Rheumatology International

, Volume 27, Issue 3, pp 225–233

Alpha-lipoic acid suppresses the development of collagen-induced arthritis and protects against bone destruction in mice


  • Eun Young Lee
    • Division of Allergy and Rheumatology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
  • Chang-Keun Lee
    • Division of Allergy and Rheumatology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
  • Ki-Up Lee
    • Division of Endocrinology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
  • Joong Yeol Park
    • Division of Endocrinology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
  • Kyung-Ja Cho
    • Department of PathologyUniversity of Ulsan College of Medicine, Asan Medical Center
  • You Sook Cho
    • Division of Allergy and Rheumatology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
  • Hee Ran Lee
    • Asan Institute for Life Science
  • Se Hwan Moon
    • Asan Institute for Life Science
  • Hee-Bom Moon
    • Division of Allergy and Rheumatology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
    • Division of Allergy and Rheumatology, Department of Internal MedicineUniversity of Ulsan College of Medicine, Asan Medical Center
Original Article

DOI: 10.1007/s00296-006-0193-5

Cite this article as:
Lee, E.Y., Lee, C., Lee, K. et al. Rheumatol Int (2007) 27: 225. doi:10.1007/s00296-006-0193-5



To test the ability of alpha-lipoic acid (LA) to attenuate the development of collagen-induced arthritis (CIA) in mice.


Mice were divided into three groups and treated with intraperitoneal administration of LA (10 or 100 mg/kg) or placebo. Clinical, histologic, and biochemical parameters were assessed. Human synovial fibroblasts and peripheral blood mononuclear cells were cocultured in various concentrations of LA to evaluate the effects on osteoclastogenesis.


LA was associated with a dose-dependent reduction of CIA, as well as preventing bone erosion and destructive changes. Intracellular reactive oxygen species in lymphocytes obtained from inguinal lymph nodes, which was significantly higher in CIA than control mice, was significantly reduced in CIA by LA. The concentrations of TNF-α, IL-1β, and IL-6 in the paws, and synovial NF-κB binding, all of which were markedly higher in CIA than control mice, were reduced by treatment with LA. In addition, LA inhibited the formation of human osteoclasts in vitro.


Amelioration of joint disease by LA was associated with reduction in oxidative stress, as well as inhibition of inflammatory cytokine activation and NF-κB DNA binding activity. Moreover, LA inhibited bone destruction in vivo and osteoclastogenesis in vitro. Collectively, these results indicate that LA may be a new adjunctive therapy for rheumatoid arthritis.


Alpha-lipoic acidRheumatoid arthritisCollagen-induced arthritisReactive oxygen speciesNF-κBOsteoclast


Rheumatoid arthritis (RA) is a common disease characterized by chronic inflammation of the synovial joints and progressive destruction of articular tissues. RA is frequently combined with osteoporosis [1, 2], which further increases morbidity in these patients. Among the factors shown to be involved in the development of RA are genetic background, various immunocompetent cells, and proinflammatory cytokines. Recent studies indicated that reactive oxygen species (ROS) play an important role in the development of RA, in that in these patients the antioxidant system is impaired and peroxidation reactions are accelerated [35]. Epidemiologic studies have shown that RA occurs in previously healthy subjects who have low levels of circulating antioxidants [6], implying a pathogenic role of increased oxidative stress in the development of RA. Although the effects of antioxidants on the development of RA have been tested, they have shown limited efficacy to date [7, 8].

Alpha-lipoic acid (LA) is a cofactor for mitochondrial α-keto-dehydrogenase complexes and participates in S–O transfer reactions. LA administration to animals has been shown to protect tissues against oxidative damage. Following its administration, LA is reduced to dihydrolipoic acid (DHLA) in various tissues [9], and both LA and DHLA have been observed to act as antioxidants towards hydroxyl radicals [10, 11] and to inhibit the oxidation of lipids and proteins [12]. In addition, LA has been shown to increase microsomal protein thiols and to protect against hemolysis and neurological disorders [13, 14], and it is used clinically to treat patients with diabetic polyneuropathy [15, 16]. To determine if LA has any role in the treatment of RA, we tested its effects on joint inflammation and erosion in an animal model of RA.

Materials and methods


Male 5-week-old DBA/1 mice were purchased from Seiwa Breeding Company for Experimental Animals (Fukuoka, Japan) and housed in specific pathogen free (SPF) cages at the Asan Institute for Life Science. All experiments were approved by the Institutional Animal Care and Use Committee of the institute.

Collagen-induced arthritis (CIA)

Bovine type II collagen (CII; Sigma Chemical Co., St Louis, MO, USA), 1 mg/ml in 0.05 M acetic acid, was emulsified in an equal volume of Freund’s complete adjuvant (CFA; Sigma Chemical Co.). Arthritis was induced by injecting mice intradermally at the base of the tail with 100 μL of this emulsion on days 0 and 21. The incidence and severity of arthritis were assessed by daily physical examination. Clinical arthritis scores were evaluated using a scale of 0–3 for each paw as previously described with minor modifications, with 0 indicating normal and 1–3 indicating minimal, moderate, and severe erythema and swelling, respectively [17]. An arthritis index score (AI) for each mouse was calculated by adding the scores for individual paws.

Suppression of CIA by LA

Before the induction of arthritis, mice were randomly divided into three groups. Group 1 mice (n = 14) were each injected intraperitoneally (i.p.) with 0.1 ml 0.9% NaCl (normal saline) on the day of primary immunization and five times per week for 7 weeks. LA (Sigma) was dissolved and diluted with 0.9% NaCl to stock concentrations of 10 and 50 mg/ml and stored in the dark. Group 2 mice (n = 13) were each injected i.p. with 10 mg/kg LA, and group 3 mice (n = 16) were each injected i.p. LA (100 mg/kg) using the same schedule as above.

Histologic evaluation of CIA

Mice were killed on day 50. After fixing in 10% formalin, the paws were decalcified in hydrochloric acid, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. A 0–3 subjective grading system (0 = normal; 1 = mild; 2 = moderate; 3 = severe) was used to evaluate periarticular inflammation (infiltration of mononuclear cells and synovial thickening) and subchondral bone erosion, as described previously with minor modifications [18]. The histopathologic score of a joint refers to the sum of the subjective scores for each of these two parameters (maximum possible score of six). Histologic scoring was performed in a double-blind manner.

Measurement of cytokine levels in mouse ankles

TNF-α, IL-1β, and IL-6 were measured using commercially available enzyme-linked immunosorbent assay (ELISA) kits (BioSource International, Inc., Camarillo, CA, USA) according to the manufacturer’s instructions. Briefly, mouse ankles were snap-frozen in liquid nitrogen, ground into powder with a mortar and pestle, and dissolved in lysis buffer (10 mM HEPES, pH 7.8, 2 mM MgCl2, 15 mM KCl 0.5 mM DTT, 1 mM phenylmethylsulfonyl fluoride (PMSF), 2 μg/ml Leupeptin, and 1% Igepal), and cytokine concentrations were measured in the tissue lysates. The concentration of each was normalized to the total amount of cellular protein in each lysate by the Bradford dye-binding assay [19].

Fluorescent measurement of intracellular ROS

Formation of oxidants in lymphocytes from regional lymph nodes was determined by fluorescence over time using 2′,7′-dichlorofluorescin diacetate (DCFH-DA; Sigma Chemical Co.), a reduced, nonfluorescent derivative of fluorescein [20]. After killing, inguinal lymph nodes were extracted, meshed, and centrifuged. The lymphocytes were isolated, washed in PBS and resuspended in 300 μl PBS, and 10 μl 20 mM DCFH-DA in ethanol were added to each milliliter of cell suspension. The cells were incubated at 37°C for 20 min in the dark, kept on ice in the dark and used within 90 min. Duplicate samples were routinely analyzed. Fluorescence was assayed by flow cytometric analysis using a Becton Dickinson FACScan (Immunocytometry Systems, San Jose, CA, USA) with CellQuest software (Becton Dickinson, Mountain View, CA, USA). The log of fluorescence intensity at 570 nm of at least 3,000 cells over 5 min was measured, and flow cytometry data were expressed as median fluorescence intensity.

Preparation of nuclear extracts of mouse joints

Mouse paws were cut just above and below the ankles, the skin was removed, and the joints were then snap frozen in liquid nitrogen and stored at −70°C. Nuclear extracts were prepared as described previously, with modifications [21]. Each tissue sample was homogenized in 4 ml buffer A (10 mM HEPES, pH 7.8, 2 mM MgCl2, 15 mM KCl 0.5 mM DTT, 1 mM PMSF, 2 μg/ml Leupeptin, and 1% Igepal), incubated on ice for 15 min, and centrifuged at 5,000×g for 1 min at 4°C. The supernatant was discarded, and each pellet was resuspended in 4 ml buffer A without 1% Igepal. Following centrifugation for 5 min in a microfuge at 4°C, the supernatants were again discarded, and buffer B (20% glycerol, 20 mM HEPES, pH 7.8, 50 mM KCl, 2 mM MgCl2, 1 mM EDTA, 0.5 mM DTT, 1 mM PMSF, 2 μg/ml Leupeptin) was added to the pellets. The samples were rocked for 30 min at 4°C and centrifuged at 4°C, and the supernatants were aliquoted and stored at −80°C. Protein concentration was measured as described [19].

Electrophoretic mobility shift assay (EMSA)

The Gel Shift Assay System (Promega, Madison, WI, USA) was used as described [22]. A double-stranded NF-κB consensus oligonucleotide probe (5′-AGTTGAGGGGACTTTCCCAGGC-3′) (Promega) was end-labeled with polynucleotide kinase and γ−32P-ATP (Amersham Biosciences, Piscataway, NJ, USA). Each binding reaction contained 1 μl of the labeled oligonucleotide and 10 μg of nuclear protein. The reactions were performed at room temperature for 30 min in binding buffer [10 mM Tris–Hcl, pH 7.5, 50 mM NaCl, 1 mM MgCl2, 0.5 mM EDTA, 0.5 mM DTT, 4% glycerol (v/v), and 1 μg poly(dI–dC) (Amersham Biosciences)]. For the negative control samples, a tenfold excess of cold consensus oligonucleotide was added. To perform supershift experiments, 40 μg of antibodies to NF-κB p50 and p65 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were added 5 min before the reactions and immediately after addition of radiolabeled probe. Samples were incubated for 15 min at room temperature and loaded onto a 4% polyacrylamide gel. After electrophoresis, the gel was transferred to Whatman paper (Whatman International, Maidstone, UK), vacuum dried and visualized by autography.

Isolation of human synovial cells

Synovial tissues were obtained at the time of total knee arthroplasty from three patients who fulfilled the American College of Rheumatology criteria for RA [23]. Synovial cells were isolated as described, with modifications [24]. Briefly, the synovial tissues were washed thoroughly with RPMI 1640 (Gibco BRL, Gaithersburg, MD, USA) and minced, and the minced tissue was incubated for 90 min at 37°C in RPMI 1640 containing 1 mg/ml of collagenase (Gibco BRL) and 0.15 mg/ml of DNase I (Sigma Chemical Co.). Each mixture filtered with a 70 μm cell strainer (Becton Dickinson, Franklin Lakes, NJ, USA), and each cell suspension (3 ml) was layered onto Ficoll/Paque (Pharmacia Biotech, Uppsala, Sweden) and centrifuged at 400×g for 30 min at 20°C. Each cell pellet was resuspended in RPMI, washed three times by centrifugation at 250×g for 10 min, and suspended in a minimum essential medium (α-MEM) (Irvine Scientific, Santa Ana, CA, USA) containing 20% heat-inactivated horse serum (Gibco BRL). The cells were sequentially subcultured for 3–5 passages and used as fibroblast-like synoviocytes (FLS).

Coculture system for osteoclastogenesis

Human blood was collected from healthy volunteers, and peripheral blood mononuclear cells (PBMC) were isolated by centrifugation over Ficoll/Paque at 400×g for 30 min. Isolated PBMC (2 × 10cells/well) were resuspended in α-MEM containing 20% horse serum and 2 ng/ml of M-CSF (Sigma Chemical Co.) and seeded in 96-well culture plates (Becton Dickinson). The following day, the adherent cells were used for subsequent cocultures with FLS. FLS (2 × 104 cells/well) were added, and the cells were cocultured for 3 weeks in α-MEM containing 20% horse serum, 100 IU/ml of benzyl penicillin, 100 mg/ml of streptomycin, and 2 ng/ml of M-CSF in the presence of 10−7 M 1,25(OH)2D3. Cultures were fed by replacing half the medium every 3 days. The adherent cells were cytochemically stained for tartrate-resistant acid phosphatase (TRAP) using a commercially available kit (Sigma Chemical Co.) as described previously [25]. The number of TRAP-positive multinucleated cells containing more than three nuclei were identified as osteoclasts and counted by light microscopy. All experiments were carried out at least three times.

Statistical analysis

Statistical analyses were performed using SPSS package (version 10.0) to calculate the means and standard deviations. In the in vivo study, group means were compared by using the Kruskal–Wallis test. In the in vitro experiments, group means were compared by analysis of variance (ANOVA) with post hoc analysis by Duncan’s multiple range tests. Dose-dependent relationships were examined by Spearman’s rank correlation analysis.


Changes in arthritis index (AI)

Arthritis was first observed about 4 weeks after initial immunization of mice with type II collagen (Fig. 1). Although the onset of clinical manifestations was not significantly delayed by LA administration, LA significantly inhibited disease severity compared with placebo-treated mice (P < 0.05). In the group treated with high dose (100 mg/kg) LA, there was a marked reduction of AI compared with the placebo-treated group (P < 0.001 on day 43).
Fig. 1

Changes in the arthritis index (AI) in mice treated with LA or placebo. Mice with CIA were treated daily with 10 mg/kg i.p. LA (n = 13), 100 mg/kg i.p. LA (n = 16), or placebo (n = 14). The AI was calculated at the sum of the measurements for the four paws of each animal. Curves were established for each group of animals. Each value is reported as the mean ± standard errors. Administration of LA significantly reduced AI compared with placebo-treated mice (*< 0.05)

Histologic findings

LA treatment markedly improved the semiquantitative histologic arthritis score in the mice with CIA (Fig. 2, P < 0.05). The histologic score was lowest in mice treated with high dose LA, and no erosion or destructive changes in bone was observed in any of the 11 mice in this group.
Fig. 2

Histologic scores in mice treated with LA or placebo. Mice with CIA were treated daily with 10 mg/kg i.p. LA (n = 9), 100 mg/kg i.p. LA (n = 11), or placebo (n = 11), and the histologic score was determined as described in Materials and methods. Data are expressed as mean ± SD. The histologic scores in LA-treated mice were significantly lower than in placebo-treated mice (P < 0.05)

In contrast to normal mice (Fig. 3a), joint sections from placebo-treated CIA mice showed severe inflammatory cell infiltrates and bone destruction (Fig. 3b). In contrast, 10 mg/kg LA-treated mice had mild inflammatory cell infiltrates and synovial thickening (Fig. 3c), whereas 100 mg/kg LA-treated mice had only minimal synovial proliferation (Fig. 3d).
Fig. 3

Representative histologic findings in joint sections from each group of mice. a Joint of a normal DBA/1 mouse on day 50. b Ankle of a placebo-treated mouse on day 50. Synovial tissue is infiltrated by numerous inflammatory cells and bony destructive change is observed. c Ankle of a 10 mg/kg LA-treated mouse after killing. Mild inflammatory cell infiltrates are present in the joint space and synovial tissue. d Ankle of a 100 mg/kg LA-treated mouse after killing. Only minimal synovial proliferation is observed

Levels of cytokines in mouse joints

Consistent with macroscopic scoring, treatment of mice with LA induced significant, dose-dependent decrease in the concentrations of TNF-α (Fig. 4a, = 0.042), IL-1β (Fig. 4b, P = 0.027), and IL-6 (Fig. 4c, P = 0.04) in their paws.
Fig. 4

Effects of LA on cytokine concentrations in CIA mouse joints. Tissue lysates were prepared from ankle joints, and cytokine concentrations were measured by ELISA. Data are expressed as mean ± SD. a TNF-α concentrations in LA-treated mice (10 mg/kg, n = 9; 100 mg/kg, n = 10) were significantly lower than in placebo-treated mice (n = 12). b IL-1β concentrations in LA-treated mice (10 mg/kg, n = 8; 100 mg/kg, n = 9) were significantly lower than in placebo-treated mice (n = 8). c IL-6 levels in LA-treated mice (10 mg/kg, n = 5; 100 mg/kg, n = 5) were significantly lower than in placebo-treated mice (n = 7)

Redox status of the mice

We also measured fluorescence of intracellular ROS by DCF in the three groups of mice, in which a shift to the right indicates increased intracellular oxidation (Fig. 5a). When we plotted median fluorescence intensity in these three groups, we found that administration of LA markedly reduced intracellular oxidation of lymphocytes from inguinal lymph nodes in mice with CIA (Fig. 5b, P = 0.025).
Fig. 5

Effect of LA on intracellular ROS formation. a Representative flow cytometry histogram showing the fluorescence signal generated by lymphocytes loaded with the redox-sensitive dye 2′,7′-DCFH in the three groups of mice. The median fluorescence intensity was 8.28 in the placebo group, 4.49 in the 10 mg/kg LA group, and 3.40 in the 100 mg/kg LA group. b ROS formation in lymphocytes from inguinal lymph nodes in LA-treated CIA mice (10 mg/kg, n = 7; 100 mg/kg, n = 5) was significantly lower than in placebo-treated mice (n = 10). Data are expressed as mean ± SD

DNA binding activity of NF-κB in CIA

LA treatment mice induced a dose-dependent reduction in DNA binding activity (Fig. 6). To confirm that this was due to specific DNA binding, we performed a competitive experiment using unlabeled NF-κB oligonucleotides (cold oligos, lane 1, Fig. 6). Supershift experiments showed that the DNA protein complex appeared to contain both p50 and p65.
Fig. 6

NF-κB binding activity in synovial tissue of mice with CIA treated with LA or placebo. EMSA results of nuclear extracts of synovial samples using a radiolabeled oligonucleotide containing an NF-κB consensus binding site are shown. The first lane shows competition with cold oligos. In supershift assays, anti-p50 Ab and anti-p65 Ab were present in the sample. LA treatment was associated with a dose-dependent reduction of NF-κB activity

A semiquantitative analysis of the intensity of DNA binding activity by EMSA in the three groups of mice showed that LA administration significantly reduced NF-κB DNA binding activity in CIA compared with placebo-treated mice (P = 0.001, Table 1). In the CIA mice treated with 100 mg/kg LA, NF-κB DNA binding activity was significantly lower than in the group treated with 10 mg/kg LA (P = 0.002 by Mann–Whitney U test).
Table 1

NF-κB DNA binding activities of synovial tissues in CIA mice


Placebo (n = 6)

10 mg/kg ALA (n = 6)

100 mg/kg ALA (n = 6)


NF-κB DNA binding activity (AU)

121.5 ± 8.6

90.5 ± 12.1

56.5 ± 15.3


Nuclear extracts were prepared from CIA mouse ankles after killing. Synovial NF-κB DNA binding activity was determined by EMSA as described in Materials and methods. Band intensity was determined by computer assisted image analysis. Data are expressed as mean ± SD

Inhibition of osteoclast formation

When we added LA, at concentrations ranging from 0.01 to 1 mM, to cocultures of human PBMC and FLS, we found that LA concentrations ≥0.25 mM significantly inhibited osteoclast formation in dose-dependent manner (Fig. 7, rho = −0.882, P < 0.01). To show that this effect was not due to cytotoxicity of LA, we measured its effect on the viability of separate cultures for FLS and PBMC, using XTT based cell proliferation assay. We found that, compared with control, each LA concentration did not affect the proliferation of PBMC and FLS after 1 and 3 weeks (data not shown).
Fig. 7

Effect of LA on in vitro osteoclast formation. FLS and PBMC were co-cultured in the presence of various concentrations of LA. After 3 weeks, the TRAP-positive multinucleated cells were counted by light microscopy. Data are expressed as mean ± SD. LA, at concentrations of 0.25, 0.05, and 1 mM significantly inhibited osteoclast formation


In this study, we assayed the effects of LA on the clinical, histologic, and biochemical parameters of mice with CIA, an animal model of RA. In this model, mice are immunized with collagen and treated with i.p. injection of LA, at doses of 10 and 100 mg/kg five times per week, with the higher dose in mice equivalent to an oral dose of 12 mg/kg in adult humans (with formula guided by FDA). We found that LA treatment reduced the severity of clinical arthritis and prevented joint destruction, as well as inhibiting synovial NF-κB activation. We also found that LA treatment reduced regional oxidative stress and decreased cytokine production in mouse joints.

LA is known to have potent antioxidant activity. In a model of streptozocin-induced diabetic neuropathy, LA was shown to reduce oxidative stress and improve distal nerve conduction [26]. More recently, in a mouse model of experimental autoimmune encephalitis (EAE), LA was found to inhibit T cell migration into the spinal cord by acting as an inhibitor of matrix metalloproteinases (MMPs), thereby suppressing disease activity [27]. To our knowledge, the work presented here is the first reporting the effects of LA on chronic inflammatory arthritis, especially RA.

Among the transcription factors activated in the presence of excessive ROS, NF-κB appears to be especially important in joint inflammation. Production of IL-1 and TNF-α by synovial macrophages is regulated by NF-κB, as is the expression of TNF-α and IL-6 in FLS [2832]. Increased cytokine production driven by NF-κB can enhance expression of vascular adhesion molecules that attract leukocytes into the joint, as well as the MMPs that help degrade the extracellular matrix. It is thought, therefore, that an inhibitor of NF-κB would be useful in the treatment of RA [22]. Our demonstration of the suppressive effect of LA on synovial NF-κB binding activity in inflamed joints is consistent with this hypothesis. In addition, the tissue concentrations of TNF-α, IL-1β, and IL-6, key cytokines in joint inflammation in RA [33], were significantly reduced by LA treatment in CIA mice. Blocking these cytokines has been used in the treatment of RA [3436].

We also found that LA prevents joint destruction, in that bone erosion was completely prevented in CIA mice treated with 100 mg/kg/day LA. To establish the mechanism of LA-induced prevention on bone erosion, we evaluated in vitro osteoclastogenesis in the presence or absence of LA. Administration of LA inhibited osteoclast formation in a dose dependent manner, an effect not due to nonspecific toxicity. Although, we did not fully investigate the mechanism by which LA inhibits osteoclastogenesis, the protective effect of LA on joint destruction may be due to its reduction of regional cytokine production (e.g. IL-1β) and inhibition of osteoclast formation [37, 38].

Although the role of antioxidants in the treatment of chronic inflammatory disease has not yet been determined, recombinant superoxide dismutase (SOD) protein has been shown to have some effectiveness in CIA. The relatively short half-life of this protein in the circulation, however, limits its clinical application to RA [39, 40]. Administration of vitamin E prevented articular destruction in an animal model of RA, but vitamin E did not change the inflammatory components of the disease (including TNF-α level and AI) or the oxidation status of the animals [8]. In humans, vitamin E had only analgesic activity [7]. Our study, which shows that LA treatment was associated with significant improvements in regional oxidative stress, inflammatory cytokines, and NF-κB DNA binding activity in a mouse model of RA, suggests that this drug may have potential therapeutic use in the treatment humans with RA.

In conclusion, we have shown here that ROS is important in an animal model of RA and that LA is effective in treating CIA. LA reduced clinical arthritis score and histologic severity of the disease. Amelioration of joint disease was associated with reduction in oxidative stress, as well as inhibition of inflammatory cytokine activation and NF-κB DNA binding activity. Moreover, LA inhibited bone destruction in vivo and osteoclastogenesis in vitro. These findings support the concept that oxidative stress is important in the pathogenesis of RA, and suggest the potential use of LA as a therapeutic agent in human RA.


This work was supported by institutional grant from Asan Institute for Life Science (2003-330), Seoul, Korea. We thank S.K. Park and S.K. Lee (Division of Nephrology, Department of Internal Medicine, Asan Medical Center, Seoul, Korea) for the technical support on EMSA. We are grateful to J.R. Kim (Asan Institute for Life Science, Seoul, Korea) for animal care.

Copyright information

© Springer-Verlag 2006