Smad2 Phosphorylation by Type I Receptor

Transforming growth factor-β (TGFβ) is a potent regulator of cell proliferation, differentiation, motility, and apoptosis. TGFβ binds to and activates serine/threonine kinase receptors that phosphorylate Smad2 and Smad3 intracellular signal transducers at two C-terminal serine residues. Here we show that substitutions of Arg-462 and Cys-463 residues, which are in proximity of the C-terminal serine residues, inhibited TGFβ type I receptor-dependent phosphorylation of the C-terminal Smad2 peptides and full-length GST-Smad2 proteins in vitro. In vivo, mutation of Arg-462 and Cys-463 inhibited TGFβ1-stimulated phosphorylation of the C-terminal serine residues in Smad2. Moreover, Smad2 with mutated Arg-462 and Cys-463 was less efficient in activation of the Smad2-responsive activin-responsive element-containing luciferase reporter ARE-luc, as compared with the wild-type protein. Thus, Arg-462 and Cys-463, which are in proximity of the C-terminal serine residues, contribute to recognition and phosphorylation of the C terminus of Smad2 by type I TGFβ receptor.

Transforming growth factor-␤ (TGF␤) 1 is a member of a large family of polypeptide growth factors that includes more than 40 members and is divided into three groups, e.g. TGF␤ isoforms, bone morphogenetic proteins (BMP), and activins. These subfamilies are characterized by similarities in structure and functional activities (1)(2)(3). TGF␤, BMPs, and activins bind to serine/threonine kinase receptors to initiate activation of the receptor kinase. Activated receptors phosphorylate receptorregulated Smad proteins that were found to have crucial importance for signaling. Five receptor-regulated Smads have been described in mammals, e.g. Smad1, Smad2, Smad3, Smad5, and Smad8 (1)(2)(3). Smad2 and Smad3 are activated by TGF␤ and activin receptors, and Smad1, Smad5, and Smad8 are activated by BMP receptors. The triggering event in Smad activation is the type I receptor-dependent sequential phosphorylation of the two C-terminal serine residues in Smads (4 -7).
The recognition of Smads by type I serine/threonine kinase receptors defines specificity in intracellular signaling. A number of reports have pointed to the importance of inter-faces between the L45 loop and the GS domain of type I receptors and the L3 loop, the adjacent basic surface, and the ␣-H1 helix in the MH2 domain of receptor-regulated Smads as determinants of specificity in recognition of Smads by receptor kinases (8 -13).
Phosphoryl groups at Ser-465 and Ser-467 of the C terminus of Smad2 were found to mediate the interaction with Smad4 (7). Crystallography studies showed that upon phosphorylation of Ser-465 and Ser-467, the C terminus of Smad2 acquires certain structural features, allowing interaction with the MH2 domain of another Smad molecule (14 -16). Phosphorylation of the two C-terminal serine residues is also important for relief of the inhibitory intramolecular interaction between the MH1 and the MH2 domains, leading to activation of Smads (6,7,17). Therefore, phosphorylation of the C terminus of receptor-activated Smads is crucial for initiation of TGF␤ signaling.
For a number of kinases, specificity of phosphorylation is determined by sequences adjacent to the site of phosphorylation (18). Consensus sequences have been established for more than 50 kinases, suggesting that substrate recognition is mediated by a complementation between amino acid residues in a substrate and in a kinase substrate-binding region in or immediate to the catalytic cleft (19). Previously we showed that the C-terminal Smad2 peptide specifically inhibited T␤R-I kinase, but not other receptor kinases (20). This suggests that the substrate recognition site of T␤R-I kinase has preferences for the C-terminal peptide of Smad2. Here we have shown that Arg-462 and Cys-463 in the C terminus of Smad2 are important for specific recognition and phosphorylation of the C-terminal serine residues of Smad2 by T␤R-I.

EXPERIMENTAL PROCEDURES
Materials, Reagents, Constructs-The constructs of constitutively active GST-T␤R-I, GST-BMPR-II, GST-Smad2, and FLAG-Smad2 were described earlier (20 -22). Mutants of GST-Smad2 and FLAG-Smad2 were generated by site-directed mutagenesis using a QuikChange kit (Stratagene). Mutations were confirmed by sequencing of the constructs, and expression of proteins was evaluated by SDS-PAGE and by immunoblotting assay. Mv1Lu, NIH3T3, and COS1 cells were obtained from ATCC (LGC, Teddington), and Smad2Ϫ/Ϫ mouse embryonic fibroblasts (MEFs) were obtained from Anita Roberts and Ester Piek. All cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum.
In Vitro Phosphorylation Assay-For phosphorylation assay, peptides were diluted in the reaction mixture (20 mM HEPES (pH 7.4), 10 mM MgCl 2 , 2 mM MnCl 2 , 1 mM dithiothreitol, 5 M ATP plus 0.5 Ci of [␥ 32 P]ATP (Redivue; Amersham Biosciences)). The peptides used in this study were synthesized using the Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry as described (20). Three lysine residues were added to the N terminus of each peptide to allow peptide binding to the P81 phosphocellulose paper or to the plastic. Phosphorylation reaction was initiated by addition of constitutively active GST-T␤R-I adsorbed on glutathione-Sepharose beads. After 20 min of incubation at 22°C, reaction mixtures were centrifuged, and the supernatants were spotted onto Whatman P81 phosphocellulose paper. Filters were washed in three changes of 0.75% phosphoric acid, rinsed in acetone, dried, and * This work was supported in part by grants from the Swedish Research Council, the Swedish Cancer Society, the Royal Swedish Academy of Sciences, the INSERM, and by support from Merck KGaA (to S. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
counted in a scintillation counter. Every experimental condition was performed in triplicate, and experiments were repeated three to four times. In some experiments, phosphorylated peptides were separated by a Tricine-polyacrylamide gel electrophoresis (23). Gels were dried and exposed by phosphorimaging. Phosphorylation of GST-Smad2 proteins in vitro was performed as described earlier (20). After SDS-PAGE, the proteins were transferred onto the nitrocellulose membrane, and phosphorylation of the C terminus was analyzed by two-dimensional phosphopeptide mapping.
Detection of Smad2 C-terminal Phosphorylation-Immunoblotting with pS2 antibodies that specifically recognize phosphorylated C-terminal Ser-465 and Ser-467 in Smad2 was performed as described earlier (20).
Luciferase Reporter Assays-The cells were transiently transfected using LipofectAMINE (Invitrogen) according to the manufacturer's recommendations or using a 25-kDa branched polyethyleneimine (Sigma) as described (24). Activin-responsive element (ARE)-luc reporter assays were performed as previously described (21). In all reporter assays, the ␤-galactosidase reporter plasmid pCMV-LacZ was used for normalization of efficiency of transfection.

T␤R-I Phosphorylates Specifically the C-terminal Peptide of
Smad2-Amino acid residues adjacent to sites of phosphorylation are for several kinases crucial for substrate recognition, e.g. Pro is required at the ϩ1 position for phosphorylation by Erk (extracellular signal-regulated kinase) 1/2 (19). To explore whether phosphorylation of Smads by T␤R-I kinase is dependent on surrounding amino acids, we performed in vitro kinase assays with peptides corresponding to the C-terminal tails of Smad2 and Smad5 as substrates (Fig. 1A). The use of peptides allowed us to exclude an influence of the N-terminal sequence, e.g. of the MH2 domain. Crystallography studies could not identify a defined structure for the C terminus of Smad, suggesting that the C terminus is not an intrinsic part of the MH2 domain (17).
We found that T␤R-I kinase phosphorylated the C-terminal A, the C-terminal sequences of peptides that were used for phosphorylation studies are shown. Complete C-terminal sequences are shown for wild-type Smad2 and Smad5; for other peptides only mutated amino acid residues are shown. B, upper panel, wild-type (Sm2S, Sm5S) and phosphorylationdeficient (Sm2A, Sm5A) peptides were phosphorylated in vitro by T␤R-I and subjected to Tricine-PAGE. Incorporated radioactivity was detected by exposure of gels by phosphorimaging. C, wild-type Smad2 peptide was not phosphorylated by serine/threonine kinase receptors BMPR-II, BMPR-IB, and T␤R-II, whereas T␤R-I phosphorylated the peptide. Migration positions of phosphorylated peptides (B and C, upper panels, 32 P) and total peptides detected after staining of gels with Coomassie Brilliant Blue (B and C, lower panels, CBB) are shown by arrows. D, wild-type Smad2 (Sm2S) and Smad5 (Sm5S) peptides at indicated concentrations were subjected to phosphorylation by T␤R-I in vitro; peptides concentrations were 31, 62, 125, 250, 500, and 1000 M. Incorporation of radioactive phosphate was evaluated and values of radioactivity plotted against concentration of peptides. E, peptides at concentration 0.5 mM, wild-type, and with substitutions of amino acids as shown in panel A were subjected to phosphorylation by T␤R-I in vitro. Incorporated radioactive phosphate was measured. Peptides are indicated. Experiments representative of three (B-D) and four (E) performed are shown.
Smad2 peptide (Sm2S), but not the Smad2 peptide with the five C-terminal serine residues substituted to alanines (Sm2A). Neither was Smad5 C-terminal peptide with mutated serines (Sm5A) phosphorylated by T␤R-I, and wild-type Smad5 peptide (Sm5S) showed very weak signal (Fig. 1B). A similar specificity of T␤R-I-dependent phosphorylation was observed with the same peptides coupled to plastic surfaces (data not shown). Furthermore, phosphorylation of the Smad2 peptide was specific for T␤R-I kinase because other serine/threonine kinases receptors, e.g. bone morphogenetic protein receptor type II (BMPR-II), bone morphogenetic receptor type I (BMPR-IB), and type II TGF␤ receptor did not phosphorylate the peptide (Fig. 1C). Preservation of the specificity of T␤R-I kinase toward a substrate was also shown previously (20). In vitro phosphorylation assay with various concentrations of the Sm2S and Sm5S peptides (31, 62, 125, 250, 500, and 1000 M) showed that the Sm2S peptide was strongly phosphorylated at a concentration of 31 M with a maximal level achieved at 500 M (Fig. 1D). Weak, but detectable, phosphorylation of Smad5 peptide was observed only at 1000-M concentration of the peptide. This suggests that under non-physiological conditions the C-terminal tail of BMP Smad5 can be phosphorylated by T␤R-I.
To explore which residue(s) in the C-terminal peptide of Smad2 defines specificity of Smad2 recognition by the T␤R-I kinase, we generated peptides with substitutions of Ser-460, Arg-462, and Cys-463 to Ala, Ile and Ala, respectively (Fig.  1A). We found that the strongest decrease of phosphorylation was observed when both Arg-462 and Cys-463 were mutated (Sm2R462I,C463A peptide; Fig. 1E). Single mutations of Arg-462 and Cys-463 decreased phosphorylation of peptides as well. We did not see changes in peptide phosphorylation when Ser-460 or when Ser-458, Ser-460, and Ser-464 were mutated (Fig. 1E). Thus, Arg-462 and Cys-463 are of importance for phosphorylation of the C-terminal peptide of Smad2 by the T␤R-I kinase.
Mutation of Arg-462 and Cys-463 Prevents Phosphorylation of the C-terminal Serine Residues in GST-Smad2 by T␤R-I-We further explored the importance of the amino acid residues in the region from Ser-460 to Cys-463 for the specific phospho- rylation of Smad2 by T␤R-I. We mutated these four residues one by one or in pairs in a Smad2-fused to GST (Fig. 2A). The GST-Smad2 constructs were expressed in bacteria (Fig. 2B), and purified proteins were phosphorylated by T␤R-I kinase in vitro. Because we were interested in analysis of phosphorylation of the C-terminal peptide, we performed two-dimensional phosphopeptide mapping. A distinct migration pattern of the C-terminal tryptic peptide allowed us to monitor phosphorylation of this particular peptide. This provides an advantage to evaluation of a total phosphorylation, which does not monitor phosphorylation of selected sites but rather of the whole molecule.
We observed a phosphopeptide with migration position corresponding to the wild-type peptide (Fig. 2C) in phosphopeptide maps of double mutants GST-Smad2S460A,V461N and GST-Smad2V461N,R462P (Fig. 2, D and E). The observed slight deviations in migrations of the peptides is because of the substitutions made and in the case of GST-Smad2V461N,R462P mutant because of abrogation of trypsin cleavage C-terminal to Arg-462. A double mutation of Arg-462 and Cys-463 abrogated GST-Smad2 phosphorylation, as we could not observe any phosphopeptide in areas of the expected migration nor did any other additional spots appear in the two-dimensional phosphopeptide maps (Fig. 2F). The variation in efficiency of phosphorylation of the full-length double mutant GST-Smad2R462P,C463I (Fig. 2F), as compared with the Sm2R462I,C463A peptide (Fig. 1E) may be explained by the presence of the upstream sequence in GST-Smad2, e.g. the MH1 and MH2 domains. GST-Smad2 constructs with Arg-462 or Cys-463 mutated separately showed low, but detectable, phosphorylation of the C-terminal peptide (Fig. 2, G and H).
Quantification of the C-terminal peptide phosphorylation, expressed as the ratio of the C-terminal peptide phosphorylation to the total phosphorylation of the respective Smad2 construct, showed arbitrary values of 0.39 for the C463I mutant, 0.56 for the R462P mutant, and 0.71 for the wild-type protein. These values, which reflect the phosphorylation efficiency of the Cterminal peptides in whole proteins, correlate with the stronger inhibitory effect of the mutation of the cysteine residue, as compared with the mutation of the arginine residue, on the phosphorylation of synthesized peptides (Fig. 1E). Thus, our data suggest that both Arg-462 and Cys-463 contribute to the T␤R-I-dependent phosphorylation of GST-Smad2 at the C terminus.
Substitution of Arg-462 and Cys-463 Inhibits Phosphorylation of the C Terminus of Smad2 by T␤R-I in Vivo-To evaluate the involvement of Arg-462 and Cys-463 in T␤R-I-dependent phosphorylation of Smad2 in vivo, we expressed FLAG-tagged Smad2, wild-type, and mutants in various cell lines. Transfected cells were treated or not with TGF␤1, and phosphorylation of the two C-terminal serine residues was monitored by immunoblotting with pS2 antibodies that recognize phosphorylated C-terminal Ser-465 and Ser-467 (Fig. 3). As a negative control, we used FLAG-Smad2 with the two C-terminal serines substituted by alanine residues (FLAG-Smad2SA). We found that none of the mutations affected the stability of proteins, whether proteins were expressed at low or at high levels (Fig. 3).
We observed that substitution of both Arg-462 and Cys-463 to either Pro-462 and Ile-463 (FLAG-Smad2R462P,C463I; RC/PI mutant) or to Ile-462 and Ala-463 (FLAG-Smad2R462I,C463A; RC/IA mutant) inhibited TGF␤1-dependent phosphorylation at the C-terminal serine residues (Fig. 3,  B-E). In Mv1Lu epithelial cells and in NIH3T3 mouse fibroblasts we did not observe any detectable C-terminal phosphorylation of RC/PI or RC/IA mutants of FLAG-Smad2; neither was there any phosphorylation of the FLAG-Smad2S465A,S467A (SS/AA) mutant as expected (Fig. 3, D and  E). We could observe some phosphorylation of the FLAG-Smad2 RC/PI and RC/IA mutants upon expression in Smad2Ϫ/Ϫ mouse embryonic fibroblasts. However, quantification of the phosphorylation showed significant reduction in efficiency of phosphorylation (Fig. 3, B and C). Moreover, analysis of time-dependence of Smad2 phosphorylation upon TGF␤1 treatment showed that the RC/PI mutant was phosphorylated at a slower rate than the wild-type FLAG-Smad2 (Fig.  4). The rate of Smad phosphorylation may be affected by various interacting proteins and could therefore be cell type-de- pendent. Our consistent results using three different cell lines of epithelial and mesenchymal origin suggest that Arg-462 and Cys-463 are of general importance for the specificity of phosphorylation of Smad2 by T␤R-I.

Substitution of Arg-462 and Cys-463 in Smad2
Results in Inhibition of TGF␤1/Smad2-dependent Transcriptional Activation-To explore the impact of the observed role of Arg-462 and Cys-463 in Smad2 phosphorylation by T␤R-I on the functional activity of Smad2, we performed luciferase reporter assay with the Smad2-responsive ARE-luc reporter. This reporter contains an ARE that mediates TGF␤/Smad2-dependent transcriptional activation (25). We found that mutation of Arg-462 and Cys-463 inhibited the transcriptional activity of Smad2 expressed in Smad2Ϫ/Ϫ mouse embryonic fibroblasts as compared with the wild-type Smad2 (Fig. 5A). Residual transcriptional activity of FLAG-Smad2 RC/PI and RC/IA mutants correlated with the levels of their TGF␤1-dependent phosphorylation (Fig. 3, B  and C). These two mutants did not show ligand-dependent transcriptional activity in Mv1Lu epithelial cells (Fig. 5B). Thus, the importance of Arg-462 and Cys-463 for recognition and phosphorylation of Smad2 by T␤R-I kinase is also reflected in the inhibition of transcriptional activation of Smad2 upon TGF␤1 treatment. DISCUSSION Here we have provided evidence that Arg-462 and Cys-463 in the C terminus of Smad2 are important for recognition and phosphorylation of the serine residues 465 and 467 by T␤R-I. Our findings expand the current view on determinants of specificity in TGF␤ signaling. Previously, it has been shown that the physical interaction between T␤R-I and Smad2 is mediated by the L45 loop and GS region of the receptor and by the L3 loop and the adjacent basic surface of Smad protein (8 -13). For Smad1, a member of the BMP Smad family, the importance of the ␣-H1 helix in the MH2 domain for interaction with ALK1 and ALK2 receptors has also been shown, in addition to the L3 loop (11). The role of the C-terminal sequences of receptorregulated Smads in determination of signaling specificity has not been addressed thoroughly. The fact that mutation of the phosphorylatable serine residues increases the interaction with T␤R-I suggests that the C terminus, or at least C-terminal phosphorylation, affects the interaction between receptor and Smad protein (4 -7).
The reported inhibition of T␤R-I by the C-terminal peptide specific for Smad2 and the inability of the Smad5-specific Cterminal peptide to inhibit the T␤R-I kinase suggest that the kinase may be able to selectively recognize the substrate even in the absence of the MH2 domain (20). Several kinases are dependent on specific sequences adjacent to the site of phosphorylation (19). Consensus sequences for at least 50 kinases have been identified (18), and residues at ϩ1 or Ϫ3 and Ϫ4 have been found to be of particular importance for recognition of substrates by kinases (19). An early attempt to identify a  (28), using Protein Data Bank ID 1IAS for T␤R-I and 1KHX for Smad2. The C-terminal peptide was modeled in the substrate-binding site of the kinase, and the phosphoserine residues were substituted by serine residues. Various structural features of the T␤R-I kinase and the C-terminal peptide are indicated by arrows. Proposed localizations of Arg-462, Cys-463, and Ser-467 of the C-terminal peptide and the activation segment of the T␤R-I kinase are indicated. The C-terminal peptide is in blue, the activation segment is green, the catalytic segment is magenta, and the phosphate-binding segment is yellow. consensus phosphorylation sequence for T␤R-I using a peptide library showed that the presence of basic charges at the Ϫ4 and Ϫ3 positions would be preferable for phosphorylation by T␤R-I (26). The presence of arginine residue at the Ϫ5 and Ϫ3 positions, compared with phosphorylatable serines in the C terminus of Smad2, is in agreement with this suggestion.
Here we have shown that T␤R-I kinase can specifically recognize the Smad2 C-terminal sequence and that Arg-462 and Cys-463 are crucial determinants for the specificity. The importance of Arg-462 and Cys-463 was observed in vitro for peptides and in the context of the whole Smad2 molecule as well as in vivo for expressed Smad2 proteins (Figs. 1-4). Some difference in the phosphorylation efficiency of peptides, as compared with full-length proteins, may be because of the upstream sequences in proteins, e.g. the MH1 and MH2 domains. However, the similarity of results obtained with peptides that do not have any of the MH2 domain structural features, and with full-length proteins, supports the notion that Arg-462 and Cys-463 are important determinants of recognition and phosphorylation of the Smad2 C terminus.
Positioning of the C-terminal peptide in the substrate-binding region of T␤R-I showed that Arg-462 and Cys-463 of Smad2 may interact with residues in the activation segment of the kinase (Fig. 6) (27). The activation segments of TGF␤and BMP-specific receptors differ, with strong similarities of the sequences within the groups of receptor. Thus, the presence of aspartic acid residues in the activation segments of T␤R-I and ActR-IB, e.g. Asp-359 and Asp-363, may provide negatively charged surfaces that interact with positively charged C-terminal residues of Smad2 (Fig. 6). Significant similarity in the activation segments has also been observed for ALK7, which is known to phosphorylate Smad2 and Smad3. The activation segments of BMPR-IA and BMPR-IB receptor kinases have sequences that differ from the TGF␤ receptors. Another group, defined on the basis of the sequences of activation segments, consists of ActR-IA and ALK1 receptors. Thus, it is tempting to suggest that the activation segments of receptor kinases provide a surface for specific recognition of the C-terminal peptide of Smad proteins. However, for the conclusive answer, the structure of the T␤R-I kinase with bound substrate has to be determined.
Our findings add a new aspect to our understanding of the determination of specificity in TGF␤ signaling. Whereas the interfaces between the L3 loop, basic surface, and ␣-H1 structures of Smad proteins and the L45 loop and GS domain of the type I receptors are involved in docking of Smads to receptors, specific recognition of the C-terminal phosphorylation sites adds to the specificity of substrate recognition. Importance of adjacent amino acid residues for specific recognition and phosphorylation of substrates is well established for many kinases. The data presented here extend this notion to the type I serine/ threonine kinases receptors and their substrates.