A Leucine-based Determinant in the Epidermal Growth Factor Receptor Juxtamembrane Domain Is Required for the Efficient Transport of Ligand-Receptor Complexes to Lysosomes*

Ligand binding causes the epidermal growth factor (EGF) receptor to undergo accelerated internalization with eventual degradation in lysosomes. The goal of this study was to investigate the molecular basis of endocytic sorting, focussing on post-internalization events. We have identified a sequence located between amino acid residues 675 and 697, encompassing a dileucine motif at residues 679 and 680, that enhances endosome-to-lysosome transport when conformational restraints in the EGF receptor carboxyl terminus are removed by truncation. The same dileucine motif is also necessary for efficient lysosomal transport of ligand-occupied full-length EGF receptors. A L679A,L680A substitution diminished the degradation of occupied full-length EGF receptors without affecting internalization but had a significant effect on recycling. Rapid recycling of mutant receptors resulted in reduced intracellular retention of occupied EGF receptors and delayed down-regulation of cell surface receptors. We propose that the L679A,L680A substitution acts primarily to impair transport of ligand-receptor complexes through an early endosomal compartment, diverting occupied receptors to a recycling compartment at the expense of incorporation into lysosome transport vesicles. We also found that mutant receptors with truncations at the distal half of tyrosine kinase domain (residues 809–957) were not efficiently delivered to the cell surface but were destroyed in an endoplasmic reticulum-associated degradative pathway.

Ligand binding causes the epidermal growth factor (EGF) receptor to undergo accelerated internalization with eventual degradation in lysosomes. The goal of this study was to investigate the molecular basis of endocytic sorting, focussing on post-internalization events. We have identified a sequence located between amino acid residues 675 and 697, encompassing a dileucine motif at residues 679 and 680, that enhances endosometo-lysosome transport when conformational restraints in the EGF receptor carboxyl terminus are removed by truncation. The same dileucine motif is also necessary for efficient lysosomal transport of ligand-occupied fulllength EGF receptors. A L679A,L680A substitution diminished the degradation of occupied full-length EGF receptors without affecting internalization but had a significant effect on recycling. Rapid recycling of mutant receptors resulted in reduced intracellular retention of occupied EGF receptors and delayed down-regulation of cell surface receptors. We propose that the L679A,L680A substitution acts primarily to impair transport of ligand-receptor complexes through an early endosomal compartment, diverting occupied receptors to a recycling compartment at the expense of incorporation into lysosome transport vesicles. We also found that mutant receptors with truncations at the distal half of tyrosine kinase domain (residues 809 -957) were not efficiently delivered to the cell surface but were destroyed in an endoplasmic reticulum-associated degradative pathway.
Epidermal growth factor (EGF) 1 and related polypeptides elicit biological responses through binding to specific cell surface receptors belonging to the ErbB family of protein-tyrosine kinases (reviewed in Ref. 1). EGF regulates the intracellular trafficking of ligand-EGF receptor (EGFR) complexes, causing accelerated internalization from clathrin-coated pits, retention in endosomes, and ultimately degradation in lysosomes (reviewed in Ref. 2). Ligand-regulated EGFR down-regulation therefore modulates cellular responses to growth factors by controlling the duration of signaling from the cell surface and by transporting activated EGFRs to intracellular compartments where they continue to signal (reviewed in Ref. 3). In addition, because EGFR is the only ErbB receptor that undergoes ligand-induced down-regulation (4), signal transduction by different members of the ErbB family may be regulated by compartmentalization in the endocytic pathway. The importance of EGFR endocytic transport to normal proliferation is exemplified by the fact that EGFRs that fail to internalize have been associated with cell transformation (5) and tumorigenesis (6).
EGFR down-regulation is a complex process regulated by many factors, including ligand occupancy, receptor aggregation, tyrosine kinase activity, endosomal acidification, and intrinsic sorting signals (2). It is also clear that the multiple transport steps in the endocytic pathway are facilitated by different molecular interactions (7)(8)(9)(10). Internalization is regulated by endocytic codes located in the EGFR carboxyl-terminal domain that are functionally interchangeable with the internalization signal of the transferrin receptor (11). Internalization also involves interactions between cytoplasmic sequences in activated EGFRs and plasma membrane clathrin AP-2 adaptor proteins (7,(11)(12)(13)(14)(15). Although a tyrosine-based signal encompassing carboxyl-terminal domain residues 973-977 mediates AP-2 binding in vitro, the physiological relevance of this interaction is unclear, because EGFRs with an internal deletion of this region undergo normal ligand-induced internalization (16). Interestingly, AP-2 also interacts with two EGFR substrates: eps15, which is constitutively associated with AP-2 in vivo (17), and SHC, which binds AP-2 in vitro (18). Another signaling intermediate required for efficient EGFR internalization is Grb2, which binds to activated EGFRs either directly or as part of a SHC-Grb2 complex (1). It has been proposed that Grb2 provides a phosphoinositide-dependent link to dynamin (19,20), a GTPase that regulates endocytosis (16,21). Taken together, these studies indicate that EGFR internalization from clathrin-coated pits is facilitated by multiple interactions occurring simultaneously.
Following internalization, ligand-receptor complexes are diverted from a recycling pathway to lysosomes (10). Endosometo-lysosome sorting presumably involves the signal-mediated transfer of ligand-receptor complexes to vesicles called endosomal carrier vesicles or multivesicular bodies (ECV/MVBs), which transport material to late endosomes (reviewed in Ref. 22). Although candidate sorting sequences have been identified in the carboxyl half of the cytoplasmic domain, a consensus has not been reached regarding the structural and enzymatic requirements for transporting EGFRs to lysosomes (9, 23, 24).
One candidate lysosomal sorting molecule, SNX-1, has been identified that binds at the distal border of the tyrosine kinase domain (see Fig. 1 and Ref. 25).
The goal of this study was to further characterize the molecular basis for selective transport of ligand-EGFR complexes to lysosomes. This was accomplished by analyzing a series of EGFR proteins with progressive truncations encompassing the entire cytoplasmic domain to test the hypothesis that cryptic sorting sequences normally masked in unoccupied full-length EGFRs would be exposed in truncated receptors. In contrast to previous studies, we analyzed sequences in the juxtamembrane domain as well as the carboxyl terminus for two reasons. First, the EGFR cytoplasmic domain has numerous consensus leucine-based signals implicated in lysosomal targeting of a number of membrane proteins (26 -29) located throughout the cytoplasmic domain. Second, the juxtamembrane domain contains a sorting signal that regulates EGFR basolateral delivery in polarized Madin-Darby canine kidney cells (30), suggesting that this region may have a broader role in vesicular transport. We have found that a sequence located between amino acid residues 675 and 697 in the cytoplasmic juxtamembrane region enhances endosome-to-lysosome transport of truncated EGFRs. Residues Leu 679 and Leu 680 , which conform to a leucine-based sorting signal, were shown to be a critical determinant for efficient lysosomal transport of cytoplasmically truncated EGFRs. This same motif was also required for the efficient transport of ligand occupied full-length EGFR complexes to lysosomes.

EXPERIMENTAL PROCEDURES
Mutagenesis-EGFR cytoplasmic truncation and substitution mutants were made using polymerase chain reaction (PCR) to modify EGFR coding sequences cloned in pCB6 ϩ , a eukaryotic expression plasmid containing a human cytomegalovirus regulatory region, transcription termination and polyadenylation signals from the human growth hormone gene, an SV40 origin of replication and early region promoterenhancer, and a neomycin resistance gene (reviewed in Ref. 31). To create cDNAs encoding receptors with cytoplasmic truncations, stop codon substitutions were made at codons for Pro 675 , Ala 698 , Val 810 , Pro 886 , Gln 958 , Tyr 974 , Leu 993 , and Val 1023 . Forward primers (listed below) were designed to anneal to sequences 5Ј to a novel restriction enzyme site (in parentheses) in the EGFR coding sequence. Reverse mutagenic primers (listed below) were designed to create a premature stop codon (in bold) 400 -500 nucleotides downstream of the forward primer, as well as a restriction site (underlined and in parentheses) compatible with the pCB6 ϩ polylinker. PCR fragments were gelpurified, digested at sites incorporated in the PCR products, and ligated to pCB6 ϩ /EGFR digested with the same restriction enzymes. cDNA-encoded proteins were named based on the carboxyl-terminal amino acid residue in the EGFR coding region (i.e. cЈ-674 has a P675STOP substitution). cDNAs with stop codon substitutions for Arg 652 and Arg 724 have been described previously (32).
The cЈ-697 cDNA containing a L679A,L680A substitution was constructed using the "megaprimer" PCR method (33). Briefly, an initial PCR reaction was performed to convert codons for Leu 679 and Leu 680 to alanine residues (underlined) using a mutagenic primer 5Ј-CTC-CCAACCAAGCTGCGAGGATCTTGAAGGAAACTGA-3Ј and a flanking primer identical to the reverse primer listed above for cЈ-697. A subsequent PCR reaction was carried out using the product of the first reaction and a second flanking primer (same as the forward primer for cЈ-674). The final PCR product was digested with BsmI and ClaI and subcloned into pCB6 ϩ /EGFR. The L679A,L680A substitution was introduced into the full-length molecule by subcloning a 188-nucleotide Eco72I-EcoRI fragment from pCB6ϩ/697 containing the L679A,L680A substitution into an EGFR cDNA ligated to a pBK-CMV phagemid (Stratagene Cloning Systems, La Jolla, CA).
PCR primers were designed using the DNASTAR software package (DNASTAR, Inc., Madison, WI). PCR amplifications were carried out using a RoboCycler 40 Temperature Cycler (Stratagene). Sequences of PCR products were verified by dideoxy chain termination DNA sequencing using a Sequenase II kit from U. S. Biochemical Corp.
Transient Transfections and Permanent Cell Lines-COS-1 monkey cells expressing endogenous EGFRs were used for transient transfections. Log-phase cells were seeded at a density of approximately 5 ϫ 10 3 cells/cm 2 24 h prior to transfection. Cells were rinsed twice with serumfree Dulbecco's modified Eagle's medium (DMEM) and then transfected with 10 g of plasmid DNA using Lipofectin reagent (Life Technologies, Inc.) according to the manufacturer's instructions. The DNA-Lipofectin mixture was replaced with DMEM supplemented with 10% fetal bovine serum and 2 mM glutamine 5 h later. When multiple dishes were transfected with the same plasmid, cells were pooled and replated 24 h later to avoid plate-to-plate variability in transfection efficiency.
NR6 cells lacking endogenous EGFRs (34,35) were used to produce permanent cell lines expressing human EGFRs. Cells were transfected with human EGFR cDNAs using Lipofectin reagent as described previously. Transfected cells were grown for 10 -14 days in selection medium containing G418 (0.8 mg/ml Geneticin; Life Technologies, Inc.). Cells were then enriched for human EGFR expression by sterile sorting on a flow cytometer (Cytofluorograph IIs; Ortho Instruments, Westwood, MA) using the EGFR-specific monoclonal antibody (mAb) EGF-R1, which detects an extracellular peptide core epitope in human EGFRs (36). EGF-R1 also cross-reacts with endogenous EGFRs expressed in COS-1 cells.
Cell Labeling, Immunoprecipitation, and Western Blots-Cells were rinsed twice with methionine-and cysteine-free minimal essential medium and then labeled with 50 Ci of Express Protein Labeling mix (1175 Ci/mmol; New England Nuclear Research Products, Wilmington, DE) per ml of methionine-and cysteine-free minimal essential medium supplemented with 10% dialyzed fetal bovine serum and 0.2% BSA. In some experiments, labeling medium was replaced with chase medium consisting of serum-free DMEM supplemented with nonradioactive methionine (0.75 mg/ml) and cysteine (1.2 mg/ml), and cells were incubated for an additional period of time before harvesting. Cells were lysed with 1% (w/v) Nonidet P-40 in 0.1 M Tris, pH 6.8, supplemented with 15% (w/v) glycerol, 2 mM EDTA, and 1 mM EGTA, 0.2 mM phenylmethylsulfonyl fluoride, and 1 M leupeptin. EGFRs were immunoprecipitated with EGF-R1 adsorbed onto protein A-Sepharose CL-4B beads (Sigma). Samples were solubilized with Laemmli buffer and separated by SDS-polyacrylamide gel electrophoresis (PAGE) (37). Gels were treated with En 3 Hance (New England Nuclear) for fluorography. Labeled proteins were quantitated by phosphorstorage autoradiography (Molecular Dynamics, Sunnyvale, CA). The percentage of radioactivity remaining was plotted as a function of time on a semi-log plot, and receptor half-lives (t1 ⁄2 values) were calculated by linear regression analysis. For Western blotting, proteins resolved by SDS-PAGE were transferred to nitrocellulose according to standard procedure (38). EGFRs were detected by immunoblotting with an affinity purified rabbit polyclonal antibody specific to amino acids 1005-1016 in the EGFR cytoplasmic domain (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Phosphotyrosine-containing proteins were detected by immunoblotting with an anti-phosphotyrosine-horseradish peroxidase conjugate (Transduction Laboratories, Lexington, KY). 125 I-EGF Cross-linking-Receptor grade mouse EGF (Toyobo Biochemicals, Osaka, Japan) was labeled with 125 I (carrier-free, Ͼ350 mCi/ml; New England Nuclear) using chloramine-T. Cells were rinsed three times with ice-cold DMEM supplemented with 0.2% BSA and then incubated with approximately 10 nM 125 I-EGF for 2 h at 4°C. Cells were rinsed again with the DMEM/BSA solution to remove unbound ligand and then incubated with 2 mM disuccinimidyl suberate (Pierce) in a solution of 0.1 M HEPES, pH 7.4, supplemented with 0.12 M NaCl, 0.05 M KCl, 8 mM glucose, and 1.2 mM MgSO 4 for 15 min at room temperature. The chemical cross-linker was quenched by a 5-min incubation with 0.05 M Tris, pH 7.4, at room temperature. Cell lysates were prepared using 1% Nonidet P-40 exactly as described above, and total cell protein was separated by SDS-PAGE.
Northern Blots-Cells were trypsinized, lysed with TRIzol reagent (Life Technologies, Inc.) for 5 min at room temperature, and then extracted with chloroform:isoamyl alcohol (24:1 v/v). RNA was precipitated with isopropanol, washed with 70% ethanol, and resuspended in distilled H 2 O treated with 0.01% diethylpyrocarbonate. 20 g of total RNA was denatured, fractionated by electrophoresis in a 1.4% agarose/ formaldehyde gel, and transferred to Biotrans (ϩ) nylon membrane (ICN Pharmaceuticals, Inc., Costa Mesa, CA) using standard techniques (39). An oligonucleotide complementary to nucleotides 2227-2247 in the human EGFR cDNA was labeled with [␣-32 P]dCTP (3000 Ci/mmol; Amersham Pharmacia Biotech) using Klenow reagent and a Random Primed DNA Labeling kit from Boehringer-Mannheim. Blots were prehybridized for 2 h at 65°C with a solution containing 1 mM EDTA, 250 mM sodium phosphate, pH 7.2, 7% SDS, 1% BSA, and 50 g/ml salmon sperm DNA. Blots were hybridized with 1 ϫ 10 6 cpm/ml of 32 P-labeled oligonucleotide in the same solution for 18 h at 65°C. Blots were washed three times with a solution of 20 mM sodium phosphate, pH 7.5, 1 mM EDTA, and 10% BSA at 65°C, air-dried, and exposed to film for autoradiography.
Treatment with Brefeldin A-Cells were preincubated in medium supplemented with vehicle (1 l of MeOH/ml) or brefeldin A (BFA) (5 g/ml) for 1 h, pulse-labeled with Express Protein Labeling mix for 30 min, and then incubated in nonradioactive chase medium for periods up to 2 h before lysis with Nonidet P-40. Labeling and chase media were also supplemented with vehicle or BFA.
Digestion with Endoglycosidase H-Pulse-labeled EGFRs collected by immunoprecipitation were solubilized with 50 l of a solution of 1% SDS and 5% ␤-mercaptoethanol in 0.1 M sodium citrate, pH 5.5. 20-l aliquots were mixed with an equal volume of distilled H 2 O and then incubated for 18 h at 37°C with 1 milliunits endoglycosidase H (endo H; 40 units/mg enzyme; Boehringer-Mannheim). Equal aliquots were subjected to sham digestions.
Internalization Assays-Cells were seeded at a density of 5 ϫ 10 5 cells/well in six-well tissue culture plates 48 h before each assay and refed with DMEM supplemented with 25 mM HEPES and 0.2% BSA (D/H/B) 24 h later. Cells were rinsed twice with cold D/H/B and then incubated with 250 ng/ml of 125 I-labeled 528 mAb Fab or 1-100 ng/ml of 125 I-labeled EGF diluted in the same medium for 2 h at 4°C. The EGFR-specific 528 mAb Fab recognizes an extracellular peptide epitope and is widely used, for example to track intracellular transport of unoccupied EGFRs (24). 528-Fabs (gift of Starla Glick, Dept. of Pediatrics, Case Western Reserve University) were iodinated exactly as described above. Cells were warmed to 37°C for periods up to 20 min and then rinsed rapidly with cold D/H/B to remove unbound 528-Fab or ligand. Cells were incubated with an acid-stripping solution containing 50 mM glycine-HCl, pH 3.0, 100 mM NaCl, 2 mg/ml polyvinylpyrrolidone, and 2 M urea for 6 min on ice. The stripping efficiency was greater than 95%. This solution was collected for ␥ counting to estimate surfacebound radioactivity, and cells were solubilized with 1 N NaOH to estimate cell-associated radioactivity. Internalization is represented as the percentage of total radioactivity (counts released from the cell surface by acid stripping plus counts remaining cell-associated after acid-stripping) associated with the interior of the cell. Internalization rates were calculated by linear regression analysis.
Recycling Assays-Cells seeded on six-well plates were rinsed with cold D/H/B and incubated with 1 ng/ml 125 I-EGF at 4°C for 1 h. Cells were then rinsed twice with cold D/H/B and allowed to internalize receptors for 10 min at 37°C. Cells were rinsed with cold D/H/B, and 125 I-EGF remaining on the cell surface was removed by a 2.5-min mild acid wash (0.2 M sodium acetate, 0.5 M NaCl, pH 4.5) as described elsewhere (23). 125 I-EGF-loaded cells were incubated with 100 ng/ml of nonradioactive EGF at 4°C for 1 h to saturate surface receptors, and then switched to 37°C for 0 -40 min to allow for receptor trafficking. At the end of each incubation period, cells were placed on ice, and media were collected to determine the amount of degraded and intact 125 I-EGF. This was followed by a 2.5-min harsh acid wash (pH 2.8) to determine the amount of surface-bound 125 I-EGF. Cells were then solubilized with 1 N NaOH to determine the amount of intracellular 125 I-EGF. To separate intact 125 I-EGF from degraded 125 I-EGF products, trichloroacetic acid and phosphotungstic acid were added to collected medium to the final concentrations of 3 and 0.3%, respectively. This mixture was incubated at 4°C for 30 min and centrifuged to collect precipitates. Precipitates, solubilized with 1 N NaOH, and supernatants were used to calculate the amount of intact and degraded 125 I-EGF, respectively. The amount of recycled 125 I-EGF was calculated by summing the radioactivity appearing on the cell surface and in the medium (intact) and was expressed as fraction of the total radioactivity present in the cell and media. The stripping efficiency of the pH 4.5 and 2.8 acid solutions was greater than 90 and 95%, respectively. The efficiency of precipitation with trichloroacetic acid and phosphotungstic acid was greater than 95%.
Ligand-induced EGFR Down-regulation-To measure down-regulation of cell surface EGFRs, cells were incubated with nonradioactive EGF for periods up to 2 h. Cells were rinsed two times with ice-cold D/H/B, and cell surface-associated EGF was removed by incubating cells for 2.5 min on ice with the same acid-stripping buffer (pH 4.5) described in the previous paragraph. Cells were rinsed two times with D/H/B and then incubated with 100 ng/ml 125 I-EGF for 1 h at 4°C. Surface-bound 125 I-EGF was removed by incubating cells with acid stripping buffer (pH 2.8) for 2.5 min on ice, and radioactivity was determined by ␥ counting. To measure EGFRs down-regulation, cells were pulse-labeled for 1 h, and then chased with serum-free DMEM containing excess nonradioactive amino acid precursors for 3 h. Cells were stimulated with 100 ng/ml nonradioactive EGF and harvested for EGFR immunoprecipitation at various time points. Labeled EGFRs were resolved by SDS-PAGE.

Expression and Stability of Human EGFRs with Cytoplasmic
Truncations-To study the relative contribution of EGFR cytoplasmic subdomains to protein stability, premature stop codons were introduced at 10 different sites throughout this region to expose cryptic sorting signals. Premature stop codons were always ligated adjacent to transcription termination and polyadenylation signals in the expression vector to rule out effects due to variability in 3Ј noncoding sequences. Three truncations were made in the carboxyl terminus (Fig. 1A). The most distal carboxyl truncation (cЈ-1022) removes consensus tyrosine NPXY-type signals located near the carboxyl-terminal end. Carboxyl truncations to residues 992 and 973 remove one or both of the known endocytic codes located in the CaIn domain, respectively (7,10). In addition to internalization, this region also regulates ligand-mediated calcium responses (7). Truncation to residue 957 (cЈ-957) deletes sequences up to the distal border of the tyrosine kinase core domain, exposing the binding site for the putative lysosomal sorting molecule SNX-1 (25). Four truncations were made within the kinase catalytic core domain (cЈ-697, cЈ-723, cЈ-809, and cЈ-885), and another two truncations were created within the juxtamembrane domain (cЈ-674 and cЈ-651).
To test the hypothesis that sorting signals that regulate EGFR transport in the endocytic pathway to lysosomes were active in truncated molecules, we first determined the steadystate expression of receptor proteins depicted in Fig. 1A in transiently transfected COS-1 cells. EGFRs were either immunoprecipitated from metabolically labeled cells (Fig. 1B) or identified by 125 I-EGF cell surface cross-linking (Fig. 1C). Both of these methods also detect endogenous monkey EGFRs, which served as an internal control for wild-type EGFR mobility on SDS-PAGE gels. These analyses showed that cDNAs coding for proteins with truncations in the carboxyl-terminal domain (cЈ-972, cЈ-992, and cЈ-1022), in the proximal half of the kinase catalytic core (cЈ-697 and cЈ-723), or in the juxtamembrane domain (cЈ-651 and cЈ-674) formed stable products that were transported to the cell surface. In contrast, low levels of cЈ-957 receptors were seen after metabolic labeling, and this receptor protein was not readily detectable following 125 I-EGF cross-linking, suggesting these receptors are probably degraded in the biosynthetic pathway. Products encoded by two other EGFR cDNAs with truncations involving the distal half of the kinase domain (cЈ-885 and cЈ-809) could not be detected by either method, suggesting these molecules are either not produced or are rapidly degraded.
To further characterize the protein stability of cytoplasmically truncated EGFRs, transiently transfected COS-1 cells were metabolically labeled for 3 h and then incubated in chase medium for either 3 or 15 h (Fig. 2). Consistent with results in Fig. 1, cЈ-809, cЈ-885 and cЈ-957 receptors were not detectable at either time point. EGFR proteins with truncations in the carboxyl-terminal domain exhibited stability similar to full-length EGFRs, as did cЈ-674 and cЈ-651 receptor proteins with truncations in the juxtamembrane domain. However, stability of two proteins truncated near the proximal border of the kinase domain, cЈ-697 and cЈ-723, was markedly reduced compared with endogenous monkey EGFRs. Analysis of EGFRs with cytoplasmic truncations therefore identified two classes of receptors with reduced protein expression: those that failed to undergo efficient transport to the cell surface (cЈ-809, cЈ-885, and cЈ-957) and those that were transported to the cell surface but had reduced stability compared with wild-type EGFRs under basal conditions (cЈ-697 and cЈ-723).
Reduced Stability of EGFR Proteins with Cytoplasmic Truncations Is Mediated by Two Distinct Post-translational Mechanisms-Based on results in Fig. 1, we hypothesized that EGFRs with truncations in the distal kinase domain (cЈ-809, cЈ-885, and c-957) were degraded in the biosynthetic pathway. To test that hypothesis, we first showed that cDNAs encoding these truncated EGFRs produced stable mRNAs as judged by Northern blot analysis (Fig. 3A). We then asked whether the fungal metabolite BFA affected protein turnover, because the endoplasmic reticulum (ER)-associated degradative pathway is BFA-insensitive (40). Transfected COS-1 cells that had under-gone a mock-treatment or a preincubation with BFA were pulse-labeled for 45 min and then incubated in chase medium for periods up to 2 h (Fig. 3B). BFA treatment had no effect on turnover of cЈ-809 receptors (Fig. 3B) or cЈ-885 receptors (not shown) and little effect on turnover of cЈ-957 receptor (Fig. 3B). The molecular weight of endogenous EGFRs was reduced in BFA-treated cells, consistent with BFA-induced blockade of Golgi-mediated carbohydrate processing. These data suggest that although a fraction of cЈ-957 receptors appear to be degraded in a post-Golgi compartment, receptors with truncations in the distal half of the kinase domain are mostly disposed of by a BFA-insensitive, ER-associated degradative pathway (41).
Similar to truncated receptors such as cЈ-973 (Fig. 3C), which exhibits normal stability, cЈ-697 (Fig. 3C) and cЈ-723 (not shown) receptors were not subject to BFA-insensitive, ER-associated degradation. In addition, cЈ-697 and cЈ-723 receptors were transported through the Golgi complex with approximately the same kinetics as wild-type EGFRs. This was shown using endo H digestion to distinguish mature EGFRs containing a mixture of complex and high mannose-type N-linked oligosaccharides from EGFR precursors containing only high mannose N-linked oligosaccharides (36,42). Because endo H only cleaves high mannose oligosaccharides (43), endo H resist-

FIG. 1. Steady-state expression of EGFRs with cytoplasmic truncations in COS-1 cells. A, schematic showing organization of transmembrane (TM) and
cytoplasmic domains for wild-type EGFR (WT) and corresponding domains encoded by cDNAs with premature stop codons. Locations of a dileucine motif (Leu679,Leu680) described in this study, the SNX-1 binding site (25), the CaIn domain (7), and conserved NPXY motifs (52) are also shown. B and C, COS-1 cells transfected with cDNAs encoding each of the proteins shown in A were assayed for EGFR expression 48 h post-transfection. B, cells were metabolically labeled for 4 h, lysed with Nonidet P-40, and immunoprecipitated with an EGFR mAb specific for an extracellular epitope common to monkey and human EGFRs. C, intact cells were chemically cross-linked with 125 I-EGF. Immunoprecipitates (B) or total cellular protein (C) were separated on 7.5% SDS-PAGE gels. Locations of full-length 170-kDa and cytoplasmically truncated EGFRs are indicated on the right. Molecular mass standards: myosin, 200 kDa; ␤-galactosidase, 116.3 kDa; phosphorylase B, 97.4 kDa. Jx, juxtamembrane. Fig. 1A were metabolically labeled for 3 h starting at 48 h post-transfection. Labeling medium was replaced with chase medium, and cells were lysed with Nonidet P-40 either 3 or 15 h later. EGFRs were immunoprecipitated with a receptor-specific mAb, and immunoprecipitates were separated on 7.5% SDS-PAGE gels. Locations of full-length 170-kDa and truncated EGFRs are indicated on the left. WT, wild type; Jx, juxtamembrane.

FIG. 2. Metabolic turnover of EGFRs with cytoplasmic truncations. COS-1 cells transfected with EGFR cDNAs shown in
ance correlates with transit through Golgi compartments where complex oligosaccharides are formed (44). Cells were pulse-labeled for 45 min and then harvested at 30-min intervals during a nonradioactive chase for immunoprecipitation with an EGFR mAb and endo H digestion. cЈ-697 receptors, truncated EGFRs exhibiting normal stability (i.e. cЈ-651 or cЈ-674), and endogenous monkey EGFRs had all acquired endo H resistance by 90 min of chase (Fig. 4). Similar results were obtained for cЈ-723 receptors (not shown). Taken together, these data suggest that cЈ-697 and cЈ-723 receptors exit the ER and undergo biosynthetic transport with normal kinetics. Once at the plasma membrane, however, these receptors exhibit reduced half-lives compared with wild-type EGFRs.
Internalization of cЈ-697 Receptors-Because cЈ-674 receptors do not have a shortened half-life, we hypothesized that EGFR residues 675-697 encompass a cryptic signal that regulates sorting in the endocytic pathway. To determine whether this region contains a cryptic internalization signal, we made permanent NR6 cell lines expressing cЈ-697 or cЈ-674 receptors. Similar to results obtained with COS-1 cells, cЈ-697 receptors exhibited reduced stability in NR6 cells compared with receptors truncated to residue 674 (not shown). The uptake of a nonactivating 125 I-labeled EGFR mAb Fab was similar in cells expressing either cЈ-697 or cЈ-674 receptors (Fig. 5). These data suggest that enhanced turnover of cЈ-697 receptors is not due to accelerated internalization. We therefore hypothesized that truncation to residue 697 unmasks a potential endosomal sorting signal able to divert basally internalized cЈ-697 receptors from a recycling pathway to lysosomes.
Role for Leu 679 -Leu 680 in cЈ-697 Receptor Degradation-To understand the molecular basis of the enhanced turnover of cЈ-697 and cЈ-723 receptors, we examined the amino acid sequence of EGFR residues 675-697 to identify potential consensus sorting signals (Fig. 6). Residues Leu 679 and Leu 680 were of particular interest because leucine-based motifs are known to regulate lysosomal trafficking of several membrane proteins (reviewed in Ref. 45). Furthermore, Leu 679 -Leu 680 and adjacent amino acids (residues 679 -683) are identical to sequences at the carboxyl terminus of an adenovirus early region 3 membrane protein called E3-13.7 (46). Because E3-13.7 re-routes recycling EGFRs to ECV/MVB lysosomal transport vesicles, we hypothesize that E3-13.7 has co-opted an intrinsic sorting signal normally used during lysosomal transport of ligand-EGFR complexes (47). We therefore tested the potential role of the Leu 679 -Leu 680 motif in the reduced stability phenotype of cЈ-697 receptors, by changing this dileucine to a dialanine using PCR- based site-directed mutagenesis. Receptor half-lives were determined by harvesting cells beginning 3 h after the end of a pulse label to allow time for transport of labeled protein to the plasma membrane. The EGFR immunoprecipitates were then quantitated by phosphorstorage autoradiography. All EGFRs examined exhibited first-order exponential decay kinetics (Fig.  7). The half-life of the unoccupied endogenous COS-1 receptor was calculated to be approximately 25 h, consistent with values reported in other studies (48). In contrast, half-lives for cЈ-723 and cЈ-697 receptors were reduced to 9 and 12 h, respectively. The half-life for cЈ-697 receptors with a L679A,L680A substitution (cЈ-697/L679A,L680A), however, was indistinguishable from the half-life for endogenous receptors (Fig. 7). These results support the hypothesis that Leu 679 -Leu 680 is part of a cryptic lysosomal sorting signal exposed in cЈ-697 and cЈ-723 receptors.
Role of Leu 679 -Leu 680 in Ligand-induced EGFR Down-regulation-To test the hypothesis that a Leu 679 -Leu 680 -based sorting signal is utilized during ligand-induced transport, we expressed the L679A,L680A substitution in the context of a fulllength EGFR (EGFR/L679A,L680A). As shown in Fig. 8A, permanent NR6 cell lines transfected with plasmids encoding either wild-type EGFR or EGFR/L679A,L680A expressed 170-kDa proteins that were reactive with EGFR-specific antibodies directed against both extracellular and cytoplasmic domains. Under normal circumstances, efficient ligand-induced EGFR down-regulation requires intrinsic tyrosine kinase activity (2). Although Leu 679 -Leu 680 lies outside of the kinase catalytic domain, it was nevertheless possible that the L679A,L680A substitution would have an adverse effect on ligand-induced kinase activity. To test that possibility, the NR6 cell lines were assayed for ligand-induced receptor autophosphorylation by Western blotting using a phosphotyrosine-specific antibody. As shown in Fig. 8B, EGF stimulation induced tyrosine phosphorylation of both proteins to a similar extent and also caused a molecular weight shift reflecting the increase in tyrosine phosphorylation. These data indicate that the L679A,L680A substitution does not interfere with ligand-induced EGFR activation.
Two approaches were used to determine the effect of L679A,L680A substitution on ligand-induced EGFR down-regulation. First, NR6 cells expressing either wild-type EGFR or EGFR/L679A,L680A were assayed for steady-state 125 I-EGF binding following a preincubation with unlabeled EGF for periods up to 2 h. Down-regulation of wild-type EGFRs reached a plateau of 60% in cells that had been preincubated with 1 ng/ml ligand for 1 h and remained at the same level with longer preincubations (Fig. 9). In contrast, EGFR/L679A,L680A receptors were down-regulated by less than 20% after 1 h of preincubation with unlabeled ligand and continued to undergo down-regulation at later time points (Fig. 9). The second approach for analyzing ligand-induced down-regulation was to measure EGFR half-lives. Cells that had been incubated in chase medium for 3 h after a 45-min pulse-label were stimulated with ligand for periods up to 2 h, and cell lysates were immunoprecipitated using an EGFR-specific mAb (Fig. 10). As shown in Fig. 10, the pool of radiolabeled receptors was rela-FIG. 6. Amino acid sequence and predicted structure of EGFR residues 675-697. A, organization of EGFR cytoplasmic domain and amino acid sequence of a putative signal encompassing Leu 679 -Leu 680 . Also shown are two other putative lysosomal sorting signals: residues 945 and 957 encompassing the binding site for SNX-1 in the kinase catalytic core domain (25) and residues 1022-1123 (Ly) in the carboxylterminal domain (23). Amino-terminal post-translational phosphorylation sites (mitogen-activated protein kinase sites Thr 669 and Ser 671 ) and acidic residues (Glu 673 ) are characteristic of other known consensus leucine-based motifs. An amphipathic helix predicted for this region is highlighted by cross-hatched bar. B, amphipathic helical representation of residues 675-697, with charged residues (ϩ/Ϫ) aligned on one side of the helix and hydrophobic residues (in italics) on the other side. The L679A,L680A substitution is also indicated. Immunoprecipitates were transferred to nitrocellulose and immunoblotted (IB) using a second EGFR-specific antibody directed against a carboxyl-terminal epitope. B, cell lysates were immunoprecipitated with a biotin-conjugated phosphotyrosine (PY) antibody. Immunoprecipitates were resolved by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with an horseradish peroxidase-conjugated phosphotyrosine antibody. tively constant in cells expressing EGFR/L679A,680A during a 2-h incubation with EGF, compared with cells expressing wildtype EGFRs where the majority of receptors were degraded. Interestingly, we consistently observed an increased amount of radioactivity associated with EGFR/L679A,680A at the 60-min time point, suggesting that these molecules may have enhanced solubility in Nonidet P-40. These data therefore indicate that the mutant EGFRs are degraded less efficiently than wild-type EGFRs following ligand occupation.
The Role of L679A,L680A on Internalization of Full-length EGFRs-We next asked whether the inability of ligand to down-regulate EGFR/L679A,680A as efficiently as wild-type EGFR was associated with differences in internalization. To measure basal internalization, cells were incubated with 250 ng/ml 125 I-Fab for 2 h at 4°C and then brought to 37°C for periods up to 15 min. Internalization is represented as the percentage of total radioactivity (cell surface plus cell interior) associated with the interior of the cell as a function of time. As shown in Fig. 11A, the ligand-independent internalization of EGFR/L679A,L680A was indistinguishable from that of wildtype EGFRs. When ligand-induced internalization was monitored by measuring 125 I-EGF uptake, both cell lines exhibited a similar internalization rate using two different concentrations of ligand (1 or 100 ng/ml), as shown in Table I. However, cells expressing the mutant EGFR did not accumulate intracellular ligand to the same extent as cells with wild-type EGFRs (Fig.  11B). The steady-state amount of ligand internalized by cells expressing the wild-type EGFR was approximately 50%, compared with approximately 30% for cells expressing EGFR/ L679A,L680A. These data suggest that the L679A,L680A substitution does not interfere with ligand-dependent internalization but that the mutant receptors have a shorter endosomal retention time than wild-type EGFRs.
The Role of L679A,L680A on Recycling of Full-length EGFRs-Because ligand-occupied mutant receptors had a longer half-life than wild-type EGFRs, we hypothesized that the reduced endosomal retention exhibited by EGFR/L679A,L680A was due to increased recycling. This hypothesis was tested by measuring 125 I-EGF recycling kinetics in NR6 cells expressing either wild-type EGFR or EGFR/L679A,L680A. Recycling was initiated by incubating cells that had been preloaded with 125 I-EGF (1 ng/ml) with an excess of nonradioactive EGF for up to 40 min at 37°C. As shown in Fig. 12, the L679A,L680A substitution was associated with increased ligand recycling compared with wild-type EGFR. Recycling rates were 4.52% min Ϫ1 for occupied mutant receptors, versus 2.72% min Ϫ1 for occupied wild-type receptors (Table I). In addition to having a faster recycling rate, the amount of recycled 125 I-EGF in cells expressing EGFR/L679A,L680A reached a maximum of approximately 55% after 20 min of incubation. In contrast, only 40% of 125 I-EGF-wild-type EGFR complexes had recycled in the same time interval. Together with data in the previous sections, this suggests that the L679A,L680A substitution inhibits the degradation of occupied full-length EGFRs by diverting ligand-receptor complexes to a recycling pathway without affecting internalization. DISCUSSION In this study, we have characterized the molecular basis for selective transport of ligand-EGFR complexes to lysosomes. By examining cytoplasmically truncated EGFRs exhibiting reduced protein stability, we identified two categories of receptor proteins with enhanced degradation phenotypes: those exhibiting enhanced degradation after reaching the cell surface and those undergoing rapid degradation shortly after biosynthesis. In addition, we demonstrated that a dileucine motif in the EGFR juxtamembrane domain is required for efficient lysosomal transport of truncated EGFRs as well as ligand occupied full-length EGFRs.
The mutant receptors cЈ-697 and cЈ-723, which terminate in the proximal half of the kinase catalytic core domain, both exhibited enhanced endosome-to-lysosome transport after delivery to the cell surface. Because cЈ-674 receptors did not have a shortened half-life, we hypothesized that EGFR residues 675-697 encompass a cryptic signal that regulates sorting in the endocytic pathway. Examination of the amino acid sequence between residues 675 and 697 revealed a potential leucine-based sorting signal at amino acids, Leu 679 -Leu 680 . To test the relative importance of Leu 679 -Leu 680 , these residues were changed to alanines in the context of a cЈ-697 truncation. The L679A,L680A substitution abolished the enhanced degradation phenotype, suggesting that Leu 679 -Leu 680 acts to divert recycling molecules to lysosomes when conformational restraints in the carboxyl-terminal regulatory domain are removed by truncation.
To test the hypothesis that Leu 679 -Leu 680 is part of a signal that regulates lysosomal sorting of ligand-receptor complexes, the L679A,L680A substitution was also studied in the context of a full-length receptor. Identical to wild-type EGFR, EGFR/ L679A,L680A mediated slow uptake of a nonactivating EGFR mAb Fab fragment consistent with constitutive membrane turnover (8). Ligand binding induced intrinsic tyrosine kinase activity and led to accelerated internalization of mutant and wild-type EGFRs to similar extents. These results indicate that Leu 679 -Leu 680 is not required for internalization of full-length EGFRs, in agreement with findings from other investigators who have shown that the EGFR juxtamembrane domain is not important for ligand-induced receptor uptake (7,11).
Although wild-type and mutant receptors facilitated the rapid accumulation of intracellular ligand, steady-state levels of accumulated ligand were substantially lower in cells expressing the mutant EGFR than in cells expressing wild-type EGFR. The reduced ability of cells expressing mutant receptors to accumulate intracellular ligand was also evident in downregulation assays monitoring the disappearance of cell surface ligand binding sites. Additionally, mutant EGFRs were not as efficiently transported to lysosomes as wild-type EGFRs, as shown by differences in the biosynthetic half-lives of ligandoccupied EGFRs. EGFR/L679A,L680A receptors did, however, exhibit faster recycling kinetics than wild-type EGFRs. These data suggest that reduced ligand accumulation and down-regulation exhibited by EGFR/L679A,L680A receptors is due to enhanced recycling of internalized receptors from endosomes.
A closer examination of the amino acid sequence surrounding Leu 679 -Leu 680 reveals that this region conforms to other known leucine-based signals by several criteria. For example, the targeting activity of leucine-based motifs is often influenced by an acidic residue four or five residues amino-terminal to the leucine pair (49). In addition, leucine-based signals may be regulated by post-translational modification, because they often have nearby phosphorylation sites (45). The putative signal encompassing Leu 679 -Leu 680 has both an appropriately distanced acidic residue (Glu 673 ) as well as nearby mitogen-activated protein kinase phosphorylation sites (Thr 669 and Ser 671 ) (Fig. 6). Although leucine-based signals implicated in lysosomal transport were originally identified in membrane proteins with relatively short (20 -30 amino acids) cytoplasmic tails, critical leucine motifs also have been found in juxtamembrane regions of other membrane proteins with more extensive cytoplasmic domains, such as the T-cell receptor CD3 ␥ subunit (29) and the insulin receptor (50).
The altered transport of the EGFR/L679A,L680A is probably TABLE I Intracellular transport kinetics for wild-type and mutant human EGFRs expressed in NR6 cells Basal and nonsaturating ligand-induced (1 ng/ml) internalization rates (K In ) were calculated by linear regression analysis of data in Fig. 11, panels A and B, respectively. Data for deriving ligand-induced K In under saturating conditions (100 ng/ml) are not shown. Ligand-induced recycling rates (K Re ) were calculated by linear regression analysis of data from Fig. 12. Ligand-induced EGFR half-lives (t 1/2 ) were calculated by linear regression analysis of values derived from PhosphorImager quantitation of data in Fig. 10 11. Internalization of EGFR-specific 125 I-Fab or 125 I-EGF by NR6 cell lines expressing human EGFRs. NR6 cells were preincubated for 2 h at 4°C with either EGFR-specific 125 I-Fabs (250 ng/ml) (A) or 125 I-labeled EGF (1 ng/ml) (B) and then switched to 37°C for times indicated in the figure. After removing surface-bound Fab or ligand by acid-stripping, cells were solubilized with 1 N NaOH to determine internalized radioactivity. Internalization is represented as the percentage of total radioactivity associated with the interior of the cell. The values are the means Ϯ S.D. (A, n ϭ 6; B, n ϭ 6). Some standard error bars are obscured by symbols. WT, wild type.
FIG. 12. 125 I-EGF recycling kinetics in NR6 cell lines expressing full-length human EGFRs. NR6 cells were preincubated for 2 h at 4°C with 125 I-labeled EGF (1 ng/ml) and then switched to 37°C for 5 min to allow for internalization. After removing surface-bound ligand by mild acid-stripping, 125 I-EGF-loaded cells were incubated with a 100-fold excess of unlabeled EGF for 1 h at 4°C and then incubated at 37°C for the indicated periods of time. At the end of each incubation, media were collected to determine the amount of intact and degraded 125 I-EGF as described under "Experimental Procedures." Surface bound 125 I-EGF was removed by a harsh acid strip, and cells were solubilized with 1 N NaOH to determine cell-associated radioactivity. The sum of intact 125 I-EGF in the media and surface-bound 125 I-EGF released by a harsh acid strip was expressed as the percentage of total radioactivity in the media and the cells at each time point. The values of are the means Ϯ S.D. (n ϭ 6). Some standard error bars are obscured by symbols. WT, wild type.