A sodium dodecyl sulfate–polyacrylamide gel electrophoresis–liquid chromatography tandem mass spectrometry analysis of bovine cartilage tissue response to mechanical compression injury and the inflammatory cytokines tumor necrosis factor α and interleukin-1β

Authors


Abstract

Objective

To compare the response of chondrocytes and cartilage matrix to injurious mechanical compression and treatment with interleukin-1β (IL-1β) and tumor necrosis factor α (TNFα), by characterizing proteins lost to the medium from cartilage explant culture.

Methods

Cartilage explants from young bovine stifle joints were treated with 10 ng/ml of IL-1β or 100 ng/ml of TNFα or were subjected to uniaxial, radially-unconfined injurious compression (50% strain; 100%/second strain rate) and were then cultured for 5 days. Pooled media were subjected to gel-based separation (sodium dodecyl sulfate–polyacrylamide gel electrophoresis) and analysis by liquid chromatography tandem mass spectrometry, and the data were analyzed by Spectrum Mill proteomics software, focusing on protein identification, expression levels, and matrix protein proteolysis.

Results

More than 250 proteins were detected, including extracellular matrix (ECM) structural proteins, pericellular matrix proteins important in cell–cell interactions, and novel cartilage proteins CD109, platelet-derived growth factor receptor–like, angiopoietin-like 7, and adipocyte enhancer binding protein 1. IL-1β and TNFα caused increased release of chitinase 3–like protein 1 (CHI3L1), CHI3L2, complement factor B, matrix metalloproteinase 3, ECM-1, haptoglobin, serum amyloid A3, and clusterin. Injurious compression caused the release of intracellular proteins, including Grp58, Grp78, α4-actinin, pyruvate kinase, and vimentin. Injurious compression also caused increased release and evidence of proteolysis of type VI collagen subunits, cartilage oligomeric matrix protein, and fibronectin.

Conclusion

Overload compression injury caused a loss of cartilage integrity, including matrix damage and cell membrane disruption, which likely occurred through strain-induced mechanical disruption of cells and matrix. IL-1β and TNFα caused the release of proteins associated with an innate immune and stress response by the chondrocytes, which may play a role in host defense against pathogens or may protect cells against stress-induced damage.

Osteoarthritis (OA) is characterized by cartilage degeneration, which results from an imbalance between matrix synthesis and matrix degradation. Development of OA secondary to traumatic joint injury occurs in ∼15–75% of patients over followup periods of 14–22 years, equivalent to an average relative risk or odds ratio of between 3 and 20 of developing OA postinjury (for review, see ref.1). Moreover, corrective surgery has little or no impact on the risk of developing OA following traumatic joint injury (1). In vitro models of joint injury have helped to understand the contribution of acute mechanical compression injury to tissue damage and degeneration commonly associated with OA. Experiments based on these models have shown that high-level stress or strain injury may lead to chondrocyte apoptosis (2), tissue fissuring and swelling (3), changes in the dynamic tissue stiffness consistent with damage to the collagen network (2), increase in type II collagen degradation, particularly at the boundaries of the tissue where the strain is highest (4, 5), and increased loss of proteoglycans (6). In addition, cytokines and mechanical injury may synergistically enhance matrix damage (7).

While the role of cytokines in OA is still debated, interleukin-1 (IL-1) and tumor necrosis factor α (TNFα) have been shown to be present in synovial fluid from OA joints (8, 9), and chondrocytes near OA lesions often possess increased levels of TNFα and IL-1 receptors (10, 11). In vitro, both IL-1 and, to a lesser extent, TNFα can inhibit the synthesis of collagens and proteoglycans, increase matrix metalloproteinase (MMP) and aggrecanase expression, increase aggrecan and collagen degradation, and enhance the production of proinflammatory mediators (12). Following joint injury, levels of cytokines, including IL-1β and TNFα, are elevated, suggesting that local production within the joint may contribute to secondary damage or repair (13). Therefore, consideration of the interactions between cytokine and mechanical tissue damage may help us to understand the molecular basis of cartilage degeneration following joint injury and secondary OA.

Recent studies have delineated changes in chondrocyte gene transcription following cartilage mechanical injury and IL-1β treatment in vitro. A time-course study of 24 genes involved in normal cartilage maintenance showed a 5–250-fold elevation in MMP-3, ADAMTS-5, and transforming growth factor β expression within 4 hours of rapid 50% compression injury, which remained elevated at 24 hours (14). Transcription of the matrix proteins aggrecan and fibronectin and of the matrix proteases MMP-1, MMP-9, and MMP-13 were also increased by 24 hours (14). Chan et al (15) analyzed transcription by gene array analysis 3 hours after rapidly applied 30 MPa compressive stress and found that transcription of the adhesion molecules and growth factors intercellular adhesion molecule 3, neural cell adhesion molecule, N-cadherin, vascular cell adhesion molecule 1 (VCAM-1), and insulin-like growth factor 1 were decreased, while the chemokine receptor CCR10, high mobility group box chromosomal protein 2, neurogranin, and ezrin were up-regulated. Saas et al (16) showed changes in gene expression after treatment of cartilage with IL-1β, with up-regulation of cytokines and chemokines, including CXCL3, CCL3, CCL5, CXCL6, GRO1, CXCL2, IL-11, IL-1β, CCL20, IL-8, cyclooxygenase 2, granulocyte colony-stimulating factor, and CCL4, as well as down-regulation of proteins, including collagens and other matrix proteins (16). Both mechanical compression injury and cytokine treatment alter normal chondrocyte gene expression in vitro.

Mass spectrometry (MS) has recently been used for protein identification and protein profiling to identify arthritis biomarkers and to study the effects of cytokine and growth factor treatments in cartilage. Searches for protein biomarkers of OA and rheumatoid arthritis (RA) in synovial fluid have led to the identification of serum amyloid A (SAA) and calgranulins A, B, and C, which are proteins that are correlated with increased severity of RA (17). Hermansson et al (18) identified the proteins inhibin βA and C-terminal telopeptide of type II collagen that were released from OA and healthy cartilage cultured in vitro. Proteomics studies of the effects of basic fibroblast growth factor and IL-1 used 35S-methionine labeling to identify new synthesis and showed increased protein expression of tissue inhibitor of metalloproteinases 1, MMP-1, MMP-3, chitinase 3–like protein 1 (CHI3L1), and SAA3 in response to these stimuli (19, 20).

Mechanical injury and exposure to the cytokines TNFα and IL-1β cause changes in tissue properties and chondrocyte behavior, and both injury and cytokine exposure may lead to cartilage degeneration. The objective of our study was to identify and compare changes in protein release from cartilage explants in response to acute mechanical compression injury and to treatment with cytokines IL-1β or TNFα. We used a systems-level MS-based proteomics method to identify and compare proteins released, including matrix protein proteolytic fragments. Proteins released may reflect both differences in new protein synthesis and differences in the release of preexisting proteins that are present in the tissue.

MATERIALS AND METHODS

Cartilage explant culture.

Joints from 2–3-week-old bovine calves were obtained from the local abattoir (Research 87, Hopkinton, MA) and explanted as described previously (2). Briefly, articular cartilage–bone cylinders measuring 9 mm in diameter were drilled from the patellofemoral groove perpendicular to the joint surface, and 2 sequential 1-mm–thick cartilage slices were obtained by microtome from the upper middle zone. Using a dermal punch, either one 6-mm–diameter disk (for untreated and cytokine-treated studies) or four 3-mm–diameter disks (for injury studies) were cored from each cartilage slice. Each 6-mm–diameter disk or each set of four 3-mm–diameter disks was placed into 1 well of a 24-well plate containing 2 ml of culture medium with 1% insulin–transferrin–selenium A (ITS; Invitrogen, Carlsbad, CA) (6). All treatments were initiated on day 7 of culture.

Mechanical injury.

Injurious compression was performed using an incubator-housed loading apparatus (6, 14). Cartilage disks measuring 3 mm in diameter were individually placed between 2 impermeable platens of a polysulfone loading chamber and subjected to a single uniaxial unconfined compression to 50% strain at a strain rate of 100%/second (velocity 1 mm/second). This compression resulted in a measured peak stress of 18.3 ± 0.7 MPa (mean ± SEM). This loading protocol was chosen based on our previous dose-response studies of peak stress (2) and strain rate (21), and a comparison of single versus multiple compression injuries (2, 21). Injured explants were placed in sets of 4 per culture well in 2 ml of medium, yielding the same cartilage volume per ml of medium as a single 6-mm–diameter disk used for cytokine-treated and untreated control explants.

TNFα, IL-1β, and posttreatment culture.

Cytokines were resuspended in 0.1% bovine serum albumin. On day 7, the 6-mm–diameter disks were left untreated or were treated with 10 ng/ml of IL-1β or 100 ng/ml of TNFα in 2 ml of medium without 1% ITS for 5 days. All cultures were subjected to a 10% medium removal and 10% medium supplementation every 24 hours (0.6 ml of medium/explant/day) until the cultures were ended on day 5 of treatment by placing the medium and explants in a −80°C environment.

Deglycosylation and reducing sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE).

Medium from 2 or more explants (8 or more for injury studies) taken from each of at least 5 different animals was pooled into a single sample for SDS-PAGE–liquid chromatography tandem MS (SDS-PAGE-LC-MS/MS). Two-milliliter aliquots of pooled medium were treated overnight at 37°C with 50 mU of chondroitinase ABC in 3 mM EDTA and dialyzed in a 7.5-kd–cutoff membrane for 2 hours against buffer containing 10 mM Tris acetate, 10 mM NaCl, 10 mM sodium acetate, and 3 mM EDTA. Samples were then dialyzed against pure water overnight at 4°C and concentrated with a SpeedVac. Concentrated samples (10–15 μl) were combined with SDS-PAGE sample buffer containing 50 mM dithiothreitol, boiled for 7 minutes, and loaded on a 4–15% gradient gel and run at 15 mA. Gels were washed with water, stained with Coomassie Blue Safestain (Invitrogen) for 30 minutes, and destained overnight.

Western blot analysis of fibronectin, type VI collagen, cartilage oligomeric matrix protein (COMP), and actin.

To verify the gel-based findings, equal amounts of concentrated medium from injured or untreated explants from 4 randomly chosen experiments were run on 4–15% gels and transferred to polyvinylidene difluoride for immunoblotting. Fibronectin blots were performed using a monoclonal antibody (1:2,000 dilution; BD Biosciences, San Jose, CA), whereas type VI collagen (AB782, 1:250 dilution; Chemicon, Temecula, CA) and actin (C11, 1:500 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) blots were performed using goat polyclonal antibodies. Anti-COMP antibody was a kind gift from Dr. Dick Heinegård (Lund University, Lund, Sweden). Appropriate secondary antibodies conjugated to horseradish peroxidase were used for detection of chemiluminescence.

Reduction, alkylation, trypsinization, and extraction from gel slices.

Each sample contained within a lane of the PAGE gel was divided into ∼30 equal slices, which were destained, reduced, alkylated, and digested in-gel (22). Each slice was reduced with 1 mg/ml of tris(2-carboxyethyl) phosphine hydrochloride for 10 minutes, alkylated with 10 mg/ml of iodoacetamide in the dark for 45 minutes, and treated with 10 ng/ml of trypsin gold (Promega, Madison, WI) in 50 mM ammonium bicarbonate for 1 hour on ice, followed by overnight incubation at 37°C. Extracted peptides were desalted using Millipore C18 ZipTips (Millipore, Bedford, MA) according to the manufacturer's instructions. Peptides from adjacent gel slices were combined to yield ∼16 peptide samples per treatment condition.

Liquid chromatography.

Capillary columns were prepared in-house using pulled, fritted capillaries (New Objective, Woburn, MA). Capillaries were packed with 5 μm of 300Å Vydac protein/peptide C18 material (Vydac, Hesperia, CA) to yield final column dimensions of 75-μm inner diameter × 160 mm, with a 10-μm tip. Samples were manually injected (Rheodyne injector; 0.5-μl internal sample loop). Chromatography was performed with an Agilent 1100 capillary pumping system (Agilent, Palo Alto, CA) attached to a passive splitter, allowing a flow rate of ∼250 nl/minute through the column. Peptides eluted with a linear gradient from 0% B to 60% B (buffer B, consisting of 1.2% volume/volume/volume in 93.8% acetonitrile, 5% water) over 120 minutes before equilibrating at starting conditions (buffer A, consisting of 1.2% acetic acid [v/v] in water) over the following 80 minutes.

Tandem mass spectrometry.

The LC column was connected to an Applied Biosystems QSTAR XL quadrupole-time-of-flight instrument (Applied Biosystems, Foster City, CA) equipped with a nanospray source and running Analyst version 1.0 software (23). Data were acquired during the entire chromatographic run (200 minutes) using an information-dependent acquisition method with a 16-second cycle time. Each cycle consisted of 1 MS scan (mass/charge [m/z] 400–1,800), followed by 3 MS/MS scans (m/z 100–2,000) on the 3 most abundant ions with counts >10, a charge of 2–4, and an m/z of 400–1,800, and excluding previous ions and isotopes for 60 seconds.

Mass spectral data analysis.

Data were analyzed using Spectrum Mill proteomics software (Rev A.03.02.060; Agilent). Raw data were extracted under default conditions and searched against bovine sequences and mammalian sequences in the NCBInr database using trypsin as the protease, allowing 2 missed cleavages, and including variable modifications of oxidized methionine and N-terminal glutamine conversion to pyroglutamic acid in the search. Peptides were considered valid with a forward–reverse score >2 and a rank 1–rank 2 score >2, a score threshold >7.67, and percentage-scored peak intensity >70% (24). Only proteins with 2 or more validated peptides and a total score >25 were considered valid for reporting. To compare identified proteins between treatment groups, the number of spectra and the summed ion intensity of peptides for each protein (total ion intensity) were used as indicators of protein amounts. Because these are semiquantitative metrics, we only considered those proteins with at least 5 additional spectra and at least a 10-fold increase in total ion intensity sufficiently different for reporting. To evaluate signs of matrix degradation, gel slices were used to estimate the molecular weight of proteins using the molecular weight marker as well as the molecular weight of some of the protein constituents.

RESULTS

Protein identification.

We identified 252 proteins that had a score of ≥25, which corresponds to MS/MS sequences that match 2 or more unique peptides from each protein. A complete list of the identified proteins is shown in Supplementary Table 1 (available on the Arthritis & Rheumatism Web site at http://www.mrw.interscience.wiley.com/suppmat/0004-3591/suppmat/). The identification statistics from the Spectrum Mill proteomics software analyses, including the number of spectra extracted and identified as well as the number of peptides and proteins identified, are shown in Table 1. Over 100 proteins and 900 peptides were positively identified from each sample. A Venn diagram illustrating the overlap of identification between samples is shown in Figure 1A. TNFα and IL-1β are grouped together as “cytokine,” given that 111 of the 153 proteins released after TNFα treatment and the 138 released after IL-1β treatment were identified in common.

Table 1. Identification statistics from Spectrum Mill proteomics software analyses of bovine articular cartilage under 4 experimental conditions*
ID statisticsControlIL-1β treatmentTNFα treatmentCompression injury
  • *

    Values are the number of extracted spectra, identified spectra, peptides, proteins identified by 2 peptides (with a score >25), and proteins identified through the bovine database (which included at least a partial sequence of each of the identified proteins) by Spectrum Mill proteomics software analysis of bovine cartilage subjected to each of the 4 experimental conditions: untreated (control), interleukin-1β (IL-1β) treatment, tumor necrosis factor α (TNFα) treatment, and compression injury. Among the 12 highest-scoring cartilage proteins were aggrecan, α1(II) collagen, perlecan, fibronectin, cartilage oligomeric matrix protein, α1(VI), α3(VI), and α1(XII) collagens, thrombospondin 1, YKL-40, link protein, and complement factor B. (See Supplementary Table 1, which is available on the Arthritis & Rheumatism Web site at http://www.mrw.interscience.wiley.com/suppmat/0004-3591/suppmat/)

Total extracted spectra15,23816,97421,52521,689
Total identified spectra3,4076,4666,0014,252
Total peptides9949311,1361,200
Proteins identified (score >25)113138153168
Proteins from bovine database106125136153
Figure 1.

Global analysis of proteins identified in bovine cartilage explants subjected to injurious compression, treatment with cytokines, or no treatment. A, Venn diagram of proteins identified in untreated, injuriously compressed, and cytokine-treated (pooled data from tumor necrosis factor α [TNFα] and interleukin-1β [IL-1β] treatments) samples of bovine cartilage. The 4 treatment conditions yielded 82 proteins in common, while a total of 119 proteins were identified by only a single experimental condition. Overlap was noted between the results of TNFα and IL-1β treatment, such that of the 153 and 138 proteins identified by these treatments, respectively, 115 were identified by both treatments. B, Pie charts showing the composition of proteins released into the medium, based on protein locations. Samples subjected to compression injury show a larger component of intracellular proteins as compared with samples subjected to the other conditions. ECM = extracellular matrix; ER = endoplasmic reticulum.

The highest-scoring and highest total-intensity proteins were predominantly those that comprise the extracellular matrix (ECM), including aggrecan, α1(II) collagen, perlecan, fibronectin, α1(VI), α3(VI), and α1(XII) collagens, and link protein, as well as thrombospondin 1, COMP, CHI3L1, and complement factor B. We also observed proteins recently described in cartilage, including cytokine-like protein C17, connective tissue growth factor, IL-17B, scrapie-responsive protein 1 (SCRG-1), and follistatin-like protein (18, 25, 26). Other proteins, such as CD109, platelet-derived growth factor receptor-like (PDGFR-like), angiopoietin-like 7, and peptidoglycan-recognition protein long (PGRP-L), represent novel cartilage proteins.

Differences between treatments.

To understand the similarities and differences between the treatment groups, systems-level differences and individual protein differences were evaluated. A global analysis of the protein composition of the medium samples categorized by protein location is shown in Figure 1B. While the composition of proteins released into the medium from both cytokine-treated and untreated samples was similar, the samples subjected to compression injury released a comparatively higher number of proteins that are typically localized to the intracellular environment.

To determine individual differences, we used spectra number and total ion intensity for comparison of the amounts of protein between samples (see Materials and Methods). Proteins found to be elevated after each treatment as compared with controls are shown in Table 2, including information about the function and location of the proteins. Proteins released in response to cytokine treatment appeared primarily to be involved in innate immunity or stress response. Proteins such as MMP-3 and CHI3L1 were elevated with all treatments as compared with controls, while proteins released in response to injury were mainly intracellular and suggestive of a loss of membrane integrity.

Table 2. Proteins identified as elevated by SDS-PAGE-LC-MS/MS analysis of medium following treatment of bovine articular cartilage with IL-1β, TNFα, or injurious compression as compared with no treatment*
Protein identifiedIL-1β treatmentTNFα treatmentCompression injuryProtein locationProtein function
  • *

    Proteins were identified as being elevated with treatment as compared with the untreated control (>10-fold increase in summed peptide ion intensity and ≥5 spectra difference), as determined by sodium dodecyl sulfate–polyacrylamide gel electrophoresis–liquid chromatography tandem mass spectrometry (SDS-PAGE-LC-MS/MS). Values are the fold increase, representing total ion intensity relative to that of the control. If no spectra were identified in the untreated sample, the order of magnitude of total ion intensity of each of the proteins is represented by + signs. The value <10 indicates that the protein was present in the sample but did not meet the above criteria for a difference.− − − indicates the protein was not identified in the sample. IL-1β = interleukin-1β; TNFα = tumor necrosis factor α; MMP-3 = matrix metalloproteinase 3; ECM = extracellular matrix; PGRP-L = peptidoglycan-recognition protein, long; ER = endoplasmic reticulum; PTM = posttranslational modification; PDI = protein disulfide isomerase; SLRP = small leucine-rich repeat proteoglycan; ERp-72 = endoplasmic reticulum protein 72; Rab GDIβ = Rab GDP-dissociation inhibitor β; PDGFR-like = platelet-derived growth factor receptor–like.

Chitinase 3–like 1 isoform 2 (YKL-40; gp38)2009914.2SecretedStress response
Stromelysin 1 (MMP-3)++++++++++++++++SecretedMetalloproteinase
Peroxiredoxin 1+++++++++++++CytoplasmRedox
Vimentin3835283IntracellularCytoskeleton
Complement factor B++++++++++++− − −SecretedComplement
Chitinase 3–like 2 (YKL-39; chondrocyte protein 39)+++++++++++<10SecretedUnknown
Clusterin11.817.8<10SecretedStress response
Extracellular matrix protein 1+++++++++++− − −ECMStructural organization
Cl inhibitor<1012.1<10SecretedProtease inhibitor
Haptoglobin+++++++++− − −SecretedHemoglobin scavenger
Vascular cell adhesion molecule 1<10+++++− − −MembraneAdhesion
Serum amyloid A3++++++++++− − −SecretedInnate immunity/stress
α1-acid glycoprotein++++++++++− − −SecretedInnate immunity
Neutrophil gelatinase–associated lipocalin23.0<10<10SecretedInnate immunity
CD14+++++<10− − −MembraneInnate immunity
Complement C1qα+++++<10− − −SecretedClassical complement pathway
Lactoferrin+++++− − −− − −SecretedInnate immunity
Complement C1r+++++− − −− − −SecretedClassical complement pathway
N-acetylmuramoyl-L-alanine amidase (PGRP-L)− − −++++− − −SecretedInnate immunity
Annexin 1− − −++++− − −Membrane associatedTrafficking
Angiopoietin-like 7<10+++++<10SecretedUnknown, possible antiangiogenic
Annexin A8− − −++++<10Membrane associatedUnknown
Complement C3− − −++++− − −SecretedComplement
Triosephosphate isomerase<10++++++++++CytoplasmGlycolysis
Lactate dehydrogenase A<10+++++++++CytoplasmMetabolism
β-actin<10+++++++++++IntracellularCytoskeleton
α2(VI) collagen12.1<1010.2ECMStructural
14-3-3 protein θ− − −+++++++++CytoplasmRegulatory
Annexin A2− − −++++++++Membrane associatedTrafficking
Proline hydroxylase α<10<1015.6ERPTM
α4-actinin− − −− − −307IntracellularCytoskeleton
Glucose-regulated protein 58 kd<10− − −+++++ERProtein folding (PDI)
Fibronectin<10<1013ECMCell-matrix interactions
Pyruvate kinase M1/M2<10<10+++++CytoplasmGlycolysis
UDP-glucose pyrophosphorylase 2− − −− − −+++++CytoplasmMetabolism
Cytoskeleton-associated protein 4<10− − −+++++IntracellularCytoskeleton
Protein disulfide isomerase A6− − −− − −+++++ERProtein folding (PDI)
Hsp90α− − −− − −+++++CytoplasmProtein folding
Glucose-regulated protein 78 kd<10− − −+++++ERProtein folding/assembly
Phosphoglycerate kinase 1<10<10+++++CytoplasmGlycolysis
Tumor-rejection antigen 1 (gp96)− − −− − −+++++ERProtein assembly
Dermatan sulfate proteoglycan 3 (epiphycan)<10<10+++++ECMSLRP and structured organization
Protein disulfide isomerase A4 (ERp-72)− − −<10+++++ERProtein folding (PDI)
Lysine hydroxylase− − −<10+++++ERPTM
Endoplasmic reticulum protein 46− − −<10+++++ERProtein folding
Glucose-6-phosphate isomerase− − −<10+++++CytoplasmMetabolism
Calreticulin (calregulin)− − −<10+++++ERProtein folding
Nucleotide diphosphate kinase− − −<10+++++CytoplasmMetabolism
FK-506 binding protein 9− − −− − −+++++ERProtein folding
Rab GDIβ− − −− − −++++CytoplasmSignaling
Parkinson's disease 7− − −<10+++++CytoplasmUnknown
Enolase 1<10− − −+++++CytoplasmGlycolysis
Peroxiredoxin 2− − −<10++++CytoplasmRedox
α1(IX) collagen<10− − −0.02ECMStructural
α1(XVI) collagen<10<100.08ECMStructural
MMP-2<10<100.09SecretedMetalloproteinase
Fibulin 4<100<10ECMStructural
PDGFR-like protein<10<100Membrane/secretedUnknown

To verify the loss of membrane integrity and release of intracellular proteins, antiactin Western blotting was performed on medium from 4 sets of injuriously compressed explants on day 5. Mechanical injury increased the release of actin into the medium, with little variability (Figure 2A), while no band was visible with the untreated control samples. Loss of membrane integrity may occur with apoptosis, mechanical lysis, or necrosis. However, the timing of the release of intracellular components differs between these mechanisms of cell death, with mechanical lysis resulting in almost immediate release and apoptosis causing release later in the progression of cell death. To more clearly delineate the time dependence of intracellular protein release, an actin immunoblot was performed on medium from posttreatment day 1 and day 4 (96 hours) following injury (Figure 2B, 2 experiments shown). Loss of actin occurred primarily in the first 24 hours. While some leaking of intracellular proteins occurred with time, this early release of actin suggests that mechanical disruption (lysis) may be a significant contributor in this unconfined compression model in addition to apoptosis.

Figure 2.

Western blot analysis of antiactin and antifibronectin in medium from 3-mm bovine cartilage explants subjected to injurious compression or no treatment (free swell). A, Antiactin blot of medium obtained on day 5 from explants from 4 different animals (each pair of control and compression-injured samples in adjacent lanes). Compression injury caused increased release of actin, suggesting a highly reproducible loss of membrane integrity in a population of chondrocytes. B, Antiactin blot of medium obtained at 24 and 96 hours from explants from 2 different animals in 2 separate experiments (with each pair of control and compression-injured samples at the 2 time points in adjacent lanes), as well as with a positive control for cell lysis. Actin was detected in samples following injury, and levels at 24 and 96 hours were similar, indicating that actin release occurred primarily within the first 24 hours after injury. C, Antifibronectin blot of medium obtained on day 5 from explants from 4 different animals (each pair of control and compression-injured samples in adjacent lanes). Injury caused an increase in fibronectin release and breakdown. Molecular weight markers are shown at the left.

Matrix protein changes and degradation.

Understanding the changes in matrix protein release and matrix degradation was a primary goal of this study. Fibronectin was the second highest–scoring matrix protein, and its release was found to be increased with injury at the threshold set for this study. Increased fibronectin release following compression injury as compared with untreated controls was verified in experiments on 4 different animals, as determined by Western blotting (Figure 2C). Elevated fibronectin release was also seen with cytokine treatment, and small amounts of fibronectin fragments were seen with cytokine and mechanical compression injury as compared with untreated controls (results not shown).

Protein proteolysis was assessed as the total ion intensity versus the approximate molecular weight, based on gel slice position and the proteins present within each slice. Type VI collagen is synthesized as a heterotrimer (α1/α2/α3[VI] collagen), which further organizes into dimers and then into tetramers to establish a loose collagen network within the pericellular matrix. In this study, type VI collagen subunits showed evidence of increased release into the medium, with possible degradation in response to injury. The highest-intensity collagen VI subunit by MS was α1(VI) collagen, with peak intensities at 100–150 kd, representing the full-length subunit, in medium from untreated and from injured explants, as well as at 50–75 kd in the compression injury sample, indicating a possible α1(VI) fragment (Figure 3A).

Figure 3.

Type VI collagen degradation in medium from bovine cartilage explants subjected to injurious compression or no treatment. A, Summed peptide ion intensity for each of the 3 type VI collagen subunits plotted against the molecular weight (determined by molecular weight marker and protein constituents). Both α1(VI) and α3(VI) collagen subunits showed increased intensity in the 50–75-kd region, suggestive of the breakdown of type VI collagen. B, Western blot analysis of anti–type VI collagen in medium obtained on day 5 from cartilage explants from 4 different animals (each pair of control and compression-injured samples in adjacent lanes). There is a stronger band at 150 kd with compression injury, as well as a doublet between 50 kd and 75 kd in which the upper band appears more intense, in samples subjected to compression injury. Molecular weight markers are shown at the left. Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

Similarly, α3(VI) collagen and α2(VI) collagen had elevated total ion intensity in the 50–75-kd region, smaller than the full-length protein subunits and suggestive of proteolysis of either the mature fibril or possibly of subunits released in nonfibrillar form. To verify this finding, Western blots of type VI collagen revealed a doublet at 50–75 kd, which may represent 1 or all of the type VI collagen subunit fragments detected by MS at greater intensity in the injury samples compared with the untreated control samples (Figure 3B). Further, the band at 150 kd likely corresponds to the subunits of the full-length type VI collagen fibril, indicating increased release following injury, either by increased synthesis or by compression-induced mechanical disruption of the matrix. Together, these data support an increased release of type VI collagen, with a possible increase in subunit proteolysis, in response to injury.

COMP release and degradation were also suggested following mechanical injury, with peaks at 40 kd and 60 kd compared with a peak at ∼100 kd in the cytokine-treated and untreated media (Figure 4A), indicative of the release of full-length protein. An immunoblot also suggested COMP degradation with injury, although the fragments were of a higher molecular weight (70–90 kd) (Figure 4B). Evidence of release of a 10-kd fragment with cytokine treatment was also noted on Western blot analysis (results not shown) and is consistent with known proteolysis (27). Additional signs of protein degradation with 1 or more of the treatments were seen with thrombospondin 1, type II collagen, aggrecan, and fibronectin (results not shown).

Figure 4.

Cartilage oligomeric matrix protein (COMP) degradation in medium from bovine cartilage explants subjected to injurious compression, treatment with cytokines (interleukin-1β [IL-1β] or tumor necrosis factor α [TNFα]), or no treatment. A, Summed peptide ion intensity for COMP plotted against the molecular weight (determined by molecular weight marker and protein constituents). The shift in the total ion intensity to a lower molecular weight with mechanical injury suggests that COMP may be degraded in response to injury. B, Western blot analysis of anti-COMP in medium obtained on day 5 from cartilage explants from 4 different animals (each pair of control and compression-injured samples in adjacent lanes). The ∼70-kd band is more pronounced in the samples subjected to mechanical compression, suggestive of COMP degradation with injury. Molecular weight markers are shown at the left. Color figure can be viewed in the online issue, which is available at http://www.arthritisrheum.org.

DISCUSSION

The purpose of the study was to characterize and compare, at the protein level, the response of cartilage tissue to injurious compression and to treatment with the proinflammatory cytokines TNFα and IL-1β, all of which are known to cause cartilage damage. Equal volumes of medium from explants subjected to injurious compression, IL-1β, or TNFα were probed for differences in protein composition and degradation as compared with untreated controls. Because of the small sulfated glycosaminoglycan–associated variations in electrophoresis of the samples, the stochastic nature of MS analyses, and the small differences in surface area between one 6-mm–diameter explant and four 3-mm–diameter explants, a strict quantitative analysis of these data is not possible. However, a comparative study focusing on large differences is useful and warranted.

In mass spectrometry, the total ion intensity is the sum of the extracted ion chromatogram values for each of the peptide peaks identified for a given protein, and it provides a semiquantitative means of comparison to identify differences at the protein level (17). Proteins released into the medium over the 5 days of culture represent a combination of new protein synthesis and secretion as well as active and passive loss of preexisting proteins within the cartilage tissue (Figure 1). New protein synthesis and secretion in response to treatment is often associated with a greater rate of release compared with untreated proteins than that of preexisting proteins, the latter likely released at a rate equal to or less than aggrecan.

As expected, the most abundant proteins released from cartilage included those comprising the ECM and, particularly, the pericellular matrix (PCM) (Table 1), including types VI, II, XII, and XI collagen, as well as proteoglycans, aggrecan, perlecan, biglycan, decorin, lumican, and chondroadherin. Other structural matrix proteins released were COMP, matrilin 1, matrilin 3, fibronectin, thrombospondin 1, and tenascin C.

Among the novel cartilage proteins identified were CD109, PDGFR-like, angiopoietin-like 7, PGRP-L, and the recently identified SCRG-1, which were found in multiple samples, except for PGRP-L, which was found only with TNFα treatment. CD109 is a 170-kd glucose-6-phosphate isomerase–linked protein that was recently identified as a member of the α2-macroglobulin and the complement C3, C4, and C5 gene family, which are proteins that contain activated thioesters (28). The similarity of CD109 to α2-macroglobulin suggests that it may serve as a protease inhibitor in cartilage or may possibly play a role in complement activity. No biologic role of PDGFR-like protein has been reported beyond its similarity to the PDGF receptor; however, we may speculate that the release of this protein may be a way of regulating the PDGF pathway in cartilage. Angiopoietin-like 7 was identified as a product of cornea stromal cells; it may serve to inhibit angiogenesis or promote phenotype maintenance (29). PGRP-L is capable of binding to and degrading peptidoglycans, which form the cell wall of bacteria, thus decreasing the inflammatory response to peptidoglycans (30). The cartilage protein SCRG-1 is expressed by mesenchymal stem cells as they differentiate into a more chondrocytic phenotype and may play a role in chondrogenesis (31). The presence of these novel cartilage proteins suggest that chondrocytes may play a role in regulating pathways as diverse as angiogenesis, chondrogenesis, and innate immunity.

The protein repertoires identified following IL-1β and TNFα treatment were similar (Table 2), including secretion of CHI3L1, CHI3L2, MMP-3, complement factor B, haptoglobin, and SAA3. CHI3L2 may be elevated in the presence of OA (32), whereas CHI3L1 was found to be elevated after IL-1 treatment in rat cartilage explants but not in human cartilage explants (32, 33). The addition of CHI3L1 counters the effects of IL-1, suggesting that it may be tissue-protective (34). Complement factor B is elevated in OA cartilage as compared with normal cartilage (18). Lactoferrin plays a role in innate immunity, and while it has not been reported to be expressed by chondrocytes, it has been identified in synovial joints together with lysozyme, RNase 7, and β-defensin 2, and β-defensin 3 (35).

SAA3, haptoglobin, and α1-acid glycoprotein (AGP) are acute-phase proteins that were increased by IL-1β and TNFα treatment. Human SAA, similar to bovine SAA3, is elevated in the synovial fluid of patients with RA (17), and SAA3 may be stimulated by IL-1β treatment of cartilage in vitro (19). In addition, in vitro treatment of cartilage with SAA increases catabolic processes, which may play an active role in cartilage degradation (36). Similarly, elevation of haptoglobin levels is associated with increased severity of RA, and AGP glycation varies in the synovial fluid of RA patients as compared with normal synovial fluid, suggesting that there may be local expression in the joint (17, 37). While elevated levels of SAA3, complement factor B, AGP, and haptoglobin in synovial fluid have been attributed to a systemic inflammatory response and synthesis by the liver in response to IL-6, IL-1, and TNFα (38), these results indicate that chondrocytes contribute to their production and may increase local concentrations within cartilage and synovium, which may be important in disease processes and immunity.

In addition to secreted proteins, membrane proteins CD14 and VCAM-1 were released into the medium in response to IL-1β and TNFα treatment, respectively. Previous studies showed increased chondrocyte expression of VCAM-1 following IL-1 and TNFα treatment (39), as well as elevated levels of VCAM-1 in synovial fluid from the joints of patients with RA (40). CD14 is expressed by chondrocytes in monolayer cultures (41) and may be important in enhancing the lipopolysaccharide response of tissues that otherwise express no or low copy numbers of CD14 (42).

The profile of proteins in medium from compression-injured explants consisted of a much higher proportion of intracellular proteins, suggesting cell death (Table 2). Loss of cell membrane integrity and concomitant release of intracellular proteins were further verified by Western blot analysis of actin release following injury (Figure 2A). Release of intracellular proteins may occur with apoptosis or necrosis via loss of membrane integrity. Previous studies using this mechanical injury model implicate apoptosis as the primary mechanism of cell death, as determined by both TUNEL assay and electron microscopy (2, 43). However, the release of endoplasmic reticulum and heat-shock proteins, as observed in our experiments, is more common during necrosis than apoptosis (44). Loss of membrane integrity with apoptosis is likely to occur over time during the 5-day culture period, while mechanical cell lysis is likely to occur at the time of injury, with intracellular proteins released rapidly in culture. Western blot analysis showed that actin release occurred primarily at early time points (Figure 2B), suggesting a role of mechanical cell lysis.

Spatial differences in local tissue deformation upon compression injury may explain the differences in cell death seen with the different methods studied. In most chondrocyte apoptosis studies by light or electron microscopy, the cut edges of explants are typically excluded from characterization, since cell death by apoptosis/necrosis is known to occur at such edges as a result of cutting alone, which complicates the interpretation of the role of loading (2, 45). Results from confined and unconfined compression injury models suggest that cell death, as determined by Live/Dead cell assay, is most abundant at the explant surface or around the periphery of the explant, where the strain is highest (4–6, 46). Chondrocytes nearest to the loading surface in a repetitive, load-controlled confined-compression study were considered to be dead by Live/Dead cell assay, but stained negative by TUNEL assay, suggesting necrotic cell death (47).

Thus, while cell death in the interior portion of mechanically injured cartilage explants appears to be dominated by apoptosis (2, 43), we hypothesize that a population of cells at the surface, on the periphery, and around the occasional blood vessel found in these immature cartilage explants may experience mechanical disruption in response to the high levels of strain. However, we cannot rule out the possibility that a population of cells may leak intracellular proteins, but may nonetheless appear to undergo apoptosis when examined by TUNEL assay or electron microscopy. Our results extend those of previous studies using this compression injury model by suggesting that there may be spatial, or at least population, differences in the chondrocyte response to the mechanical injury.

The final goal of this work was to identify evidence of increased matrix protein proteolysis fragments released into the medium, which is enabled using the SDS-PAGE-LC-MS/MS method. Inflammatory cytokines are known to increase the release of a number of proteases that can degrade matrix structural proteins, such as aggrecan, collagens, COMP, and matrilins. However, much less is known about whether matrix damage by mechanical compression injury leads to matrix proteolysis and remodeling. Evidence of injury-induced degradation of type VI collagen included an increase in the ∼75-kd region from type VI collagen subunits, as determined by MS and as verified by Western blotting (Figures 3A and B). The size of these fragments are unlikely to result from new type VI collagen subunit processing and suggest a possible proteolysis fragment similar in size to those generated from treatment of type VI collagen with MMP-9 or with a number of serine proteases (48, 49). Correlation of load-induced cell death with type VI collagen staining intensity in the PCM showed that a 1-hour compression of cartilage by 1-MPa stress caused a generalized flattening of the chondrocyte PCM in the superficial zone of immature compared with mature tissue (50). This flattening of the PCM may be associated with damage to the type VI collagen network, which may suggest that PCM injury with compression may lead to increased release of type VI collagen and type VI collagen fragments in the present study.

MS analysis indicated signs of COMP degradation (Figure 4A) with injury and possibly with IL-1β treatment. COMP release from cartilage is known to be elevated in response to exercise and to in vivo injury (51, 52), and COMP levels in serum and synovial fluid correlate with progression of arthritis (53). COMP fragments (50–90 kd) have also been observed in synovial fluid within 2 months after anterior cruciate ligament tear, but not in synovial fluid from the healthy contralateral knee joint (54). Our study suggests that injurious compression may cause the release of COMP fragments from cartilage, which may serve as a source of COMP fragments in synovial fluid in vivo. We also used molecular weight information from SDS-PAGE to identify proteolysis of thrombospondin 1, type II collagen, fibronectin, and aggrecan (results not shown), many of which are known to undergo proteolysis in response to 1 or more treatments.

In summary, we found that cytokines promote the release of proteins that are involved in inflammation and stress response, including acute-phase and complement proteins. In contrast, compression injury caused the loss of cell membrane integrity and the release of intracellular proteins, which may cause changes in gene expression and/or trigger apoptosis in neighboring cells. Release of matrix proteins and protein fragments following compression injury suggests damage to the PCM and ECM, consistent with reported decreases in tissue compression and shear stiffness (2, 6, 14, 21, 43). Damage to the matrix may also lead to changes in cell–matrix interactions, which could be responsible for apoptosis, for changes in gene expression, or for the decrease in the ability of the remaining viable cells to up-regulate biosynthesis in response to anabolic loading (2, 6, 14, 21, 43). We previously reported that several specific responses to injurious mechanical compression and cytokine treatment in vitro, using the same immature bovine cartilage as used here, were also observed in adult human cartilage (7). Ongoing studies using the present proteomics approach with adult human cartilage should further reveal similarities and differences with age and species.

AUTHOR CONTRIBUTIONS

Dr. Grodzinsky 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 design. Stevens, Grodzinsky, Tannenbaum.

Acquisition of data. Stevens, Wishnok, Chai.

Analysis and interpretation of data. Stevens, Wishnok, Chai, Grodzinsky, Tannenbaum.

Manuscript preparation. Stevens, Wishnok, Chai, Grodzinsky, Tannenbaum.

Statistical analysis. Stevens.

Acknowledgements

We thank Dr. Vadiraja Bhat for technical assistance.

Ancillary