BIOLOGICAL OVERVIEW

Identification of schnurri occurred almost simultaneously in three research groups (Arora, 1995, Grieder, 1995 and Staehling-Hampton, 1995). The search was on for downstream targets of DPP. A fly mutant for Decapentapletic receptors, either Thick veins or Punt, develops a hole in the dorsal surface of the cuticle. Such mutant flies are said to have a defect in dorsal closure. One would expect that transcription factors induced by receptor signaling would have the same mutant phenotype as the receptor mutants, known as the dorsal-open phenotype.

schnurri was isolated in an examination of uncharacterized dorsal-open mutations (Nüsslein-Volhard, 1984, Arora, 1995 and Greider, 1995). For example, one laboratory explored the role of dorsal-open mutations in the expression of labial, a target of the dpp signaling pathway. One of the unidentified mutants, schnurri, showed an interaction with dpp, a shortening of wing vein length, when placed in combination with a single mutant allele of dpp (Staehling-Hampton, 1995). Such genetic interactions often result in the identification of genes functioning in the same biochemical pathway.

One target of schnurri is bagpipe expression in the mesoderm. In schnurri mutants, bagpipe expression is reduced and visceral mesoderm development is abnormal. DPP, acting through its mesodermal receptors, induces bagpipe through schnurri. Thus schnurri mediates the action of DPP on mesodermal cells (Staehling-Hampton, 1995).

punt and schnurri function in somatic cells of the testis; they regulate a signal from somatic to germline cells that restricts proliferation of committed progenitors in the germline. In the testis, germ line stem cells and their primary and secondary spermatogonial cell progeny are surrounded by two somatic cyst cells. These two cyst cells, produced by asymmetric division of cyst progenitor cells, completely envelop the primary spermatogonial cell, forming a cyst. Cyst cells do not proliferate, but continue to envelop the germ cells throughout spermatogenesis. The primary spermatogonial cell is the mitotic founder of a cyst of secondary spermatogonial cells. There are four mitotic divisions, each with incomplete cytokinesis, producing 16 interconnected spermatogonial cells. The spermatogonial cells then undergo premeiotic S-phase, becoming spermatocytes and entering a prolonged G2 period and increasing in volume 25-fold. Subsequently, the spermatocytes undergo meiosis and spermatid differentiation. Progression through this sequence is accompanied by progressive displacement of the cyst through the tubular testis (Matunis, 1997).

By screening for mutants in which daughter germline cells fail to stop dividing, it was found that schnurri and punt are required to limit transient amplification of germ cells. Producing similar phenotypes to bag of marbles (bam) and benign gonial cell neoplasm (bgcn), mosaic analysis demonstrates that punt and schnurri act within the somatic cyst cells that surround germ cells, rather than in germ cells. shn mutant cyst cells are morphologically normal and appropriately express cyst cell markers. Thus, cyst cells themselves are not deleteriously affected by loss of shn function, suggesting that the principal role of shn function in cyst cells is to regulate the soma-to-germline signal. bam and bgcn are candidates for germ cell intrinsic components of this pathway and are likely be the targets of the punt and sch signal. Punt and Schnurii are components of a Dpp signal transduction pathway in other differentiation events. But in the testis system, a Dpp signal initiating from stem cells is probably not involved in triggering the punt/sch pathway, since a reporter gene that mimics Dpp expression shows no expression within the testis, and loss of Dpp function does not lead to overproliferation. Although the TGF-beta signal from germline cells is not known, there are two other known TGF-beta family members in Drosophila: screw and 60A. These are currently being examined to elucidate their role in this system. Thus, a signal relay operating in somatic cells regulates cell proliferating in the germline stem cell lineage (Matunis, 1997).

Decapentaplegic is crucial for embryonic patterning and cell fate specification in Drosophila. Dpp signaling triggers nuclear accumulation of the Smads Mad and Medea, which affect gene expression through two distinct mechanisms: direct activation of target genes and relief of repression by the nuclear protein Brinker (Brk). Schnurri (Shn) has been implicated as a co-factor for Mad, based on its DNA-binding ability and evidence of signaling dependent interactions between the two proteins. A key question is whether Shn contributes to both repression of brk as well as to activation of target genes. During embryogenesis, brk expression is derepressed in shn mutants. However, while Mad is essential for Dpp-mediated repression of brk, the requirement for shn is stage specific. Analysis of brk;shn double mutants reveals that upregulation of brk does not account for all aspects of the shn mutant phenotype. Several Dpp target genes are also expressed at intermediate levels in double mutant embryos, demonstrating that shn also provides a brk-independent positive input to gene activation. Shn-mediated relief of brk repression establishes broad domains of gene activation, while the brk-independent input from Shn is crucial for defining the precise limits and levels of Dpp target gene expression in the embryo (Torres-Vazquez, 2001).

Genetic evidence implicates both Shn and Mad in dpp-dependent repression of brk. In the wing disc, cells that lack Mad or shn ectopically express brk and fail to activate the Dpp-responsive genes optomotor-blind, vestigial, spalt and Dad. Abolition of shn or Mad activity results in upregulation of brk in the embryo and in the absence of shn ectopic Dpp cannot suppress brk expression. Since Shn and Mad interact directly, an attractive hypothesis is that a Shn/Mad complex is involved in the Dpp-dependent repression of brk. It has recently been suggested that Dpp signaling bifurcates downstream of Mad/Med into a Shn-dependent pathway, leading to brk repression and a Shn-independent pathway that triggers gene activation. According to this model, Shn acts primarily as a dedicated repressor that switches Mad from a transcriptional activator to a transcriptional repressor on the brk promoter. However several lines of evidence from this study are incompatible with such an interpretation (Torres-Vazquez, 2001).

A strong argument that shn has additional roles beyond brk repression comes from the fact that simultaneous loss of brk and shn activity results in a phenotype that is distinct from that of brk-null animals. If the sole function of shn is to mediate brk repression, then shn activity should be redundant in a brk mutant background. However, both at the overt phenotypic level as well as in the regulation of individual target genes, brk;shn double mutants display defects consistent with lower levels of Dpp signaling, compared with embryos that lack brk alone. These results indicate that shn participates in gene activation through brk-independent mechanisms as well. The finding that Shn is not obligately required to suppress brk transcription prior to germband elongation, while Mad is essential in this process, also argues against an exclusive role for Shn as a Mad co-repressor. In dpp- and Mad-null embryos, brk is upregulated at stage 8, while in embryos lacking shn function, derepression occurs approximately 3 hours later than the transition of brk regulation from maternal to zygotic control. Thus, brk transcription is insensitive to the absence of shn function at a time when it is responsive to Dpp and Mad. This idea is reinforced by the fact that ectopic Dpp signaling (through a constitutively activated form of Tkv called TkvA) can repress brk transcription at stage 5/6 in both wild-type and shn- animals, but not in Mad-null embryos. Collectively these data provide compelling evidence that refutes a model in which all aspects of the shn mutant phenotype result from derepression of brk transcription (Torres-Vazquez, 2001).

The unexpected result that at high levels TkvA mediates activation of brk promoter, while at low levels it causes repression reveals a possible mechanism by which Shn contributes to Mad activity. One explanation for these concentration-dependent effects of TkvA could be that the default mode of Mad action is transcriptional activation, and interaction with a co-repressor (perhaps present in limiting amounts) is crucial to bring about repression. Cells that receive very high levels of signaling could experience 'squelching', owing to excess nuclear Mad that binds to the brk promoter without recruitment of the co-repressor, thus promoting activation rather than repression. Supporting this idea, injection of TkvA into embryos that lack Mad does not induce either brk activation or repression. The increased frequency of ectopic brk expression in shn- embryos could indicate that Shn stabilizes a Mad/co-repressor complex on the brk promoter. It is worth bearing in mind that even in shn- embryos, ectopic activation did not occur independent of brk repression in the peripheral cells. Thus, it appears that Shn does not determine whether Mad acts as an activator or as a repressor, but may promote its interaction with other factors that determine the polarity of Mad transcriptional activity (Torres-Vazquez, 2001).

Analysis of Dpp-responsive gene expression in brk; shn double mutants has allowed an assessment the brk-independent input from shn to gene activation at different developmental stages in a range of tissues. Although it has not been demonstrated that each of these markers is a direct target of Dpp signaling, three categories of responses can be distinguished based on these studies. In the first group (class A), exemplified by dpp in the leading edge of the dorsal ectoderm, expression in the double mutant is indistinguishable from that in brk- embryos. Thus, shn contributes to class A gene expression primarily by relief of brk repression. Promoters belonging to class B include Dad and pnr in the dorsal ectoderm during germband extension. Expression of class B genes is downregulated in the double mutant compared with brk- embryos, but is equivalent to wild-type levels. It is inferred from this result that in the absence of Brk and Shn, Mad-mediated activation may be sufficient for expression within the normal domain, but cannot sustain the lateral expansion encountered in brk mutants. A third category of responses (class C) includes dpp and Ubx in the midgut, and sna in the primordia of the wing/haltere imaginal discs. Genes in this class show significantly reduced levels of expression in the double mutant, not only relative to brk- but also compared with wild-type animals. Class C promoters incorporate a brk-independent positive input from shn that is necessary for wild-type levels of expression. The inability of ectopic Dpp to induce sna expression in shn mutants demonstrates the essential nature of the requirement for Shn in activation of class C genes (Torres-Vazquez, 2001).

It is evident that repression of brk is crucial for expression of all three classes of genes described, and as such accounts for a significant part of the positive input from shn to gene activation. In addition, the data suggest that Mad and Shn contribute equally to repression of brk and regulation of class A genes. However, the fact that brk activity is only partially epistatic with respect to class B and C promoters, indicates that the majority of genes examined in this study integrate positive inputs from shn, as well as negative inputs from brk. The near wild-type expression of class B genes in double mutant embryos suggests that the brk-independent input from shn may be crucial at the margins of the expression domains and may be less significant in regions of the embryo that receive moderate to high levels of Dpp signaling. In contrast, the positive input from shn to class C targets appears to be important throughout the domain of expression. The observation that genes such as dCreb-A and Scr, which are repressed by dpp signaling, and which are also sensitive to loss of brk, raises the possibility that Dpp regulates their expression indirectly. In this event, the dpp target genes that mediate repression of dCreb-A and Scr would belong to classes A and C, respectively (Torres-Vazquez, 2001).

The partial restoration of dpp target gene expression in the double mutants relative to shn- embryos provides a basis for interpreting the cuticle phenotype. Homozygous brk;shn animals as well as brk;tkv mutants have an intermediate phenotype in that they show rescue of the dorsal closure defect observed in shn and tkv mutants, but they also display a reduced dorsal epidermis compared with brk-null embryos. Both dpp and pnr have been implicated in dorsal closure, which results from movement of the epidermal cells over the amnioserosa and their suturing at the midline. In light of this, the recovery of their expression in the dorsalmost ectodermal cells in the double mutants correlates well with the restoration of dorsal closure. Likewise, the compromised expression of dorsal ectodermal markers such as Dad and pnr in brk;shn embryos relative to brk null animals, provides molecular correlates for the ventralization observed in the double mutants (Torres-Vazquez, 2001).

The data presented in this study indicate that Shn can mediate both gene activation and brk repression in response to Dpp signaling. An important question is whether Shn has a Mad-independent role in activation. Shn contains a potential activation domain, and the human ortholog of Shn (PRDII-BF1) can elicit a 10-fold increase in gene expression in transfection assays. However, a Shn-Gal4 fusion protein does not activate transcription in yeast, and Shn is only marginally effective in stimulating a Dpp-responsive reporter in the absence of Mad in cell culture assays. Taken together these results suggest that Shn acts by promoting Mad binding to DNA and/or its interactions with the transcriptional machinery. There is ample precedent for such a mechanism, since several vertebrate DNA-binding Smad partners such as FAST1, OAZ, Mixer and Milk, do not have an innate ability to stimulate transcription, but potentiate gene activation by Smads in a pathway specific manner. A prediction from this data is that promoters of class B and class C genes are likely to contain binding sites for Shn as well as Mad, and that Shn increases Mad specificity by recruiting it to a subset of promoters that contain binding sites for both proteins. Analysis of gene expression in brk;tkv mutants demonstrates that for class B and class C genes Mad provides a greater brk-independent input compared with shn, consistent with the idea that Mad plays a primary role in Dpp-dependent gene activation and that shn facilitates Mad activity. Further support comes from the observation that deletion of Mad sites in the Ubx midgut enhancer had a more profound effect than abolition of Shn binding (Torres-Vazquez, 2001).

It has been shown that Mad interacts with Nejire (Nej), the Drosophila homolog of the co-activator p300/CREB binding protein (CBP). Reduction in nej activity affects the expression of ush, pnr and Ubx, and disrupts events that are dpp and shn-dependent, like tracheal migration and imaginal disc patterning. It is interesting to speculate that Shn may interact directly with Nej and stabilize complex formation between Mad/Medea and Nej (Torres-Vazquez, 2001 and references therein).

The requirement for Shn and Mad in both aspects of Dpp signaling suggests that Shn does not confer the ability to activate or repress transcription. It appears more likely that the activity of the Mad/Shn complex is modulated in a promoter specific fashion analogous to the mechanisms that convert Dl from an activator to a repressor. Similarly, the presence of binding sites for factors that bring co-repressors into proximity with Mad/Shn could permit inhibition of transcription at the brk promoter while target genes that lack these sites could be activated in the same cells. It has been shown that Smad4 interacts with the co-repressor TGIF and the co-activator CBP in a mutually exclusive manner. Thus, the ability to recruit co-activators as opposed to Smad co-repressors (such as cSki and SnoN), or more general transcriptional repressors like Groucho or CtBP, would be crucial to determining whether Dpp stimulation resulted in activation or repression of the target gene (Torres-Vazquez, 2001).

It is conceivable that in addition to repressing brk transcription, Shn and Mad could prevent residual Brk protein in the nucleus from binding to target gene promoters through steric hindrance or direct competition for common binding sites. Related anti-repression mechanisms have been postulated for Smad1 and Smad2 that interact with the transcriptional repressors Hoxc-8 and SIP1, respectively, triggering their dissociation from the osteopontin and X-Bra promoters. Although such a mechanism could potentially enhance the efficiency with which Shn and Mad antagonize brk activity, it does not account for the brk-independent input from shn observed in brk;shn embryos, since there is no Brk protein in the double mutants.

Despite the fact that shn transcripts are present from the precellular blastoderm stage onwards, loss of shn activity does not affect either brk repression or the expression of Dpp target genes until germband extension. Germline clonal analysis and ds-RNAi experiments indicate that the insensitivity of Dpp target gene expression to loss of shn during early embryogenesis is unlikely to result from perdurance of maternal message. Thus, the 'weakness' of the shn mutant phenotype may reflect a limited temporal requirement for shn in dpp signaling, rather than a lesser requirement for shn activity throughout development. The functional redundancy of shn during early patterning could be due to the presence of another protein that contributes a Shn-like activity to Dpp signal transduction. Alternatively, Mad activity alone could be sufficient for induction of early D/V patterning genes if they contain promoter elements that are more sensitive to Mad. It is also possible that the higher levels of nuclear Mad resulting from the synergy between Scw and Dpp in early embryogenesis renders the potentiation of Mad by Shn unnecessary. Finally, given the conserved nature of the BMP signal transduction pathway and the identification of Shn homologs in humans, frogs and worms, it is possible that Shn-like proteins in other systems potentiate Smad activity in an analogous manner (Torres-Vazquez, 2001).

GENE STRUCTURE

cDNA clone length - 7584 bp

Bases in 5' UTR -~200

Bases in 3' UTR - 1.2 kb

PROTEIN STRUCTURE

Amino Acids - 2528

Structural Domains

The Schnurri protein has 8 C2H2-type zinc fingers clustered in two pairs and one triplet, spread throughout the gene. Between the first and second pair is a region C-C-X 13-H-C domain which may form a single zinc finger. There is also a putative DNA binding domain, which may be regulated by phosphorylation (Arora, 1995, Grieder, 1995 and Staehling-Hampton, 1995).

A subset of zinc finger transcription factors contain amino acid sequences that resemble those of Krüppel. They are characterized by multiple zinc fingers containing the conserved sequence CX2CX3FX5LX2HX3H (X is any amino acid, and the cysteine and histidine residues are involved in the coordination of zinc) that are separated from each other by a highly conserved 7-amino acid inter-finger spacer, TGEKP(Y/F)X, often referred to as the H/C link.

Each 30-residue zinc finger motif folds to form an independent domain with a single zinc ion tetrahedrally coordinated beween an irregular, antiparallel, two stranded ß-sheet and a short alpha-helix. Each zinc finger of mouse Zif268 (which has three fingers) binds to DNA with the amino terminus of its helix angled down into the major groove. An important contact between the first of the two histidine zinc ligands and the phosphate backbone of the DNA contributes to fixing the orientation of the recognition helix. Although the two fingers of Drosophila Tramtrack interact with DNA in a way very similar to those of Zif268, there are important differences. Tramtrack has an additional amino-terminal ß-strand in the first of the three zinc fingers. The charge-relay zinc-histidine-phosphate contact of Zif268 is substituted by a tyrosine-phosphate contact. In addition, for TTK, the DNA is somewhat distorted with two 20 degree bends. This distortion is correlated with changes from the rather simple periodic pattern of amino base contacts seen in Zif268 and finger 2 of TTK (Klug, 1995 and references).

EVOLUTIONARY HOMOLOGS

In Caenorhabditis elegans, the DBL-1 pathway (named after the ligand DBL-1), a BMP/TGFß-related signaling cascade, regulates body size and male tail development. A new gene, sma-9, has been cloned that encodes the C. elegans homolog of Schnurri, a large zinc finger transcription factor that regulates dpp target genes in Drosophila. Genetic interactions, the sma-9 loss-of-function phenotype, and the expression pattern suggest that sma-9 acts as a downstream component and is required in the DBL-1 signaling pathway, and thus provide the first evidence of a conserved role for Schnurri proteins in BMP signaling. Analysis of sma-9 mutant phenotypes demonstrates that SMA-9 activity is temporally and spatially restricted relative to known DBL-1 pathway components. In contrast with Drosophila schnurri, the presence of multiple alternatively spliced sma-9 transcripts suggests protein isoforms with potentially different cell sublocalization and molecular functions. It is proposed that SMA-9 isoforms function as transcriptional cofactors that confer specific responses to DBL-1 pathway activation (Liang, 2004).

The longest conceptual SMA-9 sequence, based on the isolated cDNA clones, is 2170 amino acids (aa) in length and contains a Gln-rich N terminus, including several repeats of a QQQQL sequence of unknown function, and seven C2H2 zinc finger motifs at the C terminus clustered into two pairs and one triplet. The first pair of zinc fingers is located in the middle of the sequence, the second pair is near the C terminus and the triplet is between these two pairs. Like Shn, SMA-9 is rich in Ser and Thr. There are two acidic-residue-rich domains (ARD) that may correspond to transcriptional activation domains, one N-terminal to the zinc finger region and the other following the first pair of zinc fingers. The whole sequence contains four predicted NLSs, three at the N terminus and one at the C terminus, consistent with a function in the nucleus. Five S/TPKK motifs surround the zinc finger regions; these motifs are putative DNA-binding domains and may be regulated by phosphorylation (Liang, 2004).

A similarity search of GenBank reveals that sma-9 shares high sequence homology with a zinc finger transcription factor family that includes Drosophila Shn (BLAST E value=4e-24) and vertebrate Shn1 family members, human major histocompatibility complex-binding protein 1 (MBP1)/PRDII-BF1 (E value=2e-17) and mouse {alpha}A-crystallin-binding protein 1 ({alpha}A-CRYBP1; E value=8e-18). Similarities among them include the presence of multiple zinc fingers, NLS, ARD and S/TPKK motifs, as well as stretches of sequences rich in Gln and in Ser/Thr. In SMA-9, the first pair of zinc fingers has 77% identity to the second pair in Shn, 76% to the second pair in MBP1/ PRDII-BF1 and 74% to the second pair in {alpha}A-CRYBP1. Therefore, SMA-9, MBP1, {alpha}A-CRYBP1 and Shn may derive from a common ancestral gene, and this pair of zinc fingers may contribute to a conserved role for Shn proteins. The SMA-9 triplet of zinc fingers has 45% identity to the Shn triplet. This domain is absent from MBP1 and {alpha}A-CRYBP1, suggesting its elimination during vertebrate evolution or its acquisition in the fly-worm lineage. The SMA-9 second pair of zinc fingers has no similarity to the other family members, indicating a unique function in C. elegans. An alternative 70 aa C terminus is also unique to C. elegans (Liang, 2004).

The SHN protein is related to human transcription factors MBPI and MBPII, major histocompatibility complex-binding proteins that can also activate transcription from the HIV long terminal repeat (LTR). These proteins bind NFkappaB (Drosophila homologs Dorsal and Dorsal-related immunity factor) consensus binding sites in various promoters. (Arora, 1995, Grieder, 1995 and Staehling-Hampton, 1995).

SHN related protein PRDII/MBPI/HIV-EP1, a mammalian transcription factor, is one of a family of related zinc finger transcription factors that bind to NFkappaB-related sites. PRDII/MBPI/HIV-EP1 related proteins bind to sites in the major histocompatability complex class I genes, human interferon beta gene, beta2-macroglobulin gene, angiotensinogen gene, the alphaA-crystalin gene, alpha1-anti-trypsin gene, the recombination signal sequences of immunoglobulin and T-cell receptor gene and the HIV long terminal repeat. PRDII/MBPI/HIV-EP1 binding to NF-kappaB sites in the HIV-1 LTR can activate HIV-1 gene expression, however, no other functional role for the PRDII/MBPI/HIV-EP1 related factors has been demonstrated. PRDII/MBPI/HIV-EP1 mRNA is induced by viral infection and by growth factors, suggesting that the proteins have the potential to be regulated by growth factor or viral induced signaling (Staehling-Hampton, 1994 and references).

The cytokine TNF launches cascades of gene activation that control inflammation and apoptosis through NkappaFB and JNK/SAPK signal transduction pathways. A function for the zinc finger transcription factor kappa recognition component (KRC), a homolog of Schnurri, is described in regulating patterns of gene activation in response to proinflammatory stimuli. KRC overexpression inhibits while antisense or dominant-negative KRC enhances NFkappaB-dependent transactivation and JNK phosphorylation and consequently, apoptosis and cytokine gene expression. The effect of KRC is mediated through its interaction with the adaptor protein TRAF2, which intersects both pathways. KRC is a hitherto unrecognized participant in the signal transduction pathway leading from the TNF receptor to gene activation and may play a critical role in inflammatory and apoptotic responses (Oukka, 2002).

Schnurri-2 controls BMP-dependent adipogenesis via interaction with Smad proteins

Adipocyte differentiation is an important component of obesity, but how hormonal cues mediate adipocyte differentiation remains elusive. BMP stimulates in vitro adipocyte differentiation, but the role of BMP in adipogenesis in vivo is unknown. Drosophila Schnurri (Shn) is required for the signaling of Decapentaplegic, a Drosophila BMP homolog, via interaction with the Mad/Medea transcription factors. Vertebrates have three Shn orthologs, Shn-1, -2, and -3. This study reports that Shn-2^-/- mice have reduced white adipose tissue and that Shn-2^-/- mouse embryonic fibroblasts cannot efficiently differentiate into adipocytes in vitro. Shn-2 enters the nucleus upon BMP-2 stimulation and, in cooperation with Smad1/4 and C/EBPα, induces the expression of PPARγ2, a key transcription factor for adipocyte differentiation. Shn-2 directly interacts with both Smad1/4 and C/EBPα on the PPARγ2 promoter. These results indicate that Shn-2-mediated BMP signaling has a critical role in adipogenesis (Jin, 2006).

BMP-2 induces PPARγ expression and adipogenesis in C3H10T1/2 cells. The effects of BMP-2 on PPARγ2 promoter activity was analyzed using a PPARγ2 promoter-driven luciferase gene. Wild-type MEFs were transfected with the PPARγ2-Luc reporter, and adipocyte differentiation was induced in the presence or absence of BMP-2. The luciferase levels of wild-type cells increased 73% in the presence of BMP-2, whereas the luciferase levels of Shn-2^-/- cells were not affected by BMP-2 treatment. Thus, the PPARγ2 promoter is weakly responsive to BMP-2, and Shn-2 is required for this BMP responsiveness. The low degree of induction by BMP-2 could be due to an imbalance among the transcription factors and the promoter molecule in transfected cells (Jin, 2006).

To further examine the BMP responsiveness of the PPARγ2 promoter and the role of Shn-2, luciferase reporter assays were performed using wild-type MEFs transfected with the PPARγ2-Luc reporter and various combinations of expression plasmids for Smad1/4 and Shn-2. The PPARγ2 promoter contains C/EBP binding sites and its activity is enhanced by C/EBPα and C/EBPδ, and, therefore, the C/EBPα expression plasmid was also used. Without Smad1/4 or C/EBPα, the BMP-2-induced expression of luciferase was not observed, whereas BMP-2 enhanced luciferase expression about 2-fold in the presence of Smad1/4 and C/EBPα. When Smad1/4, C/EBPα, and Shn-2 were coexpressed together, higher BMP-2 responsiveness (3.6-fold) was observed. These results may support the speculation that the appropriate balance of these factors and the PPARγ2 promoter molecule is needed for BMP responsiveness. When Shn-2^-/- MEFs were used for similar experiments, BMP-2 enhanced luciferase expression only about 50% in the presence of Smad1/4 and C/EBPα. Exogenous expression of Shn-2 in the mutant cells significantly restored the BMP-2 responsiveness of the PPARγ2 promoter (4.5-fold). These results suggest that Shn-2, Smad1/4, and C/EBPα synergistically mediate the BMP-induced transactivation of the PPARγ2 promoter (Jin, 2006).

Smad3/4 bind to the 5'-AGAC-3' sequence, while Smad1 binds to GC-rich sequences. The mouse and human PPARγ2 promoter regions contain six AGAC sequences but not the GC-rich sequence. The AGAC sequence was also found at ten sites in the 1.2 kb promoter region of the mouse PPARγ1 gene. Among these six putative Smad binding sites in the mouse PPARγ2 promoter, four sites (sites 1, 2, 4, and 6) are conserved in the human PPARγ2 promoter. Mutant mouse PPARγ2-Luc reporters in which the AGAC sites were mutated, and the level of activation of the reporters by Shn-2, Smad1/4, ALK3QD, and C/EBPα was examined. The results indicate that three sites in the upstream region of the promoter (sites 1-3) are required for synergistic activation by these factors. Mutation of any of these three sites significantly reduced activation by Shn-2, Smad1/4, ALK3QD, and C/EBPα. The human PPARγ2 promoter lacks site 3 but has another Smad site further upstream of site 1. The presence of three Smad sites in this region of the mouse and human PPARγ2 promoters may support formation of a Smad1/4-Shn-2-C/EBPα complex to synergistically activate transcription (Jin, 2006).

Vertebrate Shn was originally identified as NF-κB site binding proteins, and the metal finger regions of Drosophila and Xenopus Shn recognize these specific DNA sequences. No NF-κB recognition sequence was found in the PPARγ2 promoter, but one sequence (5'-TCCCACCTCTCCC-3') at -94 to -82 partially resembles the Xenopus Shn binding sequence. However, mutation of this site did not affect the synergistic activation of the PPARγ2 promoter by Shn-2, Smad1/4, ALK3QD, and C/EBPα (Jin, 2006).

To examine whether Shn-2 directly binds to the PPARγ2 promoter, a DNA precipitation assay was performed. FLAG-Shn-2 was expressed in 293T cells, immunoprecipitated by anti-FLAG antibody, and eluted from the immunocomplex using FLAG peptide. The purified Shn-2 protein was mixed with ³²P-labeled PPARγ2 promoter fragments and precipitated with anti-Shn-2 antibody. The PPARγ2 promoter fragment was not detected in the immunocomplex. These results suggest that Shn-2 does not directly bind and is recruited by Smad proteins to the PPARγ2 promoter (Jin, 2006).

To investigate the interaction between Shn-2 and Smad1/4, coimmunoprecipitation assays of the exogenously expressed proteins were performed. 293T cells were cotransfected with plasmids to express FLAG-Shn-2, Myc-Smad1, and HA-Smad4, and lysates from transfected cells were immunoprecipitated with an anti-FLAG antibody. Myc-Smad1 and HA-Smad4 were coprecipitated with FLAG-Shn-2. When HA-Smad4 was deleted from this combination, FLAG-Shn-2 coprecipitated lesser amounts of Myc-Smad1. When Myc-Smad1 was deleted, HA-Smad4 was not coprecipitated with FLAG-Shn-2. These results suggest that Shn-2 interacts with the hetero-oligomers of Smad1 and Smad4 (Jin, 2006).

To determine which region of Shn-2 protein is responsible for interaction with Smad1, GST pull-down assays were performed. Two in vitro-translated Shn-2 fragments containing either the N- or C-terminal metal fingers (N1 and C1) bound to a GST-Smad1 resin, whereas the two fragments containing the central region of Shn-2 (HS and CP) exhibited only background and minor binding, respectively. Deletion of the metal finger regions from N1 and C1 abrogated the interaction with Smad1, suggesting that both metal finger regions are important for interactions with Smad1 (Jin, 2006).

The present study demonstrates that Shn-2 enters the nucleus upon BMP stimulation and plays an important role in adipocyte differentiation. The current study strongly suggests that BMP has a critical role in vivo. Upon BMP stimulation, Shn-2 is recruited to the PPARγ2 promoter via an interaction with Smad1. This is the first demonstration that Shn plays a role in vertebrate BMP/TGF-β/activin signaling. Shn-2 is required for efficient transcription of PPARγ2, possibly as a scaffold protein to form a ternary complex with Smad1/4 and C/EBPα. Interestingly, Evi-1, which is also a large protein containing two regions of metal fingers like Shn-2, interacts with and represses TGF-β/BMP-activated transcription through Smad proteins. Following TGF-β stimulation, Evi-1 and the associated corepressor CtBP are recruited to the target promoter. Thus, Shn-2 and Evi-1 interact with Smad proteins via their metal fingers and may stimulate and repress transcription by recruiting coactivator and corepressor, respectively (Jin, 2006).

Although vertebrate Shn proteins were originally identified as the NF-κB site binding proteins, the present study indicates that Shn-2 is recruited to the PPARγ2 promoter via an interaction with Smad1 and C/EBPα. This is similar to the recent report that Drosophila Shn forms a complex with Mad/Medea on the silencer element of the brinker (brk) gene to mediate Dpp-dependent brk gene silencing. The brinker silencer element contains three 5'-AGAC-3' sequences and two GC-rich sequences between them, to which Medea and Mad bind, respectively. However, the GC-rich sequence was not found between three 5'-AGAC-3' sequences in the PPARγ2 promoter. Therefore, more work is required to understand whether Smad1 in the Smad1/4 hetero-oligomers directly recognizes the DNA sequence in the PPARγ2 promoter. Interaction of Shn-2 not only with Smad1/4 but also with C/EBPα may support the idea that Shn-2 serves as a scaffold protein to form a ternary complex with various transcription factors to synergistically activate transcription. In fact, Shn-3 was reported to interact with c-Jun to activate IL-2 gene transcription (Jin, 2006).

Adipogenesis in vitro follows a highly ordered and well-characterized temporal sequence. In cultured cell models, initial growth arrest is induced by the addition of a prodifferentiative hormonal regimen and is followed by one or two additional rounds of cell division (clonal expansion). This process ceases upon induction of PPARγ2 and C/EBPα, which is concomitant with permanent growth arrest followed by expression of the fully differentiated phenotype. E2F1 induces PPARγ2 transcription during clonal expansion, whereas E2F4 represses PPARγ2 expression during terminal adipocyte differentiation. Interaction between Smad and E2F proteins has been shown for the Smad3-E2F4/5 complex mediating TGF-β-induced repression of c-myc. Therefore, Smad1/4-Shn-2 may also participate in E2F-dependent transcriptional regulation of PPARγ2 by directly interacting with E2F1/4. IFNγ decreases the expression of PPARγ2 in preadipocytes, but the mechanism remains to be elucidated. IFNγ induces the expression of Smad7, which prevents TGF-β receptor-mediated Smad3 phosphorylation. IFNγ may suppress PPARγ2 transcription by inducing Smad6, which then prevents BMP receptor-mediated Smad1 phosphorylation. FoxO1 is also known to regulate adipocyte differentiation. FoxO1 is induced in the early stages of adipocyte differentiation, and prevents adipose differentiation by upregulating multiple genes, including cell cycle inhibitors. Insulin leads to nuclear exclusion of FoxO1 and stimulates adipocyte differentiation. Smad proteins activated by TGF-β form a complex with FoxO proteins to turn on the growth-inhibitory gene p21Cip1 , and BMP-7 induces higher p21 expression than TGF-β1. By interacting with FoxO proteins, therefore, Smad1/4-Shn-2 may also regulate transcription not only of PPARγ2 but also of p21 during adipocyte differentiation (Jin, 2006).

Although more work is required to understand the role of Smad1/4-Shn-2 during adipocyte differentiation, identification of BMP signaling as the key regulatory pathway of adipogenesis in vivo may enable the development of drugs to affect this signaling pathway to suppress obesity and obesity-related diseases (Jin, 2006).

Regulation of adult bone mass by the zinc finger adapter protein Schnurri-3

Genetic mutations that disrupt osteoblast function can result in skeletal dysmorphogenesis or, more rarely, in increased postnatal bone formation. Schnurri-3 (Shn3), a mammalian homolog of the Drosophila zinc finger adapter protein Shn, is an essential regulator of adult bone formation. Mice lacking Shn3 display adult-onset osteosclerosis with increased bone mass due to augmented osteoblast activity. Shn3 was found to control protein levels of Runx2, the principal transcriptional regulator of osteoblast differentiation, by promoting its degradation through recruitment of the E3 ubiquitin ligase WWP1 to Runx2. By this means, Runx2-mediated extracellular matrix mineralization was antagonized, revealing an essential role for Shn3 as a central regulator of postnatal bone mass (Jones, 2006).

Runx2 protein levels are normal in Smurf1-deficient mice, suggesting another Nedd4 family E3 ligase might regulate Runx2 ubiquitination in vivo. Expression levels in mature osteoblasts revealed WWP1, another Nedd4 family E3 ubiquitin ligase, to be up-regulated during osteoblast differentiation in vitro. Coimmunoprecipitation experiments revealed association of Shn3 with WWP1 but not with Smurf1 or Smurf2. Additionally, endogenous WWP1 immunoprecipitates from differentiated MC3T3-E1 osteoblast cells reproducibly contained 95-kD Shn3 immunoreactive species (Jones, 2006).

An interaction between overexpressed Runx2 and WWP1 was detected in 293T cells. This was enhanced by Shn3 coexpression. The WWP1/Runx2 interaction likely occurs between the Runt domain of Runx2 and the WW domain of WWP1, because an in vitro translated fragment of WWP1 containing its WW domain coprecipitated with GST-Runx2. An interaction between endogenous WWP1 and Runx2 p65 isoform was also detected in the ROS 17/2.8 osteoblast cell line. WWP1 overexpression led to dose-dependent reductions in steady-state levels of Runx2 protein, which was reversed by brief treatment with the proteosome inhibitor MG132. Consistent with the effects of WWP1 overexpression on Runx2 protein levels, WWP1 overexpression also inhibited Runx2 function in luciferase assays, which could be potentiated by Shn3 coexpression. Finally, WWP1 promoted low levels of Runx2 ubiquitination when overexpressed in 293T cells. However, when coexpressed with Shn3, WWP1 synergized to promote Runx2 ubiquitination (Jones, 2006).

To further investigate the role of WWP1 in osteoblasts, levels of endogenous WWP1 were reduce using lentiviral delivered RNA interference (LV RNAi). RNAi-mediated knockdown of WWP1, but not the related E3 ligase Itch, led to pronounced up-regulation of BSP, OCN, and ATF4 mRNA. Runx2 mRNA levels were similar between green fluorescent protein (GFP) RNAi- and WWP1 RNAi-expressing cells, despite increased Runx2 protein levels (Jones, 2006).

To determine if Shn3-mediated repression of Runx2 function depended on WWP1, C3H10T1/2 cells were transduced with GFPi or WWP1i lentiviruses. In subsequent luciferase reporter assays, Runx2 function was enhanced in WWP1i cells. Moreover, Shn3-mediated Runx2 repression was largely WWP1-dependent (Jones, 2006).

Primary calvarial osteoblasts transduced with WWP1 or GFP RNAi-expressing lentiviruses was examined. Much like Shn3-deficient osteoblasts, WWP1 knockdown osteoblasts showed dramatic up-regulation of Runx2 targets BSP, OCN, and ATF4 at the mRNA level and elevated levels of the Runx2 protein, whereas Runx2 mRNA levels were comparable between WWP1 RNAi cells and controls. Furthermore, knockdown of WWP1 in these cultures led to increased numbers of mineralized matrix nodules (Jones, 2006).

Taken together, these data suggest a model in which the formation of a multimeric complex between Runx2, Shn3, and the E3 ubiquitin ligase WWP1 in mature osteoblasts inhibits Runx2 function. Shn3 is an integral adapter protein in this complex, as it enhances the ability of WWP1 to promote Runx2 polyubiquitination and proteasome-dependent degradation. RING domain E3 ubiquitin ligases are known to function in multimeric complexes; however, less is known about the regulation of HECT E3 ligases by protein-protein interactions. Additionally, future studies will be required to address the possibility that increased levels of Shn3/WWP1 substrates other than Runx2 may contribute to the osteosclerotic phenotypes observed in their absence (Jones, 2006).

Because the high-bone mass phenotype observed becomes more pronounced with age, it is proposed the Shn3 belongs to the small group of factors that regulate postnatal osteoblast activity. Compounds designed to block Shn3/WWP1 function may serve as therapeutic agents for the treatment of osteoporosis (Jones, 2006).

date revised: 20 December 2006

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.