Primary fibroblast cultures were established by explantation from neonatal foreskin biopsies, or from skin biopsies from healthy adults and scleroderma patients obtained under the protocols approved by the Institutional Review Board at Northwestern University. All donors or their parents/legal guardians provided written informed consent. Mouse skin fibroblasts were established by explant culture from three-week-old adiponectin-null mice and wild-type littermates 26. Fibroblasts were maintained in (D)MEM) supplemented with 10% fetal bovine serum (FBS) (Lonza, Basel, Switzerland), 50 μg/ml penicillin, and 50 μg/ml streptomycin in a humidified atmosphere of 5% CO₂ at 37°C, and studied between passages 2 to 8 27. When fibroblasts reached confluence, growth media with 10% FBS or serum-free media supplemented with 0.1% BSA were added to the cultures for 24 hours prior to TGF-β2 (Peprotech, Rocky Hill, NJ, USA), or full-length adiponectin (Bio Vendor, Karasek, Czech Republic). In selected experiments, the AMP-activated protein kinase (AMPK) inhibitor Compound C (Sigma, St Louis, MO, USA) was added to the culture 60 minutes prior to adiponectin. Toxicity was determined using lactate dehydrogenase (LDH) assays according to the manufacturer's instructions (Biovision, Milpitas, CA, USA).

Normal skin fibroblasts (3 × 10⁵) were suspended in 1.5 ml reconstitution buffer and (D)MEM. Cells were mixed with rat tail type I collagen (4 mg/ml, BD Biosciences, San Jose, CA, USA) and seeded in 12-well plates at 37°C for 48 hours to solidify the collagen plug. Epidermal keratinocytes (6 × 10⁶) were isolated from foreskin and suspended in E medium supplemented with 5 ng/ml epidermal growth factor (EGF) and seeded on the collagen plug 2829. Forty eight hours later, organotypic cultures were placed on a metal grid (BD Biosciences) and maintained at an air-medium interface by feeding with E medium every other day for five days. Metformin (1 mM) was added to the media for 24 hours followed by TGF-β (5 ng/ml). Following incubation for a further six days, cultures were harvested, RNA was isolated, and tissues were fixed in formalin. Paraffin-embedded sections (4 μm thickness) were examined by Picrosirius Red staining.

Fibroblasts were transfected with target-specific siRNA (Dharmacon, Lafayette, CO, USA) or scrambled control siRNA (10 nM). Twenty-four hours following transfection, fresh media were added to the cultures, and the incubations were continued for a further 24 hours. Knockdown efficiency was evaluated by determining endogenous mRNA levels by real-time qPCR.

At the end of each experiment, cultures were harvested, RNA was isolated using RNeasy Plus mini kits (Qiagen, Valencia, CA, USA) and examined by real-time quantitative qPCR 30. Experiments were repeated three times with consistent results. The primers used for qPCR are shown in Table 1.

Expression of AdipoR1/2 mRNA was interrogated in publicly available genome-wide expression scleroderma skin microarray datasets (GEO accession number: GSE9285) 31.

Fibroblasts at early confluence were transfected with [SBE]₄-luc plasmids harboring four copies of a minimal Smad-binding element using SuperFect Transfection kit (Qiagen) as described 32. Cultures were incubated in serum-free media containing 0.1% BSA for 24 hours, followed by TGF-β2 for a further 24 hours and harvested. Whole cell lysates were assayed for their luciferase activities using a dual-luciferase reporter assay system (Promega, Madison, WI, USA). In each experiment, Renilla luciferase pRL-TK (Promega) was cotransfected as control for transfection efficiency 33. Transient transfection experiments were performed in triplicate and repeated at least twice with consistent results.

Fibroblasts (1 × 10⁴ cells/well) were seeded onto eight-well Lab-Tek II chamber glass slides (Nalge Nunc International, Naperville, IL, USA) and incubated in serum-free Eagle's minimal essential medium (EMEM) with 0.1% BSA for 24 hours. Fresh media with adiponectin (5 ug/ml) were added, and the incubations continued for a further 24 hours. At the end of the experiments, cells were fixed, permeabilized, and incubated with primary antibodies to Type I collagen at 1:500 dilution (Southern Biotech, Birmingham, AL, USA), or to α-SMA at 1:200 dilution (Sigma, St Louis, MO, USA). Cells were then washed with PBS and incubated with secondary antibodies at 1:500 dilution (Alexa Fluor 488 and 594, Invitrogen) and viewed under a Nikon C1Si confocal microscope.

At the end of each experiment, fibroblasts were harvested and whole cell lysates subjected to Western analysis as described 30. The following antibodies were used: Type I collagen (Southern Biotech), α-SMA (Sigma), and GAPDH (Zymed, San Francisco, CA, USA). Bands were visualized using ECL reagents (Pierce, Rockford, IL, USA).

Statistical analysis was performed on Excel (Microsoft, Redmond, WA, USA) using Student t-test or analysis of variance (ANOVA). The results are shown as the means ± SEM. P <0.05 was considered statistically significant.

To investigate the regulation of fibrotic gene expression by adiponectin, foreskin fibroblasts were maintained in two-dimensional monolayer cultures. At confluence, cultures were incubated in media with an increasing concentration of adiponectin for 24 hours, and changes in gene expression were examined by real-time qPCR, Western analysis and immunocytochemistry. The results demonstrated a dose-dependent inhibition of Col1A1 and α-SMA gene expression, with a >60% reduction at 24 hours (Figure 1 and data not shown). Potent inhibition of Type I collagen and α-SMA by adiponectin was confirmed by Western analysis and immunostaining (Figure 1). Comparable results were observed in normal adult dermal fibroblasts (data not shown). Expression of both AdipoR1 and AdipoR2 mRNA in explanted fibroblasts was confirmed by real-time qPCR. Next, we investigated the effect of recombinant adiponectin in scleroderma fibroblasts. Confluent scleroderma fibroblasts (n = 4, Table 2) were incubated with adiponectin for 36 hours, and cell lysates were used for Western analysis. Results showed that adiponectin induced an approximately 40% decrease in collagen gene expression (Figure 1D).

In light of the fundamental role of TGF-β in orchestrating fibrogenesis, it was of interest to evaluate how adiponectin modulated relevant responses elicited by TGF-β. For this purpose, normal fibroblasts in two-dimensional monolayer cultures were pretreated with adiponectin followed by incubation with TGF-β for a further 24 hours. The results of real-time qPCR showed that adiponectin caused a dose-dependent attenuation of collagen and α-SMA gene expression induced by TGF-β, with an almost 50% reduction at 10 μg/ml (Figure 2A). Of note, adiponectin induced an approximately four-fold increase in the levels of the TGF-β pseudoreceptor BMP and activin membrane-bound inhibitor (BAMBI), which negatively regulates TGF-β responses. To examine the possible role of endogenous adiponectin in modulating the intensity of TGF-β responses, we used an RNAi approach. The results showed that siRNA-mediated effective knockdown of adiponectin in fibroblasts significantly increased the basal levels of Type I collagen and α-SMA mRNA and protein (1.5-fold and 1.8-fold, respectively). Moreover, adiponectin-depleted fibroblasts were hypersensitive to TGF-β treatment, with significantly enhanced stimulation of collagen and α-SMA gene expression compared to fibroblasts transfected with control siRNA, suggesting an inhibitory function for endogenous adiponectin in setting the intensity of TGF-β signaling.

In mesenchymal cells, adiponectin induces AMP kinase activity (34 and data not shown). To investigate the role of AMP kinase in modulating fibrotic gene expression, fibroblasts were incubated with the selective AMP kinase agonists 5-amino-1-β-D-ribofuranosyl-imidazole-4-carboxamide (AICAR) or metformin. The results of real-time qPCR demonstrated a potent dose-dependent inhibition of Col1A1 and Col1A2 mRNA expression, with a nearly 90% reduction at 5 mM of the AMP kinase antagonists (Figure 3). There was no evidence of cellular toxicity even at the highest concentrations of AICAR or metformin tested (data not shown). In addition to collagen, multiple genes implicated in fibrogenesis showed substantial decrease in expression. To establish the specificity of the anti-fibrotic activity of AMP kinase agonists, we examined the expression of the insulin-regulated glucose transporter GLUT4, a target gene positively regulated by AMP kinase 35. As expected, AICAR induced a substantial increase in GLUT4 mRNA expression. Both AMP kinase agonists potently attenuated the fibrotic responses induced by TGF-β (data not shown). To investigate the mechanism, transient transfection assays were performed. The results showed that adiponectin incubation resulted in disruption of canonical Smad signaling, as shown in transient transfection assays with the Smad-responsive minimal promoter (SBE)₄-luc (Figure 3D).

The organotypic raft culture (ORC) model is a three-dimensional full-thickness human skin equivalent that is a powerful approach to studying fibroblast function in the context of fibrogenesis 28. This full thickness human skin equivalent model allows us to examine fibroblast behavior where the biomechanical forces impacting the fibroblasts are relevant to the physiologically relevant context of skin 28. The three-dimensional full-thickness skin equivalents were incubated with metformin (1 mM) with or without TGF-β (5 ng/ml) for six days. Results from real-time qPCR showed that while TGF-β induced a substantial increase in fibrotic gene expression, treatment with metformin abrogated the effect (Figure 4A). Picrosirius Red staining showed that TGF-β induced a notable accumulation of strongly birefringent red collagen fibers, indicating highly cross-linked collagen, in the dermal compartment. In contrast, pretreatment of the rafts with metformin prevented collagen maturation, with a predominance of green, less cross-linked collagen fibers (Figure 4B), confirming that metformin abrogated TGF-β induced collagen protein accumulation.

To directly examine the role of AMP kinase in mediating the antifibrotic effects of adiponectin, a chemical inhibitor of AMP kinase activity was used 36. In fibroblasts preincubated with Compound C, a selective and potent AMP kinase inhibitor, the inhibitory effects of adiponectin on TGF-β-induced collagen and α-SMA mRNA and protein were completely abrogated (Figure 5).

We have shown previously that both pharmacological and endogenous ligands of PPAR-γ inhibited collagen gene expression, and abrogated the stimulation of fibrotic responses elicited by TGF-β 37. Moreover, rosiglitazone, a PPAR-γ ligand inhibited the over-expression of fibrotic genes in fibroblasts explanted from scleroderma patients 24. The anti-fibrotic activities of these ligands were blocked by the irreversible PPAR-γ antagonist GW9662, indicating that they were largely PPAR-γ-dependent 38. Adiponectin is a direct transcriptional target of PPAR-γ, and its expression in both adipocytes and fibroblasts is tightly regulated via activated PPAR-γ binding to cognate DNA recognition sequences in the adiponectin gene promoter 1139. In order to investigate the potential role of endogenous adiponectin in mediating the anti-fibrotic effects of PPAR-γ ligands, we examined the effect of prostaglandin J2 (PGJ₂) in adiponectin-null mouse skin fibroblasts. Consistent with the results using RNAi, we found that collagen and α-SMA gene expression were significantly elevated in both unstimulated and TGF-β-stimulated fibroblasts lacking adiponectin compared to wild type control fibroblasts, confirming the significant role of cellular adiponectin in modulating the intensity of TGF-β-induced fibrotic responses (Figure 6). Importantly, while PGJ₂ elicited substantial down-regulation of TGF-β responses in wild type fibroblasts, as shown previously 25, no significant PGJ₂ effect on the stimulatory response was seen in adiponectin-null fibroblasts.

We next sought to determine if the anti-fibrotic effects of adiponectin were specific for TGF-β, or more generalized for other profibrotic stimuli. To this end, fibroblasts were incubated with lipopolysaccharide (LPS), a potent ligand of Toll-like receptor 4 (TLR4). LPS induced a time-dependent stimulation of collagen and αSMA gene expression in normal fibroblasts (Figure 7 and data not shown). However, pretreatment of the cultures with adiponectin completely abrogated the stimulatory effects of LPS.

Adiponectin-induced cellular responses are mediated through activation of the adiponectin receptors AdipoR1 and AdipoR2 40. In order to investigate the adiponectin signaling axis in scleroderma, we examined AdipoR expression. Fibroblasts were explanted from skin biopsies from the affected lesional forearm of four patients with scleroderma (Table 2), and age and sex-matched healthy controls (n = 4) and grown to confluence, when total RNA was isolated and subjected to real-time qPCR. The results showed approximately 40% lower levels of AdipoR1 mRNA in scleroderma fibroblasts compared to normal fibroblasts, but the differences were not statistically significant (P = 0.22) (Figure 8). AdipoR2 levels were comparable in scleroderma and control fibroblasts.

To evaluate AdipoR1/2 mRNA expression in scleroderma skin, the expression of these genes was interrogated in a publicly available microarray dataset examining gene expression in skin 31. Biopsies clustering within the diffuse and inflammatory intrinsic subsets 31 showed an approximately 30% reduction in AdipoR1 (P <0.05), with a slight reduction in AdipoR2 (P = ns) expression compared to biopsies clustering with the normal-like subset (Figure 8B).