Salinomycin and Other Polyether Ionophores Are a New Class of Antiscarring Agent*

Background: Excessive scar formation is a debilitating complication underlying many diseases of the eye, lungs, and skin. Results: We identified salinomycin as an inhibitor of TGFβ-dependent cell signaling. Conclusion: Salinomycin and other polyether ionophores block TGFβ-driven fibroblast proliferation and myofibroblast formation. Significance: There are few effective therapies to combat excessive scarring, and salinomycin is a novel therapeutic candidate. Although scarring is a component of wound healing, excessive scar formation is a debilitating condition that results in pain, loss of tissue function, and even death. Many tissues, including the lungs, heart, skin, and eyes, can develop excessive scar tissue as a result of tissue injury, chronic inflammation, or autoimmune disease. Unfortunately, there are few, if any, effective treatments to prevent excess scarring, and new treatment strategies are needed. Using HEK293FT cells stably transfected with a TGFβ-dependent luciferase reporter, we performed a small molecule screen to identify novel compounds with antiscarring activity. We discovered that the polyether ionophore salinomycin potently inhibited the formation of scar-forming myofibroblasts. Salinomycin (250 nm) blocked TGFβ-dependent expression of the cardinal myofibroblast products α smooth muscle actin, calponin, and collagen in primary human fibroblasts without causing cell death. Salinomycin blocked phosphorylation and activation of TAK1 and p38, two proteins fundamentally involved in signaling myofibroblast and scar formation. Expression of constitutively active mitogen activated kinase kinase 6, which activates p38 MAPK, attenuated the ability of salinomycin to block myofibroblast formation, demonstrating that salinomycin targets the p38 kinase pathway to disrupt TGFβ signaling. These data identify salinomycin and other polyether ionophores as novel potential antiscarring therapeutics.

Scarring is a consequence of the normal wound healing response. However, scar formation can be exuberant, leading to hypertrophic scarring and/or fibrosis that can ultimately lead to a loss of tissue function (1)(2)(3)(4). Although there is a major knowledge gap as to why scarring sometimes proceeds out of control, hypertrophic scarring usually results from physical injury, such as laceration or surgery, or from burns, either induced thermally or chemically or by radiation (5). For example, an unfor-tunate consequence of severe thermal burns is the development of debilitating hypertrophic scars (6). Chronic inflammation and autoimmune disease can also lead to aberrant tissue reorganization and scarring (7,8). Thyroid eye disease (TED) 2 is an example of an autoimmune disease in which immune cells target the muscle and connective tissue in the ocular orbit, leading to orbital tissue remodeling and excessive scarring (3,9). Although aberrant scarring is observed in numerous pathologies, there are few, if any, effective therapies that limit or prevent scarring.
The key effector cell in scar formation is the contractile and secretory myofibroblast (10). Myofibroblasts can be derived from tissue-resident fibroblasts, epithelial-mesenchymal transitions, circulating fibrocytes, mesenchymal stem cells, or other progenitor cells (11). Myofibroblasts highly express ␣-smooth muscle actin (␣SMA), an important protein required for wound contraction, and these cells produce large amounts of extracellular matrix material, including collagen, fibronectin, and glycosaminoglycans (12,13). The contractile properties of myofibroblasts and their excessive production of extracellular matrix material such as collagen result in disruptive tissue remodeling. In addition to their contractile phenotype, myofibroblasts also secrete a variety of cytokines, including IL-6, MCP-1, and TGF␤, that recruit immune cells and lead to further myofibroblast formation (14). Myofibroblasts are fundamentally involved in scar formation, making them ideal targets for new therapeutic options to limit or even reverse scarring.
Myofibroblast differentiation is mainly driven by the cytokine TGF␤, which is produced during the natural healing process (15,16). TGF␤ is also involved in many other cellular responses, including immune suppression, cell migration, and extracellular matrix remodeling (17). Although TGF␤ is normally tightly regulated at multiple levels to limit its powerful effects, TGF␤ is highly expressed in conditions such as cancer, chronic inflammation, and fibrosis (17). TGF␤ regulates numerous cell signaling pathways while driving myofibroblast formation. One key pathway activated by TGF␤ is Smad-de-* This work was supported, in whole or in part, by National Institutes of Health Grants ES-023032 and EY-023239, and ES-001247. This work was also supported by an unrestricted grant from the Research to Prevent Blindness and by a research grant from the Rochester/Finger Lakes Eye and Tissue Bank. 1 To whom correspondence should be addressed: E-mail: Richard_Phipps@ urmc.rochester.edu. pendent signaling. Smads are a family of highly conserved transcription factors that regulate the expression of many genes that contain Smad-binding element (SBEs) in their promoter and/or regulatory regions (18). TGF␤ binding to its cognate receptor on the cell surface triggers phosphorylation of the Smad family members Smad2 and Smad3. When phosphorylated, these transcription factors bind their binding partner, the cosmad Smad4. This Smad complex is then shuttled into the nucleus, where it activates transcription of myofibroblast genes, including ␣SMA, calponin, and collagen (19 -21). The goal of this study was to identify a small molecule that could disrupt TGF␤-dependent myofibroblast formation and, thereby, mitigate excessive scarring and fibrosis. Although many attempts have been undertaken, there are few, if any, proven effective therapies for scarring. Our results reveal that the polyether ionophore, salinomycin, is a potent inhibitor of TGF␤-dependent human myofibroblast formation. Therefore, our results provide new evidence that salinomycin and other small molecules of its class are potential antiscarring compounds.

EXPERIMENTAL PROCEDURES
Cell Culture-Primary human fibroblasts were acquired and cultured as described previously (22). HEK293FT cells were obtained from the ATCC and cultured in DMEM supplemented with 10% fetal calf serum (Hyclone) and antibiotics. DMEM, minimum Eagle's medium, and hygromycin were purchased from Invitrogen. Fibroblast growth medium was purchased from Promocell. Other compounds used were salinomycin (Cayman) and narasin, monensin, and clioquinol (Sigma). Recombinant human TGF␤ was obtained from R&D Systems and was used at final concentrations of either 1, 5, or 10 ng/ml. Antibodies against ␣SMA and FLAG were purchased from Sigma. Antibodies against TAK1, phospho-TAK1, and collagen were from Santa Cruz Biotechnologies. Antibodies against Smad2, phosho-Smad2, p38, phospho-p38, cleaved PARP, and ␤-tubulin were from Cell Signaling Technology.
Development of a TGF␤-responsive Cell Line-The minimal thymidine kinase promoter was amplified by PCR with a forward primer that contained four tandem SBEs (forward, 5Ј-AGGTACCTACTAAGTCTAGACGGCAGTCTAGACGTA-CTAAGTCTAGACGGCAGTCTAGACGTAGAGCTCGGCCC-CGCCCAGCGTCTTGTC-3Ј; reverse, 5Ј-TAAAGCTTCTCGAG-ATCTGCGGCACGCT-3Ј), with restriction sites underlined. The resultant PCR product was TOPO cloned (Invitrogen), and the correct insert was verified by DNA sequencing. The construct was digested with SacI-HindIII, and the SBEx4-TK insert was purified and ligated with the pGL4.15 vector (Promega). Clones were verified by restriction digestion and then tested for TGF␤ responsiveness in transient transfections of 293FT cells. When the pSBEx4-TK-luc construct was demonstrated to be TGF␤responsive, the construct was introduced into 293FT cells for stable cell line production. Clones were selected by treatment with 200 g/ml hygromycin. Twenty clones were selected and screened for TGF␤-induced luciferase activity. One clone that was robustly responsive was subsequently used in the small molecule screen.
Plasmid DNA Transfection-Plasmid DNA was introduced into human fibroblasts by electroporation. Plasmids were electroporated with an Amaxa nucleofector (program U-025) into 1 ϫ 10 6 cells. The pcDNA3-FLAG MKK6(glu) plasmid, which encodes constitutively active MKK6 kinase S207E and T211E mutations, was obtained from Addgene (plasmid 13518 from the laboratory of Roger Davis). After transfection, cells were cultured for 12-24 h, and the culture medium was subsequently changed to treatment conditions. Western Blotting-Total protein was isolated from 0.5-2 ϫ 10 6 cells and lysed in 60 mM Tris (pH 6.8) and 2% SDS containing 1ϫ protease inhibitor mixture (Sigma). The lysates were passed through a 26-gauge needle five to six times to shear genomic DNA. Protein concentrations were determined using the detergent-compatible protein assay (Bio-Rad). Total protein (1-10 g/lane) was subjected to SDS-PAGE. Protein gels were transferred to a PVDF membrane (Millipore) and probed with antibodies as specified. Western blot band intensities were quantified using ImageLab software (Bio-Rad). Protein expression was normalized to ␤-tubulin levels.
Collagen Production Assay-Cell culture supernatant was collected and transferred (5-20 l) to a PVDF membrane using a slot blot device. The membrane was blocked and probed with a goat anti-collagen antibody (1:5000), washed, and incubated with a donkey anti-goat antibody conjugated to HRP. Band intensities were quantified using ImageLab software, and values were normalized to vehicle treatments.
Alamar Blue Viability Assay-5 ϫ 10 3 cells/well were plated in a 96-well plate (Griener) with 200 l of culture medium. Vehicle (DMSO), salinomycin, and/or TGF␤ (1 ng/ml) were added as indicated, and then 20 l of Alamar Blue reagent was added to all wells. Cells were incubated for 24 and 48 h, and then the fluorescence of the oxidized Alamar reagent was measured (excitation 470 nm, emission 480 nm). Background fluorescence was subtracted from all wells, and the fluorescence was normalized to vehicle-treated cells. The assay was performed in two different human fibroblast strains, and treatments were performed in triplicate.
BrdU Incorporation Assay-Human fibroblasts were seeded on a 96-well plate at a density of 1 ϫ 10 5 cells/well. Cells were treated in triplicate. Cell proliferation was determined using the BrdU assay following the instructions of the manufacturer. Briefly, cells were treated with a BrdU label at a 1:2000 dilution for 24 h after the initial 72-h treatment with TGF␤ with or without drugs. BrdU incorporation was measured using the BrdU cell proliferation assay kit (Calbiochem, San Diego, CA) at 450 -540 nm using a Varioskan microplate reader.
Luciferase Reporter Assays-Primary human fibroblasts were collected and electroporated as described above with pSBEx4-TK-luc and CMV-Renilla luciferase (Promega, Madison, WI). Nucleofected cells were plated and allowed to grow for 12-24 h. DMSO (vehicle), salinomycin (250 nM), and/or TGF␤ (1 ng/ml) were added to the cultures for an additional 24 -36 h. Following incubation, cells were washed two times in 1ϫ PBS and lysed directly in plates using the Dual-Glo luciferase assay buffer (Promega). Firefly and Renilla luciferase readings were measured on a Varioskan Flash luminescent plate reader (Thermo Fisher) following the instructions of the manufacturer. Lucifer-ase readings were normalized to the control vehicle-treated samples for statistical analysis.
Collagen Contraction Assay-Experiments were conducted using the collagen contraction assay from Cell Biolabs, Inc. following the protocol for the manufacturer. Briefly, 5 ϫ 10 5 cells and 500 l of collagen suspension per well were plated in a 24-well plate. After collagen gel polymerization, vehicle (ethanol), salinomycin, or narasin (250 nM) were added for 1 h before TGF␤ (5 ng/ml) treatment. Gels were detached from well and photographed at time 0, 24, 48 and 72 h. Contraction was measured by analyzing the gel area at each time point using ImageJ software. Gels treated with vehicle and TGF␤ only were set to 100%.
Wound Healing Assay-Six-well plates of human fibroblasts were grown in normal culture medium until they reached confluence. The medium was then replaced with low-serum (0.05%) medium for 24 h, and a scratch wound was made using a micropipette tip. After wounding, cells were treated with either vehicle (ethanol) or salinomycin (250 nM) for 1 h before treatment with TGF␤ (5 ng/ml). The wound area was photographed by brightfield microscopy at 0, 24, 48, and 72 h. The area of each wound was measured using ImageJ software. Wound areas at 24, 48, and 72 h were normalized to the area of the wound at time 0.
Statistical Analysis-Data were analyzed using GraphPad Prism software, and Student's t test and One-way analysis of variance were used for statistical analysis. p Ͻ 0.05, p Ͻ 0.01, and p Ͻ 0.001 were considered significant.

RESULTS
Excessive scarring results from the formation of too many myofibroblasts and the production of too much extracellular matrix material such as collagen. Because TGF␤ drives the formation of myofibroblasts in part through activating the Smad pathway, we designed a reporter construct that served as a measure of TGF␤-induced Smad activity (Fig. 1A). We inserted four tandem SBEs upstream of the minimal thymidine kinase (tk) promoter. The SBEx4-tk promoter was inserted into the pGL4.15 vector, which contains a destabilized firefly luciferase gene (Luc2P) and the hygromycin resistance gene (Hygro). The pSBEx4-tk-luc-Hygro plasmid (Fig. 1A) was then introduced into HEK293FT cells by lipofection. Stable clones were generated and screened for TGF␤-induced luciferase activity. One clone was selected that consistently gave a 15-to 20-fold induction of luciferase activity when cells were treated with TGF␤ for 24 h (Fig. 1B). To find compounds that inhibit TGF␤ activity and, therefore, may be novel anti-scarring compounds, we screened the Spectrum collection of 2300 small molecules with the cell line. Several small molecules inhibited TGF␤-dependent luciferase activity (Fig. 1C). One small molecule that was particularly potent at inhibiting luciferase activity was compound L18, which corresponded to the polyether ionophore salinomycin. Further testing of salinomycin showed a dose-dependent decrease in TGF␤-induced luciferase, where 100 nM salinomycin reduced luciferase activity ϳ4-fold, and 1 M salinomycin reduced luciferase activity levels to below baseline levels (Fig. 1D). Salinomycin, an antibiotic produced by Streptomyces albus is used as coccidiostatic agent in animal feed (23). Furthermore, salinomycin has been shown recently to have activity against cancer stem cells (24). Therefore, we proceeded to further investigate the potential of salinomycin as an antiscarring agent.
Fibroblasts are sentinel cells that respond to numerous stimuli and serve as key effector cells in many biological processes (25). One crucial function of fibroblasts is their differentiation into scar-forming myofibroblasts that are involved in wound healing but, in excess, cause scarring and fibrosis. TED is a disorder in which myofibroblasts and scar tissue accumulate in the ocular orbit, causing pain, proptosis, and, in severe cases, blindness (7). We have shown previously that human orbital fibroblasts dramatically respond to TGF␤ by forming myofibroblasts that express ␣SMA and calponin and produce high levels of collagen (26). Therefore, TED fibroblasts were an ideal model to test the ability of salinomycin to block TGF␤ function. We treated human TED fibroblasts with TGF␤ (1 ng/ml) in the presence or absence of 0.1 and 1 M salinomycin (compound L18 from the Spectrum library) or various doses of other small molecule hits from the Spectrum library for 72 h to allow the formation of myofibroblasts. After 72 h, cells were harvested and analyzed by Western blot. As expected, the fibroblasts robustly responded to TGF␤, inducing expression of ␣SMA and calponin 36-and 10-fold, respectively, over vehicle-treated cells ( Fig. 2A). Interestingly, salinomycin completely blocked the expression of ␣SMA and calponin in a dose-dependent manner ( Fig. 2A). To further characterize the effect of salinomycin, a dose-response experiment was carried out in which human fibroblasts were treated with TGF␤ and doses of salinomycin from 10 -250 nM for 72 h (Fig. 2B). At 50 nM salinomycin, expression of ␣SMA and calponin was reduced 4-fold and 3-fold, respectively, over TGF␤ treatment alone. Furthermore, at 250 nM salinomycin, ␣SMA and calponin expression was reduced ϳ16-fold and 5.2-fold, respectively, to levels at or below untreated fibroblasts (Fig. 2B). To further investigate the timing of the effect of salinomycin on TGF␤-induced myofibroblast formation, fibroblasts were treated with salinomycin (250 nM) and TGF␤ for 0, 24, 48, and 72 h before cell harvest and subsequent analysis of ␣SMA and ␤-tubulin by Western blot (Fig. 2C). TGF␤ induced ␣SMA at 24, 48, and 72 h of treatment, with a maximal induction at 72 h (13-fold over vehicle treatment). Salinomycin blocked TGF␤-induced ␣SMA expression at 24, 48, and 72 h, with more than a 20-fold inhibition at 72 h.
Another key role of myofibroblasts in scar formation is the production of collagen. To test whether salinomycin could block myofibroblast collagen production, human fibroblasts were treated with TGF␤ and salinomycin (10 -250 nM) for 72 h, and then the culture medium was collected and analyzed for collagen I levels using a specific collagen I antibody and slot blot analysis (Fig. 3A). As expected, TGF␤-induced myofibroblasts produce high levels of collagen, and salinomycin blocked production of collagen in a dose-dependent manner. Starting at 10 nM salinomycin, collagen production decreased, and, at 100 nM salinomycin, collagen production was back to baseline levels. Finally, at 250 nM salinomycin, collagen production was below baseline levels of untreated fibroblasts (Fig. 3A). Additionally, a time course experiment was performed with fibroblasts treated with TGF␤ or TGF␤ plus salinomycin (250 nM), and the culture medium was collected at 5, 24, 48, and 72 h. As expected, TGF␤ induced collagen production in a time-dependent manner in human fibroblasts (Fig. 3B). Interestingly, salinomycin reduced collagen production at 24, 48, and 72 h, with a more than 4-fold reduction in collagen production at 72 h.
Because salinomycin, a polyether ionophore, prevented formation of myofibroblasts, we next tested other ionophores for anti-myofibroblast activity (Fig. 4A). Narasin is a methylated derivative of salinomycin that is also a coccidiostat used in animal feed (27). Monensin is another polyether ionophore that is used extensively in animal feed to prevent coccidiosis (27). Finally, we chose a fourth compound, clioquinol, which is also an antiprotozoal drug and ionophore but it is structurally unrelated to salinomycin and other polyether ionophores (28). We initially tested the ability of these ionophores to inhibit the luciferase activity of our TGF␤ reporter cell line. We observed that, like salinomycin, narasin and monensin inhibited luciferase expression, whereas clioquinol did not (data not shown). To further test these four compounds, we treated human fibroblasts with TGF␤ in the absence or presence of 250 nM of sali- FIGURE 1. A small molecule screen identifies salinomycin as a potential antiscarring compound. A, the Smad-dependent reporter construct. Four tandem SBEs were inserted upstream of the minimal tk promoter. Downstream of the promoter is a destabilized version of the firefly luciferase gene (luc2P) present in the pLuc2P-Hygro plasmid that also harbors the hygromycin resistance gene. B, the reporter plasmid was introduced into the HEK293FT cell line, and individual colonies were selected with hygromycin (200 g/ml) to develop a Smad/TGF␤-dependent luciferase reporter cell line. 1 ng/ml treatment of TGF␤ for 24 h resulted in a robust increase of luciferase activity. C, the HEK293FT-luc reporter line was screened with the 2300-compound Spectrum collection of small molecules. Several hits from the initial screen that blocked TGF␤-induced luciferase activity were tested further in a Dual-Luciferase screen including a constitutive Renilla luciferase to normalize the SBE luciferase activity. Compound L18, which corresponded to salinomycin, a polyether ionophore antibiotic, inhibited the reporter construct more than 10-fold. Unt., untreated; Veh, vehicle; comp., compound. D, salinomycin (two-dimensional structure, right) exhibited a dose-dependent decrease in TGF␤-induced SBE-luciferase activity. 1 M salinomycin reduced SBE luciferase activity to below baseline levels, whereas 100 nM salinomycin reduced SBE luciferase activity more than 3-fold. Experiments were repeated three times in triplicate, and similar results were observed in all tests. *, p Ͻ 0.01 versus untreated cells; #, p Ͻ 0.01 versus TGF␤ with vehicle.
nomycin, narasin, clioquinol, or monensin for 72 h. After treatment, cell lysates were harvested and analyzed for the expression of the myofibroblast markers ␣SMA and calponin (Fig. 4B). As expected, salinomycin reduced the expression of ␣SMA and calponin to lower than vehicle levels. Interestingly, both narasin and monensin had a similar activity as salinomycin. However, the structurally unrelated compound clioquinol did not inhibit expression of ␣SMA or calponin (Fig. 4B). Therefore, these data support that salinomycin and other polyether ionophores specifically block TGF␤-induced myofibro-blast formation. To further characterize the ability of narasin to block TGF␤-induced myofibroblast formation, three doses of narasin (50, 100, and 250 nM) and salinomycin (250 nM) were used to block TGF␤ effects (Fig. 4C). As expected, TGF␤ induced ␣SMA and calponin expression by more than 10-and 5-fold, respectively, over vehicle-treated cells, and narasin effectively blocked ␣SMA and calponin expression at 50, 100, and 250 nM (Fig. 4C).
In addition to showing that salinomycin inhibits expression of ␣SMA, calponin, and collagen, we next tested the ability of salinomycin and narasin to inhibit myofibroblast contraction of a collagen gel matrix (Fig. 5A). Myofibroblast contraction is an important function of myofibroblasts in wound healing, and excessive contraction of unwanted myofibroblasts leads to debilitating scar formation (14,29). Salinomycin (250 nM) and narasin (250 nM) significantly inhibited TGF␤-induced myofibroblast contraction by 40 -80% after 2-3 days of TGF␤ treatment (Fig. 5A). As a second test of myofibroblast function, we measured cell migration after an induced scratch in a mono-   FEBRUARY 6, 2015 • VOLUME 290 • NUMBER 6 layer of fibroblasts. After the scratch was made, cells were treated with TGF␤ in the presence or absence of salinomycin. After 3 days of TGF␤ treatment alone, the open area of the wound was reduced to ϳ15% of the original wound size because of the growth and migration of myofibroblasts (Fig. 5B). Salinomycin (250 nM) significantly blocked TGF␤-induced myofibroblast growth and migration more than 3-fold.

Salinomycin Inhibits Scar-forming Myofibroblast Formation
Although salinomycin and other polyether ionophores blocked myofibroblast formation and myofibroblast function, we tested whether these effects were a result of toxicity and/or because of a block of cell proliferation. Our first analysis was to visually inspect fibroblasts treated with vehicle, TGF␤, or TGF␤ plus salinomycin (250 nM) for 72 h using brightfield microscopy (Fig. 6A). Representative images (original magnification ϫ200) from two different primary human fibroblast strains demonstrate that TGF␤ induces hallmark morphological changes of fibroblasts into myofibroblasts. However, salinomycin treatment prevented morphological changes induced by TGF␤ treatment (Fig. 6A, left panels). Additionally, visualization of salinomycin-treated cells indicated that salinomycin did not lead to cell death, and cells appeared morphologically similar to vehicle-treated fibroblasts. To further quantitatively measure that salinomycin did not affect cell viability, human fibroblasts were treated with vehicle (DMSO), 50 -250 nM salinomycin alone, TGF␤, or TGF␤ plus 250 nM salinomycin. As a positive control, fibroblasts were treated with the cytotoxic drug puromycin (5 g/ml). Treated cells were cultured in the presence of Alamar Blue reagent, which measures mitochondrial oxidation-reduction (redox) potential and, therefore, serves as a quantitative viability sensor. After 72 h of culture, Alamar Blue fluorescence was measured to assay cell viability (Fig. 6B). As expected, puromycin treatment resulted in a total loss of cell viability. However, treatment with salinomycin alone or salinomycin plus TGF␤ did not result in a loss of viability. These results demonstrate that salinomycin, at the indicated doses, does not affect human fibroblast viability. We also tested whether salinomycin modifies basal or TGF␤-induced fibroblast proliferation. Cells were treated with vehicle or 250 nm of either salinomycin or narasin for 24 h in the presence of BrdU to measure cell proliferation (Fig. 6C). Neither salinomycin nor narasin blocked basal fibroblast proliferation. As expected, TGF␤ induced human fibroblast proliferation 9-fold. Remarkably, both salinomycin and narasin were able to beneficially block TGF␤-induced proliferation 5-fold (Fig. 6C). Although we did not observe any decreases in basal fibroblast proliferation in cells treated with salinomycin, we further investigated whether salinomycin may be triggering apoptosis in fibroblasts treated with or without TGF␤. Several recent reports have demonstrated that salinomycin can induce apoptosis in cancer cells at concentrations of over 5 M (30, 31). To determine whether salinomycin triggers apoptosis in human fibroblasts at the doses needed to block myofibroblast formation, we measured production of cleaved PARP, a hallmark of apoptosis (Fig.  6D). As a positive control, cells were treated with the apoptosisinducing drug puromycin. As expected, puromycin treatment resulted in a large induction of cleaved PARP production. Salinomycin, at doses of 50 -500 nM, did not induce cleaved PARP production in human fibroblasts treated with or without TGF␤. This result suggests that salinomycin is not working through apoptosis to prevent myofibroblast formation.
To further characterize the molecular mechanisms by which salinomycin blocks TGF␤-induced myofibroblast formation, FIGURE 5. Salinomycin blocks TGF␤-induced myofibroblast contraction and migration. A, human fibroblasts were cultured in a collagen gel matrix as described in the text. After collagen gel formation, cells were treated with TGF␤ (5 ng/ml) or TGF␤ plus either salinomycin (SNC, 250 nM) or narasin (Nar, 250 nM) and allowed to contract for 72 h. Gel matrices were imaged at the 0, 24, 48, and 72 h time points. The gel area was quantified using ImageJ. TGF␤-induced contraction was set to 100%, and all other measurements were normalized to it. Representative images are shown, demonstrating that TGF␤ readily induced myofibroblast contraction, whereas salinomycin and narasin inhibit contraction by more than 80% and 90%, respectively, 72 h post-TGF␤. *, p Ͻ 0.01 versus TGF␤-treated cells. B, human fibroblasts were cultured in a 12-well culture dish until they reached a confluent monolayer. A scratch wound was then induced in the cultures, and wells were washed to remove cells and debris. Cells were treated with TGF␤ (5 ng/ml) or TGF␤ plus salinomycin (250 nM) for 72 h to allow myofibroblast migration. Open areas were quantified using ImageJ software. Areas were normalized to time 0 (area ϭ 1.0 at time 0). Images were captured at 0, 24, 48, and 72 h. After 72 h, TGF␤-treated cells had reduced the scratched area to less than 20% of the original area. Salinomycin treatment prevented myofibroblast migration 3-to 4-fold over TGF␤ only. *, p Ͻ 0.01 versus TGF␤-treated cells. Experiments were repeated in triplicate, with two different strains and representative images shown. FEBRUARY 6, 2015 • VOLUME 290 • NUMBER 6

JOURNAL OF BIOLOGICAL CHEMISTRY 3569
we analyzed the activation kinetics of some of the key mediators of TGF␤ signaling (18). Because we first identified salinomycin as a putative antiscarring molecule by its ability to block TGF␤induced Smad activity, we analyzed the levels of phospho-Smad2 by Western blot in cells treated with TGF␤ or TGF␤ plus salinomycin (Fig. 7A). TGF␤ treatment rapidly induced phosphorylation of Smad2 in human fibroblasts. Smad2 phosphorylation was completed ablated in the presence of the known TGF␤ receptor inhibitor SB-43152. However, salino-mycin treatment did not affect phospho-Smad2 levels at 1 h of TGF␤ treatment (Fig. 7A). Because we saw that salinomycin did inhibit ␣SMA expression at longer time points (24 -72 h) and because of the fact that our initial luciferase screen was performed at 24 h of treatment, we analyzed the expression of other key mediators of TGF␤-dependent myofibroblast differentiation at 24, 48, and 72 h. In addition to Smad activation, TGF␤ induced phosphorylation and activation of TGF␤-activated kinase 1 (TAK1, also called MAP3K) and the MAPK p38 FIGURE 6. Salinomycin does not affect viability or basal proliferation of human fibroblasts but does block TGF␤-induced proliferation. A, human fibroblasts were treated with vehicle (DMSO), TGF␤ (1 ng/ml), or TGF␤ plus 250 nM salinomycin (SNC) and cultured for 72 h. After culture, cell images were taken (original magnification ϫ200). Two different primary fibroblast strains at two different culture confluence states are shown. TGF␤ induces morphology changes in human fibroblasts, indicative of myofibroblast formation. Cotreatment with salinomycin effectively blocked the morphological changes induced by TGF␤, and fibroblasts appeared similar to vehicle-treated fibroblasts. B, human fibroblasts were treated with vehicle (Veh, DMSO); 50, 100, or 250 nM salinomycin; TGF␤ (1 ng/ml); and TGF␤ plus 250 nM salinomycin or puromycin (P) (5 g/ml) for 72 h in the presence of the redox-sensitive fluorescent dye Alamar Blue. After 72 h, fluorescence was measured to analyze cellular viability. RFU, relative fluorescent units. Puromycin, which served as a positive control, resulted in a total loss of cell viability, whereas salinomycin did not significantly affect cell viability at the tested doses either in the presence or absence of TGF␤. C, human fibroblasts were treated with vehicle (DMSO), 250 nM salinomycin, 250 nM narasin (Nar), TGF␤ alone, or TGF␤ plus salinomycin or narasin (250 nM) for 24 h before addition of BrdU. BrdU treatment was carried out for an additional 24 h, and then cells were fixed and stained for BrdU incorporation, which served as a measure of cell proliferation. Salinomycin and narasin did not affect basal fibroblast proliferation (first three columns). However, salinomycin and narasin significantly blocked TGF␤-induced proliferation (last three columns). Results are from a representative experiment performed in triplicate. *, p Ͻ 0.01 versus vehicle; #, p Ͻ 0.01 versus TGF␤ treatment. D, primary human fibroblasts were treated with salinomycin (50 -500 nM) in the presence or absence of TGF␤ (1 ng/ml) for 72 h, and then cell extracts were collected and analyzed by Western blot for the apoptotic marker cleaved PARP. ␤-tubulin was used as a loading control. Cells were treated with puromycin as a positive control to induce apoptosis. As expected, puromycin induced cleaved PARP levels. However, salinomycin at the doses used (50 -500 nM) did not induce PARP cleavage. (Fig. 7, B and C). The effect of salinomycin on acute TGF␤induced phosphorylation of p38 and TAK1 was analyzed at 10 -60 min. Salinomycin blocked TGF␤-induced TAK1 phosphorylation at 10 and 30 min. Phosphorylation of p38 was induced at 60 min by TGF␤, and salinomycin had only modest effects at these acute time points. At longer treatment times, TGF␤ treatment induced phospho-TAK1 2.4, 3.9, and 5.8-fold at 24, 48, and 72 h, respectively. Interestingly, salinomycin blocked phosphorylation of TAK1 at these time points, with a more than 3-fold inhibition at 72 h (Fig. 7C). Likewise, TGF␤ induced phosphorylation of p38 at 24, 48, and 72 h, whereas salinomycin blocked phosphorylation of p38, with a more than 3-fold block at 72 h. We observed similar effects on TAK1, p38, and SMAD2 phosphorylation using narasin and monensin (data not shown). These data support the concept that salinomycin and other polyether ionophores do not directly target Smad signaling but, rather, target TGF␤-induced MAPK pathways through TAK1 and p38.
To further test whether salinomycin targets the TAK1-p38 pathway to block myofibroblast formation, we used a plasmid encoding a mutant MKK6 (pcDNA3-FLAG MKK6(glu)) protein. MKK6 (also called MAP2K) is activated by TAK1 and phosphorylates p38 in response to extracellular signals such as TGF␤ or other environmental stresses. The MKK6(glu) protein, which harbors glutamate mutations at amino acid residues serine 207 and threonine 211, is constitutively active and, therefore, phosphorylates p38 independently of upstream signals such as phospho-TAK1 (32). Therefore, if salinomycin is blocking the TAK1-p38 signaling pathway, then MKK6(glu) should overcome the effect of salinomycin. We first tested MKK6(glu) using our SBE4-TK-luc reporter plasmid. The luciferase reporter plasmid was introduced into human fibroblasts along with the MKK6(glu) plasmid or a control plasmid (pcDNA3-GFP). Cells were then treated with vehicle, TGF␤, or TGF␤ plus 250 nM salinomycin. After 24 h, luciferase activity was measured (Fig. 8A). As expected, in human fibroblasts expressing the GFP plasmid, TGF␤ induced luciferase activity 2-fold, and salinomycin completely blocked TGF␤-induced luciferase activity. Interestingly, in cells expressing constitutively active MKK6(glu) plasmid, salinomycin was unable to block TGF␤induced luciferase activity (Fig. 8A). We further tested whether MKK6(glu) could attenuate the ability of salinomycin to inhibit p38 phosphorylation. Control or MKK6(glu) plasmids were introduced into human fibroblasts by electroporation, and then cells were treated with TGF␤ for 72 h. Samples were analyzed for expression of FLAG-MKK6(glu), ␣SMA, calponin, and ␤-tubulin by Western blot (Fig. 8, B and C). As expected, TGF␤ treatment induced expression of ␣SMA and calponin in both pGFP and pMKK6(glu) treated fibroblasts. Furthermore, in fibroblasts treated with pGFP, salinomycin significantly blocked TGF␤ induction of ␣SMA and calponin (Fig. 8, B and  C). Interestingly, expression of MKK6(glu) attenuated the ability of salinomycin to block TGF␤-induced ␣SMA and calponin 5-and 2-fold, respectively (Fig. 8C). Taken together, these data support the concept that salinomycin blocks the TAK1-p38 signaling pathway to prevent TGF␤ induced Smad2/3-dependent signaling and formation of scar-forming myofibroblasts (Fig. 9).

FIGURE 7. Salinomycin does not directly target TGF␤-induced SMAD2 phosphorylation but inhibits TGF␤-induced TAK1 and p38 phosphorylation.
A, human fibroblasts were treated with vehicle, salinomycin, or SB-43152 with or without TGF␤. Cells were harvested at 1 h and analyzed for phospho-SMAD2, total SMAD2, and ␤-tubulin (loading control) by Western blot. Two representative strains are shown, demonstrating that, at 1 h of treatment, TGF␤ induces phosphorylation of SMAD2 and SB-43152 completely ablates phospho-SMAD2, whereas salinomycin has no effect. B, fibroblasts were treated with TGF␤ and/or salinomycin, and cells were harvested 0, 5, 10, 30, and 60 min after TGF␤ treatment to analyze levels of phospho-TAK1, total TAK1, phospho-p38, and total p38 by Western blot. Salinomycin blocked phospho-TAK production 10 and 30 min after TGF␤ treatment. R.E., relative expression. C, Human fibroblasts were treated as in A, except that cells were harvested at the 0, 24, 48, and 72 h time points to analyze levels of phospho-TAK1, totalTAK1, phospho-p38, total p38, phospho-SMAD2, and total SMAD2 by Western blot. The relative expression level of each phosphoprotein (normalized to ␤-tubulin) as determined by densitometry is shown below each lane. Salinomycin reduced TGF␤-induced phospho-Smad2 levels more than 2-fold at 48 h. Likewise, salinomycin also inhibited phospho-TAK1 levels at 24, 48, and 72 h. Salinomycin blocked p38 phosphorylation at the 24, 48, and 72 h time points, with an 80% reduction at 24 h and more than a 3-fold reduction at 72 h. Data are representative of at least three different experiments performed with different human fibroblasts strains.

DISCUSSION
Exuberant TGF␤ signaling and excessive development of scar-forming myofibroblasts is at the root of disorders such as thyroid eye disease and idiopathic pulmonary fibrosis, which involve abnormal tissue remodeling and excessive scar formation. Unfortunately, therapeutic options to treat excessive scarring are limited, and most are not proven to be effective (33). The results presented in this report identify salinomycin and other polyether ionophores as novel small molecules that block myofibroblast formation. Salinomycin potently blocked TGF␤-induced expression of ␣SMA, calponin, and collagen, all of which are hallmarks of myofibroblasts. Salinomycin also effectively blocked myofibroblast function without affecting cell viability. The ability of salinomycin to inhibit myofibroblast function and block production of these myofibroblast markers has not been recognized previously. The knowledge gained from these studies highlights the potential of salinomycin and other polyether ionophores to serve as the basis for new anti-scarring drugs.
Heightened TGF␤ signaling is observed in cancer, fibrosis, hypertrophic scarring, and TED (3,7,34). Additionally, TED fibroblasts, which are key effector cells in the disease, both proliferate at a faster rate and more readily form myofibroblasts in response to TGF␤ than normal, non-TED fibroblasts (26,36). Therefore, TGF␤ signaling is paramount to the aberrant myofibroblast formation, collagen production, contraction, and ocular tissue remodeling underlying TED pathophysiology. Unfortunately, besides surgical removal of excessive ocular tissue, there are few, if any, effective therapies to combat the disease (7,37,38). Here we show that salinomycin blocks TGF␤-induced transformation of primary human TED fibroblasts to scar-forming myofibroblasts. Furthermore, salinomycin blocked TGF␤-induced collagen production, cell proliferation, contraction, and migration, key features of tissue remodeling and fibrosis (39,40). Therefore, salinomycin, which we show acts robustly to target TGF␤ signaling in human fibroblasts, could serve as the basis for new therapeutic options for the treatment of TED.
In our study, we screened a TGF␤-dependent Smad reporter cell line with a small molecule library consisting of bioactive drugs. Because Smad transcription factors are directly downstream of the TGF␤ receptor, we expected to identify small molecules that could act directly on the TGF␤ receptor or Smad proteins, such as the small molecule inhibitor of the TGF␤ receptor, SB-43152 (41). Our data support the hypothesis that salinomycin indirectly targets the Smad pathway by working through the TAK1-p38 MAPK pathway. The TAK1-p38 pathway functions in myofibroblast formation in part by providing a feedforward loop to stimulate Smad2/3 phosphorylation and activation (Fig. 9) (42). Additionally, TAK1 has been shown recently to be essential for myofibroblast formation, and increased activity of TAK1 has been associated with kidney fibrosis (43,44). These results are consistent with our new data showing that salinomycin blocks myofibroblast formation by impairing phosphorylation of TAK1 acutely and TAK1, p38, and Smad2 phosphorylation at later time points. Salinomycin is therefore distinguished from other TGF␤ pathway inhibitors by not affecting SMAD2 phosphorylation at early time points. The TAK1-p38 pathway has been implicated in fibrosis and excessive scarring in numerous organ systems, including the eyes, kidney, lungs, and heart (45)(46)(47)(48)(49). Recently, a p38 inhibitor, Esbriet (pirfenidone) (50), was approved for treatment of idiopathic pulmonary fibrosis in Europe, indicating the efficacy of targeting the pathway in fibrosis. Although pirfenidone does slow the disease progression of pulmonary fibrosis, it does not stop it, indicating the need for more efficacious drugs.
The activity of p38 is regulated by the upstream kinases MKK3 and MKK6 (32,51). Although these two kinases are very similar in sequence and structure, they can often have differential effects on the activation of p38 (52). Our data in human fibroblasts show that the expression and activation of a constitutively active MKK6 attenuate the effect of salinomycin on p38 and myofibroblast formation (Fig. 8). Although our data point to the ability of salinomycin to inhibit TAK1 phosphorylation and activity to prevent MKK3/6 and p38 activation, future studies aimed at dissecting the role of salinomycin in altering TAK1 signaling will be required to fully understand its mechanism of action. Likewise, although the TAK1-p38 pathways appear to be crucial targets of salinomycin in myofibroblast formation, other pathways, such as wnt/␤-catenin signaling, may also be involved because other studies have shown that salinomycin modulates wnt signaling (53).
Salinomycin, its derivative narasin, and the related polyether ionophore monensin all appear to have powerful anti-myofibroblast activity (Fig. 4). That all three possess this ability opens up the possibility that polyether ionophore function is required. Polyether ionophores preferentially bind monovalent cations such as sodium and potassium (54). Although this may be important for their coccidiostat properties, it is unclear whether this is required to block TGF␤ function. However, because only nanomolar amounts of these ionophores are required and because there are micromolar levels or more of sodium and potassium ions in culture medium, it supports an alternative property of this family of molecules. Development of new analogs of salinomycin or other polyether ionophores that alter the polyether moieties is required to fully understand the nature of their anti-TGF␤ properties.
Recent publications have reported the exciting possibility that salinomycin is a powerful therapeutic agent in combating cancer stem cells (24,35,55,56). The concept that salinomycin may target highly proliferative cells as opposed to other, more slowly growing cells is also supported by our data showing that salinomycin is not toxic to human fibroblasts at the levels needed to blunt myofibroblast formation. In addition to driving myofibroblast formation, another consequence of high TGF␤ levels in fibrosis are the unwanted fibroproliferative effects (34). Interestingly, salinomycin does not affect basal proliferation of human fibroblasts but does prevent TGF␤-induced proliferation (Fig. 4). Furthermore, although salinomycin induces apo- FIGURE 9. Potential model for the salinomycin mode of action. Active TGF␤, present at high concentrations during wound healing, exuberant scarring, or chronic inflammation, binds to the TGF␤ receptor (TGF␤R). Activation of the TGF␤ receptor triggers a range of cell signaling events, including the phosphorylation and activation of TAK1 and Smad2/3. TAK activates MKK3/6 by phosphorylation, leading to phosphorylation and activation of p38. Active p38 leads to a further increase in Smad2/3 phosphorylation and activation. These signaling pathways reinforce the myofibroblast program, leading to excessive expression of ␣SMA, calponin, and collagen to promote myofibroblast generation and scar formation. Salinomycin or other polyether ionophores can block the activation and phosphorylation of TAK1 and p38, thereby limiting the activation of Smad2/3 and blocking myofibroblast formation.
ptosis of cancer cells, we saw no indication that salinomycin induces apoptosis in human fibroblasts and myofibroblasts at the doses used in our experiments (50 -500 nM).
Interestingly, myofibroblasts present in the tumor microenvironment also play a protective role for malignant cells and aid in metastasis (25). Our results show that salinomycin can disrupt myofibroblast formation. In light of this, another potential benefit of salinomycin is that it may alter the tumor microenvironment by decreasing the number of myofibroblasts, therefore making tumor growth less favorable. Another aspect of our work is that the effects of salinomycin are mimicked by the polyether ionophores narasin and monensin. Given that these small molecules all disrupt myofibroblast formation, it may be that the polyether ionophore family of small molecules is effective in targeting cancer cells. Future studies aimed at testing whether salinomycin and other polyether ionophores alter the tumor microenvironment to promote cancer cell death may further highlight the potential of these small molecules.
In summary, our work identifies salinomycin and other polyether ionophores as novel therapeutics for prevention of excessive scarring. Future studies using animal model(s) of excessive and/or abnormal scarring are required to determine the pharmacokinetics, stability, and efficacy of these drugs in an in vivo setting. In addition to their potential use to treat a variety of fibrotic and scarring diseases such as thyroid eye disease, idiopathic pulmonary fibrosis, and hypertrophic scarring, another future use of salinomycin may be to disrupt the tumor microenvironment by destroying or preventing myofibroblast formation in the tumor region.