Role of epidermal growth factor and its receptor in mechanical stress-induced differentiation of human periodontal ligament cells in vitro

ARCHIVES OF

PERGAMON

Archives of Oral Biology 43 (1998) 987±997

ORAL BIOLOGY

Role of epidermal growth factor and its receptor in mechanical stress-induced di€erentiation of human periodontal ligament cells in vitro N. Matsuda a, *, K. Yokoyama a, S. Takeshita a, M. Watanabe a, b a

Laboratory of Cell and Stress Biology, JST at Nagasaki, 2-1303-8 Ikeda, Omura, Nagasaki 856, Japan Laboratory of Radiation and Life Sciences, School of Pharmaceutical Sciences, Nagasaki University, 1-14 Bunkyo-machi, Nagasaki 852, Japan

b

Accepted 6 June 1998

Abstract The periodontal ligament (PDL) contains precursor cells for osteoblasts and cementoblasts. It has been shown that epidermal growth factor (EGF) inhibits dexamethasone-induced di€erentiation and up-regulates EGF-receptor (EGF-R) expression, whereas EGF-R is down-regulated in the course of di€erentiation. Thus it was suggested that EGF and its receptors act as a negative regulator of osteoblastic di€erentiation in PDL cells. In order to investigate further this hypothesis, human PDL cells were now used to elucidate the role of EGF and EGF-R in their proliferation and di€erentiation under mechanical stress-loaded conditions in vitro, as the PDL regularly receives mechanical stress from occlusal forces. As a model of mechanical stress, a cyclic stretch of 9 or 18% elongation was applied to the cells with a Flexercell cell-strain unit system. Alkaline phosphatase activity and osteocalcin mRNA expression were signi®cantly induced by loading cyclic stretch for more than 4 days, whereas stretch slightly inhibited cell proliferation. Visualization of the actin stress ®bres of the cells by rhodamine phalloidin revealed that approx. 10% of the total number of cells had become aligned perpendicularly to the direction of the stretch. The e€ects of stretch on alkaline phosphatase activity and cell proliferation were totally abolished by the presence of 10 ng/ml EGF. Western blotting of EGF-R protein demonstrated that stretch-induced di€erentiation accompanied the decreased expression of EGF-R protein in the cells. However, the amount of tyrosine-phosphorylated EGF-R upon EGF stimulation was restored to the control level in stretched cells. These results suggest that the EGF/EGF-R system acts as a negative regulator of di€erentiation of PDL cells regardless of the type of di€erentiation stimuli. Also, interaction between mechanical stress and the EGF/EGF-R system may participate in the osteoblastic di€erentiation of PDL cells and thereby regulate the source of cementoblasts and osteoblasts. # 1998 Published by Elsevier Science Ltd. All rights reserved. Keywords: Periodontal ligament cells; Mechanical stress; Di€erentiation; Epidermal growth-factor receptor

1. Introduction Abbreviations: DMEM, Dulbecco's modi®ed Eagle medium, EGF(-R), epidermal growth factor (-receptor), FBS, fetal bovine serum, PBS, phosphate-bu€ered saline, RT-PCR, reverse transcription±polymerase chain reaction, SDS±PAGE, sodium dodecyl sulphate±polyacrylamide gel electrophoresis. * Corresponding author. Present address: Nagasaki University Radioisotope Center, 1-12-4 Sakamoto, Nagasaki 852-8523, Japan. Fax: +81-95-849-7153; e-mail: [email protected].

The periodontal ligament is a dense connective tissue between the root cementum and the alveolar bone that anchors the tooth and maintains the structural integrity of these mineralized tissues. Fibroblasts in the periodontal ligament, the major cell type, are regarded as multipotential, or form a heterogeneous population that can di€erentiate into either cementoblasts or osteoblasts, depending on needs and conditions. Thus,

0003-9969/98/$ - see front matter # 1998 Published by Elsevier Science Ltd. All rights reserved. PII: S 0 0 0 3 - 9 9 6 9 ( 9 8 ) 0 0 0 7 9 - X

988

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

the ligament has been regarded as the source of cementoblasts and osteoblasts (Aukhil et al., 1986; McCulloch and Melcher, 1983; Roberts et al., 1982). Indeed, cultured ®broblasts from the ligament exhibit bone cell-like properties in vitro, including a relatively large amount of alkaline phosphatase activity (Kawase et al., 1988; Nojima et al., 1990; Somerman et al., 1988), and the expression of bone-matrix proteins such as osteopontin and secreted protein, acidic and rich in cysteine (SPARC) (Ramakrishnan et al., 1995). Similar to bone-forming cultures, cells of the periodontal ligament undergo osteoblastic di€erentiation in response to dexamethasone, a synthetic glucocorticoid widely used to induce di€erentiation of osteoblasts, as characterized by elevated alkaline phosphatase activity (Matsuda et al., 1993), increased expression of osteopontin and bone sialoprotein (Ramakrishnan et al., 1995), and the formation of mineralized nodules (Cho et al., 1992; Ramakrishnan et al., 1995). The mechanisms by which cells of the periodontal ligament maintain their phenotype and di€erentiate into mineralized tissue-forming cells remain unclear. In attempts to examine the involvement of polypeptide growth factors, we have earlier investigated the role of EGF and its receptor in dexamethasone-induced in vitro di€erentiation of rat periodontal-ligament cells. We found that EGF antagonized di€erentiation and up-regulated EGF-R expression, whereas EGF-R was down-regulated in the course of di€erentiation. Therefore, the EGF/EGF-R system appears to be important as a phenotype stabilizer by functioning as a negative regulator of osteoblastic di€erentiation in these cells (Matsuda et al., 1993). Although in vivo observations on the rat (Cho et al., 1991) supported these in vitro results, a detailed study on the role of EGF/ EGF-R in cells of the human periodontal ligament cultured in conditions closer to the physiological, without dexamethasone, is still needed. Physiologically, the periodontal ligament is continuously subjected to mechanical stress caused by occlusal forces. Furthermore, remodelling of the ligament and alveolar bone occurs in response to orthodontic forces. These facts led us to speculate that responses of the ligament to mechanical stress are involved in its cell proliferation and di€erentiation. In fact, it has been shown that a variety of cells respond to mechanical stress, such as tension force (Banes et al., 1985), compression (Veldhuijzen et al., 1979) and ¯uid shear stress (Malek et al., 1993), by demonstrating signi®cant changes in their structure and function. In bone, Raab-Cullen et al. (1994) have reported, using the rat tibia 4-point bending model, that external mechanical loading induces a rapid and transient increase in mRNA expression for c-fos, a gene associated with proliferation and/or di€erentiation in bone development and fracture repair. Several in vitro studies on cultured osteoblastic cells have also demon-

strated elevated amounts of bone-related molecules, such as alkaline phosphatase, osteopontin and osteocalcin, in response to mechanical stretching (Harter et al., 1995; Mikuni-Takagaki et al., 1996; Nishioka et al., 1993). Therefore, it appears that essential functions of osteoblasts in bone remodelling are a€ected by experimentally loaded mechanical stress and that cellular responses to mechanical stress are crucial in homeostasis, adaptation to the environment, and regeneration of bone. In contrast, only a small number of studies have addressed the responsiveness of cells of the periodontal ligament to mechanical stress, in which the ability of cultured cells to proliferate in response to tension force was demonstrated (Yamaguchi et al., 1994, 1996). As we have shown that such cells maintain their phenotype and di€erentiate into osteoblastic cells through mechanisms involving the EGF/EGF-R system, it was of interest to investigate how they and the EGF/EGF-R system respond to mechanical stress. Furthermore, the interaction between mechanical stress and EGF/EGF-R has not, we believe, been reported elsewhere. Our aim now was to elucidate the role of EGF and EGF-R in the proliferation and di€erentiation of cells of the human periodontal ligament in mechanical stressloaded conditions in vitro. For this purpose, cultured ligament cells were loaded with cyclic stretch using ¯exible-bottomed culture plates (Banes et al., 1985) and their responses were monitored. Then the e€ect of EGF on these responses was examined. Finally, we evaluated changes in the amount and autophosphorylation of EGF-R in response to cyclic stretching. 2. Materials and methods 2.1. Cell culture Fibroblastic cells were obtained from explant cultures of human healthy periodontal ligament taken from a third molar that had been extracted for orthodontic reasons, as described by Matsuda et al. (1996b). The tissues were minced, put in culture dishes and incubated in DMEM (Gibco Laboratories, Grand Island, NY) supplemented with 10% FBS (Intergen Company, Purchase, NY), non-essential amino acids, 10 mM sodium pyruvate, vitamins (Gibco) and an antibiotic mixture (1 U/ml penicillin, 1 U/ml streptomycin, 1 U/ml gentamycin; Sigma Chemical Co., St. Louis, MO) at 378C in a humidi®ed atmosphere of 5% CO2±95% air. When the outgrowing cells reached con¯uency, they were trypsinized with 0.05% trypsin (1:250)±0.53 mM EDTA 4 Na (Gibco) in PBS for secondary culture. Cultures were maintained until con¯uency and passed at a 1:4 split ratio. All the experiments were done on cells of between three and seven passages.

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

2.2. Mechanical stress As an experimental model of mechanical stretch, tension force was loaded on to cells cultured on a ¯exible substratum (25 mm dia., Flex I culture plate; Flexcell International Corporation, McKeesport, PA) by applying vacuum-operated negative pressure using the Flexercell Strain Unit Model FX-2000 (Banes et al., 1985), which is capable of controlling the magnitude as well as the frequency of cell deformation. Although this system has been used in many investigations, one problem with the apparatus is that the substrate of the strain well is strained in a non-uniform manner; deformation is greatest at its periphery and least at its centre. Therefore, we aimed to examine the net response of the total number of cells to mechanical strain. Cells were subjected to 9 or 18% of maximum strain for 5 s followed by 5-s relaxation (6 cycles/min). According to the manufacturer, in these conditions the strains are distributed inhomogeneously such as ÿ5±9% or ÿ4±18%. 2.3. Cell proliferation Cells were plated on to a Flex I culture plate at concentrations of 2.5  104, 5.0  104, or 1.0  105 per well and incubated in DMEM/10% FBS for 24 h. After replacement of the medium with fresh DMEM/10% FBS, Flex I plates were placed on the Flexercell strain unit to apply cyclic stretch to the cells. At 1, 3, or 5 days of incubation under cyclic stretching, cells were harvested by trypsinization and counted in a haemocytometer. 2.4. Visualization of actin stress ®bres After 5 days of culture with or without stretch, cells were rinsed with PBS twice and ®xed by incubation with 10% neutral-bu€ered formalin for 30 min. The ¯exible substrate of 25 mm dia. was then removed from the plate and cut to 20  20 mm square. Actin stress ®bres were stained with 33 nM of rhodamine phalloidin (Molecular Probes, Inc., Eugene, OR) for 30 min. Stained cells were viewed on an LSM410 invert laser-scan microscope (Carl Zeiss, Jena, Germany). For rhodamine visualization, ¯uorescence was excited at 543 nm and emitted light was detected at 590 nm. 2.5. Alkaline phosphatase activity As one of the di€erentiation markers of the ligament cells, their alkaline phosphatase activity was examined. Ten thousand cells were plated on Flex I culture plates in DMEM/10% FBS and incubated until they reached con¯uency. After further incubation for 24 h in DMEM/1% FBS containing 50 mg/ml ascorbic acid and 10 mM b-glycerophosphate (mineralizing medium) with or without 10 nM dexamethasone (Wako Pure

989

Chemical Industries, Osaka, Japan), cells were then subjected to cyclic stretching for 2, 4 or 6 days. Alkaline phosphate activity of cell lysates was determined by using p-nitrophenyl phosphate (Wako) as a substrate (Matsuda et al., 1993). The enzyme activity was expressed as U/mg protein. For enzyme staining, cells were washed with PBS and ®xed in 10% cold neutral-bu€ered formalin for 15 min. After washing with distilled water, cells were incubated for 45 min at room temperature in alkaline-phosphatase substrate solution consisting of 0.1 mg/ml naphthol AS MX-PO4 (Dojin Chemical, Kumamoto, Japan), 0.4% N,N-dimethylformamide (Wako) and 0.6 mg/ml Fast red-violet LB salt (Sigma) in 0.1 M Tris±HCl, pH 8.3. Cells were then washed with distilled water and with 0.5 M HCl. Finally, they were stained with 0.25% alcian blue 8GX (Sigma) in 0.5 M HCl. 2.6. Osteocalcin mRNA expression Osteocalcin is a market protein of osteoblastic di€erentiation at later stages. Expression of osteocalcin mRNA by the cells after mechanical stretching was detected by RT±PCR. Total RNA was isolated from the cells by using TRIzol (Gibco BRL) based on the method of Chomczynski and Sacchi (1987). Total RNA (1 mg) was incubated with 5 U/ml of AMV reverse transcriptase XL in the presence of 50 pmol/ml of random 9 mers and 10 mM of dNTP mixture for 30 min at 558C. Synthesized cDNA solution was mixed with 20 pmol each of the forward and reverse primers and 2.5 U of Taq polymerase, followed by ampli®cation for 30 cycles. The PCR conditions were: denaturation, 958C, 1 min; annealing, 558C, 2 min 30 s; extension, 728C, 1 min 30 s. Primer sequences for osteocalcin were: forward, 5 0 -ATGAGAGCCCTCAGACTCCTC-3 0 ; reverse, 5 0 -CGGGCCGTAGAAGCGCCGATA-3 0 (Fleet and Hock, 1994). The expected product size from these primers was 294 bp. The resulting PCR products were identi®ed and visualized by electrophoretic separation on 2% agarose gel and ethidium bromide staining. All reagents for RT±PCR was purchased from Takara Biomedicals (Shiga, Japan). 2.7. Western blot Western blot was used to detect EGF-R and phosphorylated EGF-R protein in ligament cells. For EGFR protein, con¯uent cells subjected to cyclic stretching for 4 days in mineralizing medium in the presence or absence of 10 nM dexamethasone were solubilized in detergent bu€er (1% Triton X-100, 10% glycerol, 1 mM phenylmethylsulphonyl ¯uoride in 20 mM HEPES, pH 7.4). Protein (15 mg) was loaded and separated by 5% SDS±PAGE under reduced conditions

990

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

Fig. 1. Proliferation of human periodontal-ligament cells receiving a cyclic stretch of 9% elongation (A) or 18% elongation (B). Cyclic stretch was loaded to the cells from 1 day of incubation in DMEM/10% FBS. Each plot and bar represents the mean of three independent experiments and the SE of this mean, respectively.

and transferred to a polyvinylidene ¯uoride membrane (Millipore Corporation, Bedford, MA). After blocking with 3% skim milk (Difco Laboratories, Detroit, MI), the membrane was incubated successively with a polyclonal sheep antihuman EGF-R antibody (Upstate Biotechnology, Lake Placid, NY) and a peroxidaseconjugated antisheep Ig antibody (Chemicon International, Temecula, CA). For phosphorylated EGF-R protein, con¯uent cells subjected to cyclic stretching for 4 days in the presence or absence of 10 nM dexamethasone were treated with 10 ng/ml of human recombinant EGF (Upstate) for 5 or 15 min. Cell lysate was prepared as above and subjected to SDS±PAGE, followed by Western blotting with an antiphosphotyrosine monoclonal antibody (clone 4G10, Upstate), a biotinylated antimouse Ig (Amersham Corporation, Arlington Heights, IL), and an alkaline phosphatase±streptavidin conjugate (Amersham). For quanti®cation of the results, densitometric analysis was done with NIH Image 1.55 software. Experiments were repeated four times and a typical expression pattern is shown in the Results.

out cyclic stretching (Fig. 1). Control cultures grew exponentially until day 4 of incubation after which a slight decrease in the growth rate occurred. Stretchloaded cultures also grew exponentially but exhibited slightly fewer cells throughout. This decrease in cell

3. Results 3.1. Cell proliferation and morphology Proliferation of human periodontal-ligament cells was assessed by measuring growth curves with or with-

Fig. 2. Rhodamine phalloidin staining of actin stress ®bres. (A, B) Control cultures; (C, D) cultures received a cyclic stretch of 9% elongation for 5 days; (A, C) centre area; (B, D) peripheral area. Double-headed arrows indicate the direction of stretch. Original magni®cation 50.

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

991

Fig. 3. Changes in alkaline phosphatase (ALP) activity of human periodontal-ligament cells in response to mechanical stress, dexamethasone (Dex) or both in combination. Con¯uent cells in mineralizing medium were subjected to a cyclic stretch from day 0 of 9% elongation (A) or 18% elongation (B), in the presence of absence 10 nM Dex. Each plot and bar represents the mean of four independent experiments and the SE of this mean, respectively. **p < 0.01; signi®cantly di€erent from control cultures on each day by ANOVA.

growth was independent of the number of cells inoculated as well as the degree of elongation (9 or 18%). No damaged, dead or apoptic cells were observed microscopically in stress-loaded cultures. However, after 5 days of stretching at 9% elongation, the actin stress ®bres of the cells, located within approx. 0.5 mm from the peripheral edge of the specimen, became aligned perpendicularly to the direction of

the stretch. No such alignment was observed in control cultures (Fig. 2). In cultures loaded with 18% elongation for 5 days, stress-®bre alignment was also seen in cells within 0.5 mm from the periphery (data not shown). Given that the specimen was 20  20 mm, the area where cells had become aligned was calculated as approx. 10% of the entire area of the specimen.

Table 1 ANOVA of the results shown in Fig. 3

Day 2

Day 4

Day 6

Between Residual Total Dex (+) vs (ÿ) Stretch (+) vs (ÿ) 9% vs 18% Between Residual Total Dex (+) vs (ÿ) Stretch (+) vs (ÿ) 9% vs 18% Between Residual Total Dex (+) vs (ÿ) Stretch (+) vs (ÿ) 9% vs 18%

SS

df

28081 12281 40362 20895 329 0 39114 6145 45258 43902 31170 12639 69371 20003 89374 48191 28822 1080

5 18 23 1 1 1 5 18 23 1 1 1 5 18 23 1 1 1

MS

F

5616 682

8.23**

23445 329 0 7823 341

34.36** 0.48 0.00 22.92**

23445 31170 12639 13874 1111

68.68** 91.31** 37.02** 12.48**

23445 28822 1080

21.10** 25.94** 0.97

**p < 0.01; SS, sum of squares; df, degrees of freedom; MS, mean square; F, variance ratio; Dex, dexamethasone.

992

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

Fig. 4. Alkaline phosphatase staining of human periodontal-ligament cells located at 0.6 mm, approx. half radius, from the centre of the well. Con¯uent cells in mineralizing medium were subjected to a cyclic stretch for 6 days of 18% elongation in the presence of absence 10 nM dexamethasone (Dex). (A) Control; (B) Dex; (C) stretch; (D) stretch and Dex. Original magni®cation 25.

3.2. Osteoblastic di€erentiation Con¯uent cells incubated in mineralizing medium exhibited a gradual increase in alkaline phosphatase activity from day 2 to 6 of culture (Fig. 3). The addition of 10 nM dexamethasone to the mineralizing medium caused an approx. 2-fold increase in alkaline phosphatase activity as compared to control cultures throughout the incubation period, suggesting that dexamethasone induced osteoblastic di€erentiation of the cells as reported in the rat (Matsuda et al., 1993). When cells were subjected to mechanical loading (9% elongation) from day 1, they exhibited signi®cantly higher alkaline phosphatase activity than the control on day 4, and had similar levels to those of dexamethasone-treated cultures on day 6. The e€ect of the combination of cyclic stretching and dexamethasone on alkaline phosphatase activity was more potent than the individual treatments and produced a maximum level of the cellular enzyme activity on day 6 (Fig. 3A). The alkaline phosphatase activity of the cells was also increased in response to 18% elongation, reaching a maximum on day 4 (Fig. 3B). Statistical testing by multiway ANOVA demonstrated signi®cant di€erences between dexamethasone-treated and untreated groups after 2, 4 and 6 days of culture. On day 4 and 6, di€erences between stretched and unstretched groups were signi®cant. Furthermore, the di€erence between 9% elongation and 18% elongation was also signi®cant on day 4 (Table 1).

The elevated alkaline-phosphatase activity in stretchloaded cells was further con®rmed by enzyme staining of the cells. Photomicrographs of cells located 6 mm from the centre of the well are presented in Fig. 4. According to the manufacturer, the magnitude of strain at this point was approx. 3%. A part of the cultures was weakly stained in control cells (Fig. 4A), whereas stretch-loaded (Fig. 4B) and dexamethasonetreated (Fig. 4C) cultures exhibited more positive cells as well as stronger staining. In cultures treated with stretch and dexamethasone in combination, highly positive cells were abundantly and evenly observed in the entire microscopic ®eld (Fig. 4D). Expression of osteocalcin mRNA was only slight in control cells cultured in mineralizing medium for 6 days. In contrast, those subjected to stretching for 6 days markedly expressed osteocalcin mRNA (Fig. 5),

Fig. 5. Expressions of osteocalcin mRNA by unloaded or stretch-loaded periodontal-ligament cells. Con¯uent cells were incubated in mineralizing medium with or without a cyclic stretch of 9% elongation for 4 or 6 days. Total RNA was isolated from the cells and subjected to RT-PCR using primers speci®c for osteocalcin cDNA.

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

Fig. 6. E€ect of epidermal growth factor (EGF) on proliferation of human periodontal-ligament cells in conditions of cyclic stretching. Fifty thousand cells were cultured in DMEM/10% FBS for 4 days in the presence of 0±100 ng/ml EGF, with or without receiving a stretch of 9% elongation from day 1. Each column and bar represents the mean of three independent experiments and the SE of this mean, respectively. *p < 0.05; signi®cantly di€erent by ANOVA.

suggesting that cyclic stretching induced osteoblastic di€erentiation of these cells. 3.3. E€ect of EGF When cells were plated at 5  104 per dish and incubated in the presence of 10 or 100 ng/ml of EGF for 4 days, their proliferation was promoted, as shown by their increased number (Fig. 6). Furthermore, EGF was able to restore the decreased proliferation of stretch-loaded cells (1.5  105 per dish) to the unloaded control level (1.8  105 per dish). On cell di€erentiation, similar results were obtained from the alkaline phosphatase assay in the presence of EGF. In contrast to the enzyme activity of control cultures, which was only slightly reduced by 10 ng/ml EGF after 6 days of incubation, the stretch-induced elevation of alkaline phosphatase activity after 2, 4 or 6 days of incubation was totally abolished by concomitant treatment with EGF (Fig. 7). Therefore, EGF appeared to antagonize the e€ect of cyclic stretch on proliferation and di€erentiation of these cells. 3.4. EGF-R protein in mechanical stress-loaded cells The presence of 170-kDa EGF-R protein was demonstrated by Western blotting using anti-EGF-R

993

Fig. 7. E€ect of epidermal growth factor (EGF) on alkaline phosphatase (ALP) activity of human periodontal-ligament cells in conditions of cyclic stretching. Con¯uent cells in mineralizing medium were subjected to a stretch of 9% elongation from day 0 in the presence or absence of 10 ng/ml EGF. Each plot and bar represents the mean of three independent experiments and the SE of this mean, respectively. *p < 0.05, **p < 0.01; signi®cantly di€erent from control cultures on each day by ANOVA.

antibody (Fig. 8, land 1). Quantitation of the band corresponding to EGF-R protein indicated that the amount of that protein in dexamethasone-treated or stretch-loaded cultures was 72 2 6% or 40 2 9% of control cultures, respectively (Fig. 8, lane 2 and 3). In

Fig. 8. Expression of epidermal growth-factor receptor (EGFR) protein in human periodontal-ligament cells after receiving a cyclic stretch of 9% elongation in the presence or absence of 10 nM dexamethasone (Dex) for 4 days. Whole-cell lysate was subjected to Western blot with an anti-EGF-R antibody. Each column and bar represents the mean of four independent blots and the SE of this mean, respectively. **p < 0.01; signi®cantly di€erent by ANOVA.

994

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

Fig. 9. Expression of tyrosine-phosphorylated epidermal growth-factor receptor (EGF-R) protein in human periodontal-ligament cells upon EGF stimulation. Con¯uent cells, after receiving a cyclic stretch of 9% elongation in the presence of absence of 10 nM dexamethasone (Dex) for 4 days, were exposed to 10 ng/ml EGF for 5 or 15 min. Whole-cell lysate was subjected to Western blot with an antiphosphotyrosine antibody. Each column and bar represents the mean of four independent blots and the SE of this mean, respectively. *p < 0.01; signi®cantly di€erent by ANOVA.

cultures treated with stress and dexamethasone in combination, EGF-R protein was decreased to 36 2 10% of the control level (Fig. 8, lane 4). These results indicated a decrease in EGF-R protein in the di€erentiating cells, regardless of whether the di€erentiation was induced by cyclic stretching or by dexamethasone. 3.5. Autophosphorylation of EGF-R by EGF Upon binding to EGF, EGF-R, a receptor-type tyrosine kinase, is activated by autophosphorylation of its tyrosine residue (Downward et al., 1984). Western blotting with antiphosphotyrosine antibody demonstrated a thick and dense band at 170 kDa in cultures after exposure to 10 ng/ml EGF while no other band near 170 kDa responded to EGF, suggesting that the autophosphorylation of EGF-R mainly occurred upon EGF simulation. The amount of tyrosine-phosphorylated EGF-R protein was transiently increased to 6.8 2 0.3-fold of the control level within 5 min, then decreased to 3.3 2 0.3-fold of control after 15 min of exposure (Fig. 9). In cultures treated with dexamethasone for 4 days, this autophosphorylation of EGF-R was reduced and exhibited only a 3.6 2 0.5-fold increase from the control level after 5 min of exposure. In contrast, when cells were loaded with cyclic stress for 4 days and then stimulated by EGF, the amount of tyrosine-phosphorylated EGF-R protein was restored to the control level (5.52 0.9-fold at 5 min and 4.1 2 0.6-fold at 15 min), although the amount of EGF-R protein was lower than in control cultures

(Fig. 8). When the cells were stretched for a period as short as 5 or 15 min in the presence of EGF, the amount of both EGF-R and autophosphorylated EGF-R was similar to that of control cultures (data not shown). Thus, the restored level of autophosphorylation of EGF-R was seen only in the cells that had been loaded with cyclic stretch for a relatively long time. In addition, no speci®c band for tyrosine phosphorylated protein was observed in cells subjected to stretching only. 4. Discussion Addition of EGF to culture media antagonizes the spontaneous di€erentiation of rat calvarial cells (Antosz et al., 1989; Bernier and Goltzman, 1992). We have shown that EGF inhibits dexamethasone-induced di€erentiation of rat bone-marrow stromal cells (Matsuda et al., 1996a) and rat periodontal-ligament cells (Matsuda et al., 1993); the osteoblastic di€erentiation of both these types of cell was associated with decreased synthesis of EGF-R protein. Similar to those results, here we provide original evidence that EGF inhibits the e€ect of cyclic stretching on the di€erentiation and proliferation of cells from human periodontal ligament. Furthermore, stretch-induced di€erentiation also accompanied the down-regulation of EGF-R protein. These results suggest that the EGF/ EGF-R system acts as a negative regulator of di€erentiation of periodontal-ligament cells regardless of the

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

type of di€erentiation stimuli. The cellular sensors of mechanical stress and the signal-transduction pathway have yet to be identi®ed, whereas soluble inducers of osteoblastic di€erentiation, such as glucocorticoid, parathyroid hormone and 1a,25(OH)2 vitamin D3, elicit their e€ects by binding to their own intranuclear receptors (Bernier et al., 1991; Chen et al., 1986). As the EGF/EGF-R system demonstrated its functions both in mechanical strain- and dexamethasone-induced di€erentiation, it appears that the mechanisms by which EGF/EGF-R regulates the di€erentiation do not involve interactions with speci®c signals from individual inducers of osteoblastic di€erentiation. One of the earliest events triggered by the binding of EGF and EGF-R is the elevation of tyrosine kinase activity of EGF-R and subsequent tyrosine phosphorylation of EGF-R itself (Downward et al., 1984). The autophosphorylated EGF-R then delivers signals to a variety of signalling cascades via adapters such as SH2/SH3 proteins (Gergel et al., 1994). Therefore, the level of tyrosine phosphorylation of EGF-R might relate to the level of cellular response to EGF. Here, the amount of tyrosine phosphorylated protein at 170 kDa, corresponding to EGF-R, in dexamethasonetreated cultures after EGF exposure was less than in control cultures, presumably due to the decreased amount of EGF-R. Interestingly, in stretch-loaded cells, the amount of autophosphorylated EGF-R protein was higher than in dexamethasone-treated cells and was restored to the control level, although the amount of EGF-R protein itself was even lower than dexamethasone-treated cells. This result could suggest that autophosphorylation is increased by loading mechanical stretch. Further studies on the tyrosine kinase activity of EGF-R in stretch-loaded cells are needed. The e€ect of stretching was seen in cells that had been stretched for 4 days and exhibited a high level of alkaline phosphatase activity, but not in those stretched for several hours only. Therefore, the restored tyrosine phosphorylation upon EGF stimulation is not a rapid response of the cells to cyclic stretching. Rather, various intracellular changes occurring over the long term, which relate to osteoblastic di€erentiation, might be responsible for the increased phosphorylation. Thus, although the molecular mechanisms are still to be elucidated, the restored level of tyrosine-phosphorylated EGF-R protein upon EGF stimulation is a characteristic response in mechanical stress-loaded cells and might possibly be attributed to their ability to maintain a response to EGF. In work related to ours, it was found that the amount of tyrosine-phosphorylated platelet-derived growth-factor receptor upon stimulation with that factor was substantially greater in ®broblasts cultured in mechanically stressed matrices than in monolayers (Lin and Grinnell, 1993).

995

To date, several studies, including this one, have examined the changes in in vitro di€erentiation of osteoblastic cells in response to mechanical stress, but their results are the subject of controversy. Compressive pressure on two di€erent osteoblastic cell lines established from rodents, ROS 17/2.8 and MC3T3-E1 cells, resulted in an increase (Kubota et al., 1993) and a decrease (Ozawa et al., 1990) in alkaline phosphatase activity. In studies using the Flexercell system with a degree of strain almost identical to ours, elevated alkaline-phosphatase activities and/or bone matrix-protein production were reported in primary cultures from the rat alveolar bone (Carvalho et al., 1994) and in cells established from a rat (HT-3) (Nishioka et al., 1993) and a human (OHS-4) osteosarcoma (Harter et al., 1995). While our results are consistent with these, others, in striking contrast, have reported decreased alkaline-phosphatase activity and mRNA expression in cells from human periodontal ligament (Yamaguchi and Shimizu, 1994; Yamaguchi et al., 1996). This discrepancy might be attributed to di€erences in cell-culture conditions. In those studies, semicon¯uent cultures in basal medium were loaded with cyclic stretch for a certain number of days and then alkaline phosphatase activity was assayed. In contrast, in our experiments, completely con¯uent cells were incubated in medium which contained essential supplements for osteoblastic di€erentiation of the ligament cells, ascorbic acid and b-glycerophosphate (Cho et al., 1992). Progression of the osteoblastic di€erentiation of such cells from con¯uency to the formation of mineralizing nodules in the presence of ascorbic acid and b-glycerophosphate has been described by Ramakrishnan et al. (1995). Alternatively, intrinsic di€erences in the maturation stages of cultured ligament cells might result in a di€erent responsiveness to mechanical stress, as demonstrated in bone cells by Mikuni-Takagaki et al. (1996). The Flexercell system has its limitations because it produces relatively higher strain than under physiological conditions, with an inhomogeneous distribution over the cell culture. According to previous studies, 500 g of lateral orthodontic force on human upper incisors produces 24% elongation on the tension side (Yamaguchi and Shimizu, 1994), whereas 1N force on a premolar tooth produces 1% strain (Dolce et al., 1996). Although the magnitude of the human bite force varies, forces of 500±3000 g were selected as standards for clinical studies on that force (Co€ey et al., 1989). It should be noted that the periodontal ligament is subject to intermittent loading of bite forces, whereas orthodontic force is usually continuous. We applied intermittent mechanical stretching to produce 9% or 18% of elongation at the periphery of the well. These maximum strains are almost identical to those in previous investigations (Nishioka et al., 1993;

996

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997

Carvalho et al., 1994; Harter et al., 1995; Yamaguchi et al., 1996) but would be higher than physiological. In these conditions, cells are subject to the range of ÿ5± 9% or ÿ4±18% of inhomogeneous strain, according to the manufacturer. In this study, the actin stress ®bres of the cells at the periphery of the well only, which were at least 10% of the total number of cells, became aligned perpendicularly to the direction of the stretch. A recent study showed a close relation between the stress-®bre angulation and the magnitude of stretching (Takemasa et al., 1997). Therefore, as far as the alignment of the stress ®bres is concerned, our results suggest that only a part of the total number of cells received e€ective mechanical stretching. Another explanation for the limited alignment of the cells is that other stress factors, such as ¯ow of the medium as a result of cyclic stretching, could take part in cell alignment as reported in other cell types (Malek et al., 1993). Nevertheless, the alkaline phosphatase-positive cells were distributed evenly in the entire ®eld, suggesting that most of the cells were similarly induced to di€erentiate. This would be explained by the stretch-inducible production and secretion of soluble factors that are capable of regulating cell functions via autocrine action (Resnick et al., 1993; Yamaguchi and Shimizu, 1994). We aimed to detect the net response of a cell population to cyclic strain as a model of mechanical stress, because responses of the periodontal ligament to mechanical stretching in vivo may be the consequence of the total responses of the cells under heterogeneous stress distribution. In this context, we believe that the Flexecell system could be used for our purpose within its limitations. In conclusion, we demonstrate that cultured cells of human periodontal ligament undergo osteoblastic di€erentiation upon receiving a cyclic stretch. EGF was able to inhibit this mechano-inducible in vitro di€erentiation. In turn, cellular responsiveness to EGF was regulated by the activation of EGF-R. The interaction between mechanical stress and the EGF/EGF-R system may participate in the osteoblastic di€erentiation of the ligament cells and, thereby, in regulating the function of the periodontal ligament as a source of cementoblasts and osteoblasts.

Acknowledgements We are grateful to Dr M.I. Cho of the State University of New York at Bu€alo for his helpful advice during this study. This work was supported by a grant for the Regional Links Research Program at Nagasaki of Japan Science and Technology Corporation.

References Antosz, M.E., Bellows, C.G., Aubin, J.E., 1989. E€ects of transforming growth factor b and epidermal growth factor on cell proliferation and the formation of bone nodules in isolated fetal rat calvaria cells. J. Cell. Physiol. 140, 386± 395. Aukhil, M.A., Simpson, D.M., Suggs, C., Pettersson, E., 1986. In vivo di€erentiation of progenitor cells of the periodontal ligament: an experimental study using physical barriers. J. Clin. Periodontol. 13, 862±868. Banes, A., Gilbert, J., Taylor, D., Olivier, M., 1985. A new vacuum-operated stress-providing instrument that applies static or variable duration cyclic tension or compression to cells in vitro. J. Cell. Sci. 75, 35±42. Bernier, S.M., Goltzman, D., 1992. E€ect of protein and steroidal osteotropic agents on di€erentiation and epidermal growth factor-mediated growth of the CFK1 osseous cell line. J. Cell. Physiol. 152, 317±327. Bernier, S.M., Rouleau, M.F., Goltzman, D., 1991. Biochemical and morphological analysis of the interaction of epidermal growth factor and parathyroid hormone with UMR 106 osteosarcoma cells. Endocrinol. 128, 2752±2760. Carvalho, R., Scott, J., Suga, D., Yen, E., 1994. Stimulation of signal transduction pathways in osteoblasts by mechanical strain potentiated by parathyroid hormone. J. Bone. Miner. Res. 9, 999±1011. Chen, T., Hauschka, P., Cabrales, S., Feldman, D., 1986. The e€ects of 1,25-dihydroxyvitamin D3 and Dexamethasone on rat osteoblast-like primary cell cultures' receptor occupancy and functional expression patterns for three di€erent bioresponses. Endocrinol. 118, 250±259. Cho, M.I., Lin, W.L., Garant, P.R., 1991. Occurrence of epidermal growth factor-binding sites during di€erentiation of cementoblasts and periodontal ligament ®broblasts of the young rat: a light and electron microscopic radioautographic study. Anat. Rec. 231, 14±24. Cho, M.I., Matsuda, N., Lin, W.L., Moshier, A., Ramakrishnan, P.R., 1992. In vitro formation of mineralized nodules by periodontal ligament cells from the rat. Calcif. Tissue Int. 50, 459±467. Chomczynski, P., Sacchi, N., 1987. Single-step method of RNA isolation by acid guanidium thiocyanate±phenol± chloroform extraction. Analyt. Biochem. 162, 157±159. Co€ey, J.P., Williams, W.N., Turner, G.E., Mahan, P.E., Lapointe, L.L., Cornell, C.E., 1989. Human bite force discrimination using speci®c maxillary and mandibular teeth. J. Oral Rehabil. 16, 529±536. Dolce, C., Kinniburgh, A.J., Dziak, R., 1996. Immediate early-gene induction in rat osteoblastic cells after mechanical deformation. Archs. oral Biol. 41, 1101±1108. Downward, J., Parker, P., Water®eld, MD, 1984. Autophosphorylation sites on the epidermal growth factor receptor. Nature 311, 483±485. Fleet, J.C., Hock, J.M., 1994. Identi®cation of osteocalcin mRNA in nonosteoid tissue of rats and humans by reverse transcription±polymerase chain reaction. J. Bone Miner. Res. 9, 1565±1573. Gergel, J., McNamara, D., Dobrusin, E., Zhu, G., Saltiei, A., Miller, W., 1994. Identi®cation of amino acids in the Nterminal SH2 domain of phospholipase C g1 important in

N. Matsuda et al. / Archives of Oral Biology 43 (1998) 987±997 the interaction with epidermal growth factor receptor. Biochemistry 33, 14671±14678. Harter, L.V., Hruska, K.A., Duncan, R.L., 1995. Human osteoblast-like cells respond to mechanical strain with increased bone matrix protein production independent of hormonal regulation. Endocrinol. 136, 528±535. Kawase, T., Sato, S., Miake, K., Saito, S., 1988. Alkaline phosphatase of human periodontal ligament ®broblastic cells. Adv. Dent. Res. 2, 234±239. Kubota, T., Yamauchi, M., Onozaki, J., Sato, S., Suzuki, Y., Sodek, J., 1993. In¯uence of an intermittent compressive force on matrix protein expression by ROS 17/2.8 cells, with selective stimulation of osteopontin. Archs. oral Biol. 38, 23±30. Lin, Y-C., Grinnell, F., 1993. Decreased level of PDGFstimulated receptor autophosphorylation by ®broblasts in mechanically relaxed collagen matrices. J. Cell Biol. 122, 663±672. Malek, A., Gibbons, G., Dzau, V., Izumo, S., 1993. Fluid shear stress di€erentially modulates expression of genes encoding basic ®broblast growth factor and platelet-derived growth factor B chain in vascular endothelium. J. Clin. Invest. 92, 2013±2021. Matsuda, N., Kumar, N.M., Ramakrishnan, P.R., Genco, R.J., Cho, M.I., 1993. Evidence for up-regulation of epidermal growth factor receptors on rat periodontal ligament ®broblastic cells associated with stabilization of phenotype in vitro. Archs. oral Biol. 38, 559±570. Matsuda, N., Kumar, N.M., Ramakrishnan, P.R., Cho, M.I., 1996a. Role of epidermal growth factor receptor in osteoblastic di€erentiation of rat bone marrow stromal cells. J. Bone Miner. Metab. 14, 137±145. Matsuda, N., Takemura, A., Taniguchi, S., Amano, A., Shizukuishi, S., 1996b. Porphyromonas gingivalis reduces mitogenic and chemotactic responses of human periodontal ligament cells to platelet-derived growth factor in vitro. J. Periodontol. 67, 1335±1341. McCulloch, C.A.G., Melcher, A.H., 1983. Cell density and cell generation in the periodontal ligament of mice. Am. J. Anat. 167, 43±58. Mikuni-Takagaki, Y., Suzuki, Y., Kawase, T., Saito, S., 1996. Distinct responses of di€erent populations of bone cells to mechanical stress. Endocrinol. 137, 2028±2035. Nishioka, S., Fukuda, K., Tanaka, S., 1993. Cyclic stretch increases alkaline phosphatase activity of osteoblast-like cells: a role for prostaglandin E2. Bone Miner. 21, 141±150. Nojima, N., Kobayashi, M., Shinonome, M., Takahashi, N., Suda, T., Hasegawa, K., 1990. Fibroblastic cells derived

997

from bovine periodontal ligaments have the phenotypes of osteoblasts. J. Periodont. Res. 25, 179±185. Ozawa, H., Imamura, K., Abe, E., Takahashi, N., Hiraide, T., Shibasaki, Y., Fukuhara, T., Suda, T., 1990. E€ect of a continuously applied compressive pressure on mouse osteoblast-like cells (MC3T3-E1) in vitro. J. Cell. Physiol. 142, 177±185. Raab-Cullen, D.M., Thiede, M.A., Petersen, D.N., Kimmel, D.B., Recker, R.R., 1994. Mechanical loading stimulates rapid changes in periosteal gene expression. Calcif. Tiss. Int. 55, 473±478. Ramakrishnan, P., Lin, W., Sodek, J., Cho, M., 1995. Synthesis of noncollagenous extracellular matrix proteins during development of mineralized nodules by rat periodontal ligament cells in vitro. Calcif. Tiss. Int. 57, 52±59. Resnick, N., Collins, T., Atkinson, W., Bonthron, D., Dewey, C., Gimbrone, M., 1993. Platelet-derived growth factor B chain promoter contains a cis-acting ¯uid shear-stressresponsive element. Proc. Natl. Acad. Sci. USA 90, 4591± 4595. Roberts, W.E., Mozsary, P.G., Klingler, E., 1982. Nuclear size as a cell-kinetic marker for osteoblast di€erentiation. Am. J. Anat. 165, 373±384. Somerman, M.J., Archer, S.Y., Imm, G.R., Foster, R.A., 1988. A comparative study of human periodontal ligament cells and gingival ®broblasts in vitro. J. Dent. Res. 67, 66± 70. Takemasa, T., Sugimoto, K., Yamashita, K., 1997. Amplitude-dependent stress ®ber reorientation in early response to cyclic strain. Exp. Cell Res. 230, 407±410. Veldhuijzen, J., Bourret, L., Rodan, G., 1979. In vitro studies of the e€ect of intermittent compressive forces on cartilage cell proliferation. J. Cell. Physiol. 98, 299±306. Yamaguchi, M., Shimizu, N., 1994. Identi®cation of factors mediating the decrease of alkaline phosphatase activity caused by tension-force in periodontal ligament cells. Gen. Pharmacol. 25, 1229±1235. Yamaguchi, M., Shimizu, N., Goseki, T., Shibata, Y., Takiguchi, H., Iwasawa, T., Abiko, Y., 1994. E€ect of di€erent magnitude of tension force on prostagladin E2 production by human periodontal ligament cells. Archs. oral Biol. 39, 877±884. Yamaguchi, M., Shimizu, N., Shibata, Y., Abiko, Y., 1996. E€ects on di€erent magnitudes of tension-force on alkaline phosphatase activity in periodontal ligament cells. J. Dent. Res. 75, 889±894.