An increased transforming growth factor β receptor type I:Type II ratio contributes to elevated collagen protein synthesis that is resistant to inhibition via a kinase-deficient transforming growth factor β receptor type II in scleroderma

Abstract

Objective

Aberrant transforming growth factor β (TGFβ) signaling has been implicated in the pathogenesis of scleroderma (systemic sclerosis [SSc]), but the contribution of specific components in this pathway to SSc fibroblast phenotype remains unclear. This study was undertaken to delineate the role of TGFβ receptor type I (TGFβRI) and TGFβRII in collagen overexpression by SSc fibroblasts.

Methods

Primary dermal fibroblasts from SSc patients and healthy adults were studied (n = 10 matched pairs). Adenoviral vectors were generated for TGFβRI (AdTGFβRI), TGFβRII (AdTGFβRII), and kinase-deficient TGFβRII (AdΔkRII). TGFβRI basal protein levels were analyzed by ³⁵S-methionine labeling/immunoprecipitation and immunohistochemistry. Type I collagen and TGFβRII basal protein levels were analyzed by Western blot and newly secreted collagen by ³H-proline incorporation assay.

Results

Analysis of endogenous TGFβRI and TGFβRII protein levels revealed that SSc TGFβRI levels were increased 1.7-fold (P = 0.008; n = 7) compared with levels in healthy controls, while TGFβRII levels were decreased by 30% (P = 0.03; n = 7). This increased TGFβRI:TGFβRII ratio correlated with SSc collagen overexpression. To determine the consequences of altered TGFβRI:TGFβRII ratio on collagen expression, healthy fibroblasts were transduced with AdTGFβRI or AdTGFβRII. Forced expression of TGFβRI in the range corresponding to elevated SSc TGFβRI levels increased basal collagen expression in a dose-dependent manner, while similar TGFβRII overexpression had no effect, although transduction of fibroblasts at higher multiplicities of infection led to a marked reduction of basal collagen levels. Blockade of TGFβ signaling via AdΔkRII resulted in ∼50% inhibition of basal collagen levels in healthy fibroblasts and in 5 of 9 SSc cell lines. A subset of SSc fibroblasts (4 of 9 cell lines) was resistant to this treatment. SSc fibroblasts with the highest levels of TGFβRI were the least responsive to collagen inhibition via ΔkRII.

Conclusion

This study indicates that an increased TGFβRI:TGFβRII ratio may underlie aberrant TGFβ signaling in SSc and contribute to elevated basal collagen production, which is insensitive to TGFβ signaling blockade via ΔkRII.

Scleroderma (systemic sclerosis [SSc]) is characterized by excessive deposition of extracellular matrix (ECM) proteins in the skin and internal organs (1). Numerous studies have implicated transforming growth factor β (TGFβ) signaling in the pathogenesis of SSc. TGFβ signals through heteromeric receptors and downstream effectors termed SMADs. The signal cascade is initiated by ligand binding to TGFβ receptor type II (TGFβRII), a serine/threonine kinase, which then heterodimerizes with and phosphorylates a distinct serine/threonine kinase, TGFβRI. The activated TGFβRI in turn phosphorylates pathway-specific SMAD proteins (2).

There is increasing evidence that alterations in the levels of receptors and/or SMADs may contribute to abnormal ECM deposition in SSc. Elevated levels of TGFβRI and TGFβRII messenger RNA (mRNA) have been reported in cultured SSc fibroblasts and correlated with elevated collagen mRNA levels, suggesting the existence of an autocrine TGFβ signaling loop operating in SSc fibroblasts in vitro (3). This possibility was supported by subsequent studies, which confirmed elevated levels of TGFβ receptor mRNA and protein expression in SSc fibroblasts (4). Furthermore, Ihn et al (5) have shown that the blockade of TGFβ signaling via neutralizing antibodies or antisense nucleotides against TGFβ ligand reduced elevated collagen expression by SSc fibroblasts. Consistent with findings of these in vitro studies, elevated levels of TGFβ receptors were observed in vivo in localized scleroderma (6) and SSc (7). However, the role of the elevated TGFβ receptor levels in vivo remains uncertain, as illustrated by a recent study by Denton et al (8), in which fibroblast-specific expression of a kinase-deficient TGFβRII in transgenic mice unexpectedly resulted in skin and lung fibrosis.

Recently, Mori et al (9) demonstrated constitutive activation of the SMAD2/3 pathway in SSc fibroblasts. Scleroderma fibroblasts displayed elevated nuclear localization of SMAD2/3 and enhanced phosphorylation of SMAD2/3 in the absence of TGFβ. There is a discrepancy, however, between the results obtained by Mori et al (9) and those of an earlier study by Dong and colleagues (10). While elevation of SMAD3 in SSc fibroblasts was observed in both studies, Dong et al reported a marked reduction of SMAD7 levels in SSc cells, which was not reproduced by Mori et al (9). On the other hand, the protein levels of the TGFβ receptors did not appear to differ in SSc and control cell lines utilized by Dong et al. Furthermore, in contrast to the results reported by Ihn et al (5), activation of the SMAD pathway reported by Mori et al was shown to be independent of TGFβ ligand/TGFβRII signaling, since anti-TGFβ antibodies and the kinase-deficient, dominant-negative TGFβRII, respectively, had no effect on SMAD3 nuclear localization. Thus, while there is evidence that the TGFβ signaling pathway is altered in SSc fibroblasts, there is no consensus on specific alterations and their role in SSc cell function.

Given the existing contradictory findings regarding the role of TGFβ receptor/SMAD signaling in SSc, this study was undertaken to further clarify the role of this pathway in the pathogenesis of SSc. We examined the role of TGFβRI and TGFβRII in elevated collagen production by SSc fibroblasts. Our study demonstrates that up-regulation of TGFβRI in SSc fibroblasts may directly contribute to elevated collagen production by these cells.

PATIENTS AND METHODS

Patients and tissue specimens for in vitro studies.

Upon informed consent and in compliance with the Institutional Review Board for Human Studies, skin biopsy specimens were obtained from the affected areas (dorsal forearm) of 10 patients with diffuse cutaneous SSc (4 women and 6 men; median age 45 years, range 30–62 years). The duration of skin thickening was 4–6 months for the majority of patients tested (n = 7); however, 3 patients had disease durations ranging from 11 months to 3 years. All patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for SSc (11) and had not undergone any treatment for SSc at the time of biopsy. Dermal fibroblasts were cultured from the biopsy specimens as previously described (12). Biopsy specimens from healthy donors matched with each SSc patient for age, sex, and race were processed in parallel. Fibroblasts were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and utilized in passages 2–5 for all experiments. Additional SSc patient and normal donor biopsy specimens were analyzed for immunohistochemical studies.

Immunohistochemistry.

Skin biopsy specimens were obtained from 12 patients with diffuse SSc (5 men and 7 women; median age 54 years, range 33–73 years) and 5 healthy donors. Skin biopsy specimens were fixed in neutral buffered formalin and embedded in paraffin. Five micrometer–thick sections were deparaffinized with xylene and rehydrated with graded series of ethyl alcohol and phosphate buffered saline. Immunohistochemical staining of TGFβRI was performed using a rabbit polyclonal antibody against TGFβRI (v-22; Santa Cruz Biotechnology, Santa Cruz, CA) diluted 1:200 overnight at 4°C. The immunoreactivity was detected using a biotinylated secondary antibody (Vectastain ABC kit; Vector, Burlingame, CA) and diaminobenzidine as a substrate (Vector) according to the manufacturer's recommendations. The sections were then counterstained with hematoxylin. Dermal fibroblasts (TGFβRI positive versus TGFβRI negative) were counted from 6–8 randomly selected fields of vision (40× magnification). The percentage of TGFβRI-positive fibroblasts was calculated from the total number of fibroblasts counted.

Adenoviral constructs.

Replication-incompetent adenoviral vectors for full-length TGFβRII (AdTGFβRII), kinase-deficient TGFβRII (AdΔkRII), and green fluorescent protein (GFP) were generated as described previously (13). Briefly, the complementary DNA (cDNA) encoding TGFβRII (provided by Dr. R. Weinberg, Whitehead Institute for Biomedical Research, Cambridge, MA) and ΔkRII (provided by Dr. M. Schneider, Baylor College of Medicine, Houston, TX) were cloned into adenoviral shuttle vectors and linearized for in vitro ligation with the adenoviral backbone construct lacking the E1, E3, and E4 regions of the adenovirus genome. The vectors constructed for this study express GFP (AdGFP) driven by a single cytomegalovirus (CMV) promoter/enhancer, or GFP and the gene of interest under the control of two separate CMV promoter/enhancers. An adenoviral vector expressing rat full-length TGFβRI (AdTGFβRI) was generated using the method described by He et al (14). Briefly, the cDNA encoding TGFβRI (provided by Dr. Xin-Hua Feng, Baylor College of Medicine) was cloned in the shuttle vector pAdTRACK-CMV, which contains a GFP expression cassette driven by a separate CMV promoter, and was used to generate recombinant adenoviruses. An adenovirus expressing GFP alone was generated via the same method for use as a control vector.

Western blotting for TGFβRI, TGFβRII, and type I collagen.

Confluent SSc and healthy control fibroblasts were lysed in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris HCl [pH 8.0], 150 mM NaCl, 0.02% sodium azide, 0.1% sodium dodecyl sulfate [SDS], 1% Nonidet P40, 0.5% sodium deoxycholate, 1 mM phenylmethylsulfonyl fluoride). Protein concentration was quantified using the BCA Protein Assay kit (Pierce, Rockford, IL). Fifty micrograms of protein was separated via SDS–polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane (Bio-Rad, Hercules, CA) which was then blocked at room temperature using 3% milk/Tris buffered saline–Tween (TBST) for 1 hour. The blots were probed overnight with a 1:1,000 dilution of primary antibody (rabbit polyclonal antibody directed against the amino terminal of TGFβRII; generated at the Medical University of South Carolina antibody facility [15]) or goat anti–type I collagen antibody (Southern Biotechnology, Birmingham, AL) in 3% milk/TBST. Following washes with TBST, blots were incubated with horseradish peroxidase–conjugated anti-rabbit or anti-goat IgG secondary antibody. As a control for equal protein loading, membranes were stripped and reprobed for β-actin using a monoclonal antibody to β-actin (Sigma, St. Louis, MO).

Protein levels were visualized using enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ) and quantitated using NIH Image densitometry software (National Institutes of Health, Bethesda, MD). Rabbit polyclonal antibodies to TGFβRI (v-22) and TGFβRII (c-16; Santa Cruz Biotechnology) were used to visualize the expression levels of these proteins at different viral doses.

³⁵S-methionine labeling/TGFβRI immunoprecipitation.

Confluent SSc and normal dermal fibroblasts were incubated for 30 minutes in methionine-free medium followed by metabolic labeling with 200 μCi/ml ³⁵S-methionine for 3 hours prior to collection of the cells. Cells were lysed in RIPA buffer, and protein concentration was determined using the BCA Protein Assay kit. Protein (150 μg) was precleared for 1 hour with 1/10 volume protein G–Sepharose beads (Amersham Pharmacia Biotech) at 4°C. Supernatants were collected after centrifugation and incubated for 18 hours at 4°C with a 1:100 dilution of TGFβRI rabbit polyclonal antibody (v-22). Samples were again incubated with protein G–Sepharose (1/10 volume) for 1 hour at 4°C. After centrifugation, the immunoprecipitated pellet was taken through a series of 5 washes (20 mM Tris HCl [pH 7.4], 500 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.2% SDS) and resuspended in 25 μl of 2× SDS sample buffer containing dithiothreitol. Samples were boiled and electrophoresed on a 10% SDS–polyacrylamide gel. The gel was soaked in fixing solution overnight followed by fluorography with Fluoro-Hance (RPI, Mount Prospect, IL). Protein levels were quantified using a phosphorimager.

Analysis of TGFβRI and TGFβRII by reverse transcriptase–polymerase chain reaction (RT-PCR).

Confluent healthy dermal fibroblasts were transduced with varying multiplicities of infection (MOIs) of AdTGFβRI and AdTGFβRII. Forty-eight hours later, total RNA was extracted using TRIzol reagent (Gibco, Grand Island, NY). Total RNA (5 μg) was treated with 2 units DNase I (bovine pancreas; Sigma) for 15 minutes at room temperature in an 18-μl volume containing 1× PCR buffer and 2 mM MgCl₂, followed by inactivation of DNase I with 2 μl of 25 mM EDTA at 65°C for 15 minutes. Random hexamer primers (2 μl; Promega, Madison, WI) were added and annealed to the RNA according to the manufacturer's protocol. Complementary DNA was synthesized in a 50-μl reaction volume containing 5 μg of total RNA and 50 units Moloney murine leukemia virus RT (Promega) by incubating the tubes at 42°C for 45 minutes. PCR amplification was carried out in 25 μl of reaction mixture with 2.0 μl of cDNA, 0.5 mM of each primer, and TaKaRa Taq DNA Polymerase (TaKaRa Shuzo, Shiga, Japan) according to the manufacturer's instructions. The sequences of primers used for TGFβRI and TGFβRII were previously reported by McCaffrey et al (16).

Northern blot analysis and ³H-proline incorporation assay.

Radiolabeled probe for TGFβRII was prepared as previously described (15). Secreted collagenous proteins were measured in SSc and healthy control fibroblasts as previously described (12).

Statistical analysis.

Wilcoxon signed rank and Mann-Whitney tests were performed to determine statistical significance using InStat statistical analysis software (University of Reading, Reading, UK). To determine the significance of the correlation coefficient for SSc TGFβRI protein levels and SSc collagen levels after treatment with AdΔkRII, the Spearman rank correlation coefficient test was performed. Data are expressed as the mean ± SEM. P values less than or equal to 0.05 were considered significant.

RESULTS

Alteration of TGFβRI and TGFβRII protein levels in SSc fibroblasts.

Dermal SSc and healthy control fibroblasts were analyzed for TGFβRI and TGFβRII protein expression to determine whether the TGFβ pathway is altered at the level of its transmembrane signaling receptors. Expression of TGFβRI was determined in 10 pairs of SSc and healthy control fibroblasts (matched for age, sex, and race) by metabolic labeling with ³⁵S-methionine followed by immunoprecipitation with an anti-TGFβRI antibody. SSc fibroblasts from the majority of patient biopsy specimens (n = 7) expressed elevated levels of TGFβRI protein, with an average 1.71-fold increase over the levels in healthy control fibroblasts (P = 0.008) (Figure 1A).

Figure 1. Altered protein levels of transforming growth factor β receptor type I (TGFβRI [RI]) and TGFβRII (RII) in fibroblasts from patients with scleroderma (systemic sclerosis [SSc]). Matched pairs of SSc and normal control (NS) fibroblast specimens (n = 10) were analyzed for basal levels of TGFβRI by ³⁵S-methionine labeling/immunoprecipitation and for TGFβRII by Western blotting. Collagen protein expression was determined by ³H-proline incorporation assay (n = 9 pairs of specimens). A, TGFβRI and TGFβRII protein levels from all tested pairs of SSc and healthy control fibroblasts (n = 10). B, Graphic representation of the relative TGFβRI, TGFβRII, and collagen protein levels in SSc fibroblast cell lines compared with levels in normal control fibroblasts (arbitrarily set at 1). Bars show the means. C, TGFβRI:TGFβRII ratios and basal collagen levels in the SSc cell lines grouped according to TGFβRI:TGFβRII ratio <1 versus >1. Values are the mean and SEM.

Expression of TGFβRII was determined in the same pairs by Western blotting. In contrast to TGFβRI, levels of TGFβRII were modestly decreased (30%) in the majority of SSc cell lines (P = 0.031; n = 7) (Figure 1A). However, 2 SSc cell lines (SSc 8 and 9) expressed increased TGFβRII protein levels, while 1 SSc cell line (SSc 10) exhibited no change, resulting in an average TGFβRII level for all 10 SSc cell lines tested that was not significantly different from normal fibroblast TGFβRII levels. Together, these data indicate that the majority of SSc fibroblasts exhibit an increase in the ratio of TGFβRI to TGFβRII, suggesting a possible alteration in the TGFβ signal transduction pathway in these cells. Furthermore, this alteration in the TGFβ receptor ratio correlated with a significant (1.86-fold) increase in basal collagen levels (P = 0.008) as measured by ³H-proline incorporation assay (Figure 1B). The SSc cell lines (n = 6) with an increased TGFβRI:TGFβRII ratio (>1 compared with that in normal controls) showed higher basal collagen levels than the SSc cell lines (n = 3) with a TGFβRI:TGFβRII ratio <1 (Figure 1C).

In order to confirm the heterogeneous expression of TGFβRI among different SSc cell lines in vivo, immunostaining for TGFβRI was carried out in specimens from 12 SSc patients and 5 normal controls. Representative pictures from 1 SSc skin section and 1 normal skin section, reflecting the average trend of TGFβRI expression, are shown in Figure 2A. TGFβRI-positive fibroblasts and TGFβRI-negative fibroblasts were counted in the dermis of SSc patients and normal controls. This analysis revealed marked heterogeneity of TGFβRI expression among SSc skin sections from the various patients; however, the mean ± SEM percentage of TGFβRI-positive fibroblasts in SSc skin sections was higher than that in skin sections from normal controls (46 ± 5.6% versus 27 ± 6%) (Figure 2B).

Figure 2. A, Representative immunostaining for TGFβRI, comparing TGFβRI-positive fibroblasts (brown staining) (large arrows) with TGFβRI-negative fibroblasts (purple nuclear staining only) (small arrows) in SSc and normal skin sections (original magnification × 400). B, Graphic representation of the percentage of TGFβRI-positive fibroblasts among all tested SSc and normal healthy skin sections. At least 100 fibroblasts were counted for each specimen. Bars show the means. The mean ± SEM percentage of TGFβRI-positive fibroblasts in SSc skin sections was higher than that in skin sections from normal controls (46 ± 5.6% versus 27 ± 6%). See Figure 1 for definitions.

Distinct effects of an elevated TGFβRI:TGFβRII ratio on basal collagen levels in normal adult fibroblasts.

In order to delineate the functional significance of the elevated TGFβRI levels in the regulation of collagen expression, we utilized an adenovirus encoding full-length TGFβRI (AdTGFβRI). To establish experimental conditions that mimic the increased expression levels of TGFβRI in SSc fibroblasts, adult dermal fibroblasts were transduced with increasing concentrations (MOIs 2–50) of AdTGFβRI. The expression levels of TGFβRI were monitored by Western blotting (Figure 3A) and RT-PCR and normalized to GAPDH control (Figures 3B and C). The results were confirmed by Northern blot analysis (data not shown). These experiments indicated a dose-dependent increase of TGFβRI levels, with a 2-fold increase over basal levels achieved at MOI 25.

Figure 3. Dose-dependent increase in collagen protein levels following transduction of human dermal fibroblasts with adenoviral vector expressing rat full-length TGFβRI (AdTGFβRI). A, Representative blots showing TGFβRI protein levels after transduction with increasing multiplicities of infection (MOIs) of AdTGFβRI. B, Representative reverse transcriptase–polymerase chain reaction analysis of TGFβRI mRNA levels after transduction with increasing MOIs of AdTGFβRI. C, Mean and SEM TGFβRI mRNA levels in cells transduced with AdTGFβRI (n = 2 pairs of specimens). D, Representative Western blots of type I collagen protein in dermal fibroblasts transduced with increasing MOIs of AdTGFβRI (RI) or with an adenovirus expressing green fluorescent protein alone (AdGFP) as a control vector. E, Type I collagen protein levels in fibroblasts transduced with AdTGFβRI or AdGFP. Results are the mean and SEM from 3 independent experiments. Stimulation of type I collagen with AdTGFβRI at MOIs 25 and 50 was significantly increased (∗∗ = P < 0.05 versus stimulation with AdGFP). See Figure 1 for other definitions.

To determine the effect of increased TGFβRI expression on collagen expression, confluent adult dermal fibroblasts were transduced with AdTGFβRI or the control adenovirus (AdGFP) at increasing MOIs (ranging from 2 to 75). Collagen production was measured 48 hours posttransduction using Western blotting. As shown in Figure 3D, basal type I collagen protein increased in a dose-dependent manner in response to increasing TGFβRI levels. Maximal response was observed at MOI 50, which corresponds to a 3-fold increase in TGFβRI over basal expression levels. There was no appreciable effect of control AdGFP virus on collagen production at the MOIs used in this range. Interestingly, higher levels of TGFβRI expression (MOI 75) were not stimulatory (Figures 3D and E). Furthermore, MOIs ≥100 were consistently inhibitory to collagen protein levels (data not shown). These results suggest that the elevated levels of TGFβRI contribute to increased collagen production by fibroblasts. However, the stimulatory effect on collagen production is observed in only a narrow range of TGFβRI overexpression not exceeding a 3-fold increase above basal TGFβRI levels, which corresponds to the level of TGFβRI overexpression demonstrated in SSc fibroblasts.

We next examined the effects of TGFβRII levels on collagen production in dermal fibroblasts. To establish optimal experimental conditions, normal adult fibroblasts were transduced with AdTGFβRII (MOIs 2–50). TGFβRII expression levels at increasing MOIs were determined by Western blotting (Figure 4A) and RT-PCR (Figures 4B and C). The results were confirmed by Northern blot analysis (data not shown). A dose-dependent increase in TGFβRII expression levels was observed with increasing MOIs, with a 2-fold increase over basal TGFβRII levels achieved at MOI 25 (Figure 4C).

Figure 4. Lack of increase in collagen protein levels following transduction of human dermal fibroblasts with AdTGFβRII. A, Representative blots showing TGFβRII protein levels after transduction with increasing MOIs of AdTGFβRII. B, Representative reverse transcriptase–polymerase chain reaction analysis of TGFβRII mRNA levels after transduction with increasing MOIs of AdTGFβRII. C, Mean and SEM TGFβRII mRNA levels in cells transduced with AdTGFβRII (n = 2 pairs of specimens). D, Representative Western blots of type I collagen protein in dermal fibroblasts transduced with increasing MOIs of AdTGFβRII (RII) or AdGFP. E, Type I collagen protein levels in fibroblasts transduced with AdTGFβRII or AdGFP. Results are the mean and SEM from 3 independent experiments. See Figures 1 and 3 for definitions.

The effect of AdTGFβRII (MOIs 10–100) on type I collagen was then examined. In contrast to the stimulatory effect of AdTGFβRI, low concentrations of AdTGFβRII (MOIs 10–25) did not have any substantial effect on basal type I collagen levels (Figures 4D and E). Transduction of fibroblasts at higher MOIs (≥50) led to a marked reduction of basal collagen levels. Thus, increased expression of TGFβRII at levels comparable to TGFβRI overexpression did not affect collagen production. Overexpression of either receptor type at higher levels led to decreased collagen production. The significance of these inhibitory effects is currently unknown; however, since such high levels of receptor subunit expression are not observed in SSc fibroblasts, these results are not relevant to the present study.

Collagen protein synthesis in dermal fibroblasts from a subset of SSc patients is resistant to blockade of TGFβ signaling via AdΔkRII.

Given the above findings, as well as our previous finding that autocrine TGFβ signaling is important for collagen promoter activity by normal dermal fibroblasts (3), our objective was to abrogate autocrine TGFβ signaling and analyze the effect on basal collagen protein synthesis both in SSc and in healthy control fibroblasts. To effectively inhibit TGFβ signal transduction in virtually 100% of fibroblasts, we used an adenoviral vector containing a kinase-deficient TGFβRII (AdΔkRII). This ΔkRII construct, which consists of both the extracellular and transmembrane portions of TGFβRII without the serine/threonine kinase cytoplasmic domain, was previously established as a dominant-negative inhibitor of the TGFβ signal transduction pathway and was used by Mori et al to block the TGFβ pathway (9).

Fibroblasts transduced with increasing concentrations of AdΔkRII expressed decreasing collagen α2(I) steady-state mRNA levels in a dose-dependent manner, as determined by Northern blot analysis (Figure 5A). Levels of collagen mRNA from cells transduced with equal concentrations of the control virus (AdGFP) were arbitrarily set at 1. Significant inhibition of collagen α2(I) mRNA levels was achieved with MOI 200; therefore, this condition was used for the subsequent experiments.

Figure 5. Differential response of SSc fibroblasts to blockade of TGFβ signaling via an adenoviral vector for kinase-deficient TGFβRII (AdΔkRII). A, AdΔkRII decreases collagen α2(I) mRNA levels in a dose-dependent manner. Fibroblasts were transduced at the indicated MOIs for 72 hours followed by Northern blot analysis of ΔkRII and collagen mRNA. Shown are a representative Northern blot and corresponding graph for collagen mRNA. Values are mean collagen α2(I):GAPDH ratios (n = 2 experiments). B, SSc and healthy fibroblast pairs were transduced with AdGFP or AdΔkRII (MOI 200) for 72 hours in the presence or absence of exogenous TGFβ (2.5 ng/ml) for the final 48 hours. Levels of newly synthesized collagenous proteins in conditioned medium were determined via the ³H-proline assay. Top, Representative matched pair of healthy fibroblast (lanes 1–4) and SSc (lanes 5–8) specimens equally responsive to AdΔkRII. Bottom, Representative SSc cell line nonresponsive to basal extracellular matrix inhibition by AdΔkRII (lanes 1–4). C, Graphic representation of collagen protein levels after transduction with AdΔkRII relative to levels in cells transduced with AdGFP (arbitrarily set at 100%). Each triangle represents the percentage of collagen remaining for the individual cell lines derived from each SSc or normal biopsy specimen. Bars show the mean levels of collagen remaining after AdΔkRII transduction. Basal collagen levels were significantly inhibited both in healthy control cells and in SSc cell lines (P = 0.002 and P = 0.004, respectively, versus cells transduced with AdGFP). See Figures 1 and 3 for other definitions.

The matched pairs of SSc and healthy fibroblasts were transduced with equal concentrations of the control AdGFP or AdΔkRII, and basal levels of newly synthesized collagenous proteins were analyzed after 72 hours using the ³H-proline incorporation assay. Because the TGFβ pathway has the potential to affect collagen directly via transcription and mRNA stability as well as indirectly through modulating tissue inhibitor of metalloproteinases/matrix metalloproteinase expression, we chose to look at the cumulative effect of blocking these actions by analyzing collagenous proteins at the protein level.

Healthy control fibroblasts responded with a significant inhibition in basal levels of collagen and fibronectin proteins (Figure 5B, top panel). There was potent induction of collagen after 48 hours of stimulation with TGFβ (2.5 ng/ml) in cells transduced with the control AdGFP (Figure 5B, top panel) and inhibition of this stimulation in cells transduced with an equal concentration of AdΔkRII (Figure 5B, top panel, lane 4). Collagen levels were significantly inhibited, by an average of 44%, in healthy control cells (P = 0.002) (Figure 5C). A subset of SSc cell lines (5 of 9) responded similarly, with a 53% decrease in basal collagen levels upon treatment with AdΔkRII. A representative responsive SSc cell line is shown in Figure 5B (top panel). Collectively, SSc basal collagen levels in all 9 cell lines were less inhibited by AdΔkRII, with a 33% average decrease (P = 0.004), in contrast to the 44% average decrease in healthy control cells. However, these levels of inhibition were not significantly different from each other (Figure 5C).

In contrast to the significant inhibition of basal collagen in these SSc cell lines and in normal fibroblasts, a subset of SSc fibroblast cell lines (4 of 9) were nonresponsive to AdΔkRII-mediated inhibition of basal collagen levels, showing a 7% average decrease after transduction with AdΔkRII (Figure 5C). A gel representative of the resistant SSc cell lines is shown in the bottom panel of Figure 5B. The inability of AdΔkRII to down-regulate basal collagen levels was not due to inefficient viral transduction in these SSc cell lines, since both SSc and normal cells expressed similar levels of ΔkRII mRNA as determined by Northern blot analysis and similar levels of GFP expression upon visualization using fluorescence microscopy (results not shown). All SSc and normal dermal fibroblasts responded similarly to exogenous TGFβ (2.5 ng/ml), with an ∼2-fold induction of fibronectin and collagen in the presence of AdGFP. Furthermore, this induction of ECM synthesis by exogenous TGFβ was inhibited in both cell types in the presence of AdΔkRII. However, SSc fibroblasts resistant to the decrease in basal collagen by AdΔkRII were also slightly less sensitive to AdΔkRII inhibition of exogenous TGFβ stimulation compared with healthy control and responsive SSc cell lines.

Correlation of elevated SSc TGFβRI levels with resistance to AdΔkRII.

As indicated in Figure 6, there was a significant negative correlation (r = −0.73, P = 0.038) between SSc TGFβRI protein levels and responsiveness to AdΔkRII. SSc cell lines with the highest levels of TGFβRI protein expression were the least responsive to inhibition of basal collagen levels by AdΔkRII. Conversely, those SSc cell lines with lower TGFβRI levels similar to the levels expressed in normal cell lines had the greatest inhibition of basal collagen levels after transduction with AdΔkRII. Thus, overexpression of TGFβRI in SSc fibroblasts may render these cells refractory to inhibition of autocrine TGFβ signaling via the ΔkRII construct.

Figure 6. Significant correlation between elevated SSc TGFβRI protein levels and decreased responsiveness to inhibition of basal collagen levels by an adenoviral vector for kinase-deficient TGFβRII (AdΔkRII). Relative TGFβRI protein levels are plotted versus the percentage of inhibition of basal collagen levels after transduction with AdΔkRII. SSc cell lines with the highest TGFβRI levels were most resistant to inhibition of basal collagen levels by AdΔkRII. See Figure 1 for other definitions.

DISCUSSION

SSc lesional fibroblasts produce elevated levels of ECM proteins in vivo and maintain an activated phenotype in vitro (17). Alterations in the TGFβ signaling pathway have been previously proposed to contribute to this activated phenotype (3, 5). Results of the present study further strengthen this theory. We demonstrate that the majority of SSc fibroblasts express increased protein levels of TGFβRI, which correlate with increased collagen synthesis. Significantly, forced expression of TGFβRI in healthy dermal fibroblasts to mimic the elevated TGFβRI levels observed in SSc fibroblasts leads to a corresponding increase in collagen protein expression. A similar level of forced TGFβRII expression did not significantly affect collagen protein levels. SSc fibroblasts examined in the current study show variable expression of TGFβRII protein, with the majority of SSc cell lines exhibiting a decrease in TGFβRII compared with healthy control cells, suggesting that TGFβRII levels may be less critical for collagen production in SSc fibroblasts.

Recent studies of TGFβ receptor expression in SSc dermal fibroblasts have yielded conflicting results. Ihn et al reported increased TGFβRI and TGFβRII protein levels in SSc fibroblasts in vitro (5), whereas Dong et al found no differences in receptor protein levels from fibroblasts cultured in vitro or from SSc skin sections in vivo compared with levels in normal control cells (10). The study reported by Ihn et al analyzed patients with <2 years of disease, whereas the majority of patients (n = 7) in our study had disease of much more recent onset (duration of ∼4–6 months). Disease durations of the patients analyzed by Dong et al were not indicated. Taken together, these studies may suggest that TGFβ receptor levels expressed by SSc fibroblasts may differ between the early proliferative and later sclerotic stages of the disease. Such alterations in TGFβ receptor levels could serve to modulate the fibroblast behavior in these respective stages of SSc disease progression. Significantly, a marked heterogeneity with regard to TGFβRI protein expression in scleroderma skin biopsy specimens was observed in our study, which may explain discrepancies among previous reports (7, 10).

Conflicting findings also exist with regard to TGFβ receptor expression in other fibrotic diseases, including atherosclerosis, liver fibrosis, and renal fibrosis. The decreased TGFβRII:TGFβRI ratio (or increased TGFβRI:TGFβRII ratio) demonstrated by our study has also been reported in atherosclerosis (18) and in patients with liver cirrhosis in vivo (19). However, increased levels of both receptor types or exclusively elevated levels of TGFβRII have also been reported. Bobik et al (20) demonstrated elevated TGFβRI and TGFβRII in atherosclerosis, in contrast to findings by McCaffrey et al (18). A similar shift in the TGFβRII:TGFβRI ratio is reported in fibroblasts during wound healing. While TGFβRII levels are more potently induced earlier during the initial proliferative stages of wound healing and subsequently decline, TGFβRI levels increase later and remain elevated at the end of the proliferative stage with the onset of the synthetic phase, during which matrix deposition increases (21). Thus, TGFβ receptor levels may be differentially modulated during wound healing and fibrosis, thereby altering the TGFβRII:TGFβRI ratio and subsequently modulating the proliferative versus the matrix-inducing effects of TGFβ.

Interestingly, human dermal fibroblasts cultured under hypoxic conditions exhibit a decreased TGFβRII:TGFβRI ratio associated with defects in TGFβ-mediated chemotaxis (22). Therefore, the increased TGFβRI:TGFβRII ratio that we observed in the majority of SSc fibroblasts (from 7 of 10 patients) derived from patients with early disease may be a manifestation of the hypoxic environment that is present in SSc skin due to the vascular damage associated with this disease (23).

In order to determine whether autocrine TGFβ contributes to the ECM overproduction by SSc fibroblasts, we attempted to abrogate this signaling pathway utilizing an adenoviral vector expressing a kinase-deficient TGFβRII construct (ΔkRII). Overexpression of this construct has been widely used as the method of choice because it acts in a dominant-negative manner to inhibit the initiation of the TGFβ signaling cascade (24). Given the proven efficiency of this construct in both in vitro and in vivo systems, the ΔkRII construct was chosen at the onset of this study as an effective method to abrogate autocrine TGFβ signaling and thereby determine the involvement of TGFβ in basal ECM protein synthesis by healthy control and SSc dermal fibroblasts.

ECM synthesis was significantly decreased in healthy control fibroblasts transduced with the ΔkRII adenoviral vector, indicating that autocrine TGFβ signaling is responsible for ∼50% of basal fibronectin and collagen protein synthesis in normal dermal fibroblasts cultured in vitro. Interestingly, the same effective inhibition was not observed in all SSc fibroblast cell lines, which were transduced with the same concentration of vector. SSc and normal fibroblasts expressed similar levels of ΔkRII mRNA after transduction and appeared to be transduced by the AdGFP and AdΔkRII viral vectors with the same efficiency, as evidenced by similar expression levels of GFP achieved with these vectors in both cell types. In addition, exogenous TGFβ stimulation of collagen was inhibited in both cell types in the presence of AdΔkRII, indicating that the ΔkRII construct was functionally efficient in blocking exogenous TGFβ signal transduction. Therefore, this discrepancy in SSc responsiveness cannot be explained by inefficient expression of the ΔkRII construct in these cells.

Ihn et al reported that inhibition of TGFβ1 via neutralizing antibody or antisense oligonucleotide significantly down-regulated collagen promoter activity and steady-state mRNA levels in all SSc patient fibroblasts tested, suggesting that autocrine TGFβ is important in collagen overexpression in SSc (5). Despite the fact that ΔkRII did not inhibit basal collagen levels in all SSc fibroblasts tested in our study, this does not rule out the involvement of autocrine TGFβ signaling in SSc ECM expression. One possible explanation for the diminished responsiveness of SSc basal collagen levels in the presence of AdΔkRII is that a component of the TGFβ pathway downstream from TGFβRII (e.g., TGFβRI or SMAD3) is constitutively activated and therefore independent of TGFβRII signaling. The recent study by Mori et al, demonstrating that activation of the SMAD2/3 pathway in SSc fibroblasts was resistant to blockade of TGFβ signaling via AdΔkRII, supports this possibility (9). In our study, there was a significant correlation between the expression level of TGFβRI and responsiveness to ΔkRII. SSc cell lines expressing the highest levels of TGFβRI were also the most resistant to inhibition of collagen by AdΔkRII, which may suggest that an unconventional autocrine TGFβ signaling pathway via TGFβRI overexpression functions in SSc fibroblasts and may play a role in pathologic collagen expression by these cells.

In conclusion, we have demonstrated that a subgroup of SSc dermal fibroblasts manifests alterations in levels of TGFβ serine/threonine kinase receptors as well as in the response to blockade of the TGFβ pathway via a kinase-deficient TGFβRII. Furthermore, our data strongly implicate TGFβRI in ECM overproduction in at least some of the SSc fibroblasts. The mechanism responsible for the elevated levels of TGFβRI and subsequent activation of the ligand-independent TGFβ signaling pathway is not known; however, our findings are consistent with constitutive activation of SMAD signaling that occurs in SSc fibroblasts (9). In support of this view, recent studies have linked the levels of TGFβRI to the duration of SMAD2/3 accumulation in the nucleus (25, 26). Additionally, an intriguing possibility linking alterations in SMAD ubiquitination regulatory factor (SMURF) to ligand-independent activation of TGFβ/SMAD signaling in SSc has been proposed and needs to be experimentally tested (27). The mechanisms involved in aberrant TGFβ signaling in SSc merit further investigation, since, as shown here, blocking of TGFβRII may not be a feasible goal for antifibrotic therapy in all SSc patients.