Compressive force promotes Sox9, type II collagen and aggrecan and inhibits IL-1β expression resulting in chondrogenesis in mouse embryonic limb bud mesenchymal cells

The musculoskeletal system including bone, cartilage, skeletal muscles and ligaments responds to biomechanical stimuli by altering their metabolism, cellular and cytoskeletal organization, rate of proliferation and state of differentiation during development. For example, exercise induces physiological changes in muscles, bone and articular cartilage, and static force applied during orthodontic treatments induces bone remodeling, differentiation of chondrocytes and of other connective tissue cells (Asano, 1986; McNamara and Carlson, 1979; Takahashi et al., 1995). In addition, excessive or inappropriate force loading is a contributing factor in common skeletal degenerative diseases. The development of these tissues during embryonic development may also be regulated by biomechanical stimuli, contributing to the initial modeling of the three-dimensional structure of the skeleton (van Limborgh, 1982).

Indeed, mathematical modeling predicts that biomechanical forces generated by cell migration and cell division create cell-cell and/or cell-extracellular matrix (ECM) interactions which contribute to early patterning and morphogenesis in many tissues and organs, including neural tube (Jacobson and Gordon, 1976) and precartilaginous mesenchymal condensation (Edelstein-Keshet and Ermentrout, 1990; Murray and Oster, 1984; Ngwa and Maini, 1995; Totafurno and Bjerknes, 1995). During chondrogenesis, it was proposed that the osmotic changes during hyaluronic acid synthesis and degradation could produce sufficient swelling and deswelling to regulate the balance between cell-cell versus cell-ECM contacts (Oster et al., 1985).

Experimental models, though limited, have generally supported these theories. Paralyzed chicken embryos failed to form the clavicles and the quadratojugal, which is associated with the failure to activate chondrogenic markers, illustrating the prerequisite of physical movements in the initial stages of chondrogenesis (Fang and Hall, 1995; Hall, 1979; Hall and Herring, 1990). In vitro models using intermittent compressive force accelerated the maturation of chondrocytes (van Kampen et al., 1985; van’t Veen et al., 1995; Veldhuijzen et al., 1987). However, the function of static compressive force in the early molecular differentiation of chondroprogenitor cells frequently encountered in orthodontic and orthopedic treatments is as yet unknown.

Morphogenesis of the skeletal system begins with the condensation of undifferentiated embryonic mesenchymal cells (Hall, 1988). Condensation is promoted by a regulated balance between cell-ECM and cell-cell adhesion involving type I and type III collagen (von der Mark and von der Mark, 1977), tenascin (Pacifici, 1995), fibronectin (Silbermann et al., 1987), cellular receptors for these matrix proteins such as integrins (Camper et al., 1997), and other non-integrin adhesion molecules (Noonan et al., 1996), such as cell-cell adhesion molecules N-CAM and N-cadherin (Oberlender and Tuan, 1994; Tavella et al., 1994; Tsonis et al., 1994). After condensation, the differentiating chondrocytes change their cell shape and alter their cell adhesion properties. Chondrocytes express type II collagen (Mizoguchi et al., 1990; von der Mark and von der Mark, 1977), aggrecan (Vornehm et al., 1996), and associated tissue-specific glycosaminoglycans (Takahashi et al., 1996). As these matrix molecules are deposited between cells, cell-cell contacts are lost (Tavella et al., 1994). New proliferating chondrocytes establish the pattern of subsequent bone formation. At the epiphyseal growth plate, the chondrocytes become hypertrophic, the cartilage matrix becomes calcified, and the cartilage is replaced by bone in the process of endochondral ossification. At other sites, such as the formation of synovial joints, chondrocytes maintain the cartilage matrix through adulthood. Thus, skeletal development provides mechanical stability and mobility, and is likely to respond reciprocally to biomechanical stimuli in order to adapt to the range of force exerted and motion. The ECM component is conceivably the transducer of biomechanical stimuli into transcriptional controls.

Signal transduction initiated by growth factors and cytokines, and subsequent transcriptional controls contribute broadly to the regulation of cartilage and bone development. However, one such cytokine, interleukin 1-beta (IL-1β), has been demonstrated to inhibit chondrocyte differentiation by directly suppressing transcription of the cartilage specific marker type II collagen gene, and also inhibits glycosaminoglycan production leading to decreased synthesis of another cartilage specific marker, aggrecan (Chandrasekhar et al., 1994; Gibson et al., 1984; Goldring et al., 1994a,b). IL-1β also induces the expression of matrix metalloproteases, molecules involved in the degradation of cartilage matrix during hypertrophy (Yang and Gerstenfeld, 1997). In contrast to IL-1β being a negative regulator of chondrogenesis, the transcription factor Sox9 is a positive regulator in this process. Sox9 is expressed in precartilaginous condensing mesenchyme and maturing cartilage (Wright et al., 1995; Zhao et al., 1997). Sox9 binds directly to an enhancer element in the first intron of the collagen II gene (Lefebvre et al., 1996; Ng et al., 1997), and upregulates the type II collagen expression (Bell et al., 1997).

In this investigation, studies were designed to test the hypothesis that compression promoted chondrogenesis is a response of molecular determinants to biomechanical forces. Due to the complexity of differential mechanical properties among different cell populations and local matrix components, and normal movements, an in vitro three-dimensional collagen gel cell culture system was developed. We report that static compressive force promotes the expression of two cartilage specific markers, type II collagen and aggrecan. This accelerated chondrogenesis was associated with an upregulation of the positive regulator Sox9, and downregulation of the negative regulator IL-1β. An understanding of the biomechanical stimuli that regulate cartilage development and maintenance can contribute to the improved management of skeletal malformations and progressive joint diseases.

Timed-pregnant Swiss Webster mice (with day of detection of vaginal sperm plug designated as day 0 of gestation) were purchased (Harlan Sprague Dawley, Inc., Indianapolis, Indiana). At gestation day 10, pregnant mice were sacrificed and the E10 stage embryos were isolated. Fore- and hindlimb buds were microdissected in ice-cold phosphate buffered saline (PBS), placed on ice, and subsequently they were washed three times in cold PBS, dissociated in 0.25 mg/ml trypsin EDTA (Life Technologies Gibco BRL, Inc., Gaithersburg, MD) and 0.25 mg/ml collagenase type 2 (Washington Biochemical Corporation, Freehold, NJ) in 0.1 M PBS for 15 minutes at 37°C, and cell numbers were determined using a hemocytometer. Subsequently, cells were collected by centrifugation at 200 g for 15 minutes at 4°C and resuspended in Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies Gibco BRL, Inc., Gaithersburg, MD) at 4.5×10⁷ or 5.0×10⁷ cells/ml for compressed and control groups. Cells were cultured in three-dimensional collagen gel system (described below) in DMEM supplemented with 10% heat inactivated fetal bovine serum (HyClone Laboratories Inc, Logan, UT), 2.4 mg/ml of Hepes (ICN Biomedicals Inc., Aurora, OH), 0.2% bicarbonate, 2 mM glutamine (Sigma, St Louis, MO), 100 units of penicillin, 100 μg of streptomycin and 0.25 μg of amphotericin (Life Technologies Gibco BRL, Inc., Gaithersburg, MD).

In order to determine the effect of static compressive force on the differentiation of limb bud mesenchymal cells, cells were embedded and cultured in a three-dimensional collagen gel system to mimic in vivo conditions (Yasui et al., 1982) as illustrated in Fig. 1. A collagen gel stock was prepared by dissolving type I collagen from calf skin (Sigma, St Louis, MO) in 1 mM acetic acid at a concentration of 3.33 mg/ml and dialyzed against deionized water for 6 hours. The collagen gel cultures were assembled by mixing 10 volumes of 2× medium, 9 volumes of collagen stock solution and 1 volume of cell suspension to give a final cell density of 2.25×10⁶ cells/ml for compressed gels and 2.5×10⁶ cells/ml for control gel at a collagen gel concentration of 1.5 mg/ml. These gels were plated in 96-well plates and allowed to polymerized for 1 hour, after which the gels were transferred to 24-well plates so that an agarose supporting ring could be cast around the cultures. Static compressive forces were applied by the use of a weight placed on the lid of the culture plate exerted through a well insert which fitted on top of the collagen gel as illustrated in Fig. 1. The weight per unit surface area of the collagen gel was converted to compressive pressure as kiloPascals (kPa). Since our collagen gel culture system is unique, the amount of force used in our experiments; 1, 1.5, and 2 kPa was empirically determined to yield a 20-30% deformation of the gel as used in previous studies (Buschmann et al., 1995; Kim et al., 1994, 1996). Cultures were incubated at 37°C, 5% CO₂ for up to 10 days with daily medium changes. Control non-compressed cultures were assembled without force loading.

Total RNAs were isolated from both compressed and control gels at 0, 1, 3, 5, 7 and 10 days of culture, and from E9, 10, 11, 12 and 14 mouse limb buds using a guanidine isothiocyanate, phenol-chloroform extraction based total RNA isolation kit (5 prime 3 prime Inc., Boulder, CO) according to specifications from the manufacturer. Total RNA was transcribed into cDNA using SuperScript II reverse transcriptase (Life Technologies Gibco BRL, Inc., Gaithersburg, MD). Genomic DNA was also isolated from samples at day 0, 3, 5 and 7 of culture using routine phenol-chloroform extraction and ethanol precipitation methods. A series of predetermined cell numbers were used for DNA content to generate a standard curve of cell number versus total DNA content. Total cell numbers in the experimental limb bud cell cultures were calculated using this standard curve.

In order to accurately quantitate the expression of chondrogenic markers, type II collagen and aggrecan, competitive RT-PCR was performed (Siebert and Larrick, 1992). Amplimers designed for type II collagen were 5′-TGAAGACATCCGCAGCCCC-3′ and 5′-ATAATGGGAAGGCGGGAGG-3′, based on the mouse type II collagen genomic DNA sequence (Metsaranta et al., 1991) and for aggrecan 5′-GTCCCTGGTCAGCCCCGCTTG-3′ and 5′-CACTGA-CACACCTCGGAAGCA-3′ based on the mouse aggrecan cDNA sequences (Watanabe et al., 1995). Competitors bearing identical 5′ and 3′ amplification arms of type II collagen and aggrecan sequences, but with an altered internal sequence, were constructed using the PCR MIMIC construction kit (Clontech, Palo Alto, CA) according to specifications from the manufacturer. Total RNA from E10 mouse limb buds was extracted, reverse transcribed and subjected to PCR amplification for both type II collagen and aggrecan. PCR products of target templates and competitors for type II collagen and aggrecan were subcloned into pCRII vector (Invitrogen, Carlsbad, CA), amplified and transcribed into cRNA using an RNA transcription kit (Stratagene, La Jolla, CA). Standard curves of the logarithm of template versus the ratio of template to competitors were generated for type II collagen and aggrecan. Correlation formulae were deduced (Zhao et al., 1996) from the standard curves and regressions of 0.9885 and 0.9606 were calculated for type II collagen and aggrecan, respectively. PCR amplifications were carried out by using ampliTaq (Life Technologies Gibco BRL, Inc., Gaithersburg, MD) and DNA Thermal cycler 480 (Perkin Elmer, Branchburg, NJ). Thermal cycling programs for quantitative PCR were 5 minutes at 95°C for denaturation, 45 seconds at 55°C for annealing, and 2 minutes at 72°C for extension for the first cycle, followed by similar cycles changing only the denaturation step to 45 seconds and a final extension cycle of 10 minutes. Cycle numbers were empirically determined to be 27 and 32 for type II collagen and aggrecan, respectively, to optimize for signal and amplification linearity.

Semi-quantitative RT-PCR method was performed to evaluate the expression levels of Sox9 and IL-1β relative to that of β-actin. Amplimers designed for Sox9 were 5′-AGAAAGACCACCCC-GATTA-3′ and 5′-ACTGGGACG-ACATGGAGAAG-3′ based on the Sox9 cDNA sequence (Wright et al., 1995), and that for β-actin were 5′-TGCTGAT-GCCGTAACTGCC-3′ and 5′-TGAGGTAGTCCGT-CAGGTCC-3′ based on the actin cDNA sequence (Alonso et al., 1986). Twenty-eight cycles of amplification for Sox9 and 22 cycles for β-actin were empirically determined to optimize for signal and amplification linearity. Nested PCR was performed to quantitate the expression level of IL-1β, since its expression level was relatively low. Amplimers designed for IL-1β were 5′-GGGATGATGATG-ATAACCTGCTG-3′ and 5′-TTCTTGTGACCCTGAGCGACCTG-3′ for primary PCR and 5′-GGGATGATGATGATAACCTGCTGG-3′ and 5′-CCTGGGGA-AGGCATTAGAAACAGT-3′ for secondary PCR based on the IL-1β cDNA sequence (Gray et al., 1986). Forty cycles of primary amplification followed by 10 cycles of secondary amplification were also empirically determined to optimize for amplification linearity.

After PCR amplification, products were electrophoresed on 2.0% agarose gel in Tris acetate buffer, stained with ethidium bromide, imaged with Charged Coupled Device camera, then pixel depth was converted to optical density and quantitated with the aid of NIH image software package (NIH, Bethesda, MD). The amount of type II collagen and aggrecan mRNAs were calculated from the standard curve generated from competitive PCR and the amount of Sox9 and IL-1β mRNA was quantitated relative to β-actin expression.

Collagen gel supported cell cultures at day 5 and 7 were fixed with 4% paraformaldehyde in PBS for 12 hours. Subsequently, specimens were thoroughly rinsed in 0.01 M PBS, infiltrated through a graded series of sucrose and acrylamide, embedded first in 10% acrylamide gel and then in Tissue Freezing Medium (Triangle Biomedical Sciences, Durham, NC). Ten μm-thick sections were collected and subjected to either Toluidine Blue staining or indirect fluorescence immunohistochemistry for type II and X collagens (gift from Dr Sasano), IL-1β and IL-1RI (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Immunohistochemical staining was performed as previously described (Mizoguchi et al., 1990). Total cell number and the number of IL-1β positive cells were counted in a 270 μm by 310 μm area for every third or fourth serial section for 10 sections of each compressed and control sample at day 7 of culture.

IL-1β in conditioned medium from day 3, 5, 7 and 10 of culture was assayed using Quantikine M mouse IL-1β immunoassay kit (R&D System, Minneapolis, MN) according to specifications from the manufacturer.

Transcript expression levels for type II collagen, aggrecan, IL-1β and Sox9 as well as protein levels of IL-1β between compressed and control groups over the time course of study were subjected to statistical analyses using Student’s t-test. Statistical significance was deduced at P<0.05.

We developed an in vitro three-dimensional model system for culturing dissociated mouse embryonic limb bud mesenchyme in collagen gels which mimicks early in vivo chondrogenesis (Fig. 1). A static compressive force of 1 kPa was applied and evidence of chondrogenesis was assayed using histological and molecular markers. The rate and extent of chondrogenesis were compared with that of control non-compressed cultures. Control and compressed cultures exhibited the formation of cartilaginous nodules accompanied by cartilage specific matrix deposition as reflected by metachromatic Toluidine Blue staining (Fig. 2A to D). By day 5, compressed cultures contained more chondrogenic aggregates and deposited more cartilaginous matrix than controls. By day 7, compressed cultures contained cartilage lacunae indicative of maturing chondrocytes that were not seen in control cultures. In addition, cartilaginous nodule chondrocytes in compressed cultures showed aggregated linear alignment oriented perpendicular to the direction of force applied, whereas control cultures showed dispersed and more randomly distributed cells in the matrix. Cartilage ECM synthesis was characterized by positive immunocytochemical reaction to both type II and X collagen in the chondrocytes in both control and compressed cultures on days 5 and 7. Consistent with Toluidine Blue staining, there was more cartilage specific matrix accumulation in compressed cultures (Fig. 2E and F). Furthermore, type X collagen accumulation in the ECM surrounding matured chondrocytes were observed in compressed cultures at day 7 (Fig. 2G and H).

In order to quantitate the accelerated process of chondrogenesis under static compressive force, a quantitative RT-PCR method was used to assess type II collagen or aggrecan mRNA. The sensitivity and effective range of our quantitative RT-PCR methods was 0.002 to 5.95 pg/μg of total RNA. The linearity was reflected by r=0.994 and r=0.980 for type II collagen and aggrecan, respectively.

Differentiating chondrocytes went through an initial phase of cell condensation and subsequently begin to deposit cartilage specific matrix. Consistent with histological observations, matrix molecule transcripts were apparent by RT-PCR by day 3 of culture, with a major increase in transcript levels between day 3 and 5. Quantitation of type II collagen transcripts in compressed cultures by competitive RT-PCR revealed a 5-fold increase on day 3 when compared with control cultures. This upregulation was maintained and significant between control and compressed cultures through 10 days of culture; the difference at day 10 being 2-fold (Fig. 3A). Consistent with the profile of type II collagen expression, aggrecan mRNA, was also upregulated (Fig. 3B). Aggrecan mRNA levels in compressed cultures was 1.9-fold higher at 5 days than controls (which increased to a 3.4-fold elevation at day 10).

The acceleration of type II collagen expression by compressive force was found to be directly proportional to the magnitude of compressive force applied (Fig. 3C). When 1.5 and 2 kPa of force was loaded, type II collagen mRNA increased 1.5- and 3.8-fold, respectively, over transcript levels detected in samples subjected to 1 kPa of force magnitude. In comparison to the control (non-compressed) samples, a maximum force used in our experiments of 2kPa of force magnitude produced an 8.3-fold increase in type II collagen mRNA.

A collagen gel matrix under compressive force is likely to become more compact when compared with non-compressed matrix. Compaction may lead to changes in gel density and cell number per unit volume. It has been previously reported that high cell density per se promotes chondrogenesis (Solursh and Meier, 1974). Therefore, in order to examine the primary effect of compression on chondrogenesis, we should control for changes in gel density and cell number in our three-dimensional collagen gel cell culture system. Two strategies were used: (1) we monitored and compared the changes in gel density and cell number in non-compressed versus compressed collagen gels over the time course of our studies; (2) we used the observed differences to make compensatory adjustments and assayed for effects on the rate and extent of chondrogenesis.

The total cell number increased linearly in all cultures. Non-compressed cultures contained 20% more cells than compressed cultures beginning on day 3 and through day 7 (Fig. 4A). The volume of both compressed and non-compressed culture gels exhibited a phase increase from day 0 through day 3, and subsequently a gradual phase of decrease through day 7 (Fig. 4B). The volume stabilized and remained relatively constant after day 10 (data not shown). We suggest that these gel volume changes are due to the increase in cell number and/or changes in the hydration state of the gels for the first (expansion) phase, followed by a decrease in gel thickness in the second phase. Compressed gels that did not contain any cells exhibited expansion to a lesser extent (data not shown). The maximum difference in volume between compressed and non-compressed gels was 30%.

Based upon these observations, we then assayed type II collagen mRNA levels under various combinations of increased cell number and collagen gel concentration (based on the differences detected in the earlier experiments). Since the volume of non-compressed gels was approximately 30% more than compressed, we varied the collagen concentration by this amount, increasing from 1.5 to 1.95 mg collagen gel matrix/ml in control cultures. The total cell number was 20% more in non-compressed than compressed cultures. From these results, we varied the cell density by 10%; increasing from 2.25×10⁶ to 2.5×10⁶ cells/ml in control cultures. Fig. 4C shows a time course of type II collagen transcript level in non-compressed cultures with combinations of cell number and collagen gel density adjustments. No significant difference in type II collagen expression was detected among groups at each time point. Therefore, chondrogenesis advanced by compression is a direct effect attributable to the force applied and not secondarily mediated by associated changes in cell number or gel density found in compressed cultures.

We observed that compressive forces promote chondrogenesis in embryonic limb bud mesenchyme cells through experiments designed to test the increase of type II collagen as a regulated event mediated through positive regulators of type II collagen gene transcription. We analyzed the pattern of Sox9 expression during embryonic limb development in vivo. In vivo expression of Sox9 in the mouse embryo was detectable at low levels at E9. Sox9 expression peaked at E11 and then decreased through E14. At E11, Sox9 expression preceded that of the onset of an exponential accumulation of type II collagen transcripts which occurred between E12-14 (Fig. 5A). Sox9 was expressed in control and compressed cultures (Fig. 5B). Similar to in vivo profiles, Sox9 transcript levels peaked at day 3 of culture in both groups and then decreased. However, in compressed cultures, despite an overall decrease in gene expression, Sox9 transcript level was maintained at an elevated state as compared with control cultures through day 10. The ratio between compressed and control levels was 2.9 (day 5) and then reached 6.9 at day 10. Statistically significant differences in Sox9 expression in control and compressed cultures were detectable at days 7 and 10.

IL-1β is a direct suppresser of type II collagen and aggrecan gene expression in chondrocytes (Goldring et al., 1994a,b). IL-1β mRNA expression and the accumulation of the activated protein in conditioned culture media of control and compressed cultures were assayed by semi-quantitative RT-PCR and ELISA, respectively. Compressed cultures showed significantly less IL-1β transcript expression when compared with controls at days 3, 5 and 7 (Fig. 6A). The difference was largest at day 5; compressed cultures only had 1/4 of the expression level of the control value. The overall protein accumulation increased from day 3 to 7, and then stabilized after day 7 in both control and compressed groups (Fig. 6B). Peak RNA expression on day 5 (Fig. 6A) was followed by peak protein accumulation 48 hours later (Fig. 6B). However, compressed cultures showed significantly less IL-1β accumulation than that of contol cultures; compressed culture had 20% less IL-1β than control cultures at day 5 and 7. This is consistent with a lower RNA expression level beginning at day 3 prior to a detectable decrease in protein accumulation which was first observed at day 5. Since the total cell numbers in compressed cultures were slightly lower than those of controls over the course of the culture period as shown in Fig. 4, it is possible that the difference observed in IL1-β protein accumulation only reflected the difference in total cell number rather than expression level in each cell. Therefore, we counted the number of IL-1β expressing cells on the immunostained sections (Fig. 6D and F) in both control and compressed samples at day 7 of culture to standardize for the accumulation of protein per cell. The number of cells expressing IL-1β in compressed and control cultures were (2.8±0.9)×10⁶ and (2.7±0.4)×10⁶ per gel (n=3), respectively, the two values not being statistically different (P=0.74). This suggests that indeed the difference in the accumulation of IL-1β in the culture media was indicative of the expression level itself. Both IL-1β and its receptor, IL-1RI, were detectable by immunocytochemistry, demonstrating that they were coexpressed in chondrocytes at day 7 in both control and compressed groups (Fig. 6D and F). Therefore, IL-1β may exert its action by an autocrine and/or paracrine mode.

Using a unique three-dimensional collagen gel cell culture system that mimicks much of the cell adhesion interactions of embryonic tissues, we have determined that static compressive force promotes chondrogenesis during embryonic limb bud mesenchymal cell differentiation. The rate and extent of type II collagen and aggrecan expression in differentiating chondrocytes were significantly increased under compressive forces. Further, we also determined that this accelerated pattern of chondrogenesis was accompanied by up-regulation of Sox9 expression (a transcriptional activator of type II collagen), and down-regulation of IL-1β (a repressor of both type II collagen and aggrecan). The balance of signaling between a combination of intermediate activator and repressor suggests that the biomechanical stimuli and subsequent transduction resulting in differential gene activation is a tightly regulated process as summarized in Fig. 7.

Since the mesenchymal cells in our experiments are derived from embryonic mouse limb buds just prior to chondrogenesis in vivo, we assume that our in vitro model system reflects a dynamic range of signaling circuits required for normal differentiation of chondrocytes from mesenchymal cells. This culture system contains a heterogeneous population of cells including epithelial and precursor cells for bone, cartilage, muscle and other connective tissues. The acceleration of chondrogenesis observed in compressed cultures is likely due to both an increase in the number of cells differentiating into chondrocytes as well as an increase in the expression of cartilage-specific ECM proteins associated with each chondrocyte. Immunohistochemical analyses revealed that chondrocytes were more numerous and that these cells exhibited a more advanced state of differentiation as characterized by type II and X collagen deposition in the ECM under compression as compared to non-compressed controls. Increased chondrocytes may reflect increased proliferation of the prechondrocyte sub-population, or a recruitment of cells from other cell lineages to become chondrocytes in response to compressive force stimuli.

These experiments demonstrate responsiveness of the developmental process of chondrogenesis to external biomechanical stimuli. During skeletogenesis, the initial pattern of most of the skeleton is first established as cartilaginous tissue which is subsequently replaced with bone by the process of endochondral ossification. Epigenetic factors such as compressive force that stimulate chondrogenesis can have a significant impact on long term three-dimensional modeling of the structural elements in bones. Indeed, from our histological observations, the aggregated linear alignment of chondrocytes appeared to orient perpendicular to the direction of compressive force delivery and suggests that the construction of bony pillars in the epiphyseal region of long bones directed to the plane of weight bearing could also be controlled by similar biomechanical stimuli prior to and/or during bone deposition. This interpretation is consistent with Wolff’s law which states that directional modeling of bony pillars during ossification is a result of resistance to biomechanical stress.

Two morphoregulatory factors were identified to be involved in controlling chondrogenesis in response to compressive biomechanical stimuli. Sox9 expression patterns showed a similar peak level for expression in both compressed and control cultures (Fig. 5). However, a more gradual decrease in expression level was observed only in compressed cultures subsequent to the peak (Fig. 5B). Thus, a higher Sox9 mRNA level in compressed cultures was effectively sustained over the same time course when compared with controls. If Sox9 is a limiting factor in transcriptional regulation of type II collagen, higher Sox9 expression could account for the increase in type II collagen transcript accumulation that was observed at later time points. However, our data also suggest that other transcription factors play an important role in regulating type II collagen transcription, since we observed a statistically significant stimulation in type II collagen expression at the day 3 time point while no difference in Sox9 expression was detected until day 5 (Figs 3A, 4C and 5).

IL-1β expression and accumulation were suppressed in compressed cultures as a direct consequence of the biomechanical stimuli (Fig. 6). This is consistent with the promotion of chondrogenesis, since IL-1β has been shown to inhibit type II collagen transcription and glycosaminoglycan production leading to decreased aggrecan synthesis (Frisbie and Nixon, 1997). Further, IL-1β induces the cartilage degrading enzymes matrix metalloproteases (MMPs) (Arner and Tortorella, 1995; Yang and Gerstenfeld, 1997). Decreased IL-1β expression could lead to a greater accumulation of cartilage matrix, consistent with our histological and immunohistochemical findings. Immunostaining revealed coexpression of IL-1β and IL-1RI in chondrocytes, whereas neighboring non-chondrocytic cells expressed only the receptor (Fig. 6). This suggests that IL-1β may regulate chondrogenesis through either a paracrine or autocrine feedback loop, and the response of chondrocytes to compressive force could be conveyed to neighboring non-chondrocytic cells via a diffusible cytokine.

Studies of the effects of compressive force on mature articular cartilage (Buschmann et al., 1995; Lee and Bader, 1997) reveal significant differences with our results with embryonic mesenchymal cells. Dynamic, periodic loading of articular cartilage up-regulates the incorporation of ³H-proline and ³⁵S-sulfate into cartilage ECM, while static compressive force down-regulates cartilage matrix deposition (Kim et al., 1994) and aggrecan expression (Kim et al., 1996). In vitro studies with mandibular condyle cartilage, non-terminally differentiated growth plate cartilage, showed similar results. In contrast, collagen and proteoglycan biosynthesis was up-regulated in postnatal growth plate cartilage of mandibular condyle cultured under compressive forces (Copray et al., 1985). Further, compaction and low oxygen tension lead to chondrogenesis in periosteal membrane cultures (Bassette and Herrmann, 1961). Compressive stimulation on midpalatal suture cartilage in rats promoted maturation and hypertrophy of precartilaginous undifferentiated mesenchymal cells in vivo (Saitoh et al., 1997). In vivo studies in experimentally paralyzed chicken embryos (Hall and Herring, 1990) and muscular dysgenic mice (Herring and Lakars, 1981) demonstrate that decreased compression of developing bones by the associated skeletal muscles leads to decreased chondrogenesis. The effects of compressive forces on chondrogenesis are, therefore, dependent on the specific type of cartilages, the stage of cartilage maturation, the site of force loading, and the types of forces exerted upon the cartilage. The effect of static compressive force on chondrocytes differentiation from mesenchymal cells in this culture model reveals that important changes occurs during differentiation pathways between the detection of force signals and cellular responses to those epigenetic signals.

In summary, we demonstrate that biomechanical compressive stimuli accelerated the progression of chondrogenesis in mouse limb mesenchymal cells through regulating the expression of Sox9 and IL-1β as positive and negative regulators, respectively, for the expression of cartilage specific macromolecules. Possible mechanisms of how compressive stimuli are translated into biochemical signals might involve integrin or non-integrin ECM receptors, cell-cell adhesion molecules or alteration of cytoskeletal architecture mediated by MAP kinase (Banes et al., 1995; Kimura et al., 1996), rho kinase (Chrzanowska-Wodnicka and Burridge, 1996) or focal adhesion associated tyrosine kinase signal transduction pathways (Yamada and Miyamoto, 1995; Yamada and Geiger, 1997). However, the molecular mechanisms remain as yet unkown. Further investigations are necessary to clarify the molecular mechanisms by which biomechanical stimuli affect skeletogenesis through mechano-biochemical translation and transcriptional regulation. Our model system of chondrogenesis under compressive load provides the opportunity for testing the interplay of ECM or cell-cell adhesion receptor and growth factor receptors in controlling chondrocyte differentiation.

We are grateful to Dr Yasuyuki Sasano (Second Department of Oral Anatomy, School of Dentistry, Tohoku University) for the generous gift of antibodies against type II and X collagens. This work was supported by NIH grant Z01-AR41114-02-BCTB.