Antisense Inhibition of Hyaluronan Synthase-2 in Human Articular Chondrocytes Inhibits Proteoglycan Retention and Matrix Assembly*

In order to define the role of cell-associated hyaluronan in cartilage matrix retention, human articular chondrocytes as well as cartilage slices were treated with phosphorothioate oligonucleotides comprised of sequence antisense to the mRNA of human HA synthase-2 (HAS-2). As a prerequisite for these studies, it was necessary to determine which HA synthase (HAS), of three separate human genes capable of synthesizing HA, designated HAS-1, HAS-2, or HAS-3, is primarily responsible for HA synthesis in human articular chondrocytes. The copy number of each HAS mRNA expressed in cultured human articular chondrocytes was determined using quantitative (competitive) reverse transcription-polymerase chain reaction (RT-PCR). Only HAS-2 and HAS-3 mRNA expression was detected. The level of HAS-2 mRNA expression was 40-fold higher than that of HAS-3. Cultures of human articular chondrocytes and cartilage tissue slices were then transfected with HAS-2-specific antisense oligonucleotides. This treatment resulted in time-dependent inhibition of HAS-2 mRNA expression, as measured by quantitative RT-PCR, and a significant loss of cell-associated HA staining. Sense and reverse HAS-2 oligonucleotides showed no effect. The consequences of reduced HA levels (due to HAS-2 antisense inhibition) were a decrease in the diameter of the cell-associated matrix and a decreased capacity to retain newly synthesized proteoglycan. These results suggest that HA synthesized by HAS-2 plays a crucial role in matrix assembly and retention by human articular chondrocytes.

Hyaluronan (HA) 1 is a high molecular weight linear glycosaminoglycan comprised of repeating disaccharide units, ␤1,3 N-acetyl-D-glucosamine-linked ␤1,4 to D-glucuronic acid. HA is synthesized at the plasma membrane and extruded through a putative membrane pore directly into the extracellular space (1)(2)(3). Recently, three related mammalian genes have been identified as HA synthases, designated has-1, has-2, and has-3 (4 -8). These three genes display sequence homology and share common exon-intron boundaries between exons 2 and 3 (9), yet they are localized on separate chromosomes (10). Different cell types express different HAS enzymes. For example, in human dermal fibroblasts both HAS-1 and HAS-2 transcripts were expressed and both appeared to be regulated by transforming growth factor-␤ (11). These results suggest that HA synthesis may be regulated by the differential expression of these three HAS enzymes by a particular cell type or phenotype.
The extracellular matrix of cartilage contains a small amount of HA. The predominant extracellular matrix components of cartilage are collagens, types II, IX, and XI, and proteoglycans such as aggrecan (12). Nonetheless, the small amount of HA present plays an important, unique role in cartilage, serving to facilitate retention of aggrecan within the tissue. Often more than 50 aggrecan proteoglycan monomers become bound to a single filament of HA (13). These HAproteoglycan aggregates are maintained at high concentrations within the cartilage, giving the tissue its unique ability to resist compression. In addition, HA also facilitates the anchorage of HA-proteoglycan aggregates to the surface of the cells within cartilage, the chondrocytes. We have demonstrated that aggregates are tethered to the chondrocyte cell surface via interactions with CD44 HA receptors (14 -16). These same CD44 receptors also play a role in the internalization of extracellular HA leading to its degradation intracellularly, presumably within lysosomes (17).
Degenerative events that occur in cartilage, many leading to irreversible disease states such as osteoarthritis, are believed to occur due to an imbalance between turnover (catabolism) and matrix biosynthesis (anabolism), particularly related to the content of aggrecan (12). In fact, the content of aggrecan staining within cartilage is used as a criterion for disease diagnosis (i.e. safranin O staining criterion of the Mankin scale (18)). However, as discussed above, the retention of proteoglycan is also dependent on the presence of optimal concentrations of HA as well as associated functional HA receptors such as CD44. Depletion of HA has been observed in early experimental osteoarthritis in dogs (19). Reduced joint loading due to splint immobilization also resulted in a significant decrease in dog articular cartilage HA (20). In the dog model of osteoarthritis it is still uncertain whether decreased synthesis or increased degradation leads to the reduction of HA. However, in the unloaded canine cartilage model, decreased synthesis appears to be responsible for the loss of matrix (21,22). In order to probe deeper into the regulatory mechanisms that might be involved in such systems, a better understanding of which enzyme(s) are responsible for HA synthesis in human articular cartilage is essential. However, no antibodies are as yet available to quantify HAS-1, HAS-2, or HAS-3; therefore, the use of indirect methods is required. In this study, quantitative (competitive) RT-PCR was used to determine, at a minimal level, which of these three enzymes is expressed as mRNA by human articular chondrocytes. However, even after determining which enzymes are expressed, and at what levels, it would still remain unclear which enzyme is functionally or primarily responsible for HA synthesis in cartilage. To address this question, sequence-specific phosphorothioate oligonucleotides directed against the predominant mRNA expressed (HAS-2) were used to determine the effects of inhibition of this enzyme. The present study demonstrates that HAS-2 inhibition results in a reduction in HA, a reduction in cell-associated matrix size, and a reduced capacity of the chondrocytes to retain proteoglycan. Agarose was from FMC BioProducts (Rockland, ME). All other enzymes and chemicals either molecular biology grade or reagent grade materials were purchased from Sigma.

Materials
Tissue Acquisition-Human articular cartilage was obtained from the talocrural ankle joint of normal human donors. Tissue was obtained within 24 h of death through the Regional Organ Bank of Illinois, and all donors were documented as having no known history of joint disease.
Cell Culture-Full thickness slices of articular cartilage were dissected under aseptic conditions and subjected to sequential Pronase and collagenase digestion to liberate chondrocytes from tissues (23). Isolated chondrocytes were cultured for 5 days in alginate beads with daily medium changes (24). After 5 days of recovery, the chondrocytes were released from the alginate beads by treatment with 55 mM sodium citrate followed by a 20-min digestion with 0.25% trypsin/EDTA (25). The cells were washed in DMEM containing 10% FBS to inactivate trypsin activity followed by a wash with serum-free DMEM. Cells were then resuspended in serum-free DMEM/F12 containing the presence or absence of various phosphorothioate oligonucleotides and replated into 35-mm dishes as monolayer cultures at a concentration of 5 ϫ 10 5 cells/dish. To provide for optimal transfection, the phosphorothioate oligonucleotides were first mixed with 6.0 g/ml LipofectAMINE in serum-free DMEM and incubated for 20 min at room temperature to allow DNA-liposome complexes to form. The diluted complex solution (final oligonucleotide concentration of 2 M) was mixed gently and added to the chondrocytes at the time of plating. Medium containing antisense or control oligonucleotides was then removed from the cultures after 5 h of incubation and replaced with fresh DMEM containing 10% FBS (i.e. oligonucleotide-free medium) and incubation continued for varying periods.
Tissue Culture-Full thickness slices of human articular cartilage were dissected from the talocrural joint of normal human donor ankles (ϳ1 ϫ 10 ϫ 10 mm) and cultured directly in 2.0 ml of medium containing 10% FBS. Following 1 day of culture for recovery, the tissue slices were washed and medium was replaced with fresh serum-free DMEM in the presence or absence of various phosphorothioate oligonucleotides, diluted as LipofectAMINE DNA-liposomes to a final concentration of 2 M. Medium (with or without oligonucleotides) was changed every 2 days. Following 7 days of incubation the slices were removed and embedded in HistoPrep TM freezing medium. Cryostat sections (8.0 m) were prepared and stained for HA.
Cell Viability Assay-Cell viability was determined by the trypan blue dye exclusion method. After 5 min of incubation with 0.4% trypan blue, the percentage of stained cells (indicative of nonviable cells) versus stain-excluding cells were counted in four random fields of the plate.
Staining for HA-Chondrocytes in culture wells or 8-m cryostat sections of cartilage slices mounted onto glass slides were fixed with 2% paraformaldehyde buffered with phosphate-buffered saline (PBS) at room temperature for 2 h. Tissue sections were pretreated with 2 units of chondroitinase ABC (at pH 8.0) for 2 h at 37°C for unmasking. The tissue sections or cells were next treated with 0.3% H 2 O 2 in 30% methanol for 30 min at room temperature to block the internal peroxidase activity and then incubated with 1% bovine serum albumin in PBS for 1 h at room temperature. Sections or cells were then incubated with 2.0 g/ml biotinylated HA-binding protein (HABP) probe for 2 h at room temperature followed by a streptavidin-peroxidase reagent (Vectastain kit) and diaminobenzidine-containing substrate solution (SIG-MA FAST TM DAB). As a control, tissue sections or chondrocytes were pretreated with 5 units/ml Streptomyces hyaluronidase for 1 h at 60°C.
Particle Exclusion Assay-Cell-associated pericellular matrices were visualized using a particle exclusion assay (15). Briefly, following treatment of monolayer cultures of chondrocytes with or without incubation with various phosphorothioate oligonucleotides, and 24 h of recovery in oligonucleotide-free medium, the culture medium was removed and replaced with a 0.75-ml suspension of formalin-fixed erythrocytes (10 8 per ml) in PBS containing 0.1% bovine serum albumin. The particles were allowed to settle for 15 min, and the cells were photographed using a Zeiss inverted phase-contrast microscope with Varel optics.
Synthesis of Sulfated Proteoglycans-After 24 h of treatment with antisense or sense oligonucleotides, monolayer cultures of chondrocytes were pulse-labeled with [ 35 S]sulfate (20 Ci/ml) for 4 h. The cells were chased for an additional 24 h in fresh medium, at which time the culture medium pool and cell-associated matrix pool were separated and collected. The chase culture medium was removed and cell monolayer digested with 0.25% trypsin containing 1 mM EDTA, at 37°C for 30 min. The trypsin/EDTA supernatant, defined as cell-associated matrix pool, was clarified by centrifugation. Proteoglycan (PG) aggregates within each pool were dissociated by adjustment to dissociative conditions, 4 M guanidine HCl containing the following protease inhibitors at final concentrations as follows: 10 mM EDTA, 0.1 M 6-aminohexanoic acid, 5 mM benzamidine hydrochloride, 10 mM N-ethylmaleimide, and 0.5 mM phenylmethylsulfonyl fluoride for 24 h at 4°C (25). 35 S-Labeled PGs in each medium and dissociative extract were quantified by liquid scintillation counting following rapid filtration, as described previously (26). The concentration of 35 S-PGs synthesized was normalized to cellular DNA content (26).
Quantitative Competitive Polymerase Chain Reaction Using Reverse Transcriptase (RT-PCR)-Total cytoplasmic RNA was extracted from chondrocyte monolayer cultures with Trizol reagents and subjected to reverse transcription and quantitative competitive PCR. Briefly, 0.25 g of total RNA was converted to cDNA using Moloney murine leukemia virus reverse transcriptase in the presence of 0.15 M HAS-2, HAS-3, or aggrecan-specific downstream primers (HAS-2, 5Ј-TTT CTT TAT GTG ACT CAT CTG TCT CAC CGG-3Ј; HAS-3, 5Ј-CAG AAG GCT GGA CAT ATA GAG GAG GG-3Ј; aggrecan, 5Ј-CTC CAC TGC CTG TGA AGT CAC CAC-3Ј). DNA fragments that share the same primer template sequence with the target cDNA but contain a completely different, smaller, or larger intervening sequence were prepared and used as DNA internal standards (i.e. mimics) (25,27,28). Aliquots of sample cDNA mixed together with serial dilutions of DNA mimics were co-amplified as templates in the presence of downstream primers and 0.15 M upstream primers for HAS-2, HAS-3, or aggrecan (HAS-2 upstream, 5Ј-ATT GTT GGC TAC CAG TTT ATC CAA ACG G-3Ј; HAS-3 upstream, 5Ј-AGA GAC CCC CAC TAA GTA CCT CCG-3Ј; aggrecan upstream, 5Ј-GCA CCA TGC CTT CTG CTT CCG AG-3Ј) in a PCR mixture consisting of 2 mM magnesium chloride, 200 M of each deoxyribonucleotide, and 2.5 units of AmpliTaq DNA polymerase. The DNA was denatured by heating at 95°C for 2 min, followed by 23 cycles (28 cycles for HAS-3) of 1 min at 95°C, annealing at 60°C (64°C for HAS-3), and extension at 72°C for 1 min (Perkin-Elmer thermocycler). This reaction was followed by a final elongation step that lasted 5 min at 72°C. The amplified products were analyzed by electrophoresis on 1.5% agarose gels followed by staining with SYBR Green I. The stained products were scanned and quantified using a fluoroimaging system (Molecular Dynamics).
To determine that all samples contained equivalent amounts of RNA (or to normalize results due to small differences), in a separate set of reactions total RNA from samples were co-amplified in the presence of serial dilutions of an RNA internal standard (mimic) prepared for glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The GAPDH RNA mimic shares the same primer template sequence but contains a smaller intervening sequence. Samples containing 0.25 g of sample total RNA were co-reverse-transcribed with 2-fold serial dilutions of GAPDH RNA mimic in the presence of 0.15 M GAPDH-specific downstream primer (5Ј-TTA CTC CTT GGA GGC CAT GTG GGC C-3Ј). The sample and mimic cDNA products were then co-amplified in the presence of the GAPDH-specific downstream primer together with 0.15 M upstream primer (5Ј-ACT GCC ACC CAG AAG ACT GTG GAT GG-3Ј) using PCR conditions as described for HAS-2 amplification.

HAS mRNA Expression in Human Articular Chondrocytes-
Only HAS-2 and HAS-3 products were detected following RT-PCR of normal human articular chondrocyte RNA. HAS-1 mRNA expression was not detected in these chondrocytes even with the use of high cycle numbers (i.e. 35 cycles), whereas the same primer pairs successfully generated HAS-1, HAS-2, and HAS-3 products from RNA derived from human dermal fibroblasts (CCD-1093Sk, data not shown). For quantification of HAS-2 and HAS-3 mRNA expression in human articular chondrocytes, quantitative competitive RT-PCR was performed. Ten-fold dilutions of DNA internal standards (termed "mimics") were used to determine the concentration range necessary to compete for the PCR amplification of reverse-transcribed sample cDNAs (termed "targets"). Cycle number was also varied to determined the linear amplification range for both sample cDNA (target) and internal standard DNA (mimic). Shown in Fig. 1 is the co-amplification of target cDNA with 1.5-and 2-fold dilutions of mimic. As can be seen, a significantly different range of mimic concentrations was required to compete for the amplification of HAS-2 target cDNA as compared with HAS-3. The deduced copy number for HAS-2 expressed in normal human articular chondrocytes was approximately 40-fold higher than that of HAS-3 (Table I). Thus, as assessed by the level of mRNA expression, HAS-2 would appear to be the predominant enzyme expressed by human articular chondrocytes.
Inhibition of Human Chondrocyte HAS-2 by Antisense HAS-2 Oligonucleotides-In order to determine whether HAS-2 enzyme is the principal functioning HA synthase, human articular chondrocytes as well as cartilage tissue slices were transfected with HAS-2 antisense phosphorothioate oligonucleotides. The level of inhibition of HAS-2 mRNA expression in chondrocyte cultures was determined by quantitative competitive RT-PCR. As shown in Fig. 2, treatment of chondrocytes with antisense oligonucleotides resulted in an ϳ60% inhibition of HAS-2 mRNA expression (maximal inhibition obtained at 24-h time point), as compared with chondrocytes treated with the sense oligonucleotide (or compared with untreated chondrocytes, data not shown). However, at the 40-h time point (35 h of incubation in oligonucleotide-free medium), HAS-2 mRNA levels in antisense-treated cultures began to recover, approaching the levels present in sense-treated control cultures.
To determine the specificity of antisense treatment on these chondrocytes, the effects on aggrecan mRNA expression was determined by quantitative competitive RT-PCR. As shown in Fig. 3, no significant difference in the ratio of copy numbers of aggrecan was observed in antisense versus sense oligonucleotide-treated chondrocytes. Given that aggrecan expression is highly sensitive to changes in chondrocyte metabolism, these data suggest that antisense treatment also had no significant effect on chondrocyte phenotype. Like the HAS results depicted in Fig. 2, the aggrecan results were normalized to GAPDH mRNA expression levels. However, when the ratio of GAPDH copy numbers are plotted (Fig. 3), it can be seen that there were no significant changes in the GAPDH levels in antisense-and sense-treated chondrocytes over the entire time course of the experiment.
To support further the specificity of HAS-2 antisense inhibi- tion, the effect of oligonucleotide treatment on cell viability was determined by trypan blue exclusion assay. As shown in Fig. 4, no change in chondrocyte viability was observed in cells transfected with control oligonucleotides (sense or reverse) or antisense oligonucleotides. Approximately 95% of all the chondrocytes under each condition was viable.
Antisense Inhibition of HAS-2 Inhibits HA Synthesis and Accumulation-Biotinylated HABP molecules were used to visualize HA within the cell-associated matrix of chondrocytes, as well as the extracellular matrix of articular cartilage slices. As shown in Fig. 5A, chondrocytes treated with control sense oligonucleotides display prominent staining for HA within the cell-associated matrix, depicted by a double-headed arrow. Untreated chondrocytes display a similar staining pattern (data not shown). Given that the entire cell-associate matrix was removed by trypsin/EDTA treatment at the time of plating (Fig. 5A, inset), the observed accumulation of HA must repre-sent newly synthesized HA. Human articular chondrocytes treated with HAS-2 antisense oligonucleotides displayed a substantial inhibition of cell-associated HA (Fig. 5B). The expression of HA was not completely eliminated but was reduced to a level roughly comparable to the percent inhibition of HAS-2   (Fig. 2). Interestingly, however, although HAS-2 mRNA levels begin to return to control levels 40 h after antisense treatment (35 h of incubation in oligonucleotide-free medium), HA staining remains reduced at the 48-h time point (Fig. 5B,  inset). As a control, sense oligonucleotide-treated chondrocytes (like those shown in Fig. 5A) were treated with Streptomyces hyaluronidase. As can be seen in Fig. 5C, all cell-associated staining for HA was removed. However, small patches of staining remains, protected from hyaluronidase treatment. We would suggest that this hyaluronidase-insensitive (protected) HABP-stainable material represents intracellular HA that are segregated into beaded structures, resembling endosomal vacuoles.
The biotinylated HABP probe was also used to visualize HA within slices of human articular cartilage tissue. As shown in Fig. 5D, cartilage treated with control oligonucleotides displayed prominent staining for HA predominately associated with the pericellular, territorial, and near-interterritorial matrices. Seven days of treatment with HAS-2 antisense oligonucleotides resulted in a dramatic decrease in HABP-stainable HA, especially within matrix closely associated with the chondrocyte cell surface (Fig. 5E). As a control, cartilage samples identical to those in Fig. 5D were treated with Streptomyces hyaluronidase. As can be seen in Fig. 5F, hyaluronidase treatment resulted in a near complete removal of all HABP-stainable material.
Antisense Inhibition of HAS-2 Inhibits Cell-associated Matrix Assembly-One of the functions of HA in chondrocytes is to serve as a scaffold for the assembly of a cell-associated matrix (15,16). In Fig. 6A and inset, 24-h control sense or reverse oligonucleotide-treated chondrocytes, respectively, display large cell-associated matrices (identical to untreated chondrocytes, not shown). However, HAS-2 antisense oligonucleotide treatment resulted in a substantial (but not total) decrease in the diameter of cell-associated matrix (Fig. 6B). Forty three hours after an initial 5-h pulse of HAS-2 antisense oligonucleotide (i.e. 48-h time point), the cell-associated matrix diameter began to increase (Fig. 6D) but was still smaller than control cells (Fig. 6C). However, by 72 h there was no difference be-tween antisense and control oligonucleotide-treated cells (data not shown).
Effects of HAS-2 Antisense Oligonucleotides on PG Synthesis-A second important function of chondrocyte-derived HA is the retention of aggrecan. Trypsin/EDTA treatment of chondrocytes removes all cell surface-associated HA as well as the PG. When HAS-2 antisense or control oligonucleotide-treated chondrocytes were pulse-labeled with [ 35 S]sulfate, no difference in the total amount of PG synthesized was detected (Fig. 7). This result adds support to the specificity of action of HAS-2 anti- sense treatment. That is, there are no apparent effects on aggrecan mRNA expression or the biosynthesis and secretion of aggrecan. However, the cellular distribution of newly synthesized aggrecan was altered. In HAS-2 antisense oligonucleotide-treated chondrocytes, less PG accumulated (or was retained) within the cell-associated matrix as compared with control oligonucleotide-treated cultures (Fig. 7). This reduction in cell-associated PG was contrasted by an increase in PG released into the culture medium. These results suggest that HAS-2 antisense oligonucleotide-treated chondrocytes have a reduced capacity to retain newly synthesized PG. DISCUSSION The results from this study demonstrate that only HAS-2 and HAS-3 mRNAs are expressed by normal human articular chondrocytes. However, even though both messages are present, HAS-2 mRNA is expressed in large excess over that of HAS-3. The data shown in Table I represent the average of competitive RT-PCR results performed on three independent preparations of total RNA, isolated from chondrocytes derived from one representative donor. In experiments not shown, using chondrocytes from other donors, the absolute copy number of HAS-2 mRNA varied, slightly higher in some donors and lower in others. This heterogeneity in human donor cartilage samples is expected, and variations in enzymes such as HAS-2 may, in future studies, have diagnostic or prognostic importance. Nonetheless, in all of the normal donor-derived chondrocyte RNA examined, even though the absolute values of HAS-2 and HAS-3 mRNAs change, the ratio of HAS-2 to HAS-3 copy numbers remained relatively constant, from donor to donor.
The HAS-2 antisense oligonucleotide treatment used in this study resulted in a substantial inhibition of HAS-2 expression at the 8-and 24-h time points. Our previous work using CD44 antisense phosphorothioate oligonucleotides on bovine chondrocytes required between 24 and 48 h of continuous treatment before a significant inhibition could be observed (29). One important difference between the two studies was the inclusion in this study, of the liposome-forming agent LipofectAMINE. Li-pofectAMINE is a 3:1 mixture of cationic and neutral lipids.
Cationic lipid end groups bind the charged DNA molecules, and the complexes form into liposomes that subsequently interact with the cell membrane, leading to a highly efficient uptake of the bound DNA molecule into the cell (30). In addition, Lipo-fectAMINE transfection reportedly extends the half-life of oligonucleotides (31). Preliminary studies using fluoresceintagged phosphorothioate oligonucleotides plus LipofectAMINE demonstrated that a 5-h incubation period was sufficient to obtain saturable intracellular accumulation of oligonucleotide (data not shown). Without LipofectAMINE, the phosphorothioate oligonucleotides still enter the cells (chondrocytes) but require more than 24 h to reach the same level of intracellular accumulation. Thus, using LipofectAMINE, transfection times can be shorter, keeping to a minimum the time of culture with less-than-optimal medium and still provide for more than 24 h of selective gene inhibition.
Such considerations in culturing are important in order to avoid changes in phenotype, especially when working with human articular chondrocytes. In this study the cells were grown initially in a three-dimensional alginate bead system under conditions maximized for the growth of human chondrocytes (32). During the subsequent short term culture in monolayer, the cells continued to synthesize and secrete PG (Fig. 7) and continued to express aggrecan mRNA (Fig. 3). Neither aggrecan nor GAPDH mRNA levels changed significantly between the 8-, 24-, and 40-h time points. In addition, all of the chondrocytes depicted in Fig. 6 continue to maintain a rounded chondrocyte morphology together with an extensive cell-associated matrix, two key characteristics of the chondrocyte phenotype (33).
Antisense oligonucleotide inhibition of HAS-2 resulted in an inhibition in HA accumulation at the cell surface of chondrocytes as well as the pericellular matrix of cartilage tissue slices. These results confirm the expression data and suggest that HAS-2 is indeed the principal enzyme utilized by chondrocytes to synthesize HA. The level of inhibition of HAS-2 (ϳ60%) was similar to the level of inhibition that we obtained previously using CD44 antisense oligonucleotides in bovine articular chondrocytes (29). That the level of inhibition of HAS-2 achieved was not 100% was fortuitous. Given that 40% of HAS-2 mRNA was still detected in antisense-treated chondrocytes (and likely still functional), it is not surprising that a small amount of HABP staining for HA continued to be observed (Fig. 5B). The same cells also exhibited a thin particle exclusion zone of cell-associated matrix (Fig. 6B). The reduction in HA staining and the size of the cell-associated matrix appear roughly proportional to the degree of HAS-2 inhibition, again pointing to the suggestion that HAS-2 is the primary enzyme involved in chondrocyte HA production. However, additional studies such as the inclusion of HAS-3 antisense oligonucleotides will be needed to support more definitively the role of HAS-2 as the predominant HA synthase in chondrocytes.
By assuming that the size of the HA synthesized in the presence of HAS-2 antisense oligonucleotides has not changed and that the concentration of aggrecan has not been altered (from Fig. 3 and Fig. 7), the results suggest that HA concentration alone controls the diameter of the cell-associated matrix. More experiments will be required to verify these observations. However, previous studies from our laboratories (14,15,34), based mainly on the particle exclusion assay, demonstrated that three components are necessary for the assembly of a cell-associated matrix by chondrocytes, namely HA, an aggregating PG, and a cell-surface HA receptor such as CD44. Deficiency in any one of these three components is sufficient to inhibit matrix assembly in chondrocytes (and many other cell FIG. 7. The effect of HAS-2 antisense oligonucleotide treatment on PG biosynthesis and cellular distribution in cultures of human articular chondrocytes. Twenty-four h after the initial addition of sense (light gray bars) or antisense (black bars) oligonucleotides, the cultures were pulse-labeled with [ 35 S]sulfate for 4 h and chased for an additional 24 h. The medium (med) and cell-associated matrix (CM) pools were collected, and the specific incorporation of [ 35 S]sulfate in PG was determined using a rapid filtration method. The level of 35 S-PGs synthesized was normalized to cellular DNA content (following DNA analysis of the residual cell pellet). Values represent the average Ϯ S.D. of triplicate experiments; p ϭ 0.056 for differences between sense and antisense bars in medium; p ϭ 0.017 for cellassociated matrix. types). Although not shown in this study, CD44 expression was not affected by HAS-2 antisense treatment, and as discussed above, aggrecan was also not affected. Therefore, it is likely that the inhibition of HA is responsible for the altered size of the cell-associated matrix.
The maintenance of healthy cartilage requires a capacity for the turnover and repair of the extracellular matrix. A critical factor in this process is the retention of newly synthesized matrix, especially the water-soluble components of matrix such as aggrecan. The retention of aggrecan in turn, is highly dependent on its interaction with HA, including both the formation of link protein-stabilized HA-PG aggregates within the extracellular matrix and HA-CD44 interactions that serve to anchor the PG-rich matrix to the chondrocyte cell surface. Both of these retention mechanisms are likely altered when HAS-2 expression is inhibited. Thus, understanding the regulation of chondrocyte HAS enzymes may be critical to attempts at promoting cartilage tissue repair. Whether a condition analogous to this experimentally induced inhibition of HAS-2 occurs in human tissues such as during aging or disease remains to be determined. Certainly, the depletion of HA and subsequent passive loss of proteoglycan observed in joint immobilization studies (20) suggests that HA synthesis is at least dynamic and sensitive to changes in loading (or lack thereof). What factors other than loading regulate HAS expression in chondrocytes and could the control of these factors be used for beneficial changes in damaged cartilage tissue? Are changes in HAS usage by chondrocytes (i.e. usage of HAS-1 or HAS-3 instead of HAS-2) evoked by certain physiological conditions? All of these questions remain to be answered, especially now that the enzymes responsible are becoming characterized. The results from this study demonstrate the importance of these enzymes in maintaining the HA-PG-rich cartilage extracellular matrix.