TGF-β Signaling Regulates Cementum Formation through Osterix Expression

TGF-β/BMPs have widely recognized roles in mammalian development, including in bone and tooth formation. To define the functional relevance of the autonomous requirement for TGF-β signaling in mouse tooth development, we analyzed osteocalcin-Cre mediated Tgfbr2 (OCCreTgfbr2fl/fl) conditional knockout mice, which lacks functional TGF-β receptor II (TβRII) in differentiating cementoblasts and cementocytes. Strikingly, OCCreTgfbr2fl/fl mutant mice exhibited a sharp reduction in cellular cementum mass with reduced matrix secretion and mineral apposition rates. To explore the molecular mechanisms underlying the roles of TGF-β signaling through TβRII in cementogenesis, we established a mouse cementoblast model with decreased TβRII expression using OCCM-30 cells. Interestingly, the expression of osterix (Osx), one of the major regulators of cellular cementum formation, was largely decreased in OCCM-30 cells lacking TβRII. Consequently, in those cells, functional ALP activity and the expression of genes associated with cementogenesis were reduced and the cells were partially rescued by Osx transduction. We also found that TGF-β signaling directly regulates Osx expression through a Smad-dependent pathway. These findings strongly suggest that TGF-β signaling plays a major role as one of the upstream regulators of Osx in cementoblast differentiation and cementum formation.

positive regulation of DKK1 in cementoblasts 10 . However, the upstream regulators of Osx in cementoblast differentiation and cementum formation remain unclear.
TGF-β s have widely recognized roles in hard tissue formation during mammalian development and exhibit versatile regulatory functions in the body 11,12 . Recent advances in molecular and genetic studies using gene targeting in mice have enable a better understanding of TGF-β signaling in bone and in the signaling networks underlying osteoblast differentiation and bone formation [13][14][15][16] . TGF-β regulates the proliferation and differentiation of osteoblasts both in vitro and in vivo; however, the effects of TGF-β on osteoblast differentiation depend on the extracellular milieu and the differentiation stage of the cells 17 . TGF-β stimulates proliferation and early osteoblast differentiation while inhibiting terminal differentiation 17,18 . However, the net effects of postnatal TGF-β signaling on cementoblast differentiation and cementum formation remain unclear.
In this study, we demonstrated the role of TGF-β signaling in cementogenesis using a conditional knockout mouse and in vitro cellular approaches. Our observations suggest that TGF-β signaling plays a major role in establishing the correct formation of cellular cementum through regulation of Osx expression.

Targeted inactivation of TβRII in cementoblasts results in a sharp reduction in cellular cementum.
To study the role of TGF-β signaling in in vivo differentiation of cementoblasts, we generated mice conditionally inactive for Tgfbr2, the gene encoding Tβ RII. Tgfbr2 fl/fl mice, having loxP sites at introns 1 and 2 of Tgfbr2 19 , were crossed with mice expressing Cre recombinase under the control of the mouse osteocalcin (OC) promoter 20 . We previously reported LacZ expression in the teeth of ROSA26 OC-Cre double-transgenic mice. We confirmed that the Cre recombinase produced from the OC-Cre promoter was active in differentiating cementoblasts and cementocytes of the developing tooth 21 . Interestingly, conditional Tgfbr2 knockout (OC Cre Tgfbr2 fl/fl ) mice displayed a sharp reduction in cellular cementum mass compared to the WT controls (Tgfbr2 fl/fl ), as observed by SEM at 6 weeks of age (Fig. 1A). Furthermore, H-E staining of tissue sections of the mandibular first molars at 16 weeks of age confirmed the tooth phenotypes following further development (Fig. 1B). Dramatic changes in the gross cellular cementum area of the mandibular first molar were observed, as shown in the measured areas of the mesial root (MR) and distal root (DR) (Fig. 1C). There was a largely reduced cementum area (approximately 2% to 8% of WT control) in both roots of the OC Cre Tgfbr2 fl/fl mice, while no remarkable difference in total root length was observed between WT control and conditional knockout mice. Reduced matrix formation and mineralization rates of cellular cementum with inactivation of TβRII in cementoblasts. We next asked whether the inactivation of Tβ RII in cementoblasts altered the levels of extracellular matrix proteins and mineralization important for the regulation of cellular cementum. As described previously, cellular cementum is formed by Osx-positive mesenchymal progenitor cells or cementoblasts in the apical root 9 . Osx was found to be expressed in cementoblasts and cementocytes at 3 weeks postnatally, and then its expression increased sharply 4 to 6 weeks postnatally; these observations correlate closely with the formation of cellular cementum 8 . Immunohistochemical staining showed that Bsp and DMP1, key markers of cementum, were highly expressed in the cementocytes of the developing cementum in Tgfbr2 fl/fl mice at 4 weeks. Inactivation of Tβ RII in cementoblasts led to dramatic reductions in Bsp and DMP1 in the cementocytes and in cementum mass in OC Cre Tgfbr2 fl/fl mice ( Fig. 2A). Moreover, it was remarkable that the expression level of DMP1 was reduced in the cementocytes with a sharp reduction in cementum mass in OC Cre Tgfbr2 fl/fl mice at 16 weeks of age (Fig. 2B). To address whether Tβ RII plays a role in controlling the mineralization rate in cementogenesis, a fluorochrome labeling assay was used. As shown in Fig. 2C, the fluorochrome labeling lines appeared thicker in the cementum area of Tgfbr2 fl/fl mice compared to OC Cre Tgfbr2 fl/fl mice. The distance between the two lines, reflecting the rate of cellular cementum formation, was much shorter in OC Cre Tgfbr2 fl/fl mice (1.9 μ m/day) than in Tgfbr2 fl/fl mice (10.0 μ m/day) (Fig. 2D).

Inactivation of TβRII results in reduced Osx expression and impaired differentiation of cementoblasts.
There was a definite decrease in Tβ RII expression in the cementum-lining cells of OC Cre Tgfbr2 fl/fl mice at 4 weeks of age (Fig. 3A). However, not all of the cementum-lining cells were Tβ RII-negative; some had intact Tβ RII, implying that OC-negative cells exist along with OC-positive cementoblast progenitor cells in the lining. We noticed that the sharp reduction in cementum mass in OC Cre Tgfbr2 fl/fl mice was very similar to that obtained with a conditional deletion of Osx using the 2.3-kb Col1a1-Cre 9 . We hypothesized that deletion of Tβ RII inhibits cementoblast differentiation partly through a decrease in Osx protein. As shown in Fig. 3B, the number of Osx-positive cells was obviously decreased in the cellular cementum of OC Cre Tgfbr2 fl/fl mice at 4 weeks of age. To investigate the molecular mechanism of altered TGF-β signaling in cementogenesis, we employed OCCM-30 cementoblast-like cells, which were derived from OC-positive root-surface cells of a developing mouse tooth and have been reported to display cementoblast-like characteristics 5 . To determine whether the clonally derived OCCM-30 cell line is a suitable model of cementogenesis, we investigated its temporal sequence of development in culture and examined the time course of extracellular matrix accumulation (Supplementary Figure 2). This cementoblast-like cell system progressed through distinct stages of cementoblast development in a manner analogous to in vivo cementum formation, as demonstrated by the temporal sequence of development characterized by distinct initial proliferation, intermediate differentiation, and terminal matrix-formation stages. The initial stage of development was characterized by cell proliferation as evidenced by high level of Cyclin D1 that decreased sharply after OM treatment. Importantly, the intermediate phase was defined as a differentiation stage evidenced by sharp increases in transcriptional factors, including Osx and Runx2, which participate in the processes of both osteogenesis and odontogenesis 6,8,22 . The final phase of OCCM-30 maturation was defined by matrix formation with progressive increases in extracellular matrix proteins including Bsp and Opn. Based on this observation, we established OCCM-30 cementoblasts lacking functional Tβ RII (shTbr2) by stably expressing shRNA for mouse Tgfbr2 (Supplementary Figure 3A). Inactivation of Tβ RII in OCCM-30 cells significantly decreased the cell proliferation rate (Supplementary Figure 3B). Morphologically, the original OCCM-30 cells remained round while the shTbr2 cementoblasts developed a larger, elongated and spindle-like shape. Furthermore, these morphological changes showed a much clearer distinction upon differentiation by OM treatment for 4 days (Supplementary Figure 3C).
Western blot analysis revealed consistently lower expression of Osx in shTbr2 OCCM-30 cells compared with the expression seen in the controls at various time points of differentiation while no difference in the expression of Runx2 was observed (Fig. 3C). Studies using mice lacking the Alpl gene have demonstrated that ALP is required for proper cementum formation 23 . We found that the enzymatic activity of ALP, as measured by ALP staining, was nearly non-existent in shTbr2 cementoblasts (Fig. 3D). The level of Alpl gene expression was also decreased in shTbr2 cementoblasts, as determined by real-time qPCR (Fig. 3E). To examine the status of cementoblast functional differentiation, we assayed the expression of other genes using real-time qPCR. The mRNA levels of Osx and other genes associated with cementogenesis were significantly lower in shTbr2 cementoblasts than in the controls, with the exception of no observed significant difference in Runx2 expression (Fig. 3E). Moreover, the expression levels of the genes for extracellular matrix proteins, including Bsp (Bsp), Ocn (Ocn), collagen type I (Col1a1 and Col1a2), and DMP1 (Dmp1), were all reduced as a result of decreased Tβ RII expression in cementoblasts.

Enforced Osx expression partially rescues the impaired differentiation of cementoblasts result-
ing from inactivation of TβRII. OC Cre Tgfbr2 fl/fl mice display a striking similarity to mice with a conditional deletion of Osx established with the 2.3-kb Col1a1-Cre with regard to the absence of a statistically significant reduction in Osx levels, suggesting that TGF-β signaling regulates Osx expression via a transcriptional mechanism. To test this hypothesis, Osx expression was enforced in control and shTbr2 OCCM-30 cells by retroviral infection. Equivalent expression of Osx was achieved in both cell types (Fig. 4A). Although ALP activity was not fully recovered by enforced Osx expression in shTbr2 OCCM-30 cells, Osx expression promoted significant ALP activity in both OCCM-30 control and shTbr2 cells (Fig. 4B). Similarly, enforced Osx expression significantly induced the expression of cementogenesis-associated genes, including Alpl, Col1a1, Col1a2, in control cementoblasts. The mRNA expression levels of Osx target genes were partially recovered by enforced Osx expression in shTbr2 OCCM-30 cells. However, Alpl and Bsp expression was not recovered, implying that they have other upstream regulators (Fig. 4C). Therefore, when highly expressed via viral transduction at equivalent levels in control and shTbr2 OCCM-30 cells, Osx expression could partially rescue the impaired differentiation of cementoblasts resulting from inactivation of Tβ RII.

TGF-β signaling regulates Osx expression through a Smad-dependent pathway in cementoblasts.
The TGF-β system transmits signals via Smad mediators to regulate a variety of biological processes.
The rescue of Tβ RII-deficient cementoblasts by enforced Osx expression suggests that Osx expression is regulated through Smad2/3 downstream of TGF-β signaling and is responsible for at least some of its effects in promoting cementoblast differentiation. To determine whether the regulation of Osx expression by Tβ RII is dependent on Smad mediators, we examined the involvement of Smad3 in Osx expression using a pharmacological inhibitor of Smad3, SIS3 24 . As shown in Fig. 5A, expression of Osx and DMP1 induced by OM treatment decreased sharply upon Smad3 inhibition by SIS3, while no difference in Runx2 was observed. To determine whether Smad3 regulates Osx transcription, Osx promoter activity was also examined using SIS3 in control and shTbr2 OCCM-30 cells. As shown in Fig. 5B, Osx promoter activity in control cementoblasts was decreased in a concentration-dependent manner by SIS3. Most significantly, treatment of control OCCM-30 cells with 5 μ M SIS3 resulted in Osx promoter activity that was as low as the basal level of promoter activity in shTbr2 OCCM-30 cells. As expected, no significant difference was seen in shTbr2 OCCM-30 cells treated with SIS3.
It is still unclear which mechanism mediates transcriptional regulation of Osx during cementoblast differentiation. To address the hypothesis that the Smad2/3 downstream effectors of TGF-β signaling regulate transcription through direct binding to the promoter region of Osx, we performed ChIP-qPCR analysis using OCCM-30 cells treated with OM. As shown in Fig. 5C, significantly increased recruitment of Smad2/3 to the regulatory regions of Osx was observed in OCCM-30 cells treated for 1 day with OM, compared with untreated cells. However, the association of Smad2/3 with the regulatory regions of Runx2 was modest and not statistically significant. Additionally, no association with the regulatory regions of Alpl was detected upon OM treatment for 1 day. As expected, the association of Smad2/3 with the regulatory regions of Osx was significantly decreased in shTbr2 cementoblasts compared with control cells, and no significant associations with the regulatory regions of Runx2 and Alpl were observed (Fig. 5D). This ChIP-qPCR data supports the concept that Tβ RII regulates Osx expression directly through the interaction with Smad2/3 in the regulatory regions of Osx, while no direct binding in the regulatory regions of Runx2 and Alpl was observed.
Scientific RepoRts | 6:26046 | DOI: 10.1038/srep26046 Discussion Molecular and genetic animal models established with spatiotemporal gene targeting technologies are useful in studies evaluating critical developmental factors, including growth factors, signaling molecules and transcriptional factors, in order to overcome disadvantages such as death in utero. In the present study, we analyzed mice that having loxP sites at introns 1 and 2 of Tgfbr2 under the control of mouse osteocalcin promoter. We found a sharp reduction in cellular cementum in developing teeth. Interestingly, these phenotypes are very similar to those observed with a conditional deletion of Osx with the 2.3-kb Col1a1-Cre in the absence of a statistically significant reduction in Osx levels, as previously reported 9 . In addition to an in vivo model, in vitro cementoblast models showed that endogenous TGF-β signaling controls cementoblast differentiation by directly regulating Osx expression and that this process is mediated through a Smad signaling pathway. The expression of Osx in cementoblasts lacking Tβ RII was largely reduced; conversely, enforced expression of Osx partially rescued cells lacking Tβ RII in vitro. The current study using immunohistochemistry assays to evaluate the protein expression levels of DMP1 and Bsp in tissues from Tβ RII conditional knockout mice, coupled with in vitro assays using cementoblasts with a shRNA system, supports a similar mechanism by which TGFβ signaling controls cementogenesis via the transcriptional activity of Osx.
Osx was originally defined as a zinc finger-containing transcription factor essential for osteoblast differentiation and bone formation during bone homeostasis 6,7 . Recent studies have demonstrated that Osx accelerates cementoblast differentiation in part through down-regulation of Wnt/β -catenin signaling by activating DKK1 10 . Recently, we also reported that Osx regulates odontoblast differentiation, maturation and root elongation 8,25 . These findings strongly indicate that Osx is a site-specific regulator in tooth root formation. The regulation of Osx in osteogenesis has been documented using various in vitro and in vivo models. Osx expression is activated by BMPs through Runx2-dependent and -independent mechanisms involving Dlx5 and Msx2 26,27 . These mechanistic reports uncovered both genetic and epigenetic mechanisms of regulation of Osx expression. These studies also showed that BMP-2 induction of Osx required the transcriptional activation of Dlx5 by p38 MAPK-mediated phosphorylation 27,28 . Moreover, additional pathways may act in parallel to, or independent of Osx transcription to post-transcriptionally modulate the function of Osx through phosphorylation, regulation of protein stability, and transcriptional activity [29][30][31] . However, in contrast to our understanding of bone biology with BMPs, the molecular mechanisms by which the expression of Osx can be regulated by TGF-β in cementogenesis remain unclear. In the present study, we first documented that Osx expression during cementoblast differentiation and cementum formation might be controlled directly by endogenous TGF-β signaling.
TGF-β is a potent multifunctional regulator of cell growth and differentiation. Although nearly all cells synthesize and respond to TGF-β , hard tissues including bone and cartilage are particularly rich in this growth factor. In previous studies, mammalian TGF-β isoforms have osteoinductive activity [32][33][34][35] , whereas transgenic mice overexpressing TGF-β 2 in bone exhibited an osteoporotic phenotype characterized by increased osteoblast and osteoclast activities and impaired matrix mineralization by osteoblasts 36 . It is well established that TGF-β signaling suppresses osteoblast differentiation and maturation in vivo and in vitro [13][14][15] . As reported previously with a mouse model for endogenous TGF-β signaling in osteoblasts, dominant negative inhibition of TGF-β responsiveness under the control of the rat osteocalcin promoter fragment leads to decreased bone remodeling and increased trabecular bone mass 13 . Exogenous TGF-β also exerts inhibitory effects on cementoblast differentiation, as evidence by decreased Bsp and Ocn expression in vitro as previously described 37 . However, the roles of endogenous TGF-β in cementum formation have not been fully elucidated. Despite the similar properties of osteoblasts and cementoblast in vitro, the discrepancy between the physiological roles of TGF-β in bone development and cementum formation in tooth development may be due to differences in the complexity of the in vivo physiological environment and developmental status. The effects of TGF-β on osteoblast differentiation depend on the extracellular milieu and the differentiation stage of the cells 17 . Furthermore, the anabolic signaling network for the coupling of bone resorption and formation is orchestrated by the interaction of local osteotropic factors, such as TGF-β and insulin-like growth factor, with various systemic hormones like PTH 16 . Although osteoblasts and cementoblasts have many features in common and similar regulation mechanisms, cementum seems to have unique properties during postnatal development and regulation. There is little evidence for cementum remodeling, while bones are continuously formed and resorbed through a process of bone remodeling. A system consisting of osteoblasts, osteocytes, bone-lining cells, and osteoclasts is required for proper functioning of bone 3 . However, cementum seems to function without such a complex cell system, with signals from the adjacent periodontal ligament likely influencing cementoblast function.
During tooth development, Runx2 and Osx are both highly expressed in the dental mesenchyme at early stages during which the crown develops. However, from the bell stage to the postnatal stage, only Osx was expressed during root developments, while the expression of Runx2 declined sharply 38 . When mice haploinsufficient for Runx2 were examined in order to characterize its function on cementum, the results were found to be normal, even in the presence of serious bone defects 39 . Moreover, patients with cleidocranial dysplasia, caused by mutations involving the Runx2 gene in humans, show no statistically significant difference in cementum formation compared with normal subjects 40 . These findings support the idea that Runx2 functions primarily embryonically at earlier stages in tooth development, while Osx participates at later stages, specifically in tooth root development, including the cementum formation stage. Interestingly, the expression level of Osx was sensitively altered by the modulation of TGF-β signaling by treatment with shRNA for Tgfbr2, while the gross level of Runx2 was not altered. Given the central role of TGF-β signaling in Osx expression during cementogenesis, these results indicate that the regulation of Osx expression is mechanically more critical than that of Runx2 during cementogenesis.
In summary, we demonstrated that OC Cre Tgfbr2 fl/fl mice, with conditionally disrupted Tβ RII in cementoblasts during tooth development, exhibit a sharp reduction in cellular cementum. Furthermore, the functional lack of Tβ RII in cementoblasts accompanies a decrease in the expression of Osx, a key regulator of cellular cementum formation, and consequently results in decreased ALP activity and decreased expression of genes associated with cementogenesis in vitro. These results demonstrate a direct effect of TGF-β signaling on endogenous cementum formation. Furthermore, the mechanistic analysis indicates that TGF-β signaling controls cementoblast differentiation by directly regulating Osx expression through a Smad-dependent pathway. Our findings provide important information for understanding the mechanism of cementoblastic differentiation during tooth root formation, thereby supporting the future development of therapies and periodontium regeneration strategies.

Materials and Methods
Mouse strain and tissue preparation. All procedures were performed in accordance with the National Institutes of Health Guidelines on the Use of Laboratory Animal. Experimental protocols and animal care methods were approved by the Animal Welfare Committee of Chonbuk National University. OC Cre Tgfbr2 fl/fl mice were generated and genotyped as described previously 41 . For histologic analysis, mice were sacrificed, and their mandibles were carefully dissected. The tissues were fixed in 4% paraformaldehyde and decalcified in 10% ethylenediaminetetraacetic acid solution for 2 to 4 wk at 4 °C. The decalcified tissues were dehydrated through a graded ethanol series, embedded in paraffin, and sectioned at 5-μ m thickness. Slides were stained with hematoxylin and eosin (H-E) for observation and further analysis. For statistical analysis, six independent littermates were used in each study.
Immunohistochemistry. Immunohistochemistry was performed as described previously 42  Double-fluorochrome labeling. Six-week-old animals were injected with calcein (20 mg/kg of body weight intraperitoneally; Sigma-Aldrich) at 7 and 2 days before being sacrificed. Fifty-micrometer cross-sections were cut perpendicularly, passing through the midsection of the mandibular first molar, and were viewed under a model LSM510 confocal laser scanning microscope (Carl Zeiss, Ostalbkreis, Germany). We calculated the average cementum mineral apposition rate using eight separate measurements of the distance between the two fluorescent labels in each section using the analySIS Pro imaging system (Soft Imaging System) and then divided that measurement by the number of days between injections (5 days). construct was also transfected as a control. The luciferase Osx promoter plasmid + 1269/-91 was kindly provided by Dr. Mark Nanes (Emory University, Atlanta, GA, USA). Transfection experiments were performed with Lipofectamine TM 3000 (Invitrogen) according to the manufacturer's instructions. After 24 h, transfected cells were harvested for whole cell lysate preparation or cultured with OM for further differentiation. Viral particles were generated by transfecting a 293T-based amphotropic retroviral packaging cell line, Phoenix with a plasmid expressing shRNA for mouse Tgfbr2 (OriGene Technologies, Rockville, MD, USA) using Lipofectamine TM LTX and PLUS reagent (Invitrogen). Supernatants containing viral particles were collected between 48 and 72 h after transfection, filtered through a 0.45-μ m filter, and used immediately. Subconfluent OCCM-30 cells were infected overnight with the retroviral particles expressing shRNA for mouse Tgfbr2 in the presence of Polybrene. Forty-eight hours later, transduced OCCM-30 cells were selected from growth medium containing 10 μ g/ml puromycin.
ChIP-qPCR. ChIP assays were performed using a ChIP assay kit from Millipore (Billerica, MA, USA) according to the manufacturer's protocol with minor modifications. In brief, cells were cross-linked by 1% formaldehyde for 5 min at room temperature. After a wash in phosphate buffered saline (PBS), an aliquot of the cross-linked chromatin was sonicated and incubated overnight with anti-Smad3 (1:100, Cell Signaling) or normal IgG. After incubation with protein G agarose beads for 2 h, and several washes, DNA-protein complexes were eluted. The cross-links were reversed by overnight incubation at 65 °C in the presence of 25 mM NaCl and subsequent digestion with RNase A and proteinase K, followed by purification of the ChIP-DNA. The ChIP-DNA was then amplified by qPCR; all reactions were carried out in triplicate. The sequences of the primers used for ChIP-qPCR are listed in Supplementary Table 1. The total amounts of binding by the Smad3 antibody were determined by expressing the amount of DNA obtained from each immunoprecipitation (IP) sample as a percentage of total DNA 43 .
Alkaline phosphatase (ALP) activity and staining. Quantitative alkaline phosphatase activity was determined using an assay based on the hydrolysis of p-nitrophenylphosphate (p-NPP) to p-nitrophenol (p-NP). Cell layers were washed twice with ice-cold phosphate-buffered saline (PBS) and lysed in 50 mM Tris-HCl buffer (pH 7.0) containing 1% (v/v) Triton X-100 (Sigma Aldrich) and 1 mM PMSF (Sigma Aldrich). Whole cell lysates were assayed by adding 1 mg/ml of pNPP as a substrate in 0.1 M glycine buffer (pH 10.4) containing 1 mM ZnCl 2 (Sigma Aldrich) and 1 mM MgCl 2 (Sigma Aldrich) to each tube for 15 min at 37 °C. Reactions were stopped by adding NaOH (final 0.6 N), and the absorbance was measured spectrophotometrically at 405 nm. Enzyme activity was expressed as OD405/min/mg of protein or a percentage of control. The concentration of protein in each cell lysate was measured using a DC Protein Assay ® (Bio-Rad Laboratories, Hercules, CA, USA). For ALP staining, cells were fixed in 10% formalin, incubated with 0.1% Triton X-100 for 5 min, and then stained with the Leukocyte Alkaline Phosphatase kit (Sigma Aldrich), according to the manufacturer's protocol.
Proliferation assay. Proliferation rates of OCCM-30 cells were measured using the Cell Counting Kit-8 (Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer's instructions. In brief, cells were cultured in 24-multiwell plates and treated with 10 μ l/well of the kit solution. Absorbance was measured spectrophotometrically at 450 nm.
RNA preparation and real-time qPCR. For preparation, 5.0 × 10 5 cells were seeded in a 60 mm culture dish and cultured for 3 weeks under mineralization differentiation induction conditions. Total RNA was prepared using an RNeasy Mini kit (QIAGEN, Valencia, CA, USA) according to the manufacturer's instructions, and cDNA was synthesized from 2 μ g of total RNA using Superscript II reverse transcriptase (Invitrogen). Real-time PCR was performed with SYBR Green PCR Master Mix (Applied Biosystems, Warrington, Cheshire, UK) following the manufacturer's protocols. Reaction conditions comprised 40 cycles of 15 sec of denaturation at 95 °C and 1 min of amplification at 60 °C. All reactions were run in triplicate, and expression was normalized to that of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (Gapdh). Relative levels of transcript expression were quantified using the Δ Δ Ct method. The calculation was performed using the Ct value of Gapdh to normalize the Ct value of the target gene in each sample and obtain the Δ Ct value, which then was used to compare different samples. Relative mRNA expression was compared in a histogram. Specific primers sets used in the analysis are listed in Supplementary Table 2.
Scanning electron microscopy (SEM). The samples were fixed in 4% paraformaldehyde for 2 hours at room temperature and incubated in 8% formaldehyde for 2 days at 4 °C. The samples were dehydrated in an ascending ethanol series (30%, 50%, 70%, 80%, 90%, one time each and twice in 100%). After critical point drying, according to a standard procedure using liquid carbon dioxide, the samples were sputter-coated with platinum and examined with a scanning electron microscope (JSM-6400; JEOL, Tokyo, Japan) under 20-kV conditions. Luciferase activity. Luciferase activity was determined using the Dual-Luciferase reporter assay system (Promega), according to the manufacturer's instructions. Light intensity was measured with a luminometer, and the luciferase activity was divided by that of the control reporter to normalize for transfection efficiency.
Statistical analysis. Data are presented as mean ± standard error (SE) of three or more separate experiments, as indicated. Normal data with equal variance were analyzed using Student's t-test (for single comparison) or one-way analysis of variance (for multigroup comparisons) with Tukey's procedure. Significance was assigned for p ≤ 0.05 as indicated.