Switching of Sox9 expression during musculoskeletal system development

The musculoskeletal system, which comprises muscles, tendons, and bones, is an efficient tissue complex that coordinates body movement and maintains structural stability. The process of its construction into a single functional and complex organization is unclear. SRY-box containing gene 9 (Sox9) is expressed initially in pluripotent cells and subsequently in ectodermal, endodermal, and mesodermal derivatives. This study investigated how Sox9 controls the development of each component of the musculoskeletal system. Sox9 was expressed in MTJ, tendon, and bone progenitor cells at E13 and in bone at E16. We detected Sox9 expression in muscle progenitor cells using double-transgenic mice and myoblastic cell lines. However, we found no Sox9 expression in developed muscle. A decrease in Sox9 expression in muscle-associated connective tissues, tendons, and bones led to hypoplasia of the cartilage and its attachment to tendons and muscle. These results showed that switching on Sox9 expression in each component (muscle, tendon, and bone) is essential for the development of the musculoskeletal system. Sox9 is expressed in not only tendon and bone progenitor cells but also muscle progenitor cells, and it controls musculoskeletal system development.


Connection between muscle progenitors and tendon-bone progenitors. Although Sox9 is
observed throughout tendon and bone progenitors 17,18 , little is known about the connection between muscle progenitors and Sox9 + tendon-bone progenitors. To observe this connection, we performed immunofluorescence staining with antibodies against desmin, which is a marker of the myotendinous junction (MTJ) 11 , and Sox9, which is a marker of tendon and bone progenitor cells 17,18 . In situ hybridization of Scx 17,18 and alkaline phosphatase staining 11 allowed us to distinguish tendon progenitors from bone progenitors. We analyzed the connection at the presumed locations in five regions: the lateral pterygoid muscle attachment to the condyle of the mandible (Fig. 1a-d,f), the triceps brachii muscle attachment to the olecranon (Fig. 1e), the intercostal muscle attachment to the ribs (Fig. 1g-i), the deltoid muscle attachment to the scapula (Fig. 1j-l), and the temporal muscle attachment to the coronoid process of the mandible (Fig. 1m-o). The progenitor cells expressing Sox9 crossed from the tendon anlage to the bone anlage, and the most forward migrating cells made contact with the desmin-accumulating MTJ (Fig. 1).
Sox9 is essential for chondrocyte differentiation and cartilage formation 2 . It is temporally expressed in tendons during the early stage of development but not in developed tendon cells 17 . To clarify the role of Sox9 expression during tendon and bone development, we analyzed the fluorescence intensity of immunohistochemical staining. The fluorescence intensity versus distance plot showed switching of Sox9 expression. At E13, the fluorescence intensity was>100 in the tendon and bone regions (Fig. 2b). At E16, the fluorescence intensity was>100 in the bone region but <100 in the tendon (Fig. 2d). During detailed observation of the connection between muscle progenitors and tendon-bone progenitors, we noticed Sox9 expression in a part of the muscle. The fluorescence intensity of Sox9 expression was>50 in the MTJ region at E13 but <50 in the MTJ region at E16 (Fig. 2b,d). The occupancy rate of Sox9 expression in the MTJ at E13 was high compared to that in the MTJ at E16 (E13: 37.56 ± 6.02%, E16: 0.40 ± 0.45%, P < 0.05) ( Fig. 2e-g).
These results demonstrated that Sox9 was expressed not only in the tendon and the bone but also in a part of the muscle: the MTJ (Fig. 2). [20][21][22] . To determine whether Sox9 is expressed in muscle progenitor cells in vivo, we of double immunofluorescence staining with antibodies against desmin, which is a marker for muscle progenitors 11 ,and Sox9 17,18 . In previous studies of head development 23,24 , we defined the cells in the desmin + area as the cranial paraxial mesoderm (CPM), from which the head muscle originates (Fig. 3a-c). Cranial neural crest cells (CNCs), from which the head tendon and bone originate, were located in the superficial layer of each pharyngeal arch and were wrapped around muscle progenitor cells 23 (Fig. 3a-c). At E10, we found two masses comprising desmin + muscle progenitor cells in the first pharyngeal pouch (Fig. 3a). A mass of muscle progenitor cells in the CPM had few Sox9 + progenitor cells compared to the number of CNCs (CNC: 88.92%±2.24%; CPM: 25.77%±0.21%; P < 0.001) (Fig. 3d). On the other hand, Sox9 was also expressed in some muscle progenitor cells in the limb ( Fig. 3e-h). At E10, we could not identify desmin + muscle progenitor cells in the limbs, which showed Sox9 expression. At E12, we detected desmin + muscle in the limbs, and some muscle cells showed Sox9 expression ( Fig. 3f,g). The limb muscle anlage had few Sox9+ progenitor cells compared to the tendon and bone anlagen (muscle: 24.21%±7.18%; tendon and bone: 96.70%±0.75%; P < 0.01) (Fig. 3h). Therefore, Sox9 was expressed in some muscle progenitor cells in vivo (Fig. 3).

Sox9 expression in muscle progenitor cells in vivo. Some researchers have found Sox9 expression in myoblastic cells in vitro
To determine whether Sox9 is expressed in the muscle, we generated double-transgenic Sox9 creERT2 / Rosa26-loxP-stop-loxP-tdTomato reporter mice 19 . We detected the enrichment of tdTomato + cells in the masseter and intercostal muscles of mice induced at E9 (Fig. 4b,f,d,h) and few tdTomato + cells in the masseter muscle in mice induced at E15 (Fig. 4c,g). We could not identify tdTomato + cells in the intercostal muscle in mice induced at E15 (Fig. 4e,i). The number of tdTomato + cells in the masseter and intercostal muscles in mice induced at E9 was high compared to that in mice induced at E15 (masseter: P < 0.01, Fig. 4k; intercostal: P < 0.05, Fig. 4i). The tdTomato + area was located between the nuclei of the muscle (Fig. 4f,h). In mice induced at E9, we could clearly identify tdTomato + cells in the pancreas (Fig. 4j). The lumber vertebrae showed the enrichment of tdTomato + cells in mice induced at E15 (Fig. 4m). These results clearly showed that Sox9 was expressed in muscle progenitor cells.

Sox9 expression in muscle progenitor cells in vitro.
We noticed that Sox9 was expressed in the cytoplasm of muscle progenitor cells in the myotome (Fig. 5a-d). Sox9 expression in sex cells translocates from the cytoplasm to the nucleus at the onset of male sexual differentiation 25 . To identify whether Sox9 expression during myogenesis translocates from the cytoplasm to the nucleus, we investigated the mouse myoblast C2C12 cell line. In the control group (undifferentiated C2C12 cells), Sox9 was clearly expressed in the cytoplasm ( Fig. 5e-g,l,m). Reverse transcription-polymerase chain reaction (RT-PCR) showed Sox9 messenger RNA (mRNA) expression in the control group (Fig. 5h). In myogenic induction, Sox9 was expressed in the nucleus ( Fig. 5i-k,o,p). In addition, in the control group, the immunofluorescence intensity of Sox9 in the cytoplasm was high compared to that in the nucleus (Fig. 5l-n). In contrast, myogenic induction resulted in Sox9 expression in the nucleus of C2C12 cells ( Fig. 5o-q).

Discussion
Several studies have revealed the function of Sox9 in tendon and bone development [2][3][4][12][13][14][15][16][17][18][19] . Nevertheless, the fundamental question of whether Sox9 is expressed in developing muscles has been largely neglected. To reveal whether Sox9 controls the development of three components (muscles, tendons, and bones) of the musculoskeletal system, we studied the expression of Sox9 in developing muscle. This study demonstrated that muscle progenitors show Sox9 expression. Therefore, we revealed that Sox9 controls all the main components of the musculoskeletal system.
Our finding that the development of muscles, tendons, and bones is controlled by the switching of Sox9 expression provides new perspectives on the development of the musculoskeletal system. In the early stage, Sox9 was expressed in progenitor cells of all components of the musculoskeletal system. Subsequently, it was detected in the MTJ, tendon, and bone. In the late embryonic stage, bones showed Sox9 expression. Therefore, it is necessary to switch on Sox9 expression in each component (muscle, tendon, and bone) for the development of the musculoskeletal system (Fig. 8). We considered that the simplification of the functioning of transcription factors allows muscles, tendons, and bones to easily establish the complex structure of the musculoskeletal system. www.nature.com/scientificreports www.nature.com/scientificreports/ It is well known that tendons and bones have the same origin. Tendons and bones are derived from the sclerotome, the lateral plate mesoderm, and the neural crest [2][3][4][5] . Tendon progenitor cells in the trunk develop from the syndetome, which is in the rear of the sclerotome 7,8 . In addition, in the sclerotome, Scx and Sox5 are coexpressed in the multipotent cell group, which can differentiate into cartilage and tendons, and are then expressed specifically in tendon and cartilage regions 7 . Sox9 and Scx were also detected in the subpopulations of tendon/ligament progenitor cells and chondroprogenitor cells 7,13 . In this study, muscle progenitor cells showed Sox9 expression both in vivo and in vitro. Therefore, the components of the musculoskeletal system (muscle, tendon, and bone) originate from Sox9 + progenitor cells. However, little is known about whether muscles, tendons, and bones have the same origin. This hypothesis is supported by the differentiation of somites into muscles, tendons, and bones 23 .
Previous studies on myoblastic cell lines in vitro have suggested that muscles, tendons, and bones have the same origin because they retain the capacity to differentiate into a chondrogenic, osteoblastic lineage or tenogenic lineage. According to Bettex-Galland and Wiesmann 29 , L6 myoblasts differentiate into chondrocytes under the influence of demineralized bone. Katagiri et al 20 . showed that bone morphogenetic protein-2 (BMP2) converts C2C12 cells into an osteoblast lineage. L6 myoblasts show Sox9 expression, which plays a role in chondrogenesis 21 . According to Uemura et al 30 ., myostatin promotes tenogenic differentiation of C2C12 myoblastic cells. C2C12 and L6 appear to be stem cells that can differentiate into each component of the musculoskeletal system.
A few researchers have reported that muscle shows Sox9 expression. Rat L6 myoblastic cells show Sox9 expression, but after several days of culture, there is a decline in the level of Sox9 21 . We demonstrated that muscle progenitors show Sox9 expression in vivo and in vitro and that Sox9 expression in C2C12 cells translocate from the cytoplasm to the nucleus during muscle induction. Since Sox9 represses muscle gene expression 22 , myoblasts may retain their undifferentiated state. To retain an undifferentiated state, muscle progenitor cells expressed Sox9 in the cytoplasm (Fig. 8).
The Wnt1 gene plays an important role in the development of the neural crest and its derivatives, and the Wnt1 Cre transgenic mouse line is widely used to investigate neural crest development. To clarify the role of Sox9 in the development of neural crest derivatives, Mori-Akiyama et al 4 . performed gross and histological observations using Wnt1Cre;Sox9 flox/flox mice. The analysis showed that all cartilage and endochondral bones, such as the anterior part of the cranial base, Meckel's cartilage, the malleus, the incus, and the nasal capsule, were missing. Although the bone features of Wnt1Cre;Sox9 flox/flox mice have been described 4 , little is known about the muscles and tendons of these mice. We found that Wnt1Cre;Sox9 Flox/+ mice have a cleft palate, deficient Meckel's cartilage, a small mandibular body, and hypoplastic condylar cartilage (Fig. 7). Moreover, these mice show hypoplasia of masticatory muscles and their tendons (Fig. 7). Because previous studies showed that Sox9 plays an important role in tendon and bone development 18,19 , it is easy to understand why tendon and bone hypoplasia is observed in Wnt1Cre;Sox9 Flox/+ mice. The reason why muscle hypoplasia occurs in Wnt1Cre;Sox9 Flox/+ mice is that defects in Sox9 expression in muscle connective tissues cause hypoplasia of muscle fibers (Figs. 6 and 7) 27,28 . www.nature.com/scientificreports www.nature.com/scientificreports/

Conclusion
Sox9 is expressed in progenitor cells of all components of the musculoskeletal system. However, the muscles and tendons do not express Sox9 during the late embryonic period. Therefore, it is essential to switch on Sox9 expression in each component (muscle, tendon, and bone) during the development of the musculoskeletal system (Fig. 8). In addition, a decrease in Sox9 expression in muscle-associated connective tissues, tendons, and bones leads to hypoplasia of the muscle, tendon, and bone. Therefore, Sox9 controls the development of each component of the musculoskeletal system.   d, f, h, j) E9 and (c, e, g, i, m) E15 and analyzed at E18. The number of tdTomato + cells in muscles induced at E9 and analyzed at E18 was much larger than that induced at E15 and analyzed at E18 (k, i; masseter: P < 0.01; intercostal: P < 0.05). Sox9, SRY-box containing gene 9. (2020) 10:8425 | https://doi.org/10.1038/s41598-020-65339-9 www.nature.com/scientificreports www.nature.com/scientificreports/ R26 tdTomato (B6;129S6-Gt(ROSA)26Sor tm14(CAG-tdTomato)Hze /J) mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA). All mice were bred under specific-pathogen-free conditions. Wnt1 Cre and Sox9 creER mice were mated with R26 tdTomato mice to generate Wnt1 Cre ; R26 tdTomato (Sox9 creER ; R26 tdTomato ) mice. Wnt1 Cre transgenic mice were mated with Sox9 flox/flox mice to generate Wnt1 Cre ; Sox9 flox/+ mice. We performed PCR to genotype each strain according to the instructions of the Jackson Laboratory. A female mouse was housed with a male mouse overnight, and noon of the day when we observed the vaginal plug was designated as E0.5.

Methods
To trace the lineages of Sox9-expressing cells in developing muscle, we generated double-transgenic Sox9 creERT2 /Rosa26-loxP-stop-loxP-tdTomato reporter mice in whom Cre expression in Sox9 + progenitor cells could be induced at different developmental stages by administration of tamoxifen (Tam). www.nature.com/scientificreports www.nature.com/scientificreports/ Histological analysis. We fixed fetal tissues in 4% phosphate-buffered paraformaldehyde (PFA). When we made the paraffin blocks, we decalcified the specimens using 10% ethylenediaminetetraacetic acid (EDTA) for 7 days at room temperature. We prepared the paraffin blocks using standard methods and cut a series of 5-μm-thick tissue sections using a sliding microtome (Leica Biosystems, Wetzlar, Germany). When we made the frozen blocks, we incubated the tissue samples overnight in 30% sucrose in phosphate-buffered saline (PBS). Then, we embedded the tissue samples in Tissue-Tek OCT compound (Sakura Finetek, Japan). Next, we cut 15-μm-thick sections using a CM1950 cryomicrotome (Leica Biosystems, Wetzlar, Germany) and stained the sections with Hoechst 33342 (1:1000 dilution; Thermo Fisher Scientific, Waltham, MA, USA) to visualize the nuclei using an Axio Imager wide-field fluorescence microscope (Zeiss, Oberkochen, Germany). We analyzed the images using ImageJ (NIH, Bethesda, MD, USA). For 3D reconstruction, we loaded the digital images of the hematoxylin and eosin (H&E)-stained serial sections into Amira (Visage Imaging, Inc., Richmond, Australia) by using a voxel size appropriate for the section thickness.
Double-staining of ALP and desmin. We stained the sections using an ALP staining kit (Primary Cell, Hokkaido, Japan) according to the manufacturer's instructions. Briefly, we rinsed the sections in running distilled water for 1 min, after which we added 50 mL of staining solution dropwise to each section. Then, we incubated the sections for 3 h at room temperature until the ALP staining was a bright intense blue, and then we washed the sections with PBS. Next, we incubated them in 3% hydrogen peroxide with methanol for 30 min, subjected them to several additional washings with PBS, and then incubated them in 3% bovine serum albumin for 1 h to block nonspecific binding. Subsequently, we treated the sections with a primary antibody against desmin (1:1000 dilution; Abcam, Cambridge, UK) and incubated them overnight in a moisture chamber at 37 °C. Then, we applied a secondary antibody using EnVision TM + Dual Link System-horseradish peroxidase (HRP; Dako, Tokyo, Japan) www.nature.com/scientificreports www.nature.com/scientificreports/ at room temperature. Finally, after several more washes with PBS, we treated the sections with ImmPACT DAB (Funakoshi) to detect any reaction and then inspected them after counterstaining with hematoxylin.
RNA in situ hybridization. The anti-sense probe for Scx has been described previously 31 . We labeled the probe with digoxigenin (DIG RNA labeling mix; Roche, Rotkreuz, Switzerland) and performed hybridization by following a standard protocol. Briefly, we fixed the sections for 10 min using 4% PFA, digested them using 1 μg/ mL proteinase K (Roche) for 5 min, and then fixed them again for 5 min. Next, we performed acetylation for 10 min in a solution containing triethanolamine, hydrochloric acid, and acetic anhydride. We preblocked the sections using hybridization buffer (50% formamide, 5x saline sodium citrate [SSC], 50 μg/mL yeast tRNA, 1% sodium dodecyl sulfate [SDS], and 50 μg/ml heparin) and subsequently incubated them with an Scx anti-sense probe diluted to 1 ng/μL in hybridization buffer. After washing away the unbound probes using SSC buffer, we detected the probes in the sections using antidigoxigenin antibody conjugated to ALP (Roche) and BM purple (Roche).
Tamoxifen treatment. We dissolved tamoxifen (T5648; Sigma-Aldrich, St. Louis, MO, USA) in ethanol and then diluted it in corn oil (C8267; Sigma-Aldrich) at a concentration of 10 mg/mL, as described previously 32 . Then, we injected 1.5 mg or 3 mg of tamoxifen into the peritoneal cavity of pregnant mice at E9 or E15, respectively, and coinjected 1 mg/40 g of progesterone (P8783; Sigma-Aldrich).
Cells were washed with warm PBS (37 °C), fixed in 4% paraformaldehyde at room temperature for 30 min and then subjected to immunofluorescence staining for myotubes and 4,6-diamidino-2-phenylindole (DAPI) staining for nuclei. The primary antibody was rabbit anti-sox9 antibody (1:1000, Merck Millipore), and the secondary antibody was donkey anti-goat IgG Alexa Fluor 555 (1:1000, Thermo Fisher Scientific). To verify the validity of the immunohistostaining results, we performed RT-PCR on the control group (undifferentiated C2C12 cells).
Statistical analysis. All statistical analyses were performed by using SPSS Statistics 21.0 (IBM, Armonk, NY, USA). P-values were calculated using Student's t-test. The between-group differences for which P < 0.05 were considered statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001 are used throughout the report). Error bars show the standard deviation of the mean.