Scleraxis is a transcriptional activator that regulates the expression of Tenomodulin, a marker of mature tenocytes and ligamentocytes

Tenomodulin (Tnmd) is a type II transmembrane glycoprotein predominantly expressed in tendons and ligaments. We found that scleraxis (Scx), a member of the Twist-family of basic helix-loop-helix transcription factors, is a transcriptional activator of Tnmd expression in tenocytes. During embryonic development, Scx expression preceded that of Tnmd. Tnmd expression was nearly absent in tendons and ligaments of Scx-deficient mice generated by transcription activator-like effector nucleases-mediated gene disruption. Tnmd mRNA levels were dramatically decreased during serial passages of rat tenocytes. Scx silencing by small interfering RNA significantly suppressed endogenous Tnmd mRNA levels in tenocytes. Mouse Tnmd contains five E-box sites in the ~1-kb 5′-flanking region. A 174-base pair genomic fragment containing a TATA box drives transcription in tenocytes. Enhancer activity was increased in the upstream region (−1030 to −295) of Tnmd in tenocytes, but not in NIH3T3 and C3H10T1/2 cells. Preferential binding of both Scx and Twist1 as a heterodimer with E12 or E47 to CAGATG or CATCTG and transactivation of the 5′-flanking region were confirmed by electrophoresis mobility shift and dual luciferase assays, respectively. Scx directly transactivates Tnmd via these E-boxes to positively regulate tenocyte differentiation and maturation.


Expression of Tnmd in tenocytes in vivo and in vitro.
Tnmd is predominantly expressed in dense connective tissue, such as tendons and ligaments 1,2,8,9,18 . We compared the expression of Scx and Tnmd in the developing mouse embryos at E12.5 and E13.5 by whole-mount in situ hybridisation ( Fig. 1a-f). At E12.5, Tnmd was detected in the developing axial tendons along the cervical and thoracic spine, but its expression was low in the limbs (Fig. 1a). In the developing forelimb, Tnmd was detected in the primordia of the triceps brachii tendon and extensor digitorum communis tendon (Fig. 1b), while Scx was widely detected in the tendon primordia of the forelimb (Fig. 1c). At E13.5, Tnmd was expressed in the axial tendons along the entire spine and limbs (Fig. 1d). Tnmd and Scx were coexpressed with the developing tendons (Fig. 1e,f); however, Scx expression was also detected in the developing joint capsules at high levels ( Fig. 1f).
We also analysed the expression of Tnmd along with other differentiation markers, including Scx, collagen alpha-2(I) chain (Col1a2), and collagen alpha-1(II) chain (Col2a1), in various mesenchymal cells (Fig. 1g,  Supplementary Fig. 1). Tnmd mRNA was detected at high levels in tenocytes and to a lesser extent in C3H10T1/2 cells, whereas Tnmd expression was undetectable in fibroblastic NIH3T3 cells, osteoblastic MC3T3-E1 cells, chondrogenic ATDC5 cells, or primary costal chondrocytes by northern blot analysis (Fig. 1g). In contrast, Scx, a marker of both tendon progenitors and tenocytes 22 , was expressed in tenocytes, C3H10T1/2 cells, MC3T3-E1 cells, differentiated ATDC5 cells, and chondrocytes but not in NIH3T3 cells, suggesting that not only tendon-derived cells but also skeletal cells express Scx in vitro (Fig. 1g). We also detected the expression of Twist1 in rat tenocytes, NIH3T3 cells, and C3H10T1/2 cells (Fig. 1h, Supplementary Fig. 2). Interestingly, the mRNA level of Tnmd dramatically decreased during serial passaging of rat tenocytes, while increased mRNA levels of Scx were observed in the fourth or later culture passages of tenocytes and high levels of Col1a2 were detected during serial passages until the 10 th culture (Fig. 1i, Supplementary Fig. 3). These results clearly suggest that Tnmd is a specific marker gene for tenocytes both in vivo and in vitro.

Indispensable role of Scx in expression of Tnmd gene and protein in tendons and ligaments.
Mouse Scx consists of two exons (Fig. 2a). To generate Scx-deficient mice, we used TALEN-mediated technology rather than a conventional homologous recombination-based strategy using embryonic stem cells. To disrupt the Scx gene using TALENs, we designed the TALEN recognition sequences to be within exon 1 of the mouse Scx locus (Fig. 2a), so that most of the Scx protein would be lost due to a frameshift mutation after creation of a double-stranded break by TALENs. TALEN mRNAs produced by in vitro transcription were microinjected into the cytoplasm of fertilized oocytes. After in vitro culture of the injected oocytes overnight, two-cell embryos were SCientiFiC RepoRTS | (2018) 8:3155 | DOI: 10.1038/s41598-018-21194-3 transferred into pseudopregnant female mice. Fourteen pups with deletion mutations were identified from among 18 newborn mice by genotyping and direct sequencing of the amplified DNA (Fig. 2b). We obtained eight founder mice with deletion mutations that would yield a disruptive frameshift mutation and a premature stop codon (#1, #3, #4, #6, #8, #11, #13, and #14), thereby inactivating the Scx protein (Fig. 2b). We crossed founder #13 with a 11-base pair (bp) deletion, both resulting in frameshift and premature stop codons shortly downstream, with a C57BL/6 wild-type mouse to obtain the heterozygote through germline transmission. Heterozygous Scx +/− and homozygous Scx −/− mice were viable at two weeks, while Scx −/− mice exhibited hypoplastic tendon formation (Fig. 2d,f) compared to wild-type mice (Fig. 2c,e), as previously reported 10,21 . (a-f) Expression patterns of Tnmd (a,b,d,e) and Scx (c,f) in embryos at E12.5 (a-c) and E13.5 (d-f) are shown. Whole-mount in situ hybridisation with antisense probes for these two genes were performed. Lateral (a,d) and dorsal views of the forelimbs (b,c,e,f) are shown, respectively. Red, black, and blue arrowheads indicate the developing triceps branchii tendon, joint capsules between proximal and middle phalanx, and extensor digitorum communis, respectively. Open red, black, and blue arrowheads indicate the developing extensor carpi radialis brevis/longus tendon, joint capsules between the metacarpus and proximal phalanx, and extensor carpi radialis tendon, respectively. Asterisks (b,c,e,f) indicate the extensor digitorum communis tendon. (g) Total RNA was prepared from NIH3T3 cells on day 3, C3H10T1/2 cells on day 3, MC3T3-E1 cells on day 14, differentiated ATDC5 cells on day 21 (Diff. ATDC5), primary rat costal chondrocytes on day 24 (Chondrocytes), and a secondary culture of rat limb tendon-derived tenocytes (Tenocytes To examine the expression of Tnmd in Scx +/− and Scx −/− mouse tissues, we performed in situ hybridisation in frozen tissue sections ( Fig. 3a-o). In the hindlimb of a Scx +/− embryo at E18.5 and a Scx +/− neonate at postnatal day 1 (P1), expression of both Col1a1 (Fig. 3a,j) and Scx (Fig. 3b,c) was detected in the tendons and ligaments of the knee (Fig. 3a,b) and ankle (Fig. 3c,j). Similar to Scx, Tnmd was expressed in Col1a1-positive tendons and ligaments of the knee (Fig. 3d,e) and ankle (Fig. 3j,k) of the Scx +/− neonate at P1. In the Scx −/− neonate at P1, Col1a1 expression persisted in the tendons and ligaments (Fig. 3g,m), whereas Tnmd expression was nearly absent (Fig. 3h,n). Expression of Tnmd was undetectable in the anterior and posterior cruciate ligaments and distal region of the Achilles tendon of Scx −/− at P1 (Fig. 3h,n). However, only faint expression of Tnmd was detected in the quadriceps femoris tendon, patella ligament, and proximal portion of the Achilles tendon of Scx −/− at P1 (Fig. 3h,n). We performed immunostaining to examine the localization of Tnmd proteins in the Achilles tendon ( Fig. 3p-s). In the ankle of Scx +/− and Scx −/− at P1, hyaline cartilage of calcaneus was positively stained with an anti-Chmd antibody (Fig. 3p-s) and localization of Col1 was observed in the Achilles tendon (Fig. 3p,r). In good agreement with the expression pattern of mRNA, the Tnmd protein was localized to the Achilles tendon of Scx +/− at P1 (Fig. 3q), whereas in the Scx −/− neonate, faint or no expression of Tnmd was detected in the proximal or distal portion of the Achilles tendon, respectively (Fig. 3s).
We isolated tenocytes from rat limb tendons (Fig. 4a) and depleted Scx in these cells by RNA interference. In tenocytes transfected with siScx-1 or siScx-2, the level of Scx was decreased to less than 25% of that of the control at 72 h after lipofection (Fig. 4b). The level of Tnmd was markedly decreased to 17% and 18% by gene silencing of Scx with siScx-1 and siScx-2, respectively (Fig. 4b). The expression level of Col1a2 was slightly decreased to 80% by siScx-2, but no significant decrease in Col1a2 expression was detected in cells transfected with siScx-1 (Fig. 4b).
These results suggest that Tnmd expression depends on Scx in postnatal mature tenocytes.
Determination of transcription start sites in mouse Tnmd gene. Tnmd was cloned as a related gene of Chmd, which was identified as a cartilage-derived angiogenesis inhibitor 1,2 . Mouse Tnmd was mapped to chromosome XqE3 (nucleotides 133,851,207-133,865,578 in the UCSC Genome Browser [GRCm38/mm10 assembly]). Based on the genomic and cDNA sequences (GenBank AF219993) of mouse Tnmd 2 , we determined the exon-intron boundaries ( Table 1) and found that Tnmd consists of seven exons spanning approximately 15 kb (Fig. 5a). Protocadherin (Pcdh19) is located more than 160-kb upstream from the first exon of Tnmd. Tetraspanin 6 (Tspan6) is located immediately downstream of Tnmd, followed by sushi repeat-containing protein, X-linked 2 (Srpx2) and synaptotagmin-like 4 (Styl4) (Fig. 5a). In situ hybridisation with antisense probes for Col1a1 (a,d,g,j,m), Scx (b,c), and Tnmd (e,h,k,n) or a sense probe for Tnmd (f,i,l,o) was performed on semiserial sections prepared from the legs of a Scx +/− embryo at E18.5 (a,b), a Scx +/− neonate at P1 (c-f,j-l) or a Scx −/− neonate at P1 (g-i,m-o). Sagittal sections of the knee joint (a,b,d-i) and the ankle (c,j-o) are shown. Black and hollowed arrowheads in (g,h,m, and n) indicate the tendons/ligaments that were positive or negative staining with the Tnmd antisense probe, respectively. (p-s) Double immunostaining of Chmd (green) and Col1 (red) (p,r) or Tnmd (red) (q,s) was performed on frozen sections prepared from Scx +/− (p,q) and Scx −/− (r,s) at P1. Sagittal sections of the ankle are shown. White and hollowed arrowheads in (r,s) indicate the regions of Achilles tendon that were positive and negative for staining with the Tnmd antibody, respectively. acl, anterior cruciate ligament; At, Achilles tendon; ca, calcaneus; fe, femur; pa, patella; pcl, posterior cruciate ligament; pl, patella ligament; qft, quadriceps femoris tendon. Scale bars, 500 μm. As an initial step in the identification and characterisation of the mouse Tnmd promoter, we analysed the transcription start site using CapSite hunting technology. We performed nested PCR using CapSite cDNA derived from mouse embryos at E16 and examined six independent cDNA clones. Subsequent sequence analysis revealed two different transcription start sites located at −58 and −84 bp from the translational start site of mouse Tnmd (Fig. 5b). Designating the distal transcriptional start site as +1, a putative TATA box was found −25 bp upstream of the distal transcription start site (Fig. 5b). We also found that half of the cDNA clones amplified from CapSite cDNA contained a 90-nucleotide insertion between the cDNA sequences corresponding to exon 2 and exon 3 (data not shown). The inserted region encoding a 30-amino acid sequence was determined to be a single independent exon, located 348-bp upstream of exon 3 in mouse Tnmd. The boundary sequences of this inserted region of the genome follow the GT-AG rule. To date, Tnmd transcripts containing the insertion sequence have only been found in cDNA clones amplified from mouse 16-day embryo CapSite cDNA.
Scx is a member of the Twist family of bHLH transcription factors, which functions through dimerization with E-proteins and binding to E-boxes (CANNTG) 24 . Both Scx and Twist1 were co-expressed with Tnmd in tenocytes (Fig. 1g,h) and upregulated the expression of Tnmd in chick tenocytes, whereas overexpression of Myog, a myogenic bHLH factor, resulted in downregulation of the mRNA levels of Tnmd in chick tenocytes 25 . To examine whether Scx is directly involved in the transactivation of mouse Tnmd, we performed dual luciferase assays in rat tenocytes by co-transfecting the luciferase reporter containing five E-boxes (pGL4.10 −1030/+84), three E-boxes (pGL4.10 −525/+84), or two E-boxes (pGL4.10 −295/+84) with various combinations of expression vectors for FLAG-tagged mouse Scx (fmScx) or Twist1 (fmTwi) and/or its heterodimeric partners E12 (fmE12) or E47 (fmE47) tagged with FLAG ( Fig. 8). E12 and E47 are products of two alternatively spliced mRNAs and have nearly identical sequences, except in the stretches encoding the DNA-binding domain 26 . The luciferase activities of pGL4.10 −1030/+84, pGL4.10 −525/+84, or pGL4.10 −295/+84 cotransfected with fmScx and/or fmE12 or fmE47 were significantly higher than that in the control (Fig. 8a,b,c). Similar luciferase activity of pGL4.10 −1030/+84 was observed when cotransfected with fmTwi and/or fmE12 or fmE47 (Fig. 8d). We then tested whether Scx and E12 or E47 directly interact with these E-boxes identified in the promoter region using electrophoretic mobility shift assays (EMSA). EMSA was performed with nuclear extracts containing fmScx and FLAG-tagged human E12 (fhE12), fmE12, or fmE47 using biotin-labelled oligonucleotides containing the E-box and/or mutated E-box ( Table 2). The expected molecular weight of each translated protein was confirmed by western blotting (data not shown). Of the five consensus E-box sequences, a specific shift band was detected when fmScx and fhE12, fmE12 or fmE47 was incubated with a biotin-labelled E1E2 or E5 oligonucleotide (black arrowheads in Fig. 9a-c). In the presence of an anti-FLAG antibody, a supershifted band was clearly detected (hollowed arrowheads in Fig. 9a-c), suggesting that Scx and E12 or E47 heterodimer complexes directly interact with the oligonucleotide E1E2 or E5. To identify which E-box on E1E2 is responsible for the specific binding of Scx/E12, we performed EMSA with mutated oligonucleotides (Table 2, Fig. 9d). Both a specific shift band and supershifted band were detected when fmScx and fhE12 were incubated with the biotin-labelled M1E2 oligonucleotide containing a mutated E1 sequence and normal E2 sequence. No specific bands were detected with the biotin-labelled M1M2, E1M2, or M5 oligonucleotides (Fig. 9d).
We also tested whether Twist1 and/or E12 or E47 bind to E2 (CATCTG) and E5 (CAGATG) by EMSA. EMSA was performed with nuclear extracts containing fmTwi and fhE12, fmE12, or fmE47, using biotin-labelled oligonucleotides containing the E-box and/or the mutated E-box (Table 2). Specific shift bands were detected when fmTwi and fmE12 or fmE47 were incubated with a biotin-labelled E1E2 or E5 oligonucleotide (black arrowheads in Fig. 10a,b). In the presence of an anti-FLAG antibody, supershifted bands were clearly detected (hollowed arrowheads in Fig. 10a,b). Both specific shift bands and supershifted bands were detected when fmTwi and fhE12 were incubated with the biotin-labelled M1E2 oligonucleotide containing a mutated E1 sequence and normal E2 sequence. No specific bands were detected with the biotin-labelled M1M2 or E1M2 oligonucleotides (Fig. 10a). These results suggest that Twist1 and/or E12 or E47 heterodimer complexes directly interact with the oligonucleotide E2 or E5.

Discussion
Here, we demonstrated that Scx transactivates the promoter of mouse Tnmd, which consists of seven exons encoding a type II transmembrane protein, marking tenocytes and ligamentocytes. Silencing of Scx caused marked downregulation of Tnmd in cultured tenocytes. Loss of Scx function in vivo nearly abolished Tnmd expression in both tendons and ligaments during musculoskeletal development and growth. Of the five E-boxes around the TATA box, CAGATG and CATCTG are preferential binding sites for Scx, as demonstrated by EMSA. Dual luciferase assays in Tnmd-expressing tenocytes revealed enhancer activity upstream of the promoter region. Thus, Scx directly transactivates Tnmd via these E-boxes to positively regulate tenocyte differentiation and maturation.
We previously reported that chick Tnmd is present at early stages of tendon/ligament formation and is persistently expressed in mature tendons and ligaments at high levels 8 . In mice, our whole-mount in situ hybridisation analysis revealed that Scx expression in the developing tendon and ligament primordia precedes the expression of Tnmd. Later in musculoskeletal development, Scx and Tnmd are coexpressed with mature tendons and ligaments. Thus, in mice, Scx is an early marker gene expressed in both tendon/ligament progenitors and tenocytes/ligamentocytes, whereas Tnmd acts as a late marker gene to indicate mature tenocytes/ligamentocytes. Consistent with our previous report that Scx is expressed in a subpopulation of skeletogenic Sox9 + progenitor cells that contribute to the formation of future enthesis 18 , Scx was also detected in skeletal cells, including chondrogenic ATDC5, osteogenic MC3T3-E1, and chondrocytes, in vitro. In contrast, Tnmd is not expressed in these skeletogenic cells and is highly specific to mature tenocytes. This is in good accordance with our in situ hybridisation data showing that Tnmd is detected in the developing tendons and ligaments in vivo. Unlike Col1a2, Tnmd expression was dramatically downregulated during serial passaging of tenocytes, suggesting that Tnmd is a good indicator of mature phenotypes in cultured tenocytes. In our rotator cuff injury model using rats, local administration of FGF-2 resulted in biomechanical and histological improvement of the repaired rotator cuff by promoting growth of tenogenic progenitor cells in association with a significant increase in Tnmd-positive cells in the midsubstance of the tendon 27 . A strong positive correlation between the location of the aligned collagen fibre orientation and expression levels of Tnmd mRNA was observed in this model 27 . In chick embryos, Tnmd expression was not detected in oval immature tenocytes, but was observed in elongated mature tenocytes of the embryonic Achilles tendon 8 . Taken together, a high level of Tnmd expression is a good indicator of the maturation of tenocytes localized between aligned collagen fibres under both physiological and pathological conditions. Interestingly, we also detected the expression of Tnmd in C3H10T1/2 cells, which is an undifferentiated mesenchymal cell line derived from mouse embryonic fibroblasts. C3H10T1/2 cells have the potential to differentiate into tenocytes 28 . As previously reported, Tnmd is also expressed in cultured tendon stem/progenitor cells (TSPC) 19 and Scx-expressing bone marrow stromal cells 29 . Loss of Tnmd results in reduced self-renewal and augmented senescence of tendon/ligament progenitor cells without affecting the multipotential of TSPC 19 . Thus, a low level of Tnmd expression may be required for maintenance of the tenogenic potential in stem/progenitor cells in vitro.
Scx was originally isolated from a mouse E14.5 cDNA library, using the yeast two-hybrid system as a novel partner of bHLH protein that dimerizes with E12 24 . Additionally, Scx binds to the mouse muscle creatine kinase enhancer (CCCAGATGTGGCTGCTCCC) as a heterodimer with E12 24 . It was also reported that Scx acts as a transcriptional activator for Col1a1 in tendons via binding to the E-box site CACGTG 30 and Aggrecan (Acan) in osteosarcoma-derived ROS17/2.8 cells 31 . Scx/E47 heterodimers bind to CAGGTG to regulate Col2a1 expression together with Sox9 and p300 32 . In this study, we analysed five E-boxes (E1: CACTTG; E2: CATCTG; E3: CAATTG; E4: CAAATG; and E5: CAGATG) upstream and downstream of the TATA box in the mouse Tnmd promoter. EMSA revealed that E2 (CATCTG) and E5 (CAGATG) are Scx-binding E-boxes. Li et al. performed chromatin immunoprecipitation, to show that Scx binds to the region containing both E4 (CAAATG) and E5 (CAGATG) 28 . In this study, we determined that Scx does not bind to E4 by our EMSA analysis using nuclear extract of HEK293T cells expressing Scx. We also found that Twist1 binds to these Scx-binding E-boxes and transactivates the genomic region (−1030 to +84). Thus, not only Scx but also Twist1 regulate Tnmd expression via E2 (CATCTG) and E5 (CAGATG) in the promoter region.
Mouse Scx gene consists of two exons and is located within the third intron of block of proliferation 1 (Bop1) transcribed in the opposite orientation 33   In Scx-deficent embryos, severe defects were observed in force-transmitting and intermuscular tendons; however, muscle-anchoring tendons and ligaments were not affected 21 . Homozygous Scx Cre knock-in mice exhibited defective maturation of tendons and ligaments as well as entheseal and sesamoid cartilage in which Scx was transiently expressed during development 10 . Similar phenotypes were observed in Scx null mice generated by TALEN-mediated technology. Our in situ hybridisation analysis revealed that Tnmd expression was nearly Figure 9. Preferential binding of Scx/E12 or Scx/E47 to E2 and E5. Double-stranded synthetic oligonucleotides containing the E-box were end-labelled with biotin and used as a probe for EMSA. The oligonucleotides sequences used in EMSA (E1E2, M1M2, E1M2, M1E2, E3, E4, E5, and M5) are shown in Table 2. Results shown in (a), (b), (c) and (d) are from assays using oligonucleotides containing the consensus E-box sequences (E1E2, E3, E4, and E5) and mutated oligonucleotides (M1M2, E1M2, M1E2, and M5), respectively. Biotin-labelled probes were incubated with nuclear extracts from HEK293T cells transfected with the empty vector pcDNA3 as a control, or expression vectors encoding the FLAG fusion of human E12 (fhE12), mouse E12 (fmE12), mouse E47 (fmE47), and mouse Scx (fmScx). The binding reactions were performed in the presence or absence of non-labelled oligonucleotides as binding competitors (Comp) or an anti-FLAG antibody (Ab). The positions of shifted and super-shifted bands are shown with black and hollowed arrowheads, respectively. absent in both tendons and intracapsular ligaments, such as the anterior cruciate ligaments of Scx null mice at P1. Similarly, Tnmd was minimally detected in the distal portion of the Achilles tendon of P1 neonates lacking Scx. Silencing of Scx in tenocytes isolated from 2-week old rat limb tendons also resulted in significant downregulation of Tnmd. These results suggest that Tnmd expression in tenocytes is dependent on Scx both in vivo and in vitro.
Gene and protein expression of Tnmd are nearly absent in Scx null embryos, suggesting a crucial role for Scx in Tnmd expression. However, a loss of Tnmd expression was observed in the tenth cultures of tenocytes, although the cells maintained Scx expression at a high level. Scx + skeletogenic cells, such as MC3T3-E1 and ATDC5 cells, also do not express Tnmd. Enhancer activity was increased in the upstream region (−1030 to −295) in tenocytes, but not in NIH3T3 cells not expressing Tnmd. As shown in whole-mount in situ hybridisation, Scx expression precedes that of Tnmd during embryogenesis. These results suggest that Scx is necessary for the induction of Tnmd expression in mature tenocytes, but not sufficient in immature tenocytes and skeletogenic cells. This is also consistent with our previous finding that retroviral overexpression of Scx in the chick hindlimb resulted in significant upregulation of Tnmd in tendons but did not induce ectopic Tnmd expression outside the tendinous tissue 8 . We speculate that some additional transcription factors are required for the expression of Tnmd. Further studies are underway to elucidate other transcriptional factors that coordinately regulate the transcription of Tnmd with Scx in tendons and ligaments.  Table 2. Results shown in (a) and (b) are from assays using the oligonucleotides containing the consensus E-box sequences (E1E2 and E5) and mutated oligonucleotides (E1M2, M1E2, and M5), respectively. Biotin-labelled probes were incubated with nuclear extracts from HEK293T cells transfected with the empty vector pcDNA3 as a control, or expression vectors encoding the FLAG fusion of human E12 (fhE12), mouse E12 (fmE12), mouse E47 (fmE47), and mouse Twist1 (fmTwi). Binding reactions were performed in the presence or absence of non-labelled oligonucleotides as binding competitors (Comp) or an anti-FLAG antibody (Ab). The positions of shifted and super-shifted bands are shown with black and hollowed arrowheads, respectively. Cell culture. Mouse limb and tail tendons were isolated from 4-week-old male mice and seeded onto 100-mm cell culture dishes. Tenocytes outgrown from the tendon were passaged once or twice and grown in minimum essential medium Eagle alpha modification (α-MEM) supplemented with 10% foetal bovine serum (FBS). Rat tenocytes were isolated from limb tendons of 7-or 14-day-old male Wistar rats. Minced tendons were incubated with 0.1% ethylenediaminetetraacetic acid (EDTA) (Dojin, Tokyo, Japan) at 37 °C for 20 min and digested with 0.05% trypsin containing 0.53 mM EDTA (Gibco, Grand Island, NY, USA) at 37 °C for 5 min followed by digestion with 0.1% collagenase (Roche, Basel, Switzerland) at 37 °C for 10 min. Tenocytes were grown in α-MEM supplemented with 10% FBS (Cambrex, East Rutherford, NJ, USA) and 50 μg/mL kanamycin (Sigma-Aldrich, St. Louis, MO, USA) on dishes coated with type I collagen (Koken, Tokyo, Japan). Rat chondrocytes were isolated from rib cartilages of 4-week-old Wistar rats. Cartilage minces were incubated with 0.1% EDTA at 37 °C for 20 min and digested with 0.15% trypsin (Difco, Detroit, MI, USA) at 37 °C for 1 h and 0.1% collagenase at 37 °C for 3 h. Chondrocytes were grown in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium (DMEM/F12 medium; Asahi Techno Glass, Haibara, Japan) supplemented with 10% FBS (Sigma-Aldrich). NIH3T3 cells were grown in DMEM (Sigma-Aldrich) supplemented with 10% FBS (Biological Industries, Beit Haemek, Israel) and 50 μg/mL kanamycin. C3H10T1/2 cells were grown in DMEM (Sigma-Aldrich) supplemented with 10% FBS (Sigma-Aldrich). MC3T3-E1 cells were grown in α-MEM supplemented with 10% FBS. ATDC5 cells were grown in DMEM/F12 medium (Asahi Techno Glass) supplemented with 5% FBS (Hana-Nesco Bio Corp., Tokyo, Japan), 10 μg/mL human insulin (Roche), 10 μg/mL human transferrin (Roche), and 3 × 10 −8 M sodium selenite (Sigma-Aldrich), as described previously 36 . Cells were incubated at 37 °C in a 5% CO 2 atmosphere.
Quantitative RT-PCR (qRT-PCR) analysis. Total RNA was extracted from rat tenocytes, using an RNeasy Plus Mini kit (Qiagen, Hilden, Germany). Two-hundred nanograms of total RNA was used to synthesize cDNA with a PrimeScript RT reagent kit (Takara Bio, Shiga, Japan). qRT-PCR was performed using SYBR Premix Ex Taq II (Takara Bio) on a StepOne instrument (Life Technologies, Carlsbad, CA, USA). Relative mRNA expression was normalized to that of 18 S rRNA and calculated using the 2 −ΔΔCT method. Specific primers for qRT-PCR are listed in Table 3.
Determination of transcriptional start sites. The transcriptional start sites of the mouse Tnmd gene were determined by the CapSite hunting method with 16-day-old mouse embryo CapSite cDNA (Nippon Gene, Toyama, Japan), according to the manufacturer's instructions. CapSite cDNAs were used as a template for the first round of polymerase chain reaction (PCR) with a forward primer (1RC: 5′-CAAGGTACGCCACAGCGTATG-3′) and mouse Tnmd gene-specific reverse primer (mTnmd-R1; 5′-CACTGGTAGGAAAGTGAAGATCTTC -3′). Samples were amplified for 35 cycles under the following conditions: denaturation for 30 s at 94 °C, annealing for 30 s at 55 °C, and extension for 30 s at 72 °C. Products of the first round of PCR were used as the template in the second round of nested PCR. Nested PCR was performed with a forward primer (2RC: 5′-GTACGCCACAGCGTATGATGC-3′) paired with the mouse Tnmd gene-specific reverse primers (mTnmd-R2: 5′-CACCGTCCTCCTCAAAGTCCTG-3′; mTnmd-R3: 5′-CATTCTCATCTATTTCTTCCTCTGG-3′; and mTnmd-R4: 5′-CACTGACTGTTCAAAGAAAGTTG-3′). Samples were amplified for 35 cycles under the following conditions: denaturation for 30 s at 94 °C, annealing for 30 s at 57 °C, and extension for 30 s at 72 °C. The amplified products were cloned into pCRII-TOPO (Invitrogen) and sequenced with the 310 Genetic Analyzer (Applied Biosystems).

Construction of expression vectors.
The entire coding sequence of mScx was amplified from mScx cDNA (Cserjesi et al. 24 ) using LA Taq polymerase (Takara) with a forward primer (mScxF) and reverse primer (mScxR) for mScx and subcloned into pCRII-TOPO (mScxpCRII; Invitrogen). FLAG-tagged mScx was then amplified from mScxpCRII with the forward primer containing the coding sequence of mScx followed by the FLAG peptide coding sequences with a NotI site (FLAGmScxF) and a specific reverse primer (FLAGmScxR) using PrimeStar DNA polymerase (Takara). FLAG-tagged mScx was inserted into the EcoRV site of pcDNA3 vector (Invitrogen) to construct pcDNA3-FLAG-mScx. For construction of FLAG-tagged mouse E12 (fmE12) and mouse E47 (fmE47), a DNA fragment from the second codon to the stop codon were amplified with a forward primer containing a NotI site (FLAGmE12/47 F) and reverse primer with a NotI site (FLAGmE12/47 R). To construct FLAG-tagged mTwist1 (fmTwi), a DNA fragment from the second codon to the stop codon was amplified with a forward primer containing a NotI site (FLAGmTwist1F) and reverse primer with a NotI site (FLAGmTwist1R). The amplified fragments were inserted in place of the mScx fragment of pcDNA3-FLAG-mScx. The mouse E12 (BC018260), mouse E47 (BC006860), and mouse Twist1 (BC083139) cDNA clones for PCR amplification were purchased from Open Biosystems. Primers used for construction of expression vectors are shown in Supplementary Table 1 Table 3. Primers for qRT-PCR. activities were measured using PicaGene Dual Sea Pansy Luminescence kit (TOYO INC MFG CO., LTD., Tokyo, Japan) and microplate luminometer (Berthold Technologies, Bad Wildbad, Germany; Centro XS 3 LB960).
Reporter construct activity was normalized by comparison with the activity of the Renilla luciferase construct. All experiments were performed in triplicate.
Electrophoretic mobility shift assay (EMSA). The sequences of the oligonucleotides containing the E-box and/or mutated E-box are shown in Table 2. Double-stranded oligonucleotide probes were labelled with biotin using the Biotin 3′End DNA Labeling kit (Thermo Scientific, Waltham, MA, USA). Nuclear proteins were isolated using a Nuclear Extract kit (Active Motif, Carlsbad, CA, USA) from HEK293T cells or Lenti-X HEK293T Cells (Takara Bio) transfected with pcDNA3 or pcDNA3.1 vectors expressing fmScx, fmTwi, fhE12, fmE12, or fmE47 proteins following the manufacturer's protocol. pcDNA3.1 vector with fhE12 was kindly provided by Dr. Eric N. Oloson 39 . DNA-protein binding was assayed with a Gelshift Chemiluminescent EMSA kit (Active Motif) following the manufacturer's protocol. Briefly, a total of 2 μL of nuclear proteins were incubated with 100 fmol of biotin-labelled DNAs for 30 to 90 min at room temperature in binding buffer containing 1 μg of poly d(I-C), 5 mM MgCl 2 , and 2.5% glycerol. For competition experiments, a 200-fold or 400-fold molar excess of unlabelled oligonucleotide was included in the binding reaction. Antibody supershifts were carried out by adding 5 μg of anti-FLAG M2 monoclonal antibody (Sigma-Aldrich) together with nuclear extracts prior to incubation with DNA. Protein-DNA complexes were resolved by electrophoresis on 5% or 4% polyacrylamide gel in Tris-borate-EDTA buffer, transferred onto Nytran SPC membrane (GE Healthcare), and detected with streptavidin-horseradish peroxidase conjugate and chemiluminescent reagent.
Statistical analysis. P-values were calculated by one-way analysis of variance using the SPSS software package (SPSS 21.0, SPSS, Inc., Chicago, IL, USA). The data were considered statistically significant at a P value < 0.05.