Joint immobilization prevents murine osteoarthritis and reveals the highly mechanosensitive nature of protease expression in vivo

Authors


  • The work was completed at the Kennedy Institute of Rheumatology when it was a division of Imperial College London; the Institute became a division of the University of Oxford on August 1, 2011.

Abstract

Objective

Mechanical joint loading is critical for the development of osteoarthritis (OA). Although once regarded as a disease of cartilage attrition, OA is now known to be controlled by the expression and activity of key proteases, such as ADAMTS-5, that drive matrix degradation. This study was undertaken to investigate the link between protease expression and mechanical joint loading in vivo.

Methods

We performed a microarray analysis of genes expressed in the whole joint following surgical induction of murine OA (by cutting the medial meniscotibial ligament). Gene expression changes were validated by reverse transcriptase–polymerase chain reaction in whole joints and microdissected tissues of the joint, including the articular cartilage, meniscus, and epiphysis. Following surgery, mouse joints were immobilized, either by prolonged anesthesia or by sciatic neurectomy.

Results

Many genes were regulated in the whole joint within 6 hours of surgical induction of OA in the mouse. These included Arg1, Ccl2, Il6, Tsg6, Mmp3, Il1b, Adamts5, Adamts4, and Adamts1. All of these were significantly regulated in the articular cartilage. When joints were immobilized by prolonged anesthesia, regulation of the vast majority of genes was abrogated. When joints were immobilized by sciatic neurectomy, regulation of selected genes was abrogated, and OA was prevented up to 12 weeks postsurgery.

Conclusion

These findings indicate that gene expression in the mouse joint following the induction of OA is rapid and highly mechanosensitive. Regulated genes include the known pathogenic protease ADAMTS-5. Targeting the mechanosensing mechanisms of joint tissue may offer new strategies for disease modification.

Mechanical stimuli are arguably the most important etiologic factors in osteoarthritis (OA) development. Not only does disease arise in joints where the cartilage has sustained direct trauma (e.g., intraarticular fracture) or indirect trauma (e.g., meniscal injury), but mechanical factors are believed to explain, at least partly, the associations of the disease with aging and obesity (1, 2).

Experimental joint immobilization has been used historically to examine the role of mechanical factors in joint homeostasis and arthritis development in large animals. Joints can be immobilized by casting the limb in a rigid material, which prevents flexion of the joint, or by interfering with the nerve supply to the limb, e.g., by spinal cord or peripheral nerve transection. In nonarthritic joints, non–weight-bearing immobilization leads to reversible cartilage thinning (3, 4, 5, 6). Such thinning of the cartilage is not the same as OA, since it is usually reversible. Moreover, it is not associated with a breech in the integrity of the superficial layer, although in certain circumstances severe thinning may be irreversible upon rapid remobilization (7). Similar findings are seen in human subjects, where cessation of weight-bearing activity (after spinal cord injury) results in a significant decrease in the volume of the articular cartilage (8, 9, 10).

Joint immobilization protects against the development of OA. When OA was induced in dogs either by iodoacetate injection or by transection of the cruciate ligament, casting of the joint (either weight bearing or non–weight bearing) prevented cartilage degradation, thus confirming the crucial role of joint movement on the progression of degenerative arthritis (11, 12, 13).

Surgical destabilization of the medial meniscus (DMM) in the mouse has become a well-established model of OA in recent years, since it produces robust cartilage degradation, osteophyte formation (14), and pain (15, 16) and can be combined with genetic modification for interrogating disease pathways (17–20). It is now well established that OA is not simply due to repeated wear and tear, leading to attrition of the articular surfaces, but that it requires activation of inflammatory response genes, including those that drive catabolic activity in the joint. These enzymes lead to breakdown of the major extracellular matrix components of cartilage, namely, type II collagen and the proteoglycan, aggrecan. In mice, deletion of the aggrecan degrading-enzyme, ADAMTS-5, and the collagenase, matrix metalloproteinase 13 (MMP-13), substantially protects the joint against OA (14, 20, 21). These proteases are likely to be important in human disease also (22–24). Regulation of ADAMTS-5 and MMP-13 has been demonstrated in microdissected rat cartilage 4 weeks following surgical induction of OA (25), and we have previously shown regulation of both enzymes in whole joint extracts from mice obtained 2 weeks following surgical DMM (18).

A link between the expression of pathogenic proteases in the joint in vivo and mechanical loading has not yet been established. The aim of this study was to determine when and where pathogenic genes are expressed in experimental OA joints and to determine the effect of joint immobilization on gene regulation and the development of OA.

MATERIALS AND METHODS

Animals.

Animal experiments were performed following ethical and statutory approval in accordance with local policy. Mice were kept in an approved animal-care facility, with strict compliance to care and usage protocols. All mice were housed 3–6 per cage in standard individually ventilated cages and maintained under 12-hour light/12-hour dark conditions at an ambient temperature of 21°C. Animals were fed a certified mouse diet (RM3; Special Dietary Systems) and water ad libitum. C57BL/6 mice were purchased from Harlan UK.

Surgical procedures.

Male mice (ages 10–12 weeks) were anesthetized by intraperitoneal injection of a 1:2 mixture of fentanyl/fluanisone (Hypnorm; VetaPharma) and midazolam (Hypnovel; Roche), at a dose of 10 ml/kg body weight. Mice undergoing surgery for later experiments were anesthetized by inhalation of isoflurane (3% induction and 1.5–2% maintenance) in 1.5–2 liters/minute oxygen. All animals received a subcutaneous injection of buprenorphine (Vetergesic; Alstoe Animal Health) after surgery. The mice were fully mobile within 2–4 hours after surgery using fentanyl/fluanisone and midazolam or within 4–5 minutes after withdrawal of isoflurane.

Induction of OA by DMM was performed as previously described (20). Mice used as sham controls were anesthetized and prepared as for DMM. The right knee was opened using the same medial parapatellar approach, and the meniscus was identified but the meniscotibial ligament was not released (i.e., capsulotomy alone). Left (contralateral) knees in both the mice subjected to DMM and the sham-operated mice were left as unoperated controls. Some joints were included from mice that had not undergone surgery on either side (naive mice). OA was scored using a validated histologic scoring system (20). Results were expressed as the summed score, which was calculated by adding together the 3 highest total section scores for any given joint (minimum of 8 sections per joint, taken 80μ apart).

For sciatic neurectomy, the posterior hind quarter of the right (ipsilateral) limb was shaved, the skin was prepared with iodine, and a small cutaneous incision was made ∼5 mm lateral to the tail. The sciatic nerve was exposed and carefully dissected, causing minimal trauma to the surrounding muscle. A 2–4–mm segment of nerve was excised. For femoral neurectomy, the ventral portion of the right (ipsilateral) thigh was prepared as described above. An incision was made over the femoral nerve. The neuromuscular bundle was carefully divided between the inguinal ligament and the knee by blunt dissection. The femoral nerve was cut, and an 8–10–mm section was removed.

TaqMan Low-Density Array (TLDA) microfluidic cards.

Complementary DNA (cDNA) was generated from whole-joint RNA using a Promega Reverse Transcription system or, for microdissected tissue, a High Capacity cDNA kit (Applied Biosystems) according to the recommendations of the manufacturer. Reactions were performed in a DNA Engine DYAD Peltier thermal cycler (MJ Research). TLDA microfluidic cards were custom designed to order from Applied Biosystems. All thermocycling was carried out on a 7900HT System (Applied Biosystems) using an ABI Prism 7900HT Sequence Detection System (SDS 2.4; Applied Biosystems). A total of 100 μl of reaction mixture containing 50 μl cDNA template in nuclease-free water and 50 μl of TaqMan 2× Universal PCR Master Mix (Applied Biosystems) was added to each TLDA loading port. The TLDA card was centrifuged and sealed. Ct values were obtained using RQ Manager and Data Assist (Applied Biosystems) and using the 2math image method after manually choosing the Ct threshold.

Immunohistochemistry.

Paraformaldehyde-fixed joints from naive, unoperated mice and from mice 3 days postsurgery (DMM or sham) were decalcified in formic acid for 2 weeks and then embedded in paraffin. Sections were deparaffinized and treated with hyaluronidase (1 mg/ml for 30 minutes at 40°C) (for arginase 1) or with chondroitinase ABC (1 × 10−5 IU for 2 hours at 37°C). Primary antibodies for arginase 1 (Santa Cruz Biotechnology) were used at a dilution of 1:500. For aggrecanase activity, a neoepitope antibody recognizing the FREEE aggrecan terminus (kindly provided by Amanda Fosang, University of Melbourne, Melbourne, Victoria, Australia) was used at a dilution of 1:4,000. Antigen signal was amplified using a Vectastain Elite ABC kit (Pk-6100; Vector) and visualized using a diaminobenzidine peroxidase substrate kit (SK-4100) according to the recommendations of the manufacturer (Vector). Sections were counterstained with hematoxylin.

Statistical analysis.

Group means were compared by analysis of variance, and the difference between groups was determined by post hoc testing with Bonferroni adjustment. P values less than 0.05 were considered significant.

RESULTS

Microarray analysis of genes expressed in the mouse joint following surgical induction of OA.

We induced OA in mice by DMM. Using this model, we and others have shown that there is a robust degradation of articular cartilage over 4–12 weeks in male C57BL/6 mice (14, 18, 20). We determined the gene response in mouse joints at early time points after the induction of OA by extracting RNA from whole joints (after the skin and muscle had been removed) 6 hours and 1, 3, and 7 days after sham surgery or DMM (each performed in triplicate). An Affymetrix ST1 gene array was performed. The full list of regulated genes is available online at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE26475.

Supplementary Table 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131, provides a list of those genes that were up-regulated >1.4 fold (Supplementary Table 1A) or down-regulated <1.4 fold (Supplementary Table 1B) on the microarray. Further analysis of the microarray data, including unsupervised hierarchical and K-means clustering, has been performed previously (26). The most strongly regulated genes included Arg1, Tsg6 (also known as Tnfaip6), Ccl2, and Saa3. Some MMP family members were weakly induced at these early time points following DMM in the mouse, including Mmp3, Adamts1, and Adamts4. Levels of Adamts5 were not significantly elevated in this semiquantitative assessment.

Next, we selected a number of genes for further analysis: those that were either significantly regulated in the microarray (at any time point) or considered to play a putative role in OA pathogenesis. TLDA microfluidic cards were prepared for quantitative validation of the selected genes. Table 1 shows 21 significantly regulated genes, as well as 2 genes of interest that were not regulated in the first 7 days after surgery in the mouse (Mmp13 and Acan). The gene intensity levels as measured on the microarray are also given, and provide an approximate measure of the abundance of messenger RNA for each gene in the naive mouse joint. Quantitative reverse transcriptase–polymerase chain reaction (RT-PCR) showed that there was a robust regulation of all of the genes selected from the microarray. It also revealed regulation of Adamts5, albeit at only 1.56 fold above the levels in sham-operated control mice (Table 1). The modest fold increase in Adamts5 was consistent with the findings of our prior study of mice 2 weeks after DMM (18), and was similar to the fold change determined in articular cartilage 4 weeks after surgical joint destabilization in the rat (25).

Table 1. Genes regulated in whole joints 6 hours and 4 weeks following surgery in mice subjected to DMM versus sham-operated mice*
GeneMicroarray intensity in naive mice6 hours postsurgery6 hours postsurgery4 weeks postsurgery
Sham-operated mice versus naive miceDMM-operated mice versus naive miceDMM-operated mice versus sham-operated micePDMM-operated mice versus sham-operated miceP
  • *

    Values are the mean ± SD or mean fold change (n = 6 mice subjected to destabilization of the medial meniscus [DMM] for Adamts5 analysis; n = 3 mice for all other groups). RNA was extracted from whole joints after skin and muscle had been removed. The genes selected for validation by reverse transcriptase–polymerase chain reaction (RT-PCR) included those that were most highly regulated on the microarray analysis, as well as those that had a putative role in osteoarthritis. Intensity levels from the microarray indicate gene abundance within the mouse joint. RT-PCR was performed using custom-made TaqMan microfluidic cards. NS = not significant; ND = not done.

Arg111.830.487 ± 0.70179.864 ± 6.412163.992≤0.0011.288NS
Ccl222.037.192 ± 0.928212.277 ± 2.90029.516≤0.0011.451NS
Saa315.551.928 ± 0.88526.538 ± 3.39413.765≤0.001ND
Il64.192.567 ± 0.66424.678 ± 5.9599.614≤0.0011.299NS
Il1b53.341.003 ± 0.3847.143 ± 0.6657.122≤0.0011.171NS
Tsg637.265.073 ± 0.81433.680 ± 0.5036.639≤0.0011.919NS
Timp1141.330.962 ± 0.3205.793 ± 0.0496.022≤0.0012.009≤0.005
Ccl724.395.814 ± 1.18029.503 ± 3.6905.074≤0.001ND
Wisp2128.841.648 ± 0.3117.510 ± 1.3934.557≤0.0011.244NS
Tnfrsf12a (Tweakr)50.931.440 ± 0.3836.447 ± 0.0764.477≤0.0011.299NS
Mmp321.051.574 ± 0.2266.655 ± 0.6594.228≤0.0016.528≤0.005
Ptgs2 (Cox-2)21.653.899 ± 1.59216.051 ± 1.4374.117≤0.0011.708NS
Podoplanin75.112.252 ± 0.2668.726 ± 2.1873.874≤0.001ND
Adamts163.781.605 ± 0.5125.802 ± 1.6983.615≤0.0011.454NS
Adamts425.271.392 ± 0.1955.001 ± 0.6323.592≤0.011.132NS
Cd68389.790.899 ± 0.2902.404 ± 0.1612.674≤0.010.978NS
Cd1422.82.548 ± 0.1566.756 ± 0.5042.651≤0.001ND
Mmp1943.132.306 ± 0.3565.563 ± 0.2322.412≤0.050.998NS
Inhibin A66.941.093 ± 0.0962.251 ± 0.6042.053≤0.010.892NS
Adamts5171.531.144 ± 0.0181.782 ± 0.1951.560≤0.011.024NS
Acan257.851.136 ± 0.0470.927 ± 0.1790.816NS0.763≤0.005
Mmp131,900.740.629 ± 0.2710.445 ± 0.1730.707NS1.190NS
Col2a11,219.920.919 ± 0.1520.505 ± 0.1420.549≤0.050.899NS

The reproducibility between biologic replicates (3 separate mouse joints per experimental group) was excellent, as shown for Arg1, Ccl2, Il6, Tsg6, Timp1, and Ptgs2 (Figure 1A). The strongest gene response was seen for most genes 6 hours after DMM. At this time point, there was also a modest increase in some highly regulated genes in the sham-operated mouse joints (e.g., Ccl2 and Tsg6). Strong induction of genes at later time points was rarely associated with an increase in gene expression in sham-operated mouse joints. When we analyzed these genes over longer periods, we found that the majority had returned to basal levels within 2–4 weeks (Table 1). The regulation of a small number of genes, including Timp1 and Mmp3, persisted at low levels.

Figure 1.

Reverse transcriptase–polymerase chain reaction (RT-PCR) of selected genes in whole joints and microdissected joint tissue of mice subjected to surgical destabilization of the medial meniscus (DMM) or sham surgery. A, Male C57BL/6 mice (10 weeks of age) underwent either DMM (triangles) or sham surgery (squares). RNA was extracted from whole joints (on the animal's right side) 6 hours, 3 days, and 7 days after surgery. RT-PCR was performed using TaqMan microfluidic cards. Fold changes were normalized to ribosomal protein S18 and expressed relative to levels in naive, unoperated mice (circles). Symbols represent individual mice; horizontal lines indicate the mean (n = 3 mice per group). B, Immunohistochemical analysis for arginase 1 and for the aggrecanase-generated neoepitope (FREEE) was performed on joints from naive mice, sham-operated mice, and mice subjected to DMM, 3 days after surgery. A preimmune FREEE control is also shown. M = meniscus. Bars = 200 μm. C, Articular cartilage (cart), menisci (men), and epiphyses (bone) were microdissected from the joints of sham-operated mice (Sh) and mice subjected to DMM, 6 hours after surgery. RT-PCR was performed using TaqMan microfluidic cards. Fold changes were normalized to ribosomal protein S18 and expressed relative to levels in the sham tissue. Symbols represent individual mice; horizontal lines indicate the mean (n = 5–10 mice per group). Whiskers indicate the SEM. ∗ = P < 0.05; ∗∗∗ = P < 0.001, by one-way analysis of variance followed by Bonferroni post hoc multiple comparison test. NS = not significant.

To confirm that some of the genes that were regulated at early time points after surgery were being translated, we performed immunohistochemical analysis for arginase 1, and for the aggrecanase-generated neoepitope indicative of ADAMTS activity. Figure 1B shows staining for arginase on day 3 after sham surgery or DMM in the mouse (compared to naive, unoperated mouse joints). These results demonstrate arginase 1 staining in several tissue types in the mouse joint, including the inflamed synovium, chondrocytes, and bone marrow. Aggrecanase neoepitope (FREEE) was apparent in the articular cartilage in tissue specimens from mice subjected to DMM compared to those from naive mouse joints (Figure 1B). Some staining was also observed in the sham-operated mouse joints, although this appeared to be largely in the meniscus. A preimmune control for the FREEE antibody was used to confirm specificity.

Gene expression in microdissected tissue from the mouse joint.

The whole-joint analysis was highly reproducible and yielded good quantities of RNA. It is also advantageous because it allows analysis of all of the tissue types of the joint, including nonresident cells, e.g., inflammatory cells that infiltrate the joint in response to injury or disease. Nonetheless, we believed that it was important to validate the expression of some of the genes in separate tissue types in the mouse joint. We therefore performed RT-PCR on RNA extracted from microdissected mouse articular cartilage, menisci, and epiphyses (after articular cartilage had been removed). An illustration of the validation of this technique, showing macroscopic pieces of mouse articular cartilage following dissection, and the histologic appearance of the mouse joint after the articular cartilage has been removed from the medial tibial plateau, is available online at http://www.kennedy.ox.ac.uk/pubs/vincent1. We confirmed that there was no significant contamination of the tissues by activated synovium by showing that although Cd68, a marker of activated macrophages, was regulated in whole joints following DMM in the mouse, there was no Cd68 regulation in the cartilage, meniscus, or epiphysis when they were analyzed separately (data are available online at http://www.kennedy.ox.ac.uk/pubs/vincent1).

The articular cartilage from the tibial plateaus and femoral condyles of 3 mouse joints were pooled in order to produce sufficient RNA (100 ng) to run the microfluidic cards. Sufficient RNA was extracted from 2 menisci (1 joint), and 1 tibial epiphysis. Samples were processed as described in Materials and Methods, and microfluidic cards were analyzed for each of the separate samples. Of the 20 up-regulated genes shown in Table 1, all except Cd68 were regulated in the articular cartilage (data not shown). Figure 1C shows the levels of 3 genes in each of the tissue types following DMM in the mouse. Results were normalized to ribosomal protein S18 and expressed relative to levels in the sham-operated mouse joints. Adamts5 was regulated in the articular cartilage, at just above 2 fold compared to the sham-operated control mice. Ccl2 was regulated in all compartments. Mmp3 was strongly regulated in the articular cartilage.

Gene induction after DMM in the mouse is dependent on joint mobilization.

Most genes were regulated soon after mice were active following DMM, and because many of the genes were expressed in mechanosensitive joint tissue, we examined whether weight bearing was essential for gene induction. Mice were separated into 2 groups. In the first group, surgical DMM was performed on mice under anesthesia for a short time (∼30 minutes). Since mice quickly recover after the anesthetic has worn off, this group was active for ∼3.5 hours before the joints were harvested. In the second group, surgical DMM was performed on mice under prolonged anesthesia, and the mice were paralyzed for the duration of the experiment (4 hours). The short duration of the experiment (4 hours rather than 6 hours) was necessary to comply with the anesthetic limits stipulated in our animal license. RNA was extracted from animals subjected to DMM and from naive, unoperated animals either with or without anesthetic to ensure that neither anesthetic directly affected gene regulation.

Gene expression was analyzed by RT-PCR using microfluidic cards as described above. In the active mice, the profile of gene expression was similar to that seen 6 hours after DMM. However, in mice that were completely paralyzed after surgery, gene regulation in response to joint destabilization was strongly blunted, reaching statistical significance in 15 of 20 genes (75%) (Table 2). For many of these genes, suppression was incomplete, implying that some induction following DMM was not dependent on joint movement. Some of this may be due to the damage induced by the surgical procedure per se. Nonetheless, these data indicate that the vast majority of the genes we examined in the mouse joint following DMM are mechanosensitive. Figure 2 shows the effect of paralysis on the gene regulation of Arg1, Ccl2, Il6, Tsg6, Timp1, and Ptgs2.

Table 2. Gene regulation 4 hours after DMM in mice under short-term anesthesia or prolonged anesthesia (joint immobilization)*
GeneMice under short-term anesthesiaMice under prolonged anesthesia (paralyzed joint)P
  • *

    Values are the mean ± SD fold change in induction 4 hours after surgical destabilization of the medial meniscus (DMM) compared to naive, unoperated mice (n = 3 mice per group). Reverse transcriptase–polymerase chain reaction was performed on joints 4 hours after DMM with a short-term anesthetic (30 minutes) or with a prolonged anesthetic that immobilized mice for the duration of the experiment (4 hours). All genes were down-regulated in mice that were paralyzed after surgery, with the majority reaching statistical significance. NS = not significant.

Arg153.200 ± 13.8793.517 ± 4.707≤0.01
Ccl2258.197 ± 16.09147.757 ± 2.804≤0.0001
Saa326.058 ± 1.1997.380 ± 3.512≤0.001
Il622.980 ± 2.5123.407 ± 3.011≤0.001
Il1b7.403 ± 1.9391.487 ± 0.967≤0.01
Tsg621.670 ± 1.4832.710 ± 0.587≤0.0001
Timp16.167 ± 0.5621.187 ± 0.405≤0.0001
Ccl724.220 ± 3.5621.380 ± 0.121≤0.0001
Wisp23.660 ± 0.8491.417 ± 0.317≤0.01
Tnfrsf12a (Tweakr)5.230 ± 1.5502.517 ± 0.928NS
Mmp36.843 ± 1.2392.693 ± 0.602≤0.01
Ptgs2 (Cox-2)12.640 ± 1.7841.707 ± 1.462≤0.001
Podoplanin3.487 ± 1.2291.503 ± 1.196NS
Adamts15.321 ± 0.6281.608 ± 0.414≤0.001
Adamts46.450 ± 1.9094.780 ± 0.844NS
Cd682.033 ± 0.2801.323 ± 0.390NS
Cd1410.087 ± 0.3394.897 ± 1.262≤0.001
Mmp193.060 ± 0.6592.327 ± 0.612NS
Inhibin A3.450 ± 0.7990.480 ± 0.296≤0.01
Adamts51.671 ± 0.0331.339 ± 0.093≤0.01
Figure 2.

Response of selected genes in mice subjected to surgical destabilization of the medial meniscus (DMM) and joint immobilization by prolonged anesthesia. Male mice (10 weeks of age) underwent DMM either under short-acting inhaled anesthesia (isoflurane) or under prolonged anesthesia (fentanyl/fluanisone and midazolam). RNA was extracted from the mouse joints after 4 hours, and reverse transcriptase–polymerase chain reaction was performed. Symbols represent individual mice; horizontal lines indicate the mean (n = 3 mice per group). All genes were significantly up-regulated upon surgical DMM (P < 0.05).

Joint immobilization by sciatic neurectomy abrogates the responses of selected genes, including Adamts5, and protects against OA in mice.

In order to exclude changes in gene expression that could be due to differences in anesthesia and to relate changes in mechanosensitive gene expression with subsequent disease outcome, we performed sciatic and/or femoral neurectomies in mice at the same time as surgical DMM. The sciatic nerve, which takes its roots from L4, L5, S1, S2, and S3, supplies the posterior muscle groups of the hind limb and foot. Following sciatic neurectomy, mice bore weight on the operated limb, but there were significant alterations in gait as assessed by the catwalk test (results are available online at http://www.kennedy.ox.ac.uk/pubs/vincent1). Specifically, the knee was immobilized (held in full extension), and walking was enabled by flexion at the hip. The femoral nerve takes its roots from L3–L4 and innervates the quadriceps muscles. Femoral neurectomy resulted in a subtle alteration in gait, leading to an exaggerated, slightly “slapping” step. Combining sciatic and femoral neurectomies resulted in dragging of the limb with weight being born periodically on the operated side. Gait analysis by catwalk testing was not possible in these latter 2 sets of mice, but dynamic gait can be visualized in the supplementary video, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

The levels and type of activity of the mice were assessed 12 weeks after surgery using the Laboratory Animal Behavior Observation Registration and Analysis System. This analysis showed that mice that had undergone DMM alone had reduced activity 12 weeks after surgery compared to sham-operated control mice. Reduction in activity is a sensitive measure of painful behavior and correlates with other pain indices in this model (15). Mice that had undergone sciatic neurectomy at the same time as DMM had activity levels similar to those of sham-operated mice. Mice that underwent femoral neurectomy (with or without sciatic neurectomy) had reduced activity levels, similar to those in mice that had undergone DMM alone (results are available online at http://www.kennedy.ox.ac.uk/pubs/vincent1).

To determine how joint immobilization by sciatic neurectomy affected the expression of genes after DMM, we extracted RNA from the joints of mice 6 hours after DMM, which was performed either alone or in combination with sciatic neurectomy. Of the 20 up-regulated genes shown in Table 1, 9 (45%) were significantly suppressed following sciatic neurectomy, and 11 (55%) were unchanged (Table 3). These changes were not identical to the global suppression of genes seen when the joint was immobilized by prolonged anesthesia, presumably reflecting the difference between weight-bearing and non–weight-bearing joint immobilization. No genes were regulated when sciatic and/or femoral neurectomies were performed in the absence of joint surgery (data not shown). Those genes that were suppressed when sciatic neurectomy was performed at the same time as DMM included Adamts5, Ccl2, Arg1, and Il6. Unaffected genes included Timp1, Tsg6, and Ptgs2 (Table 3 and Figure 3A).

Table 3. Gene regulation 6 hours after surgery in mice with joints immobilized by sciatic neurectomy after DMM*
GeneDMM-operated mice versus sham-operated miceDMM-operated mice with sciatic neurectomy versus sham-operated miceP
  • *

    Values are the mean ± SD fold change 6 hours after surgery. Reverse transcriptase–polymerase chain reaction was performed on tissue specimens extracted from the mouse joints 6 hours after sham surgery, destabilization of the medial meniscus (DMM) alone, or DMM and sciatic neurectomy (performed at the same time). NS = not significant.

Arg1164.329 ± 13.19462.649 ± 3.959≤0.001
Ccl220.169 ± 0.2761.093 ± 0.110≤0.001
Saa313.780 ± 1.7843.565 ± 0.630≤0.001
Il68.605 ± 2.0770.729 ± 0.159≤0.001
Mmp34.440 ± 0.5981.264 ± 0.257≤0.001
Il1b6.514 ± 1.4673.946 ± 0.906≤0.01
Adamts13.7159 ± 0.2522.077 ± 0.249≤0.01
Adamts43.593 ± 0.4541.049 ± 0.869≤0.01
Adamts51.580 ± 0.1290.776 ± 0.031≤0.05
Tsg66.639 ± 0.0986.758 ± 0.693NS
Ccl75.075 ± 0.6355.784 ± 0.359NS
Wisp24.572 ± 0.8464.319 ± 0.973NS
Tnfrsf12a (Tweakr)4.477 ± 0.0534.839 ± 0.418NS
Ptgs2 (Cox-2)4.096 ± 0.3953.788 ± 0.164NS
Podoplanin3.876 ± 0.9735.340 ± 0.330NS
Timp13.008 ± 0.0293.139 ± 0.551NS
Cd142.590 ± 0.2072.124 ± 0.006NS
Mmp192.363 ± 0.1693.097 ± 0.074NS
Inhibin A2.037 ± 0.5532.453 ± 0.409NS
Cd682.689 ± 0.3912.398 ± 0.119NS
Figure 3.

Joint immobilization by sciatic neurectomy modulates gene expression and protects against the development of osteoarthritis in mice. A, Male C57BL/6 mice (10 weeks of age) underwent sham surgery or surgical destabilization of the medial meniscus (DMM) with or without sciatic neurectomy (ScN). Results are shown for 3 genes that were significantly suppressed when sciatic neurectomy was performed (Arg1, Il6, and Ccl2) and for 3 genes that were unaffected by sciatic neurectomy (Tsg6, Timp1, and Ptgs2). Symbols represent individual mice; horizontal lines indicate the mean (n = 3 mice per group). B, DMM or sham surgery was performed with or without femoral neurectomy (FN) and/or sciatic neurectomy, and histologic analysis was performed 12 weeks after surgery. Representative results are shown. LFC = lateral femoral condyle; LTP = lateral tibial plateau; M = meniscus; MFC = medial femoral condyle; MTP = medial tibial plateau. Bars = 500 μm in whole joint sections; bars = 200 μm in lateral and medial sections. C, Mice were subjected to femoral and/or sciatic neurectomy as indicated. Summed cartilage scores were calculated for the joints of mice subjected to DMM, joints of sham-operated mice, and contralateral control joints. Bars show the mean ± SEM (n = 10 mice per group). ∗∗∗ = P < 0.001. NS = not significant.

We next examined the course of OA in mice in which DMM had been performed in combination with neurectomy. Figure 3B shows representative sections used for histologic assessment 12 weeks after surgery, and Figure 3C shows the summed scores. There was significant cartilage degradation in the joints of mice subjected to DMM compared to the joints of sham-operated mice or the contralateral, unoperated limbs, as expected. No change in disease was seen when DMM was combined with femoral neurectomy, but strong suppression of disease (>80%) was seen when sciatic neurectomy was performed, either with DMM alone or in combination with femoral neurectomy and DMM.

DISCUSSION

There are a number of important conclusions that arise from these data. The first is that full movement of the mouse joint is essential for the development of OA following surgical destabilization. The findings of the sciatic neurectomy experiments point toward a neutral or protective effect of weight bearing through the fully extended leg (much akin to splinting the limb with a rigid cast), and suggest that dynamic flexion of the mouse joint is critical for disease development. This is consistent with the findings of other studies showing accelerated disease in destabilized rat joints following forced flexion/extension exercise (27).

The second conclusion is that gene expression in the mouse joint, which includes known pathogenic genes such as Adamts5, is highly mechanosensitive. Again, the expression of a proportion of these genes, including Adamts5, is abrogated in the protected sciatic neurectomized joints, indicating that dynamic flexion of the mouse joint is essential for their induction. These experiments also confirm the mechanosensitive nature of the tissues of the mouse joint and identify the articular cartilage as being the principal source of Adamts5.

Finally, and perhaps most surprisingly, the induction of genes following surgical destabilization in mice occurred within hours of joint use following surgery. It is most unlikely that this response is principally due to the reaction to surgery. Three pieces of evidence support this statement. First, there was a very significant difference between the gene expression profiles of the joints of mice subjected to DMM and the joints of sham-operated mice even though the difference in surgical damage between these 2 procedures is minimal. Second, gene expression was largely suppressed by complete immobilization of the mouse joint, which would not have been expected if the surgical procedure per se was driving gene expression. Third, gene regulation was detected within 6 hours in the articular cartilage, meniscus, and bone in the mice, indicating that these are most likely primary tissue responses.

Taken together, these findings suggest that expression of Adamts5 in the articular cartilage following DMM in the mouse occurs via a direct mechanosensing mechanism. In other words, the sudden loss of joint protection afforded by the meniscus is sufficient to trigger mechanosensitive pathways in cartilage that lead to pathogenic gene expression and subsequent disease in the mouse. The fact that gene regulation was strongest within hours and was significantly reduced within the ensuing couple of weeks suggests that mechanoadaptation leading to an increase in the mechanical threshold for gene induction occurs in the destabilized mouse joint over this time.

While OA is regarded as being an insidious disease, it is important to remember that aggrecan loss has been demonstrated early in this model (neoepitope cleavage is clearly apparent by 2 weeks [18], and inhibition of cartilage loss in ADAMTS-5–null mice is apparent by 4 weeks [20]). We suggest that the expression of proteases within the first few hours of surgery is disease initiating, a proposition that is supported by the presence of aggrecanase-generated neoepitope in the joint tissue within the first few days of surgery. This proposal is also supported by the early detection of neoepitope in the synovial fluid of rats following meniscal tear, which is maximal 4 days after surgery and decreases rapidly thereafter (28). Curiously, Mmp13, which also plays a role in murine OA (14), was not regulated in the first 7 days after surgery in the present study. Subsequent analyses have shown induction of Mmp13 starting 2 weeks following DMM and have detected regulation in the articular cartilage beyond this time point (Gardiner MD, et al: unpublished observations). This is consistent with the later protection seen in MMP-13–null mice, which show a reduction of disease at 8 weeks, but not 4 weeks, after surgery (14).

The role of the synovium in this model of OA is unclear. We see significant synovitis both in the joints of mice subjected to DMM and those of sham-operated mice early after surgery and lasting for about 10 days. Joint immobilization does not appear to change the level of synovitis (by semiquantitative assessment) even though it substantially reduces inflammatory gene regulation in the joint. The strong chemokine response that is seen may contribute to the inflammatory cell infiltration. Indeed, there is visibly less inflammation in mice deficient in CCL2 and those deficient in its receptor, CCR2, after DMM, although, interestingly, these mice are not protected against OA (Vincent TL, et al: unpublished observations). From these observations and others, we think it is unlikely that the synovium is a major source of pathologic proteases, although inflammatory cells could modify disease in other ways (such as by recruiting repair cells into the joint [29]). A more in-depth analysis of the resident and infiltrating inflammatory cells, which is currently very challenging, would be required to explore this further.

A number of other strongly regulated genes were identified, but their roles in the mouse joint remain unclear. Some of these, such as Tsg6, Saa3, Timp1, and Ptgs2, were identified by Appleton et al following a microarray of rat articular cartilage 4 weeks after surgical induction of OA (25). Some have also been identified in a microarray study of isolated meniscal cells obtained from human OA samples (30). Arginase 1 is an enzyme whose primary role is in the liver, where it hydrolyzes arginine into ornithine and urea (31). It has been described as a marker for the M2 subclass of macrophages and may have other roles, such as the ability to compete with the production of nitric oxide, by removing arginine on which the synthesis of nitric oxide is dependent, or promotion of collagen synthesis by generation of polyamines (32). Its role in chondrocytes and other tissues of the joint has yet to be explored.

Of the inflammatory cytokines identified in this study, interleukin-1 (IL-1) is not considered to be a major driver of OA (33), and there are conflicting reports on the role of IL-6 (34, 35). Tsg6 is known to be chondroprotective in mice following the induction of inflammatory arthritis (36, 37), and may exert an antiinflammatory effect by inhibiting leukocyte extravasation (38). Tsg6 may have additional unknown effects in cartilage by nature of the fact that it binds strongly to hyaluronan, an abundant component of the extracellular matrix. Activin A is a homodimer of inhibin βA and is a transforming growth factor β family member. Activin A–null mice develop a severe skeletal phenotype which results in perinatal death, indicating its importance in bone and cartilage development (39). We have previously demonstrated that activin is highly expressed by OA tissue, and that it is also made by injured articular cartilage, a process which is largely fibroblast growth factor (FGF) dependent (40). Its role in mature cartilage is unknown.

Little is known about how mechanical thresholds are set or activated in vivo. We have previously shown that FGF-2 is a mechanotransducer in cartilage in vitro (41, 42), and this pathway is chondroprotective in vivo; FGF-2–null mice develop accelerated disease and have higher joint levels of Adamts5 (18). In vitro FGF-2 is a strong inducer of activin (40), Timp1 (43), and cyclooxygenase 2 (Chong K, et al: unpublished observations). Taken together, our data suggest that FGF-2 may be involved in driving gene induction upon uniaxial joint loading but is unlikely to be the mechanism by which pathogenic proteases are induced upon dynamic flexion of the joint. The role of FGF-2 in mechanotransduction in vivo is being explored and will be addressed in a future publication.

Finally, what are the clinical implications of these findings? Human joint immobilization (while weight bearing) can be achieved by casting or by external fixation. Indeed, the latter, in combination with joint distraction (off-loading the joint mechanically), has been used with success to treat established OA (44, 45). Our findings suggest that immobilization immediately following acute destabilizing injuries could delay the initiation of disease and provide a critical window for introducing disease-modifying therapies prior to remobilization.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Vincent had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Burleigh, Boruc, Saklatvala, Vincent.

Acquisition of data. Burleigh, Chanalaris, Gardiner, Driscoll, Boruc.

Analysis and interpretation of data. Burleigh, Chanalaris, Gardiner, Saklatvala, Vincent.

Acknowledgements

Dr. Gardiner developed the technique of microdissecting murine joints when he was a PhD student in the laboratory of Professor Hideaki Nagase, Kennedy Institute of Rheumatology, Imperial College London.

Ancillary