Src and fibroblast growth factor 2 independently regulate signaling and gene expression induced by experimental injury to intact articular cartilage




To investigate whether cartilage injury activates protein tyrosine kinases distinct from fibroblast growth factor (FGF)–related signaling, and whether they contribute to injury-induced gene responses.


Phosphokinases and protein tyrosine phosphorylation were assayed by Western blotting of cartilage lysates. Immunoprecipitation and Western blotting with 4G10 antibody and immunoprecipitation kinase assay were carried out. Tyrosine-phosphorylated proteins on silver-stained gels of injured cartilage lysates were identified by mass spectrometry. Messenger RNA induction in cartilage explants was assessed by quantitative reverse transcriptase–polymerase chain reaction.


Protein tyrosine phosphorylation occurred within seconds of injury to the surface of intact articular cartilage, as did activation of MAPKs and IKK. Activation did not reoccur upon reinjury of cultured explants. The prominent tyrosine-phosphorylated proteins focal adhesion kinase, paxillin, and cortactin were identified as substrates of Src family kinases. The Src family kinase inhibitor PP2 blocked injury-induced tyrosine phosphorylation. It did not prevent activation of the MAPKs and IKK but differentially inhibited 8 of 10 inflammatory response genes that were induced by injury. In contrast, FGF signaling blockade with PD173074 reduced all MAPK and IKK activation by ∼50% and inhibited a different subset of genes but had no effect on Src-like signaling.


Injury to the surface of intact articular cartilage activates Src-like kinases as well as MAPKs and IKK (implying NF-κB activation). FGF-2 contributes to MAPK/IKK activation but not to Src-like signaling, suggesting that the latter is a parallel pathway that also regulates the injury-induced inflammatory gene response.

Articular cartilage has little or no capacity for repair following injury (1). Damaged cartilage is prone to degeneration. For example, scoring of the articular surface of rabbit knees leads, within weeks, to cartilage degeneration similar to that observed in osteoarthritis (OA) (2). In humans, joint injury, either directly to the articular surface or via destabilizing ligamentous or meniscal tears, is a risk factor for the development of OA (3). It is likely that joint tissue microinjury, secondary to congenital or acquired biomechanical factors, accounts for further and perhaps a larger proportion of cases of progressive OA. In an attempt to understand the response of chondrocytes to cartilage injury, several groups of investigators have studied the effects of high or repeated impact loading, strain, or cutting of cultured cartilage or chondrocyte monolayers (4–6).

Our group has studied the effects of either dissecting cartilage from the intact articular surface or scoring the surface, because these are simple, reproducible modes of experimental injury that can be delivered rapidly to amounts of tissue sufficient for biochemical investigation. Within minutes, this type of sharp injury to the intact articular surface activated the 3 MAPKs, namely, JNK, p38 MAPK, and ERK (7, 8). Some degradation of IκBα was also apparent, implying possible activation of NF-κB. The injured cartilage rapidly released a factor that strongly activated ERK in cultured chondrocytes and was identified as basic fibroblast growth factor 2 (FGF-2), which was stored in the pericellular matrix (7, 9, 10). Although FGF-2, which signals via a receptor tyrosine kinase (predominantly FGF receptor 1 [FGFR-1]) accounted in part for ERK activation seen upon injury, this did not explain the strong JNK activation or activation of NF-κB. This pattern of signaling pathway activation is typically associated with inflammatory stimuli such as interleukin-1 (IL-1), tumor necrosis factor α (TNFα), or microbial products acting via Toll-like receptors.

The mechanism by which sterile tissue injury activates inflammatory signaling is obscure. Soluble ligands such as heat-shock proteins and high mobility group proteins released by injured cells can directly activate the pathways (11). However, no soluble JNK- or NF-κB–activating factor released from cartilage upon injury has been reported.

We previously observed that one response of freshly dissected cartilage was to produce activin A, a cytokine related to transforming growth factor β. We used an inhibitor of the FGFR tyrosine kinase (PD173074, also known as SB402451) to show that this action was FGF dependent (7, 12, 13). The effect of the Src family protein tyrosine kinase inhibitor PP2 on induction of activin A was also tested, because Src family kinases are commonly activated by growth factor receptor tyrosine kinases such as FGFR (12, 14). PP2 also inhibited activin A production. We therefore examined whether or not cartilage injury activated protein tyrosine kinase activity distinct from that activated by FGFR, and, if there was Src activation, whether it was downstream of FGF-2. Here, we show that cartilage injury rapidly stimulates Src-like protein kinase activity independently of FGF-2, and that this activity contributes to the overall gene response to injury. We also show that the injury activates IKK within seconds, and that FGF serves to amplify the MAPK and IKK signals.



Porcine articular cartilage was dissected from the metacarpophalangeal (MCP) joints of freshly slaughtered 3–6-month-old pigs obtained from a local abattoir.


The general laboratory reagents used were the best grade available from either Sigma-Aldrich or BDH, unless stated otherwise. Dulbecco's modified Eagle's medium (DMEM; BioWhittaker) was supplemented with 25 mM HEPES. EasyTides γ32P-ATP was obtained from Perkin-Elmer.

Human FGF-2 was obtained from Peprotech. Glutathione S-transferase (GST)–IκBα and recombinant human IL-1α were obtained in-house. Rabbit polyclonal antibodies to phosphorylated activating transcription factor 2 (ATF-2), JNK, p38 MAPK, and paxillin were purchased from Cell Signaling Technology. Mouse monoclonal antibodies to IKKγ/α were obtained from BD PharMingen. Mouse anti-phosphotyrosine antibody 4G10 conjugated to agarose beads (4G10 beads), horseradish peroxidase (HRP)–conjugated 4G10 antibody (4G10-HRP antibody), rabbit polyclonal antibodies to phosphorylated Tyr576 on focal adhesion kinase (FAK), and mouse monoclonal antibody to cortactin were from Upstate Biotechnology. The murine monoclonal antibody to phosphorylated ERK was from Sigma, and rabbit polyclonal antibody to ERK was obtained from Santa Cruz Biotechnology. HRP-linked anti-mouse or anti-rabbit IgG was obtained from Dako Cytomation. An electrochemiluminescence Western blot detection system was purchased from Amersham Biosciences. The FGFR tyrosine kinase inhibitor PD173074 was provided by Stephen Skaper (GlaxoSmithKline). PP2 was from Calbiochem.

Preparation and culture of cartilage explants and porcine primary chondrocytes.

Cartilage was dissected into DMEM containing 25 mM HEPES buffer and washed. The cartilage explants were cut into small pieces (∼3–4 mm3) and cultured as indicated for up to 48 hours. Primary chondrocyte monolayers were prepared as previously described (7).

Western blotting.

Cartilage specimens were snap-frozen and kept at −80°C prior to lysis with radioimmunoprecipitation assay (RIPA) buffer, sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), and Western blotting (12).

Immunoprecipitation kinase assay for IKK.

Cartilage lysates were incubated in a nondissociative buffer overnight with IKKγ antibody, prior to assay of IKK activity on GST–IκBα(1–54) (15). Autoradiography of proteins and quantification of substrate phosphorylation were performed with a Fujifilm FLA-5100 phosphorimager. Data were analyzed using advanced image data analysis software (AIDA version 4.10; Raytest).

Immunoprecipitation and purification of tyrosine-phosphorylated proteins.

A previously performed protocol was adapted (16) and carried out at 4°C. Clarified RIPA lysate (600 μl) was incubated for 30 minutes with protein A–agarose beads in phosphate buffered saline and then centrifuged briefly. The supernatant was incubated with 4G10 beads overnight. Beads were then washed 5 times before incubation with 20 μl of 0.2M phenyl phosphate in RIPA. The phenyl phosphate–bead slurry was transferred to a 0.2-μm cellulose acetate spin column (Spin-X; Corning Costar). Eluate was either concentrated with an SDS-PAGE Clean-Up Kit (GE Healthcare) or boiled with SDS sample buffer and sonicated prior to electrophoresis.

Mass spectrometry.

Silver-stained protein bands (17, 18) were cut from gels, digested with trypsin, prepared, and analyzed (19, 20). Tandem mass spectra were recorded with a quadrupole/orthogonal acceleration time-of-flight mass spectrometer (Waters) interfaced to a Waters CapLC capillary chromatograph.

RNA isolation and quantitative reverse transcriptase–polymerase chain reaction (RT-PCR).

Total RNA and complementary DNA were prepared from cartilage (12). Amplification by real-time PCR was performed with SYBR Green PCR Master Mix (Applied Biosystems/Life Technologies) and specific primers (MWG Biotech). A RotorGene 6000 thermocycler and RotorGene 6000 software were from Corbett Research. Expression levels of the genes of interest were normalized to GAPDH. The melting curve was used to verify products.

The primers used were as follows: for GAPDH, sense 5′-CATGGAGAAGGCTGGGGCTC-3′, antisense 5′-ATGAGGTCCACCACCCTGTT-3′; for IL-1α, sense 5′-GAGGCAGTGAAATTTGACATGG-3′, antisense 5′-GGCAATGAACAACTTTGGATGGG-3′; for IL-6, sense 5′-ATGCTTCCAATCTGGGTTCAA-3′, antisense 5′-CACAAGACCGGTGGTGATTCT-3′; for IL-8, sense 5′-GGCTCCCAAGAATTTCTCAGTA-3′, antisense 5′-CAGCAGCCTAGGGTTGCAAG-3′; for monocyte chemotactic protein 1 (MCP-1), sense 5′-TGCCCAGCCAGATGCAATTA-3′, antisense 5′-GACCCACTTCTGCTTGGGTT-3′; for CCL5, sense 5′-CCAGCAGCAAGTGCTCCAT-3′, antisense 5′-ACACCTGGCGGTTCTTTCTG-3′; for inhibin βA (21), sense 5′-GACATCCGGACTGCCTGCGAGCAG-3′, antisense 5′-GTAGCCGGACGGAGCGATTAGCCAGTC-3′; for serum amyloid A2 (SAA2), sense 5′-AGAGCCTACTCGGACATGAGAGA-3′, antisense 5′-CCCCGGGCATGGAAGTAC-3′; for cyclooxygenase 2 (COX-2), 5′-CCGACAGCCAAAGACACTCA-3′, antisense 5′-CGGAGGTGTTCAGGAGTGTGA-3′; for inducible nitric oxide synthase (iNOS), sense 5′-TTCGAGCACATCTGCAGG-3′, antisense 5′-GTCAGGAGGGATTTCGAA-3′; for hyaluronan synthase 1 (HAS-1), sense 5′-CTGGGTGGCCTTCAACGT-3′, antisense 5′-GATGCAGGACACACAGTGGAA-3′; for matrix metalloproteinase 1 (MMP-1) (22), sense 5′-CCGTTTGTTCTTACTCCAGGGAA-3′, antisense 5′-GGGTACATCAGAGCCCCAATGTC-3′; for MMP-3 (23), sense 5′-GGAGTTCCTGATGTTGGTTACTTC-3′, antisense 5′-CAAAACTTTTCCAGGTCCGTCAAA-3′; for MMP-13, sense 5′-GTGACAGGCAGACTTGATGA-3′, antisense 5′-AAGCGTGAGCCAACAGACCA-3′; for ADAMTS-4, sense 5′-ACACGCCTCCGATACAGCTT-3′, antisense 5′-GTAGAACGTGGCGTTGAAGGA-3′; for ADAMTS-5, sense 5′-CGGCTTCTTCGCAGTCAAG-3′, antisense 5′-CCAGCTCAGCAGCGGTTT-3′.

Statistical analysis.

One-way analysis of variance was used to compare ≥3 sets of samples. P values less than 0.05 were considered significant. Data were analyzed using GraphPad Prism and Microsoft Excel.


Rapid induction of cell signaling, including protein tyrosine phosphorylation, by cartilage injury.

The MCP joints of porcine trotters were opened and equilibrated for 1 hour at 37°. Cartilage was then swiftly dissected from the surface into serum-free DMEM and kept for increasing periods of time prior to snap-freezing, lysis, and Western blotting with 4G10-HRP antibody. Several tyrosine phosphorylation bands were induced (Figure 1A). Prominent among these bands were one at 170 kd, a complex in the 100–120-kd region, a doublet at 80 kd, and one at 60 kd. The lanes were equally loaded, as judged by staining of ERK protein (Figure 1A). In a similar dissection experiment, lysates were incubated with 4G10 antibody, and the immunoprecipitates were incubated with phenyl phosphate, which selectively elutes phosphorylated proteins. SDS-PAGE (4–12% gradient gel) of the eluates and Western blotting with 4G10-HRP antibody was carried out. The same major bands were seen and were visible within 15 seconds of injury (Figure 1B). The induction of tyrosine phosphorylation was independent of immersion of the tissue in culture medium, because scoring of the articular surface induced the same bands (data not shown). Protein tyrosine phosphorylation was maximal by 10 minutes (Figure 1A), and this time point was chosen for further studies.

Figure 1.

Activation of intracellular signaling pathways following dissection of articular cartilage. A–E, Porcine metacarpophalangeal joints were opened and equilibrated at 37°C prior to dissection (diss) of cartilage (∼0.2 gm) from the articular surface directly onto dry ice (time 0) or into serum-free Dulbecco's modified Eagle's medium (DMEM) for the times indicated prior to snap-freezing and lysis. F, Cartilage explants (0.5 gm) were rested for 48 hours in serum- free DMEM and then either lysed immediately (time 0) or treated (re-cut into 2 × 2–mm squares, incubated with medium conditioned with freshly re-cut explants [CM], 50 ng/ml fibroblast growth factor [FGF], or 100 ng/ml interleukin-1α), and cultured for 30 minutes and then lysed. A, Lysates subjected to 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and Western blotting (WB) with 4G10 antibody. B, Lysates incubated with 4G10 beads overnight prior to elution, electrophoresis (NuPAGE Bis-Tris discontinuous 4–12% gel), and Western blotting with horseradish peroxidase–conjugated 4G10 antibody (1:20,000). Lysates were separately Western blotted for ERK to ensure equal amounts of cellular protein. C and D, Lysates subjected to Western blotting for phospho-JNK, phosphorylated activating transcription factor 2 (phospho–ATF-2), phospho–p38 MAPK, phospho-ERK, and ERK. E and F, IKK was immunoprecipitated (IP) from lysates, and IKK activity was assayed on glutathione S-transferase (GST)–IκBα. Autoradiography of PVDF membranes is shown. Membranes were also blotted for IKKα to ensure equal immunoprecipitation of IKK complex. Measurement of IKK activity (integral) from 0 point was normalized to 100%, and activity in the other samples was expressed as a percentage of this.

Using kinase assay, we previously demonstrated that JNK was activated 10 minutes following dissection of cartilage (8). Although phospho-JNK was visualized by Western blotting of cartilage lysates, extracts from large amounts of tissue were required (Figure 1C). We therefore routinely used antibodies to the phosphorylated form of transcription factor ATF-2 to examine JNK activation. ATF-2 is a well-known substrate of JNK (24) and was easily visualized in small amounts of extract. In some cells, ATF-2 has been reported to be a substrate for both p38 MAPK and JNK (25), but it was a substrate only for the latter in cartilage, because the increase in phospho–ATF-2 in IL-1–stimulated cartilage was not inhibited by a p38 MAPK inhibitor (data not shown).

In an experiment similar to that shown in Figure 1A, Western blotting of cartilage lysates showed phosphorylation of ATF-2, p38 MAPK, and ERK (Figure 1D). An increase in phosphorylation was evident by 1 minute and was maximal by 5 minutes. To investigate whether the canonical NF-κB pathway was also activated by injury, the IKK complex was immunoprecipitated from the lysates, and its activity was assayed on GST–IκBα. Increased activity was seen within 30 seconds of dissection of cartilage from the articular surface and was maximal by 5 minutes (Figure 1E). Activity was similarly induced if the articular surface was simply scored (data not shown).

To investigate whether cutting cartilage adapted to culture caused activation of protein kinases, explants were rested for 48 hours in serum-free DMEM and then either re-cut into small squares and left for 30 minutes or treated with IL-1 for 30 minutes. Although IL-1 activated IKK in the explants as expected, re-cutting the cartilage did not (Figure 1F). Re-cutting the cartilage rested in culture also failed to activate JNK or induce the tyrosine phosphorylation pattern (data not shown).

We questioned whether the damaged fresh cartilage released any soluble mediator with IL-1–like activity that would explain the pattern of kinase activation seen upon injury. DMEM was conditioned by rapidly dicing freshly dissected cartilage in it for 30 minutes. This medium failed to activate IKK in the rested explants (Figure 1F) or to activate JNK or induce the tyrosine phosphorylation pattern (data not shown). We previously observed that re-cutting rested cartilage released FGF-2 from the pericellular pool, and that this activated ERK (and to a lesser extent p38 MAPK) in the explants. Taken together, these findings suggested that 1) neither FGF-2 nor any other soluble factor released by cartilage upon injury appeared to be responsible for the tyrosine phosphorylation pattern, activation of IKK, and activation of JNK and 2) injury to the intact articular surface was necessary for activation of these intracellular signaling pathways, because this response could not be induced by repeating the insult in vitro.

Identification of injury-induced tyrosine-phosphorylated proteins.

To identify the tyrosine-phosphorylated proteins, immunoprecipitation with 4G10 antibody was scaled up. This enabled sufficient amounts of the proteins of interest to be seen on a silver-stained gel and identified by mass spectrometry. Smaller amounts of sample on the same gel were subjected to Western blotting with 4G10 antibody to allow alignment of tyrosine-phosphorylated and silver-stained bands (Figure 2A). The 2 upper bands of the 100–120-kd cluster were identified as the tyrosine kinase focal adhesion kinase (FAK). Three peptides likely to be from FAK were identified, the second of which had the highest probability of being from FAK alone (Figure 2B). A band on silver-stained gel that aligned with the regulated 60-kd band on Western blotting was identified by mass spectrometry as paxillin. Both FAK and paxillin are known to be tyrosine phosphorylated, and both were observed only in the lanes showing results for cartilage 10 minutes after dissection (Figure 2A).

Figure 2.

Identification and verification of tyrosine-phosphorylated proteins from injured cartilage lysates. A, Porcine metacarpophalangeal joints were opened and equilibrated at 37°C. Cartilage (∼12 gm) was dissected either into a tube on dry ice (time 0) or into serum-free medium for 10 minutes. Lysates were incubated with 4G10 beads prior to elution in phenyl phosphate solution. One-sixth of the material was Western blotted with 4G10 antibody. The rest of the material was electrophoresed on the same gel, and the gel was divided and silver-stained. Alignment of the gel pieces is shown. Silver-stained gel bands were cut out, trypsin-digested, and analyzed by tandem mass spectrometry. Bands at 100–120 kd and 60 kd were identified as focal adhesion kinase (FAK) and paxillin, respectively. B, Tandem mass spectra for 1 of the 2 peptides likely to be from FAK (TLLATVDETLPVLPASTHR) are shown. C, 4G10 antibody immunoprecipitates of cartilage lysates were stained with antibodies to total FAK, antibody to paxillin, or antibody to cortactin. Each membrane was reprobed with horseradish peroxidase–conjugated 4G10. D, Cartilage (∼0.5 gm) was dissected into lysis buffer (time 0) or serum-free Dulbecco's modified Eagle's medium and cultured for increasing periods of time prior to lysis. Western blotting of lysates with antibodies to phospho-Y576 FAK or phospho-Y118 paxillin and ERK is shown.

FAK and its substrate paxillin, a cytoskeletal adaptor protein, are both substrates of Src. FAK requires phosphorylation by Src on Tyr576 and Tyr577 for its full activation, while tyrosine phosphorylation of 2 sites on paxillin (Tyr118 and Tyr31) is Src dependent (26–28). To validate the findings, antibodies to FAK and paxillin were used for Western blotting of 4G10 antibody immunoprecipitates. The 100–120-kd complex on Western blotting with 4G10 antibody aligned with FAK, and the 60-kd band aligned with paxillin (Figure 2C). Western blotting of cartilage lysates with antibodies to phospho-FAK and phospho-paxillin was also carried out. This confirmed phosphorylation of Tyr576 on FAK and Tyr118 on paxillin within 10 minutes of cartilage injury (Figure 2D). When the 4G10 immunoprecipitates were Western blotted for cortactin (another Src substrate), a doublet of 80 kd and 85 kd was seen in the sample obtained from injured tissue (Figure 2C). This doublet aligned with the 80-kd and 85-kd doublet visible on Western blotting with 4G10 antibody. The 170-kd tyrosine-phosphorylated band was not identified.

Dependence of injury-induced protein tyrosine phosphorylation on Src but not FGF-2.

Mass spectrometry and Western blotting showed that 3 well-recognized Src substrates were phosphorylated upon injury to cartilage. Therefore, we examined the effect of inhibiting Src family kinases with the compound PP2 on injury-induced tyrosine phosphorylation. We first injected PP2 or 0.1% DMSO alone into the MCP joints. After 1 hour, the joints were opened, and cartilage was dissected into DMEM containing the inhibitor. The explants were either lysed immediately or after 20 minutes and were Western blotted for tyrosine-phosphorylated proteins. The explants left for 20 minutes after dissection showed, as expected, strong tyrosine phosphorylation when compared with those lysed immediately (Figure 3A). PP2 prevented tyrosine phosphorylation caused by injury, again implicating Src family kinases in this signaling (Figure 3A). PP2 did not affect the induction of phospho-ERK by injury (Figure 3A), which is partly due to release of FGF-2 (see below).

Figure 3.

Dependence of injury-induced tyrosine phosphorylation on Src-like signaling but not FGF-mediated signaling. A, Porcine metacarpophalangeal (MCP) joints either were not injected (−) or were injected with DMEM containing 10 μM Src family kinase inhibitor PP2 or vehicle (veh; 0.1% DMSO), followed by equilibration at 37°C. Cartilage (∼0.5 gm) was then dissected either directly into lysis buffer (time 0) or into medium alone, medium containing 0.1% DMSO, or PP2 and cultured for 20 minutes prior to lysis. Lysates were assayed for phosphotyrosine content by Western blotting with 4G10 antibody. The membrane was stripped and reprobed with antibodies to phospho-ERK and then ERK to ensure equal cellular protein loading. B, Porcine MCP joints either were not injected or were injected with medium containing 250 nM FGF receptor inhibitor PD173074 (PD) or medium alone (veh) and equilibrated at 37°C. Cartilage (∼0.5 gm) was then dissected either immediately into a tube on dry ice (time 0) or into serum-free medium alone or medium containing PD173074 and cultured for 5 minutes prior to lysis. Lysates were assayed for phosphotyrosine content by Western blotting with 4G10 antibody. The membrane was stripped and reprobed with antibodies to ERK to ensure equal cellular protein loading. Samples were re-run and Western-blotted for phospho-ERK. See Figure 1 for other definitions.

We were unable to confirm with available phospho-Src antibodies (to human Src Tyr416, Tyr418, and Tyr215) whether phosphorylation of known phosphorylation sites occurred upon injury: all antibodies caused strong nonspecific staining in the region of interest. Western blotting of 4G10 antibody immunoprecipitates with antibodies to c-Src also failed to demonstrate a band regulated by injury. However, Western blotting of lysates from cultured primary porcine chondrocytes with antibodies to the nonhematopoietic Src members (Src, Yes, Fyn, and Hck) showed that all 4 Src members were detectable (data not shown). We concluded that Src-like signaling was induced upon cartilage injury, but that any phosphorylation involved in this activation, and the member(s) of the Src family activated by injury, could not be determined.

Because cartilage injury causes release of FGF-2 from a pericellular pool (7, 10), and because FGFs activate Src in other cells of mesenchymal origin, we tested whether the injury-induced Src-like activity was attributable to release of FGF. In contrast to PP2, PD173074 had no perceptible effect on the induction of tyrosine phosphorylation following injury (Figure 3B). As a control, samples were also blotted for phosphorylated ERK, which was partially inhibited by the FGFR inhibitor, as expected. In another experiment, PD173074 also had no effect on the increased phosphorylation of FAK seen upon injury (data not shown). These experiments suggested that Src-like activity following injury was independent of FGF-2.

Amplification of the 3 MAPKs and IKK by FGF-2, but not Src, following injury to the intact articular surface.

To investigate whether Src-like activity could be responsible for the regulation of inflammatory signaling (i.e., MAPK and IKK) activated by cartilage injury, PP2 was injected into the joints prior to dissection. Cartilage dissection caused phosphorylation of ATF-2 after 10 minutes, but PP2 caused no reproducible inhibition of this phosphorylation (Figure 4A). As shown in Figure 4A, the phospho–ATF-2 levels were variable. Phosphorylation of paxillin on Tyr118 was also seen, and this was inhibited in the presence of PP2 (Figure 4A), suggesting that this phosphorylation was dependent on Src, as expected. In a separate experiment to examine the effect of PP2 on IKK activation following cartilage injury, cartilage that had been kept for increasing times after dissection was lysed, and its activity was assayed on GST–IκBα. There was strong phosphorylation of the substrate within 5 minutes of injury, but this was not inhibited by PP2 (Figure 4B). These experiments suggested that the Src-like signaling seen in cartilage following injury was not responsible for activation of MAPK and NF-κB pathways but instead may represent a separate, parallel pathway activated by FGF-independent mechanisms.

Figure 4.

Dependence of injury-induced IKK/MAPK activation on FGF receptor (FGFR) ligation but not Src-like signaling. A–C, Porcine metacarpophalangeal joints were equilibrated at 37°C and either were not injected or were injected with medium containing either vehicle (veh; 0.1% DMSO), 10 μM PP2 (A and B), or 250 nM PD173074 (PD) (C). After 1 hour, ∼0.2 gm of cartilage was dissected either directly into a tube on dry ice (time 0) or into medium alone or into medium containing vehicle or inhibitor for the times indicated, prior to lysis. The membranes were stripped and reprobed for ERK. In A and C (left panel), lysates were Western blotted. In B and C (right panel), lysate IKK activity was measured by kinase assay; quantitation of IκBα phosphorylation is shown as a percentage of the time 0 measurement. Membranes were probed for IKKα to ensure equal immunoprecipitation of IKK complex. In C (right panel), the mean IKK activity of cartilage left for 5 minutes (n = 3 similar experiments) was normalized to 100%, and quantitation of IKK activity of the other samples was expressed as a percentage of this. Bars show the mean ± SD. ∗ = P = 0.02 versus cartilage left for 5 minutes without inhibitor by one-way analysis of variance. D, Shown is a schematic presentation of the signaling response to intact articular cartilage injury. Following injury, there is activation of Src family kinases, MAPKs, and IKK. FGF signaling contributes to ∼50% of IKK/MAPK signaling seen following injury but does not appear to be responsible for the Src-like signaling. See Figure 1 for other definitions.

We investigated whether blockade of FGF-2 signaling would have effects on JNK, p38 MAPK, and IKK in addition to the effect on ERK. In experiments similar to that shown in Figure 3B, the FGFR inhibitor or vehicle alone was injected into joints. As expected, the cartilage explants that were left for 5 minutes after dissection showed strong phosphorylation of ATF-2 and p38 MAPK as well as increased IKK activity when compared with those lysed immediately (Figure 4C). The activation of all of these pathways was reduced by ∼50% in the presence of PD173074. This observation suggested that the FGF-2 released from cartilage upon injury not only is responsible in part for ERK activation but also augments the activity of JNK, p38 MAPK, and IKK. FGF-2 is not normally considered to be an activator of JNK or canonical NF-κB. As previously discussed, FGF-2 does not activate JNK or IKK in rested cartilage explants, and it appears that its role in signaling amplification is evident only upon injury to an intact articular surface.

The difference in the effects of the FGFR inhibitor on Src-like signaling and MAPK phosphorylation after injury (Figures 3B and 4C, respectively) is indirect evidence that Src-like signaling does not lie downstream of inflammatory signaling such as NF-κB. As summarized in Figure 4D, FGF-2 is present in the pericellular matrix bound to the heparan sulfate chains of the proteoglycan perlecan. It is released upon injury and augments MAPK and IKK signaling. The Src-dependent tyrosine phosphorylation that follows injury is a separate, FGF-independent pathway. It could be that a single damage-sensing mechanism activates Src-like signaling, IKK, and MAPKs, and that activation of IKK and MAPKs is subsequently enhanced by FGF-2 release. An alternative explanation is that different signals may be responsible for the Src-like activity and activation of the inflammatory signaling pathways.

Contribution of both Src and FGF-2 to gene expression following cartilage injury.

We previously showed that inflammatory signaling following dissection of articular cartilage induced expression of mRNA for IL-1α and IL-1β and secretion of intracellular proIL-1α protein (8) and induced activin A mRNA and protein secretion (12). Given the diversity of intracellular signaling following injury, the extent and strength of gene expression were explored. A panel of 15 inflammatory response gene mRNAs known to be induced by IL-1 were examined. To confirm the response of these genes in porcine cartilage to IL-1, rested explants were stimulated with IL-1α for either 4 hours or 20 hours. RNA was extracted, and mRNA expression was measured by quantitative RT-PCR (Table 1).

Table 1. Induction of inflammatory response gene mRNA by IL-1α and cartilage injury*
 Fold change at 4 hoursFold change at 20 hours
  • *

    Messenger RNA expression for 15 inflammatory response genes was assessed by quantitative reverse transcriptase–polymerase chain reaction. GAPDH was used as the housekeeping gene. Cartilage explants were rested for 24 hours and then stimulated for either 4 hours or 20 hours with 20 ng/ml interleukin-1α (IL-1α). Alternatively, cartilage was dissected either immediately into TRIzol (hour 0) or into culture medium for 4 hours or 20 hours. Comparisons were performed by one-way analysis of variance with Dunnett's correction. Each experiment was repeated 3 times. Values are the mean ± SD fold change compared with hour 0. MCP-1 = monocyte chemotactic protein 1; SAA2 = serum amyloid A2; COX-2 = cyclooxygenase 2; iNOS = inducible nitric oxide synthase; HAS-1 = hyaluronan synthase 1; MMP-1 = matrix metalloproteinase 1.

  • P < 0.001 versus hour 0, by one-way analysis of variance (ANOVA) with Dunnett's correction.

  • P < 0.01 versus hour 0, by one-way ANOVA with Dunnett's correction.

  • §

    P < 0.05 versus hour 0, by one-way ANOVA with Dunnett's correction.

 IL-1α8.0 ± 3.571.8 ± 16.94.8 ± 2.316.0 ± 9.4
 IL-61,11.6 ± 91.76.3 ± 1.735.3 ± 14.32.3 ± 0.6
 IL-823.9 ± 6.70.5 ± 0.480.2 ± 15.81.0 ± 0.4
 MCP-17.9 ± 3.56.8 ± 1.45.9 ± 2.03.4 ± 0.5
 CCL511.1 ± 8.32.4 ± 0.921.6 ± 5.31.4 ± 0.6
 Inhibin βA5.5 ± 0.78.9 ± 1.61.5 ± 0.32.5 ± 0.5
 SAA27.4 ± 4.63.2 ± 2.417.8 ± 7.521.6 ± 22.5
Synthetic enzymes    
 COX-25.0 ± 1.95.4 ± 1.84.1 ± 1.91.8 ± 1.0
 iNOS81.3 ± 1981.3 ± 0.21,105 ± 1665.3 ± 2.2
 HAS-13.2 ± 2.00.94 ± 0.51.4 ± 2.00.6 ± 0.3
 MMP-111.3 ± 5.668.2 ± 24.631.9 ± 14.52,81.1 ± 97.5
 MMP-352.5 ± 21.225.9 ± 7.963.3 ± 21.541.0 ± 10.4
 MMP-135.4 ± 2.81.6 ± 0.52.1 ± 0.50.3 ± 0.1
 ADAMTS-435.5 ± 7.72.7 ± 1.1§6.5 ± 3.21.0 ± 0.6
 ADAMTS-59.1 ± 4.51.7 ± 0.22.3 ± 1.11.2 ± 0.1

To investigate whether these same genes were regulated by injury, cartilage was dissected either immediately into RNAlater (Ambion) or into serum-free DMEM for 4 hours or 20 hours prior to RNA extraction. Eight of the 15 IL-1–inducible genes were significantly induced at 4 hours following dissection (expression of a further gene, SAA2, was significantly elevated at 20 hours) (Table 1). The genes regulated by injury therefore appeared to be a subset of IL-1–sensitive genes.

To examine whether Src-like signaling was important in the gene response to injury, 10 genes that were induced ≥1.7-fold by injury were selected for further investigation. The effects of intraarticular injection of PP2 were compared with those of vehicle alone, 4 hours after injury (20 hours for SAA2). The presence or absence of vehicle did not significantly affect the response of the genes (data not shown). Induction of each gene by injury was normalized to 100% in the presence of vehicle, and the percent inhibition with PP2 was determined (Figure 5A). Eight of the 10 genes tested were significantly inhibited, and 5 were inhibited by at least 50%. This was consistent with Src-like activity being a significant signal for the gene response to articular cartilage injury.

Figure 5.

Dependence of the inflammatory gene response to intact articular cartilage injury on Src and fibroblast growth factor 2 (FGF-2). In 2 separate experiments, porcine metacarpophalangeal joints were injected with medium containing vehicle (0.1% DMSO) and 10 μM Src family kinase inhibitor PP2 (A) or 250 nM FGF receptor inhibitor PD173074 (PD) (B) and equilibrated at 37°C. Cartilage (∼0.2 gm) was then dissected either immediately into RNAlater (hour 0) or into medium containing 0.1% DMSO or the respective inhibitor and cultured for 4 hours (or 20 hours in the case of serum amyloid A2 [SAA2]) prior to RNA extraction. Quantitative reverse transcription–polymerase chain reaction was carried out for the housekeeping gene GAPDH (used for normalization) and 10 inflammatory response mRNAs. Bars show the mean ± SD percent change after treatment with PP2 (A) or PD173074 (B) compared with the sample injected with vehicle alone, normalized to 100% for each gene (broken line). For each mRNA sample, 3 separate experiments were performed. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001 versus vehicle alone, by one-way analysis of variance with Dunnett's correction. NS = not significant; IL-1α = interleukin-1α; MCP-1 = monocyte chemotactic protein 1; COX-2 = cyclooxygenase 2; MMP-1 = matrix metalloproteinase 1.

Because FGF-2 contributed to MAPK/IKK signaling following injury, we determined that it was likely that the FGFR inhibitor would also interfere with this gene response. In an experiment similar to that shown in Figure 5A, the same mRNAs were assessed 4 hours after cartilage dissection in the presence of the FGFR inhibitor PD173074. Expression of 7 of the 10 injury-regulated genes was significantly reduced in the presence of the FGFR inhibitor (Figure 5B). In contrast to the effects of the Src inhibitor, the expression of IL-1α, MCP-1, and SAA2 was not reduced by FGFR inhibition, while the transcription of aggrecanases did appear to be partially dependent on FGF.


Inflammation is usually considered as a process in vascularized tissue that increases blood flow and causes diapedesis of leukocytes. However, as shown in Figure 4D, we demonstrated that even in avascular tissue such as cartilage, trauma causes intracellular signaling and gene expression characteristic of an inflammatory response. The signaling included activation of Src or Src-like kinases as well as MAPKs and IKK.

A striking feature of the response to injury was that activation of Src, IKK, and JNK occurred only upon damaging an intact articular surface; cutting cartilage adapted to culture (while still releasing FGF-2) did not cause these effects. Furthermore, no detectable soluble factor was released from the injured cartilage that reproduced the effects on cultured chondrocytes. It is possible that a soluble factor is unstable and thus undetectable. Alternatively, it is possible that the chondrocytes directly sense injury to the matrix, and that there is propagation of a signal by direct cell interactions. It is intriguing that whatever the sensing mechanism, it is an alarm that is not reset when the tissue is cultured.

This is the first report of Src activation and its effect on the gene response following injury of intact cartilage. Phosphorylation of FAK has been observed upon mechanical stimulation of cultured chondrocytes and was suggested to be integrin dependent (29). Src-like signaling appears to contribute to injury-induced gene expression independently of classic inflammatory signaling pathways such as NF-κB and the MAPKs. Src family kinases have been implicated in NF-κB activation by bacterial lipopolysaccharide and TNF family members (30, 31) but have also been reported to activate inflammatory gene transcription independently (32).

This injury response is not restricted to porcine articular cartilage. Preliminary experiments suggest that a similar response also occurs following injury to other connective tissue. For example, Src, IKK, and MAPK activation follows dissection of porcine synovium (Watt FE and Saklatvala J: unpublished observations). Similar signaling or gene expression is also seen in the connective tissue of mice, for example, following dissection of mesentery or in response to avulsion injury of the hip osteochondral junction (Ismail HM, et al: unpublished observations).

FGF-2 appears to be an important contributor to the response in cartilage, although its role is complex: FGF-2 suppresses IL-1–induced aggrecanolysis in vitro but at the same time does not inhibit the induction of several IL-1–induced genes (33). FGF-2 was shown to be chondroprotective in a murine model of OA (34), and yet here we observed that it amplifies inflammatory signaling and gene expression immediately upon cartilage injury.

The inflammatory gene response that follows cartilage injury is not simply the same as that caused by IL-1α in culture. For example, IL-8, HAS-1, and iNOS were all induced by IL-1 but not by cartilage injury. Interestingly, IL-1α, inhibin βA (the subunit of activin A), and MMP-1 mRNA were all regulated more strongly by injury than by IL-1 alone; this may illustrate that amplification by FGF-2 or Src-like signaling is important in gene expression following injury. Src and FGF-2 appear to contribute to this gene response independently; some Src-dependent genes (IL-1α, MCP-1, and SAA2) were not regulated by FGF-2. Given that pharmacologic inhibitors of signaling pathways may not be completely specific, these observations further support the signaling scheme presented in Figure 4D.

Is an injury-induced gene response important in vivo, and, specifically, in posttraumatic OA? In a recently published study by our group, 21 genes were significantly up-regulated in murine whole joint and articular cartilage immediately following surgical destabilization of the medial meniscus, which causes OA (35). All of the genes induced by dissection of porcine cartilage were among these and included IL-1, IL-6, COX-2, inhibin βA, MMP-3, ADAMTS-4, MCP-1 (CCL2), and SAA3 (the murine equivalent of SAA2). The exception was MMP-1 (which is not functional in the mouse). Interestingly, joint immobilization abrogates much of the murine gene response and protects against OA, implicating this immediate tissue injury response in the development of posttraumatic OA, at least in the mouse.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Watt had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Watt, Didangelos, Peirce, Vincent, Wait.

Acquisition of data. Watt, Ismail, Didangelos, Peirce, Wait.

Analysis and interpretation of data. Watt, Ismail, Wait, Saklatvala.


We thank Shajna Begum, Maria J. Smolinska, and Lesley Rawlinson for technical advice and support. The IKK assay protocol was adapted from a protocol described by M. Bebien (Kennedy Institute of Rheumatology).