Introduction
Phosphorylation of receptors and/or intracellular signalling molecules is the driving force in most signal transduction pathways. In addition, the actions of these signalling molecules are finely tuned by other post-translational modifications. Of these, the role of ubiquitination has been extensively studied in various signalling pathways. Another type of post-translational modification, sumoylation, has recently been shown to be important for certain signalling pathways1. Sumoylation is mainly observed in nuclear and perinuclear proteins but has so far been shown only for a few types of cell-surface proteins, including ion channel proteins and glutamate and kinate receptors. On page 654 of this issue, Kang et al.2 report that TGF-
type I receptor (TRI) is sumoylated when the receptor is phosphorylated, causing enhancement of TGF- signalling. This is the first growth factor receptor whose function has been shown to be modified by sumoylation.
TGF- binds to two Ser-Thr kinase receptors: TRI (also known as ALK-5) and type II (TRII) receptors. When a ligand binds, TRII transphosphorylates TRI, leading to activation of intracellular signalling pathways, including the canonical Smad pathway and other non-Smad pathways3 (Fig. 1). Phosphorylation of TRI occurs on the Gly–Ser rich region (GS region) located at the amino-terminal boundary of the Ser-Thr kinase domain, as well as on other Ser and Thr residues. Activated type I receptor kinase binds to the receptor-regulated Smads (R-Smads) Smad2 and Smad3, inducing phosphorylation of their carboxy-terminal SSXS motif. R-Smads then interact with Smad4, translocate into the nucleus and regulate transcription of target genes. Interaction between TRI kinase and R-Smads requires the L45 loop and GS region of TRI to physically bind to the L3 loop and adjacent 
-helix1 sequence in the C-terminal MH2 domain of R-Smads, causing phosphorylation of the SSXS motif. Thus, the phosphorylation cascade from the receptors to Smads is crucial for TGF- signalling.
Figure 1: Proposed model of R-Smads phosphorylation by T
RI.
![Figure 1 : Proposed model of R-Smads phosphorylation by T|[beta]|RI.](/ncb/journal/v10/n6/thumbs/ncb0608-635-f1.jpg)
TGF- induces formation of a heterotetrameric receptor complex composed of TRII and TRI. TRII kinase phosphorylates TRI and conjugation of SUMO-1 occurs on TRI. R-Smads are then phosphorylated by TRI, form complexes with Smad4 and translocate into the nucleus where they regulate transcription of target genes. Smad signalling pathways, and possibly non-Smad pathways, are enhanced by sumoylation of TRI.
Covalent attachment of ubiquitin to Lys residues, is catalysed by three types of enzyme: E1 activating enzyme, E2 conjugating enzyme and specific E3 ubiquitin ligases. This controls the turnover of many signalling proteins. For example, ubiquitination of TGF- receptors is induced by E3 ubiquitin ligases (for example, Smurf1 and Smurf2) and regulates the amplitude of TGF- signalling through degradation of the receptors by the ubiquitin proteasome system4, 5. An inhibitory Smad (Smad7) interacts with the Smurfs and recruits them to TRI for ubiquitin-dependent degradation of the receptor complexes. Ubiquitination also occurs on Smad proteins by the action of Smurfs and other E3 ligases6.
A number of ubiquitin-like proteins, including the small ubiquitin-like modifier (SUMO), use a system similar to ubiquitination. SUMO is activated by the E1 activating enzyme consisting of the Aos1/Uba2 heterodimer and transferred to the E2 enzyme Ubc9, followed by ligation to the substrates1. E3 ligases for sumoylation have broad specificity, compared with those for ubiquitination, and several enzymes, including the PIAS family proteins, function as E3 ligases for sumoylation. This does not necessarily promote degradation of target proteins, and the functional consequences of sumoylation vary, depending on target proteins. Smads are also post-translationally modified by sumoylation; in the case of Smad4, sumoylation protects it from ubiquitin-dependent degradation6.
In mammals, seven type I receptors transduce signals for ligands of the TGF- family. Of these, TRI/ALK-5, ActRIB/ALK-4 and ALK-7 are structurally related to each other. Among members of the TGF- family, TGF-1, 2 and 3 bind to TRI, whereas activins bind to ALK-4, and nodal binds to ALK-4 and ALK-7 (ref. 7). Although TRI and ALK-4 both phosphorylate Smad2 and Smad3, their biological activities seem to be different. For example, TGF- has potent growth inhibitory activity on epithelial cells, whereas activins do so less potently8. Interestingly, Kang et al.2 report that sumoylation occurs only on TRI but not on ALK-4 (nor on ALK-7 as it lacks the specific sumoylation residue), enhancing activation of the TRI kinase and subsequent phosphorylation of Smad proteins. These findings suggest that sumoylation of the kinase domain induced by ligand stimulation is unique to TRI, which may be responsible for specific effects of TGF-, for example potent growth inhibition and stimulation of cancer metastasis.
How does sumoylation amplify the signalling activity of TRI? Kang et al.2 found that the kinase activities of both TRII and TRI are required for sumoylation of TRI. Sumoylation may occur on multiple Lys residues, but Lys 389 is the primary sumoylation site. As Lys 389 is located on the surface of TRI kinase and has the same orientation as the GS region and the L45 loop, sumoylation may increase the affinity of TRI kinase for R-Smads. Smad7 also binds tightly to the TRI kinase domain and competes with R-Smads for interaction with TRI. Furthermore, Smad7 recruits Smurfs and the GADD34–PP1 (protein phosphatase 1) complex to the receptors4, 5, 9, inducing degradation and dephosphorylation of the receptors, respectively. An intriguing possibility is that when TRI is sumoylated, R-Smads may be able to interact with TRI with higher affinity than Smad7. Phosphorylation of R-Smads by TRI is facilitated by some intracellular proteins, including Dab2 (ref. 10). Thus, another possibility is that binding of these Smad co-activators may be facilitated by sumoylation of TRI.
A mis-sense mutation of human TGFBR1 (encoding TRI protein) mutating Ser 387 to Tyr has been found in breast and head-and-neck cancers11, 12. Ser 385 in rat TRI (corresponding to Ser 387 in humans) localizes close to the Lys 389 sumoylation site. Although several loss-of-function mutations in TRI have been reported, functional importance of these mutations has not been fully elucidated. Interestingly, the authors found that the S385Y mutation does not significantly affect kinase activity of TRI, but decreases its sumoylation. Thus, transcription by the S385Y mutant was attenuated, although less potently than the K389R mutant. Accordingly, Ras-transformed mouse embryonic fibroblasts (MEFs) expressing the S385Y mutant developed fewer metastases than those expressing wild-type TRI in an animal model2. This seems to contrast with the findings that the S387Y mutation in human TGFBR1 was observed in metastatic tumours, but this may be due to bidirectional effects of TGF- signalling on progression of cancer13.
The discovery of TRI sumoylation may provide a better understanding of the unique action of the TGF- signalling pathway. However, there also remain several important questions to be addressed. First, sumoylation of TRI occurs on a Lys residue that is not in a consensus sumoylation motif 
Kx(D/E), where is a large hydrophobic residue1. Thus, identification of the E3 ligase will be important in understanding the mechanism of TRI sumoylation. In the consensus Kx(D/E) motif, phosphorylation of Ser close to this motif contributes to sumoylation. It will therefore be interesting to study how phosphorylation of TRI regulates the sumoylation of Lys 389. Second, sumoylation of TRI results in accelerated activation of the TRI kinase and downstream Smad signalling, but it is also possible that TRI sumoylation may regulate non-Smad signalling pathways. Transcriptional activities were only partially suppressed, but metastasis of the Ras-transformed MEFs was markedly suppressed by the K389R and S385Y mutations. As such, whether non-Smad pathways are modulated by these mutants remains an important question. Third, the fate of TRI protein may be altered by sumoylation. Kang et al.2 have shown enhanced activation of R-Smads at the early phase of receptor activation. However, as sumoylation has been reported to induce endocytosis of the kinate receptor subunit GluR6 (ref. 14), it is possible that sumoylation may also regulate endocytosis and recycling of the TGF- receptor complexes.
In advanced cancers, TGF- is known to stimulate progression of cancer through induction of epithelial–mesenchymal transition (EMT), as well as its action on the tumour microenvironment13. TGF- is also involved in the development of various fibrotic diseases. Monoclonal antibodies to TGF- and small-molecule inhibitors for TRI are currently being assessed in preclinical and clinical trials for use in the treatment of cancer and fibrotic diseases15. Although better tissue-penetration can be achieved by small-molecule inhibitors than by monoclonal antibodies, small-molecule inhibitors of TRI that have been developed so far also inhibit the kinase activities of ALK-4 and ALK-7 (ref. 15). Thus, compounds that specifically inhibit TGF- signalling may be more valuable for therapeutic purposes. Inhibition of ubiquitination by proteasome inhibitors is a recently developed, effective way to treat certain diseases, including multiple myeloma and rheumatoid arthritis. As the mechanism by which TRI is sumoylated may differ from that of other proteins, it may be possible to specifically block TGF- signalling by inhibition of TRI sumoylation, providing a new strategy for treatment of diseases induced by aberrant TGF- signalling.
