Mechanical loading is known to modify articular cartilage structure through a mechano-adaptive homeostatic response (1, 2) and to contribute to osteoarthritis (OA) (3–5). Despite these fundamental roles, the specific mechanical stimuli that are detrimental or beneficial to joint health remain undefined. This lack of understanding is at least partly due to a paucity of in vivo models in which specific components of mechanical loading can be controlled noninvasively.
Most studies examining the effects of mechanical loading on articular cartilage and chondrocytes have been performed in vitro. Previous investigators have applied specific forces to chondrocyte monolayers or, for example, 3-dimensional scaffolds (6–8). However, as with most in vitro models, these disrupt crucial interactions between cells and their extracellular matrix (ECM) in intact tissues. While the use of articular cartilage explants diminishes these concerns (9–11), explant models fail to retain relationships between distinct joint tissues and are complicated significantly by dissection and culture.
In vivo models preserve these chondrocyte–ECM interactions and those of intact cartilage with other tissues. The most extensively used in vivo models modifying mechanical loads involve surgical joint destabilization. Initial studies used canine anterior cruciate ligament transection in the Pond-Nuki model (12) or rabbit knee meniscectomy (13). These produced major advances and highlighted a need for similar models in smaller, genetically modifiable species, such as mice.
The development of surgical murine models now allows induction of lesions with varying severity (14); several models are considered representative of human OA (15, 16). However, these are severely disadvantaged by the need for surgery, which increases infection risk, causes trauma, and directly affects joint cell metabolism. The intransience of the modified loading that current models utilize also means that the models are not useful for studying responses to specific episodes of loading or discrete components of the mechanical milieu. Availability of a model that, once characterized, allows application of adjustable noninvasive loading is obviously an attractive prospect.
Responses to differing loading regimens were first examined in rabbits by Radin et al (17). In those studies, magnitude and frequency of applied loads were changed, and joints later analyzed (18). Damage was induced in other rabbit models using a single adjustable impact (19, 20). Development and characterization of similar nonsurgical models in mouse knee joints would offer greater advantages, as mice are often the animal of choice in such studies and their use allows examination of responses in mutant and transgenic animals.
The advantage of nonsurgical joint loading is also inherent in exercise and disuse models (21, 22). However, these are hampered by reliance on a relatively uncontrolled mechanical input as the stimulus. There is, therefore, a need for a nonsurgical murine model that allows for adjustment of specific components of the applied load, including its magnitude, frequency, and loading cycle shape.
In this report we describe a newly developed nonsurgical model for loading of mouse knee joints. The loading is applied through natural points of articulation, with the contralateral joint experiencing normal gait between loading episodes, providing an internal control. The model has allowed us to demonstrate induction of focal articular cartilage lesions. Another attractive characteristic is its controllability, which enabled us to distinguish between the short- and long-term effects of single- or multiple-loading episodes on articular cartilage integrity. Additionally, we describe histologic changes in various other joint tissues, induced by adjusting the number of exposures to a specific loading pattern. With its versatility and modifiability, this model may allow discrimination between joint responses to physiologic and pathologic loading, as well as a more thorough investigation of the interaction between genetic and mechanical influences in OA initiation and progression.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
The in vivo model described herein offers opportunities to noninvasively apply specific mechanical loading elements without surgery, and to study their effects in the intact murine knee joint. The advantages of our model are the adjustability and reproducibility of the episodes of applied loading. We have shown that articular cartilage injury can be induced after 1 loading episode, thereby offering a new model of localized articular cartilage damage. In addition, we found that the application of multiple loading episodes induces lesions that resemble those observed in OA; these lesions increase in severity without further specific applied loading. This provides an opportunity to investigate articular cartilage responses to lesion induction and the mechanisms involved in the progression of OA-like lesions.
Using a specific regimen of loading episodes, articular cartilage lesions were reproducibly restricted to the lateral femur. The likelihood that these were a direct product of localized loading is increased when the supporting results of micro-CT imaging performed during loading application are considered. These showed close apposition across the lateral compartment, with increases from 2N (holding load) to 9N (peak loads) in episodes of compressive loading, indicating that load-related “contact” likely coincides with articular cartilage lesions in the lateral femur. This predominant lateral damage is probably due to the design of the apparatus, in which the knee is held in a valgus position by the custom-made cup during loading.
Our studies also displayed a relative protection of the directly opposed tibial articular cartilage, which is likely to share a virtually identical mechanical challenge. There are many possible explanations for this. The tibial articular cartilage might have distinct structural properties that make it inherently less prone to trauma. Additionally, the tibia may be relatively protected because of the applied loading direction; micro-CT imaging results support the idea that our model probably loads the tibia in a direction that is relatively in accordance with normal use. In contrast, the femoral epiphysis likely receives nonaxial, nonphysiologic compressive loads. This restricted localization of femoral lesions is potentially advantageous from a research perspective. This compartment is not commonly affected by either spontaneous or surgically induced OA (14, 33), and our model may provide a new means for exploring relationships between genetic and load-induced susceptibility to cartilage damage. Though protection against overt lesions in other compartments may be considered advantageous, a lack of histologic lesions might not necessarily signify a lack of biochemical changes, and the use of opposing tibiae as internal controls should be undertaken with caution. The femoral compartment in contralateral, nonloaded joints is a suitable control.
Our model also allows the number of loading episodes to be varied. This profoundly influenced the character of articular cartilage lesions; they increased only in mean lesion grade, but not maximum, following repetitive loading with only habitual use. Lesions induced by the application of multiple loading episodes progressed by becoming more extensive (anteroposterially) without an apparent need for further loading, but not more severe. In contrast, lesions induced by a single application of loading were severe, but not susceptible to such progression. This leads us to hypothesize that a threshold or “point of no return” is reached after 2 weeks of repeated loading, instigating lesion progression independently of additional applied loading. This new model may allow the nature of this threshold to be defined and its underlying mechanisms to be identified. It may also allow adjustment of joint loading to further optimize the induction of cartilage lesions.
Our data suggest that PG loss identifies articular cartilage that is vulnerable to progression resembling that seen in OA (27, 34, 35). By using transgenic and mutant mice, our model could enable elucidation of important processes, as well as new targets for limiting articular cartilage lesion induction and, vitally, for slowing OA progression. Other models are limited by permanent alteration of joint mechanics, whereas the controllability, adjustability, and temporary nature of the applied loading make our model very versatile.
OA influences the whole joint; osteophyte formation and synovial inflammation and fibrosis are late OA hallmarks. Our observations revealed localized medial osteophyte formation, near the cruciate ligament insertion into the femoral groove. This resembled changes that follow instability induced by collagenase injection (36). More prolonged loading (5 weeks) promotes additional osteophyte formation on the lateral femur. To our knowledge, few studies have considered the location of murine osteophytes. Blaney-Davidson et al (36) found that the location was highly dependent upon the model. Future studies may help to define the exact mechanical and molecular cues that control osteophyte formation.
OA is also often associated with synovial lining cell hyperplasia and fibrosis (37–39). However, these features were seen only in joints that received continued loading, and since synovium appeared to rapidly return to normal upon loading cessation, this suggests that synovium has acute sensitivity to mechanical stimuli. Mechanical loads control both synovial fluid and lining cell behavior in vivo (40, 41), and this model provides a new tool for understanding synovial pathophysiology.
Damage or transection of cruciate ligaments usually induces OA. In our model, loading caused marked changes in the cruciate ligament within 2 weeks. The changes resembled those found in OA joints (42), where they are speculated to be the product of high mechanical stress. It is possible that these changes influence OA progression, and our model may make it possible to test such hypotheses. The response of murine menisci to excessive loading resembles secondary ossification. In loaded joints, meniscal chondrocytes become hypertrophied and blood vessels invade (after 2 weeks) before a marrow cavity surrounded by ossified matrix is formed (5 weeks) (43–45). Together with osteophyte formation, synovial hyperplasia and fibrosis, and cruciate ligament changes, this load-induced meniscal ossification indicates the diversity of the specific end points (OA hallmarks) that might be selected for measurement in our model. Their relevance to OA, combined with the likelihood that this model can be modified to accentuate specific desired characteristics, strengthen its potential for use in studying the joint as an organ. The rapid appearance of these OA hallmarks strongly implicates mechanical influences in their development. This model may therefore help identify specific features of the joint loading environment that drive the appearance of these late changes in human OA.
Translating these findings to human OA should be undertaken with great caution, as mouse models have many shortcomings. These include those inherent to all animal models and those specific to mice, such as limitations associated with size, weight, and gait. Further limitations specific to mice include distinct epiphyseal and metaphyseal trabecular bone architecture and mass, and lack of intracortical bone remodeling, which will affect the mechanical environment of articular cartilage during episodes of loading. Mouse articular cartilage composition and organization also hinder direct comparison with human OA. These and many other additional limitations are, however, inevitable consequences of the use of animal models.
In conclusion, we have developed a model for mechanical loading in mouse knee joints, which offers opportunities to study the effects of various loading magnitudes and regimens on joint health and disease. Because the mouse has ready possibilities for genetic manipulation, the interaction between loading and genetics can be more fully investigated. Another attractive feature in this model is its controllability. We have found that identification of short- and long-term effects of single or multiple loading episodes on articular cartilage integrity enables discrimination between spontaneously progressing and nonprogressing lesions, and this model can also be used to regulate the effects of loading on various other joint tissues. The versatility of this model should allow discrimination between joint responses to physiologic and pathologic loading and will, over a longer period of time, enable better elucidation of the interactions between age and genetic and mechanical influences in OA initiation and progression.