SEARCH

SEARCH BY CITATION

Introduction

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Rheumatoid arthritis (RA) is a chronic, systemic, immune-mediated inflammatory disease associated with decreased life expectancy and quality of life. RA is characterized by chronic inflammation and synovial hyperplasia leading to destruction of cartilage and bone. A better understanding of the pathophysiology of RA has led to significant improvements in its treatment. However, many questions remain with respect to the pathogenesis of RA, and disease remission is achieved in only a minority of patients. Therefore, there is still a need to develop novel antirheumatic therapies. Animal models of RA complement descriptive studies of well-defined patient samples and in vitro studies and represent important tools for the development and evaluation of new treatment options.

This review article addresses the immunologic characteristics of and similarities and differences between a variety of animal models of human RA. Commonly used animal models such as collagen-induced arthritis (CIA), adjuvant-induced arthritis (AIA), and a model of spontaneous arthritis using tumor necrosis factor (TNF)–transgenic mice are described. We also discuss the less frequently used models of streptococcal cell wall (SCW)–induced arthritis, proteoglycan-induced arthritis (PGIA), and K/BxN-transgenic mice, including the serum transfer–induced (STIA) model, to indicate the variety of animal models with specific characteristics that are available and that could be used to study specific components and stages of the inflammation process. Furthermore, we describe the effects of some targeted interventions, comparing animal studies with clinical trials. Finally, the importance of choosing a suitable animal model of arthritis for screening the preclinical efficacy of newly developed therapeutics is addressed.

Animal models of rheumatoid arthritis

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Collagen-induced arthritis in mice.

CIA is a frequently used animal model of human RA. CIA can be initiated in DBA/1 mice by intradermal injection of an emulsion consisting of bovine type II collagen in Freund's complete adjuvant (CFA). A booster injection with collagen, administered intraperitoneally 21 days later, is usually followed by the development of chronic arthritis mainly affecting the hind paws (1, 2). Mouse CIA shares many clinical, histologic, and immunologic features with human RA (3). These shared features include symmetric joint involvement, synovitis, and cartilage and bone erosions (Table 1). Even the production of rheumatoid factor has been reported. A clear difference between CIA in mice and human RA is that mice develop antibodies directed to collagen, while this is not the case in a large proportion of patients with RA (Table 1). More recently, a protocol has been developed for the induction of CIA in nonsusceptible C57BL/6 mice, which can be used in genetically modified mouse strains. In this model, 2 intradermal injections with chicken type II collagen in CFA on day 0 and day 21 results in the development of arthritis (4, 5). CIA induced in C57BL/6 mice is characterized by reduced severity of arthritis, reflected by lower clinical scores and reduced paw swelling compared with that observed in DBA/1 mice. The clinical score better reflects disease progression than does the caliper-derived score used to measure joint swelling, because the digits of the paws are mainly affected.

Table 1. Similarities and differences between animal models of arthritis and RA*
Animal modelSimilarities to RADifferences from RA
  • *

    RA = rheumatoid arthritis; CIA = collagen-induced arthritis; MHC = major histocompatibility complex; NSAIDs = nonsteroidal antiinflammatory drugs; PGIA = proteoglycan-induced arthritis; AIA = adjuvant-induced arthritis; SCW = streptococcal cell wall; STIA = serum transfer–induced arthritis; Tg = transgenic; TNF = tumor necrosis factor.

CIA in miceSymmetric joint involvement, peripheral joints affected, synovitis, cartilage and bone erosions, inflammatory cell infiltrate, pannus formation, erythema, edema, genetically regulated by MHC and non-MHC genesFormation of antibodies to collagen, greater incidence in males, periostitis, poor responses to NSAIDs, not characterized by exacerbations and remissions
CIA in ratsHigher susceptibility in females, symmetric joint involvement, peripheral joints affected, synovial hyperplasia, inflammatory cell infiltrate, genetically regulated by MHC and non-MHC genes, production of rheumatoid factorNot characterized by exacerbations and remissions
PGIA in miceDevelopment of polyarthritis, presence of rheumatoid factor, deposition of immune complexes in the joint, persistent joint inflammationDevelopment of ankylosing spondylitis, not characterized by exacerbations and remissions
AIA in ratsSymmetric joint involvement, inflammatory cell infiltrate, cartilage degradation, synovial hyperplasia, genetic linkage, T cell dependenceDamage to cartilage less severe than in RA, bone destruction more prominent; no rheumatoid factor produced, gastrointestinal tract and skin affected
SCW-induced arthritis in miceCharacterized by exacerbations and remissionsNone specified in publications
Polyarticular SCW-induced arthritis in ratsSymmetric joint involvement, synovial hyperplasia, inflammatory cell infiltration, relapsing inflammationNo rheumatoid factor produced
Monarticular SCW-induced arthritis in ratsCharacterized by exacerbations and remissionsNone specified in publications
STIA in miceInflammatory cell infiltrate, synovial hyperplasia, pannus formation, cartilage destructionNone specified in publications
K/BxN-Tg miceSymmetrically affects small peripheral jointsDistal interphalangeal joints often affected, no systemic manifestations, no production of rheumatoid factor, arthritis does not remit
Human TNF–Tg miceSynovial hyperplasia, presence of an inflammatory cell infiltrate, pannus formation, cartilage destruction, and bone resorptionNo production of rheumatoid factor

Susceptibility to RA and CIA is linked to the expression of specific class II major histocompatibility complex molecules (2, 6). Earlier studies showed that mice with the H-2q haplotype (e.g., DBA/1 mice) had the highest susceptibility, whereas the H-2b haplotype (e.g., C57BL/6 mice) conferred the lowest responsiveness to immunization with collagen and adjuvant. Class II MHC haplotypes and antigen presentation of specific collagen peptides to T cells do influence the responsiveness of mice, but the adapted induction schedule and the use of a higher concentration of Mycobacterium tuberculosis have proven that CIA can be induced in mice strains that were previously considered to be unresponsive (4, 7).

The immune system is intimately involved in the development and maintenance of inflammation in the CIA model (Table 2). B cells, and specifically antibodies recognizing collagen, appear to be the main players influencing pathology. T cells are thought to play a role in the induction phase of the disease, supporting the activation of collagen-specific B cells, but have a less substantial role in the chronic phase. However, the development of chronic arthritis requires both T cells and B cells to generate a sufficient immune response. Collagen-specific antibodies produced by activated B cells bind collagen in the joint, resulting in complex formation, complement activation, and triggering of a local inflammatory response by which monocytes, granulocytes, and T cells are attracted to the joint. This further instigates immune activation, resulting in production of cytokines and mediators driving inflammation and destruction of cartilage and bone (2, 8). Of note, neutrophils are abundantly present in the synovial tissue of mice with CIA (9), which is in contrast to RA synovium, where macrophages constitute the major phagocytic cell population (10).

Table 2. Involvement of immune cells, complement, and major histocompatibility complex (MHC) in animal models of arthritis*
 Mouse CIARat CIARat AIAMouse PGIAMouse SCW-induced arthritisRat polyarticular SCW-induced arthritisRat monarticular SCW-induced arthritisMouse STIAK/BxN-transgenic mouseHuman TNF–transgenic mouse
  • *

    CIA = collagen-induced arthritis; AIA = adjuvant-induced arthritis; PGIA = proteoglycan-induced arthritis; SCW = streptococcal cell wall; STIA = serum transfer–induced arthritis; TNF = tumor necrosis factor; PG-PS = peptidoglycan–polysaccharide; G6PI = glucose-6-phosphate isomerase; IA = intraarticular; NR = not reported; APCs = antigen-presenting cells; NK = natural killer.

AntigenType II collagenType II collagenHsp65, peptide 180-186ProteoglycanSCW componentsPG-PS 10SPG-PS 100PG6PI-specific autoantibodiesG6PIHuman TNF
Monocyte/macrophagesYesYesYesYesYesYes, starting during acute phase (∼5 days after injection)Yes (after initial neutrophil phase, a few days after IA injection)YesYesYes
Dendritic cellsYesNRYesNR; B cells are dominant APCsNRNRNRNRYesNR
GranulocytesYesYesYesYesNRNRYes (during initial and reactivation response)YesYesYes
T cellsYes, CD4+, type II collagen reactive, mainly during inductionYesYes, synovialYes, CD4+ T cellsYes (only during reactivation phase), main role for CD4+ T cellsYes, only during chronic phaseYes (during reactivation phase)No (can increase severity but are not crucial for disease induction)Yes, autoreactive to G6PINo
B cells, antibody productionYes, production of complement fixing type II collagen–reactive antibodiesYes, production of antibodies to type II collagenYesYes, auto-antibodies required for initiation of diseaseSCW-specific antibodies are detectedYes, only during chronic phase; minimal/no antibody response to PG-PSNRNo (G6PI-specific autoantibodies crucial, B cells in recipient mice not crucial)Yes, produce G6PI- specific auto-antibodiesNo
NK cellsDampen CIANRNRNRNRNRNRNRNRNR
ComplementYesYesNoYesNRYesYesYesYesNR
Specific MHCNoYesNoNoNRNoNRNoYesInfluences severity of arthritis

Cytokines influence disease induction and development in a time-dependent manner (2) (Table 3). There is an initial Th1 cell–type response in draining lymph nodes, with production of Th2 cytokines such as interleukin-4 (IL-4) and IL-10 being significantly suppressed. At the time of the clinical onset of arthritis, IL-1, IL-10, TNFα, transforming growth factor β (TGFβ), and IL-6 can be detected at the site of inflammation in the joint. The involvement of interferon-γ (IFNγ) is not completely clear, because contradictory results have been reported in studies involved in either blocking IFNγ or injecting this cytokine (6). DBA/1 mice deficient in IFNγ or its receptor were found to be highly susceptible to the development of CIA (6). More recently described cytokines such as IL-17, IL-21, IL-23, IL-32, and IL-33 have all been reported to aggravate CIA (11, 12).

Table 3. Involvement of cytokines in animal models of arthritis*
 Mouse CIARat CIARat AIAMouse PGIAMouse SCW-induced arthritisRat polyarticular SCW-induced arthritisRat monarticular SCW-induced arthritisMouse STIAK/BxN-Tg miceHuman TNF–Tg mice
  • *

    CIA = collagen-induced arthritis; AIA = adjuvant-induced arthritis; PGIA = proteoglycan-induced arthritis; SCW = streptococcal cell wall; STIA = serum transfer–induced arthritis; Tg = transgenic; TNFα = tumor necrosis factor α; IL-1β = interleukin-1β; NR = not reported; IFNγ = interferon-γ; MCP-1 = monocyte chemotactic protein 1; MIP-1α = macrophage inflammatory protein 1α.

TNFαYesYesYes, detectable 4 days post-injectionYesYes/negative, minor role, only expressed during initial joint swellingYesYes, during reactivationYes/negative, varying resultsNoYes
IL-1βYesYesYes, detectable 4 days post-injectionYesYes, involved in cartilage breakdown and inflammatory cell influxYesYes, during reactivationYesYesYes
IL-4No (can dampen inflam-mation)No (can dampen inflammation)Not until later stage (can dampen response)No (can dampen inflam-mation)NoYesYesNoYesNR
IL-6YesYesYes, detectable 4 days post-injectionYesYesYesYesNoNoNo
IL-10No (can dampen inflam-mation)No (can dampen inflammation)No (can dampen response)No (can dampen inflam-mation)No (can dampen inflammation by influencing TNF levels)NoNoNRNRNR
IL-12No (protects from inflam-mation)NRNRYesYesNRNRNoNoNR
IL-17YesYesYesNoYes, required to switch from an acute to a chronic reactionYesYesNRNRNR
IL-21YesNRYesNRNRNRNRNRYesNR
IL-23YesYesNRNRYes, chronic stageNRNRNRNRNR
IL-32YesNRNRNRNRNRNRNRNRNR
IFNγContradictory findings, possible role in regulating T cellsYesYesYesNRYes/no, conflicting reportsNoNo (can dampen response)NRNR
MCP-1YesYes, recruitment of monocytes, plays role in development of arthritisYesYesNRNRYes (during reactivation phase, up-regulated via IL-4)NRNRNR
MIP-1αYesYesYesYesNRNRYes (reactivation phase)NRNRNR
MIP-2YesYesNRYesNRNRYes (reactivation phase)NRNRNR

Collagen-induced arthritis in rats.

Three different cartilage-derived proteins can induce arthritis in rats: type II collagen, type XI collagen, and cartilage oligomeric matrix protein. In this review article, we focus on CIA initiated with type II collagen, which is the model that is used most often.

Intradermal injection of autologous collagen in an emulsion with Freund's incomplete adjuvant leads to the development of a severe polyarthritis in DA rats and Lewis rats, beginning ∼2 weeks after immunization. The onset of arthritis is characterized by swelling of the paws of both the fore limbs and hind limbs. Swelling can persist for a few weeks, decreases gradually, and reappears, resulting in chronic arthritis with severe consequences such as malformation of bone structures (13).

The rat CIA model is useful in addressing the effects of compounds in the late, chronic stage of arthritis. It has many similarities with human RA, such as higher susceptibility in females, but it is self-limiting and not characterized by exacerbations and remission (Table 1). In a recent study, a gene array analysis of rat CIA and human RA revealed that there are significant differences in the inflammatory mechanisms between human RA and rat CIA (14). Still, this model has been shown to be valid for evaluation of the effect of new treatments (3).

B cells play an important role in rat CIA, because adoptively transferred antibodies can induce arthritis in recipient rats (15) (Table 2). Collagen-specific antibodies have the ability to form immune complexes in the joint, and they can interact with complement components and initiate inflammation. The role of complement has been demonstrated by complement depletion in rats, using cobra venom factor. In these rats, arthritis onset was delayed until complement levels returned to normal (16). T cells are involved, but transfer of these cells alone induces only a mild response (17). However, T cells are essential, because athymic nude rats that lack T cells are resistant to the induction of arthritis. Neutrophils play a role in rat CIA, because depletion of this cell population results in decreased swelling and inflammation (18). Susceptibility to arthritis is linked to specific MHC genes (19). Large numbers of class II MHC–positive cells are present in the joint, and it has been hypothesized that these antigen-presenting cells (APCs) interact with and activate CD4+ T cells that are present in the joint, resulting in continued inflammation (20, 21) (Table 2).

TNFα and IL-1β are key cytokines in rat CIA (Table 3). The kinetics of these cytokines during CIA development has been investigated using immunohistochemical analyses of synovium. These studies showed that both TNFα and IL-1β are detectable more than a week before leukocyte infiltration and disease onset (22). Fibroblast-like synoviocytes and blood vessel endothelial cells were the producers of these cytokines during the prearthritis phase. After the onset of clinical disease, macrophages were the primary source, and TNFα-producing cells were more abundant than cells producing IL-1β (22). Another study, however, showed no clear increase in TNFα in rat paw tissue extracts, as measured by enzyme-linked immunosorbent assay during the course of CIA, although IL-1β levels were increased (23).

Adjuvant-induced arthritis.

AIA is initiated in Lewis rats by intradermal injection of CFA at the base of the tail (24, 25). The genetic background of rats is important, in that both MHC and non-MHC genes contribute to their susceptibility to AIA (26). Specific trait loci are associated with the severity of disease (26). AIA is characterized by a rapid onset and progression to polyarticular inflammation. Signs of arthritis usually develop between day 10 and day 14 following the injection of CFA. Usually the disease is severe and leads to permanent joint malformations, including ankylosis. Joint swelling, lymphocyte infiltration, and cartilage degradation are shared features with human RA (27) (Table 1). However, damage to cartilage is less severe than that in RA, whereas bone destruction is more prominent.

In rats with AIA, activated T cells can be detected in the inflamed joints (Table 2). T cells infiltrating the joint originate from various compartments, including the spleen, draining lymph nodes, Peyer's patches, and the recirculating T cell pool (28). The antigen that induces the immune response has been shown to be Hsp65, with the responsible epitope being peptide 180–186 (29). However, other components in CFA are important as well, because injection of only Hsp65 cannot trigger arthritis (30). Of note, the injection of mineral oil only can result in oil-induced arthritis; however, only a transient arthritis develops, in contrast to the chronic erosive arthritis seen in AIA (20, 31).

In contrast to the ability of heat-shock protein (HSP) to induce an inflammatory response in rats, HSPs are reported to have immunomodulatory capacities in patients with juvenile idiopathic arthritis, highlighting a clear difference between animal models of arthritis and patients with RA (32, 33).

Cytokines expressed in the joint during the early stages of inflammation include IL-17, IFNγ, and TNFα, accompanied by cytokines involved in macrophage stimulation, specifically IL-1β and macrophage inflammatory protein 1α (MIP-1α) (34) (Table 3). As inflammation progresses in the joint, increased levels of IL-4, IL-6, monocyte chemotactic protein 1 (MCP-1)/CCL2, and TGFβ can be detected (34). TNFα, IL-1β, IL-21, and IL-17 are all involved in the pathology of this disease, because their blockade ameliorates disease (35–37). AIA is not joint-specific but is associated with granuloma formation in various organs and tissues, such as the spleen, liver, bone marrow, skin, and eyes (38).

Proteoglycan-induced arthritis.

Proteoglycan isolated from human cartilage can be used to induce arthritis in susceptible mouse strains. PGIA can be induced only in BALB/c and C3H mice, with susceptibility to PGIA varying between mice from different suppliers (39). PGIA is not MHC-specific, because other mice with the same class II alleles are not susceptible to PGIA (40). Development of polyarthritis, deposition of immune complexes, and the presence of rheumatoid factor are shared features with human RA. Proteoglycan, emulsified with an adjuvant, is intraperitoneally injected on days 0 and 21 and optionally also on day 42, resulting in the development of an erosive arthritis (40–42). In the earlier literature, CFA is mentioned as the adjuvant (43), whereas dimethyldioctadecylammonium bromide (DDA) is used in the more recent studies. Use of DDA results in an earlier onset of arthritis and increased arthritis severity, without granuloma formation and tissue irritation, as seen with CFA (40).

After injection of the emulsion, strong B cell and T cell responses develop (44). B cells have a dual role in PGIA, because they function as autoantibody-producing cells and are crucial in their role as APCs, and they activate proteoglycan-specific T cells (Table 2) (44). The adoptive transfer of arthritis to recipient mice requires both B cells and T cells in order to be successful; neither B cells nor antibodies alone can confer arthritis to recipient mice (45). It is proposed that subsequently produced antibodies form immune complexes with proteoglycan in mouse cartilage locally in the joint, resulting in complement deposition and further immune cell activation as well as cytokine and chemokine production (39).

The hallmark cytokines of arthritis, TNFα and IL-1β, are expressed during the effector phase of PGIA, as determined in messenger RNA (mRNA) isolated from joints (Table 3) (46). Furthermore, increased levels of MIP-1α, MIP-1β, MIP-2, and MCP-1 mRNA are detected. IFNγ influences PGIA development, because neutralization of IFNγ or its absence in IFNγ-deficient mice results in a decrease in arthritis severity (47). IL-4 and IL-10 can regulate and dampen the arthritis (39). PGIA has been used less often than CIA, due to the fact that human cartilage has to be obtained and extensively processed in order to prepare the needed proteoglycan fraction. PGIA induced with proteoglycan from other species has also been reported (40), which might facilitate the use of PGIA as a model to study new therapeutics for arthritis.

Streptoccocal cell wall–induced arthritis in mice.

Various versions of the SCW-induced arthritis model in mice have been described in susceptible strains such as BALB/c, DBA/1, and C3H mice (48). Originally, a model similar to rat monarticular SCW-induced arthritis was developed. In this model, mice were injected once intraarticularly in one knee joint with SCW fragments, thereby sensitizing the joint. Subsequent intravenous injection of SCW fragments could result in a flare reaction, as seen in the rat model (49). More recently, modified versions of this model have been described in which mice either are injected once intraarticularly or receive multiple intraarticular injections, resulting in continued inflammation (50). The initial intraarticular injection of SCW fragments induces acute joint inflammation, which resolves within a week. Repeated injections (≥3) result in the development of chronic arthritis (50).

Macrophages and B cells play a role in SCW-induced arthritis in mice, whereas T cells are involved only during the reactivation phase (Table 2). In contrast to its role in other experimental models of arthritis, TNFα plays a minor role in SCW-induced arthritis and is involved only in the initial joint swelling. IL-1β, however, is heavily involved in inflammatory cell influx and cartilage degradation and is important during both the acute and chronic phases of the disease (Table 3).

Streptococcal cell wall–induced arthritis in rats.

Purified peptidoglycan–polysaccharide (PG-PS) polymers produced from Streptococcus pyogenes (group A) possess a high inflammatory capacity and are capable of inducing arthritis in rats. Depending on the fragments of PG-PS used and the route of administration, 2 models of arthritis can be initiated (51, 52). These models are less frequently used to study therapeutic effects, possibly due to their high costs. However, SCW-induced arthritis has advantages over the AIA model, in that these models provide a possibility to study acute or flare reactions in arthritis.

Polyarticular SCW-induced arthritis.

The polyarticular model is induced by intraperitoneal injection of PG-PS 10S in female Lewis rats, resulting in an acute inflammatory response and swelling of the joints. The joint inflammation progresses during the first 5 days and is followed by a period of remission, after which spontaneous reactivation occurs, resulting in chronic arthritis. This model provides the opportunity to study both the early and the more chronic phases of arthritis (53). The initial response is not T cell dependent (Table 2). Monocytes are involved in both the acute phase and the further development of arthritis (54). During the chronic phase, the activation status of T cells, B cells, and monocytes seems to correlate with the severity of arthritis (54). Production of TNFα, IL-1β, and IL-6 correlates with the degree of inflammation in arthritic rats (55) (Table 3). No clear correlation exists between the severity of arthritis and antibody production to PG-PS, which is minimal (8). Two strains of rats with the same MHC differ in susceptibility, indicating that there is no association between disease development and specific MHC alleles. Similarities of this model to human RA include symmetric joint involvement, synovial hyperplasia, inflammatory cell infiltration, and relapsing inflammation (27) (Table 1).

Monarticular SCW-induced arthritis.

The reactivation model of arthritis encompasses a model whereby an intraarticular injection of PG-PS 100P in the hind ankle joint of female Lewis rats induces an initial surge of inflammation. This subsides in a few days, with diffuse infiltration of the synovium containing lymphocytes and monocytes remaining in the joint. After the initial intraarticular injection, the joints are sensitized, and inflammation in the joint can be reactivated by intravenously administered PG-PS 100P, lipopolysaccharide, or superantigen (8).

The monarticular SCW-induced arthritis model seems to be mediated by Th2 cells, because swelling in this model develops within hours. IL-4 is reported to be crucial in disease development, whereas blockade of IFNγ and IL-10 does not affect joint swelling (56) (Table 3). The response is primarily neutrophilic at 6 hours and changes to a mainly mononuclear cell infiltrate by 48 hours. Granulocytes are also involved during the reactivation phase, because antibodies to neutrophils inhibited swelling for 3 days after reactivation (57). T cells are mainly involved in the reactivation phase of arthritis, because depletion of T cells by monoclonal antibodies rendered rats unresponsive to intravenously injected PG-PS (58) (Table 2).

IL-1, TNFα, and IL-4 are essential during the reactivation phase (59). The combined action of IL-4 and TNFα enhances T cell adhesion molecules and may thereby regulate part of the transition from an acute to a chronic T cell–mediated phase (56). Chemokines such as CCL2/MCP-1, MIP-1α, and MIP-2 all play a role in the response after reactivation (Table 3).

The monarticular SCW-induced arthritis model is well suited for studying the effect of therapeutics over time, because it mimics flares of arthritis similar to those observed in patients with RA (Table 1). Furthermore, the animals can be monitored for a longer period of time, because the disease is less severe compared with other models of arthritis. Of note, the kinetics of the expression of cytokines such as IL-1β, IL-6, and TNFα suggest that the optimal time point for evaluating the efficacy of treatment is 3 days after intravenous challenge (60).

Transgenic mouse models of RA

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

In addition to the induced-arthritis models, arthritis develops spontaneously in some genetically modified mice (8, 61). These are mice that are either deficient in (knockout) or transgenic for a specific gene of interest. These models give valuable information regarding the role of genes in the inflammatory process and provide a tool with which to study the effect of therapeutics in mice in which joint inflammation develops spontaneously. Two well-known models are the K/BxN mouse model and the human TNF–transgenic mouse model.

K/BxN mice and serum transfer–induced arthritis.

Arthritis develops spontaneously in K/BxN mice, which are generated after crossing KRN-C57BL/6 T cell receptor–transgenic mice with NOD mice. In K/BxN mice, arthritis is caused by an immune response to the enzyme glucose-6-phosphate isomerase (G6PI). Symptoms of disease can be observed at ∼4–5 weeks of age (62). Autoreactive T cells recognize a peptide derived from G6PI presented by APCs on I-Ag7 class II MHC molecules and activate B cells to produce G6PI-specific autoantibodies. T cells are critical in the initiation phase of the disease, because injection of an anti-CD4 antibody can block arthritis only when it is given, at the latest, 5 days before disease onset. Injection of CD4-specific antibody at a later time point does not affect arthritis development (Table 2). IL-1 is required in this model, but TNFα is not needed for development of K/BxN arthritis, nor is IL-6 (Table 3).

In another variant of this model, immunization with G6PI induced a T cell–dependent arthritis in genetically unaltered mice (63). An important difference between this variant and K/BxN arthritis is spontaneous remission after the maximum clinical score is reached, 14 days after immunization (64).

The autoantibodies produced by activated B cells are a crucial factor in K/BxN arthritis and are the active components used in the STIA model. A transient arthritis can be induced by intraperitoneal injection of serum from K/BxN mice, resulting in the influx of inflammatory cells, hyperplasia of synoviocytes, pannus formation, and cartilage destruction in the recipient mice (65). Inflammation in the joints resolves, unless mice are repeatedly injected with K/BxN serum. STIA can be induced in a variety of mice strains and is not MHC-restricted (66).

Neutrophils have a crucial role in the initiation phase of STIA (Table 2). Immune complexes formed by G6PI, complement components, and IgG1 antibodies present in the injected serum bind to Fcγ receptor type III on neutrophils, resulting in neutrophil activation and production of mediators such as TNFα (67). This results in vascular permeability, allowing antibodies and cells to enter the joint cavity. Mast cells are activated after G6PI immune complexes, allowing further release of (vasoactive) mediators.

Antibodies that have reached the joint cavity can bind to the cartilage surface, trigger complement, and activate a wide range of immune cells. This will ultimately result in a chronic condition with immune cell infiltrates (67). B cells and T cells are not required for arthritis induction after serum injection, because arthritis can be induced in recombination-activating gene 1 (RAG-1)–knockout mice lacking B cells and T cells (65). However, more severe and persistent arthritis develops in mice that lack B cells only, indicating that T cells can influence disease progress in the later stages. IL-1β is the main cytokine necessary for STIA (Table 3). The role of TNFα is not clear, because arthritis still develops in a proportion of TNFα-knockout mice after serum transfer (68).

The STIA model has not been used as often as the CIA and AIA models for validation of therapeutics. One reason may be that this model highlights only the pathologic effect of antibodies and does not involve B cells and T cells, although it does show erosive synovitis that is dependent on neutrophils, mast cells, macrophages, and mediators of inflammation. A main advantage of this model is that it can be used in various (knockout) mouse strains.

Human TNF–transgenic mice.

Human TNF–transgenic (Tg197) mice express high levels of soluble and transmembrane human TNFα (69) and undergo the spontaneous development of arthritis. Synovial hyperplasia and inflammatory cell infiltrates can be observed from 3–4 weeks of age, and mice have fully developed disease at 10 weeks of age. Features in common with human RA include synovial hyperplasia, the presence of an inflammatory cell infiltrate, pannus formation, cartilage destruction, and bone resorption (Table 1) (70). The human TNF transgenics were generated using mice with MHC haplotypes that are less favorable for arthritis development, namely, H-2K and H-2B, which might bring into question whether MHC has an important role in this model. However, MHC can influence disease development, because backcrossing to the arthritis-susceptible DBA/1 background resulted in earlier onset and the development of more severe disease.

Mice generated by crossing human TNF–transgenic mice with RAG-1–knockout mice, which have no T cells or B cells, still develop erosive arthritis, indicating that arthritis in this model is not dependent on T cells or B cells (Table 2) (71). Fibroblast-like synoviocytes obtained from synovium of human TNF–transgenic mice can induce synovitis, cartilage damage, and bone damage when they are transplanted into normal mice, further exemplifying the relative lack of involvement of immune cells in this model of arthritis (72).

In this model, the destructive effect of excess TNFα and the relationship between TNF and IL-1β in the pathogenesis of arthritis were studied (73). The model has been used very effectively in the assessment of anti-TNF therapies but can also be used for testing other biologics and small molecules (69).

From animal models toward the clinic

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

Both mouse and rat models of experimental arthritis have been extensively used to evaluate the effect of therapeutics in the preclinical development phase. Positive results obviously facilitate progression to clinical trials. A wide variety of animal models for arthritis exist, each of which has its own similarities to human disease (Table 1) (27).

Conventional medication for the treatment of arthritis consists of nonsteroidal antiinflammatory drugs (NSAIDs), corticosteroids, and disease-modifying antirheumatic drugs including methotrexate (MTX). The majority of these drugs did not follow a classic drug development route for the treatment of arthritis, because they were already being prescribed for other diseases. However, the therapeutic efficacy of these drugs has been described in animal models of arthritis, including AIA and CIA (3). For example, oral treatment with the NSAID indomethacin at a dosage of 1–3 mg/kg/day in the rat CIA model resulted in 50–90% inhibition of arthritis (74). In mice with CIA, MTX injected intraperitoneally from the time of the booster injection until the end of the experiment resulted in clear inhibition of the development of arthritis, as shown by reduced clinical and histology scores (75). In the next section, we will discuss various targeted treatments that are currently approved for the treatment of RA and compare the outcome of treatment in animal studies with results obtained in clinical trials.

TNFα inhibitors.

TNFα is considered a key factor involved in the immune process driving RA. The role of TNFα and the effect of blocking its activity were investigated in various animal models. A hamster antibody specific for murine TNF was capable of reducing clinical manifestations and histologic scores in a mouse CIA model, also after disease onset (76). Spontaneously developed arthritis in human TNF–transgenic mice was blocked by treatment with a human TNFα–specific antibody (69). Treatment from birth prevented joint disease, and the joint histology of treated human TNF–transgenic mice matched that of wild-type, nonarthritic mice. Administration of the TNF inhibitors etanercept (fusion protein of TNFR type II [p75] and the Fc region of human IgG1), infliximab (chimeric mouse/human monoclonal TNFα antibody), and adalimumab (fully human anti-TNFα antibody) showed efficacy when given prophylactically (treatment started at 3 weeks of age). For the first 2 TNF blockers, efficacy was demonstrated when treatment was started at 6 weeks of age (77). A soluble recombinant human TNFR Fc fusion protein (TNFR:Fc) that can bind and neutralize TNFα was able to reduce both the incidence and severity of CIA in mice when it was injected before the onset of disease (78). Treatment with TNFR:Fc also resulted in decreased clinical scores in mice with established CIA. In rat AIA, treatment with PEGylated TNFR was effective in reducing paw swelling by 50% when given at the onset of clinical signs of arthritis (79). In rat CIA, treatment that was started after the appearance of clear clinical manifestations resulted in a mean reduction in ankle joint swelling of 46% (80).

These positive results were consistent with those from clinical trials that were successful (81, 82), resulting in US Food and Drug Administration approval of TNFα inhibitors, including infliximab, adalimumab, and etanercept, for the treatment of RA. The mechanism of action and clinical profiles of these and new-generation therapeutics blocking the action of TNFα have previously been reviewed in detail (82).

Two new TNF inhibitors, golimumab (a fully human anti-TNFα monoclonal antibody) and certolizumab pegol (a PEGylated Fab′ fragment of a humanized anti-TNF monoclonal antibody), have more recently been shown to be effective in the treatment of RA (83–88). Thus, TNFα has clearly been proven to be a good target for RA treatment and has fulfilled the expectations raised during the preclinical studies performed in animal models.

IL-1 inhibitors.

IL-1 is well known for its role in animal models of arthritis, specifically for its influence on bone metabolism (89, 90). Drugs focused on blocking the actions of IL-1 have been developed, using either an IL-1 receptor antagonist (IL-1Ra) or antibodies against IL-1 or its receptor. In animal models, both IL-1Ra and the antibodies have shown efficacy. In a rat CIA model, continuous administration of IL-1Ra suppressed both inflammation and bone resorption (91). This result was partially reproducible in rat AIA, demonstrating an effect on bone destruction only (91). In the rat polyarticular SCW-induced arthritis model, treatment with IL-1Ra that started just prior to reactivation and continued over 5 days resulted in inhibition of joint swelling (92). In mice with CIA, treatment that started either before or during established disease proved effective. Treatment of mice with AIA and mice with immune complex–mediated arthritis with IL-1Ra was effective, as was administration of an IL-1 antibody (93, 94). When IL-1Ra was used, continuous administration resulting in sufficient blood levels of the drug was a prerequisite for therapeutic success (91).

These encouraging results led to the development of anakinra, an IL-1Ra. However, anakinra has shown only limited efficacy in patients with RA (90). It has been proven to reduce inflammation and bone erosion in patients with RA but not to the same extent as TNFα-blocking drugs (95) or other biologics (96). There has been a debate as to whether the limited effectiveness of anakinra was related to its pharmacokinetic properties.

New drugs targeting the IL-1 pathway have been developed more recently. One of these is AMG 108, a fully human monoclonal antibody binding to IL-1R type I. In a phase II trial, only patients injected with the highest dose (250 mg, administered subcutaneously once a month for 6 months) in combination with MTX achieved significantly more responses according to the American College of Rheumatology criteria for 20% improvement in disease activity (ACR20) (97) and ACR50 (40.4% and 20.2%, respectively) compared with the placebo control (29.1% and 8.4%, respectively); there was no difference in ACR70 responses (98). The notion that the IL-1 pathway may not be the best therapeutic target in RA was further supported by the experience with ACZ885 (canakinumab), a fully human monoclonal antibody that can bind and neutralize human IL-1β. In a human IL-1β–dependent mouse model of joint inflammation, ACZ885 reduced swelling and cartilage damage (99). However, in a proof-of-concept clinical trial, it was demonstrated that intravenous injections of ACZ885 on day 1 and day 15 resulted in clinical improvement in only a subset of the patients with RA, but this clinical improvement did not reach statistical significance (99).

Taken together, the outcomes of IL-1–targeting therapy in patients with RA have been disappointing, in spite of success in the treatment of autoinflammatory syndromes (100), and do not correlate with the results observed in animal models of RA. For IL-1Ra, this could perhaps be ascribed to the short half-life of the molecule and the high dose that must be administered in order to occupy all IL-1 receptors and prevent IL-1 binding (101). However, the recent results using antibodies against IL-1 and IL-1R, respectively, indicate that the role of IL-1 in animal models may be quite different from that in human RA.

IL-6 inhibitors.

IL-6 is abundantly expressed in both the synovial fluid and serum of patients with RA (102); its expression in synovial tissue is significantly correlated with clinical signs and symptoms of arthritis (10). It has a variety of functions, including initiation of the acute-phase response, activation of vascular endothelial cells, and induction of B cell and T cell differentiation (89, 103).

Studies in animals have clearly suggested the involvement of IL-6 in RA. In IL-6–knockout mice, CIA was less severe and had a delayed onset compared with CIA in wild-type mice (104). Similarly, injection with methylated bovine serum albumin induced AIA in wild-type mice but failed to induce joint swelling in IL-6–knockout mice (105). The potency of IL-6 receptor blockade was subsequently tested in animal models of arthritis. One intraperitoneal injection of a monoclonal antibody recognizing the IL-6 receptor on day 0 or day 3 suppressed the development of murine CIA (106). Treatment with an antibody targeting IL-6 itself was also shown to reduce the incidence and severity of CIA in mice (107).

These positive results prompted an investigation into the potency of IL-6 as a target in patients with RA. Tocilizumab is a humanized monoclonal antibody recognizing the 80-kd component of the IL-6 receptor. Over the past years, many clinical trials have been performed proving the efficacy of tocilizumab in RA (108–110), and tocilizumab has now been approved for the treatment of RA in Japan and Europe.

T cell–targeted therapy with abatacept.

T cell activation and proliferation are important factors in the initiation and perpetuation of arthritis. The possibility of treating arthritis by interfering with the interaction between the costimulatory molecules CD28 on T cells and CD80 (B7-1) and CD86 (B7-2) on APCs has raised interest in the field of rheumatology (111, 112). CTLA-4 (CD152) is a homolog of CD28. It binds to both CD80 and CD86 with higher affinity than CD28, thereby preventing T cell activation. This knowledge has led to the development of soluble CTLA-4Ig, which consists of the extracellular portion of CTLA-4 fused to the modified Fc portion of human IgG1. Studies in animal models proved its efficacy in dampening the manifestations of arthritis. Administration of CTLA-4Ig in mice at the time of the first collagen injection prevented the development of CIA (113). When treatment was delayed until arthritis had developed, CTLA-4Ig was still able to ameliorate disease. In a rat model of CIA, prophylactic treatment with CTLA-4Ig prevented the development of arthritis (114). Treatment after the onset of disease in G6PI-induced arthritis resulted in a reduction of clinical scores, further exemplifying that CTLA-4Ig can mediate its effect when inflammation is ongoing (115).

Abatacept is a CTLA-4Ig molecule developed for clinical application. In clinical trials, abatacept was effective both as monotherapy (116) and when tested in combination with MTX (117). In a study in patients who had previously received anti-TNF drugs, abatacept could induce ACR criteria responses in a significant proportion of the patients involved in the trial (118). Abatacept has been approved for the treatment of moderate-to-severe RA that has responded inadequately to other treatments.

Conclusions

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

We have provided an overview of the immunologic factors involved in animal models of RA (Tables 2 and 3) and have summarized similarities and differences between these animal models and human RA (Table 1). We described the relationship between efficacy data in preclinical in vivo studies and the experience in clinical trials. It is clear from these data and the literature that there are no definitive guidelines regarding which animal model will have the most value for predicting outcome in clinical trials in RA. This can be attributable to several reasons, as follows: 1) studies performed in inbred animals are compared with a heterogeneic patient population; 2) the components and functioning of the immune system are different between animals and humans, and, therefore, therapeutics can act differently in animal models of arthritis versus human RA; and 3) timing of treatment is crucial, because cytokines and cells may play a different role during different stages of the disease process. Animals can be treated in the very early stages of disease, whereas patients with RA are included in clinical trials when chronic synovial inflammation is already present.

Thus, treatments that are effective in animal models of arthritis, such as IL-1 blockade, may fail in RA. Still, animal models remain a very important tool for preclinical screening of new therapeutics in pharmaceutical development, and thorough consideration as to which animal model should be used is required. The choice of the animal model is dependent on the specific therapeutic target to be tested. In addition, it is recommended to mimic the variability in the human situation as much as possible, which includes using >1 experimental arthritis model (taking the heterogeneity of human disease into consideration), preferably also in >1 species. Furthermore, a therapeutic rather than a prophylactic approach must be taken when testing the efficacy of new drugs (3). Patients with RA who present with signs and symptoms have disease manifestations comparable with active, chronic disease in animal models. The tables summarizing the characteristics of each model provided in this review can aid researchers in choosing the most suitable animal model for their novel targets and therapeutics in development.

It should be noted that therapeutics that are discontinued due to disappointing results in animal models of arthritis might still be potent in dampening inflammation and bone damage in patients with RA. Typically, programs evaluating drugs that fail in animal models of arthritis are discontinued for obvious reasons. Therefore, the negative predictive value of the animal models is not known. Further knowledge on human immunology and development of tools to aid this research might enable testing circumstances that bear more resemblance to the human situation (119).

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES

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.

REFERENCES

  1. Top of page
  2. Introduction
  3. Animal models of rheumatoid arthritis
  4. Transgenic mouse models of RA
  5. From animal models toward the clinic
  6. Conclusions
  7. AUTHOR CONTRIBUTIONS
  8. REFERENCES