By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
The mouse is an optimal model organism in which gene–environment interactions can be used to study the pathogenesis of osteoarthritis (OA). The gold standard for arthritis research in mice is based on histopathology and immunohistochemistry, which are labor-intensive, prone to sampling bias and technical variability, and limited in throughput. The aim of this study was to develop a new technique that assesses mouse cartilage by integrating quantitative volumetric imaging techniques.
A novel mouse model of OA was generated by cruciate ligament transection (CLT) and evaluated by histopathology and immunohistochemistry. Knee joint specimens were then imaged using a new technique that combines high-resolution micro–computed tomography (micro-CT) and phase-contrast optics followed by quantitative analyses. A comparative analysis was also performed in a previously established mouse model of OA generated by destabilization of the medial meniscus (DMM).
Phase-contrast micro-CT achieved cellular resolution of chondrocytes and quantitative assessment of parameters such as articular cartilage volume and surface area. In mouse models of OA generated by either CLT or DMM, we showed that phase-contrast micro-CT distinguished control and OA cartilage by providing quantitative measures with high reproducibility and minimal variability. Features of OA at the cellular or tissue level could also be observed in images generated by phase-contrast micro-CT.
We established an imaging technology that comprehensively assessed and quantified the 2-dimensional and 3-dimensional changes of articular cartilage. Application of this technology will facilitate the rapid and high-throughput assessment of genetic and therapeutic models of OA in mice.
Osteoarthritis (OA) is characterized by degeneration of articular cartilage, intraarticular inflammation with synovitis, and remodeling of periarticular and subchondral bone. It is a major cause of disability and is one of the most common musculoskeletal disorders (1). The prevalence of the disease is also increasing as a consequence of demographic shifts and lifestyle (2). New treatment modalities are direly needed, and animal models with reliable and informative end points have been a major focus of research efforts. Mouse models that mimic clinical OA are of importance, because genetic analysis can be easily exploited, leading to the development of new therapeutic modalities and pharmacologic treatments. As is true for most other soft tissue diseases, the primary assessment of all existing mouse models of OA has been limited to standard histologic scoring (3), which is time-consuming and subject to artifacts and interobserver variability. It provides only semiquantitative analysis of two-dimensional (2-D) sections that may generate biased 3-D representation.
High-resolution micro–computed tomography (micro-CT) imaging has been established as the gold standard for assessing the morphology of calcified tissue, such as trabecular and cortical bone in small animals, at high resolution (4–7). However, this approach does not effectively image uncalcified soft tissue such as cartilage. In situ imaging of soft tissue using equilibrium partitioning with an ionic contrast agent via micro-CT was shown to be successful in rabbits and rats using Hexabrix, an iodine-based staining reagent (8, 9). However, this procedure, when used in tissue other than blood vessels, required dissection, with the potential for introducing artifacts in small animals, and resolutions have been inadequate for imaging most soft tissue in mice.
To date, osmium tetroxide has been one of the most successful contrast stains used for micro-CT imaging of soft tissue and embryos (10, 11). In addition, treatment with ruthenium hexamine trichloride (RHT) has yielded homogeneous, reproducible preservation of epiphyseal chondrocytes (12). To further increase the quality of cartilage imaging, we applied phase-contrast optical imaging. Phase-contrast imaging at optical wavelengths has been an important technique for thin biologic samples without staining. Diffraction causes shifts in the relative phase of incident radiation (including x-rays) at the interfaces of tissue. In the context of imaging, x-rays arrive at the detector out of phase with respect to others. Any detector sensitive to the x-ray propagation direction is therefore capable of detecting tissue surface even when the x-ray absorptions of the tissue are similar (13, 14). This provides the possibility of distinguishing tissue of interest from remaining soft tissue with similar absorbance.
Here, we present a new approach that combines contrast staining, phase-contrast optics, high-resolution micro-CT, and 3-D graphic quantification. As an example, we investigated the visualization and quantification of articular cartilage and subchondral bone in a mouse model of OA, at cellular resolution. To generate a robust and reliable mouse model of OA, we performed cruciate ligament transection (CLT) in the mice, with minimal injury to surrounding tissue. OA induced by this procedure mimics posttraumatic OA in humans, because ligament tear is a common cause of OA. We assessed our model using classic histology and immunohistochemistry. After validating our mouse model using established technology, we scanned the knee joints in situ and analyzed the samples in 3 different ways. First, we observed the samples in 2-D views for pathologic changes. Second, we reconstructed the joints in 3-D views and observed OA traits comprehensively; this provided direct visualization of cartilage degeneration and bone remodeling. Last, to further enhance the reliability and objectiveness of the method, we developed algorithms that facilitated the quantification of volume and area measurements based on the 3-D platform. Because destabilization of the medial meniscus (DMM) in a genetically modified mouse is commonly performed to evaluate the impact of therapeutic modifications in OA (15, 16), we compared our analytic approach and our mouse model of OA with the DMM model, using histology scores and quantification of cartilage volume and area. This technology provides a novel and robust end point for assessment of OA and potentially other soft tissue diseases, with high throughput.
MATERIALS AND METHODS
FVB/N and 129SvEv mice were purchased from The Jackson Laboratory. The strains were used because they are common background strains for transgenic and knockout mouse lines. All studies were performed with approval from the Baylor College of Medicine Institutional Animal Care and Use Committee. All mice were housed under pathogen-free conditions in groups of 4–5 per cage. Mice had free access to feed and water. To avoid potential postmenopausal bone loss, all of the mice used in the experiments were male.
Cruciate ligament transection surgery.
Eight-week-old male FVB/N mice were anesthetized by intraperitoneal injection of 0.2 ml/10 gm 1.2% tribromoethanol (Sigma-Aldrich), and the knees were prepared for aseptic surgery. Appropriate skin preparation for a sterile procedure consisted of removing the hair and cleansing with surgical soap, followed by a 70% alcohol wash. The cleaning procedure was repeated 3 times. To provide a better view of the knee joint space, a Feather #11 disposable scalpel (Feather Safety Razor) was used to provide a skin-deep ∼5-mm incision in the medial side of the joint parallel to the patellar tendon. A Beaver EdgeAhead Safety Sideport MVR 0.90-mm 20-gauge microsurgical scalpel (Beaver-Visitec) was inserted into the knee joint space from the side of the patellar tendon. Transection of the anterior and posterior cruciate ligaments was performed in one group of mice (CLT group), and no transection was performed in another group of mice (sham surgery group). The success of surgery was evaluated by determining joint laxity in response to valgus and varus transection of the ligaments. When the surgery was performed properly, it was associated with no blood loss. The skin was closed by application of 5.0 monofilament sutures (Ethicon), and the sutures were removed 2 weeks after surgery.
Destabilization of the medial meniscus.
The DMM model of OA was generated using standard approaches that have been previously described (16).
Histology and immunohistochemistry.
The whole knee joints were fixed with 4% paraformaldehyde (Sigma-Aldrich) overnight at 4°C on a shaker. Then they were decalcified in 14% EDTA for 5 days at 4°C on a shaker. After dehydration using alcohol gradients and infiltration with xylene and paraffin, samples were embedded in paraffin. Paraffin-embedded joints were sectioned at 6 μm on a sagittal plane. Samples were stained with Safranin O–fast green using standard protocols. Immunohistochemical analysis was performed using rabbit anti-mouse type X collagen (a generous gift from Dr. Greg Lunstrum, Shriners Hospital for Children, Portland, OR) and mouse anti-human matrix metalloproteinase 13 (MMP-13) monoclonal antibody (Millipore) as primary antibodies, and one dropper bottle of horseradish peroxidase polymer conjugates (Invitrogen) as secondary antibody. HistoMark TrueBlue (KPL) was used as a developing reagent. Samples without primary antibody treatment served as negative controls.
Phase-contrast micro-CT scanning.
Whole knee joints were dissected and fixed with 0.7% RHT (Sigma-Aldrich), glutaraldehyde (0.2% volume/volume; Polysciences), cacodylic acid (0.05M; Sigma-Aldrich), pH 7.4, for 24 hours to allow complete penetration. Afterward, the samples were washed 3 times for 30 minutes with RHT (0.7%), cacodylic acid (0.05M), pH 7.4. Samples were then fixed for 24 hours with RHT (0.7%), osmium tetroxide (1%; Polysciences), and cacodylic acid (0.1M), pH 7.4. After dehydration using alcohol gradients and infiltration with xylene, samples were embedded in paraffin. Before scanning, excess paraffin was removed, and the samples were mounted and stabilized on an Xradia sample holder. Samples were then scanned using an Xradia μXCT system.
Reconstruction and analysis of micro-CT data.
Data were reconstructed using Xradia software and were transformed into Dicom files. Reconstruction involves correction for beam hardening and for center shift effects caused by differences between the center of sample rotation and the center of the detector. Samples were analyzed using TriBON software (Ratoc).
Parametric data between 2 groups were compared by Student's t-test, and one-way analysis of variance followed by Tukey's post hoc comparison was performed to compare parametric data between multiple groups. Histologic grades were compared by Wilcoxon's rank test. All analyses were performed using SPSS software. P values less than 0.05 were considered significant.
Histologic and functional outcome in mouse model of CLT.
Anterior cruciate ligament and posterior cruciate ligament injury are common causes of OA in humans (17, 18). To generate a mouse model of OA that mimics this clinical condition, we performed anterior and posterior CLT in mice using a microsurgical approach that left the patellar tendon intact. Joint laxity in response to valgus and varus stress confirmed transection of the ligaments. In mice that underwent sham surgery, a similar invasive procedure was conducted but without actual ligament transection (Figure 1A). Examination of continuous histologic sections collected 24 hours after transection confirmed that the patella was intact, and that no visible damage was evident on either the joint surface or the meniscus in both the CLT group and the sham surgery group. To assess the severity of OA generated by our technique microscopically, histologic grading was performed in a blinded manner by 2 independent observers, using standard Osteoarthritis Research Society International (OARSI) scores (3).
The limbs of mice in the sham surgery group demonstrated insignificant levels of OA, whereas histologically significant OA was observed over time in the limbs of mice in the CLT group (Figures 1B and C). In addition, accelerated cartilage hypertrophy was observed in OA joints, as demonstrated by increased immunohistochemical staining using anti–MMP-13 and anti–type X collagen antibodies (Figure 1D). These data showed that this surgical model produced the pathognomonic features of OA on standardized histologic assessment and may serve as a clinically relevant OA model.
Morphologic evidence of OA assessed by phase-contrast micro-CT.
Phase-contrast micro-CT was performed, and a scanning protocol was established to image cartilage at ultra-high resolution. Intact knees were fixed and infiltrated with RHT (a reagent that increases contrast in soft tissue and preserves chondrocyte morphology) and osmium tetroxide to create high contrast in the cell membrane (12, 19). Samples were embedded in paraffin and later scanned in situ to visualize all tissues.
Scanning of the OA joints was performed at different resolutions (Table 1). At 10-μm resolution, an overview of the entire joint showed high-density materials including meniscus and bone. At 4-μm resolution, the cartilaginous area in the joint could be imaged. At 2.03-μm resolution, a single mouse knee joint condyle was shown, providing soft tissue visualization and cellular detail. At 1.54-μm resolution and 0.54-μm resolution, a limited area of joints could be imaged to show changes in cell distribution, as demonstrated by cell lacunae (Figure 2A). At the tissue level, OA joints displayed loss of uncalcified cartilage. At the cellular level, OA joints showed chondrocyte clustering, while sham-transected joints showed evenly distributed chondrocytes (Figure 2A). After comparing different resolutions, we concluded that cartilage could be clearly imaged at 4 μm and 2.03 μm resolution. In the interest of throughput, all joints were scanned at 4 μm resolution in subsequent analyses to assess the morphologic signs and symptoms associated with OA progression.
Table 1. Conditions used for cartilage phase-contrast micro–computed tomography scanning*
Scan time, hours:minutes
Source voltage, kV
Source power, watts
Detector distance from sample, mm
Source distance from sample, mm
No. of images obtained
Exposure time for each image, seconds
Articular cartilage loss due to mechanical stress is the first step in the development of OA. Once articular cartilage completely disappears, bone-on-bone contact in the joint causes pain. Loss of cartilage triggers cellular proliferation that results in hypertrophic bone remodeling. Phase-contrast micro-CT was capable of visualizing cartilage loss (Figure 2B). In some OA knees, we also observed free-floating cartilage fragments (Figure 2C).
Metaplasia transforms meniscus from a cartilaginous tissue to an osseous tissue. This process causes uneven weight distribution in the knee and excessive forces in specific areas, leading to more-localized articular cartilage loss. In almost all knee joints in the CLT group, at least 1 meniscus underwent changes in size, cellular architecture, and density. Histologic observations corroborated the phase-contrast micro-CT findings (Figure 2D). Chondrocyte hypertrophy above the tidemark, which is best shown by increased expression of MMP-13 and type X collagen, is a biomarker of altered cartilage and bone remodeling in OA. It leads to subchondral bone remodeling, which is characterized by erosion, denudation, and in severe cases, microfracture. These signs can occur only regionally within the knee joint and can easily be missed by histologic sampling. For example, erosion and denudation were visualized especially in the edge of cartilage by analysis of the whole joint using phase-contrast micro-CT (Figure 2E). Microfractures, even in cases that were not easily identified by histologic analysis, were detected at multiple sites by scanning through OA knee joints (Figures 2F and G). Osteophyte formation near the articular surface (Figure 2E) and heterotopic ossification in patellar tendons (Figure 2H) were also observed in severe cases of OA.
Three-dimensional reconstruction of the OA knee joint.
To assess cartilage in a volumetric manner, 3-D reconstruction of the CT images was performed. After 3-D reconstruction, data analysis was performed using TriBON software. High-density regions (bone and fixative-saturated lipid-rich tissue) were assigned a mask, and a hole-filling algorithm was used to include marrow space with surrounding cortical and trabecular bone (mask 1). Due to similarities in the density of cartilage and the surrounding soft tissue, the joint surfaces of femoral and tibial cartilage were defined manually, and each was assigned a mask (mask 2 and mask 3, respectively). Cartilage volume was defined by subtracting mask 1 from mask 2/3, resulting in mask 4 (femoral cartilage volume) and mask 5 (tibial cartilage volume), respectively (Figure 3). The 3-D–reconstructed joint revealed cartilage volume loss, meniscus metaplasia, bone remodeling, and osteophyte formation in CLT joints (see Supplementary Video 1, which is available at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). The joint was also imaged in various planes, such as the coronal and sagittal planes, as commonly done for the scoring of OA (see Supplementary Video 2, which is available at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131).
Quantification of OA cartilage volume and surface.
To objectively assess changes in articular cartilage and subchondral bone in a quantitative manner, the volumes of masks 4 and 5 were calculated. Significant decreases in the tibial and total articular cartilage volumes were observed 1 and 2 months after CLT. Significant loss of femoral articular cartilage volume was not detected until 2 months after surgery, suggesting that loss of tibial articular cartilage was a more prominent feature in this OA model. Three months after surgery, all mice in the CLT group showed significant loss of cartilage at all condyles (Figure 4A).
Cartilage surface area changes may also reflect cartilage loss. In order to calculate cartilage surface area, mask 4/5 underwent 2-D dilation and was subtracted from mask 4/5 and mask 1, leaving only the joint cartilage surface. One month after surgery, no significant surface area loss was observed (Figure 4B). This piece of data, together with quantification of cartilage volume, showed that during the first month of OA development, the cartilage became thinner, but subchondral bone coverage was not lost. Two months after surgery, there was significant loss in the femoral, tibial, and total articular cartilage volumes in the CLT group compared with the sham surgery group (Figure 4B). Three months after surgery, all condyles had significant loss of cartilage surface area, showing progression of cartilage surface loss in this mouse model of surgically induced OA (Figure 4B).
To further quantify the progression of OA, the bone surface area that was covered by cartilage in OA joints was calculated. This was calculated by the shared area between mask 1 and the dilated region of mask 4/5. One month and two months after surgery, most bone surface areas covered by cartilage were significantly decreased in the CLT group (Figure 4C). Three months after surgery, all condyles showed a significant increase in the bone-exposed surface (Figure 4C).
Osteophyte formation and heterotopic bone formation are hallmarks of late-stage OA and represent a pathologic sign of bone remodeling. Osteophyte and heterotopic bone numbers and volume were quantified by 3-D labeling of objects larger than 5,000 voxels from mask 1 (bone and high-density objects). Volumes were deselected for the femur, tibia, fibula, and menisci. The remaining volumes were assigned as osteophytes. No osteophytes or heterotopic bone formation was detected in the control knee joints. In the CLT group, both the number and volume of osteophytes increased as the length of time after surgery increased (Figures 4D and E). Most joints in the CLT group developed osteophytes ∼2 months after surgery. Three months after surgery, the mice had a large number of osteophytes, suggesting severe remodeling of bone and soft tissue and late-stage OA.
Comparison between CLT and the previously established DMM model.
The DMM model of OA is considered to represent mild OA. It has been used to investigate genetic contributions and treatment approaches to OA. To compare the different features of OA induction by DMM versus CLT and the effects of different mouse strains on OA development, we compared our CLT model with a DMM model generated in both FVB/N and 129SvEv strains. The 129SvEv strain is commonly used to generate knockout mouse models and was the first strain of mice in which DMM was performed. One month after DMM, the OA that developed in FVB/N mice was slightly less severe than that in 129SvEv mice, as evaluated by OARSI scoring (Figures 5A and E, and data not shown). In addition, when mice that underwent DMM surgery were evaluated by phase-contrast micro-CT quantification, they showed loss of medial cartilage volume and area (Figures 5B–D, F–H). Interestingly, DMM also caused loss of lateral femoral cartilage volume and area, while little loss of lateral tibial cartilage was observed (Figures 5B–D, F–H). This might be caused by the differential thickness and shape between femoral and tibial cartilage. It underscores the contention that OA changes in the medial compartment do affect lateral articular cartilage volume. No osteophyte formation was detected in the DMM model (data not shown).
After DMM was performed, the development of OA appeared to be less severe in FVB/N mice compared with 129SvEv mice (Figures 5B–D, F–H). This is probably due to size and activity differences between the 2 strains. One month after surgery, the maximum histology scores of the medial condyle in joints that underwent DMM were comparable with those in joints that underwent CLT (Figures 1B, 5A, and 5E). Interestingly, cartilage surface area losses in joints subjected to DMM were more severe than those in joints of mice in the CLT group (Figures 4B and 5G). The variability (as assessed by the standard deviation) of both the histology scores and micro-CT quantification of joints that underwent DMM was slightly less than that of joints that underwent CLT, suggesting that DMM might be a more consistent method (Figures 1B, 4A–C, and 5).
The results of this study establish that phase-contrast micro-CT is capable of visualizing signs of OA such as articular cartilage loss, meniscus metaplasia, subchondral bone remodeling, and osteophyte formation. It also represents the first time that changes in mouse OA knee joints have been assessed quantitatively using a 3-D imaging approach. The changes in mouse OA knee joints observed by phase-contrast micro-CT correlate well with those observed using traditional methods such as histologic grading and functional studies. Because phase-contrast micro-CT data are available in all possible sectioning planes and are acquired from intact specimens with the potential for cellular resolution, the approach provides the power of cellular analysis in histology and whole joint analysis in functional studies, but with higher fidelity, higher throughput, and lower variability, in a quantitative manner.
When this technique was used to evaluate the DMM model of OA, histology scoring could not detect statistically significant changes in the lateral condyle of the joint that underwent DMM, while phase-contrast micro-CT detected femoral cartilage loss. This observation suggests that phase-contrast micro-CT is a more sensitive method for detecting disease progression. At the same time, subchondral bone could be easily assessed. On the basis of these features of phase-contrast micro-CT analysis, we believe that it will have a substantial impact on the analysis of OA pathogenesis and development of treatments by bringing to bear the significant advantages associated with mouse models, including lower costs and the capacity for genetic manipulation.
Compared with traditional standard-resolution micro-CT, which is the standard for imaging mineralized tissue from mice (20, 21), phase-contrast micro-CT protocols require optimization of tissue processing and data acquisition for different soft tissues. Different analysis protocols and landmarks will still need to be defined when analyzing new types of tissue. Until now, when applied in low-resolution mode without tissue staining, phase-contrast micro-CT has served much like standard micro-CT for evaluating bone remodeling in joint diseases (7). When combined with ionized contrast staining such as Hexabrix (8), phase-contrast micro-CT may provide sufficient resolution to assess proteoglycan loss in cartilage in mouse OA models, which would be a future application for this approach in the context of OA modeling. With other contrast reagents such as Microfil (22, 23), it may also allow imaging of neovascularization.
Compared with micro–magnetic resonance imaging (micro-MRI), which is widely used for soft tissue imaging, phase-contrast micro-CT still does not show soft tissue boundaries as effectively (24, 25). However, this problem may be solved by investigating new contrast staining reagents that mark the tissue of interest. A good candidate reagent for cartilage matrix is phosphotungstic acid, because it binds lysine and arginine residues (26). Scanning in different media such as air or agarose may also increase soft tissue contrast as well as expedite tissue processing. Importantly, phase-contrast micro-CT provides imaging resolution that is an order of magnitude higher than that provided by micro-MRI. This presents opportunities to quantify soft tissue volume and surface area in small animals, opening avenues for studying developmental phenotypes using mouse models. Moreover, the samples for phase-contrast micro-CT can be preserved long term, allowing reanalysis if necessary. Finally, the shorter scanning time of phase-contrast micro-CT allows higher analysis throughput compared with micro-MRI.
Our understanding of OA is based largely on histologic and immunohistochemical studies. A more comprehensive and quantitative imaging modality such as phase-contrast micro-CT, which allows observation of cellular features of OA in the context of the whole knee joint, may shed new light on the pathogenesis of OA.
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. Lee 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. Ruan, Heggeness, Lee.
Acquisition of data. Ruan, Dawson, Jiang, Gannon.
Analysis and interpretation of data. Ruan, Gannon, Lee.
We thank Terry Bertin and Ayelet Erez for suggestions and comments on the manuscript, Greg Lunstrum (Shriners Hospital for Children) for providing type X collagen, and Stephen Henry (MD Anderson Cancer Center) for suggestions regarding immunohistochemistry.