Increased versican content is associated with tendinosis pathology inthe patellar tendon of athletes with jumper’s knee

A. Scott1, Ø. Lian2,3, C. R. Roberts4, J. L. Cook5, C. J. Handley6, R. Bahr2, T. Samiric6, M. Z. Ilic6, J. Parkinson6,

D. A. Hart7, V. Duronio1, K. M. Khan1,8,9

1Department of Medicine, University of British Columbia, Vancouver, BC, Canada, 2Department of Sports Medicine, Oslo SportTrauma Research Center, Norwegian School of Sport Sciences, Oslo, Norway, 3Kristiansund Hospital, Kristiansund, Norway,4Centre for Blood Research and Faculty of Dentistry, University of British Columbia, Vancouver, BC, Canada, 5Centre for PhysicalActivity and Nutrition Research, Deakin University, Melbourne, Australia, 6School of Human Biosciences, La Trobe University,Melbourne, Australia, 7Departments of Surgery, Microbiology & Infectious Diseases, and Medicine, University of Calgary, Calgary,AB, Canada, 8Department of Family Practice, University of British Columbia, Vancouver, BC, Canada, 9Centre for Hip Health,Vancouver Coastal Health and Research Institute, Vancouver, BC, CanadaCorresponding author: Karim K. Khan, UBC Department of Family Practice, University of British Columbia, DavidStrangway Building, 320-5950 University Boulevard, Vancouver, BC, Canada V6T 1Z3. Tel: 604 827 4129, Fax: 604 827 4184,E-mail: kkhan@interchange.ubc.ca

Accepted for publication 2 August 2007

Expansion of the extracellular matrix is a prominent butpoorly characterized feature of tendinosis. The presentstudy aimed to characterize the extent and distribution ofthe large aggregating proteoglycan versican in patients withpatellar tendinosis. We obtained tendon from tendinopathypatients undergoing debridement of the patellar tendon andfrom controls undergoing intramedullary tibial nailing.Versican content was investigated by Western blotting andimmunohistochemistry. Microvessel thickness and densitywere determined using computer-assisted image analysis.Markers for smooth muscle actin, endothelial cells (CD31)and proliferating cells (Ki67) were examined immunohisto-chemically. Western blot analysis and immunohistochem-

ical staining revealed elevated versican content in theproximal patellar tendon of tendinosis patients(P5 0.042). Versican content was enriched in regions offibrocartilage metaplasia and fibroblast proliferation, aswell as in the perivascular matrix of proliferating micro-vessels and within the media and intima of arterioles.Microvessel density was higher in tendinosis tissue com-pared with control tissue. Versican deposition is a prominentfeature of patellar tendinosis. Because this molecule is notonly a component of normal fibrocartilagenous matrices butalso implicated in a variety of soft tissue pathologies, futurestudies should further detail both pathological and adaptiveroles of versican in tendons.

Tendinopathy is a common source of chronic mus-culo-skeletal pain (Hazleman et al., 2004). The histol-ogy of tendinosis – the underlying pathology in manycases of tendinopathy – is well described and certainhistopathological features of tendinosis appear to beconsistent across tendons (Khan et al., 1999), but themolecular changes in tendon composition have beeninvestigated less extensively. Imaging abnormalitiesinclude hypoechoic areas on greyscale ultrasound andhigh signal on magnetic resonance imaging (MRI),corresponding to an increase in glycosaminoglycans(Cook et al., 2000). Cross-sectional studies of tendonfrom competitive athletes have demonstrated thatsuch abnormal tendon morphology on imaging –even when not associated with symptoms – is asso-ciated with a 4.2-times greater risk of developing intosymptomatic tendinopathy compared with a sonogra-phically normal tendon (Cook et al., 2000). Thus,changes in the extracellular matrix may represent anearly step in the pathogenesis of tendinosis.

Unlike the pathology of osteoarthritis, which ischaracterized by a loss of glycosaminoglycan anddecreased DNA content, biochemical assays of ten-dinosis have demonstrated an increased presence ofglycosaminoglycan and DNA. Chondroitin sulfate isthe major glycosaminoglycan in tendinosis tissue andthe accompanying increase in DNA content has beenprimarily attributed to fibroblasts and vascular cells(endothelial and smooth muscle cells) (Khan et al.,1996; Fenwick et al., 2002). Therefore, the excessivechondroitin sulfate found in biochemical assays oftendinosis could be produced by resident fibroblasts(tenocytes) or by vascular cells, or both (Riley, 2005).Furthermore, the core proteins to which the glyco-saminoglycan is likely bound have not yet beenexamined to our knowledge.Versican is the most abundant of the large chon-

droitin sulfate proteoglycans in tendon (Yoon& Halper, 2005). Immunohistochemical studies ofnormal tendon and of collagenase-digested tenocyte

Scand J Med Sci Sports 2008: 18: 427–435 Copyright & 2007 The Authors

Journal compilation & 2007 Blackwell MunksgaardPrinted in Singapore . All rights reservedDOI: 10.1111/j.1600-0838.2007.00735.x

427

arrays have found versican in the endotendonaround vessels, and in the pericellular matrix aroundtenocytes (Bode-Lesniewska et al., 1996; Ritty et al.,2003). In an experimental model of subfailure liga-ment injury, versican mRNA in resident fibroblastswas reported to be acutely upregulated (Provenzanoet al., 2005). This finding may be consistent withversican’s role in creating a hydrated, anti-adhesivematrix that promotes cellular proliferation andmigration following injury (Kinsella et al., 2004).Versican expression is also upregulated in bovinetendon by in vitro compressive loading (Robbinset al., 1997), a mechanical stimulus which has beenhypothesized to occur at the inferior patellar poleduring loaded knee flexion (Almekinders et al., 2003;Hamilton & Purdam, 2004). Thus, both injury andaltered mechanical loading could lead to distinctpatterns of increased versican synthesis in tendon,and the distribution of versican in tendons mayprovide insight into the underlying mechanisms.Therefore, we aimed to characterize the quantity

and distribution of versican in the patellar tendon ofathletes presenting with tendinosis. We hypothesizedthat versican would be present in excess in patellartendinosis, and that it would be distributed in theextracellular matrix surrounding tenocytes, endothe-lial cells and/or smooth muscle-containing cells.

MethodsSubjects

Tendon tissue was obtained from patients with documentedpatellar tendinosis (n5 21) and from control participants(n5 10). The patients (18 males and three females) wereathletes who had experienced pain for longer than 3 monthsand were unable to participate in sport at their pre-injury level.The average Victorian Institute of Sports Assessment (VISA)score for patients was 42 (15–65), indicating significant func-tional limitation (05 very poor function and 1005perfectfunction and no pain) (Visentini et al., 1998). All patients inthe present study also had MRI abnormalities at the proximalaspect of the patellar tendon confirmed by a board-certifiedmusculoskeletal radiologist and a specialist orthopedic sur-geon. The control group (seven males and three females) wasselected from patients undergoing intramedullary nailing forlow-energy tibia fractures and who had no current or previousknee pain. None of the patients or controls had previoussurgical treatment in or around the knee, corticosteroidinjections in or around the knee, serious traumatic injuryaffecting the knee, or any rheumatic or degenerative kneecondition. The average age of subjects was 30 years (24–34years, n5 21) in the patient group and 29 years (19–43 years,n5 10) in controls. All subjects gave their written, informedconsent. The study was approved by the local Committees forResearch Ethics in Australia, Norway and Canada.

Biopsy procedure

The surgical technique and biopsy handling were identical inthe two groups. Biopsies were taken from the proximal bone–tendon junction, and the tendon tissue was excised using a full-

thickness wedge-shaped incision, widest at the patellar pole(1 cm) and narrowing distally (2–3 cm in length). Biopsies wereeither immediately frozen in liquid nitrogen and stored at� 801, or chemically fixed in Zamboni’s solution for 4–24 hand then washed in 0.1M phosphate-buffered NaCl, pH 7.2,with 15% sucrose (w/v) (PBS) and 0.1% natriumazide. Afterfixation, the biopsies were stored in PBS at 4 1C, dehydrated,paraffin embedded and serially sectioned at 5mm thickness.

We applied Bonar’s classification system (Khan et al., 1999)to confirm the presence of tendinosis in patients, and itsabsence in controls, using tendon sections stained with Alcianblue and hematoxylin and eosin (H&E). Distinct features oftendinosis (fibrocartilage metaplasia, neovascularization,fibroblastic hypercellularity) were independently identified bytwo investigators (A. S., J. L. C.) blinded to group. The inter-rater reliability for identifying these features was 100%.Patient biopsies that yielded tissue sections at least 1 cm2 insize (10 of 21 biopsies) were used for more extensive qualita-tive observations regarding the relation of versican expressionto features of tendinosis.

Detection and quantification of versican in tissue samples

Western blot methodology

Tendon samples were extracted with 4M guanidium chlorideand 50mM sodium acetate buffer (50mM sodium acetate,100mM disodium sulfate; pH 5.8) containing proteinaseinhibitors for 72 h at 4 1C. The protease inhibitor cocktailwas prepared as a 10 � concentrate and added to the bufferedguanidine hydrochloride solution to give final concentrationof 8mg/L soybean trypsin inhibitor (SBTI), 0.1M aminohex-anoic acid, 0.01M ethylenediaminetetra-acetic acid (EDTA),1.0mM benzamidine hydrochloride, 0.1mM phenylmethylsul-fonyl fluoride (PMSF), 2.0mM iodoacetic acid and 0.02% (w/v)sodium azide. The proteoglycans present in the extract wereisolated by ion-exchange chromatography on Q-Sepharose,deglycosylated with chondroitinase ABC and keratanase, andseparated by sodiumdodecyl sulfate-polyacrylamide electro-phoresis as described previously (Samiric et al., 2004). Hand-ling and processing of patient and control biopsy materialwere identical. The gels were then analyzed by Westernblotting using the LF99 antibody (a kind gift of Dr. LarryFisher, NIH) and 2B1 (Seikagaku, Japan) to versican. TheLF99 antibody was raised against a peptide corresponding tothe N-terminus of versican (LHKVGKSPPVRC) (Bernstein etal., 1995). This epitope is cleaved during versican maturation,and therefore represents newly synthesized proteoglycan. The2B1 antibody was raised against proteoglycans extracted froma human yolk sac tumor, and later identified as specificallyrecognizing an epitope close to the C-terminal domain in theversican core protein (Isogai et al., 1996). Samples equivalentto 2mg wet weight of tissue were applied to each well.

Quantitative immunohistochemistry

The distribution of versican was examined in tissue sectionsusing the 2B1 antibody followed by a commercially availableavidin–biotin–HRP visualization system (Vector Labs, VectorLaboratories, Burlington, Canada). Stained tissue sectionswere systematically scanned at �200 magnification by a tech-nician who was unaware of the labeling code using the BLISSmicroscope workstation (Bacus Laboratories, Lombard, Illi-nois, USA). Image Pro Plus (version 4.5.1.22 from MediaCybernetics, San Diego, California, USA) was used to high-light positive DAB end-reaction product using the same RGB

Scott et al.

428

threshold file for all sections. Values are expressed as theaverage number of positive pixels per field.

Labeling of proliferating cells and vascular cell types

Proliferation of endothelial and smooth muscle cells arekey events in angiogenesis – a potentially relevant featureof tendinosis (Kristoffersen et al., 2005). Therefore, weused monoclonal antibodies against markers for endothelialcells (CD31; DAKO JC70A, DAKO, Mississauga, Canada),smooth muscle actin (a-SMA; Biomeda 1A4, Biomeda, Bur-lingame, CA, USA) and proliferating cells (Ki67; BD B56, BDBiosciences, Mississauga, Canada) to label cell types in normaland patellar tendinosis tissue. CD31 (platelet endothelial celladhesion molecule) has been commonly used to identifymicrovascular endothelial cells (Biberthaler et al., 2003). a-SMA has been shown to identify the contractile cytoskeletonin the media of arterioles, as well as pericytes and myofibro-blasts (Skalli et al., 1986, 1989; Darby et al., 1990). Ki67 is anuclear protein expressed in proliferating cells, with peakexpression during the M phase (Landberg et al., 1990).

Determination of microvessel parameters

Microvessels (defined here as vessels with diameter o50 mm orsmaller) from the 10 most proximal viewing fields (�40objective, 0.35mm2 total area) were outlined using digitized,systematically scanned micrographs of H&E sections. Allindividual vessel profiles were manually traced in Image ProPlus and counted with the digital counting tool. Vessel densitywas defined as the number of profiles per millimeter square.Vessel thickness and lumen diameter were determined bymeasuring vessel profiles in the transverse plane. Thicknesswas defined by the external borders of the adventitia, whereasthe lumen diameter was defined by the internal borders of theintima. To obtain a sufficient number of vessels for analysis ofthickness and lumen (n5 303), both circular and oblique vesselprofiles were included as long as the plane of section passedcompletely through the vessel; diameters were thus obtainedusing the least diameter method, which minimizes measure-ment error due to oblique sectioning (Clarkson et al., 1980).

Statistical analysis

To compare the amount of versican immunostaining in con-trol and patient samples, the average number of positive pixelswas subjected to a t-test with significance pre-determined ata5 0.05, after ensuring the data were normally distributedand did not violate the assumption of equal variance (Levene’stest). Results are expressed as mean � SD. Vessel density wascompared in the same way. Vessel diameters were not nor-mally distributed, and the distribution appeared to differbetween patients and controls; therefore the non-parametricKolmogorov–Smirnov test was used to test the hypothesis thatthe distributions were drawn from different populations.Analysis was carried out using SPSS version 13.0 (SPSSInc., Chicago, USA).

ResultsComparison of versican expression in normal tendon andtendinosis

Western blot analysis of proteoglycan samplesextracted from patellar tendon samples demonstratedthat there were clearly increased levels of versican

present in pathological tissue (Fig. 1). Most of theversican labeled by the 2B1 antibody appeared to becatabolic products of the proteoglycan (Samiricet al., 2004). Labeling with the LF99 antibody wasprominent in gels obtained from patient samplesbut was barely detectable in normal tendon, andappeared to preferentially identify a high molecularweight form of versican.Semi-quantitative immunohistochemistry indepen-

dently confirmed that versican was present in bothnormal and pathological tendon, but was signifi-cantly increased in patients (2.63 � 105 � 9.0 � 104)compared with controls (1.97 � 105 � 6.0 � 104,P5 0.042).

Versican distribution in patellar tendinosis – qualitativeobservations

Tendons from patients with patellar tendinosisdemonstrated three main patterns of versican expres-sion (Fig. 2). A proximal–distal pattern of thesetendinosis features could be detected in nine of the10 patient biopsies, which had sufficient tissue forextensive qualitative observation (as described in‘Methods’).In the most proximal part of the tendon, there

were prominent regions of fibrocartilage (Zone 1).These regions extended into the tendon proper up toseveral millimeters in length and width. In Zone 1,versican was more prominent in tendinosis tissuethan in control tissue in the vicinity of more rounded,chondrocytic-appearing tenocytes, sometimes inalternating layers consistent with classic descriptionsof fibrocartilage.

250

98

6450

36

N P

148

N P

200

97

68

43

(a) (b)

Fig. 1. Detection of versican in normal and pathologicaltendon. Western blots of purified tendon proteoglycanextracts (�2 mg/lane) were probed with (a) 2B1 (versicanC-terminal domain) or (b) LF-99 (N-terminal domain).Lanes represent normal (N) and pathological (P) humanpatellar tendon samples. Molecular weights are shown inkilodalton. Similar results were obtained in gels from fivedifferent patients.

Excess versican in human patellar tendinosis

429

Bordering this zone distally in all cases was aregion of vascular hyperplasia and endothelial pro-liferation (Zone 2), in which versican was alsoenriched (Fig. 2).Moving further distally, Zone 3 was characterized

primarily by increased tenocyte density but otherwiseby normal tendon structure (Fig. 2). In these regions,a greater quantity of versican was detected, whichwas distributed evenly throughout the extracellularmatrix of the tendon proper, i.e., there was no focalincrease in versican in Zone 3.Finally, large regions of the anterior and distal

patellar tendon in patients with tendinosis werefrequently normal and demonstrated no general orfocal increase in versican (Fig. 2). As in uninjuredtendon from control samples, tendon from patellartendinosis patients contained versican predominantlyaround microvessels in the endotendon running long-

itudinally with the principal line of action of thetendon (Bode-Lesniewska et al., 1996).

Association of versican with vascular hyperplasia

In the hypervascular region (Zone 2), versicanformed an expanded glycosaminoglycan-rich matrixpenetrated by CD311 microvessels exhibiting posi-tive Ki67 nuclear labeling in the intima (Fig. 3(a–c)).In microvessels from both normal and tendinosissamples, versican expression was typically comple-mentary to a-SMA (i.e. absent from the tunicamedia) (Fig. 3(d,e)). However, in some larger arter-ioles of tendinosis biopsies, versican was present inassociation with medial and/or intimal hyperplasia(Fig. 4), giving the vessel walls an irregular, thick-ened appearance. In patellar tendinosis tissue sam-ples, a-SMA-positive fibroblasts (i.e. myofibroblasts)

Fig. 2. Representative micro-graphs demonstrating zones ofversican deposition in associationwith specific features of tendinosispathology; magnetic resonanceimaging (MRI) from a jumper’sknee patient is shown for orienta-tion. Zone 1 contains banded dis-tribution of versican (brownstaining) in association with chon-drocytic cells (Hematoxylin coun-terstain), bordered distally bymicrovascular hyperplasia (Zone2). Zone 2 is interspersed andbordered by tenocyte hyper-cellularity and a generalized in-crease in versican content(Zone 3). In surrounding normaltendon versican staining is fainterand is localized to the endotendon.Scale bar on micrographs5 50mm,on MRI5 1 cm.

Scott et al.

430

were present in areas of versican-rich tissue, oftenadjacent to proliferating microvessels (Fig. 5), butthis was not the case in normal tendon.

Quantitation of microvessel density and thickness

Vascular density was 5.6 � 9.7 vessels/mm2 in con-trols, and 15.6 � 16.7 vessels/mm2 in patients(P5 0.026). The distribution of vessel diameterswas significantly different in patients (P5 0.035)(Fig. 6) due to an increased presence of small-diameter microvessels (capillaries, venules and arter-ioles). The distribution of lumen diameters was notdifferent in patient and control samples.

Discussion

The major novel finding of this study is that eleva-tions in versican levels contribute to the increasedextracellular matrix volume in symptomatic patellartendinosis. Versican expression was associated withdistinct regions of fibrocartilage, hypervascularity,hypercellularity, as well as morphologically normaltendon. Thus, several distinct cell populations andprocesses appear to contribute to the expression ofversican in the patellar tendon with features oftendinosis. This finding extends previous work,which demonstrates increased expression of aggre-can, biglycan and type II collagen in tendinosislesions (Corps et al., 2006; Maffulli et al., 2006).

Fig. 3. Association of versicanwith vascular proliferation (a) dif-fuse versican labeling (brown)around an abnormally dense capil-lary bed within the tendon proper,with more intense labeling directlyadjacent to the vessels (arrow). (b)CD31 labeling endothelial cellsserial to (a). (c) Higher power viewof the same region showing Ki67labeling of endothelial cells in themicrovascular walls. (d) Intenseversican labeling (asterisk) in aregion of arteriolar proliferation.(e) Serial to (d), confirming thepresence of smooth muscle actin(brown) in the intima media. Scalebars5 50 mm.

Fig. 4. Versican distribution in ar-terioles from normal tendon andtendinosis tissue. (a) Normal arter-iole – versican is localized to thesurrounding connective tissue, andto a single layer of endothelial cells(tunica intima). (b) Abnormalarteriole – versican surrounds an irre-gularly thickened endothelial layerand extends into the smooth musclelayer (tunica media), and is alsomarkedly increased in the outerlayers. Scale bar5 50 mm.

Excess versican in human patellar tendinosis

431

Turning first to the zone of fibrocartilage at theinferior pole of the patellar tendon in patient samples(Zone 1), versican was more prominent in the matrixsurrounding rounded, chondrocytic cells. Directlyadjacent to the region of chondroid metaplasia, a

zone of hypervascularity (Zone 2) was consistentlyobserved in patient samples. This was characterizedby increased versican expression both in the perivas-cular matrix and within the vessel walls. This prox-imal–distal pattern of fibrocartilage bordered byvascular tissue has been noted previously in thesupraspinatus tendon (Archambault et al., 2007).Fukuda et al. (1990) examined en-bloc sections offull-thickness rotator cuff tears and reported chon-droid metaplasia in the proximal remnant of thetendon, whereas the distal torn end was characterizedby hypervascular granulation tissue. This is consis-tent with a hypothesis that fibrocartilage metaplasiamay initially develop as an adaptive response tocompression under the acromion, but that this adap-tation weakens the tendon’s ability to withstandtensile loads, thereby leading to microtears adjacentto the site of fibrocartilage formation. A similarprocess has been postulated for the tibialis posteriorwhere it is compressed by the medial malleolus(Petersen et al., 2004). These data will therefore beof interest to those who have hypothesized thatpatellar tendinopathy may have a compressive etiol-ogy (Almekinders et al., 2003). Fibrocartilage tendsto develop within areas of tendon (distinct from theenthesis) that are subjected to compression or shear,for example at the deep surface of wrap-aroundtendons such as the supraspinatus or tibialis poster-ior. However, fibrocartilage metaplasia is also some-times observed as a pathological dysregulation of cellphenotype, for example in scar tissue (Boor &Ferrans, 1985).

Fig. 5. Association of versicanwith microvascular hyperplasiaand myofibroblasts. (a) Versicanin association with an abnormalmicrovascular bed (arrow), andremodeling tissue (asterisk). (b)CD31 confirms that the upperstructure is vascular, whereas thelower structure is not. (c) Intenselabeling of smooth muscle actin(a-SMA) in the vascular structure.a-SMA is also positive in scatteredmyofibroblasts in the adjacentremodeling area (rectangle). (d)Higher power view of rectangulararea in (c) Scale bars5 50 mm.

Fig. 6. Quantitation of vessel diameters (a) and lumendiameters (b) in the proximal 0.35mm2 of the patellartendon. Morphometric analysis demonstrates increased pre-sence of microvessels in patellar tendinosis.

Scott et al.

432

Versican was increased in association with micro-vessels in two overlapping patterns. First, and mostcommonly, there were often dense beds of capillariesand/or arterioles, and the entire region demonstratedexcess staining for versican. Proliferating endothelialand smooth muscle cells could be identified in theseregions. This pattern is consistent with versican’s rolein facilitating proliferation and migration in angio-genesis. The second abnormality was the increasedpresence of versican in association with intimal andmedial hyperplasia of arterioles. We have recentlydocumented increased vascular endothelial growthfactor (VEGF) expression by endothelial cells inpatellar tendinosis tissue (A. S., O. L., R. B., V. D.,D. A. H., K. K., unpublished observations). Inaddition to causing endothelial cells to proliferate,VEGF has been reported to upregulate their expres-sion of versican (Cattaruzza et al., 2002). Takentogether, these findings suggest a possible causallink between the two findings and a potential roleof VEGF in causing or perpetuating the increasedversican expression in Zone 2.Neovessels were occasionally seen in proximity to

versican-rich tissue populated by myofibroblasts(a-SMA-positive fibroblasts). The presence of myofi-broblasts is in keeping with prior studies of tendino-sis (Khan et al., 1996) and of other healing softtissues (Schmitt-Graff et al., 1994), and their associa-tion with versican confirms the importance of thisproteoglycan in soft-tissue repair and remodeling(Roberts, 1995). The factors leading to increasedversican expression in healing wounds may includehypoxia (leading to hypoxia-induced-factor-1-a andVEGF expression), cytokines including transforminggrowth factor-b and PDGF-A and -B (Evanko et al.,1998, 2001; Nikitovic et al., 2006), and compressiveor tensile loads (Mao et al., 1998; Grande-Allenet al., 2004; Milz et al., 2005).A third zone of versican expression was typically

seen distal and/or superficial to Zones 1 and 2, andthis was characterized by a relatively even elevationin versican levels in association with increased teno-cyte density. This could be interpreted as evidenceof adaptation. In animals and in humans, exercisecan induce adaptive improvements in the materialproperties of the tendon, including increased stiffnessand cross-sectional area (Woo et al., 1980; Kjaer,2004; Yoon & Halper, 2005). Chondroitin sulfatecontent is correspondingly increased in the tendonsof animals subjected to treadmill running vs seden-tary controls (Woo et al., 1980; Banes et al., 1999;Kjaer, 2004; Yoon & Halper, 2005). Conversely,the increased versican in Zone 3 could represent asecondary effect due to diffusion of cytokines fromadjacent areas.In the current study, Western blot analysis demon-

strated that versican content in patient samples

consisted both of newly synthesized, high molecularweight forms and of lower weight catabolic products.The regulation of versican expression and turnover iscomplex and includes alternate splicing of mRNAfrom the versican gene to generate at least threepossible isoforms (Rahmani et al., 2006). Versicandegradation involves proteolytic cleavage into dis-tinct entities, and regulation of its catabolism byaggrecanases and matrix metalloproteases, and bytheir endogenous inhibitors (Kinsella et al., 2004;Rahmani et al., 2006). Future studies should there-fore examine these regulatory processes in relation tothe development of tendinosis pathology.In this study, we did not examine the distribution

of other potentially relevant components of theextracellular matrix (e.g., aggrecan, biglycan, dec-orin, lumican, collagens, fibronectin, hyaluronan). Inparticular, aggrecan is likely to contribute to theregions of fibrocartilagenous metaplasia observedhere (Milz et al., 2005). Aggrecan mRNA levelswere increased in Achilles tendinosis (Corps et al.,2006) and in the ruptured supraspinatus (Lo et al.,2005) whereas versican levels were unchanged com-pared to controls (Corps et al., 2006). The discre-pancy between findings at the protein and mRNAlevels in these studies is consistent with a report fromcartilage showing that versican protein can remain inthe matrix long after mRNA levels have declined(Matsumoto et al., 2006). The complex regulation ofversican and other matrix molecules in the develop-ment and progression of tendinosis clearly requiresfurther research.In summary, we found that versican is present at

elevated levels in human patellar tendinopathy. Ver-sican was enriched in distinct regions of the tendontissue, specificially in areas of fibrocartilage, thosewith fibroblast proliferation, as well as in the peri-vascular matrix of proliferating microvessels andwithin the media and intima of arterioles. We con-clude that future studies should address both patho-logical and adaptive mechanisms underlying theelevated versican levels detected in the present studiesof patellar tendinosis.

Perspectives

The current study deepens our understanding ofpatellar tendinosis and the clinical rationale for itstreatment. Formerly known as ‘‘tendinitis,’’ a lack ofknowledge regarding the underlying pathologicalprocesses has hampered the development of effectivetreatments (Maffulli et al., 1998; Kjaer, 2004; Riley,2004). In laboratory studies, the transformation ofnormal, tensile tendon to fibrocartilage, and viceversa, is related to the dynamic balance of compres-sive, shear and tensile forces experienced by teno-

Excess versican in human patellar tendinosis

433

cytes at specific regions in the tendon (Robbinset al., 1997; Malaviya et al., 2000; Milz et al.,2005). Tensile (eccentric) loading of the patellartendon using a decline board gives superior resultsto standard knee flexion exercises in the treatmentof patellar tendinosis (Young et al., 2005) and thisclinical success has been suggested to result fromincreased tensile vs compressive loading oftenocytes in the posterior, proximal tendon withthe decline board (Young et al., 2005). Therefore,appropriate tensile loads delivered in such a way asto minimize compression may be able to reverse theprocess of fibrocartilage metaplasia in the patellartendon. Future developments in imaging technologymay allow the dynamic visualization of versican andother matrix components in tendinopathy patients,and thus allow a direct assessment of the molecularresponse to rehabilitation.

Key words: versican, patellar tendon, tendinosis, ex-tracellular matrix.

Acknowledgements

This work was funded by grants from the Canadian Institutesof Health Research (CIHR; MOP-77551) and the Worker’sCompensation Board of British Columbia (grant RS0203-DG-13). The Oslo Sports Trauma Research Center hasbeen established at the Norwegian University of Sport &Physical Education through generous grants from the RoyalNorwegian Ministry of Culture, the Norwegian OlympicCommittee & Confederation of Sport, Norsk Tipping, andPfizer. Mr. Scott is recipient of a CIHR Post-Doctoral Fellow-ship. Dr. Hart is the Calgary Foundation-Grace GlaumProfessor in Arthritis Research and supported by IGH ofCIHR. Dr. Duronio is a Michael Smith Foundation forHealth Research Senior Scholar. Dr. Khan is a CIHR NewInvestigator.

References

Almekinders LC, Weinhold PS, MaffulliN. Compression etiology intendinopathy. Clin Sports Med 2003:22: 703–710.

Archambault JM, Jelinsky SA, Lake SP,Hill AA, Glaser DL, Soslowsky LJ.Rat supraspinatus tendon expressescartilage markers with overuse. JOrthop Res 2007: 25: 617–624.

Banes AJ, Horesovsky G, Larson C,Tsuzaki M, Judex S, Archambault J,Zernicke R, Herzog W, Kelley S, MillerL. Mechanical load stimulatesexpression of novel genes in vivo and invitro in avian flexor tendon cells.Osteoarthrit Cartilage 1999: 7:141–153.

Bernstein EF, Fisher LW, Li K, LeBaronRG, Tan EM, Uitto J. Differentialexpression of the versican and decoringenes in photoaged and sun-protectedskin. Comparison by immuno-histochemical and northern analyses.Lab Invest 1995: 72: 662–669.

Biberthaler P, Wiedemann E, Nerlich A,Kettler M, Mussack T, Deckelmann S,Mutschler W. Microcirculationassociated with degenerative rotatorcuff lesions. In vivo assessment withorthogonal polarization spectralimaging during arthroscopy of theshoulder. J Bone Jt Surg Am 2003:85-A: 475–480.

Bode-Lesniewska B, Dours-ZimmermannMT, Odermatt BF, Briner J, Heitz PU,Zimmermann DR. Distribution of thelarge aggregating proteoglycanversican in adult human tissues. J

Histochem Cytochem 1996: 44:303–312.

Boor PJ, Ferrans VJ. Ultrastructuralalterations in allylamine cardiovasculartoxicity. Late myocardial and vascularlesions. Am J Pathol 1985: 121: 39–54.

Cattaruzza S, Schiappacassi M,Ljungberg-Rose A, Spessotto P,Perissinotto D, Morgelin M, MucignatMT, Colombatti A, Perris R.Distribution of PG-M/versicanvariants in human tissues and de novoexpression of isoform V3 uponendothelial cell activation, migration,and neoangiogenesis in vitro. J BiolChem 2002: 277: 47626–47635.

Clarkson PM, Kroll W, McBride TC.Maximal isometric strength and fibertype composition in power andendurance athletes. Eur J Appl PhysiolOccup Physiol 1980: 44: 35–42.

Cook JL, Khan KM, Kiss ZS, PurdamCR, Griffiths L. Prospective imagingstudy of asymptomatic patellartendinopathy in elite junior basketballplayers. J Ultrasound Med 2000: 19:473–479.

Corps AN, Robinson AH, Movin T,Costa ML, Hazleman BL, Riley GP.Increased expression of aggrecan andbiglycan mRNA in Achillestendinopathy. Rheumatology (Oxford)2006: 45: 291–294.

Darby I, Skalli O, Gabbiani G. Alpha-smooth muscle actin is transientlyexpressed by myofibroblasts duringexperimental wound healing. LabInvest 1990: 63: 21–29.

Evanko SP, Johnson PY, Braun KR,Underhill CB, Dudhia J, Wight TN.Platelet-derived growth factorstimulates the formation of versican-hyaluronan aggregates and pericellularmatrix expansion in arterial smoothmuscle cells. Arch Biochem Biophys2001: 394: 29–38.

Evanko SP, Raines EW, Ross R, GoldLI, Wight TN. Proteoglycandistribution in lesions of athero-sclerosis depends on lesion severity,structural characteristics, and theproximity of platelet-derived growthfactor and transforming growthfactor-beta. Am J Pathol 1998: 152:533–546.

Fenwick S, Harrall R, Hackney R, BordS, Horner A, Hazleman B, Riley G.Endochondral ossification in Achillesand patella tendinopathy.Rheumatology (Oxford) 2002: 41:474–476.

Fukuda H, Hamada K, Yamanaka K.Pathology and pathogenesis of bursal-side rotator cuff tears viewed from enbloc histologic sections. Clin OrthopRelat Res 1990: 254: 75–80.

Grande-Allen KJ, Calabro A, Gupta V,Wight TN, Hascall VC, Vesely I.Glycosaminoglycans andproteoglycans in normal mitral valveleaflets and chordae: association withregions of tensile and compressiveloading. Glycobiology 2004: 14:621–633.

Hamilton B, Purdam C. Patellartendinosis as an adaptive process: a

Scott et al.

434

new hypothesis. Br J Sports Med 2004:38: 758–761.

Hazleman B, Riley G, Speed CA. Softtissue rheumatology. Oxford: OxfordUniversity Press, 2004.

Isogai Z, Shinomura T, Yamakawa N,Takeuchi J, Tsuji T, Heinegard D,Kimata K. 2B1 antigencharacteristically expressed onextracellular matrices of humanmalignant tumors is a large chondroitinsulfate proteoglycan, PG-M/versican.Cancer Res 1996: 56: 3902–3908.

Khan KM, Bonar F, Desmond PM,Cook JL, Young DA, Visentini PJ,Fehrmann MW, Kiss ZS, O’Brien PA,Harcourt PR, Dowling RJ, O’SullivanRM, Crichton KJ, Tress BM, WarkJD. Patellar tendinosis (jumper’s knee):findings at histopathologicexamination, US, and MR imaging.Victorian Institute of Sport TendonStudy Group. Radiology 1996: 200:821–827.

Khan KM, Cook JL, Bonar F, HarcourtP, Astrom M. Histopathology ofcommon tendinopathies. Update andimplications for clinical management.Sports Med 1999: 27: 393–408.

Kinsella MG, Bressler SL, Wight TN.The regulated synthesis of versican,decorin, and biglycan: extracellularmatrix proteoglycans that influencecellular phenotype. Crit Rev EukaryotGene Expr 2004: 14: 203–234.

Kjaer M. Role of extracellular matrix inadaptation of tendon and skeletalmuscle to mechanical loading. PhysiolRev 2004: 84: 649–698.

Kristoffersen M, Ohberg L, Johnston C,Alfredson H. Neovascularisation inchronic tendon injuries detected withcolour Doppler ultrasound in horseand man: implications for research andtreatment. Knee Surg SportsTraumatol Arthrosc 2005: 13: 505–508.

Landberg G, Tan EM, Roos G. Flowcytometric multiparameter analysis ofproliferating cell nuclear antigen/cyclinand Ki-67 antigen: a new view of thecell cycle. Exp Cell Res 1990: 187:111–118.

Lo IK, Boorman R, Marchuk L,Hollinshead R, Hart DA, Frank CB.Matrix molecule mRNA levels in thebursa and rotator cuff of patients withfull-thickness rotator cuff tears.Arthroscopy 2005: 21: 645–651.

Maffulli N, Khan KM, Puddu G. Overusetendon conditions: time to change aconfusing terminology. Arthroscopy1998: 14: 840–843.

Maffulli N, Reaper J, Ewen SW,Waterston SW, Barrass V. Chondralmetaplasia in calcific insertionaltendinopathy of the Achilles tendon.Clin J Sports Med 2006: 16:329–334.

Malaviya P, Butler DL, Boivin GP, SmithFN, Barry FP, Murphy JM, Vogel KG.An in vivo model for load-modulatedremodeling in the rabbit flexor tendon.J Orthop Res 2000: 18: 116–125.

Mao JJ, Rahemtulla F, Scott PG.Proteoglycan expression in the rattemporomandibular joint in responseto unilateral bite raise. J Dent Res1998: 77: 1520–1528.

Matsumoto K, Kamiya N, Suwan K,Atsumi F, Shimizu K, Shinomura T,Yamada Y, Kimata K, Watanabe H.Identification and characterization ofversican/PG-M aggregates in cartilage.J Biol Chem 2006: 281: 18257–18263.

Milz S, Benjamin M, Putz R. Molecularparameters indicating adaptation tomechanical stress in fibrous connectivetissue. Adv Anat Embryol Cell Biol2005: 178: 1–71.

Nikitovic D, Zafiropoulos A, Katonis P,Tsatsakis A, Theocharis AD,Karamanos NK, Tzanakakis GN.Transforming growth factor-beta as akey molecule triggering the expressionof versican isoforms v0 and v1,hyaluronan synthase-2 and synthesis ofhyaluronan in malignant osteosarcomacells. IUBMB Life 2006: 58: 47–53.

Petersen W, Hohmann G, Pufe T, TsokosM, Zantop T, Paulsen F, Tillmann B.Structure of the human tibialisposterior tendon. Arch Orthop TraumaSurg 2004: 124: 237–242.

Provenzano PP, Alejandro-Osorio AL,Valhmu WB, Jensen KT, Vanderby RJr Intrinsic fibroblast-mediatedremodeling of damaged collagenousmatrices in vivo. Matrix Biol 2005: 23:543–555.

Rahmani M, Wong BW, Ang L, CheungCC, Carthy JM, Walinski H,McManus BM. Versican: signaling totranscriptional control pathways. CanJ Physiol Pharmacol 2006: 84: 77–92.

Riley G. The pathogenesis oftendinopathy. A molecular perspective.Rheumatology (Oxford) 2004: 43:131–142.

Riley GP. Gene expression and matrixturnover in overused and damagedtendons. Scand J Med Sci Sports 2005:15: 241–251.

Ritty TM, Roth R, Heuser JE. Tendoncell array isolation reveals a previously

unknown fibrillin-2-containingmacromolecular assembly. Structure2003: 11: 1179–1188.

Robbins JR, Evanko SP, Vogel KG.Mechanical loading and TGF-betaregulate proteoglycan synthesis intendon. Arch Biochem Biophys 1997:342: 203–211.

Roberts CR. Is asthma a fibrotic disease?Chest 1995: 107: 111S–117S.

Samiric T, Ilic MZ, Handley CJ.Characterisation of proteoglycans andtheir catabolic products in tendon andexplant cultures of tendon. Matrix Biol2004: 23: 127–140.

Schmitt-Graff A, Desmouliere A,Gabbiani G. Heterogeneity ofmyofibroblast phenotypic features:an example of fibroblastic cellplasticity. Virchows Arch 1994: 425:3–24.

Skalli O, Pelte MF, Peclet MC, GabbianiG, Gugliotta P, Bussolati G, RavazzolaM, Orci L. Alpha-smooth muscle actin,a differentiation marker of smoothmuscle cells, is present inmicrofilamentous bundles of pericytes.J Histochem Cytochem 1989: 37:315–321.

Skalli O, Ropraz P, Trzeciak A,Benzonana G, Gillessen D, GabbianiG. A monoclonal antibody againstalpha-smooth muscle actin: a newprobe for smooth muscledifferentiation. J Cell Biol 1986: 103:2787–2796.

Visentini PJ, Khan KM, Cook JL, KissZS, Harcourt PR, Wark JD. The VISAscore: an index of severity of symptomsin patients with jumper’s knee (patellartendinosis). Victorian Institute ofSport Tendon Study Group. J Sci MedSport/Sports Med (Australia) 1998: 1:22–28.

Woo SL, Ritter MA, Amiel D, SandersTM, Gomez MA, Kuei SC, Garfin SR,Akeson WH. The biomechanical andbiochemical properties of swinetendons – long term effects of exerciseon the digital extensors. ConnectTissue Res 1980: 7: 177–183.

Yoon JH, Halper J. Tendonproteoglycans: biochemistry andfunction. J Musculoskelet NeuronalInteract 2005: 5: 22–34.

Young MA, Cook JL, Purdam CR, KissZS, Alfredson H. Eccentric declinesquat protocol offers superior results at12 months compared with traditionaleccentric protocol for patellartendinopathy in volleyball players. Br JSports Med 2005: 39: 102–105.

Excess versican in human patellar tendinosis

435