Journal of Dermatological Science 64 (2011) 199–209

Keloid fibroblasts are more sensitive to Wnt3a treatment in terms of elevatedcellular growth and fibronectin expression

Alvin Wen Choong Chua a,b,c,*, Shu Uin Gan c, Yixin Ting a,b, Zhenying Fu c, Che Kang Lim d,Colin Song b, Kanaga Sabapathy e, Toan Thang Phan c,f

a Skin Bank Unit, Singapore General Hospital, Singaporeb Department of Plastics, Reconstructive and Aesthetic Surgery, Singapore General Hospital, Singaporec Department of Surgery, National University of Singapore, Singapored Department of Clinical Research, Singapore General Hospital, Singaporee National Cancer Centre, Singaporef Faculty of Dentistry, National University of Singapore, Singapore

A R T I C L E I N F O

Article history:

Received 20 April 2011

Received in revised form 25 August 2011

Accepted 14 September 2011

Keywords:

Keloid fibroblasts

Wnt

SFRP1

Cellular growth

Fibronectin

A B S T R A C T

Background: Current evidence suggests the potential role of Wnt signalling in keloids pathogenesis but

such literature remains scanty. We hypothesize that Wnt signalling is upregulated in keloid fibroblasts

(KFs) and this promotes cellular growth, migration and extracellular matrix (ECM) production in such

fibroblasts.

Objectives: To verify the downregulation of secreted frizzled-related protein 1 (SFRP1), a Wnt inhibitor

and test KFs sensitivity to Wnt3a treatment compared to NFs in terms of activation of Wnt/b-catenin,

cellular growth, migration and ECM expressions. Next, to investigate if ectopic expression of SFRP1 and

treatment of quercetin in KFs can reverse their phenotypes.

Methods: Quantitative Real-time PCR and western blotting were used to verify SFRP1 expression in NFs

and KFs. The fibroblasts were tested with Wnt3a conditioned media and its effects were tested for (1) the

cells’ sensitivity to direct Wnt signalling via the activation of TCF reporter assay and protein expression of

b-catenin, (2) cellular growth, (3) cell migration and (4) expressions of ECM components. Finally KFs

were stably transduced with SFRP1 and treated with 2 doses of quercetin.

Results: Lower levels of SFRP1 were confirmed at mRNA and protein levels in KFs which partly explained their

sensitivity to Wnt3a treatment in terms of higher Wnt activation, cellular growth and fibronectin expression.

Interestingly, Wnt3a did not promote higher cell migration rate and increase in collagen I expression. Ectopic

expression of SFRP1 and quercetin treatment was able to mitigate Wnt3a-mediated phenotype of KFs.

Conclusions: Using SFRP1 or inhibitors of Wnt signalling might be one of the therapeutic solutions to treat

keloid scarring.

� 2011 Japanese Society for Investigative Dermatology. Published by Elsevier Ireland Ltd. All rights

reserved.

Contents lists available at SciVerse ScienceDirect

Journal of Dermatological Science

jou r nal h o mep ag e: w ww .e lsev ier . co m / jds

1. Introduction

Keloid scars represent a pathological response to cutaneousinjury and occur only in humans. This condition is foundpredominantly in people of African, Asian and Hispanic descentand is reported to be under-studied relative to other chronic skindisorders [1]. Though not malignant in nature, keloids are fibrotictumors of the dermis that form during a protracted wound-healing

* Corresponding author at: Singapore General Hospital, Skin Bank Unit c/o,

Plastics, Reconstructive and Aesthetic Surgery, Block 6 Level 6, Outram Road,

Singapore 169608, Singapore. Tel.: +65 6321 4974; fax: +65 6220 9340.

E-mail address: alvin.chua.w.c@sgh.com.sg (A.W.C. Chua).

0923-1811/$36.00 � 2011 Japanese Society for Investigative Dermatology. Published b

doi:10.1016/j.jdermsci.2011.09.008

process and to date, there is no satisfactory treatment for thisfibroproliferative disease [2,3].

There is increasing evidence to show that the canonicalWnt/b-catenin signalling is involved in keloid pathogenesis [4].Sato first reported that TGF-b induced the upregulation ofWnt/b-catenin signalling in hypertrophic scar and keloid fibro-blasts (KFs) [5]. Gene profiling of normal scar fibroblasts and KFs[6] as well as immunohistochemical examination of active areas ofkeloid tissues [7] revealed the decreased expression of secretedfrizzled-related protein 1 (SFRP1), a known Wnt inhibitor. Inaddition, initial microarray profiling of normal skin fibroblasts(NFs) and KFs performed in our laboratory also revealed significantdown-regulation of SFRP1 in KFs (unpublished data).

SFRP1 is a secreted glycoprotein containing a cysteine-richdomain which negatively regulates Wnt signalling at the level of

y Elsevier Ireland Ltd. All rights reserved.

Table 1Profiles of normal and keloid fibroblasts.

Fibroblast lines Male/female (M/F) Ethnicity Age of donor (years) Origin

Normal skin

NFM06 M Malay 26 Groin

NFS05 F Malay 25 Breast

NFS06 M Chinese 22 Abdomen

Keloid

KF12 F Chinese 20 Earlobe

KFS01 F Malay 16 Earlobe

KFS04 M Chinese 17 Earlobe

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209200

the plasma membrane where it binds and sequesters Wntmolecules from their respective membrane receptors, the frizzled(Fzd) family [8]. In this study, we verified the down-regulation ofSFRP1 in KFs at both the mRNA and protein levels and hypothesizethat this significant loss of SFRP1 would result in KFs being moresensitive to direct Wnt activation compared to NFs. This sensitivityof KFs to Wnt/b-catenin signalling might account for keloidpathogenesis as it would bring about elevated cellular growth[9,10], migration rate [10,11] and production of extracelluarmatrix (ECM) components [2,12]. The above hypothesis was testedby treating NFs and KFs with Wnt3a conditioned media (CM) – aknown activator of the canonical Wnt/b-catenin signallingpathway [13,14]. Wnt3a was reported to promote cellular growthand motility in NIH3T3 fibroblasts [13,15], human mesenchymalstem cells [16] and lung fibroblasts [17]. In addition, lunghomogenates from patients with idiopathic pulmonary fibrosisexpressed higher mRNA levels of fibronectin [18] which is also anelevated ECM component reported in keloids [19,20]. To furtherprove that SFRP1 or rather, the loss of SFRP1 is one of themechanisms implicated in keloid formation, KFs were stablytransduced with SFRP1 to determine if the fibroproliferativecharacteristics of KFs can be reversed.

Finally, we investigated if quercetin had inhibitory effects onb-catenin expression and cellular characteristics in KFs. Querce-tin, a strong dietary antioxidant, was found to inhibit Wnt/b-catenin signalling in SW480 colon cancer cells [21] and growth inacute lymphoblastic leukaemia cell lines [22]. Our grouppreviously reported the inhibition of KFs’ growth by quercetinthrough the suppression of insulin growth factor (IGF)-1 [23] andTGF-b [24] signalling pathways. It would therefore be interestingto determine if quercetin acts on the Wnt/b-catenin signallingas well.

2. Materials and methods

2.1. Patients and samples

Three respective samples of keloid and normal skin weretaken with consent from surgical procedures. The criteria fordiagnosis of keloid are based on definitions suggested by aninternational clinical recommendations on scar management[25]. A brief detail of the donor information and site of tissueorigin is given in Table 1. Approval was given by the NationalUniversity Hospital and SingHealth Centralised InstitutionalReview Boards for the collection and use of human skin tissuesand cells for this research in adherence to the Declaration ofHelsinki Principles.

2.2. Primary cell culture of fibroblasts

NFs and KFs were derived from their respective excisedspecimens using explant method. Briefly, normal skin or keloidtissue was immersed overnight at 4 8C in 2.5 mg/ml of Dispase II

(Roche) in Dulbecco’s Modified Eagle Medium (DMEM, Gibco,Invitrogen) after removal of subcutaneous tissue using a surgicalscalpel and a pair of forceps. The following day, the entireepidermis was gently peeled off for separate culture ofkeratinocytes. To obtain NFs or KFs, the remaining dermis werecut into small square pieces (3–4 mm) and placed onto 100 mmtissue culture plates with minimal volume (3 ml) of growthmedia consisting of DMEM with 10% (v/v) fetal calf serum (FCS),100 U/ml of penicillin and 100 mg/ml streptomycin (P/S) [allfrom Gibco, Invitrogen]. This was to allow the small pieces of thetissue to adhere to the plate. Fresh growth media wasreplenished 1–2 days after the tissue placement and fibroblastoutgrowth was checked every 2–3 days. All culture conditionswere maintained in 5% CO2 atmosphere at 37 8C and fibroblastswere sub-cultured when 80–90% confluency was reached. Onlycells from the second to fifth passages were used for allexperiments.

2.3. Culture of cell lines and transient transfection of HEK293T cells

HEK 293T and control L-cells were maintained in DMEM/10%FCS while Wnt3a-secreting L-cells were cultured in the samemedia with 350 mg/ml of G418. For over-expression of SFRP1 inHEK293T, pCMV6-Entry control vector and pCMV6-SFRP1 plasmidwith Myc/Flag tag at C-terminal (both from Origene Technologies)were transfected into HEK293T using Lipofectamine/Plus reagent(Invitrogen).

2.4. Collection of conditioned media (CM) from Wnt3a-secreting L-

cells and control L-cells

Wnt3a-secreting L-cells and control L-cells in respective T150flasks (80–90% confluent) were subcultured in the ratio of 1:10 intonew T150 flasks and maintained in DMEM/10% FCS withoutantibiotics at 37 8C for 4 days. This was followed by the same freshmedia change and the media conditioned for another 3 days. Thefirst and second batches of the CM were combined, steri-filteredand stored at either 4 8C or �30 8C, ready to be used for treatmentof fibroblasts. Media conditioned with Wnt3a-secreting L cellswere termed Wnt3a L-cell CM and for L-cells – L-cell CM whichserves as a control to Wnt3a L-cell CM treatment [26] for allexperiments.

2.5. Quantitative Real time PCR (Q-PCR)

Total RNAs of NFs and KFs were extracted with Trizol(Invitrogen) and quantified by a spectrophotometer. cDNAs weregenerated from the total RNAs using Superscript III reversetranscript kit (Invitrogen). Quantitative Real-time PCR wasperformed using the LightCycler 480 (Roche) on a mixture ofcDNAs, required primers and SYBR Green dye according tomanufacturer’s recommendation. A housekeeping gene, GAPDH

was used as an internal control. Measurements were repeated

Table 2Oligonucleotides sequences of primer pairs for SFRP1, CCN2, COL1A1, FN1 and GAPDH.

Gene (human) Forward primer Reverse primer

SFRP1 50-CAA GAA GAA GAA GCC CCT GA-30 50-AAG TGG TGG CTG AGG TTG TC-30

CCN2 50-GGC CCA GAC CCA ACT ATG ATT AG-30 50-CTG CAG GAG GCG TTG TCA TT-30

COL1A1 50-CAA AGG TCC CCG TGG TGA GA-30 50-CAG CAA TAC CTT GAG GCC CG-30

FN1 50-GCG CCG GCT GTG CTG CAC AGG-30 50-GCC TGG GGA CAG CGG TGC CC-30

GAPDH 50-GCC AAG GTC ATC CAT GAC AAC-30 50-GTC CAC CAC CCT GTT GCT GTA-30

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209 201

thrice and the fold change values between keloid and normalsamples were calculated using DCt method [27]. Primer sequencesof GAPDH and genes of interest are listed in Table 2.

2.6. Luciferase reporter assay of fibroblasts

NFs were seeded at a density of 2 � 105 cells per well in 6-wellplates in DMEM/10% FCS and transiently transfected usingLipofectamine/Plus reagent (Invitrogen) with 2 mg each of TOP-flash or FOPflash reporter (Upstate, Millipore) and 0.5 mg b-galactosidase-expressing vector (gift from Iqbal Dulloo NationalCancer Centre, Singapore) as an internal control. After treatmentwith required culture media, transfected NFs were lysed formeasurements of luciferase and b-galactosidase activities by aluminometer (EG&G Berthold Lumat LB 9507). The luciferaseactivity was normalised to b-galactosidase activity.

2.7. Cellular growth/viability assays

All NFs and KFs were seeded at a density of 104 cells per well in24-well plates or 2 � 103 cells per well in 96-well plates usingDMEM/10% FCS with P/S. These cells were subsequently exposed toserum free DMEM for 24 h to allow them to achieve basal statebefore being treated with test culture media. Cell growth wasassessed by MTS assay using CellTiter-96 Aqueous-One prolifera-tion kit (Promega) on a photometric plate reader (BIO-RAD,Benchmark PlusTM) at 492 nm.

Separately, transduced KFs with control vector (EMW) orvector with SFRP1 (EMW-SFRP1) were tested for cell viabilityusing propidium iodide (PI) staining. In total, 104 cells wereacquired per sample based on GFP-positive population with dataacquisition and analyses performed on cell cytometer (BeckmanCyan ADP).

Finally, to verify the above MTS and flow cytometry results, thecells were probed in western blots using proliferating cell nuclearantigen (PCNA) antibody which is a marker for cell growth andproliferation.

2.8. Migration assay and live cells time-lapse imaging

Migration assay was performed using the scratch method andlive cell time-lapse imaging. NFs and KFs were grown to confluencyin 12-well plates before a scratch wound was manually introducedinto each well using a yellow pipette tip. The cultures were washedtwice with phosphate buffered saline (PBS), replaced with freshcontrol L-cell CM and Wnt3a L-cell CM. The plate was subsequentlytransferred to an incubating chamber (5% CO2, 37 8C) within amicroscope using phase contrast mode (Nikon Eclipse C1 PlusConfocal Microscope) for time-lapse imaging. Images of cellmigration were taken at 30 min intervals for duration of 24 h,collected and analysed with NIS-Elements software (Nikon). Cellmigration was measured by percent closure of the scratch zoneselected, as follows:

Percent closure ð%Þ ¼migrated cell surface area

total surface area� 100

Three zones were randomly selected from each well to calculatethe average percent closure for each cell type and treatment.

2.9. Lentiviral transduction of SFRP1 into KFs

SFRP1 gene was cloned into lentiviral vector encoding IRES-linked GFP (EMW). Approximately 5 � 106 cells were seeded ineach 10-cm tissue culture plates 24 h before transfection. Calciumphosphate precipitation method was used for packaging virusencoding either GFP (control vector) or SFRP1 and GFP. 10 mg oflentiviral vector, 7.5 mg helper plasmid pPax2, and 2.5 mg ofMD2G envelope plasmids (Addgene) were used for packaging. Thetransfection media were replaced with complete medium after 12–14 h incubation. The viral supernatant was filtered through a 0.45-mm filter, and the titer of supernatant on 293T cells wasdetermined using flow cytometry. KF cells were infected at anMOI of 1–5. The transduction efficiency was visualized andassessed by the GFP positivity.

2.10. Immunoblotting

For extraction of total cell lysates (CLs), cultured cells were lysedin RIPA buffer (Santa Cruz Biotechnology) containing proteaseinhibitors. Conditioned media (CM) of respective cell cultures wereconcentrated using Amicon Ultra-4 PL 10 or Amicon Ultra-15 PL 10centrifugal filters (Millipore). Proteins from CL and CM werequantified using Pierce BCA protein assay kit (Thermo Scientific)according to manufacturer’s instruction. Equal amounts of protein(between 20 and 60 mg) in 5� loading dye were heated at 95 8C andseparated on 8% or 12% SDS–polyacrylamide gels and electroblottedonto nitrocellulose membrane. After blocking with 5% non-fat milkat room temperature for 1 h, blots were incubated overnight withthe specific antibodies against SFRP1 – 1:500 dilution (R&DSystems); b-catenin – 1:1000 (Cell Signalling); fibronectin –1:5000 (Epitomics); collagen I, II, III – 1:500 (Monosan) and PCNA(C-20) – 1:1000 (Santa Cruz Technology). The blots were subse-quently incubated with appropriate horseradish-peroxidase conju-gated secondary antibodies (Thermo Scientific) and visualized witha chemiluminescence-based photoblot system (Amersham Bios-ciences). GAPDH (FL-335) – 1:1000 (Santa Cruz Biotechnology) wasused as a primary antibody to probe for all CLs and ponseau stainingwas used for all CM blots to check for equal protein loading (byweight). These served as controls for normalisation in thesubsequent relative integrated densities measurements of theprotein expressions (Section 2.11).

2.11. Measurement of relative integrated densities and statistical

analysis

Relative integrated densities of protein expressions inwestern blots were quantified using ImageJ 1.43 m (US NationalInstitutes of Health) for three independent experiments. Allstatistical analyses in the entire study were performed usingGraphPad InStat version 3.06 using one-way Anova–Student–Newman–Keuls multiple comparisons test or Student’s t-test(two-tailed).

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209202

3. Results

3.1. Keloid fibroblasts expressed lower levels of SFRP1 compared to

normal fibroblasts

mRNA expression of SFRP1 was found to be down-regulated onthe average by 85 times in the KF group compared to the NF group(p < 0.05 for n = 3 in each group) when analysed by Q-PCR method(Fig. 1A). Concurrently, mRNA expression of collagen, type 1, alpha1 (COL1A1), fibronectin 1 (FN1) and connective tissue growth factor(CCN2) were also analysed for both types of fibroblast which re-affirmed the phenotype of KFs in this study as the tested KFsexpressed significantly higher levels of these pro-fibrotic factors.KFs were previously reported to be up-regulated for COL1A1 [28–30], FN1 [19,20] and CCN2 [31,32].

In western blot analyses of NFs and KFs (3 lines in each group),higher expression of SFRP1 was observed in conditioned media(CM) of NFs compared to KFs (Fig. 1B). The positive control (PC)used was from CM of HEK293 transiently transfected with pCMV6-SFRP1 plasmid with Myc/Flag tag which were probed and verifiedwith both c-Myc and SFRP1 antibodies (Supplementary Fig. S1).

3.2. Activation of Wnt/b-catenin signalling was enhanced in KFs

compared to NFs after Wnt3a treatment

Fig. 2A demonstrated that b-catenin/T-cell factor (TCF)-induced transcriptional activities were higher in KFs comparedto NFs after Wnt3a CM treatment. The above results wereconfirmed by western blot analyses of their b-catenin expressions.It was observed that Wnt3a L-cell CM increased b-cateninexpression in all our lines of NFs and KFs relative to its controlmedia (L-cell CM) as represented by 2 lines each of NFs and KFs inFig. 2B. KFs generally expressed higher levels of b-catenin

Fig. 1. Repression of SFRP1 expression in keloid fibroblasts. (A) mRNA fold change of

normal fibroblasts (NFs) and keloid fibroblasts (KFs) [3 different lines in each group]

for expression levels of SFRP1, COL1A1, FN1 and CCN2 analysed by Quantitative Real

time PCR. All values represent mean � standard deviation (SD) and p values obtained

were based on unpaired Student’s t-test (2-tailed) analysis (n = 3 each) with *p < 0.05

and ***p < 0.005. (B) Western blot expression of SFRP1 in conditioned media (CM) of

NFs and KFs with CM of HEK293T over-expressed with 1 mg of pCMV6-SFRP1 plasmid

used as positive control (PC).

compared to NFs for both types of treatment. Taken together,the results suggest that there is higher amplification of Wnt/b-catenin signalling in KFs compared to NFs.

3.3. Increase in cellular growth and fibronectin expression were more

pronounced in KFs compared to NFs after Wnt3a treatment

As Wnt/b-catenin signalling was reported to promote cellulargrowth and proliferation in fibroblasts [13,33,34], we thereforetested if Wnt3a would similarly bring about enhanced growth inKFs compared to NFs.

First, it was demonstrated using MTS assay (Fig. 3A) thatthere was more pronounced increase (dotted line for NFS05versus solid line for KFS04) in Day-4 cellular growth in KFscompared to NFs after treatment with Wnt3a (relative to controlL-cell CM). Statistical comparison of percentage increase incellular growth in both types of fibroblast (3 lines in each group)

Fig. 2. Wnt/b-catenin signalling is up-regulated in keloid fibroblasts. (A) Luciferase

activity of NFs and KFs co-transfected with reporter gene constructs with

SUPERTOPflash (2 mg) or its mutant SUPERFOPflash (2 mg) and b-galactosidase

plasmid (0.5 mg) as an internal control. Ratios of luciferase/b-galactosidase

activities after treatment with L-cell CM and Wnt3a L-cell CM represent b-

catenin/T-cell factor (TCF)-induced transcriptional activities, expressed by

mean � SD of duplicates. (B) Western blot expression of b-catenin in NFs and KFs

(2 representative lines in each group) treated with L-cell CM and Wnt3a L-cell CM;

relative integrated densities of the western blot bands were quantified using ImageJ

software.

Fig. 3. Higher proliferation rate in keloid fibroblasts after Wnt3a treatment. (A) Day-4 MTS assays of representative lines of NF (NFS05) and KF (KFS04) treated with control

L-cell CM and Wnt3a L-cell CM; absorbance values represent mean � SD of quadruplicates. (B) Percentage increase of Day 4 MTS absorbance values in NFs and KFs (3 different

lines in each group) due to Wnt3a L-cell CM treatment relative to control L-cell CM treatment; values represent mean � standard error of means (SEM) of percentage change (n = 3

each). (C) Western blot expression of PCNA, fibronectin, collagen I and GAPDH in representative lines of NF (NFS05) and KF (KFS04) after 2-day treatment with serum free DMEM, L-

cell CM and Wnt3a L-cell CM. (D) Average relative integrated densities of PCNA/GAPDH, fibronectin/GAPDH and collagen I/GAPDH expressions in western blots of NFs and KFs

(3 different lines in each group) treated with same sequence of culture media shown in (C); values represent mean � SD. Statistical analyses were performed using one-way

Anova–Student–Newman–Keuls multiple comparisons test or Student’s t-test (2-tailed) with *p < 0.05, **p < 0.01 and ****p < 0.0001.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209 203

after treatment with Wnt3a at Day 4 (Fig. 3B) revealed that theincrease in cellular growth of KFs was significantly higher(p < 0.01).

Next in western blot analyses, while higher expression of PCNAwas observed in both NFs and KFs after Wnt3a treatment (Fig. 3Cand D), the quantified relative integrated densities showed that thelevel of PCNA increase was higher and more significant in the KF

group (p < 0.01 for KFs versus p < 0.05 for NFs, n = 3). Similarly, itwas observed that the increase in protein expression of fibronectinwas significant only in KFs but not in NFs. However, no significantchange was observed in protein expression of collagen I for bothNFs and KFs after Wnt3a treatment. Treatment of samples withDMEM without serum served as negative controls in the westernblot analyses and the increase of PCNA, fibronectin and collagen I

Fig. 4. No significant change in migration rates of normal and keloid fibroblasts after Wnt3a treatment. (A) Phase contrast images of representative lines of NF (NFS06) and KF

(KFS01) in scratch test assay at T = 0 h and T = 24 h after treatment with L-cell CM and Wnt3a L-cell CM, respectively; scale bar = 200 mm. (B) Average percentage of scratched

area closed after treatment with control L-cell CM and Wnt3a L-cell CM on NFs and KFs (3 different lines in each group) from T = 0 h to T = 24 h measured at 6-h intervals;

values represent mean � SD.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209204

expressions with L-cell CM was possibly due to the effects of serumwhich was used in the preparation of the CM (Section 2.4).

3.4. Marginal decrease in cell migration rate was observed in NFs and

KFs after Wnt3a treatment

Cell migration and closure in scratch test assays were shownin representative lines of NF and KF at T = 0 h and T = 24 h inFig. 4A. Analyses of the area of cell closure after migration for bothNFs and KFs revealed that in general, KFs had a higher migrationrate compared to NFs, concurring with previous reports [10,11].As shown in Fig. 4B, Wnt3a brought about a marginal decrease incell migration for both NFs and KFs, starting at T = 6 h. However,this difference brought about by Wnt3a treatment was notsignificant at all time points based on Student’s t-test (two tailed)analysis.

3.5. Keloid fibroblasts stably transduced with SFRP1 displayed lower

cellular growth and viability

Fig. 5A confirmed the stable transduction of the 3 lines of KFwith control lentivirus encoding GFP (EMW) or lentivirus

encoding GFP and SFRP1 (EMW-SFRP1). SFRP1 protein washighly secreted in serum free CM of all SFRP1-tranduced KFs.However, only KFS01 and KF12 expressed lower levels ofb-catenin in western blots after the SFRP1 transduction(Fig. 5B).

MTS assays performed on these stably transduced fibroblasts atDay 4 also showed that 2 lines of KFs – KFS01 and KF12 withectopic expression of SFRP1, displayed significantly lower cellulargrowth (Fig. 6A) compared to their controls when cultured ingrowth media – DMEM/10% FCS. However, PI staining assays(Fig. 6B) performed on all 3 lines of SFRP1-transduced KFs based onGFP-positive population, showed that the overall average cellviability of these cells was significantly lower than their controlcounterparts (p < 0.05). Western blot analyses also showed that allKFs transduced with SFRP1 expressed lower levels of PCNAcompared to their controls (Fig. 6C) with KFS04 showing the leastdecrease.

In addition, when treated with Wnt3a L-cell CM, the lowercellular growth observed after ectopic expression of SFRP1 becamesignificant in KFS04 while in KF12, the effect was more significantat p < 0.01 (Fig. 6D) when compared to treatment with controlL-cell CM.

Fig. 5. Over-expression of SFRP1 and its control in keloid fibroblasts. (A)

Fluorescence microscopy of a representative line, KF12 stably expressed with

EMW-SFRP1 and EMW constructs respectively. Both EMW constructs contain GFP

tags which indicate that the cells were more than 90% transfected; scale

bar = 100 mm. Next, western blot expressions of SFRP1 in CM of KFS01, KFS04

and KF12 over-expressed with EMW and EMW-SFRP1 constructs, respectively. (B)

Western blot expression of b-catenin in cell lysates of KFS01, KFS04 and KF12

overexpressed with EMW and EMW-SFRP1 constructs; relative integrated densities

of the b-catenin/GAPDH expression were quantified using ImageJ software.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209 205

3.6. SFRP1-transduced keloid fibroblasts expressed lower protein

levels of fibronectin but not collagen I

Western blot analyses in Fig. 7A showed that there was down-regulation of fibronectin in the SFRP1-transduced KFs with KFS01exhibiting the maximum decrease while there was only marginaldecrease in KFS04 when cultured in growth media – DMEM/10%FCS. However, there was no change in collagen I expression in allthe SFRP1-transduced KFs compared to their controls. In addition,it was observed that there was wider reduction of fibronectin

expression in SFRP1-transduced KFS04 as well as in SFRP1-transduced KF12 when they were treated with Wnt3a L-cell CM(Fig. 7B).

3.7. Quercetin inhibited cellular growth as well as expressions of b-

catenin and fibronectin in keloid fibroblasts

MTS assay in Fig. 8A showed that quercetin brought aboutsignificant reduction in cellular growth of KFS01 – a line withhighest activation of Wnt/b-catenin signalling, in a dose-depen-dent manner by Day 4 after Wnt3a treatment. Similarly, westernblot analyses also revealed that quercetin was able to mitigateWnt3a treatment with dose-dependent reduction of fibronectin,b-catenin and PCNA expressions but this was not observed forSFRP1 expression (Fig. 8B and C).

4. Discussion

There are a growing number of studies which associate thecanonical Wnt/b-catenin signalling to several fibro-proliferativediseases [35]. For example, Cheon et al. [36] reported that therewas sustained elevated b-catenin levels in patients’ hyperplasticscar tissue after more than 2 years of the initial injury. Morerecently, immunohistochemical examination of lung biopsysamples from patients with systemic sclerosis-associated ad-vanced pulmonary fibrosis and in vitro investigation of normalhuman lung fibroblasts implicated Wnt/b-catenin signalling tolung fibrosis [17]. Interestingly, down-regulation of SFRPs whichare known extracellular antagonists of the Wnt/b-catenin signal-ling [37], was also implicated with fibrosis [38,39]. The probe forSFRP1 in our Q-PCR and western blot studies confirmed its loss,both at the mRNA and protein levels in KFs when compared withnormal skin fibroblasts. Our mRNA data (Fig. 1A) corroborated withthe work by Smith et al. [6] which compared KFs with that ofnormal scar fibroblasts, suggesting that silencing of SFRP1 may beimportant in the fibrosis signature shown by KFs [7]. This silencingof SFRP1 was further demonstrated in our study where negligibleamount of SFRP1 was detected in CM of KFs compared to NFs(Fig. 1B).

The subdued amount of secreted SFRP1 by KFs to sequester Wntligands [40] might be one of the explanations for KFs being moresensitive to Wnt3a treatment in terms of higher activation levels ofTCF/b-catenin activity (Fig. 2A), increased expression of b-catenin(Fig. 2B) and increased cellular growth (Fig. 3A–D). While theabove data for NFs and KFs corroborated with several studieswhich demonstrated the ability of Wnt3a to promote cellulargrowth and proliferation in murine and human fibroblasts/mesenchymal cells through the Wnt/b-catenin signalling pathway[13,15,16,33,34,41], we have further demonstrated here that KFswere more susceptible to such exogenous treatment. Similarly, itwas observed that there was a higher degree of fibronectin up-regulation in KFs compared to NFs (Fig. 3C and D) with Wnt3atreatment. This can be explained by the fact that fibronectin is aknown Wnt/b-catenin signalling target gene in mesenchymal cells[36,42,43] and it is also known to be up-regulated in hyperplasticwounds [36], hypertrophic scars and keloids [19,20]. In this studyhowever, there was little or no increase of collagen I in both NFsand KFs after Wnt3a treatment (Fig. 3C and D). This data suggeststhat collagen I is not a robust target of Wnt/b-catenin signalling inhuman skin fibroblasts, similar to what was reported in humanlung fibroblasts [17].

Next, it was found that both NFs and KFs did not respond toexogenous Wnt3a in terms of migration rate and in fact a slightdecrease in their motility was observed, albeit an insignificant one.This result was initially unexpected as it was reported thatexogenous Wnt3a stimulated motility in NIH3T3 [15], rat bone

Fig. 6. SFRP1 transduction of keloid fibroblasts reduced their cellular growth and viability. (A) Day-4 MTS assays of KFS01, KFS04 and KF12 after culture in DMEM/10% FCS

growth media; absorbance values represent mean � SD (n = 6 each group). (B) Average percentage of viable cells from 3 lines of KFs transduced with EMW and EMW-SFRP1 based

on propidium iodide (PI) staining in flow cytometry performed in triplicates, expressed by mean percentage � SEM (n = 9). (C) Western blot expression of PCNA in cell lysates of

KFS01, KFS04 and KF12 overexpressed with EMW and EMW-SFRP1 constructs after 3-day culture in DMEM/10% FCS growth media; relative integrated densities of PCNA/GAPDH

were quantified using ImageJ software. (D) Day-4 MTS assays of KFS04 and KF12 after culture in control L-cell CM and Wnt3a L-cell CM, respectively; absorbance values represent

mean � SD (n = 4 each group). Statistical analyses were performed using Student’s t-test (2-tailed) with *p < 0.05 and **p < 0.01.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209206

marrow mesenchymal stem cells [41] and invasion capacity ofhuman mesenchymal stem cells [16]. Interestingly, it was alsoreported that while mouse mesenchymal stem cells (mMSCs)responded to Wnt3a treatment in terms of increase in prolifera-tion, there was however a corresponding decrease in invasioncapacity [44]. This impaired invasion phenomenon in mMSCs wasattributed in part to the down-regulation of membrane type 1-matrix metalloproteinase (MT1-MMP), MMP-2 and MMP-9 whichparadoxically are known Wnt targets in cancer and immune cells[45–47]. In human lung fibroblasts, while increased b-cateninsignalling enhanced migration, there was no correspondingincrease in levels of Wnt targets, MMP-1, 2 and 9. These factorsare known to be responsible for lung fibroblasts enhanced motilityand therefore the specific targets of exogenous Wnt3a in thisaspect remain to be elucidated [17]. Taken together, the abovestudies suggest that the targets of Wnt/b-catenin signalling arecell-type, tissue-type as well as context dependent [48]. In ourcase, we speculate that Wnt3a might have brought about marginaldecrease in MT1-MMP and MMP-2 expressions as these factors areknown to influence migration rate in keloids [11,49]. Furtherinvestigation on the above hypothesis will be performed in future

studies. We believe that the involvement of Wnt/b-cateninsignalling on the motility of KFs is a complex mechanism toelucidate as it is through the interplay of MMPs and inhibitors ofmetalloproteinases (TIMPs).

In this study, the stable transduction of SFRP1 in KFs had alsoindirectly helped verify the effects of Wnt3a treatment on them.After it was confirmed that over-expression of SFRP1 did bringabout reduction in b-catenin expression (Fig. 5B) in 2 KF lines(KFS01 and KF12), these 2 lines correspondingly demonstratedlower cellular growth in MTS assays (Fig. 6A) as well as lowerprotein expressions of PCNA and fibronectin (Figs. 6C and 7A).SFRP1 did not seem to have an effect on the expression of collagen Iin KFs after transduction.

While we could not fully explain the reason for KFS04’sinsensitivity to the forced expression of SFRP1, what we couldspeculate was the inherent SFRP1 expression in this line (Figs. 1Band 5A). However, PI staining based on GFP-positive populationshowed that the overall viability of the 3 lines transduced withSFRP1 was significantly lower than their controls (Fig. 6B). Inaddition, KFS04 also showed reduction in PCNA expression afterSFRP1 transduction (Fig. 6C). While we cannot discount off-target

Fig. 7. SFRP1 transduction of keloid fibroblasts reduced expression of fibronectin but not collagen I. (A) Western blot expression of fibronectin and collagen I in cell lysates of

KFS01, KFS04 and KF12 over-expressed with EMW and EMW-SFRP1 constructs after 3-day culture in DMEM/10% FCS growth media. (B) Western blot expression of

fibronectin in cell lysates of KFS04 and KF12 over-expressed with EMW and EMW-SFRP1 constructs after 3-day culture in control L-cell CM and Wnt3a L-cell CM, respectively.

Relative integrated densities of fibronectin/GAPDH were quantified using ImageJ software.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209 207

effects of SFRP1 especially in KFS04, these results suggest thatover-expression of SFRP1 worked in part through the canonicalWnt/b-catenin pathway by sequestering endogeneous Wntligands to bring about down-regulation of b-catenin. This de-stabilization of b-catenin in turn brought about an decrease incellular growth and fibronectin expression (Figs. 6 and 7).

Finally, on top of confirming quercetin’s inhibitory effect onKFs’ cellular growth in a dose-dependent manner (based on MTSassay at Day 4) and decrease in PCNA expression (Fig. 8A–C)after Wnt3a treatment, we also demonstrated for the first timethat this potent antioxidant did reduce KFs b-catenin andfibronectin protein expressions in a dose-dependent manner(Fig. 8B and C). Quercetin did not seem to reduce SFRP1 proteinexpression in KFs (Fig. 8B), suggesting that this anti-oxidantprobably acts downstream of the ligand–receptor interactions ofthe Wnt/b-catenin pathway where SFRP1 is known to target as

an extracellular antagonist. It was previously reported thatquercetin caused decreased levels of b-catenin and TCF productin nucleus of SW480 colon cancer cells [21] which might also bethe case in KF cells. As quercetin is known to be a multi-target orbroad spectrum kinase inhibitor [50], our reported data hereusing quercetin might be a result of direct or indirect targetingof the Wnt/b-catenin pathway in KFs.

In summary, KFs have lower levels of SFRP1 compared to NFswhich might contribute to KFs sensitivity to Wnt3a treatment.Use of exogenous Wnt3a in skin fibroblasts which resulted indirect activation of canonical Wnt signalling pathway, inducedhigher cellular growth rate and fibronectin expression but not cellmigration and collagen I expression. Targeting the Wnt/b-cateninpathway with the use of SFRP1 or Wnt inhibiting agents mightbe one of the therapeutic solutions to prevent or treat keloidscarring.

Fig. 8. Inhibitory effects of quercetin on cellular growth and protein expressions of keloid fibroblasts. (A) Day-4 MTS assays of a representative line, KFS01 treated with control

L-cell CM, Wnt3a CM L-cell, Wnt3a L-cell CM + 5 mg/ml quercetin and Wnt3a L-cell CM + 20 mg/ml quercetin; absorbance values represent mean � SD of quadruplicates. (B)

Representative western blot expressions of fibronectin, b-catenin, PCNA, SFRP1 and GAPDH in cell lysates of KFS01 treated in the same sequence of culture media and reagents

shown in (A). (C) Average relative integrated densities of fibronectin/GAPDH, b-catenin/GAPDH and PCNA/GAPDH expressions in western blots of NFs and KFs (3 different lines in

each group) treated with same sequence of culture media shown in Fig. 8A; values represent mean � SD. Statistical analyses were performed using one-way Anova–Student–

Newman–Keuls multiple comparisons test with *p < 0.05, **p < 0.01 and ***p < 0.001.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209208

Acknowledgements

This study was supported by research grants from theSingapore General Hospital, Department of Clinical Research(DCR/P30/2009 and DCR/P33/2009). We are also grateful to Ms.Teo Pei Yun and Mr. Ridzwan Bin Rahmad for their technicalassistance in this project.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in

the online version, at doi:10.1016/j.jdermsci.2011.09.008.

References

[1] Bock O, Schmid-Ott G, Malewski P, Mrowietz U. Quality of life of patients withkeloid and hypertrophic scarring. Arch Dermatol Res 2006;297:433–8.

[2] Niessen FB, Spauwen PH, Schalkwijk J, Kon M. On the nature of hypertrophicscars and keloids: a review. Plast Reconstr Surg 1999;104:1435–58.

[3] Butler PD, Longaker MT, Yang GP. Current progress in keloid research andtreatment. J Am Coll Surg 2008;206:731–41.

[4] Chua AW, Ma D, Gan SU, Fu Z, Han HC, Song C, et al. The role of R-Spondin2 inkeratinocyte proliferation and epidermal thickening in keloid scarring. J InvestDermatol 2010.

[5] Sato M. Upregulation of the Wnt/beta-catenin pathway induced by transform-ing growth factor-beta in hypertrophic scars and keloids. Acta Derm Venereol2006;86:300–7.

[6] Smith JC, Boone BE, Opalenik SR, Williams SM, Russell SB. Gene profiling ofkeloid fibroblasts shows altered expression in multiple fibrosis-associatedpathways. J Invest Dermatol 2008;128:1298–310.

[7] Russell SB, Russell JD, Trupin KM, Gayden AE, Opalenik SR, Nanney LB, et al.Epigenetically altered wound healing in keloid fibroblasts. J Invest Dermatol2010;130:2489–96.

[8] Gumz ML, Zou H, Kreinest PA, Childs AC, Belmonte LS, LeGrand SN, et al.Secreted frizzled-related protein 1 loss contributes to tumor phenotype ofclear cell renal cell carcinoma. Clin Cancer Res 2007;13:4740–9.

[9] Calderon M, Lawrence WT, Banes AJ. Increased proliferation in keloid fibro-blasts wounded in vitro. J Surg Res 1996;61:343–7.

[10] Lim CP, Phan TT, Lim IJ, Cao X. Stat3 contributes to keloid pathogenesis viapromoting collagen production, cell proliferation and migration. Oncogene2006;25:5416–25.

[11] Fujiwara M, Muragaki Y, Ooshima A. Keloid-derived fibroblasts showincreased secretion of factors involved in collagen turnover and dependon matrix metalloproteinase for migration. Br J Dermatol 2005;153:295–300.

[12] Tuan TL, Nichter LS. The molecular basis of keloid and hypertrophic scarformation. Mol Med Today 1998;4:19–24.

[13] Yun MS, Kim SE, Jeon SH, Lee JS, Choi KY. Both ERK and Wnt/beta-cateninpathways are involved in Wnt3a-induced proliferation. J Cell Sci 2005;118:313–22.

[14] Chen S, McLean S, Carter DE, Leask A. The gene expression profile induced byWnt 3a in NIH 3T3 fibroblasts. J Cell Commun Signal 2007;1:175–83.

[15] Kim SE, Choi KY. EGF receptor is involved in WNT3a-mediated proliferationand motility of NIH3T3 cells via ERK pathway activation. Cell Signal 2007;19:1554–64.

[16] Neth P, Ciccarella M, Egea V, Hoelters J, Jochum M, Ries C. Wnt signalingregulates the invasion capacity of human mesenchymal stem cells. Stem Cells2006;24:1892–903.

[17] Lam AP, Flozak AS, Russell S, Wei J, Jain M, Mutlu GM, et al. Nuclear {beta}-catenin is increased in SSc pulmonary fibrosis and promotes lung fibroblastmigration and proliferation. Am J Respir Cell Mol Biol 2011.

[18] Konigshoff M, Balsara N, Pfaff EM, Kramer M, Chrobak I, Seeger W, et al.Functional Wnt signaling is increased in idiopathic pulmonary fibrosis. PLoSOne 2008;3:e2142.

[19] Babu M, Diegelmann R, Oliver N. Fibronectin is overproduced by keloidfibroblasts during abnormal wound healing. Mol Cell Biol 1989;9:1642–50.

A.W.C. Chua et al. / Journal of Dermatological Science 64 (2011) 199–209 209

[20] Kischer CW, Hendrix MJ. Fibronectin (FN) in hypertrophic scars and keloids.Cell Tissue Res 1983;231:29–37.

[21] Park CH, Chang JY, Hahm ER, Park S, Kim HK, Yang CH. Quercetin, a potentinhibitor against beta-catenin/Tcf signaling in SW480 colon cancer cells.Biochem Biophys Res Commun 2005;328:227–34.

[22] Kawahara T, Kawaguchi-Ihara N, Okuhashi Y, Itoh M, Nara N, Tohda S.Cyclopamine and quercetin suppress the growth of leukemia and lymphomacells. Anticancer Res 2009;2(9):4629–32.

[23] Phan TT, See P, Tran E, Nguyen TT, Chan SY, Lee ST, et al. Suppressionof insulin-like growth factor signalling pathway and collagen expressionin keloid-derived fibroblasts by quercetin: its therapeutic potential usein the treatment and/or prevention of keloids. Br J Dermatol 2003;148:544–52.

[24] Phan TT, Lim IJ, Chan SY, Tan EK, Lee ST, Longaker MT. Suppression oftransforming growth factor beta/smad signaling in keloid-derived fibroblastsby quercetin: implications for the treatment of excessive scars. J Trauma2004;57:1032–7.

[25] Mustoe TA, Cooter RD, Gold MH, Hobbs FD, Ramelet AA, Shakespeare PG, et al.International clinical recommendations on scar management. Plast ReconstrSurg 2002;110:560–71.

[26] Ueda Y, Hijikata M, Takagi S, Takada R, Takada S, Chiba T, et al. Wnt/beta-catenin signaling suppresses apoptosis in low serum medium and inducesmorphologic change in rodent fibroblasts. Int J Cancer 2002;99:681–8.

[27] Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 2001;25:402–8.

[28] Uzawa K, Marshall MK, Katz EP, Tanzawa H, Yeowell HN, Yamauchi M. Alteredposttranslational modifications of collagen in keloid. Biochem Biophys ResCommun 1998;249:652–5.

[29] Uitto J, Perejda AJ, Abergel RP, Chu ML, Ramirez F. Altered steady-state ratio oftype I/III procollagen mRNAs correlates with selectively increased type Iprocollagen biosynthesis in cultured keloid fibroblasts. Proc Natl Acad Sci US A 1985;82:5935–9.

[30] Kauh YC, Rouda S, Mondragon G, Tokarek R, diLeonardo M, Tuan RS, et al.Major suppression of pro-alpha1(I) type I collagen gene expression in thedermis after keloid excision and immediate intrawound injection of triamcin-olone acetonide. J Am Acad Dermatol 1997;37:586–9.

[31] Xia W, Kong W, Wang Z, Phan TT, Lim IJ, Longaker MT, et al. Increased CCN2transcription in keloid fibroblasts requires cooperativity between AP-1 andSMAD binding sites. Ann Surg 2007;246:886–95.

[32] Igarashi A, Nashiro K, Kikuchi K, Sato S, Ihn H, Fujimoto M, et al. Connectivetissue growth factor gene expression in tissue sections from localized sclero-derma, keloid, and other fibrotic skin disorders. J Invest Dermatol 1996;106:729–33.

[33] Torii K, Nishizawa K, Kawasaki A, Yamashita Y, Katada M, Ito M, et al. Anti-apoptotic action of Wnt5a in dermal fibroblasts is mediated by the PKAsignaling pathways. Cell Signal 2008;20:1256–66.

[34] Baksh D, Tuan RS. Canonical and non-canonical Wnts differentially affect thedevelopment potential of primary isolate of human bone marrow mesenchy-mal stem cells. J Cell Physiol 2007;212:817–26.

[35] Bowley E, O’Gorman DB, Gan BS. Beta-catenin signaling in fibroproliferativedisease. J Surg Res 2007;138:141–50.

[36] Cheon S, Poon R, Yu C, Khoury M, Shenker R, Fish J, et al. Prolonged beta-catenin stabilization and tcf-dependent transcriptional activation in hyper-plastic cutaneous wounds. Lab Invest 2005;85:416–25.

[37] Kawano Y, Kypta R. Secreted antagonists of the Wnt signalling pathway. J CellSci 2003;116:2627–34.

[38] He W, Zhang L, Ni A, Zhang Z, Mirotsou M, Mao L, et al. Exogenouslyadministered secreted frizzled related protein 2 (Sfrp2) reduces fibrosis andimproves cardiac function in a rat model of myocardial infarction. Proc NatlAcad Sci U S A 2010;107:21110–5.

[39] Kaplan FS, Pignolo RJ, Shore EM. The FOP metamorphogene encodes a noveltype I receptor that dysregulates BMP signaling. Cytokine Growth Factor Rev2009;20:399–407.

[40] Esteve P, Bovolenta P. The advantages and disadvantages of sfrp1 and sfrp2expression in pathological events. Tohoku J Exp Med 2010;221:11–7.

[41] Shang YC, Wang SH, Xiong F, Zhao CP, Peng FN, Feng SW, et al. Wnt3a signalingpromotes proliferation, myogenic differentiation, and migration of rat bonemarrow mesenchymal stem cells. Acta Pharmacol Sin 2007;28:1761–74.

[42] Gradl D, Kuhl M, Wedlich D. The Wnt/Wg signal transducer beta-catenincontrols fibronectin expression. Mol Cell Biol 1999;19:5576–87.

[43] Howard JC, Varallo VM, Ross DC, Roth JH, Faber KJ, Alman B, et al. Elevatedlevels of beta-catenin and fibronectin in three-dimensional collagen culturesof Dupuytren’s disease cells are regulated by tension in vitro. BMC Muscu-loskelet Disord 2003;4:16.

[44] Karow M, Popp T, Egea V, Ries C, Jochum M, Neth P. Wnt signalling in mousemesenchymal stem cells: impact on proliferation, invasion and MMP expres-sion. J Cell Mol Med 2009;13:2506–20.

[45] Takahashi M, Tsunoda T, Seiki M, Nakamura Y, Furukawa Y. Identification ofmembrane-type matrix metalloproteinase-1 as a target of the beta-catenin/Tcf4 complex in human colorectal cancers. Oncogene 2002;21:5861–7.

[46] Hlubek F, Spaderna S, Jung A, Kirchner T, Brabletz T. Beta-catenin activates acoordinated expression of the proinvasive factors laminin-5 gamma2 chainand MT1-MMP in colorectal carcinomas. Int J Cancer 2004;108:321–6.

[47] Wu B, Crampton SP, Hughes CC. Wnt signaling induces matrix metalloproteinaseexpression and regulates T cell transmigration. Immunity 2007;26:227–39.

[48] Arce L, Yokoyama NN, Waterman ML. Diversity of LEF/TCF action in develop-ment and disease. Oncogene 2006;25:7492–504.

[49] Imaizumi R, Akasaka Y, Inomata N, Okada E, Ito K, Ishikawa Y, et al. Promotedactivation of matrix metalloproteinase (MMP)-2 in keloid fibroblasts andincreased expression of MMP-2 in collagen bundle regions: implications formechanisms of keloid progression. Histopathology 2009;54:722–30.

[50] Kerr DJ, La Thangue NB. Signal transduction blockade and cancer: combinationtherapy or multi-targeted inhibitors? Ann Oncol 2004;15:1727–9.