J Biol Chem, Vol. 275, Issue 9, 6075-6079, March 3, 2000

Smad6 Is a Smad1/5-induced Smad Inhibitor
CHARACTERIZATION OF BONE MORPHOGENETIC PROTEIN-RESPONSIVE ELEMENT IN THE MOUSE Smad6 PROMOTER^*

Wataru Ishida^§, Toshiaki Hamamoto, Kiyoshi Kusanagi, Ken Yagi, Masahiro Kawabata, Kazuhiko Takehara^§, T. Kuber Sampath^¶, Mitsuyasu Kato, and Kohei Miyazono

From the  Department of Biochemistry, Cancer Institute of the Japanese Foundation for Cancer Research (JFCR), and Research for the Future Program, Japan Society for the Promotion of Science, 1-37-1 Kami-ikebukuro, Toshima-ku, Tokyo 170-8455, Japan, the ^§ Department of Dermatology, Kanazawa University School of Medicine, Takara-machi 13-1, Kanazawa 920-8640, Japan, and ^¶ Creative BioMolecules, Inc., Hopkinton, Massachusetts 01748

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Smad6 is an inhibitory Smad that is induced by bone morphogenetic proteins (BMPs) and interferes with BMP signaling. We have isolated the mouse Smad6 promoter and identified the regions responsible for transcriptional activation by BMPs. The proximal BMP-responsive element (PBE) in the Smad6 promoter is important for the transcriptional activation by BMPs and contains a 28-base pair GC-rich sequence including four overlapping copies of the GCCGnCGC-like motif, which is a binding site for Drosophila Mad and Medea. We generated a luciferase reporter construct (3GC2-Lux) containing three repeats of the GC-rich sequence derived from the PBE. BMPs and BMP receptors induced transcriptional activation of 3GC2-Lux in various cell types, and this activation was enhanced by cotransfection of BMP-responsive Smads, i.e. Smad1 or Smad5. Moreover, direct DNA binding of BMP-responsive Smads and common-partner Smad4 to the GC-rich sequence of PBE was observed. These results indicate that the expression of Smad6 is regulated by the effects of BMP-activated Smad1/5 on the Smad6 promoter.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Members of the transforming growth factor- (TGF-)¹ superfamily, including TGF-s, activins, and bone morphogenetic proteins (BMPs), bind to type II and type I serine/threonine kinase receptors and transduce intracellular signals through Smad proteins (1, 2). Smads can be classified into three subclasses based on their structure and function, i.e. receptor-regulated Smads (R-Smads), common-partner Smads (Co-Smads), and inhibitory Smads (I-Smads). Among the various R-Smads in mammals, Smad1, Smad5, and presumably Smad8 are activated by BMP receptors, whereas Smad2 and Smad3 are activated by TGF- or activin receptors. R-Smads then form complexes with the Co-Smad Smad4 and translocate into the nucleus, where they regulate transcription of various target genes. I-Smads, including Smad6 and Smad7, stably interact with activated type I receptors and compete with R-Smads for activation by the receptors (3-5). Smad6 has also been reported to compete with Co-Smads for complex formation with R-Smads (6). Smad6 and Smad7 inhibit BMP signaling (6, 7), whereas Smad7 is more potent than Smad6 in inhibiting the effects of TGF- and activin (6, 8, 9).

R-Smad-Co-Smad complexes regulate transcription of target genes in the nucleus by directly binding to DNA, interacting with other DNA-binding proteins, and recruiting transcriptional coactivators or corepressors (2). Smad3 and Smad4 bind to the CAGA box containing "GTCT" and its complementary "AGAC" sequences through their N-terminal Mad homology 1 domain (10, 11). In contrast, Mad, a Drosophila R-Smad activated by a Drosophila BMP-like factor Decapentaplegic, and Medea, a Drosophila Co-Smad, have been shown to bind to the GCCGnCGC motif in the vestigial and tinman gene promoters (12, 13). Smads interact with various transcription factors, including FAST1 (14), c-Jun-c-Fos (15), and PEBP2/Runx (16), and exert specific effects in various cell types. Moreover, Smads recruit transcriptional coactivators p300/CBP, which activate transcriptional responses through acetylation of core histones (2). Under certain conditions, transcriptional corepressors, e.g. TGIF and c-Ski, bind to the Smad complexes and inhibit transcription by recruiting histone deacetylases (17, 18).

Signaling by the TGF--like factors is regulated through multiple mechanisms. The activity of BMPs is regulated extracellularly by antagonists that directly bind to BMPs, e.g. Chordin, Noggin, and Cerberus family proteins (19, 20). The activity of Nodal is regulated by Antivin and Lefty, which are members of the TGF- superfamily (21, 22). At the receptor level, a pseudoreceptor, BAMBI, has been shown to inhibit the signaling by TGF-, activin, and the BMPs (23). In the cytoplasm, I-Smads play important roles in regulating Smad signaling. In addition, transcriptional corepressors TGIF, c-Ski, and SnoN associate with Smad2/3 and Smad4 in the nucleus, rendering cells resistant to the effects of TGF- (17, 18, 24-26). Importantly, expression of some of these signal-regulating molecules is controlled through the activities of TGF--like factors themselves (23, 25, 27).

Expression of I-Smads is regulated by various extracellular stimuli, including peptide growth factors (28). Interferon- induces the expression of Smad7; cells treated with interferon- thus become resistant to the effect of TGF- (29). I-Smads are also induced by TGF- superfamily signaling (4, 28, 30, 31), and I-Smads may thus terminate TGF-/BMP signaling through a negative feedback loop after certain lag time periods. However, it remains to be determined whether R-Smads-Co-Smads are directly involved in the induction of I-Smads or whether other non-Smad pathways participate in this process. To elucidate the molecular mechanism of the autocrine switch-off signaling by Smad6, we have characterized the promoter region of the mouse Smad6 gene. We have also generated a BMP-responsive reporter construct using the Smad6 promoter. Our results demonstrated that Smad6 is induced by Smad1/5 through the activation of the BMP receptors.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Isolation of the Mouse Smad6 Promoter-- The 5'-region of the mouse Smad6 gene was cloned from a 129SV mouse genomic library (Stratagene) using a 701-bp SacI-SacI fragment of the mouse Smad6 cDNA (3) corresponding to a part of the 5'-untranslated region and N-terminal region of the mouse Smad6 as a probe. Cloning and purification of the Smad6 promoter was performed as described (32).

Cell Culture and DNA Transfection-- P19 mouse embryonal carcinoma cells (gift of T. Momoi), human HaCaT keratinocytes (gift of N. E. Fusenig), Mv1Lu mink lung epithelial cells, C2C12 mouse mesenchymal cells, and COS7 cells (American Type Culture Collection) were cultured as described (7, 18, 33). For transient transfection, 60-80% confluent cells in 6-well plates were transfected using FuGENE6 transfection reagent (Roche Molecular Biochemicals). The original constructions of the constitutively active type I receptors and Smads have been described (3, 34).

Construction of Luciferase Promoters and Luciferase Assay-- Reporters containing the promoter region of the mouse Smad6 were generated by inserting the DNAs into pGL2-Basic (Promega). Reporters containing three copies of the GC-rich sequences flanked by two adenine residues were constructed by inserting the DNAs into 90COLXLUC (gift of S. Harada) containing the core promoter of the mouse collagen X (COLX) gene in pGL2-Basic (35). Oligonucleotides containing the wild-type or mutant GC-rich sequences were prepared by polymerase chain reaction. After transient transfection of DNAs, cells were incubated for 24 h, and luciferase activity in the cell lysates was determined using a luminometer. Luciferase activities were normalized to sea-pansy luciferase activity under control of the thymidine kinase promoter.

Gel Mobility Shift Assay-- Gel mobility shift assay was performed as described previously (34, 36). Briefly, whole-cell extracts were prepared from COS7 cells transfected with FLAG-tagged Smad5, Myc-tagged Smad4, and a constitutively active BMP type IB receptor (ALK6QD) expression constructs. For supershift analysis, antibodies were added to the whole-cell lysates. Whole-cell lysates were added to a premix solution containing poly(dI-dC), and probes were subjected to [-³²P]dCTP Klenow labeling. The final concentration of NaCl in the samples was adjusted to 50 mM using hypotonic and hypertonic lysis buffers. Complexes were then resolved on a 4% polyacrylamide gel, and analyzed by autoradiography.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Cloning of the Mouse Smad6 Promoter-- We obtained three overlapping clones covering the 5' part of the mouse Smad6 gene, using a part of the mouse Smad6 cDNA as a probe. Fragments obtained from the 9.5-kilobase pair region of the 5'-untranslated region of the mouse Smad6 gene (Fig. 1A) were inserted into the pGL2-Basic luciferase reporter and examined for responsiveness to BMP signaling using the BMP type II receptor (BMPR-II) and a constitutively active form of BMP type IB receptor (ALK6QD). We used P19 cells to determine the responsiveness to BMPs, because this cell line is highly transfectable and demonstrates up-regulation of the Smad6 mRNA in response to BMPs.² Similar to the mouse Smad7 promoter recently reported by Nagarajan et al. (37), a TATA box could not be found in the 5'-untranslated region of the Smad6 gene. However, multiple Sp1 sites were observed in this region (data not shown), which is common in most promoters lacking a TATA box (38). Moreover, we observed a basal transcription activity, which was more than 100-fold compared with the values obtained from the parental pGL2-Basic vector, indicating the presence of a transcription initiation site in this region. As shown in Fig. 1B, decreases in luciferase activity were observed after deletions of nucleotides 3213 to 2384, 2383 to 1912, and 1911 to 1697 (numbers from the ATG start codon). Among them, the decrease in luciferase activity was most prominent after deletion of nucleotides 1911 to 1697. These results suggest that the Smad6 promoter may contain multiple BMP-responsive regions, one of which resides in the nucleotide 1911 to 1697 region.

View larger version (34K): [in this window] [in a new window]   Fig. 1.   Cloning of the mouse Smad6 promoter. A, cloning of mouse genomic fragments encoding the Smad6 promoter is shown. A 701-bp fragment from the mouse Smad6 cDNA was used as a probe for screening a mouse genomic library. B, deletions of the Smad6 promoter reveal multiple BMP-responsive regions. Fragments of the mouse Smad6 gene promoter were inserted into pGL2-Basic, and transcriptional activation activity was determined by luciferase assay (Luc) in the presence of BMPR-II and ALK6QD (0.5 µg of DNA for each) using P19 cells. Results are shown as fold-induction compared with the control cultures in the absence of BMPR-II and ALK6QD. Error bars represent standard deviations. *, p = 0.01; **, p = 0.005.

Proximal BMP-responsive Element (PBE) in the Smad6 Promoter-- Because the proximal region of the Smad6 promoter containing nucleotides 1911 to 1697 responded to BMPs, we further examined the structure of this region. Nucleotide sequencing of PBE revealed the presence of a 28-bp GC-rich sequence (Fig. 2A), which contained three overlapping GCCGnCGC-like sequences in the 5'-portion and another GCCGnCGC-like sequence in the 3'-portion. The GCCGnCGC motif has been shown to bind Drosophila Mad and Medea (12, 13). Moreover, our current study revealed that this motif binds to mammalian Smad1 and Smad4 and that artificial promoter reporter constructs containing multiple repeats of the GCCGnCGC motif respond to BMPs (39). Because there are no Smad binding motifs except a CAGA box, which is a binding site for Smad3/Smad4 (10, 11), we generated various reporter constructs with or without the GC-rich sequence and CAGA box and examined responsiveness to BMPs using BMPR-II and ALK6QD in P19 cells (Fig. 2B). Deletion of the GC-rich sequence, but not the CAGA box, resulted in a slight decrease in transcriptional activity.

View larger version (32K): [in this window] [in a new window]   Fig. 2.   Characterization of PBE in the Smad6 promoter. A, nucleotide sequence of the Smad6 promoter PBE is shown. The 28-bp GC-rich sequence is indicated by a filled box and the GCCGnCGC-like motifs are shown by arrows. The CAGA box is shown as an open box. B, transcriptional activation was examined using reporter constructs without the GC-rich sequence or the CAGA box. Luciferase assay (Luc) was performed in the presence of BMPR-II and ALK6QD (0.5 µg of DNA for each) using P19 cells, and the results are shown as fold-induction compared with the control cultures in the absence of BMPR-II and ALK6QD. Results are shown from one experiment repeated three times with essentially the same results.

The GC-rich Sequence Is Responsible for Responsiveness to BMP Stimulation-- To further study the functional importance of the GC-rich sequence in the Smad6 PBE, we constructed a heterologous luciferase reporter termed 3GC2-Lux, containing three tandemly repeated GC-rich sequences fused to the COLX promoter inserted into pGL2-Basic. Constitutively active forms of BMP type I receptors, including ALK2 (ALK2QD), ALK3/BMPR-IA (ALK3QD), and ALK6/BMPR-IB (ALK6QD), only weakly induced transcriptional activation of 3GC2-Lux in P19 cells (Fig. 3A). Similarly, BMP-responsive R-Smads, i.e. Smad1 and Smad5, induced weak transcriptional activation. However, in the presence of Smad1 or Smad5, the BMP type I receptors strongly induced transcriptional activation of 3GC2-Lux. In contrast, the constitutively active forms of TGF- or activin type I receptors (TR-I/ALK5TD or ActR-IB/ALK4TD, respectively) did not activate the luciferase reporter in the presence or absence of the TGF-/activin-responsive R-Smads, Smad2 or Smad3.

View larger version (32K): [in this window] [in a new window]   Fig. 3.   The GC-rich sequence is a BMP-responsive element. AD, effects of constitutively active type I receptors (ALK 2QD, 3QD, 6QD, 4TD, and 5TD; 0.5 µg of DNA) and R-Smads (Smad 1, 2, 3, and 5; 0.3 µg of DNA) on 3GC2-Lux were studied in P19 cells (A), HaCaT keratinocytes (B), Mv1Lu cells (C), and C2C12 cells (D). E and F, effects of OP-1/BMP-7, activin, and TGF- on 3GC2-Lux were examined in the presence and absence of R-Smads (0.3 µg of DNA) in P19 cells (E) or Mv1Lu cells (F). In E and F, +, ++, and +++ represent 50, 150, and 500 ng/ml for BMP7, 0.25, 0.75, and 2.5 ng/ml for TGF-, and 10, 30, and 100 ng/ml for activin, respectively.

Similar responsiveness to BMP type I receptors and BMP-responsive R-Smads was observed in other cell types, including HaCaT human keratinocytes (Fig. 3B), Mv1Lu mink lung epithelial cells (Fig. 3C), and C2C12 mouse mesenchymal cells (Fig. 3D). ALK2QD and ALK6QD were more potent in the transcriptional activation than ALK3QD in most cell types, although ALK3QD was more potent than the other BMP type I receptors in HaCaT cells (Fig. 3B). In addition, Smad5 was more potent in the transcriptional activation than Smad1 in most experiments, even though expression levels of the Smad1 and Smad5 proteins were comparable when determined by immunoblotting (data not shown).

We also examined the effects of various ligands on transcriptional activation of 3GC2-Lux in P19 cells (Fig. 3E). OP-1/BMP-7 induced transcriptional activation of 3GC2-Lux, which was dramatically enhanced by transfection of Smad1 or Smad5. Smad5 was more potent than Smad1 in the transcriptional activation, similar to the results using constitutively active type I receptors. In contrast to OP-1/BMP-7, TGF- or activin did not induce transcriptional activation in the presence and absence of Smad2 or Smad3.

Interestingly, 3GC2-Lux was activated by TGF- in Mv1Lu cells (Fig. 3F), whereas it was not activated by the constitutively active form of TR-I/ALK5 (ALK5TD) (Fig. 3C) or by Smad2/3 (Fig. 3F). Moreover, Smad1/5 enhanced the transcriptional activity of TGF- in mink lung cells. Transcriptional activation of 3GC2-Lux by TGF- was not observed in other cell types examined in the present study (data not shown). These results suggest that TGF- may bind to a type I receptor other than ALK5 in mink lung cells and activate 3GC2-Lux through Smad1/5 but not through Smad2/3. In agreement with this notion, TGF- was shown to activate Smad1 in certain cell types (40). We have also found that TGF- binds to ALK1 in human umbilical endothelial cells and activates Smad1 and Smad5.³

We next mutated the GC-rich sequence in 3GC2-Lux (Fig. 4A) and compared their transcriptional responses with those of the wild-type 3GC2-Lux. Mutation in the GCCGnCGC-like sequence in the 5' portion of PBE (3GC2 mut1-Lux) resulted in the loss of transcriptional activation by BMP stimulation in the presence and absence of Smad1/5 (Fig. 4B). In contrast, mutation in the GCCGnCGC-like sequence in the 3'-portion of PBE (3GC2 mut2-Lux) rather enhanced the transcriptional activation by BMP stimulation compared with the wild-type 3GC2-Lux. These results indicate that the GCCGnCGC-like sequence in the 5' portion of PBE is important for the BMP-responsiveness of PBE.

View larger version (42K): [in this window] [in a new window]   Fig. 4.   The GCCGnCGC-like motif in the 5' portion of PBE is responsible for the responsiveness to BMP signaling. A, structures of wild-type (wt) and mutant (mut) 3GC2-Lux are shown. Nucleotide mutations introduced into 3GC2 mut1-Lux and 3GC2 mut2-Lux are shown by lowercase underlined letters. B, effects of ALK6QD (0.5 µg of DNA) and Smad1/5 (0.3 µg of DNA) on wild-type or mutant 3GC2-Lux were studied in P19 cells. C, DNA binding of Smad5 and Smad4 to the GC-rich sequence is shown. A gel mobility shift assay was performed using whole-cell lysates from the cells transfected with or without FLAG-tagged Smad5 (F-Smad5), Myc-tagged Smad4 (M-Smad4), and ALK6QD. Wild-type (wt) or 3GC2 mutant (mut1 or mut2) DNAs were used as probes. Supershifts of the bands were examined in the presence of FLAG (F), Myc (M), or Smad4 (S) antibody (Ab).

DNA Binding of Smad5 and Smad4 to the GC-rich Sequence-- We then examined the direct DNA binding of Smad5 and Smad4 to the GC-rich sequence of PBE. As shown in Fig. 4C, Smad5 or Smad4 alone bound to DNA only weakly. When both Smad5 and Smad4 were transfected into COS7 cells and stimulated by cotransfection with ALK6QD, they formed DNA-binding complexes in a gel mobility shift assay (Fig. 4C, open arrowhead). Incubation of the gel shift mixture with FLAG antibody toward Smad5, or Myc or Smad4 antibody to Myc-Smad4, resulted in supershifts of DNA-binding complexes (Fig. 4C, closed arrowheads). Binding of Smad5 was stronger than binding of Smad1 (data not shown), consistent with the higher transcriptional activity of Smad5 over Smad1 on 3GC2-Lux. Smad5 and Smad4 bound to the 3GC2 mut2 DNA but not to the mut1 DNA. Novel bands were observed when the 3GC2 mut1 DNA was used as a probe, but these bands were not significantly supershifted by the antibodies to Smad5 or Smad4. Moreover, binding of Smad5 and Smad4 to the wild-type 3GC2 probe was prevented by the unlabeled 3GC2 wild-type or mut2 DNA but only weakly by the mut1 DNA (data not shown). These results suggest that BMP-responsive R-Smads and Smad4 bind to the GCCGnCGC-like motif preferentially at the 5' portion in the GC-rich sequence, which may be crucial for efficient transcriptional activation of the Smad6 promoter.

Signaling of the members of the TGF- superfamily is negatively regulated by multiple mechanisms. These regulatory molecules are often induced by the TGF--like factors themselves. Although Smads are the central molecules in the TGF- signaling pathways, it has not been fully determined whether such negative feedback regulation occurs as a direct effect of Smads or indirectly through other non-Smad pathways. Recently, expression of the mouse Smad7 has been reported to be induced by direct binding of Smad3 activated by TGF-/activin receptors and Smad4 to the CAGA box (37). Our present findings revealed that the Smad6 mRNA is directly induced by Smad1/5 activated by the BMP receptors. Because Smad6 preferentially inhibits the effects of BMPs, our results indicate that Smad6 terminates BMP signaling of Smad1/5-Smad4 through a negative feedback loop.

	ACKNOWLEDGEMENTS

We thank Dr. T. Momoi for P19 cells, Dr. N. E. Fusenig for HaCaT cells, Dr. S. Harada for 90COLXLUC, and Yasufumi Yuuki, Yumi Sasaki, and Aki Hanyu for technical help.

	FOOTNOTES

^* This study was supported by grants-in-aid for scientific research from the Ministry of Education, Science, Sports and Culture of Japan.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

To whom correspondence should be addressed. Tel.: 81-3-5394-3866; Fax: 81-3-3918-0342, E-mail: mit-ind@umin.ac.jp.

² W. Ishida, unpublished data.

³ T. Imamura, unpublished data.

	ABBREVIATIONS

The abbreviations used are: TGF-, transforming growth factor-; BMP, bone morphogenetic protein; BMPR, BMP receptor; bp, base pair(s); Co-Smad, common-partner Smad; COLX, type X collagen; I-Smad, inhibitory Smad; OP, osteogenic protein; PBE, proximal BMP-responsive element; R-Smad, receptor-regulated Smad.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES