Proteomic analysis of articular cartilage vesicles from normal and osteoarthritic cartilage




Articular cartilage vesicles (ACVs) are extracellular organelles found in normal articular cartilage. While they were initially defined by their ability to generate pathologic calcium crystals in cartilage of osteoarthritis (OA) patients, they can also alter the phenotype of normal chondrocytes through the transfer of RNA and protein. The purpose of this study was to analyze the proteome of ACVs from normal and OA human cartilage.


ACVs were isolated from cartilage samples from 10 normal controls and 10 OA patients. We identified the ACV proteomes using in-gel trypsin digestion, nanospray liquid chromatography tandem mass spectrometry analysis of tryptic peptides, followed by searching an appropriate subset of the Uniprot database. We further differentiated between normal and OA ACVs by Holm-Sidak analysis for multiple comparison testing.


More than 1,700 proteins were identified in ACVs. Approximately 170 proteins satisfied our stringent criteria of having >1 representative peptide per protein present, and a false discovery rate of ≤5%. These proteins included extracellular matrix components, phospholipid binding proteins, enzymes, and cytoskeletal components, including actin. While few proteins were seen exclusively in normal or OA ACVs, immunoglobulins and complement components were present only in OA ACVs. Compared to normal ACVs, OA ACVs displayed decreases in matrix proteoglycans and increases in transforming growth factor β–induced protein βig-H3, DEL-1, vitronectin, and serine protease HtrA1 (P < 0.01).


These findings lend support to the concept of ACVs as physiologic structures in articular cartilage. Changes in OA ACVs are largely quantitative and reflect an altered matrix and the presence of inflammation, rather than revealing fundamental changes in composition.

Articular cartilage vesicles (ACVs) are 50–150-nm membrane-bound extracellular organelles that are found in normal articular cartilage (1). They were initially characterized in reference to their role in the pathologic mineralization of cartilage, in studies which mirrored studies of matrix vesicles derived from growth plate cartilage and other normally mineralizing tissues (2). ACVs concentrate enzymes, ions, and substrates necessary for mineral formation (1). Isolated ACVs generate pathologic calcium-containing crystals identical to those found in arthritic human joints (1, 3). Articular cartilage, however, does not typically undergo matrix mineralization, except in pathologic conditions such as osteoarthritis (OA) (4).

While a primary role for ACVs in pathologic mineralization seems plausible, the high quantity of ACVs in normal healthy articular cartilage remains puzzling (5). Few structures in nature have only a single pathologic function, and the energy expenditure required for the formation of ACVs is unlikely to be wasted. It has been postulated that in growth plate cartilage, matrix vesicles may participate in matrix repair, in addition to matrix mineralization (6). We recently demonstrated that ACVs contain RNA (7), like other types of extracellular vesicles (8). ACVs specifically transfer their labeled RNA and protein to intact naive primary chondrocytes via simple coculture. Importantly, exposure of normal chondrocytes to small quantities of intact ACVs induces markers of chondrocyte hypertrophy, such as those seen in OA cartilage (7). Thus, during early OA, ACVs may be released from the matrix by matrix-degrading enzymes and interact directly with chondrocytes to promote chondrocyte hypertrophy.

The contents and functions of ACVs, however, remain poorly elucidated. It is not known whether ACVs, like growth plate matrix vesicles, are formed through zeiotic blebbing (9). It has also been suggested that ACVs are products of stressed or apoptotic cells (10), and would thus be significantly altered in OA cartilage. Proteomic analysis of exosomes (11) and several types of growth plate matrix vesicles (9, 12) revealed important information relevant to the functions and mechanisms of formation of these vesicles. In this study, we characterized the ACV proteome and compared the proteomes of ACVs derived from OA and normal human articular cartilage.



Human OA hyaline articular cartilage was obtained from deidentified, discarded pathologic specimens at the time of surgery for total knee replacement for OA (n = 10). None of the specimens contained visible crystal deposits in the cartilage. Snap-frozen normal adult human cartilage from knees of adult donors with no clinical joint disease (n = 10) was obtained from the National Disease Research Interchange and the Musculoskeletal Transplant Foundation. All visible cartilage was cleaned of adherent bone and stored at –70°C until use. Previous work has demonstrated that there are no significant differences between ACVs derived from fresh or from frozen cartilage (13). All experiments with human tissue were approved by the Institutional Review Boards of the Zablocki Veteran Affairs Medical Center, Milwaukee and the Medical College of Wisconsin.

ACV isolation.

ACVs were isolated from whole cartilage as previously described (1). Briefly, hyaline articular cartilage was minced and weighed under sterile conditions. Cartilage pieces were incubated in Dulbecco's modified Eagle's medium with 0.1% hyaluronidase (1 ml/gm wet weight cartilage) for 5 minutes to remove surface hyaluronate, and for 10 minutes with 0.5% trypsin (1 ml/gm cartilage). Trypsin inhibitor (0.2% soybean trypsin inhibitor, 1 ml/gm cartilage) was added to inactivate any remaining trypsin. All incubations were performed at 37°C with 5% CO2 with stirring.

After washing, cartilage pieces were incubated for 45 minutes with 0.2% bacterial collagenase type II (2.5 ml/gm cartilage; Worthington). Additional media was added so that the final collagenase concentration was 0.05% in a total of 10 ml media/gm cartilage, and cartilage was incubated overnight with stirring. The mixture was filtered and centrifuged at 500g for 15 minutes to remove cells, then at 37,000g for 15 minutes to remove large cell fragments and organelles. The supernatant was then centrifuged at 120,000g for 60 minutes to pellet the ACV fraction. ACVs were resuspended in Hanks' balanced salt solution and protein concentrations were determined by Lowry assay (14). Electron microscopy of ACVs isolated in this manner has revealed a biphasic population of membrane-bound vesicles, 50–150 nm in diameter, that are morphologically distinguishable from cellular debris (15), and are identical to ACVs that have spontaneously elaborated into the conditioned media of healthy primary chondrocyte monolayers (16). Their small size and lack of DNA (7) distinguish them from apoptotic bodies.

Proteome analysis.

Protein from each ACV preparation (50 μg) was polymerized into a gel in the cap of an Eppendorf tube by adding 100 μl acrylamide/bis (30% acrylamide:2.67% bis-acrylamide), 2 μl of 10% ammonium persulfate, and 2 μl TEMED (17). Gel pieces were transferred to the corresponding Eppendorf tubes in 1 ml of 40% methanol and 7% acetic acid, and incubated for 30 minutes. Gel pieces were washed in 50% acetonitrile in water, followed by 50% acetonitrile in 100 mM ammonium bicarbonate (pH 8.0). Gels were then dried under vacuum using a Savant SpeedVac. Two hundred microliters of ammonium bicarbonate (100 mM) (pH 8.0) containing 1 μg trypsin (Promega) was added to each gel piece and incubated overnight at 37°C. Each gel piece was then extracted twice with 70% acetonitrile in 0.1% formic acid, and the extracts of each gel were pooled together and dried. Twenty microliters of GuHCl (6M) in 5 mM potassium phosphate and 1 mM dithiothreitol (pH 6.5) was added to each dried sample, sonicated, and peptides were extracted using a C18 ZipTip (Millipore). The extracted peptides were collected and dried. Five microliters of formic acid (0.1%) in MS-grade water containing 5% acetonitrile was added to each sample.

Samples were subjected to nanospray liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis using an LTQ mass spectrometer (Thermo-Fisher), coupled to a Surveyor high-performance liquid chromatography system equipped with a Micro AS auto sampler (Thermo-Fisher). The instrument was interfaced with an Aquasil, C18 PicoFrit capillary column (75 μm × 10 cm; New Objective). The mobile phases consisted of 1) 0.1% formic acid containing 5% acetonitrile (solvent A) and 2) 0.1% formic acid in 95% acetonitrile (solvent B). A 180-minute linear gradient was used. The ions eluted from the column were electrosprayed at a voltage of 1.75 kV. The respective proteomes were determined from searching the MS/MS data against the human subset of the Uniprot database.

Bioinformatic analysis.

Data were analyzed using Visualize software (written by Dr. Brian Halligan, Medical College of Wisconsin, Milwaukee). Proteins were considered definitely present if the search results showed at least 2 matched peptides per protein and a false discovery rate (FDR) of ≤5%. Proteins that satisfied the above criteria were further subjected to Holm-Sidak analysis for multiple comparison testing in order to detect those present at significantly different levels between groups. P values less than 0.05 were considered significant.


More than 1,700 different gene products were identified in one or more of the 20 ACV samples. Approximately 170 proteins were represented by >1 peptide and having an FDR of ≤5%, which satisfied our stringent criteria. We eliminated several proteins that resulted from blood contamination (including hemoglobin and unique red cell proteins), which is unavoidable in these specimens. Table 1 lists the remaining proteins, including their accession numbers, number of peptides identified per protein, scan count, protein coverage, and the pI of the proteins.

Table 1. The articular cartilage vesicle proteome*
Accession no.DescriptionPeptide countScan count% coveragepI
  • *

    This table represents a subset of the identified proteins that were represented by at least 2 peptides and had a false discovery rate of ≤5%. Peptide count refers to the number of peptides unique to the protein. Scan count is the number of times that one or more of the peptides unique to the protein were counted. Peptide coverage is the percentage of the protein covered by the peptides present. pI refers to the isoelectric point of the protein.

 P68032α-cardiac actin85920.4245.25
 P68104Elongation factor 1-alpha 147912.5548.81
 P06396Gelsolin precursor66011.7655.954
 O00560Syntenin 12138.3897.042
 P04264Cytokeratin 13107.7767.792
 Q05639Elongation factor 1-alpha 23787.3438.817
 P07437Tubulin β chain224.5054.788
 P04350Tubulin β-4 chain224.5054.788
 Q13885Tubulin β-2A chain224.4944.788
 Q9BVA1Tubulin β-2B chain224.4944.788
 P68371Tubulin β-2C chain224.4944.8
 Q13509Tubulin β-3 chain224.4444.835
 P02760α1-microglobulin/bikunin precursor223.6935.98
 Q9BQL6Kindlin 1222.9545.966
 Q9UPN3Microtubule-actin crosslinking factor 1270.7735.284
Cytoplasmic enzymes     
 P61626Lysozyme C precursor (EC
 Q06830Peroxiredoxin-1 (EC
 P04406GAPDH (EC
 P06733α-enolase (EC
 P14618M1/M2 isozymes of pyruvate kinase (EC
 P00338L-lactate dehydrogenase A chain (EC
 P32119Peroxiredoxin-2 (EC
 P00558Phosphoglycerate kinase 1 (EC
 P62937Peptidyl-prolyl cis-trans isomerase A (EC
 P12724Eosinophil cationic protein precursor (EC 3.1.27.-)2316.87510.142
 P60174Triosephosphate isomerase (EC
 Q9UKU9Angiopoietin-like 274814.4027.137
 P18669Phosphoglycerate mutase 1 (EC
 Q8N0Y7Probable phosphoglycerate mutase 4 (EC
 Q9UKZ9Procollagen C–proteinase enhancer 246711.3258.241
 P50148Guanine nucleotide–binding protein G(q)2210.1985.612
 O95837G-protein α subunit 1422010.1415.842
 P34096Ribonuclease 4 precursor (EC 3.1.27.-)2129.5248.801
 P13929β-enolase (EC
 P09104γ-enolase (EC
 P04899Adenylate cyclase–inhibiting G α protein2377.6275.362
 P06744Glucose-6-phosphate isomerase (EC
 Q05524α-enolase, lung specific (EC
 P28330Long-chain acyl-CoA dehydrogenase (EC
 Q6UXX5Inter-α inhibitor H5–like protein6714.6468.821
 Q96PE3Type I inositol-3,4-bisphosphate 4-phosphatase (EC
 P178586-phosphofructokinase, liver type (EC
 P13637Na+/K+ ATPase 3 (EC
 P05023Na+/K+ ATPase 1 (EC
Extracellular matrix     
 P10915Proteoglycan link protein261,11070.3397.003
 O15335Chondroadherin precursor2153052.3689.203
 P21810Biglycan precursor1936850.0007.087
 P12109Collagen α1(VI) chain precursor475,60937.5495.311
 P20774Mimecan precursor (osteoglycin)99133.2215.478
 P51888Prolargin precursor151,00132.4619.189
 P12111Collagen α3(VI) chain precursor1088,07130.0696.458
 O75339Cartilage intermediate-layer protein 1 precursor411,42228.0418.156
 P49747Cartilage oligomeric matrix protein precursor181,68626.2884.352
 Q92743Serine protease HtrA1 precursor (EC 3.4.21.-)1117525.6257.525
 P12110Collagen α2(VI) chain precursor366,97625.6135.836
 O43854Integrin-binding protein DEL-198325.0006.917
 P08493Matrix Gla protein precursor2923.3019.497
 P02751Fibronectin precursor3676421.5005.48
 P04004Vitronectin precursor926218.6195.587
 P02649Apolipoprotein E precursor58815.7735.659
 P51884Lumican precursor41814.7936.226
 P07585Decorin precursor239.4718.296
 P01009α1-antitrypsin precursor258.8525.401
 Q8IUL8Cartilage intermediate-layer protein 2 precursor81187.8727.992
 P02671Fibrinogen α-chain precursor61277.3905.745
 P16112Aggrecan core protein precursor255,5297.0814.11
 P01023α2-macroglobulin precursor5156.9206.067
 P02458Collagen α1(II) chain precursor82315.0788.227
 Q14055Collagen α2(IX) chain precursor3414.0648.976
 P00488Coagulation factor XIIIA chain precursor (EC
 P07996Thrombospondin 1 precursor4423.3334.728
 P20849Collagen α1(IX) chain precursor252.9328.488
 Q03692Collagen α1(X) chain precursor2232.6479.492
 P35443Thrombospondin 4 precursor26562.3934.451
 P24821Tenascin precursor (TN)232.1814.8
 P20908Collagen α1(V) chain precursor221.4694.948
 Q96QB0Collagen α2(V) chain precursor251.3346.123
 P13611Versican core protein precursor2160.9724.436
 P35555Fibrillin 1 precursor280.5574.817
Growth factors     
 Q16674Melanoma-derived growth regulatory protein precursor31728.2448.564
 P55107Bone morphogenetic protein 3b precursor265.0219.329
 O75888CD256 antigen294.8009.445
 Q8NI99Angiopoietin-like 6222.7668.281
Inflammatory components     
 P01834Igκ chain C region32348.1135.627
 P02743Serum amyloid P–component precursor1239734.9786.161
 P01857Igγ 1 chain C region65423.9397.952
 P02647Apolipoprotein A-I precursor41221.3485.59
 P02748Complement component C9 precursor53212.5225.448
 P60827Complement C1q TNF-related protein 8 precursor2511.4509.667
 P01859Igγ 2 chain C region3611.0437.299
 P05090Apolipoprotein D precursor269.5245.062
 P06727Apolipoprotein A-IV precursor4108.5865.297
 P04220Igμ heavy chain disease protein257.9285.139
 P01871Igμ chain C region256.8286.403
 P01031Complement C5 precursor8296.6236.172
 P01860Igγ 3 chain C region245.8627.296
 P01861Igγ 4 chain C region235.8107.037
 P01877Igα 2 chain C region265.0005.773
 P01876Igα 1 chain C region264.8166.143
 P07357Complement component C8 α chain precursor234.7956.118
 Q9BXJ3Complement C1q tumor necrosis factor–related protein 4 precursor254.5597.933
 P13671Complement component C6 precursor252.6776.407
 P07355Annexin II2643368.3437.302
 P04083Annexin I1739157.3916.726
 Q08431Lactadherin precursor (milk fat globule/endothelial growth factor VIII)171,00154.2647.877
 P08758Annexin V1842048.9034.932
 P14555Phospholipase A2, membrane precursor (EC
 P04271S100 protein, β chain38545.0554.523
 P09525Annexin IV76233.3335.873
 P06702S100 A9 protein2728.0705.78
 P08133Annexin VI118623.5125.44
 P6310414-3-3 protein ζ/δ2420.0004.728
 P10909Clusterin precursor637916.4815.945
 P215895′-nucleotidase precursor (EC
 P27105Protein 7.2b32914.6347.589
 P60903S100 A10 protein2314.4336.827
 O15232Matrilin 3 precursor59612.7576.312
 Q9HCJ1Progressive ankylosis protein homolog (ANK)2210.9767.558
 O75131Copine III3396.3315.631
 P50995Annexin XI2136.1397.254
 P43007Neutral amino acid transporter A (SATT)274.5115.938
 Q96FN4Copine II2174.3805.761
 P62805Histone H476556.86311.762
 P84243Histone H3.32528.88911.649
 P28001Histone H2A (10 subtypes)32321.70511.202
 O60814Histone H2B (11 subtypes)2817.60010.134
 Q00056Homeobox protein Hox-A4 (Hox-1D) (Hox-1.4)2556.2509.764
 O60687Sushi repeat–containing protein SRPX2 precursor78120.2156.888
 O75340Probable calcium-binding protein ALG-238017.8015.17
 O43293Death-associated protein kinase 3 (EC
 Q7Z7G0Target of Nesh-SH3 precursor (Tarsh)5255.3959.252
 P08582Melanotransferrin precursor3384.6075.715
 Q9H4G0Band 4.1–like protein 1 (neuronal protein 4.1)2102.9515.455
 Q9Y2E4Disco-interacting protein 2 homolog C231.9287.023
 P62834Ras-related protein Rap-1A precursor31113.5876.321
 P61224Ras-related protein Rap-1B precursor31113.5875.639
 P29992Guanine nucleotide–binding protein α 11 subunit35110.0285.538
 P17252Protein kinase C α type (EC
 Q9UJ30Protein kinase C β type (EC

The ACV proteome.

As shown in Table 1, ACV proteins fell into several categories, as delineated by cellular location and function. Extracellular matrix (ECM) proteins, including collagens, proteoglycans, and other matrix proteins, were prominent among the types of proteins present. The presence of multiple chains of type VI collagen and several members of the small leucine-rich proteoglycan family confirmed the pericellular location of ACVs. Other ECM proteins, such as DEL-1, are of particular interest, as their function in cartilage has not been thoroughly investigated and their association with ACVs may have important implications for understanding their roles in cartilage. Key chondrocyte growth factors, including bone morphogenetic protein 3 (BMP-3), angiopoietin-related protein, and melanoma inhibitory protein (cartilage-derived retinoic acid–sensitive protein [CD-RAP]), were also present in ACVs.

The presence of phospholipid-binding, integral membrane proteins and some signaling pathway components supports the notion of the membrane-rich composition of ACVs. The annexin family of proteins is particularly prominent in the ACV proteome, and is also well represented in the growth plate matrix vesicle proteome (9, 18). ANK, a putative pyrophosphate transporter, is present in ACVs. Milk fat globule/endothelial growth factor VIII, also known as lactadherin, is also a prominent component of exosomal vesicles (19). It has been observed in articular cartilage (20), but lacks an assigned function. Clusterin participates in membrane recycling and cell adhesion (21), and copine III is a ubiquitously distributed protein that may also have a kinase activity (22). Cell-signaling proteins, such as the Ras proteins, are common to several types of extracellular vesicles (11). Isoforms of protein kinase C were also present in ACVs.

Cytoskeletal components, including actins and actin-capping factors, were present in ACVs, supporting the theory that ACVs bud from chondrocyte microvilli (9). Various forms of tubulin were also present. Enzymes in the ACV proteome included coagulation factor XIIIA, a transglutaminase enzyme previously found in ACVs, which may have a role in mineralization (23), and peptidyl-prolyl cis-trans isomerase, which is involved in protein folding, signal transduction, trafficking, assembly, and cell cycle regulation. Peroxiredoxin belongs to the class of antioxidant enzymes (24).

While several nuclear histones were present, proteins from mitochondria and other intracellular organelles were not seen in the ACV proteome. Specifically, tetraspanins, which are characteristic markers of multivesicular bodies from which exosomes are generated (11), were also absent from the ACV proteome.

Differences between OA and normal ACVs.

While the majority of proteins were shared by ACVs from normal and OA cartilage, a few were seen exclusively in either OA ACVs or normal ACVs (Table 2). Six proteins were found only in normal ACVs, while 9 proteins were seen exclusively in OA ACVs. Many of the proteins that were exclusive to OA, such as fibrinogen, complement, immunoglobulins, and apolipoproteins, are typical markers of inflammation. Interestingly, chondrocytes are capable of generating apolipoprotein A-I, and the N-terminal proteolytic product of this protein comprises the amyloid deposits found in aging knee menisci (25). Immune complexes and complement have been found in human OA cartilage (26) and in a rabbit model of OA (27). While these inflammatory components could certainly originate from synovial fluid, chondrocytes are capable of synthesizing components of the classical complement pathway, including C1, C2, and C4 (28).

Table 2. Comparison of the proteomes of articular cartilage vesicles derived from normal and osteoarthritic cartilage*
Accession no.DescriptionTotal peptidesTotal coverageNormal run fractionOA run fractionNormalized ratioLog normalized ratioNormalized PH-S
  • *

    Proteins that were significantly different between groups by Holm-Sidak analysis for multiple comparisons are included. Total peptides refers to the number of unique peptides belonging to any one protein. Total coverage refers to the percent of the protein covered by the peptides. Normal run fraction is the number of runs that contained peptides characteristic of that protein in the normal articular cartilage vesicles (ACVs). Osteoarthritis (OA) run fraction is the number of runs that contained peptides characteristic of that protein in OA ACVs. The normalized ratio is the ratio of total scan count in the OA group:normal group, taking into account the number of runs for each specimen group. H-S is the P value for the normalized Holm-Sidek analysis for multiple comparisons.

OA only         
 P02671Fibrinogen α-chain precursor55.080402.29 × 10−330
 P02649Apolipoprotein E precursor515.770602.87 × 10−240
 P01857Ig γ 1 chain C region623.940504.89 × 10−136.16 × 10−11
 P01031Complement C5 precursor86.620203.20 × 10−73.65 × 10−5
 P60174Triosephosphate isomerase (EC × 10−62.24 × 10−4
 P01834Ig κ chain C region348.110303.78 × 10−64.08 × 10−4
 P215895′-nucleotidase precursor (EC × 10−43.00 × 10−2
 P00738Haptoglobin precursor36.400203.01 × 10−43.03 × 10−2
 P06727Apolipoprotein A-IV precursor48.590205.69 × 10−44.89 × 10−2
Normal only       
 Q9UKU9Angiopoietin-related protein 2 precursor611.566005.51 × 10−157.11 × 10−13
 P68104Elongation factor 1-alpha 1412.552002.10 × 10−62.29 × 10−4
 Q14055Collagen α2(IX) chain precursor34.061006.10 × 10−56.45 × 10−3
 Q16674Melanoma-derived growth regulatory protein precursor328.243001.43 × 10−41.47 × 10−2
 P13611Versican core protein precursor20.972003.36 × 10−43.31 × 10−2
 P57053Histone H2B (11 subtypes)217.601003.36 × 10−42.95 × 10−2
Up-regulated in OA       
 P02768Serum albumin precursor816.58106016.214.023.98 × 10−94.70 × 10−7
 Q15582Transforming growth factor β–induced protein ig-h3 precursor2039.095010013.933.808.94 × 10−860
 O43854Integrin-binding protein DEL-1925.0050704.182.065.75 × 10−86.67 × 10−6
 P02743Serum amyloid P–component precursor1234.98401003.301.722.83 × 10−250
 P04004Vitronectin precursor918.6280902.811.495.93 × 10−147.53 × 10−12
 Q92743Serine protease HtrA1 precursor (EC 3.4.21.-)1125.6370902.301.204.84 × 10−75.42 × 10−5
 P12111Collagen α3(VI) chain precursor10830.071001002.001.001.59 × 10−1890
 P12109Collagen α1(VI) chain precursor4739.011001001.940.952.58 × 10−1210
 P12110Collagen α2(VI) chain precursor3628.361001001.790.845.72 × 10−1190
Down-regulated in OA         
 P10915Proteoglycan link protein2670.341001000.68−0.568.25 × 10−111.01 × 10−8
 Q08431Lactadherin precursor (MFG-E8)1754.261001000.64−0.651.43 × 10−121.78 × 10−10
 P02458Collagen α1(II) chain precursor84.8480500.56−0.851.21 × 10−51.29 × 10−3
 P10909Clusterin precursor616.48100800.52−0.953.38 × 10−104.09 × 10−8
 P26038Moesin712.3180900.50−1.011.45 × 10−41.49 × 10−2
 P21810Biglycan precursor1950.00100800.48−1.064.17 × 10−125.17 × 10−10
 P51888Prolargin precursor1532.46100900.43−1.211.11 × 10−380
 O75339Cartilage intermediate-layer protein 1 precursor4128.041001000.41−1.293.17 × 10−610
 P16112Aggrecan core protein precursor257.081001000.40−1.332.08 × 10−2430
 O15335Chondroadherin precursor2155.431001000.39−1.364.52 × 10−260
 Q92954Lubricin87.26100500.33−1.582.45 × 10−113.01 × 10−9
 Q8IUL8Cartilage intermediate-layer protein 2 precursor87.8770600.33−1.622.49 × 10−82.91 × 10−6
 O60687Sushi repeat–containing protein SRPX2 precursor720.2260300.32−1.651.37 × 10−61.52 × 10−4
 P49747Cartilage oligomeric matrix protein precursor1826.29100900.27−1.905.21 × 10−1420
 P02751Fibronectin precursor3621.501001000.24−2.081.88 × 10−750
 P35443Thrombospondin 4 precursor22.39100900.21−2.261.67 × 10−730
 P35241Radixin47.3860300.19−2.422.00 × 10−72.30 × 10−5
 O15232Matrilin 3 precursor512.7680200.14−2.878.07 × 10−161.00 × 10−13
 P29992Guanine nucleotide–binding protein α 11 subunit310.0340100.14−2.899.34 × 10−59.76 × 10−3
 P15311Ezrin34.4460200.10−3.279.34 × 10−101.11 × 10−7
 P07996Thrombospondin 1 precursor43.3360100.09−3.423.32 × 10−73.75 × 10−5
 P98160Perlecan62.1650100.09−3.517.08 × 10−108.50 × 10−8

A small group of proteins were present only in normal ACVs. This heterogeneous group of proteins included angiopoietin-related protein, a member of the tumor necrosis factor α family, which may play a role in cartilage development (29). Elongation factor 1-alpha 1 participates in protein translation and microtubule formation. One chain of type IX collagen was also present. Melanoma inhibitory protein (CD-RAP) was present only in normal ACVs, as was versican, a large proteoglycan that is present in normal articular cartilage (30).

We used analytic methods specifically designed to detect significant differences in protein levels between groups in order to analyze quantitative differences between proteins in ACVs from normal and OA cartilage. Proteins found in higher levels in normal ACVs than OA ACVs included ECM proteins, such as type II collagen, and several proteoglycans (including biglycan, proline/arginine-rich end leucine-rich repeat protein, aggrecan, and perlecan), as well as cartilage oligomeric matrix protein, fibronectin, and thrombospondin. Interestingly, lubricin levels were also significantly higher in normal ACVs than in OA ACVs. This finding is consistent with observed losses of lubricin in OA cartilage (31). Levels of cartilage intermediate-layer proteins 1 and 2 were also higher in normal ACVs than in OA ACVs. It is expected that the matrix-destructive milieu of OA might decrease quantities of these important proteins. Levels of some members of the actin-binding family of proteins, such as moesin, radixin, and ezrin, were also higher in normal ACVs than in OA ACVs. These findings are less readily explained.

Levels of 9 proteins were significantly increased in OA ACVs. A dramatic increase in transforming growth factor β (TGFβ)–induced protein βig-H3 was noted, supporting the idea of a putative role for TGFβ in OA. Levels of DEL-1 were also increased in OA ACVs. The functions of DEL-1 in cartilage and its involvement in OA are largely unstudied. Vitronectin binds the αvβ3 integrin, and is present in the pericellular matrix of fetal human articular cartilage. Levels of vitronectin were also higher in OA ACVs. Some evidence points to a role for ligands of the αvβ3 integrin in modulating interleukin-1β expression in chondrocytes (32). Serine protease HtrA1 levels were increased in OA ACVs, and this protease has recently been implicated in aggrecan degradation in OA (33). Increased levels of type VI collagen were observed in OA ACVs, and have also been noted in OA cartilage (34).


We characterized the proteome of human ACVs, and found 170 proteins that satisfied our stringent criteria for inclusion in the more closely examined group. This group of proteins certainly does not represent an exhaustive list of ACV proteins. Several classes of proteins known to be present in or on ACVs, such as integrins, discoidin domain receptors (35), plasma cell membrane glycoprotein 1, and alkaline phosphatase, were represented in the full proteome, but were not present in this select group. In general, integral membrane proteins are often underrepresented in proteomic studies. One reason is the relative paucity of lysine and arginine, which are the cleavage sites of trypsin (36). Additional studies to confirm the presence or absence of proteins on the full list will be necessary.

The ACV proteome reflects important structural information about these organelles. ECM proteins represented the most prevalent protein type in ACVs. The concentration of pericellular ECM proteins in ACVs supports findings from histologic studies of their pericellular location (37). The presence of cell adhesion proteins, such as fibronectin, vitronectin, and numerous small leucine-rich proteoglycans, suggests that ACVs are firmly rooted in matrix. The number of phospholipid-binding and membrane-associated proteins demonstrates the importance of membrane in this organelle. The presence of actin and its associated cytoskeletal components in the ACV proteome suggests that, similar to matrix vesicles, ACVs are likely formed from zeiotic blebbing of cell membrane microvilli (12).

The ACV proteome also reflects potential functions of ACVs. Interestingly, the presence of several potent mineralization inhibitors, including matrix Gla protein, TGFβ-induced 68-kd protein/TGFβ-inducible gene h3, and serine protease HtrA1, supports the hypothesis that mineralization may not be the primary function of ACVs. Several ACV-associated ECM proteins mediate collagen fibril formation, and their presence may contribute to a matrix repair function for ACVs, such as that proposed for growth plate matrix vesicles (6). ACVs also possess some cell-signaling machinery, and could possibly respond to external signals. Growth factors, such as BMP-3, and regulatory proteins, including CD-RAP and factor XIIIa, may contribute to their ability to modulate the chondrocyte phenotype (38).

The ACV proteome differs from that of other extracellular membrane–limited particles and shares some characteristics with the proteomes of matrix vesicles that have been derived from growth plate cartilage (12). However, some unique proteins noted in the matrix vesicle proteome, such as aquaporin, glycoprotein HT7, and scavenger receptor type B, were not detected in the ACV proteome. Interestingly, matrix vesicles derived from an osteoblastic cell line had fewer proteins in common with ACVs (9), suggesting that these immortalized bone cells may produce very different extracellular vesicles. The ACV proteome also shares fewer similarities with exosomal proteomes. Lactadherin and ubiquitin are common to both ACVs and exosomes (39). The heterotrimeric G proteins, Hsp70 and Hsp90, and members of the tetraspanin family that are particularly characteristic of exosomal vesicles (11) were not seen in ACVs.

We were able to compare and relatively quantify proteins in ACVs derived from OA with those from normal cartilage. Surprisingly, few proteins were unique to one or the other type of ACV. These findings support the hypothesis that ACVs are constitutively produced by chondrocytes, and are not preferentially generated by damaged or dying cells. ACVs are clearly affected by the environment in which they reside. Differences in ECM proteins in OA and normal ACVs largely reflect known changes in OA ECM and increased catabolism. The presence of complements and immunoglobulins in OA ACVs warrants further consideration. While proteomic techniques would seem a logical methodology to investigate important changes in cartilage proteins in OA, cartilage is quite refractory to 2-dimensional gel-based protein analysis (40). This is due to the high concentrations of strongly charged proteoglycans which run anomalously on sodium dodecyl sulfate polyacrylamide gels, and to the highly crosslinked state of cartilage matrix, which is typically not soluble in sample buffers (41).

Our studies are not without limitations. While all the cartilage was from adult subjects, we were unable to age-match the OA and normal samples, and aging-related changes may confound the observed differences between OA and normal ACVs. Further studies will be necessary to address the effect of age on ACVs. Additionally, because of the stringent inclusion criteria, we have only discussed a small group of the proteins found in ACVs. Certainly, trypsin-resistant proteins, or proteins with low prevalence and/or low molecular weight, may be underrepresented with these techniques (36). All of this work was done with collagenase-released ACVs, and it is possible that the enzyme exposure required to isolate ACVs from whole cartilage might alter their protein profiles. For example, Xiao and colleagues showed some decrease in ECM proteins in collagenase-released matrix vesicles, compared to those that are released into conditioned media (9). The ACVs we studied exhibited multiple ECM proteins, but a direct comparison of the proteome of ACVs released nonenzymatically into chondrocyte-conditioned media and those obtained enzymatically from whole cartilage was not possible, given the limitations of frozen cartilage samples.

In conclusion, ACVs contain many proteins that reflect their possible functions and mechanisms of production. The present results reinforce the physiologic nature of ACVs. The demonstrated association of a few proteins, such as DEL-1 and lactadherin, with ACVs may have ramifications for understanding their role in cartilage. The observed differences in OA ACVs, particularly the presence of inflammation markers, provide additional support for a role of inflammation in OA. Further analysis of the ACV proteome will lead to a better understanding of the important roles of ACVs in cartilage physiology and disease.


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. Rosenthal 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. Rosenthal, Gohr, Ninomiya.

Acquisition of data. Rosenthal, Gohr, Ninomiya, Wakim.

Analysis and interpretation of data. Rosenthal, Gohr, Ninomiya, Wakim.


We thank Dr. Brian Halligan for the Visualize software, and the Biotechnology and Bioengineering Center at the Medical College of Wisconsin for access to their computer cluster.