Scleraxis is required for maturation of tissue domains for proper integration of the musculoskeletal system

Scleraxis (Scx) is a basic helix-loop-helix transcription factor that is expressed persistently in tendons/ligaments, but transiently in entheseal cartilage. In this study, we generated a novel ScxCre knock-in (KI) allele, by in-frame replacement of most of Scx exon 1 with Cre recombinase (Cre), to drive Cre expression using Scx promoter and to inactivate the endogenous Scx. Reflecting the intensity and duration of endogenous expression, Cre-mediated excision occurs in tendinous and ligamentous tissues persistently expressing Scx. Expression of tenomodulin, a marker of mature tenocytes and ligamentocytes, was almost absent in tendons and ligaments of ScxCre/Cre KI mice lacking Scx to indicate defective maturation. In homozygotes, the transiently Scx-expressing entheseal regions such as the rib cage, patella cartilage, and calcaneus were small and defective and cartilaginous tuberosity was missing. Decreased Sox9 expression and phosphorylation of Smad1/5 and Smad3 were also observed in the developing entheseal cartilage, patella, and deltoid tuberosity of ScxCre/Cre KI mice. These results highlighted the functional importance of both transient and persistent expression domains of Scx for proper integration of the musculoskeletal components.


Results
Establishment of Scx Cre KI mice. On mouse chromosome 15, Scx gene, consisting of two exons, is located within the third intron of block of proliferation 1 (Bop1), which is transcribed in the opposite orientation 9,22 . The targeting strategy placed Cre expression under the control of Scx promoter and led to the inactivation of Scx allele by in-frame replacement of most of its exon 1, which encodes for most of the coding region, including a b-HLH region. The linearised Scx CreNeo targeting vector (Fig. 1a) was electroporated into KY1.1 embryonic stem (ES) cell line (C57BL/6J × 129S6/SvEvTac) 23 . ES cell clone with the correct targeting event was microinjected into blastocysts to obtain chimeric mice. We generated heterozygous mice by crossing these chimeras with C57BL/6 wild type mice and confirmed the successful germ line transmission of Scx CreNeo targeted allele. Heterozygous Scx CreNeo/+ KI mice were viable, whereas homozygous Scx CreNeo/CreNeo KI mice were embryonic lethal. This is consistent with the previous report that conventional Scx knockout mice die around embryonic day (E)10.5 24 . To obtain Scx Cre/+ KI mice, we deleted the flippase (FLP) recombinase target (FRT)-flanked neomycin-resistance (Neo) cassette by mating Scx CreNeo/+ KI mice with mice expressing FLP recombinase (FLPe) 25 . The correctly targeted event of Scx Cre/+ KI mice was confirmed by PCR of tail tip DNA and Southern blot analysis (Fig. 1b,c). Expression of Scx was not detected in a Scx Cre/Cre KI mouse embryo at E12.5 (Fig. 1d). Thus, we successfully disrupted endogenous Scx by inserting a Cre in-frame into the ATG start site of mouse Scx gene.

Cre-mediated recombination in Scx Cre KI; Rosa-tdTomato.
Cre-mediated recombination was examined by crossing Scx Cre/+ KI mice with Rosa-tdTomato reporter mice 26 . Although endogenous expression of Scx is detectable around E9.5, Cre-mediated tdTomato expression in Scx Cre/+ ; Rosa-tdTomato was first detected around E16.5 (data not shown). In the hind limbs of Scx Cre/+ ; Rosa-tdTomato neonates, we detected tdTomato expression in the Achilles tendon and in tendons and ligaments associated with knee joint and heel (Fig. 2a-c). Tail tendons of Scx Cre/+ ; Rosa-tdTomato neonates were also positive for tdTomato (Fig. 2d). Moreover, at postnatal day (P)14, more intense expression of tdTomato was observed in body tendons and ligaments (Fig. 2e,f,g). In the trunk of 2-week-old Scx Cre/+ ; Rosa-tdTomato mice, intense tdTomato expression was observed in tendons of longissimus muscle (Fig. 2e), central tendon of diaphragm (Fig. 2f), and costal tendons (Fig. 2g). In the hindlimb of Scx Cre/+ ; Rosa-tdTomato neonates, tdTomato-positive cells were found in patella ligament, anterior and posterior cruciate ligaments, quadriceps femoris tendon, patella cartilage, femur cartilage, and meniscus ( Fig. 2h-j). In the intervertebral region of Scx Cre/+ ; Rosa-tdTomato neonates, tdTomato expression was detected in the outer annulus fibrosus on the ventral side (Fig. 2i). In the hindlimb of Scx Cre/Cre ; Rosa-tdTomato neonates, the number of tdTomato-positive cells in patella cartilage increased (Fig. 2k). Mosaic Cre-mediated recombination could have occurred because of heterogeneity in the endogenous Scx mRNA levels and time window of Scx expression in the Scx-expressing cells.
Tendon deficiency and skeletal abnormalities in Scx Cre/Cre KI neonates. Heterozygous Scx Cre/+ KI mice were viable, fertile, and displayed no apparent developmental defects (Fig. 3a,c,e,g). Consistent with previous findings 7 , morphological defects in force-transmitting and intermuscular tendons were evident at birth, in Scx Cre/Cre KI neonates (Fig. 3b,d,f,h). The diaphragm of Scx Cre/Cre KI neonates is functional and permits normal breathing. The forelimb autopod of Scx Cre/Cre KI neonates was locked in a dorsal flexure (Fig. 4b), compared to that of Scx Cre/+ KI neonates (Fig. 4a). The deltoid tuberosity (DT) and tibial tuberosity observed in Scx Cre/+ KI neonates (Fig. 4a,k,m) were missing in Scx Cre/Cre KI neonates (Fig. 4b,l,n). The rib cage, transverse process of lumbar, patella cartilage, and entheseal cartilage of calcaneus of Scx Cre/Cre KI neonates were smaller than that of Scx Cre/+ KI neonates ( Fig. 4c-j).

Defective maturation of tendons, ligaments, and annulus fibrosus of intervertebral discs in
Scx Cre/Cre KI neonates. Tnmd is related to a cartilage-derived angiogenesis inhibitor gene product, chondromodulin (Chmd) 27 , and is a marker of mature tenocytes and ligamentocytes 10,11,28 . Tnmd expression is positively regulated by Scx 6 . Scx was heterogeneously expressed in the developing leg tendons (Fig. 5a), whereas uniform expression of Tnmd was observed (Fig. 5b). Higher levels of Scx expression in the tendons were detected near the myotendinous junction (Fig. 5a). We confirmed inactivation of endogenous Scx expression in Scx Cre/Cre at E15.5 by in situ hybridisation (Fig. 5c,d). As previously reported 7 , in Scx null mice, gene expression of Tnmd and type XIV collagen was not detectable in forelimb tendons. However, during postnatal growth, Tnmd expression in other dense connective tissues, such as ligaments and annulus fibrosus of intervertebral discs, has not been evaluated yet. Thus, we investigated localisation of Col1 and Tnmd in tendons and ligaments of Scx Cre/+ and Scx Cre/Cre KI mice, by in situ hybridisation (Fig. 5e-h) and double immunostaining (Figs 6 and 7). In the hindlimb of a Scx Cre/+ neonate, tendons and ligaments were positive for Col1 and Tnmd (Fig. 6a,c,e,g). In the knee joint of a homozygous Scx Cre/Cre KI neonate, patella ligament, anterior and posterior cruciate ligaments, and anchoring tendons were positive for Col1, but Tnmd expression was almost absent (Fig. 6b,d). Similarly, in the heel of a homozygous Scx Cre/Cre KI neonate, tendons and ligaments, including the Achilles and extensor digitorum longus tendons, were positive for Col1 but negative for Tnmd (Fig. 6f,h). The meniscus was positive for Col1 but negative for Tnmd at E16.5, despite finding tdTomato positive cells in the meniscus of Scx Cre/+ ; Rosa-tdTomato mice (Figs 2j and  6i,j). The outer annulus fibrosus of intervertebral discs in Scx Cre/+ KI neonates and the tendinous region of the  Arrows in (f) indicate central tendon of diaphragm. Arrowheads in (g) indicate ribs. Dotted line in (f) encloses the outer boundary of diaphragm. (h-k) Sagittal sections of knee joint (h,j,k) and vertebral column (i) prepared from Scx Cre/+ ; Rosa-tdTomato (h-j) and Scx Cre/Cre ; Rosa-tdTomato (k) neonates. An inset in (i) shows magnified image of ventral annulus fibrosus. Arrows in (h,j,k) indicate tdTomato-positive chondrocytes in patella, meniscus, and femur, respectively. acl, anterior cruciate ligament; At, Achilles tendon; cl, cruciate ligament; fe, femur; lu, lung; pa, patella; pcl, posterior cruciate ligament; me, meniscus; np, nucleus pulposus; nt, neural tube; pl, patella ligament; qft, quadriceps femoris tendon; ti, tibia; vb, vertebral body. Scale bars, 200 μm (h-k); 100 μm (inset in (i)). diaphragm in wild type mice were also positive for Col1 and Tnmd (Fig. 7a,c,e), but Tnmd expression was almost absent in that of Scx Cre/Cre KI mice (Fig. 7b,d,f). These results suggest that loss of Scx causes defective maturation of tendons, ligaments, and outer annulus fibrosus of intervertebral discs.
Defective maturation of knee joint of Scx Cre/Cre KI mice. Tnmd expression, observed in cruciate ligaments of Scx Cre/+ ; Rosa-tdTomato mice, was missing in Scx Cre/Cre ; Rosa-tdTomato mice (Fig. 8a,b). In knee joints of 2-week-old Scx Cre/+ KI mice, tendons and ligaments developed well (Fig. 8c), whereas defective maturation was observed in knee joints of homozygous Scx Cre/Cre KI mice (Fig. 8d). The size of the patella was also significantly small and entheseal cartilage was defective in a Scx Cre/Cre KI mouse (Fig. 8d). To analyse endochondral ossification in the patella, immunostaining using antibodies against type II collagen (Col2), osteocalcin (Ocn), and sclerostin was performed. In a Scx +/+ mouse, a central part of the patella was vascularised to be replaced by bone stained with Ocn, but Col2-positive staining indicates the presence of cartilaginous matrix (Fig. 8e,f,h,i). At this stage, the patella of a Scx Cre/Cre KI mouse was Col2-positive without any indication of vascularisation and ossification (Fig. 8g,j), suggesting a delay in the endochondral ossification of the patella. In the entheseal region of patella tendon and ligament of a Scx +/+ mouse, sclerostin-positive cells were observed in Col2-and Ocn-positive calcified fibrocartilage (Fig. 8e,f,h,i). Col2-positive unmineralised fibrocartilage was also present at an adjacent region in (a-n) Skeletons of Scx Cre/+ (a,c,e,g,i,k,m) and Scx Cre/Cre (b,d,f,h,j,l,n) KI mice at P0. Lateral (a,b) and dorsal (c,d) views of ribs and lateral views of lumbar (e,f), knee joint (g,h), heel (i,j), humerus (k,l), and lower leg (m,n) are shown. Arrows in (e,f) indicate transverse processes of lumbar. Arrows in (g,h) indicate patella. Arrows in (k,m) indicate deltoid tuberosity and tibial tuberosity, respectively. fe, femur; fi, fibula; hu, humerus; pa, patella; ra, radius; ti, tibia; ul, ulna. patella tendon and ligament of a Scx +/+ mouse (Fig. 8f). Such a structure was not observed in the Scx Cre/Cre KI mouse. Taken together, Scx is also required for maturation of tendons, ligaments, and the patella.
Decreased Sox9 expression and phosphorylation of Smad1/5 and Smad3 in the developing entheseal cartilage, patella, and deltoid tuberosity of Scx Cre/Cre KI mice. Entheseal cartilage around the tendon/ligament attachment sites and sesamoid bones such as patella are derived from Scx/Sox9 double positive progenitors 2,4 . Blitz et al. reported that activation of TGF-β and BMP signaling is required for specification and differentiation of these progenitors, respectively 4,29 . Thus, we examined how loss of Scx affects expression of Sox9 and activation of these signaling pathways in the developing forelimb and hindlimb of heterozygotes and homozygotes at E13.5. Activation of TGF-β and BMP signaling pathways was monitored by phosphorylation of Smad1/5 and Smad3 that are intracellular downstream mediators 30 . As shown in Fig. 9, mature chondrocytes and progenitor cells of the patella and the deltoid tuberosity were visualized as Chmd positive (green) and Sox9 positive regions (red), respectively. In a Scx Cre/Cre KI mouse, Sox9 expression was markedly decreased in the developing patella and entheseal cartilage (Fig. 9b,h), compared with that in a Scx Cre/+ KI mouse (Fig. 9a,g). Sox9 positive progenitors in the developing deltoid tuberosity of a Scx Cre/+ KI mouse (Fig. 9g) were ,e,f,i) or Col1 (green) (c,d,g,h,j) was performed in Scx Cre/+ (a,c,e,g) and Scx Cre/Cre (b,d,f,h) KI neonates and wild type (wt) embryos at E16.5 (i,j). Frozen sagittal sections of knee (a-d,i,j) and heel (e-h) are shown. Tnmd (a,e) and Col1 (c,g) were detected in tendons and ligaments of Scx Cre/+ , whereas Col1-positive tendons and ligaments of Scx Cre/Cre (d,h) were negative for Tnmd (b,f). Arrows and arrowheads in (a-d) indicate tendons and ligaments respectively, whereas arrows in (i,j) indicate menisci. acl, anterior cruciate ligament; At, Achilles tendon; ca, calcaneus; EDL, extensor digitorum longus tendon; fe, femur; fi, fibula; me, meniscus; pa, patella; pcl, posterior cruciate ligament; pl, patella ligament; qft, quadriceps femoris tendon; ti, tibia. Scale bars, 200 μm. absent in a Scx Cre/Cre KI mouse (Fig. 9h). Similarly, phosphorylation of Smad1/5 and Smad3 observed in a Scx Cre/+ KI mouse (Fig. 9c,e,i,k) was markedly decreased especially in the prospective patella and deltoid tuberosity of a Scx Cre/Cre KI mouse (Fig. 9d,f,j,l).
Gene expression of Col14a1, Decorin, Mohawk, Tnmd and Egr1 in Scx Cre/Cre KI mice. In addition, we investigated the expression of other tendon-associated genes such as Col14a1, Decorin (Dcn), Mohawk (Mkx), Tnmd, and Early growth response 1 (Egr1) in Scx Cre/Cre KI embryos at E15.5. Col14a1 is a member of the fibril-associated collagens with interrupted triple helices (FACIT) collagen family 31 . Dcn is a small leucine-rich proteoglycan involved in regulating collagen fibrillogenesis 32 and Mkx is a family member of atypical homeobox genes that are expressed in developing tendons 33,34 . Egr1 is reported to be a transcription factor regulating the expression of Col1a1 in tendon development 35 . In a Scx Cre/Cre KI mouse, Col14a1 expression was almost absent in the developing triceps brachii tendons (Fig. 10a,b). In a wild type mouse, we found expression of Dcn in the developing tendon, dermis, and soft connective tissues (Fig. 10c), which are consistent with previous findings 32 . In Scx Cre/Cre KI embryos, we found that Dcn expression was almost absent in triceps brachii tendons (Fig. 10d) and also found generally weak expression of Mkx (Fig. 10e,f). Around the pelvic bone and lumbar at E15.5, Tnmd expressing tendons were also positive for Egr1 (Fig. 10g,i). In Scx Cre/Cre KI embryos, expression of Tnmd and Egr1 was lost and decreased in theses tendons, respectively (Fig. 10h,j). These findings suggest that loss of Scx affects the expression of genes related to tendon maturation.

Discussion
We successfully generated Scx Cre KI mice that can be used as genetic tools to reveal in vivo function of Scx-expressing tissue domains integrating the musculoskeletal components. In Scx Cre KI mice, Cre expression was driven by Scx promoter and endogenous Scx was inactivated. Cre-mediated recombination efficiency of Scx Cre KI mice reflected the intensity and duration of endogenous Scx expression. Cre-mediated excision was mainly observed in domains with persistent Scx expression, such as tendons, ligaments, and annulus fibrosus of intervertebral discs. In homozygous Scx Cre/Cre KI mice, Tnmd, a marker of mature tenocytes and ligamentocytes, was almost absent. In addition, other tendon markers such as Col14a1, Dcn, Mkx, and Egr1 were downregulated in homozygotes. Moreover, regardless of the short duration of endogenous Scx expression, defective maturation was observed in Scx-expressing domains that are involved in integration of the musculoskeletal components.  Scx is localised in the intron 3 region of Bop1 gene on chromosome 15. Bop1 and Scx are a pair of bidirectional overlapping coding genes related to cellular proliferation and differentiation, respectively 7,22 . The first reported Scx knockout failed to form mesoderm and died during early stages of embryogenesis 24 , probably because the neomycin phosphotransferase gene linked to phosphoglycerate kinase promoter (PGK-Neo) cassette affected the expression of Bop1. Later, Murchison et al. reported generation and characterisation of a conditional allele in the Scx locus with FRT-Neo cassette, by flanking the first exon, which includes most of the coding region, with loxP site 7 . As was the case with homozygotes of Scx knockout with a Neo allele, we too failed to obtain viable Scx CreNeo homozygous pups, but we successfully established a new line of Scx Cre KI mice by mating a Scx CreNeo heterozygote with a flippase-expressing FLPe mouse. Homozygous Scx Cre/Cre KI mice with tendon defects were viable and similar to Scx knockout mice generated by Murchison et al. 7 , suggesting that the replacement of most of the exon 1 by Cre does not cause embryonic lethality, unlike as observed on retention of PGK-Neo cassette in this genomic region. Cre-mediated recombination in a Scx Cre/+ ; Rosa-tdTomato mouse was less efficient than that in a ScxCre-L or a ScxCre-H transgenic mouse mated with a Rosa-tdTomato reporter mouse. Under the control of the endogenous Scx promoter/enhancer in KI mice, Cre expression gradually increased to reach levels sufficient for recombination, thus enabling Cre-mediated tdTomato expression not in the Scx + progenitor cell population but in the differentiated cell populations of tendons, ligaments, the annulus fibrosus of the intervertebral discs, and patella cartilage. More tdTomato positive cells were observed in Scx Cre/Cre ; Rosa-tdTomato mice than Scx Cre/+ ; Rosa-tdTomato, suggesting that the number of cells expressing Cre above the threshold increases in homozygotes due to a gene dosage effect. In the Cre-loxP system, recombination occurs only when Cre protein is accumulated above the recombination threshold. Mosaic and later Cre-mediated recombination is considered to be observed due to heterogeneity in the endogenous Scx mRNA levels and a narrow time window of Scx expression.
We previously reported that Scx positively regulates the expression of Tnmd in tenocytes both in vivo and in vitro 6 . Tnmd expression in Scx-positive cells in the periodontal ligament is also upregulated by lentiviral overexpression of Scx and downregulated by knockdown of endogenous Scx 36 . In this study, immunostaining of Scx Cre/Cre homozygotes with anti-Tnmd antibody revealed that Scx was necessary for the expression of Tnmd in most parts of the developing tendinous and ligamentous tissues. As previously reported, Tnmd acts as a positive regulator of postnatal tendon growth, maturation of collagen fibres, and cellular adhesion in the periodontal ligament 17 . Loss of Tnmd resulted in abated tenocyte proliferation leading to reduced tenocyte density, greater variation in collagen fibril diameters, and increase in maximal fibril diameters 16 . Thus, ablation of Tnmd expression in tendons, ligaments, and annulus fibrosus of intervertebral discs of homozygous KI mice suggests defective phenotypes in these tissues, other than the previously reported defects in force-transmitting and intermuscular tendons.
In the spinal column, an intervertebral disc lies between adjacent vertebrae and acts as a shock absorber. Intervertebral discs consist of inner and outer annulus fibrosus of sclerotome origin and notochord-derived nucleus pulposus. The inner annulus fibrosus has cartilaginous matrix associated with Col2 fibres, whereas the outer annulus fibrosus has thick multiple layers of dense connective tissue containing predominantly Col1. We previously reported that Scx + /Sox9 + progenitor population gives rise to both inner and outer annulus fibrosus and that Sox9 in this population is indispensable for the formation of inner annulus fibrosus 2 . In this study, we found that Tnmd expression in the outer annulus fibrosus was missing in Scx Cre/Cre KI neonates. Thus, we demonstrated, for the first time, that Scx is necessary for the maturation of outer annulus fibrosus.
Scx is also transiently expressed in Sox9-expressing entheseal and sesamoid cartilage during early stages of chondrogenic differentiation, as previously reported 1,3,37 . This is consistent with our previous findings from lineage analysis using ScxCre transgenic mice that entheseal and patella chondroprogenitors were positive for Scx 3 . In the Scx-expressing domain, mice lacking Sox9 lose the cartilaginous domain that contributes to the establishment of tendon/ligament attachment sites 2,4 . In this study, we found that expression of Sox9 in Scx + /Sox9 + chondroprogenitors is markedly decreased. Even though Scx expression in Scx + /Sox9 + chondroprogenitors is only transient, loss of Scx caused defective formation of cartilaginous elements arising from the Scx + /Sox9 + progenitor population. Moreover, activation of BMP and TGF-β signalling pathways in these cartilaginous elements was diminished as evidenced by decreased phosphorylation of Smad1/5 and Smad3. These findings indicate that transient expression of Scx in chondroprogenitors is functionally important in the process of entheseal cartilage formation.
Muscle contractions in utero generate mechanical forces that are essential for normal embryonic development, through modulation of cell signalling and gene expression. As reported previously, Scx is the mechanical stress responsive gene that is upregulated in the periodontal ligament in response to tensile force and maintains its fibrogenic state by inhibiting mineralisation 36 . Conversely, removal of tensile force in the tendon results in a decrease in Scx expression 38 . The severe defective phenotypes, initially in force-transmitting and intermuscular tendons and later in ligaments of Scx Cre/Cre KI mice lacking Scx, raised the possibility that Scx expression in response to mechanical force during embryonic and postnatal growth might be required for the proper development and structural maintenance of these tissues. Further investigation of target genes regulated by tensile force-responsive Scx may reveal how Scx participates in the regulation of formation and maintenance of tendons and ligaments during development and growth, in the presence or absence of mechanical stress. Such studies are now underway.

Materials and Methods
Animals and embryos. Mice were purchased from Japan SLC (Shizuoka, Japan) or from Shimizu Laboratory Supplies (Kyoto, Japan). The generation and establishment of ScxGFP transgenic strains have been reported previously 3 . For analysis of Cre activity, Rosa-CAG-LSL-tdTomato (Rosa-tdTomato) mice obtained from Jackson laboratory were used 26 . Cre-mediated recombination was monitored by red fluorescence of tdTomato expression driven by CAG promoter. All animal experimental protocols were approved by the Animal Care Committee of the Institute for Frontier Life and Medical Sciences, Kyoto University and the Committee of Animal Experimentation, Hiroshima University, and conformed to institutional guidelines for the study of vertebrates. All methods were carried out in accordance with relevant guidelines and regulations.
Generation of Scx Cre KI mouse strain. The Scx CreNeo targeting vector was designed to insert the coding region of Cre, along with a translational stop codon followed by FRT-flanked PGK-Neo cassette, in place of the ATG codon of Scx coding region, with simultaneous deletion of most of the Scx exon 1 (Fig. 1a). For construction of the targeting vector, a fused fragment, containing 706 bp Scx genomic region and Cre (706ScxCre) 3 , was cut with BglII from pBluescript SK (+) (Stratagene). 706ScxCre was inserted into the BamHI site of pPE-7neoW-F2LF with PGK-Neo cassette flanked by two FRT sites, so that the PGK promoter was oriented in the opposite direction to the Scx promoter. The 486 bp Scx genomic fragment downstream of BglII site in the first exon and first intron was amplified by PCR using a forward primer with HindIII and BamHI sites (Scx-547F: 5′-AAGCTTGGATCCAGATCTGCACCTTCTGCCTCAG-3′) and a reverse primer with HindIII and NotI sites Scientific RepoRts | 7:45010 | DOI: 10.1038/srep45010 (Scx-intR2: 5′-AAGCTTGCGGCCGCGGAGGGGTAGTGGCAC-3′). The BamHI site within Scx-547F was introduced for genotyping using Southern blot analysis. The HindIII-digested amplified fragment was inserted into the HindIII site of pPE7neoW-F2LF. PGK-Neo cassette with homology arms was then used for modification of bacterial artificial chromosome clone (RP23-415D19, CHORI) by using Red/ET recombination system (Gene Bridges) to generate ScxBAC-CreNeo, in which Cre was introduced in-frame with ATG of Scx exon 1 and replaced with a 548 bp region. The approximately 0.5 kb homology arms, homologous to extreme 5′ and 3′ ends of the genomic region to be retrieved from ScxBAC-CreNeo, were amplified by PCR. The 5′ homology arm was amplified with forward primer (F1: 5′-AAGCTTCGTTTGCCATCCAGGTCATAACCC-3′) and reverse primer (R1: 5′-ATCGATCCTTGTAAACTACCCACTTGTACCTAAG-3′). The 3′ homology arm was amplified with forward primer (F2: 5′-ATCGATGACAGGCCAGGCCGTGTTCCTGTG-3′) and reverse primer (R2: 5′-CTCGAGCCACATGGCCACTTCACCTCCTTCCTAC-3′). The resultant amplified 5′ and 3′ homology arms were subcloned into HindIII-ClaI and ClaI-XhoI sites of pMCS-DTA, respectively. For generation of the Scx CreNeo targeting vector, the modified genomic region with 1.7 kb short and 6.3 kb long arms was retrieved from ScxBAC-CreNeo and transferred to pMCS-DTA. The PacI-linearised Scx CreNeo targeting vector was electroporated into KY1.1 ES cell line (F1 hybrid of C57BL/6J and 129S6/ SvEvTac) 23 . Total 151 ES cell clones were obtained by positive and negative selection and then screened by PCR using forward primer (F3: 5′-GGATCCTCCTGGGCCCACAGGTCATAG-3′) and reverse primer (Cre-R1: 5′-CTTGCGAACCTCATCACTCGTTGCATCG-3′) (Fig. 1a). We obtained one correctly targeted ES clone that was microinjected into the blastocysts of BDF1 hybrid to generate chimeric mice.
Genotyping. Mice were genotyped using tail tip genomic DNA by PCR or Southern blot analysis.
Northern blot analysis. Total RNA was extracted from wild type and Scx Cre/+ KI embryos at E12.5. Fifteen micrograms of the total RNA was denatured with 6% formaldehyde, fractionated using 1% agarose gel electrophoresis, and transferred onto Nytran membranes with Turboblotter system. A specific cDNA probe for Scx was labelled with [α-32 P]-dCTP. Hybridisations were performed as previously described 6,21 . Histology. Scx Cre/+ ; Rosa-tdTomato and Scx Cre/Cre ; Rosa-tdTomato neonates were fixed in 4% paraformaldehyde dissolved in phosphate-buffered saline (PFA/PBS) at 4 °C for 3 h, immersed in 20% sucrose/PBS, frozen, and cryosectioned to a thickness of 8 μm. After washing with PBS, nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma). Cre-mediated tdTomato expression was detected with a fluorescent microscope. For histological evaluation of knee joint during postnatal growth, paraffin sections prepared from 2-week-old Scx Cre/+ and Scx Cre/Cre KI mice were rehydrated and stained with haematoxylin and eosin. The images were captured under the Leica DMRXA microscope equipped with the Leica DC500 camera (Leica Microsystems).
Mouse Col14a1 cDNA was also amplified using primers described previously 7 . For in situ hybridisation on frozen sections, mouse embryos were fixed with 4% PFA/PBS for 3 h, and treated in 20% sucrose before embedding in Tissue-Tek OCT compound (Sakura Finetek). Seven-micron-thick frozen sections of embedded embryos were prepared and postfixed with 4% PFA/PBS for 10 minutes at room temperature and carbethoxylated twice in 0.1% DEPC/PBS. Sections were treated in 5 × SSC, and hybridisation was performed at 58 °C with DIG-labelled antisense RNA probes. To detect DIG-labelled RNA probes, immunological detection was performed using an anti-DIG antibody conjugated with alkaline phosphatase (Anti-DIG-AP Fab fragment, Roche) and BM purple (Roche). Images were captured using an IX70 microscope equipped with a DP80 camera (OLYMPUS).

Skeletal preparation.
Mouse neonates were dehydrated with ethanol. After removing skin and soft tissues, the neonates were stained with 0.015% Alcian blue 8GX (Sigma) and cleared with 2% potassium hydroxide. The neonates were then stained with 0.05% Alizarin Red S (Wako) in 1% KOH and cleared with 1% KOH.