Rheumatoid arthritis (RA) is a multisystem inflammatory autoimmune disease that destroys the surrounding joints. The basic pathological change that occurs in RA is chronic synovitis, which is accompanied by synovial cell proliferation, inflammatory cell infiltration and the formation of vasospasms that invade the cartilage and bone tissue of the subsynovial layer, causing joint destruction (1-4). When the disease progresses to an advanced stage, the joint tissue is severely damaged and all joint function is lost. Furthermore, to a certain extent, the lungs (5), cardiovascular system (6,7), nervous system (8) and other organs (9,10) are selectively affected, which seriously impairs the patients' quality of life. RA is a disease that occurs globally. It may occur in adults of any age but is mainly diagnosed in middle-aged females (11,12). For the treatment of RA, timely and effective control of disease progression are urgently required. Early diagnosis and treatment may prevent bone and joint damage, as well as reduce disability and suffering, which lays a foundation for improving the quality of life of affected patients (13).

To date, the etiology and pathogenesis of RA have remained to be fully elucidated. Due to its complex pathogenesis, there is currently no ideal drug that is able to completely cure RA. Accordingly, animal models of RA are an important resource for studying and exploring the pathogenesis of RA, as well as developing effective anti-inflammatory drugs. These animal models may be classified in several ways according to species (mainly rat and mouse), disease type (genetically engineered, induced or spontaneous) and inciting agent (chemicals, collagen or exogenous polysaccharides/proteins/proteoglycans) (14). Commonly used models include collagen-induced arthritis (CIA) (15), proteoglycan-induced arthritis (16), Staphylococcus aureus-induced arthritis (17,18) and genetically engineered arthritis mice (such as K/BxN mice) (19,20). In addition, chimera models are frequently used to test drugs with specific targets by transferring corresponding human tissue samples to nonobese diabetes (NOD)/severe combined immunodeficiency (SCID) mice. For instance, the human RA synovium-cartilage-NOD/SCID mouse chimera model (arthritis/SCID mouse chimera model) (21,22) may be used to test the mechanism of synovial invasion of cartilage and bone and the efficacy of related drugs. General practical considerations in the use of various rodent disease models have been reviewed by Bolon et al (14) and Williams (23). Caplazi et al (24) discussed mouse models of RA, providing a wider perspective regarding systemically induced mouse models of RA as well as the value of polyarthritis and spontaneously occurring and genetically engineered models of RA.

The present review mainly focused on providing a detailed introduction to the CIA mouse model. This model is always the first choice and the most widely used, as it may be generated rapidly and inexpensively and is similar in pathogenesis to human RA (25). However, this model has several critical features that are frequently overlooked by researchers. For instance, the use of the model has a limited scope and the arthritis induction rate varies depending on the genetic background of the mice (26). Furthermore, there are several controversies regarding the modeling process, such as when drugs should ideally be administered and how to choose the best administration method. The present review provides detailed knowledge related to these topics. With this information, it is expected that researchers who are new to the field or unfamiliar with this knowledge are able to avoid unnecessary errors and select the appropriate model to obtain reliable results.

In 1977, Trentham et al (27) reported for the first time that the immunization of rats with a human, chicken or rat type II collagen (CII) emulsion in complete Freund's adjuvant (CFA) led to the development of erosive polyarthritis, accompanied by an autoimmune response to cartilage. This CIA model was reproduced in mice and monkeys in 1980 and 1986, respectively (28,29). Since CII is the main protein in articular cartilage, the immune response generated by CII mainly targets the joints. The production of CII-specific antibodies in mice is an important feature that has also been reported in RA (30). The clinical and histological appearance of CII-induced arthritis in mice indicates that this is an ideal animal model for the investigation of various immunogenetic traits in RA (31). For instance, mice immunized with CII also produce rheumatoid factors (26,32). Furthermore, the pathological features of both RA and CIA consist of marked synovitis with cartilage degradation and bone erosion (32).

Susceptibility to RA is closely related to certain allelic subtypes of human leukocyte antigen (HLA)-DR and HLA-DQ (23,33). The DR and DQ molecules expressed by these subtypes are able to present autoantigen peptides such as CII 260-273 and may be recognized by a specific T-cell receptor (TCR), thereby activating T cells (34). The susceptibility of mice to CIA is also closely related to major histocompatibility complex (MHC) class II molecules. The mouse histocompatibility 2, class II antigen A, beta 1 (IA) gene is homologous to HLA-DQ, and the mouse histocompatibility 2, class II antigen E alpha (IE) gene is homologous to the HLA-DR gene (35). It has been indicated that polymorphisms of the β1 chain of the IA molecule determine susceptibility to CIA and that the IE molecule is also involved in the regulation of CIA-related pathological processes, which are related to the incidence and severity of CIA (36). Sequence analysis of IA, IE, DQ and DR indicated that in humans and mice, the MHC class II molecule β1 chain expressed by the RA/CIA allelic subtypes frequently had the conserved sequence glutamine-lysine/arginine-arginine-alanine-alanine, known as the shared epitope. The spatial structure formed by the side chain molecules of these adjacent amino acid residues is the key binding site of the antigenic peptide. The MHC II molecules expressed by H-2q mice are IAq and IEq, and the MHC class II molecule that recognizes the presented CII antigen is IAq; therefore, IAq mice (such as DBA/1 and B10.Q mice) are frequently used to generate CIA animal models (31,37).

Accordingly, the establishment of CIA requires heterogeneous (bovine or chicken) CII to immunize susceptible mice two times. The disease-related CII antigenic peptide must contain a core peptide. CII is a homotrimer composed of three A1 chains containing 1,018 amino acids each. Within these chains, the CII 260-270 segment contains functional amino acids that are able to bind to MHC II molecules and be recognized by the TCR; they are known as the core antigen peptides and are related to the occurrence of RA and CIA. The core peptide is easily degraded at room temperature, which is why it is necessary to work at 4˚C when emulsifying collagen. As the H-2q haplotype has strong susceptibility to CII core peptides, it is able to induce strong T-cell proliferation and CIA production (38). Excessive activation of T cells, increased secretion of cytokines and production of CII-specific antibodies are the major factors involved in the pathogenesis of CIA (39). The CIA mouse model is mainly induced by CD4⁺ T cells and MHC class II-restricted T cells (40). In this disease model, T helper type 1 (Th1) cytokines are secreted and damage is mediated by the cellular immune response. T cells interact with antigen-presenting cells, activate lymphocytes and stimulate monocytes/macrophages to release numerous inflammatory factors, such as IL-1β and TNF-α, leading to CIA inflammation, hyperplasia of the synovial lining, neoangiogenesis, pannus formation and the destruction of cartilage and bone (41,42).

Of note, Bolon et al (14) and Williams (23), among others, reported that immunization of mice with heterologous type II collagen usually leads to a relatively acute and self-remitting form of arthritis. By contrast, immunization with autologous collagen results in more severe and prolonged arthritis that is probably more reminiscent of human RA. However, due to the low affinity of a specific epitope of murine collagen (CII256-270) for IAq, autologous type II collagen is less arthritogenic than heterologous collagen, resulting in a low level of CII-specific T-cell activation. The use of heterologous CII protein is still recommended as the first choice from a comprehensive perspective (43). In addition, CII-responsive T cells can only be induced when the amount of heterologous CII is large enough. An appropriate amount of CII should be used when immunizing mice; the optimal CII concentration is specified further below.

The CIA mouse model also has certain drawbacks. The joints of mice are small and CIA in mice exhibits a variable disease pattern. By contrast, rat arthritis models offer much larger specimen sizes and the distribution and extent of inflammatory changes in rat CIA joints are more reproducible (14). Therefore, modeling with mice requires more standardized modeling protocols and operations to ensure the replicability of the model as much as possible. Mice have the advantage of costing less than rats (44); thus, it is prudent to use a CIA mouse model in the initial screening of anti-RA drugs unless the purpose of the experiment has specific requirements for specimen size or other restrictions that would preclude mice.

In general, among the various mouse models, the CIA model is most similar to RA in terms of pathogenesis and clinical characteristics. It is relatively stable and is an ideal and internationally recognized arthritis model for studying the pathogenesis of RA and screening drugs for RA treatment.

There are three commonly used methods for administration: Intraperitoneal administration (61), intravenous administration (62) and oral administration (63). The medication route is selected according to the purpose of the experiment and the characteristics of the drug. In general, Traditional Chinese Medicine is administered orally by dissolving the medicine in drinking water or by gavage (64).

The time of administration should be given special consideration and depends on whether the researcher aims to detect the preventive or therapeutic effect of the drug. Normally, the time of the first immunization is referred to as D0 and the time of the second immunization of DBA/1 mice is D21. In most studies, mice are administered injections as preventive treatments on D0-10 and as therapeutic treatments on D21-25 (65-69). Based on the pathogenesis of the CIA model as described above, the specific development of RA pathology began after the first immunization. The second immunization only enhanced the symptoms (63). Approximately 10% of CIA mice develop symptoms of joint swelling, which can be visually observed between the first and second immunizations. It further verifies that the onset of pathological signs occurs during this period. Accordingly, the other 90% were observed 4-5 days after the second immunization (D25-D26 of the experiment) (70). Consequently, it is recommended that mice are administered preventive treatments from D0 and therapeutic treatments from D21; regardless of the drug's preventive or therapeutic effects, earlier administration of the drug is more efficient. Furthermore, the observation time should not be too long. If the lymphocyte reaction is being monitored, the best time-point is D31. If joint damage and deformation are being monitored, D31-D41 is an appropriate range. When the observation time exceeds 42-60 days, the mouse heals and symptoms of arthritis, such as redness and swelling, disappear (26).

Consideration should be given to the extent of animal suffering (76). In the process of CIA modeling, mice exhibit ulceration and inflammatory reactions in the tail skin due to stimulation from the emulsion at the base of the tail. Therefore, it is necessary to perform the injections as slowly as possible and the depth of the needle should be appropriate. If the needle is placed too superficially, it may penetrate the skin. This leads to leakage of the emulsion, which may increase the possibility of ulceration in the tail of the mouse. In general, mice experience substantial pain due to the process of CIA model generation and the symptoms of inflammation itself. Therefore, the researcher should be as gentle as possible with the experimental animals, provide stable keeping conditions and minimize errors and pain during the operations. Humane end-points with severity limits should be incorporated into protocols to limit suffering (76). When taking samples from sacrificed mice, researchers should keep their workspace away from the cages of the remaining animals to prevent panic and psychological tension caused by the mice witnessing the suffering of their peers. This panic may even cause changes in immune cells (77). In addition, it is worth mentioning that numerous small animals lose their lives in research and researchers should respect the animals. Operators should strictly follow the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health. In fact, one of the intentions of the present review is to help experimenters minimize errors and reduce the number of experimental animals used. The experiments should therefore be carefully planned and designed together with a primary investigator or supervisor, ensuring that each individual experiment is ethically approved beforehand.

The CIA mouse model has been used in numerous studies on RA etiology, pathogenesis, drug screening, transgenic mice and immunotherapy worldwide. Research on CIA and the standardized application of CIA animal models are of great significance in these fields. However, in certain studies, irregularities still exist in the establishment and application of this model. The present review examined the molecular basis and limitations of the CIA model in detail and aimed to provide a reference for the investigation and use of CIA. All of the small and easily overlooked but important details, including the reagents and protocol required for modeling, are provided in this report.

It is worth mentioning that scientists are also exploring humanized RA animal models, which may be used to verify the efficacy and mechanisms of anti-rheumatic drugs that involve humanized antibodies. At present, it is thought that the most stable model animals for humanized RA are humanized bone marrow/liver/thymus mice. In this model, 6- to 8-week-old NOD.Cg-Prkdc^scid Il2rg^tm1wjl/SzJ (NSG) immunocompromised mice receive a thymus/liver implant, as in the SCID-hu mouse model, followed by a second human hematopoietic stem cell transplant (78). The advantage of this system is the full reconstitution of the human immune system in the periphery. This model is stable for 12 to 18 weeks (79,80). Accordingly, in theory, it is most desirable to induce CIA in BLT mice and then assess the efficacy of humanized antibodies in them. However, several problems were encountered in the experiment. Frist, as the genetic background of NSG mice is NOD (MHC II H-2g), the success rate of CIA establishment is unknown. Second, the emergence of BLT mice is still relatively novel, meaning that the development of the model itself is not very mature, and that there are uncertainties associated with it. In addition, the cost of BLT mice is quite high right now. Therefore, much work remains to be performed to overcome the challenges associated with this model.