A single internalization signal from the di-leucine family is critical for constitutive endocytosis of the type II TGF-β receptor

Clathrin-coated-pit-mediated endocytosis is a major mechanism by which cells regulate the level of cell-surface receptors (reviewed by Mukherjee et al., 1997; Sorkin and Waters, 1993). For many receptors, signal termination involves their internalization or sequestration, which may also lead to a more prolonged step of downregulation. The endocytosis of most membrane proteins is generally mediated by short, specific signal sequences in their cytoplasmic domains (reviewed by Mukherjee et al., 1997; Sorkin and Waters, 1993). Receptor endocytosis can be either constitutive or ligand-activated; in the latter case, ligand-mediated post-translational modifications such as phosphorylation or ubiquitination may be involved (Bonifacino and Weissman, 1998; Sandoval and Bakke, 1994).

Structurally, internalization signals fall into three broad groups: tyrosine-based signals, di-leucine-based signals, and a third group of varied signals with no clear relation. Tyrosine-based signals typically contain the sequences YXXZ or NPXY, where X is any amino acid and Z is a hydrophobic amino acid (Bonifacino and Dell’Angelica, 1999; Trowbridge, 1991). Di-leucine signals include a pair of small hydrophobic amino acids, at least one of which is a leucine, with some additional context requirements (Kirchhausen, 1999; Sandoval and Bakke, 1994). The variable third group is less defined and includes motifs of clustered acidic residues (Tugizov et al., 1999; Voorhees et al., 1995), the ubiquitination-dependent endocytosis signal of the growth hormone receptor (Govers et al., 1999), and a series of variable cytoplasmic sequences shown to be essential for internalization of their respective receptors (Morelon and Dautry-Varsat, 1998; Setiadi et al., 1995). The different groups of signals may also differ in their interactions with clathrin-coated pits. Thus, YXXZ signals were reported to bind to the μ2 subunit of AP2 (Boll et al., 1996; Bonifacino and Dell’Angelica, 1999), the clathrin associated assembly protein complex specific for the plasma membrane (Kirchhausen et al., 1997; Marks et al., 1997; Schmid, 1997). By contrast, NPXY signals were suggested to bind directly to clathrin (Kibbey et al., 1998). The binding of peptides containing di-leucine to the μ chains of AP2 and AP1 yielded controversial results (Bremnes et al., 1998; Ohno et al., 1995; Rodionov and Bakke, 1998); these signals were also proposed to bind to the β subunit of AP1 (Rapoport et al., 1998) and to the AP3 adaptor complex proposed to be involved in lysosomal sorting (Honing et al., 1998). In addition, proteins such as Eps15 and arrestin were recently shown to couple specific receptors to AP2 (Benmerah et al., 1996; Goodman et al., 1996; Zhang et al., 1996).

The transforming growth factor β (TGF-β) superfamily of ligands mediate a broad range of biological responses. Depending on the cell type, TGF-β can induce growth inhibition or stimulation, cell differentiation and production of extracellular matrix (Heldin et al., 1997; Kingsley, 1994; Massague, 1998; Massague and Weis Garcia, 1996). TGF-β signals via two related receptors termed types I and II (TβRI and TβRII). These receptors are single spanning transmembrane serine-threonine kinases and are sufficient to induce TGF-β responses (Franzen et al., 1993; Lin et al., 1992). TβRI and TβRII are homodimeric prior to ligand binding (Gilboa et al., 1998; Henis et al., 1994). TGF-β binding to TβRII increases its association with TβRI into heterotetramers, leading to phosphorylation of TβRI by TβRII (Wells et al., 1999; Wrana et al., 1994). The activated TβRI phosphorylates Smad2 or Smad3, which are translocated to the nucleus together with Smad4, and alters the transcription of a large repertoire of genes (Heldin et al., 1997; Massague, 1998; Piek et al., 1999). This mode of signal transduction is unique in its directness (Nakao et al., 1997).

Several phenomena suggest that a tight control of the TβRII cell surface level is a key factor in determining the cellular response to TGF-β. Thus, loss of TβRII expression can lead to uncontrolled cell proliferation and correlates with pathological phenomena such as colorectal and lung carcinoma (de Jonge et al., 1997; Lu et al., 1996). By contrast, TβRII overexpression may enhance its functional association with TβRI, resulting in ligand-independent activation (Feng and Derynck, 1996). It is therefore important to characterize the factors governing TβRII endocytosis. However, the endocytic fate of the native TGF-β receptors has not been explored, mainly because the large number of proteins that bind TGF-β precludes the use of iodinated ligand to follow the internalization of specific receptors. The endocytosis studies conducted thus far employed exclusively chimeric receptors comprised of the extracellular domain of the GM-CSF receptors (α or β) fused to the transmembrane and cytoplasmic domains of the TGF-β receptors, using the GM-CSF ligand to follow the internalization (Anders et al., 1997; Anders et al., 1998; Dore et al., 1998). Although this chimeric system yielded significant insights into the role of the interactions between the cytoplasmic domains of the receptors in their internalization, these conclusions may not be valid for the endocytosis of the endogenous TGF-β receptors, because the foreign extracellular domain and the foreign bivalent ligand (GM-CSF) can alter the conformation and interactions of the receptors. Indeed, the chimeric receptors failed to interact with native TGF-β receptors (Anders et al., 1997). Therefore, the present study aimed to characterize the endocytosis of the full length, intact TβRII and to identify the internalization signal(s) governing this process. To this end, we employed TβRII epitope-tagged at the extracellular terminus, allowing it to be labeled with antibodies on live cells to follow endocytosis of the receptors without involvement of the ligand. A series of similarly tagged truncation, deletion or alanine-substitution mutants was made to explore the role of specific domains and short peptide sequences in TβRII internalization. Our results demonstrate that TβRII undergoes constitutive endocytosis in the absence of ligand via clathrin-coated pits. This process is dependent on a short sequence (residues I₂₁₈I₂₁₉L₂₂₀), which conforms to the di-leucine family of internalization signals.

9E10 (α-myc) mouse ascites directed against the myc epitope tag (Evan et al., 1985) was purchased from Harvard Monoclonals. IgG fractions and Fab′ fragments were prepared as described (Henis et al., 1994). Fluorophore-labeled antibodies were obtained from Jackson ImmunoResearch Laboratories. Primers used in mutagenesis were from GIBCO BRL. Mutagenesis was performed either with recombinant Pfu DNA polymerase or with the Quickchange mutagenesis kit (both from Stratagene). Restriction enzymes and T4 DNA ligase were from New England Biolabs.

COS-7 cells (CRL 1651, American Type Culture Collection) were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS), 100 U/ml penicillin, 100 μg/ml streptomycin and 4 mM glutamine. Media and tissue culture reagents were from Biological Industries, Beit Haemek, Israel.

COS-7 cells were transiently transfected by the DEAE dextran method (Seed and Aruffo, 1987) using pcDNA1 or pcDNA3 (Invitrogen) containing myc-TβRII (TβRII containing a myc tag at the extracellular terminus; Henis et al., 1994) or the mutant receptors generated in the present study. Assays were performed 48 hours after transfection.

The truncation, deletion and alanine substitution mutants of myc-TβRII employed in the current study are depicted in Fig. 1. For generating truncation mutants, myc-TβRII was subcloned from pcDNA1 (Henis et al., 1994) into pcDNA3 by digestion with EcoRI and XhoI, and served as a template for PCR site-directed mutagenesis. The procedure employed a forward primer recognizing the T7 promoter region of pcDNA3, and the following primers (containing a stop codon and a NotI restriction site) as the reverse primers (the consensus stop codon is depicted in bold letters, and the NotI restriction site is underlined): (1) S199 mutant, 5′-TTTTCC-TTTTGCGGCCGCCTATGAACTCAGCTTCTGCTGCCG-3′; (2) E211 mutant, 5′-TTTTCCTTTTGCGGCCGCCTACTCCATCAGC-TTCCGCGTCTTGC-3′; (3) A217 mutant, 5′-TTTTCCTTTT-GCGGCCGCCTAGGCACAGTGCTCGCTGAACTC-3′; (4) L220 mutant, 5′-TTTTCCTTTTGCGGCCGCCTAVAGGATGATGGCA-CAGTGCTC-3′; (5) P243 mutant, 5′-TTTTCCTTTT-GCGGCCGCCTAGGGCAGCAGCTCTGTGTTGTG-3′. The PCR products were digested with EcoRI and NotI and inserted into pcDNA3 digested with the same enzymes.

To generate the Δ(200-242) deletion mutant, a three-step PCR procedure was employed. The first reaction used the T7 promoter primer and a reverse primer recognizing the sequences flanking the deleted region (5′-GTCCAGCTCAATGGGTGAACTCAGCT TCTGCTGCCG-3′). The second reaction employed a forward primer recognizing the sequences flanking the deleted region (5′-AGAAGCTGAGTTCACCCATTGAGCTGGACACCCTG-3′) and a reverse primer recognizing the region encoding the L₃₈₅-N₃₇₇ segment of TβRII (5′-AGGATGTTGGAGCTCTTGAGGTC-3′). These PCR reactions generated products with the desired deletion at the 3′ and 5′ ends, respectively. The third reaction used the above two PCR products as a template. The T7 promoter primer served as the forward primer, and a primer recognizing the cDNA sequence encoding L₃₈₅-N₃₇₇ was employed as the reverse primer (5′-AGGATGTT-GGAGCTCTTGAGGTC-3′). The final PCR product was digested with EcoRI and BglII and inserted into myc-TβRII in pcDNA1 digested similarly, replacing the relevant part of the receptor with the mutated sequence.

The ΔIIL mutant was generated with the Quickchange mutagenesis kit (Stratagene), using myc-TβRII in pcDNA1 as the template with two mutagenic primers (5′-CGGTCATCTTCGGCACAGTGC-TCGCTGAAC-3′ and 5′-GAGCACTGTGCCGAAGATGACC-GCTCTGAC-3′) The F to A (F/A) mutation was inserted both into wild-type (wt) myc-TβRII and into ΔIIL (to generate the ΔIIL-F/A mutant) using the Quickchange mutagenesis kit (Stratagene). Myc-TβRII or the ΔIIL mutant in pcDNA1 served as templates; the mutagenic primers used were 5′-GTGCTCGCTGGCCTCC-ATGAGCTTCCGCG-3′ and 5′-CTCATGGAGGCCAGCGAGCA-CTGTGCC-3′. The 3A alanine substitution mutant was generated with the same mutagenesis kit, using myc-TβRII in pcDNA1 as template; the mutagenic primers were 5′-GAGCACTGTGCCGCT-GCCGCAGAAGATGACCGCTCTGAC-3′ and 5′-CGGTCATCTT-CTGCGGCAGCGGCACAGTGCTCGCTGAAC-3′. The sequences of all the mutants were verified by DNA sequencing.

Forty-eight hours after transfection, COS-7 cells grown on glass coverslips were incubated (30 minutes, 37°C) in serum-free DMEM to allow digestion of serum-derived ligands. After washing twice with cold Hank’s balanced salt solution (HBSS; Sigma) supplemented with 20 mM Hepes (pH 7.2) and 2% BSA Fraction V (Boehringer Mannheim) (HBSS/Hepes/BSA), the cells were incubated in the same buffer supplemented with normal goat IgG (200 μg/ml, 30 minutes, 4°C) to block nonspecific binding. This was followed by successive incubations (1 hour, 4°C, with three washes between incubations) with α-myc IgG or Fab′ followed by FITC-conjugated goat-anti-mouse (FITC-GαM) Fab′ in the same buffer (the specific concentrations are specified in the figure legends). Keeping the cells at 4°C through all labeling steps ensures that antibodies cannot penetrate inside and only cell-surface proteins are labeled. Subsequently, labeled cells were either kept at 4°C (control) or incubated at 37°C for different periods. The cells were fixed by successive incubations in methanol (5 minutes, −20°C) and acetone (2 minutes, −20°C), and mounted with mowiol (Hoescht) containing 29 mM n-propyl gallate (Sigma). Fluorescence digital images were acquired with a Leica DMR microscope (×100 or ×63 oil immersion objective) coupled to a CCD camera (SenSys, Photometrics). Images were exported to and processed with Photoshop (Adobe).

Internalization of myc-TβRII or its mutants was quantified by a point-confocal measurement of the fluorescence intensity of myc-TβRII (labeled by fluorescent antibodies) remaining at the cell surface after incubation at 37°C for various periods of time.

This method is advantageous because its microscopic nature enables the selection of intact, live cells for the measurement. The intensity measurement was performed using a fluorescence photobleaching recovery setup described previously (Henis and Gutman, 1983) under non-bleaching illumination conditions. This setup uses a laser beam (488 nm) focused on the cell surface to a small spot (×63 oil immersion objective, 0.85 μm Gaussian radius). The fluorescence excited in this well-defined focus passes through an additional pinhole located at the image plane of the microscope, making it a true one-spot confocal setup and enabling the measurement of fluorescence exclusively from the plane of the plasma membrane. The fluorescence intensity is measured quantitatively by a cooled photomultiplier tube in a photon-counting mode. For these measurements, cells were transfected with myc-TβRII constructs and the cell surface receptors were labeled at 4°C with α-myc followed by FITC-GαM Fab′, as described above for immunofluorescence microscopy. The cells were fixed (either immediately, or following incubation at 37°C for various periods to enable endocytosis), mounted, and taken for the point-confocal measurements of surface receptor density (150-200 cells measured and averaged for each sample). Labeling the surface receptors in the cold yielded smooth uniform staining, whereas further incubation at 37°C shifted part of the labeling into a vesicular pattern representing endocytic vesicles, as evident from their deeper focal plane in the point-confocal measurement and from the inability of acid stripping to remove the vesicular staining (see Results; Fig. 2; Fig. 3).

The fluorescence intensity in the uniformly labeled regions, which is proportional to the level of labeled receptors remaining at the cell surface, can be measured by focusing the laser beam at the cell surface, away from the vesicles. Because dissociation of antibodies from the cell surface during the 37°C incubation may also contribute to the reduction in the cell surface fluorescence with time, the results were normalized to the decrease in intensity following incubation for similar periods at 37°C under conditions that blocked the endocytosis of the receptors (cells treated with hypertonic medium; see below).

This reduction was below 15% in all cases. At each time point, the percentage of cell surface receptors that underwent internalization is given by the percent reduction in the surface associated fluorescence intensity. For wt TβRII, the results were compared with those obtained using internalization measurement by fluorescence-activated cell sorter (FACS). Transfected cells were detached from the dish using PBS containing 5 mM EDTA (15 minutes, 37°C), washed and incubated with α-myc antibodies in HBSS/Hepes/BSA as described above. Cells were then incubated at 37°C for different periods to allow endocytosis, cooled on ice and the antibody-labeled receptor complexes remaining at the cell surface were labeled in the cold with FITC-GαM, as described above (see Immunofluorescence microscopy). The mean fluorescence intensity of 10⁴ cells/sample was measured by FACS (FACSorter flow cytometer, Becton Dickinson) and the background of similarly labeled untransfected cells was subtracted. Internalization was quantified by the reduction in surface-associated fluorescence, as described earlier (Morelon and Dautry-Varsat, 1998).

The treatments employed were incubation in hypertonic medium or acidification of the cytosol (Hansen et al., 1993). Hypertonic treatment, which blocks coated-pit-mediated endocytosis by dispersing the underlying clathrin lattices (Heuser and Anderson, 1989), was performed by a 30 minute pre-incubation of the transfected cells (48 hours post-transfection) with 0.45 M sucrose in serum-free DMEM. The treated cells were incubated with antibodies at 37°C for internalization studies as described in the previous two sections, except that 0.45 M sucrose was added at all steps. Cytosol acidification, which blocks coated-pit-mediated endocytosis by preventing the pinching-off of coated vesicles (Sandvig et al., 1987), was performed as described by us earlier (Fire et al., 1995). Transfected cells (48 hours after transfection) were incubated in Hepes-buffered DMEM (pH 7.2) containing 30 mM NH₄Cl (30 minutes, 37°C), followed by 5 minutes (37°C) in potassium-amiloride (KA) buffer (0.14 M KCl, 2 mM CaCl₂, 1 mM MgCl₂, 1 mM amiloride HCl, 20 mM Hepes pH 7.2). This buffer was then replaced by cold KA containing 2% BSA, which was used for all subsequent incubations and washes.

The extracellularly tagged myc-TβRII and mutants derived from it enabled us to follow the constitutive endocytosis of the full-length TβRII, using anti-tag antibodies. The experiments were performed in COS-7 cells, based on the demonstration that transient expression of internalization-competent membrane proteins in CV1-derived cells is an appropriate system to analyze the nature and strength of the interactions between internalization signals and the endocytic machinery (Fire et al., 1995; Lazarovits and Roth, 1988). Figure 2 depicts an experiment where cells transfected with myc-TβRII were labeled in the cold (to allow only extracellular surface labeling) by fluorescent monovalent Fab′ fragments, and incubated at 37°C for different periods of time to allow internalization. The initial distribution of the fluorescence (prior to incubation at 37°C) was homogeneous, typical of cell surface labeling (Fig. 2A). Incubation at 37°C led to the progressive appearance of a vesicular pattern typical of endocytic vesicles (Fig. 2B-D); indeed, acid stripping of the cells removes the membrane-associated uniform labeling but not the vesicular staining, demonstrating that the latter represents endocytic vesicles (Fig. 2E). To determine whether the internalization of TβRII is mediated via clathrin-coated pits, we employed two independent treatments known to disrupt coated-pit-mediated internalization. As shown in Fig. 2F, hypertonic treatment to disrupt the clathrin lattice structure (Hansen et al., 1993; Heuser and Anderson, 1989) blocked myc-TβRII internalization; similar results (data not shown) were obtained following cytosol acidification, which blocks the pinching off of clathrin-coated vesicles (Hansen et al., 1993; Sandvig et al., 1987).

To quantify TβRII internalization, we employed a point-confocal method (see Materials and Methods) that measures the level of antibody-labeled receptors remaining at the cell surface after various periods of incubation at 37°C. The percentage of the reduction in the level of the tagged receptors in the plasma membrane with time (corrected for the minor contribution of antibody dissociation) yields a direct measure of the internalization rate. We have chosen this method because it proved to be more sensitive and accurate than the measurement of the internalization by FACS (see below), mainly owing to its microscopic nature (enabling the selection of undamaged intact cells for the measurement and avoiding cells with extreme atypical high labeling levels) and owing to the collection of light only from the well-defined narrow focal plane of the plasma membrane. The results obtained for the internalization of wt myc-TβRII are depicted in Fig. 3A, along with a picture of a representative cell showing schematically typical locations of measurement (Fig. 3B). These results demonstrate that myc-TβRII undergoes constitutive endocytosis with a halftime (t_1/2) of ∼15 minutes. The endocytosis rate, defined as the fraction of the surface receptor population internalized per minute, was 0.022 minutes⁻¹, derived from the initial linear part (up to 20 minutes) of the internalization curve as described earlier (Zwart et al., 1996). Analogous results (t_1/2 ∼ 15 minutes) were obtained using a FACS-based internalization assay (see Materials and Methods). These values are in accordance with those obtained in studies using GM-CSF/TGF-β receptor chimeras (Dore et al., 1998).

Sorting signals governing the intracellular trafficking of membrane proteins are normally localized to their cytoplasmic domains, where they interact with cellular components involved in these processes. Thus, if TβRII endocytosis is mediated by internalization signals of its own, truncation of its cytoplasmic tail is expected to block it. We therefore prepared a myc-tagged mutant (S199; Fig. 1) similar to a TβRII truncation mutant described earlier (Wieser et al., 1993), which contains only the first 10 amino acids of the cytoplasmic tail and was shown to undergo normal processing and transport to the cell surface. Immunofluorescence internalization experiments similar to those performed with myc-TβRII (Fig. 4) show that S199 fails to shift into a vesicular endocytic pattern even after prolonged incubation at 37°C (Fig. 4C,D). Quantitative measurement of its internalization (as in Fig. 3) showed no detectable internalization after 30 minutes at 37°C. In many receptors the internalization signals are located in the region proximal to the transmembrane domain. This was postulated to be true also for TβRI (Yao et al., 2000), which exhibits a significant degree of structural homology to TβRII.

We therefore proceeded to examine if this region is essential for the internalization of TβRII. To this end, we prepared the myc-tagged truncation mutant P243 (see Fig. 1), truncated right at the start of the kinase homology domain immediately after residue P₂₄₃, and the Δ(200-242) mutant (where the sequence between S₁₉₉ and P₂₄₃ was deleted; Fig. 1). As shown in Fig. 4, incubation at 37°C shifted P243 to a vesicular staining pattern characteristic of endocytic vesicles, whereas Δ(200-242) remained evenly distributed at the cell surface. Internalization measurement by the point-confocal method (as described in Fig. 3) showed that P243 endocytosis was indistinguishable from that of myc-TβRII, whereas Δ(200-242) endocytosis was undetectable even after 30 minutes. As in the case of myc-TβRII, hypertonic treatment or cytosol acidification (not shown) completely blocked P243 endocytosis. The normal internalization of P243 versus the complete absence of endocytosis of S199 and Δ(200-242) suggests that the internalization signal(s) reside in the region between these two residues. Because the entire kinase domain and the remainder of the cytoplasmic tail are still present in the Δ(200-242) mutant, these findings suggest that the signal(s) that target TβRII for coated-pit-mediated endocytosis are enclosed within the sequence proximal to the transmembrane domain and preceding the kinase domain. Furthermore, they indicate that the kinase domain and the following cytoplasmic tail are not necessary for the constitutive endocytosis of TβRII.

In view of the above results we focused the search for the TβRII internalization signal(s) on the cytoplasmic region of the receptor enclosed between residues S₁₉₉ and P₂₄₃. Sequence analysis of the region pointed to three potential endocytosis signals: (1) a potential di-leucine motif, L₂₄₁L₂₄₂; (2) a cluster of acidic residues (EDDRSD) between E₂₂₁ and D₂₂₆; and (3) an isoleucine-leucine-based motif, I₂₁₈I₂₁₉L₂₂₀. In addition to these ‘classical’ endocytosis motifs, residues with bulky aromatic side-chains (e.g. F₂₁₂) might serve as internalization signals when found in the right structural conformation. To identify which of the potential signals is involved, we prepared a series of myc-tagged truncation mutants shorter than P243, gradually removing the candidate signals. The truncation mutant L220 (Fig. 1), which lacks both L₂₄₁L₂₄₂ and the E₂₂₁-D₂₂₆ acidic cluster, was internalized as efficiently and with the same rate as myc-TβRII (Fig. 5B; for quantification, see Fig. 5G). By contrast, a further truncation excising I₂₁₈I₂₁₉L₂₂₀ (the A217 mutant; Fig. 1) drastically reduced the internalization (Fig. 5D,G); the endocytosis rate (calculated from the quantitative measurement shown in Fig. 5G) was significantly lower than for the wt receptor (endocytosis rate of 0.005/minute for the A₂₁₇ mutant, compared with 0.022/minute for the wt receptor). Deletion of six more residues just before F₂₁₂ (E211 mutant; Fig. 1) essentially eliminated endocytosis (Fig. 5F,G). These results suggest that the sequence I₂₁₈I₂₁₉L₂₂₀ is the major internalization motif responsible for the constitutive endocytosis of TβRII. The residual low level of internalization observed for the A217 mutant, which is abolished in the E211 mutant, indicates that the short sequence immediately preceding I₂₁₈I₂₁₉L₂₂₀ also contributes to the regulation of TβRII internalization, perhaps as an extension of the IIL motif.

The above results were obtained using truncated TβRII mutants and, although the I₂₁₈I₂₁₉L₂₂₀ sequence functions as an internalization signal in the context of the truncated protein, it is possible that it is masked in the full length receptor. We therefore examined the effects of deleting or mutating these three amino acids on the endocytosis of the full-length TβRII. The results of internalization experiments employing the ΔIIL deletion mutant (myc-TβRII with a deletion of the I₂₁₈I₂₁₉L₂₂₀ triad; Fig. 1) and the 3A alanine substitution mutant (the IIL triad replaced by three alanine residues; Fig. 1) are depicted in Fig. 6. They clearly demonstrate that the internalization of both mutants is greatly impaired (Fig. 6D,F); quantitative analysis of their endocytosis relative to the wt receptor (Fig. 6I) indicates that they exhibit only residual internalization. To examine the possible contribution of F₂₁₂ (a bulky aromatic residue) to the internalization, we used two mutants where this residue was replaced by alanine. Mutation of F₂₁₂ in myc-TβRII to alanine (mutant designated as F212A) had no measurable effect on the internalization (Fig. 6I), suggesting that its role is at most marginal. Adding this mutation to the receptors lacking I₂₁₈I₂₁₉L₂₂₀ (the ΔIIL-F/A mutant; Fig. 1) reduced the already low internalization only marginally, within the experimental error (Fig. 6I).

The cell surface level of TβRII is tightly controlled. The importance of this regulation is emphasized by the role of TβRII as the primary ligand-binding receptor in the TGF-β signaling cascade, and by the constitutive nature of its Ser/Thr kinase activity (Brand and Schneider, 1996; Massague, 1998). For many membrane receptors, internalization processes play key roles in regulating their surface expression levels. Although many structural requirements of TβRII domains for activation and signaling were characterized (Alevizopoulos and Mermod, 1997; Brand and Schneider, 1996; Massague, 1998; Zhu and Sizeland, 1999), data on TGF-β receptor endocytosis were available only for chimeric constructs with the extracellular domain of GM-CSF receptors (Anders et al., 1997; Anders et al., 1998; Dore et al., 1998). Even in the case of homomeric complexes, the former studies followed exclusively the internalization of complexes that are mediated by the binding of the dimeric GM-CSF ligand. The experiments reported in the current study were designed to investigate the endocytosis of the native full-length receptors, and follow the internalization of the TβRII population present naturally at the cell surface (mainly in homodimers; Henis et al., 1994). Furthermore, the endocytosis motif(s) that target TGF-β receptors (or other receptors from the Ser/Thr kinase family) for internalization were previously unknown. In the current study, we investigated these issues by employing extracellularly myc-tagged TβRII and a series of mutants derived from it to characterize TβRII internalization.

The extracellular location of the myc tag renders the tagged cell-surface receptors accessible to labeling from the outside upon incubation of live cells with fluorescent antibodies. Employing a point confocal assay to measure the reduction in the plasma membrane-associated labeling following internalization, we were able to demonstrate for the first time that the full-length TβRII undergoes constitutive endocytosis, characterized by an initial endocytosis rate of 0.022/minute (Fig. 2; Fig. 3). This rate is in agreement with the rate reported for GM-CSF receptor/TβRII chimera, measured by GM-CSF internalization (0.03/minute; Dore et al., 1998). In accordance with the report on the chimeric receptors (Anders et al., 1997), TβRII internalization appears to be mediated via the clathrin-coated pit pathway, as demonstrated by its blockade following hypertonic or cytosol acidification treatments (Fig. 2), known to disrupt coated-pit-mediated endocytosis (Hansen et al., 1993; Heuser and Anderson, 1989; Sandvig et al., 1987).

To identify the region(s) in TβRII where the endocytosis signal(s) might reside, we generated a series of myc-TβRII mutants with successive truncations of the cytoplasmic tail, and studied their internalization. The results of these experiments (Fig. 4) implicated a relatively short sequence, between S₁₉₉ and P₂₄₃, in targeting the receptor for endocytosis; this was validated by the failure of the deletion mutant Δ(200-242) to undergo internalization (Fig. 4). Because the deleted segment contained several potential endocytosis motives, we employed a series of truncation, deletion and alanine substitution mutants to elucidate the role of the potential signals in the internalization (Fig. 5; Fig. 6). These studies have clearly identified a single motif, I₂₁₈I₂₁₉L₂₂₀, as the major endocytosis signal of TβRII. An amino acid several residues upstream (F₂₁₂) may have some contribution to the residual internalization of the truncated mutants, but this role is negligible when F₂₁₂ is replaced by alanine in the context of the full receptor (Fig. 6). A possible explanation for this minor difference is that it becomes fully or partially exposed upon truncation of the receptor.

The I₂₁₈I₂₁₉L₂₂₀ motif belongs do the di-leucine-based family of trafficking signals. These signals are seemingly of a loose nature, with a doublet of small hydrophobic residues as a common element. I₂₁₈I₂₁₉L₂₂₀ is a triad of hydrophobic residues and may represent an overlap of two adjacent signals or an essential element of a larger sequence requirement. In a recent review (Kirchhausen, 1999), the consensus sequence for di-leucine internalization signals has been described as (−)(2-4)XLL, where X is usually a polar residue and (−) is often a negatively charged residue or a phosphoserine. In accordance with the loose sequence requirements presented by this class of internalization signals, the leucine residues may be substituted by other small hydrophobic amino acids such as Ile, Met or Val (Kirchhausen, 1999; Sandoval and Bakke, 1994). Indeed, the −4 position relative to I₂₁₈I₂₁₉L₂₂₀ in TβRII is occupied by the negatively charged E₂₁₄. Furthermore, S₂₁₃, which immediately precedes this residue, was shown to be constitutively autophosphorylated (Luo and Lodish, 1997), enhancing the negative charge. This constitutive phosphorylation is in accordance with the constitutive nature of TβRII endocytosis (Fig. 2). By contrast, in ligand-activated signals such as those of CD3γ (Dietrich et al., 1998), furin (Teuchert et al., 1999), or the cation-dependent mannose 6-phosphate receptor (Schweizer et al., 1997), a post-translational modification in the vicinity of the signal may serve as an on/off switch for its function, possibly by mediating its exposure. Thus, although the I₂₁₈I₂₁₉L₂₂₀ motif is responsible for the internalization of singly-expressed TβRII, it is still possible that other potential motifs become exposed and active following ligand-mediated heterocomplex formation with TβRI. Interestingly, when we subjected the region S₁₉₉-P₂₄₃ to a protein structure analysis using PredictProtein program (Rost, 1996; Rost and Sander, 1993; Rost and Sander, 1994), the hydrophobic I₂₁₈I₂₁₉L₂₂₀ sequence was found to reside in an extended (sheet) conformation in the center of an exposed hydrophilic globular domain. This domain was formed with the participation of the highly hydrophilic sequence EDDRSD (positions 221-226). The analysis showed this particular conformation to be impaired either by deletion of the IIL residues or by their substitution with three alanines. In summary, the current study demonstrates for the first time that the full-length, intact TβRII undergoes constitutive endocytosis via clathrin-coated pits. Our findings demonstrate that a specific sequence, I₂₁₈I₂₁₉L₂₂₀, which belongs to the family of di-leucine internalization signals, is the key mediator of the internalization process. This is the first identification of an endocytosis signal for receptors from the TGF-β receptor family. The system of the epitope-tagged TβRII and the battery of mutants of these receptors that vary in their ability to undergo endocytosis can serve in future studies to investigate the effects of ligand-mediated interactions with TβRI on internalization.

This work was supported by grants from the Israel Science Foundation administered by the Israel Academy of Sciences and Humanities, and from the Israel Cancer Research Fund. It is in partial fulfillment of the requirements for a PhD thesis by Marcelo Ehrlich, to be submitted to the Senate of Tel Aviv University.