Synoviocytes protect cartilage from the effects of injury in vitro

Equine cartilage samples were extracted from cadaveric stifle joints within sixteen hours of death from 12 different horses (ages 3–5 years) that died for reasons unrelated to the musculoskeletal system and factors not influencing this study. Using aseptic technique (sterile gloves, instruments and media) osteochondral plugs 5 mm in diameter (varying heights) were harvested from the trochlear ridges of the left and right limbs using a customized cylindrical osteotome (Sontec Instruments, Centennial CO). Due to the non-uniform swelling of the cartilage that occurs during culture when the cartilage remains attached to the subchondral bone, the cartilage was removed from the subchondral bone at the calcified and non-calcified cartilage junction immediately after extraction from the joint. To allow for recovery of the chondrocytes from tissue harvest the explants were cultured for forty-eight hours in full growth medium (DMEM (Invitrogen, Carlsbad, CA), 10% fetal bovine serum (v/v) (Hyclone, Logan, UT), 20 μg/ml ascorbic acid (Sigma Aldrich, St Louis, MO), 1% penicillin and streptomycin (v/v) (Invitrogen, Carlsbad, CA), 10 mM HEPES (Invitrogen, Carlsbad, CA), 0.1 mM non-essential amino acids (Invitrogen, Carlsbad, CA), 0.4 mM proline (Sigma Aldrich, St Louis, MO), 0.25 μg/ml amphotericin B (Invitrogen, Carlsbad, CA)) prior to commencement of the experiments.

Synoviocytes were isolated from synovial tissue harvested from the stifle of 6 different horses than those used for cartilage harvest (ages 3–5) that also died for reasons unrelated to the musculoskeletal system and to this study. Synovial tissue was minced and incubated with media containing collagenase type II (Worthington Biochemical, Lakewood, NJ) in a spin column flask for 4 hours. Synoviocytes were then filtered from the media using a 40 μm filter and culture expanded in full growth medium (Ham’s F12(Invitrogen, Carlsbad, CA), 10% FBS (v/v) (Hyclone, Logan, UT), 1% antibiotic and antimycotic (v/v) (Invitrogen, Carlsbad, CA) and 10 mM HEPES (Invitrogen, Carlsbad, CA)) to passage 3 to ensure >95% of the cells will express a fibroblastic phenotype (Figure 1) [21] with minimal presence of lymphocytes, natural killer cells and macrophages [22] followed by cryopreservation until experiment commencement.

Just prior to injury, the thickness of each cartilage plug (plane perpendicular to the articular surface; average thickness 1.33 mm) was measured using digital calipers. Cartilage plugs were injured as previously defined [23]. In brief, cartilage plugs were removed from media and placed into a sterile polysulphone loading chamber consisting of a well aligned coaxially with an impermeable platen (10 mm diameter) in the absence of media. Using a H1KS benchtop universal testing machine (Tinius Olsen, Horcham PA) an initial compressive tare load of 4 N was applied. After creep equilibrium was attained for 5 seconds, an injurious compression at a rate of 100% strain/second was applied until 60% final cartilage strain was achieved. Compression was then released at the same rate and plugs were immediately placed into the appropriate culture condition.

Synoviocyte and cartilage were co-cultured using a transwell plate with a 0.4 μm microporous insert (Corning Inc., Corning, NY). Cartilage and synoviocytes were randomly paired for co-culture to form 2 sets of experiments. One set of experiments involved the culture of synoviocytes with cartilage for gene expression analysis at 1, 2, 4 and 8 days in co-culture. The other set of experiments involved the culture of synoviocytes with cartilage for histologic and immunologic evaluation at days 8, 16 or 32 in culture. The time points to evaluate gene expression were determined based on previous studies that have evaluated changes in gene expression in cartilage after injury [18, 20]. For histologic evaluation we extended the duration of culture because in previous studies [23] we found injured cartilage required culture for at least 28 days to develop histologic changes. Each experiment was conducted in duplicate, and all experiments repeated a total of 6 times using cartilage from a new horse for each experiment and synoviocytes from 6 different horses total; synoviocytes from each horse were used for one gene expression and one histology experiment. Each experiment consisted of four conditions; injured and control (uninjured) cartilage cultured alone or with synoviocytes. Experiments for gene expression analysis included the fifth condition of synoviocytes cultured alone. Forty-eight hours prior to commencement of experiment synoviocytes were recovered from liquid nitrogen and seeded into the bottom well of a 24-well transwell system at a seeding density of 5000 cells/cm² and cultured in growth media composed of half cartilage DMEM media and half synoviocyte F12 media. Immediately after injury, cartilage plugs were added to the top well of the transwell system, uninjured control cartilage samples were maintained in parallel. All samples remained in culture for 1, 2, 4, 8, 16 or 32 days, with media changes every 2–3 days.

For gene expression analysis, cartilage and synoviocytes were removed from incubation at 1, 2, 4 and 8 days and immediately placed on ice. The cartilage plugs were rinsed with ice cold RNase free PBS then snap frozen in Trizol (Invitrogen, Carlsbad, CA) and stored at −80°C. Similarly the synoviocytes were rinsed with ice cold RNase free PBS, treated with 350 ul of RLT lysis buffer (RNeasy minikit, Qiagen, Valencia CA) and stored in −80°C. For histology and immunohistochemistry (IHC) cartilage samples at days 8, 16 and 32 were removed from culture and sectioned across the diameter perpendicular to the fissures when present (arbitrarily in the middle of the sample when not) and always along a plane perpendicular to the superficial surface. One half of each sample was placed into 10% neutral buffered formalin for histologic analysis and the remaining half was embedded in OCT, snap frozen in liquid nitrogen and stored at −80°C for immunohistochemistry.

To extract RNA from synoviocytes, samples in RLT lysis buffer were thawed and passed through a QIAshredder homogenization column (Qiagen, Valencia, CA) followed by RNA extraction using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions with on column genomic DNA digest using RNase free DNase (Qiagen, Valencia, CA). To extract RNA from cartilage, the plugs were pulverized and homogenized in Trizol (Invitrogen, Carlsbad, CA) on ice using a tissue homogenizer. The homogenized sample was then allowed to incubate at room temperature for 20 minutes in Trizol followed by centrifugation at 12,000 x g for 12 minutes at 4°C. The supernatant was transferred to a new microcentrifuge tube, mixed with 400 ul chloroform, incubated at room temperature for 3 minutes and then centrifuged at 12,000 × g for 15 minutes at 4°C. The aqueous phase was transferred to a new microcentrifuge tube, mixed with 200 ul chloroform, incubated at room temperature for 3 minutes then centrifuged at 12,000 × g for 15 minutes at 4°C. The aqueous phase was transferred to a new microcentrifuge tube to which 500 ul isopropanol was added. The samples were incubated at room temperature for 10 minutes then centrifuged at 12,000 × g for 10 minutes at 4°C to pellet the RNA. The RNA fraction was cleaned with 70% ethanol and re-suspended with RNase free water followed by digestion of genomic DNA using RNase free DNase (Qiagen, Valencia, CA).

Synoviocyte mRNA was evaluated for expression of cyclooxygenase 2 (cox-2), interleukin 1β, 6, 10 (IL-1β, -6, -10), interleukin 1 receptor antagonist protein (IRAP), matrix metalloproteinase 1, 3, 13 (MMP-1,-3,-13), transforming growth factor β (TGF-β), tissue inhibitor of metalloproteinases 1 (TIMP-1), aggrecanase 1, 2 (ADAMTS4, 5) and fibroblast growth factor 2 (FGF2). Chondrocyte mRNA was evaluated for collagen types 1 and 2 (Col1 and Col2), aggrecan, IL-1β, -6, -10, TNF-α, TGF-β, MMP-1, -3, -13, ADAMTS4, 5 and TIMP-1. All expression levels were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as screening of all samples determined no variability in GAPDH expression. Equal concentrations of RNA (5 ng) from each sample were reverse transcribed into cDNA using superscript First-Strand Synthesis for RT-PCR (Invitrogen, Carlsbad, CA). Relative gene expression levels were determined by semi-quantitative real time PCR using TaqMan based probes and primers purchased from the Lucy Whittier Molecular and Diagnostic core facility at University of California, Davis, or from the Orthopaedic Research Center at Colorado State University, Fort Collins (Table 1). All real time PCR assays were run on the ABI Prism 7000 Sequence Detection System (Applied Biosystems, Forster City, CA). Relative gene expression (using the delta delta Ct method [24]) was determined by comparing gene expression levels to baseline, where baseline for synoviocytes is equal to expression in synoviocytes cultured alone, and for injured cartilage cultured alone or in co-culture baseline is equal to control cartilage cultured alone or in co-culture respectively.

Formalin fixed samples were paraffin embedded, and sectioned (5 μm thick) using a Leica RM2255 microtome (Leica Biosystems, Buffalo Grove IL). One slide each from an injured and control sample was stained with Hematoxylin and Eosin (H&E) to evaluate cellular pathologic changes or Safranin O Fast Green (SOFG) to detect changes in regional glycosaminoglycan (GAG) content. All images of slides were obtained using Q-capture (QImaging, Surrey BC).

Immunohistochemical staining was conducted for the detection of aggrecan and collagen type II. Briefly, frozen embedded samples were sectioned (8 μm thick) and mounted onto adhesive slides using the CryoJane® Tape-Transfer system (Instrumedics Inc., St Louis, MO). Sections were probed with mouse antibodies raised against aggrecan at a 1:20 dilution (Novus Biologics, Littleton, CO) or collagen type II using undiluted supernatant (Hybridoma Bank, Iowa City, IA), followed by probing with a goat anti mouse secondary antibody conjugated with horseradish peroxidase at a 1:500 dilution (Jackson Immunoresearch West Grove, PA) and antibody detection with VECTOR® NovaRED™ (Vector laboratories, Burlingame, CA). For negative controls, additional sections were probed with normal mouse serum at a concentration equal to that of the primary antibody.

All histological and IHC sections were blindly evaluated using a previously established grading scale detailing the severity of OA characteristics in equine cartilage [25]. In brief, each region was assigned a grade for each OA characteristic that corresponds to the severity by which the sample deviates in histologic appearance compared to normal articular cartilage where 0 = normal or no change, 1 = slight, 2 = mild, 3 = moderate. To eliminate the inclusion or artifact from tissue harvest in the analysis, the outer 5% of the tissue around the left and right edges and below the deep zone were not graded. One section was graded for each sample. For each OA characteristic, scores for each region (superficial, middle, deep and fissures) were evaluated separately then summed together for each slide, the maximum score for each of the 4 regions was 3 creating a maximum score for all regions combined of 12. Sections stained with H&E were used to visualize generalized chondrocyte cell death (determined by the amount of lacunae lacking discernable nuclei), localized cell death/focal cell loss (the presence of distinct regions devoid of discernable nuclei) and chondrocyte cluster formation (lacunae containing more than one nuclei). Changes to the ECM were determined using both histologic and IHC stained sections. Safranin O Fast Green stained sections were used to identify proteoglycan content of control and treated cartilage. To identify ECM molecules more specifically, sections stained by IHC were used to assess regions containing aggrecan and collagen type II in control and treated cartilage.

Statistical analysis was conducted to determine differences in IHC and histologic staining results accounting for both injury and location using Proc GLIMMIX. Predictive F-values were used to determine statistical differences between injury and control or between regions, with specific comparisons made using a least squares means procedure, both with significance set at α = 0.05.

In synoviocyte samples, mRNA expression of IL-1β, IL-1RA, IL-10 and TNF-α was not detected. There were no significant differences in the expression of cox-2, MMP-3,-13, IL-6, FGF2 and TGFB in synoviocytes cultured with control versus injured cartilage samples (Figure 2A). Synoviocyte expression of MMP-1 (p = 0.0334), ADAMTS4 and 5 (p = 0.0004 and p = 0.0011), and TIMP-1 (p = 0.0010) were significantly affected by culture condition. Synoviocytes cultured in the presence of control cartilage had a significantly higher relative expression of ADAMTS4 and 5 (p < 0.0001 and p = 0.0005) and TIMP-1 (P = 0.0007) compared to synoviocytes cultured with injured cartilage (Figure 2A). Synoviocytes cultured with injured cartilage had significantly higher expression of MMP-1 (p = 0.0095) and significantly lower expression of ADAMTS4 and 5 (p = 0.007 and p = 0.0037) and TIMP-1 (p = 0.0022) compared to baseline synoviocyte culture (Figure 2A). Expression of MMP-1 in synoviocytes was significantly affected by both main effects treatment and duration in culture where synoviocytes cultured in the presence of injured cartilage had significantly greater expression at two days in culture compared to synoviocytes cultured with injured cartilage (p = 0.0339) including baseline (synoviocytes cultured alone) (p < 0.0001) (Figure 2B).

In cartilage samples we were unable to detect mRNA expression of IL-1β, IL-1RA, IL-10, TNF-α, FGF2 or Col1, and there were no significant differences in expression levels of MMP-1, -3, -13, ADAMTS4, TGFβ and IL-6 between cartilage treatments or compared to baseline (Figure 3A). Injured cartilage co-cultured with synoviocytes had significantly higher expression of collagen type 2 (p = 0.0027) and ADAMTS5 (p = 0.0401) compared to injured cartilage cultured alone (Figure 3A) and had significantly higher expression of type 2 collagen (p <0.0001) and cox-2 (p = 0.0283) compared to control cartilage cultured with synoviocytes (Figure 3A). Injured cartilage cultured alone had significantly higher expression of aggrecan compared to control cartilage (p = 0.0371) (Figure 3A). TGF-β levels were significantly highest in injured cartilage cultured in the presence of synoviocytes at day two compared to all other injured cartilage samples (p <0.0001) and control cartilage cultured with synoviocytes (p <0.0001) (Figure 3B).

Histologic sections stained with H&E showed increased chondrocyte cell death in injured cartilage cultured with or without synoviocytes compared to both uninjured controls (p < 0.0001). There was no significant difference in generalized cell death between injured cartilage samples cultured with and without synoviocytes (Figure 4A). Independent of treatment, the incidence of cell death in all cartilage samples was significantly increased at day 32 compared to days 8 and 16 (p < 0.0001) (Figure 4B). Injured cartilage cultured alone had a significantly higher incidence of focal cell loss at day 16 and 32 compared to all other treatment groups (p = 0.0258) (Figure 4C). Chondrocyte cluster formation was significantly more severe in injured cartilage cultured alone compared to all other treatments (p < 0.0001) (Figure 5A). The size of chondrocyte clusters (based on number of nuclei) was significantly affected by both treatment and duration in culture. Chondrocyte clusters in the injured cartilage cultured alone were significantly larger at days 16 and 32 compared to the clusters formed in any other treatment (p = 0.0130) (Figure 5B, C).

Sections stained with SOFG revealed effects of both treatment and duration in culture (p = 0.0058) (Figure 6A). At day 8, there was a significant reduction in total SOFG staining in all treatment groups compared to control cartilage cultured alone (Figure 6A). At day 16 there was a trend of reduced SOFG staining in samples cultured in the presence of synoviocytes, and at day 32 there was no significant difference in SOFG staining between any treatment (Figure 6A).

The results from IHC staining show no significant differences in aggrecan content between any samples (Figure 7). Immunohistochemical staining for collagen type II content indicate collagen was not affected by treatment, however duration in culture had a significant impact with day 32 samples having the most severe reduction compared to all other treatment groups (p = 0.0460) (Figure 6B).

A summary of all histology and immunohistologic changes can be seen in Figure 7.