By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Please be advised that we experienced an unexpected issue that occurred on Saturday and Sunday January 20th and 21st that caused the site to be down for an extended period of time and affected the ability of users to access content on Wiley Online Library. This issue has now been fully resolved. We apologize for any inconvenience this may have caused and are working to ensure that we can alert you immediately of any unplanned periods of downtime or disruption in the future.
J. A. Jazayeri, School of Biomedical Sciences, Charles Sturt University, Boorooma Street, PO Box 588, Wagga Wagga, NSW 2678, Australia. E-mail: firstname.lastname@example.org
This study is investigating the role of leukaemia inhibitory factor (LIF) in the development of inflammation and joint damage in the mouse K/B×N serum transfer arthritis model. LIF knock-out (LIF−/−) mice were generated by mating heterozygote females (LIF+/−) with heterozygote males. Arthritis was induced in 8–20-week-old LIF knock-out mice (LIF−/−) by intraperitoneal injection of pooled K/B×N sera (50 µl) on days 0 and 2. Clinical disease was scored daily for 6 days. Safranin-O and haematoxylin-stained sections were scored for synovitis, joint space exudate, cartilage degradation and bone damage. RNA was extracted from ankle joints and used to investigate gene expression levels of tumour necrosis factor (TNF)-α, interleukin (IL)-1, LIF, LIF receptor, oncostatin M (OSM), OSM receptor, IL-6 and their common receptor subunit gp130 by quantitative reverse transcription–polymerase chain reaction (qRT–PCR). The results show that wild-type mice developed severe clinically overt polyarthritis. In contrast, LIF−/− mice showed a more than 50% reduction in clinical arthritis severity. Significantly lower histological scores were observed in LIF−/− mice compared to wild-type disease controls. LIF−/− mice had histopathological scores that were similar to normal healthy mice. IL-6 subfamily cytokine and receptor subunit expression remained unchanged. The expression levels for IL-6 were reduced significantly in all the diseased mice, whether wild-type or LIF−/− mice (P < 0·001), compared to healthy wild-type mice. We conclude that LIF contributes to the development of disease in the K/B×N serum transfer model of arthritis. These results provide further evidence for the role of LIF in inflammation and cartilage bone resorption and provide impetus to test the effects of LIF blockade as a therapeutic strategy in rheumatoid arthritis.
Cytokines have been implicated strongly in the pathogenesis of rheumatoid arthritis (RA), and improved disease outcomes have been observed with the introduction and subsequent widespread usage of cytokine antagonists and cytokine modulators . Cellular immune mechanisms are also important in disease development. Diverse strategies utilizing synthetic and biological disease-modifying anti-rheumatic drugs (DMARDs) to target specific cytokines and cellular immune mechanisms have proved recently to be therapeutically successful.
In addition to interleukin (IL)-1, tumour necrosis factor (TNF)-α and IL-6, whose proinflammatory roles are well established in RA [1,2], a large number of other cytokines have been detected in rheumatoid synovium and are believed to contribute to disease pathogenesis. These include interferon (IFN)-γ, leukaemia inhibitory factor (LIF) and oncostatin M (OSM). Elevated quantities of these cytokines have been shown to be present in the rheumatoid joint [3–7]. LIF, OSM and IL-6 act directly on chondrocytes to stimulate proteoglycan (PG) and collagen resorption and to inhibit PG and collagen synthesis in vivo. In RA, cytokines such as IL-1 and TNF-α contribute significantly to chondral resorption, although not all the observed cartilage catabolic activity in synovial fluid can be accounted for by the activities of these two cytokines alone . In-vitro and in-vivo studies, together with experiments performed with murine LIF binding protein, have implicated LIF in joint inflammation and chondral damage [8–11]. Despite these findings, the role of LIF in RA remains controversial .
LIF is a pleiotropic cytokine which belongs to the IL-6 cytokine subfamily. It plays a pivotal role in several physiological systems including proliferation, differentiation and cell survival, blastocyst implantation and the mediation of inflammation. LIF shares the same receptor complex with other IL-6 subfamily cytokines. The receptor complex is comprised of a LIF receptor (LIF-R) subunit and a signal transduction element, notably glycoprotein 130 (gp130), which is common to all the receptors engaged by the IL-6 cytokine subfamily . LIF binds to the LIF-receptor (LIF-R) α-chain and the LIF/LIF-R complex is then able to bind to gp130 to form a LIF–LIF-R–gp130 signalling complex . Conversely, OSM, which is structurally similar to LIF, binds to gp130 and signals through either a gp130–LIF-R or gp130 and OSM receptor (OSM-R) heterodimer . LIF, OSM and other IL-6 subfamily cytokines, including IL-6 itself, are expressed in the cartilage and synovial membrane and also regulate matrix metalloproteinase (MMP) expression in the joint . Using both human and bovine chondrocytes, OSM stimulation has been shown to increase most of the major MMPs implicated in cartilage degradation, in particular MMP-1, 3 and 13 . In addition, Lotz et al. have shown that LIF is expressed in cartilage and synovium and can contribute to the pathogenesis of arthritis .
Among the animal models of arthritis, the K/B×N serum transfer model is a well-established and widely accepted model of inflammatory arthritis [17,18]. It is characterized by symmetrical involvement of peripheral joints, pannus formation, synovial hyperplasia and bone and cartilage degradation. In these respects, and histologically, it closely resembles RA. In its early phase of pathogenesis the model is dependent on the functions of T and B cells specific to the self-antigen glucose-6-phosphate isomerase (GPI). However, the disease can be induced by transfer of anti-GPI immunoglobulin (Ig)Gs from K/B×N transgenic mice with spontaneous arthritis into lymphocyte-deficient mice, thereby implicating anti-GPI autoantibodies as the main arthritogenic agent in the K/B×N serum transfer model .
We have shown previously that LIF is found in excess in the synovial fluids of RA patients and that it can stimulate proteoglycan release and contribute to cartilage degradation in vitro[10,11,19]. Varghese et al. have shown that LIF and OSM stimulate collagenase-3 expression in osteoblasts . To our knowledge, there are relatively few in-vivo studies that address the role of LIF in the pathogenesis of RA and dissect the immunopathology of the disease process. In this study, we sought to examine the role of LIF in the pathogenesis of inflammation in K/B×N serum transfer arthritis by induction of disease in LIF-deficient (LIF−/−) and LIF-sufficient (LIF+/+) mice. The extent and severity of inflammation and joint tissue damage was determined subsequently using routine parameter measurements.
Materials and methods
All experiments involving mice were approved by the Faculty of Pharmacy and Pharmaceutical Sciences Animal Ethics Committee, Monash University. Mice, strain C57BL/6, were obtained from the Howard Flory Institute of Physiology and Medicine (Melbourne, Australia) and were used to generate the LIF−/− genotype and as wild-type disease controls.
Generation of LIF−/− mice
LIF significantly enhances blastocyst formation and deficiency in LIF renders the females infertile. Therefore, for breeding purposes, LIF−/− were generated by mating heterozygote females (LIF+/−) with heterozygote males. For screening purposes, polymerase chain reaction (PCR)-quality genomic DNA was prepared from the tail biopsies following the HotShot method described by Truett et al. . Two separate PCR reactions were performed. The first PCR was a multiplex, and was carried out for amplifying two fragments; a 400 base pairs (bp) fragment of β-galactosidase (LacZ) and a generic 200 bp fragment of control DNA. The control DNA was present in all the genomic DNA samples, irrespective of their genotype. Hence, the control DNA was amplified in the multiplex PCR in order to confirm the presence of the DNA template in each reaction mixture. The second PCR was carried out to amplify a 200 bp fragment of the LIF gene. The results were compiled together to decide the genotype.
K/B×N serum transfer arthritis model
Arthritis was induced in 8–20-week-old healthy male mice (n = 9 per group of disease-positive control and diseased LIF−/−) by injection of 50 µl of K/B×N serum intraperitoneally (i.p.) on days 0 and 2. Normal healthy C57BL/6 male mice (n = 9) were used to measure the changes attributable to normal growth. All mice were monitored daily for changes in body weight and clinical disease severity, which was scored 0 (no signs of inflammation) to 5 (severe redness and swelling), as described by Santos et al. . The maximum score per mouse was thus 20. Hindpaw and ankle joint thickness were measured by digital microcallipers (Mitutoyo, Kawasaki-shi, Japan) on days 0 and 6 and the results were expressed as the difference between day 0 and day 6 measurements in mm (Fig. 1c,d). The experiments were repeated in two independent batches.
The overall extent of illness of the groups was evaluated by measuring the area under the curve (AUC). The curve refers here to the line graph plotted between the clinical scores on the y-axis over the course of the study, i.e. time measured in days on the x-axis (Fig. 1a). The AUC is a more accurate measure of the difference in clinical score between groups than the total score of each group, which is calculated by summing all the scores obtained on the clinical scale over the time-frame being analysed . This is because AUC can be calculated even when a few random data points are missing.
Resected ankle joint tissues from healthy control, diseased control and LIF−/− mice (n = 9/group) were examined by histopathology. Briefly, joint tissues were fixed in formalin and decalcified in 15% ethylenediamine tetraacetic acid (EDTA)–Tris buffer (BDH Chemicals, Sydney, Australia). Six-µm-thick sagittal knee sections were cut, stained with Safranin-O and then counterstained with fast green/iron haematoxylin (ICN Biomedicals, Aurora, OH, USA). Histopathological scoring was performed as described by Santos et al. . Each section was scored from 0 to 3 with respect to four different aspects of inflammation, namely: synovitis (hypercellularity of the synovial membrane including pannus formation), joint space exudate (determined by the presence of leucocytes, discretely or in clusters in the joint space) and cartilage degradation (loss of Safranin-O staining in the articular cartilage, with a score of 0 for completely stained cartilage and a score of 3 for completely unstained tissue). Sections were also scored for bone damage (estimated extent of subchondral bone loss). Both the clinical and histological scorings were performed by a scorer who was blinded to the mouse genotype.
Extraction of RNA from the whole joint
The ankle joint was pulverized by placing it between two metallic surfaces, cooled previously in liquid N2, and hitting the top with a hammer. Samples were not allowed to thaw at any stage of pulverization. Total RNA was extracted from pulverized ankle joint tissue using the RNeasy plus Mini Kit (Qiagen, Doncaster, Vic, Australia), according to the manufacturer's protocol. The concentration of total RNA was determined by measuring absorbance at 260 nm (1 unit at 260 nm = 40 µg RNA/ml). The 260 : 280 ratios for all samples were greater than 1 : 8. Samples were stored at −80°C until use.
Quantitative real-time PCR
Real-time PCR was performed using SYBR Green Mastermix (Applied Biosystems, Sydney, Australia) on a Rotor-GeneTM 3000 (Corbett Research, Sydney, Australia). The real-time PCR was carried out in a total volume of 12·5 µl and contained 6·25 µl of SYBR Green master mix [including the buffer, deoxyribonucleotide triphosphate (dNTPs), DNA polymerase, MgCl2], 0·8 µM of each primer, 2·25 µl PCR-grade water and 3 µl of cDNA. All samples were profiled for the expression levels of LIF, LIF-R, OSM, OSM-R, IL-6, gp130, IL-1α and TNF-α in triplicate, under the following PCR conditions: 95°C 10 min; then [95°C 25 s, 60°C 30 s, 72°C 20 s] for 40 cycles (see Table 1 for primers). To determine the efficiency of the PCR reaction, a pool of two to three randomly chosen cDNA samples was diluted serially twofold and run in real time to construct a calibration curve plotting log of the template amount on the x-axis and the generated Ct value on the y-axis. The amplification efficiencies of all the primers were calculated to be similar to that of the reference gene, with any difference being less than 5%. The results were analysed using the ΔΔCT method with the expression of the target genes normalized with β-actin as the housekeeping gene.
Table 1. Primers used in genotyping and real-time polymerase chain reaction experiments.
Amplicon size (bp)
Gp130: glycoprotein 130; bp: base pairs; IL: interleukin; LIF: leukaemia inhibitory factor; LIF-R: leukaemia inhibitory factor receptor; OSM: oncostatin M; OSM-R: oncostatin M receptor; TNF: tumour necrosis factor.
Used as a positive control
The significant differences in the samples were evaluated using either the Kruskal–Wallis test for analysing clinical scores, histopathological scores and AUC values or one-way analysis of variance (anova) for evaluating parameters such as change in ankle joint and hindpaw thickness and ankle joint thickness, and the real-time PCR data. The Kruskal–Wallis test was followed by Dunn's multiple-comparison post-test, whereas one-way anova was followed by Bonferroni multiple-comparison post-test to compare all the groups with each other (GraphPad Prism version 5).
Generation of LIF−/− knock-out mice
The generation of LIF−/− genotype was confirmed by PCR reactions in which the presence of a 400 bp LacZ amplicon (in multiplex PCR) and the absence of the LIF 200 bp gene product (in single-plex PCR) were indicative of the LIF−/− mouse genotype. In addition, as for control, a 200 bp fragment was amplified and detected in the multiplex PCR reactions using LIF+/+, LIF+/− and LIF/– DNA samples as templates. LIF−/− mice strains thus generated were healthy and active with no sign of inflammation in any of their joints. LIF−/− mice originated from LIF+/− heterozygotes mice with no known defects in innate immunity. However, LIF−/− mice were found to have a slightly lower body weight (16–18 g) compared to the wild-type offspring used in the experiment (24–26 g).
K/B×N serum-transfer arthritis is delayed and less severe in LIF−/− mice
Daily scoring of arthritis severity showed a progressive incremental increase in clinical scores for disease wild-type control. The clinical score was based on the extent of visible inflammation in distal joints and paws, particularly the extent of swelling and redness in the affected areas. LIF−/− mice developed a delayed and less severe form of arthritis (AUC value = 7·694) compared to the wild-type mice (AUC value = 26·92) (Fig. 1a). Scores for normal healthy wild-type mice are shown for comparison (AUC value = 0·8889) (Fig. 1a). On plotting the AUC values for each group, a statistically significant difference was found between the healthy wild-type and diseased wild-type mice (P < 0·001) (Fig. 1b). Also, the AUC scores were significantly lower for the diseased LIF−/− mice compared to the diseased wild-type (P < 0·05) (Fig. 1b). Similarly, hindpaw and ankle joint thickness measurements were substantially lower at day 6 in LIF−/− mice compared to wild-type mice (Fig. 1c, P < 0·01 and Fig. 1d, P < 0·05 respectively).
With respect to parameters of both clinical and histopathological inflammation, it is apparent that LIF−/− mice had a much less severe arthritis compared to wild-type control mice (Fig. 2). K/B×N serum-treated wild-type mice (positive control) exhibited significant paw inflammation associated with severe histopathological synovitis [Fig. 2b(ii) and b(i)]. In contrast, in LIF−/− mice, paw inflammation was mild [Fig. 2a(ii)] and resembled that of normal healthy mice [Fig. 2c(ii)]. In addition, LIF−/− mice had minimal histopathological manifestations of arthritis [Fig. 2a(i)]. Normal healthy mouse histology shows no evidence of synovitis and joint exudate as well as intact cartilage [Fig. 2c(i)].
Resected ankle joints were also scored for extent of synovitis, joint space exudate, cartilage destruction and bone damage (Fig. 2d). The histological scores for all four components of inflammation were appreciably and statistically significantly higher in wild-type mice compared to LIF−/− mice (P < 0·001, Fig. 2d). Scores in the LIF-deficient mice were similar to those in healthy wild-type controls.
Quantitative real-time PCR
Gene expression levels for some of the IL-6 subfamily cytokines and their receptor subunits were examined. These include LIF, LIF-R, OSM, OSM-R, gp130 and IL-6. In addition, and for comparison, the expression levels of two well-established proinflammatory cytokines, notably IL-1 and TNF-α, were also determined. Total RNA from the ankle joints was isolated and subjected to quantitative reverse transcription–PCR (qRT–PCR) in triplicate. The expression levels of LIF, LIF-R, OSM, OSM-R, gp130, IL-1 and TNF-α, although varied, did not differ significantly between the three groups of mice (Fig. 3a–f,h). However, a marked and statistically significant (P < 0·001) reduction in IL-6 gene expression was observed in the diseased LIF knock-out mice and disease control group mice in comparison to healthy wild-type mice (P < 0·001, Fig. 3g).
The role of LIF in models of inflammatory arthritis and patients with these diseases remains a matter of conjecture. Some experimental studies have suggested that LIF has the potential to promote or amplify inflammation and tissue damage, whereas others have shown the opposite. The presence of increased concentrations of LIF in RA synovial fluid (RASF) and the pro-catabolic effect of LIF on cartilage resorption, together with an attenuating effect of LIF binding protein with respect to RASF-stimulated cartilage degradation, supports the possibility that LIF contributes to tissue damage in RA . At the same time, the observation that LIF-deficient mice manifest more severe arthritis in a complete Freund's adjuvant (CFA) model suggests otherwise . A close approximation to the cellular and structural pathology of rheumatoid arthritis (RA) is achieved in the K/B×N arthritis model. Other key features of RA, such as autoantibody production, polyclonal B cell activation and hypergammaglobulaemia, have also been observed in K/B×N arthritis . In the light of these attributes, in this study the K/B×N arthritis model for RA was induced in wild-type and LIF knock-out mice in order to investigate further the role of LIF in inflammatory arthritis.
The induction of K/B×N serum-transfer arthritis in LIF-sufficient wild-type mice resulted in the rapid development of joint disease, characterized by severe synovitis, considerable joint space exudate and marked structural joint damage with compelling histopathological evidence of chondral and bone resorption. Although LIF-deficient (LIF−/−) mice were also susceptible to arthritis induction, disease development was delayed, progressed more slowly and was considerably less severe on day 6 after induction. The clinically apparent attenuation of disease severity in LIF−/− mice was corroborated by the histological findings. The better outcome with respect to the preservation of joint structure in LIF−/− mice is not surprising, as in previous studies we and others have shown that LIF induces cartilage and bone resorption in pig, goat and human cartilage and in mouse calvariae [9–11,19,24].
We used qRT–PCR to profile the expression levels of not only LIF and its receptor, but also other members of the IL-6 subfamily cytokines, notably IL-6 and OSM, together with their common receptor subunit gp130, and also TNF-α and IL-1. These were performed in resected ankle joints from healthy wild-type, diseased control and diseased LIF knock-out mice. The results show that the levels of all cytokines and receptors profiled remained unchanged except for IL-6. In both the LIF-sufficient (disease control) and LIF-deficient (knock-out) mice with K/B×N arthritis, there was a statistically significant reduction in IL-6 gene expression compared to control mice without arthritis. The reason for this is not clear. Our findings with respect to the mRNA concentrations of IL-6 subfamily cytokines and their receptor subunits do not explain the diminished disease in LIF-deficient mice. IL-6 has been shown to exert a co-factor role, but not a pivotal or key orchestrating role in the pathogenesis of K/B×N arthritis. However, IL-6−/− mice have been found to be protected against joint inflammation and destruction in collagen-induced arthritis (CIA) as well as antigen-induced arthritis (AIA) [25,26]. Importantly, this protection was observed despite the fact that the levels of TNF-α and IL-1 remained unchanged in these models. Further experiments are needed to clarify the discrepancy in IL-6 gene expression in both LIF-sufficient and -deficient mice.
The possibility that the less severe disease observed in LIF−/− mice may be a consequence of the delayed development of arthritis in these animals cannot be discounted. For ethical reasons we were unable to extend the end date of the experiments to determine if LIF−/− mice might have developed a severe form of arthritis over a longer period of time. However, the significant differences observed up to day 6 suggest strongly that LIF has an important role in the induction and development of K/B×N arthritis. It should be noted that in contrast to the collagen-induced model of arthritis (CIA), TNF-α and IL-1 inhibitors do not attenuate disease severity to any great degree in the K/B×N mouse model . Thus, the K/B×N model may be useful for exploring alternative pathogenetic mechanisms that may have applicability to those RA patients who fail to respond to TNF blockade either primarily or secondarily. Further experiments using LIF and other IL-6 cytokine subfamily antagonists may be useful to elucidate more fully the exact contribution of LIF in the K/B×N mouse model and, by implication, in human RA.
Not all studies support a protagonist action for LIF in inflammatory arthritis. Some studies in vitro have indicated that LIF may inhibit rather than promote MMP activity. For example, the observations reported by Banner et al. in a CFA model of arthritis indicate that LIF knock-out mice develop more severe disease, suggesting that LIF exerts a predominantly anti-inflammatory action in this model . Indeed, in LIF-deficient mice with CFA-induced arthritis, the disease severity of the arthritis was enhanced in spatial extent, amplitude, cellular infiltrate and IL-1β production. Moreover, local injection of LIF into the hindpaw was found to diminish the pain response and attenuate IL-1 secretion caused by CFA. Similarly, Zhu et al.  observed suppression of cutaneous inflammation that had been induced by local injection of an adenoviral vector encoding LIF, suggesting an anti-inflammatory role for LIF in some skin disorders.
In conclusion, the data reported in this study demonstrate clearly that LIF contributes to the induction and progression of inflammatory arthritis and joint tissue damage in the K/B×N serum transfer arthritis model. They confirm and extend previous observations concerning the role of LIF in inflammation and cartilage and bone resorption in inflammatory arthritis and provide further impetus to test the effects of LIF blockade as a potential therapeutic strategy in RA in general and in the subset of patients who fail to respond to TNF inhibitors in particular.
This work was funded by generous contributions from the L. E. W. Carty Foundation, the Helen Macpherson Smith Trust and the CASS Foundation (Melbourne, Victoria). We also wish to thank Professor Colin Pouton for his continuous support and guidance during the course of this project.