Mechanical Regulation of Mitogen-activated Protein Kinase Signaling in Articular Cartilage*

Articular chondrocytes respond to mechanical forces by alterations in gene expression, proliferative status, and metabolic functions. Little is known concerning the cell signaling systems that receive, transduce, and convey mechanical information to the chondrocyte interior. Here, we show that ex vivo cartilage compression stimulates the phosphorylation of ERK1/2, p38 MAPK, and SAPK/ERK kinase-1 (SEK1) of the JNK pathway. Mechanical compression induced a phased phosphorylation of ERK consisting of a rapid induction of ERK1/2 phosphorylation at 10 min, a rapid decay, and a sustained level of ERK2 phosphorylation that persisted for at least 24 h. Mechanical compression also induced the phosphorylation of p38 MAPK in strictly a transient fashion, with maximal phosphorylation occurring at 10 min. Mechanical compression stimulated SEK1 phosphorylation, with a maximum at the relatively delayed time point of 1 h and with a higher amplitude than ERK1/2 and p38 MAPK phosphorylation. These data demonstrate that mechanical compression alone activates MAPK signaling in intact cartilage. In addition, these data demonstrate distinct temporal patterns of MAPK signaling in response to mechanical loading and to the anabolic insulin-like growth factor-I. Finally, the data indicate that compression coactivates distinct signaling pathways that may help define the nature of mechanotransduction in cartilage.

Articular cartilage serves as a bearing surface for joints (1,2). In this capacity, it is routinely exposed to mechanical loading. Articular cartilage is nearly unique among mammalian tissues in that it lacks a blood supply. This removes it from the minute-to-minute hormonal regulation available to other more vascularized tissues. In this environment, mechanical forces may play an important role in regulating cellular functions within tissues (3).
Under normal physiological conditions, in vivo joint loading can result in peak dynamic mechanical stresses on cartilage as high as 15-20 megapascals (150 -200 atmospheres) (4). These peak stresses occur over very short durations (Ͻ1 s) and therefore lead to cartilage compressive strains of only 1-3%. In contrast, sustained (static) physiological stresses of ϳ3.5 megapascals applied to cadaveric knee joints for 5-30-min durations resulted in compressive strains of various knee cartilages as high as 40 -45% (5). Several lines of evidence indicate that cartilage is responsive to these wide ranges of mechanical strain and stress (6 -13). In vivo, acute (14,15) and chronic (16) injurious compressive overloads can lead to cartilage degeneration. In vitro, static compression within the physiological range can reversibly inhibit the synthesis of critical components of the cartilage matrix. Such static compressive forces can down-regulate the gene expression and production of type II collagen, aggrecan core protein, and link protein (11,13,17). In contrast, cyclically applied hydrostatic pressure (18) and compressive strain (19) can stimulate aggrecan core protein and protein synthesis. Articular cartilage biosynthetic activity is also regulated by cell signaling molecules. Insulin-like growth factor-I (IGF-I) 1 is the predominant anabolic growth factor in synovial fluid (20) and stimulates the synthesis of both proteoglycans and collagen (21). Although static compression reduces cartilage IGF-I content (22), the short time course of inhibition of cartilage biosynthesis by static compression is not consistent with an effect that is mediated entirely by IGF-I or other soluble growth factors. Rather, it is thought that specific mechanotransduction mechanisms mediate the chondrocyte biosynthetic machinery under load (11,(22)(23)(24). This study seeks to characterize such signaling mechanisms, including identification of pathways, time courses of pathway coactivations, and possible differences between mechanical and biochemical signal transduction.
Mechanical forces are complex multicomponent stimuli. The relatively simple case of a ramp-and-hold static compression of cartilage can result in transient interstitial fluid expression, cell deformation (25), increased osmolarity (26), decreased extracellular pH (27), changes in fixed charge density (28), and altered transport of soluble factors within the tissue (22). Each of these physical phenomena may potentially act as a "mechano-ligand" activating or inhibiting one or more signaling pathways. Prior studies have often focused on the effects of one or a few of these components of compressive loading such as negative fixed charge density (26) or interstitial pH (22,29,30). Recent studies have also demonstrated that mechanical stretching of chondrocytes can increase nitric oxide production (31) and alter membrane transport phenomena, resulting in proliferative changes (32).
In many cell types, protein kinases are critical mediators of the cellular responses involved in the minute-to-minute regulation of tissue function. Little is known of the role of signaling by protein kinases in cartilage in response to load. The existing evidence for MAPK involvement in chondrocyte mechanotransduction is particularly scant. Interestingly, however, previous kinetic studies found that the chondrocytic response to static load is unlikely to be due to the intermediary action of cytokines or growth factors (22). These findings suggest that a mechanism of direct mechanical regulation of cartilage exists. At present, published data linking any MAPK pathway activation to mechanotransduction events in cartilage are limited to the response to fluid shear flow applied over plated chondrocyte monolayers using fluid velocities that are much higher than those known to occur during cartilage loading in vivo (33,34). Moreover, the temporal organization of simultaneously activated pathways has, to our knowledge, not been investigated. This is particularly true for the analysis of chondrocyte signaling under physiologically relevant loads applied to cells in situ within their native matrix. Such information would have potential relevance to the pre-pathological states leading up to the formation of degenerative joint disease. To gain further insight into the mechanisms of mechanotransduction in chondrocytes, we have studied the role of MAPK pathways in response to mechanical compression of chondrocytes within their native cartilage. We have demonstrated that activation of the ERK1/2, p38, and JNK pathways is mechano-dependent and that these pathways are differentially activated with respect to each other and with respect to IGF-I.

EXPERIMENTAL PROCEDURES
Materials-Articular cartilage was obtained from freshly harvested intact juvenile bovine femorotibial joints (Arena, Inc., Hopkinton, MA). The protease inhibitors leupeptin, pepstatin A, aprotinin, and phenylmethylsulfonyl fluoride were from Sigma. Phosphorylation state-specific ERK1/2, anti-p38 and anti-SEK1 polyclonal antibodies; phosphorylation state-independent anti-ERK1/2 and anti-p38 antibodies; and horseradish peroxidase-conjugated goat anti-rabbit antibody were from Cell Signaling Technology, Inc. (Beverly, MA). Protran TM nitrocellulose was from Schleicher & Schü ll. The enhanced chemiluminescence detection substrate (ECL) was from PerkinElmer Life Sciences. All other chemicals were from Sigma and were the purest grade available.
Cartilage Explant Preparation and Compression for Dose-response Studies-Disks of articular cartilage (3 ϫ 1 mm, diameter ϫ thickness) were harvested from the femoropatellar groove of newborn calves and were incubated in Dulbecco's modified Eagle's medium with 10 mM HEPES, 0.1 mM nonessential amino acids, 100 units/ml penicillin/ streptomycin, an additional 0.4 mM proline, and 20 g/ml ascorbate (basal medium) containing 10% fetal bovine serum (FBS) for 2 days. Disks were subjected to graded levels of unconfined uniaxial static mechanical compression in incubator-housed compression chambers containing fresh basal medium plus designated amounts of FBS. Compression was expressed as a percentage of the original cut thickness of the disks (1 mm). Loading conditions included no compression and 0 (held at 1 mm), 12, 25, 35, and 50% compression. In each experiment, 12 cartilage disks were simultaneously compressed within a loading chamber designated for each loading condition, unless noted otherwise. Following compression for the indicated time periods, the disks from within an individual compression chamber were pooled, rinsed with serum-free medium, blotted dry, and flash-frozen in liquid nitrogen. This termination procedure was repeated for each compression chamber.
To evaluate the effect of loading in the presence of low serum concentrations (2%) and in the absence of serum, experiments were performed as described above with the exception that cartilage disks were incubated in basal medium containing 2% FBS for 2 days following harvest and then segregated into two groups of 48 disks each. One group was cultured for an additional 24 h in fresh basal medium containing 2% FBS. The other was cultured in fresh basal medium containing 0.01% bovine serum albumin. All cartilage disks were then subjected to graded compressive loads (no compression and 0, 12, 25, 35, and 50% compression; eight disks per load condition) for 4 h in basal medium. Following compression, disks from within an individual compression chamber were pooled, rinsed with basal medium, and flashfrozen in liquid nitrogen. This termination procedure was repeated for each compression chamber.
Cartilage Explant Preparation and Compression for Time Course Studies-In short-term time course experiments, cartilage disks were harvested and incubated in basal medium containing 2% FBS as described above. Cartilage disks were maintained at no compression and at 0 or 50% of the original cut thickness for 10, 20, 40, or 60 min. In long-term time course experiments, cartilage disks were maintained at either 0 or 50% of the original cut thickness and compressed for 4, 8, 12, or 24 h (eight disks per time condition). Following compression, the disks from within each compression chamber were pooled, rinsed with serum-free medium, briefly blotted dry, and flash-frozen in liquid nitrogen. This termination procedure was repeated for each compression chamber.
IGF-I Treatment of Cartilage Explants-Cartilage was harvested as described above with the exception that explants were cut into disks of 3 ϫ 0.5 mm (diameter ϫ thickness) to facilitate IGF-I diffusive penetration into the full thickness of the disks. After incubation for 2 days in basal medium containing 2% FBS, the medium was removed, and all disks were serum-starved in basal medium for 24 h. The cartilage disks were then incubated in basal medium in the presence or absence of 300 ng/ml human recombinant IGF-I (PeproTech, Inc., Rocky Hill, NJ) for graded time periods (10, 20, and 40 min and 4, 12, and 24 h) under no compression. At the completion of incubation, disks comprising each individual IGF-I/time combination were pooled, rinsed with serum-free medium, blotted dry, and flash-frozen in liquid nitrogen.
Tissue Preparation and Immunoblotting-Cartilage disks were pulverized under liquid nitrogen using a Bessman tissue pulverizer (Fisher) and then homogenized using a Polytron device (Brinkmann Instruments) for 45 s in buffer (20 mM Tris (pH 7.6), 120 mM NaCl, 10 mM EDTA, 10% glycerol, 1% Nonidet P-40, 100 mM NaF, 10 mM Na 4 P 2 O 7 , 1 mM phenylmethylsulfonyl fluoride, 2 mM Na 3 VO 4 , 40 g/ml leupeptin, 1 M pepstatin A, and 10 g/ml aprotinin) at a ratio of 100 l/10 mg of tissue. Homogenates were extracted by end-over-end rotation for 1 h at 4°C and clarified by centrifugation at 13,000 ϫ g for 60 min. Supernatants were quantified for protein concentration using the BCA assay (Pierce). Aliquots containing 40 g of protein suspended in Laemmli buffer were resolved by SDS-PAGE (10% resolving gel); transferred to Protran nitrocellulose membranes; and blocked with 5% bovine serum albumin in 10 mM Tris (pH 7.6), 150 mM NaCl, and 0.1% Tween 20 (Tris-buffered saline/Tween) for 2 h at 37°C. Membranes were incubated with phosphorylation state-specific antibody (1:1000) or phosphorylation state-independent antibody (1:1000) overnight at 4°C, washed with Tris-buffered saline/Tween (3 ϫ 5 min), incubated with horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (1:2000) for 1 h at room temperature, and again washed with Trisbuffered saline/Tween (5 ϫ 10 min). For the ECL reaction, immunoblots were developed for 1 min in Western Lightning Chemiluminescence Reagent Plus (PerkinElmer Life Sciences). The respective phosphoproteins or total proteins were visualized using enhanced chemiluminescence on x-ray film (Blue Sensitive autoradiographic film, Marsh Biomedical Products, Inc., Rochester, NY) and then quantified using a computing densitometer (Amersham Biosciences). Band intensities were determined by calculating the sum of all pixel optical density values within selected bands minus the background (volume integration). Band intensity profiles are denoted on individual figures by use of the arbitrary term "Volume" and are meant to identify total optical density units of band intensity minus the appropriate background values. Note that specific phosphorylation state-independent antibodies were utilized in conjunction with the corresponding phosphorylation state-specific antibodies on immunoblots to demonstrate equivalency of protein loading between lanes for each phosphoprotein analyzed. In each case, protein loading was found to be equivalent, or minor differences were not correlative with phosphorylation trends between lanes.
Statistical Analysis-For strain dose dependence and time course experiments, the density of bands within given blots was normalized to that of the most common test condition between the various experiments of that type. Two-way ANOVAs of strain and time were then used to determine whether significant changes were present. In case of significance, two-tailed Student's t tests and one-way ANOVAs were performed to determine where the significance occurred. Dependent values created by normalization were removed. Statistical analyses were performed using the SYSTAT Version 9.0 statistical software package (SYSTAT Software Inc., Richmond, CA). Note that each figure depicts a representative immunoblot and its accompanying band intensity plot for each type of experiment performed. However, all available data from individual experiments were utilized in the generation of statistical analyses.

Mechanical Compression Activates the ERK1/2 MAPK Pathway in Bovine Cartilage
Explants-Cartilage explants were subjected to graded amounts of compression ranging from 0% (cut thickness) to 50%, and the activation state of the ERK1/2 signaling pathway was assessed by immunoblot analysis using phosphorylation state-specific anti-ERK1/2 antibodies. Both the ERK1 (p44) and ERK2 (p42) kinases were phosphorylated in a dose-dependent manner within the 0 -50% compression range (Fig. 1). This dose-dependent response was observed in n ϭ 10 separate experiments. The pattern and degree of loadinduced ERK1/2 activation were similar under the two serum conditions except at the maximal compression tested (50%). These data suggest that mechanical compression activates the ERK1/2 pathway and that this activation does not depend on the presence of exogenous serum growth factors. The maximal magnitudes of the increases in ERK phosphorylation were 2.2and 3.7-fold for ERK1 and ERK2, respectively, at 50% compression for 4 h in the absence of serum. The effect of compression was consistently greater for ERK2 than for ERK1. This difference was observed at the various time points and compression levels tested in all experiments. The total ERK1/2 protein levels were assessed as described under "Experimental Proce-dures" and were found to be equivalent among lanes (Fig. 1, gels, lower panels).
Effect of Duration of Compression on Compression-induced ERK1/2 Phosphorylation-To determine whether the dose-response pattern of compression-induced ERK1/2 activation (Fig.  1) is dependent on the duration of the compressive stimulus, further dose-response experiments were carried out for shorter (10 min and 1 h) and longer (24 h) time periods. Dose-response analyses were performed using phosphorylation state-specific ERK1/2 immunoblot assays as described above. We found a nearly linear dose-response relationship between ERK1/2 phosphorylation and the magnitude of static compression after 10 min of compression ( Fig. 2A). After a 1-h compression period, ERK1/2 phosphorylation showed a dose-response curve with a stronger plateau at the higher compressions (Fig. 2B). At 24 h, the dose-response pattern changed dramatically, with a significantly lower sensitivity to compression at the lower compressive strains (Fig. 2C). At all three time points, the maximal increase in ERK1 and ERK2 phosphorylation was observed at the highest magnitude of compression. Replicate experiments (n ϭ 10) showed similar ERK1/2 phosphorylation patterns.
Time Course of ERK1/2 Phosphorylation in Response to Mechanical Compression-Cartilage explants were subjected to no compression (NC) or 0 or 50% compression for 10, 20, 40, and 60 min and then subjected to phosphorylation state-specific ERK1/2 immunoblot analysis. In shorter time course experiments, the phospho-ERK1/2 levels in the 50% compressed samples were highest at the earliest time point tested (10 min) and then decreased over the next 50 min (Fig. 3A). Similar results were obtained in n ϭ three experiments. In longer time course experiments (n ϭ two experiments), cartilage explants were analyzed by phosphorylation state-specific ERK1/2 immunoblotting following compression times of 4, 8, 12, and 24 h (Fig.  3B). The phospho-ERK2 levels were highest at the earliest time point (4 h) and decreased with time in a bimodal fashion (Fig.  3B). In the 8-h interval between the first three time points of the experiment (4 -12 h), the phospho-ERK2 levels decreased by 82% of the total phospho-ERK2 decrease over the entire time interval of the experiment (4 -24 h). This period of rapid decay was followed by an ϳ12-h period of slower loss of phospho-ERK2 levels. Interestingly, at 50% compression, even after 24 h, the phospho-ERK2 levels remained elevated in comparison with samples held at 0% compression. These data indicate that one effect of static compression on cartilage explants is to produce a sustained ERK2 activation. As in the earlier doseresponse experiments (Figs. 1 and 2), ERK1 phosphorylation was increased to a lesser degree than ERK2 phosphorylation by prolonged static compression.
Time Course of ERK1/2 Phosphorylation in Response to IGF-I-To compare mechanically induced ERK1/2 activation with activation by a known biochemical inducer of the ERK pathway, we investigated the time course of ERK1/2 phosphorylation in response to addition of IGF-I. Cartilage explants were harvested as described above and incubated in the absence or presence of 300 ng/ml IGF-I for the same time periods used for ERK1/2 activation in mechanically loaded explants. This concentration of IGF-I was chosen on the basis of previously reported dose-response studies demonstrating that, under the conditions of these experiments, 300 ng/ml is reproducibly on the upper plateau of the dose-response curve for IGF-I stimulation of [ 3 H]proline and [ 35 S]sulfate incorporation, indices of protein and glycosaminoglycan synthesis, respectively (22,35). IGF-I produced an initial rapid increase in ERK1/2 phosphorylation, followed by a rapid decrease to the phospho-ERK1/2 levels present in untreated explants (Fig. 3C). This Shown is the dose-dependent phosphorylation of ERK1 and ERK2 in intact cartilage disks in response to increasing levels of static compressive strain. Cartilage explants were compressed and held for 4 h using increments of compression (0 -50%) based on the original cut thickness of the specimens. Compression was performed in the absence (gels, left panels; graph, pERK1/Ϫ and pERK2/Ϫ) and presence (gels, right panels; graph, pERK1/ϩ and pERK2/ϩ) of 2% fetal bovine serum. Following compression, explants were processed as described under "Experimental Procedures" and analyzed for ERK1/2 phosphorylation using a phosphorylation state-specific anti-ERK1/2 antibody (gels, upper panels) and compared with the expression of total ERK1/2 (gels, lower panels). NC indicates no compression, and 0% indicates holding the cartilage disk thickness at the original cut thickness of 1 mm. Note that total ERK1/2 protein loading was approximately equivalent among lanes (gels, lower panels). Each data point (band) corresponds to 12 individually compressed and pooled cartilage plugs. ANOVA of the normalized immunoblot results (111 independent bands from n ϭ 10 replicate experiments) revealed a significant change in phospho-ERK1 (pERK1) and phospho-pERK2 (pERK2) activation as a result of graded compressions (p Ͻ 0.001 for both). pattern of transient ERK1/2 activation is in contrast to the sustained activation produced by compressive stress (Fig. 3B). Similar results were obtained in n ϭ two experiments.
Effect of Compression on p38 MAPK Phosphorylation-The p38 MAPK pathway is an alternative to the ERK1/2 MAPK pathway in the response mechanism used by many cell types to adapt to environmental changes. To determine whether the p38 pathway is involved in mechanical regulation of cartilage, explants were subjected to graded levels of compression from 0 to 50%, and tissue extracts were prepared as described under "Experimental Procedures." The activation state of p38 was assessed by immunoblot analysis using a phosphorylation state-specific anti-p38 antibody. After a 10-min loading period, compression produced a "dose-dependent" increase in phosphorylated p38, with a maximal stimulation at 50% compression, the highest magnitude of compression tested. Maximal stimulation was ϳ5-fold compared with non-compressed samples. When analyzed 1 h after the onset of loading, the phosphorylated p38 MAPK levels remained close to base line, with maximal levels ϳ4-fold lower than those observed at 10 min and with a marginal dose-response (Fig. 4). Similar trends were observed in n ϭ five experiments.
Time Course Responsiveness of p38 MAPK Phosphorylation-Cartilage explants were statically compressed to either 0 or 50% of the original cut thickness for graded time periods, and phospho-p38 was assessed by immunoblot analysis. At 0% compression, the phospho-p38 levels remained low over the entire duration of the short-term (10 -60 min) (Fig. 5A) and long-term (20 min to 24 h) (Fig. 5B) experiments. The data of Figs. 4 and 5 (A and B) collectively indicate that, at 50% compression, maximal p38 phosphorylation occurred between 10 min and 4 h (Fig. 5, A and B), followed by a decline in phospho-p38 to levels close to those at 0% compression by 24 h (Fig. 5B). These findings suggest that mechanical compression activates the p38 MAPK pathway in a transient fashion.
Effect of Compression on SEK1 Phosphorylation-JNKs, together with the p38 kinases, constitute the SAPK subfamily of MAPKs (36). These kinases are often activated in response to events involving cellular adaptation to environmental changes. Such changes include immune challenge, transformation, apoptosis, and response to pro-inflammatory cytokines. To determine whether the JNK pathway is similarly activated by static compressive loading, cartilage explants were subjected to graded levels of compression for 10 or 60 min and then analyzed by immunoblotting using a phosphorylation state-specific anti-SEK1/MKK4 antibody as described under "Experimental Procedures." SEK1/MKK4 was selected as the member of this pathway for analysis because it is an immediate upstream specific activator of JNK1, -2, and -3 (37). Direct analysis of JNK phosphorylation was found to be unreliable because the available phosphorylation state-specific anti-JNK antibodies appeared not to recognize bovine JNK family members. Mechanical compression resulted in a dose-dependent increase in SEK1 phosphorylation (Fig. 6). The time course of SEK1 phosphorylation differed from that of ERK1/2 activation (Fig. 2, A  and B) and p38 activation (Fig. 4). SEK1 phosphorylation was relatively low at 10 min and much higher by 1 h. DISCUSSION Several lines of research have demonstrated that mechanical forces regulate cartilage metabolism (6 -13). Little is currently known, however, about the pathways used by chondrocytes to

FIG. 2. ERK1/2 phosphorylation versus compression after increasing durations of compression.
Cartilage explants were subjected to graded degrees (0 -50%) of static compression and analyzed for ERK1/2 phosphorylation at 10 min (A), 1 h (B), and 24 h (C). ERK1 (pERK1) and ERK2 (pERK2) phosphorylation was analyzed using a phosphorylation state-specific antibody (gels, upper panels), and loading of equivalent amounts of total ERK1/2 protein was monitored by immunoblot analysis using a phosphorylation state-independent antibody (gels, lower panels). Note the equivalency of total ERK1/2 protein loading among lanes (gels, lower panels). ANOVA of the normalized Western results (111 independent bands from n ϭ 10 replicate experiments) showed a significant change in phospho-ERK1 and phospho-ERK2 activities as a function of compression duration (p Ͻ 0.001 for both). NC, no compression. transduce mechanical signals. This work was undertaken to determine whether elements of the MAPK signal transduction pathways are regulated by mechanical forces in chondrocytes.
This study was designed to investigate the mechanism of mechanical signal transduction in articular chondrocytes in situ within their native matrix within a physiological range of com- Cartilage explants were compressed to 50% of their original cut thickness and held for timed increments of 10 -60 min (A) and 4 -24 h (B) and compared with cartilage disks exposed to maximal concentrations (300 ng/ml) of IGF-I for 10 min to 24 h (C). Samples were harvested and analyzed for ERK1/2 phosphorylation on immunoblots using a phosphorylation state-specific anti-ERK1/2 antibody (A-C, gels, upper panels), and loading of equivalent amounts of total ERK1/2 protein was assessed (A and C, gels, lower panels). Note the equivalency of total ERK1/2 protein loading among lanes (A and C, gels, lower panels). Note that, in C (graph), the following designations are used: phospho-ERK1 (pERK1) without IGF-I (E), phospho-ERK2 (pERK2) without IGF-I (ᮀ), phospho-ERK1 with IGF-I (F), and phospho-ERK2 with IGF-I (s). NC, no compression. pressive strains (5). Our data provide direct evidence that the ERK1/2, p38, and JNK (SEK1) pathways are activated by static compressive loading of cartilage.
This study shows that static mechanical compression stimulates phosphorylation of both ERK1 and ERK2 and that this occurs in a strain dose-dependent manner (Fig. 1). The compression conditions that induced phosphorylation of ERK1/2 in this study have been reported previously to cause a dose-dependent inhibition of chondrocyte proteoglycan synthesis and protein synthesis (11,13,17,22). It was somewhat surprising to find that ERK1/2 is activated by static compression, an inhibitor of chondrocyte anabolism, because the ERK1/2 pathway is often linked to growth factor action and is usually associated with cellular proliferation. Indeed, our data demonstrate that, in chondrocytes in situ within their native matrix, IGF-I also activates ERK1/2. Previous studies examining the effect of IGF-I on bovine calf cartilage under the culture conditions employed in the present study showed that IGF-I is anabolic for this tissue (22,35).
Static compression maximally phosphorylated ERK1 and ERK2 at 10 min, the earliest time point tested (Fig. 3A). These findings indicate that activation of ERK1/2 in chondrocytes in native cartilage is a rapid response mechanism to mechanical stimulation. The time course data also suggest that the ERK1/2 signal is a relatively proximal component of mechanoreceptor signaling in chondrocytes. Two recent studies have also linked the activation of the ERK1/2 pathway to fluid shear flow, an early consequence of tissue pressurization (33,34).
This work further demonstrates that static compression is capable of maintaining a sustained level of ERK2 activation even up to 24 h (Fig. 2). There are therefore two kinetically separate signaling events that take place following static compression in chondrocytes. The initial ERK2 phosphorylation is of relatively high magnitude and short duration and is followed by a more prolonged plateau of sustained phosphorylation at a lower magnitude. Such activation kinetics have not, to our knowledge, been observed before in chondrocytes. These distinct activation phases may have distinct effects on cellular pathways. The importance of such time-dependent effects has been demonstrated in other cell types. For example, in rat pheochromocytoma (PC12) cells, the decision to divide or differentiate in response to epidermal growth factor versus nerve growth factor is thought to depend upon the respective transient or sustained ERK activation (38 -40). Similarly, in CCL39 hamster lung fibroblasts, biphasic (i.e. transient and sustained) ERK activation induced by thrombin results in progression to mitosis, whereas transient ERK activation (ϳ10 min) induced by serotonin does not (41). These data suggest that the mechanism by which MAPK pathways modulate the cellular response to mechanical signals is based, in part, on the temporal architecture of MAPK activation. It is also possible that the differential effects of static compression and IGF-I on chondrocyte biosynthesis are mediated in part by the differential response pattern of ERK1/2 activation in these cells.
Articular cartilage in vivo is exposed to both biomechanical and soluble biochemical factors (e.g. growth factors). The addi- FIG. 4. p38 pathway activation in response to mechanical compression. Samples were subjected to increasing levels of compression (0 -50%), followed by immunoblot analysis using a phosphorylation state-specific anti-p38 (pp38) antibody. Compression series were carried out at either 10 min (gel, left panel; graph, q) or 1 h (gel, right panel; graph, f). The samples used in this experiment were identical to those used in Fig.  2 (A and B). See Fig. 2 (A and B, gels, lower panels) for protein loading equivalency. Note that the higher than expected band intensity plotted for the 0% compression point at 10 min (q) is likely a result of a nonspecific speck, inherent to film exposure, and can be seen in the corresponding lane in the gel above. ANOVA of the normalized p38 data (23 independent bands from n ϭ five replicate experiments) showed that p38 activity varied significantly as a function of strain rate (p Ͻ 0.01). NC, no compression. tive or synergistic action of these distinct classes of stimuli has been shown to regulate cartilage by mechanisms that are not simply the result of mechanical regulation of growth factor transport. Static compression has been reported to inhibit matrix biosynthesis within 1-2 h of load application in bovine articular cartilage (13,22). In contrast, IGF-I does not increase matrix biosynthesis until after 12 h in chondrocytes within intact cartilage (22) or in chondrocytes isolated from their matrix and not subjected to IGF-I diffusion through cartilage matrix (42). In this study, the IGF-I response contrasted with that of mechanical compression, where IGF-I induced an early and transient activation of ERK1/2 with no sustained activation (Fig. 3C). Based on the measured diffusivity for IGF-I in bovine cartilage (D ϳ 10 Ϫ7 cm 2 /s) (43), we estimate that ϳ50% of the cells in the disk would be exposed to IGF-I within a characteristic diffusion time constant of 10 min. Thus, although diffusion of IGF-I is not as fast as the application of mechanical stimulation, we do not believe that diffusion is severely rate-limiting in the interpretation of activation kinetics for times Ͼ10 min. Further studies will be required to define the specific relationship, if any, between ERK1/2 activation and downstream chondrocyte biosynthetic activity for both IGF-I and mechanical compression.
The early time course of p38 pathway activation was initially similar to that of ERK1/2 pathway activation, with peak phosphorylation occurring at the earliest time points tested (10 -60 min) (Figs. 4 and 5). This suggests that, like ERK1/2, the p38 signal is one of the more proximal signaling events in chondrocyte mechanotransduction. However, unlike the ERK1/2 activation time course, p38 activation is followed by a rapid signal decay phase and does not possess the sustained activation observed with ERK2.
There are four known isoforms (␣, ␤, ␥, and ␦) that make up the p38 subfamily of MAPKs. It is not currently known whether FIG. 5. Time course analysis for p38 activation response to compression. Cartilage explants were held to 0% (graphs, q) or compressed to maximal levels (50%) (graphs, f) and held for between 10 and 60 min (A) or between 20 min and 24 h (B). Following compressions, samples were subjected to immunoblot analysis using a phosphorylation state-specific anti-p38 (pp38) antibody (gels). The samples used in A were identical to those used in Fig. 3A, and samples used in B were identical to some of those used in Fig. 3B. See Fig. 3A (gels, lower panel) for protein loading equivalency. ANOVA of the p38 activation data (23 independent bands from n ϭ five replicate experiments) revealed a trend as a function of strain duration, but this was not statistically significant (p ϭ 0.087). NC, no compression. the phosphorylation state-specific anti-p38 antibody used in this study detects all four of these isoforms, and available evidence indicates that some of the isoforms comigrate upon 10% SDS-PAGE. The phosphorylation state-specific anti-p38 antibody does not cross-react with either ERK1/2 or JNKs. Thus, these results may reflect the activation of one or more p38 MAPK isoforms.
Little is known of the functions of the SAPK pathways in chondrocytes. Our data demonstrate that mechanical stimulation activates SEK1, one of the central components of the JNK pathway core signaling module. Previously, it has been shown that pro-inflammatory cytokines such as interleukin-1 activate the p38 and JNK pathways (SAPKs) (44) and later the ERK1/2 pathway (45) in articular chondrocytes. Interestingly, the effects of interleukin-1 and static mechanical loading on cartilage anabolism are similar (11,12) in that both reduce proteoglycan synthesis by chondrocytes (46 -48). Taken together, these data raise the possibility that the mechanism by which static loading and interleukin-1 inhibit matrix biosynthesis involves SAPKs as common elements in their signal transduction pathways. A potential role of SAPKs in the regulation of chondrocytes by compression is consistent with the fact that SAPKs are the major regulators of adaptation to environmental stresses in several other cell types (36).
Activation of the JNK pathway in chondrocytes demonstrates another concerted MAPK event in response to static compressive loading, but it displays a time course distinct from that for both ERK1/2 and p38. SEK1 phosphorylation lagged behind its ERK1/2 and p38 counterparts, with maximal activation 1 h after the onset of compression compared with 10 min for ERK1/2 and p38 (Fig. 6). This relative delay in SEK1 phosphorylation, coupled with the fact that SEK1 phosphorylation is the step prior to JNK activation, suggests that JNK activation may be a relatively temporally distal component of the chondrocytic mechanotransduction machinery.
Such multiple pathway activation may help specify the type of response to static compression. This is consistent with the emerging view of MAPK signaling as a system of modules within a signaling network that functions as an integrative whole (49,50). Such a model may be particularly applicable to the structurally related MAPKs.
To further interpret the kinetics of activation of the ERK1/2, p38, and JNK (SEK1) pathways in response to compression, we can compare the results of this study with the kinetics of the intratissue mechanical forces and flows caused by compression. As described previously (13,23), application of a ramp-andhold compression to the 3-mm cartilage diameter disks causes an initial transient intratissue pressurization and fluid flow within the matrix immediately following compression and during a 5-10-min period of stress relaxation. After stress relaxation has ended, fluid flow ceases, and intratissue pressure returns to zero (i.e. that of the medium) in the final equilibrium compressed state of the tissue. It is interesting to speculate that the rapid activation of ERK1/2 and p38 may be due to cell deformation, fluid flow, pressurization, or their combined effects during the initial mechanical transient, in which these physical signals are reminiscent of the mechanical components of low frequency "dynamic" compression (13). In contrast, the delayed activation of SEK1 observed by 1 h after compression occurs at a time at which there is only static compression, with no fluid flow or pressurization. Alternatively, the rapid activation and subsequent decrease in ERK1/2 and p38 may correspond to desensitization or accommodation of the cells that may occur over time in response to a sustained static compression. Further experiments are needed to determine whether one or more of these specific mechanical signals are related to the observed kinetics of activation.
We have studied this MAPK pathway activation with activation state-specific antibodies raised against the dually phosphorylated MAPK subtypes, ERK1/2 and p38. Dual phosphorylation of both the Thr and Tyr residues in the conserved -TXY-motif of MAPK family members is the only known activation mechanism for the ERK1/2, p38, and JNK proteins (51). This approach permits the direct visualization of regulatory phosphorylation events that have occurred in chondrocytes within intact cartilage matrix and does not rely on interpretation of phosphorylation events that may occur in kinase assays in vitro using heterologous substrates such as myelin basic protein. FIG. 6. SEK1 phosphorylation in response to cartilage compression. Cartilage disks were subjected to various degrees of compressive strain at either 10 min (gel, left panel; graph, q) or 1 h (gel, right panel; graph, f), followed by Western blot analysis of JNK pathway activation using a phosphorylation state-specific anti-SEK1 (pSEK1) antibody. The samples used in this experiment were identical to those used in Fig. 2 (A and B). See Fig. 2 (A and B, gels, lower panels) for equivalency of protein loading. NC, no compression.
A component of mechanically induced signal pathway activation in chondrocytes may be indirect, involving mechano-dependent secretion of soluble intermediates such as interleukins (52). Such a mechanism may contribute to the sustained phase of ERK2 activation that is reported here. Studies aimed at the synergy between mechanical and biochemical actions on cartilage metabolism have just recently begun to address the issue of signal integration in articular cartilage. Further studies will be required to elucidate the cooperative pathways that regulate chondrocyte function.