|Home | About | Journals | Submit | Contact Us | Français|
A growing body of evidence suggests the involvement of connective tissue growth factor (CTGF) in the development and maintenance of fibrosis and excessive scarring. As the expression of this protein requires an intact actin cytoskeleton, disruption of the cytoskeleton represents an attractive strategy to decrease CTGF expression and, consequently, excessive scarring. The small heat-shock-related protein (HSP20), when phosphorylated by cyclic nucleotide signaling cascades, displaces phospho-cofilin from the 14-3-3 scaffolding protein leading to activation of cofilin as an actin-depolymerizing protein. In the present study, we evaluated the effect of AZX100, a phosphopeptide analogue of HSP20, on transforming growth factor-β-1 (TGF-β1)-induced CTGF and collagen expression in human keloid fibroblasts. We also examined the effect of AZX100 on scar formation in vivo in dermal wounds in a Siberian hamster model. AZX100 decreased the expression of CTGF and type I collagen induced by TGF-β1, endothelin, and lysophosphatidic acid. Treatment with AZX100 decreased stress fiber formation and altered the morphology of human dermal keloid fibroblasts. In vivo, AZX100 significantly improved collagen organization in a Siberian hamster scarring model. Taken together, these results suggest the potential use of AZX100 as a strategy to prevent excessive scarring and fibrotic disorders.
Transforming growth factor-β (TGF-β) is a multifunctional protein that regulates cell growth, differentiation, motility, and extracellular matrix production in the normal wound-healing process, but has also been implicated in excessive scar formation and fibrotic disorders (Yang et al., 2003; Leask and Abraham, 2004; Lu et al., 2005; Phan et al., 2005). TGF-β1 (and possibly TGF-β2) seems to be important for optimal wound healing in the first week after wounding, although persistent expression later in wound healing suggests it may be critical in the formation of hypertrophic scars (Lu et al., 2005; Kryger et al., 2007).
In fibroblasts, TGF-β1 stimulates the expression of connective tissue growth factor (CTGF). CTGF has been proposed to mediate some of the profibrotic effects of TGF-β, including fibroblast proliferation and the production of extracellular matrix proteins such as collagen and fibronectin (Duncan et al., 1999; Leask and Abraham, 2004, 2006). Elevated levels of CTGF were found in dermal fibrotic lesions such as scleroderma (Abraham et al., 2000) and keloids (Igarashi et al., 1996; Khoo et al., 2006), as well as in fibrotic lesions of other organs (Ito et al., 1998; Paradis et al., 1999). To further support the role of CTGF in fibrosis, it has been demonstrated that simultaneous co-injection of CTGF and TGF-β caused sustained fibrotic response in vivo, whereas injection of TGF-β alone caused only a transient response (Mori et al., 1999).
A growing body of evidence implicates cytoskeletal reorganization in the modulation of CTGF expression (Muehlich et al., 2007). Whereas stabilization of the actin cytoskeleton increases CTGF expression, actin depolymerization or increased monomeric actin decreases the expression of CTGF (Ott et al., 2003; Muehlich et al., 2007). In addition, increased levels of cAMP have been shown to abolish TGF-β-induced reorganization of the actin cytoskeleton as well as TGF-β-induced CTGF expression (Kothapalli et al., 1998; Santibanez et al., 2003). We have demonstrated that phosphorylation of the small heat-shock-related protein, heat shock protein 20 (HSP20), on serine 16, is one of the downstream events in the cAMP signaling pathway (Beall et al., 1999; Woodrum et al., 2003). A phosphorylated peptide analogue of the region surrounding the phosphorylation site of HSP20 (WLRRAS*APLPGLK) in which the serine is phosphorylated (S*), when attached to a protein transduction domain (PTD; to allow penetration into the cell) mimics the effects of activation of cyclic nucleotide signaling pathways (Komalavilas et al., 2008). This peptide, referred to as AZX100, relaxes smooth muscles from a variety of different species (Flynn et al., 2003; McLemore et al., 2004) and inhibits smooth muscle migration (Tessier et al., 2004). AZX100 also prevents a form of vascular scarring, intimal hyperplasia, in an organ culture model of human saphenous vein (Tessier et al., 2004). In addition, treatment of cultured cells with this peptide leads to changes in cellular morphology (loss of actin stress fibers and increases in monomeric g-actin) similar to the changes observed with activation of the cAMP signaling pathway (Dreiza et al., 2005; Komalavilas et al., 2008).
At least one of the mechanisms by which AZX100 exerts these cellular effects has been recently identified. Phosphopeptide analogues of HSP20, but not scrambled or nonphosphorylated control peptides bind to the intracellular scaffolding protein, 14-3-3 (Dreiza et al., 2005). The binding of phosphorylated HSP20 competes with the binding of another protein, phospho-cofilin, to 14-3-3 (Dreiza et al., 2005). When not bound to 14-3-3, cofilin becomes dephosphorylated and activated as an actin-depolymerizing protein. This leads to disruption of the actin cytoskeleton. The cytoskeleton represents, therefore, a promising therapeutic target to reduce the expression of CTGF and excessive scar formation. Hence, we postulated that AZX100 might ameliorate the fibrotic effects of TGF-β1, and possibly other mediators of fibrosis. To date, there are no therapeutic drugs available that target the cytoskeleton as a means to prevent excessive scarring and fibrosis.
The concentration of TGF-β1 necessary to increase CTGF expression in keloid fibroblasts after 24 hours was determined. No significant increase (P<0.05) in the CTGF expression was observed when cells were stimulated with 0.62 ng ml−1 TGF-β1. At 1.2 and 2.5 ng ml−1, TGF-β1 significantly (P<0.05) increased CTGF expression in a dose-dependent manner (Figure 1a and b).
We next evaluated the effect of AZX100 on the expression of CTGF and type I collagen in TGF-β1-stimulated cells. At 50 µm, AZX100 decreased the TGF-β1-induced expression of both CTGF and collagen by 33.6 ± 11.1 and 34.2 ± 18.5%, respectively (P<0.05) when cells were stimulated with 1.2 ng ml−1 of TGF-β1 (Figure 1c–e). Lower concentrations of AZX100 had no effect (data not shown). A comparable reduction of CTGF and collagen (24 ± 7.4 and 32.4 ± 11.1%, respectively) was observed with AZX100 (50 µm) treatment at a higher dose of TGF-β1 (2.5 ng ml−1; Figure 1c–e). The experiments were conducted with cells from three different patients, and consistent results were obtained.
To ensure that the PTD is necessary, but not sufficient for AZX100 activity, the following control peptides were used: the HSP20 phosphopeptide (without a PTD) and the PTD (YARAAARQARA) alone. Neither the HSP20 phosphopeptide nor the transduction domain alone had an effect on TGF-β1-induced expression of CTGF and collagen expression (Figure 1c–e).
Besides TGF-β, many other substances released at the wound-healing site have profibrotic activity and stimulate the expression of CTGF, including lysophosphatidic acid (LPA), thrombin, and endothelin (ET). Hence, the effect of AZX100 on CTGF and collagen expression induced by these additional profibrotic agents was examined. The expression of CTGF and collagen was 1.7-fold higher after ET (50 nm) stimulation compared to untreated (control) cells, whereas LPA (25 µm) stimulation led to a 2.0 and 2.1 times higher CTGF and collagen expression, respectively. Thrombin, on the other hand, had no effect on the expression of CTGF or collagen (Figure 2a–c). AZX100 treatment significantly inhibited the ET and LPA-induced increases in CTGF and collagen (Figure 2d–f). Thus, AZX100 inhibits CTGF and collagen expression induced by multiple profibrotic agonists.
As an intact actin cytoskeleton is needed for CTGF expression, we next examined the effect of AZX100 on stress fiber formation. Staining of the actin cytoskeleton with Alexa 450 phalloidin revealed that keloid fibroblasts had robust stress fibers, which were enhanced by TGF-β1 stimulation (Figure 3a, c and e). AZX100 treatment led to decreased stress fiber staining and altered the morphology of the cells that were either untreated (Figure 3b) or stimulated with TGF-β1 at 1.2 ng ml−1 (Figure 3d). The effect of AZX100 was not as strong on cells stimulated with 2.5 ng ml−1 of TGF (Figure 3f). No signs of cell detachment were observed after 3 days, showing that the effect of the peptide is not associated with cell death (Figure 3g). Similar decreases in stress fibers have been observed in response to AZX100 treatment when other mediators (LPA or serum) are used to stimulate stress fiber formation in 3T3 fibroblasts (Dreiza et al., 2005).
The effect of AZX100 treatment on the expression of α-smooth muscle actin (α-SMA), a hallmark of the differentiation of fibroblasts into myofibroblasts (Tomasek et al., 2002), was also evaluated. Keloid-derived fibroblasts constitutively expressed α-SMA, which was further increased upon TGF-β1 stimulation (Figure 3h–i). AZX100 treatment inhibited the TGF-β1-induced expression of α-SMA without affecting the expression of β-actin (Figure 3h–i). AZX100 also led to significant decreases in basal α-SMA expression.
Because AZX100-induced disruption of the actin cytoskeleton has been correlated with cofilin dephosphorylation due to the displacement of cofilin from its binding site to the protein 14-3-3 (Dreiza et al., 2005), the effect of AZX100 treatment on cofilin phosphorylation in TGF-β1-stimulated keloid fibroblasts was evaluated. We observed that the phospho-cofilin/cofilin ratio was increased 1.5 and 2.3 times after stimulation with TGF-β1 at 1.2 and 2.5 ng ml−1, respectively. AZX100 treatment decreased this ratio by 27.0 ± 4.1 and 24.4 ± 8.8% in cells stimulated with TGF-β1 at 1.2 and 2.5 ng ml−1, respectively (Figure 4).
The classical pathway activated by TGF-β1 implicated in increased CTGF expression involves the phosphorylation of SMAD2/3, which forms a complex with SMAD4 that translocates to the nucleus and binds CTGF promoters at the Smad consensus sequence, CAGAC (Leask and Abraham, 2004). To evaluate the potential involvement of this pathway in the effect of AZX100, we investigated whether AZX100 treatment had any effect on the phosphorylation of SMAD3. As expected, TGF-β1 increased SMAD3 phosphorylation (Figure 5). AZX100 did not alter TGF-β1-induced SMAD3 phosphorylation. This finding suggests that the observed effect of AZX100 on the cytoskeleton and CTGF is likely independent of SMAD signaling pathway.
A Siberian hamster model of scarring was selected because unlike normal laboratory rodents, this animal responds to environmental stressors, as evidenced by changes in behavior, body mass, and cortisol levels, which impact wound healing (Castro and Matt, 1997; Detillion et al., 2004). In addition, it has a broad upper back that provides a surgical site that is inaccessible to chewing and scratching by the animal. Incisions were treated with a single injection of either 1mm AZX100 or saline immediately after wound closure. Samples were harvested at 7, 14, and 21 days, processed, and stained with Masson’s trichrome. Representative images for control and treated incisions at all time points are shown in Figure 6a–f. As in other rodent studies of the neodermis (Shah et al., 1995), histological slices are shown at a magnification of 200 times. Masson’s trichrome stain was selected to reveal collagen fibers and more closely compare our results with previous published scoring scales (Shah et al., 1995; Beausang et al., 1998), which are in use in clinical trials conducted by Renovo (www.renovo.com). Slides were blindly examined by a pathologist for collagen orientation, density, and maturity (Figure 6g). For all categories, a score of zero corresponds to normal skin, thus the lower the score, the better the appearance of the scar (Beausang et al., 1998). Collagen fiber orientation was assessed by its similarity to the normal basket-weave pattern of adjacent control skin; however, the differences between AZX100 and control groups were not significant. Collagen fiber density was assessed by its similarity to the density of adjacent normal skin, and there is a significant improvement in collagen fiber density with AZX100 treatment at 21 days (P<0.05). Collagen fiber maturity was assessed by fiber thickness and length compared to that of adjacent normal skin, and there is also a significant improvement in collagen fiber maturity with AZX100 treatment at 21 days (P<0.05). The individual collagen scores were also combined into a total collagen score. At 7 and 14 days, both treated and control scars appeared similar with comparable collagen composite scores. However, at 21 days, a significant improvement emerged in the treated group relative to the control group, with collagen composite scores of 2.9 ± 0.3 and 4.6 ± 0.5, respectively.
In the present study we investigated the potential use of the peptide AZX100 for the prevention of excessive scarring and fibrotic disorders. AZX100 decreased TGF-β1-induced expression of CTGF and type I collagen in cultured keloid fibroblasts (Figure 1). As CTGF appears to be involved in inducing collagen expression (Duncan et al., 1999), the effect of AZX100 on collagen expression likely results from its ability to decrease the expression of CTGF. AZX100 did not alter TGF-β1-induced SMAD signaling (Figure 5), hence the effects of AZX100 appear to be SMAD independent.
The reduction in CTGF expression caused by AZX100 could be attributed to its effect on the actin cytoskeleton. AZX100 displaces the actin accessory protein cofilin from its binding site on the 14-3-3 adapter protein. Such displacement leads to dephosphorylation and activation of cofilin, which, in turn, mediates actin depolymerization and stress fiber disruption (Gohla and Bokoch, 2002; Dreiza et al., 2005). In a previous study (Dreiza et al., 2005), we showed that cofilin displacement occurs in the presence of 10 and 25 µm of AZX100; in this study, the concentration of AZX100 was higher (50 µm), and thus, enough to prevent the binding of cofilin to 14-3-3. Cofilin dephosphorylation by AZX100 was confirmed in this study in TGF-β1-stimulated cells (Figure 4). It seems, therefore that SMAD-independent cytoskeleton changes may be a major determinant of TGF-β1- induced CTGF production. Indeed, several studies have shown that disruption of the cytoskeleton by cytochalasin D reduces the stimulated expression of CTGF in renal fibroblasts and rat mesangial cells (Hahn et al., 2000; Heusinger-Ribeiro et al., 2001). Recently, Muehlich et al. (2007) have proposed a molecular mechanism for such a phenomenon. Using techniques that involve coexpression of mutant actins and different CTGF promoter constructs, these authors observed that the CTGF gene promoter contains an actin-sensitive site, which may be activated by monomeric actin. Their data indicate an inverse relationship between CTGF expression and depolymerized actin.
Another important observation of the present study is that AZX100 reduces the expression of α-SMA (Figure 3). This protein is a hallmark of the differentiation of fibroblasts into myofibroblasts, a process that appears to contribute substantially to the development of fibrosis (Desmouliere et al., 2005). Although the precise mechanism by which AZX100 reduces α-SMA expression in keloid fibroblasts remains to be determined, it should be mentioned that α-SMA expression increases with force tension (Wang et al., 2003). Thus, the ability of AZX100 to reduce α-SMA expression may also be related to the cytoskeleton-disrupting activity of AZX100.
Besides TGF-β, a multitude of cytokines, growth factors, and proteins are released at the wound-healing site. Many of these compounds stimulate profibrotic activity, such as LPA and ET (Hahn et al., 2000; Shi-Wen et al., 2004). Thus, it is important for a potential antifibrotic drug to be able to inhibit the effects of multiple profibrotic mediators. AZX100 was also able to reduce CTGF expression induced by LPA and ET (Figure 2). As the signaling pathways used by these molecules are diverse, the reduction of ET- and LPA-induced CTGF expression by AZX100 suggests that AZX100 is not specific for a certain pathway, but instead affects CTGF expression induced by the activation of several pathways, presumably because its mechanism of action is downstream and related to the cytoskeleton.
One of the fundamental problems associated with protein-based therapeutics is that large molecules such as proteins and peptides do not cross cell membranes. By using a PTD as a “carrier”, the attached active peptide can be transported across cell membranes (Schwarze et al., 1999; Flynn et al., 2003; Snyder and Dowdy, 2004). The PTD used in this study is an optimized version of TAT (Ho et al., 2001). We observed that without the PTD, the free HSP20 phosphopeptide had no effect on TGF-β1-stimulated expression of CTGF and collagen, demonstrating that the PTD is necessary for intracellular delivery of this active moiety of HSP20 (Figure 1).
Consistent with our in vitro observations regarding the molecular mechanisms of AZX100, in vivo data demonstrate a significant impact on collagen fiber orientation, density, and maturity (Figure 6). This improvement from a single injection in a rodent model of dermal scarring is comparable to another molecule currently under development to improve clinical scar appearance, TGF-β3 (Shah et al., 1995). Although TGF-β3 has been shown to have in vivo efficacy in dermal scarring with multiple injections in a rodent model, the molecular mechanism(s) of action are not clearly known. TGF-β3 is believed to promote epidermal homeostasis by preventing keratinocyte apoptosis in culture and reducing phorbol acetate-induced c-Jun N-terminal kinase activity (Lee et al., 1999). In addition, TGF-β3 has been shown to promote palatal shelf adhesion by inducing chondroitin sulfate proteoglycan expression in mice (Gato et al., 2002). Clinical trials using recombinant TGF-β3 (Justiva) in various dermal scarring indications have shown a nominal but significant improvement in scar appearance (see Renovo Annual Report, http://www.renovo.com/documents.asp?c_id=52), although the underlying molecular mechanisms are not clear.
Although many different pharmacologic agents have been studied to improve adult scarring, several of them have failed to show clinical efficacy despite in vitro and in vivo preclinical efficacy. For example, IFN-γ has been shown to decrease collagen synthesis in a wide variety of cells, including human dermal fibroblasts, human chondrocytes, and rat myofibroblasts (Lee et al., 1999; Phan et al., 2002; Ragoowansi et al., 2003; Amadeu et al., 2004; Liu et al., 2004; Wong, 2005; Al-Attar et al., 2006; Davison et al., 2006; Uysal, 2006). These results were confirmed using murine models of fibrosis (Sun et al., 1993; Katzung, 1996), but mixed results were obtained in the clinic (Berman and Flores, 1998; Davison et al., 2006).
To the best of our knowledge, therapeutic modalities like AZX100, in which a relevant phosphoprotein motif is directly delivered inside the cell and modulates downstream events that regulate scar formation, are previously unreported. Exploiting posttranslational (for example, phosphorylated) targets in this pathway may lead to finer, more specific control of collagen deposition than other pharmacologic agents that act through receptor-based modulation of signaling cascades, which typically amplify multiple enzymatic activities. The AZX100 molecule bypasses upstream events, thus potentially offering a treatment for abnormal wound-healing processes that respond to multiple upstream events. Taken together, these findings suggest that AZX100 may present a molecularly targeted, proteomic therapy for improving dermal scarring.
The peptides used in this study were AZX100 (YARAAARQARAWLRRAS* APLPGLK; S* denotes phosphoserine), the PTD (YARAAARQARA), and the free HSP20 phosphopeptide (WLRRAS* APLPGLK, without a transduction domain). Peptides were synthesized by UCB-Bioproducts-Lonza (Cambridge, MA) or at Arizona State University using an automated peptide synthesizer (Apex 396; Advanced ChemTech, Louisville, KY), and were purified by fast protein liquid chromatography (Akta Explorer; Amersham Pharmacia Biotech, Piscataway, NJ).
Human keloid fibroblasts were isolated from three different patients in accordance with the Declaration of Helsinki Principles 1975 and with protocols approved by the Human Subjects Institutional Review Board at Stanford University and maintained as previously described (Phan et al., 2005; Xia et al., 2006). Cells were grown at 37°C and 10% CO2 atmosphere in DMEM (Mediatech, Harndon, VA) containing 10% fetal bovine serum (Gibco, Carlsbad, CA) and additional penicillin and streptomycin (1%) in 10 cm2 dishes. When the cells reached 70% confluence, they were growth-arrested by reducing the concentration of fetal bovine serum in the media to 0.5% for 48 hours before the experiment.
Cells were either nonstimulated (control) or stimulated for 24 hours with TGF-β1 (R&D systems, Minneapolis, MN; 0.6–5 ngml−1) in the presence or absence of AZX100 (5–50 µm). In separate experiments, the cells were stimulated with other profibrotic mediators: thrombin (Sigma, St Louis, MO; 50–150 nm), ET (Sigma; 10–100 nm), or LPA (Sigma; 10–50 µm).
At the end of the experiments, adherent cells were rinsed with phosphate-buffered saline, and lysed using UDC buffer (8 M urea, 10mm dithiothreitol, 4% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate). Equal amounts of protein (20 µg per lane) were loaded on 15, 10, or 4–20% SDS–PAGE gels; proteins were electrophoretically separated, and then transferred to Immobilon membranes (Millipore, Billerica, MA). Membranes were probed overnight at 4°C with the following primary antibodies: rabbit anti-CTGF (Torrey Pines Biolabs, Houston, TX), rabbit anti-collagen (Cortex Biochem, San Leandro, CA), mouse anti-cofilin (BD Biosciences, San Jose, CA), rabbit anti-phosphorylated cofilin (a gift from Dr James Bamburg), mouse anti-α-SMA (Sigma), mouse anti-GAPDH (Chemicon, Temecula, CA), and rabbit anti-β-actin (Sigma). After washing, the membranes were incubated with appropriate secondary antibodies (Li-Cor) for 1 hour at room temperature. The protein–antibody complexes were visualized using the Odyssey direct infrared fluorescence imaging system (Li-Cor).
Human keloid fibroblasts were grown on glass coverslips in six-well dishes at a density of 2.0×105 cells per well. On the following day, cells were treated as indicated above, fixed with 4% paraformaldehyde for 1 hour, permeabilized with 0.1% Triton X for 15 minutes, and incubated for 1 hour with Alexa 350 phalloidin (Invitrogen, Eugene, OR) to stain the actin cytoskeleton. Images were obtained using a fluorescence microscope (Axiovert 200; AxioVision software, Carl Zeiss, Thornwood, NY).
All procedures were carried out with approval from the Arizona State University Institutional Animal Care and Use Committee. Siberian hamsters (4–6 months) were obtained from Kathleen Matt. The animals were anesthetized using ketamine–xylazine–acepromazine (21, 2.4, and 0.03mgkg−1, i.p.). The area on the upper back was shaved, scrubbed with chlorhexidine, and a single linear incision (2 cm in length) extending through the dermis, epidermis, and into the subcutaneous fat was made on the upper back. The incision was then closed with interrupted 4–0 nylon sutures. AZX100 (100 µl of 1mm) or saline was injected under the skin on each side of the incision immediately after closure (n=7–8 animals per group). Sutures were removed after 1 week, and animals were killed at 7, 14, and 21 days after surgery by inhalation of carbon dioxide. These time points are consistent with previous studies by Shah et al. (1995). The healing wounds and adjacent skin (as a control) were excised, histologically processed, and stained by Masson’s trichrome method to reveal collagen fibers. Each specimen was analyzed by light microscopy under multiple magnifications by a dermal pathologist, who blindly scored (in duplicate) the specimens for dermal collagen fiber orientation, density, and maturity using an established scoring scale (Beausang et al., 1998).
All protein expression data are presented as means ± SD. The Western blot bands were quantified by densitometry, and protein expression was normalized by the loading control (glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or β-actin) expression. One-way analysis of variance followed by Tukey’s test was used to compare experimental groups, with a significance level of P<0.05. Histology data were analyzed as the average of the duplicate scores. Group means and standard error values were calculated for the collagen fiber orientation, density, and maturity, as well as the total collagen score (sum of all three categories). Statistical analysis was performed with SigmaStat software (version 3.11.0; Systat Software Inc., San Jose, CA). Between-group differences at each time point were evaluated by the Mann–Whitney rank-sum test. Exact, two-sided P-values <0.05 were considered significant.
We thank Dr Jie Li for histological scoring. This work was supported by NIH RO1HL58027, STTR Grant R42 HL071309-04 and VA Merit Review award to C.M. Brophy. Oak Foundation supported the work of DPL, GPY, and MTL.
CONFLICT OF INTEREST
The patents for the cell permeant phosphopeptide analogues of HSP20 are owned by Arizona State University and the Veteran’s Administration. The intellectual property has been licensed to Orthologic Inc., Tempe, AZ and the authors (EF, PK, CF, AP, and CB) have financial interest in this technology.