Scleraxis and osterix antagonistically regulate tensile force-responsive remodeling of the periodontal ligament and alveolar bone

The periodontal ligament (PDL) is a multifunctional fibrous tissue that physically connects the cementum covering the tooth root to the cortical surface of the alveolar bone (Beertsen et al., 1997). Despite its osteogenic potential, as evidenced by its high level of alkaline phosphatase (ALP) activity (Yamashita et al., 1987), the PDL between the cementum and the alveolar bone is fibrous and maintains its width unmineralized under both physiological and orthodontic conditions (Beertsen et al., 1997). The PDL senses multidirectional mechanical forces, such as mastication, speech and orthodontic tooth movement (Mabuchi et al., 2002; Pavlin and Gluhak-Heinrich, 2001). Under physiological conditions, the position of teeth in their sockets is maintained by establishing a dynamic equilibrium between bone resorption and apposition at the PDL-bone interface exposed to a variety of mechanical stimuli (Pavlin and Gluhak-Heinrich, 2001; Takano-Yamamoto et al., 1994). On application of orthodontic force, proliferation of osteogenic cells and mineralization of the extracellular matrix (ECM) occur on the tension side, whereas the compressed region within the PDL shows increased osteoclastic activity (Beertsen et al., 1997; Terai et al., 1999). Hence, the PDL is a mechanoresponsive tissue that is essential for not only the maintenance of its space and the tooth socket but also tooth movement.

The PDL contains a variety of cell populations, consisting of fibroblasts, osteoblasts, osteoclasts, cementoblasts, endothelial cells, sensory cells and progenitor/stem cells (Beertsen et al., 1997; Seo et al., 2004), thus enabling the PDL to perform supportive, remodeling, sensory, nutritive and homeostatic functions. It appears that a certain population of PDL cells is tensile force-responsive and has the unique ability to switch cellular differentiation state into either fibroblastic or osteogenic, depending on the position of the cells in the PDL. However, it remains unclear how PDL cells regulate the balance between fibrogenesis and osteogenesis by transducing mechanical force into the biological mediators.

Osteogenic differentiation is regulated by runt-related transcription factor 2 (Runx2) and osterix (Osx; Sp7 – Mouse Genome Informatics) (Komori et al., 1997; Nakashima et al., 2002). Osx is a zinc-finger-containing transcription factor that regulates the differentiation of pre-osteoblasts into fully functional osteoblasts and cementoblasts (Cao et al., 2012; Nakashima et al., 2002). Unlike bone formation, the molecular mechanisms governing ligament formation are not fully understood. Scleraxis (Scx) is a basic helix-loop-helix transcription factor that is predominantly expressed in the tendon/ligament cell lineage (Brent et al., 2003; Cserjesi et al., 1995; Schweitzer et al., 2001; Sugimoto et al., 2013a,,b). Scx is reportedly required for the formation and maturation of force-transmitting and intermuscular tendons (Murchison et al., 2007). The expression of the type I collagen and tenomodulin (Tnmd) in tenocytes is positively regulated by Scx (Murchison et al., 2007; Shukunami et al., 2006). Cellular adhesion in the PDL is enhanced by Tnmd overexpression and decreased by a loss of Tnmd (Komiyama et al., 2013). Mechanical forces also modulate the expression of Scx in tendons in vivo and in vitro (Maeda et al., 2011; Scott et al., 2011).

In our present study, taking advantage of ScxGFP-transgenic (Tg) mice that express enhanced green fluorescent protein (GFP) under the control of the promoter/enhancer of the mouse Scx gene (Sugimoto et al., 2013b), we demonstrated that Scx and Osx are significantly upregulated on the tension side in parallel with the activation of bone morphogenetic protein (BMP) signaling. Scx is strongly expressed in Scx⁺/Osx⁺ and Scx⁺/Osx⁻ cells that are localized to the unmineralized middle zone of the PDL. By contrast, Osx is highly expressed in Scx⁻/Osx⁺ PDL cells in the perialveolar zone, in which new bone formation takes place. Under osteoinductive culture conditions, lentiviral overexpression of Scx in PDL cells inhibited mineralization without affecting Osx mRNA levels. Thus, the counteracting effect of Scx on Osx-driven osteogenesis regulates tensile force-responsive PDL remodeling, in which the fine balance of Scx⁺/Osx⁻, Scx⁺/Osx⁺ and Scx⁻/Osx⁺ cell populations contributes to the maintenance of the physiological junctional attachments between teeth and bones in a position-specific manner in response to mechanical stress.

We analyzed Scx expression in dental tissue (Fig. 1A) using ScxGFP Tg mice, which express GFP in the tendon and ligament cell lineages. To visualize calcified tissues, the frozen sections were immunostained with osteocalcin (Ocn; Bglap – Mouse Genome Informatics), which is detected in tooth and periodontal tissues (Takano-Yamamoto et al., 1994). In 2-, 3- and 4-week-old ScxGFP Tg mice, alveolar bone and dentin were immunostained intensely with the anti-Ocn antibody (Fig. 1B-E). In an unerupted maxillary second molar of a 2-week-old ScxGFP Tg mouse, only PDL cells in the distal region of the root expressed Scx at low levels, as monitored by GFP expression (arrows in Fig. 1B,C). At the third postnatal week, Scx became detectable throughout the PDL, except for the periapical region, during eruption of molar teeth (Fig. 1D). In a second molar of a 4-week-old ScxGFP Tg mouse, Scx was detected in the PDL and odontoblast-predentin layer (arrowheads in Fig. 1E). These results are consistent with endogenous Scx expression detected by in situ hybridization (supplementary material Fig. S1).

For a more detailed analysis, we compared the expression of periostin (Pstn; Postn – Mouse Genome Informatics), a major ECM component of the PDL (Ma et al., 2011), with the expression of Scx. In a maxillary second molar of a 6-week-old ScxGFP Tg mouse, Pstn was localized to the gingival lamina propria (red arrowheads in Fig. 1F), PDL and Sharpey's fibers entering the alveolar bone and cementum (Fig. 1F). Scx was detected in the Pstn-positive PDL (Fig. 1F,G). Interestingly, PDL cells expressed Scx at varied levels depending on their position within the PDL. Weak Scx expression was observed in the periapical PDL (arrows in Fig. 1F,G), whereas Scx was highly expressed in the cervical PDL (asterisks in Fig. 1F,G). PDL cells expressing Scx at a higher level exhibited a more flattened and elongated morphology in the cervical region (Fig. 1J,K). These results suggest that the stress-strain levels within the PDL in response to physiological mechanical forces affect Scx expression in PDL cells.

Tenocytes and ligamentocytes express Scx and Tnmd similarly (Sugimoto et al., 2013a). We compared the marker gene expression of PDL cells with that of tenocytes isolated from limb tendons (supplementary material Fig. S2). Scx and Pstn were expressed in PDL cells and tenocytes at similar levels. Both tenascin C (Tnc) and type I collagen (Col1a1) were expressed in tenocytes, but Tnc was undetectable in PDL cells. The expression of Col1a1 was higher in PDL cells than in tenocytes. Osx was expressed in PDL cells, but its expression was undetectable in tenocytes.

Tendons and ligaments are hypovascular dense connective tissue that is made up of regular bundles of collagen fibers, whereas bone, especially in newly forming areas, is highly vascularized (Benjamin and Ralphs, 2000; Docheva et al., 2005; Maes et al., 2010; Shukunami et al., 2008). CD31⁺ vascular endothelial cells were observed scarcely in the patella ligament (Fig. 2A) and Achilles tendon (Fig. 2B). By contrast, the PDL was penetrated by blood vessels (Fig. 2C), suggesting that the PDL is distinct from ligament binding to adjacent bones in terms of its anti-angiogenic property. It has been suggested that progenitor cells expressing alpha-smooth muscle actin (αSMA) reside in the perivascular regions of the PDL and differentiate into osteoblasts, cementoblasts and fibroblasts (Roguljic et al., 2013; San Miguel et al., 2010). Consistent with previous findings, Scx⁺ cells were not localized in close proximity to the perivascular regions (empty arrowheads in Fig. 2D,E). The expression of αSMA in perivascular cells in the PDL did not overlap with the expression of Scx (Fig. 2F,G). This suggests that Scx-expressing PDL cells are comparatively mature fibroblastic cells that maintain the ligamentous tissue of erupted teeth. Osx was detectable throughout the PDL as well as the pulp (Fig. 2H). At this stage, almost all Scx⁺ PDL cells were positive for Osx (Fig. 2H-L). Osx was also expressed at a higher level in Scx⁻ perivascular cells in the periapical PDL and alveolar bone (white arrowheads in Fig. 2M,N).

To investigate the characteristics of Scx⁺ cell populations in the PDL after tooth root formation is completed, we analyzed ALP activity and Osx expression in a 12-week-old ScxGFP Tg mouse (Fig. 3). In the leg (Fig. 3A), ALP activity was detectable in cartilaginous and bony tissues, but undetectable in Scx⁺ tendons and ligaments (Fig. 3B). Osx was expressed in immature osteoblasts (arrowheads in Fig. 3C) and prehypertrophic/hypertrophic chondrocytes (data not shown), but was undetectable in Scx⁺ cells in the tendons and ligaments of the leg (Fig. 3C,D). In a maxillary second molar of a 12-week-old ScxGFP Tg mouse, ALP activity was detected in cells throughout the PDL as well as in dental pulp cells (Fig. 3E). Osx was expressed in osteoblasts, cementoblasts, odontoblasts and PDL cells (Fig. 3F). In the periapical region, some Osx⁺ cells were detected in the PDL that comprised cells with weak or no Scx expression (Fig. 3G,H). In the oblique fibers of the PDL, overlapping Scx and Osx expression was detected in a subset of fibroblastic cells (yellow arrowheads in Fig. 3I,J) that were neither perivascular nor osteoblastic/cementoblastic cells. These results indicate that Scx⁺ PDL cells in vivo retain an osteogenic phenotype with high ALP activity and Osx expression, different from fibroblasts in tendons and ligaments.

Under physiological conditions, Scx was expressed at a high level in elongated PDL cells under tensile force as a result of stretching of the PDL by trans-septal fibers between the molars (Fig. 1). This observation raises the possibility that tensile force on the PDL positively regulates Scx expression. We then examined Scx expression during experimental tooth movement by monitoring GFP expression in ScxGFP Tg mice (Fig. 4A,B). At 48 h after the experimental tooth movement, the distance between the first and second molars increased by 107.8±15.4 μm (mean±s.d., n=4). In transverse sections of maxillary first molars with experimental tooth movement, Scx upregulation was observed clearly in the PDL on the tension sides (arrowheads in Fig. 4D) compared with the corresponding regions on the contralateral control side (Fig. 4C). By the insertion of an elastic band, widening of the PDL space and elongated fibroblastic cells were observed on the tension side (Fig. 4E,F). High levels of Scx expression were induced in PDL cells by tensile force exerted by experimental tooth movement (arrowheads in Fig. 4G,H). To quantify the increase in Scx expression, we calculated the proportion of cells expressing Scx at a high level (Scx^high) to the total pool of cells expressing Scx (n=4) (supplementary material Table S1). The mean proportion of Scx^high cells was significantly increased on the experimental side (48.7%) compared with the control side (27.5%) (Fig. 4I). In RT-qPCR analysis, significant upregulation of Scx was detected in periodontal tissues on the experimental side (1.26-fold, P=0.02) compared with the control side (supplementary material Fig. S3). Upregulation of Pstn expression was also detected in the tensioned regions of the experimental side (1.57-fold, P=0.01) (supplementary material Fig. S3). Because bone formation is facilitated on the tension side of alveolar bone, it has been proposed that the PDL on the tension side is exposed to more osteogenic factors than in its unloaded state (Henneman et al., 2008). Consistent with this notion, phosphorylation of Smad1/5 for the BMP pathway was significantly increased on the tension side (Fig. 5A-H). Osx⁺ cells were also increased on the tension side (Fig. 5I-P). Upon tensile loading, the proportion of cells positive for pSmad1/5 and Osx in Scx⁺ PDL cells was increased 1.91- and 1.39-fold, respectively (Fig. 5Q). Furthermore, the number of Scx^high/Osx⁺ PDL cells (white arrowheads in Fig. 5I,M) was also increased in response to tensile force (2.49-fold, Fig. 5Q). Thus, it can be concluded that both Scx and Osx are tensile force-responsive transcription factors.

To elucidate the role of Scx in the PDL in response to osteogenic stimuli, we performed lentiviral overexpression of Scx in PDL cells (Fig. 6A). Successful overexpression of Scx in LV-Scx-infected cells on day 10 was confirmed by GFP expression and RT-PCR (Fig. 6B-D). Mineral deposition was monitored in osteo-inducing cultures on day 25 (Fig. 6E). Notably, formation of the calcified nodules in PDL cells overexpressing Scx was significantly suppressed compared with that in Lv-Vec-infected cells (Fig. 6E). The expression of Runx2, Osx, osteopontin (Opn; Spp1 – Mouse Genome Informatics) and Ocn was significantly increased in PDL cells in response to osteogenic stimuli (Fig. 6F,G). Although the expression levels of Runx2 and Osx were not affected by Scx overexpression (Fig. 6F), Opn and Ocn expression was significantly downregulated by Scx overexpression under osteo-inducing conditions (Fig. 6G). Among the non-osteogenic genes examined here, endogenous Scx, matrix Gla protein (Mgp), Pstn and Tnmd were significantly upregulated by Scx overexpression under non-inducing conditions (supplementary material Fig. S4). Under osteo-inducing conditions, Pstn and Tnmd expression was significantly decreased compared with that under non-inducing conditions (gray bars in supplementary material Fig. S4), whereas the expression level of Pstn recovered to that observed in Lv-Vec-infected cells under the non-inducing conditions by Scx overexpression (black bar of Pstn in supplementary material Fig. S4). The level of Mgp in response to osteoinduction was further increased by Scx overexpression (gray and black bars in supplementary material Fig. S4). We then knocked down Scx by RNA interference (Fig. 7A). In PDL cells transfected with siScx-1 or siScx-2, the level of Scx was decreased to less than 10% on day 6 (Fig. 7B). Although transient knockdown by siRNA did not significantly affect mineralization on day 6 (data not shown), gene silencing of Scx resulted in a marked increase of Ocn expression in PDL cells cultured under osteogenic conditions (Fig. 7C). No significant increase of Ocn expression by Scx siRNA was detected under non-inducing conditions (Fig. 7C). Taken together, these findings suggest that tensile force-responsive Scx has an inhibitory action on mineralization by regulating the expression of ECM molecules.

PDL cells have a unique differentiation potential in response to mechanical stimuli. In this study, we demonstrated that the Scx⁺/Osx⁻, Scx⁺/Osx⁺, Scx⁻/Osx⁺ PDL cell populations contribute coordinately to tensile force-induced remodeling of the PDL to maintain the junction between the cementum of the tooth and the alveolar bone. The Scx overexpression and knockdown experiments demonstrate that Scx negatively regulates the expression of osteogenic genes (Opn and Ocn) only under osteo-inducing conditions. The balance between fibrogenesis and osteogenesis in the tensile force-loaded PDL is regulated antagonistically by Scx and Osx in a position-specific manner (Fig. 8).

Scx is expressed predominantly in developing tendons and ligaments exposed to mechanical loading (Schweitzer et al., 2001; Sugimoto et al., 2013a). The gradual and temporary loss of tensile loading results in the reversible loss of Scx expression in tendons (Maeda et al., 2011), whereas Scx expression is enhanced by cyclic loading in vitro (Scott et al., 2011). Our in vivo study using ScxGFP Tg mice revealed that Scx is induced in association with PDL maturation and is strongly expressed in elongated fibroblastic cells when they were exposed to tensile stress transmitted from collagen fibers running between the tooth and alveolar bone. During experimental tooth movement, Scx and Osx expression was significantly upregulated on the tension side, in which bone formation compensates for tooth movement away from the bone under tensile force. These results suggest that Scx is a tensile force-inducible gene in the PDL under physiological and orthodontic conditions.

We reported that Scx positively regulates the expression of Tnmd, which has anti-angiogenic activity in its C-terminal cysteine-rich domain (Kimura et al., 2008; Oshima et al., 2004). In contrast to ligaments responsible for bone-to-bone connections, the PDL, localized between the alveolar bone and the cementum of the tooth, is exceptionally well vascularized, reflecting the high metabolic turnover of cellular and extracellular constituents. However, our double-immunostaining study revealed that Scx and CD31 were expressed in a mutually exclusive way within the PDL. Scx^high PDL cells are found away from the apical region of the developing tooth root, in which active angiogenesis takes place. Scx overexpression resulted in the upregulation of Tnmd in PDL cells, suggesting that Scx enhances the mature ligamentocyte phenotype. As reported previously, cellular adhesion is enhanced by Tnmd overexpression and decreased by a loss of Tnmd in the PDL (Komiyama et al., 2013). In the PDL, the Scx⁺ cell population represents a group of mature ligamentocytes and its expression level varies, depending on the extent of tensile stress.

Osx is a zinc-finger-containing transcription factor that regulates the differentiation of pre-osteoblasts into fully functional osteoblasts (Nakashima et al., 2002). Genetic evidence suggests that Osx expression during tooth root formation is also closely associated with cementum formation (Cao et al., 2012). The number of Osx⁺ cells in the PDL increase sharply in 4- to 6-week-old mice, whereas few Osx⁺ cells are detected in the PDL of 6-month-old mice (Cao et al., 2012). Consistent with these findings, a number of Osx⁺ cells were found in the developing PDL in 6-week-old mice, but Osx expression was decreased in the PDL of 12-week-old mice, the root formation of which is completed. In the experimental tooth movement model, Osx is upregulated in the PDL on the tension side undergoing bone formation. The Osx⁺ cell population represents a group of cells committed to osteoblasts or cementoblasts in the PDL, thus contributing to the active remodeling of periodontal tissues.

Ectopic mineralization within ligaments causes tissue dysfunction. Ossification of the posterior longitudinal ligament of the spine causes spinal pain and, in severe cases, spinal cord compression (Inamasu et al., 2006). Despite the mechanical loading of mastication under physiological conditions or orthodontic forces during tooth movement, the PDL maintains its constant width unmineralized throughout the lifetime of an organism (Beertsen et al., 1997). Analysis of Scx-deficient mice revealed that Scx is essential for the condensation and differentiation of the progenitor cells for force-transmitting and intermuscular tendons, but no apparent morphological abnormality has been reported in developing ligaments (Murchison et al., 2007). However, we found a novel inhibitory action of Scx on mineralization by overexpression and knockdown experiments in PDL cells maintained under osteo-inducing conditions. Scx overexpression inhibited PDL mineralization without affecting Runx2 or Osx mRNA levels, which were upregulated by osteoinduction. Conversely, the upregulated expression of Ocn by osteoinduction was increased further in PDL cells by gene silencing of Scx. Thus, in the osteogenic environment of the tensile force-loaded PDL, Scx might act as a negative regulator of alveolar bone formation to maintain the width of the PDL and to prevent ankylosis, that is, the fusion of the tooth root with the surrounding alveolar bone.

Molecules that negatively regulate mineralization are thought to play key roles in maintaining the homeostasis of the PDL trapped between the cementum and the alveolar bone. It has been reported that Mgp, asporin, msh homeobox 2 (Msx2) and twist-related protein 1 (Twist1) act as inhibitors of the mineralization of the PDL (Hashimoto et al., 2001; Kaipatur et al., 2008; Komaki et al., 2007; Murshed et al., 2004; Yamada et al., 2007; Yoshizawa et al., 2004). Scx is also a member of the twist subfamily of basic helix-loop-helix transcription factors (Atchley and Fitch, 1997). The inhibitory action of Pstn on mineralization was reported using an odontoblastic cell line (Ma et al., 2011). We found that not only Scx but also Pstn were significantly upregulated on the tensile force-loaded site of the PDL. Under non-inducing conditions, Mgp and Pstn were significantly upregulated by Scx overexpression. Pstn was significantly downregulated in PDL cells by osteoinduction, whereas its expression was recovered by Scx overexpression. These results suggest that Scx also participates in keeping the PDL unmineralized in concert with previously reported molecules.

Scx⁺/Sox9⁺ progenitors contribute to the establishment of the junction between cartilage and tendon/ligament (Blitz et al., 2013; Sugimoto et al., 2013a). Interestingly, the PDL and alveolar bone are also derived from Scx⁺/Sox9⁺ progenitors (unpublished data). In this study, we demonstrated that the Scx⁺/Osx⁻, Scx⁺/Osx⁺ and Scx⁻/Osx⁺ PDL cell populations participate coordinately in remodeling and maintaining the junction between the cementum of the tooth and the alveolar bone. As reported previously, αSMA⁺ progenitor/stem cells in the perivascular regions of the PDL can differentiate into osteoblasts, cementoblasts and fibroblasts (Roguljic et al., 2013; San Miguel et al., 2010). However, expression of αSMA in the PDL did not overlap with Scx expression, suggesting that fibroblastic Scx⁺ PDL cells respond to the mechanical stress to quickly participate in PDL remodeling. The increase in the number of the pSmad1/5⁺ PDL cells and Osx⁺ PDL cells on the tension side suggests that PDL cells become more osteogenic through the activation of BMP signaling upon tensile force loading. Osx-driven osteogenesis in the PDL is counteracted by the increased expression of Scx, which facilitates ligamentogenic fibroblast maturation and inhibits osteogenic mineralization. The Scx^high/Osx⁺ cell population appears on the tension side of the PDL that does not calcify, whereas the Scx⁻/Osx⁺ cells and osteogenic PDL cells with weak Scx expression reside close to the alveolar bone. The balance of Scx and Osx activities can be a determinant of the decision whether PDL cells follow either the fibroblastic or osteogenic differentiation pathway. Further studies are now underway to explore how the position-specific upregulation of Scx is regulated in the tensile force-responsive PDL.

C57BL/6 mice and Wistar rats were purchased from Shimizu Laboratory Supplies (Kyoto, Japan). The generation and establishment of ScxGFP transgenic strains have been reported previously (Sugimoto et al., 2013b). Scx expression was monitored by GFP expression. All animal experimental procedures used in this study were approved by the Animal Care Committee of the Institute for Frontier Medical Sciences, Kyoto University, Japan, and conformed to institutional guidelines for the study of vertebrates.

Nine-week-old Scx-GFP Tg male mice were subjected to experimental tooth movement. These mice were anesthetized by intraperitoneal injection of sodium pentobarbital (30 mg/kg) (Kyoritsu Seiyaku, Tokyo, Japan). Experimental tooth movement was achieved by the interproximal insertion of a piece of orthodontic elastic band between the upper first and second molars on the left side, according to the method described by Waldo and Rothblatt (1954). The contralateral right side was used as a control. At 48 h after insertion, a maxillary impression was taken using silicon impression material under anesthesia. The impressions were filled with dental stone, and the distance between the first and second molars was measured using a dial tension gauge (Mitutoyo, Kanagawa, Japan).

For hematoxylin and eosin (H&E) staining, sections were stained with Gill's hematoxylin (Vector Laboratories) and 0.25% eosin (Sigma). To detect ALP activity, non-decalcified frozen sections were covered with a 2% Nitro-Blue tetrazolium chloride/5-bromo-4-chloro-3′-indolyphosphatase p-toluidine salt stock solution (Roche) diluted in ALP buffer at pH 9.5 (100 mM Tris-HCl, 100 mM NaCl, 50 mM MgCl₂) and then incubated for 5 min at 37°C in the dark. PDL cells were fixed with 95% methanol for 20 min and stained with 1% Alizarin Red S (Wako) at pH 6.4 for 16 h.

Anesthetized mice were perfused with 4% paraformaldehyde in phosphate-buffered saline (PFA/PBS) containing 20% sucrose, and their upper jaws or legs were dissected. The specimens were fixed in 4% PFA/PBS containing 20% sucrose for 3 h, embedded in SCEM (Section-Lab), and frozen in n-hexane cooled with dry ice. Undecalcified frozen sections at a thickness of 4 μm were obtained according to Kawamoto's film method using tungsten carbide blades, either TC-65 (Leica Microsystems) or SL-T35 (Section-Lab), and adhesive films (Section-Lab) (Kawamoto, 2003). After washing with ethanol and PBS, the sections were fixed in 4% PFA/PBS for 5 min and decalcified with 0.25 M ethylenediaminetetraacetic acid (EDTA)/PBS for 1 h. Sections for the detection of Osx, αSMA or pSmad1/5 were boiled in 10 mM sodium acetate at pH 6 for 10 min. After blocking with 3.2% skim milk/PBS, the sections were incubated with primary antibodies for 16 h, washed and then incubated with appropriate secondary antibodies conjugated with Alexa Fluor 488 or 594 (Life Technologies). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (Sigma). The primary antibodies used were anti-GFP (rat IgG2a) (Nakarai Tesque, 04404-84; 1:1000), anti-GFP (rabbit IgG) (MBL, 598; 1:1000), anti-CD31 (BD, 553370; 1:1000), anti-osteocalcin (anti-Ocn) (Takara Bio, M173; 1:800), anti-Osx (Abcam, ab22552; 1:800), anti-periostin (anti-Pstn) (BioVendor, RD181045050; 1:800), anti-αSMA (Abcam, ab5694; 1:500) and anti-pSmad1/5 (Cell Signaling, #9516S; 1:100). The images were captured under a Leica DMRXA microscope equipped with a Leica DC500 camera (Leica Microsystems). After acquisition of the fluorescent images, HE staining was performed on the same sections.

Images of the specimens subjected to experimental tooth movement were acquired under the same conditions using a GFP filter with a 1 s exposure. To determine high expression levels using the brightness of a color given the RGB values, areas with high brightness were selected automatically in the green channel image using Adobe Photoshop CS3 (Adobe Systems). Scx^high cells were defined as cells with a brightness >2.5-fold higher than the background level (dentin area) in green channel images. Nuclei labeled with DAPI were counted.

PDL cells were isolated from 4-week-old male Wistar rats. The maxillary and mandibular first, second and third molars were extracted and washed with PBS. The PDL attached to the middle of the root surface was scraped off, placed onto 35-mm cell culture dishes (BD) and maintained in MF-start (TOYOBO). At confluence, the cells outgrown from the PDL were passaged twice and grown in minimum essential medium Eagle alpha modification (α-MEM) supplemented with 10% fetal bovine serum (FBS). Cells were maintained in a humidified atmosphere of 5% CO₂ in air.

For osteoinduction, PDL cells were maintained in α-MEM supplemented with 10% FBS, 1 μM dexamethasone, 10 mM β-glycerophosphate, 50 μg/ml ascorbic acid and 200 ng/ml of recombinant human (rh) BMP6 (differentiation medium) for 3 days, and then cultured in differentiation medium without rhBMP6 for another 21 days, as reported previously (Hakki et al., 2014; Yoshizawa et al., 2004).

Lentiviral particles were produced using HIV-based lentiviral vector constructs purchased from SBI. The full coding DNA fragment of mouse Scx was inserted into the pCDH-CMV-MCS-EF1-GreenPuro vector (SBI). The lentiviral particles were concentrated with PEG-it virus precipitation solution (SBI), and the multiplicity of infection (MOI) in NIH3T3 cells was titrated using a Global Ultra Rapid Lentiviral Titer Kit (SBI). For PDL cells, 1.0 MOI of the lentiviral particles were used.

Small interfering RNA (siRNA) oligonucleotide duplexes were purchased from GE Healthcare Life Sciences. Scx was knocked down using siScx-1 (catalog number J-113656-09) and siScx-2 (catalog number J-113656-10) included in the ON-TARGET plus rat Scx siRNA-Set of 4 (catalog number LQ-113656-00-002). For the control experiment, siGENOME non-targeting siRNA Pool number 1 (catalog number D-001206-13-05) was used. Transfection of siRNA into PDL cells was performed with DharmaFECT 1 transfection reagent (GE Healthcare Life Sciences) according to the manufacturer's instructions.

Total RNA was extracted from PDL cells using an RNeasy Plus Mini Kit (QIAGEN). Two hundred nanograms of total RNA were used to synthesize complementary DNA (cDNA) with a PrimeScript RT reagent Kit (Takara Bio). RT-PCR was performed with Takara Ex Taq (Takara Bio) and specific primers for Scx (forward, 5′-GCAGCGGCACACAGCGAAT-3′; reverse, 5′-AGGCGCAGCGTCTCAATCT-3′). RT-qPCR was performed using SYBR Premix Ex Taq II (Takara Bio) on a StepOne instrument (Life Technologies). Relative mRNA expression was normalized to 18S rRNA and calculated using the 2^−ΔΔCT method. Specific primers for RT-qPCR are listed in Table 1.

P-values were calculated by t-test or one-way analysis of variance using the SPSS software package (SPSS 21.0). Data were considered statistically significant for a P-value <0.05.

We thank Ms H. Sugiyama for her valuable secretarial help.