Experimental Models of Osteoarthritis

Margaret M. Smith

Christopher B. Little

Although the etiopathogenesis of osteoarthritis (OA) is still the subject of intense debate and research, its pathology is well established.¹ The pathologic process of the OA joint is characterized by extensive fibrillation of articular cartilage in those regions subjected to high contact stress accompanied by sclerosis of subchondral bone. At the joint margins, osteophytosis and bone remodeling generally occur, the synovial capsule becomes fibrotic, and the lining is usually inflamed.²^,³ As a consequence of the synovitis, the metabolism of type B lining cells, the synoviocytes, is disturbed, resulting in the biosynthesis of hyaluronan with a reduced molecular weight.⁴ Because the rheologic properties of hyaluronan depend on its molecular size and its concentration,⁵ a decrease in either leads to a decline of the viscoelastic and lubricating ability of synovial fluid, imposing additional mechanical stresses on articular cartilage and subchondral bone. There is also evidence that blood flow in the intraosseous vasculature of OA joints may be impaired owing to the presence of lipid emboli and fibrin thrombi.⁶^,⁷ The ischemia so arising may compromise osteocyte viability, leading to necrosis and bone remodeling.⁶

The complex pathobiologic changes of human OA normally take several decades to develop and may be influenced by a multitude of factors, such as genetic predisposition, hormonal status, occupation, and body mass index, among others. The need to clarify the molecular events that occur in the various joint tissues at the onset and during the progression of OA has necessitated the use of models, which, although imperfect, can exhibit many of the pathologic features that characterize the human disease. In vitro studies using cell and tissue culture models have proven invaluable in defining specific molecular and cellular events in degradation of joint tissues such as cartilage. However, to fully understand the complex inter-relationship between the different disease mechanisms, joint tissues, and body systems, studying OA in animal models is necessary. The realization by workers in the field that OA is not a normal physiologic consequence of the aging process and that it might be amenable to therapeutic intervention provided a major stimulus to the development and use of animal models to test hypotheses concerning the pathogenesis of this disorder. This experimental approach represents one of the cornerstones for the discovery of new anti-OA drugs, and agents have emerged from such animal model studies that are now the subject of clinical evaluation. This chapter will concentrate on the in vivo animal models used to study OA.

Animal models of inflammatory arthropathies (rheumatoid arthritis [RA]), such as collagen-induced arthritis (CIA) in mice, have proven predictive of clinical efficacy, as therapies that are beneficial in CIA have moved into clinical use with proven benefit in RA treatment in humans (e.g., anti-TNF and anti-IL-1treatments).⁸^,⁹ To date, however, none of the available animal models of OA can truly be said to be similarly “predictive,” as no anti-OA therapies have yet been proven in human trials. In this chapter, we review the literature on the most frequently used animal models of OA, particularly with regard to their use for determining the pathophysiology of the disease process. Many of these models have been used to evaluate various forms of treatment, but the vast amount of literature relating to this will not be exhaustively reviewed in this chapter. We have included reports on OA models that have appeared since the previous edition was published either if they describe a novel pathophysiological mechanism or method for evaluation of a previously published model, or they include an entirely new model. In the interests of brevity, many of the older citations have been deleted and readers are referred to previous editions for details.¹⁰ We have confined our review to models of OA in weight-bearing appendicular joints and thus have not included the temporomandibular or spinal facet joints although the disease processes are likely to be similar. The models are divided by the method of induction rather than the animal used. It is important to keep in mind inherent
physiological differences that may exist between species that could influence the comparison with human disease. For example, adult rodents (rats, mice) do not express MMP-1, a major collagenase implicated in human disease,¹¹^,¹² and their aggrecan core protein lacks the extended keratan sulfate binding region of other species including humans.¹³

Induction of Osteoarthritis by Injection of Compounds into Joints

A variety of agents, when they are injected intra-articularly into animal joints, elicit pathologic changes that show some resemblance to those seen in OA. Whereas two or three daily injections of physiologic saline into rabbit joints produced a decrease in the aggregation of cartilage proteoglycans,¹⁴ five daily injections of 10% saline provoked synovial hyperplasia, cartilage hypertrophy, and osteophyte formation after 2 months.¹⁵ Corticosteroids are known to have detrimental effects on the metabolism of human cartilage,¹⁶ and when high doses of these agents are administered intra-articularly to mice¹⁷ or rabbits,¹⁸^,¹⁹ experimental arthropathies generally result. In hydrocortisone-injected rabbit joints, cartilage was depleted of hyaluronan, which may lead to a reduction in the ability of proteoglycans to aggregate and subsequent diffusion out of the cartilage matrix.²⁰ Studies in horses have further demonstrated the potential detrimental effect of intra-articular corticosteroids on long-term cartilage metabolism and biomechanical properties.²¹^,²²^,²³ This chondrodestructive effect of some corticosteroids in animal models has raised the question of their indiscriminate use in clinical practice; however, these studies using normal joints do not mimic the disease situation, where the benefit of modulating the inflammatory process may outweight the detrimental effects. Co-administration of hyaluronan with corticosteroids was found to decrease proteoglycan degradation and loss from equine cartilage.²⁴

Postmenopausal women receiving estrogen replacement therapy have a decreased OA incidence,²⁵ radiologic progression of their disease²⁶ particularly in large joints,²⁷ and decreased levels of cartilage and bone collagen breakdown biomarkers.²⁸ This is consistent with the observation that ovariectomy increased cartilage collagen breakdown and erosion in rats,²⁹^,³⁰ inflammatory arthritis severity in DBA/1 mice,³¹ and surgically induced OA in sheep.³² However, intra-articular administration of estrogen induced degenerative changes in cartilage that resembled human knee OA when it was given at 0.3 mg/kg/day in mice³³ and rabbits.¹⁸^,³³ It was noted in subsequent studies³⁴ that estrogen receptors were upregulated in the femoral but not in the tibial cartilages of rabbit knees with estradiol-induced OA.

Models of OA directed toward selective degradation of the cartilage extracellular matrix have been developed by the use of intra-articular injections of proteolytic enzymes such as papain, trypsin, hyaluronidase, and collagenase in mice, rats, and rabbits (see review by Pritzker³⁵). The mechanisms responsible for cartilage degradation in these models, particularly those in which papain or trypsin was injected, may deviate significantly from those that normally occur in the human disease because these proteins elicit an acute inflammatory reaction that may also contribute to cartilage destruction. Intra-articular injection of collagenase has been acknowledged to provoke additional joint instability by degrading the surrounding capsule and ligaments as well as by directly cleaving the cartilage collagen.³⁶ Studies with C57Bl₁₀ mice injected intra-articularly with bacterial collagenase demonstrated correlations between the degree of instability of the joint, the amount of cartilage damage, and the size of the osteophytes formed.³⁷^,³⁸^,³⁹ Activation of the synovial macrophages also plays a pivotal role in osteophyte formation and fibrosis in this model.⁴⁰

Intra-articular monosodium iodoacetate, an inhibitor of cellular glycolysis, has been used to induce OA-like cartilage changes in joints of hens, mice, guinea pigs, rats, rabbits, and horses.¹⁰ Cartilage lesions appeared in all species with marked loss of safranin O staining, indicating depletion of proteoglycans that is associated with an increase in aggrecan proteolysis by aggrecanases (ADAMTS) and matrix metalloproteinases (MMPs).⁴¹^,⁴²^,⁴³ This model was used to show that cartilage lesions in rat joints could be detected by scanning acoustic microscopy⁴⁴ and ultrasonography⁴⁵ and that the lesions could be ameliorated by MMP-inhibitors⁴⁶ or by exercise in guinea pigs.⁴²

Models of cartilage degeneration have been induced by intra-articular injection of specific cytokines, such as interleukin-1 (IL-1)⁴⁷^,⁴⁸ and transforming growth factor-β,³⁹^,⁴⁹^,⁵⁰ in both rabbits⁴⁸^,⁵⁰ and mice.³⁹^,⁴⁷^,⁴⁹^,⁵¹ Intra-articular IL-1 injection in mice following immunization with methylated bovine serum albumin induces a transient inflammatory arthropathy more characteristic of RA rather than OA.⁵²^,⁵³ Many other compounds, such as carrageenan and zymosan, when they are given by intra-articular injection, elicit degenerative changes in joint tissues; however, these substances also elicit an early acute synovitis and hence are relevant to inflammatory arthropathies more than to OA. In both rabbits⁵⁴ and horses,⁵⁵ intra-articular injection of autogenous cartilage particles led to cartilage degeneration over time. The interaction of these immunogenic particles with the synovial macrophages was considered to be responsible for the ensuing synovitis and subsequent joint disease. Oral quinolones induce cartilage lesions in young animals of a number of species, including rats,⁵⁶ guinea pigs,⁵⁷ dogs,⁵⁸ and rabbits.⁵⁹ The pathologic change in cartilage may be associated with chelation of magnesium⁶⁰^,⁶¹ and the changes are reported to be similar to human OA except that osteophytes do not develop in this model.⁵⁹

Immobilization Models of Osteoarthritis

That articular cartilage requires joint motion and loading to maintain its normal composition, structure, and function is now widely accepted.⁶²^,⁶³^,⁶⁴ Immobilization of a limb has been shown to induce atrophic changes within the articular cartilage of the joint, which include thinning,⁶³^,⁶⁴ increased hydration,⁶²^,⁶³^,⁶⁵^,⁶⁶ reduced proteoglycan content,⁶²^,⁶³^,⁶⁴^,⁶⁵^,⁶⁶^,⁶⁷^,⁶⁸^,⁶⁹^,⁷⁰ altered proteoglycan structure,⁶²^,⁶³^,⁶⁵^,⁶⁸ decreased proteoglycan
synthesis,⁶²^,⁶⁴^,⁶⁶^,⁶⁷^,⁷¹ and increased collagen content and synthesis.⁶⁵^,⁶⁸^,⁷²^,⁷³ Increased synthesis of prothrombin by chondrocytes may contribute to the cartilage remodeling and thinning observed in immobilized rat joints.⁷⁴ Because many of these cartilage changes are similar to those described for human OA joints, limb immobilization has been used as a model particularly for studying the process of cartilage degeneration and more recently evaluation of potential biomarkers⁷⁵ and anti-arthritic therapies such as chondroitin sulfate.⁷⁶ However, fundamental morphologic differences in cartilage have been noted that may diminish the usefulness of the immobilization models. In OA cartilage, chondrocytes proliferate into clones or nests, often remaining active into the late stages of the disease. In contrast, chondrocytes within cartilage of immobilized joints do not form clones but undergo necrosis, particularly if there is no residual movement within the splint.⁷⁷ This response of cartilage most likely results from impaired nutrition of the chondrocytes in the immobilized joint because of the absence of movement and loading.⁷⁸ If a limited range of motion is allowed in the immobilized limb, however, the extent of cartilage degeneration is markedly reduced.⁶⁶ Early work with immobilization models of OA in various species including rat, rabbit, and dog has been reviewed by Troyer⁷⁹ and in the previous edition of this chapter.¹⁰

Models of Spontaneous Osteoarthritis

Naturally occurring OA is known to manifest spontaneously in a number of animal species. Selected studies using mice, rats, guinea pigs, and macaques are summarized in Table 5-1. Some breeds of dogs, including Labradors, German shepherds, and beagles, also develop OA with age, but this is generally secondary to hip dysplasia.¹⁰⁷ In the spontaneously developing OA in dogs, a common pathologic feature of the early disease is a greater degree of synovitis and fibrosis of the joint capsule than is reported for human OA.¹⁰⁸^,¹⁰⁹ A spontaneous decrease in afferent joint nerves with aging in Fisher 344 rats, and worsening of age-related OA with joint denervation in this species, suggested that loss of neuromuscular control and subsequent aberrant loading may be associated with spontaneous OA.¹¹⁰

Many inbred strains of mice develop spontaneous OA with age, including STR/ORT, BALB/c, DBA/1, and C57BL/6 (see Table 5-1). The incidence of the disease varies with strain and sex but can be as high as 90% in aged mice. Characteristic features of OA progression in these models include joint space narrowing, osteophyte formation, focal cartilage lesions, and decreased staining for cartilage proteoglycan. Although aging is a high-risk factor for OA in these models, the morphologic changes arising from the aging process alone have not been satisfactorily addressed. As with age-related OA in rats, loss of joint innervation has been implicated in spontaneous OA in C57BL6Nia mice.¹¹¹ In some strains of mice such as DBA/1 and STR/ORT, OA is largely confined to the male gender and has been shown to depend on the presence of testosterone and aggressive behavior.⁸⁰ The development of refined immunohistochemical techniques, computer-assisted digital image analysis, and advances in molecular biology investigation methodologies (in situ hybridization, genome wide microarray, etc.) has allowed the small amount of joint tissue from mice to be thoroughly investigated. Upregulation of chondrocyte expression of different matrix molecules, cytokines and MMPs, and proteolysis of collagen and aggrecan by MMPs and ADAMTS enzymes in a similar manner to human OA has been demonstrated in spontaneous OA in STR/ort mice.⁸²^,⁸³^,⁸⁴^,¹¹²^,¹¹³

Research with spontaneous OA in guinea pigs has mostly been confined to the Hartley or Dunkin-Hartley strain (see Table 5-1), which has been widely used as an OA model in recent years. These animals have visible cartilage lesions by 3 months of age and a high incidence of bilateral knee OA by 12 months with subchondral bone sclerosis.⁹⁰ Magnetic resonance imaging (MRI) has been successfully used to study this species,⁹⁴ and there is enough joint tissue available for mRNA expression studies.¹¹ Disease progression has been slowed by diet restriction⁹¹ and exercise modification¹¹⁴ demonstrating the role of mechanical factors in the disease process. It has been suggested that altered subchondral bone remodeling⁹⁷ and abnormalities in the cruciate ligament¹¹⁵ may precede and lead to cartilage erosion in this animal model. However, changes in chondrocyte metabolism with ATP depletion and increased NO production have also been found preceding and in association with OA onset.¹¹⁶ This naturally occurring model has proven useful for evaluating potential therapies.¹¹⁷^,¹¹⁸

Both rhesus (Macaca mulatta) and cynomolgus (Macaca fascicularis) macaques show a high incidence of spontaneous OA (see Table 5-1); its epidemiology and joint pathology resemble those of OA in humans.¹⁰⁰ OA changes have been noted separately from those due to aging⁹⁸ and, like the guinea pig (see above), both cartilage metabolic changes¹⁰⁵^,¹¹⁹ and subchondral bone mineralization abnormalities¹⁰⁶ are implicated in the disease. In female macaques ovariectomy worsens the cartilage lesions and this can be inhibited by estrogen replacement therapy.¹²⁰ The joints are also of a sufficient size for radiologic, histologic, and biochemical studies, and the prevalence allows the use of age-matched non-OA controls.⁹⁹ However, because these animals are free ranging and have an extended life span (18 to 30 years), studies may require years for completion and may be influenced by uncontrollable environmental factors. Ethical and financial considerations make widespread use of this model unlikely.

Osteoarthritis in Genetically Modified Mice

The mouse is a convenient species for genetic modification (GM), and a number of spontaneous and engineered mutants have been studied that have, among their phenotypic abnormalities, an increased incidence of spontaneous OA (recently reviewed by Helminen et al.¹²¹). A selection of GM mice that have an increased incidence of OA or OA-like joint disease is given in Table 5-2. Homozygous deficiency in the major structural components of cartilage such as collagen II and its associated
“minor” collagens IX and XI, or components of the proteoglycan aggregates (aggrecan and link protein), are usually lethal during embryogenesis or in the early postnatal period. However, it is noteworthy that while heterozygous deficiencies in the collagen network result in spontaneous OA-like changes in cartilage with aging, haplo-insufficiency of aggrecan or link protein does not cause joint cartilage abnormalities although intervertebral disc degeneration may be seen.¹⁵⁷^,¹⁵⁸^,¹⁵⁹^,¹⁶⁰^,¹⁶¹ Conversely, GM mice may also have a reduced incidence of spontaneous OA, or show protection in models of induced arthritis, and these animals are invaluable for advancing our

understanding of the pathways involved in arthritis development and defining novel targets for disease therapy.¹⁶²^,¹⁶³^,¹⁶⁴ While GM mice with increased spontaneous OA are extremely useful for investigating the role of specific proteins or mutations in the pathophysiology of OA, their utility as more universal models for evaluating therapy of OA in general, as opposed to the disease induced by the specific genetic abnormality they carry, must be questioned.

TABLE 5-1 SELECTED STUDIES OF SPONTANEOUS OA IN ANIMALS

Species	Age	Findings	Reference
DBA/1 mice	0-6 months old	OA development at 4 months old only in males	80
STR/ORT mice	5-50 weeks	OA in 85% of all male mice; MMP and aggrecanase activity increase and colocalize with advancing OA; MMP-2, -3, -7, -9, -13, and -14 gene expression upregulated in AC at all ages but only MMP-3 and -14 protein detected by immunolocalization; collagen cleavage evident only where AC surface fibrillation occurred; chondrocyte apoptosis by TUNEL correlated with severity of OA lesions	81,82,83,84,85,86
C57BL/6 mice	18 months old ± running	Increased incidence and severity of OA changes in mice run 1 km/day, therefore running accelerated OA development; collagen degradation absent in areas of chondrocyte death	87, 88
C57 mice	6 and 8-12 months old	Some heat shock proteins, interleukin-6, and interferon expression were upregulated; expression of other heat shock proteins and interleukin-1 was unchanged in AC	89
Hartley guinea pigs	2-30 months old	OA AC lesions visible by 3 months increasing to >50% of medial tibial AC bilaterally by 1 year old, with subchondral sclerosis; severity of OA lesions was reduced by 40% at 9 months, 56% at 18 months on restricted diet. At 12 months OA on unprotected medial tibial plateau, lateral not affected; increased volume of AC and subchondral bone. Despite gross OA changes only moderate and focal collagen ultrastructure network disruption; no fibre thickening. MRI showed AC thickness increased over first 6 months then decreased, T2 relaxation times increased with time so more predictable AC OA indicator. At onset of OA, AC PG and collagen decreased, AC small and large PG synthesis decreased, water increased and AC PG degradation was unchanged. Collagenase 1 and 3 mRNA expression varied with age and compartment; focal areas of collagenase 1 and 3 proteins in matrix in AC at OA lesions; initial bone density higher	11, 90–97
Rhesus macaque	5-25 years old	Young animals with OA had increased PG levels whereas old had decreased; collagen correlated with age in both normal and OA but lower in OA AC. OA changes progressive through life with high prevalence. OA frequency increased with age and parity (in females); epidemiology and histology resembled OA in man. Increased OA by histology correlated with increased collagen extractability	98,99,100,101
Cynomolgus macaque	5-30 years old	High prevalence of OA lesions, subchondral bone changes common and severe, showing before AC changes; subchondral bone thickness of medial tibia correlated with severity of OA lesions and increasing weight; prevalence and severity of OA lesions increased with age. All chondrocytes in OA lesions stained positive for α1, α3 and α5β1 integrins; in normal cells only α5β1. Increasing OA associated with reduced response to IGF-1 by chondrocytes. The mineralized plate beneath the AC thickened, overmineralized calcified AC and subchondral bone worse with age and OA	102,103,104,105,106
AC = articular cartilage; IGF = insulin growth factor; MMP = matrix metalloproteinase; PG = proteoglycan.

TABLE 5-2 TRANSGENIC MODELS OF OA IN MICE

Genetic Modification	Resulting Phenotype	Reference
Spontaneous deletion in Col2a1 gene giving defect in the C-propeptide (Dmm)	Chondrodysplasia and early onset OA	122, 123
150-bp deletion mutation in the Col2a1 gene (Del1)	Develop OA in knee joints; cathepsin K upregulated at lesion sites; knee OA increased with exercise	124,125,126,127
A large internal deletion in the Col2a1 gene	Chondrodysplasia and increased incidence of OA in 15-month-old animals	128
Heterozygous deletion of Col2a1	Normal cartilage PG content and thickness, increased superficial zone fibrillation (73% v 21%) in 15-month-olds, reduced with running	127, 129
Arg519Cys mutation in Col2a1	Chondrodysplasia and generalized OA by 2 months of age	130, 131
Truncated Col9a1	Mild chondrodysplasia, with increased incidence of OA; loss of PG by 4-6 months and progressive fibrillation and erosion of AC by 12-18 months	132, 133
Deletion of Col9a1	AC fibrillation, chondrocyte proliferation, and osteophytosis by 9 months	134
1-bp deletion in Col11a1 resulting in effective deletion (Cho)	Heterozygous mice have loss of AC PG, increased type II collagen degradation and AC fibrillation with increased MMP-13 in knee by 3-6 months	135,136,137
Single or double knockout of biglycan and fibromodulin (fdn); double knockout of lumican and fdn	Tendon mineralization, gait abnormality, increased joint laxity, and increased OA evident by 6 months of age	138,139,140,141,142
ADAM-15 knockout	Increased loss of proteoglycan and AC erosion with synovial hyperplasia in knee joints at 12-14 months of age in male mice	143
MMP-14 knockout	Bone development and growth abnormalities with early onset of marked synovial hyperplasia and arthritis at 1-2 months of age	144
Interleukin-6 knockout	Increased knee OA in males but not females associated with decreased aggrecan synthesis and bone mineral density	145
Mig-6 knockout	Abnormal gait by 1 month, early onset osteophytes and AC degradation	146
Unknown spontaneous mutation (B6C3F1)	Ankylosing OA—tiptoe walking, swelling of ankle joints, and radiographic and histologic findings of OA AC changes by 9 months old in 80% of males	147
Unknown spontaneous mutation mapping to chromosome 10 (twy)	Autosomal recessive trait, tip-toe walking mouse with multiple osteochondral lesions and longitudinal ligament calcification; decreased AC PG staining and presence of degenerated collagen fibers by 4-8 weeks	148, 149
Defect in a copper transporting gene (Blotchy)	Affects elastin and collagen cross-linking resulting in spontaneous OA	150
Tissue specific BMP receptor type 1a deficiency	Under Gdf5 control no BMP receptor in developing joints—retention of webbing, failure of joint formation, and premature AC proteoglycan loss and erosion	151, 152
alpha-1 integrin knockout	Earlier onset of OA with increased PG loss, AC degradation and synovial hyperplasia from 4-10 months but not different from wildtype at 12-15 months	153
Postnatal overexpression of MMP-13	Aggrecan depletion, collagen proteolysis, fibrillation, and erosion of articular cartilage along with synovial hyperplasia after 5 months	154
Transgenic overexpression of bovine growth hormone	Chondrocyte and synovial hypertrophy with cartilage fibrillation at 6 months	155
Truncated kinase-deficient TGF-β type II receptor	Musculoskeletal developmental abnormalities and progressive chondrocyte hypertrophy with cartilage fibrillation by 6 months of age	156
AC = articular cartilage; BMP = bone morphogenic protein; mig-6 = mitogen inducible gene 6; PG = proteoglycan; TGF-β = transforming growth factor β.