SEARCH

SEARCH BY CITATION

Keywords:

  • TGFβ signaling;
  • Xenopus;
  • Drosophila;
  • D-V patterning;
  • Smad;
  • Schnurri

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

The bone morphogenetic protein (BMP) signaling pathway is a conserved and evolutionarily ancient regulatory module affecting a large variety of cellular behaviors. The evolutionary flexibility in using BMP responses presumably arose by co-option of a canonical BMP signaling cascade to regulate the transcription of diverse batteries of target genes. This begs the question of how seemingly interchangeable BMP signaling components elicit widely different outputs in different cell types, an important issue in the context of understanding how BMP signaling integrates with gene regulatory networks to control development. Because a molecular understanding of how BMP signaling activates different batteries of target genes is an essential prerequisite to comprehending the roles of BMPs in regulating cellular responses, here we review the current knowledge of how BMP-regulated target genes are selected by the signal transduction machinery. We highlight recent studies suggesting the evolutionary conservation of BMP target gene regulation signaling by Schnurri family zinc finger proteins. Developmental Dynamics 238:1321–1331, 2009. © 2009 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

Bone morphogenetic proteins (BMPs) comprise the largest subgroup of the transforming growth factor β (TGFβ) superfamily of extracellular signaling molecules. Because BMPs were first identified for their bone-inducing properties (Wozney et al.,1988), their importance has been underscored by over 10,000 entries in the current literature. Among their many duties, BMPs control cell type specification, differentiation, pluripotency, apoptosis, proliferation, and tissue morphogenesis. An abbreviated list of developmental events in which BMPs function includes growth of the mouse egg cylinder, specification and development of germ cells, dorsal–ventral (D-V) patterning of the body axes, induction of epidermis, early patterning of the central nervous system, neural crest and various placodes, and development of craniofacial structures, skin accessory organs, eyes, limb buds, and virtually every organ of the body. Being involved in so many different developmental events, it comes as little surprise that BMPs are also an evolutionarily ancient group of proteins. BMP signaling is at least 700 million years old, predating the evolution of the bilateria (see Fig. 1). Ligands and downstream components of the signaling cascade have been found both in Cnidaria (metazoans that include Hydra, the anemone Nematostella, jellyfish, and corals), and also in the sponges, the most ancient metazoan clade (Suga et al.,1999; Hayward et al.,2002; Finnerty et al.,2004; Reinhardt et al.,2004; Rentzsch et al.,2006,2007; Reber-Müller et al.,2006; Nichols et al.,2006; Adamska et al.,2007; Zoccola et al.,2009). Thus far, no TGFβ superfamily signaling components have been reported in taxa “basal to” the sponges, including the single-celled choanoflagellates, fungi, or protozoa (Nichols et al.,2006; King et al.,2008). Therefore, TGFβ signaling was probably invented at or near the origins of the metazoa, ∼700 million years ago (Peterson et al.,2008) and subsequently co-opted to play roles in an increasing variety of biological processes.

thumbnail image

Figure 1. A phylogenetic tree depicting the appearance of bone morphogenetic protein (BMP) signaling cascade components during evolution. Animals within these categories that have been examined for the presence or absence of BMP signaling components using either Metazome (metazome.org) or Blast searches (blast.ncbi.nlm.nih.gov/Blast.cgi or compagen.zoologie.uni-kiel.de/blast/blast_cs.html). Using Metazome, mammals examined include Homo sapiens, Mus musculus, Rattus norvegicus, Canis familiaris, and Monodelphis domestica. Birds include the chicken Gallus gallus, while amphibians included Xenopus tropicalis. Teleost fishes include Gastrocleus aculeatus, Oryzias latipes, Takifugu rubripes, and Danio rerio. Urochordates, cephalochordates, and echinoderms examined were the ascidian Ciona species savignyi and intestinalis and the amphioxus Branchiostoma floridae and the sea urchin Strongylocentrotus purpuratus, respectively. Insects consisted of Drosophila melanogaster, Anopheles gambiae, Aedes aegypti, Bombyx mori, and Tribolium casteneum. Nematode worms included both Caenorhabditis elegans and briggsae, while molluscs are represented by the limpet Lottia gigantean. Cnidaria are represented by Nematostella vectensis. We also blast searched the genome of the cnidarian Hydra magnapapillata using http://hydrazome.metazome.net/cgi-bin/gbrowse/hydra/. We identified brinker candidate genes in nondrosophilid insect species belonging to Culex, Anopheles, Pediculus, and Apis using NCBI blast. Neither blast searches of the Daphnia pulex genome (http://genome.jgi-psf.org/Dappu1/Dappu1.home.html) nor the NCBI EST database found a brinker in this species. BMP R-Smads were also identified using NCBI blast in the poriferan Oscarella carmela. Compagen was used to blast search the lamprey Petromyzon marinus, the choanoflagellate Monosiga brevicola, the poriferan Amphimedon queenslandica, and the placozoan Trichoplax adhaerans (not shown as precise position in the tree remains controversial). Branch lengths herein do not attempt to accurately depict the evolutionary divergence times or genetic distances between different sister groups.

Download figure to PowerPoint

With the great diversity in cellular responses elicited by BMPs, one commonly asked question is: how can seemingly interchangeable BMP ligands elicit such a wide range of different outputs? A partial answer to this question comes from early embryologists—how a cell responds to external stimuli is not encoded in the stimulus per se, but is a function of the responding cell's “developmental history.” Differential gene induction by BMPs depends on the specific nature of the “machinery” expressed in each responding cell type. Thus, the major focus of the current review is to illuminate some of the many ways the core (canonical) machinery of the BMP signal transduction cascade couples with cell-type specific transcription factors to differentially regulate target gene expression.

THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

The TGFβ superfamily can be divided into two branches based on the components used in each signaling cascade. One branch contains TGFβ1, the ligand for which the superfamily is named, and also the activins and nodals. The other branch contains the BMPs, the growth and differentiation factors (GDFs) and anti-Mullerian hormone (AMH). It is noteworthy that some ligands named as BMPs/GDFs signal by means of the TGFβ branch (e.g., GDF1/Vg1), while other TGFβ superfamily members appear to function as inhibitors of signaling (e.g., Lefty). A “core” signal transduction cascade applies to both branches and has been elucidated in great detail (Fig. 2; Shi and Massagué,2003; Feng and Derynck,2005; Miyazono et al.,2005; Ross and Hill,2008). At nearly every step, the signaling cascades of the two branches each contain branch-specific components, presumed to insulate these signaling cascades from one another.

thumbnail image

Figure 2. The canonical BMP signaling cascade.

Download figure to PowerPoint

The signaling ligands are disulfide-linked dimers and bind receptors consisting of two subunits (referred to as receptor “types” I and II). Both receptor subunits contain cytoplasmic serine/threonine kinase domains at their C-termini. Ligand binding facilitates/stabilizes type I-type II interactions and active signaling complexes have a (type I)2(type II)2 subunit stoichiometry. The coming together of type I and type II receptor subunits permits efficient transphosphorylation of the type I receptor subunit by the type II receptor, which is a constitutively active kinase. Phosphorylation of the type I subunit stimulates its own kinase activity and, in turn, the type I subunit phosphorylates cytoplasmically localized signal transducers, the Smads. Thus, the ligand-receptor interaction ultimately facilitates receptor type I subunit activation, thereby transmitting the signal to the cell's interior by means of the Smads. While there is some evidence for “noncanonical” BMP signaling, cascades that may be Smad-independent (reviewed by Moustakis and Heldin,2005), the canonical pathway by means of the Smads seems to be the major pathway for BMPs to influence gene expression and thus will be the focus of this review.

Upon activation, the type I receptors phosphorylate the Smads on their C-terminal SSXS sequence motif and this modification results in activation of the Smad proteins. The TGFβ and BMP branches each have Smads dedicated for signaling. Smads 2 and 3 are used in the TGFβ branch, whereas Smads 1, 5, and 8 (also known as Smad9) transmit BMP signals. All of these Smads are referred to as receptor-regulated Smads (R-Smads) to differentiate these from Smad4. This Smad (often referred to as a “Co-Smad”) is shared by R-Smads for both the TGFβ and BMP cascades, lacks the C-terminal SSXS motif, and therefore is not phosphorylated by the type I receptors. Finally, there are also inhibitory Smads (I-Smads) 6 and 7, which function to block TGFβ superfamily signaling at several levels. Both the R-Smads and Smad4 are structurally composed of highly conserved N-terminal Mad Homology (MH) 1 and C-terminal MH2 domains. The MH1 domain, which houses DNA-binding activity, is separated from the MH2 domain by a “linker” region that is less well conserved and thought to be largely unstructured and flexible. The MH2 domain functions in transcriptional activation and also is required for Smad multimerization. R-Smads form complexes with Smad4 in a 2:1 ratio, although R-Smad homotrimers lacking Smad4 (Correia et al.,2001) may also play a role in signaling. C-terminal Smad phosphorylation accomplishes at least two major ends. Phosphorylation elicits a conformational change, freeing an interaction between the MH1 and MH2 domains that is inhibitory to DNA binding. This “opening” of the R-Smads permits their interaction with Smad4 and their accumulation in the nucleus where they gain access to the genome.

While this cursory overview of signal transduction explains the transmission of BMP signals from the extracellular environment, through the receptors, to Smad complexes in the nucleus, it does not sufficiently address how activation of this pathway leads to target gene identification. Much has been published on this subject and we will discuss various strategies for direct target gene recognition in response to BMP stimulation. This first requires some discussion of the DNA binding properties of the Smads.

IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

Current evidence, much from overexpression experiments performed in cell culture, Xenopus, and zebrafish embryos, suggests that all three BMP R-Smads are functionally interchangeable (Suzuki et al.,1997; Dick et al.,1999; Kramer et al.,2002). Phenotypic analyses of mouse Smad knockouts also suggests that Smads 1, 5, and 8 are functionally redundant (Hester et al.,2005; Arnold et al.,2006; Pangas et al.,2008; Retting et al.,2009). Consistent with this notion, all three BMP R-Smads share a very high degree of amino acid sequence similarity. Smads 1 and 5 are nearly identical and most of their divergence from Smad8 is in the linker separating the MH1 and MH2 domains. These observations suggest that the three BMP R-Smads evolved by means of gene multiplication from a single Smad1/5/8 precursor (Fig. 1). This notion is further supported by the observation that the cnidarian Nematostella contains genes encoding only four Smads, with high amino acid sequence similarity to Smads1/5/8, Smads2/3, Smad4, and the I-Smad, Smad6. The Nematostella smad1/5/8 gene and vertebrate smad8 genes also share linkage to the nearby alg5 gene, suggesting local synteny. Echinoderm (Strongylocentrotus; sea urchin), cephalochordate (Branchiostoma; amphioxus), and urochordate (Ciona; ascidian) genomes also all contain a single smad1/5/8 ortholog. On the other hand, the genomes of the fishes and tetrapods contain (at least) three distinct smad1, smad5, and smad8 genes, as does the genome of the lamprey Petromyzon, a jawless vertebrate (agnathan) that is considered the outgroup to all other vertebrates. These observations are consistent with the hypothesis that the three BMP R-Smads arose during the two rounds of whole genome duplication (and gene loss) that occurred before the emergence of agnathan fishes at or near the origin of the vertebrata (Fig. 1).

Smads identify gene targets by direct binding to cis-acting DNA regulatory elements. The MH1 domain of Drosophila Mad (Smad1/5/8) was found to bind DPP responsive regulatory regions in the vestigial, labial, and Ultrabithorax (Ubx) genes and interacts with the consensus sequence 5′-GCCGNCGC-3′ (Kim et al.,1997). Vertebrate BMP R-Smads seem to have a similar GC-rich binding site preference (Kusanagi et al.,2000; Ishida et al.,2000) but many variations in sequence have since been reported. Binding site selection assays performed for the Smad3 (R-Smad for the activin/nodal/TGFβ branch) and Smad4 MH1 domains suggest that these proteins prefer binding the DNA motif 5′-GTCTAGAC-3′ (Zawel et al.,1998). An X-ray structure of the Smad3 MH1 domain co-crystalized with this DNA motif (Shi et al.,1998) shows that Smad3 binding occurs on this palindrome through the “half site” 5′-GTCT-3′ (reverse complement is 5′-AGAC-3′), and this sequence is often referred to as a “Smad-binding element” (SBE). Of interest, the crystal structure, combined with sequence alignments for all R-Smads and Smad4, suggests that all signaling Smads should have similar DNA-binding specificities because the amino acid residues in a β-hairpin that directly contacts the DNA bases are invariant between the human R-Smads and Smad4 (Shi et al.,1998).

Consistent with this view, the affinity of the isolated Smad1 MH1 domain binding to the SBE (half site) was determined to be ∼5 × 10−7 M (Shi et al.,1998), and the affinities of the Smad3 and Smad4 MH1 domains are similar, being ∼1.1 × 10−7 M, and ∼3 × 10−7 M, respectively (Shi et al.,1998). Unfortunately, similar affinity measurements are lacking for binding of these MH1 domains to the GC-rich motif (both types of Smad binding motifs can be considered degenerate forms of the motif 5′-GNCN-3′). Furthermore, fly Medea (Smad4) was also shown to bind several GC-rich elements in the tinman promoter (Xu et al.,1998).

Notwithstanding these experimentally observed differences in DNA binding site preference, the BMP R-Smads generally prefer GC-rich binding elements, whereas the other Smads prefer the SBE. GC-rich boxes also differentially favor BMP activation in cell culture transfections and reporter gene induction is inefficient with SBE sites alone (Korchynskyi and ten Dijke,2002; Katagiri et al.,2002; López-Rovira et al.,2002). Perhaps the similarity in in vitro affinities of isolated MH1 domains to DNA motifs does not accurately reflect the preferences of the full-length proteins complexed in R-Smad/Smad4 heterotrimers. It has also been suggested that Smad1's preference for GC-rich motifs over SBEs might be due to a series of highly basic amino acids adjacent to the β-hairpin residues found in all three BMP R-Smads that is lacking in both the R-Smads specific to the activin/nodal/TGFβ branch and also Smad4 (Shi et al.,1998; Shi,2001).

A challenge for Smads in mediating specific targeting of BMP and TGFβ responsive genes is posed by the relatively weak affinities of the Smads for these sequence motifs, being several orders of magnitude lower than most sequence-specific DNA-binding proteins. Due to the Smads' low affinity and selectivity for their binding sites, they are not expected to be capable of strongly differentiating between proper targets of gene activation and random DNA. This may explain why relatively long concatemers of Smad-binding sites are often needed to obtain transcriptional activation in cell culture (Jonk et al.,1998; Kusanagi et al.,2000). The promoters of some BMP inducible genes (e.g., bambi, and the id and vent family genes) have numerous Smad-binding sites (López-Rovira et al.,2002; Korchynskyi and ten Dijke,2002; Karaulanov et al.,2004), presumably to more efficiently recruit the Smads to enhance transcriptional regulation. While isolated MH1 domains of Smad3 do not show cooperative binding to palindromic SBEs (Shi et al.,1998), it is reasonable to believe that full-length Smads would behave differently. Smad MH2 domains promote heterotrimerization (Correia et al.,2001; Qin et al.,2001; Chacko et al.,2001), and thus these MH1 domains, physically “tethered” to one another in trimeric Smad complexes, should be capable of binding SBE and GC-rich sites organized in clusters with higher affinities than those shown for isolated MH1 domains on single 4-bp elements. There is also evidence suggesting that not all relative orientations of GC-rich/SBE GNCN motifs, in dimer or multimer form, are equivalent in Smad binding affinity (Johnson et al.,1999; Gao and Laughon,2006,2007). While it can be debated as to whether Smads binding alone is sufficient for transcriptional activation of the target sites, the fact that SBE/GC-rich sites occur in relative isolation in some promoters, and that Smads interact weakly with single binding sites, suggests that Smad-DNA interactions are typically stabilized by other sequence-specific DNA binding partners. The early demonstration that FoxHI (Fast1) protein directly interacts with Smad2/4 complexes in the mix.2 promoter, and that interactions of Smad2/4 with Nodal target genes are enhanced by FoxHI provided strong support for this idea (Chen et al.,1996;1997), which has held considerable sway over models for regulation of many BMP/TGFβ target genes.

TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

Because the importance of cooperative interactions between Smads and other sequence-specific DNA-binding factors was first recognized, several transcription factors mediating BMP signaling have been identified. These proteins are believed to partner with the Smads in strengthening Smad-DNA interactions (as suggested by the FoxHI example) and thereby play a major role in recruiting Smads to genes containing binding sites for these partnering transcription factors. One such Smad partner first implicated in mediating BMP signaling is Oaz (OE/EBF-associated zinc-finger protein, also known as Znf423). This multi-zinc finger protein has been proposed to mediate signaling by means of a BMP response element (BRE) in the Xenopus vent2 gene (Hata et al.,2000). Oaz binds the BRE in vitro, and activates BMP-dependent reporter gene expression in transient transfection assays. An oaz knockout mouse shows brain midline (Cheng et al.,2007) and olfactory neuronal defects (Cheng and Reed,2007) that may be linked to impaired BMP signaling. However, these mutant mice fail to show defects in earlier embryogenesis associated with reductions in BMP signaling, suggesting other transcription factors may be involved in mediating BMP signaling earlier in development. Consistent with this view, the spatiotemporal patterns of oaz and vent2 expression in Xenopus tail bud stage embryos are also nonoverlapping. A potential complication is the presence in vertebrate genomes of another zinc finger gene, znf521, which is closely related to oaz. It is currently unclear whether znf521 is functionally redundant with oaz because the expression of this gene is down-regulated in response to BMPs in mesenchymal cells and Znf521 appears to antagonize osteoblast differentiation (Wu et al.,2009). It should be noted that Drosophila contains a single oaz ortholog showing local synteny with the chromosomal regions containing vertebrate oaz genes. Loss of this gene results in defects in posterior spiracle development (Krattinger et al.,2007), but the direct link between fly oaz and DPP signaling in this process remains to be established. Based on these observations, Oaz may not be a ubiquitous/universal DNA-binding Smad partner but, rather, Oaz may act as a specialized tissue-specific transcription factor mediating BMP signaling.

Runx2 is another important transcription factor regulating the expression of BMP targets. Runx2's role in BMP signaling has been revealed in studies of chondrocyte and osteocyte differentiation. Loss of runx2 leads to severe abrogation of bone development (reviewed in Karsenty,2008). Genes such as osteocalcin, osteopontin, col10, and smad6, which are involved in osteoblast formation, are all directly regulated by binding of Runx2 together with Smad1 on the promoters of these genes (Zhang et al.,2000; Drissi et al.,2003; Wang et al.,2007). Because both Runx2 and BMP signaling are critical for development of these cell types, these data suggest that Runx2 may widely coordinate the expression of BMP target genes involved in osteoblast differentiation. However, whether all Runx2 targets in osteoblasts also incorporate BMP R-Smad input is unknown at present.

Among the other DNA-binding Smad partners that have received attention are Zeb2 (also known as Sip1) and Zeb1 (also known as δEF1; Sekido et al.,1996; Verschueren et al.,1999). These closely related zinc finger proteins appear to form an inhibitor/activator pair and are orthologs of Drosophila zfh1 (Postigo,2003; Postigo et al.,2003). Zeb2 is expressed in early Xenopus and zebrafish embryos and appears to negatively regulate expression of at least some BMP target genes (van Grunsven et al.,2006; Delalande et al.,2008). The interplay between these proteins and TGFβ signaling is complex and they also function in regulating activin/nodal signaling (Verschueren et al.,1999). Smad1 protein has also been found to physically interact with HoxC8, Nkx3-2, YY1, β-catenin/Lef1 complex, and Gata factors 4, 5, and 6 (Shi et al.,1999; Kim and Lassar,2003; Benchabane and Wrana,2003; Brown et al.,2004; Lee et al.,2004; Hu and Rosenblum,2005).

All the transcription factor partners for the BMP R-Smads described here have sequence-specific DNA binding activity and presumably direct Smad binding to specific BMP target genes to mediate BMP signaling in certain cell types. It remains unclear whether there exist ubiquitous DNA-binding Smad partners that regulate batteries of BMP-responsive genes. And because Smad partners recognize different DNA motifs, it seems unlikely that there will be a cis-regulatory “code” (beyond the Smad motifs discussed earlier) that is common to recognition of all BMP targets.

SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

In Drosophila, activation of most DPP (BMP) targets occurs through an indirect, double repression mechanism operating at the level of transcriptional regulation (Fig. 3A, top). The gene brinker encodes a repressor of direct BMP target genes transcription (Campbell and Tomlinson,1999; Jaźwińska et al., 1999a,b; Minami et al.,1999) and its widespread expression is controlled by yet to be characterized transcriptional activators (Yao et al.,2008). DPP/BMP signaling alleviates Brinker-mediated repression by inhibiting brinker transcription (Fig. 3A, bottom; Campbell and Tomlinson,1999; Jaźwińska et al., 1999a,b; Minami et al.,1999). A critical transcription factor mediating this DPP-dependent repression of brinker is the zinc finger DNA-binding protein Schnurri (Shn; German for whiskers; Arora et al.,1995; Grieder et al.,1995; Staehling-Hampton et al.,1995). shn encodes a large protein (∼275 kDa) composed of eight zinc fingers (arranged in four clusters), which interacts with Smads (Mad and Medea, orthologs of vertebrate Smads 1/5/8 and 4, respectively) to mediate DPP signaling (Dai et al.,2000; Udagawa et al.,2000). In response to DPP/BMP signaling (Fig. 3A, bottom), Schnurri represses the transcription of brinker by binding to its promoter region together with Smads (Pyrowolakis et al.,2004). This, in turn, frees up Brinker-repressed BMP targets, allowing them to be up-regulated. Indeed, a majority of the genes regulated by DPP signaling in Drosophila contain Brinker binding sites and require alleviation of Brinker repression.

thumbnail image

Figure 3. Various modes of Schnurri docking to bone morphogenetic protein (BMP) target genes. A: In Drosophila, in response to DPP signaling, Schnurri binds to the regulatory regions of brinker, bag of marbles, and gooseberry by means of docking to SMM (“Schnurri/Mad/Medea”) sites. These sequence motifs are directional: they are composed of a GC-rich site for Mad (Smad1) binding followed by a “Smad-binding element” (SBE) site for Medea (Smad4) binding. Schnurri only docks when these two motifs are in this orientation and only when separated by 5 bps. B: Fly Schnurri interacts with the Ubx B enhancer to drive Ubx transcription in the midgut. Schnurri binding to this gene also occurs by means of two interactions, one by means of a direct contact of Schnurri zinc fingers with NFκB-like sequence motifs in the Ubx gene, and the other by means of Schnurri docking to fly Smads bound to canonical sites located less than 100 bp distally. C: In mouse, Shn2 interacts with the pparg2 promoter region by means of docking to two protein intermediaries, the Smad complex bound to an SBE in the promoter proximal region, and C/EBP bound to a canonical CCAAT box located approximately 90 bp distally. D: As in the case of Drosophila Schnurri, vertebrate Shn1 proteins can bind to SMM elements as in the case of the BMP-response elements (BREs) found in the Xenopus vent2 and id3 promoter proximal regions. These BREs have the same architecture as the fly SMM sites but in vertebrate cells both fly and vertebrate Shn1 act as transcriptional activators to mediate BMP signals.

Download figure to PowerPoint

Analysis of brinker regulation by Schnurri has revealed an interesting mechanism explaining how receptor activation of the Smads can be used for repression of gene transcription. The regulatory region of the brinker gene contains a “silencer element,” the brkSE, which was identified as a 16-bp cis-acting element that responds to DPP signaling (Pyrowolakis et al.,2004). This motif is also present in the DPP-responsive genes bag of marbles and gooseberry. Recently, eleven copies of this motif, referred to as SMM sites, were found as part of cis-regulatory modules distributed across 16kb of the brinker upstream regulatory region and play a complex role in brinker repression (Yao et al.,2008). An extensive mutagenesis analysis has shown that the brkSE is comprised of single Mad and Medea consensus binding sites separated by a precise 5-bp spacer region. The length of this spacer was shown to be critical for Schnurri docking to this site by means of protein–protein interactions with both Mad and Medea (Pyrowolakis et al.,2004; Gao et al.,2005). Currently, there is no evidence that Schnurri binds directly to the brkSE and thus the current model supposes that the geometry of the Mad/Medea complex on the DNA, spaced by 5-bp, is the critical determinant that creates a surface for Schnurri docking (Pyrowolakis et al.,2004; Gao et al.,2005; Gao and Laughon,2006,2007). It remains a puzzle why such precise spacing is required for Schnurri docking as Schnurri interacts with the Smad MH2 domains and these are separated from their DNA-bound MH1 domains by the linker region, which is thought to be flexible. One possibility is that the linker domains in the [R-Smad]2:[Smad4]1 complex are not as accommodating as is widely presumed.

In addition to this repressive function, Shn may also act as a transcriptional activator (Fig. 3B). Shn is required for Ultrabithorax (Ubx) expression in the midgut, and binds to the Ubx B midgut enhancer together with Mad and Medea (Dai et al.,2000). Shn binding to the Ubx gene occurs by means of two interactions. The Shn zinc finger domains directly bind the enhancer by means of NF-κB-like DNA sequence motifs. Shn activates a Ubx B reporter gene in cell culture and mutation of the NF-κB-like Shn-binding site reduces Ubx B reporter expression in fly embryos (Dai et al.,2000). Shn also contacts the fly Smads bound to sites located less than 100 bp from the NF-κB-like elements and shown to be important for DPP regulation. Whereas it is currently unclear how many genes are direct targets for Schnurri-dependent activation of transcription in Drosophila, some of Schnurri's functions have been shown to be independent of brinker de-repression (Torres-Vazquez et al.,2000). It is also noteworthy that Brinker does not operate in BMP signaling in most other organisms as the brinker gene appears to be an evolutionary innovation of insects. brinker orthologs are present in Anopheles and Culex (both mosquitos), Pediculus (louse), Apis (honeybee), Tribolium (beetle), and Bombyx (silkworm), but not in Daphnia (waterflea, an arthropod and outgroup to the insects). Thus far, no brinker-like gene has been identified outside the insects (see Fig. 1). Of interest, Pfam's algorithms for identifying protein domains group Brinker (see pfam.sanger.ac.uk/family?id=BrkDBD) with Pogo element transposases, suggesting that Brinker may have evolved from an insect transposable element.

As brinker is absent in more distant members of the Ecdysozoa, such as the nematodes, this suggests that nematode BMP signaling probably does not involve a double repression mechanism. Therefore it is interesting that Caenorhabditis elegans (and briggsae) does have a Schnurri family member, encoded by the gene sma-9 (Savage-Dunn et al.,2003; Liang et al.,2003). This gene is predicted to produce multiple proteins containing up to 7 zinc fingers (organized in 3 clusters) by means of alternative splicing. sma-9 was identified in a mutant screen for small body length, a screen that also identified several other genes encoding BMP pathway components, including dbl-1 (dpp and bmp ligand-1, a worm BMP ligand), and sma-3 (a Smad transducer) (Savage-Dunn et al.,2003). SMA-9 was shown to function downstream of DBL-1 signaling by epistasis experiments (Liang et al.,2003), and to antagonize BMP signaling in dorsal–ventral patterning of the worm's mesoderm (Foehr et al.,2006). In addition, through the use of SMA-9 fusions to either the herpes virus VP16 transcriptional activation or engrailed transcriptional repression domains, phenotypic rescue experiments have shown that SMA-9 may act as both a transcriptional activator and repressor (Liang et al.,2007). These findings parallel results showing that fly Schnurri has a dual role in mediating activation and repression (Yao et al.,2006). The analogies to Drosophila may become clearer when the mechanistic details of C. elegans's SMA-9 protein interactions with the BMP R-Smads and genomic targets are further elucidated.

SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

Most vertebrate genomes, including the lamprey's, contain three Schnurri-related genes and thus triplication of the Schnurris, like the Smads, also seems to have occurred during the agnathan radiation (Fig. 1). The mammalian Schnurri family members (4–5 zinc fingers organized in 2–3 clusters) are variously known as Shn1/Hivep1/Mbp1/Zas1/PrdIIBFI (Fan and Maniatis,1990; Baldwin et al.,1990), Shn2/Hivep2/Mbp2/Zas2 (Nomura et al.,1991; van't Veer et al.,1992), and Shn3/Hivep3/Krc/Mbp3/Zas3 (Wu et al.,1993). These Shn proteins have mainly been studied for their possible roles in regulating an assortment of genes including those encoding collagen type IIA, α-A-crystallin, α-1-antitrypsin, β interferon, MHC H2-Kb and HIV genes (see review by Wu,2002), but have not been implicated in BMP signaling until very recently (Yao et al.,2006; Jin et al.,2006).

Evidence supporting a role for vertebrate Schnurris in BMP signaling comes from two independent routes. First, mice deficient for Shn2 have defects in bone homeostasis (Saita et al.,2007). shn2 mutant mice show deficiencies both in osteoblasts' ability to make bone and osteoclasts' ability to resorb bone (Saita et al.,2007). Overexpression of Shn2 in osteoblasts enhanced BMP-induced differentiation (Saita et al.,2007) and BMPs play a role in stimulating both of these cell types. In addition to bone defects, shn2 KO mice also have a reduction in the amount of white adipose tissue (Jin et al.,2006). The defect in adipogenesis has been linked to a reduction in the ability of the gene encoding peroxisome proliferator activator gamma 2 to be transcriptionally stimulated by BMPs. Interestingly, Shn2 interacts with the pparg2 promoter in a manner mechanistically different from the interaction of Shn with the Drosophila brkSE: Shn2 docks to Smad1, presumably with Smad4 bound to a typical SBE(s) in the promoter proximal region, while also contacting the protein C/EBPα, another factor known to play a role in adipogenesis, bound to a canonical CCAAT site located distally (Fig. 3C; Jin et al.,2006). Thus, the model suggests that Shn2, by means of interactions with DNA-bound Smad1/4 and C/EBPα, acts as a transcriptional activator to induce pparg2 in response to BMP signaling.

Another link between vertebrate Schnurris and BMP signaling comes from examination of the cross-species behavior of a BRE from the Xenopus vent2 gene (Candia et al.,1997; Rastegar et al.,1999; Hata et al.,2000; Karaulanov et al.,2004; von Bubnoff et al.,2005). The BRE mediates BMP induction during frog gastrulation and is also found in the promoters of two other BMP target genes, id3 and bambi (Karaulanov et al.,2004; von Bubnoff et al.,2005). Transgenic Drosophila lines bearing a vertebrate BRE-driven lacZ reporter gene have provided insight into how vertebrate target gene expression is regulated by BMPs. This transgene contained several binding sites for Grainyhead (Grh), a transcriptional activator that drives broad expression in the fly embryo and epithelial sheets of the larval imaginal discs. The transgene surprisingly exhibited patterns of expression during early embryogenesis reminiscent of fly DPP target genes (Yao et al.,2006) suggesting that the DPP-induced repression mechanism might have analogies to the activation mechanism in the frog (Yao et al.,2006).

In stark contrast to the repressive function of Schnurri, acting on the BRE (or brkSE) in Drosophila, overexpression of either human Shn1 or fly Shn in frog embryos activates the expression of a vent2 BRE luciferase reporter (Yao et al.,2006). Human Shn1 protein fragments supershift a BRE DNA fragment together with the Smads (Yao et al.,2006) just as in the situation with the Drosophila brkSE (Pyrowolakis et al.,2004). In addition, as in the fly brkSE situation, modifications to the frog vent2 BRE that alter the spacing between its putative BMP R-Smad and Smad4 binding sites, without altering the sites themselves, similarly disrupt its ability to function as a BMP response element in frog embryos (Yao et al,2006). Finally, human Shn1 protein can rescue fly schnurri mutant embryos as effectively as a fly schnurri, suggesting that human Schnurri can function as a repressor in flies just as Drosophila Schnurri could activate a reporter by means of the vent2 BRE in Xenopus (Fig. 3D; Yao et al.,2006). Thus, it has been postulated that the presence or absence of coactivators and corepressors is the critical component to determining Schnurri's functions in the two systems. Thus far there is little information regarding the nature of the cofactors influencing Schnurri function. Recently, however, it was revealed that direct physical interaction of C-terminal binding protein (CtBP), a well-known short-range acting corepressor, contributes to Schnurri's ability to repress brk transcription (Yao et al.,2008). In mammals Shn1 was shown to interact with MSX2-interacting nuclear target protein (MINT; also known as SHARP and SPEN), which further recruits SMRT/NcoR to negatively regulate the collagen type IIA (col2a1) gene (Yang et al.,2005). To our knowledge this is the only example of a Schnurri protein acting to repress gene transcription in vertebrates. However, whether Shn1's repression of the col2a1 gene requires active BMP signaling is currently not known.

It is noteworthy that like the mouse shn2 mutant, shn3 mutants also have bone defects. However, in contrast to shn2, shn3 mutants have an increase in adult bone mass and this is due to increased osteoblast activity (Jones et al.,2006). Shn3 functions in osteoblasts as a negative regulator of Runx2 protein stability (Jones et al.,2006). No evidence is currently available directly linking Shn3 to transcriptional regulation of BMP target genes and thus it remains unclear whether all three vertebrate Schnurris are capable of being functionally redundant.

Based on these pieces of evidence, it is clear that Schnurris interact with BMP target genes in a variety of ways: (1) Shn1 can dock onto Smads 1 and 4 bound to brkSE-type elements/BREs in both Drosophila (Fig. 3A) and Xenopus (Fig. 3D); (2) Shn2 binds the pparg2 promoter in adipocytes by means of interactions with Smads and C/EBP (Fig. 3C), which are separately bound to their sites less than one hundred base pairs apart; and (3) Schnurris can bind DNA directly by means of their clustered Zn finger domains and also contact Smads bound to nearby SBE or GC-rich elements, as in the case of the fly Shn interaction with the Ubx B enhancer (Fig. 3B). In each case, Schnurris are likely using the services of bound coactivators and corepressors to modulate transcription. These observations suggest that Schnurris act as scaffolds, permitting the interaction of a wide variety of factors and DNA elements. As the three vertebrate Schnurris are also widely expressed (Wu,2002; Durr et al.,2004), it is attractive to suggest that through these diverse interactions Schnurri may coordinate BMP regulation of a diversity of downstream BMP targets. This highly adaptive mechanism of Schnurri function may have helped to maintain conserved BMP signaling during evolution while recruiting other transcription factors into BMP signaling to increase its diversity. While we entertain this notion, we are also reminded that Schnurri genes are not present in cnidarians or sponges, indicating that not all BMP signaling uses Schnurri. Alternatively, functional Schnurri homologs in early-diverging metazoans may be sufficiently diverged in sequence that simple sequence comparisons fail to identify them.

WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

How important are Schnurri proteins for BMP signaling in vertebrate cells? The suggestion that Schnurris play roles as scaffolding proteins, interacting with a variety of transcription factors to regulate many BMP targets, is testable. Schnurri loss-of-function analyses should yield critical information on this. The three Schnurri genes are coexpressed weakly and uniformly during Xenopus early embryogenesis (Durr et al.,2004; our unpublished observations), however, later patterns of spatial expression has not been explored. There is some evidence (reviewed in Wu,2002) suggesting that mouse Shns1 and 2 may be expressed more broadly and Shn3 may have a more restricted pattern, but his will require further analysis. However, redundancy between the three Shn genes will likely require complex knockout combinations. Therefore, multiple knockout/knockdown approaches in a variety of species such as mouse, Xenopus, and zebrafish will be necessary and informative.

Another area to be explored is the use of the recently developed methods of genome-wide analysis of transcription factor binding, most notably ChIP-on-chip and ChIP-seq methodologies. Application of these approaches to the binding patterns of the BMP R-Smads across the genome will likely yield valuable information on two major themes. First, this would be expected to provide a larger number of candidate genes for direct regulation by BMP signaling. And second, using de novo methods for identifying enriched sequence motifs, we may learn something about how Smads regulate BMP targets. For example, do the Smads mostly use GC-rich and SBE motifs to regulate targets, or is binding to transcription factor partners without direct Smad binding to DNA the more predominant mode of action of the Smads? This may seem like an odd question but a recent study, the first of this kind on the Smads, suggests this might be the case (Chen et al.,2008). ChIP-seq was used to examine the genome-wide binding patterns of 13 different transcription factors in mouse embryonic stem cells, including Smad1. The study concluded that Smad1 interacts with the genome primarily through composite Sox2-Oct4 sites in mES cells. As Oct4, Sox2, and Nanog also bound these sites in mES cells, it seems that Smad1 may bind these sequences as part of a complex with these other transcription factors, without binding DNA itself. It is not clear whether these researchers searched for GC-rich/SBE motifs in the regions bound by Smad1. If Smad1 does indeed bind DNA in other cell types primarily by means of protein–protein interactions, and not through SBE/GC-rich motifs, then the current dogma for how Smads recognize their genomic targets will require significant modification.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES

We thank Drs. Kavita Arora, Kohei Miyazono, Tom Schilling, Rob Steele, and Rahul Warrior referees for discussions and critical comments that substantially improved the manuscript. As the literature on TGFβ signaling field is vast, the authors would like to apologize to members of the TGFβ community for the inevitable insufficiencies in citing relevant studies herein.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. THE CANONICAL BMP SIGNALING CASCADE: FROM THE CELL MEMBRANE INTO THE NUCLEUS
  5. IT'S A SMAD WORLD: DNA RECOGNITION BY THE SMADs
  6. TRANSCRIPTION FACTOR PARTNERS FOR SMADS IN THE REGULATION OF BMP TARGET GENE EXPRESSION
  7. SCHNURRI, A TRANSCRIPTIONAL PARTNER WITH MULTIPLE PERSONALITIES
  8. SCHNURRI IS AN EVOLUTIONARILY CONSERVED REGULATOR OF BMP TARGET GENES IN VERTEBRATES
  9. WHERE DO WE GO FROM HERE? LOOKING INTO THE CRYSTAL BALL
  10. Acknowledgements
  11. REFERENCES