Fibrochondrogenic potential of synoviocytes from osteoarthritic and normal joints cultured as tensioned bioscaffolds for meniscal tissue engineering in dogs

Meniscal tears are a common cause of stifle lameness in dogs. Use of autologous synoviocytes from the affected stifle is an attractive cell source for tissue engineering replacement fibrocartilage. However, the diseased state of these cells may impede in vitro fibrocartilage formation. Synoviocytes from 12 osteoarthritic (“oaTSB”) and 6 normal joints (“nTSB”) were cultured as tensioned bioscaffolds and compared for their ability to synthesize fibrocartilage sheets. Gene expression of collagens type I and II were higher and expression of interleukin-6 was lower in oaTSB versus nTSB. Compared with nTSB, oaTSB had more glycosaminoglycan and alpha smooth muscle staining and less collagen I and II staining on histologic analysis, whereas collagen and glycosaminoglycan quantities were similar. In conclusion, osteoarthritic joint—origin synoviocytes can produce extracellular matrix components of meniscal fibrocartilage at similar levels to normal joint—origin synoviocytes, which makes them a potential cell source for canine meniscal tissue engineering.

solution (Ethidium homodimer and Calcein AM Live/Dead Viability Assay, Invitrogen, Carlsbad, CA) for 20 minutes at 37˚C, 5% CO2, 95% humidity. Cells were then visualized in at least five regions of the bioscaffolds, (and two in the center and three on the periphery, at approximately the 2, 6, and 10 o'clock positions) using a laser microscope (Eclipse Ti-u Laser Microscope, Nikkon, Japan). The number of viable (green) and non-viable (red) cells per each field counted by hand. Due to the complex three -dimensional nature of the bioscaffolds, these cell counts provided an estimate of cell viability.
Immunohistologic Analysis: Two TSB per dog were fixed in 10% buffered formalin, paraffin embedded, and tissue blocks cut in 4µm sections for histologic and immunohistologic analysis. All slides were labelled with randomly generated acquisition numbers and analyzed in a blinded fashion. Sections were stained with Hematoxylin and Eosin ("H&E"), Masson's Trichrome, and Toluidine Blue. Cell morphology and general ECM architecture was assessed using H&E; organization and intensity of collagen staining was described using Masson's Trichrome, and intensity of GAG staining was assessed using Toluidine Blue.
Each of these histologic observations was assigned a score (Table 2). Then a histologic intensity coefficient was calculated for each ECM component, as follows: [[(Extracellular matrix staining intensity score) x (percentage area coverage of positive staining score)] + [(Intracellular staining intensity score) x (percentage positive staining cells score)]]/2 (Table 1).
Tissue Weight: One TSB per dog was lyophilized and a dry weight obtained. Samples were digested in 1.0ml Papain Solution (2mM Dithiothreitol and 300ug/ml Papain) at 60ºC in a water bath for 24 hours. This papain digest solution was used to obtain double stranded DNA (dsDNA), GAG and collagen content of the bioscaffolds.
DNA Quantification: Double stranded DNA quantification assay (The Quant-iT PicoGreenTM Assay, Invitrogen) was performed per manufacturer's instructions; double stranded DNA extracted from bovine thymus was used to create standards of 1,000, 100, 10, and 1 ng/mL. Standard and sample fluorescence was read by a fluoromoter (Qubit, Invitrogen) at 485nm excitation/ 528nm emission, and dsDNA concentration was determined based on the standard curve.
Biochemical ECM Analysis: Glycosaminoglycan content was determined by the di-methylmethylene blue sulfated glycosaminoglycan assay (Farndale et al., 1986) using a spectrophotometer (Synergy HT-KC4 Spectrophotometric Plate Reader and FT4software, BioTec, Winooski, VT). Collagen content was determined by Erlich's hydroxyproline assay, as described by Reddy et al. (Reddy and Enwemeka, 1996). Hydroxyproline content was converted to collagen content using the equation: μg hydroxyproline x dilution factor/ 0.13 = μg collagen (Ignat'eva et al. 2007), because hydroxyproline consists of approximately 13% of the amino acids in human meniscal collagen (Fithian, et al., 1990). Collagen and GAG content were standardized to tissue dry weight as percentage of dry weight, to compare the experimental neotissues to previously reported normal meniscal ECM content (Eyre and Wu, 1983). Total GAG and collagen content were also reported in μg/neotissue to measure total synthetic activity over the course of 30 days in each TSB. GAG and collagen content were additionally standardized to dsDNA content using the following equations: [μg GAG/ ug dsDNA](Li and Pei, 2011) and [μg collagen/ ug dsDNA] to identify chondrogenic cellular activity of each tested cell origin.
Real-Time RT-PCR: One TSB per dog was snap frozen in liquid nitrogen and stored at -80 °C. Total RNA was isolated using the phenol-chloroform extraction (Chomczynski P, 1986) with slight modifications. Samples were pulverized using a liquid nitrogen-cooled custom-made stainless steel pulverizer and homogenized in trizol (Trizol, Qiagen Sciences, 0.025mL/mg of tissue) and mixed with chloroform. The aqueous phase was then treated with isopropanol to precipitate nucleic acids. RNA of samples was purified using on-column DNAse digestion (RNeasy, Qiagen Sciences).
First-strand cDNA synthesis was performed from 400 ng total RNA(SuperScript III First-Strand Synthesis System, Invitrogen Life Technologies, Carlsbad, CA) and Oligo-(dT)20 primers.
To control for possible genetic DNA contamination, non reverse-transcribed samples were also processed. Pre-designed primers and probes (Taq-Man® Primers and Probes, Applied Biosystems Inc., Foster City, CA) were obtained for each of the genes of interest: IL-1β, IL-6, TNF-α, SOX-9 (an embryonic chondrogenic transcription factor), collagen type I α1, collagen type II α1, aggrecan, and the reference gene GAPDH (see Appendix 1). All assays were confirmed to amplify their targets at 95% or greater efficiency using RNA from tissues of interest. Quantitative real-time PCR was performed (StepOnePlus RT-PCR System, Applied Biosystems Inc.) using a proprietary reagent system (TaqMan Gene Expression Master Mix, Applied Biosystems Inc.) Controls included template-free negative controls and non reverse-transcribed negative controls. All samples were run in triplicates and all negative controls were run in duplicates for 40 cycles (15 seconds at 95°C, 1 minute at 60°C) after 2 minutes of incubation with Uracil-DNA Glycosylase at 50°C, and 10 minutes at 95°C of enzyme activation.
Quantitative gene expression was determined in triplicates using the comparative CT method (Schmittgen TD, 2008). The gene GAPDH was used as internal control (housekeeping gene). Threshold cycles (CT) for each gene were defined by recording the cycle number at which fluorescence reached a gene-specific threshold. Fold changes for gene expression data were calculated using the following formula: fold change = 2 -ΔΔCT = [(CTgene of interest -CThousekeeping gene GAPDH)nTSB -(CTgene of interest -CThousekeeping gene GAPDH)oaTSB].

Statistical Methods
A D'Agostino & Pearson omnibus normality test was performed on all data to test for normality. Cell harvest data was non-parametric data and was analyzed with a Wilcoxon matched-pairs signed rank test, and data reported as median and interquartile range. Significance The effect of osteoarthritis (osteoarthritic versus normal joint status) on gene expression and ECM composition was analyzed using a 2-tailed Student's t-test, assuming unequal variances. The effect of osteoarthritis on the histologic scoring of TSB extracellular matrix formation was analyzed using a non-parametric Mann-Whitney U-test; ranking of the histologic scores was performed using a Kruskal-wallis analysis on ranks followed by a Fisher's exact test.
Significance was declared at P < 0.05. Data were analyzed using Statistical Analysis System, version 9.3 (SAS Institute Inc., Cary, NC).

Cell Harvest:
The mean age of dogs with stifle osteoarthritis was 4.7 years (range: 2-8 years). Breeds represented included: Golden Retriever (1), American Staffordshire Terrier (2), Labrador Retriever (3), Australian Shepherd (1), Rottweiler (2), Boston Bull Terrier (1), Goldendoodle (1), and mixed breed (1), with 7 neutered males, 4 spayed females, and one intact female dog. As observed by a Diplomate of the American College of Veterinary Surgeons -Small Animal, all dogs had marked villous synovial hyperplasia and osteophytosis, and grade 1-2 Outerbridge cartilage lesions of the medial femoral condyle and tibial plateau (Outerbridge, 1961). Cell yield from arthroscopic synovial debris was 1.9 x10 6 ±3.7 x10 5 cells per joint, and cells were 99.5% ±0.002 viable at harvest. Mean age of dogs with normal stifles was 4.3 years (range: 3-6 years); breeds represented included: Red Tick Hounds (4), Labrador Retriever (1), and American Staffordshire cross (1), with 3 female intact dogs, 2 male intact dogs, and one neutered male. Cell yield per joint was 1.4 x10 7 ±2.6 x10 6 per joint and cells were 99.5% ±0.01 viable. As the entire stifle joint synovial membrane could be harvested post mortem, a greater volume of tissue and thus greater cell numbers were obtained from the normal joints versus arthroscopic harvest of the osteoarthritic joints (P=0.01).

Cell Culture and Cell Characterization:
At 4 th passage, cells were transferred into eight 150cm 2 flasks in order to have enough TSB for tissue analyses. This, however, resulted in greater cell seeding numbers for nTSB versus oa TSB. Thus, normal joint-origin synoviocytes were seeded at 1.49 x10 7 cells per flask, wherease 6.52x10 6 osteoarthritic joint-origin cells were seeded per flask. At 4 th passage, normal joint-origin cells were 99.0 ±0.4% viable compared with 98.8 ±0.4% viability of osteoarthritic joint-origin cells (P=0.85). Culture duration from tissue harvest to hyperconfluent cell membrane formation and synthesis of TSB was 37.6 days and similar for both cell origins (range 20-49 days).
During the first week of tensioned bioscaffold culture, the culture media phenol red pH indicator changed to yellow by the time the 24 hour media change was required, indicating marked increase in media acidity. In addition, during the first 7-10 days of culture, approximately 2-3 bioscaffolds per normal and osteoarthritic joint unraveled or slipped off their wire hoops (no group differences observed), and were not analyzed in this study. The typical appearance of intact nTSB and oaTSB is pictured in Fig. 2; thickness of TSB was 2-3mm. At harvest, tensioned bioscaffolds from normal dogs had a dry weight of 39.3mg (range 27.5-50.4mg), which was more than for oaTSB (23.6mg, range 10.2-50.1mg; P= 0.008).
Mean estimated cell viability of nTSB and oaTSB was similar, with 78% of cells viable (range: 72-86%). Cell viability was not associated with peripheral versus central location on the TSB. Laser microscopy revealed cells with fusiform, fibroblastic cytoplasm, oriented parallel with the vector of tension, as well as the presence of acellular, circular regions in the bioscaffolds.
Hematoxylin and eosin staining revealed highly cellular bisocaffolds, with layers of fibroblastic cells organized in parallel, as sheets or bands, or variably arranged in whorls, with eosinophilic ECM (Fig.3). Subjectively, both nTSB and oaTSB had heterogeneous extracellular matrix architecture and cell distribution, with regions of dense cellularity and regions of dense extracellular matrix (Fig. 3).
Percent dsDNA content was used to quantify tissue cellularity. Despite an initial higher seeding cell count at 4 th passage, dsDNA accounted for 0.11 ± .02% of nTSB dry weight, versus 0.21 ± 0.03% of oaTSB dry weight (P = 0.01).
Immunohistologically, oaTSB had more ASM positive cells than nTSB; the median histologic score for nTSB was 6 versus 9 for oaTSB (p= 0.0102, Fig.4). Nine of 12 oaTSB had the highest possible ASM histologic scores of 9, whereas none of the nTSB achieved a perfect score of 9 (P=0.009, Fig.4). In 50% of all bioscaffolds, ASM positive cells were concentrated around the bisocaffold periphery and around the margins of what appeared to be spontaneously forming circular defects ranging from 70-600µm (Fig.3). These circular defects corresponded with the acellular regions viewed on laser microscopy. The other 50% of bioscaffolds did not contain circular defects, nor did ASM expression seem to be geographically localizable. Gene Expression: The oaTSB had a greater gene expression of type I collagen (7-fold increase; P = 0.04) and type II collagen (71-fold increase; P = 0.02) and a lower gene expression of interleukin-6 (19-fold decrease; P = 0.001) versus nTSB. No significant changes were observed for relative expression of SOX-9 (P=0.72), aggrecan (P=0.84), and tumor necrosis factor-α (P=0.77; Table 2). Interleukin-1β was not expressed at detectable levels in any bisocaffolds.
Glycosaminoglycan Content: The total GAG content of oaTSB was lower than the GAG content of nTSB (P=0.02; Table 3). After adjustment for dry weight or DNA content, no significant group differences were observed.
Collagen Content: There was no difference in quantified total collagen content of oaTSB and nTSB (Table 3). Similar results were observed after adjustment for dry weight or DNA content.
Masson's Trichrome staining revealed collagen deposited in bands, sheets, and whorls, containing and surrounded by numerous fibroblastic cells lined in parallel with the orientation of the collagen (Fig.6). A significant difference in the median type I collagen histologic scores of nTSB and oaTSB could not be detected, which were 7.5 and 6.0, respectively (P=0.11, Fig.4). However, 4 of 6 nTSB had a type I collagen histologic score greater than 7.5, versus only 1 of 12 oaTSB had a collagen score of 7.5 (P=0.02, Fig.4). Histologically, nTSB had more type II collagen than oaTSB (Fig.6); median type II collagen histologic scores were 4.0 in nTSB and 2.5 in oaTSB, (P= 0.03, Fig. 4). None of the oaTSB had a score greater than 2.5 whereas 5 of 6 nTSB had a collagen type II histology score of 2.75 (P=0.0007, Fig 4).
Synovial Macrophage Content: Based on immunohistochemistry, no macrophages (Type A synoviocytes) were found in any bisocaffolds (Fig.7).

Discussion
Previous studies comparing in vitro canine synoviocyte fibrochondrogenesis in monolayer culture (Warnock et al., 2011), and canine synoviocyte chondrogenesis in micromass culture (Krawetz et al., 2012) concluded that osteoarthritic synoviocytes had inferior in vitro fibrochondrogenic potential, compared with normal synoviocytes. Fiorito and workers came to a similar conclusion in a study comparing in vitro chondrogenesis of human synoviocytes grown in pellet culture, as determined by histologic analysis (Fiorito et al., 2005). In contrast, with the culture conditions in the present study, especially providing conditions for self-tensioning, cells originating from osteoarthritic joints increased type I and II collagen gene expression, and oaTSB contained similar total collagen content, as compared to nTSB. While tissue dry weight and thus total GAG content of oaTSB was lower than nTSB, a significant difference in GAG content standardized to dry weight and cellularity could not be detected between oaTSB and nTSB.
Histologic analysis using toluidine blue, a semi-quantitative measure of GAG, revealed more GAG deposition in oaTSB than nTSB. Thus, the greater dry weight of nTSB versus oaTSB was likely due to unmeasured ECM components such as fibronectin, type III and VI collagen, and  , 1990;Price et al., 1996). These findings also indicate that given the chance to self-tension, autologous, diseased synoviocytes can produce the ECM components of fibrocartilage in vitro at a comparable level of normal joint-origin synoviocytes.
The unstable mechanical environment and inflammatory environment of the cranial cruciate ligament deficient joint favors synovial intimal hyperplasia and synovial membrane and joint capsule fibrosis (Bleedorn et al., 2011;Buckwalter, 2000;Oehler et al., 2002;Smith et al., 1997), all of which were encountered in the osteoarthritic joints in the present study. The in vivo pathogenic synovial hyperplasia may have accounted for the collagen gene upregulation seen in oaTSB. Rat and human osteoarthritic synoviocytes spontaneously express TGFβ-1 and its receptor (Fiorito et al., 2005;Mussener et al., 1997), which is a pro-collagen and chondrogenic growth factor (Daireaux et al., 1990;Leask and Abraham, 2004;Miyamoto et al., 2007;Pangborn and Athanasiou, 2005a, b;Pei et al., 2008b). Upregulation of TGFβ-1 and its receptor may also be a plausible mechanism for oaTSB collagen gene upregulation. Collagen II upregulation seemed to occur independently of SOX-9 expression, a finding duplicated in cultured human osteoarthritic chondrocytes (Aigner et al., 2003). Additionally, decreased expression of IL-6 gene may be a mechanism for the observed upregulation of type II collagen genes in oaTSB; IL-6 has been found to inhibit chondrogenic differentiation of murine marrow mesenchymal cells (Wei et al., 2013). Further research is required to confirm the mechanism of hyaline chondrogenic ECM formation in canine TSB, through immunohistochemistry of TGFbeta-receptor and SMADfamily protein expression (Xu et al., 2012).
Despite equal quantities of non-specific collagen in nTSB and oaTSB, immunohistologic analysis revealed less type I and type II collagen in oaTSB, particularly in the ECM. Post translation regulation by prolyl-4-hydroxylases (Grimmer et al., 2006) or ECM degradation by IL-6 has also been found to increase gingival fibroblast synthesis of type I collagen in vitro (Martelli-Junior et al., 2003), and increase type I collagen synthesis by tenocytes in vivo (Andersen et al., 2011). It is possible that the decreased IL-6 gene expression in oaTSB synoviocytes also decreased type I collagen formation as seen on histologic analysis. One weakness of our study was that expression of type I and II collagen was not corroborated with a Western blot, nor quantified via ELISA, to further our understanding of this discrepancy between histologic collagen expression and collagen gene expression. Additionally we did not characterize the percentage and type of mesenchymal progenitor cells present in normal versus osteoarthritic synovium; difference in number and chondrogenic potential of these cells may have also accounted for a difference in collagen ECM formation.
Other osteoarthritic cell types, such as chondrocytes, have reduced cell proliferation compared to normal cells in monolayer culture (Acosta et al., 2006). In contrast, oaTSB contained more dsDNA per dry weight than nTSB, despite the lower harvest cell yield and lower cell seeding density at 4 th passage of osteoarthritic joint-origin synoviocytes. There was an intrinsic weakness of our study; by clinical necessity, synovium from osteoarthritic joints was harvested using a different technique (arthroscopy) than the normal joints (arthrotomy), and more synoviocytes can be obtained via arthrotomy. Although cell growth kinetics was not the focus of this study, cell culture media containing 17.7% FBS likely provided mitotic stimuli to support and increase oaTSB cellular proliferation. The markedly hyperplastic state of the synovium in vivo may also have primed the osteoarthritic cells to continue to proliferate in vitro. Cell viability was high at harvest and at the start of 4 th passage, but declined in all TSB; due to the long culture period, cell mortality may have been caused by senescence. Additionally, as evidenced by media color changes, inadequate nutrient delivery to TSB in the culture wells and daily shifts in pH may have also led to nTSB and oaTSB cell mortality. This cell mortality may have affected ECM formation in both groups: the collagen content of nTSB (12%) and oaTSB (16%) did not reach that of the healthy meniscus, at 60-70% of dry weight (McDevitt and Webber, 1990), although the GAG content of nTSB (1.7%) and oaTSB (2%) did approximate the 2-3% GAG per dry weight of the whole meniscus (McDevitt and Webber, 1990;Stephan et al., 1998).
Consistent with prior studies (Warnock et al. 2013), all oaTSB and nTSB in the present study were negative for any macrophages, which have been reported to contaminate human osteoarthritic synoviocyte monolayer cultures and reduce in vitro chondrogenic activity (Pei et al., 2008a) by contributing to the inflammatory milieu. In the present study, 4 passages and long term culture as TSB likely eliminated any non-adherent cells, including synovial macrophages (Krey et al., 1976). Synovium from osteoarthritic joints has also been found to express inflammatory cytokines (Fiorito et al., 2005). Both nTSB and oaTSB expressed similar RNA quantity of the TNFα gene, indicating an inflammatory response in in vitro culture (Lindroos et al., 2010), independent of the diseased status of the cell origin. Paradoxically, IL-6 expression was decreased in oaTSB. Although the exact reason for this is unclear, decreased IL-6 gene expression may represent the response of synoviocytes from osteoarthritic joints to the change in environment; from the high motion, inflamed stifle containing multiple injured cell types (ligament, cartilage, meniscus, synovium) to the static tension of TSB culture and high FBS concentration cell culture media.
Decreased IL-6 in oaTSB may have reflected better mechanical homeostasis (Asparuhova et al., 2009;Chan et al., 2011;Gardner et al., 2012) in the cells in oaTSB: the majority of cells in oaTSB were uniformly positive for ASM, while 10-50% of nTSB cells were ASM positive.

may explain increased ASM in the cells of oaTSB.
In the present study, staining for ASM was positively associated with the formation of circular defects, indicating that the ECM was not strong enough to prevent tears from forming during ASM-mediated self-tensioning (Kambic et al., 2000;Vickers et al., 2004;Warnock et al., 2013).
Given the higher dsDNA content of oaTSB and the high cellularity of the TSB, these defects may have also been caused by increased cell turnover.
Conclusion: When cultured as TSB in high concentrations of FBS, osteoarthritic jointorigin synoviocytes can produce ECM components of meniscal fibrocartilage at similar levels to normal joint-origin synoviocytes. Potential reasons for this include increased collagen and decreased IL-6 gene expression and the greater GAG and ASM staining in oaTSB compared with nTSB Osteoarthritic joint-origin synoviocytes are a viable cell source toward meniscal tissue engineering. Further investigation of culture environments to optimize cell viability and ECM formation and strength are justified due to the promising data reported here.
Acknowledgements: the authors give a profound thanks to Jesse Ott, for technical assistance with performing the assays used in this study.   Hyperconfluent cell sheets.
Representative example of a hyperconfluent cell sheet just prior to harvest for formation of a tensioned synoviocyte bioscaffold. A) gross appearance of the hyperconfluent cell sheet in monolayer culture, and B) phase contrast photomicrograph of the hyperconfluent cell sheet, 10X objective magnification, bar= 100μm.

Figure 4
Histology scores for tensioned synoviocyte bioscaffolds Histology scores for type I collagen, typeII collagen, glycosaminoglycan, and alpha-smooth muscle actin in normaljoint-origin tensioned synoviocyte bioscaffolds versus osteoarthritic joint-origin tensioned synoviocyte bioscaffolds, showing the median and interquartile range.
Histologic scores for collagens type 1 and 2 were calculated as follows: Histologic score=