Differential Requirement for Type I and Type II Transforming Growth Factor β Receptor Kinase Activity in Ligand-mediated Receptor Endocytosis*

Transforming growth factor β (TGFβ) superfamily polypeptides regulate cell growth and differentiation by binding to single pass serine/threonine kinases referred to as TGFβ type I and type II receptors. Signal propagation is dependent upon heteromeric (type I-type II) complex formation and transphosphorylation of the type I receptor by the type II receptor. While many of the phosphorylation events necessary for receptor signaling have recently been characterized, the role of TGFβ receptor kinase activity in modulating receptor endocytosis has not been addressed. To that end, we have used chimeric receptors consisting of the extracellular domain of the granulocyte/macrophage colony-stimulating factor α and β receptors spliced to the TGFβ type I and type II transmembrane and cytoplasmic domains to address the specific role of type I and/or type II receptor kinase activity in TGFβ receptor internalization, down-regulation, and signaling. To inactivate chimeric receptor kinase activity, point mutations in the ATP binding site were made at amino acids 232 and 277 in the type I and type II receptor, respectively. Either of these mutations abolished plasminogen activator inhibitor 1 protein expression stimulated by granulocyte/macrophage colony-stimulating factor activation of chimeric heteromeric type I-type II TGFβ receptors. They did not, however, modulate TGFβ signaling stimulated through the endogenous TGFβ receptor. Although TGFβ receptor signaling was dependent upon the kinase activity of both chimeric receptors, the initial endocytic response was distinctly regulated by type I and/or type II receptor kinase activity. For instance, while heteromeric receptor complexes containing a kinase-inactive type I receptor were endocytosed similarly to wild type complexes, the kinase activity of the type II TGFβ receptor was necessary for optimal internalization and receptor down-regulation. Furthermore, these responses were shown to occur independently of type II receptor autophosphorylation but require a type II receptor capable of transphosphorylation.

Transforming growth factor ␤ (TGF␤) superfamily polypeptides regulate cell growth and differentiation by binding to single pass serine/threonine kinases referred to as TGF␤ type I and type II receptors. Signal propagation is dependent upon heteromeric (type I-type II) complex formation and transphosphorylation of the type I receptor by the type II receptor. While many of the phosphorylation events necessary for receptor signaling have recently been characterized, the role of TGF␤ receptor kinase activity in modulating receptor endocytosis has not been addressed. To that end, we have used chimeric receptors consisting of the extracellular domain of the granulocyte/macrophage colony-stimulating factor ␣ and ␤ receptors spliced to the TGF␤ type I and type II transmembrane and cytoplasmic domains to address the specific role of type I and/or type II receptor kinase activity in TGF␤ receptor internalization, downregulation, and signaling. To inactivate chimeric receptor kinase activity, point mutations in the ATP binding site were made at amino acids 232 and 277 in the type I and type II receptor, respectively. Either of these mutations abolished plasminogen activator inhibitor 1 protein expression stimulated by granulocyte/macrophage colony-stimulating factor activation of chimeric heteromeric type I-type II TGF␤ receptors. They did not, however, modulate TGF␤ signaling stimulated through the endogenous TGF␤ receptor. Although TGF␤ receptor signaling was dependent upon the kinase activity of both chimeric receptors, the initial endocytic response was distinctly regulated by type I and/or type II receptor kinase activity. For instance, while heteromeric receptor complexes containing a kinase-inactive type I receptor were endocytosed similarly to wild type complexes, the kinase activity of the type II TGF␤ receptor was necessary for optimal internalization and receptor down-regulation. Furthermore, these responses were shown to occur independently of type II receptor autophosphorylation but require a type II receptor capable of transphosphorylation.
The transforming growth factor ␤ (TGF␤) 1 superfamily of proteins regulate a number of diverse biologic processes (1)(2)(3). While the cellular response can be as distinct as growth stimulation or growth inhibition, it appears as though a similar receptor system is utilized for both pathways. Understanding how the receptors are regulated for one family of proteins will ultimately extend the knowledge for the entire superfamily. The model most commonly accepted for receptor activation requires oligomerization of a type I and type II TGF␤ receptor (4 -7). This occurs through ligand binding to a type II receptor and recruitment of a type I receptor into a dimeric and/or tetrameric complex (7)(8)(9)(10)(11). The serine/threonine kinase activity of the type I receptor is then activated by specific type II receptor phosphorylations in the juxtamembrane region of the type I receptor (12)(13)(14)(15)(16). This cascade of receptor interactions and phosphorylations ultimately results in the propagation of the TGF␤ signal to downstream effectors in which the Smad family of proteins has a fundamental role (17)(18)(19)(20)(21)(22)(23)(24).
Although a great deal of information has been generated documenting potential receptor interactions required for TGF␤ signaling, the endocytic fate of the receptor-ligand complex is essentially unexplored. Typically, once growth factor receptors bind ligand, they are endocytosed through structures referred to as clathrin-coated pits (25)(26)(27). The endocytic process usually requires the intrinsic enzymatic (kinase) activity of the receptor and is mediated via defined elements routinely found in the cytoplasmic domain of the receptor (28 -34). Although no canonical sequences have been identified, a structure representing a tight turn conformation has been proposed as a common determinant for an internalization signal (34). While much of our current understanding surrounding growth factor receptor endocytosis derives from studies performed on the epidermal growth factor and insulin receptor tyrosine kinases, relatively little has been done investigating these processes in the TGF␤ receptor superfamily. Since the signaling mechanism, intrinsic receptor kinase activity, and biology of the two receptor systems differ, it is unknown whether the paradigms developed for the receptor tyrosine kinases will be operative in the TGF␤ serine/threonine receptor family.
TGF␤ receptors have been previously reported to undergo down-regulation after ligand binding in some cell types but not in others (6,(35)(36)(37). While this might simply represent cell type differences, it is now possible to evaluate these earlier studies in the context that both heteromeric and homomeric TGF␤ receptor interactions have been documented on the cell surface (38,39). For instance, ligand binding to homomeric type II receptor oligomers might result in a distinct endocytic response from that observed following activation of signaling competent type I-type II receptor heteromers. In that regard, our recent studies have shown that while type I-type I, type II-type II, or  type I-type II TGF␤ receptor oligomerization in mesenchymal  AKR-2B cells results in the internalization of bound ligand,  only signaling-competent type I-type II TGF␤ receptor heteromers are down-regulated (40). Although that study demonstrated distinct endocytic responses of heteromeric and homomeric TGF␤ receptors, it did not address the potential regulatory role for receptor serine/threonine kinase activity, nor did it determine whether the endocytic and signaling responses were independently regulated.
In the present paper, we have employed chimeric receptors consisting of the granulocyte/macrophage colony-stimulating factor (GM-CSF) ␣ and ␤ receptor ligand binding domain fused to the transmembrane and cytoplasmic domain of kinase-inactive type I and type II TGF␤ receptors to examine the role of TGF␤ receptor kinase activity in receptor trafficking and signaling. Consistent with previous reports in epithelial cells (7,8), we find an obligate requirement for both type I and type II TGF␤ receptor kinase activity in mediating heteromeric receptor signaling in mesenchymal cells. However, in contrast to that observed for receptor signaling, a differential requirement for receptor kinase activity in modulating the endocytic response of the receptor complex was observed. For instance, while internalization and down-regulation occurred independently of the type I receptor kinase, type II receptor-transphosphorylating activity was needed for optimal endocytosis. Thus, in addition to activating downstream effector molecules, type I receptor phosphorylation (by the type II TGF␤ receptor) is similarly required to promote internalization and down-regulation of the TGF␤ receptor complex. Construction of Mutant Receptors-The ␤ I K232R mutation was generated using the Transformer TM site-directed mutagenesis kit version 2 (CLONTECH, Palo Alto CA). The mutagenic primer was 5Ј-GAAGAAGTTGCTGTTAgGATATTCTCCTCTAGA, and the selection primer was 5Ј-TGACTGGTGAGgcCTCAACCAAGT, where the mutagenic bases are in lowercase type. The mutation and the rest of the receptor sequence was verified by automated DNA sequencing. The beta I receptor was religated into the pHa expression plasmid through the SalI site (4). To generate the ␤ II K277R mutation, an HpaI/AccI cassette from pCMV-T␤RII HA K to R plasmid was first placed in pCR TM II (InVitrogen) and then ligated back into pHa following KpnI and BamHI digestion.
The ␣ II P525L mutation was generated using the QuickChange TM (Stratagene, La Jolla CA) mutagenesis kit. Mutagenic primers were 5Ј-GGGACCACGACCtAGAGGCCCGTC and 5Ј-GACGGGCCTCTaG-GTCGTGGTCCC. Mutant constructs in pCR TM II were verified by automated DNA sequencing and ligated into pHa following XbaI and BamHI digestion.
Isolation of Clones-Parental AKR-2B cells were plated at 1 ϫ 10 5 cells/well in a six-well dish (22 cm 2 ) 24 h prior to transfection. Cells were rinsed with serum-free DMEM and then incubated for 6 h in 2 ml of DMEM with transfection solution consisting of 2-4 g of expression plasmid DNA and 2 l/g TransIT TM LT2 (PanVera Corp., Madison, WI) in a final volume of 100 l with Opti-MEM (Life Technologies). Cells recovered for 16 h in 5% FBS/DMEM and then were placed in selective medium (5% FBS/DMEM with 400 g/ml Geneticin and 135 g/ml hygromycin B) for 24 h before splitting 1:40 by surface area. 2-3 weeks later, well separated colonies were isolated and expanded.
Fluorescence-activated Cell Sorting (FACS)-Cells were detached in DMEM containing 40 mM EDTA, 20 mM HEPES, pH 7.2. After washing with 5% FBS/DMEM and antibody buffer (PBS supplemented with 2% FBS, 0.02% NaN 3 , pH 7.4), approximately 5 ϫ 10 5 cells were incubated with primary monoclonal antibody at 5 g/ml (anti-human GM-CSF ␣ receptor, Santa Cruz Biotechnology catalog no. SC 458; anti-human GM-CSF ␤ receptor, Santa Cruz catalog no. 457; or control mouse ascities, Sigma catalog no. M-8273) for 1 h at 4°C with rocking. Cells were washed twice with antibody buffer and incubated with a 1:50 dilution of secondary fluorescein isothiocyanate-conjugated antibody (Sigma catalog no. F-2012). Following a 45-min incubation at 4°C, the cells were washed; fixed in 500 l of PBS containing 1% paraformaldehyde, pH 7.4; and filtered through a 40-m nylon filter prior to flow analysis using a Beckton Dickinson FACS Vantage with PCLYSIS version 1.1 software.
Plasminogen Activator Inhibitor-1 Production-Cells were plated in six-well tissue culture dishes at 2 ϫ 10 5 cells/well 24 h before treatment. The serum containing medium was removed, and the cultures were placed in 1.0 ml of serum-free DMEM lacking methionine but supplemented with the indicated growth factors. Following a 2-h treatment at 37°C, wells were pulsed for 2 h with 50 Ci/ml [ 35 S]Met/Cys Promix (Amersham Pharmacia Biotech) and processed by washing once with PBS; three times with 10 mM Tris, 0.5% deoxycholate, 50 g/ml phenylmethanesulfonyl fluoride (Sigma), pH 8.0; twice with 2 mM Tris, pH 8.0; and once with PBS (41). Matrix proteins were eluted from the wells by the addition of 100 l the 2ϫ Laemmli buffer containing 10% ␤-mercaptoethanol, separated by 8% SDS-polyacrylamide gel electrophoresis, and processed for fluorography.
Transient Transfections-Cells were plated in six-well dishes (9.6 cm 2 ) at 1.5 ϫ 10 5 per well 24 h prior to transfection. Three g of 3TP-Lux, 0.5 g of pCMV-␤-galactosidase, and 7 l of TransIT LT2 (Mirus Corp, Madison WI) were combined with Opti-MEM (Life Technologies) to a final volume of 100 l, and transfection was performed as described previously (4). Cultures were then stimulated in 5% FBS/ DMEM for 24 h in the presence or absence of TGF␤ or GM-CSF, and luciferase activity was determined following normalization for transfection efficiency with ␤-galactosidase.
Internalization Assays-Cells were plated in six-well dishes (9.6 cm 2 / well) at 1.5 ϫ 10 5 cells/well in 5% FBS/DMEM. Following 24 h at 37°C, internalization assays were initiated by incubation at 4°C for 2-4 h in binding buffer (0.3 ml DMEM containing 200 mM HEPES, pH 7.4, 25 mg/ml bovine serum albumin) supplemented with 100 pM 125 I-GM-CSF (119 Ci/g; NEN Life Science Products). Once equilibrium had been reached, the plates were washed two times with binding buffer containing 75% horse serum and then placed at 37°C in 5% FBS/DMEM for various times to promote receptor endocytosis. To calculate the percentage of internalization (i.e. specific cpm in cell/specific surface cpm), the cultures were returned to 4°C, and the remaining surface-bound ligand was removed by acid washing (PBS, pH 3.0) and the internalized ligand was determined by cell lysis in 0.2 M NaOH, 40 g/ml salmon sperm DNA. All time points contained parallel plates with 25-fold excess cold GM-CSF to document specificity of binding.
Potassium depletion of cells was performed essentially as described by Larkin et al. (42) and Sorkin et al. (43). Following an initial 5-min hypotonic shock with DMEM/H 2 O (1:1) at 37°C, cultures were incubated for 10 min in buffer A (50 mM HEPES, 100 mM NaCl, pH 7.4) and 30 min in buffer B (50 mM HEPES, 100 mM NaCl, 1 mM CaCl 2 , 2.5% bovine serum albumin, pH 7.4) at 37°C. Cells were then placed at 4°C (in buffer B) for 15 min and replaced with buffer B containing 100 pM 125 I-GM-CSF. Control cells were treated similarly but did not undergo initial hypotonic shock, and subsequent incubations in buffers A and B contained 10 mM KCl.
Down-regulation Assays-Similar conditions as described above for internalization assays were used to determine receptor down-regulation. The primary difference being that the cultures were first incubated at 37°C with 5% FBS/DMEM containing unlabeled GM-CSF (520 pM or 10 ng/ml) for the indicated times. Surface bound ligand was removed by acid washing (PBS, pH 3.0), and the remaining cell surface receptor binding determined by incubation at 4°C for 2-4 h with 100 pM 125 I-GM-CSF. The plates were then washed two times with cold 75% horse serum, 25% binding buffer, and specifically bound 125 I-GM-CSF was determined. Control studies had shown that acid washing removed 90 -95% of receptor-bound ligand without affecting subsequent binding (data not shown).

RESULTS
Chimeric Receptor Expression-Once the chimeric receptor cDNAs were mutated and their sequences were confirmed, they were stably transfected into mouse fibroblast AKR-2B cells in ␣ and ␤ pairs to generate high affinity ligand-dependent heteromeric complex formation. The designation ␣I␤II, for example, represents clones expressing chimeric receptors consisting of the ligand binding domain of the GM-CSF ␣ receptor fused to the transmembrane and cytoplasmic domain of the type I TGF␤ receptor and the ligand binding domain of the GM-CSF ␤ receptor fused to the transmembrane and cytoplasmic domain of the type II TGF␤ receptor (4). The 600 series of clones contain a wild type chimeric type I TGF␤ receptor co-expressed with a kinase-inactive chimeric type II TGF␤ receptor, and the 700 series contain a wild type chimeric type II TGF␤ receptor co-expressed with a kinase-inactive chimeric type I TGF␤ receptor. Individual clones were isolated by ring subcloning and initially screened for membrane expression of the chimeric receptors by FACS (Fig. 1). Our previous work has shown that the parental AKR-2B cell line does not show specific staining for these receptors (4). Once clones were shown to express the chimeric receptors, their ability to specifically bind radiolabeled GM-CSF was determined (data not shown). The parental AKR-2B cells showed no significant binding, which increased to Ͼ85% specific binding when wild type chimeric ␣I␤II receptors were expressed (4). For each of the mutated chimeric receptors, similar amounts of specific GM-CSF binding ranging from 70 to 90% of total binding was seen.

Lack of GM-CSF-induced Signaling in Clones Expressing
Kinase-dead Chimeric Receptors-Once expression and ligand binding of the chimeric receptors was confirmed, we wished to determine whether the chimeric kinase-dead receptors responded in a similar fashion to that reported for kinase-inactive endogenous TGF␤ receptors (2,44). This was addressed by examining the ability of the chimeric and endogenous TGF␤ receptors to stimulate expression of the extracellular matrixassociated protein plasminogen activator inhibitor-1 (PAI-1). As shown in Fig. 2, clone A105 (expresses wild type chimeric receptors) stimulates PAI-1 production when treated with either GM-CSF or TGF␤ (activates chimeric or endogenous TGF␤ receptors, respectively). However, cells expressing type I or type II kinase-dead chimeric TGF␤ receptors do not induce PAI-1 protein when treated with GM-CSF at either 10 or 100 ng/ml. This does not represent a general signaling defect in the TGF␤ pathway(s), since the addition of TGF␤ to activate the endogenous TGF␤ receptors results in PAI-1 expression similar to that observed in the parental cell line (Fig. 2). In that regard, identical ligand-dependent results are seen when the cultures are examined for their ability to form colonies in soft agar (data not shown).
Transfection of kinase-dead TGF␤ receptors has been previously shown to function as dominant/negative mediators of TGF␤ action (45)(46)(47)(48)(49)(50). This inhibitory activity is presumably a reflection of the requirement for TGF␤ receptor oligomerization in signal propagation (4 -7). Moreover, the cytoplasmic domains of the type I and type II TGF␤ receptors have been shown to form both receptor heteromers and homomers in overexpressing COS cells and yeast two-hybrid screens (38,51). Fig. 2, however, shows that signaling through endogenous TGF␤ receptors occurs independently of the expression of kinase-inactive (signaling-incompetent) chimeric TGF␤ receptors. To address this potential paradox, the studies shown in Fig. 3 were performed. Representative chimeric kinase-inactive type I and type II TGF␤-expressing clones (A706 and A608, respectively) were stimulated with TGF␤, GM-CSF, or TGF␤ plus GM-CSF, and the effect on PAI-1 protein secretion was determined. As expected, treatment with GM-CSF at either 10 or 100 ng/ml was unable to activate the kinase-dead chimeric receptors, while the addition of TGF␤ stimulated PAI-1 protein expression through endogenous TGF␤ receptor signaling. In order to determine whether PAI-1 expression was regulated by the oligomerization state of both the chimeric and endogenous receptor system, cultures were simultaneously treated with GM-CSF and TGF␤, and synthesis of PAI-1 protein was determined. The results show independent TGF␤ receptor signaling regardless of the heteromerization state of the kinase-dead receptors. Thus, in either the absence or presence of GM-CSF, kinase-dead chimeric receptors are unable to form a functional association (i.e. act as a dominant/negative) with endogenous TGF␤ receptors.
Internalization of Kinase-dead Chimeric Receptors-Previous work from our laboratory has shown that both homomeric and heteromeric TGF␤ receptor combinations internalize ligand at similar rates in a clathrin-dependent manner (40). However, the receptor elements or activities regulating this response have not been identified or characterized. Since the kinase activity of the chimeric type I and type II TGF␤ receptor is required for signaling (Figs. 2 and 3) and internalization of tyrosine kinase receptor family members is dependent upon receptor kinase activity (31,52,53), we wished to determine the role(s) of TGF␤ receptor serine/threonine kinase activity in ligand-mediated internalization (Figs. 4, 5, and 8) and downregulation (Figs. 6 and 9). To control for clonal bias, the kinetics of ligand internalization for two (A700s) or three (A600s) individual clones over 1 h 37°C incubation is shown (Fig. 4). Although each of the clonal families containing a kinase-dead TGF␤ receptor is signaling-incompetent (Figs. 2 and 3), differential effects on ligand internalization are observed. While clones expressing a kinase-inactive type I receptor (in the context of a wild type type II receptor; the A700 family) internalize labeled GM-CSF similar to the wild type A105 clone, expression of a kinase-inactive type II receptor (in the context of a wild type type I receptor; the A600 family), diminishes both the rate and extent of ligand internalization (Fig. 4).
The finding that internalization occurs independently of type I TGF␤ receptor kinase activity but requires a functional type II receptor kinase for optimal internalization could reflect a unique endocytic mechanism for each kinase-inactive receptor complex. Since we have previously shown that wild type chimeric TGF␤ receptors are internalized through a clathrin-dependent mechanism (40), we next determined whether the internalization of kinase-inactive type I or type II TGF␤ receptors occurred through a similar mechanism. As shown in Fig. 5, when clathrin-dependent internalization is inhibited by cytosolic K ϩ depletion (42,43), both type I and type II kinaseimpaired chimeric TGF␤ receptor heteromers are unable to internalize bound ligand. Thus, although the A600 clones show a diminished ability to internalize ligand, this is not a consequence of diverting to an alternative clathrin-independent pathway but reflects an intrinsic role(s) of the type II receptor kinase.
Down-regulation of Cell Surface Binding following Ligand Binding- Fig. 4 shows that the kinase activity of only the type II receptor is required for optimal internalization, yet both the type I and type II receptor kinases are required for signaling ( Figs. 2 and 3). These findings indicate that TGF␤ receptor endocytosis is not simply a reflection of receptor signaling but is a process controlled by distinct regulatory mechanisms. To address this question further, we next determined whether type I and/or type II TGF␤ receptor kinase activity modulated the levels of ligand binding following a preincubation of cells with GM-CSF (i.e. down-regulation). As shown in Fig. 6, receptor down-regulation over 4-h ligand stimulation was determined for multiple clones (identical to the experiment performed in Fig. 4) of wild type (kinase active) chimeric TGF␤ receptor heteromers as well as chimeric heteromers consisting of a kinase-dead type I (A700s) or type II (A600s) TGF␤ receptor. Similar to what we observed for ligand internalization (Fig.  4), clones expressing a kinase-dead type I chimeric TGF␤ receptor (A700s) also down-regulated surface binding as completely as the unmutated wild type heteromers, while clones expressing an inactive kinase in the type II TGF␤ receptor (A600s) were impaired in their ability to down-regulate cell surface receptors (Fig. 6). These results support the hypothesis that inactivation of the type II TGF␤ receptor has a dominant effect on both internalization and receptor down-regulation. Moreover, the data show that the endocytic response to heteromeric TGF␤ receptor complex formation is regulated, at least in part, by the kinase activity of the type II TGF␤ receptor.
Type II Receptor Transphosphorylation, but Not Autophosphorylation, Modulates Heteromeric Receptor Endocytosis-The type II TGF␤ receptor has both auto-and transphospho- Clones from the A600 (A608, A615, A618; ␣I␤II K277R) (Ⅺ), A700 (A706, A708; ␣II␤I K232R) (q), or wild type heteromer A105 (␣I␤II) (f) transfection groups were incubated with 100 pM radiolabeled GM-CSF for 2 h at 4°C. Unbound ligand was removed, and the cells were placed at 37°C for the indicated times when the remaining surface-bound ligand was removed by acid stripping and counted. The remaining cell-associated label was solubilized and counted as internalized ligand. A 25-fold excess of unlabeled GM-CSF was used to determine nonspecific binding. Each curve represents the average Ϯ S.E. of two (A700s) or three (A600s) independent clones. The A105s and each of the clones within a particular group were assayed three separate times in duplicate.
rylating activity (8,18). Moreover, the only known substrate for the type II receptor kinase is the type I TGF␤ receptor. Since the K277R mutation in the ATP binding site would abolish both activities, we wished to determine whether either function could account for the decreased endocytic activity seen in the 600 series clones. To that end, Carcamo et al. (12) have described a type II receptor mutation that has autophosphorylating activity in vitro and in vivo but fails to transphosphorylate an associated type I receptor. When the identical (proline to leucine at amino acid 525) mutation was made in the chimeric type II TGF␤ receptor, we also found the heteromeric receptor complex unable to stimulate expression of PAI-1 protein following ligand binding (Fig. 7A). A similar result is observed when luciferase activity is measured from the TGF␤-responsive 3TP-Lux reporter plasmid (Fig. 7B). Although the addition of TGF␤ increased luciferase expression 30 -40-fold through endogenous TGF␤ receptors, stimulation of the mutant P525L chimeric receptor (in the same cell clone) did not increase luciferase activity. Since mutation at amino acid 525 in the type II receptor had a similar effect on chimeric receptor signaling as that reported for the endogenous TGF␤ receptor, we next examined whether ligand internalization or receptor down-regulation were also modified. As shown in Figs. 8 and 9, internalization (Fig. 8) and down-regulation (Fig. 9) in the P525L clones were diminished to a similar extent as that observed for the kinase-inactive A600 cultures (compare Figs. 4 and 8 and Figs. 6 and 9). Thus, although TGF␤ receptor signaling and endocytosis are distinctly regulated activities (Figs. 2-4 and 6), they are both highly dependent upon the transphosphorylating activity of the type II receptor kinase.

DISCUSSION
Regulated control of the endocytic response constitutes one of the earliest cellular mechanisms for responding to environmental stimuli. Class I receptors, such as the low density lipoprotein receptor are constitutively endocytosed, while class II receptors, such as the epidermal growth factor and insulin receptor undergo ligand-dependent endocytosis. While it is well established that receptor tyrosine kinase activity is required for optimal internalization of full-length receptors (31,52,53), this obligate requirement can be partially overcome in truncated receptors (30,54). These findings suggest that kinase activation removes an inhibitory signal or exposes a motif, which then allows internalization and down-regulation. While this complex relation of kinase activation and receptor endocytosis in tyrosine kinase receptors has been extensively investigated, the relationship between TGF␤ receptor activation and endocytosis has not been similarly examined. There are two likely reasons for this: first, quantitative 125 I-TGF␤ binding studies are compromised by a high degree of nonspecific binding; second, the natural occurrence of both heteromeric and homomeric TGF␤ receptor interactions makes any analysis problematic. In that regard, the chimeric system is ideally suited to address both of these concerns (4). For instance, our recent studies have shown that heteromeric and homomeric TGF␤ receptor complexes in mesenchymal AKR-2B cells have distinct endocytic fates (40). While these results suggest a requirement for receptor cross-talk, the regulatory role(s) of the type I and/or type II TGF␤ receptor kinase in the endocytic process has not been addressed. Since each of these receptor kinases have such distinct roles in TGF␤ receptor activation, it is likely that novel paradigms will need to be developed defining the mechanisms whereby phosphorylation regulates TGF␤ receptor endocytosis and trafficking.  6. Receptor down-regulation is regulated by the kinase activity of the type II TGF␤ receptor. Clones from the A600 (A608, A615, A618; ␣I␤II K277R) (f), A700 (A706, A708; ␣II␤I K232R) (Ⅺ), or wild type heteromer (A105, ␣I␤II; A110 ␣II␤1) (q) transfection groups were treated with 10 ng/ml (520 pM) GM-CSF for the indicated times, and receptor down-regulation was performed. Following acid treatment to remove bound ligand, specific surface binding of radiolabeled GM-CSF (100 pM) was determined after a 2-h incubation at 4°C in the presence or absence of 25-fold excess unlabeled GM-CSF. Percentage of control binding represents the percentage of zero time-specific binding observed following GM-CSF treatment for the indicated time. Each curve represents the average Ϯ S.E. of two (wild type heteromers and A700s) or three (A600s) independent clones each of which were assayed twice in duplicate (i.e. each data point represents 8 or 12 independent analyses). Analysis of variance and Student's t test for the 2-and 4-h points show significant differences between the A600s and the wild type heteromers (p values of 0.02 and 0.03 at 2 and 4 h, respectively).
To initially address the question of whether TGF␤ receptor kinase activity is linked to internalization and down-regulation, cell lines were generated stably expressing chimeric receptors containing mutations in the putative ATP binding site for both the type I and type II receptors (55,56). Lysine to arginine mutations at position 232 in the type I receptor and 277 in the type II receptor were engineered into the chimeric receptors and transfected into parental AKR-2B fibroblasts in the heteromeric combinations ␣I␤II K277R and ␣II␤I K232R. Once stable clones expressing the receptor combinations were isolated (Fig. 1), each clone's ability to signal TGF␤-dependent responses was assayed. As shown in Figs. 2 and 3, each of the clones expressing the kinase-inactive receptors did not induce secretion of PAI-1 protein in response to GM-CSF treatment. This was not due to a general lesion in the TGF␤ signaling pathway, since activation of the endogenous TGF␤ receptors resulted in PAI-1 protein expression. A similar response for all the clones was seen when the ability to form colonies in soft agar was measured (data not shown).
It has been previously shown that kinase-inactive TGF␤ receptors can function as dominant/negatives to inactivate TGF␤ signaling both in vitro and in vivo (45)(46)(47)(48)(49)(50). In addition, both heteromeric and homomeric complexes of the type I and type II TGF␤ receptor cytoplasmic domains have been observed in yeast two-hybrid screens and overexpressing COS cells (38,51). Since we did not observe functional association between the kinase-inactive chimeric receptors and the endogenous TGF␤ receptors (i.e. inactivation of TGF␤-dependent signaling) (Figs. 2 and 3), this suggested that either the level of chimeric receptor expression was not great enough for inhibition or the chimeric and endogenous TGF␤ receptors associated into separate signaling complexes. In support of the latter possibility, ligand binding to the endogenous TGF␤ receptors does not result in the heterologous down-regulation of the chimeric receptors (data not shown). Moreover, when both receptor families are activated by simultaneous treatment with GM-CSF and TGF␤ (Fig. 3), the oligomerization state of one family does not affect the signaling activity of the other complex. Although these studies do not directly document the specific receptor interactions formed, they are consistent with the hypothesis, recently proposed by Luo and Lodish (5), that the TGF␤ receptor associations formed in vivo are linked through interactions between their extracellular (not cytoplasmic) domains. Since the chimeric and endogenous receptors only share transmembrane and cytoplasmic domains, there would be no association (i.e. no dominant/negative effect) regardless of whether the receptor families were activated by ligand.
Mutation of the intrinsic kinase activity of the tyrosine kinase family of receptors abolishes their ability to internalize radiolabeled ligand (30,31). When similar studies were per-FIG. 7. Expression of PAI-1 protein is not induced by chimeric type II receptors harboring the P525L mutation. A, clones 13 and 22 expressing a wild type chimeric ␣1 receptor and a chimeric ␤2 receptor with a mutation of proline 525 to leucine were treated for 4 h in methionine-free DMEM alone (Con) or supplemented with 10 ng/ml GM-CSF (GM) or TGF␤2 (TGF ␤). The extracellular matrix-associated PAI-1 protein was analyzed as described under "Experimental Procedures." B, wild type heteromeric clone A105 or P525L clones 13 and 22 were transiently transfected with the 3TP-Lux reporter as described previously (4). Following recovery, arrested cells were stimulated for 24 h in 5% FBS/DMEM alone (clear bar), 5% FBS/DMEM supplemented with 10 ng/ml GM-CSF (gray bar), or 10 ng/ml TGF␤2 (black bar). Luciferase activity was then determined on normalized samples. The data are depicted as the -fold increase in luciferase activity relative to the mean of the control-treated 525 clones and represent the mean Ϯ S.D. of two separate experiments for each clone done in duplicate. formed on the chimeric receptors containing a kinase-inactive type I or type II TGF␤ receptor, differential effects on ligand internalization were observed (Fig. 4). While internalization was unaffected by the absence of a functional type I receptor kinase (A700 clones), cultures expressing a kinase-inactive type II TGF␤ receptor (A600 clones) showed a diminished rate and extent of internalization. Although the underlying mechanism(s) regulating this response is currently unknown, it has been postulated that phosphorylation by receptor tyrosine kinases induces conformational changes necessary for revealing internalization motifs (30,34); perhaps similarly acting elements are exposed in the type I TGF␤ receptor following activation by the type II receptor kinase.
Ligand internalization is routinely followed by a decrease in cell surface receptor binding referred to as down-regulation. As shown in Fig. 6, treatment of heteromeric wild type chimeric TGF␤ receptors with GM-CSF results in a 60 -80% downregulation of surface binding by 2-4 h. Consistent with that observed for the internalization studies in Fig. 4, inactivation of the type I TGF␤ receptor kinase (A700s, ␣II␤I K232R) did not affect receptor down-regulation. Since type I receptor kinase activity is required for cellular signaling (Figs. 2 and 3 and data not shown), yet no effect is observed on heteromeric TGF␤ receptor internalization and down-regulation in the absence of a functional type I receptor kinase (Figs. 4 and 6), this supports the hypothesis that receptor endocytosis is not dependent upon, or the result of, TGF␤ receptor signaling. However, in contrast to that observed in the A700 clones, mutation of the type II TGF␤ receptor kinase (A600s, ␣I␤II K277R) decreased both internalization and receptor down-regulation by approximately 50% (Figs. 4 and 6). While these data show a primary regulatory role for the type II TGF␤ receptor kinase, it is of interest that the A600 clones have residual endocytic activity. This suggests involvement of other receptor elements, substrates, and/or receptor interactions in addition to kinase activity in regulating TGF␤ receptor endocytosis.
The signaling activity of the type II TGF␤ receptor has been recently shown to be both positively and negatively regulated by various phosphorylations (18,57,58). In addition to these autophosphorylations, the type II receptor associates with, transphosphorylates, and activates the type I receptor. No other kinase has been shown to phosphorylate the type I receptor in vivo, nor have other substrates been reported for the type II receptor. Since Figs. 4 and 6 demonstrated that the endocytic response of the TGF␤ receptor complex was dependent upon the type II receptor kinase, we next wished to determine whether this was a reflection of either the auto-or transphosphorylating activity of the type II receptor. To address this question, a proline to leucine mutation at amino acid 525 was made in the chimeric type II receptor, which had been previously shown to abolish type II receptor transphosphorylating activity but have no effect on autophosphorylation (12). When the endocytic response of these clones was examined, ligand internalization and receptor down-regulation was affected similarly to that seen in cultures containing a kinaseinactive type II receptor (compare Figs. 4 and 6 with Figs. 8 and 9, respectively). The data are consistent with a model whereby phosphorylation of the type I receptor (or some other substrate) by the type II TGF␤ receptor is necessary for efficient receptor down-regulation and trafficking. Direct test of this model will include 1) identifying the particular site(s) in the type I receptor phosphorylated by the type II receptor necessary for appropriate receptor cross-talk and/or 2) determining whether a substrate (in addition to the type I TGF␤ receptor) for the type II receptor kinase is required for endocytosis similar to that proposed for the epidermal growth factor receptor (31).
The present results add several new concepts to our understanding TGF␤ receptor interactions including 1) documenting the requirement for both type I and type II TGF␤ receptor kinase activity in chimeric receptor signaling; 2) providing additional evidence that the chimeric and endogenous TGF␤ receptors are functionally independent; 3) showing that internalization mediated through heteromeric TGF␤ receptors can occur independent of type I receptor kinase activity; 4) demonstrating that the endocytic and signaling activities of the TGF␤ receptors are distinctly regulated; and 5) determining that ligand internalization and receptor down-regulation are controlled, at least in part, by the transphosphorylating activity of the type II TGF␤ receptor. It is becoming increasingly clear that the paradigms developed for other receptor families may need to be modified as we further characterize the signaling and endocytic activities of the TGF␤ receptor superfamily.