Alpha 5 Integrin Mediates Osteoarthritic Changes in Mouse Knee Joints

Osteoarthritis (OA) is one of most common skeletal disorders and can affect synovial joints such as knee and ankle joints. α5 integrin, a major fibronectin receptor, is expressed in articular cartilage and has been demonstrated to play roles in synovial joint development and in the regulation of chondrocyte survival and matrix degradation in articular cartilage. We hypothesized that α5 integrin signaling is involved in pathogenesis of OA. To test this, we generated compound mice that conditionally ablate α5 integrin in the synovial joints using the Gdf5Cre system. The compound mice were born normally and had an overall appearance similar to the control mice. However, when the mutant mice received the OA surgery, they showed stronger resistance to osteoarthritic changes than the control. Specifically the mutant knee joints presented lower levels of cartilage matrix and structure loss and synovial changes and showed stronger biomechanical properties than the control knee joints. These findings indicate that α5 integrin may not be essential for synovial joint development but play a causative role in induction of osteoarthritic changes.


Introduction
Osteoarthritis (OA) is one of the most common skeletal diseases and involves pathological changes in synovial joint components such as articular cartilage and synovium. Pathological changes seen in articular cartilage in OA include irregularity and loss of articular surface, loss of proteoglycan matrix and alterations of collagen fibrils and fibers, and cleft and erosion of articular cartilage [1]. These changes result from many cellular events including cytoskeletal changes, proliferation, matrix synthesis and degradation, cell senescence and apoptosis, as well as hypertrophy [2]. In addition to alterations of cartilage structure and function, thickening of subchondral bone and ectopic bone formation (osteophyte) occur in OA [3,4]. Synovial changes, including synovitis and hyperplasia of synovial cells, are found in synovial joints from the early stages of osteoarthritis, and have been studied as causative events and therapeutic targets of this condition [5,6]. Continuous and extensive efforts have been made to understand the biology of joint homeostasis and the pathogenesis of OA for the development of therapy for this disease, though no effective disease-modifying osteoarthritis drug has been discovered.
Alpha 5 integrin (α5 integrin) is a fibronectin receptor that mediates a variety of biological phenomena including mesoderm induction, vascular development, and neural crest development [7][8][9][10]. Human articular chondrocytes express α5β1 together with many other heterodimers of integrins such as α1β1, αvβ5, αvβ3 and α3β1 [11,12]. Among these integrins, α5β1 has been demonstrated to play roles in synovial joint development [13] and in the regulation of chondrocyte survival, matrix degradation, and expression of cytokines and non-cartilaginous collagens [14][15][16][17][18]. Based on these findings, we hypothesized that α5 integrin signaling is involved in pathogenesis of OA. To test this hypothesis, we generated conditional knockout mice lacking α5 integrin expression specifically in synovial joints through use of the GDF5Cre system [19,20] and analyzed the pathological and biomechanical changes of knee joints after OA surgery. The findings suggest that α5 integrin mediates the progression of OA, possibly targeting synovial cells.

Materials and Methods Mice
All animal experiment procedures were approved by the Institutional Animal Care and Use Committee of the Children's Hospital of Philadelphia. The Institute maintains veterinary personnel who supervise our work and answer any question we may have related to the use and care of vertebrate animals. When pain/distress are observed, the animals will be treated with Buprenex (0.1-2.0 mg/kg), as well as crushed or wet food. If pain or distress continues, we will euthanize the mouse regardless of the scheduled endpoints. The criteria that determine discomfort/distress/pain are any three of the following signs: Abnormal posture, slow, careful or abnormal (waddling) gait, low activity, slow eating, cowering or vocalizing while handling, change in eye or coat appearance and weight loss. We strictly follow the American Veterinary Medical Association Panel on Euthanasia for adult mice and NIH Euthanasia Guidelines for mouse embryos and neonates. Specifically, mice older than 14 days will be euthanized by CO 2 inhalation. We will assure euthanasia by one of the following criteria: Cervical dislocation after no response to tail or toe pinch, no respiration or heartbeat after thirty seconds continuous monitoring, or rigor mortis. We will give anesthesia (isoflurane inhalation) to mouse embryos older than E15 and neonates up to 14 days prior to decapitation. Mice will be anesthetized by inhalation of 1-5% of isoflurane during OA surgery and tail clipping for genotyping.
α5 integrin was conditionally knocked out of the synovial joints by generation of the triple compound transgenic mouse that encodes Cre under the control Gdf5 expression (Gdf5Cre) [19], flox α5 integrin [9] and α5 null alleles [8]. First we mated the Gdf5Cre mouse with the α5 null mouse. The resulting double-het (Gdf5Cre; α5null/wt) mouse was mated with homozygous floxed α5 integrin mouse (α5fl/fl) to generate the triple compound mouse (Gdf5Cre; α5null/fl, hereafter called α5 CKO). The genotyping for the Gdf5Cre, α5 null and flox α5 alleles and the excised allele of the floxed α5 have been described elsewhere [9]. The Gdf5Cre; α5fl/wt, α5fl/ null or α5fl/wt were used as the control since they do not show significant differences in OA phenotype after OA surgery. To monitor the fate of Gdf5-lineage cells, we generated the Gdf5Cre;Rosa-GFP compound mouse. To determine localization of α5 integrin, we used C57BL/6j (Jackson Lab, Bar Harbor, ME, USA) mice.

OA surgery
Under anesthesia, we transected the medial collateral ligament and removed the medial meniscus in the right knee in both α5 CKO and control littermate male mice (n = 9/group) at 3 months of age as described previously [21]. Knees were collected and examined by histological and histochemical means 1 month after surgery.

Histology and immunostaining
The knee joints were fixed with 10% formalin or 4% paraformaldehyde overnight and decalcified with EDTA for 2 weeks. The paraffin sections were stained with hematoxylin and eosin or Safranin O staining. To detect apoptotic cells, we performed TUNEL staining using the Apop- Tag

Histological evaluation of osteoarthritic changes
The slides of knee joints were stained with hematoxylin and eosin or Safranin O in increments of five slides and inspected by two independent researchers. The extent of changes in medial tibial plateau and medial femoral condyle of articular cartilage was evaluated using Mankin and OARSI scoring systems [22]. The extent of changes in the synovium was evaluated following the scale from a previous report [23]: 0, normal (synovial lining 1-3 cells thick); 1, mild inflammation (synovial lining 4 or 5 cells thick, increased cellularity); 2, moderate inflammation (synovial lining 6-8 cells thick and/or increase in cellularity); and 3, severe inflammation (synovial lining >9 cells thick and dense cellularity). The synovium adjacent to the meniscus toward the femur and the tibia were examined separately (4-5 sections/mouse).

Nanoindentation
A separate group of mice were used for atomic force microscopy (AFM)-based nanoindentation, following our previously established procedures [24].
Briefly, the condyles from hind knees were dissected immediately after euthanasia by carbon dioxide, and mounted on steel AFM sample disks. Through this process, the joints were kept in sterile phosphate buffered saline (PBS) with protease inhibitors to minimize post-mortem degradation. Nanoindentation was performed on the surfaces of medial condyle cartilage using a borosilicate microspherical tip (R 5 μm, nominal spring constant k 7.4 N/m, AIO-TL tip C, NanoAndMore) and a Dimension Icon AFM (BrukerNano). On each condyle, at least 10-15 different locations were tested up to an indentation depth of~1 μm at an approximately 10 μm/s indentation rate. The effective modulus, E ind , was calculated by fitting the loading portion of each indentation force-depth curve to the elastic Hertz model via least squares linear regression, where R is the tip radius, and ν is Poisson's ratio (~0.1 for cartilage) [25].

Statistics
Student's t-tests or two-way factorial ANOVA followed by Bonferroni post-hoc multiple comparison tests were used to identify the differences. The threshold for significance for all tests was set as p<0.05.

Ablation of α5 integrin in synovial joints in mice
The GDF5Cre; α5fl/mice were born normally and had an overall appearance similar to the control mice. Histological analysis of the knee joints of the GDF5Cre; α5fl/mice did not show significant abnormalities in the structure of synovial joints including articular cartilage (Fig 1A  vs 1D) and synovium (Fig 1G vs 1J) compared to the control mice. In the control mice, expression of α5 integrin was strongly expressed in articular cartilage at P0 (S1 Fig) and gradually decreased, but remained in the articular surface ( Fig 1B and S1 Fig) at 4 weeks of age. α5 expression of the synovium was very strong (Fig 1H). In contrast, the α5 CKO mouse did not show positive staining for α5 integrin in articular cartilage (Fig 1E and S1 Fig) and presented much weaker staining in synovium (Fig 1K) while the perichondrium adjacent to growth plate showed α5 staining similar to the control mice (S1 Fig).

Analyses of osteoarthritic changes in the α5 CKO articular cartilage
Next we compared the osteoarthritic changes between the α5 CKO and the control mice after OA surgery. Interestingly, we observed significantly milder histological osteoarthritic changes in the tibial medial plateau and femoral medial condyle of the α5 CKO mice (Fig 2B, 2D and 2F) than those in the control mice (Fig 2A, 2C and 2E). Both matrix loss and structural changes were less evident in the α5 CKO articular cartilage compared to the control articular cartilage. Histological evaluation by Mankin and OARSI score methods demonstrated that the α5 CKO articular cartilage in the OA surgery side showed significantly lower values than the control articular cartilage (Fig 2G and 2H).
To examine whether milder histological changes in the α5 CKO articular cartilage was related to the changes in cartilage biomechanical properties, we performed AFM-nanoindentation assessment on the femoral condyles with or without surgery. The α5 CKO articular cartilage with OA surgery showed a significantly higher effective indentation modulus (Eind) than the control (Fig 3, p = 0.0482). In addition, the average of the Eind of the α5 CKO articular cartilage without surgery was found to be marginally higher than the control (p = 0.0563, Fig 3).
To gain more insights into understanding pathological changes in articular cartilage in the α5 CKO mice, we examined typical molecular changes associated with OA. The α5 CKO group (Fig 4A and 4C) contained a smaller number of TUNEL-positive cells than the control group (Fig 4B and 4D) (S2 Fig, TUNEL). Semi-quantitative analysis for number of TUNEL positive cells confirmed fewer TUNEL-positive cells in the a5 CKO group than the control group ( Fig  4E). Expression of collagen 10, a hypertrophic marker, was found in the eroded deep zone of the control articular cartilage (S2E and S2F Fig, arrows). The α5 CKO articular cartilage had

Analysis of synovial changes in the α5 CKO mice
The knee joints of the control mice showed more evident hyperplasia of synovium (Fig 5A and  5C) compared to the α5 CKO mice (Fig 5B and 5D). The control synovial lining tissue had multi-layer of cells with high cellularity (Fig 5C, double headed arrow) containing CD31-positive vessels (Fig 5E, arrows). The difference in synovial change between the control and the α5 CKO group was confirmed by semi-quantitative analysis (Fig 6). Moreover, the margin of medial tibia in control mice contained ectopic cartilage (Fig 5G), while that of the α5 CKO mice did not (Fig 5H).

α5 integrin expression in the OA joints
We examined changes in α65 integrin expression after OA surgery. The staining of α5 integrin was weaker in the OA articular cartilage than that in the normal articular cartilage (Fig 7A vs  7B). In contrast, α5 integrin was found throughout the hyperplastic lesion of the synovium (Fig 7F). These findings suggest that ablation of α5 integrin in the synovium results in reduction in synovial changes. To examine how Gdf5-lineage cells behave in OA, the Gdf5;Rosa-GFP reporter mice received OA surgery. Four weeks after surgery we examined which cells express GFP to monitor the distribution of GDF5-lineage cells in the knee joints.

Discussion
Results in this study demonstrated that while α5 integrin may not be essential for synovial joint development and structural maturation, it can play an important role in osteoarthritic changes in articular cartilage and synovial tissue. Osteoarthritis is the most common joint disease and causes clinical symptoms, including pain and stiffness. The results from this study provide important insights into the mechanism of osteoarthritis and would pave the way toward the development of new and effective therapeutic drugs to prevent or ameliorate trauma-induced OA.

The role of α5 integrin in chondrocytes
We observed here that ablation of α5 integrin reduced matrix loss and structural changes in articular cartilage after OA surgery. The α5 CKO articular cartilage showed smaller values of the modified Mankin score and OARSI score (Fig 2H and 2G). Furthermore, the α5 CKO articular cartilage also had a higher nanoindentation modulus (Fig 3), indicating a better load bearing function than wild-type cartilage. At the cellular level, the number of apoptotic cells was significantly smaller in the α5 CKO than that in the control (Fig 4). The finding suggests that α5 integrin may regulate osteoarthritic responses through mediating the apoptotic signal. Pulai et al. [14] have demonstrated that incubation with the α5β1 antibodies induces cell death of articular chondrocytes in suspension culture and concluded that α5β1 integrin provides matrix survival signals. In contrast, we observed a decrease in apoptotic cells in the α5 CKO articular cartilage, suggesting that α5 integrin participates in cell apoptosis. The discrepancy can be accounted for through the following explanations. First, previous findings are based on in vitro culture experiments while the results in this study were from in vivo experiments. Articular chondrocytes behave differently in vitro than in vivo. Furthermore, the articular chondrocytes Previous reports have demonstrated that α5 β1 integrin mediates matrix degradation induced by fibronectin fragments in cultured chondrocytes [15]. Furthermore, the expression of inflammatory-related molecules, IL-1β, IL-6, IL-8, PGE2 and nitrite (NO) were stimulated in articular chondrocytes when they are treated with the functional antibodies of α5β1 [17]. The phenotype of the α5 CKO mice observed in this study is likely consistent with these findings. The levels of fibronectin [26,27] and fibronectin fragments (30-200 kD) are increased in OA articular cartilage and synovial fluid [28]. Fibronectin fragments are strong stimulators of MMPs [29]. Therefore, the absence of α5 integrin may attenuate fibronectin fragment-induced matrix degradation in the α5 CKO mice. It has been demonstrated that α5 integrin expression increased in articular cartilage after partial menisectomy OA surgery in rats [30,31] while our results showed reduction in its expression (Fig 7). A decrease in α5 integrin staining was observed particularly in the region that showed severe structural defects. Thus the reduction in α5 integrin expression seen in our model may be associated with reduction in cell metabolic activity or cell death. This assumption is consistent with our findings that articular cartilage at the surgery side contained many apoptotic cells (Fig 4) and a lower level of GFP expression in the GDF5;Rosa-GFP mice compared to the articular cartilage at the contralateral side (S3 Fig).

The role of α5 integrin in synovial changes
The OA model we used in this study is a hemisectomy of medial meniscus in addition to incision of medial collateral ligament, which is a more severe condition than the DMM model that involves transection of the medial meniscotibial ligament [32]. We observed obvious synovial changes at an earlier time point than in the DMM model. The synovium and fat pad expressed α5 integrin at a higher level than articular cartilage. The Gdf5Cre system induces Cre activity in articular cartilage but also does so in other joint components, including synovium and fat pad [33]. We demonstrated that thickened synovial tissue is composed of the cells labeled with Gdf5Cre ( S3 Fig) and showed strong staining for α5 integrin. Thus, deletion of α5 integrin may result in a marked impact on synovium. Indeed, synovial changes in the control mice were much more evident than in the α5 CKO mice. Synovial changes observed in the control group include hyperplasia of synovium (Fig 5A and  5C), an increase in vessels (Fig 5E), and appearance of ectopic cartilaginous lesions (Fig 5G). In the α5 CKO group, thickening of the synovium (Fig 5B and 5D and Fig 6) and ectopic cartilage formation (Fig 5H) were to a lesser extent, indicating that α5 integrin mediates these changes directly or indirectly. α5 integrin has been demonstrated to regulate cell proliferation via cell adhesion or spreading [34,35]. Furthermore, α5 and α v integrins contribute to embryonic vascular development and tumor angiogenesis [9,36]. Thus, contribution of α5 integrin to hyperplasia of the synovium and vessel formation may be direct. Previous reports have shown that α5 integrin is involved in up-regulation of MMPs (MMP-1, 3 and 13) and inflammatory cytokines (IL-1α and IL-1β) in synovial cells in response to fibronectin fragments [37,38], suggesting that α5 integrin-mediated up-regulation of MMPs and inflammatory cytokines in synovial cells may result in stimulation of loss of matrix and destruction of cartilage. Requirement of α5 integrin has been reported in growth plate chondrocytes in suspension culture [39]. Therefore, absence of α5 integrin would inhibit the growth of ectopic cartilaginous lesions. We need further investigation to elucidate how α5 integrin participates in induction of ectopic cartilage formation.

Conclusions
In summation, our results demonstrate that ablation of α5 integrin in synovial joints reduces susceptibility to osteoarthritic changes in mice and suggest that α5 integrin is a stimulator of osteoarthritis.