Dysregulation of transforming growth factor β signaling in scleroderma: Overexpression of endoglin in cutaneous scleroderma fibroblasts

Abstract

Objective

As an initial approach to understanding the basis of the systemic sclerosis (SSc; scleroderma) phenotype, we sought to identify genes in the transforming growth factor β (TGFβ) signaling pathway that are up-regulated in lesional SSc fibroblasts relative to their normal counterparts.

Methods

We used gene chip, differential display, fluorescence-activated cell sorter, and overexpression analyses to assess the potential role of TGFβ signaling components in fibrosis. Fibroblasts were obtained by punch biopsy from patients with diffuse cutaneous SSc of 2–14 months' duration (mean 8 months) and from age- and sex-matched healthy control subjects.

Results

Unexpectedly, we found that fibroblasts from SSc patients showed elevated expression of the endothelial cell–enriched TGFβ receptor endoglin. Endoglin is a member of the nonsignaling high-affinity TGFβ receptor type III family. The expression of endoglin increased with progression of disease. Transfection of endoglin in fibroblasts suppressed the TGFβ-mediated induction of connective tissue growth factor promoter activity.

Conclusion

SSc is characterized by overproduction of matrix; that is, genes that are targets of TGFβ signaling in normal fibroblasts. Our findings suggest that lesional SSc fibroblasts may overexpress endoglin as a negative feedback mechanism in an attempt to block further induction of profibrotic genes by TGFβ.

Scleroderma (systemic sclerosis [SSc]) is a chronic systemic fibrosing disorder of unknown etiology characterized by progressive scarring and fibrosis of the skin and certain internal organs (1). In addition to fibrosis, SSc is characterized by immunologic abnormalities and vascular dysfunction, which manifests itself clinically as Raynaud's phenomenon (1). The pathologic phenotype of SSc is complex and, consequently, the etiology of the disease and effective therapies to combat this disorder remain elusive.

Due to this complexity, widely divergent theories have been proposed to explain the SSc phenotype. Owing to the role of transforming growth factor β (TGFβ) in matrix synthesis and fibroblast differentiation, this growth factor was proposed over a decade ago to play a key role in the pathogenesis of SSc (2). Anti-TGFβ strategies seem to be effective at blocking the onset of fibrosis in some animal models of acute (i.e., resolving) disease; however, these models may not have direct relevance to the human disease (for review, see ref. 3). In particular, SSc is a chronic fibrotic disorder, and the role played by TGFβ in maintaining persistent fibrosis is not understood.

Evidence implicating TGFβ as a profibrotic growth factor in SSc is scant and somewhat contradictory (for review, see ref. 4), chiefly relying on approaches that stain for proteins in tissue sections of involved skin from SSc patients or in vitro cell culture. For example, although some early studies showed elevated TGFβ levels in fibrotic skin, other studies showed no such elevation (5–9). However, a more recent study aimed at addressing this controversy showed that levels of TGFβ messenger RNA (mRNA) were elevated in the leading, inflammatory edge of the lesion, but not in the fibrotic lesion itself (10). These findings suggest that TGFβ may play a role in the initiation and expansion, but not maintenance, of the SSc fibrotic lesion. Intriguingly, TGFβ is elevated in the substantially less severe limited form of SSc but not in diffuse cutaneous SSc (6). Discrepancies between the earlier studies may be due to differences in the stage of the disease or in the area of the biopsy from which the fibroblasts were taken.

Adding further complexity to the potential involvement of TGFβ in SSc are the observations that SSc fibroblasts do not secrete more TGFβ than normal cells, and that SSc fibroblasts are not more sensitive to TGFβ treatment (11–14). In fact, SSc fibroblasts seem to be less sensitive to TGFβ treatment than their normal counterparts are (11–14). Data concerning the levels of TGFβ receptors (TGFβR) in the pathology of fibrosis are similarly confusing. Some studies have shown no elevation of TGFβ binding to the cell surface of SSc cells (11, 12), whereas one study showed elevated levels of TGFβR in SSc fibroblasts (15). Although the precise role of TGFβ and TGFβ signaling in SSc remains to be defined, the studies thus far suggest that TGFβ is not correlated with the maintenance of fibrosis, but rather, is associated with the initiation of fibrosis.

We have initiated studies aimed at evaluating on a molecular level the role of TGFβ signaling in persistent fibrosis. Although not normally expressed in fibroblasts unless induced by TGFβ, the connective tissue growth factor (CTGF) protein is constitutively overexpressed in SSc lesions and in fibroblasts cultured from lesional areas of SSc patients (14, 16). The TGFβ induction of CTGF in normal fibroblasts depends on the TGFβ-signaling mediators, Smads (16). However, the overexpression of CTGF in SSc fibroblasts from established lesions was found to be independent of the action of Smads (16, 17). In addition, since the TGFβ/Smad target gene plasminogen activator inhibitor was not elevated in SSc fibroblasts relative to normal fibroblasts (16), the maintenance of the SSc phenotype may not be due to a general elevation of Smad-dependent signaling.

Although these studies collectively suggest that TGFβ and its downstream signaling events may not be generally stimulated in lesional SSc fibroblasts, the precise role of TGFβ signaling in the pathology of persistent fibrosis is not yet fully understood. In the present study, we investigated the potential role of TGFβ signaling in the fibrotic phenotype in SSc. We found that the endothelial-enriched high-affinity TGFβ receptor endoglin (18) is up-regulated in SSc fibroblasts and that the expression of endoglin increases with progression of disease. One intriguing finding was that overexpression of endoglin in fibroblasts blocks the accumulation of activated, nuclear Smads and suppresses the ability of TGFβ to induce the promoter of a target gene, the profibrotic cytokine CTGF. Together with the previous observation that endoglin is a functional TGFβ receptor in fibroblasts (19), these results suggest that SSc fibroblasts induce the expression of endoglin to suppress the TGFβ induction of gene expression in a negative feedback loop.

MATERIALS AND METHODS

Cell culture.

Skin samples from 8 female patients who met the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for SSc (20) were initially analyzed for array, reverse transcription–polymerase chain reaction (RT-PCR), and Western blot analyses. An additional patient was subsequently included in the study for Western blot analysis. All patients had diffuse cutaneous SSc of 2–14 months' duration (average 8 months; average initial skin score 32 of a possible 66). No patient had detectable lung or kidney involvement. We used a separate group of 5 patients with similar characteristics for differential display and cell sorter analysis. Patients underwent two 5-mm punch biopsies of involved forearm skin (8 cm above the ulnar styloid on the extensor aspect of the nondominant forearm). Six healthy female control subjects matched for sex and age with the SSc patients were biopsied in the identical location. All study subjects gave written informed consent for the procedures.

Dermal fibroblast cultures were grown from the skin biopsy tissues as described elsewhere (21). Briefly, biopsy specimens were cut into small sections and seeded into T25 plastic flasks. Cells were grown in Dulbecco's modified Eagle's medium with a 1% solution of penicillin (100 units/ml), streptomycin (100 μg/ml), and Fungizone (0.025 mg/ml) (Life Technologies, Gaithersburg, MD) and 10% fetal bovine serum (Life Technologies). Control and scleroderma fibroblasts were grown simultaneously and used before passage 5. Human dermal fibroblasts (American Type Culture Collection, Rockville, MD) were cultured as previously described (16, 21).

RNA extraction and gene chip and differential display analyses.

Total RNA was extracted using TRIzol (Life Technologies). Twice-purified poly(A+) RNA (Oligotex mini RNA kit; Qiagen, Chatsworth, CA) was used to synthesize double-stranded complementary DNA (cDNA) (Superscript Choice System; Life Technologies). cDNA was synthesized with a modified oligo(dT) primer incorporating a T7 polymerase promoter site. Following phenol–chloroform extraction and ethanol precipitation, cDNA was transcribed in vitro using T7 polymerase (T7 Megascript System; Ambion, Austin, TX) with biotin-labeled CTP and UTP.

Labeled sample RNA was purified and hybridization to a proprietary Roche DNA microarray (Roche Molecular Biochemicals, Indianapolis, IN) was performed as described elsewhere (22). Test hybridizations were performed to assess the quality and reliability of cDNA signals. The differential display study between affected and unaffected skin of SSc patients was conducted as described previously (14).

Semiquantitative RT-PCR.

To confirm the presence of mRNA for endoglin in scleroderma fibroblasts, semiquantitative RT-PCR was performed as described elsewhere (23). Total RNA was extracted according to the manufacturer's directions for RNAzol B (Tel-Test, Friendswood, TX) from fibroblasts from 2 confluent T175 flasks. The cDNA was made from 4 μg of RNA in 9 μl of diethyl pyrocarbonate–treated water to which was added 2 μl of a 0.5 μg/ml mixture of random-hexamer primers and 0.5 μl of RNasin (Promega, Madison, WI). After heating RNA to 65°C for 5 minutes, 5 μl of 1.25 mM dNTPs (Roche), 4 μl of 5× reverse transcriptase buffer, and 1 μl of Moloney murine leukemia virus reverse transcriptase (University of California, San Francisco, Cell Culture Facility) were added and incubated at 37°C for 90 minutes. The reaction was stopped by heating to 95°C for 5 minutes and then quenching on ice.

A competitor plasmid containing the hypoxanthine guanine phosphoribosyltransferase (HPRT) cDNA engineered with an intervening sequence was added to the amplification reaction to allow internal normalization of constitutive proteins (23). Dilution of the input cDNA to the amount at which the sample HPRT (which can be distinguished from the plasmid HPRT competitor by size) is normalized across samples allows the quantitative assessment of the input endoglin cDNA. The sequences for the HPRT primers were 5′-CCTGCTGGATTACATCAAAGCACTG-3′ and 5′-ACCGAATAATTAGTCAGCT-3′. The endoglin primers were 5′-AATGCCATCCTTGAAGTCCATGTC-3′ and 5′-GGAATTGTAGGCCAAGTGCAG-3′ (Life Technologies).

Ten microliters of cDNA was added to 10 μl of 10× PCR buffer (15 mM MgCl₂, 100 nM Tris, pH 9.0, 500 mM KCl, 1% Triton X-100), 2 μl each of 20 μM sense and antisense primers, 1 μl of 20 mM dNTPs, 74.5 μl of water, and 0.5 μl of Taq polymerase (Chiron, Emeryville, CA) and cycled 35 times at 94°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds, with a final extension at 72°C for 10 minutes. PCR products were visualized by agarose gel electrophoresis and ethidium bromide staining.

Western blot analysis.

Fibroblasts from cell monolayers were lysed in 1% Triton X-100, 0.1% sodium dodecyl sulfate (SDS), and 2 mM phenylmethylsulfonyl fluoride in Dulbecco's phosphate buffered saline without Ca⁺⁺ or Mg⁺⁺ (PBS). Supernatants were clarified by centrifugation (10,000g for 10 minutes at 4°C). Aliquots containing 15 μg of total protein per lane were separated using an 8–10% polyacrylamide gel. Samples were transferred to nitrocellulose (Schleicher & Schuell, Keene, NH), visualized with Ponceau Red (Sigma, St. Louis, MO), and blocked with 5% nonfat milk in PBS (Bio-Rad, Richmond, CA). The filters were incubated with anti-human endoglin antibodies (anti-CD105) at a 1:1,000 dilution (PharMingen, San Diego, CA) for 1 hour at room temperature. The blots were washed and incubated for 1 hour with a 1:20,000 dilution of horseradish peroxidase–conjugated anti-mouse secondary antibody (Amersham, Arlington Heights, IL). Antigens were visualized using a chemiluminescence kit (Amersham or Pierce, Rockford, IL).

Immunohistochemistry.

Fibroblasts were seeded on coverslips at 1.5 × 10⁵ cells/ml in 6-well dishes with 4–5 12-mm round coverglasses (Fisher Scientific, Fair Lawn, NJ). When the cells were confluent, the coverslips were removed and washed with PBS. Monolayers were fixed (100% acetone for 10 minutes at −20°C), blocked with 10% goat serum in PBS for 15 minutes, and then incubated with anti-endoglin (PharMingen) at a dilution of 1:400 for 1 hour at room temperature or overnight at 4°C. Washing was performed by gently adding 200 μl of PBS per coverslip and removing it by light vacuuming 4 times. Fluorescein isothiocyanate (FITC)–labeled anti-mouse antibody (Sigma) at 1:200 was then added and incubated for 1 hour at room temperature.

The coverslips were then washed with PBS, mounted in FITC-Guard (Testog, Chicago, IL), and viewed using a Zeiss Axioskop fluorescence microscope (Carl Zeiss, Thornwood, NY). Images were captured and analysis was performed on a Macintosh computer using the NIH Image program (developed at the National Institutes of Health, Bethesda, MD; available online at http://rsb.info.nih.gov/nih-image/) with a Zeiss Optronix charge-coupled device camera and framegrabber.

Flow cytometry.

Endoglin (CD105) expression and total TGFβ cell surface binding were assessed by flow cytometry. The reagents used were FITC-labeled anti-CD105 (endoglin; Research Diagnostics, Flanders, NJ) and biotinylated TGFβ1 (R&D Systems, Minneapolis, MN). Single-cell suspensions were prepared from confluent monolayers of normal and SSc fibroblasts. Cells were incubated with primary antibodies/reagent at a concentration of 10 μg/ml for 60 minutes at 4°C, washed twice, and incubated with secondary reagents for a further 30 minutes at 4°C when required.

Cells were then washed and fixed in 1% paraformaldehyde in PBS and the fluorescence intensities of the cell population were analyzed by flow cytometry using FACSCalibur (Becton Dickinson, Mountain View, CA). Isotype-matched irrelevant primary monoclonal antibody was used as a control for nonspecific binding, and the average fluorescence intensities were determined by subtracting background values from experimental values.

Transfection and reporter assays.

NIH3T3 and dermal fibroblasts were cultured in 6-well plates and transfected with Lipofectamine Plus (Life Technologies) as previously described (16, 21). TGFβ1 (Invitrogen, San Diego, CA) was used at 2.5 ng/ml. The full-length CTGF reporter construct has been described elsewhere (21). The endoglin expression vector was a kind gift from C. Bernabeu (Centro de Investigaciones Biologicas, Madrid, Spain), the Smad-3 and Smad-4 expression vectors were from J. Massague (Howard Hughes Medical Institute, New York, NY). Expression values were controlled for variations in transfection efficiencies by cotransfecting test DNA constructs with a cytomegalovirus/β-galactosidase plasmid and assaying cells for β-galactosidase activity (Clontech, Palo Alto, CA). After adjusting for differences in transfection efficiency between data points, expression values are reported as the mean ± SEM.

For studies involving nuclear localization of Smads, NIH3T3 cells were cultured in 100-mm dishes and transfected with empty expression vector or expression vector encoding endoglin as described above. Nuclear extracts were prepared using a commercial kit (Pierce), and equal amounts of protein (20 μg) were subjected to SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis as described above, using a 1:250 dilution of a rabbit anti–Smad-3 antibody and a 1:2,000 dilution of a horseradish peroxidase–conjugated anti-rabbit antibody (both from Zymed, South San Francisco, CA).

RESULTS

Differential gene expression by scleroderma fibroblasts.

To identify mRNA that were up-regulated in SSc fibroblasts relative to normal fibroblasts, mRNA were extracted from 3 normal control subjects and 8 patients with diffuse cutaneous SSc. The SSc and normal mRNA pools were then transcribed into cDNA probes, which were then hybridized to a proprietary 250-gene matrix and TGFβ signaling gene array, resulting in 3 independent (replicate) hybridization experiments. The results of these experiments are shown in Table 1.

Table 1. Up-regulation of genes in scleroderma fibroblasts*

Gene	Hybridization intensity
	Normal fibroblasts	Scleroderma fibroblasts	Magnitude of increase
* Array studies were performed as described in Materials and Methods. Hybridization intensity values (in arbitrary units) were standardized to the expression of β-actin. TGFβ = transforming growth factor β; NS = not significant; TGFβR = TGFβ receptor; TIMP = tissue inhibitor of metalloproteinases.
Experiment 1
β-actin	3,916	3,916	–
TGFβ3	42	35	NS
TGFβRI	100	Absent	–
TGFβRII	34	Absent	–
TIMP-3	206	514	2.5
Type IV collagen	194	696	3.6
Endoglin	Absent	172	>100
Experiment 2
β-actin	1,885	1,885	–
TGFβ3	27	30	NS
TGFβRI	61	73	NS
TGFβRII	44	46	NS
TIMP-3	221	358	1.6
Type IV collagen	67	196	2.9
Endoglin	Absent	50	>50
Experiment 3
β-actin	3,681	3,681	–
TGFβ3	79	75	NS
TGFβRI	Absent	Absent	–
TGFβRII	112	107	NS
TIMP-3	295	546	1.9
Type IV collagen	88	170	1.9
Endoglin	Absent	46	>46

Overall, only a small subset of genes was expressed aberrantly in samples from SSc fibroblasts compared with control cell lines. Tissue inhibitor of metalloproteinases 3, previously identified as an overexpressed transcript in scleroderma fibroblasts (24), was induced from 1.6–2.4-fold in SSc fibroblasts (Table 1). Type IV collagen was also up-regulated in SSc fibroblasts. Conversely, TGFβRI, TGFβRII, and TGFβ3 mRNA, although readily detected, were not differentially expressed between patient and control samples. However, to our surprise, the tissue-restricted, endothelial-enriched TGFβ receptor endoglin, previously implicated in angiogenesis (18, 25–27), was significantly overexpressed in SSc fibroblasts. Similar results were obtained using differential display analysis between fibroblasts cultured from unaffected and affected areas of the skin of SSc patients (data not shown).

To confirm our gene chip data, we analyzed mRNA that had been directly prepared from fibroblasts obtained from the same subjects used for the gene chip experiments. Using semiquantitative RT-PCR for endoglin, we confirmed that SSc fibroblasts possessed elevated levels of endoglin mRNA as compared with control fibroblasts from healthy subjects (Figure 1). In contrast to the array studies, low levels of endoglin mRNA were found in normal fibroblasts, possibly reflecting the fact that the RT-PCR method was more sensitive at detecting low levels of transcripts than was the gene chip method. Owing to the previously reported role of endoglin as a known TGFβ receptor in endothelial cells and in fibroblasts, and given its known role as a specific mediator of angiogenesis in endothelial cells and its previously reported tissue-restricted pattern of expression (18, 19, 25–27), we decided to further examine the potential role of this TGFβ receptor in the SSc phenotype.

Figure 1. Enhanced endoglin mRNA expression in scleroderma fibroblasts. A, Reverse transcription–polymerase chain reaction (RT-PCR) of endoglin in fibroblasts cultured from 3 healthy control subjects (C1–C3) and from 3 patients with scleroderma (SD1, SD4, and SD7). B, The specimens were normalized to hypoxanthine guanine phosphoribosyltransferase (HPRT) using a competitor construct as described in Materials and Methods. Equal amounts of HPRT RT-PCR product are shown.

Abundant expression of endoglin protein in scleroderma fibroblasts.

To verify our results showing enhanced expression of endoglin mRNA in fibroblasts cultured from SSc lesions relative to the expression in skin from healthy control subjects, we performed Western blot analysis of protein extracts harvested from cultured dermal fibroblasts from affected areas of 8 SSc patients and 3 healthy controls. These were the same subjects used in the original array analysis. Our results showed that 7 of the 8 SSc patients possessed endoglin levels that were substantially higher than those in the healthy subjects (Figure 2A). Endoglin protein was observed in healthy individuals, consistent with the low level of endoglin mRNA detected by RT-PCR analysis, but not by gene chip analysis (compare Table 1 data with Figure 2). Intriguingly, the differences in endoglin levels between normal and SSc fibroblasts seemed to be more pronounced at the protein, rather than mRNA, level. These results suggest that there might be posttranscriptional mechanisms operating that result in enhanced endoglin protein expression in SSc dermal fibroblasts.

Figure 2. Up-regulation of endoglin protein in scleroderma fibroblasts. A, Western blots of fibroblast extracts from 3 healthy control subjects (C1–C3) and 8 patients with scleroderma (SD1–SD8). Cells were grown simultaneously under identical culture conditions. Anti-CD105 (antiendoglin) antibodies were used as described in Materials and Methods. B and C, Immunohistochemistry with fluorescein isothiocyanate–labeled antiendoglin antibodies, demonstrating increased staining of endoglin in scleroderma dermal fibroblasts (C) compared with healthy control dermal fibroblasts (B).

To confirm our Western blot data, we performed immunohistochemistry with an antiendoglin antibody on fibroblasts cultured from healthy subjects and from lesional areas of SSc patients. When normal dermal fibroblast cultures were examined with an antiendoglin antibody, several cells expressed substantial levels of endoglin. However, in general, normal fibroblasts lacked endoglin staining (Figure 2B). However, when cells from scleroderma patients were stained with the antiendoglin antibody, all cells showed a bright membrane-localizing signal consistent with a generalized overexpression of endoglin throughout the cell population (Figure 2C). Thus, the small amount of endoglin mRNA and protein observed in cultures of normal fibroblasts seemed to reflect endoglin expression in only a few cells; however, endoglin overexpression is not a general characteristic of normal dermal fibroblasts. Conversely, the overexpression of endoglin in fibroblasts from lesional skin of SSc patients reflected a characteristic of the overall cell population; both greater amounts of endoglin per cell and greater numbers of endoglin-expressing cells were observed in SSc fibroblasts compared with healthy control fibroblasts. Collectively, these results suggest that a general characteristic of SSc fibroblasts as a population, relative to their normal counterparts, is a greatly enhanced number of cells than express endoglin at their cell surface.

To verify these results, we performed fluorescence-activated cell sorter analysis with normal and SSc fibroblasts using an antiendoglin antibody and TGFβ1 (Figure 3). Three additional healthy subjects and 3 additional SSc patients were used for these analyses. When we compared the average fluorescence intensities of fibroblasts from healthy subjects with those from affected skin of SSc patients, we found, as expected, an overexpression of endoglin protein on the cell surface of SSc fibroblasts (Figure 3B). Consistent with these observations more TGFβ receptors, as measured by TGFβ1 binding to the cell surface, were observed on SSc fibroblasts relative to normal fibroblasts (Figure 3A).

Figure 3. Flow cytometry analysis of transforming growth factor β (TGFβ) receptor expression. Dermal fibroblasts from 3 healthy control subjects (normal) and 3 patients with diffuse cutaneous systemic sclerosis (SSc) (scleroderma) were analyzed as described in Materials and Methods. TGFβ receptor levels were determined by overall binding of A, avidin–fluorescein isothiocyanate–labeled TGFβ1 protein or B, antiendoglin antibody. Fluorescence intensities of stained cells were determined by FACSCalibur, and the average fluorescence intensities (AFIs) were determined by subtracting the background values (gray plots) from the experimental values (black plots). SSc fibroblasts exhibited elevated levels of cell surface endoglin compared with controls.

Correlation of endoglin protein expression with progression of SSc.

To determine whether expression of endoglin in SSc might reflect a role in the onset or maintenance of the fibrotic phenotype, we assessed whether there was a correlation between the duration of disease and greater expression of endoglin. Three SSc patients who initially showed markedly elevated expression of endoglin mRNA in our gene chip analyses underwent additional biopsies at 6–11 months after the initial mRNA analysis and detection of endoglin message. Owing to patient availability, a new patient, SD10, with characteristics similar to those of the initial patient pool (see Materials and Methods) was added to the study. (Patient SD1 was no longer available for this study.)

There was progression of the SSc between the time of the initial biopsies and the second biopsies, as indicated by elevations in the skin scores between the initial biopsies and the followup biopsies performed 6–11 months later (Figure 4). We then cultured fibroblasts obtained from the initial and followup biopsies and subjected whole cell protein extracts to Western blot analysis with an antiendoglin antibody. We found that the fibroblasts cultured from the later biopsy samples showed elevated expression of endoglin protein. Thus, the expression of endoglin in SSc fibroblasts seemed to correlate with disease progression and, hence, may represent a relatively late response in the development of fibrosis in SSc lesions.

Figure 4. Increase in levels of endoglin with disease progression and increased fibrosis in scleroderma patients. Western blots of 3 pairs of scleroderma fibroblasts (from patients SD7, SD4, and SD10) obtained at 2 different time points (shown in months [m]) during the course of the disease were performed. Blots were scanned into the computer using the NIH Image program, and the band intensity was plotted on a histogram. Bars show the relative densitometry units for the 6 fibroblast samples. Skin scores (SS) are shown across the bottom. NA = not available.

Suppression of TGFβ signaling in fibroblasts by the overexpression of endoglin.

To determine the effect of endoglin overexpression on fibroblasts, we tested the ability of an expression vector encoding endoglin to modulate the TGFβ induction of a target promoter, from the cytokine CTGF (16, 21), in transfected mouse NIH3T3 (Figure 5A) and primary human dermal (Figure 5B) fibroblasts. We found that relative to cotransfection of empty expression vector, cotransfection of an endoglin expression vector with a CTGF promoter/secreted enhanced alkaline phosphatase (SEAP) reporter construct suppressed the ability of TGFβ1 to induce CTGF promoter activity (Figures 5A and B). Previously, we demonstrated that the TGFβ induction of CTGF in fibroblasts is Smad-dependent and that transfection of Smad-3 and Smad-4 activates the CTGF promoter in the absence of added TGFβ ligand (16, 17). In the present study, we found that transfection of an expression vector encoding endoglin was not able to suppress the ability of Smad-3 and Smad-4 to activate the CTGF promoter (Figures 5A and B). These results suggest that the effect of endoglin in suppressing the action of TGFβ on fibroblasts is upstream of the TGFβ receptor–mediated phosphorylation of Smads.

Figure 5. Blocking of transforming growth factor β (TGFβ) induction of connective tissue growth factor (CTGF) promoter activity by transfection of endoglin into fibroblasts. A full-length CTGF promoter/secreted enhanced alkaline phosphatase (SEAP) reporter plasmid (16, 20) was transfected into A, mouse NIH3T3 fibroblasts and B, primary human dermal fibroblasts with empty expression vector (0.5 μg/well) or with expression vectors encoding endoglin (ENDO) or Smad-3 or Smad-4 (1 μg/well). To control for variations in transfection efficiencies between samples, cells were also transfected with a cytomegalovirus promoter/β-galactosidase reporter plasmid (0.25 μg/well). After transfection, cells were maintained in serum-free medium for 18 hours. Cells were then incubated for an additional 24 hours with or without the addition of TGFβ1 (2.5 ng/ml), as indicated. Cells transfected with Smad-3 and Smad-4 were not treated with TGFβ. Media were then assayed for SEAP activity, and cell layers were assayed for β-galactosidase activity. Values are the mean ± SEM of a representative experiment (n = 6) and were adjusted to control for transfection efficiency as previously described (16, 21).

Binding of TGFβ to type I and type II TGFβR results in phosphorylation of receptor-regulated Smads. These Smads then migrate into the nucleus, where they activate transcription (for review, see ref. 28). Thus, to support the idea that overexpression of endoglin might act to suppress TGFβ signaling at the level of TGFβR, we reasoned that overexpression of endoglin should prevent the TGFβ1-mediated activation of Smads, as visualized by their translocation and accumulation in the nuclei of cells. Owing to the greatly enhanced transfection efficiency of NIH3T3 fibroblasts relative to primary dermal fibroblasts, the former cell type was used for this experiment.

We transfected either empty expression vector or vector encoding endoglin into NIH3T3 fibroblasts. Eighteen hours after transfection, we treated cells for an additional 30 minutes with or without 2.5 ng/ml of TGFβ1. Nuclear extracts were then prepared and equal amounts of protein were subjected to SDS-PAGE and Western analysis with an anti–Smad-3 antibody (Figure 6). We found that, as expected, the addition of TGFβ1 to cells transfected with empty expression vector resulted in the accumulation of activated nuclear Smad-3. Conversely, transfection of expression vector encoding endoglin inhibited the accumulation of activated nuclear Smad-3. Collectively, our data suggest that, if overexpressed in fibroblasts, endoglin may act to suppress the signaling pathway stimulated by TGFβ1.

Figure 6. Blocking of the accumulation of nuclear Smad-3 by transfection of endoglin into fibroblasts. NIH3T3 fibroblasts were transfected with either empty expression vector or expression vector encoding endoglin. Eighteen hours after transfection, cells were incubated with or without 2.5 ng/ml transforming growth factor β1 (TGFβ1) for 30 minutes, as indicated. Nuclear extracts were then prepared, and equal amounts of protein were subjected to Western blot analysis with an anti–Smad-3 antibody as described in Materials and Methods.

DISCUSSION

The ability of TGFβ to stimulate collagen production in normal fibroblasts has led to the proposal that this factor mediates the fibrosis associated with scleroderma (2). However, the precise role of TGFβ in sustained/chronic fibrosing disorders remains a subject of controversy (for review, see ref. 4). In previous studies, we have found that, although TGFβ-induction of the profibrotic marker CTGF in fibroblasts requires Smads, constitutive overexpression of CTGF in chronic fibrosis seems to be Smad-independent (16). Furthermore, promoter activity of the TGFβ/Smad target plasminogen activator inhibitor was not elevated in SSc fibroblasts (16), suggesting that Smad-dependent TGFβ signaling is not generally activated in persistent, chronic fibrotic lesions.

In an attempt to further clarify the role of TGFβ signaling in fibrosis, we used gene chip and differential display analyses to identify genes that were up-regulated in fibrotic areas of skin from SSc patients compared with skin from healthy subjects. We found no obvious differences in TGFβ3 levels between normal and SSc fibroblasts. Similar findings have been reported using distinct experimental approaches (11, 12). In contrast to previous reports that demonstrated increases in both TGFβRI and TGFβRII mRNA in scleroderma fibroblasts (15), we found no consistent bias in the expression of these mRNA when we compared SSc with normal fibroblasts by array or differential display analysis.

Unexpectedly, we found that endoglin, a TGFβ binding protein, was consistently overexpressed in dermal cutaneous fibroblasts cultured from involved areas of skin from SSc patients relative to normal fibroblasts. Consistent with our protein expression data, we found moderate elevation of TGFβ binding to the cell surface of SSc fibroblasts relative to normal fibroblasts. Although based on relatively few patients, endoglin expression appeared to increase with disease progression, suggesting that endoglin might represent a potential marker for staging SSc.

Endoglin is a 180-kd homodimeric glycoprotein that is expressed primarily and abundantly on endothelial cells, although some expression on other cell types, including monocytes, macrophages, and fibroblasts, has been noted (29, 30). Endoglin has been repeatedly demonstrated to be a functional TGFβ binding protein in endothelial cells and fibroblasts (refs.18 and19; for a recent review, see ref. 31). Endoglin binds TGFβ1 and TGFβ3 with high affinity, but apparently does not, by itself, transduce a signal intracellularly (18, 32). Signals are processed through interactions with other members of the TGFβ receptor complex, including TGFβRI and TGFβRII. In endothelial cells, endoglin is likely to be involved with angiogenesis, since prominent endoglin expression has been demonstrated in neovascular states, including the enhanced vascularity of psoriasis (33) and angiogenesis in malignant melanoma (25). Furthermore, endoglin knockout mice show a profound failure to develop blood vessels in utero; both endothelial and smooth muscle components are affected (26). Finally, loss-of-function mutations in the human endoglin gene are associated with a vascular dysplastic syndrome, hereditary hemorrhagic telangiectasia type 1 (27).

Until this study, endoglin overexpression has not been examined in patients with fibrosis. Furthermore, the function of endoglin in fibroblasts, other than as a demonstrated TGFβ receptor (19, 31), is unknown. In this study, we found that, if overexpressed in fibroblasts, endoglin suppressed TGFβ induction of promoter activity of a target (CTGF) promoter. We further found that overexpression of endoglin blocked the accumulation of activated nuclear Smad-3 but was not able to suppress the transactivation of the CTGF promoter by Smad-3 and Smad-4. Thus, endoglin might act in fibroblasts to modulate TGFβ signaling by acting as a molecular sink that regulates or reduces the total pool of TGFβ accessible to activating, signal-transducing receptors.

Intriguingly, we found that endoglin expression increased with disease progression. The fibrosis of SSc is characterized by marked overexpression of matrix genes that are normally the target of TGFβ signaling. However, SSc fibroblasts are less responsive to TGFβ in terms of matrix induction than their normal counterparts are (refs.11–14; for review, see ref. 34). Our results suggest that the up-regulation of endoglin in SSc fibroblasts may represent a normal biologic compensatory mechanism that regulates TGFβ signaling and, in the context of scleroderma, may be used in alleviating the fibrotic SSc phenotype by suppressing further induction by TGFβ of profibrotic matrix genes in SSc fibroblasts.