Autoantibodies to Fibrillin-1 Activate Normal Human Fibroblasts in Culture through the TGF-β Pathway to Recapitulate the “Scleroderma Phenotype”¹

Scleroderma or systemic sclerosis (SSc)³ is a chronic, multisystem disease characterized by widespread cutaneous and visceral fibrosis. Although the etiopathogenesis of SSc is unknown, it is believed to be a complex disorder of autoimmune origin based on familial and clinical overlap with other autoimmune diseases and the presence of circulating, highly disease-specific autoantibodies to nuclear and nucleolar Ags (1). Pathological findings in clinically affected SSc skin indicate that fibroblasts are activated to produce excessive amounts of collagens (COL) and other extracellular matrix (ECM) components, likely via activation of TGF-β-dependent pathways (2). The ligand TGF-β binds the TGF-β transmembrane receptor complex and induces phosphorylation of TGF-βRI by the TGF-βRII (TGFBRII) kinase, which in turn phosphorylates Smad3. Phosphorylated Smad3 (p-Smad3) translocates to the nucleus to activate transcription of target genes in the nucleus, such as COL (3). Furthermore, turnover of the ECM components in SSc also appears to be influenced by a family of matrix metalloproteinases (MMP), along with tissue inhibitors of MMP (TIMP) (4).

Recently, we have demonstrated that microsatellite and single nucleotide polymorphisms in and around the fibrillin-1 gene (FBN1) on chromosome 15q are associated with susceptibility to SSc (5, 6). A partial, in-frame duplication of the same gene (fbn1) is the cause of the tight skin (tsk) phenotype in a murine model of human SSc (7). Fibrillin-1 is the major constituent of 10- to 12-nm microfibrils in the ECM and recently has been demonstrated to be an important regulator of TGF-β (8). Mutations in FBN1 are known to cause Marfan syndrome, an autosomal dominant condition characterized by skeletal, ocular, and cardiovascular complications (9). In cultured fibroblasts from SSc patients, fibrillin-1 is synthesized and secreted normally into the ECM, but the fibrillin-1-containing microfibrils are unstable and turn over more rapidly when compared with microfibrils from control fibroblasts (10). First-degree relatives of patients with SSc also have the same microfibril instability, suggesting that this is an inherent genetic defect (11). Theoretically, a breakdown of fibrillin-1-containing microfibrils could lead to release of TGF-β sequestered in microfibrils (8). In addition, neoepitopes of unstable fibrillin-1 could be revealed to the immune system resulting in an autoimmune response. Indeed, we have recently found that a significant fraction of SSc patients have circulating autoantibodies to portions of fibrillin-1, which are SSc-specific (12, 13, 14). These autoantibodies were detected with a recombinant peptide expressing the proline-rich region of fibrillin-1 (12, 13). Interestingly, the tsk mouse also has been found to produce anti-fibrillin Abs as well (15). Therefore, we tested the hypothesis that affinity-purified human autoantibodies to fibrillin-1 from SSc patients might induce activation of normal human dermal fibroblasts using microarray screening of ECM gene expression profiles and quantitative RT-PCR, immunoblotting, and immunohistochemistry assays of certain ECM components known to play important roles in the profibrotic phenotype of the SSc fibroblast.

Our previous studies indicated that some autoantibodies to fibrillin-1 in SSc sera, but not all, are directed against a 30-kDa glutathione fusion protein containing the proline-rich C region of human fibrillin-1 (11, 12). The recombinant peptide (aa 378–495) in this study was cloned by RT-PCR from normal human fibroblast RNA and expressed in the pBADThio (Invitrogen Life Technologies) vector as a thioredoxin fusion protein with a C-terminal polyhistidine tag. The recombinant proteins were purified from Origami B (Novagen) lysates by immobilized metal affinity chromatography.

The sera from six SSc patients who had previously been shown to contain autoantibodies against the proline-rich region of fibrillin-1 (12, 13) were used for purification of this Ab by affinity chromatography (SulfoLink; Pierce) on a column coupled with the recombinant fibrillin-1 peptide. The affinity columns used for this purification would be expected to bind Abs of any class of Ig. Purified autoantibodies were tested by Western blot against the recombinant fibrillin-1 peptide and fibrillin-1 containing cellular lysates from normal fibroblasts. They were also were examined for binding to fibrillin-1 containing microfibrils in cultures of normal human fibroblasts using indirect immunofluorescence (see below) using goat anti-human IgG-FITC.

The primary cultures of fibroblasts from normal skin biopsies of two healthy individuals (ages 20 and 60 years) with no histories of autoimmune diseases were established by mincing and placing tissues into 60-mm culture dishes and maintained in DMEM with 10% FCS, antibiotic and antimycotic. Early passage fibroblast cell strains were plated at a density of 2.5 × 10⁵ cells in 35-mm dishes, and grown until they reached 80% confluence. Culture media were changed with FCS-free DMEM containing purified anti-fibrillin-1 autoantibodies (100 μg/ml), normal human IgG (100 μg/ml), or an equal volume of PBS. Cellular lysates and medium from cultured fibroblasts were harvested for gene and protein detections after 1, 6, 12, and 24 h.

Oligonucleotide microarrays containing 16,659 human genes from the Qiagen/Operon version 1.1 were used in gene profiling, primarily of ECM transcripts. The microarray chips were obtained from Center for Genome Information, University of Cincinnati. The 3DNA Array 350 Detection kit (Genisphere) was used for probe labeling and hybridization following manufacturer’s protocol. Briefly, total RNA was extracted from fibroblasts with a TRIzol kit (Invitrogen Life Technologies), and cDNA was synthesized by reverse transcriptase in the presence of Cy3-dUTP or Cy5-dUTP. Labeled cDNA probes were purified using a GFX column (Amersham Pharmacia Biotech). After denaturing, the probe mixture was applied to the microarray slides. The hybridizations were conducted at 55°C for 16 h. The hybridized slides were washed and detected using a ScanArray Lite array scanner (Packard Instrument), and gene expression was quantified and normalized with QuantArray software (Packard Instrument) as described in our previous report (16).

Quantitative real-time RT-PCR was performed using the ABI 7900 Sequence Detector (Applied Biosystems) as described previously (16). Specific primers and probes were purchased from Assays on Demand or Assays by Designs from Applied Biosystems. Total RNAs were extracted from cultured fibroblasts as described previously, and cDNAs were synthesized using Superscript II reverse transcriptase (Ivitrogen Life Technologies). Each sample was measured in duplicate. The PCR master mix (TaqMan Universal PCR Buffer and primers/probe) was added directly to each well of the cDNA plate. Transcripts were assayed on an ABI 7900 with optimized thermal cycling conditions. The resulting data were analyzed using SDS software (Applied Biosystems).

After a 24-h treatment with affinity-purified anti-fibrillin-1 autoantibodies, normal human IgG, or normal culture media, the fibroblasts were washed with PBS and fixed with 100% methanol at 4°C for 2 min. The fibroblasts were washed with PBS again and incubated with p-Smad3 polyclonal Abs. After washing, the fibroblasts were incubated with anti-rabbit Ab conjugated with FITC and washed with PBS. 4′,6′-Diamidino-2-phenylinodole was used for nuclear counterstaining. The images of fibroblasts with fluorescence-labeled proteins were acquired with a digital camera of the fluorescence microscopy (Nikon).

Cellular lysates from cultured fibroblasts were obtained using PBS-TDS (0.1% Triton X-100, 12.1 mM sodium deoxycholate, 3.5 mM SDS in PBS, pH 7.4) containing a mixture of protease inhibitors (0.2 mM PMSF, 1.0 mM EDTA, 0.5 μg/ml leupeptin and pepstain). Protein concentrations were determined by a spectrophotometer. Equivalent amounts of protein were loaded on each lane. Resolved proteins were transferred onto nitrocellulose membranes and incubated with 1/1,000 dilution of primary Ab (anti-human autoantibodies against fibrillin-1, rabbit polyclonal anti-human COL type 1, mouse monoclonal anti-human Smad3, and rabbit polyclonal anti-human p-Smad3). The murine anti-human fibrillin-1 mAb, COL type I polyclonal Ab, Smad3 mAb, and p-Smad3 polyclonal Abs, respectively, were purchased from Lab Vision, Biodesign International, Zymed Laboratories, and Santa Cruz Biotechnology. The peroxidase-conjugated secondary Abs (1/10,000 dilution) were used to detect specific protein bands with the ECL system (Amersham Pharmacia Biotech). The intensity of the bands was quantitated using a Storm 860 scanner and ImageQuant software (Molecular Dynamics).

Normalized microarray data from two fibroblast strains were analyzed by fold changes of selected ECM transcripts after exposure to anti-fibrillin-1 autoantibodies vs media alone vs normal human IgG at four time points (1, 6, 10, and 24 h). The genes selected were those that had been previously reported to be over-expressed in SSc fibroblasts (2, 16, 17, 18) and to show 2-fold or greater increased expression in both strains in response to antifibrillin-1 autoantibody as compared with the media or IgG-treated fibroblasts. These genes were considered markers of fibroblast activation and were selected for validation by quantitative RT-PCR. Quantitative RT-PCR was performed in duplicates for each RNA sample and paired t tests were used to assess differences in transcript levels of the selected genes. A transcript was considered significantly differentially expressed by RT-PCR at p < 0.05.

Anti-fibrillin-1 autoantibodies from pooled SSc sera were isolated by affinity chromatography using a recombinant fibrillin-1 peptide expressing the proline-rich region of fibrillin-1 (aa 378–495). Western blot showed that affinity purified anti-fibrillin-1 autoantibodies specifically reacted to the recombinant fibrillin-1 peptide (Fig. 1,A). To determine whether the autoantibody preparation bound fibrillin-1, the purified anti-fibrillin-1 autoantibody was used as a primary Ab on Western blots containing fractionated dermal fibroblast lysates. A mouse mAb to fibrillin-1 was used as a positive control primary Ab on an identical Western blot (Fig. 1,B). The blots showed that both the affinity purified anti-fibrillin-1 autoantibodies and mouse mAb to fibrillin-1 recognized the same high molecular mass band that corresponded to the molecular mass of fibrillin-1 on denaturing PAGE (Fig. 1,B). Finally, indirect immunofluorescence with the purified fibrillin-1 autoantibody preparation showed that it also bound to extracellular microfibrils in cultured human fibroblasts (Fig. 1 C). These data provide evidence that anti-fibrillin-1 autoantibodies recognize native fibrillin-1.

Microarrays were used to screen for genes whose transcription was altered by treatment with anti-fibrillin-1 autoantibodies purified from SSc sera (Fig. 2). Among the ECM components, the expression of the COL type I α 2 gene (COL1A2) was increased by 2.1-fold after 1 h of Ab treatment, and remained elevated at up to 24 h. COL3A1, COL4A2, and COL6A3 transcripts showed over 2-fold increased expression after 10 h of exposure to anti-fibrillin-1 Abs. MMP3 and TIMP3 genes showed >2-fold increased expression after 6 h of exposure. In contrast, fibrillin-1 transcripts showed over 2-fold increase only after 24 h of stimulation. Connective tissue growth factor (CTGF) also showed increased expression. There was no change in the expression of the TGF-β1 transcripts in response to autoantibody stimulation.

Several of the above genes were then selected for validation with quantitative RT-PCR. Comparisons of anti-fibrillin-1 autoantibody with normal human IgG-stimulated cells showed these changes to be specific (Fig. 3,A). The quantitative RT-PCR confirmed the increased expression of COL1A2, COL3A1, MMP1, TIMP3, and FBN1 (Fig. 3,B). In addition, the transcript levels of TGFBRII, CTGF, latent TGF-β binding protein 1 (LTBP1), and LTBP3 also showed increased responses to anti-fibrillin-1 autoantibodies by RT-PCR (Fig. 3,C). At the protein level, immunoblot analysis showed increased levels of both type I collagen and fibrillin-1 in the lysates of normal human fibroblasts after 24 h of treatment with anti-fibrillin-1 autoantibodies (Fig. 4).

Assessment of p-Smad3 and nonphosphorylated Smad3 with Western blots in cellular lysates of normal dermal fibroblasts showed increased amounts of p-Smad3 upon addition of anti-fibrillin-1 Abs after 1 h that persisted up to 24 h (Fig. 5,A). Immunofluorescence showed that 24 h after the addition of antifibrillin-1 autoantibodies, p-Smad3 staining was increased in the nucleus of fibroblasts compared with untreated controls or those treated with normal human IgG or media (Fig. 5,B). The addition of anti-TGF-β1-neutralizing Abs to the fibroblast cultures in combination with the anti-fibrillin-1 autoantibodies resulted in a significant blunting of stimulation of COL1A2, COL3A1, FBN1, CTGF, and TGFBRII transcripts (Fig. 6).

Our recent reports of a specific autoantibody response directed against a portion of an important structural and regulatory ECM molecule, fibrillin-1, in SSc raises the possibility of a pathogenic role for this autoantibody in the fibrosis characteristic of this disease (12, 13). The studies described here show an altered gene and protein expression profile of cultured normal human dermal fibroblasts in the presence of anti-fibrillin-1 autoantibodies as compared with normal IgG or media alone.

The most striking feature of this profile was increased gene expression of COL, MMPs, and TIMPs that seem to recapitulate the pattern of activated fibroblasts in SSc (2). An exception is up-regulation of MMP1, which is characteristically under-expressed in lesional SSc fibroblasts (2), although its pattern of expression in the early phase of SSc is unknown. Fibrillin-1, the target autoantigen, also showed increased transcript expression only after 24 h. Quantitative RT-PCR confirmed the microarray results for these transcripts. In agreement with the array and RT-PCR results, protein levels of COL type 1 and fibrillin-1 also were increased as measured by Western blot analysis. Taken together, these observations suggest a potential profibrotic effect of anti-fibrillin-1 autoantibodies in cultured normal human dermal fibroblasts in vitro.

The altered expression of fibrillin-1 demonstrated by all of these measures is of considerable interest. This is because fibrillin-1 has a pivotal role in regulation TGF-β signaling in the ECM, and elevated levels of immunoreactive TGF-β and its signal transduction components have been reported in SSc (2, 19). TGF-β is synthesized complexed to latent-associated peptide (LAP), and together the TGF-β:LAP complex is termed the small latent complex (20). During the secretory process, LTBP1 becomes disulfide-linked to the LAP forming the large latent complex (21, 22, 23). TGF-β itself remains noncovalently bound within this complex of LAP and LTBP1, and when secreted from cells, LTBP1 becomes immobilized in the ECM in a covalent manner involving tissue transglutaminase-mediated cross-linking of a region in the N-terminal sequence of LTBP1 (24, 25). LTBP1 has been immunolocalized to fibrillin-1-containing microfibrils in several tissues (26, 27, 28, 29). It has been proposed that the N-terminal region of LTBP1 is cross-linked to the ECM and while its C-terminal region is bound to the N-terminal region of fibrillin-1, and that interactions between LTBP1 and fibrillin-1 may stabilize latent TGFβ complexes in the ECM (29). Thus, the ECM can serve as a reservoir for latent, inactive TGF-β. Once released, TGF-β activates expression of ECM components including, interestingly, fibrillin-1 itself (30). Several of these genes including COL1A2, COL3A1, COL6A1, COL6A3, and TIMP1 genes, are up-regulated through Smad3-dependent mechanism (31).

Although our studies demonstrated that expression of the TGF-β1 gene (TGFB1) was not increased by treatment with anti-fibrillin-1 autoantibodies, some TGF-β-related genes, including TGFBRII, CTGF, LTBP1, and LTBP3 showed increased expression in cultured fibroblasts as early as 1 h after anti-fibrillin-1 autoantibody treatment, which preceded the subsequent fibrillin-1 transcript up-regulation (Fig. 3 C). TGFBRII is an upstream regulator and CTGF is an important downstream element involved in the TGF-β signaling pathway (32), whereas LTBP 1 and LTBP 3 are TGF-β binding proteins (33). The expression patterns of the TGFBRII, CTGF, and LTBP genes support a scenario involving the early activation of TGF-β.

It is of interest that the epitope (aa 378–495) recognized by the anti-fibrillin-1 autoantibodies from SSc sera contains the proline-rich region of fibrillin-1, which is in close proximity to the region where fibrillin-1 interacts with LTBP (8, 20). However, it should be kept in mind that these autoantibodies are polyclonal, and it is conceivable that the affinity purified anti-fibrillin-1 preparation we used contains autoantibodies that cross-react with other proteins. An amino acid basic local alignment search tool search shows that the 118-aa recombinant fibrillin-1 peptide we used has considerable homology to multiple other ECM proteins including fibulin-2, fibulin-3, LTBP3, LTBP2, and fibulins-1 through 5.

An attractive disease mechanism is that, anti-fibrillin-1 Abs may alter or disrupt the interaction between fibrillin-1 and LTBP (Fig. 7). Subsequently, released TGF-β induces activation of its signaling pathway with perturbation of ECM regulation (Fig. 7). The increased phosphorylation of Smad3 and nuclear translocation of p-Smad3 seen in the fibroblasts upon addition of anti-fibrillin-1 autoantibodies supports this hypothesis, given that Smad3 is known to be phosphorylated in response to TGF-β stimulation. Further evidence for this comes from the observation that the addition of anti-TGF-β1 Abs together with anti-fibrillin-1 autoantibodies blunts the up-regulation of markers of fibroblast activation induced by anti-fibrillin-1 Abs (Fig. 6).

In summary, these studies demonstrate a potential profibrotic effect of anti-fibrillin-1 autoantibodies from SSc patients at least in the cell culture system. Previously, we found no correlation of the presence of anti-fibrillin-1 autoantibodies and the severity of skin fibrosis in SSc patients (13); however, we also have documented variability of anti-fibrillin-1 autoantibody expression over time in different patients (14). Also, not all patients with SSc produce autoantibodies to fibrillin-1, but some may subsequently be found to produce other autoantibodies or have other abnormalities that promote fibrosis. Nevertheless, for the complex autoimmune disease, scleroderma, these data demonstrate a potential pathogenic role of this disease-specific autoantibody. The data raise the provocative possibility that the prominent fibrosis that characterizes this disorder may be driven or exacerbated, at least in part and/or in some patients, by anti-fibrillin-1 autoantibodies.

The authors have no financial conflict of interest.