Ligand-induced internalization and increased cell calcium are mediated via distinct structural elements in the carboxyl terminus of the epidermal growth factor receptor.

Signals that can mediate ligand-induced receptor internalization and calcium regulation are present in a 48-amino acid "calcium-internalization" domain in the C' terminus of the epidermal growth factor (EGF) receptor. The basis of calcium and internalization regulation signalled by this 48-amino acid sequence was analyzed using deletion and substitution mutant receptors. Cells expressing truncated receptors containing either the NH2- or COOH-terminal portion of the 48-residue domain displayed high affinity EGF-dependent endocytosis and receptor down-regulation. These endocytosis-competent EGF receptor mutants that lacked any autophosphorylation site were unable to increase the concentration of intracellular calcium. To investigate the role of self-phosphorylation in EGF-induced calcium mobilization, phenylalanine was substituted for the single autophosphorylated tyrosine residue in this region of an internalization-competent truncated receptor. The receptor-mediated calcium response was abolished, while ligand-dependent receptor internalization was unimpaired. These results demonstrate that EGF-dependent receptor endocytosis and calcium mobilization are separate events. Tyrosine self-phosphorylation is required for increased [Ca2+]i, while structural features distinct from autophosphorylation are required for receptor internalization.

In this article we investigate whether the structural requirements for EGF-induced receptor internalization and calcium responses are conferred by separate signals. Permanent cell lines expressing EGF receptors with deletions or specific mutations within the CaIn domain were analyzed for EGFinduced increases in [Ca"], and for ligand-induced internalization and down-regulation. Sequences critical for receptor internalization are distinct from a requirement of tyrosine self-phosphorylation for EGF-induced increases in [Ca'+],. These results indicate that ligand-induced increases in [Ca'+]{ and endocytosis are mediated by distinct events which depend on different sequence motifs, each of which is located in the regulatory C' terminus of the EGF receptor.

Construction and Expression of Mutant Receptors-
The expression plasmids containing the hEGF receptor cDNA, the kinase-inactive mutant M'", and C'-terminal truncated EGF receptors (~' 9 9 1 , c'973, and c'957) have been described (1,5 ) . A truncation mutant (~'1000) was obtained by removing the coding sequence for the last 186 amino acids of the EGF receptor. To generate the fusion plasmid f958/993-1022, a SalI site was created at residues 959 and 992 by site-directed mutagenesis according to the method of Kunkel (10). The sequences between the two SalI sites were deleted, and residues Val and Asp were introduced at the junction. In the plasmid c'1000F"''', the Tyr at residue 992 was substituted by Phe. All the mutant sequences were confirmed by the dideoxy chain termination method. Transfection and selection of monoclonal mouse B82 L cells expressing the various EGF receptor mutants were carried out as previously described (1). Colonies were screened using "'I-EGF binding (11).
Receptor Dorun-regulation~cells expressing normal or mutant EGF receptors were treated with 50 nM EGF for various times at 37 "C. The extent of receptor down-regulation was determined by ",-'I-EGF binding after acid stripping (12). Nonspecific binding, estimated in the presence of a 200-fold molar excess of unlabeled EGF, was less than 5% and was subtracted. Measurements were made in at least two independent clonal B82 cell lines expressing each receptor.
Internalization of EGF Receptors-The specific internalization rate (ke) in B82 cells expressing normal or mutant receptor was measured as described by Lund et al. (13). The affinity and capacity of the endocytic system for ligand-bound receptors was evaluated by plotting ke against the net internalization velocity measured at varying ligand plot (13).

concentrations. This generates a saturation of internalization (Satin)
Measurement of Intracellular Calcium-Cells were grown on coverslips in complete culture medium, then shifted to serum-free medium containing 0.5 mg/ml of bovine serum albumin a t least 15 h before experiments. Cells were loaded with the fluorescent CaY+ indicator fura-2 via the acetoxymethyl ester for 40-60 min at room temperature; quantitation was performed as described (1,14). Cells were exposed to 0.5% fetal bovine serum as a positive control. To document cell viability, 5 FM ionomycin was added to demonstrate ability to increase [Ca"],.
Irnrnunodetection of Autophosphorylated EGF Receptors-Cells were incubated without or with 50 nM EGF for 2 min a t 37 "C. After washing, cells were solubilized and the receptors were immunoprecipitated with 1 pM monoclonal mouse anti-EGF receptor 528 antibody (15). The immunoprecipitates were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blot analysis (9). T o detect receptor autophosphorylation, blots were incubated with "."I-PYZO monoclonal antiphosphotyrosine antibody a t 1:lOOO dilution (4). Membranes were then washed followed by autoradiography. EGF receptors were detected on the same blots by incubation with a rabbit anti-hEGF receptor antibody directed against the N'-terminal 13 amino acids of the hEGF receptor diluted 1:lOOOO in the same blocking buffer. The washed membranes were developed by the alkaline phosphatase-conjugate detection system (Promega).

Characterization of the Internalization Domain of the EGF
Receptor-Previous analysis of mutant EGF receptors with truncated C' termini indicated that receptor internalization could be restored by a region spanning residues 973-1022 (5), implying that these sequences are a critical component of the internalization signal. Examination of the primary amino acid sequence of this domain suggests distinct regions. The proximal portion (residue 973-991), rich in acidic residues, is highly negatively charged, while the distal portion (residue 992-1022) is a Ser/Thr-rich region. It has been established that there is a single autophosphorylation site at Tyr-992 (9) and there are sites of serine phosphorylation in the Ser/Thr-rich region (16). To evaluate the contribution of each subregion to the process of ligand-induced internalization and down-regulation, two mutant EGF receptors were constructed. One was truncated at residue 991 (~'991); the other was truncated at residue 958 with residues 993-1022 fused at the C' terminus (f958/993-1022). Plasmids expressing either transcription unit were transfected into mouse B82 cells that do not express the endogenous EGF receptor gene (17). Clonal cell lines expressing 2.5-10 x 10" receptors/cell were selected for characterization. Immunostaining with anti-EGF receptor serum confirmed expression of mutant receptors of the expected molecular weight (see Fig. 4) and receptors exhibited similar binding affinities for '"I-EGF. Both receptors exhibited biphasic Scatchard plots, with K,, values of 0.23 and 8.0 nM for c'991 and 0.16 and 6.6 nM for f958/993-1022 EGF receptors. The tyrosine kinase activity of these mutants was assayed by Western blotting extracts of EGF-stimulated cells using a monoclonal antiphosphotyrosine antibody. Both receptors showed enhanced phosphorylation of cellular substrates in vivo (data not shown), but did not undergo self-phosphorylation (see Fig. 4). These data indicate that the c'991 and f958/ 993-1022 receptors are processed normally, and are activated in response to EGF in uiuo.
The ability of these receptors to undergo ligand-induced endocytosis and down-regulation was investigated. Lund et al. (13) have developed quantitative methods for measurement of occupancy dependent changes in internalization rates. In-ternalization of EGF receptors is a saturable second order process that is induced by receptor occupancy. Internalization via a high affinity pathway requires the intrinsic protein tyrosine kinase activity of the receptor, while low affinity constitutive internalization is independent of kinase activity and of cytoplasmic receptor sequences (18). Data are presented as Satin plots, which are analogous to Scatchard plots, for holo and mutant EGF receptors. As shown in Fig. 1, the Satin plot of the holoreceptor is biphasic, indicating entry through both high affinity and constitutive endocytic pathways (13,18). In contrast, kinase-active receptors truncated to residue 973 show a slope of near 0, indicating these receptors lack sequences required for high affinity internalization. Similar results were observed for c'957, kinase-active EGF receptors (5). Addition of the sequences of the distal part of the CaIn domain (residues 993-1022) to the kinase core, f958/ 993-1022, restored high affinity endocytosis (Fig.  1). Similarly, EGF receptors containing the proximal portion of the CaIn domain (~' 9 9 1 ) also exhibited biphasic Satin plots (Fig.  l), indicating that sequences located between residues 973 and 991 are also capable of conferring high affinity endocytosis. Both the distal and proximal halves of the CaIn domain thus contain sequences that contribute to occupancy induced high affinity internalization of holo EGF receptors.
In agreement with the requirements for high affinity internalization, fusion of the 993-1022 sequence to the kinase core restored ligand-induced down-regulation (Fig. 2). Down-regulation of c'991 receptor was significantly greater than that of c'957 or c'973 EGF receptors, but was less effective than that observed for holoor f958/993-1022 EGF receptors. The distal residue (993-1022) of the CaIn domain thus appear more potent in determining receptor down-regulation than the proximal residues (973-991). Together, these results suggest that the C' terminus of the EGF receptor contains at least two independent signals responsible for ligand-induced endocytosis and down-regulation.
Uncoupling of Ligand-Induced Calcium Response from Receptor Internalization-The identification of two internalization elements in the C' terminus of the EGF receptor led  (13). The curve through the data was fit to a two-component internalization model (high affinity and constitutive internalization) using the simplex algorithm (18). Similar results were obtained in four independent experiments using different permanent transfections. us to further evaluate their role in the EGF-dependent calcium regulation conferred by the CaIn domain (5).
Because the EGF-induced calcium response is rapid and transient (1,19), accurate continuous measurements of intracellular calcium concentration are necessary. Ratio imaging of individual cells loaded with the fluorescent dye fura-2 was used to visualize the rapid effects of EGF on [Ca"], with the results of fluorescence quantitations being converted to [Ca'-],. B82 cells expressing holo EGF receptors showed a rapid response to EGF with [Ca"], increasing from 200 to 600 nM within 20 s (Fig.  3). C'-terminal truncation to c'1022 or to c'1000 resulted in similar kinetics of response to EGF, but with a slightly diminished increased [Ca"],, to 74% of that induced by the holoreceptor. In contrast, there was no EGF-induced change in [Ca"], in cells with receptors expressing either part of the CaIn domain (c'991 and f958/993-1022) ( Fig. 3 and Table I).
While the c'973 EGF receptor fails to signal EGF-induced changes in [Ca"],, previous studies using '"'Ca'' efflux or the fluorescent indicator indo-1 on cell populations gave conflicting results for the c'991 EGF receptor (5). As shown in Fig.  3, data obtained from the measurements of single cells using fura-2 imaging assay were in agreement with the '"Ca" efflux data, and indicated that c'991 EGF receptors did not cause an increase in [Ca"],. These results indicate that EGF-induced receptor internalization and increased [Ca'l'], are conferred independently, and that increased [Ca'l-], is not required for receptor down-regulation. This apparent independent mediation of these two EGF-induced early events was also TAHI.E I FXF-induced changes in fCo"], R82 cells expressing the indicated mutant EGF receptors were analyzed for EGF-induced changes in [Ca'.], as described in the legend to Fig. 3 . Results of repeated experiments are expressed as the mean 2 estimated standard deviation relative to the response in cells expressing holo EGF receptors. Each experiment contained internal controls using serum and ionomycin responses to validate cell viahility, dye loading, and ability to respond.  lanes 3 and 4 ) , f958/993-1022 (lunes 5 and 6 ) , and c'1000F""' (lanes 7 and 8 ) EGF receptors were treated without or with 50 nM EGF for 2 min. Cells were lysed and EGF receptors were quantitatively immunoisolated with 528 monoclonal IgG, electrophoresed on 7 . 5 5 sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and transferred to Immobilon. Upper ponel, identification of receptor tyrosine phosphate using"'1-PY20 monoclonal antiphosphotyrosineantibody. Lower panel, identification of EGF receptor using anti-EGF receptor serum.
investigated by treating cells with the membrane permanent compound BAPTA/AM, which is hydrolyzed to the active Ca" chelator BAPTA upon entry into cells (20,21). Although intracellular Ca" buffering by BAPTA was effective in blocking Ca" transients, down-regulation of holoreceptors was not affected (data not shown), providing further support for the conclusion that ligand-induced receptor internalization is independent of increased [Cas+],.
EGF-stimulated [Ca"], Requires EGF Receptor Autophosphorylation-The dramatic consequence of deletion of amino acid residues 1000-991 to extinguish EGF-stimulated increases in [Ca'"], indicated this region contained a potent signal for regulating [Ca'.],. Because 8 out of 9 amino acids in this region were included in f958/993-1022 EGF receptors which gave no calcium response to EGF, the amino acid at 992 (Tyr) was deduced to be the essential signal. Because Tyr-992 is the most N'-terminal tyrosine self-phosphorylation site in the EGF receptor (9), we tested the h-ypothesis that receptor self-phosphorylation was necessary for increased [Ca"],. A mutant EGF receptor truncated at residue 1000 and containing a Phe substitution at 992 (~'1000 F!'"") lacked detectable tyrosine self-phosphorylation. As shown in Fig. 4, cells expressing holo and ~'1000 EGF receptors exhibited EGF-induced self-phosphorylation in uiuo, whereas EGF Receptor Internalization and Calcium Response c'1000 F"!'2 and f958/993-1022 EGF receptors did not. The c'991 EGF receptors have previously been documented to lack any detectable self-phosphorylation in vivo (9). The ~'1000 F"'" EGF receptor was internalized and down-regulated to the same extent as c'1000 receptor (Figs. 1 and 2), but it was not competent to increase [Cay+], (Fig. 3 and Table I). These data indicate that self-phosphorylation is not required for EGF receptor internalization but is essential for coupling receptor activation to mechanisms that increase [Ca2+],.

DISCUSSION
Endocytosis of low density lipoprotein, transferrin, and cation-independent mannose 6-phosphate receptors is independent of ligand binding but is determined by specific cytoplasmic motifs that have been proposed to adopt a tight turn structure (22)(23)(24)(25). This class of receptors is constitutively clustered in coated pits and continuously rapidly internalized. In contrast, a second class of receptors that serve to activate signal transduction pathways as a consequence of ligand binding undergo ligand-induced conformational changes which alter their intrinsic enzymatic activity, cell surface distribution, internalization, and trafficking in cells. For example, occupancy induced internalization of EGF and insulin receptors requires both intrinsic protein tyrosine kinase activity and specific sequence motifs (5,18,26,27). Deletional analysis of the EGF receptor revealed that the sequences which are required for occupancy induced internalization via the high affinity saturable pathway are located in the C' terminus distal to the core kinase domain (5).
Analysis of progressive C'-terminal truncations revealed that deletion of residues 1022-973 abolished ligand-induced internalization, down-regulation, and EGF-induced increases in [Ca'.],. In this article we present evidence that this 48-amino acid segment contains two distinct regions that independently specify high affinity internalization and down-regulation. Neither of these two regions, located between residues 973-991 and residues 993-1022, contain sites of tyrosine self-phosphorylation. The 993-1022 sequence appears more potent in the process of receptor down-regulation than the 973-991 sequence. The function of both regions require the intrinsic protein tyrosine kinase activity of the EGF receptor because neither sequence functions when receptor kinase activity is lost through an M'" substitution.' Although receptor selfphosphorylation may facilitate exposure of these sequences in the context of the holoreceptor, it is not required for their function.
The ability of EGF receptors to effect increases in [Ca'+], requires a site of tyrosine self-phosphorylation. Tyr-992, the most proximal of the five self-phosphorylation sites in the C' terminus, is shown to be essential for this function in the context of C'-truncated EGF receptors; and it is likely that other self-phosphorylation sites may serve this function in EGF holoreceptors (28). The self-phosphorylated C' terminus of the EGF receptor is the binding site for phospholipase C-T (28), an association that occurs via the SH2 domains of phospholipase C-T (29). Phospholipase C-7 which is activated by holo EGF receptors in vivo (30), increases inositol trisphosphate, a central mediator of calcium mobilization (31). The necessity of a single site of tyrosine self-phosphorylation in the EGF receptor truncated to residue 1000 for increased '' C-P. Chang, unpublished data.
[Ca'+], fits the postulate that interaction with phospholipase C-T is essential to increase [Ca"],. The regulatory C' terminus of the EGF receptor contains multiple sites of tyrosine selfphosphorylation which may couple activated EGF receptors to phospholipase C-T and increase [Ca"],; Tyr-992 is demonstrated to be one such site.
Increased [Ca2+]; is a distinct process which is not required for ligand-induced high affinity internalization and receptor down-regulation, nor for mediating growth responses. Although the sequence determinants for calcium regulation are distinct from those for occupancy induced endocytosis, both require the intrinsic protein tyrosine kinase activity of the EGF receptor. It will now be of particular interest to define precisely the sequences that signal receptor internalization, and to define the basic machinery with which these sequences interact.