All patients gave their written informed consent. Limited cutaneous SSc was defined by skin thickening in areas solely distal to the elbows and knees, with or without facial involvement; diffuse SSc was defined by the presence of skin thickening proximal, as well as distal, to the elbows and knees, with or without facial or truncal involvement (LeRoy et al 1988). Sixty patients were included in the study: 20 (16 females and 4 males, mean age 53.5 ± 15.2 years) with diffuse SSc and antitopoisomerase 1 antibodies, 20 (all females, 56.7 ± 12.8 years) with limited cutaneous SSc and anticentromere antibodies and 20 (14 females and 6 males, mean age 50.9 ± 10.9 years) with diffuse SSc and no antitopoisomerase 1 or anticentromere antibody. For these 3 groups, disease durations were measured from the onset of first non-Raynaud’s phenomenon symptom (11.3 ± 9.6 years, 9.4 ± 9.5 years, 5.6 ± 5.4 years in the 3 groups, respectively). Interstitial lung disease, assessed by chest high-resolution computed tomodensitometry and abnormal pulmonary function test values (total lung capacity or vital capacity less than 80% of predicted values and/or diffusing capacity <75% of that predicted) was present in all patients with antitopoisomerase 1 antibodies, in 13 patients without antitopoisomerase 1 or anticentromere antibody and none of the patients with anticentromere antibodies. Pulmonary arterial hypertension, defined as systolic pulmonary artery pressure was >40 mm Hg (Mukerjee et al 2003) as measured by Doppler echocardiography, was diagnosed in 4 patients with antitopoisomerase 1 antibodies, 2 patients with anticentromere antibodies and 3 patients without antitopoisomerase 1 or anticentromere antibody. None of the patients included had cancer or another connective tissue disease. None of the patients received corticosteroids or immunosuppressive therapy. Sixty unrelated healthy age-, gender- and parity-matched donors served as controls.

Serum samples were obtained from patients and controls and stored in aliquots at −80 °C until tested. Mean serum IgG and IgM concentrations determined by nephelometry were respectively: 10.6 ± 2.9 and 1.4 ± 0.8 mg/ml in patients with anticentromere antibodies, 13.9 ± 3.9, 1.7 ± 0.74 mg/ml in patients with antitopoisomerase 1 antibodies, 14.4 ± 3.3 and 1.9 ± 1.0 mg/ml in patients without specific autoantibody and 12.0 ± 2.8 and 1.6 ± 0.7 in healthy controls. Serum IgG and IgM concentrations did not differ significantly between patients in the different groups and controls. A therapeutic preparation of normal human IgG, intravenous immunoglobulin (IVIg) and normal human polyclonal IgM (LFB, Les Ulis, France) served as internal standards.

Normal human dermal fibroblasts were cultured from histologically normal skin obtained in 3 donors undergoing resection of basocellular carcinoma. Biopsy specimen were cut into small sections and seeded into Petri dishes and then into 162 cm² plastic flasks. Fibroblasts were grown in Dulbecco’s modified Eagle’s medium (DMEM) with 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, 10 mM Hepes, 100 U/ml penicillin, 100 μg/ml streptomycin (Life Technologies Ltd, Auckland, NZ) at 37 °C in 5% CO₂. When confluent, the cells were detached using 0.05% trypsin with 2 μM EDTA.

HEp-2 cells were obtained from EuroBio (Les Ulis, France) in 162 cm² flasks in DMEM and cultured at 37 °C in 5% CO₂ supplemented with 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, 10 mM hepes, 20 mM sodium pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin (Life Technologies Ltd). When confluent, the cells were detached using 0.05% trypsin, 2 μM EDTA.

For each patient and healthy control, antinuclear antibodies were investigated by indirect immunofluorescence on HEp-2 cells and considered positive at a dilution ≥1/160. Anticentromere antibodies were characterized by a pattern of discrete dots in cells lined up on the metaphase plate in dividing interphase cells. Antitopoisomerase 1 antibodies were detected by ELISA with a commercial kit (Inova Diagnostics Inc, San Diego, CA, USA) using purified calf thymus topoisomerase 1 and sera diluted 1:100.

IgG and IgM reactivities were analyzed using a semi-quantitative immunoblotting technique with normal human dermal fibroblasts and HEp-2 cells. Cellular protein extracts were performed in a buffer containing 4% sodium dodecyl sulfate, 1.45 M 2-mercaptoethanol, 125 mM Tris/HCl pH 6.8, 1 μg/ml aprotinin, 1 μg/ml pepstatin and 1 μg/ml leupeptin and then sonicated 4 × 30 s (Chanseaud et al 2003; Garcia de la Pena-Lefebvre et al 2004). Equal amounts of solubilized proteins were subjected to preparative sodium dodecyl sulfate–polyacrylamide gel electrophoresis through 10% polyacrylamide gels. The proteins were then transferred onto nitrocellulose membranes for 1 h at 0.8 mA/cm² using a semi-dry electroblotter model A (Ancos, Hojby, Denmark). After blocking with phosphate buffer saline −0.2% Tween^® for 90 min, the membranes were incubated with sera for 4 h at room temperature at 200 μg/mL for IgG and 20 μg/mL for IgM in a cassette miniblot system (Immunetics Inc., Cambridge, MA, USA). They were then extensively washed before being incubated with secondary rabbit antihuman Fc-γ of antihuman Fc-μ antibody coupled to alkaline phosphatase (Dakocytomation, Golstrup, Denmark) for 90 min at room temperature. Immunoreactivities were revealed using the nitroblue tetrazolium-5-bromo-4-chloro-3-indolyl-phosphate substrate (Sigma, St. Louis, MO, USA) (Garcia de la Pena-Lefebvre et al 2004; Tamby et al 2005). Immunoreactivities were quantified by scanning the membranes with a densitometer (Epson Perfection 1200S, Seiko Epson Corporation, Nagano-Ken, Japan). The membranes were then stained with colloidal gold (Protogold;^® Biocell, Cardiff, UK) and subjected to a second densitometric analysis to quantify transferred proteins. This approach allows the immunoreactivity repertoires to be compared by referring to their respective protein peaks corrected for electrophoretic migration defects by superimposing corresponding protein peaks by means of computer analysis. Standard IgM or IgG preparations included in each blot, allow the rescaling of the membranes transferred with a given protein extract and to adjust for the intensity of labelling on different membranes (Chanseaud et al 2003; Garcia de la Pena-Lefebvre et al 2004). All experiments were performed twice.

Rabbit polyclonal antibody raised against amino acids 685–765 of human DNA topoisomerase 1 recombinant protein (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was tested in immunoblots during 90 min at room temperature at dilutions of 1:500, and 1:1000 for reactivity with fibro-blasts proteins extracts. Immunoreactivities were assayed using an alkaline phosphatase conjugated goat antirabbit IgG antibody (Santa Cruz) and revealed using nitroblue tetrazolium-5-bromo-4-chloro-3-indolyl-phosphate) (Sigma).

To identify the target antigens, we performed an additional SDS-PAGE protein electrophoresis with a fibroblast extract in 10% polyacrylamide gels and dissected bands of interest. We then performed a passive adsorption onto a polyvinyle difluoride (PVDF) membrane (Messer et al 1997) and not an electrotransfer, in order to favor the protein binding onto the membrane in a nondestructive manner for subsequent analysis of the separated proteins of interest. The proteins were derivatized and sequenced by automated Edman’s degradation using a PE Applied Biosystems 492 cLC protein sequencer (Perkin Elmer, Wellesley, MA, USA) allowing the identification and quantification in the range of femtomoles of the amino acids released.

Because of the large number of reactivities identified in the blots, we decided to submit the data to multivariate statistical analysis. Therefore we used IGOR software (Igor Pro 3.16, Wavemetrics Inc, Lake Oswego, OR, USA) with specially designed software packages (Nobrega et al 1993), and Stat-view software (Statview 5.0; SAS institute Inc., Cary, NC, USA). Densitometry curves of IgG and IgM reactivities of each patient and control on each of the different protein extracts were divided into sections surrounding individual peaks of immunoreactivity on each substrate. In order to discriminate between groups of individuals, the peak amplitudes corresponding to these sections were submitted to principal component analysis (Nobrega et al 1993; Mouthon et al 1996). Principal component analysis data were then subjected to linear discriminant analysis, and the individual linear discriminant analysis factor 1 coordinates of each patient or healthy control in the 2 groups were compared using a paired Wilcoxon test. Repertoires of reactivities of patients in each group were compared two by two by using the same procedure, except that the individual linear discriminant analysis factor 1 coordinates of each patient in the 2 groups were compared using a Mann-Whitney test.

IgG and IgM reactivity patterns with fibroblast antigens were homogeneous in terms of protein bands recognized and intensity of reactivities among healthy individuals and patients in each group. As depicted in Figure 1, IgG from healthy individuals bound to 5 to 8 protein bands in the fibroblast extract, with expression of three main immuno-reactivities of approximately 53, 44, and 33 kDa, respectively (Figure 1A). Two other minor IgG reactivities were directed against low molecular weight proteins, and some patients expressed individual specific reactivities. IgG from patients in the different groups also bound to 6 to 10 protein bands in the fibroblast extract including immunoreactivities directed toward the same 53, 44, and 33 kDa bands. In addition to these bands, IgG from patients with limited cutaneous SSc and anticentromere antibodies (Figure 1B) bound with low intensity to a 80 kDa antigen. IgG from patients with diffuse SSc and antitopoisomerase 1 antibodies (Figure 1C) expressed a high intensity reactivity with the 44 kDa protein band as compared with those expressed by IgG from patients in other groups and healthy controls. Interestingly, almost all patients with diffuse SSc and antitopoisomerase 1 antibodies bound to two 90 and 65 kDa protein bands, whereas IgG from patients with limited cutaneous SSc and anti anticentromere antibodies and healthy controls did not. Finally, IgG from several patients without specific autoantibody bound to the two 90 and/or 65 kDa fibroblasts protein bands, whereas other patients in this group did not (Figure 1D). IgG from several patients in each group expressed individual specific reactivities.

IgM from patients and healthy individuals bound to three to five protein bands in the fibroblast extract and with a high intensity reactivity with the 53kDa protein band (data not shown).

IgG and IgM reactivities from patients in each group and healthy controls were compared by performing principal component analysis followed by linear discriminant analysis and non parametric tests, as previously described. IgG reactivities from patients with diffuse SSc and antitopoisomerase 1 antibodies with fibroblasts antigens different significantly from those of healthy controls, whereas no significant difference was observed between limited cutaneous SSc with anticentromere antibodies and their controls and diffuse SSc without antitopoisomerase 1 or anticentromere antibody and their controls (Table 1).

IgM reactivities from patients with limited cutaneous SSc with anticentromere antibodies were significantly different from those of healthy controls, whereas no significant difference was observed between patients in other groups and their respective healthy controls (Table 1).

Multiparametric analysis of serum reactivities from patients compared two by two differed in the case of IgG from patients with diffuse SSc with antitopoisomerase 1 antibodies vs patients with diffuse SSc without antitopoisomerase 1 or anticentromere antibody and in the case of IgM from patients with diffuse SSc and antitopoisomerase 1 antibodies vs patients with limited cutaneous SSc with anticentromere antibodies (Table 2).

In order to better characterize the cell specificity of AFA, we further performed experiments with HEp-2 cells, the reference source of nuclear antigens. IgG reactivity patterns were relatively homogeneous among healthy individuals and among patients in each group when tested toward HEp-2 cells protein extracts. IgG from patients with antitopoisomerase 1 antibodies recognized 3–6 protein bands in HEp-2 cells extract, including one major band of 100-kDa, whereas IgG from patients with anticentromere antibodies recognized 4–8 bands, including two major bands of 65 and 80 kDa, whereas none of them bound to the 100-kDa band (Figure 2). IgG from patients without antitopoisomerase 1 or anticentromere antibody recognized 6–10 HEp-2 cells bands, including the 100-kDa protein band in 5/20. IgM reactivity patterns with HEp-2 cells protein extracts were relatively homogeneous among healthy individuals and among patients in each group. No difference was observed between IgM reactivities of patients in each group and healthy individuals (data not shown).

Principal component analysis of the data discriminated between the IgG reactivity patterns of SSc patients in each group and healthy controls as well as between patients in the different groups compared two by two (Tables 1 and 2). IgM reactivity patterns with HEp-2 cells protein extracts were relatively homogeneous among healthy individuals and among patients in each group. No significant difference was observed between IgM reactivities of patients in each group and controls or between patients in each group tested two by two, with the exception of patients with antitopoisomerase 1 antibodies versus healthy individuals (Tables 1 and 2). We thus demonstrated that IgG reactivities from SSc patients in the different groups were not specific to fibroblasts antigens but also bound HEp-2 cells antigens.

We incubated the fibroblast protein extract with rabbit polyclonal antibodies raised against amino acids 685–765 of human DNA topoisomerase 1 recombinant protein as indicated in Materials and methods. This antibody bound to the same 90 kDa protein band that was recognized by serum IgG from patients with antitopoisomerase 1 antibodies (data not shown). We also tested the polyclonal goat antitopoisomerase 1 antibody on HEp-2 protein extract and observed that this Ab bound to the 100 kDa protein band (data not shown).

We further sequenced the 90-kDa protein band, as described in Materials and methods. DNA-topoisomerase I was identified by sequencing the 10 to 25 N-terminal amino acids: SQIEADFRLNDSHKH of this protein. Thus, the N-terminal sequence of the protein started at position 10, suggesting a lysis of the nine first amino acids during preparation of the protein sample (data not shown). Thus, AFA from patients with diffuse SSc with or without antitopoisomerase 1 antibodies bind to fibroblast topoisomerase 1.