Introduction
Arthritis is a clinical entity characterized by joint swelling, stiffness and spontaneous or motion-related pain. Impairment of physical function, that is, disability, is the major complication of arthritis and the degree of its reversibility depends on both our potential to interfere with the active disease process as well as with joint damage, given that the latter is currently not reversible [
1].
Our current knowledge on the pathogenesis of chronic inflammatory joint diseases such as rheumatoid arthritis (RA) is still limited. While evidently there are a variety of important genetic factors involved [
2] and a large proportion of patients develops autoantibodies consistent with the involvement of an autoimmune response, neither the trigger(s) of disease are known nor are the pathways leading to chronicity fully understood. Therefore, animal models of arthritis are very useful in dissecting these pathways, although each of them may depict only a facet of RA or other forms of inflammatory joint disease. To a large extent these models rely on the assessment of disease activity, which is hampered by the inability to obtain subjective symptomatological information and, therefore, is at best semi-quantitative in nature. Also, since physical function, the most important endpoint in patients with arthritis, is usually assessed by questionnaires [
3], disability cannot be evaluated in a similar way in experimental animal models. Therefore, tools which could address disability to allow comparison between different models or allow us to improve our understanding of the effects of therapeutic interventions would be highly valuable.
In the present study we have characterized the effects of varying degrees of severity of pristane-induced arthritis (PIA) in rats on physical function using the CatWalk method. Arthritis can be induced in susceptible rat strains such as Dark Agouti (DA) or Lewis by a single intradermal injection of the chemically inert mineral oil pristane. The induced disease closely mimics human RA as it fulfils many of the clinical criteria of RA including a symmetrical involvement of peripheral joints, the presence of rheumatoid factor and anti-RA33 antibodies [
4], destruction of cartilage and bone, and a chronic disease course [
5]. Arthritis in the DA rat develops suddenly and dramatically around two weeks after pristane-injection. An episode of severe and destructive arthritis in the peripheral joints follows and gradually subsides several weeks later. Subsequently, a chronic relapsing disease develops in most of the animals which can reach almost as high a severity as during the first arthritic episode and does not subside [
6].
The CatWalk-system was originally developed for studying rats with spinal cord injuries [
7], and is a video-based system for automated gait analysis that enables to measure a number of variables pertaining to gait pattern and weight-bearing. CatWalk has recently been used to assess gait changes caused by carrageenan-induced arthritis [
8,
9] and adjuvant-induced arthritis [
10]. Similar systems employing a treadmill instead of a flat surface for walking were applied for characterizing adjuvant arthritis in the rat and collagen-induced arthritis in the mouse [
11,
12]. Furthermore, thermal imaging of paws can provide reliable information about the degree of joint inflammation in murine and rat models of arthritis [
13,
14].
In the present study arthritic rats were followed from the preclinical and early stages of arthritis, over the maximum severity to the remission after the acute phase of PIA. It will be shown that severe arthritis dramatically affects the walking pattern and thus physical function in this experimental model. Importantly, since gait changes are detectable already before the onset of clinically visible symptoms of arthritis such as erythema and swelling, automated locomotor analysis may be of particular interest to define the earliest processes involved in a very severe and debilitating disease such as PIA.
Materials and methods
Animals
Dark Agouti (DA) rats (originating from Zentralinstitut für Versuchstierzucht, Hannover, Germany) were bred and maintained under Specific Pathogen Free (SPF) conditions in the animal facility of the Institute of Biomedical research, Medical University of Vienna, Austria. Rats were housed in groups of two to three in plastic cages with 12 h light/dark-cycles. Standard rodent chow and water was provided ad libitum. Experiments were performed on age- and sex-matched animals. All experiments were approved by the local ethical committee.
Induction and evaluation of pristane-induced arthritis
Arthritis was induced in rats at the age of 8 to 12 wk by a single intradermal injection at the base of the tail with 150 μl of pristane oil (2,6,10,4-tetramethylpentadecane; Sigma-Aldrich, St. Louis, USA). Control rats received intradermal injection of sterile phosphate-buffered saline (PBS; PAA Laboratories, Pasching, Austria). Rats were randomly allocated to pristane- or PBS-groups. Arthritis development was monitored by a blinded observer in all four limbs using a semiquantitative scoring system as previously described [
15]. Briefly, one point was given for each swollen or red finger/toe (erythema or swelling of distal or proximal interphalangeal joint or both), one point was given for each swollen or red metacarpophalangeal/meta-tarsopahalangeal joint, and one to five points were given for a swollen wrist/ankle, depending on severity (the maximum score per limb and rat was 15 and 60, respectively). The rats were scored at days 0, 6, 9, 11, 12, 14, 16, 19, 24, 29 and 35 after injection.
Assessment of pain- and arthritis-induced gait changes with the CatWalk method
The CatWalk system (Noldus Information Technology, Wageningen, The Netherlands) has been described in detail elsewhere [
7]. In short, it is a video-based system for automated gait analysis using an enclosed walkway with a glass floor where light enters along one long edge. At the placement of a paw, the light projected into the glass floor is scattered and produces an illuminated print. Using this method it is possible to measure a number of gait and weight bearing-variables. To avoid possible misinterpretation, the analysis was performed using a threshold value for skin-floor contact of 44 arbitrary units (a.u, possible range 0 to 255), that is, all pixels brighter than 44 were used.
In this study the following variables were analysed:
1.
Print area:The total floor area contacted by the paw during stance phase, that is, the area that would be blackened if the animal's paw had been painted with ink.
2.
Duration of stance phase:Since the absolute duration of stance or swing phase depends on the animal's walking speed, these variables are transformed to a fraction of total step duration according to the following formula: fraction stance phase = stance phase duration/(stance phase duration + swing phase duration), where the stance phase duration is the time of contact of one paw with the floor during a single stepcycle and the swing phase duration is the time of non-contact with the floor during the stepcycle.
3.
Regularity index:Interlimb coordination is complete if only normal step sequence patterns occur during uninterrupted locomotion (that is, when the animal places the four paws one after another). The regularity index grades the degree of interlimb coordination as follows: RI = (NSSP × 4/PP) × 100 (%) wherein NSSP represents the number of Normal Step Sequence Patterns and PP the total number of Paw Placements. In healthy animals this value is 100%.
Testing was performed between 09:00 and 16:00 h on age-matched DA. 1F rats. All animals were first allowed to habituate in and to cross the walkway before recording of runs. Testing was conducted by the same person. At each test occasion the animals were allowed to walk until a satisfactory recording was acquired. A satisfactory recording was defined as uninterrupted locomotion for a sufficiently long distance of the walkway to get at least three to four paw-placements by each non-arthritic paw. If the animal stopped to explore the environment by rearing, turned around to walk in the opposite direction or if it refused to walk far enough, the recording was repeated. A >5% reduction of print area or shortening of stance phase time versus pre-operative values was considered a significant change in locomotor behaviour.
Histological examinations
Paws from rats 11 and 35 days after injection of pristane or PBS were collected and decalcified with ethylenediaminetetraacetic acid (EDTA). Serial paraffin-embedded tissue sections of hind and front paws were analysed by H&E staining, by tartrate-resistant acid phosphatase (TRAP) staining (leukocyte acid phosphatase kit; Sigma-Aldrich) for the identification of osteoclasts, and by toluidine blue staining for visualizing cartilage. Areas of synovial inflammation, erosions of cartilage and bone, and number of osteoclasts were determined by histomorphometry using an Axioskop 2 microscope (Zeiss, Oberkochen, Germany) and the OsteoMeasure analysis software (Osteometrics, Decatur, GA, USA).
Statistical analysis
For each walk of an animal, the mean was calculated per paw for print area, stance and swing duration. Regularity index was also obtained. Mean values for all variables were then calculated for pristane- and PBS-injected rats. Comparisons between gait variables in control and pristane groups were made by ANOVA tests. Where significant effects were found, subsequent comparisons were performed using Dunnett's test. Comparisons with pre-morbid values within groups (gait variables, arthritis scores, and histological assessments) were statistically tested using paired t-tests. To analyse correlations between arthritis score area under the curve (AUC) and area under the curve of gait variables, Pearson's correlation coefficients were calculated. All data are expressed as mean ± Standard error of the mean (SEM).
Discussion
Recognition of early arthritis is one of the most important aspects in rheumatology, since it allows interference with the disease process before massive damage has occurred. On the other hand, information on the character of the initial pathologic events is limited, since the affected structures are not directly accessible in patients with very early disease and imaging techniques such as magnetic resonance imaging or sonography do not provide insights into the cellular processes involved. Therefore, animal models are very useful to dissect the very early arthritic processes in the joint. Pristane-induced arthritis (PIA) in the rat very closely mimics human RA since it fulfils many of its clinical criteria. However, PIA and other inducible animal models of arthritis have a rapid disease onset and progress quickly, making the dissection of initial phases of arthritis difficult. Indeed, once signs and symptoms of arthritis become discernible, not only full blown inflammation but also destructive events are likely to have already occurred [
16,
17]. This has been analysed by performing serial histological examinations in animals that will develop arthritis before the onset of the symptoms. However, it is difficult to envisage performing serial pre-emptive analyses in a consistent manner and, therefore, other means have to be sought. Thermal imaging was able to detect inflammatory changes already prior to onset of joint swelling in murine collagen-induced arthritis but not in rat PIA [
13,
14]. In the present study we have shown that the onset of inflammation in PIA is mirrored by changes in the animals' gait behaviour as detected by semi-automated locomotor analysis several days before clinically discernible signs and symptoms are apparent.
In a recent study of adjuvant-induced arthritis the experiments were terminated before the arthritic condition had become advanced [
10]. The current study also provides insight into processes of remission of arthritis and their consequences for locomotion. Arthritis caused by injection of pristane caused a change in several variables typical of the way rats walk, indicating an unwillingness or inability to use the affected paws. The area touching the floor of the affected paws decreased markedly, and so did the time they rested on the floor. This was apparent about 11 to 12 days after pristane-injection, with maximum effects observed around Days 20 to 24 after pristane-injection and then gradually returning towards control values. Depending on their type, abnormalities as revealed by the CatWalk system in heavily arthritic paws were seen 0.9 to 4.3 days before clinically discernible signs of arthritis. Importantly, however, histological examination revealed that already at that time and before onset of clinically apparent arthritis significant inflammatory changes occurred in the synovial membrane, supporting previous notions on an inflammatory pre-clinical phase of arthritis [
18,
19].
Interestingly, in line with observations in the pre-arthritic phase of Tumor necrosis factor (TNF) α transgenic mice [
16], in pre-clinical PIA there was not only inflammation of the synovial membrane, but particularly prominent inflammation of the tendon sheaths. Thus, these data suggest that periarticular tendonitis is perhaps the earliest event occurring in this model of RA.
It is assumed that the behavioural changes detected by the CatWalk system in the very early phase of arthritis are due to nociception, that is, detection of noxious stimuli and the subsequent transmission of encoded information to the brain [
20] rather than due to physical impairment caused by swelling and destructive events within the joint. Pain is a perceptual process that arises in response to nociception. Although rats may not process nociceptive stimuli in the same way humans do, information gained from analysis using the CatWalk system suggests that PIA in the rat can be used to model pain in humans with inflammatory joint disease. Our data suggest that inflammation-induced noxious stimuli arise earlier than clinically visible symptoms of arthritis. As pain on motion is a major aspect of complaint in arthritic patients [
21‐
23] elucidation of the processes involved in the initial phases of arthritis which may differ from those in later stages is essential [
24]. Therefore, measuring the early clinically relevant symptoms in arthritis models may allow searching for novel therapeutics to interfere not only with pain, but particularly with the earliest events of inflammation and subsequent joint destruction and to improve physical function and quality of life of patients with RA.
Conclusions
Using the PIA rat model of inflammatory polyarthritis we show that gait changes in affected paws observed by the CatWalk system, such as reduction of print area and shortening of the stance phase, precede the onset of clinically discernible arthritis symptoms and mirror histopathological changes during the course of PIA. Implementation of this technique in PIA reveals a beginning tenosynovitis in animals with changes in locomotive behaviour but without externally visible arthritis symptoms. Gait analysis can thus be used to pinpoint the exact timing for histological assessment of the initial inflammatory changes in animal models of inflammatory arthritis and also allows for comparing disease course and gait changes in different models in an objective manner.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MHH participated in the design of the study, in acquisition and analysis of data, and drafted the manuscript. RH and BN helped with data collection. HR helped to edit the manuscript. JS and GS were involved in the design of the study and helped with interpretation of data and manuscript preparation. All authors read and approved the final manuscript.