Systemic sclerosis (SSc) is a connective tissue disorder of great clinical heterogeneity. Its pathophysiology remains unclear. Our aim was to evaluate the relative roles of reactive oxygen species (ROS) and of the immune system using an original model of SSc. BALB/c and immunodeficient BALB/c SCID mice were injected s.c. with prooxidative agents (hydroxyl radicals, hypochlorous acid, peroxynitrites, superoxide anions), bleomycin, or PBS everyday for 6 wk. Skin and lung fibrosis were assessed by histological and biochemical methods. Autoantibodies were detected by ELISA. The effects of mouse sera on H2O2 production by endothelial cells and on fibroblast proliferation, and serum concentrations in advanced oxidation protein products (AOPP) were compared with sera from patients with limited or diffuse SSc. We observed that s.c. peroxynitrites induced skin fibrosis and serum anti-CENP-B Abs that characterize limited SSc, whereas hypochlorite or hydroxyl radicals induced cutaneous and lung fibrosis and anti-DNA topoisomerase 1 autoantibodies that characterize human diffuse SSc. Sera from hypochlorite- or hydroxyl radical-treated mice and of patients with diffuse SSc contained high levels of AOPP that triggered endothelial production of H2O2 and fibroblast hyperproliferation. Oxidized topoisomerase 1 recapitulated the effects of whole serum AOPP. SCID mice developed an attenuated form of SSc, demonstrating the synergistic role of the immune system with AOPP in disease propagation. We demonstrate a direct role for ROS in SSc and show that the nature of the ROS dictates the form of SSc. Moreover, this demonstration is the first that shows the specific oxidation of an autoantigen directly participates in the pathogenesis of an autoimmune disease.
Systemic sclerosis (SSc)4 is a connective tissue disorder of unknown etiology characterized by vascular hyperreactivity, fibrosis of skin and visceral organs, and immunological alterations, including a distinct pattern of autoantibodies in the sera (1). The mechanisms that determine the clinical manifestations of the disease remain unclear (2). Several reports have suggested that reactive oxygen species (ROS) are involved in the pathogenesis of SSc (3, 4, 5, 6, 7, 8, 9). Indeed, skin fibroblasts from SSc patients spontaneously produce large amounts of ROS that trigger collagen synthesis (7, 10). In addition, autoantibodies found in SSc patients against the platelet-derived growth factor receptor expressed on fibroblasts also induce the production of ROS (11). Recently, we have demonstrated that sera from patients with SSc could induce not only the production of ROS by endothelial cells but also the hyperproliferation of fibroblasts (6).
However, no direct proof for the involvement of oxidative stress in SSc pathogenesis has been brought forth to date. Moreover, the origin and nature of the oxidative stress remain to be elucidated. Environmental factors, and in particular silica dust, may be involved, which generate hydroxyl radicals (OH·) (12). Iterative ischemia-reperfusion that occurs frequently in SSc patients before the development of fibrosis may also be a source of superoxide anions (O2) through the activation of the xanthine/xanthine oxidase pathway (5).
To better understand the role of ROS in the fibrotic, vascular, and autoimmune processes that characterize SSc, s.c. injections of various types of ROS-inducing agents were performed on the backs of normal and SCID mice. Four different types of ROS-generating substances were used to reproduce the vast majority of oxidative stresses that could occur in humans: hypochlorous acid (HOCl) produced during the neutrophil burst; OH· that mainly originate from hydrogen peroxide (H2O2) subsequent to the Fenton reaction in macrophages; O2), the major ROS produced during ischemia-reperfusion injuries by endothelial cells; and peroxynitrites (ONOO−) generated by the combination of O2) and NO during reperfusion of ischemic tissues and inflammatory process. This procedure allowed us to directly demonstrate the involvement of ROS in the induction of the disease, to precisely characterize the nature and origin of ROS involved in the induction of diffuse or limited SSc, to study the mechanism leading to local or systemic involvement, and to determine the role of the immune system in the disease process.
Materials and Methods
Animals, cells, and chemicals
Specific pathogen-free, 6-wk-old female BALB/c, DBA/1, (NZB x NZW)F1, and BALB/c SCID mice were purchased from Harlan Sprague-Dawley and maintained with food and water ad libitum. All mice were housed in autoclaved cages with sterile food and water. They were given humane care according to the guidelines of our institution. HUVECs were obtained by digestion of umbilical cords with 0.1% collagenase. NIH 3T3 fibroblasts were obtained from American Type Culture Collection. HEp-2 cells were obtained from EuroBio. All cells were cultured as reported previously (13). All chemicals were from Sigma-Aldrich except H2-DCFDA (2′,7′-dichlorodihydrofluorescein diacetate; Molecular Probes), bleomycin (Bellon Laboratories), human IgG (IVIg, Tegeline; LFB), anti-DNA topoisomerase 1 Abs (Santa Cruz Biotechnology), and anti-B220-PE, anti-CD11b-FITC, anti-CD4-allophycocyanin Cy7, anti-CD8-PE Cy7 Abs, and anti-CD3 mAb (BD Pharmingen).
Patients and healthy controls
A total of 20 patients with SSc and 20 healthy controls were enrolled in the study. All participants gave written informed consent. SSc was defined according to LeRoy and Medsger criteria and the American Rheumatism Association (1, 14, 15). Among the 20 patients with SSc, 10 had limited cutaneous SSc with no interstitial fibrosis and 10 had diffuse cutaneous SSc with interstitial lung fibrosis and a total lung capacity (TLC) of lower than 75% of predicted values. Limited cutaneous SSc was defined by skin thickening only in areas distal to the elbows and knees, and diffuse cutaneous SSc was defined by the presence of proximal, as well as distal skin thickening, to the elbows and knees (15). Interstitial lung disease was assessed in all patients by chest high-resolution computed tomodensitometry and pulmonary function test values. After collection and centrifugation, serum aliquots of 1-ml serum were stored at −80°C until use.
ROS preparation and injections
The 6-wk-old female BALB/c, DBA/1, (NZB x NZW)F1, and BALB/c SCID mice were randomly distributed into experimental and control groups (n = 10 per group). A total of 100 μl of solution generating ROS (HOCl, O2, ONOO−, or OH·) or bleomycin was injected s.c. into the shaved back of the mice, using a 27-gauge needle, everyday for 6 wk. All agents were prepared extemporaneously. Control groups received injections of 100 μl of sterilized PBS.
Generation of HOCl.
HOCl was produced by adding 166 μl of NaClO solution (2.6% as active chlorine) to 11.1 ml of KH2PO4 solution (100 mM (pH 7.2)) (16). HOCl concentration was determined by spectrophotometry at 292 nm (molar absorption coefficient = 350 M−1 cm−1).
Generation of O2.
Generation of ONOO−.
A 0.6 M H2O2 solution was prepared by mixing 8.8 M H2O2 with 0.7 M HCl. Five milliliters of the solution obtained were mixed with 5 ml of NaNO2 (0.6 M) on ice. Then 10 ml of 1.2 M NaOH were added. H2O2 in excess was removed by adding MnO2. The concentration of ONOO− in the solution was determined at 300 nm (molar absorption coefficient = 1670 M−1 cm−1) (19, 20).
Generation of OH·.
Preparation of bleomycin.
A solution of bleomycin (100 μg/ml) was prepared in PBS.
Two weeks after the end of the injections, the animals were sacrificed by cervical dislocation. Serum samples were collected from each mouse and stored at −80°C until use. Lungs and kidneys were removed from each mouse and a skin biopsy was performed on the back region, involving the skin and the underlying muscle of the injected area. Samples were stored at −80°C for determination of collagen content or fixed in 10% neutral buffered formalin for histopathological analysis.
Treatment with N-acetylcysteine (NAC)
BALB/c mice were injected s.c. with a solution generating HOCl everyday for 6 wk, as described, and simultaneously i.p. treated with either NAC (150 mg/kg per injection diluted in PBS; Bristol-Myers Squibb) (n = 10) or PBS (n = 10) three times per week for the same 6 wk. A control group (n = 10 mice) received both s.c. and i.p. injections of sterilized PBS. The experimental procedure described was applied.
Fixed lung, kidney, and skin samples were embedded in paraffin. A 5-μm thick tissue section was prepared from the midportion of paraffin-embedded tissue and stained with H&E. Slides were examined by standard brightfield microscopy (Olympus BX60) by a pathologist who was blinded to the animal’s group assignment. Dermal thickness was measured under a light microscope of stained sections.
Collagen content in skin and lung
Collagen content assay was based on the quantitative dye-binding Sircol method (Biocolor) (23). Skin taken from the site of injection and lung pieces from each mouse were diced using a sharp scalpel, put into aseptic tubes, thawed and mixed with pepsin (1:10 weight ratio) and 0.5 M acetic acid. Collagen extraction was performed overnight at room temperature under stirring. The solution was then centrifuged at 20,000 × g for 20 min at 4°C and 50 μl of each sample were added to 1.0 ml of Syrius red reagent. Tubes were rocked at room temperature for 30 min and centrifuged at 20,000 × g for 20 min. The supernatants were discarded and 1.0 ml of the 0.5 M NaOH was added to the collagen-dye pellets. The concentration values were read at 540 nm on a microplate reader (Fusion; PerkinElmer) vs a standard range of bovine collagen type I concentrations (supplied as a sterile solution in 0.5 M acetic acid).
Isolation of fibroblasts from the skin of mice and proliferation assays
Skin fragments from the back of mice submitted to HOCl (n = 7), bleomycin (n = 7), or PBS (n = 5) injections for 6 wk were collected 1 wk after the last injection. Skin samples were digested with Liver Digest Medium (Invitrogen) for 1 h at 37°C. After three washes in complete medium, cells were seeded into sterile flasks. Isolated fibroblasts were cultured in DMEM/Glutamax-I supplemented with 10% heat-inactivated FCS and antibiotics at 37°C in humidified atmosphere with 5% CO2. For proliferation assay, primary fibroblasts (2 × 103 per well) were seeded in 96-well plates and incubated with 150 μl of culture medium with 10% FCS at 37°C in 5% CO2 for 48 h. Cell proliferation was determined by pulsing the cells with [3H]thymidine (1 μCi/well) during the last 16 h of culture. Results were expressed as an absolute number of cells in cpm.
Immunohistochemistry of lung tissue section
Frozen lung tissue sections of 5-μm thickness were fixed in acetone for 5 min at −20°C, washed with PBS, and blocked with 2% normal mouse serum for 30 min at room temperature. Slides were then stained for 1 h at room temperature with a 1/50 dilution of FITC-labeled rat anti-mouse B220 (clone RA3-6B2; BD Pharmingen), with a 1/50 dilution PE-labeled hamster anti-mouse CD3 (clone 145-2C11; BD Pharmingen), or with the respective FITC- and PE-labeled control isotype (BD Pharmingen). After extensive washing in PBS, slides were observed with an Olympus microscope equipped with an epifluorescence system and pictures taken at magnification ×400 with a digital camera.
Serum autoantibodies detection
Levels of anti-DNA topoisomerase 1 IgG, anti-CENP-B IgG, anti-dsDNA IgG, anti-cardiolipin IgG Abs, anti-annexin V IgG Abs and of IgM rheumatoid factors were measured using standard ELISA in mouse serum diluted 1/50. Serum IgG reactivities were also analyzed by immunoblotting technique with whole cell HEp-2 and HUVEC protein extracts for mice and patients (serum diluted 1/50 in both cases). Levels of anti-DNA topoisomerase 1 IgG Abs were detected by ELISA using purified calf thymus DNA topoisomerase 1 bound to the wells of a polystyrene microwell plate (Inova Diagnostics). Levels of anti-CENP-B IgG Abs were measured by ELISA using microplates coated with full-length recombinant CENP-B (Pharmacia Diagnostics). Levels of anti-dsDNA IgG, anti-cardiolipin IgG Abs, anti-annexin V IgG Abs, and IgM rheumatoid factor were measured using standard ELISA (24). Calf thymus DNA (5 μg/ml), cardiolipin (50 μg/ml), annexin V (2 μg/ml), or i.v. human IgG (10 μg/ml, Tegeline; LFB) were coated onto ELISA plates (precoated with protamine sulfate when dsDNA was used as substrate) overnight at 4°C. All plates were blocked with PBS-1% BSA and washed, and 100 μl of 1/50 mouse serum were added and allowed to react for 1 h at room temperature. After five washes, bound Abs were detected with alkaline phosphatase-conjugated goat anti-mouse IgG or IgM Abs, and the reaction was developed by adding p-nitrophenyl phosphate. Optical density was measured at 405 nm using a Dynatech MR 5000 microplate reader (Dynex Technology) and the optical density in blank wells (no Ag coated) was subtracted.
HEp-2 cells and HUVEC were harvested in the presence of EDTA. Whole cell protein extracts were prepared in 125 mM Tris-HCl (pH 6.8) with 4% SDS, 1.45 M 2-ME, and 1 μg/ml each of aprotinin, pepstatin, and leupeptin on ice and sonicated four times during 30 s. Nuclear protein extracts were prepared as previously described (25). Briefly, cells were incubated in 10 mM HEPES, 1.5 mM MgCl2, 0.5 mM DTT for 10 min on ice and lysed with 10 strokes of Dounce homogenizer. The homogenate was centrifuged for 20 min at 25,000 × g. The pellets of nuclei were resuspended in 3 ml of buffer with 20 mM HEPES, 25% glycerol, 0.42 M NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.5 mM PMSF, 0.5 mM DTT and lysed with 10 strokes of Dounce homogenizer. The suspension was then centrifuged for 30 min at 25,000 × g, and the supernatant was dialyzed. The dialysate was finally centrifuged, and the supernatant used for quantification. Equal amounts of loading buffer with solubilized proteins (140 μl/gel) were subjected to 10% SDS-PAGE, transferred onto nitrocellulose membranes, and incubated for 4 h at room temperature with 1/50 mouse sera using a Cassette Miniblot System (Immunetics) (26). The membranes were then extensively washed and incubated with alkaline phosphatase-conjugated goat anti-mouse IgG Ab. Immunoreactivity was revealed with NBT/5-bromo-4-chloro-3-indolyl phosphate.
Flow cytometric analysis of spleen cell subsets
Spleen cell suspensions were prepared after hypotonic lysis of erythrocytes. Cells were incubated with the appropriate labeled Ab at 4°C for 45 min in PBS with 0.1% sodium azide and 5% normal rat serum to block nonspecific binding. Cell suspensions were then subjected to four-color analysis on a FACSCanto flow cytometer (BD Biosciences). The mAbs used in this study were anti-B220-PE, anti-CD11b-FITC, anti-CD4- allophycocyanin Cy7, and anti-CD8-PE Cy7 Abs.
In vitro spleen cell proliferation
Spleen cells were isolated by gentle disruption of spleens and erythrocytes lysed by hypotonic shock in potassium acetate solution. Spleen cells were cultured in RPMI 1640 supplemented with antibiotics, Glutamax (Invitrogen Life Technologies), and 10% heat-inactivated FCS (Invitrogen Life Technologies) as complete medium. The proliferation assay was conducted in 96-well flat-bottom plates. Briefly, spleen cell suspensions (2 × 105 cells) were cultured in complete medium for 48 h in the presence of 10 μg/ml LPS (Boehringer Mannheim) or with precoated anti-CD3 mAb (2.5 μg/ml). Cell proliferation was determined by pulsing the cells with [3H]thymidine (1 μCi/well) during the last 16 h of culture and measuring the radioactivity incorporated by liquid scintillation counting.
Total serum IgG and IgM Ab concentrations
Levels of total mouse IgG and IgM Abs were measured using standard ELISA.
Endothelial cells (8 × 103 per well) were seeded in 96-well plates (Costar; Corning) and incubated with their respective growth medium alone for 12 h at 37°C in 5% CO2. Culture medium was removed after 12 h and cells were preincubated with 50 μl of H2-DCFDA diluted 1/1000 in PBS to test cellular H2O2 production. After 30 min, 50 μl of mouse serum were added and H2O2 production was monitored spectrofluorimetrically for 6 h (Fusion; PerkinElmer). Results were expressed in arbitrary units per minute and per millions of cells.
Determination of advanced oxidation protein product (AOPP) concentrations in sera
AOPP concentration was measured by spectrophotometry as previously described (27). In test wells, 200 μl of serum diluted 1/20 in PBS were distributed onto a 96-well plate, and 20 μl of acetic acid was added. Next, 10 μl of 1.16 M potassium iodide were added. In standard wells, 10 μl of 1.16 M potassium iodide was added to 200 μl of chloramine-T solution followed by 20 μl of acetic acid. Calibration used chloramine-T within the range from 0 to 100 μmol/L. The absorbance was immediately read at 340 nm on a microplate reader (Fusion; PerkinElmer). AOPP concentration was expressed as 0–100 μmol/L of chloramine-T equivalents.
H2O2 production, fibroblast proliferation, and collagen synthesis induced by various types of oxidized proteins
DNA topoisomerase 1 (extracted from placenta (28)), human polyclonal IgG (IVIg, Tegeline; LFB) collagenase, pepsin, and BSA were oxidized with 1 mM HOCl or 10 mM OH· for 1 h at room temperature. Proteins were then dialyzed overnight against PBS and tested for AOPP content. For H2O2 production assay, endothelial cells (8 × 103 per well) were incubated with either oxidized or nonoxidized protein solutions (125 μg/ml) and the production of H2O2 was assessed spectrofluorometrically using H2-DCFDA. For fibroblast proliferation assay, NIH 3T3 fibroblasts (4 × 103 per well) were seeded in 96-well plates (Costar) and incubated with 50 μl of one of the oxidized or nonoxidized protein preparations (500 μg/ml) and 150 μl of culture medium without FCS at 37°C in 5% CO2 for 48 h. Cell proliferation was determined by pulsing the cells with [3H]thymidine (1 μCi/well) during the last 16 h of culture. Results were expressed as absolute number fibroblasts as cpm. For type I collagen mRNA synthesis assay, NIH 3T3 fibroblasts (106 per well in 4 ml of complete medium) were seeded in 6-well plates and incubated with 50 μl of one of the oxidized or nonoxidized protein preparations (500 μg/ml) for 24 h. After the incubation period, cells were washed three times with PBS, and total RNA was extracted from fibroblasts with TRIzol Reagent (Invitrogen). Type I collagen mRNA was then assayed by a standard two-step RT-PCR as previously described (29). The following PCR primers were used: for type I collagen (forward) 5′-TGTTCGTGGTTCTCAGGGTAG-3′ and (reverse) 5′-TTGTCGTAGCAGGGTTCTTTC-3′; and for β-actin (forward) 5′-TGGAATCCTGTGGCATCCATGAAAC-3′ and (reverse) 5′-TAAAACGCAGCTCAGTAACAGTCCG-3′. The PCR products were subjected to electrophoresis on a 2% agarose gel and detected by ethidium bromide staining. Amplicon intensity was quantified using a scanner densitometer (Vilber Lourmat). Results were expressed as densitometry units normalized for β-actin expression for collagen type I transcripts.
Determination of oxidized DNA topoisomerase 1 concentrations in the skin of mice exposed to ROS
DNA topoisomerase 1 was extracted from areas of skin injected with prooxidative agents, bleomycin, or PBS as previously described (28). AOPP was measured in the obtained extracts as described.
Determination of TGF-β1 in the skin of mice exposed to ROS
Skin samples taken from the site of injections from each mouse injected with HOCl, bleomycin, or PBS were diced using a sharp scalpel, put into aseptic tubes, thawed, and mixed with 500 μl of RIPA (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% Triton, 0.5% sodium desoxycholate, 0.5% SDS, 0.1% H2O, antiproteases). Equal amounts of loading buffer with solubilized proteins (30 μg/well) were subjected to 15% SDS-PAGE, transferred onto PVDF membranes, and incubated for 1 h at room temperature with anti-TGF-β1 Abs (Promega) using a Cassette Miniblot System (Immunetics). The membranes were then extensively washed and incubated with an anti-Fc γ-chain-specific goat anti-rabbit IgG Abs coupled to HRP (DakoPatts). Immunoreactivity was revealed with ECL (Amersham Biosciences).
AOPP depletion assays
Sera from mice treated with HOCl (n = 10) or PBS (n = 10), or sera from patients with SSC (n = 10) or healthy controls (n = 10) were incubated for 5 min at 37°C with various amounts of 2-ME or PBS as indicated in the each experiment. Sera were then immediately tested for AOPP concentration and for the ability to induce H2O2 production by endothelial cells and fibroblast proliferation as described. Control experiments were performed with the same amount of 2-ME but without sera, to ensure that the results were not linked to a direct effect of 2-ME on H2O2 production by endothelial cells or on fibroblast proliferation. Endothelial cell viability was assessed by Hoechst method immediately after the measure of H2O2 production by spectrophotometry.
DNA topoisomerase 1 depletion assays
Sera from patients (n = 5) were diluted 1/2 in PBS, and 200 μl of the serum dilution was incubated with 5 μl of rabbit polyclonal anti-human DNA topoisomerase 1 Ab (Santa Cruz Biotechnology) or irrelevant sera as controls for 2 h at 4°C. Then, 50 μl of protein G (Sigma-Aldrich) was added and incubated with the serum dilution overnight at 4°C. After the incubation period, the suspensions were centrifuged 5 min at 10,000 × g to remove protein G, and the supernatants were tested for AOPP content and for their ability to induce H2O2 production by endothelial cells and to induce fibroblast proliferation.
All quantitative data were expressed as mean ± SEM. Data were compared using the Mann-Whitney nonparametric test or Student’s t paired test. When analysis included more than two groups, one-way ANOVA was used. A value for p < 0.05 was considered significant.
The s.c. injection of ROS-generating agents induced dermal fibrosis in mice
Histopathological analyses of skin biopsy samples showed an increase in the dermal thickness of BALB/c mice treated with solutions generating HOCl, ONOO−, OH·, and bleomycin compared with sham-injected mice (p = 0.006, p = 0.007, p = 0.011, and p = 0.002, respectively, for dermal thickness) (Fig. 1,a and b). These results were corroborated by the measure of the concentration of acid- and pepsin-soluble type I collagen content per milligram of skin directly exposed to agents generating ROS or bleomycin. The concentration of type I collagen was significantly higher in the skin of BALB/c mice directly exposed to agents generating HOCl, ONOO−, OH·, and bleomycin than the concentration measured in skin of mice injected with PBS (p = 0.002, p = 0.001, p = 0.008, and p = 0.008, respectively) (Fig. 1,c), whereas the solution generating O2 had no effect (p = 0.27 vs PBS). Identical results were observed in DBA/1 and in (NZB x NZW)F1 mice injected (see Supplemental Fig. 1).5 In addition, fibroblasts isolated from the skin of mice submitted to treatment with agents generating HOCl displayed a higher proliferation rate than fibroblasts obtained from mice injected with PBS or bleomycin (p = 0.01 vs PBS and p = 0.003 vs bleomycin) (Fig. 1,d). The concentration of collagen content in skin was significantly decreased in mice injected with agents producing HOCl and simultaneously treated with the antioxidative agent N-acetylcysteine (NAC), suggesting that the accumulation of collagen observed upon HOCl injection was associated with an oxidative stress (p = 0.01 vs mice exposed to HOCl and treated with PBS) (Fig. 1 e).
The s.c. production of HOCl and OH· induced a systemic reaction with lung fibrosis and renal involvement
We next investigated whether s.c. production of ROS could trigger a systemic reaction as observed in the diffuse form of human SSc. Histopathological analysis of lung tissue from BALB/c mice injected with agents generating HOCl, OH·, or bleomycin showed thickening of the pulmonary interalveolar septa accompanied by cell infiltrates, whereas mice treated with agents generating ONOO−, O2, or PBS showed no signs of fibrosis or inflammation (Fig. 2,a). Moreover, in mice injected with agents generating HOCl, OH·, and bleomycin, the concentration of type I collagen in the lung was higher than the concentration found in mice injected with PBS (p = 0.0048, p = 0.024, and p = 0.024, respectively) (Fig. 2,b). By contrast, injection of agents generating ONOO− or O2 did not increase the level of collagen synthesis in lung vs PBS (p = 0.28 for ONOO−, p = 0.17 for O2). As observed for the skin, the concentration of collagen content in lung was significantly decreased in mice injected with agents producing HOCl and simultaneously treated with NAC (p = 0.0185 vs mice exposed to HOCl and treated with PBS) (Fig. 1,e). Immunohistochemistry analysis of lung tissue sections from BALB/c mice injected with agents generating HOCl or bleomycin were next performed to analyze the characteristics of the cell infiltrates observed in the lung of these mice. Staining with anti-mouse B220 or anti-mouse CD3 showed that most cells were consisting of T lymphocytes (Fig. 2 d).
In addition, kidneys from mice injected with HOCl or OH·-producing agents displayed abnormal accumulation of collagen in the interstitium with some foci of inflammatory cells (see Supplemental Fig. 2).5 Intimal fibrosis with intima-media thickening and a decrease of lumen diameter were observed in small renal arteries of mice treated with either bleomycin or an HOCl- or OH·-producing agent. By contrast, no signs of fibrosis or vascular damage were observed in kidneys from mice treated with an ONOO−- or O2-producing agent, or with PBS. In addition, kidneys from mice exposed to agents generating HOCl, OH·, or ONOO− displayed myocyte necrosis.
Exposition to an oxidative stress triggered a systemic autoimmune response
Because features of autoimmunity are observed in patients with SSc in addition to fibrosis, we investigated the presence of autoantibodies in the serum of mice injected with ROS-generating agents or bleomycin by immunoblotting experiments with whole cell protein extracts of HUVEC and HEp-2 cells. Serum Abs from mice submitted to the action of HOCl, OH·, ONOO−, or bleomycin but not O2 or PBS also bound to a 85-kDa protein band present in both endothelial and HEp-2 cell extracts. The same 85-kDa protein band was also recognized by the serum from a patient with limited cutaneous SSc and anti-centromere Abs (patient 1, Fig. 3,a). Serum Abs from mice submitted to the action of HOCl, OH·, or bleomycin, but not other ROS or PBS, bound to several bands including two 100-kDa protein bands present in both endothelial (Fig. 3,a) and HEp-2 cell extracts (data not shown). The same 100-kDa bands were recognized by the serum from a patient with diffuse cutaneous SSc and anti-DNA topoisomerase 1 IgG Abs (patient 2, Fig. 3 a).
In addition to these immunoblotting experiments, we investigated the presence of autoantibodies by ELISA experiments with recombinant DNA topoisomerase 1, the major target of autoantibodies in patients with diffuse cutaneous SSc and recombinant CENP-B, the main centromeric Ag targeted by autoantibodies in patients with limited cutaneous SSc. We detected higher levels of anti-CENP-B Ab in the sera of mice exposed to ROS (except O2) and bleomycin compared with mice injected with PBS (p = 0.004 for ONOO−, p = 0.001 for HOCl, p = 0.004 for OH·, and p = 0.001 for bleomycin) (Fig. 3,b). Sera from mice exposed to agents generating HOCl, OH·, and bleomycin contained a significant level of anti-DNA topoisomerase 1 Abs (p = 0.001, p = 0.006, and p = 0.006 vs mice treated with PBS, respectively) (Fig. 3 c). Consequently, both immunoblotting and ELISA experiments detected the presence of anti-CENP-B Ab in the sera of mice injected with agents generating HOCl, OH·, ONOO−, or bleomycin and confirmed the presence of significant levels of anti-DNA topoisomerase 1 Abs in the sera of mice injected with agents generating HOCl, OH·, or bleomycin, but not ONOO−.
We next tried to detect the presence of autoantibodies usually found in other connective tissue diseases but not in SSc in the sera of experimental mice (Fig. 3 d). We detected no significant levels of rheumatoid factors, anti-annexin V, or anti-cardiolipin Abs, regardless of the prooxidative product injected. Anti-DNA IgG Abs were detected in the sera from mice exposed to agents generating O2 but in no other group of mice (p = 0.017 as compared with mice receiving PBS). Therefore, all the prooxidative agents tested induced an autoimmune response, but only the exposition to agents generating HOCl, OH·, and ONOO− elicited an immune response characteristic of the SSc phenotype.
We then analyzed the spleen cell subpopulations in mice exposed to HOCl (n = 7), PBS (n = 5), or bleomycin (n = 7) for 6 wk. The s.c. injections of HOCl increased the number of total spleen cells and of spleen B cells when compared with PBS-injected mice (p = 0.039 for total spleen cells and p = 0.019 for spleen B cells) (see Supplemental Fig. 3, a and b).5 By contrast, no significant difference was observed in the number of CD11b+, CD4+, or CD8+ spleen cells between the animals exposed to HOCl or PBS (see Supplemental Fig. 3b).5 We next investigated the rate of proliferation of splenocytes after stimulation with an anti-CD3 mAb or with LPS. When stimulated by an anti-CD3 mAb, splenocytes isolated from mice submitted to HOCl displayed a higher proliferation rate than splenocytes obtained from mice injected with PBS (p = 0.032) (see Supplemental Fig. 3c).5 LPS exerted a higher pro-proliferative effect on splenocytes from mice injected with HOCl than on splenocytes from mice injected with PBS (p = 0.041) (see Supplemental Fig. 3c).5
We finally measured total serum IgM and IgG Ab concentrations in these mice. No significant difference was observed in total serum IgG Ab concentrations between mice treated with HOCl or PBS (see Supplemental Fig. 3d),5 whereas mice exposed to HOCl displayed higher serum IgM Ab concentrations than mice exposed to PBS (p = 0.005) (see Supplemental Fig. 3d).5
Taken together, our results confirm that HOCl injections induce an immune activation involving both B and T lymphocytes in BALB/c mice. This immune activation leads to T cell infiltration of the lung and to the production of autoantibodies, especially of anti-DNA topoisomerase I Abs.
Sera from mice and patients with SSc induced H2O2 production by endothelial cells
Because some mice s.c. injected with prooxidative agents developed not only a local reaction with skin sclerosis, but also systemic involvement with lung fibrosis and autoimmunity, we hypothesized that soluble factors generated at the site of ROS production and subsequently carried by the peripheral blood could mediate distal fibrotic processes. We thus tested the ability of the sera of these mice to induce the production of H2O2 by endothelial cells in vitro. The sera of mice exposed to agents generating HOCl, OH·, and to a lesser extent, ONOO− induced a higher production of H2O2 by endothelial cells than the sera of mice treated with PBS (p = 0.015, p = 0.004, and p = 0.032) (Fig. 4 a).
We tested sera from 20 patients with SSc and from 20 healthy individuals for their ability to mediate endothelial cell activation and H2O2 production. There were 10 patients who had limited cutaneous SSc and no lung involvement, 10 patients suffered from diffuse cutaneous SSc with pulmonary fibrosis (with TLC <75%). All the sera from patients with limited and diffuse SSc induced a higher production of H2O2 by endothelial cells than sera from healthy subjects (p < 0.0001 in each case) (Fig. 4,a). Nevertheless, the sera from patients with diffuse SSc and lung involvement induced a higher release of H2O2 by endothelial cells than the sera from patients with limited SSc and no lung fibrosis (p = 0.03) (Fig. 4 a). Thus, s.c. generation of ROS resulted in the production of some serum-soluble factors capable of activating endothelial cells and inducing the production of other ROS types.
Sera containing AOPP
We then investigated which circulating factors were involved in the propagation of the oxidative stress from skin to lung. We hypothesized that the soluble factors possibly related to the systemic oxidative stress could be proteins oxidized in the skin subsequently carried by the peripheral blood to the lung where they could induce the fibrotic process. Consistent with this hypothesis, we observed that sera from mice exposed to agents generating HOCl or OH· and sera from patients with diffuse cutaneous SSc and lung fibrosis SSc contained AOPP. The sera from mice injected with agents generating HOCl and OH· contained higher amounts of AOPP than the sera from mice treated with PBS (p = 0.024 and p = 0.006, respectively) (Fig. 4,b). In contrast, the sera from mice exposed to other types of ROS or bleomycin did not contain higher levels of AOPP than those from mice injected with PBS. Sera from patients with diffuse SSc and lung fibrosis (with TLC <75%) also contained higher amounts of AOPP than sera from healthy subjects (p = 0.04) (Fig. 4,b), whereas serum AOPP concentrations in patients with limited cutaneous SSC and no lung fibrosis did not differ from concentrations observed in healthy individuals (p = 0.88) (Fig. 4 b).
Oxidized DNA topoisomerase 1 induced ROS production, fibroblast proliferation, and type I collagen synthesis
To further characterize the nature of AOPP associated with the systemic oxidative stress observed in mice exposed to agents generating HOCl or OH·, we tested the ability of a variety of in vitro-synthesized AOPP to reproduce fibroblast proliferation and H2O2 production by endothelial cells and type I collagen synthesis. We tested albumin and IgG, given their high concentration in serum, and unrelated proteins like collagenase and pepsin for their ability to trigger H2O2 production by endothelial cells, fibroblast proliferation, and collagen synthesis. In addition, we also tested DNA topoisomerase 1 because this protein is a specific autoantigen in SSc and is able to be cleaved in an oxidation reaction (30). We observed that the properties of these in vitro-synthesized AOPP were dependent on the nature of the protein used. Indeed, AOPP generated from DNA topoisomerase 1 oxidized by HOCl or OH· induced a higher production of H2O2 by endothelial cells than AOPP derived from IgG, collagenase, albumin, or pepsin oxidized by HOCl or OH· (p < 0.0001 in all cases) (Fig. 5,a). Moreover, AOPP resulting from the oxidation of DNA topoisomerase 1 oxidized by HOCl or OH· induced the highest rate of fibroblast proliferation (p < 0.0001 in all cases) (Fig. 5,b) and the highest rate of type I collagen mRNA synthesis in vitro (p < 0.05 in all cases) (Fig. 5 c).
The s.c. injections of HOCl- and OH·-generating agents induced local formation of DNA topoisomerase 1-derived AOPP but not of TGF-β1
We then hypothesized that oxidized DNA topoisomerase 1 was formed in the skin areas where ROS were generated and was subsequently responsible for the development of the distal fibrotic lesions observed that may not be depending on cytokines such as TGF-β. To test this hypothesis, we performed selective DNA topoisomerase 1 extraction from the dorsal skin of mice exposed to HOCl or OH· because those two types of ROS induced the highest concentrations of AOPP in serum (Fig. 5,a). Oxidized DNA topoisomerase 1, as assayed by the measure of AOPP concentration in the DNA topoisomerase extract, was found in skin of mice treated with agents producing HOCl and OH·, but not in the skin of mice injected with bleomycin or PBS (Fig. 5,d). By contrast, the skin extracts from mice treated with bleomycin, but not those from mice injected with HOCl, contained TGF-β1 (Fig. 5 e).
Reduction of serum AOPP and DNA topoisomerase 1 decreased the production of H2O2 by endothelial cells and fibroblast proliferation
To directly investigate the involvement of AOPP and oxidized DNA topoisomerase 1 in the spreading of systemic fibrosis in this model, we reduced serum AOPP by using the reducing agent 2-ME. By adding 5 μM 2-ME to sera from mice exposed to agents producing HOCl, we obtained a significant decrease in AOPP concentrations (p = 0.033 vs similar sera without 2-ME) (Fig. 6,a). Conversely, no significant decrease of AOPP concentrations was observed in the sera from PBS-treated mice (p = 0.99 vs similar sera without 2-ME) (Fig. 6 a).
We next tested the sera depleted of AOPP for the ability to induce endothelial H2O2 production as compared with homologous nondepleted sera. AOPP depletion decreased H2O2 production in mice treated with a solution inducing HOCl (p = 0.001 vs similar sera without 2-ME) (Fig. 6 b). Control experiments performed with 2-ME alone showed no changes in endothelial cell viability and in H2O2 production (data not shown).
Finally, we tested the effects of all depleted and nondepleted sera on the proliferation of fibroblasts. Sera from mice treated with HOCl tended to exert a higher pro-proliferative effect on NIH 3T3 fibroblasts than sera from mice receiving PBS (p = 0.05 vs serum from PBS-treated mice) (Fig. 6,c). After AOPP depletion, sera from mice treated with HOCl lost their higher ability to stimulate fibroblast proliferation (p = 0.037 vs similar sera without 2-ME) (Fig. 6 c). 2-ME alone did not modify fibroblast proliferation (data not shown).
Similar results were observed with sera from SSc patients after addition of 5 μM 2-ME ex vivo (Fig. 6, a–c). In addition, because higher volumes of serum were available from patients than from mice, we performed additional experiments with human sera and mixed the sera with 1 and 10 μM 2-ME, respectively. AOPP concentrations and serum properties (induction of H2O2 production and of fibroblast proliferation) were related to the dose of 2-ME used, in a dose-dependent fashion.
Previous experiments showed that oxidized DNA topoisomerase 1 exerted higher effects than other AOPP. To assess the involvement of this protein in the spreading of the disease, we depleted DNA topoisomerase 1 protein from human sera by immunoprecipitation using an anti-human DNA topoisomerase 1 Ab. The SSc sera depleted of DNA topoisomerase 1 (n = 5) showed a decrease in whole AOPP concentration, in the production of H2O2 by endothelial cells, and in the proliferation of fibroblasts (Fig. 6 d).
Role of the immune system in the development of SSc in mice
We next investigated whether skin and lung fibrosis observed in BALB/c mice exposed to HOCl was dependent on a simultaneous activation of the immune system. For that purpose, immunodeficient SCID mice were injected s.c. with a solution generating HOCl. As observed in normal mice, immunodeficient SCID mice developed more skin and lung fibrosis than SCID mice exposed to PBS as assessed by collagen content (p = 0.001 and p = 0.029 for skin and lung, respectively) (Fig. 7, a and b). These results indicate that the immune system was not required for the initiation of SSc in mice. However, SCID mice exposed to a solution generating HOCl showed significantly lower pulmonary fibrosis than wild-type BALB/c, with no differences in the extent of skin fibrosis (p = 0.428 and p = 0.0158 for skin and lung collagen content respectively) (Fig. 7, a and b). This decrease in lung fibrosis was related to the immune system and not to differences in serum AOPP levels (Fig. 7,c) or endothelial H2O2 production potentials (Fig. 7 d) between the strains. Taken together, these results suggest that the immune system is not required for the development of skin fibrosis, but rather for the full development of lung fibrosis through synergistic interactions with the direct toxicity of oxidized proteins.
In this study, we produced and described original animal models of SSc for both the limited and diffuse forms of SSc that support the direct role of ROS in both forms of the disease (3). These tools offer new opportunities for studying this chronic and fatal human disease of unknown etiology.
We observed three different phenotypes in mice submitted to s.c. oxidative stress, depending on the type of ROS injected. The s.c. injections of agents generating O2 did not induce any feature of SSc. By contrast, s.c. injections of agents generating OH· or HOCl induced cutaneous and lung fibrosis and kidney involvement along with the production of serum anti-DNA topoisomerase 1 Abs; all features that characterize diffuse cutaneous SSc in humans. The injections of agents generating ONOO−, conversely, induced limited cutaneous fibrosis and the production of anti-CENP-B Abs in the absence of either lung involvement or the presence of serum anti-DNA topoisomerase 1 Abs. This disease shares numerous similarities to the limited cutaneous form of SSc. To our knowledge, this experimental demonstration is the first to show the two different clinical forms of human SSc in mice.
Using these animal models, we discovered a new mechanism involving AOPP, which accounts for the systemic propagation of the disease. AOPP were first described as new markers of oxidative stress in patients with uremia (27). Recently, these oxidized protein products have been shown to act as true inflammatory mediators. They are able to trigger the oxidative burst of neutrophils and monocytes and to stimulate dendritic cells in vitro (31, 32). In addition, they promote renal fibrosis in a remnant kidney model (33).
Our experiments suggest that AOPP, generated in high amounts after exposition of the skin to agents generating hypochlorite or hydroxyl anions, are directly involved in the spreading of the fibrosis from the skin into visceral organs via the systemic circulation, thus leading to the development of the diffuse form of SSc and anti-DNA topoisomerase 1 Abs. These observations are consistent with previous studies, which demonstrate a direct link between the concentration of AOPP in the serum of SSc patients and the development of lung fibrosis (6). Nevertheless, future research is necessary to fully understand the exact molecular mechanisms of the damages induced by AOPP.
Our study also demonstrated a new role for the protein DNA topoisomerase 1 in SSc. This protein is known to be the selective target of the immune response in patients with diffuse cutaneous SSc (34). Such anti-DNA topoisomerase 1 Abs are neither detected in patients with limited cutaneous SSc nor in those with other connective tissue disorders. Interestingly, Casciola-Rosen et al. (30) have shown that OH·, generated by the Fenton reaction, can cleave DNA topoisomerase 1. It was postulated that this modification could increase DNA topoisomerase 1 antigenicity and thus may be required for the development of anti-DNA topoisomerase 1 Abs. Our observations corroborate the findings of Casciola-Rosen et al. (30). In our animal model, we found that exposition to HOCl or OH·, generated via the Fenton reaction from H2O2, induced high amounts of oxidized DNA topoisomerase in the skin along with concomitantly high serum levels of anti-DNA topoisomerase 1 Abs. However, in addition to this known antigenic property, we found a direct pathogenic role for oxidized DNA topoisomerase 1 in SSc. Indeed, high amounts of oxidized DNA topoisomerase 1 were detected in the skin of mice after exposition to HOCl. We showed in vitro that oxidized DNA topoisomerase 1 was capable of triggering H2O2 production by endothelial cells and fibroblast proliferation, fully recapitulating the effects of whole AOPP contained in the sera of HOCl-treated mice and of patients with diffuse cutaneous SSc. These results were not reproduced with other oxidized proteins, suggesting that the observed phenotypes were solely due to the toxic properties of oxidized DNA topoisomerase 1. As expected, when we depleted DNA topoisomerase 1 in SSc sera, ROS production and fibroblast proliferation decreased close to baseline levels. Taken together, these results suggest that DNA topoisomerase 1, exposed to a specific oxidation, is not only the target of the immune response in SSc but is a crucial factor responsible for the diffusion of the oxidative stress and the induction of tissular damages in this mouse model and in SSc patients.
The pathogenic role of oxidized DNA topoisomerase 1 per se was confirmed by the development of SSc in SCID mice exposed to HOCl. These results obtained in mice with no operational immune system suggest that B and T lymphocytes are not required for the development of the disease (35, 36, 37). However, the extent of the HOCl-induced pulmonary fibrosis was lower in SCID mice than in immunocompetent mice, indicating that the immune system synergizes with the direct effects of oxidized proteins for the full development of the systemic disease (see Supplemental Fig. 4).5 This conclusion is consistent with the fact that, in mice exposed to HOCl, the lung fibrosis is associated with a T cell infiltrate, similar to that observed in human SSc (38).
Thus far, the classical model available to study SSc has been the bleomycin model. Bleomycin is a well known profibrotic agent that causes pulmonary fibrosis after tracheal instillation (39) and skin and lung fibrosis following s.c. injections (40). However, there are many features of bleomycin treatment that are not consistent with human SSc. First, whereas bleomycin treatment does not modify the phenotype of fibroblasts, fibroblasts obtained from SSc patients display a high rate of spontaneous proliferation. Our animal models reproduce this latter phenotype following treatment with agents generating HOCl. Second, in the bleomycin-induced SSc, systemic fibrosis is caused by the intrinsic chemical properties of the drug and involves the generation of cytokines such as TGF-β (41, 42). This cytokine, able to stimulate fibroblasts to produce matrix proteins in vitro, is not involved in our models and seems not to be involved human SSc. Although the data regarding the presence of this cytokine in the skin of SSc patients are contradictory (43, 44), it has been clearly shown that fibroblasts from patients with SSc do not secrete more TGF-β than normal cells (45). Thus, we believe that we have created a more representative model of human SSc than the bleomycin model.
In conclusion, we provide new animal models recapitulating the different clinical forms of human SSc. Using these models, we demonstrate a novel, direct role for ROS in the induction of SSc. We show that the nature of the ROS involved determines the extent of fibrosis and the specificity of the autoantibodies produced, thus dictating the form of the resulting SSc. In addition, we demonstrate for the first time that specific oxidation of an autoantigen, in this case DNA topoisomerase 1, can participate in the pathogenesis of an autoimmune disease by not only inducing a breach in tolerance, but also by exerting a direct toxic effect ultimately responsible for the initiation and systemic propagation of the disease.
We are indebted to Dr. Andrew Wang for reviewing the manuscript and to Agnes Colle for typing the manuscript.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by a grant from “Fondation Philippe” (to F.B.), grants from Actelion and the “Association des Sclérodermiques de France” (to A.S.), and a grant from the “Fondation pour la Recherche Médicale” (to P.G. and N.K.).
Abbreviations used in this paper: SSc, systemic sclerosis; AOPP, advanced oxidation protein product; NAC, N-acetylcysteine; O2, superoxide anion; OH·, hydroxyl radical; ONOO−, peroxynitrite; ROS, reactive oxygen species; TLC, total lung capacity.
The online version of this article contains supplemental material.