenfold dilution of the antibody by microinjection (cf. Füchtbauer et al., 1985) yields a cytoplasmic concentration of 2.7×10⁻⁶M for 3C5. Assuming an average protein concentration in BHK cells of 100 mg/ml, 0.2% are equivalent to a UA/hsc70 concentration of 3×10⁻⁶M. As the antibody, of the IgG type, has two binding sites, the effective 3C5 concentation was at least equal or in excess to UA/hsc70.

Due to the pleiotropic effects ascribed to this chaperoning protein, one might have expected the 3C5-injected cells to be in poor health. However, at least for the short period (up to 3 hours after injection) considered in this study and with the antibody concentration used, we could not observe adverse effects on cellular morphology, adhesion, microfilament organization, Golgi, mitochondrial or nuclear structure in light and electron microscopic analyses. Moreover, 24 hours after the injection experiments Trypan Blue exclusion was identical between the 3C5-, control-serum-injected and non-injected cells (data not shown). This correlates well with the findings by Imamoto et al. (1992), who demonstrated a rapid effect of injected polyclonal UA/hsc70 antibodies on import of nuclear proteins, while RNA synthesis indicative of overall cellular vitality remained unaffected.

We analysed the consequences of 3C5 injection on receptor-mediated endocytosis in two different cell systems. In the first one, we used transfected hamster cells that express a human Fc receptor in high density. Previously, we had verified that these cells internalize the appropriate ligand (ahlgG) via coated vesicles, apparently using the normal endocytotic pathway. However, the uptake process is somewhat slower than expected: approximately 75% of the ligand is endocytosed within 20 minutes, while for example the same amount of LDL is endocytosed by fibroblasts within 10 minutes. Moreover, we found evidence that in the transfected cells, at least a fraction of the FcRII receptor molecules is recycled, while this is not the case in their natural environment, i.e. in human immunocompetent cells. These observations may be correlated with the fact that the Fc receptor is grossly overexpressed in BHK clone 2/14 cells as compared with the normal situation (cf. Höning et al. 1991). Keeping this in mind, we also analysed the effect of 3C5 on CV-mediated endocytosis in another, natural system in parallel, and chose the LDL uptake in fibroblasts as a well known example.

Our results show that in both systems, the injection of 3C5 antibody molecules caused a drastic decrease of ligand endocytosis. On the basis of our findings that this antibody inhibits the uncoating of coated vesicles in vitro, one is tempted to speculate that this is also the case in the injected cells. However, in the present study we cannot directly prove this. Moreover, the two cell systems used may differ in their response to antibody-based interference with UA/hsc70 activity, in particular in the endocytotic step being affected. In 3C5-injected BHK 2/14 cells, coated structures accumulated in the cortical region, and the ligand remained associated with clathrin-positive areas for much longer than in control cells. In human fibroblasts, gold-LDL was found to be collected in coated pits, but it was either not taken up with the normal velocity, or returned to the surface within minutes. In both systems, however, our experiments strengthen the view that UA/hsc70 is directly involved in receptor-mediated endocytosis, presumably in the dynamics of clathrin-associated structures. The identification of the actual step being blocked by antibody injection has to await further, extensive ultrastructural analysis of 3C5-injected cells, and the investigation of additional cell systems.

We thank Drs J. Frey and W. Engelhardt (Bielefeld) for clone 2/14 cells, Dr M. Melkonian (Köln) for providing his confocal laser scanning microscope, Dr H. Reinke (Bielefeld) for peptide sequencing, S. Henning and Ch. Wiegand for devoted technical assistance, and R. Klocke and I. Demesvary for expert typing. This study was supported by the Deutsche Forschungsgemeinschaft (SFB 223).