Zebrafish Dkk3a Protein Regulates the Activity of myf5 Promoter through Interaction with Membrane Receptor Integrin α6b*

Background: Dkk3a regulates the promoter activity of myf5 through an unknown receptor. Results: Itgα6b is a high affinity Dkk3a receptor through its β-propeller domains. Conclusion: Dkk3a regulates the promoter activity of myf5 through interaction with receptor Itgα6b. Significance: This study discovered a novel mechanism for Dkk3a to regulate the promoter activity of myf5 during myogenesis. Myogenic regulatory factor Myf5 plays important roles in muscle development. In zebrafish myf5, a microRNA (miR), termed miR-3906 or miR-In300, was reported to silence dickkopf-3-related gene (dkk3r or dkk3a), resulting in repression of myf5 promoter activity. However, the membrane receptor that interacts with ligand Dkk3a to control myf5 expression through signal transduction remains unknown. To address this question, we applied immunoprecipitation and LC-MS/MS to screen putative membrane receptors of Dkk3a, and Integrin α6b (Itgα6b) was finally identified. To further confirm this, we used cell surface binding assays, which showed that Dkk3a and Itgα6b were co-expressed at the cell membrane of HEK-293T cells. Cross-linking immunoprecipitation data also showed high affinity of Itgα6b for Dkk3a. We further proved that the β-propeller repeat domains of Itgα6b are key segments bound by Dkk3a. Moreover, when dkk3a and itgα6b mRNAs were co-injected into embryos, luciferase activity was up-regulated 4-fold greater than that of control embryos. In contrast, the luciferase activities of dkk3a knockdown embryos co-injected with itgα6b mRNA and itgα6b knockdown embryos co-injected with dkk3a mRNA were decreased in a manner similar to that in control embryos, respectively. Knockdown of itgα6b resulted in abnormal somite shape, fewer somitic cells, weaker or absent myf5 expression, and reduced the protein level of phosphorylated p38a in somites. These defective phenotypes of trunk muscular development were similar to those of dkk3a knockdown embryos. We demonstrated that the secreted ligand Dkk3a binds to the membrane receptor Itgα6b, which increases the protein level of phosphorylated p38a and activates myf5 promoter activity of zebrafish embryos during myogenesis.

The Dickkopf (Dkk) 2 family consists of dkk1-4, and soggy (sgy), also named dkk3-related gene (dkk3l) (1). Dkks contain two conserved cysteine-rich domains: an N-terminal cysteinerich domain unique to Dkks and a C-terminal cysteine-rich domain related to the colipase fold. All Dkks share 37-50% protein identity, but they contain a variable linker region to separate the N-and C-terminal conserved cysteine-rich regions (1). In addition to a secretory form of Dkk, which contains a signal sequence, an intracellular form of Dkk was also reported (2). Dkk1, Dkk2, and Dkk4 inhibit Wnt signaling through binding to LRP5/6 by the transmembrane protein Kremen, resulting in internalization of LRP5/6, which prevents Wnt and Frizzled receptor from forming an active complex with LRP5/6 (3,4). Dkk2 can also activate the Wnt pathway in certain situations, depending on the cell type, the presence of Wnt ligands, and levels of LRP5/6 (5-7). Dkk3 is the most divergent member of the Dkk family in terms of DNA structure, protein function, and evolutionary impact (8). For example, it has been reported that Dkk3 does not physically interact with LRP5/6 or Kremen (3,9). Unlike the other Dkk members, Dkk3 does not regulate Wnt signaling in various activity assays, including Wnt-dependent secondary axis induction in Xenopus embryos and Wnt1/ Fz8 signaling in cultured cells (1,5,6). On the other hand, some studies have demonstrated that Dkk3 does regulate Wnt signaling, including weak inhibition of Wnt7A signaling in PC12 cells (10) as well as inhibition of Wnt activity in the osteocarcinoma Saos-2 cell line (11). Thus, whether Wnt signaling is regulated by Dkk3 is still controversial.
The dkk3 gene is identical to the reduced expression in immortalized cells gene (REIC) (12), and REIC/Dkk3 has been shown to act as a tumor suppressor or antioncogene because the expression of dkk3 is decreased in a variety of cancer cell lines (13)(14)(15)(16). Thus, human Dkk3 provides a new strategy against some human malignant tumors. For example, the * This work was supported by National Science Council, Republic of China Grant 100-2313-B-002-043-MY3. We are grateful to the staffs of TC5 Bio-Image Tools, Technology Commons, College of Life Science, NTU for help with the CLSM. □ S This article contains supplemental Figs. S1-S4. 1 To whom correspondence should be addressed: Inst. of Molecular and Cellular Biology, National Taiwan  forced expression of REIC/Dkk3 inhibited cell growth in HeLa and liver cancer cell lines (17). Furthermore, overexpression of dkk3 suppresses the growth of cells and the invasive capacity of cancer cell lines (11,15,18,19). Apart from the study of Dkk3 function in cell lines, dkk3 is reported to be expressed in many organs during mouse embryonic development, including neural epithelium, limb bud, bone, and heart, particularly in regions of epithelial-mesenchymal transformation (20). The dkk3 gene is also widely expressed in adult tissues with the highest levels found in the heart and brain (1). In zebrafish, two dkk3 genes, dkk3a/dkk3r (NCBI Reference Sequence NM_001159283.1) and dkk3b (NCBI Reference Sequence NM_001089545.1), have been reported (21). The dkk3a is strongly expressed in various neuronal structures of the head, whereas dkk3b is expressed mainly in the endocrine cells of the pancreas and brachial arches. Nevertheless, the function of Dkk3 during embryonic development is extremely limited. Myf5, a myogenic regulatory factor, plays important roles in the specification and differentiation of muscle cells during myogenesis. Recently, we found a novel microRNA, named miR-I300 (miR-3906), which binds to the 3Ј-untranslated region (UTR) of the mRNA of dkk3a (22). Knockdown of dkk3a with dkk3a morpholino (MO) resulted in the down-regulation of myf5 expression, suggesting that Dkk3a is involved in the upstream positive regulation of myf5. Furthermore, Hsu et al. (23) discovered that secreted Dkk3a can activate the phosphorylation of p38a and that phosphorylated p38a (p-p38a) causes stabilization of Smad4 through sumoylation. Stabilization of Smad4 enables formation of a Smad2/3a/4 complex, which then enters the nucleus to activate the myf5 promoter.
Although ␤-transducin repeat-containing proteins (24) and dynein light chain Tctex-1 (25) have been identified to associate with intracellular Dkk3 in human ovary, the membrane receptor that interacts with extracellular ligand Dkk3a to control zebrafish myf5 expression through p38a signal transduction is totally unknown. In this study, we found that Integrin ␣6b (Itg␣6b) is a membrane receptor of secretory Dkk3a and that the interaction of Dkk3a and Itg␣6b drives the downstream signal transduction to regulate myf5 promoter activity in somites during the development of zebrafish embryos.
Immunoprecipitation-Recombinant baculovirus containing plasmids pVL1392-dkk3a-FLAG and pVL1392-EGFP-FLAG were individually transfected into Sf21 cells. The recombinant proteins Dkk3a-FLAG and EGFP-FLAG were purified by anti-FLAG beads (Sigma). Immunoprecipitation of Dkk3a-FLAG and EGFP-FLAG with extracts isolated from zebrafish embryos at 16 hpf followed the protocols described in the Pierce Crosslink Immunoprecipitation kit (Thermo). The resultant immunoprecipitates were analyzed by SDS-PAGE followed by silver staining and in-gel digestion.
In-gel Digestion and LC-MS/MS Analysis-The immunoprecipitates shown on SDS-PAGE were cut into many gel pieces about 1 mm in length, and they were used to perform in-gel digestion and LC-MS/MS according to procedures described by Chiang et al. (26) using Mascot Distiller (Matrix Science, UK). The resultant MGF file was searched using the Mascot search engine (v2.2, Matrix Science) with the following conditions. 1) The protein database was set as Swiss-Prot. 2) Taxonomy was set as Danio rerio (zebrafish). 3) One trypsin missed cleavage was allowed. 4) The peptide mass tolerance was set at Ϯ0.5 Da, and the fragment mass tolerance was set at Ϯ0.5 Da. 5) Carbamidomethyl (Cys) was chosen as a fixed modification. 6) Oxidation (Met) and deamidation (Asn and Gln) were chosen as variable modifications.
Cell Surface Binding Assay-The conditional media used to culture Sf21 cells infected by recombinant baculovirus containing pVL1392-dkk3a-FLAG to produce extracellular Dkk3a-FLAG were concentrated about 40-fold using Amicon Ultra (Millipore). The concentration of secreted Dkk3a-FLAG was Dkk3a Regulates myf5 through Interaction with Itg␣6b determined by Western blotting using anti-FLAG antiserum. For cell surface binding experiments, HEK-293T cells were separately transfected with pCMV-itg␣6b-EGFP and pCMV (served as control) using Lipofectamine 2000 (Invit-rogen). After transfection for 48 h in a 6-well plate, cells were incubated with Dkk3a-FLAG proteins for 30 min at 4°C. We discarded the conditional media, washed cells with PBS, and fixed them with 4% paraformaldehyde. Then cells were stained with anti-FLAG-Alexa Fluor 555 (Cell Signaling Technology) and observed under confocal microscopy (Zeiss LSM 780).
In Vitro Transcription, Whole-mount in Situ Hybridization (WISH), Fluorescence Microscopy Observation, Western Blot Analysis, and Luciferase Activity Detection-The synthesis of capped mRNAs, labeling of probes for WISH, observation of embryos under fluorescence microscopy, extraction of total proteins from embryos, Western blotting, and quantitative measurement of luciferase activity followed the methods described by Hsu et al. (23).
Knockdown Experiments-The antisense MOs designed specifically to knock down itg␣6b (itg␣6b MO) and dkk3a (dkk3a MO) were CGGTTAAAGTCCTGTAAGATTCCAT and GAGGCTGAATCCGAGCAGAAACATG, respectively. The itg␣6b negative control MO (itg␣6b control MO) and dkk3a  negative control MO (dkk3a control MO) were designed as CGGTTAAAATCATGTAAAATTCAAT and GACGCT-CAATCCGACCACAAAGATG, respectively (the mutatedmismatched nucleotides are underlined). All MOs were prepared at a stock concentration of 1 mM and diluted to the desired concentration for microinjection into each embryo.

Itg␣6b Is a High Affinity Dkk3a
Receptor-To search for the membrane receptor bound by the secreted Dkk3a, we first produced recombinant Dkk3a fused with the FLAG reporter using the baculovirus expression system in insect cells (Fig. 1), and the protein function of dkk3a was not altered by adding a FLAG reporter at the C terminus (supplemental Fig. S1). This recombinant Dkk3a-FLAG was immunoprecipitated with membrane proteins extracted from zebrafish embryos at 16 hpf. The precipitated proteins were analyzed by SDS-PAGE with silver staining (Fig. 2). We further applied LC-MS/MS to reveal the putative proteins bound by Dkk3a-FLAG, including cell membrane proteins, mitochondrial proteins, endoplasmic reticulum proteins, and cytoskeleton proteins (Table 1). Among them, we focused on the membrane proteins and then screened them by an in vivo luciferase assay of myf5 promoter. Because Itg␣6b was capable of up-regulating myf5 promoter activity (see Fig. 4), we proposed that Itg␣6b might be a likely membrane receptor bound by Dkk3a. To further confirm this, we used a cell surface binding assay, which demonstrated that Dkk3a and Itg␣6b were co-expressed at the cell membrane of HEK-293T cells (Fig. 3). Data from the cell surface cross-linking experiment combined with cross-linking immunoprecipitation also demonstrated the high affinity of Itg␣6b for Dkk3a (see Fig. 5C). Based on this evidence, we suggested that Itg␣6b is a high affinity Dkk3a receptor.
Binding between Ligand Dkk3a and Membrane Receptor Itg␣6b Up-regulates the Promoter Activity of myf5-To further confirm the impact of binding between ligand Dkk3a and membrane receptor Itg␣6b on myf5 expression in zebrafish embryos, we injected several constructs. The luciferase activity of embryos injected with control plasmid pZmyf5 6.3R in which the reporter was driven by the upstream 6.3 kb of myf5 served as 100%. Compared with control embryos, the luciferase activity of embryos injected with excessive dkk3a mRNA and itg␣6b mRNA increased 223 and 217%, respectively (Fig. 4). Interestingly, when dkk3a and itg␣6b mRNAs were co-injected into embryos, the luciferase activity was up-regulated as high as 396%, which was 4-fold greater than that of control embryos (Fig. 4), suggesting that binding between ligand Dkk3a and membrane receptor Itg␣6b results in synergistic enhancement of myf5 promoter activity. In contrast, the luciferase activity of embryos injected with dkk3a MO alone, itg␣6b MO alone, dkk3a MO plus itg␣6b MO, dkk3a MO plus itg␣6b mRNA, or itg␣6b MO plus dkk3a mRNA was decreased to 49 -69% of control embryos (Fig. 4), suggesting that the binding between ligand Dkk3a and its receptor, Itg␣6b, results in up-regulation of myf5 promoter activity.
The ␤-Propeller Repeat Domains of Itg␣6b Are Key Segments Bound by Dkk3a to Regulate myf5 Promoter Activity-To determine which segment of Itg␣6b acts as a key motif in regulating myf5 promoter activity, we analyzed the domain structure of Itg␣6b using the NCBI database. The data showed a total of five ␤-propeller repeat domains at the N-terminal region of Itg␣6b (Fig. 5A). Because the ␤-propeller repeat domain has been reported as a ligand-binding region of the integrin ␣ subunit for signal transduction (27), we designed three constructs, namely Itg␣6b⌬1 lacking five repeat domains, Itg␣6b⌬2 lacking the first repeat domains, and Itg␣6b⌬3 lacking the last four repeat domains (Fig. 5A), and performed an in vivo myf5 promoter luciferase assay. Compared with the embryos injected with pZmyf5 6.3R alone, the luciferase activity driven by myf5 promoter in the embryos injected with excessive itg␣6b mRNA was increased up to 227% (Fig. 5B). In contrast, the luciferase activity of embryos injected with deletion clones (itg␣6b⌬1 mRNA, itg␣6b⌬2 mRNA, and itg␣6b⌬3 mRNA) remained unchanged (Fig. 5B), suggesting that the ␤-propeller repeat domains of Itg␣6b are required for the regulation of myf5 promoter activity by Itg␣6b. The cell surface cross-linking experiment combined with co-immunoprecipitation also detected the binding between the fusion proteins Itg␣6b-MYC and Dkk3a-FLAG

profile of putative genes and their encoded proteins obtained by LC-MS/MS analysis
After the Dkk3a-FLAG fusion protein was purified by anti-FLAG beads, it was immunoprecipitated with zebrafish membrane proteins.

Location
Gene ID Common name   . dkk3a binds to itg␣6b and regulates myf5 promoter activity. Plasmid DNA, mRNA, and MO were microinjected individually or together (as indicated by ϩ) into the embryos to carry out the transient luciferase assay. The luciferase activity of the embryos injected with only pZmyf5 6.3R in which the luciferase reporter was driven by the upstream 6.3 kb of zebrafish myf5 promoter was measured independently three times, and the average value served as 100%. Compared with the embryos injected with pZmyf5 6.3R alone, the luciferase activity was increased in the embryos injected with dkk3a mRNA alone, itg␣6b mRNA alone, or dkk3a mRNA combined with itg␣6b mRNA. In contrast, the luciferase activity was decreased in the embryos injected with dkk3a MO alone, itg␣6b MO alone, dkk3a MO together with itg␣6b MO, dkk3a mRNA together with itg␣6b MO, or itg␣6b mRNA with dkk3a MO, suggesting that ligand Dkk3a binds to its receptor, Itg␣6b, resulting in up-regulation of myf5 promoter activity. The data were presented as the average Ϯ standard deviation from measurements collected from three independent experiments and two measurements for each experiment. ** and *** indicate the significant differences of values at p Ͻ 0.01 and p Ͻ 0.001 levels, respectively.

Dkk3a Regulates myf5 through Interaction with Itg␣6b
NOVEMBER 16, 2012 • VOLUME 287 • NUMBER 47 FIGURE 5. The ␤-propeller repeat domains of Itg␣6b are key segments that interact with Dkk3a to control the promoter activity of myf5. A, schematic drawing of three different deletions of the Itg␣6b domain. B, the mRNA encoding for each deletion construct was co-microinjected into the one-celled stage of embryos with pZmyf5 6.3R in which the luciferase reporter was driven by the upstream 6.3 kb segment of zebrafish myf5 promoter to carry out the transient luciferase assay. The luciferase activity driven by injected construct(s) (as indicated by ϩ) was measured in three independent experiments. The average value of the embryos injected with only pZmyf5 6.3R served as 1 (100%). Compared with the embryos injected with pZmyf5 6.3R alone, the luciferase activity was increased in the embryos injected with pZmyf5 6.3R and itg␣6b mRNA. In contrast, the luciferase activity was decreased in the embryos injected with mRNAs encoding for three different deletions of ␤-propeller repeat domains of Itg␣6b. The data were presented as the average Ϯ standard deviation from measurements collected from three independent experiments and two measurements for each experiment. *** indicates a significant difference at p Ͻ 0.001.

Dkk3a Regulates myf5 through Interaction with Itg␣6b
pared with the non-injected wild-type embryos at 16 hpf (Fig.  6A), all injected embryos exhibited similar abnormal phenotypes, such as incomplete muscle development and irregular shape of somitic cells with no clear boundary (Fig. 6, B, C, and D, and Table 2). However, the number of somites in all defective embryos was unaffected. Furthermore, we injected NLS-EGFP and Lyn-Tomato mRNA, which were used to label cell nuclei and cell membrane, respectively, into wild-type embryos or embryos injected as noted above. Compared with the wild-type embryos (Fig. 6AЈ, 1-3), the embryos injected with itg␣6b MO alone (Fig. 6BЈ, 1-3) and dkk3a MO alone (Fig. 6CЈ, 1-3) exhibited fewer cells within somites (Fig. 6E). Compared with the embryos injected with either itg␣6b MO alone or dkk3a MO alone, much fewer somitic cells and smaller somites were observed in the embryos co-injected with itg␣6b MO plus dkk3a MO (Fig. 6DЈ,1-3), suggesting that a negative synergistic effect occurred in the context of somitic cell number and somite size. Nevertheless, the expression of the somite differentiation marker myod in all defective embryos was unaffected (Fig. 7), suggesting that knockdown of Itg␣6b and Dkk3a results in a decreased number of cells within somites, thus causing defects in trunk muscular development similar to those identi-fied above. The rescue experiment also indicated that the inhibition of itg␣6b MO and dkk3a MO was specific (supplemental Figs. S2 and S3).
Knockdown of Itg␣6b and Dkk3a Results in the Down-regulation of myf5 Expression-To investigate and characterize myf5 expression induced by knockdown of itg␣6b and dkk3a in the somites of embryos, we injected itg␣6b MO alone, dkk3a MO alone, and itg␣6b MO plus dkk3a MO separately into onecelled fertilized eggs. These embryos were then subjected to WISH to examine the expression of myf5 at 16 hpf. The noninjected embryos (WT) served as the control group, which showed myf5 expression in presomitic mesoderm (indicated by a bracket) and five somites from ϪII, ϪI, to 0 and from 1 to 2 (Fig. 7A). We noted a weak expression of myf5 in the mature somites (1 and 2) (Fig. 7A, black arrowheads) but a strong expression of myf5 in the newly forming somites (0, ϪI, and ϪII), presenting a V-shape (Fig. 7A, red arrowheads). However, in the itg␣6b MO-injected, dkk3a MO-injected, and itg␣6b MO-plus dkk3a MO-injected embryos, myf5 was expressed normally in presomitic mesoderm, but it was significantly down-regulated in the newly forming somites (0, ϪI, and ϪII), presenting a horizontal shape, and absent in the mature somites  (1 and 2) (Fig. 7, B, C, and D, and Table 2). These data indicated that ligand Dkk3a regulates myf5 expression through receptor Itg␣6b in zebrafish. Interestingly, we noticed that the defective phenotypes of myf5 expression and its occurrence percentage caused by co-injection of itg␣6b MO plus dkk3a MO were similar to those of embryos injected with either itg␣6b MO alone or dkk3a MO alone, suggesting that no negative synergistic effect occurred under these conditions. In addition, we also used WISH to detect the expression patterns of myod in the WT, itg␣6b MO-injected, dkk3a MO-injected, and itg␣6b MO-plus dkk3a MO-injected embryos. Results showed that myod was expressed normally in the mature somites (from 0 to 6) but absent in the newly forming somites in all injected embryos (Fig. 7, E, F, G, and H), suggesting that the decrease or absence of myf5 expression in somites did not result from a lack of somites.
Knockdown of Itg␣6b and Dkk3a Results in the Decrease of the Protein Level of p-p38a-To study whether the phosphorylation of p38a is affected by knockdown of either receptor Itg␣6b or ligand Dkk3a, we extracted total proteins of the 16-hpf embryos derived from one-celled fertilized eggs injected with either itg␣6b MO or dkk3a MO. The total proteins were analyzed by Western blot. Compared with the wild-type embryos, the amounts of p38a protein in the itg␣6b MO-and dkk3a MOinjected embryos were relatively unchanged; i.e. the p38a protein levels relative to ␣-tubulin levels among wild-type embryos, itg␣6b MO-injected embryos, and dkk3a MO-in-jected embryos were 1:0.92:0.99, respectively (Fig. 8A). However, the protein level of p-p38a from embryos injected with either itg␣6b MO or dkk3a MO was greatly reduced. In this case, the p-p38a levels relative to ␣-tubulin levels among wildtype embryos, itg␣6b MO-injected embryos, and dkk3a MOinjected embryos were 1:0.32:0.29, respectively (Fig. 8A). On the other hand, we found that the amounts of p38a protein extracted from wild-type embryos, from embryos injected with dkk3a mRNA alone, and from embryos injected with dkk3a mRNA plus itg␣6b MO were quite close. Under these conditions, the p38a levels relative to ␣-tubulin levels among wildtype embryos, dkk3a mRNA-injected embryos, and dkk3a mRNA-plus itg␣6b MO-injected embryos were 1:0.97:0.99, respectively (Fig. 8B). However, although the protein level of p-p38a was increased in the dkk3a mRNA-injected embryos, it was reduced in the embryos injected with dkk3a mRNA plus itg␣6b MO (Fig. 8B). Results showed that protein levels of p-p38a relative to ␣-tubulin levels among wild-type embryos, dkk3a mRNA-injected embryos, and dkk3a mRNA-plus itg␣6b MO-injected embryos were 1:1.73:0.23, respectively (Fig. 8B). Similarly, the protein level of p-p38a was increased in the itg␣6b mRNA-injected embryos but reduced in the embryos injected with itg␣6b mRNA plus dkk3a MO (Fig. 8C). Specifically, the protein levels of p-p38a relative to ␣-tubulin levels among wildtype embryos, itg␣6b mRNA-injected embryos, and itg␣6b mRNA-plus dkk3a MO-injected embryos were 1:1.62:0.10, respectively (Fig. 8C). This line of evidence indicated that the   2). We also used WISH to detect the expression patterns of myod in the WT, itg␣6b MO-injected, dkk3a MO-injected, and itg␣6b MO-plus dkk3a MO-co-injected embryos and found that myod was expressed normally: it was present in the mature somites (from 0 to 6) but absent in the newly forming somites (E, F, G, and H), suggesting that the decrease or absence of myf5 expression in somites did not result from a lack of somites. Scale bar: 50 m.
ligand Dkk3a activates the phosphorylation of p38a through binding the receptor Itg␣6b, which in turn up-regulates myf5 expression in somites during muscle development.
The Defective Expression of myf5 Caused by Either itg␣6b MO or dkk3a MO Can Be Rescued by Introducing Either p38a mRNA or smad4 mRNA-We asked whether myf5 defects caused by itg␣6b MO could be rescued by overexpression of itg␣6b mRNA. In the itg␣6b MO-injected embryos, 70% exhibited defects of myf5 expression, but 30% (n ϭ 133) did not (supplemental Fig. S4). However, when itg␣6b MO was co-injected with itg␣6b(wobble) mRNA, the percentage of embryos that exhibited no defect of myf5 expression increased to 72% (n ϭ 95) (supplemental Fig. S4), suggesting that the defect induced by itg␣6b MO was specific. Interestingly, when itg␣6b MO was we co-injected with p38a mRNA and smad4 mRNA, the percentage of embryos that exhibited no defect of myf5 expression increased to 63 (n ϭ 104) and 64% (n ϭ 98), respectively (supplemental Fig. S4). Similarly, when dkk3a was knocked down, we observed 74% of embryos (n ϭ 141) with a defect in myf5 expression (supplemental Fig. S4). However, the percentages of defects in the embryos injected with MO and excessive mRNAs of dkk3a(wobble), p38a, and smad4 were reduced to 37 (n ϭ 96), 42 (n ϭ 98), and 41% (n ϭ 103), respectively (supplemental Fig. S4). This line of evidence suggested that Dkk3a interacts with Itg␣6b to regulate myf5 expression through the downstream p38a and Smad4 pathways.
Conclusion-Based on the collective evidence shown in this study, we hypothesized a model to depict the impact of binding between ligand Dkk3a and receptor Itg␣6b on the activation of zebrafish myf5 promoter in somites during embryogenesis (Fig.  9). In summary, we demonstrated that secreted Dkk3a binds to the membrane receptor Itg␣6b, which results in enhancement of the phosphorylation of p38a, which in turn increases the protein level of p-p38a and activates the myf5 promoter. This finding further demonstrates that Dkk3a regulates the promoter activity of zebrafish myf5 through the p38a signaling pathway.

Dkk3a and Itg␣6b Are Co-expressed in Zebrafish Embryonic
Somites during Developmental Stages-Integrins are important membrane receptors that are located in the extracellular matrix of animal cells. Integrins play important roles in cell adhesion, migration, gastrulation, morphogenesis, growth, proliferation, differentiation, apoptosis, and signaling transduction during development. The cytosolic integrins interact with actin filaments through binding cellular cytoskeleton proteins, such as vinculin, talin, ␤-actin, paxillin, and tensin (28), and activate many cytosolic signal transduction pathways (29). In this study, we demonstrated that zebrafish Dkk3a regulates muscle development through its interaction with the ␣ form of integrins, in this case Itg␣6b. In fact, the expression pattern and function of Itg␣6b are largely unknown except for mouse itg␣6b, which is expressed in the early developmental stages and is continuously presented in lateral plated mesoderm, heart, head, and forelimbs at embryonic day E9.5 (30). In somites, expression of mouse itg␣6b is decreased in dermomyotome and sclerotome but elevated in myotome. In early epaxial myotome, the expres- Western blot analysis was performed using a specific antibody against p38a, p-p38a, or ␣-tubulin, which served as an internal control. A, the Western signals of p38a and p-p38a levels from WT embryos, 12-ng itg␣6b MO-injected embryos, and 2-ng dkk3a MO-injected embryos are shown. The p38a and p-p38a levels relative to ␣-tubulin levels are also indicated. B, the Western signals of p38a and p-p38a levels from WT embryos, 200-pg dkk3a mRNA-injected embryos, and 200-pg dkk3a mRNA-plus 12-ng itg␣6b MO-injected embryos are shown. The p38a and p-p38a protein levels relative to ␣-tubulin levels are indicated. C, the Western signals of p38a and p-p38a levels from WT embryos, 200-pg itg␣6b mRNA-injected embryos, 200-pg itg␣6b mRNA-plus 2-ng dkk3a MO-injected embryos are shown. The p38a and p-p38a protein levels relative to ␣-tubulin levels are also indicated.

Dkk3a Regulates myf5 through Interaction with Itg␣6b
NOVEMBER 16, 2012 • VOLUME 287 • NUMBER 47 sion pattern of mouse itg␣6b mRNA is similar to that of myf5 mRNA (30). Interestingly, our data also showed that zebrafish itg␣6b mRNA is expressed in somites during muscle development and is co-localized with myf5 mRNA, indicating that the expression pattern of zebrafish Itg␣6b is similar to that of mice. Therefore, it is reasonable to speculate that the zebrafish ligand Dkk3a interacts with receptor Itg␣6b to regulate myf5 expression in somites during muscle development.
Ligand Dkk3a Binds Receptor Itg␣6b to Regulate the Phosphorylation of p38a-It has been demonstrated that the integrin family is able to interact with outer membrane ligands to activate downstream signaling, including tyrosine kinases, such as focal adhesion kinase (31), c-Abl (32), and Src family members (33,34); serine/threonine kinases, including mitogen-activated protein kinases (35,36) and protein kinase C (37,38); the lipid kinase phosphatidylinositol 3-kinase (39); and small GTP-binding proteins, including Ras, Rac, Rho, and CDC42 (40). Recently, Hsu et al. (23) demonstrated that knockdown of dkk3a results in a decrease in the protein level of p-p38a, indicating that Dkk3a activates myf5 promoter activity through reg-ulation of the phosphorylation of p38a. In this study, we go further to reveal that zebrafish Dkk3a binds to membrane receptor Itg␣6b. When itg␣6b was knocked down, we found that the protein level of p-p38a was greatly decreased. This result supports that of Segat et al. (41), who demonstrated that overexpression of human Itg␣6b activates the phosphorylation of p38a, in turn causing proliferation of mouse primary culture chondrocytes. Therefore, we suggested that ligand Dkk3a activates the phosphorylation of p38a through its interaction with receptor Itg␣6b in somites during muscle development.
Ligand Dkk3a Regulates myf5 Promoter Activity through Interaction with Its Receptor, Itg␣6b-Myf5 plays an important role in muscle cell specification during muscle cell development to ensure that muscle cell proliferation and differentiation to myoblasts can be successively processed (42)(43)(44). Hsu et al. (22) knocked down dkk3a by injecting dkk3a-specific MO in embryos, which then exhibited such abnormal phenotypes as small head, incomplete muscle development, shortened tail, short axial extension, and irregular shape with vague boundary somites. These phenotypes are similar to those induced by FIGURE 9. Schematic model of Dkk3a regulation of zebrafish myf5 promoter activity through binding receptor Itg␣6b. This study demonstrated that secreted Dkk3a binds to the membrane receptor Itg␣6b and activates the phosphorylation of p38a, consequently increasing the protein level of phosphorylated p38a. In the downstream pathway reported by Hsu et al. (22), Dkk3a causes the sumoylation (SUMO) of Smad4 through activation of phosphorylated p38a. Stabilized Smad4 then enables the formation of a Smad2/3a/4 complex, which then enters the nucleus to activate the myf5 promoter.
knockdown of myf5 as reported by Chen and Tsai (45) because knockdown of dkk3a results in the decrease of myf5 expression in somites. This line of evidence indicates that Dkk3a is involved in the positive regulation of myf5 expression. In this study, we found that knockdown of itg␣6b results in embryos at 16 hpf having defective somites similar to those induced by knockdown of dkk3a. Additionally, myf5 expression in somites was decreased. On the other hand, when either excessive dkk3a mRNA or itg␣6b mRNA was injected into embryos, myf5 promoter activity was activated. Furthermore, when dkk3a mRNA and itg␣6b mRNA were combined and co-injected into embryos, myf5 promoter activity was greatly up-regulated, suggesting that dkk3a and itg␣6b produce a synergism that strongly up-regulates myf5 expression. Recently, Wilschut et al. (46) demonstrated that knockdown of Itg␣6b causes down-regulation of myf5 and inhibits muscle cell proliferation in porcine primary muscle stem cells, indicating that Itg␣6b is very important for muscle cell proliferation. Taken together, we proposed that ligand Dkk3a activates myf5 promoter through interaction with membrane receptor Itg␣6b in somites during muscle development.

A Plausible Regulatory Model Showing How Ligand Dkk3a Interacts with Membrane Receptor Itg␣6b to Control myf5
Expression in Zebrafish Embryos-Based on the evidence shown in this study and the report of Hsu et al. (23), we have formulated the following hypothesis. First, when secreted ligand Dkk3a binds to the membrane receptor Itg␣6b, phosphorylation of p38a is activated, consequently increasing the protein level of p-p38a. Second, such increase in p38a phosphorylation and protein level causes Smad4 sumoylation to occur followed by the formation of a Smad2/3a/4 complex that enters the nucleus and activates the myf5 promoter (Fig. 8). Still, we cannot rule out other proteins that may also be involved in the downstream signaling transduction through Dkk3a-integrin interaction. In fact, integrins receive or transmit signals in two different ways: inside out and outside in (47). In the inside-out signaling pathway, the conformation of integrins might be changed to increase affinity with outer membrane ligands if cytosolic proteins, such as talin or kindlins, bind integrins through the ␤ tail. Based on the mass spectrometry data of Dkk3a-associated putative proteins as shown in Table 1, we found that one such interactive protein is zebrafish talin 1. Thus, it is highly likely that Itg␣6b may process inside-out signaling through binding of cytosolic talin 1, resulting in an increase in its affinity to outer membrane ligand Dkk3a. On the other hand, in the outside-in signaling pathway, integrins can regulate signaling pathways, including growth factors like TGF␤. When TGF␤ is secreted to the outer membrane, it interacts with latency-associated peptide and latent TGF ␤-binding protein to become inactivated. Nevertheless, when this inactivated complex interacts with integrin, TGF␤ becomes active through protease digestion or thrombospondin 1 cleavage (48). It has been reported that Dkk3a can regulate Smad4 stability through TGF␤ downstream signaling (23). Thus, it is equally likely that Dkk3a can interact with Itg␣6b in a manner similar to TGF␤, enabling Dkk3a to activate downstream signaling after thrombospondin 1 cleavage. Indeed, Table 1 shows mass spectrometry data indicating that type I thrombospondin domain-containing 7A-like is a Dkk3a-interacting candidate protein.