Bone morphogenetic proteins (BMPs) control numerous processes in developing and adult organisms by binding to receptors on various cell types. The signals are relayed to the nucleus by signaling molecules called Smads. How the Smads are recruited to BMP receptors has been difficult to establish. In a paper appearing in this issue of Journal of Cell Science, Shi and colleagues show that a lipid-binding protein present in intracellular endosomal compartments is the crucial link (Shi et al., 2007). The protein also appears to act as a switch that can shut off signaling.
BMPs - some are also known as growth and differentiation factors (GDFs) - form the largest group within the transforming growth factor β (TGF-β) family, which also includes TGF-β, activins and inhibins, Nodal, and Mullerian-inhibiting substance (Miyazawa et al., 2002). BMPs were originally identified as cytokines that induce bone and cartilage tissue in vivo; since then they have been shown to be required in soft tissue in the embryo, and the adult organism for regulation of proliferation, differentiation and apoptosis. Nearly, all bind as dimers to two distinct receptor types, type I and II, which contain serine/threonine kinase activities in their intracellular domains. Upon ligand binding, two type-I receptors and two type-II receptors form a complex. The type-II receptor then phosphorylates the type-I receptor, which in turn phosphorylates the downstream signaling molecules, Smad proteins.
Smads include receptor Smads (R-Smads), common Smads (Co-Smads) and inhibitory Smads (I-Smads) (Shi and Massague, 2003). Type-I receptors for TGF-β, activins and Nodal recognize and phosphorylate the R-Smads Smad2 and Smad3; type-I receptors for BMP and GDF recognize and phosphorylate R-Smads Smad1, Smad5 and Smad8. The common Co-Smad Smad4 associates with phosphorylated R-Smads, translocating with them to the nucleus to regulate transcription.
A key issue is how the various Smads are recruited to the activated receptor complex for phosphorylation. In the case of Smad2 and Smad3, this involves a protein called Smad anchor for receptor activation (SARA) (Tsukazaki et al., 1998), which recruits them to the vicinity of the receptor. Phosphorylation of Smad2 increases its affinity for Smad4 and decreases its affinity for SARA - it therefore leads to release of Smad2, unmasking its nuclear import signal and rapid accumulation of activated Smad2 in the nucleus. Interestingly, SARA contains a phosphatidylinositol 3-phosphate [PtdIns(3)P]-binding FYVE domain (Stenmark et al., 1996), which localizes it predominantly to the early endocytic compartment (Stenmark and Aasland, 1999). How can an early endosomal protein present Smad2 and/or Smad3 to the phosphorylated receptor, if the latter is on the plasma membrane? Activated receptors internalized into the early endocytic compartment might encounter the SARA-Smad complex there, initiating signals from this compartment (Panopoulou et al., 2002; Hayes et al., 2002). Alternatively, small amounts of SARA are known to localize to the plasma membrane by binding to the receptor; the SARA-Smad complex might encounter activated receptors there, initiating signals at this location. Following internalization into the early endosome, signaling could then continue in a SARA-enriched environment (Di Guglielmo et al., 2003).
Regardless of how and where activation occurs, SARA is clearly an essential component of Smad-dependent signaling by TGF-β and activin A. But how about BMPs? How are Smad1, Smad5 and Smad8 recruited to activated BMP receptor complexes? Shi and co-workers now provide the answer (Shi et al., 2007), showing that a FYVE-domain protein, endofin (endosome associated FYVE domain protein) functions as a Smad anchor for receptor activation in the BMP pathway.
Endofin was originally identified as a ubiquitously expressed FYVE-domain protein that localizes to the early endocytic compartment (Seet and Hong, 2001). Because of its similarity to SARA, endofin has been tested for interactions with Smad2 and involvement in TGF-β signaling; however, the outcome was negative on both counts (Seet and Hong, 2001). Shi et al. show convincingly that endofin interacts with unphosphorylated Smad1 through its putative Smad-binding domain (SBD). Immunofluorescence and protein-fragment complementation indicates this interaction occurs in the early endocytic compartment. Mutations in the FYVE domain and SBD of endofin decrease BMP signaling by 30% and 20%, respectively. Furthermore, BMP-induced phosphorylation of Smad1 is moderately affected endofin knockdown. In addition, Shi et al. show that endofin regulates the expression of BMP-dependent genes in vitro and in vivo (Shi et al., 2007).
As in the case of SARA, there is an open issue concerning where Smad1 phosphorylation occurs. Given the predominant localization of endofin, one could conclude that Smad1 phosphorylation occurs in the early endosome after internalization of the ligand-receptor complex. However, recent data provide evidence that Smad1 phosphorylation is independent of trafficking, and internalization is required for continuation of signaling and transcriptional regulation of BMP-dependent genes (Hartung et al., 2006). Apparently, more work is required concerning this issue.
Shi et al. also implicate endofin in dephosphorylation of the BMP type-I receptor. SARA and endofin each have a consensus binding site for the type I serine/threonine protein phosphatase (PP1). In the case of SARA, the interaction with PP1c recruits the phosphatase to the type-I receptor (ALK5) for dephosphorylation both in the absence of ligand (because the type-II receptor is constitutively active it may occasionally phosphorylate the type-I receptor) and in its presence (Bennett and Alphey, 2002). Smad7, an I-Smad, binds to ALK5 and recruits GADD34, a regulatory and targeting subunit of the PP1 holoenzyme. This complex in turn binds to the catalytic subunit of PP1, which is presented by SARA (Shi et al., 2004). Endofin is different: Smad7 is not involved and GADD34 binds directly to the type-I receptor, ALK3. Therefore, endofin fulfills a dual role in BMP signaling: (1) it interacts with Smad1, thereby presenting it to the receptor for phosphorylation; (2) it interacts with PP1c, leading to dephosphorylation of the BMP type-I receptor. Further work is required to address the fine-tuning of this regulation.FIG1
Recently, endofin was shown to interact with Smad4 and facilitate TGF-β signaling without affecting BMP signaling (Chen et al., 2007). This would appear to contradict the new results; the reasons for the discrepancy are unclear. Interestingly, Chen et al. presented evidence that endofin interacts with several type I receptors, including ALK5 (a TGF-β receptor), ALK4 (an activin A receptor), ALK3 and ALK6 (BMP receptors) and ALK1 (a TGF-β/BMP receptor). They also suggested that SARA and endofin can form a complex. Thus, endofin might function more widely in TGF-β family signaling.
Since both SARA and endofin are enriched in the early endosomal compartment, the endocytic pathway is strongly implicated in the control of TGF-β signaling. Indeed, in the case of ALK5, the route of endocytosis can modulate signaling (Di Guglielmo et al., 2003; Felberbaum-Corti et al., 2003). There are five known main internalization routes: (1) clathrin-coated vesicles, (2) caveolae, (3) macropinocytosis/phagocytosis, (4) non-clathrin, non-caveolar pathways and, (5) the recently identified APPL pathway (Miaczynska et al., 2004). The route seems to determine which downstream proteins the ligand-receptor complex encounters, dramatically influencing the signaling output.
When the TGF-β ligand-receptor complex is internalized in clathrin-coated vesicles, signaling is enhanced by the abundant availability of SARA-Smad2/Smad3 complexes on early endosomes (Di Guglielmo et al., 2003). However, when it is internalized by caveolae, where Smad2 and/or Smad3 are not present, the result is attenuation of TGF-β signaling as a consequence of ubiquitylation and degradation of the complexes via interaction with Smad7-Smurf2 in caveosomes (Di Guglielmo et al., 2003). Internalization of BMP-2-receptor complexes in clathrin-coated vesicles is important for Smad-dependent cascades (Hartung et al., 2006), which is consistent with the early endocytic localization of endofin (Shi et al., 2007). Interestingly, Smad-independent signaling depends on internalization of BMP-2-receptor by caveolae (Hartung et al., 2006).
Another interesting point raised by the involvement of endofin in BMP-receptor signaling is the fact that it can recruit TOM1 to endosomes. TOM1 is a member of a protein family characterized by a Vps27-Hrs-Stam (VHS) domain, and TOM1-endofin interaction is required for the recruitment of clathrin by endofin (Seet and Hong, 2005). The VHS domain is considered to have a general membrane-targeting/cargo-recognition role in vesicular trafficking, and in GGA proteins interacts with receptor tails. TOM1 interacts with ubiquitin and Tollip, a protein that is involved in IL1 signaling and also binds polyubiquitin chains (Yamakami et al., 2003; Brissoni et al., 2006). TOM1 thus links polyubiquitylated proteins to clathrin. TOM1 is itself mono-ubiquitylated, and this may function as a signal for recruitment of other ubiquitin-binding proteins, including ubiquitin-interaction motif (UIM)-containing proteins.
What is the relevance of this? In the early endocytic compartment, receptors destined for lysosomal degradation are sorted from those which will be recycled back to the plasma membrane. The sorting machinery recognizes proteins destined for lysosomes in a number of ways, one of which is by recognizing conjugated ubiquitin. Ubiquitylated proteins are recognized by the Hrs-STAM-EPS15 complex and retained in specialized microdomains on the endosomal membrane, which become stabilized by the recruitment of clathrin by Hrs. Consequently, Hrs recruits the endosomal sorting complex required for transport (ESCRT) complexes involved in multivesicular body (MVB) formation, the endosomal membrane invaginates, ubiquitin is removed and recycled back to the cytosol; upon this the MVB is formed and finally fuses with the lysosome (Raiborg and Stenmark, 2002). Overexpression of endofin inhibits EGF degradation (Seet and Hong, 2001), suggesting that endofin has a role in early-to-late-endosome trafficking. However, overexpressed endofin may simply sequester PtdIns(3)P and, therefore, this issue needs to be reinvestigated. Since endofin interacts with the type-I receptors of the TGF-β family, such as ALK5, ALK4, ALK3, ALK6 and ALK1 (Chen et al., 2007), a key issue will be to examine its potential role in the degradation of TGF-β family receptors.
Endofin thus clearly plays a key role at several levels of BMP signaling. First, it binds to Smad1, enhancing its phosphorylation and signaling. Second, it binds to Smad4, presenting it to Smad2 and Smad1. Third, it regulates ALK receptor phosphorylation by recruiting PP1c. A possible fourth involvement may be in the degradation of TGF-β family receptors. Its implication in BMP-receptor signaling opens new avenues for dissection of the molecular mechanisms involved in the regulation of this pathway.
The comments of Theodore Fotsis, Harald Stenmark, Danny Huylebroeck and Aris Moustakas are gratefully acknowledged.