Sequential appearance of epidermal growth factor in plasma membrane-associated and intracellular vesicles during endocytosis.

Receptor-mediated internalization of epidermal growth factor (EGF) occurs by a process involving initially clathrin-coated pits on the cell surface and the subsequent formation of ligand-containing endosomes. Using a modified acid wash technique, cell surface-bound EGF was removed. Utilizing sucrose density centrifugation, the residual cell-associated EGF was separated into plasma membrane-associated and intracellular vesicle-associated forms. Using these procedures we have identified a transient form of cell-associated EGF that is still attached to the plasma membrane but not accessible to the extracellular fluid. This form of EGF appears to be the precursor for endosomic EGF. We suggest that this intermediate form represents the receptor-ligand complex shown by electronmicroscopy to be located in narrow-necked plasma membrane invaginations (Willingham, M. C., and Pastan, I. (1980) Cell 21, 67-77).

Peptide hormones, such as epidermal growth factor (EGF),' appear to enter cells by way of endocytosis (1-3). Electronmicrographs have revealed narrow-necked invaginations of the plasma membrane containing labeled cell surface receptors (4-6). These structures were proposed to represent an intermediate step in the transfer of receptor-ligand complexes between clathrin-coated pits on the cell surface and endosomes within the cell. We previously demonstrated (7) a possible biochemical correlate of such an intermediate step; specifically, the existence of two forms of intracellular EGF, one associated with plasma membrane (EGFpMV) and another associated with intracellular vesicles (EGFv). Further, acidstabile EGF appeared to accumulate more rapidly in the plasma membrane fraction than in the vesicular fraction.
Here we attempt to determine if EGFpMV is formed and degraded independently of EGFv or, conversely, if it represents an intermediate stage in the formation of cytosolic ~ ~ * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 'The abbreviations used are: EGF, epidermal growth factor; EGFcs, epidermal growth factor bound to cell surface receptors; EGFpMV, epidermal growth factor bound to receptors located in a plasma membrane-associated acid stable environment; EGFV, epidermal growth factor bound to receptors located in a light density vesicle; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; HEB, Hepes-buffered Eagle's medium containing 0.1% bovine serum albumin.
vesicles. Narrow-necked structures, such as those observed by Pastan and co-workers (4-6) might provide a sequestered environment in which EGF could not be dissociated by a low pH wash but would still be attached to the plasma membrane. The results demonstrate that upon internalization EGF first appears in a plasma membrane-associated state and then is transferred to a vesicular fraction. We speculate that EGFpMV represents EGF in narrow-necked plasma membrane-associated invaginations.

EXPERIMENTAL PROCEDURES
Materials-EGF was obtained from Biomedical Technologies (Cambridge, MA) and sodium ['251]iodide from Amersham Corp. All other chemicals were purchased from Sigma.
Cell Culture"l321N1 human astrocytoma cells were grown in Dulbecco's modified Eagle's medium (low glucose) containing 5% fetal calf serum in a humidified incubator at 8% COZ. Cells were subcultured at a density of lo4 cells/cm2 4 days prior to the experiment.
EGF lodination-EGF was iodinated with chloramine T (8) to a specific activity of 0.5 Ci/pmol EGF.
EGF Binding ASS~~S--'~~I-EGF was diluted to a specific activity of 0.1-0.2 Ci/pmol and used at a final concentration of 0.75 nM EGF. Binding was performed either at 37 or 4 "C.
When binding was performed at 4 'C, cells were first cooled in Hepes-buffered Eagle's medium containing 0.1% bovine serum albumin (HEB) for 30 min. lZ5I-EGF was then added for an additional 2 h. Cells were washed twice and HEB with 0.75 nM nonlabeled EGF was added. Cells were incubated at 37 "C for the time indicated, dishes were placed on ice, and cells were washed twice with HEB, incubated with 5 ml of 50 mM sodium acetate, pH 4.5 (2), or HEB for 5 min, washed twice with HEB, and treated for 20 min with 0.25 mg/ml concanavalin A. Cells were lysed by hypotonic shock, i.e. washed once with 1 mM Tris, 2 mM EDTA, pH 7.5, incubated for 20 min with Tris-EDTA, and scraped with a rubber policeman. Lysates were adjusted to 1 ml, layered on top of a sucrose step gradient (3.2 ml15%, 4.2 ml38%, 4.2 ml60% sucrose), and centrifuged for 30 min at 35,000 rpm using a SW40 rotor. The interfaces were separated and filtered and radioactivity was determined. Plasma membrane accumulated at the 38%/60% sucrose interface (7); intracellular vesicles accumulated at the 15%/38% sucrose interface (7). When, prior to lysis, the cells were washed with sodium acetate, pH 4.5, the 38%/ 60% interface represents EGFpMV and the 15%/38% interface represents EGFv; when cells were washed with HEB, the 38%/60% interface represents total plasma membrane-associated EGF. EGFcs was calculated as total plasma membrane-associated EGF minus EGFpMV. Membrane-associated "'I-EGF in the two interfaces accounted for 95% of the total membrane-associated 1251-EGF on the gradient.
When binding was performed at 37 "C, cells were incubated with 5 ml of HEB containing 0.75 nM '"1-EGF. Dishes were then placed on ice, and cells were washed twice with HEB and incubated for 5 min with 5 ml of 40 mM sodium acetate, 150 mM NaCl, pH 5.1. Cells were washed twice with HEB and either kept on ice or incubated at 37 "C in HEB with 0.75 nM EGF for the time indicated. Dishes were placed on ice, and cells were washed twice, incubated with concanavalin A, and further incubated as described above.
When total cell-associated EGF was determined, cells were incubated with 12'I-EGF, dishes were placed on ice, and cells were washed four times with HEB, incubated for 5 min with the appropriate buffer, washed twice again with HEB, and solubilized in 0.1 N NaOH. Radioactivity was determined in a Beckman y-counter.

RESULTS AND DISCUSSION
We have used two independent methods to measure the endocytosis of EGF. Others had shown that, whereas EGF bound to the cell surface is rapidly dissociated by washing with low pH buffers, internalized EGF remains cell-associated EGF in 9). Exposure of cells to EGF at 37 "C leads to a time-and temperature-dependent formation of acid-stabile (internalized) EGF. If cells are exposed to EGF at 4 "C, essentially all cell-associated EGF is acid-labile (cell surface). Upon warming to 37 "C, the portion of acid-stabile EGF increases rapidly as endocytosis occurs. EGF associated with the cell surface and that associated with intracellular membrane vesicles also can be distinguished using sucrose density centrifugation techniques (7, 10, 11). In this procedure cross-linking of cell surface proteins with concanavalin A allows efficient separation of plasma membrane (identified by adenylate cyclase activity) and intracellular membrane vesicles due to their different densities (11). In general the two procedures produce agreeable results. However, in the early phase of the internalization process a substantial portion of the acid-stabile EGF was found associated with the plasma membrane fraction (7). We have designated this form of EGF EGFpMV.
The goal of the present work was to determine if EGFpMV serves as a precursor for the acid-stabile EGF associated with intracellular membrane vesicles (EGFV). Fig. 1 shows a simplified scheme of the proposed sequential pathway of EGF from the cell surface, EGFcs, to plasma membrane-associated vesicles, EGFPMV, and subsequently to intracellular vesicles, internalization In A, cells were incubated for 5 min at 37 "C in HEB containing 0.75 nM '"I-EGF. Culture dishes were then placed on ice, and the cells were washed, incubated for 5 min with 1 ml of the buffer indicated, washed, and solubilized in 1 ml of 0.1 N NaOH. In B, cells were incubated on ice for 5 min with the buffer indicated, washed, incubated for 20 min at 37 "C with HEB containing 'T-EGF, placed on ice, washed, incubated with 40 mM sodium acetate for 5 min, washed, and solubilized in 1 ml of 0.1 N NaOH. Experiments were done twice in triplicate, mean values f S.D. are given. EGF appearance on the cell surface, in plasma membrane-associated and in free vesicles. Cells were incubated for 2 h at 4 "C in HEB containing 0.75 nM '"I-EGF. Cells were then washed and the medium was changed to HEB containing 0.75 nM nonlabeled EGF. Cells were incubated at 37 "C for the time indicated, culture dishes were placed on ice, and the cells were incubated with 5 ml of 40 mM sodium acetate, pH 4.5, or HEB for 5 min and then lysed. EGFpMV (0) and EGFv (A) were separated on sucrose step gradients. EGFcs (V) was calculated as total plasma membrane-associated lz5I-EGF minus EGFPMv. EGFv. The scheme indicates only those putative features of the overall pathway we focus on in this study. To determine if EGFpMV and EGFv are formed sequentially, it was necessary to determine if EGFv could still increase when cell surface acid-labile EGF (EGFcs) was removed, and if under these conditions an increase in EGFv was accompanied by a decrease in EGFpMV. Fig. 2 shows a detailed time course of the formation of EGFv and EGFpMV, which were separated on a sucrose step gradient (7). Cells were exposed to T -E G F at 4 "C to allow binding to the receptor. Internalization occurred during a subsequent incubation at 37 "C in the presence of nonlabeled EGF. In one set of cultures surface-bound EGF was removed by a low pH wash (3), while in a parallel set cells were washed at neutral pH to measure the total amount of cell-associated EGF. EGFcs was defined as the difference in EGF associated with the plasma membrane fraction of the sucrose gradients from neutral and acid-washed cells. The rate of loss of EGFcs and the rate of increase in EGFv and EGFpMv were determined from the data of two independent experiments, one of which is shown in Fig. 2.
EGFcs disappeared at a rate of 4.8 f 0.2 fmol/lO min/dish reaching a nondetectable level within 8-10 min. Concomitantly, the amount of EGFpMV increased at a rate of 2.2 f 0.9 fmol/lO min/dish reaching a maximum after approximately 8 min. The amount of EGFV increased at a rate of 0.9 f 0.1 fmol/lO min/dish, reaching a maximum significantly later than EGFpMV at 25 min. The additional loss of EGFcs was due to concomitant dissociation of the receptor-ligand complex estimated at a rate of 1.6 fmol/lO min/dish = 7.3 x 10-4/s.)2 The increase in EGFv between 8 and 25 min was associated with a decrease in EGFpMV (0.5 f 0.1 fmol/lO min/ dish), suggesting the latter as a source for EGFV. After reaching a plateau, vesicular EGF declined at a rate of 0.40 f 0.05 fmol/lO min/dish due to extrusion of lz5I-EGF and its degradation products.' Similar experiments using slightly different wash or preincubation conditions confirmed these results. These experiments demonstrate that the amount of EGFv increases after the cell surface has been depleted of EGF.
The protocol described above involves separating the binding and internalization reactions by making use of the differential effects of reduced temperature on these reactions. However, the possibility exists that EGFpMV is an artifact of the temperature transition from 4 to 37 "C. Thus, we attempted to demonstrate transfer of EGF between the two acid-stabile states during continuous exposure of cells to EGF at 37 "C.
This required a procedure for the rapid and complete removal of the cell surface-bound EGF without altering the process of endocytosis.
Under the standard conditions (2) used to dissociate the cell surface EGF receptor complex, i.e. hypotonic buffer, pH 4.5, subsequent internalization is inhibited (Table IB). Increasing the pH to 5.2 and using an isotonic buffer resulted in removal of cell EGFcs (Table IA) without preventing subsequent internalization of EGF. The latter conditions were used to remove cell surface-bound EGF after different periods of incubation with EGF at 37 "C. The subsequent kinetics of * C. Hertel and H. Affolter, manuscript in preparation.
appearance of EGFpMv and EGFv were studied. Fig. 3A (solid  symbols) shows the time-dependent increase in EGFPMv, while iodo-EGF is continuously present (solid symbols). The overall kinetics are faster than in the experiment shown in Fig. 2, because the cells were not pretreated at 4 "C (see also Table   I, A and B). Once EGFcs is removed (open symbols), the amount of EGFpMv decreases. Fig. 3B shows the rapid increase in EGFV. Furthermore, in contrast to EGFpMV, a further increase in EGFv is observed after depleting EGFcs. The results demonstrate that the formation of EGFpMV is dependent on the continuous presence of EGF on the cell surface, whereas EGFv can accumulate, at least transiently, in the absence of EGF on the cell surface. This additional increase in EGFv was most probably derived by transfer from EGFpMV.
The present communication describes results from two independent approaches that demonstrate a sequential translocation of surface bound EGF (EGFcs), first to a plasma membrane-associated acid-stabile form (EGFpMV) and then to an intracellular vesicular form (EGFV). We speculate that