Internalization and Down-regulation of the Human Epidermal Growth Factor Receptor Are Regulated by the Carboxyl-terminal Tyrosines”

The C terminus of the epidermal growth factor receptor (EGF-R) contains three tyrosines (Y 1068, Y1148, and Y1173) which correspond to the major autophosphorylation sites. To investigate the role of the tyrosines in internalization and down-regulation of the EGF-R, mutational analysis was performed with receptors in which 1,2, or all 3 tyrosines were changed to phenylalanines. The triple point mutant EGF-R, expressed in NIH-31’3, exhibited low autophosphorylation in vivo, low biological and reduced kinase ac- tivities. Single and double point mutants were down-regulated, as well as wild type EGF-R in response to EGF showing a half-life of about 1 h. Degradation of the triple point mutant, however, was impaired and resulted in a half-life of 4 h in the presence of EGF. EGF-dependent down-regulation of surface receptors was decreased in the triple point mutant EGF-R as was internalization and degradation of EGF. The specific rate of internalization of the triple point mutant was reduced. By contrast, intracellular processing of ligand previously internalized at 20 “C was similar between wild type and mutant receptors. Taken together the data indicate that the delay in degradation observed in cells expressing the triple point mutant EGF-R can be attributed mainly to a slower removal from the cell surface. Our results show that in the full-length EGF-R ail three C-terminal tyrosines are necessary for rapid internalization, suggesting that autophosphorylation is required for efficient EGF-dependent receptor endocytosis. indicate that rapid


Internalization and Down-regulation of the Human Epidermal Growth Factor Receptor Are Regulated by the Carboxyl-terminal Tyrosines"
(Received for publicat.ion, December 6,1990) Kristian HelinS and Laura Beguinot The C terminus of the epidermal growth factor receptor (EGF-R) contains three tyrosines (Y 1068, Y1148, and Y1173) which correspond to the major autophosphorylation sites. To investigate the role of the tyrosines in internalization and down-regulation of the EGF-R, mutational analysis was performed with receptors in which 1,2, or all 3 tyrosines were changed to phenylalanines. The triple point mutant EGF-R, expressed in NIH-31'3, exhibited low autophosphorylation in vivo, low biological and reduced kinase activities. Single and double point mutants were downregulated, as well as wild type EGF-R in response to EGF showing a half-life of about 1 h. Degradation of the triple point mutant, however, was impaired and resulted in a half-life of 4 h in the presence of EGF. EGF-dependent down-regulation of surface receptors was decreased in the triple point mutant EGF-R as was internalization and degradation of EGF. The specific rate of internalization of the triple point mutant was reduced. By contrast, intracellular processing of ligand previously internalized at 20 "C was similar between wild type and mutant receptors. Taken together the data indicate that the delay in degradation observed in cells expressing the triple point mutant EGF-R can be attributed mainly to a slower removal from the cell surface. Our results show that in the full-length EGF-R ail three C-terminal tyrosines are necessary for rapid internalization, suggesting that autophosphorylation is required for efficient EGF-dependent receptor endocytosis.
The EGF' receptor (EGF-R) is a cell surface glycoprotein able to bind EGF and transforming growth factor 01, thereby mediating their mitogenic effects (1,2). Binding of EGF results in activation of rapid events including autophosphorylation, changes in membrane potential, phosphatidylinositol turnover, elevation of cytoplasmic Ca2+ and intracellular pH, and phosphorylation of cellular substrates (2-4).
Another very early event triggered by binding of EGF is down-regulation of both ligand and receptor (5,6). Prior to activation the EGF-R is diffusely distributed on the plasma membrane. Upon EGF addition the receptor accumulates in ." *This work was supported by grants from the Danish Cancer Society and the Danish Biotechnolo~ Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate t.his fact.
3 Recipient of a Danish Biotechnology Program fellowship.
For recycling receptors which are selectively localized in coated pits such as receptors for low density lipoprotein and transferrin, a consensus sequence important for internalization has been identified (16)(17)(18)(19). These receptors, however, permanently reside in coated pits, while growth factor receptors, like the EGF-R, only accumulate in the pit.s upon ligand binding to undergo endocytosis (20,21).
Although recent deletion analysis delimits a region essential for EGF-dependent receptor internali~tion (22), the molecular signals inducing internalization and degradation upon EGF binding are still poorly understood. There is substantial experimental evidence that points to a central role of phosphorylation in inducing receptor down-regulation. The EGF-R intrinsic tyrosine kinase activity is essential for receptor degradation (23,24); however, while one group has reported that the kinase-negative receptor is able to internalize in response to EGF (24, 25), another finds no internalization at all (22,23). In addition, the second group finds that microinjection of antiphosphotyrosine antibodies prevents EGF-induced endocytosis (26). Phosphorylation of EGF-R by other cellular kinases seems also to affect the normal pathway of receptor internalization. In the absence of EGF or transforming growth factor-a, activation of protein kinase C by phorbol esters induces phosphorylation of the EGF-R on Thr-654 (27, 28) and its internalization without degradation (29, 30). Furthermore, it has recently been reported that mutations of 2 amino acids of the EGF-R, threonine 669 and serine 671, which can be phosphorylated by a cellular kinase different from protein kinase C, also affect receptor internalization (31).
To directly test the role of autophosphorylation on EGFdependent down-regulation, we have used EGF-R mutants with point mutations in the three C-terminal autophosphorylation sites. The biological activity of the point mutants is gradually reduced in parallel with the substitution of the tyrosine residues.2 In this paper we report that mutation of all three autophosphorylation sites which results in a mutant with low autophosphorylation, reduced kinase activity, and low biological activity, drastically reduces EGF-dependent receptor internalization. These results indicate that in the full-length receptor all 3 C-terminal tyrosines are necessary to mediate rapid receptor internalization and suggests that K. Helin, T. Velu, P. Martin, W. C . Vass, G. Allevato, D. R. Lowy, and L. Beguinot, (1991) Oncogene, in press. phosphorylation of these sites may expose a domain required for internalization.

EXPERIMENTAL PROCEDURES
Materials--"'I-EGF (100-180 pCi/pg) was from Amersham Corp. "'I-Protein A (70-100 pCi/pg) and [35S]methionine (1100 Ci/mmol) were from Du Pont-New England Nuclear. Mouse EGF was from Bethesda Research Laboratories, fibronectin from Upstate Biotechnology Inc. Transferrin, cell culture reagents, serum, and G418 were purchased from Gibco. Bio-Rad kit and SDS-polyacrylamide gel electrophoresis reagents were from Bio-Rad. Staphylococcus aureus was from The Enzyme Center, and all other reagents were from Sigma. Polyclonal anti-EGF-R antibody Ab2913 was previously described (32). Affinity purified polyclonal antiphosphotyrosine antibody, a-Ptyr 4387 was kindly provided by Drs. P. P. Di Fiore and D.
Mutant Construction Transfection and Cell Culture-The EGF-R mutants were made by oligonucleotide site-directed mutagenesis of Tyr-1173, -1148, and -1068 into Phe as previously described.' A mutant with deletion of the last 123 amino acids was constructed (Dc123) which lacks, in addition to the 3 main Tyr, the fourth residue of autophosphorylation, Tyr-1086 (34, 35). Wild type and mutant cDNAs were cloned in the PC011 expression vector containing the neo' gene as described (36).
NIH-3T3 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 nM glutamine, penicillin, and streptomycin and 10% heat-inactivated newborn calf serum. Transfection, G418 selection, and generation of NIH-3T3 cells overexpressing the various EGF-R mutants have been described.' NIH-3T3 cells expressing between 1 X lo5 and 1.5 X lo6 wild type or mutant hEGF-R (MI 31-YYF, MI 37-FYY, MI 33-YFF, MI 34-FFF, and Dc123) per cell were used. For binding studies, wild type and MI 34 NIH-3T3 expressing, respectively, 4.5 and 3 X lo5 receptors/cell were also used as described in the legend for figures. Results were similar with cells expressing low and high numbers of EGF-R.
Immunoblotting-Cells were plated on fibronectin-treated dishes and when confluent starved overnight in DMEM supplemented with 5 pg/ml transferrin and lo-' M sodium selenite. The cells were treated with or without 50 nM EGF for 5 min and immediately lysed in 1% Triton X-100 containing 2 mM phenylmethylsulfonyl fluoride, 1% aprotinin, 5 mM sodium orthovanadate (Na,VO,), and 10 mM sodium pyrophosphate. Total proteins were determined by the method of Bradford (44) (Bio-Rad). Equal amounts of proteins (100 pg) were separated on a 7% SDS-polyacrylamide gel (371, electrophoretically transferred to nitrocellulose, and exposed to either anti-EGF-R antibody Ab2913 or antiphosphotyrosine antibody a-Ptyr 4387 and "' 1protein A according to the method of Towbin et al. (38).
Metabolic Labeling and EGF-R Half-life-Cells were plated at 5 X 10fi/35-mm dishes in complete medium and 8 h later metabolically labeled for 16 h with 250 pCi/ml of [35S]methionine in 1 ml of methionine-free DMEM with 5% fetal calf serum. Cells were washed twice, incubated in complete medium with or without 50 nM EGF for appropriate times at 37 "C, and lysed as previously described (7,29). Cell lysates with equal numbers of counts were immunoprecipitated using polyclonal anti-EGF-R antibody Ab2913 (32) and S. aureus. The immunoprecipitated proteins were analyzed on 7.5% SDS gels and developed by autoradiography. Quantitation of immunoprecipitated receptors from gels and of [35S]methionine incorporation in cell lysates by trichloroacetic acid was performed as described (29).
EGF Receptor Binding and Down-regulation-Cells were plated in 24-multiwell dishes at 1 X 105/well 24 h before binding. Monolayers were washed twice in serum-free DMEM and incubated for the indicated time with 100 nM EGF at 37 "C in 0.5 ml of binding medium (DMEM, 20 mM Hepes, pH 7.5, 0.1% bovine serum albumin). Cells were placed on ice and cell-associated EGF was removed by acid wash (0.2 M acetic acid, 0.5 M NaC1, pH 4.0) by the procedure of Haigler et al. (39). The monolayers were carefully washed once with ice-cold DMEM containing 10% newborn calf serum, twice with binding medium (7), and then incubated for 2 h on ice in binding medium containing 0.25 nM "'I-EGF (final concentration). After the incubation, cells were washed four times in ice-cold binding medium and solubilized with 1 ml of 1 M NaOH. Radioactivity was counted in a mini y-counter. Results are the average of triplicate determinations. Nonspecific binding, estimated in the presence of 1 pg/ml unlabeled EGF, was <5% and subtracted.
EGF Internalization, Degradation, and Specific Internalization Rate-Cells were plated in 24-multiwell dishes and cultured as de-scribed above. Monolayers were washed twice with ice-cold serumfree DMEM and once with binding medium and kept on ice for 2 h in 0.5 ml of binding medium containing 1 nM of lZ5I-EGF. Medium was then removed and cells were rapidly warmed and incubated at 37 "C for the indicated time. Cells were again placed on ice, incubation times with ice-cold binding medium. Surface-associated EGF was medium was removed and saved, and monolayers were washed four determined by acid-stripping (39) and internalized EGF was determined by measuring the cell-associated radioactivity after solubilization in 1 ml of 1 M NaOH. The amount of degraded (trichloroacetic acid soluble) and intact EGF (precipitable by trichloroacetic acid) was determined by precipitating the incubation medium with 10% trichloroacetic acid and counting supernatant and pellet. As control, the proportion of trichloroacetic acid-precipitable counts in the native lz5I-EGF was determined to be 90-95%.
To study the chase of internalized EGF, monolayers were incubated with 10 nM lz5I-EGF for 1 h at 20 'C followed by 10 min at 37 "C to allow internalization but not degradation (25, 40). Surface-bound EGF was then removed by an acid wash, cells were washed as described above, incubated at 37 "C for different times with 100 nM unlabeled EGF, and then processed as above.
Internalization rate was measured according to Wiley and Cunningham (41). Briefly, 1 nM lz5I-EGF was given to cells in binding medium pre-equilibrated at 37 "C, and the cells were incubated at 37 "C. At 2-min intervals, plates were rapidly rinsed on ice, and surface-associated and internalized EGF was determined by acid stripping as described above.

In Vivo Autophosphorylation of Wild Type and Mutant EGF
Receptors-To clarify the role played by autophosphorylation in EGF-R internalization and down-regulation, we have used EGF-R mutants in which the three main C-terminal autophosphorylation sites (Y1173, Y1148, and Y1068) were mutated into phenylalanines as single, double, or triple mutants. Human EGF-R mutants, constructed by in vitro site-directed mutagenesis and named MI 31 (F1173), MI 37 (F1068), MI 33 (F1173-F1148), and MI 34 (F1173-Fl148-Fl068) were transfected in NIH-3T3 cells expressing low numbers of endogenous mouse receptors ( 4 x 104/cell). Scatchard analysis showed similar binding constants for wild type and point mutant receptors. As recently described, the EGF-dependent mitogenic and transforming activity of these mutant EGF-R was gradually decreased in parallel with the removal of the main autophosphorylation sites2 I n vitro and in vivo kinase activity was also affected by tyrosine substitutions suggesting that autophosphorylation modulates EGF-induced kinase activity.', NIH-3T3 cells expressing mutant and wild type receptors were stimulated with EGF, cell extracts blotted on nitrocellulose filters, and the phosphorylated receptor revealed by an anti-Ptyr antibody, as shown in Fig. 1 (panel A ) . As a control the amount of receptor was determined by an anti-EGF-R antibody and shown to be similar for all the mutants (Fig. 1,  panel B ) . Tyrosine 31 (lanes 2 ) , MI 33 (lanes 3 ) , and MI 34  (lanes 4 ) were treated without (-) or with (+) 50 nM EGF for 5 min at 37 "C prior to lysis. One hundred micrograms of total protein lysates were separated on a 7% SDS-polyacrylamide gel and transferred to a nitrocellulose filter. A, phosphotyrosine containing EGF-R were identified with a polyclonal a-Ptyr antibody coupled to "..' Ilabeled protein A. R, total EGF-R were identified with a polyclonal CY-EGF-R antibody coupled to '""Iprotein A. Sizes are shown in kilodaltons. Autoradiography was for 38 h for the a-Ptyr ( A ) and 16 h for the a-EGF-R blot ( R ) . Autoradiograms exposed in the linear range were scanned and quantified. The content of phosphotyrosine incorporated in the EGF-stimulated EGF-R bands ( A ) was then normalized for the total amount of EGF-R present in each lysate ( B ) . MI 31 (YYF) showed a 2.5-fold decrease, and MI 33 (YFF) and MI 34 (FFF) EGF-R a 5-fold decrease in relative phosphotyrosine content compared to wild type EGF-R.  60 (lane 4 ) , and 120 min (lane 5) at 37 "C, or immediately lysed (lane 1 ). EGF-Rs were immunoprecipitated with a polyclonal a-EGF-R antibody, separated by 7.5% SDS-polyacrylamide gel, and subjected to autoradiography as described under "Experimental Procedures." R, the relative number of residual receptors for each mutant is plotted as a function of incubation time with EGF. Quantitation was performed both by direct scanning of autoradiograms exposed in the linear range and by scintillation counting of the solubilized '"S-labeled receptor's band.

EGF-dependent phosphatidylinositol hydrolysis and cytoplasmic [Ca'+] elevation:'
EGF-dependent Half-life of Wild Type and Mutant EGF Receptors-To determine if EGF-dependent degradation of EGF-R was affected by the tyrosine substitutions, mutant cells were metabolically labeled with [:''S]methionine and EGF was added for different times. The EGF-R was immunoprecipitated and the amount of residual receptor determined (Fig. 2). In the absence of EGF, no difference in the stability of mutant receptors was observed, the half-life being approximately 10-12 h in NIH-3T3 cells (Table I). In the presence of EGF, similar half-lives (approximately 1 h) were observed for wild type and point mutants with 1 (1173F and 1068F) or 2 tyrosine substitutions (1173F-1148F). These results are in agreement with our previous finding with the corresponding deletion mutants (9) and with the data on single mutants by Honegger et al. (42). In contrast, degradation of the triple point mutant was significantly delayed in  EGF-R (0) were incubated with 100 nM EGF at 37 "C for the indicated time. The EGF was removed by an acid wash, and residual EGF-Rs at the cell surface were evaluated by incubating the cells on ice for 2 h in binding medium containing 0.25 nM ""I-EGF. Cells were then processed as described under "Experimental Procedures." Results are the average of triplicate determinations and are expressed as percentage of binding a t time 0. The presented experiment has been carried out twice with cells expressing low numbers of EGF-R (1 X 10' and 1.5 X 10')). Identical results were also obtained twice with cells expressing high numbers of receptors (4.5 and 3 X 10').
the presence of EGF, resulting in a prolonged half-life (4 h compared to 1 h), (Fig. 2 and Table I). This result suggests that phosphorylation of the receptor on all 3 tyrosines is important for the rapid disappearance of the receptor. A small but consistent change was also observed with the comparable deletion mutant Dc123, which lacks all four autophosphorylation sites together with the last 123 amino acids, whose half-life was 1.5 h. In agreement with this finding, Chen et al. (22) have reported that a similar deletion mutant is removed slightly less efficiently from the surface when compared with wild type. The relative efficient degradation of the deletion mutant compared to the triple point mutant suggests either that conformational changes in the C terminus of the truncated receptor or compensatory autophosphorylation sites allow sufficient processing of the deletion mutant.
EGF-R Down-regulation of Wild Type and Triple Point Mutant EGF-R-Several possibilities could account for the prolonged half-life of the triple point mutant EGF-R. The initial internalization step might be affected so that mutant receptors were removed from the membrane more slowly; alternatively, lysosomal delivery might be impaired so that receptors would recycle several times or accumulate in an intracellular compartment. T o investigate whether endocytosis was impaired in the mutant receptor, three types of internalization experiments were performed.
As shown in Fig. 3   Internalization and degradation of EGF by wild type and MI 34 (FFF) EGF receptors. Wild type ( A ) and MI 34 ( B ) cells were kept on ice for 2 h with 1 nM lz5I-EGF. Medium was removed, cells were rapidly warmed to 37 "C and incubated for the indicated time. EGF bound to the surface (0) was determined by an acid wash, internalized EGF (0) was determined by solubilizing the cells with 1.0 M NaOH and degraded EGF (W) in the supernatant was measured as trichloroacetic acid-soluble counts. Initial counts bound on the surface were 5250 and 3650 cpm for the wild type and MI 34 cells, respectively, using cells expressing high numbers of receptors. Each point is the average of triplicate measurements. Values are expressed as percentage of initial bound counts at time 0. The results have been confirmed twice with cells expressing both high and low numbers of EGF receptors.
on the plasma membrane measured by lZ5I-EGF binding.
The wild type receptor was rapidly down-regulated; after 30 min only 20% of the receptors were still present on the plasma membrane. In contrast, the triple point mutant was removed much more slowly; after 1 h approximately 60% of receptors were still present at the cell surface. Single (MI 31 and MI 37) and double point mutants (MI 33) were downregulated as well as the wild type receptor (data not shown) in agreement with having half-lives similar to wild type receptor. In response to 100 nM 4/3-phorbol12-myristate 13-acetate, however, the triple point mutant was down-regulated with kinetics similar to wild type hEGF-R," suggesting that the impaired down-regulation was EGF-specific.
Internalization, Degradation, and Internalization Rate of EGF-Direct internalization and degradation of the ligand was then measured under conditions where 10-30% of total receptors were occupied. Cells, preincubated with 1 nM lZ5I-EGF for 2 h on ice, were washed and incubated a t 37 "C for different times up to 4 h and residual surface EGF, internalized EGF, and degraded EGF in the supernatant were determined. As shown in Fig. 4, in wild type expressing cells EGF was rapidly removed from the cell surface and accumulated inside the cells with a maximum (80%) after 30 min. After 2 h 60-70% of the total counts were found in the medium as trichloroacetic acid-soluble counts, indicating efficient EGF degradation (Fig. 4, panel A ) . In cells expressing the triple point mutant receptor, internalization of EGF proceeded at a much slower rate: after 30 min, more than 50% of the ligand K. Helin and L. Beguinot, unpublished observations. was still at the cell surface (Fig. 4, panel B ) . The slower internalization was reflected in a more persistent accumulation of ligand inside the cells. After 2 h, more than 45% of the radioactivity was still cell associated and less than 40% was found degraded in the medium. Ligand dissociation was similar for wild type and triple mutant EGF-R under these conditions and could not explain the difference in ligand removal from the cell surface between wild type and triple point mutant. After 10 min, dissociation accounted for 30 and 37% of total counts, respectively, in wild type and MI 34, after 30 min, 42 and 50% of total.
In order to directly localize the receptor, a time course similar to that shown in Fig. 4 was performed by immunofluorescence using a monoclonal anti-EGF-R antibody (EGF-Rl). After 1 or 2 h of EGF treatment, the triple mutant EGF-R showed a more heterogeneous pattern of localization, diffuse cell surface fluorescence, as well as localization in peripheral endocytic vesicles with little staining in vesicles in the perinuclear r e g i~n .~ In contrast, after 2 h little fluorescence was observed in cells expressing the wild type receptor and all localized in lysosomes, in agreement with the short halflife of the normal receptor in the presence of EGF.
To better quantitate the initial delay in internalization of the triple point mutant receptor, cells were incubated with 1251-EGF at 37 "C and surface bound versus internalized EGF was measured in the first 10 min (Fig. 5). Calculation of the internalization rate constant (41) revealed a 2-fold reduction for the mutant receptor compared to wild type (0.038 min" versus 0.075 min"). The reduced internalization rate of MI 34 mutant was observed both at lower (1 nM, Fig. 5) and higher concentrations of EGF ( 5 nM, data not shown), suggesting that it is an intrinsic property of the mutant receptor, independent of the proportion of occupied receptors.
Taken together these data show that the triple point mutant was internalized more slowly and this effect resulted in a prolonged half-life in the presence of the ligand. This suggests that phosphorylation at the proper sites in the full-length receptor is important for rapid internalization leading to rapid degradation.
Intracellular Processing-We also tested whether intracellular processing of the mutant receptor was affected by the substitution of the C-terminal tyrosines. Cells were preincubated with '251-EGF for 1 h at 20 "C to induce internalization but not degradation (40), surface-bound EGF removed by an Specific internalization rate of occupied wild type ( WT) and MI 34 (FFF) EGF receptors. Wild type (0) and MI 34 (0) cells pre-equilibrated at 37 "C were incubated with 1 nM "'I-EGF at 37 "C. At 2-min intervals plates were rapidly rinsed at 0 "C, and surface-associated and internalized lZ5I-EGF was determined as described under "Experimental Procedures." Each point is the average of triplicate determinations. The data are presented as a ratio of internalized EGF/surface EGF versus time. Internalization rate constant for the wild type EGF-R was 0.075 min" and 0.038 min" for the FFF receptor. Identical internalization constants were obtained with 5 nM lZ51-EGF and with cells expressing both low and high numbers of EGF receptors. incubated with 10 nM lz5I-EGF at 20 "C, followed by 10 min incubation at 37 "C. Cells were placed on ice and surface-bound EGF was removed by an acid wash. The cells were incubated at 37 "C in binding medium containing 100 nM unlabeled EGF for the indicated times and then chilled again on ice. Trichloroacetic acid-precipitable (0) and nonprecipitable (B) counts, surface-associated (0) and internalized (0) counts were determined as described under "Experimental Procedures." Internalized counts at time 0 were 8540 and 2580 cpm for wild type and MI 34 cells, respectively. Values are expressed as percentage of initial internalized counts. Each point represents the mean of triplicate measurements. The experiment was performed twice with cells expressing high amounts of EGF-R and repeated with similar results for cells with low numbers of EGF-R.

Time (mln)
acid wash, and cells warmed to 37 "C in the presence of 100 nM unlabeled EGF. Removal of surface EGF and addition of excess cold EGF were used to directly follow only the previously internalized ligand and to avoid rebinding of lZ51-EGF. As shown in Fig. 6, in cells expressing the mutant receptors the total amount of internalized lz51-EGF was %fold less than for wild type cells. This is in agreement with the previous experiments (quantitation is given in the legend for Fig. 6) and indicates that at 20 "C the internalization of the mutant receptor was also impaired. However, the kinetics of disappearance of cell-associated radioactivity and of reappearance of EGF in the medium, as trichloracetic acid-soluble and -precipitable counts, were similar in both wild type and mutant cells. This finding suggests that once internalized, EGF was processed similarly and that under these conditions apparently intact EGF was released from both types of cells. The EGF released in the medium comigrated with native lZ5I-EGF on SDS gel (data not shown). Whether these trichloroacetic acid-precipitable counts represent biologically active or a partially inactive form of EGF (11) is not clear at present since it has proven difficult to perform binding studies with such a low amount of lz5I-EGF. While it is possible to conclude a similar processing of internalized ligand and to suggest a similar processing for wild type and mutant receptors, we cannot directly demonstrate recycling of EGF and of receptors.
In the presence of monensin, a 20-30% decrease of surfacebound EGF and a similar increase of intracellular EGF was observed both in wild type and mutant cells.5 This result also indicates that the behavior of wild type and mutant EGF-R were similar and suggests that some recycling was occurring for both types of receptors. While Felder et al. (25) showed a difference in intracellular processing between wild type and kinase-negative EGF receptors, indicating that in response to EGF only the kinase-negative mutant was recycling, we find no significant difference in the intracellular processing between the wild type and the triple point mutant receptors. Our results complement the findings of Chen et al. (23) and Honegger et al. (24) that EGF-R kinase activity is essential for receptor degradation and the additional observation of Chen et al. (22) that a kinase-negative EGF-R is unable to undergo ligand-induced internalization.
A likely possibility is that EGF-dependent phosphorylation of the C-terminal tyrosines, induced by activation of the kinase, allows exposure of an internalization domain which is apparently required for rapid receptor internalization (22). Deletion mutants lacking more than this extreme C terminus, as the one used by Chen et al. (22), might not require this regulatory function and could be in a conformation which allows internalization. Indeed, in the 123 amino acids Cterminal deletion, Dc123, which lacks all four autophosphorylation sites, EGF-induced internalization and degradation occurred quite efficiently (Fig. 2), suggesting that the internalization domain may be sufficiently exposed to serve its role. Alternatively, since the deletion mutant bears at least one compensatory site for autophosphorylation, phosphorylation on Tyr-992 (43) (or on an as yet unidentified Tyr) could be a sufficient signal for this purpose. Phosphorylation of this residue does not appear to take place in the triple point mutant.' Interestingly Tyr-992 is located just downstream of the putative internalization domain (22).
It is also possible that kinase activity is essential since rapid EGF-R internalization requires phosphorylation of substrates. The triple point mutant has a lower kinase activity in vitro and in vivo toward substrates'* and this low activity might account for the reduced internalization. Regardless of the molecular mechanism, however, we can conclude that in the full-length protein the C-terminal tyrosines are important for rapid internalization.
In conclusion, the C-terminal tyrosines seem to play a dual role in the regulation of the EGF receptor activity. We have recently reported that they are important to positively modulate the overall biological activity of the intact EGF-R and to activate its kinase function. Activation of EGF-R via autophosphorylation also serves as a signal for down-regulation of the mitogenic signal, triggering the removal of receptors from the surface and thereby creating a delicate control mechanism for EGF-R action.