Regulatory role of heme oxygenase 1 in inflammation of rheumatoid arthritis




To examine the expression and pathogenetic roles of heme oxygenase 1 (HO-1), an inducible heme-degrading enzyme with antiinflammatory properties, in rheumatoid arthritis (RA).


HO-1 expression in synovial tissue from patients with RA, patients with osteoarthritis, and patients with noninflammatory joint diseases was determined by immunoblotting and immunohistochemistry. Effects of various agents, such as hemin (a chemical inducer of HO-1), small interfering RNA (siRNA) specific for HO-1, HO-1 expression vector, and antirheumatic agents, on HO-1 expression in RA synovial cell lines were analyzed by real-time reverse transcription–polymerase chain reaction (PCR) and immunoblotting. Cytokine synthesis was evaluated by real-time PCR and enzyme-linked immunosorbent assay.


HO-1 was expressed more abundantly in the lesions of synovial tissue from patients with RA than in those from the other patient groups. Hemin, auranofin, and HO-1 expression vector induced HO-1 and reduced expression of tumor necrosis factor α (TNFα) messenger RNA, lipopolysaccharide (LPS)–induced secretion of interleukin-6 (IL-6) and IL-8, and expression of cyclooxygenase 2 in the synovial cell lines. Treatment with HO-1–specific siRNA augmented the synthesis of TNFα, IL-6, and IL-8 and canceled the suppressive effects of auranofin on TNFα secretion. When hemoglobin, as a scavenger of carbon monoxide, was added to auranofin-treated synovial cell lines, LPS-dependent production of IL-6 and IL-8 was increased.


Our data demonstrate that HO-1 is expressed in RA synovial tissues and plays a regulatory role in the development of inflammation. The pharmacologic effects of auranofin depend, in part, on the levels of HO-1, suggesting that HO-1 induction is a novel therapeutic strategy for RA.

Rheumatoid arthritis (RA) is a systemic inflammatory disease that affects multiple joints (1, 2). The inflammatory tissues of RA patients are metabolically active, leading to increased oxygen consumption, while the oxygen supply is deficient because the intraarticular pressure exceeds the capillary perfusion pressure in the chronically inflamed joints during movement (3). Therefore, hypoxia has been implicated in the development of RA (4). Indeed, oxygen tension in the synovial fluid or synovial tissue is lower in patients with RA than in those with other pathologic conditions (4). Hypoxia-inducible factor 1α (HIF-1α), a hypoxia-inducible transcription factor, is strongly expressed in RA synovial tissue in response to the hypoxic condition, whereas none or little HIF-1α is expressed in synovial tissue from healthy individuals (5). These findings suggest that stress proteins under the control of HIF-1α are involved in the pathogenesis of RA.

In this study, we focused on heme oxygenase 1 (HO-1), the expression of which is regulated by HIF-1α (6, 7). HO-1 plays cytoprotective roles in various conditions, including inflammatory disorders (8, 9). HO-1 is an inducible isoform of HO, which degrades heme to biliverdin, Fe2+, and carbon monoxide. Biliverdin is subsequently converted to bilirubin by biliverdin reductase. Fe2+ stimulates synthesis of ferritin. Both bilirubin and ferritin act as antioxidants (10), whereas carbon monoxide suppresses apoptosis and the synthesis of inflammatory mediators such as proinflammatory cytokines, nitric oxide, and prostaglandins (11–13). Thus, multiple biochemical actions of the heme degradation products and their metabolic derivatives contribute to the cytoprotective functions of HO.

To maintain homeostasis, the human body up-regulates HO-1 in conditions such as ischemia, atherosclerosis, and inflammation (7, 14). These cytoprotective functions can be applied for therapeutic purposes. We previously reported that chemical induction of HO-1 suppresses lupus nephritis (15) and that adenovirus vector–mediated gene transfer of HO-1 complementary DNA (cDNA) has beneficial effects on lipopolysaccharide (LPS)–induced lung injury (16), influenza viral pneumonia (17), bleomycin-induced pulmonary fibrosis (18), Pseudomonas chronic respiratory infection (19), and elastase-induced emphysema (20) in mouse models. Similarly, favorable outcomes have been achieved with therapies using HO-1 chemical inducers or with gene transfer in other animal models, including ischemia/reperfusion-induced injuries of the heart (9) and liver (21), LPS-induced ocular inflammation (22), and allogenic heart allotransplantation (23).

Conversely, lack of HO-1 is associated with severe chronic inflammation, as shown in studies with HO-1–deficient mice (24) and in a patient with HO-1 deficiency (25). Interestingly, of the disease-modifying antirheumatic drugs (DMARDs), gold agents induce Nrf2-dependent HO-1 transcription in vitro (26). In contrast, up-regulation of HO-1 expression is found in rat adjuvant arthritis (27). Furthermore, we recently found that the serum level of HO-1 is excessively increased in patients with some inflammatory disorders such as adult-onset Still's disease and hemophagocytic syndrome, suggesting that HO-1 is involved in the development of systemic inflammation (28). Therefore, it is important to determine the roles of HO-1 in the inflammatory process of RA.

In the present study, we examined the expression of HO-1 in synovial tissue from patients with RA, and assessed the relationship between HO-1 expression level and function in synovial cell lines. We also assessed the pharmacologic effects of corticosteroids and DMARDs, including gold agents, on the expression of HO-1 in these cell lines.



Synovial tissue samples were obtained during arthroscopy, arthroplasty, or synovectomy from 15 patients with RA (all women, mean ± SD age 57.1 ± 15.1 years) who fulfilled the 1987 revised criteria of the American College of Rheumatology (formerly, the American Rheumatism Association) for the classification of RA (29). All patients had been treated with one or a combination of the following agents: corticosteroids (9 patients), nonsteroidal antiinflammatory drugs (7 patients), and DMARDs such as methotrexate (8 patients), sulfasalazine (4 patients), bucillamine (3 patients), and D-penicillamine (1 patient). None of the patients had received cytotoxic drugs or gold agents. A total of 17 tissue samples was obtained from the knees (7 patients), elbows (4 patients), hips (1 patient), wrists (3 patients), and feet (2 patients). Patients with other joint diseases, i.e., osteoarthritis (OA) (6 women, mean ± SD age 68.2 ± 9.4 years), osteochondritis dissecans (OCD) (a 16-year-old man), and meniscal injury (2 women, ages 62 years and 43 years), served as controls. The study was approved by the ethics committee of our institute, and all subjects in the study gave their written informed consent.

Isolation of RA synovial cells and establishment of the cell lines.

Synovial tissues were minced aseptically, and then digested enzymatically with 1.0 mg/ml collagenase (Wako, Osaka, Japan) in Dulbecco's modified Eagle's medium (DMEM) (Sigma-Aldrich, St. Louis, MO) for 2 hours at 37°C. Single-cell suspensions harvested from the tissues were filtered through a nylon mesh (0.22 μm). The cells were then plated in culture dishes containing DMEM supplemented with 100 units/ml penicillin, 0.1 mg/ml streptomycin (Sigma-Aldrich), and 10% fetal calf serum (Equitech-Bio, Kerrville, TX) and allowed to adhere to the dishes at 37°C under an atmosphere of 5% CO2. Adherent synovial cells were removed by adding trypsin-EDTA (Sigma-Aldrich). The collected synovial cells were used at the second to sixth passages for subsequent experiments. Under these conditions, immunohistochemistry using antifibroblast antigen (clone AS02; Calbiochem, San Diego, CA) indicated that more than 95% of cells were fibroblast-like synoviocytes (results not shown). Data from 6 separate experiments using distinct cell lines derived from different donors were utilized for statistical analysis.


Specimens of synovial tissues were embedded in Tissue-Tek OCT compound (Sakura Finetechnical, Tokyo, Japan) at −80°C prior to sectioning. Five-micrometer–thick sections of the specimens were cut, placed on microscope slides, and fixed in 4% paraformaldehyde (Wako). After microwave treatment for 5 minutes at 500W in 1 mmole/liter EDTA buffer at pH 8.0, the sections were incubated for 10 minutes in Tris buffered saline (TBS) containing 0.4% Triton X and then blocked with a blocking agent (Dako, Glostrup, Denmark) for 30 minutes. Thereafter, optimally diluted anti–HO-1 polyclonal antibody (1:500 dilution; Biomol, Plymouth Meeting, PA), anti-CD68 monoclonal antibody (mAb) (1:100 dilution; Dako), or antifibroblast antigen (1:20 dilution; Calbiochem) was applied to the tissue overnight at 4°C. Rabbit IgG (1:10,000 dilution) or mouse IgG1 (1:100 dilution) was used as negative control (Dako). The sections were subsequently incubated with alkaline phosphatase–labeled anti-mouse or rabbit Ig antibody (Dako) for 60 minutes. The reactivity was visualized with the new fuchsin substrate system (Dako). The sections were counterstained with hematoxylin.


Bacterial LPS (Escherichia coli serotype 0111:B4; Wako) was solubilized in deionized distilled water (DDW). Hemin (Sigma-Aldrich) was dissolved in Tris–0.1N NaOH. A solution of auranofin (Wako), dexamethasone (Sigma-Aldrich), and D-penicillamine (Sigma-Aldrich) was prepared in DMSO, ethanol, and DDW, respectively. Desferoxamine mesylate (DFO; Sigma-Aldrich) was prepared as a stock solution of 10 mM in DDW. Bilirubin (Wako) and hemoglobin (Hgb; Sigma-Aldrich) were dissolved in 0.2N NaOH, neutralized to pH 7.4, using 1N HCl and DDW, respectively. The controls for each drug were the same concentrations of solvents that were used for individual drugs.

Plasmid construct.

Full-length HO-1 cDNA was generated by reverse transcription–polymerase chain reaction (RT-PCR) with a panel of primers, consisting of sense 5′-CACCGGCCGGATGGAGCG-3′ and antisense 5′-CTGCATTCACATGGCATA-3′. The PCR products were subcloned into pcDNA3.1 (Invitrogen, Carlsbad, CA). The authenticity of HO-1 cDNA–containing plasmid was confirmed by DNA sequencing. We performed electroporation with Gene Pulser II (Bio-Rad, Richmond, CA). After the cells transfected with the HO-1 cDNA–containing plasmid or control plasmid (pcDNA3.1) were seeded into 12-well plates for 48 hours, they were exposed to 10 ng/ml LPS for 24 hours. At the end of the cultures, the culture supernatants were collected.

Small interfering RNA (siRNA) oligonucleotides.

The sequences of human HO-1–specific siRNA were determined on the basis of the AA-N19 rule (30), and were 5′-UGCUGAGUUCAUGAGGAACd(TT)-3′ (sense) and 5′-GUUCCUCAUGAACUCAGCAd(TT)-3′ (antisense). As a control siRNA, we used a corresponding nonsilencing siRNA, with the sequences 5′-UUCUCCGAACGUGUCACGUd(TT)-3′ (sense) and 5′-ACGUGACACGUUCGGAGAAd(TT)-3′ (antisense). A control siRNA-labeled fluorescein was used to determine the transfection efficacy of siRNA. All synthetic RNA oligonucleotides were synthesized and purified at Qiagen (Valencia, CA).

Transfection with siRNA duplexes in vitro.

RA synovial cells (4 × 104 cells/well) were seeded into 24-well plates for 24 hours prior to transfection with siRNA oligonucleotides. The cells had been transfected for 24 hours with 0.2 μg/well siRNA by using the RNAiFect transfection reagent according to the guidelines provided by the manufacturer. Under these conditions, we confirmed that the transfection efficacy of fluorescein-labeled siRNA oligonucleotides into the cells was more than 95%, by using fluorescence microscopic observation (Olympus, Tokyo, Japan) and FACScan analysis (Becton Dickinson, San Jose, CA) (results not shown). Thereafter, the cells were exposed to 0.1 μg/ml auranofin or 10 ng/ml LPS for 6 hours or 24 hours. At the end of the cultures, the cell lysates, messenger RNA (mRNA), and culture supernatants were collected.

Immunoblotting analysis.

Freshly isolated synovial tissues or cultured cells were treated for 30 minutes on ice with lysis buffer (137 mM NaCl, 20 mM Tris HCl, 50 mM NaF) supplemented with a protease inhibitor (Sigma-Aldrich). The supernatants were recovered by centrifugation at 15,000 revolutions per minute for 30 minutes and stored in aliquots at −80°C until used. Aliquots of the protein from each sample were boiled and resolved electrophoretically on a 4–20% gradient of polyacrylamide gel (Daiichi Kagaku, Tokyo, Japan) and transferred onto a polyvinylidene difluoride membrane. After having been blocked with 5% skim milk–TBS for 16 hours at 4°C, the membrane was incubated with optimally diluted anti–HO-1 murine mAb (1:1,000 dilution; Stressgen, Victoria, British Columbia, Canada), anti–cyclooxygenase 2 (anti–COX-2) mAb (1:1,000 dilution; BD Transduction Laboratories, Lexington, KY), or anti-actin goat polyclonal IgG (1:1,000 dilution; Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour at room temperature, and subsequently for 1 hour of incubation with horseradish peroxidase (HRP)–conjugated anti-mouse secondary antibody (1:5,000 dilution; Amersham Biosciences, Piscataway, NJ) or rabbit anti-goat IgG HRP conjugate (1:5,000 dilution; Zymed, South San Francisco, CA). The signals were developed by using the enhanced chemiluminescence detection system (Amersham Life Sciences, Little Chalfont, UK) and exposure to Kodak Biomax film for 1–5 minutes (Kodak Imaging Systems, Rochester, NY). Amounts of the blotted protein were measured densitometrically by using Scion image analysis software and an image processing software (NIH Image Engineering, Bethesda, MD).

RNA extraction, RT-PCR, and quantitative real-time RT-PCR.

Total RNA was isolated from cells by using TRIzol (Invitrogen). The cDNA was generated from the RNA with reverse transcriptase (SuperScript II; Invitrogen). For the PCR, 1 μl of cDNA was incubated with 9.375 μl DDW, 2 μl of dNTP, 2.5 μl of 10× PCR buffer, and 0.125 μl of Taq and a primer pair for HO-1 (forward 5′-CAGGCAGAGAATGCTGAG-3′, reverse 5′-GCTTCACATAGCGCTGCA-3′), GAPDH (forward 5′-ACAGTCAGCCGCATC-3′, reverse 5′-AGGTGCGGCTCCCTA-3′), and β-actin (forward 5′-TCCTGTGGCATCCACGAAACT-3′, reverse 5′-GAAGCATTTGCGGTGGACGAT-3′). Cycling conditions included 30 cycles of amplification for 30 seconds at 94°C, 30 seconds at 55°C, 1 minute at 72°C, and a final extension phase consisting of 1 cycle of 10 minutes at 72°C. Ten microliters of the PCR products and 2 μl of loading buffer were run on a 1.5% agarose gel stained with ethidium bromide.

Panels of primers of human HO-1, tumor necrosis factor α (TNFα), and GAPDH cDNA were purchased from PE Applied Biosystems (Foster City, CA). Real-time PCR was performed by using a TaqMan Universal Master Mix (PE Applied Biosystems), and the data were analyzed by the ABI Prism 7700 sequence detection system (PE Applied Biosystems). Briefly, 1:50 amounts of cDNA derived from 1 μg of total RNA, 200 nmoles/liter of probe, and 800 nmoles/liter of primers were incubated in 25 μl at 50°C for 2 minutes and 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. The analysis system determined the number of cycles at which the amplified DNA in the sample exceeded the threshold during the PCR. Standard curves of each cDNA were generated by serial dilutions of the conventional RT-PCR–amplified products. Gene expression levels of the individual samples were calculated on the basis of the standard curves. The data on HO-1 and TNFα were standardized to expression of GAPDH in the same samples.

Enzyme-linked immunosorbent assay (ELISA).

Concentrations of interleukin-6 (IL-6) and IL-8 in the culture supernatants were determined by ELISA using optimal pairs of capture and detecting biotinylated antibodies, which were purchased from R&D Systems (Minneapolis, MN).

Statistical analysis.

Comparisons of ≥3 populations were made using the Kruskal-Wallis test. Comparisons of 2 independent data sets were made using the Mann-Whitney U and chi-square tests. P values less than 0.05 were considered statistically significant.


HO-1 expression in rheumatoid synovial tissues.

Previous reports have shown that HIF-1α, one of the transcription factors of the HO-1 gene, is abundantly expressed in RA synovial tissues (5), possibly implicating HO-1 in RA lesions. We first examined the expression of HO-1 in the synovial tissue from patients with RA, patients with OA, and patients with noninflammatory joint diseases. Immunoblotting techniques revealed that substantial amounts of HO-1 were expressed in the synovial tissues from all patients with RA (Figure 1A). It has been reported that HO-1 protein is found in chondrocytes from OA patients (31). Semiquantitative analysis using a densitometer revealed that the HO-1 expression level, which was corrected for the amount of actin as an internal control, was significantly higher in RA patients than in OA patients (Figure 1B).

Figure 1.

Expression of heme oxygenase 1 (HO-1) proteins in synovial tissues from patients with rheumatoid arthritis (RA) compared with patients with osteoarthritis (OA) (n = 6 per group; all women). A and B, The HO-1 expression level was semiquantitatively determined using a densitometer. Values below each lane in A indicate the HO-1 protein level corrected to the level of actin as an internal control. Bars in B show the mean and SEM HO-1 expression level relative to actin, in individual groups. = P < 0.05 by Mann-Whitney U test. The respective age, site of joint tissue, duration of disease, and preoperation medications were as follows. RA patients: lane 1, 52 years, wrist, 7 years' duration, D-penicillamine and steroids; lane 2, 55 years, elbow, 26 years' duration, methotrexate (MTX) and nonsteroidal antiinflammatory drugs (NSAIDs); lane 3, 62 years, knee, 27 years' duration, bucillamine, steroids, and NSAIDs; lane 4, 32 years, wrist, 9 years' duration, bucillamine, steroids, and NSAIDs; lane 5, 73 years, knee, 27 years' duration, MTX; lane 6, 62 years, elbow, 29 years' duration, MTX and steroids. OA patients: lane 1, 73 years, knee, 20 years' duration, NSAIDs; lane 2, 70 years, knee, 15 years' duration, NSAIDs; lane 3, 75 years, hip, 3 years' duration, no medication; lane 4, 67 years, hip, 3 years' duration, no medication; lane 5, 50 years, hip, 18 years' duration, no medication; lane 6, 74 years, knee, 14 years' duration, no medication. C, Representative immunohistochemistry findings in synovial tissue from patients with RA and those with OA (original magnification × 100). Control tissues were obtained from a healthy part of the knee joints in patients with osteochondritis dissecans. HO-1–positive cells are stained red. D, Representative immunostaining of RA synovial tissues with mouse IgG, anti–HO-1, and anti-CD68 monoclonal antibody. HO-1– and CD68-positive cells are stained red (original magnification × 40 and × 100). No positive cells were found in the tissues stained with mouse IgG and rabbit IgG (not shown).

Histochemical studies demonstrated that HO-1–expressing cells were densely concentrated in the lining and sublining layers, especially in the parafollicular areas of RA synovial tissues (Figures 1C and D), in conjunction with predominant aggregation of CD68-positive cells (Figure 1D). A minority of the HO-1–expressing cells also showed staining with a fibroblast-specific mAb (results not shown). In contrast to the findings in RA synovial tissue, HO-1–expressing cells were sparse in the synovial tissues of OA patients and rarely detected in noninflammatory synovia obtained from patients with OCD or meniscal injury (Figure 1C). Thus, these results indicate that HO-1 expression is up-regulated in the joint lesions of RA.

Effects of HO-1–specific siRNA on proinflammatory cytokine synthesis by synovial cells.

The expression level of HO-1 in synovial cells from RA patients was reduced after in vitro culture. However, low, but significant, levels of HO-1 mRNA and HO-1 protein were detectable in established synovial cell lines by means of RT-PCR and immunoblotting techniques, respectively (Figures 2A and 3B).

Figure 2.

Effects of HO-1 small interfering RNA (siRNA) oligonucleotides on synthesis of proinflammatory cytokines. After 24 hours of resting cultures of RA synovial cell lines, the cells were transfected with 0.2 μg of HO-1–specific or control siRNA oligonucleotides by using the RNAiFect transfection reagent, and 30 hours later, mRNA was extracted from the cells. Expression of HO-1 (A) and tumor necrosis factor α (TNFα) (B) mRNA was analyzed by conventional and/or real-time reverse transcription–polymerase chain reaction. Values are the mean and SEM fold change, considering 1 as the value for control siRNA samples. Thereafter, the cells were stimulated with 10 ng/ml lipopolysaccharide for an additional 18 hours. Concentrations of interleukin-6 (IL-6) (C) and IL-8 (D) in the cell culture supernatants were determined using specific enzyme-linked immunosorbent assays. Values are the mean and SEM ng/ml. All analyses were carried out on 6 RA tissue samples. = P < 0.01 by Mann-Whitney U test. See Figure 1 for other definitions.

Figure 3.

Effects of hemin and HO-1 cDNA–containing plasmid on cytokine production by RA synovial cell lines. A, The cell lines treated with hemin were harvested at 6 hours to determine the HO-1 mRNA level by real-time reverse transcription–polymerase chain reaction. Values are the mean and SEM fold change in 6 patients with RA, considering 1 as the value for control samples. = P < 0.05 versus control samples, by Kruskal-Wallis test and Mann-Whitney U test. B, HO-1 protein levels were determined by immunoblotting after 24 hours or 48 hours of hemin or HO-1 cDNA–containing plasmid treatment, respectively. C and D, After 12-hour pretreatment with hemin (solid bars) or in the absence of hemin (open bars), the cells were stimulated with 10–1,000 ng/ml lipopolysaccharide (LPS) for another 24 hours. After 48-hour pretreatment with HO-1 cDNA–containing plasmid or control vector, the cells were stimulated with 10 ng/ml LPS for another 24 hours. The culture supernatants were harvested to measure amounts of interleukin-6 (IL-6) (C) and IL-8 (D) by enzyme-linked immunosorbent assay. Values are the mean and SEM ng/ml in 6 patients with RA. # = P < 0.01, by Mann-Whitney U test. See Figure 1 for other definitions.

To investigate the role of endogenously expressed HO-1 in RA synovial cells, we examined the effects of HO-1–specific siRNA on proinflammatory cytokine synthesis by the cell lines. Real-time PCR analysis revealed that treatment with siRNA oligonucleotides led to a 70% reduction in the HO-1 mRNA level in the synovial cell lines (Figure 2A). Under these conditions, the spontaneous mRNA expression of TNFα was significantly increased in comparison with that in the controls (Figure 2B). Furthermore, the cell lines transfected with HO-1–specific siRNA secreted larger amounts of IL-6 and IL-8, in response to LPS, than those treated with the control siRNA (Figures 2C and D). Thus, targeting the HO-1 gene augmented proinflammatory cytokine synthesis by synovial cell lines.

Effects of chemical induction or forced expression of HO-1 on proinflammatory cytokine synthesis by synovial cells.

We next examined the effects of hemin, a potent HO-1 inducer, on proinflammatory cytokine production by RA synovial cell lines. As expected, hemin treatment enhanced levels of both the mRNA and the protein of HO-1 in the RA synovial cell lines, in a dose-dependent manner (Figures 3A and B). The kinetic study revealed that the protein level reached its peak at 12–24 hours (results not shown). Because no toxic effect was seen at a range of 10– 100 μM hemin, we used 100 μM hemin in subsequent experiments. We found that 12-hour pretreatment with hemin significantly suppressed secretion of IL-6 and IL-8 by the cell lines, irrespective of the LPS concentration used to trigger the cytokine production (Figures 3C and D).

To show the direct involvement of HO-1 expression in regulation of the cytokines, HO-1 cDNA–containing plasmid was transfected into the RA synovial cell lines. The transfected cells excessively expressed HO-1 proteins (Figure 3B) and secreted significantly lower amounts of IL-6 and IL-8 in response to LPS when compared with synovial cell lines treated with a control vector (Figures 3C and D). Taken together, these results show an inverse correlation between the HO-1 expression level and proinflammatory cytokine production by synovial cell lines, suggesting a regulatory role of HO-1 in the synthesis of these proinflammatory cytokines.

Effect of DMARDs on HO-1 expression in synovial cells.

Because our data suggested that HO-1 induction suppresses inflammatory responses in RA, we sought to determine whether antirheumatic agents would induce HO-1 expression in synovial cells. We examined the influences of auranofin, D-penicillamine, and dexamethasone on HO-1 expression by RA synovial cells. Dosages of all agents were determined on the basis of their therapeutic blood concentrations (32). We found that auranofin strongly induced HO-1 expression at both the mRNA and the protein level, in a concentration-dependent manner (ranging from 0.01 μg/ml to 1.0 μg/ml) (Figures 4A and 5B). A concentration of auranofin higher than this range was too toxic to the cells, because of the presence of the DMSO vehicle. In contrast, neither D-penicillamine nor dexamethasone affected HO-1 expression at the therapeutic concentrations tested (Figures 4B and C).

Figure 4.

Effect of disease-modifying antirheumatic drugs on expression of mRNA for HO-1 in RA synovial cell lines. After the synovial cell lines had been cultured with auranofin (A), dexamethasone (B), or D-penicillamine (C) for 6 hours, the expression levels of HO-1 mRNA were analyzed by real-time reverse transcription–polymerase chain reaction. Values are the mean and SEM fold change in 6 patients with RA, considering 1 as the value for control samples. = P < 0.001 versus control samples, by Kruskal-Wallis test and Mann-Whitney U test. See Figure 1 for definitions.

Figure 5.

Effects of auranofin and HO-1 small interfering RNA (siRNA) on expression of mRNA for tumor necrosis factor α (TNFα) and protein levels of cyclooxygenase 2 (COX-2) in RA synovial cell lines. A, TNFα mRNA levels were determined in cell lines treated with auranofin. Values are the mean and SEM fold change from 6 independent experiments, considering 1 as the value for control samples. = P < 0.01 versus control samples, by Kruskal-Wallis test and Mann-Whitney U test. B, HO-1 and COX-2 protein levels were determined by immunoblotting in synovial cells after treatment with auranofin for 24 hours. C and D, Levels of HO-1 mRNA (C) and TNFα mRNA (D) were determined by real-time reverse transcription–polymerase chain reaction after pretreatment with HO-1 or control siRNA at 0.2 μg for 24 hours and culture with 0.1 μg/ml of auranofin for another 6 hours. Values are the mean and SEM fold change in 6 patients with RA, considering 1 as the value for control samples (stimulated with control siRNA). # = P < 0.001, by Kruskal-Wallis test and Mann-Whitney U test. See Figure 1 for other definitions.

Effect of HO-1–specific siRNA on the suppression of TNFα synthesis by auranofin.

We found that auranofin significantly suppressed the spontaneous TNFα mRNA expression level (Figure 5A) and COX-2 expression (Figure 5B) by the cell lines, in a dose-dependent manner. Interestingly, the expression levels of TNFα mRNA and COX-2 protein were negatively correlated with the level of HO-1 protein, suggesting that the antirheumatic properties of auranofin are dependent on HO-1 induction.

To examine the possibility that HO-1 plays a role in the pharmacologic effects of auranofin, we treated the cell lines with HO-1–specific siRNA oligonucleotides in the presence of auranofin. As expected, these oligonucleotides almost completely abrogated the HO-1–inducing effects of auranofin (Figure 5C). Furthermore, the suppression of TNFα synthesis by auranofin was reversed by the addition of the HO-1–specific siRNA (Figure 5D). These findings suggest that the pharmacologic effects of auranofin are, at least in part, dependent on HO-1 induction.

Effects of heme degradation products on proinflammatory cytokine synthesis.

As mentioned above, the antiinflammatory properties of HO-1 are known to be mediated by the heme degradation products Fe2+, carbon monoxide, and bilirubin (33, 34). To determine which product is involved in the suppression of cytokine synthesis, we examined the effects of DFO, which chelates Fe2+, Hgb (a carbon monoxide scavenger), and bilirubin on LPS-induced cytokine synthesis by the cell lines. The concentration of each agent was chosen in accordance with previously reported specifications (35, 36). We found that the suppressive effects of auranofin on secretion of IL-6 and IL-8 were canceled by the addition of Hgb, but not by the addition of DFO (Figures 6A and B). Bilirubin did not affect the production of these cytokines. In addition, Hgb enhanced LPS-induced IL-6 secretion even in the absence of auranofin. These results suggest that carbon monoxide, not Fe2+ or bilirubin, is responsible for the suppressive effects of endogenously expressed or chemically induced HO-1 on cytokine synthesis.

Figure 6.

Effects of desferoxamine mesylate (DFO), bilirubin, and hemoglobin on cytokine production by rheumatoid arthritis (RA) synovial cell lines. The cells were stimulated with 10 ng/ml lipopolysaccharide for 24 hours after 12-hour pretreatment with 100 μM DFO, 1 μM bilirubin, or 1–100 μM hemoglobin in the presence or absence of auranofin. Concentrations of A, interleukin-6 (IL-6) and B, IL-8 were determined by enzyme-linked immunosorbent assay. Values are the mean and SEM fold change in 6 patients with RA, considering 1 as the value for control samples. = P < 0.01; # = P < 0.05, by Kruskal-Wallis test and Mann-Whitney U test.


This study demonstrated that HO-1 is abundantly expressed in the joint tissues of patients with RA. Although the joint tissues of patients with OA also expressed HO-1 proteins, as was shown in previous studies (31), the expression level was much less than that in patients with RA. In contrast, very little HO-1 was detected in noninflammatory synovial tissues. HO-1 expression was also up-regulated in joint lesions in a model of rat adjuvant arthritis (27).

Several pathologic factors have been suggested to be involved in overexpression of HO-1 in RA lesions. In addition to superoxides and proinflammatory cytokines, hypoxia may play an important role in HO-1 expression in the lesions (5). Indeed, expression of HO-1 in joints is similar to that of HIF-1α. Whereas HIF-1α was aberrantly expressed in RA synovia, expression of the transcription factor was lower in OA synovia and was absent from synovia of healthy controls (5). In RA synovial tissues, the distribution of HO-1–expressing cells was similar to that of HIF-1α–expressing cells, with HO-1–expressing cells being predominantly found in the lining and sublining layer at sites of CD68-positive macrophage-like synovial cells. In addition, immunohistochemical analysis using serial sections suggested that some of the CD68-positive cells expressing HO-1 were also stained with antifibroblast antigen. Consistent with these findings, recent studies have shown that a substantial proportion of synovial cells coexpressed CD68 and fibroblast-specific markers (37).

The expression level of HO-1 was variable among the synovial tissue samples from RA patients in our study. Clinical backgrounds may affect the expression level of HO-1. However, all of the RA patients in this study had joints with progressively destructive disease, classified as stage IV of Steinbrocker's therapeutic criteria (38). The HO-1 expression level in the synovial tissues did not correlate with age, disease duration, and laboratory data, including rheumatoid factor and C-reactive protein level, in the RA patients studied.

Because we showed that a gold agent potently induced HO-1 expression in vitro in RA synovial cell lines in this study, it appears that therapeutic agents may affect the expression level of HO-1 in synovial lesions. However, the effects of gold therapy in vivo remain undetermined, because all of the patients studied had never received gold agents. Previous medications and clinical responsiveness to DMARDs did not affect local HO-1 expression in the joint lesions in our study. We have not yet specified any clinical factors that affect the HO-1 expression levels in joint lesions. The present findings suggest that HO-1 overexpression is one of the universal features of RA synovium, and it will be important to determine the clinical factors involved in up-regulation or variance of HO-1 expression in RA joints.

The majority of HO-1–expressing cells in RA joint lesions are macrophage-like synovial cells, but some of the cells also bear a fibroblast marker. RA synovial tissues contain a mixture of macrophage-like and fibroblast-like cells. During consecutive culture of the mixed cell populations in vitro, fibroblast-like cells bearing an immunologic marker of fibroblasts become predominant in the cell lines (data not shown). In spite of the phenotypic changes that occur during culture, previous studies have shown that fibroblast-like cell lines generated by procedures similar to ours maintain several important features of RA synovial cells (39–44). The cell lines secrete cytokines and growth factors, including IL-6, transforming growth factor β, and fibroblast growth factor (39), rapidly restore rheumatoid phenotypes in response to IL-1 or TNFα (40, 41), show an anchorage-independent proliferation without contact inhibition (42, 43), and express transcription factors that regulate DNA synthesis as RA synovial cells in vivo (44). Therefore, it is plausible that the regulatory and functional mechanisms of HO-1 in the fibroblastic cell lines reflect those observed in RA synovial tissues.

Our study also suggests that HO-1 plays a regulatory role in the progression of RA inflammation, since beneficial effects of HO-1 have been shown in other inflammatory conditions (8). We demonstrated herein that suppression of endogenously expressed HO-1 resulted in significant enhancement of spontaneous TNFα and LPS-induced IL-6 and IL-8 synthesis, whereas HO-1 induction led to reduced synthesis of the proinflammatory cytokines and COX-2 expression. Interestingly, auranofin is known to suppress proinflammatory cytokines (45). The suppressive effects of auranofin in the present study were partially reversed by the addition of HO-1–specific siRNA, suggesting that some antirheumatic effects of auranofin depend on HO-1 induction. HO-1 expression is also induced by simvastatin (46), which has been shown to have antirheumatic effects on arthritis (47).

The antiinflammatory effects of HO-1 depend on the biochemical properties of heme degradation products, namely, carbon monoxide, Fe2+, and bilirubin (12). In particular, carbon monoxide has been shown to be highly protective in several rodent disease models, mimicking the actions of HO-1 (48). In vitro studies have revealed that carbon monoxide modulates intracellular signal transduction systems, including the JNK signaling pathway, leading to attenuation of the activation of activator protein 1 (10), the p38 MAPK system, and the guanylyl cyclase system generating cGMP (12, 49). These biochemical effects result in reduced production of proinflammatory cytokines such as TNFα and enhanced synthesis of IL-10, an antiinflammatory cytokine (8). In concordance with these previous findings, our present data also showed that carbon monoxide generated by heme degradation was responsible for HO-1–dependent suppression of proinflammatory cytokine synthesis.

The other heme degradation product, bilirubin, which functions as a superoxide scavenger, also may contribute to HO-1–dependent suppression of synovial inflammation. In contrast, Fe2+ mediates the formation of hydroxyl radicals, which antagonize the antioxidant effects of bilirubin (50, 51).

Several investigators have pointed out that an extremely high level of HO-1 is toxic in some situations, because it generates excessive Fe2+ (51). Indeed, a recent study revealed a pathologic role of HO-1 in the progression of adjuvant arthritis in a rat model (27), in that HO-1 expression was up-regulated in the joint lesions and that treatment with an HO inhibitor, tin protoporphyrin IX, suppressed leukocyte infiltration, hyperplastic synovitis, erosion of articular cartilage and osteolysis, and the production of inflammatory mediators. In that experimental rat model, HO-1 induced vascular endothelial growth factor (VEGF), which is involved in angiogenesis in inflammatory lesions in the joint. However, we failed to show that chemical induction of HO-1 augmented VEGF synthesis in RA synovial cells in vitro (results not shown).

Although it remains to be determined what causes such a discrepancy in the roles of HO-1 between human in vitro studies using RA synovial cell lines and rat adjuvant arthritis studies, it can be speculated that the difference arises from variations in the HO-1 expression level between these 2 systems. Therefore, when considering the therapeutic application of HO-1, it is important to optimize the expression level by pharmacologic induction or gene therapy.

In summary, our data demonstrate that HO-1 expressed in RA synovial tissues plays regulatory roles in the progression of RA inflammation. Pharmacologic HO-1 induction and HO-1 gene therapy should be considered as novel therapeutic strategies for inflammatory disorders such as RA.