Objectives Increasing evidence suggests that oxidative stress may play a key role in joint destruction due to rheumatoid arthritis (RA). The aim of this study was to elucidate the role of nuclear factor erythroid 2-related factor 2 (Nrf2), a transcription factor that maintains the cellular defence against oxidative stress, in RA.
Methods The activation status of Nrf2 was assessed in synovial tissue from patients with RA using immunohistochemistry. Antibody-induced arthritis (AIA) was induced in Nrf2-knockout and Nrf2-wild-type control mice. The severity of cartilage destruction was evaluated using a damage score. The extent of oxidative stress, the activation state of Nrf2 and the expression level of Nrf2 target genes were analysed by immunhistological staining. The expression of vascular endothelial growth factor (VEGF)-A was examined on mRNA and protein using the Luminex technique. A Xenogen imaging system was used to measure Nrf2 activity in an antioxidant response element-luciferase transgenic mouse during AIA.
Results Nrf2 was activated in the joints of arthritic mice and of patients with RA. Nrf2-knockout mice had more severe cartilage injuries and more oxidative damage, and the expression of Nrf2 target genes was enhanced in Nrf2-wild-type but not in knockout mice during AIA. Both VEGF-A mRNA and protein expression was upregulated in Nrf2-knockout mice during AIA. An unexpected finding was the number of spontaneously fractured bones in Nrf2-knockout mice with AIA.
Conclusion These results provide strong evidence that oxidative stress is significantly involved in cartilage degradation in experimental arthritis, and indicate that the presence of a functional Nrf2 gene is a major requirement for limiting cartilage destruction.
Statistics from Altmetric.com
An increase in reactive oxygen species (ROS) plays an important role in the pathogenesis of rheumatoid arthritis (RA),1 2 and antioxidants and antioxidative enzymes have been shown to reduce cartilage damage in animal models of RA.3 4
Over the course of evolution, cells have developed complex mechanisms to defend against oxidative stress. A battery of genes encoding detoxifying and antioxidative enzymes is orchestrated upon exposure to ROS. This coordinated response is regulated through a cis-acting element, the antioxidant response element (ARE), within the regulatory regions of these ‘safeguard’ genes. Activation of nuclear factor erythroid 2-related factor 2 (Nrf2) and the resultant binding to the ARE initiate or enhance the transcription of genes coding for antioxidative enzymes.5,–,8
Although oxidative stress is thought to be involved in the pathogenesis of RA, there is no conclusive experimental evidence supporting the idea that defective antioxidant responses in joints lead to increased cartilage destruction. To test the hypothesis that deficiency in Nrf2-mediated antioxidant defences plays a central role in the pathogenesis of RA, we studied a mouse model of RA in wild-type and in Nrf2-deficient mice.
Details regarding materials, animals, histological examination of arthritis damage, human synovial membrane tissues, immunohistochemical protocols, in vivo Nrf2 activity imaging, quantitation of protein and mRNA concentration and statistical analysis are shown in the online supplement.
Antibody-induced arthritis in mice
Arthritis was induced using an arthritogenic monoclonal antibody cocktail purchased from MD-Bioscience (Zurich, Switzerland) according to the manufacturer's protocol. Details of the protocol are given in the online supplement.
Nrf2 is activated in the synovial membrane of patients with RA and mice with antibody-induced arthritis
Joint tissues from healthy and wild-type mice with antibody-induced arthritis (AIA) were examined for expression of Nrf2 using anti-Nrf2 antibody immunohistochemistry. Nrf2 expression was upregulated in the synovial cells and subintimal adipocytes from mice with AIA (figure 1B) but was not those from untreated mice (figure 1A). A score of positive immunostaining for Nrf2 in control mice and those with AIA is shown in figure 1C. The protein expression of Nrf2 was considerably enhanced in synovial cells of Nrf2-wild-type mice with AIA (figure 1C, right graph) compared with controls (figure 1C, left graph). To confirm these results in humans, we also examined the synovial membranes of patients with RA and of healthy donors for expression of Nrf2. As in mice, Nrf2 expression was upregulated in human arthritis synovial cells and in subintimal adipocytes (figure 1E), but not in those of healthy donors (figure 1D). A positive immunostaining score for Nrf2 in healthy subjects and patients with RA is shown in figure 1F. As in mice, the protein expression of Nrf2 was considerably enhanced in synovial cells of patients with RA (figure 1F, right graph) compared with healthy donors (figure 1F, left graph).
Nrf2 is activated in the paws of mice with AIA
ARE-luciferase mice were used to measure the activity of Nrf2 in the hind paws during AIA. Untreated ARE-luciferase mice at age 6–8 weeks showed no luciferase activity. Ten days after antibody injection, the luciferase activity detected in the hind paws of the AIA mice rose significantly compared with controls (day 0, before treatment; figure 1G). This onset of activity corresponded with the incidence of the AIA described by the manufacturer of the AIA model. The luciferase activity increased steadily over time until day 19, when it reached a plateau with no significant changes until the end of the experiment (day 24, figure 1H). Interestingly, this AIA-induced ARE activation appears to be similar to that seen in the NF-κB-luciferase mice in the same arthritis model.9 Luciferase activity was detectable in the forepaws, tail and ears as well as in the hind paws. The time-dependent progression of the measurements is shown in figure 1H.
Haeme oxygenase-1, γ-glutamyl cysteine synthetase and thioredoxin is expressed in the joints of mice after induction of arthritis
Joint tissue from wild-type and Nrf2-knockout mice with and without AIA was examined for expression of haeme oxygenase-1 (HO-1; figure 2A) and γ-glutamyl cysteine synthetase (γ-GCS; figure 2C) using specific antibodies. A positive immunostaining score for HO-1 and γ-GCS is shown in figure 2B,D. The protein expression of both HO-1 and γ-GCS is considerably enhanced in the synovial cells of Nrf2-wild-type mice with AIA (figure 2A,C, bottom left; B and D third bar) compared with controls (figure 2A,C, top left; figure 2B,D first bar). In contrast, the protein upregulation of Nrf2 target genes was not observed in Nrf2-knockout mice with AIA (figure 2A,C, bottom right; figure 2B,D fourth bar). HO-1 and γ-GCS were also upregulated in adipocytes of the subintima (figure 2A,B). Furthermore, thioredoxin upregulation in wild-type mice with AIA was also shown using a luminex QuantiGene multiplex assay (figure 2E).
Destruction of articular cartilage is intensified in Nrf2-knockout mice with AIA
To mimic the clinical scenario of RA, mice were subjected to AIA. AIA developed rapidly in the mice and clinical signs (periarticular erythema and oedema) of the disease appeared in their hind paws with 100% incidence of AIA by day 21. There was no macroscopic evidence of hind paw erythema or oedema in the sham groups (data not shown). The rate and the absolute gain in body weight in the normal mice and the AIA mice were comparable for the first week (data not shown). There was no evidence of pathology in the control wild-type and Nrf2-knockout mice (figure 3A, top left and right). Histological evaluation of the joints at day 21 showed signs of severe arthritis including pannus formation and hyperplasia which was more pronounced in the joints of Nrf2-knockout mice (figure 3A, bottom right) than in AIA wild-type mice (3A, bottom left). Pannus formation, hyperplasia, erosion of cartilage and infiltration of the knee joint were scored and the level of cartilage damage in the different treatment regimes is shown in figure 3B–E. The scoring criteria are shown in table S1 in the online supplement. Nrf2-wild-type and Nrf2-knockout mice without AIA had no or very weak pannus formation, hyperplasia, erosion of cartilage and infiltration of the knee joint (figure 3B–E, first two columns). Both Nrf2-wild-type and Nrf2-knockout mice had cartilage damage as a result of AIA (figure 3B–E, last two columns), but the Nrf2-knockout mice had significantly more hyperplasia (figure 3B, p=0.04), pannus formation (figure 3C, p=0.036), erosion of cartilage (figure 3D, p=0.048) and infiltration (figure 3E, p=0.032).
Nrf2-knockout mice develop excessive articular cartilage damage during AIA
Safranin-O staining of cartilage of Nrf2-wild-type and Nrf2-knockout mice was of similar intensity (figure 4A, upper pictures). Safranin-O staining of the articular cartilage in AIA wild-type mice was reduced compared with healthy mice (figure 4A, bottom left), which indicates a loss of proteoglycan from the cartilage matrix. By contrast, reduction of safranin-O staining of AIA Nrf2-knockout mice cartilage was significantly more pronounced than that of AIA Nrf2-wild-type mice (figure 4A, bottom right). Scoring of the colour intensity is shown in figure 4B.
Articular cartilage of Nrf2-knockout mice has increased oxidative damage during AIA
One consequence of oxidative stress is membrane lipid peroxidation. We therefore stained slices of mouse hind knees by immunohistochemistry using an antibody against 4-hydroxy-nonenal (HNE) and evaluated the colour intensity by score ranking. HNE staining of cartilage and the synovial membrane of Nrf2-wild-type and Nrf2-knockout mice without AIA had no positive cells (figure 4C,E, upper images and figure 4D,F, first two columns). In Nrf2-wild-type mice, AIA caused lipid peroxidation in cartilage (figure 4C, bottom left and figure 4D, third column) and the synovial membrane (figure 4E, bottom left and figure 4F, third column). Genetic disruption of Nrf2 increased the macromolecular oxidative damage during the inflammatory response in the cartilage (figure 4C, bottom right and figure 4D, fourth column) and in the synovial membrane (figure 4E, bottom right and figure 4F, fourth column).
Nrf2-knockout provokes spontaneous fractures during AIA
An unexpected finding was the number of spontaneously fractured bones in the AIA Nrf2-knockout mice. A fibrous callus is shown in figure 5A. These spontaneous fractures were seen in 70% of the AIA Nrf2-knockout mice (figure 5B).
Level of VEGF-A mRNA and VEGF-A protein in mice with and without AIA
The luminex technique was used to measure vascular endothelial growth factor (VEGF)-A mRNA and protein in knee joints of mice with and without AIA. mRNA expression of VEGF-A was increased in both Nrf2-wild-type mice (1.7-fold vs wild-type control) and Nrf2-knockout mice arthritic joints (1.5-fold vs wild-type control, figure 6A). The protein concentration of VEGF-A was increased in the joints of AIA Nrf2-knockout mice compared with wild-type control mice, but not in the AIA Nrf2-wild-type mice (figure 6B).
We have shown that the oxidative stress responsive transcription factor Nrf2 is increased in the joints of patients with RA. The increase in Nrf2, which occurs in response to oxidative stress within RA joints, is not sufficient to fully counteract the progression of the disease. We hypothesise, however, that a deficiency in the Nrf2-mediated antioxidant defences plays a central role in the pathogenesis of RA.
We used an anticollagen monoclonal antibody/lipopolysaccharide-induced arthritis model for the present study. There was a significantly higher expression of Nrf2 in the joints of arthritic mice (figure 1A–C) and in those of patients with RA compared with the control groups (figure 1D–F). Since unstressed cells have a very low steady state level of Nrf2 protein, these results indicate that an oxidative stress-induced activation of Nrf2 occurs during RA. In agreement, xenogen imaging using arthritic ARE-luciferase mice showed increasing Nrf2 activity over time during the progression of arthritis (figure 1G,H). We next examined the expression of the Nrf2 target genes HO-1, γ-GCS and thioredoxin which are known to be upregulated under arthritic conditions and to be protective against joint destruction.10,–,12 As expected, we were able to confirm this upregulation in Nrf2-wild-type mice with AIA. On the other hand, Nrf2-knockout mice are unable to upregulate these genes after AIA induction. In mice without AIA a weak staining of HO-1 and GCS was detectable, which might be due to the fact that these genes have activating transcription factors other than Nrf2 (figure 2A–E). The positive staining of Nrf2 target genes such as HO-1 and γ-GCS in subintimal adiposites shown in figure 2A,C is congruent with the positive staining for Nrf2 in figure 1B,E. This might lead to the conclusion that the subintima is also under oxidative stress. This suggests that oxidative stress is a factor contributing to RA, and that Nrf2 could play an important role in alleviating its effects.
Consistent with this, Nrf2 deficiency significantly enhanced the severity of arthritis in our RA model. The difference between Nrf2-knockout mice and Nrf2-wild-type mice with regard to the level of joint destruction was clear on histological sections stained with H&E and safranin-O (figure 3 and 4A,B). We evaluated hyperplasia, pannus formation, erosion of cartilage and infiltration in the joint and found a significantly higher rate of destruction in all four categories in the Nrf2-knockout mice compared with wild-type mice (figure 3). We hypothesise that an increased oxidative burden resulting from defective oxidative defence may lead to the increase in cartilage destruction in arthritic Nrf2-knockout mice. To test our hypothesis, we measured the oxidative stress in the cartilage of mice with AIA. Because of the highly reactive nature of ROS, it is difficult directly to demonstrate their presence in vivo. It is more appropriate to measure the ‘footprints’ of ROS—for example, their effects on various lipids, proteins and nucleic acids. Accordingly, evidence for oxidative stress in RA has in many cases been generated by approaches that detect oxidant-induced changes to these molecules.1 We measured the amount of lipid peroxidation by immunohistochemistry using an antibody against HNE. HNE is a specific product of lipid peroxidation that occurs in response to oxidative stress, including RA and osteoarthritis.13 14 The results showed significantly more oxidised lipids in the cartilage of Nrf2-knockout mice with AIA than in wild-type mice of the same group (figure 4C–F). Further studies are needed to ascertain clearly whether the higher rate of cartilage destruction results from more inflammation, less oxidative stress defence, or both.
An unexpected finding was the number of spontaneously fractured bones in the Nrf2-knockout mice with AIA (figure 5A,B). Hinoi and colleagues reported a negative effect of Nrf2 on osteoblast and chondrocyte differentiation. They showed an Nrf2-dependent inhibition of the runt-related transcription factor 2-dependent osteocalcin promoter activity and decreased mRNA expression of several chondrocyte differentiation markers such as type II collagen, type X collagen and osteopontin.15 16 However, further research is required to elucidate a possible function of Nrf2 in bone formation.
Recently, Biniecka and colleagues reported that increased oxidative stress correlates with the level of hypoxia and VEGF expression in the joints of patients with RA.13 Since Nrf2 is the major regulator of the oxidative stress response, we examined VEGF-A expression on the mRNA and protein levels in our mouse model of RA. In both Nrf2-knockout and Nrf2-wild-type mice, VEGF-A mRNA expression was upregulated during AIA (figure 6A). Surprisingly, the VEGF-A protein level was upregulated only in Nrf2-knockout mice with AIA but not in wild-type mice of the same group (figure 6B). We therefore assume that Nrf2 might have an inhibitory effect on VEGF-A mRNA translation and might negatively upregulate the mechanism that regulates VEGF-A protein synthesis.17
Our data show for the first time that mice that have undergone targeted deletion of the Nrf2 gene are significantly more vulnerable to the pathological joint changes associated with AIA than wild-type controls. We provide evidence that the higher levels of cartilage degradation is due to the inability of Nrf2-knockout mice to upregulate endogenous antioxidants. These results suggest that the presence of a functional Nrf2 gene is a major requirement for limiting cartilage destruction during AIA. Our data provide strong evidence that oxidative stress is significantly involved in the cartilage degradation activated in experimental arthritis and that Nrf2 is a critical regulator of the oxidative stress defence. This is the first study to suggest a transcriptional regulatory mechanism of cellular antioxidant defence in experimental RA.
Finally, the activities of Nrf2 make it a promising candidate for the development of novel therapies for treating RA. Since a recent report has shown that anti-tumour necrosis factor (TNF)-α therapy does not suppress ROS production in patients with RA, we would expect Nrf2 activation, alone or in combination with anti-TNFα, to have a more beneficial effect than anti-TNFα alone.18
The authors thank Susanne Echterhagen, Inka Geurink, Christiane Jaeschke, Ursula Mundt, Michaela Nicolau, Marie Pradella, Angela Rüben, Sonja Seiter and Lian Shen for their excellent technical assistance.
CJW, AF, AG, DV, SL and TP contributed equally to the work.
Funding This research project received funding from the ‘Deutsche Forschungsgemeinschaft’ Pu 214/3-2, Pu 214/4-2 and Pu 214/5-2 and SFB617 and was also supported by the START-Program of the Faculty of Medicine, RWTH Aachen University.
Patient consent Obtained.
Ethics approval The study was approved by the institutional review board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.