Oncogene (2006) 25, 5416–5425 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Stat3 contributes to keloid pathogenesis via promoting collagen production, cell proliferation and migration

CP Lim1, T-T Phan2,3, IJ Lim2 and X Cao1

1Institute of Molecular and Cell Biology, Proteos, Singapore; 2Department of Surgery, National University of Singapore, Singapore and 3Department of Bioengineering, National University of Singapore, Singapore

Keloids,partially considered as benign tumors,represent extracellular matrix (ECM), promotion of angiogenesis, the most extreme example of cutaneous scarring that and the migration and proliferation of keratinocytes and uniquely afflicts humans as a pathological response to fibroblasts. In the event of excessive wound healing scar wound healing. It is characterized by excessive deposition formation occurs, ranging from hypertrophic scars to of collagen and other extracellular matrix components by keloids. The term ‘keloid’ was coined in 1806 to describe dermal fibroblasts. Upon cutaneous injury,cocktails of the crab claw-like appearance of the scar (Yang et al., chemokines,cytokines and growth factors are secreted 2003). Keloids are strictly defined as scars that spread temporally and spatially to direct appropriate responses beyond the boundaries of the original wound and do not from neutrophils,macrophages,keratinocytes and fibro- regress spontaneously (Cosman et al., 1961). In contrast, blasts to facilitate normal wound healing. Signal transdu- hypertrophic scars do not develop beyond the periphery cer and activator of transcription 3 (Stat3) is an oncogene of wound, with spontaneous softening and flattening and a latent transcription factor activated by various from remodeling over time (Ragoowansi et al., 2001). cytokines and growth factors. We investigated the possible These scars afflict humans exclusively, and the inciting role of Stat3 in keloid scar pathogenesis by examining skin trauma can range from small injuries such as ear skin tissue and cultured fibroblasts from keloid-scarred piercing and abrasions to more severe trauma such as patients. We observed enhanced expression and phosphor- surgery or burns. ylation of Stat3 in keloid scar tissue,and in cultured Keloids are aggressive in nature and Africans keloid fibroblasts (KFs) in vitro. Increased activation of and Asians appear to be more susceptible (Tuan and Janus kinase (Jak)2,but not Jak1,was detected in KFs, Nichter, 1998). Treatment and management of keloid and suppression of Jak2 by its inhibitor repressed Stat3 scars have been difficult, owing to their recurrence, and Y705 phosphorylation. Inhibition of Stat3 expression and although a myriad of treatments are available none are phosphorylation by short interfering RNA or Cucurbitacin fully effective (Tuan and Nichter, 1998; Alster and I resulted in the loss of collagen production,impaired Tanzi, 2003; Mustoe, 2004). Keloid scars are typified by proliferation and delayed cell migration in KFs. We show, overexuberant deposition of collagen and other ECM for the first time,a role of Stat3 in keloid pathogenesis. components such as fibronectin by keloid fibroblasts Inhibitors of Stat3 may be useful therapeutic strategies (KFs) (Diegelmann et al., 1979; Babu et al., 1989), with for the prospective treatment of keloid scars. overproduction of collagen types I and III mRNA in Oncogene (2006) 25, 5416–5425. doi:10.1038/sj.onc.1209531; keloid tissues (Naitoh et al., 2001). Keloid fibroblasts published online 17 April 2006 reportedly displayed an increased rate of proliferation upon wounding (Calderon et al., 1996). At the cellular Keywords: Stat3; keloid; scar; fibroblasts; collagen; level, increased levels of cytokines and growth factors, migration or their receptors, and sensitization of mitogenic response to these secreted factors after wounding in KFs have been suggested to play a role in keloid pathogenesis (Calderon et al., 1996). Transforming growth factor (TGF)-b1 and TGF-b2 were found to Introduction be elevated in KFs (Lee et al., 1999), and increased sensitivity to TGF-b1 in KFs increased fibronectin Wound healing is a regulated, yet complex, event (Babu et al., 1992) and collagen production, and cell including the breakdown of fibrin clots, degradation of proliferation (Younai et al., 1994). Platelet-derived growth factor (PDGF)a receptor expression was shown to be enhanced in KFs, which corresponded to increased Correspondence: Dr X Cao, Cell Signaling Lab, Institute of Molecular mitotic response upon exposure to all three PDGF and Cell Biology, 61 Biopolis Drive, Proteos, Singapore 138673, isoforms (Haisa et al., 1994). Overexpression of insulin- Singapore. E-mail: [email protected] like growth factor-I receptor detected in KFs appeared Received 10 October 2005; revised 25 January 2006; accepted 20 to enhance their invasiveness, but not fibroproliferation February 2006; published online 17 April 2006 (Yoshimoto et al., 1999). Elevated interleukin (IL)-6 Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5417 (Xue et al., 2000) and vascular endothelial growth factor Results (Wu et al., 2004) expression have also been observed in KFs. Enhanced Stat3 expression and phosphorylation in keloid Signal transducer and activator of transcription 3 scar tissue vs normal skin tissue in vivo (Stat3) is an oncogene and a latent transcription factor A thin layer of epidermis underlies a very thin layer of that is involved in diverse processes, including cell keratin in normal skin. In contrast, the keratin and proliferation and migration, inflammation, immune epidermal layers of keloid tissue are thicker, as shown by response and cell survival. It is activated by various the hematoxylin–eosin staining of paraffin sections growth factors and cytokines, including PDGF and IL-6 (Figure 1a). The dermis is also considerably thicker, (Darnell, 1997), by Tyr705 phosphorylation, which with compacted and irregular connective tissue com- enables the dimerization via its phosphotyrosine residue pared to normal skin, which appears more uniformly and reciprocal SH2 domain of its dimer partner. Stat3 stacked and regular. can homodimerize, or heterodimerize with Stat1, and To examine the expression of Stat3, cryosections from the dimer then translocates to the nucleus, binds to normal skin and skin overlying keloid scar were target DNA sequences and regulates the transcription of subjected to immunofluorescence using a polyclonal target genes (Darnell et al., 1994). Whereas Tyr705 Stat3 (C-20) antibody, performed in parallel with anti- phosphorylation is essential for Stat3 activation, Ser727 rabbit immunoglobulin (Ig)G as controls. As illustrated phosphorylation was shown to be required for maximal in Figure 1b, Stat3 could be detected in both normal gene activation (Wen et al., 1995). Tyr705 phosphoryla- skin and keloid tissue, whereas weak fluorescence was tion in cytokine receptors that lack intrinsic tyrosine observed with anti-rabbit IgG antibody. Stat3 expres- kinase activity is mediated by receptor-associated Janus sion was easily detected in the epidermal layer in both kinases (Jaks) (Darnell, 1997). Overexpression of a skin types, which is predominantly populated with constitutively activated Stat3 caused cell transformation keratinocytes. In the dermal layer, the connective tissue and tumorigenesis in mice (Bromberg et al., 1999). is interspersed with cellular components comprised Activation of Stat3 has been detected in various cancers primarily of fibroblasts, which are involved in the and tumor-derived cell lines (reviewed by Bowman et al., deposition of collagen fibers and other ECM compo- 2000). On the other hand, Stat3 knockout resulted in nents. A small population of fibroblast cells in the embryonic lethality (Takeda et al., 1997), prompting dermal layer of normal skin also showed Stat3 expres- further conditional tissue- or cell-specific knockout sion, whereas a much higher number of fibroblasts in analyses (Takeda and Akira, 2000). In particular, Stat3 keloid samples 25, 26, 31 and 48 showed Stat3 expression disrupted in keratinocytes using Cre recom- expression (Figure 1b and data not shown). binase driven by keratin 5-specific promoter showed Activation of Stat3 was also examined using phos- impaired wound healing and growth factor-dependent phoStat3 (pStat3) monoclonal antibody, performed in migration of Stat3-deficient epidermal cells in mice, parallel with anti-mouse IgG as controls. Weak pStat3 although the development of epidermis and hair follicles staining was observed in normal skin, whereas activa- appeared normal and proliferation remained unaffected tion of Stat3 was significantly increased in both dermis (Sano et al., 1999). and epidermis in keloid 26 (Figure 1c). In this paper, we hypothesized a role for Stat3 in To further verify the data above, tissue lysates of keloid scar pathogenesis as Stat3 was previously shown normal skin from five individuals and keloid scars from to function in wound healing (Sano et al., 1999). eight individuals were investigated for phosphorylation Moreover, Stat3 is a key molecule activated by various and expression of Stat3. Three samples from normal cytokines and growth factors, ligands that are known to skin showed low Tyr705 Stat3 phosphorylation, whereas be secreted by various cell components in response two showed moderate Tyr705 Stat3 phosphorylation to tissue injury during wound healing. Our results (Figure 1d). In contrast, all keloid tissue lysates, except showed an enhancement of Stat3 expression and/or for one, showed moderate to very high Tyr705 Stat3 phosphorylation in keloid tissues in vivo,aswellasin phosphorylation. The degree of Tyr705 Stat3 phosphor- cultured proliferating KFs. We detected concomitant ylation correlated well with the level of Stat3 expression. increased activation of Jak2, but not Jak1, in KFs As for Ser727 Stat3 phosphorylation, six out of eight compared to normal dermal fibroblasts. Inhibition of keloid tissue samples showed an increase in Ser727 Stat3 Tyr705 Stat3 phosphorylation was observed by Jak2 phosphorylation compared to all five normal skin inhibitor, AG 490, corroborating our hypothesis that samples. Jak2 mediates Stat3 phosphorylation. Finally, we demonstrated that the increased collagen production, Increased Stat3 activation in keloid fibroblasts compared cell proliferation and migration in KFs were suppressed to normal skin fibroblasts by Stat3 short interfering RNA (siRNA) and an The epidermis is populated with 95% keratinocytes and inhibitor of Jak2/Stat3, Cucurbitacin I (Blaskovich 5% non-keratinocyte cells comprising melanocytes, et al., 2003). Taken together, these data reveal an Langerhans cells and Merkel cells (Edmondson et al., important role of Stat3 in the pathogenesis of keloid 2003), whereas fibroblasts are the predominant cells in scars. Stat3 inhibition suppresses KF activity, and may the dermis, along with some endothelial and mast cells. serve as a therapeutic target for the treatment of keloid As we observed increased Tyr705 Stat3 phosphorylation fibrosis. in the dermal cells in keloid tissue compared to normal

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5418

Figure 1 Stat3 expression and activation are elevated in keloid skin tissues. (a) Hematoxylin–eosin stain of paraffin tissue sections of normal skin (NS1) and keloid scar (KS28). D, E and K indicate dermis, epidermis and keratin, respectively. (b) Cryosections of normal skin and keloid scar tissue were stained with either anti-rabbit immunoglobulinG (a-rabbit IgG) or polyclonal anti-Stat3 (C-20) (a- Stat3) and counterstained with 4,6-diamidino-2-phenylindole (DAPI). (c) Cryosections were probed with anti-mouse IgG (a-mouse IgG) or monoclonal anti-phosphoStat3 (pStat3) (a-pStat3) and counterstained with DAPI. Pictures were taken together with phase contrast and all scale bars represent 50 mM.(d) Equal amounts of tissue lysates from normal skin (NS) and keloid samples (KS) were subjected to Western blot analysis with anti-pY705 Stat3 antibody. The blot was stripped and reprobed with anti-pS727 Stat3, and subsequently with anti-Stat3 and anti-actin.

skin, we further investigated the activation of Stat3 in presented here as representatives. In NF5, very weak primary fibroblast cells derived from normal skin Tyr705 Stat3 phosphorylation was observed at day 1, and keloid tissues. Two normal fibroblast (NF) and increased at day 2 and sustained until day 5 (Figure 2b, three KF strains were examined for Stat3 activation. left panel). In contrast, strong Tyr705 Stat3 phosphor- Tyr705 phosphorylation of Stat3 was elevated from 2.3- ylation was observed in KF48 at day 1, which increased to 4.3-fold, with an average of three-fold, in KFs from day 2 to day 4, and decreased to day 1 level at day compared to NFs (Figure 2a). Tyr phosphorylation of 5. The Ser727 Stat3 phosphorylation profile was similar Stat1 and Stat5 was also examined, but was undetect- to Tyr705 Stat3 phosphorylation in both NF5 and able in any of the samples. Signal transducers and KF48. In addition, an elevated Stat3 expression was activators of transcription 1 protein expression was observed in KF48 compared to NF5. This kinetics in present, whereas Stat5 expression was weak, in all of the NF5 and KF48 is similarly observed in NF4 and KF8 samples. (Figure 2b, right panel), and KF8 also showed an To further examine the kinetics of Tyr705 and Ser727 enhanced pY705 Stat3 and pS727 Stat3 compared to Stat3 phosphorylation, NFs and KFs were cultured for NF4. 5 days under normal growth conditions. Five different Stat3 phosphorylation was further examined in sets of NF and KF were examined, and two are serum-free condition. We observed strong constitutive

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5419

Figure 3 Jak2 mediates Tyr705 Stat3 phosphorylation. (a) Stat3 phosphorylation by Jak2, but not Jak1. Equal amounts of total cell lysates from normal fibroblast, NF103 and keloid fibroblast, KF48, were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and Western blot analysis was Figure 2 Enhanced Stat3 phosphorylation and expression in performed using phosphoJak1 (pJak1) or phosphoJak2 (pJak2) keloid fibroblasts (KFs). (a) Western blot analyses of total cell antibodies. The blot was stripped and reprobed with Jak1 and Jak2 lysates from normal fibroblasts (NFs) and KFs with equal amounts antibodies, respectively, and subsequently with anti-actin for of protein, and antibodies as indicated. The graph on the upper normalization. Arrows indicate the position of the protein bands. right illustrates the pY705 Stat3 level normalized to Stat3 (b) NF103 and KF48 were serum-starved for 24 h and incubated expression, whereas the graph on the lower right depicts an with either dimethyl sulfoxide (DMSO) or 50 mM AG 490 for 7 average of the normalized pY705 Stat3 s.d. of the KFs and NFs another 27 h. Equal amounts of total cell lysates were resolved in presented. (b) Normal fibroblasts, NF5 and NF4, and KFs, KF48 SDS–PAGE and blotted with pY705 Stat3 antibody in Western and KF8, were cultured in serum-free media for 2 days before blot analysis. The blot was stripped and reprobed with Stat3 and stimulating with 10% fetal bovine serum. Total cell lysates were actin antibodies. harvested at day 1–5, and phosphorylations of Stat3 were analysed by Western blot after loading equal amounts of protein. (c) NF4 and KF48 fibroblasts were incubated in serum-free Dulbecco’s modified Eagles’s medium for 2 days before harvesting daily over 5 days to investigate the serum-independent phosphorylation of examined whether Jaks contribute to the enhanced Stat3. Equal amounts of total cell lysates were subjected to Western pTyr705 Stat3 in KF. Nearly confluent NFs and blot analysis with antibodies as indicated. Arrowheads indicate the KFs were maintained in serum-free media and har- position of the protein bands. vested, and activation of endogenous Jak1 and Jak2 was examined in Western blot analysis using anti-phospho- Jak1 (pJak1) or anti-phosphoJak2 (pJak2) antibody. Tyr705 and Ser727 Stat3 phosphorylation in KF48 As shown in Figure 3a, increased pJak2 was detected throughout the 5 days, compared to very weak in KF48 compared to NF103, whereas pJak1 signal phosphorylation in NF4 (Figure 2c). Although actin hardly detectable in either normal fibroblasts or expression was similar in all samples, Stat3 expression KFs, although Jak1 expression was present in both was elevated in KF48 compared to NF4. Nevertheless, a cell types. We also examined Src and epidermal growth dramatic increase in the Tyr705 Stat3 phosphorylation factor receptor (EGFR), as they are reported to be in KF48 was clearly observed after normalization to able to phosphorylate Stat3 also. However, phosphoSrc Stat3 expression. Overall, phosphorylations and expres- (Tyr416) and phosphoEGFR (Tyr845) signals sion of Stat3 were augmented in KFs and tissue lysates were undetectable in both NF and KF (data not compared to the samples from normal skin. Tyr705 and shown). Ser727 Stat3 phosphorylations in KFs also occur in a To confirm that Jak2 mediates Stat3 Tyr phosphor- constitutive and serum-independent manner. ylation, NF103 and KF48 were incubated with tyrphos- tin AG 490, a Jak2 tyrosine kinase inhibitor. pTyr705 Stat3 was inhibited by AG 490, but not by dimethyl Jak2, but not Jak1, activation is increased in keloid sulfoxide (DMSO) control, as shown in Figure 3b. This fibroblasts corroborates the finding that Jak2 mediates Tyr705 Janus kinase (Jak)1 and Jak2 are known Tyr kinases Stat3 phosphorylation in both normal and keloid skin that phosphorylate Stat3 in cytokine signaling. We fibroblasts.

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5420 Inhibition ofStat3 inhibits collagen production and Stat3 target genes Secretion of collagen by fibroblasts occurs during the process of tissue remodeling in wound healing, and enhanced collagen production has been implicated in fibrosis (Singer and Clark, 1999). Indeed, collagen production was elevated in six out of eight keloid tissue samples examined, whereas the other two showed a similar expression to those from five normal skin samples (Figure 4a). To investigate whether Stat3 plays any role in keloid pathogenesis, KFs were infected with retrovirus expres- sing Stat3 siRNAs. Of the four different targets for Stat3 siRNA selected, three displayed an inhibitory effect on Stat3 expression, namely Stat3 siRNA 1, 2 and 4, but not Stat3 siRNA 3, in a 293T-based amphotropic packaging cell line (Figure 4b, left panel). Inhibition of Stat3 expression by these three Stat3 siRNAs correlated with the inhibition of collagen production in KF48 (Figure 4b, right panel). Inhibition of Stat3 was also evaluated by incubating KF48 for 30 min with various doses of Cucurbitacin I, an inhibitor of Jak2/Stat3 activation (Blaskovich et al., 2003), or with DMSO as a control, followed by further incubation in normal growth media for 48 h. As shown in Figure 4c, a dose- dependent inhibition of pTyr705 Stat3 and collagen production was observed by Cucurbitacin I, with effective concentration as low as 1 mM. A dose-depen- dent decrease of Stat3 expression was also observed, although actin expression was rather similar in all samples. In addition, inhibition of Stat3 also resulted in a decreased expression of two Stat3 target genes (Bromberg et al., 1999), cell cycle regulator cyclin D1 and antiapoptotic Bcl-XL (Figure 4c). Taken together, Figure 4 Inhibition of Stat3 decreases collagen production and inhibition of Stat3 resulted in diminished collagen Stat3 target genes. (a) Tissue lysates from normal skin (NS) and production and Stat3 target genes. keloid samples (KS) were analysed for collagen production in Western blot analysis using equal amounts of protein. (b) Four targets of Stat3 short interfering RNA (siRNA) were cloned into Inhibition ofStat3 decreases cell proliferation pSuper.retro.puro vector and transfected into amphotropic 293T- Enhanced fibroblast proliferation is a characteristic of based retroviral packaging cell line, including vector alone. KF48 keloid tissue. In agreement with this, KFs grown in cells were subsequently infected with retrovirus harvested 48 h after normal culture conditions proliferated faster than NFs, as transfection. Total cell lysates from both amphotropic packaging observed in KF48 and KF107 compared to NF2 and cell line (left panel) and KF48 (right panel) were harvested and equal amounts of protein were analysed for Stat3, actin and NF4 (Figure 5a). To investigate whether Stat3 is involved collagen expressions in Western blot analyses. The inhibition of in the enhanced KF proliferation, KF48 was infected with collagen was further illustrated in the graph at the bottom right retrovirus expressing Stat3 siRNAs, and cell proliferation panel after normalization to actin by densitometry. (c) KF48 cells was examined. As shown in Figure 5b, inhibition of Stat3 were left untreated (0 mM), or treated with dimethyl sulfoxide (DMSO) or various doses of Cucurbitacin I (1, 5, 10 or 20 mM) for by Stat3 siRNA 4 decreased cell proliferation of KF48, 30 min, followed by further incubation in 10% fetal bovine serum/ compared to a non-infected control, vector control and Dulbecco’s modified Eagles’s medium for 48 h before harvesting. Stat3 siRNA 3, which do not inhibit Stat3. Cell Equal amounts of total cell lysates were resolved by sodium proliferation was also investigated by treating KF48 cells dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) with various doses of Cucurbitacin I or with DMSO and examined for the phosphorylation and expression of Stat3, expression of cyclin D1, Bcl-XL, actin and production of collagen control before incubating in normal growth media and in Western blot analyses. Arrows/arrowheads indicate the position examined daily over 4 days. Cucurbitacin I concentration of the protein bands. as low as 1 mM was able to inhibit cell proliferation of KF48 compared to DMSO and untreated control (0 mM), whereas Cucurbitacin I exposure at 5 mM and above and fibroblasts to resurface areas of skin loss. We seemed to cause cell death (Figure 5c). further examined the migration potential of NFs and KFs using the scratch assay. Cells were first cultured to Increased cell migration in keloid fibroblast full confluence, and treated with 10 mg/ml mitomycin C Wound healing is a complex multi-step process. It for 2 h before scratch-wounding with a yellow pipette involves the migration and proliferation of keratinocytes tip. Pictures were taken at 0 and 14 h later. As observed

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5421 transmembrane assay using the 8.0 mM pore size polyvinylpropylene-free polycarbonate Transwell mem- branes, performed in triplicates. Three non-overlapping fields per replicate were captured and quantified as the mean of the total migrating cells in the triplicates. Once again, the three KF samples showed significant in- creased migration compared to the three NF samples (Figure 6c).

Cell migration in keloid fibroblasts is inhibited by Stat3 short interfering RNA and Cucurbitacin I To examine the role of Stat3 in cell migration of skin fibroblasts, KF48 cells were infected with retrovirus expressing Stat3 siRNAs and scratch assays were performed as described previously. In Figure 7a, Stat3 siRNA 1, 2 and 4, all of which inhibit Stat3 expression, were able to impede cell migration of KF48, whereas vector alone or Stat3 siRNA 3, which did not inhibit Stat3 expression and served as a negative control, had no rescuing effect. Three non-overlapping fields were acquired, and graphically illustrated as mean7s.d. in Figure 7b. Transmembrane cell migration assays were also performed in triplicates as described earlier, and quantified as mean7s.d. (Figure 7c). In agreement with the scratch assay, Stat3 siRNA 1, 2 and 4 were able to reduce the cell migration of KF48, whereas vector and Stat3 siRNA 3 did not. The right panels in Figure 7b and c depict immunoblots of Stat3 and actin to show the efficiencies of inhibition by the Stat3 siRNAs. The role of Stat3 in cell migration was also examined using Cucurbitacin I in the scratch assay. KF48 cells were first incubated with 10 mg/ml mitomycin C for 2 h, before treating for 30 min with DMSO or 1 mM Cucurbitacin I, the effective dose for inhibiting Stat3 activation as observed in Figure 4c. The scratch wounding was performed, followed by further incuba- tion in normal growth media. Pictures were taken at 0, 22 and 48 h after wounding. As shown in Figure 7d, inhibition of Stat3 activation by Cucurbitacin I decel- Figure 5 Inhibition of Stat3 decreases cell proliferation. (a) Increased cell proliferation in keloid fibroblasts (KFs) compared erated the migration of KF48, compared to DMSO to normal fibroblasts (NFs). Normal fibroblasts, NF2 and NF4 control. Quantification of the mean7s.d. of three non- and KFs, KF48 and KF107, were seeded at 3000 cells/well in 96- overlapping fields is depicted in Figure 7e, and showed a well plates in quadruplicates in normal growth media, and cell significant reduction in cell migration at 22 h and a proliferation was examined daily up to 6 days using XTT cell proliferation kit according to the manufacturer’s instruction. (b) delayed migration at 48 h by Cucurbitacin I compared KF48 fibroblasts were seeded in 96-well plates, infected with to DMSO control (*Po0.001). These results are in retroviral Stat3 short interfering RNAs (siRNAs) the next day for agreement with our recent report showing defective cell 18 h, before changing to 10% fetal bovine serum (FBS)/Dulbecco’s migration in Stat3-deficient mouse embryonic fibro- modified Eagles’s medium (DMEM). Cell proliferation was blasts (Ng et al., 2006). examined daily from the day of infection. (c) KF48 fibroblasts were seeded in 96-well plates and either left untreated (0) or treated with dimethyl sulfoxide (DMSO) or various doses of Cucurbitacin I, and cell proliferation was examined daily over 4 days. Discussion in Figure 6a, all three NFs, NF2, NF4 and NF103, Keloid scars do not occur spontaneously, but are the showed a slower migration pattern towards the midline result of excessive wound healing after cutaneous of scratch wound compared to three KFs, KF43, KF48 injuries, exhibiting abnormalities in cell migration, and KF107. Three non-overlapping fields were captured proliferation, inflammation, synthesis and secretion of and quantified as mean7s.d., and results are presented collagen and other ECM components, and remodeling in Figure 6b. All three KF samples displayed enhanced of the wound matrix (Singer and Clark, 1999). In this cell migration rates compared to the three NF samples paper, we investigated the role of Stat3 in keloid examined. Cell migration was also examined by the pathogenesis in both normal skin and keloid tissue

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5422

Figure 6 Increased cell migration in keloid fibroblasts (KFs) compared to normal fibroblasts (NFs). (a) Three NFs, NF2, NF4 and NF103, and three KFs, KF43, KF48 and KF107, were grown in 60 mm dishes until confluent, and treated with 10 mg/ml mitomycin C for 2 h, before a scratch wound was introduced using a yellow pipette tip. Cells were incubated for a further 14 h in normal growth media. Pictures were taken at 0 and 14 h after wounding. (b) Pictures from three non-overlapping fields from (a) were taken and cells migrated into the wound site were quantified. Data are presented as mean7s.d. *Po0.05 indicates KF43 vs NF4 and NF103, whereas **Po0.005 and ***Po0.001 refer to three NFs vs KF48 and KF107, respectively, by Student’s t-test. (c) Cell migration was also examined using Transwell assay, performed in triplicates. KF48 cells were seeded in serum-free Dulbecco’s modified Eagles’s medium (DMEM) at the top of the insert with 10% fetal bovine serum (FBS)/DMEM as chemoattractant in the bottom chamber. Cells from three non-overlapping fields from each replicate that migrated from top to bottom of the insert after 48 h were counted and the mean of the triplicates7s.d. was presented in the graph. *Po0.001 and **Po0.005 indicate three NFs vs KF43, and KF48 and KF107, respectively, by Student’s t-test.

sections, as well as in fibroblast cell cultures. Our data granulation tissue, neovascularization) and tissue remo- showed that Stat3 expression and phosphorylation are deling, events that are governed by a well-orchestrated enhanced in keloid tissue sections, tissue lysates and temporal and spatial secretion of various chemokines, fibroblasts (Figures 1b–d and 2), and Stat3 Tyr cytokines and growth factors by different cell compo- phosphorylation is mediated by Jak2 (Figure 3). Inhibi- nents (Singer and Clark, 1999; Werner and Grose, tion of Stat3 using Stat3 siRNA or by an inhibitor of 2003). Stat3 is a key molecule activated in response to Jak2/Stat3 activation, Cucurbitacin I, decreased the most of these ligands. Here, we observed enhanced collagen production, proliferation and migration of expression and phosphorylation of Stat3 in keloid KFs, to a level comparable to that of NFs (Figures 4–7). tissues and fibroblasts, suggesting a lack of a stop signal Fibroblasts are responsible for the synthesis, deposi- or its downregulation, and/or an incessant stimulation tion and remodeling of the ECM. We report here that by cytokines or growth factors. Indeed, some cytokines overexpression and phosphorylation of Stat3 plays a that activate Stat3 have been reported to be enhanced in novel role in the excessive collagen deposition by keloid tissue such as IL-6 (Xue et al., 2000). On the other fibroblasts, leading to fibrotic scars such as keloids. hand, whether there is any mutation in Stat3 in keloid Inhibition of Stat3 expression by siRNA or Cucurbita- tissues that predisposes to its overexpression remains to cin I showed a corresponding decline in collagen be determined. As Stat3 is involved in all the different expression (Figure 4b and c), suggesting that collagen phases of wound healing, we speculate that normal production may be regulated by Stat3 at the transcrip- regulation of Stat3 is required throughout the process, tional level. Preliminary scanning of the promoter and dysregulation of Stat3 signaling will tip the balance region of both collagen type I a1, COL1A1, and towards impaired wound healing and fibrosis. Activa- collagen type III a1, COL3A1, revealed some putative tion of Stat3 by proinflammatory cytokines also Stat3 binding sites (CP Lim and X Cao, unpublished). It suggests a role of Stat3 in the progression of scarring will be interesting to investigate whether Stat3 has a at the inflammation phase, as fetal tissues that lack direct effect on the transcription of procollagen type I inflammation undergo perfect scarless wound healing, and III, the two collagens that are found to be highly and scarring ensued in the presence of an acute synthesized in keloids (Abergel et al., 1985; Naitoh et al., inflammatory infiltrate (Yang et al., 2003). 2001). Owing to their autonomous nature, keloids have also The process of cutaneous wound healing can be been considered as benign tumors. Stat3, on the other arbitrarily divided into three phases – inflammation, hand, possesses oncogenic potential, as overexpression tissue formation (re-epithelialization, formation of of a constitutively dimerized/active Stat3 mutant

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5423

Figure 7 Decreased cell migration owing to inhibition of Stat3. (a) KF48 cells were infected with retrovirus expressing Stat3 short interfering RNA (siRNA), and scratch-wound assay was performed as described previously in Figure 6a. Pictures were taken at 0 and 15 h after wounding. (b) Data represent mean7s.d. of cells migrated into the wounding site from three non-overlapping fields from (a). *Po0.005 refers to Stat3 siRNA 1, 2 and 4 vs vector, by Student’s t-test, and NS indicates not significant. Inhibition of Stat3 by Stat3 siRNAs is shown in the Western blot at the right panel. (c) KF48 cells were infected with retrovirus harboring Stat3 siRNAs and similar cell migration assay using Transwell inserts was performed as mentioned in Figure 6c. *Po0.001 in comparison to vector by Student’s t-test, and n.s. indicates not significant. Inhibition of Stat3 by Stat3 siRNAs is shown in the Western blot at the right panel. (d) KF48 cells were pretreated with 10 mg/ml mitomycin C for 2 h, followed by either dimethyl sulfoxide (DMSO) or 1 mM Cucurbitacin I for 30 min, before cells were wounded using yellow pipette tips. Cells were further incubated in 10% fetal bovine serum (FBS)/ Dulbecco’s modified Eagles’s medium (DMEM) normal growth media, and pictures were taken at 0, 22 and 48 h after wounding. (e) Data represent mean7s.d. of cells migrated into wounding sites from three non-overlapping fields from (d). *Po0.001 by Student’s t- test.

(Stat3C) caused cellular transformation and tumor Inhibitors of Stat3 expression or phosphorylation were formation (Bromberg et al., 1999), and Stat3 was shown able to suppress KF collagen production, cell prolifera- to be essential for developing skin cancer in mice (Chan tion and migration to a level comparable to NFs. et al., 2004). Recently, overexpression of Stat3C in A549 Besides siRNA and Cucurbitacin I, the efficacy of other carcinoma cells followed by microarray analyses re- Stat3 inhibitors could additionally be explored, includ- vealed that Stat3 regulated genes are common to both ing other pharmacological agents such as Cucurbitacin wound healing and cancer, including cell invasion/ Q, which is an analog of Cucurbitacin I and a selective migration, angiogenesis and remodeling of ECM, based Stat3 inhibitor (Sun et al., 2005), STA-21 (Song et al., on the concept that tumors are ‘wounds that do not 2005), Stat3 decoy oligonucleotide (Leong et al., 2003), heal’ (Dauer et al., 2005). This supports our observation or phosphotyrosyl peptides (Turkson et al., 2001), all of that hyperactive Stat3 plays a role in wound healing which either exhibited antitumor activity, increased gone awry, as seen in keloids. However, there is still a apoptosis and inhibited proliferation or suppressed cell distinct difference between wound healing and skin transformation of cancer-derived or v-Src-transformed cancer, as cancer cells metastasize and scars do not. cell lines. These potential cancer therapeutic agents that In summary, we have shown for the first time a novel target the inhibition of Stat3 may also be useful for the role of Stat3 in the pathogenesis of keloid scars. prospective clinical treatment of keloid scars.

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5424 Materials and methods diluted 1:200 in FDB to 1 mg/ml for monoclonal pStat3 (B7), polyclonal Stat3 (C-20), anti-mouse IgG or anti-rabbit IgG Normal and keloid tissue/fibroblasts antibodies. Tissue sections were washed again as described and Ethical approvals from the National University of Singapore incubated for 1 h at room temperature, with either FluoroLink Institutional Review Board and National Healthcare Group Cy3-labeled goat anti-rabbit IgG (PA43004; Amersham Domain Specific Review Boards, and informed consent from Biosciences, Pistacaway, NJ, USA) or anti-mouse IgG-Cy3 patients, were obtained before surgical excision of skin and (AP124C; Chemicon International, Tencala, CA, USA) keloid tissue. All patients had received no prior treatment for diluted at 1:400 in FDB. Sections were washed again and keloid lesions. A full history was taken and physical mounted with Vectashield mounting medium containing examination was performed in addition to color-slide photo 4,6-diamidino-2-phenylindole (DAPI) (H-1200; Vector documentation of all keloid lesions. The procedure for the Laboratories Inc., Burlingame, CA, USA). Pictures were isolation and culture of fibroblasts from skin specimens was as taken with Leica DM4000 B microscope. described previously (Lim et al., 2002). Isolated fibroblasts were grown in Dulbecco’s modified Eagles’s medium (DMEM) XTT proliferation assay containing 4500 mg/ml glucose, supplemented with 10% fetal Cells seeded at 3000 cells/well in 96-well plates and grown in bovine serum (FBS), 2 mML-glutamine, 100 U/ml penicillin normal growth conditions, or infected with retrovirus expres- and 100 ng/ml streptomycin. Normal skin and keloid scar sing Stat3 siRNAs, were assayed for proliferation using the tissue, and normal and keloid-derived fibroblasts, used in this XTT sodium 30-[1-(phenylaminocarbonyl)-3,4-tetrazolium]-bis work are listed in Supplementary data. (4-methoxy-6-nitro) benzene sulfonic acid hydrate prolifera- For the isolation of tissue lysates for Western blot analysis, tion kit (Roche, Mannheim, Germany) according to the frozen tissue specimens were lysed in lysis buffer containing manufacturer’s instructions. The readings were taken at 20 mM Tris-HCl, pH 7.5, 1% Triton X-100, 100 mM NaCl, 450 nm and referenced against 690 nm, 2 h after the addition 0.5% Nonidet P-40 and 1 mg/ml protease inhibitor cocktail of the XTT reagent. Each sample was performed at (Boehringer Mannheim, Germany). Equal amounts of proteins quadruplicates in two to three independent experiments. Data were resolved by sodium dodecyl sulfate–polyacrylamide gel are presented as mean7s.d. electrophoresis in Western blot analysis. Migration assay Antibodies and reagents For migration assay using the scratch method, cells were Polyclonal anti-phosphoTyr705 Stat3, monoclonal anti- grown to confluence and incubated with 10 mg/ml mitomycin C phosphoSer727 Stat3, anti-pJak1 (Tyr1022/1023), anti-pJak2 for 2 h, before a scratch wound was introduced using a yellow (Tyr1007/1008), anti-phosphoSrc (Tyr416), anti-phospho- pipette tip. Cells were further incubated in normal growth EGFR (Tyr845) and anti-Bcl-XL antibodies were purchased media and pictures were taken at 0, 14, 15, 22 or 48 h after from Cell Signaling Technology, whereas monoclonal pStat3 wounding. Three non-overlapping fields were captured by (B7) (sc-8059), polyclonal Stat3 (C-20) (sc-482), monoclonal Axiovert 135 microscope from Zeiss (Jena, Germany) and cells cyclin D1 (sc-20044), anti-mouse IgG (sc-2025) and anti-rabbit migrated into the wound site were counted and presented as IgG (sc-2027) antibodies were acquired from Santa Cruz mean7s.d. Biotechnology Inc. (Santa Cruz, CA, USA). Monoclonal Migration assay was also performed using the Transwell anti-Stat3 antibody was purchased from BD Transduction 8.0 mM pore size polycarbonate membrane (Corning Inc., Laboratories (Lexington, KY, USA). Polyclonal anti-actin Acton, MA, USA). 1 Â 104 cells/well were seeded at the top antibody and mitomycin C were purchased from Sigma (Saint of the insert in 0.2 ml serum-free DMEM, with 0.5 ml 10% Louis, MO, USA) and monoclonal anti-human collagen I, II, FBS/DMEM in the bottom chamber as chemoattractant. III was purchased from MONOSAN (Uden, Netherlands). After 24 h, cells were washed once in cold PBS, before AG 490 was obtained from Calbiochem (San Diego, CA, fixing with 4% paraformaldehyde in PBS for 15 min at USA), whereas Cucurbitacin I was acquired from Galchimia room temperature. Cells were permeabilized with 0.1% (A Corun˜ a, Spain). Triton X-100 in PBS for 3 min at room temperature, and stained with hematoxylin for 10 min at room temperature, Hematoxylin–eosin staining followed by destaining with water thrice. Cells from the Tissue sections were fixed with 4% paraformaldehyde in top of the inserts were removed by cotton swabs, and inserts phosphate-buffered saline (PBS), embedded in paraffin and were left to dry. The inserts were mounted onto slides sectioned. Tissue sections were stained with hematoxylin for 5– and three non-overlapping fields of cells migrated to the 8 min, washed with running water, quickly destained with acid bottom of the inserts were captured using Leica DM4000 B ethanol containing 1% hydrochloric acid and 70% ethanol, microscope. washed with running water and counterstained with eosin– ethanol for 3–5 min. Tissue sections were then rehydrated by Retroviral Stat3 short interfering RNA Four different targets gradual immersion in 70, 80, 95 and 100% ethanol, cleared against Stat3 (Genbank Accession number NM_139276) were with xylene and finally mounted in entellan. Pictures were selected and cloned into BglII and XhoI into pSUPER. taken with Leica DM4000 B microscope. retro.puro from OligoEngine, Seattle, WA, USA and se- quenced. The four targets are named Stat3 siRNA 1 Immunofluorescence (nucleotides (nt) 461–480), Stat3 siRNA 2 (nt 1264–1283), Cryosections were fixed with 4% paraformaldehyde in PBS for Stat3 siRNA 3 (nt 364–383) and Stat3 siRNA 4 (nt 1662– 15–30 min at room temperature, washed thrice with 0.1% 1681). The corresponding primers with a 9-nt short hairpin in Triton X-100 in PBS for 10 min each, and blocked with FDB the middle (bold) are F1 50-GATCCCCAGTCGAATGTT (Fluorescence Dilution Buffer) containing 5% normal goat CTCTATCATTCAAGAGATGATAGAGAACATTCGACT 0 0 serum, 2% FBS, 2% bovine serum albumin, 1 mM CaCl2 and TTTTTC-3 and R1 5 -TCGAGAAAAAAGTCGAATGTT 1mM MgCl2 in PBS. Tissue sections were washed as above and CTCTATCATCTCTTGAATGATAGAGAACATTCGACT incubated for 1 h at room temperature, with primary antibody GGG-30,F250-GATCCCCGGCGTCCAGTTCACTACTA

Oncogene Increased Stat3 phosphorylation and expression in keloid CP Lim et al 5425 TTCAAGAGATAGTAGTGAACTGGACGCCTTTTTC-30 infection, or XTT cell proliferation or migration assay, which and R2 50-TCGAGAAAAAGGCGTCCAGTTCACTACTA was performed by replating the cells into 8.0 mM Transwell TCTCTTGAATAGTAGTGAACTGGACGCCGGG-30,F3 membrane and fixed 24 h later as described above. 50-GATCCCCGATTGGGCATATGCGGCCATTCAAGAGA TGGCCGCATATGCCCAATCTTTTTC-30 and R3 50-TCG AGAAAAAGATTGGGCATATGCGGCCATCTCTTGAA Acknowledgements TGGCCGCATATGCCCAATCGGG-30, and F4 50-GATC CCCGCGTCCATCCTGTGGTACATTCAAGAGATGTACC We are grateful to Dr Garry Nolan for the retroviral Phoenix ACAGGATGGACGCTTTTTC-30 and R4 50-TCGAGAA packaging cell line. We thank Dr Birgit Lane for helpful AAAGCGTCCATCCTGTGGTACATCTCTTGAATGTAC discussions, Zheng Qi, Guo Ke, Li Jie and Bin Qi from the CACAGGATGGACGCGGG-30. These vectors containing Histology Unit at the Institute of Molecular and Cell Biology, Stat3 siRNA or empty vector were transfected into 293T- Singapore, for their kind assistance. The research work at the based amphotropic packaging cell line (Phoenix-Ampho) for Institute of Molecular and Cell Biology was supported by the 7–9 h, and the amphotropic retroviruses were harvested 48 h Agency of Science, Technology and Research (A*STAR), later, pelleted to remove non-adherent cells and cellular debris, Singapore. This research was also supported by A*STAR- and filtered through a 0.45 mM cellulose acetate membrane. Biomedical Research Council (BMRC) grant 02/1/21/19/106 The KFs were infected overnight in the presence of 4 mg/ml (to Lim IJ and Phan TT). Xinmin Cao is also an adjunct polybrene, and replaced with fresh 10% FBS/DMEM the professor at the Department of Biochemistry, National following day. Cells were assayed for Western blot 48 h after University of Singapore, Singapore.

References

Abergel RP, Pizzurro D, Meeker CA, Lask G, Matsuoka LY, Mustoe TA. (2004). BMJ 328: 1329–1330. Minor RR et al. (1985). J Invest Dermatol 84: 384–390. Naitoh M, Hosokawa N, Kubota H, Tanaka T, Shirane H, Alster TS, Tanzi EL. (2003). Am J Clin Dermatol 4: 235–243. Sawada M et al. (2001). Biochem Biophys Res Commun 280: Babu M, Diegelmann R, Oliver N. (1989). Mol Cell Biol 9: 1316–1322. 1642–1650. Ng DC, Lin BH, Lim CP, Huang G, Zhang T, Poli V et al. Babu M, Diegelmann R, Oliver N. (1992). J Invest Dermatol (2006). J Cell Biol 172: 245–257. 99: 650–655. Ragoowansi R, Cornes PG, Glees JP, Powell BW, Moss AL. Blaskovich MA, Sun J, Cantor A, Turkson J, Jove R, Sebti (2001). Br J Plast Surg 54: 504–508. SM. (2003). Cancer Res 63: 1270–1279. Sano S, Itami S, Takeda K, Tarutani M, Yamaguchi Y, Miura Bowman T, Garcia R, Turkson J, Jove R. (2000). Oncogene 19: H et al. (1999). EMBO J 18: 4657–4668. 2474–2488. Singer AJ, Clark RA. (1999). N Engl J Med 341: 738–746. Bromberg JF, Wrzeszczynska MH, Devgan G, Zhao Y, Pestell Song H, Wang R, Wang S, Lin J. (2005). Proc Natl Acad Sci RG, Albanese C et al. (1999). Cell 98: 295–303. USA 102: 4700–4705. Calderon M, Lawrence WT, Banes AJ. (1996). J Surg Res 61: Sun J, Blaskovich MA, Jove R, Livingston SK, Coppola D, 343–347. Sebti SM. (2005). Oncogene 24: 3236–3245. Chan KS, Sano S, Kiguchi K, Anders J, Komazawa N, Takeda K, Akira S. (2000). Cytokine Growth Factor Rev 11: Takeda J et al. (2004). J Clin Invest 114: 720–728. 199–207. Cosman B, Crikelair GF, Gaulin JC, Lattes R. (1961). Plast Takeda K, Noguchi K, Shi W, Tanaka T, Matsumoto M, Reconstr Surg 27: 335–358. Yoshida N et al. (1997). Proc Natl Acad Sci USA 94: 3801– Darnell Jr JE. (1997). Science 277: 1630–1635. 3804. Darnell Jr JE, Kerr IM, Stark GR. (1994). Science 264: Tuan TL, Nichter LS. (1998). Mol Med Today 4: 19–24. 1415–1421. Turkson J, Ryan D, Kim JS, Zhang Y, Chen Z, Haura E et al. Dauer DJ, Ferraro B, Song L, Yu B, Mora L, Buettner R et al. (2001). J Biol Chem 276: 45443–45455. (2005). Oncogene 24: 3397–3408. Wen Z, Zhong Z, Darnell Jr JE. (1995). Cell 82: 241–250. Diegelmann RF, Cohen IK, McCoy BJ. (1979). J Cell Physiol Werner S, Grose R. (2003). Physiol Rev 83: 835–870. 98: 341–346. Wu Y, Zhang Q, Ann DK, Akhondzadeh A, Duong HS, Edmondson SR, Thumiger SP, Werther GA, Wraight CJ. Messadi DV et al. (2004). Am J Physiol Cell Physiol 286: (2003). Endocr Rev 24: 737–764. C905–C912. Haisa M, Okochi H, Grotendorst GR. (1994). J Invest Xue H, McCauley RL, Zhang W. (2000). J Surg Res 89: Dermatol 103: 560–563. 74–77. Lee TY, Chin GS, Kim WJ, Chau D, Gittes GK, Longaker Yang GP, Lim IJ, Phan TT, Lorenz HP, Longaker MT. MT. (1999). Ann Plast Surg 43: 179–184. (2003). Wound Repair Regen 11: 411–418. Leong PL, Andrews GA, Johnson DE, Dyer KF, Xi S, Mai JC Yoshimoto H, Ishihara H, Ohtsuru A, Akino K, Murakami R, et al. (2003). Proc Natl Acad Sci USA 100: 4138–4143. Kuroda H et al. (1999). Am J Pathol 154: 883–889. Lim IJ, Phan TT, Bay BH, Qi R, Huynh H, Tan WT et al. Younai S, Nichter LS, Wellisz T, Reinisch J, Nimni ME, Tuan (2002). Am J Physiol Cell Physiol 283: C212–C222. TL. (1994). Ann Plast Surg 33: 148–151.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene