Implications of the Wnt5a/CaMKII Pathway in Retinoic Acid-Induced Myogenic Tongue Abnormalities of Developing Mice

Although proper tongue development is relevant to other structures in the craniofacial region, the molecular details of muscle development in tongue remain poorly understood. Here, we report that pregnant mice treated with retinoic acid (+RA) produce embryos with tongue malformation and a cleft palate. Histological analyses revealed that at E14.5, the tongues of +RA fetuses failed to descend and flatten. Ultrastructural analysis showed that at perinatal stage E18.5, the myofilaments failed to form normal structures of sarcomeres, and arranged disorderly in the genioglossus. The proliferation and levels of myogenic determination markers (Myf5 and MyoD) and myosin in the genioglossus were profoundly reduced. Wnt5a and Camk2d expressions were down-regulated, while levels of Tbx1, Ror2, and PKCδ were up-regulated in the tongues of +RA fetuses. In mock- and Wnt5a-transfected C2C12 (Wnt5a-C2C12) cells, Wnt5a overexpression impaired proliferation, and maintained Myf5 at a relative high level after RA treatment. Furthermore, Wnt5a overexpression positively correlated with levels of Camk2d and Ror2 in C2C12 cells after RA exposure. These data support the hypothesis that the Wnt5a/CaMKII pathway is directly involved in RA-induced hypoplasia and disorder of tongue muscles.

However, the aberrant morphogenesis and the molecular mechanisms that regulate myogenic development in the tongues of developing mice exposed to high-dose RA remain unclear. Here, we show that the Wnt5a/CaMKII pathway is implicated in RA-induced abnormal tongue myogenic development in embryonic mice. At E14.5 stage, myogenic cell proliferation in genioglossus is reduced, accompanied by down-regulations of Myf5 and MyoD. Subsequently at E18.5 stage, myofilaments fail to form normal sarcomere structures and were disordered arranged in the genioglossus. Compared to the fetal mice from control pregnant mice without RA treatment, Wnt5a was positively correlated with Camk2d level and inversely correlated with the levels of Tbx1, Ror2 and PKCd in the tongues of fetal mice from pregnant mice exposed to high-dose RA. Using C2C12 cells, we further showed that stable expression of Wnt5a is closely linked to the proliferation and differentiation C2C12 cells. The positive association of Wnt5a with Camk2d and Ror2 in C2C12 cells in response to RA stimulation support our in vivo finding and indicate a direct involvement of the Wnt5a/CaMKII pathway in RAinduced tongue malformation.

Results
Fetuses of retinoic acid-treated pregnant mice develop tongue malformation. Morphology and ultrastructure of the tongues of E14.5, E15.5 and E18.5 mouse fetuses were examined by hematoxylin and eosin (HE) staining, immunohistochemical staining and transmission electron microscopy ( Figure 1). At E14.5, fetal mice exposed to excess RA developed tongue deformities (Figure 1a). The tongues of control mouse fetuses were flat, and descended as a result of genioglossus muscle contraction. In addition, the bilateral palatal shelves moved upward, growing horizontally and maintained contact with the tongue (Figure 1a i). By contrast, the tongues of RAexposed fetuses remained at a higher position, a consequence of failed flattening and descent (Figure 1a ii). Moreover, the bilateral palatal shelves extended vertically along both sides of the tongue, forming a cleft (Figure 1a ii).
We then examined tongue muscle development immunohistochemically, using myosin heavy chain as a marker of differentiation. At E15.5, a large number of myotubes were positive for myosin in both control and mutant group. Compared to fetuses from control pregnant mice (Figure 1b i and ii), the expressions of myosin in the tongue body and genioglossus were apparently decreased in RAaffected fetuses ( Figure 1b iii and iv). At early stage of E14.5, only some myotubes were positive for myosin. In RA-affected fetuses, myosin staining in myotubes became weaker compared to control fetuses (Supplementary Figure 1).
According to transmission electron microscopy, the genioglossus of normal control mouse fetuses contained definitive sarcomere structures in myofibrils, i.e., complete light and dark bands, and clear Z-lines and M-lines (Figure 1c i), as well as the tongue body at E18.5 (Supplementary Figure 2); myofibrils were arranged orderly in the longitudinal (anteroposterior) direction (Figure 1c i). By contrast, significant defects including hypoplasia and muscle derangement were observed in genioglossus (Figure 1c Figure  2), RA exposure cause hypoplasia in the myocyte, only structures of myofilaments were detected, while these myofilaments failed to form classic sarcomere structures. Furthermore, in the sagittal section of genioglossus that were anteroposteriorly arranged, we noted the presence of transverse myofilament bundles adjacent to the longitudinal ones within the same myocyte, a characteristic of muscle derangement (Figure 1c ii, black arrow). We also observed that a great amount of myofilament bundles were arranged transversely among obliquely arranged ones (Supplementary Figure 2) in sagittal sections of genioglossus of 1RA fetuses.
Thus, excessive RA exposure caused hypoplasia and disordered arrangements in tongue, especially in the genioglossus muscle, the major muscle responsible for tongue descent, resulting in high position of tongue.
Retinoic acid inhibits proliferation and myogenic determination of tongue muscle. Next, we determined if excess RA affects cell proliferation and myogenic determination of fetal tongue muscles. Myosin expression in the tongue body (iii) and the genioglossus muscle (iv) were decreased in the 1RA fetus versus control fetus. A higher magnification of the tongue intrinsic muscle from control fetuses showed multinucleated myotubes expressing high levels of myosin ((i), arrow). In 1RA fetuses, myosin expression in myotubes ((iii) arrow) became weaker. In a transverse section of the genioglossus muscle, myosin was expressed at high levels in the control myotubes ((ii), arrowhead); in 1RA fetuses, myosin staining was very weak. Bar 5 40 mm. Upper right inserts in (i and iii) show images at lower magnifications, Bar 5 100 mm. (c): Transmission electron microscopy examination of sagittal sections of the genioglossus. In control (i), myofibrils (white arrow) and sarcomeres were arranged longitudinally with integral I and A bands, and clear Z-and M-lines. In 1RA (ii), the classic structures of sarcomeres of myofibrils were not found, myofilament bundles were arranged longitudinally (white arrow) and transversely (black arrow) in a single cell. Bar 5 500 nm. Cont.: control mouse fetus; 1RA: RA-exposed mouse fetus.PS: palatal shelves; T: tongue. We measured the cell proliferation in the genioglossus of mouse fetuses collected from control and RA-exposed pregnant mice. At E14.5, the numbers of BrdU-staining positive cells/cm 2 in genioglossus of mouse fetuses exposed to RA (Figure 2a ii) were 35.3% of those observed in control fetuses (Figure 2a i, P , 0.01). To identify if RA exposure impairs myogenic determination in fetal tongue, we measured mRNA and protein levels of Myf5 and MyoD, recognized early myogenic markers, in the tongues of mouse fetuses exposed to excess RA or vehicle alone. Myf5 and MyoD proteins were readily detected in the genioglossus of control group (Figures 2b i and 2c i), while in RA-treated group (Figures 2b ii and 2c ii), their corresponding levels were apparently lower than control. qRT-PCR analyses revealed that RA exposure decreased the mRNA levels of Myf5 (64.6%, P , 0.05) and MyoD (55.8%, P , 0.05) (Figure 2d).
Thus, Wnt5a expression is positively correlated with Camk2d and negatively correlated with Ror2, Tbx1 and PKCd expressions in the tongues of fetal mice exposed to excess RA.
In vitro evidence for crosstalk between Wnt5a and Camk2d in mediating the effects of RA on myogenic development of tongue muscles. C2C12 cell, originally derived from myoblast cell clones of adult C3H mouse leg muscle 17 , is an established skeletal muscle progenitor cell line that provides an ideal model system for studying skeletal muscle differentiation in vitro 18 . Since Wnt5a expressed at low level in C2C12 cells (unshown data), we established a stable C2C12 cell line expressing HA-tagged Wnt5 (HA-Wnt5a-C2C12 cells) using retroviral mediated gene delivery. Immunoblotting ( Figure 4a) and qRT-PCR (Figure 4b) confirmed the overexpression of Wnt5a in HA-Wnt5a-C2C12 cells.
RA does not affect the proliferation (the Optical density at 450 nm) of wild-type C2C12 cells, as measured by the CCK-8 assay (data not shown). By contrast, when HA-Wnt5a C2C12 cells were treated with RA (10 mM) for 24 h, 36 h, and 48 h, the cell proliferation decreased by 20% (P 5 0.057), 21% (P , 0.05) and 27% (P , 0.05), respectively, compared to HA-Wnt5a-C2C12 cells treated with vehicle alone (Figure 4c). Thus, the level of Wnt5a is associated with HA-Wnt5a-C2C12 cell proliferation in response to RA.
The mRNAs of Myf5 and MyoD in differentiated control-and HA-Wnt5a-C2C12 cells were readily detected ( Figure 4d and   Camk2d, Ror2 and PKCd in the tongues of E14.5 mouse fetuses. The mRNA levels of Tbx1, Ror2 and PKCd in the 1RA fetuses increased 62%, 52% and 55%, respectively, as compared to controls; the mRNA levels of Wnt5a and Camk2d in the 1RA fetuses decreased 24% and 38%, respectively, compared to controls. Experiments were performed in triplicate, and ** denotes P , 0.01.

www.nature.com/scientificreports
Ror2 and PKCd in HA-Wnt5a-C2C12 cells (Figure 4i). Thus, our data support direct relationships between Wnt5a and Camk2d, and Wnt5a and Ror2. This hypothesis was further supported by results shown in Figure 4i. Wnt5a remained at relatively high levels in HA-Wnt5a-C2C12 cells, even after RA treatment. Meanwhile, Camk2d mRNA was significantly higher in HA-Wnt5a-C2C12 cells compared to control C2C12 cells. However, RA treatment decreased Ror2 mRNA level in HA-Wnt5a-C2C12 cells to a greater extent than that observed in control C2C12 cells (Figure 4i), its level was about 2fold of its change in RA-treated-C2C12 against C2C12 cells (Figures 4 h and i). Similarly, the mRNA levels of Tbx1 and PKCd were slightly more down-regulated in RA-treated HA-Wnt5a-C2C12 cells than their expression changes in HA-Wnt5a-C2C12 cells against C2C12 cells (Figures 4 h and i). Thus, the implicated inverse relationships between Wnt5a and Tbx1, and Wnt5a and PKCd in this process require further investigation.

Discussion
Normal tongue development requires contractions and homing of the tongue and other facial muscles 1 . Early movement of mouse tongue is adapted to functional activities such as suckling, swallowing, and chewing 1 . Prenatal exposure to excess RA triggers cleft palate formation in mouse fetuses 9,19,20 , while the malformation of tongue was rarely reported 21,22 . The aberrant morphogenesis of tongue and underlying molecular and cellular mechanisms that regulate tongue development in the presence of excess RA in fetal mice remain unclear. In current work, we also observed that the excessive RA exposure during pregnancy led to the failure of flattening and descent of tongue in mouse fetuses (Figure 1a ii), which is associated with defective subcellular changes of the tongue muscle including the hypoplasia and deranged tongue muscles, at E18.5 (Figure 1c ii).
How does excess RA induce tongue malformation in fetal mice? RA plays important roles in mediating germ cell development and neural differentiation of embryonic stem cells [23][24][25] . RA activates myogenesis by up-regulating myogenic markers MyoD and myogenin in somites, and myf5 in presomitic and somitic mesoderm in developing zebra fish embryos 26 . According to our data, at E14.5, the proliferations of genioglossus cells were greatly suppressed (Figure 2a) by excess RA. Based on IHC and qRT-PCR assays of the tongues of RA-exposed fetal mice at E14.5, which showed the down-regulation of Myf5 and MyoD, two myogenic determination markers (Figures 2b, c and d), we concluded that RA interferes with tongue myogenic differentiation. Genioglossus is the major muscle involved in tongue descent in fetal mice. Fetal tongue descent occurs before E14.5, indicating that muscle fibers are already mature and can contract in the genioglossus. At E18.5, failure of myofilaments to assemble into normal sarcomeres and derangement in the genioglossus of fetal mice lead to tongue malformation.
Wnt5a is involved in the developments of face, ears, genitals, limbs, distal digits 27 , and early embryonic myogenesis in mice 15 . We previously reported that Wnt5a-deficient mice developed cleft palate and distorted tongue 11 . In addition, the serine-threonine Ca 21 / calmodulin-dependent protein kinase II (CaMKII) 28 emulates the effects of Wnt5a during the myogenic development of chick wing bud, and is involved in Wnt5a-induced myogenic determination 29 . Loss of Wnt5a, or loss of Tbx1 together with reduced RA synthesis, results in cardiac hypoplasia 12,13 , so that Wnt5a may control tongue size, fungiform papilla patterning and development through interacting with the Ror2 receptor 16 . Based on the studies, we selectively examined the expression levels of Wnt5a, Camk2d, Ror2, Tbx1 and PKCd in the tongues of fetal mice with or without RA exposure. At E14.5, RA exposure decreased Wnt5a mRNA level in the tongues of fetal mice, while reducing Camk2d mRNA and increasing Ror2, Tbx1 and PKCd mRNA levels ( Figure 3).
We also show that Wnt5a is involved in the RA-mediated inhibition of proliferation and differentiation of C2C12 cells. Since Wnt5a protein is not detected in C2C12 cells, and the growth of C2C12 cells is unaffected by RA (10 mM, data not shown), we established a stable C2C12 cell line over-expressing HA-tagged Wnt5a. RA decreased the rate of cell proliferation in HA-Wnt5a-C2C12 cells in a timedependent manner (Figure 4b and Figure 4c), and decreased mRNA levels of Myf5 and MyoD (Figure 4d). Notably, Wnt5a over-expression markedly increased the levels of Myf5 and MyoD mRNA following RA exposure as compared to control cells (Figure 4e), indicating that Wnt5a antagonizes the suppressive effects RA on Myf5 and MyoD expression.
Our study also suggests a crosstalk between Wnt5a and Camk2d in tongue myogenic development following RA exposure. RA-induced down-regulation of Wnt5a mRNA in C2C12 cells was associated with reductions of Camk2d and Ror2 mRNAs, and up-regulations of Tbx1 and PKCd mRNAs (Figure 4f). In HA-Wnt5a-C2C12 cells, however, RA-mediated suppression of Wnt5a mRNA level was associated with up-regulations of Tbx1, Camk2d, Ror2 and PKCd (Figure 4g). Moreover, Wnt5a over-expression down-regulated the basal mRNA levels of Camk2d, Ror2 and PKCd as compared to wildtype C2C12 cells (Figure 4h). The relative high levels of Wnt5a in HA-Wnt5a-C2C12 cells (Figure 4h, i), with or without RA treatment (10 mM), allowed us to observe a positive correlation between Wnt5a and Camk2d, and inverse correlations of Wnt5a with Tbx1, Ror2 and PKCd. Based on the observation that RA-induced greater changes (2fold) in Ror2 mRNA level in HA-Wnt5a-C2C12 cells compared to wild-type C2C12 cells (Figure 4h, Figure 4i), we infer a direct association of Wnt5a with Camk2d and Ror2 expression. The concomitant up-regulations of Wnt5a, Camk2d, and Ror2 might promote normal development of genioglossus cells by antagonizing the effect of excess RA. The potential inverse correlations of Wnt5a mRNA level with Tbx1 and PKCd mRNAs in this process requires further investigation.
To sum up, our work shows that excess RA induces tongue malformation in fetal mice. At E14.5, tongues of fetal mice from pregnant females treated with RA failed to descend and flatten. Myofilaments of the genioglossus and tongue body failed to form normal structures of sarcomeres, resulting in hypoplasia of muscle fibers. Thus, the contraction force of genioglossus muscles will be decreased. Meanwhile, in genioglossus, a great amount of myofilament bundles ranged transversely in anteroposteriorly arranged ones, resulting in derangement in muscle fibers. So, the contracting directions of mutant genioglossus muscles will be inconsistent. RA exposure also impaired proliferation, determination and differentiation of the genioglossus cells. Taken together, the contraction effects of genioglossus will surely be decreased, and then cause high position of tongue. In vitro, Myf5 and MyoD were markedly down-regulated by RA in the genioglossus of fetal mice and in cultured C2C12 cells. Wnt5a plays a potentially important role in the normal development of mouse tongue, and regulates the rate of proliferation in C2C12 cells. Over-expression of Wnt5a antagonized RA-mediated downregulation of Myf5 and MyoD in C2C12 cells. Results from both in vivo and in vitro studies indicate a crosstalk between Wnt5a, Camk2d, and Ror2 in myogenic development of the tongue in response to RA. Wnt5a/CaMKII pathway is involved in RA-induced abnormal myogenic development of tongue in fetal mice. While the potential negative correlation of Wnt5a with Tbx1 and PKCd in tongue malformation following RA exposure needs further investigation, our study provides novel insights into RA-mediated birth defects and tongue-related diseases.

Methods
Animal experiment. Female ICR mice (10 , 12-week-old) were crossed with fertile males (ICR) overnight; the appearance of a vaginal plug was designated as GD 0.5 (gestation day 0.5). Pregnant mice at GD 10 were randomly divided into treatment and control groups, and respectively given all-trans retinoic acid (RA, 100 mg/kg) dissolved in edible oil, and edible oil only by gavage 10 . Pregnant mice were euthanized at E14.5, E15.5, or E18.5.
Histochemical staining. At embryonic day 14.5 (E14.5), the heads of mouse fetuses were fixed in 4% paraformaldehyde (PFA), embedded in paraffin, sliced into 5-mm sections and fixed onto polylysine-coated slides. Tissue sections were then stained with hematoxylin and eosin by standard procedures. Transmission electron microscopy (TEM). Muscle samples from the tongue body and genioglossus at E18.5 were immediately fixed in 2.5% glutaraldehyde for 2 h, post-fixed with 2% osmium tetroxide for 2 h, deyhdrated in a graded ethanol series, and embedded in Epon 812. The genioglossus was sectioned sagittally, and the tongue body was sectioned horizontally. Ultrathin sections were obtained using an Ultramicrotome (Leica EM UC6, Germany). All sections, mounted on copper grids, were contrasted using uranyl acetate and lead citrate before being viewed using a transmission electron microscope (JEM-2000EX). Cell culture. The murine skeletal muscle cell line C2C12 was purchased from the cell bank of Type Culture Collection of The Chinese Academy of Science (Shanghai, China). C2C12 myoblasts were cultured in high-glucose Dulbecco's modified Eagle's medium (DMEM, Invitrogen, USA) supplemented with 10% fetal bovine serum (FBS, Invitrogen, USA) at 37uC with 5% CO 2 as previously described 30 . At 48 h, C2C12 myoblasts (,80% confluent) were induced to differentiate using high-glucose DMEM supplemented with 2% horse serum (Hyclone, USA) and treated immediately with RA (10 mM final concentration, Sigma-Aldrich, USA) dissolved in dimethyl sulfoxide (DMSO) for experimental group cells and with DMSO for control group cells. Media containing fresh RA were changed every 2 days for experimental group cells. Cells were collected 4 days post induction of differentiation. RNA extraction and qRT-PCR were performed as described above.
Wnt5a overexpression. Plasmid construction and viral infection. Plasmids were constructed using standard methods; sequences were verified by restriction digestion and/or sequencing. Mouse full-length Wnt5a cDNA fused with hemagglutinin (HA) tag was cloned from total RNAs extracted from C2C12 cells using standard PCR protocol. HA-Wnt5a sequence was subcloned into the pLNCX retroviral vector. For viral packaging, 3 3 10 6 293T cells were incubated in a 10-cm tissue culture plate containing 10 mL DMEM 1 10% FBS without antibiotics overnight. Retroviral plasmid (2 mg), packaging plasmid (2 mg), envelope plasmid (2 mg), and FuGENEH 6 transfection reagent (18 mL; Promega, USA) were added to 600 mL serum-free OPTI-MEM (Invitrogen, USA); after a 30-min incubation at room temperature, the mixture was gently added to 293T cells dropwise. After 12 , 15 h, the transfection mixture was removed and cells were incubated in 10 mL of DMEM supplemented with 10% FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin. Seventy-two hours later, the virus-containing media were harvested, centrifuged at 1, 250 rpm for 5 min and filtered through a 0.22 mm filter to remove cells. The collected viruses were stored at 280uC. For viral infections, cells plated the day before infection were incubated with media containing appropriate amount of retroviruses in the presence of polybrene (6 mg/ml, Sigma-Aldrich, USA) for 6 hrs. Forty-eight hours later, the infected cells were selected using G418 (600 mg/ml) for 14 days.
Cell proliferation of C2C12 cells. C2C12 cells and Wnt5a-overexpressing C2C12 cells were incubated with or without RA for 24, 36 and 48 h. Cell proliferation was analyzed using Cell Counting Kit-8 (CCK-8, Dojin Laboratories, Kumamoto, Japan), according to manufacturer's instructions.
Statistical analysis. IHC assays were performed in triplicate; data were analyzed using unpaired Student's t-test. For qRT-PCR, three independent biological samples were used to establish statistical significance, and the data were analyzed using two independent sample tests of double DCt values. Statistical analyses were conducted using SPSS 13.0 software, and were considered significant at P , 0.05.