VERSICAN KNOCKDOWN REDUCES INTERZONE AREA DURING EARLY STAGES OF CHICK SYNOVIAL JOINT DEVELOPMENT

Abstract

Much has been learned regarding factors that specify joint placement, but less is known regarding how these molecular instructions are translated into functional joint tissues. Previous studies have shown that the matrix chondroitin sulfate proteoglycan, versican, exhibits a similar pattern of expression in the embryonic joint rudiment of chick and mouse suggesting conserved function during joint development. In the present study versican’s importance in developing joints was investigated by specific inhibition of its expression in the early joint interzone, tissue that gives rise to articular cartilages and joint cavity. In ovo microinjection of adenoviral shRNA constructs into the HH25 chick wing was employed to silence endogenous versican protein in developing appendicular joints. Results showed statistically significant (12–14%) reduction of non-chondrogenic elbow joint interzone area in whole mount specimens at HH36 in response to versican knockdown. Attenuated expression of key versican-associated molecules including hyaluronan, tenascin, CD44, and link protein was also noted by histochemical and immunohistochemical analysis. Versican knockdown also lowered collagen II expression in presumptive articular chondrocytes indicating possible delay in chondrogenesis. Results suggest that versican functions interactively with other matrix/cell surface molecules to facilitate establishment or maintenance of early joint interzone structure.

INTRODUCTION

Synovial joint formation within the embryonic limb begins with flattening and packing of non-chondrogenic mesenchymal cells to form the joint interzone (Craig et al. 1987; Archer et al. 2003). Joint interzones in the developing embryo are comprised of a central intermediate layer sandwiched between two chondrogenous laminae lining the epiphyseal ends of the long bone anlagen (Craig et al. 1987). Articular cartilages may be derived from cells of the intermediate zone as the outer layers differentiate into chondrocytes and add to the growing cartilage template (Archer et al. 1994; Ito and Kida 2000). Articular cartilages differ from other chondrocytes forming the long bone template in that the former retain their phenotype throughout the life of the organism, tend to be less mitotic, and pack more densely during embryonic joint development (Lufti 1974; Howlett 1979). Articular cartilage imparts strength and functional resilience to joints while resisting ossification, unlike skeletal cartilage tissue, which undergoes cellular hypertrophy, apoptosis and replacement by osteoblasts (Jurand 1965; Pacifici et al. 2005).

Synovial joint morphogenesis hinges strongly on the process of cavitation or the partitioning of the interzone tissue bridging joint-forming skeletal structures (Pacifici et al. 2005). Several studies have proposed mechanical movement as the cause of cavitation or maintenance of joint cavity structure (Fell and Canti 1934; Hamburger and Waugh 1940). Studies have also indicated that cavitation may be due to increased expression of hyaluronan and its cell surface receptor CD44 in the interzone (Craig et al. 1990; Archer et al. 1994; Edwards et al. 1994; Pitsillides et al. 1995). Indeed, oligosaccharide-mediated blockade of hyaluronan function in the interzone (Dowthwaite et al. 1998) and conditional deletion of hyaluronan synthase 2 (Matsumoto et al. 2009) compromised normal synovial cavity formation.

Synovial joint differentiation is critically dependent on the coordinated activity of several key signaling molecules and transcription factors. GDF5 is one of several paracrine signals that may be necessary for synovial joint initiation as its expression is detected in early chick and mouse joint interzones and even throughout early mesenchymal condensations (Storm and Kingsley 1996). Mutant mice lacking GDF5 showed joint deformities such as brachypodism and defects in overall skeletal growth (Storm et al. 1994, Settle et al. 2003). Wnt9a is expressed in the presumptive joint and its overexpression in chick embryo limbs drives ectopic expression of other interzone markers including GDF5, chordin and CD44 (Hartmann and Tabin 2001), providing strong evidence that Wnt signals govern initial joint positioning. Moreover, conditional ablation of β-catenin in embryonic limb resulted in absence of synovial joints (Guo et al. 2004). Collectively, these data indicate that Wnt9a signaling through the canonical β-catenin dependent pathway is sufficient for synovial joint specification and maintenance of the mesenchymal nature of interzone cells by preventing differentiation into chondrocytes (Hartmann and Tabin 2001; Guo et al. 2004).

Chondroitin sulfate proteoglycans (CSPGs), core proteins covalently linked to glycosaminoglycan (GAG) side chains, interact with numerous molecules via action of their GAG side chains or terminal domains (Ruoslahti 1989). The versican CSPG plays an important role in development of limb cartilage primordia and is highly expressed in the limb core matrix during chondrogenesis (Shinomura et al. 1990) and synovial joint formation (Shibata et al. 2003; Snow et al. 2005; Shepard et al. 2007). Versican’s role in the process of endochondral ossification may be facilitation of precartilage mesenchymal condensation (Williams et al. 2005; Kamiya et al. 2006; Shepard et al. 2008; Hudson et al. 2010), a requisite step for chondrogenic differentiation (Hall and Miyake 2000). Following initial differentiation of cartilage precursor tissues, versican is down-regulated within them, but continues to be highly expressed in the surrounding perichondrium and joint interzone (Shibata et al. 2003; Snow et al. 2005; Shepard et al. 2007)

Alternative splicing of versican leads to generation of four isoforms expressed in a tissue dependent and developmentally-regulated manner. All isoforms contain the globular N-terminal G1 and C-terminal G3 domains (Zimmermann and Rouslahti 1990). The full length versican isoform (V0) consists of two chondroitin sulfate (CS-α and CS-β) attachment sites, V1 contains CS-β and V2 only CS-α, while the V3 isoform contains neither GAG attachment site (Shinomura et al. 1993). The N-terminal G1 domain of versican consists of an immunoglobulin-like motif followed by two-link protein modules that bind hyaluronan, an interaction further stabilized by association with link protein to create a loose, highly hydrated extracellular environment (Le Baron et al. 1992, Lee et al. 1993; Wight 2002; Wu et al. 2005). The C-terminal G3 domain consists of a C-type lectin-like domain, two epidermal growth factor-like repeats and a complement regulatory domain and interacts with several extracellular matrix (ECM) molecules including fibrillin-1, fibulin-1, and tenascins (Aspberg et al. 1997; Olin et al. 2001; Isogai et al. 2002).

Previous studies showed conserved immunolocalization of versican across the joint interzone through progressive stages of chick and mouse limb development that eventually resolve to the immediate articular surface (Snow et al. 2005; Shepard et al. 2007; Capehart 2010). This dynamic pattern of versican expression leads to the hypothesis that versican may function in joint interzone formation and/or maintenance subsequent to its role in early stages of cartilage differentiation. A recent elegant study utilizing conditional prx1-Cre driven deletion of versican from mouse limb mesenchyme (Choocheep et al. 2010) described a limb phenotype in which distal joint formation was compromised, resulting in improper orientation of autopod joint structure. Because of widespread cartilage template defects including overall delay in chondrogenesis and dislocated regions of cartilage hypertrophy, it is possible that the joint phenotype could be attributed in part to versican deletion from the entire skeletal template in addition to specific effects on joint differentiation. In order to further address versican’s role in early stages of joint formation it would be advantageous to misexpress versican specifically in the presumptive joint structure. The present study was designed to extend our understanding of versican function in early synovial joint morphogenesis through use of recombinant adenoviral short hairpin RNAs (shRNA) targeted to the synovial joint rudiment in order to assess impact of versican knockdown on initial stages of joint development.

MATERIALS AND METHODS

Recombinant shRNA Adenoviral-shRNA Preparation

A shRNA template targeting chick versican (NM_204787) mRNA at nucleotides 320–338 (adeno-320; 5’- TATCTTCGAATCAAATGGT-3’) was selected using Ambion siRNA target finder (www.ambion.com/techlib/misc/siRNA_finder.html; Applied Biosystems, Austin TX). A second shRNA sequence targeting versican mRNA at nucleotides 5334–5352 (adeno-5334; 5’-AGCCTGACATGACTGCTTC-3’) was also utilized and overlapped a published siRNA sequence used previously to silence versican in vitro (nucleotides 5337–5355; Sheng et al. 2006). The −320- and −5334 target sequences are located in versican G1 and CS-β domains, respectively, with −320 targeting all isoforms and −5334 the V0/V1 isoforms predominant in limb. A non-coding sequence (5’-GACTTGTTACTGTTTCGAC-3’) was also used to construct a control adeno-shRNA. Oligonucleotides containing the 5’-TTCAAGAGA-3’ loop sequence were prepared (Integrated DNA Technologies, Coralville IA) and the pSilencer adeno 1.0-CMV system (Applied Biosystems, Austin TX) used to engineer replication incompetent shRNA adenoviral constructs. Oligonucleotide templates were ligated into the pSilencer shuttle vector and verified by sequencing.

For each recombinant shRNA adenovirus, corresponding shuttle and adenoviral-lacZ plasmids were linearized using Pac1 restriction endonuclease and co-transfected into low density HEK-293 cells (ATCC, Manassas VA) in DMEM-10% heat inactivated fetal bovine serum (FBS)-1% penicillin/streptomycin (Invitrogen, Carlsbad CA). After three rounds of expansion in the packaging HEK-293 line, cells were lysed, centrifuged, and adenoviruses purified using the Adeno-X Virus Mini Purification Kit (Clontech, Mountain View CA) according to manufacturer’s instructions. Titers (~5×10⁹ ifu/ml) were determined by adenoviral lacZ reporter staining following the Adeno-X Rapid Titer Kit (Clontech) protocol. Adenoviral aliquots were maintained at −70°C for long-term storage.

Determination of Gene Silencing by qRT-PCR

Chick embryonic fibroblasts (CEFs) were isolated from Hamburger & Hamilton (1951) stage 35 (HH35) dermal tissues in PBS and digested with 0.25% Trypsin/EDTA (Invitrogen) for 10 minutes at 37°C. Cells were centrifuged and media replaced with fresh DMEM/F12 containing 2% FBS. Three samples at 2×10⁶ cells were used for each adenovirus and infected at a MOI (multiplicity of infection) of 100. After infection, cells were seeded in 35mm plates and supplemented with 2 ml DMEM/F12-2% FBS. Media were replaced after 24 hours with fresh media containing 10% FBS. Total RNA was extracted from replicates after 2 days using the RNA aqueous-4 PCR kit (Applied Biosystems, Austin TX).

Superscript III First Strand Synthesis kit (Invitrogen) was used to reverse transcribe equivalent concentrations of total RNA from each of the three samples using oligodT priming. The cDNAs were diluted three-fold and used for real time PCR with 3 replicates run for each of the three samples. Primer sets utilized were: 320F: 5’-AACGTCAGTCCTTCCATGCT-3’, R: 5’-TGAATGGGTTGGAACAGACA-3’, 5334F: 5’-CCTTTTGAAAGCAACCCAGA-3’, R: 5’-TGTGCCAGAAGCCAAAGAAG-3’. Real-time PCR reactions (Applied Biosystems ABI 7300) used RT² Real-Time SYBR Green /Rox PCR master mix (SA Biosciences, Frederick MD) according to manufacturer’s instructions. β-actin primers (F: CACAGATCATGTTTGAGACCTT, R: CATCACAATACCAGTGGTACG; Accession # L08165; DeBoever et al. 2008) were used to amplify endogenous β-actin as an internal control. Threshold cycle data was analyzed relative to untreated control samples using MS Excel and followed Pfaffl (2001) for relative quantification calculations. Student t-test was also performed to evaluate statistical significance.

In ovo versican knock down in the developing elbow region at HH35 for 2 replicate sample pools (n=5) co-injected with adeno-320 + adeno-5334 at HH25 (see below) was also assessed relative to control adeno-shRNA. RNA was extracted from lysates utilizing the RiboPure Kit (Applied Biosystems), cDNA transcribed, and assayed by real-time PCR. Microarray analysis of replicate RNA samples relative to control shRNA treatment was also performed through the University of North Carolina Genomics Core (Chapel Hill, NC) using the 4X44K chick gene expression array (#G2519F, Agilent Technologies, Wilmington DE) with Lowess normalization and fold-change in expression determined for molecules relevant to the present study.

In ovo Microinjection

Fertilized chick eggs (SPF; Charles River, Wilmington MA) were incubated in a 37.5°C humidified egg incubator until HH25. Eggs were ‘windowed’ and eggshell and vitelline membranes carefully removed using tungsten needles. Presumptive joint regions of accessible wings (usually right wing) were injected with adeno-shRNA control, adeno-320, adeno-5334 or a 1:1 combination of adeno-320 and -5334 viruses (adeno-320 + adeno-5334). Uninjected wings served as matched contralateral controls (CLC). Each microinjection delivered approximately 7.5 × 10⁶ virions targeted to the joint region using pulled-glass needles inserted into a micromanipulator (No.778; Narishige International, East Meadow NY) connected to a pneumatic pump (Model 820; WPI, Sarasota FL). Following injection, windowed eggs were sealed with plastic tape and incubated until desired developmental stages were reached (HH34-36). Embryos were harvested from eggs and preserved using appropriate fixatives. BSL2 animal use procedures were approved by the ECU IACUC and IBC.

Whole Mount LacZ Reporter Histochemistry

Adenoviral β-galactosidase reporter staining was adapted from Kern et al. (2007). Briefly, embryos were fixed briefly in ice cold 4% paraformaldehyde, rinsed 3 times with PBS, and washed in 0.02% sodium deoxycolate, 0.01% Tergitol–type NP-40 in PBS overnight at 4°C. Following permeablization, specimens were incubated in 0.02% magnesium chloride, 0.1% potassium ferrocyanide, 0.1% potassium ferricyanide, and either 0.1% X-gal (Invitrogen) or Red Gal (Research Organics, Cleveland OH) in PBS at 37°C in the dark. Specimens were postfixed with 4% paraformaldehyde, washed and stored in PBS at 4°C or embedded in paraffin.

Whole Mount Alcian Blue/ Alizarin Red Histochemistry

Alcian blue/Alizarin red whole-mount histochemistry was used to detect gross morphological changes in elbow morphology at HH36 using a modification of Kuczuk and Scott (1984). Specimens were fixed in 95% ethanol overnight at room temperature, followed by overnight staining in 0.2 % alcian blue in acidified ethanol (pH 1.0). Embryos were destained in 95% ethanol, macerated with 2% potassium hydroxide (KOH) until translucent (usually 3–5 hours) and stained with 0.01% Alizarin Red in 1% KOH. Samples underwent serial KOH/glycerol washes (80:20, 60:40, 40:60, 20:80) and were stored in 100% glycerol.

Histo- and Immunohistochemistry on Tissue Sections

Paraformaldehyde-fixed paraffin (7 µm) or cryostat sections (10µm) were immunostained overnight at 4°C with primary antibodies: rabbit anti-chick versican (1:50 dilution; Zanin et al. 1999; Shepard et al. 2007, 2008), rabbit anti-phosphohistone H3 (Ser10) (1:100; Cell Signaling, Danvers MA) or monoclonal mouse IgG supernatants (1:5; Developmental Studies Hybridoma Bank, Iowa City IA) anti-chick collagen II (II-II6B3), tenascin (M1B4), CD44 (1D10), link protein (9/30/8A4), and TGF-B3 (active form; D-B3). Secondary antibodies used were fluorescein- or rhodamine-conjugated goat anti-mouse or rabbit antibodies (1:200; Cappel, MP Biomedicals, Santa Ana CA) for 2 hours at room temperature. Histochemical localization of hyaluronan utilized biotinylated hyaluronic acid binding protein (HABP) (1:100; Cape Cod, East Falmouth MA) and fluorescein -strepavidin (1:200; Vector Labs, Burlingame CA). Negative control samples were routinely processed with omission of primary reagents. Control specimens were also incubated with irrelevant monoclonal mouse IgG supernatants or non-immune rabbit IgG. Prior to primary reagent incubation sections were treated with citrate-based Antigen Unmasking solution (Vector Labs) per manufacturer’s instructions followed by 1mg/mL bovine testicular hyaluronidase (Sigma Chemical; St Louis MO), except those reserved for HABP or CD44 staining which were treated with 0.25U/mL chondroitin ABC lyase (Sigma). Samples were then blocked with 3% bovine serum albumin and 1% normal goat serum (Sigma). Immunostained sections were post-fixed in 80% and 50% ethanol, equilibrated in PBS, and mounted in 10% PBS-90% glycerol containing 100 mg/mL 1,4-diazabicyclo (2,2,2) octane (DABCO; Sigma). Sections were viewed using an Olympus BX-40 microscope equipped with epifluorescence optics and images captured using a SPOT-RT camera and software (Diagnostic Instruments, Sterling MI). Matched control and experimental images were processed with equal adjustment of fluorescence levels for brightness and/or contrast for the entire image.

TUNEL Assay

TUNEL (Terminal deoxynucleotidyl transferase dUTP nick end labeling) assay for apoptotic cells was performed using ApopTag Peroxidase In Situ Apoptosis Detection Kit (Chemicon, Temecula CA) according to manufacturer’s instructions. Briefly, deparaffinized sections were incubated in 20 µg/mL proteinase K 15 min at room temperature, washed and endogenous peroxide quenched in 3% hydrogen peroxidase. Sections were equilibrated in terminal transferase buffer and incubated 1 hour in reaction buffer containing TdT enzyme and digoxigenin-11-dUTP at 37°C. Following serial PBS washes, specimens were incubated in anti-digoxigenin peroxidase conjugate 30 min at room temperature. Specimens were PBS washed and incubated in diaminobenzidine substrate (Sigma) until characteristic brown cells were visible.

RESULTS

To better understand versican function during early stages of synovial joint development, we employed an adenoviral mediated shRNA interference strategy to facilitate site directed cleavage and degradation of mRNA transcripts. Quantitative real-time PCR assessment of gene silencing activity of adeno-shRNA constructs on chick embryonic fibroblasts (CEF) in vitro resulted in an average 70% knockdown in versican mRNA levels with adeno-320 shRNA virus and 50% knockdown with adeno-5334 shRNA after 2 days of infection as compared to untreated negative controls (Figure 1A). Both experimental CEF groups showed significant (p<0.05) knockdown of versican mRNA. Furthermore, in ovo co-injection of adeno-320 + adeno-5334 into presumptive chick elbows at HH25 led to an average reduction of 65% in versican mRNA levels four days following injection when compared with adeno-control shRNA (Figure 1B).

FIGURE 1.

Real time PCR evaluation of versican knockdown. A) Histogram representing reduction of versican mRNA in HH35 chick embryonic fibroblasts (CEF) in vitro two days following infection with versican shRNA adenoviruses. qRT-PCR results demonstrate 70% mRNA knockdown with the adeno-320 shRNA and 50% with adeno-5334 shRNA (3 biological replicates per treatment). P-values obtained from Student t-test indicate significant reduction in versican transcript levels compared to the untreated control CEF. B) Real-time PCR assessment of versican mRNA knockdown at HH35 in ovo following co-injection of presumptive elbow with a 1:1 combination of adeno-320 + adeno-5334 shRNA (combined 1 and 2) at HH25. Data show an average reduction of 65% in versican mRNA levels in ovo for combined adeno-320 + adeno-5334 injections (3 replicates per sample) as compared with adeno-shRNA control.

Sites of recombinant adeno-shRNA infection in the limb could be readily detected and verified by histochemical staining for the β-galactosidase reporter encoded by adenoviral constructs at HH34 (Figure 2), demonstrating that developing joint tissues could be reproducibly targeted by microinjection into presumptive joint at HH25. Different wing joint primordia were utilized for versican knockdown; however, reproducibility was best achieved through microinjection directed to the presumptive dorsal aspect of the elbow which could be easily targeted due to its characteristic flexure at HH25. Care was taken to compare contralateral control and experimental tissue sections taken from similar areas within the developing joint structure and elbow images were rotated such that the humerus was in a similar position in all figure panels to aid orientation.

FIGURE 2.

Adeno-control shRNA injection at HH25 does not reduce versican in the developing chick elbow joint at HH34. A) Versican immunolocalization in the developing elbow joint interzone in specimen treated with control adeno-shRNA. Note strong versican signal throughout the interzone ECM. B) Areas of adeno-control shRNA expression in panel A are indicated by lacZ reporter staining. C) Representative control section in which primary antibody was omitted. In other control preparations little or no specific immunoreactivity was observed following incubation with either irrelevant mouse IgG monoclonal supernatant or non-immune rabbit IgG. D) Higher magnification of matrix versican immunolocalization in humeroulnar and radiohumeral joint interzones in contralateral control (clc) elbow. E) Higher magnification of versican localization shown in boxed region of panel A. F) Corresponding higher magnification of lacZ staining shown in same area of panel B. h, humerus; r, radius; u, ulna. Scale bar in A=100µm for panels A–C and D=25µm for D–F.

Little or no difference in versican reactivity was noted with adeno-control shRNA treated samples relative to uninjected contralateral controls at HH34 (Figure 2A, B, D–F) with versican localization spanning the radiohumeral and humeroulnar joint interzone and presumptive articular surface ECM as reported previously (Shepard et al. 2007). Similar levels of versican reactivity between experimental and control specimens verified that adenoviral infection was without non-specific effects on veriscan expression and furthermore that the X-gal precipitate used to visualize sites of adenoviral infection did not interfere with immunodetection of antigens. Sections incubated with control immunoreagents showed little or no specific reactivity with joint tissues (Figure 2C).

In order to corroborate shRNA-mediated reduction in versican mRNA levels as gauged by qRT-PCR, adenoviral-shRNA knockdown of versican protein in developing joint tissues was assessed by immunostaining with chick versican antibody. Reduction in versican staining in HH36 synovial joint interzones and presumptive articular chondrocytes at sites of adenoviral infection relative to contralateral control wings was visible on paraffin sections of developing wrist (Figure 3A–C) or humeroulnar joint interzones (Figure 3D–F) microinjected with adeno-320 + adeno-5334. Consistent viral titers (approximately 7.5 × 10⁶ virions) were routinely injected during experiments. Some variability in versican immunoreactivity in adeno-320 and adeno-5334 treated specimens and lacZ reporter staining was observed, suggesting differences in adenoviral uptake and consequent versican reduction among individual specimens. As significant versican knockdown could be detected after two days in vitro and was effective for at least five days post-injection, analysis of effects on early joint structure in ovo was performed primarily between HH34-36 (e8-10), encompassing the period in which the joint interzone and joint line (site of incipient cavitation) are established.

FIGURE 3.

Versican immunolocalization in sections of the developing wrist/elbow interzone (A–F) and whole mount Alcian blue/Alizarin red histochemistry of wings (G–K) at HH36 injected with adeno-shRNA constructs at HH25. A) Versican localization in the contralateral control (clc) interzone ECM between ulna (u) and ulnar carpal (c). B) Versican is reduced (arrow) in the wrist joint interzone of adeno-320 versican shRNA treated wing. C) Same section as panel B shows adeno-320 expressing interzone cells (arrow) as detected by lacZ reporter histochemistry. D) Versican localization in the humeroulnar (h, u) interzone of the contralateral control wing. E) Versican immunoreactivity is reduced in regions of the elbow interzone (white arrowhead) injected with adeno-5334 shRNA. F) LacZ reporter staining showing areas of adeno-5334 expression that correlate with reduced versican in panel E. G) Control shRNA injected wing shows normal elbow morphology and interzone has similar dimensions to corresponding contralateral control (bracket). H) Adeno-5334 versican shRNA expression results in slight reduction of elbow interzone (dark arrow) area relative to control in this specimen. I) Co-injection of combined adeno-320 + adeno-5334 shRNAs (comb) results in dramatic alteration in elbow interzone spacing (dark arrowhead). J) Higher magnification of adeno-320 + adeno-5334 shRNA treated interzone in panel I. K) LacZ reporter histochemistry using RedGal shows adeno-320 + adeno-5334 shRNA expression (dark arrows) along interzone and developing articular surface of the humerus. Scale bars= 25µm A–C, 100 µm D–F, 1mm G–K.

Overall effects of shRNA transfection on synovial joint morphology in HH25 elbows were evaluated at HH36 using whole mount Alcian Blue and Alizarin Red histochemistry. Wings injected with the adeno-control shRNA virus showed little or no difference in interzone area when compared to their contralateral counterparts (Figure 3G). Relative to adeno-control and uninjected contralateral limbs, a generalized decrease in elbow interzone spacing in greater than 50% of viable HH36 specimens was observed with adeno-320, adeno-5334, and co-injected adeno-320 + adeno-5334 shRNAs. Effects ranged from relatively minor (Figure 3H) to more severe diminution of elbow interzone area (Figure 3I, J). LacZ reporter staining of whole mount specimens at HH36 again showed successful delivery of adenoviral constructs to the central and outer (chondrogenic) laminae of the elbow interzone at HH25 (Figure 3K). Estimated interzone area between epiphyseal ends of the humerus, radius, and ulna was performed using the area measurement tool of Image J software (NIH, Bethesda MD). An overall 14% average decrease in interzone area between following co-injection of adeno-320 + adeno-5334 shRNAs was noted as compared to contralateral control (CLC) wings (Table 1). Early stage elbow regions independently injected with adeno-5334 and -320 shRNA showed 12% and 14% reductions, respectively. Significance of differences in interzone areas between adeno-injected and CLC wings resulting from control and experimental treatments were examined using matched pairs analysis and t-test (STATDISK; Triola, 2005). A statistically significant decrease in elbow interzone area was noted for wings injected with adeno-320 and adeno-5334 as well as the combined adeno-320 + adeno-5334 shRNAs when compared with respective contralateral controls (Table 1).

Table 1.

Estimated whole mount elbow interzone area in square-millimeters (mm²) +/− SD at HH36 in versican shRNA-treated embryos at HH25. Embryo number (n) represents whole mount Alcian blue and Alizarin Red stained specimens analyzed using Image J software (NIH, Bethesda MD). T-test was performed to confirm whether reduction between average area of shRNA-injected versus contralateral (CLC) wing interzones was statistically significant. P value < 0.05 was deemed significant.

shRNA (n)	CLC	Injected	Average reduction	P-value
Control (6)	0.117±0.04	0.121±0.04	0	0.36
320 (7)	0.124±0.02	0.107±0.01	14%	0.03
5334 (7)	0.121±0.03	0.106±0.02	12%	0.04
320+5334 (8)	0.113±0.02	0.098±0.03	14%	0.02

Several possibilities existed as the underlying mechanism(s) leading to reduction in interzone area in response to localized decreases in versican. TUNEL assay was used to assess apoptosis at sites at versican knockdown. In agreement with a previous report (Kavanaugh et al. 2002), overall levels of apoptosis in the joint interzone were low in uninjected contralateral control, control adeno-shRNA and combined 320/5334 adeno-injected specimens. Similar numbers of TUNEL-positive cells appeared in elbow interzone sections (n=8–11 at HH36; not shown) suggesting that increased apoptosis in response to adeno-shRNA treatments was not the causative factor for reduction of interzone area. In addition, little or no difference in cell proliferation was observed between control and adeno-shRNA treated wings at HH34/36. Low numbers of mitotic cells were found scattered within epiphyseal cartilages and elbow interzones (Figure 4A–D), with contralateral control wings having similar numbers of mitotic cells as lacz-positive regions of adeno-injected specimens.

FIGURE 4.

Expression of combined adeno-320 + adeno-5334 versican shRNAs does not correlate with changes in mitosis in the HH34 humeroulnar interzone following injection at HH25 (contralateral control, panels A, C, E, G and versican adeno-shRNA treatment, panels B, D, F, H). A) Contralateral control (clc) displaying well-organized cell layers within central interzone lamina (bracket). h, humerus. u, ulna. B) LacZ reporter histochemistry shows localization of adeno-320 + adeno-5334 shRNA (comb) expression in the elbow interzone. Bracketed area shows disorganization of interzone cells. C) Phosphohistone H3 (ser10) labeling of control interzone in boxed area of panel A shows low number of mitotic cells (arrowheads) within the control elbow joint. D) Phosphohistone H3 (ser10) labeling of section in boxed area of panel B identifies similar number of mitotic cells in interzone areas expressing adeno-320 + adeno-5334 shRNA. E) Higher magnification phase contrast image of bracketed area of panel A and C shows typical laminar organization of control interzone cells (small arrows). F) Higher magnification phase contrast image of bracketed area in panel B and D shows disruption of interzone cell alignment following adeno-320 + adeno-5334 treatment. G) Higher magnification image of boxed area in A and C shows cell associated TGFβ3 immunolocalization in the control interzone (large arrows denote representative labeled cells). H) Higher magnification of boxed area in B and D shows similar interzone TGFβ3 localization noted in panel G (large arrows) following adeno-320 + adeno-5334 shRNA treatment. Scale bar in A = 100µm for panels A–D and E= 25µm for E–H.

As Choocheep et al. (2010) presented evidence that versican may sequester TGFB family members in the interzone of the mouse autopod, we also examined TGFB3 expression in double labeled specimens of the chick humeroulnar interzone at HH34. Cell associated TGFB3 was noted in both adeno-320 + adeno-5334 co-injected specimens and uninjected controls, particularly toward the peripheral aspect (Figure 4G, H). No clear difference in staining intensity or numbers of TGFB3-expressing interzone cells between control and versican knockdown samples was observed in our study.

At HH34/36, the extent of early cavitation of the elbow interzone was variable among control as well as experimental specimens, thus it was difficult to ascertain with certainty specific impacts of versican knockdown on initial stages of synovial cavity formation. Numbers of distinguishable cell layers comprising the interzone between opposing epiphyses of the humeroulnar joint were similar with only insignificant differences (p=0.41) between experimental adeno-shRNA treated specimens and matched contralateral controls at HH34/36 (14.8+/−2.7 and 15.9+/−1.4, respectively; n= 6 sections from 5 specimens for each). Interestingly, however, the typical laminar structure of the central interzone was often less organized in versican knockdown specimens with cell alignment not as well defined relative to similar regions of control samples (Figure 4A, B&E, F).

Since reduction in the number of proliferative cells or increased apoptosis did not offer a suitable explanation for decreased interzone spacing at early stages of joint development, a third alternative was investigated. As versican has numerous interactions with other molecules implicated previously in skeletal/joint formation (Pacifici et al 2005; Khan et al 2007; Pitsilledes and Ashhurst 2008), targeted reduction of versican protein in the elbow joint interzone could impact localization and/or accumulation of other matrix and cell surface molecules that might compromise early interzone structure. Microarray analysis to analyze changes in gene expression in the elbow joint region at HH35 in response to targeted versican misexpression showed that in addition to reduced versican, down regulation of several matrix-associated transcripts were evident, including tenascin C, collagen II (alpha I), link protein, hyaluronan synthase 2 and the hyaluronan receptor, CD44 (Table 2). Overall changes in mRNA were less than 2-fold, ranging from −0.2 to −1.0, however, these molecules are also expressed in tissues associated with the adjacent skeletal template such that effects of lowered versican on gene expression specifically in the developing joint could be masked as we were unable to completely separate all surrounding tissues from the elbow joint rudiment during dissection.

Table 2.

Fold change (+/−) of versican and selected skeletal/joint-associated mRNA at HH35 as determined by microarray analysis following co-injection of adeno-320+5334 shRNA at HH25. Fold-change derived from average log ratios from two arrays as compared to adeno-control shRNA treatment.

Molecule (Gene ID)	Fold change
Versican/PG-M (NM_204787)	−0.61
Collagen IIA- alpha 1 precursor (NM_204426)	−0.18
Tenascin C (X73833)	−0.37
Hyaluronan synthase 2 (NM_204806)	−1.13
CD44 (CR386646)	−0.30
Link protein 1 (NM_205482)	−0.39
TGFB3 (NM_205454)	+0.20

Effect of versican knockdown on expression patterns of these molecules were investigated in order to determine if changes were specifically associated with developing joint structures. Embryos injected with experimental adenoviral shRNAs at HH25 and examined at HH34/36 were histochemically stained for lacZ reporter activity to localize sites of viral infection, sectioned, and subsequently processed for immunolocalization studies. As with versican immunodetection (Figure 2), control adeno-shRNA injections showed little or no difference in collagen II, tenascin, hyaluronan, CD44 or link protein staining patterns when compared to uninjected contralateral wings at HH34/36 (not shown).

Localization of collagen II in the cartilage ECM in the developing humeroulnar joint at HH36 showed strong uniform labeling along the presumptive articular surfaces in control limbs (Figure 5A) with signal extending into the interzone, suggesting ongoing chondrogenesis in the outer interzone lamina that may contribute to the articular chondrocyte population (Archer et al. 1994; Ito and Kida 2000). In contrast, collagen II staining along the periphery of the humeral epiphysis was dampened in intensity and did not extend as far into the interzone in adeno-5334 infected areas (Figure 5B, C). Double labeling of this same section for versican (Figure 3E) verified attenuation of versican signal in this same location. Tenascin C has also been suggested as an important ECM constituent necessary for establishment of articular cartilages (Koyama et al. 1995; Pacifici 1993, 1995) and combined adeno-320 + adeno-5334-shRNA treatment resulted in attenuation of pericellular tenascin associated with distal humeral chondrocytes of the presumptive articular surface along the interzone boundary in areas of adenoviral infection (Figure 5D–F).

FIGURE 5.

Impact of versican reduction on expression of joint-related molecules at HH34 (G–L) and HH36 (A–F, M–O) following microinjection at HH25. A) Collagen II (coll II) immunoreactivity is strong in the ECM along the developing articular surface of the humerus (h) and ulna (u) in the contralateral control (clc) elbow. Inset shows higher magnification of boxed area in which collagen II staining extends into the interzone (small arrowhead). B) Collagen II staining is reduced along the incipient articular region in response to adeno-5334 versican shRNA treatment. Higher magnification inset shows less collagen II signal extending into the interzone in the same area of the humerulnar joint. C) LacZ reporter staining of panel B shows region of adeno-5334 shRNA expression in the interzone. Higher magnification inset confirms adeno-5334 expression along the articular surface. D) Tenascin (tn) stains the chondrocyte matrix intensely along the articular surface of the humerus (large arrow) in the control elbow. E) Attenuation of tenascin signal is noted along the articular edge of specimens co-injected with adeno-320 + adeno-5334 versican shRNA (comb). F) LacZ reporter localization of panel E shows adeno-320 + adeno-5334 shRNA along the articular edge and interzone adjacent to the humerus. Bracket outlines central humeroulnar interzone. G) Hyaluronan (ha) is highly expressed in the epiphyses and around well organized mesenchyme of the central humeroulnar interzone (bracketed area) in contralateral control limb. H) Adeno-320 shRNA treatment reduces hyaluronan levels along the articular boundary and interzone (boxed area). Note interzone lacking well aligned cellular arrangement. I) LacZ reporter staining of panel H shows adeno-320 expression (boxed area) in the articular region and interzone. Dashed line outlines developing articular edge of the humerus. J) CD44 is also highly expressed (small arrow) along the presumptive articular cartilage and interzone between humuerus, ulna, and radius (r) of the control elbow. K) CD44 immunostaining is less intense in same area of adeno-320 injected elbow (boxed area). L) Adeno-320 expression in panel K is confirmed by lacZ reporter staining (boxed area). M) Link protein surrounds epiphyseal chondrocytes (large arrowhead) of the humerus just deep to the developing articular region (small asterisk) of control elbow joints. N) Co-injection of adeno 320 + adeno-5334 versican shRNA (comb) resulted in reduced link protein immunoreactivity in the same location just deep to the articular zone. Faint immunostaining is also noted in the adjacent interzone (large asterisk). O) LacZ reporter staining of panel N shows adeno-320 + adeno-5334 expression in epiphyseal and articular chondrocytes of the humerus as well as within adjacent interzone. Scale bars in panel C = 100µm for panels A–C, G–L & 25µm in insets in A–C. F = 25µm in D–F, M–O.

Hyaluronan and its CD44 receptor have been previously demonstrated as crucial to joint morphogenesis (Craig et al. 1990; Archer et al. 1994; Edwards et al. 1994; Pitsillides et al. 1995; Dowthwaite et al 1998; Matsumoto et al. 2007) and versican may interact with both (reviewed in Wight 2002; Wu et al. 2005). Because the chick CD44 antibody (ID10) did not work well on paraffin-sectioned specimens, cryosectioned material was utilized for immunodetection of CD44 and histochemical localization of hyaluronan. As seen in Figure 5G, strong pericellular hyaluronan signal was observed in both cartilages and interzone of the HH34 humeroulnar joint rudiment in the control wing. Hyaluronan signal was particularly notable within the central interzone where mesenchyme was aligned in the typical stratiform orientation parallel to the future joint line (Dowthwaite et al. 1998). A marked decrease in expression of hyaluronan was observed along the distal humerus in future articular and adjacent interzone regions expressing the adeno-320 shRNA (Figure 5H, I). As noted previously, the laminar organization of the interzone also appeared disrupted in response to reduction in versican protein. CD44 was also highly expressed in cells of the HH34 elbow interzone of the uninjected control (Figure 5J) and was visibly reduced at sites of adeno-320 expression within the joint region (Figure 5K, L).

Versican interaction with hyaluronan is stabilized by link protein, which has greater affinity for versican than aggrecan (Shi et al. 2004). Transient expression of link protein has also been reported in the pre-cavitation interzone of the mouse autopod (Capehart 2010) so it was of interest to investigate its expression in early stages of chick joint morphogenesis. Little link protein was observed in the control humeroulnar interzone at HH36, but low-level cell associated immunoreactivity was found within the humeroulnar interzone alongside areas of adeno-320+ adeno-5334 expression (Figure 5M-0). As expected, link protein was highly expressed in the cartilage matrix of humeral and ulnar epiphyses. Unlike collagen II and tenascin, however, little link protein expression was observed in the presumptive articular region immediately adjacent to the central interzone (Figure 5M). Relative to the same region of the contralateral control, link protein expression in epiphyseal chondrocytes of the humerus subjacent to the future articular surface was notably reduced in response to adeno-320 + adeno-5334 co-injection at HH25 (Figure 5N, O).

DISCUSSION

Joint morphogenesis hinges on a series of physical and molecular processes that involve specification and formation of the joint interzone. The hyalectin proteoglycan, versican, is continuously expressed throughout the matrix of the joint interzone during early stages of chick limb development while at progressively later stages (HH41) it persists in articular chondroctyes (Shepard et al. 2007). The intent of this study was to better understand matrix versican’s function during early establishment of the synovial joint rudiment. We approached this question by attempting to analyze the morphologic and molecular consequences of ectopically silencing versican protein levels in the synovial joint during early stages of joint development (HH25) and assessing the effects at HH34-36 using adenovirally-encoded shRNAs prepared by our laboratory in combination with in ovo microinjection. In our study using a transient knockdown system to target versican specifically within developing joint tissues, we chose to focus primarily on the impacts of versican silencing on the elbow joint. This was done in part as a precaution since microinjection into the future autopod could potentially damage the apical ectodermal ridge (AER) or progress zone and lead to mechanically induced skeletal deformities (Saunders 1948), and because reproducibly targeting adenoviral constructs to this joint location was better achieved. Histochemical staining of HH36 embryos resulted in significant reduction in the joint interzone area between the opposing epiphyses in experimental shRNA-treated elbows when compared with controls. In addition to diminution in elbow template spacing in whole mount specimens, we noticed reduced immunoexpression of other synovial joint and cartilage markers in tissue sections including collagen II, tenascin, CD44, hyaluronan, and link protein at sites of shRNA transfection. This was often accompanied by varying degrees of disorganization of the typical well-ordered alignment of interzone cells that occurs parallel to and along the future joint line (Dowthwaite et al. 1998).

A recent study utilizing a conditional versican knockout mouse reported ‘tilting’ of joints of the hindlimb autopod (Choocheep et al. 2010) which was ascribed to close packing of mesenchymal cells at the joint interzone and perhaps disruption of chondrocyte columns within the epiphyseal growth plate of the digits. In the mutant, the authors also observed aberrant nodes of chondrocyte hypertrophy and slight delays in chondrocyte differentiation when compared with wild-type hindlimb autopod. The joint phenotype and delays in chondrogenesis were attributed to disruption in versican-dependent localization of TGFβ family members. Interestingly, the authors did not notice a difference in the more proximal elbow or knee joints and concluded that the phenotype was only observed in the autopod, since TGFβ signaling is more prevalent in that area (Choocheep et al. 2010). TGFβ3 has been utilized successfully in a recent joint regeneration protocol (Lee et al. 2010) therefore we elected to examine TGFβ3 expression in early avian joint tissues. In the present study, little or no difference in immunodetectable TGFβ3 was detected between control and versican knockdown specimens in the elbow joint interzone and there was actually slight upregulation of TGFβ3 mRNA in experimental elbow regions (Table 2). Although versican expression patterns in developing joint tissues appear highly conserved between chick and mouse (Snow et al. 2005, Shepard et al. 2007; Capehart 2010) it is possible that species-specific differences may occur with regard to versican/TGFβ interaction in joints proximal to the autopod.

After injection with versican shRNAs, we noted marked reduction in expression of collagen II among articular chondrocytes concomitant with decreased levels of versican. Since collagen II is a well-established marker for chondrocyte differentiation (Linsenmeyer et al. 1973), decrease in collagen II occurring from versican knockdown at sites of articular cartilage formation could translate into a delay or inhibition in establishment of the future articular layer. Versican has been implicated previously in early stages of cartilage differentiation (Williams et al. 2005; Kamiya et al. 2006; Shepard et al. 2008; Choocheep et al. 2010) so it is entirely possible that it facilitates transition of outer interzone laminae into the articular cartilage phenotype. The expression pattern of the glycoprotein tenascin, which is co-distributed with collagen II in the epiphysis, was also impacted by versican knockdown in the elbow joint region, especially in the presumptive articular cartilage matrix lining the epiphyseal ends of the distal humerus. A previous study has shown that in cartilaginous areas where reduction in tenascin occurs, the areas become particularly vulnerable to chondrocyte maturation leading to osteogenesis (Pacifici et al. 1993). In our experiments, we noticed reduced levels of tenascin among articular chondrocytes concurrent with versican silencing. It is intriguing to think that normal versican levels may deter articular chondrocytes from transitioning into bone by helping to maintain adequate tenascin in the extracellular environment. Such activities for versican along the developing articular cap would suggest that its role is different, but complementary to that of aggrecan, which provides joint cartilages with their unique functional characteristics and is largely expressed in a non-overlapping pattern with versican as joint formation proceeds (Shepard et al. 2007).

Link protein shows greater affinity toward interaction with the G1 domain of versican than the G1 domain of aggrecan to form stable versican-hyaluronan-link protein complexes (Shi et al. 2004). In the present study little link protein was observed in early chick interzone tissue of control specimens at HH36 but was predominant in the epiphyseal cartilage matrix, just deep to the incipient articular surface. Link protein immunoreactivity, however, was markedly reduced in this location in response to versican knockdown. Link protein was expressed in the interzone of the prx1-Cre conditional versican mutant (Choocheep et al. 2010) and was suggested to stabilize hyaluronan in the absence of versican. In the chick elbow, weak link protein immunoreactivity was noted in the versican deficient interzone in keeping with the previously reported compensatory mechanism for sustaining hyaluronan.

Hyaluronan and its cell surface receptor CD44 have long been proposed to possess key roles in synovial joint morphogenesis. Furthermore, a recent publication has shown that versican is involved in regulating hyaluronan and CD44 activities in extracellular signal transduction pathways (Suwan et al. 2009). Another study using conditional deletion of hyaluronan synthase 2 (Has2) knockout mice led to defective synovial joint cavities along with defects in chondrocyte maturation confirmed by reduction in extracellular aggrecan (Matsumoto et al. 2009). Defective joints were also reported in cases where interaction between hyaluronan and CD44 was interrupted using hyaluronan oligosaccharides (Dowthwaite et al. 1998). Interestingly, ADAMTS-1 has been suggested to proteolyze versican in the early mouse interzone, yielding a fragment containing the hyaluronan-binding G1 domain that may function independently of the intact proteoglycan (Capehart 2010). In the present study versican knockdown yielded visible deficiencies in both CD44 and hyaluronan at sites of adeno-shRNA infection in the early elbow region. Down regulation of CD44 and hyaluronan synthase 2 transcripts also lends support to versican’s ties to the hyaluronan pathway. It is not known at present whether versican elicits its effects on mRNA expression of these molecules and the others investigated herein through direct interaction with hyaluronan and/or CD44 signaling or perhaps through an alternative mechanism such as the TGFβ-smad pathway. In any case, versican misexpression may have impacts at the transcript or functional level in keeping with a growing appreciation of ECM influence on gene expression as well as on direct matrix interactions. Clearly, much remains to be learned regarding versican’s interplay with the hyaluronan pathway during joint morphogenesis.

In the present study, reduction in non-chondrogenic interzone spacing between opposing epiphyses comprising the elbow joint did not appear due to reduced proliferation or increased apoptosis either as a generalized response to adenoviral infection or to lowered versican protein. The reduction in interzone area appeared due to alterations in alignment and layering of interzone cells as a result of reduction of matrix constituents including hyaluronan and versican. Alternatively, reduction of collagen II and tenascin along the developing articular surface or link protein in the epiphyseal cartilage matrix could perhaps alter signaling to the interzone through an unidentified mechanism, thus impacting matrix structure. It was not possible to draw unequivocal conclusions regarding effects of versican knockdown on early joint cavitation events due to variability in degree of interzone cell separation among contralateral control or shRNA-treated specimens at HH34-36. In addition, due to the variable size of lacZ-positive infection sites and level of versican reduction within the developing joint rather than its complete deletion from the entire rudiment, it was necessary to restrict comparative observations in tissue sections to the local level in which adenoviral mediated shRNA expression occurred. As such, results still show that versican misexpression within the developing joint structure can alter early aspects of joint morphology. From our examination of several factors impacted by versican knockdown, it appears that the reduced interzone area phenotype observed in whole mount specimens is likely due to a collective effect on interzone matrix structure as a result of localized versican silencing. Taken together, results suggest that reduced versican in combination with attenuated expression of hyaluronan, CD44, and perhaps articular tenascin may destabilize the ECM leading to alteration of interzone area. In addition, reduction in collagen II levels within the outer edges of the interzone along the forming articular surface suggests that articular cartilage differentiation may be compromised or at least delayed in response to lowered versican.