The Type II Transforming Growth Factor- (cid:98) Receptor Autophosphorylates Not Only on Serine and Threonine but Also on Tyrosine Residues*

The type I and type II receptors for transforming growth factor- (cid:98) (TGF- (cid:98) ) are structurally related transmembrane serine/threonine kinases, which are able to physically interact with each other at the cell surface. To help define the initial events in TGF- (cid:98) signaling, we characterized the kinase activity of the type II TGF- (cid:98) receptor. A recombinant cytoplasmic domain of the re- ceptor was purified from Escherichia coli and baculovi-rus-infected insect cells. Anti-phosphotyrosine Western blotting demonstrated that the type II receptor kinase can autophosphorylate on tyrosine. Following an in vitro kinase reaction, the autophosphorylation of the cytoplasmic domain and phosphorylation of exogenous substrate was shown by phosphoamino acid analysis to occur not only on serine and threonine but also on tyro- sine. The dual kinase specificity of the receptor was also demonstrated using immunoprecipitated receptors ex- pressed in mammalian cells and in vivo 32 P labeling showed phosphorylation of the receptor on serine and tyrosine. In addition, the kinase activity of the cytoplasmic domain was inhibited by the tyrosine kinase inhib- itor tyrphostin. Tryptic mapping and amino TGF- as specificity suggest a role for in (cid:98) receptor m M m M kinase reactions were terminated by addition of an equal volume of 2 (cid:51) sample buffer. samples for 5 prior electrophoretic separation by SDS-PAGE. The gel was then fixed, dried, and autoradiographed. Phosphoamino Acid Analysis— Kinase reactions were done as described above and the 32 P-phosphorylated products were transferred to polyvinylidene difluoride membranes (Bio-Rad) after separation by SDS-PAGE. The reaction products visualized by autoradiography were cut from the membrane and subjected to phosphoamino acid analysis at 100 °C as described affinity-purified 6 IIK each cycle C18

The importance of protein phosphorylation in various signal-ing events that regulate cell proliferation has been well documented. Most mitogenic growth factors interact with transmembrane tyrosine kinases or receptors that associate with cytoplasmic tyrosine kinases, which, as a result of ligand-induced autophosphorylation, trigger signaling cascades that involve multiple phosphorylation events (1,2). Initiated by tyrosine phosphorylation, these cascades involve several serine/ threonine kinases as well as a dual specificity kinase, MAP 1 kinase kinase, which activates its substrate MAP kinase by phosphorylation on both tyrosine and threonine residues (3,4).
In contrast to many mitogenic growth factors, transforming growth factor-␤ (TGF-␤) induces an antiproliferative effect in many cell types, including epithelial, endothelial and hematopoietic cells (5)(6)(7). In addition to its growth modulatory activity, TGF-␤ has a wide range of effects on extracellular matrix synthesis, cell-substrate adhesion, cell differentiation, and migration (5)(6)(7)(8). TGF-␤, which exists as three isoforms encoded by separate genes (9,10), is considered a prototype for the many structurally related members of the TGF-␤ superfamily, which play important roles in diverse cell differentiation and developmental processes.
Until the recent cloning and characterization of the cell surface receptors for TGF-␤, little was known about the mechanisms of signal transduction by this growth factor or related factors. Cross-linking studies had previously shown the presence of several cell surface TGF-␤ binding proteins (for reviews, see , with most cells expressing three types of high affinity cell surface binding components known as types I, II, and III receptors. Studies on mutant cell lines lacking functional type I or type II receptors showed that these two receptor types mediate most if not all TGF-␤ responses and that both receptor types are required for full responsiveness to TGF-␤ (11)(12)(13)(14)(15). The type III receptor, also known as betaglycan, is not required for TGF-␤ signaling but may contribute to presentation of the ligand to the type II receptor (16,17).
The type I and type II TGF-␤ receptors are structurally related transmembrane kinases with a cytoplasmic segment consisting largely of a kinase domain, which has a predicted specificity for serine and threonine (reviewed in [11][12][13]. In fact, the serine/threonine kinase activity of these receptors has been experimentally verified both in vitro (18 -20) and in vivo (21,22). In addition to the TGF-␤ type II receptor, several other type II receptors for TGF-␤ related ligands have been charac-terized, including two types of type II activin receptors (18,23,24), a Caenorhabditis elegans type II receptor that binds BMP-2 and BMP-4 (25), a mammalian BMP-2/4 type II receptor (26,27), and a Drosophila type II receptor that binds the related Dpp gene product (28,29). A number of type I receptors have also been cloned (20, 30 -38). They are generally smaller than the type II receptors, have a defined cysteine pattern in their extracellular domains, and contain a highly conserved SGSGSGLP sequence immediately upstream of the cytoplasmic kinase domain. In contrast to the type II receptors, which define their own specificity of ligand binding, many type I receptors have their specificity of ligand binding largely determined by the coexpressed type II receptor. For example, the type I receptors Tsk7L (39) and TSR1 (31) bind TGF-␤ or activin depending on the coexpressed type II receptor. On the other hand, the ALK-5/R4 receptor is primarily a functional type I receptor for TGF-␤ (32,34).
The type I and type II receptors cooperate in signal transduction, and both receptor types are required for full responsiveness to TGF-␤ (20, 40 -43). Type II and type I receptors physically interact with each other, and such heteromeric complex formation is required for efficient ligand binding to type I receptors (20,21,39,40,44). The type II and type I receptors also exist as homomeric receptor complexes at the cell surface (45,46). These findings led to the proposal that the two receptor types form a heteromeric, probably tetrameric, type II/type I receptor complex (12,47,48), which mediates TGF-␤ signaling. In this complex, the cytoplasmic domains of the type II receptor are constitutively phosphorylated on serine and threonine, due to ligand-independent autophosphorylation and to phosphorylation by other cytoplasmic kinases (21,22). Furthermore, the type II receptor kinase phosphorylates the cytoplasmic domain of the type I receptor on serine and threonine (21,22) and the phosphorylation of both types of cytoplasmic domain contributes to the stability of the heteromeric complex (44). The existence of multiple type II receptors with defined ligand binding specificity and various type I receptors with an ability to bind different ligands, depending on the nature of the co-expressed type II receptor, suggests the existence of a complex signaling system in which combinatorial interactions may provide a substantial degree of flexibility in the cellular responses to TGF-␤ and related factors.
To gain insight into the initial events in the signaling of TGF-␤ and related factors, we have further characterized the kinase activity of the type II TGF-␤ receptor. Using the cloned type II TGF-␤ and activin receptors, it has previously been shown that their kinase domains are able to autophosphorylate on serine and threonine (18 -22, 49). However, a detailed comparison of kinase domain sequences of these and related receptors indicates some structural similarities with tyrosine kinases (50). 2 In addition, endogenous activin type II receptor purified from mammalian cells exhibited not only serine and threonine but also tyrosine kinase activity (51). In contrast, the recombinant type II receptors for activin and TGF-␤ have been reported to only have serine and threonine kinase activity (18 -22, 49). Because of this apparent contradiction, we have studied the autophosphorylation activity of the type II TGF-␤ receptors. We show that the cytoplasmic domain of this receptor phosphorylates itself and exogenous substrates not only on serine and threonine but also on tyrosine residues. We have also localized the autophosphorylated tyrosine residues in the cytoplasmic domain of the type II receptor. Replacement of these tyrosines by phenylalanines strongly inhibits the kinase activity of the type II receptor. Our results establish the type II TGF-␤ receptor as a dual specificity kinase, which is autophosphorylated not only on serine and threonine but also on tyrosine, and suggest a dual specificity activity for other members of this receptor kinase family.

EXPERIMENTAL PROCEDURES
Plasmid Construction and Mutagenesis-Plasmid pGST-IIK was designed to express in Escherichia coli the C-terminal 374 amino acids of the type II TGF-␤ receptor cytoplasmic domain as a glutathione Stransferase (GST) fusion protein. The corresponding coding region of the human type II TGF-␤ receptor cDNA was amplified using the polymerase chain reaction (PCR), incorporating flanking EcoRI restriction sites. The PCR primers used were 5Ј-GGGGCCGAATTCCGGCAG-CAGAAGCTGAGTTC-3Ј and 5Ј-GGGGCCGAATTCGAGCTATTTGG-TAGTGTTTAGG-3Ј. The EcoRI fragment was then ligated into the EcoRI site of pGEX2T (Pharmacia Biotech Inc.), thus generating pGST-IIK and the sequence of this insert was verified using the Sequenase kit (U. S. Biochemical Corp.).
Expression plasmid pVL1393-(His) 6 IIK was constructed to express the cytoplasmic kinase domain of the type II receptor in the baculovirus expression system. The same receptor cDNA fragment as in pGST-IIK was ligated into the EcoRI site of the baculovirus expression vector pVL1393(His) 6 to allow expression of the cytoplasmic domain with an N-terminal (His) 6 extension. pVL1393(His) 6 was constructed by inserting oligonucleotide linkers encoding the sequence Met-Ser-(His) 6 into the BamHI and EcoRI cut plasmid pVL1393 (Invitrogen). The linkers used were 5Ј-GATCCTATAAATATGTCGCATCATCATCATCATCATG-GTTCCATGG-3Ј and 5Ј-AATTCCATGGAACCATGATGATGAT-GATGATGCGACATATTTATAG-3Ј.
Plasmid pIIR-myc (46) expresses the full-length human type II TGF-␤ receptor with a C-terminal Myc epitope tag when transfected into mammalian cells.
In vitro mutagenesis of the cytoplasmic domain was done using the Sculptor kit (Amersham) according to the manufacturer's recommendations. The PCR product used in the construction of pGST-IIK was subcloned into M13mp18 to mutagenize the sequence encoding the cytoplasmic domain of the type II receptor. The mutated inserts were then ligated back into pGEX2T to generate the GST-fusion proteins in E. coli.
Type II TGF-␤ Receptor Kinase Domain Expression and Purification-The GST-IIK fusion protein was prepared as described (52). Briefly, 1 ml of an overnight culture of E. coli DH5␣ cells transformed with pGST-IIK was used to inoculate 1 liter of LB medium containing 50 g/ml ampicillin. The culture was grown to an A 600 nm of 1.0, and expression of the fusion protein was induced with 0.2 mM isopropyl-1thio-␤-D-galactopyranoside. After an additional 5 h, the culture was harvested by centrifugation and the cells, resuspended in 30 ml of NETN (100 mM NaCl, 5 mM EDTA, 20 mM Tris, pH 7.4, 0.5% Nonidet P-40) containing protease inhibitors, were lysed by one freeze-thaw cycle followed by sonication for 2 min. The lysate was centrifuged, and 200 l of glutathione-Sepharose 4B (Pharmacia) was added to the cleared supernatant and incubated in suspension for 1 h. Following adsorption, the beads were pelleted by centrifugation, washed three times with 30 ml of NETN, and resuspended at a concentration of 50% in NETN.
To express the (His) 6 -tagged cytoplasmic domain in insect cells, pVL1393-(His) 6 IIK was cotransfected with PharMingen Baculogold linearized baculovirus DNA into SF9 insect cells. Plaque purification and recombinant virus screening were carried out as described (53). For the production of fusion protein, cells were harvested 48 -52 h after infection, pelleted, and lysed by resuspension in insect cell lysis buffer (10 mM HEPES, pH 7.4, 10 mM NaCl, 1 mM EDTA). After 20 min on ice, the suspension was cleared for 10 min in a microcentrifuge. The expressed fusion protein was purified by absorption through its (His) 6 -sequence using Co 2ϩ -chelate affinity chromatography. Briefly, the cleared cell lysate was incubated with Co 2ϩ -Sepharose 6B beads in 20 mM sodium phosphate, pH 8.0, 300 mM NaCl, 10% glycerol for 30 min at 4°C. The beads were washed in the same buffer containing 15 mM imidazole. The adsorbed protein was then eluted in buffer containing 100 mM imidazole and stored at Ϫ20°C.
Transient Expression and Immunoprecipitation of Myc-tagged Type II Receptor-Plasmid pIIR-myc, which drives the expression of a Myctagged human type II receptor (46), was transiently transfected (54) into 293 cells using 25 g of DNA/10-cm diameter plate. Cells were metabolically labeled using [ 35 S]Cys and [ 35 S]Met, and the 35 S-labeled type II receptor was immunoprecipitated as described (46). The anti-Myc monoclonal antibody 9E10 was obtained from Dr. J. M. Bishop (University of California, San Francisco). Anti-phosphotyrosine immunoprecipitations were carried out using the PY20 (Zymed) and 4G10 (Upstate Biotechnology, Inc.) antibodies according to the manufacturers' recommendations.
In Vitro Kinase Assays and Kinase Inhibitor Studies-Ten l (0.1-0.5 g) of GST-IIK fusion protein, bound to glutathione-Sepharose and washed in 500 l of kinase buffer (25 mM HEPES, pH 7.4, 10 mM MnCl 2 ), was used for each reaction. Kinase reactions were carried out for 15 min in a final volume of 30 l of kinase buffer. When appropriate, histone 2B (Boehringer Mannheim) was added as substrate at a final concentration of 100 ng/l. The kinase reactions using purified (His) 6 -IIK protein were carried out similarly. Kinase assays were also carried out on full-length receptors expressed in mammalian cells. 293 cells were transfected with pIIR-myc and lysed, and the receptors were immunoprecipitated 72 h after transfection. The immune complexes absorbed to protein A-Sepharose were then washed with GST kinase buffer (33% glycerol, 0.1% Triton X-100, 25 mM HEPES, pH 7.4, 10 mM MnCl 2 , 1 mM NaVO 4 ). The kinase reactions were initiated by addition of ATP to 10 M and 1 l of [␥-32 P]ATP (DuPont NEN; 3000 Ci/mmol) and allowed to proceed for 5 min at room temperature. Kinase inhibitors (Life Technologies, Inc.) were incorporated when appropriate and used according to the manufacturer's guidelines and as described previously (55). The kinase reactions were terminated by addition of an equal volume of 2 ϫ sample buffer. The samples were boiled for 5 min prior to electrophoretic separation by SDS-PAGE. The gel was then fixed, dried, and autoradiographed.
Phosphoamino Acid Analysis-Kinase reactions were done as described above and the 32 P-phosphorylated products were transferred to polyvinylidene difluoride membranes (Bio-Rad) after separation by SDS-PAGE. The reaction products visualized by autoradiography were cut from the membrane and subjected to phosphoamino acid analysis at 100°C as described (56).
Peptide Mapping and Sequencing-100 g of affinity-purified (His) 6 -IIK fusion protein was phosphorylated in vitro as described above in a total volume of 100 l in the presence of unlabeled ATP at a concentration of 1 mM. In parallel, 1-5 g of (His) 6 -IIK fusion protein was autophosphorylated in the presence of [␥-32 P]ATP. After completion of the assay, 400 l of 50 mM ammonium bicarbonate, 20 mM EDTA containing 1 g of modified trypsin, which does not undergo autoproteolysis (Promega), was added to the mixture which contained approximately 200,000 cpm of the 32 P-labeled reaction product. Digestion was carried out for 2 h at 37°C as described (57), and the reaction products were separated by SDS-PAGE, transferred to nitrocellulose, and treated with trypsin. To resolve the tryptic peptides by HPLC, the protein digest was loaded onto a reverse phase C18 column (25 cm ϫ 4.6 mm, Vydac, Hesperia CA) and peptides were eluted with a linear gradient of acetonitrile in 0.1% trifluoroacetic acid/water over 1 h. The elution profile of the 32 P-labeled peptides was determined by Cerenkov counting of individual fractions. 100 l of each fraction was dried down and subjected to phosphoamino acid analysis. Peptide peaks that contained phosphotyrosine were sequenced by Edman degradation on a protein sequencer (model 492; Appied Biosystems, Inc., Foster City, CA) to establish their sequence and the location of the phosphorylated tyrosines. These peaks were usually a mixture, but allowed us to establish the sequences of individual peptides. The amount of radioactivity corresponding to each cycle of Edman degradation and comparison with predicted sequences of the tryptic peptides allowed us to assign the phosphorylation to particular amino acids. Finally, the peptides corresponding to the established tyrosine-phosphorylated sequence were synthesized, and their elution positions on the C18 column confirmed the sequence identity of the tyrosine-phosphorylated tryptic peptides.
In Vivo 32 P Labeling of the Receptor and Phosphoamino Acid Analysis-293 cells were transiently transfected with plasmid pIIR-myc as described above, and the transfected cells were labeled with 1 mCi/ml of [ 32 P]phosphate in Dulbecco's modified Eagle's medium, 3 g/liter glucose without phosphate, for 12 h. The cells were then lysed, and the type II receptors were immunoprecipitated using the anti-Myc antibody as described above for 35 S-labeled receptors. Following SDS-PAGE, the 32 P-labeled receptors were transferred to polyvinylidene difluoride membranes, visualized by autoradiography, and processed for phosphoaminoacid analysis as outlined above.
Functional Assays-Plasmid p800Luc contains a TGF-␤-responsive promoter for plasminogen activator inhibitor type I (PAI-1), which controls the expression of luciferase (58). This reporter plasmid was used to score TGF-␤-induced gene expression in transient transfection assays. Plasmid pCAL2, which contains a TGF-␤-responsive cyclin A promoter (Ϫ516 to ϩ245), is a reporter plasmid to score TGF-␤-induced growth inhibition (59). Plasmid pRK␤Gal, which expresses ␤-galactosidase under the control of the cytomegalovirus promoter, was used to normalize for transfection efficiency (59).
The reporter assays were performed following transient co-expression of TGF-␤ receptor and reporter expression plasmids in Mv1Lu cells as described (59). Briefly, cells (ϳ3 ϫ 10 6 ) were harvested and electroporated at 960 microfarads and 750 V/cm in a Bio-Rad electroporation apparatus, typically using 10 g of p800luc or pCAL2, 10 g of pRK␤GAL, 10 g of receptor expression plasmid(s), and 10 g of pBluescript II SKϩ (carrier DNA). After electroporation, cells were allowed to recover for 4 h in Eagle's minimal essential medium with Earle's BSS supplemented with NEAA and 10% fetal bovine serum. The cells were then treated with or without 10 ng/ml TGF-␤1 in the same medium but containing 0.2% fetal bovine serum. After 24 h (for PAI-1 assay) or 48 h (for cyclin A assay), the cells were harvested and lysed in reporter lysis buffer (Promega) and cell lysates were assayed for luciferase and ␤-galactosidase activities. The luciferase assay was carried out using Analytic Luminescence Laboratory's assay reagents, and ␤-galactosidase was assayed in Galacto Light Plus kit (Tropix). Both luciferase and ␤-galactosidase activity was measured in luminometer Monolight 2010. The luciferase activity, which reflects the promoter activity of cyclin A or PAI-1, was normalized to ␤-galactosidase to account for transfection efficiency.

Expression of the Cytoplasmic Domain of the Type II TGF-␤
Receptor in E. coli and Insect Cells-To study the kinase activity of the type II TGF-␤ receptor, we expressed its cytoplasmic domain in E. coli and insect cells. In E. coli, the sequence of the C-terminal 374 amino acids of the type II TGF-␤ receptor including the kinase domain, was expressed as a fusion to an N-terminal GST segment that confers high affinity of the fusion protein for glutathione-Sepharose (52). The affinity-purified protein consisted of about 90% full-length GST-IIK fusion protein with its degradation products as major contaminants (Fig. 1A).
The cytoplasmic domain of the type II TGF-␤ receptor was also expressed in baculovirus-infected insect cells. For this purpose, we constructed an expression vector encoding the C-terminal 374 amino acids of the type II TGF-␤ receptor preceded by an N-terminal (His) 6 tag. This fusion protein, (His) 6 -IIK, was purified from infected insect cell lysates by absorption chromatography based on the high affinity of the (His) 6 sequence for Co 3ϩ -chelate resin (Fig. 1B).
The two types of purified fusion proteins were used to characterize the kinase activity of the type II receptor in vitro, as discussed below. Whereas both proteins had similar properties, the (His) 6 -IIK protein from insect cells had a higher specific kinase activity (data not shown). This difference in specific activities is likely due to the fact that a large fraction of the GST-IIK protein expressed in E. coli may have been obtained as an inactive protein in inclusion bodies.
Auto-and Substrate Phosphorylation on Serine, Threonine, and Tyrosine-Initially, the presence of phosphotyrosine in the type II receptor cytoplasmic domain was tested by Western blot analyses using the anti-phosphotyrosine monoclonal antibody PY20 (Zymed). GST-IIK fusion protein reacted with the antiphosphotyrosine antibody, whereas the degradation products containing the GST sequence did not (Fig. 2, lanes 1). The tyrosine-phosphorylated GST-IIK migrated slightly slower than the non-tyrosine phophorylated protein (data not shown), as often observed with differentially phosphorylated proteins. Since E. coli lacks detectable tyrosine kinase activity, any phosphotyrosine should have resulted from its intrinsic kinase activity. This was confirmed by testing a kinase-inactive version of GST-IIK in which the lysine at position 277 in the ATP binding site was replaced by arginine. This mutated fusion protein prepared in a similar way to GST-IIK was not phosphorylated on tyrosine as assessed by Western blot analysis (Fig. 2, lanes 2), indicating that tyrosine phosphorylation was dependent on the kinase activity of the cytoplasmic domain of the type II receptor.
Purified GST-IIK protein and baculovirus-derived (His) 6 -IIK were subjected to in vitro kinase assays in the presence of [␥-32 P]ATP. As illustrated with GST-IIK, the fusion protein displayed kinase activity and autophosphorylated (Fig. 3A,  lane 1). Autophosphorylated GST-IIK protein could be immunoprecipitated using anti-phosphotyrosine antibody 4G10 (Fig.  3A, lane 2), further documenting the ability of the cytoplasmic domain to autophosphorylate on tyrosine. GST-IIK or (His) 6 -IIK, 32 P-labeled in autophosphorylation reactions, was subjected to phosphoamino acid analysis. In addition to phosphoserine as predominant phosphoamino acid and phosphothreonine, phosphotyrosine was clearly detected (Fig.   3B), confirming the dual specificity of the type II TGF-␤ receptor kinase in vitro.
Finally, the kinase specificity of the type II TGF-␤ receptor cytoplasmic domain was tested using exogenous substrates. Histone 2B, enolase, poly(Glu-Tyr), and casein were efficiently phosphorylated by both the GST-IIK and (His) 6 -IIK fusion proteins (Fig. 4A), and phosphoamino acid analysis of 32 Plabeled histone 2B revealed its phosphorylation primarily on serine but also on threonine and tyrosine (Fig. 4B), a pattern similar to that seen in autophosphorylation reactions.
The Type II TGF-␤ Receptor Immunoprecipitated from Transfected Mammalian Cells Phosphorylates Serine, Threonine, and Tyrosine Residues-The experiments described above indicate that the recombinant type II TGF-␤ receptor cytoplasmic domain has dual kinase specificity in vitro. To verify that the full-length type II receptor expressed in mammalian cells had a similar kinase specificity, we performed immunoprecipitations of Myc epitope-tagged type II receptors, transiently expressed in transfected 293 cells. An expression vector for the human type II TGF-␤ receptor with the C-terminal epitope tag (46) was transfected into 293 cells. These cells have, based on cross-linking of cell surface receptors with 125 I-TGF-␤, low endogenous levels of type II TGF-␤ receptors (30). The immunoprecipitated receptor had the expected molecular mass of about 69 kDa (Fig. 5, lane 2) and was autophosphorylated in an in vitro kinase assay in the presence of [␥-32 P]ATP (Fig. 5, lane  4). Similar immunoprecipitations carried out using cells transfected with untagged receptor (Fig. 5, lanes 1 and 3) or a kinase-deficient receptor point mutant (Ref. 60 and data not shown; see below) or using untransfected cells (data not shown) revealed no detectable levels of phosphorylation and demonstrated that the observed kinase activity was due to the immunoprecipitated type II receptor.
Phosphoamino acid analysis of autophosphorylated Myctagged type II receptor showed the presence of predominantly phosphoserine with less phosphothreonine and phosphotyrosine (Fig. 6, panel 1), a pattern consistent with the activity of the cytoplasmic domain expressed in E. coli or baculovirusinfected cells. Histone 2B could also be phosphorylated by the immunoprecipitated type II receptor kinase. Again, phosphoamino acid analysis showed that the 32 P-labeled histone 2B contained phosphorylated serine, threonine and tyrosine (Fig.  6, panel 2), further documenting the dual kinase specificity of the type II TGF-␤ receptor.
The Type II TGF-␤ Receptor Is Phosphorylated on Tyrosine in Vivo-The type II TGF-␤ receptor is known to be constitutively autophosphorylated on serine and threonine (21,22). We thus determined whether the receptor expressed in vivo can also be phosphorylated on tyrosine. The transfected Myc-tagged type II receptor was 32 P-labeled in vivo and immunoprecipitated using the tag-specific antibody (Fig. 7, left lane). Phosphoamino acid analysis of the gel-purified type II receptor band revealed the presence of a low level of phosphotyrosine, in addition to the abundant phosphoserine (Fig. 7), thus confirming the dual kinase specificity of the receptor in vivo. Parallel experiments using the kinase-inactive point mutant of the type II receptor expressed in transfected cells, revealed a much lower level of phosphorylation of the receptor band, presumably due to phosphorylation by cytoplasmic kinases (21,49), resulting in a much lower level of phosphoserine and no detectable tyrosine phosphorylation (data not shown).
Sensitivity of the Kinase Activity to Tyrphostin and Other Kinase Inhibitors-The type II receptor kinase was tested for its sensitivity to a panel of kinase inhibitors. The kinase activity of the immunoprecipitated Myc-tagged type II receptor was partially inhibited by staurosporine, an inhibitor of many serine/threonine kinases, and methyl 2,5-dihydroxycinnamate, but not by several other kinase inhibitors, such as genistein and lavendustin A (Fig. 8A). Interestingly, the kinase activity of the type II receptor was strongly inhibited by tyrphostin, a competitive inhibitor of substrate binding to some tyrosine kinases (Fig. 8B). The latter result further supports the finding that the receptor phosphorylates on tyrosine and suggests that, in contrast to most or all serine/threonine kinases, the active site of the enzyme can accommodate a tyrosine as substrate.
Localization of the Phosphorylated Tyrosines in the Type II Receptor Cytoplasmic Domain-To evaluate the biological importance of the autophosphorylation of the type II TGF-␤ receptor on tyrosine, we identified the phosphorylated tyrosine residues in the cytoplasmic domain. For this purpose, purified (His) 6 -IIK protein was autophosphorylated in vitro. Whereas most of the protein was phosphorylated using unlabeled ATP, a fraction was phosphorylated with [␥-32 P]ATP in a separate reaction. The autophosphorylated (His) 6 -IIK protein was di-  1 and 3) or Myc-tagged (lanes 2 and 4) type II receptor. Immunoprecipitations were carried out using an anti-Myc monoclonal antibody. In lanes 1 and 2, the cells were metabolically 35 S-labeled prior to immunoprecipitation and SDS-PAGE, whereas in lanes 3 and 4, the cells were unlabeled and subjected to immunoprecipitations followed by in vitro 32 P-kinase reaction and SDS-PAGE. RII marks the type II receptor, which has size heterogeneity due to differences in glycosylation (46), and several degradation products including one prominent one. gested with trypsin, and tryptic peptides, obtained following HPLC separation, were assayed for the presence of phosphotyrosine using phosphoamino acid analysis. The three HPLCfractionated peptide peaks, which contained phosphotyrosine, were then subjected to Edman degradation. As outlined under "Experimental Procedures," the combination of cold and radioactive sequencing, confirmed by the HPLC migration of corresponding chemically synthesized peptides, allowed the localization of the autophosphorylated tyrosines at positions 259, 336, and 424 in individual peptides of the cytoplasmic domain of the receptor (Fig. 9). The presence of phosphoserine and/or phosphothreonine in these phosphoamino acid analyses is due to contaminating phosphorylated peptides.
Effect of Mutagenesis of the Phosphorylated Tyrosines on the Kinase Activity-To assess the effect of tyrosine autophosphorylation of the type II receptor cytoplasmic domain on the kinase activity, we constructed mutants of the GST-IIK fusion protein, in which the three tyrosines that were autophosphorylated in vitro were individually replaced by phenylalanines. Furthermore, we made a mutant GST-IIK protein in which all three tyrosines were replaced by phenylalanine. All mutated fusion proteins, as well as the kinase-defective version of GST-IIK with the Lys to Arg replacement in the ATP binding site, were purified from E. coli lysates. Equal quantities of wild type and mutant GST-IIK proteins were subjected to anti-phosphotyrosine Western blot analysis (Fig. 10A). Consistent with our observations in Fig. 2A, wild type GST-IIK was autophosphorylated on tyrosine (Fig. 10A, lane 1), whereas the kinaseinactive point mutant was not (Fig. 10A, lane 6). The Tyr to Phe mutation at position 336 did not greatly affect tyrosine autophosphorylation (Fig. 10A, lane 3), whereas the Tyr mutations at positions 259 and 424 diminished the reactivity of the GST-IIK fusion protein with anti-phosphotyrosine (Fig. 10A, lanes 2  and 4). Finally, replacement of all three tyrosines by phenylalanines abolished the reactivity with anti-phosphotyrosine (Fig. 10A, lane 5), similarly to the kinase-defective point mutant of GST-IIK (Fig. 10A, lane 6).
This decreased tyrosine phosphorylation could in principle be due to a specific decrease in the number of tyrosine phosphorylation sites resulting from the mutation of the tyrosine while maintaining the kinase activity, or could result from a generally impaired kinase activity, which is largely on serine and threonine. To distinguish between these possibilities, we evaluated the kinase activity of the different GST-IIK proteins in autophosphorylation assays in the presence of [␥-32 P]ATP (Fig. 10B). The single tyrosine mutation at position 336 did not affect the kinase activity of the cytoplasmic domain. In contrast, the single mutations at positions 259 and 424 decreased the kinase activity, which is consistent with the result of the anti-phosphotyrosine Western blot, and the triple tyrosine mutation resulted in a greatly impaired kinase activity similar to the Lys to Arg mutation in the kinase-defective GST-IIK (Fig.  10B). Finally, to verify the effect of these mutations on the kinase activity of the type II receptor made by mammalian cells, we expressed the wild type and the point-mutated kinasedefective type II receptor as well as the mutated receptor with the three tyrosines replaced by phenylalanine in 293 cells. In vitro autophosphorylation assays of the immunoprecipitated receptors confirmed that the receptor with the triple tyrosine mutation did not have detectable kinase activity, similarly to the kinase-inactive point mutant (Fig. 10C, lanes 5 and 6). Whereas the mutation of Tyr 259 decreased receptor autophosphorylation (Fig. 10C, lane 2) consistent with the results using the E. coli-derived cytoplasmic domain, we did not observe a major decrease in kinase activity as a result of the Tyr 424 mutation (Fig. 10C, lane 4). The basis of the discrepancy between the latter result and the decreased activity resulting from the same mutation in the E. coli-derived fusion protein is unclear. Finally, we were unable to efficiently express the Tyr 336 -mutated receptor in 293 cells (Fig. 10C, lane 3). In summary, the results obtained using fusion proteins produced in E. coli suggest that autophosphorylation on Tyr 259 and Tyr 424 is functionally important for the kinase activity of the type II TGF-␤ receptor. The results using full-size receptors expressed in mammalian cells support this notion but are not totally in agreement, in part due to the discrepancy with the Tyr 424 mutation and due to the influence of these point mutations on the expression levels.
Role of Tyrosine Autophosphorylation in Type II Receptor Signaling-To evaluate the role of tyrosine autophosphoryla-  by tyrphostin (B). A, affinity-purified (His) 6 IIK was subjected to in vitro 32 9. Phosphoamino acid analysis and sequence of tryptic peptides. A tryptic digest of autophosphorylated type II receptor cytoplasmic domain was separated by HPLC. Phosphoamino acid analysis was carried out for each peak fraction, and three of them, shown here, were found to contain phosphotyrosine (Y). Edman degradation of these peak fractions revealed the phosphotyrosine containing tryptic peptide sequence shown. Each of these peptides contained a single tyrosine, which therefore corresponds to a phosphotyrosine at the indicated position. However, the phosphotyrosine containing peptide was usually contaminated with another peptide that comigrated on HPLC, which explains the presence of phosphoserine or phosphothreonine, even though some of the sequences shown did not contain serine or threonine. S, phosphoserine; T, phosphothreonine; Y, phosphotyrosine. tion of the type II TGF-␤ receptor in vivo, we first evaluated the effect of tyrphostin on TGF-␤ signaling in Mv1Lu cells that have endogenous type II and type I receptors. As a measure of the TGF-␤ signaling ability, we used the PAI-1 luciferase reporter assay in which the luciferase gene is expressed under control of the TGF-␤-inducible PAI-1 promoter (58). As is apparent from Fig. 11, the TGF-␤-induced expression from the PAI-1 reporter was strongly inhibited in a dose-dependent way by increasing concentrations of tyrphostin. This inhibition of signaling by the receptors is consistent with the inhibition of the type II receptor kinase by tyrphostin in vitro (Fig. 8).
We also evaluated the ability of the triple mutated type II receptor, in which the three tyrosine targets of autophosphorylation were replaced by phenylalanine, in comparison with the wild type type II receptor. Thus, expression plasmids for both receptors were transiently transfected at two concentrations (0.5 or 5 g) in DR26 mutant Mv1Lu cells, which lack type II receptors. To measure TGF-␤ responsiveness in these transient transfection assays, we used both the PAI-1 luciferase assay which scores TGF-␤-induced gene expression (58), and the cyclin A reporter assay in which decreased luciferase expression from the cyclin A promoter correlates with growth inhibition (59). As shown in Fig. 12, transfection of wild type type II receptor restored TGF-␤ responsiveness in Mv1Lu-DR26 cells in both assays, and the response was somewhat higher at 5 g of transfected plasmid than at 0.5 g. Surprisingly, the triple Tyr mutant of the type II receptor also allowed TGF-␤ signaling as assessed in both assays, although the PAI-1 response at 5 g of transfected plasmid was reproducibly lower than for wild type receptor, and the cyclin A response of the triple mutant was lower at 0.5 g of transfected plasmid. DISCUSSION The receptors for TGF-␤ and related proteins constitute a distinct family of transmembrane proteins with predicted serine/threonine kinase activity (reviewed in Refs. 12 and 13). The experiments described here demonstrate that the type II TGF-␤ receptor phosphorylates not only on serine and threonine but also on tyrosine residues, and should therefore be considered as a dual specificity kinase. The evidence for this conclusion is obtained from a variety of experiments. The kinase domain expressed as a fusion protein in E. coli or in insect cells autophosphorylates not only on serine and threonine but also on tyrosine. Because E. coli does not have detectable intrinsic tyrosine kinase activity and since an inactivated point mutant of the kinase did not react with anti-phosphotyrosine antibody, we conclude that the tyrosine phosphorylation resulted from autophosphorylation of the receptor kinase. In addition, the cytoplasmic domain phosphorylates exogenous substrates such as histone 2B on serine, threonine, and tyrosine. We also showed that the receptor immunoprecipitated from transfected mammalian cells phosphorylates on serine, threonine, and tyrosine residues, similarly to the cytoplasmic domain expressed in E. coli or insect cells. Furthermore, phosphoamino acid analysis of the in vivo 32 P-labeled, transfected receptor revealed phosphoserine and a small amount of phosphotyrosine, thus supporting the significance of this dual kinase specificity in vivo. This phosphorylation did not require the presence of ligand (data not shown), which is in agreement with the constitutive activity and autophosphorylation of the type II TGF-␤ receptor (21,22). Finally, the kinase activity of the receptor can be inhibited by tyrphostin, a competitive inhibitor of tyrosine phosphorylation, suggesting that this kinase, unlike standard serine/threonine kinases, can accommodate a tyrosine residue as substrate at its active site.
The amino acid sequences of the kinases of the TGF-␤ receptor family predict, based on specific motifs (18,19,50,61), a kinase specificity for serine and threonine. Accordingly, previous results using recombinant receptor proteins have demonstrated that the type II receptors for TGF-␤ or activin as well as the T␤RI/ALK-5/R4 type I receptor autophosphorylate on serine and threonine (19 -22, 49). Whereas phosphothreonine is the major phosphorylated amino acid following in vitro assays (19), the type II TGF-␤ receptor autophosphorylates in vivo primarily on serine (21,22). Interestingly, the endogenous activin receptor purified from cells has been reported to have dual kinase specificity (51), but this was not confirmed in a study using the cloned type II activin receptor (49). The results described here demonstrate that the type II TGF-␤ receptor is indeed a dual specificity kinase. This discrepancy between the current and previous results is probably due to the low levels of phosphotyrosine compared with phosphoserine and phosphothreonine, combined with the lability of phosphotyrosine at high temperatures during the hydrolysis reaction of the phosphorylated cytoplasmic domain. Accordingly, we performed the hydrolyses for the phosphoamino acid analyses at 100°C instead of the more standard 110°C. Furthermore, the ability of the receptor kinase to autophosphorylate on tyrosine in vitro may be considerably attenuated because the tyrosines are already phosphorylated when expressed in E. coli, insect cells, or mammalian cells.
A number of reports have suggested or demonstrated the dual specificity of various kinases (Refs. 62 and 63; reviewed in Ref. 64). In general, their levels of tyrosine phosphorylation are low compared with serine and threonine, and dual specificity is usually only demonstrated in autophosphorylation reactions. Indeed, an in vivo function for dual kinase specificity has to date only been demonstrated for the MAP kinase kinases (3,4). Based on their primary structure, dual specificity kinases appear to be indistinguishable from serine/threonine kinases and map throughout the kinase family tree (50). Accordingly, the type II and type I receptors for TGF-␤ superfamily members have been classified as serine/threonine kinase receptors based on their sequence and kinase activity (19,20,49). However, they also show similarity in their kinase domains with tyrosine kinases (50). For example, a CW motif in subdomain XI that is highly conserved in tyrosine kinases is also conserved in the type II and type I receptors. The current report, together with the sequence conservation of the kinase domains, suggests that perhaps all these receptors are dual specificity kinases with the ability to autophosphorylate on tyrosine(s).
Sequencing of tryptic phosphopeptides led to the localization of three autophosphorylated tyrosines in the type II TGF-␤ receptor: Tyr 259 in kinase subdomain I, Tyr 336 in subdomain V, and Tyr 424 in subdomain VIII. Tyr 259 is conserved among the type II TGF-␤ receptors from different species, but not among other type II receptors (data not shown). Replacement of this tyrosine by phenylalanine decreased the kinase activity of the cytoplasmic domain expressed in E. coli, but not in mammalian cells. The basis for this reproducible discrepancy is unclear. This tyrosine is in the ATP-binding site of the kinase, and phosphorylation of residues in this region is known to be an important factor in the inhibition of kinase activity of cyclindependent kinases (65,66). Tyr 336 is well conserved among not only the different type II but also the type I receptors. In addition to tyrosine, phenylalanine is also found in the corresponding position in other type II receptors. Accordingly, replacement of Tyr 336 by phenylalanine did not affect the kinase activity of the type II TGF-␤ receptor. Finally, Tyr 424 is absolutely conserved in all type II and type I receptors characterized so far. Its replacement by phenylalanine strongly decreased the kinase activity of the type II receptor. This tyrosine is located two amino acids upstream from the signature sequence APE in kinase subdomain VIII. The sequence between subdomain VII and the APE sequence represents a target for regulatory phosphorylations in several kinases and can function as an activation loop (67). For example, phosphorylation of Thr 197 in this loop of protein kinase A (68) and Thr 183 and Tyr 185 in the corresponding sequence in MAP kinase (69) contribute to activation of these kinases. Based on structural information, phosphorylation of this loop alters the conformation and increases the activity of the kinase (70,71). Similarly, the dual specificity kinase glycogen synthetase kinase 3 undergoes tyrosine autophosphorylation on the corresponding residue in subdomain VIII and this phosphorylation enhances its kinase activity (72). Thus, for both the MAP kinases and for glycogen synthetase kinase, tyrosine phosphorylation upstream from the APE sequence enhances the kinase activity and plays an autoregulatory role. Likewise, the conserved Tyr 424 is located closely upstream from the APE motif in subdomain VIII of the type II TGF-␤ receptor, and its replacement by phenylalanine strongly inhibits the kinase activity. Taken together, the autophosphorylation of the tyrosines in the kinase domain of the type II TGF-␤ receptor, and possibly all serine/threonine ki- FIG. 12. Signaling activities of wild type and mutant type II receptor with three tyrosines replaced by phenylalanine. Mv1Lu-DR26 cells lacking endogenous functional type II receptor were transfected with the reporter plasmids p800Luc or pCAL2, and pRK␤Gal, and treated with the indicated concentrations of TGF-␤. The luciferase activities were determined as in Fig. 11. A, PAI-1-luciferase expression assay. B, cyclin A-luciferase expression assay. nase receptors, illustrates the dual specificity of the kinase activity and suggests an autoregulatory role similar to that seen in various other kinases. The putative regulatory role of these tyrosines may explain the strong inhibition of replacement of all three tyrosines on the kinase activity of this receptor.
Whereas the ability of the type II receptor to autophosphorylate on tyrosines is clearly illustrated and the tyrosine to phenylalanine mutations inhibit the kinase activity in vitro, the role of the tyrosine autophosphorylation is less unambiguous. Clearly, the inhibition of the signaling activity by tyrphostin is consistent with the in vitro inhibitory effect on the kinase activity of the type II receptor. However, the triple tyrosine mutant of the type II receptor has signaling activity in reporter assays, and this activity is only slightly lower than that of the wild type receptor. This result thus indicates that the triple mutated type II receptor is biologically active. However, these data have to be interpreted with caution, since a primary role of the type II receptor is to phosphorylate and activate the type I receptor, which has an effector role in signaling. Therefore, a low level of activity of the type II receptor may be sufficient to allow signaling by the heteromeric receptor complex. In addition, we tested only two responses, and other responses may show a greater sensitivity to the impaired type II receptor kinase activity.
By analogy with tyrosine kinase receptors (2), the tyrosine autophosphorylation also raises the possibility that, in addition to an autoregulatory role of the kinase activity, the sites of tyrosine phosphorylation may also act as docking sites for signaling proteins. However, this is unlikely since the three tyrosine phosphorylation sites are located in functional kinase domains and not in flanking domains or inserts as is the case for the known tyrosine kinase receptor docking sites. Another possibility is that the type II and/or type I receptor phosphorylate target proteins on tyrosine. However, TRIP-1, which can associate with the type II receptor in vivo, is only phosphorylated on serine and threonine (60). Similarly, Smad2 and Smad3, downstream mediators of TGF-␤ signaling, which associate with the heteromeric receptor complex, are also serineand threonine-phosphorylated (73,74). Future studies will have to determine whether any substrate proteins are tyrosinephosphorylated in vivo by the type II TGF-␤ receptor or by related receptors and will hopefully reveal the biological significance of the tyrosine autophosphorylation in the regulation of receptor activity in vivo.