ORIGINAL ARTICLE

Syndecan-2 and Decorin: Proteoglycans With aDifference—Implications in Keloid Pathogenesis

Anandaroop Mukhopadhyay, PhD, Man Yi Wong, BSc, Sui Y. Chan, PhD, Dang V. Do, BSc, Audrey Khoo, BSc,Chee T. Ong, BSc, Han H. Cheong, BSc, Ivor J. Lim, FRCS, and Thang T. Phan, MD, PhD

Background: Growth factors and cytokines involved in the wound healingprocess seem to be immobilized at the cell surface and extracellular matrixvia binding with proteoglycans, making them important modulators of celldynamics. Our aim was to investigate the expression of two proteoglycans,namely syndecan-2 and decorin, and to elucidate their role in the pathogen-esis of an aberrant wound healing process leading to keloid scar.Methods: Intrinsic expression of syndecan-2, fibroblast growth factor(FGF)-2, and decorin in keloid tissue was investigated using Westernblotting and immunohistochemistry. Normal and keloid fibroblasts weretreated with serum to see the effects of serum growth factors on theexpression of syndecan-2 and decorin. The role of epithelial-mesenchymalinteractions in modulating syndecan-2, FGF-2, and decorin expression wasinvestigated using an established two-chamber serum-free coculture model.Finally, the antifibrotic effect of decorin was investigated by studying itseffect on the expression of extracellular matrix components.Results: Syndecan-2 and FGF-2 were upregulated in keloid tissue; decorinwas downregulated. Normal and keloid fibroblasts treated with serum led toincrease in syndecan-2 and decrease in decorin expression. Under cocultureconditions, syndecan-2 was shed in the conditioned media. FGF-2 was alsoupregulated under coculture conditions and, when added to fibroblast mo-nocultures, increased shedding of syndecan-2. Decorin levels were upregu-lated under coculture conditions only in normal cocultures. Decorin was alsoable to decrease extracellular matrix proteins, highlighting its importance asan antifibrotic agent.Conclusion: Syndecan-2 and FGF-2 are not only overexpressed in keloidtissues but may interact with each other resulting in the shedding ofsyndecan-2, which in turn might activate a whole cascade of events respon-sible for a keloidic phenotype. In addition, decorin had an antifibrotic effectand could well be used as a potential therapeutic agent for keloids.Key Words: Syndecan-2, FGF-2, Decorin, Fibroblasts, Keloid Scar, Kera-tinocytes, TGF-�, Epithelial-mesenchymal interactions.

(J Trauma. 2010;68: 999–1008)

Keloids and hypertrophic scars are fibroproliferative disor-ders that may follow local skin trauma or inflammatory

skin disorders and encompass lacerations, tattoos, burns,injections, ear piercings, vaccinations, bites, acne, abscesses,or surgery.1 A keloid scar extends beyond the confines of theoriginal wound, does not regress spontaneously, grows inpseudotumor fashion with distortion of the lesion, and tendsto recur after excision. It is known to occur only in humansand leads to various functional and esthetic abnormalities.Numerous treatment modalities have been explored, but noneof the treatments are effective; this is partly because the exactmechanisms underlying keloid pathogenesis remain elusive.To identify new targets for therapeutic intervention, it isimperative to identify key modulators of keloidogenesis.

The transforming growth factor (TGF)-� family ofgrowth factors has been widely studied and is thought to playa central role in tissue regeneration by regulating fibroblastproliferation, differentiation, and matrix production.2–4 How-ever, various other soluble and insoluble effectors are equallyimportant in wound repair and, thus, may play a part in keloidformation. One such effector group comprises cell surface pro-teoglycans, namely the syndecans, which are a family of trans-membrane heparan sulfate proteoglycans that are differentiallyexpressed during development and wound repair.5,6 Because oftheir unique function of integrating signaling from circulatinggrowth factors and extracellular matrix (ECM) proteins withother cellular receptors such as the integrins, they have oftenbeen referred to as “tuners of transmembrane signaling.”7 Theyexist as four isoforms: syndecan-1, syndecan-2, syndecan-3, andsyndecan-4, each having characteristic cell- and tissue-specificdistribution throughout development.8

Syndecan-2, also called fibroglycan, is especially abun-dant in mesenchymal cells9 and has been shown to play animportant role in diverse biological processes. Syndecan-2modulates cell behavior by affecting cell adhesion and signal-ing.10 Chen et al.11 demonstrated syndecan-2 to regulate TGF-�signaling and reported that syndecan-2 could regulate signalingof TGF-� by binding to it through the core protein of itsectodomain. It has been shown to interact with various othercytokines and growth factors, such as fibroblast growth factor(FGF)-2 and vascular endothelial growth factor, and mediatetheir effects on promoting mesenchymal tissue growth andangiogenesis.12–14 Syndecan-2 has been implicated in varioustypes of cancers, with increased expression demonstrated invarious cancer cell lines.15 Owing to these properties, syndecan-2has recently become a subject of widespread research.

Submitted for publication November 26, 2008.Accepted for publication October 1, 2009.Copyright © 2010 by Lippincott Williams & WilkinsFrom the Departments of Surgery (A.M., D.V.D., A.K., C.T.O., I.J.L., T.T.P.),

Bioengineering (T.T.P), and Pharmacy (M.Y.W., S.Y.C., H.H.C.) and NUSTissue Engineering and Stem Cell Research Program (T.T.P.), NationalUniversity of Singapore, Singapore.

Supported by the Biomedical Research Council, Singapore, grants 03/1/21/19/251, 04/1/21/19/338, and 05/1/21/19/390, the National Medical ResearchCouncil, Singapore, grant NMRC/1026/2005, and a scholarship from theNational University of Singapore (to M.A.).

The authors state no conflict of interest.Address for reprints: Phan TT, MD, PhD, Department of Surgery, National

University of Singapore, 10 Kent Ridge Crescent, Singapore 119260; email:surptt@nus.edu.sg.

DOI: 10.1097/TA.0b013e3181c4070d

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010 999

In addition to heparan sulfate proteoglycans, solublechondroitin and dermatan sulfate proteoglycans such asdecorin have also been shown to play a role in cell-matrixinteractions. Decorin is an abundant component of bone andskin ECM and associates with types I, II, and VI collagenfibrils and fibronectin in vivo.16 Decorin is expressed abun-dantly in normal fibroblasts (NF) in which its expression hasbeen shown to be enhanced by interleukin (IL)-1 and dexa-methasone and inhibited by TGF-�.17–19 Unlike the synde-cans, decorin inhibits the biological activity of TGF-� bybinding with the core protein in various cell types20 and, thus,has been explored as a potential antifibrotic agent. Its antifi-brotic effect has been demonstrated in experimental kidney,lung, and muscular fibroses.21–23

In this study, we studied intrinsic syndecan-2, FGF-2,and decorin expression in keloid tissue by performing immu-nohistochemical analysis and Western blot assays. We furtherinvestigated the role of serum on the expression profile ofsyndecan-2 and decorin by stimulating the NF and keloidfibroblasts (KF) with 10% fetal calf serum (FCS). The role ofFGF-2 in shedding of the syndecan-2 ectodomain and itsimplications in modulating cell behavior was investigated bytreating keloid fibroblast cultures with recombinant FGF-2and immunoblotting the conditioned media for the expressionof syndecan-2. The role of epithelial-mesenchymal interac-tions in modulating syndecan-2, FGF-2, and decorin expres-sion dynamics was studied using an established two-chamberserum-free coculture model. In addition, the antifibrotic effectof decorin was studied by investigating its effect on theexpression of ECM components collagen, fibronectin, and�-smooth muscle actin (SMA).

MATERIALS AND METHODS

Keloid Keratinocyte and Fibroblast DatabaseKeratinocytes and fibroblasts were randomly selected

from a specimen bank of keratinocyte/fibroblast strains de-rived from excised keloid and normal skin specimens (Table1). All patients had received no previous treatment for thekeloids before surgical excision. A full history was taken andexamination performed by a plastic surgeon, complete withcolor slide photographic documentation, before taking in-formed consent before excision. Approval by the NationalUniversity of Singapore-Institutional Review Board was ob-tained before excision of human tissue and collection of cells.

Primary Cell CultureKeratinocyte Culture From Keloids

Keloid specimens after excision were repetitivelywashed in phosphate buffer saline containing 150 �g/mLgentamicin and 7.5 �g fungizone, until the washing solutionbecame clear. The epidermis was then scored after dividingthe tissues into 5 mm � 10 mm pieces. Dispase (5 mg/mL)was added in Hank’s balanced salt solution before overnightincubation of the tissue at 4°C. The epidermis was carefullyscraped off with a scalpel the next day and placed in 0.25%trypsin/0.1% glucose/0.02% ethylenediaminetetraacetic acid for10 minutes in the incubator. Trypsin action was quenched byaddition of Dulbecco’s modified Eagle’s medium (DMEM)/

10% FCS. The suspended cells were transferred into tubes andcentrifuged at 1,000 rpm for 8 minutes. The cells were seeded inkeratinocyte culture medium (80 mL DMEM supplementedwith 20 mL FCS, 10 ng/mL epidermal growth factor, 1 � 10�9

M cholera toxin, and 0.4 �g/mL hydrocortisone) at 1 � 105

cells/cm2 for 24 hours before changing to keratinocyte growthmedium. The cell strains were maintained and stored in liquidnitrogen tanks. Only cells from second and third passages wereused for the experiments.

Fibroblast Culture From KeloidsRemnant dermis from the keloids was minced and

incubated in a solution of collagenase type 1 (0.5 mg/mL)and trypsin (0.2 mg/mL) at 37°C for 6 hours. Cells werepelleted and grown in tissue culture flasks. Fibroblast cellstrains were maintained and stored in liquid nitrogen tanksuntil use. Only cells from the second and third passageswere used for all experiments.

TABLE 1. Patient Database

Gender Age (yr)

Normal

NF/NK1 F Chinese 27 Breast —

NF2 M Indian 32 Abdominal —

NF/NK4 M Chinese 3 Earlobe —

NF8 F Malay 28 Breast —

NF/NK9 F Chinese 24 Breast —

NF/NK5 F Chinese 35 Breast —

NF/NK3 F Chinese 12 Groin —

NS31 F Malay 13 Groin —

NS32 M Chinese 20 Forearm —

NS33 M Chinese 29 Thigh —

NS34 F Chinese 35 Breast —

NS35 M Malay 33 Forearm —

Keloid

KF6 F Chinese 16 Earlobe 1

KF46 M Chinese 36 Forearm 1

KF3 F Malay 20 Earlobe 1.5

KS/KF/KK48 F Indian 23 Earlobe 1.5

KF/KK31 F Chinese 17 Earlobe 0.5

KF/KK32 M Indian 28 Chest 2

KF/KK43 M Chinese 34 Elbow 1

KF/KK16 F Chinese 19 Earlobe 1.5

KS/KF/KK 25 M Indian 25 Forearm 1

KF/KK30 F Chinese 30 Face 1

KF/KK45 F Malay 23 Forearm 1.5

KF/KK24 M Malay 21 Chest 2

KF/KK29 F Indian 18 Earlobe 1

KS50 M Malay 10 Face 1

KS51 F Chinese 18 Earlobe 1.5

KS52 M Indian 21 Elbow 1

KS53 M Malay 28 Face 1.5

KS54 F Malay 16 Forearm 1

KS55 M Chinese 25 Earlobe 2

KS56 F Chinese 22 Earlobe 1

KS57

NS, normal skin; KS, keloid scar.

Mukhopadhyay et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010

© 2010 Lippincott Williams & Wilkins1000

Normal Human Fibroblasts and KeratinocytesFibroblasts and keratinocytes were isolated from normal

adult skin samples removed in plastic surgery procedures. In-formed consents were obtained with National University ofSingapore-Institutional Review Board approval.

Keratinocyte-Fibroblast CocultureNormal keratinocytes (NK) and keloid keratinocytes

(KK) were seeded at a density of 1 � 105 cells/cm2 on6-well Transwell (Costar, Singapore, Singapore) clearpolyester membrane inserts with a 0.4-�m pore size and anarea of 4.5-cm2. Ten days before coculture, cells weremaintained in serum-free keratinocyte growth mediumuntil they were 100% confluent in a monolayer. The mediumwas then changed to DMEM supplemented with 10% heat-inactivated FCS and penicillin/streptomycin (25 units each).The cells were raised to the air-liquid interface to allow thekeratinocytes to stratify and reach terminal differentiation.

NF and KF obtained from randomly selected normaland keloid strains were seeded in 6-well plates at a density of1 � 104 cells/mL in DMEM/10% FCS for 24 hours and thenin serum-free medium for another 48 hours. Cells on both themembrane inserts and the wells were washed twice withphosphate buffer saline to remove the old medium beforecombination of the inserts and plates for coculture in serum-free DMEM. Controls comprised one series each of non-cocultured NF and KF.

Serum Growth Factor Stimulation of NF and KFWhen fibroblasts are exposed to serum, the signal is

interpreted to be a physiologic wounding signal.24 The serumstimulation model is an in vitro model used to mimic the earlyphases of wounding. NF and KF were seeded in 6-well platesat a density of 1 � 104 cells/mL in 10% FCS for 24 hours andsubsequently seeded in serum-free medium for another 48hours. The fibroblasts were exposed to either DMEM supple-

Figure 1. Elevated levels of syndecan-2 and FGF-2, and lower levels of decorin in tissue extracts obtained from keloid tissue.Expression of syndecan-2, FGF-2, and decorin in normal and keloid skin tissues were analyzed by Western blot assay. Studygroups comprised normal skin and keloids. Tissue specimens were sonicated in lysis buffer, and 100 �g of total protein ex-tracts were subjected to SDS-PAGE and Western blotting. Blots were hybridized with anti-syndecan-2 polyclonal antibody. (A)Lanes 1–5: Keloid scar (K50, K51, K52, K53, and K54). Lanes 6–8: Normal skin (N31, N32, and N33). Equal loading was con-firmed by blotting with �-actin. (B) Lanes 1–8: Keloids (K50, K51, K52, K53, K54, K55, K56, and K57). Lanes 9–13: Normalskin (N31, N32, N33, N34, and N35). Equal loading was confirmed by blotting with �-actin. (C) Lanes 1–3: Keloids (K51,K52, and K53). Lanes 4–6: Normal skin (N31, N32, and N33). Equal loading was confirmed by blotting with �-actin.

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010 Syndecan-2 and Decorin in Keloid Pathogenesis

© 2010 Lippincott Williams & Wilkins 1001

mented with 10% FCS or DMEM alone for 24 hours beforebeing harvested for analysis.

Treatment of NF and KF With rhFGF-2 andrhDecorin

Different strains of NF and KF were seeded in 6-wellplates at a density of 1 � 104 cells/mL in DMEM/10% FCSfor 24 hours and then in serum-free medium for another 48hours. The cells were subsequently treated with varyingconcentrations of rhDecorin (0 ng/mL, 250 ng/mL, 500 ng/mL, 1,000 ng/mL, and 2,000 ng/mL; R&D Systems) andrhFGF-2 (10 ng/mL; R&D Systems). The cell lysate andconditioned media from these treated cultures were collectedand analyzed by Western blotting.

ImmunohistochemistryParaffin sections were dewaxed and antigens retrieved

by immersing slides in 0.01 mol/L citrate buffer, pH 6.0,heating in a microwave oven (high for 2.5 minutes, low for 5minutes), cooling at 4°C for 20 minutes, and then washing inwater for 5 minutes. Endogenous peroxidase was blocked in3% H2O2, and nonspecific binding blocked for 20 minutes(Universal R.T.U vectastain kit; Vector Labs). Sections wereincubated with antibodies specific for FGF-2 (Santa CruzBiotechnology) diluted 1:50 for 1 hour. After washing, slideswere incubated in universal secondary antibody provided inthe vectastain kit. Slides were washed in Tris-buffered NaCl(TBS), with 0.05% Tween-20, pH 7.5, and then MilliQ H2O.The reaction product was developed with a 3, 3�-diamino-benzidine tetrahydrochloride substrate kit (Zymed), and sec-tions were counterstained with hematoxylin. All wash stepswere in TBS/0.05% Tween-20. Antibodies were diluted in1% bovine serum albumin/TBS. Nonimmune rabbit antibodyfor FGF-2 of the appropriate immunoglobulin isotype wasused as negative controls.

Preparation of Normal Skin and Keloid TissueExtracts

Keloid and normal skin tissue extracts were preparedby sonicating 120 mg of skin tissues in 200 �L of lysis buffercontaining 20 mmol/L Tris-HCL (pH 7.5), 1% Triton X-100,100 mmol/L NaCl, 0.5% Nonidet P-40, and 1 mg/mL pro-tease inhibitor cocktail (Boehringer Mannheim, Mannheim,Germany), followed by centrifugation at 13,000 rpm for 10minutes. Protein concentration of the tissue extracts wasdetermined by the Bradford method, and 100 �g of proteinsubjected to SDS-PAGE and Western blotting.

Western BlotTotal protein extracts from keloid and normal skin

tissue were obtained from freshly excised specimens. NF orKF and keratinocytes under different culture conditions werelysed in cell lysis buffer containing 20 mmol/L Tris-HCl (ph7.5), 1% Triton X-100, 100 mmol/L NaCl, 0.5% NonidetP-40, and 1 mg/mL protease inhibitor cocktail (antipaindihydrochloride, aprotinin, bestatin, chymostatin, E-64, eth-ylenediaminetetraacetic acid-Na, leupeptin, Pefabloc-SC,pefstatin, phosphormidon; Boehringer Mannheim, Mann-heim, Germany). The lysates were then centrifuged at 13,000

rpm for 15 minutes to remove cell debris. The supernatant(whole cell extract) was then quantified, and 100 �g ofprotein was electrophoresed on 8% and 14% sodium dodecylsulfate-polyacrylamide gels (SDS-PAGE) using the Protein IIsystem (Bio-Rad). The proteins were then electrophoreticallytransferred to a nitrocellulose membrane (Bio-Rad) in freshlyprepared transfer buffer (25 mmol/L Tris base, 190 mmol/Lglycine, and 20% methanol). The blots were subsequentlyprobed with anti-human syndecan-2 rabbit polyclonal anti-body (Santa Cruz Biotechnology), anti-human FGF-2 rabbitpolyclonal antibody (Santa Cruz Biotechnology), anti-humandecorin mouse monoclonal antibody (R&D systems), mono-clonal antibodies to collagen I, III (Monosan antibodiesTheNetherlands), �-SMA (Sigma), or fibronectin (BD Transduc-tion Laboratories), followed by treatment with their respec-tive secondary antibodies. The blots were visualized with achemiluminescence-based photoblot system (Amersham Bio-sciences). Supernatants (conditioned media) were concen-trated from 10 mL to 100 �L using a Centricon centrifuge(Millipore, MA) and then equal volumes were loaded andsubjected to Western blotting.

Statistical AnalysisStatistical significance between groups was assessed

using either Student’s t test or a one-way analysis of variancewith Tukey’s post hoc test. All tests were performed usingMiniTab software. Mean value of data from five different cell

Figure 2. Increased localization of FGF-2 in the basal layerof epidermis and dermis in keloid tissue. Paraffin sections ofnormal (A, C) and keloid (B, D) tissues were prepared andlabeled with antibodies specific for FGF-2 and observed withmicroscopy (�40). In each panel, the insert shows the sametissue labeled with a nonimmune rabbit antibody of the ap-propriate immunoglobulin isotype instead of the specificFGF-2 antibody. Keloid tissue had more intense labeling inthe basal layer of the epithelium and the dermis comparedwith normal skin. The localization of FGF-2 (or its lack in thenegative control inserts) is shown by an arrow. The dermisand epidermis are represented by “D” and “E,” respectively.

Mukhopadhyay et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010

© 2010 Lippincott Williams & Wilkins1002

populations in each group was used. In all results, the cellpopulation representative of other populations studied hasbeen shown.

RESULTS

Keloid Scar Intrinsically Expresses HigherLevels of Syndecan-2 and FGF-2, and LowerLevels of Decorin

The expression profile of syndecan-2, FGF-2, anddecorin in keloids and normal skin tissue was studied using aWestern blot assay. Although normal skin expressed lowlevels of syndecan-2 (Fig. 1, A) and FGF-2 (Fig. 1, B),decorin was abundantly expressed (Fig. 1, C). In contrast,keloid scar samples had abundant levels of syndecan-2 (Fig.1, A), FGF-2 (Fig. 1, B) and low levels of decorin (Fig. 1, C).Although decorin was downregulated in keloid tissue, a32-kDa band was observed in this sample, which was com-pletely missing in normal tissue.

Increased Localization of FGF-2 Is Seen in theBasal Epidermis and Dermis of KeloidsCompared With Normal Skin

FGF-2 was observed to be localized both in the dermisand epidermis of keloid scars in higher levels when compared

with normal skin, where basal levels of expression wasobserved (Fig. 2, A to D).

Serum Growth Factors Upregulate Syndecan-2and Downregulate Decorin Expression in KF

Stimulation with 10% FCS is interpreted by fibroblastsas a wound healing signal; thus, NF and KF were treated with10% FCS to investigate if syndecan-2 and decorin wereinvolved in the initial phases of wound healing. Both NF andKF showed increased levels of syndecan-2 (Fig. 3, A) onstimulation with serum growth factors, suggesting a possiblerole of syndecan-2 in the initial wound healing mechanism.However, decorin expression was downregulated on stimu-lation with serum (Fig. 3, B).

Coculture of KK With KF and NK With NFIncrease Levels of Syndecan-2 and Decorin inthe Conditioned Media, Respectively

To investigate if epithelial-mesenchymal interactionsmodulate syndecan-2 expression, a coculture model wasset up wherein NK and KK were cocultured with NFand KF, respectively. Fibroblasts were subsequently lysed,and total cell extracts were subjected to Western blot.Although no difference was observed in cellular levels ofsyndecan-2 in both NF and KF cocultured with their

Figure 3. Serum growth factors upregulated syndecan-2 expression and downregulate decorin expression in NF and KF. Ex-pression of syndecan-2 and decorin by fibroblast after exposure to 10% FCS was detected by Western blot assay. Fibroblastswere seeded in 6-well plates at a density of 1 � 104 cells/mL in DMEM/10% FCS for 24 hours and, subsequently, in serum-free medium for another 48 hours. The fibroblasts were then subjected to either DMEM/10% FCS or serum-free DMEM for 24hours. Fibroblasts were lysed, and 100 �g total protein cell lysates were subjected to SDS-PAGE and Western blotting. Blotswere hybridized with anti-syndecan-2 polyclonal antibody. (A) Lane 1: NF4�DMEM; lane 2: NF4�DMEM/10% FCS; lane 3:NF14�DMEM; lane 4: NF14�DMEM/10%FCS; lane 5: KF45�DMEM; lane 6: KF45�DMEM/10%FCS; lane 7: KF49�DMEM;and lane 8: KF49�DMEM/10% FCS. Equal loading was confirmed by blotting with �-actin for cell extracts. (B) Lane 1:NF4�DMEM; lane 2: NF4�DMEM/10% FCS; lane 3: NF14�DMEM; lane 4: NF14�DMEM/10%FCS; lane 5: KF45�DMEM;lane 6: KF45�DMEM/10%FCS; lane 7: KF49�DMEM; and lane 8: KF49�DMEM/10% FCS. Equal loading was confirmed byblotting with �-actin for cell extracts.

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010 Syndecan-2 and Decorin in Keloid Pathogenesis

© 2010 Lippincott Williams & Wilkins 1003

respective keratinocytes (Fig. 4, A), an increased level ofsyndecan-2 was observed in the conditioned media ofKK/KF cocultures compared with KF in monoculture (Fig.4, B), suggesting the possible cleavage of syndecan-2 inKF on coculture with KK. Although conditioned mediaobtained from the coculture of NF with NK demonstratedhigher levels of decorin compared with NF monocultures(Fig. 4, C), no significant difference in decorin levels wasobserved in the conditioned media of cocultured KF whencompared with KF in monoculture (Fig. 4, D) suggesting aloss of KK control of underlying KF decorin expression.

Cocultured Keratinocytes and FibroblastsDemonstrate Increased Expression of FGF-2Compared With Cells in Monoculture

To investigate the effect of epithelial-mesenchymalinteractions on the dynamics of FGF-2 expression, NK andKK were cocultured with NF and KF, respectively, with celllysates collected from both groups and blotted for FGF-2expression. An increase in FGF-2 expression was observed incocultured keratinocytes (Fig. 5, A). In addition, coculturedKK expressed higher levels of FGF-2 when compared withcocultured NK (Fig. 5, B), suggesting a more active mecha-

Figure 4. Increased syndecan-2 and decorin is observed in keloid and normal cocultures, respectively. (A) Cell lysates ob-tained from NF and KF cocultured with NK and KK, respectively, were assayed by Western blot for syndecan-2 expression. Celllysate from NF and KF in monoculture were used as controls. No difference in the expression of syndecan-2 was observed incocultured fibroblast cell extracts compared with fibroblasts in monoculture. Lane 1: NF1 noncocultured; lane 2: NK1/NF1;lane 3: NF2 noncocultured; lane 4: NK3/NF2; lane 5: KF 45 noncocultured; lane 6: KK45/KF45; lane 7: KF 48 noncocultured;and lane 8: KK48/KF48. (B) Conditioned media from keloid fibroblast cocultures were assayed by Western blot for syndecan-2expression. Conditioned media from KF in monoculture were used as controls. Equal volumes of conditioned media were sub-jected to SDS-PAGE. Conditioned media from KF cocultures shed more syndecan-2 into the conditioned media when com-pared with the KF monocultures. Lane 1: KF48 noncocultured; lane 2: KK48/KF48; lane 3: KF45 noncocultured; lane 4: KK45/KF45; lane 5: KF24; and lane 6: KK24/KF24. NF and KF were cocultured with NK and KK, respectively, and conditioned mediawas collected and analyzed for the expression of decorin. (C) Lane 1: NF4�DMEM; lane 2: NF1�DMEM; lane 3:NF14�DMEM; lane 4: NF3�DMEM; lane 5: NK1/NF1; lane 6: NK2/NF2; lane 7: NK14/NF14; and lane 8: NK5/NF5. (D) Lane1: KF29�DMEM; lane 2: KF48�DMEM; lane 3: KF49�DMEM; lane 4: KF25�DMEM; lane 5: KK48/KF48; lane 6: KK45/KF45;lane 7: KK25/KF25; and lane 8: KK49/KF49.

Mukhopadhyay et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010

© 2010 Lippincott Williams & Wilkins1004

nism within KK compared with NK. FGF-2 upregulation wasalso observed in cocultured fibroblasts (Fig. 5, C), whencompared with those in monoculture, highlighting a possibleautocrine effect of FGF-2 on keratinocytes and a paracrineeffect on the underlying fibroblasts.

rhFGF-2 Increases Syndecan-2 Cleavage and,Thus, Its Detectable Levels in Keloid FibroblastConditioned Media

The role of FGF-2 in the cleavage of fibroblast syndecan-2resulting in its increased levels in conditioned media was inves-tigated by treating KF with exogenous FGF-2 and analyzing theconditioned media for expression of syndecan-2 by Westernblot. A significant increase in the levels of syndecan-2 wasobserved compared with untreated KF (Fig. 6).

Decorin Decreases the Expression of ECMComponents in Both NF and KF

To investigate the effect of decorin on the expression ofECM components, decorin at various concentrations (250ng/mL, 500 ng/mL, 1,000 ng/mL, and 2,000 ng/mL) wasadded to both NF and KF, after which cell lysates andconditioned media were analyzed for collagen, fibronectinand �-SMA expression. A decrease in the expression of theECM components was observed both in the cell lysate (Fig.7, A and B) and conditioned media (Fig. 7, C).

DISCUSSIONIt has been found that most of the growth factors and

cytokines involved in the wound healing process to be im-

Figure 5. Cocultured NK and KK and fibroblasts express increased levels of FGF-2 and compared with monocultured cells.Cell lysates from monocultured NK and KK and fibroblasts were extracted and analyzed by Western blot for the expression ofFGF-2. Similarly, lysates from cocultured NK and KK and fibroblasts were lysed, and cell extracts were analyzed by Westernblot for the expression of FGF-2. (A) Lane 1: NK1; lane 2: NK1/NF1; lane 3: KK48; and lane 4: KK48/KF48. (B) Lane 1: NK1/NF1; lane 2: NK2/NF2; lane 3: NK3/NF3; lane 4: KK48/KF48; lane 5: KK45/KF45; and lane 6: KK49/KF49. Equal loading wasconfirmed by blotting with �-actin for cell extracts. (C) Lane 1: NF1; lane 2: NK1/NF1; lane 3: NF2; lane 4: NK2/NF2; lane 5:KF48; lane 6: KK48/KF48; lane 7: KF49; and lane 8: KK49/KF49.

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010 Syndecan-2 and Decorin in Keloid Pathogenesis

© 2010 Lippincott Williams & Wilkins 1005

mobilized at the cell surface and ECM through binding withproteoglycans.25 These proteoglycans in turn are known toplay important roles in various physiologic processes. Be-sides providing mechanical strength by filling in the spacebetween the collagen and elastin fibers by absorbing water,they also influence collagen formation, cell proliferation, cellmigration, and cell adhesion during wound healing.26

Syndecan-2, a cell surface proteoglycan, plays an im-portant role in the transmodulation of ECM assembly.Because keloid scars are characterized by an aberrant orga-nization of the ECM assembly, we investigated if syndecan-2was being abnormally expressed in keloid tissues when com-pared with normal skin tissue. Western blot analysis of thetissue extracts demonstrated that syndecan-2 was indeedupregulated in keloid tissue and that it could be one of thefactors responsible for the excess deposition of ECM com-ponents.

To understand the underlying biology of syndecan-2 inwound repair and scar formation, we investigated the expres-sion profile of syndecan-2 at different stages of the woundhealing process using in vitro models. We used the serumstimulation model to understand the response of fibroblasts toserum, the soluble fraction of clotted blood that is normallyencountered by cells in vivo in the context of a wound. Weobserved that NF and KF when exposed to serum growthfactors expressed increased amounts of syndecan-2, suggest-

ing its possible role in the early stages of the healing processwhen the wound bed is exposed to serum factors and in themodulation of the ensuing events by interaction with some ofthe macrophage-derived growth factors, such as FGF-2, vas-cular endothelial growth factor, and epidermal growth fac-tor.7 The hypothesis that syndecan-2 might be an importantplayer in early phase of the healing process is furtherstrengthened by numerous studies reporting syndecan-2 to beinduced by proinflammatory cytokines, such as IL-1� andIL-1� and TGF-�, which are abundant in the early woundenvironment.14,27,28

Strikingly enough, it has been observed that duringdevelopment, syndecan-2 is localized at sites where intensiveepithelial-mesenchymal interactions shape and transform theepithelia and the mesenchyme into morphologically and func-tionally differentiated tissues, suggesting a possible role ofsyndecan-2 in modulating epithelial-mesenchymal interac-tions.10 Previous findings from our group have shown thatepithelial-mesenchymal interactions play an important role inkeloid pathogenesis with KK modulating NF and KF growthand proliferation,28 influencing collagen expression by bothNF and KF, inducing NF to secrete collagen in a keloid-likemanner,29 and using the insulin-like growth factor system ofmitogens as inductive ligands for this process. Thus, toinvestigate if these interactions in anyway affected expres-sion of syndecan-2, NF and KF were cocultured with NKand KK, respectively, and the cell lysates and conditionedmedia blotted. Surprisingly, although we observed thatthere was no difference in the expression of syndecan-2from cell lysates, a definite increase in syndecan-2 levelswas seen in the conditioned media obtained from KF cocul-tured with KK. This strongly suggests that syndecan-2 wasbeing shed from the keloid fibroblast cell surface into theconditioned media under the influence of the overlying ker-atinocytes. Syndecan-1 and -4 have been previously shown tobe shed from the surface of mouse epithelia and SVEC4–10cells in vitro, and factors released during stress, injury andcancer have been demonstrated to upregulate this process.30,31

Previous studies by Fears et al.32 highlighted the role ofsyndecan-2 shedding, induced by various factors, to promoteangiogenesis. Of the various factors studied, FGF-2 wasobserved to stimulate syndecan-2 shedding from microvas-cular endothelial cells. When we added exogenous rhFGF-2 toour keloid fibroblast cultures, syndecan-2 shedding was alsoobserved, with increased amounts of secreted syndecan-2 foundin the conditioned media.

The production and secretion of heparin-bindinggrowth factors, such as FGF, have been implicated in variousmacrophage-mediated processes including tumor angiogene-sis, arteriogenesis and capillary sprouting, wound healing,and inflammation.14 It would appear that the syndecan-FGFsystem could also be an important player in initiating aberrantwound healing processes leading to fibrosis. BecauserhFGF-2 increased syndecan-2 shedding in KF, we investi-gated if FGF-2 was intrinsically upregulated in keloid tissuescompared with normal skin and, more importantly, in KK/KFcocultures, where increased levels of syndecan-2 had beenobserved in conditioned media. We found increased expres-

Figure 6. Exogenous rhFGF-2 stimulates shedding of synde-can-2 from fibroblast cell surface into the conditioned me-dia. The rhFGF-2 (10 ng/mL) was added to the fibroblastcultures and conditioned media assayed for the expressionof syndecan-2. Conditioned media from untreated KF wereused as controls. Equal volumes of conditioned media weresubjected to SDS-PAGE. KF treated with FGF-2 shed moresyndecan-2 when compared with untreated fibroblasts. Lane1: KF43 noncocultured; lane 2: KF43�FGF-2; lane 3: KF29noncocultured; and lane 4: KF29�FGF-2.

Mukhopadhyay et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010

© 2010 Lippincott Williams & Wilkins1006

sion and localization of FGF-2 in keloid epidermis anddermis with compared with normal skin sections. In addition,fibroblasts cocultured with NK and KK expressed increasedlevels of FGF-2 when compared with fibroblasts in single cellculture. Interestingly, KK/KF cocultures expressed increasedlevels of FGF-2 when compared NK/NF. One arm of keloi-dogenesis resulting from epithelial-mesenchymal interactionswould, thus, appear to involve increased FGF-2 levels, whichin turn stimulate the shedding of syndecan-2 from fibroblasts.This shed syndecan-2 would modulate not only the activity ofFGF-2 but also other profibrotic growth factors.

FGF-2 has also been demonstrated to decrease decorinmRNA expression in osteosarcoma cells. The reduced ex-pression of decorin corresponds with increased collagen pro-duction by these cell types.33 Because keloids are marked byincreased production of type 1 collagen, we investigated theexpression of decorin in keloid tissues. As expected, a de-creased expression of decorin was observed in keloid tissueswhen compared with normal skin. Interestingly, a 32-kDaband was observed in keloid tissues probably because of thedegradation of decorin to an inactive fragment by variousmatrix metalloproteinases that are known to be present inkeloids.34 Previous studies have shown matrix metallopro-teinases to degrade decorin to an inactive fragment, whichis then unable to further bind to TGF-�, releasing it from

the ECM with resultant increased activity in vivo.35 Fur-ther studies will be required to assess if this mechanismplays a role in keloid pathogenesis. When stimulated with10% serum, a reduction in decorin was observed in bothNF and KF. This may be because of negative feedbackfrom increased levels of TGF-� and other growth factors inthe serum.

We investigated the effect of epithelial-mesenchymalinteractions in modulating the expression of decorin in ourcoculture model and observed that although NK/NF cocul-tures expressed increased amounts of decorin compared withmonocultures, KK/KF cocultures did not show any signifi-cant difference in decorin levels compared with monocul-tures. This would suggest that the overlying keratinocytesmight have directly or indirectly blocked the mechanismsresponsible for modulating fibroblast decorin expression.FGF-2 demonstrated above to be upregulated in coculturesmight well be one of the factors responsible, and furtherstudies will be required confirm this hypothesis.

Finally, we explored if exogenous administration ofdecorin could be used as an antifibrotic agent. NF and KFwere treated with decorin at different concentrations, andexpression of different ECM components was investigated. Itwas observed that collagen was downregulated on treatmentin a dose-dependent manner, both in the cell lysate and the

Figure 7. Effect of decorin on collagen, fibronectin, and �-SMA expression. NF or KF were cultured in the absence or pres-ence of different concentrations decorin. Cells were harvested, lysed, and both cell lysate and conditioned media collected forWestern blot analysis for collagen, fibronectin, or �-SMA. (A) Lane 1: NF1�DMEM; lane 2: NF1�decorin (250 ng/mL); lane 3:NF1�decorin (500 ng/mL); lane 4: NF1�decorin (1,000 ng/mL); and lane 5: NF1�decorin (2,000 ng/mL). (B) Lane 6:KF48�DMEM; lane 7: KF48�decorin (250 ng/mL); lane 8: KF48�decorin (500 ng/mL); and lane 9: KF48�decorin (1,000 ng/mL). (C) Lane 1: KF48�DMEM; lane 2: KF48�decorin (250 ng/mL); lane 3: KF48�decorin (500 ng/mL); and lane 4:KF48�decorin (1,000 ng/mL).

The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010 Syndecan-2 and Decorin in Keloid Pathogenesis

© 2010 Lippincott Williams & Wilkins 1007

conditioned media. In addition, �-SMA, a phenotypic markerfor myofibroblasts, and fibronectin were also downregulatedon treatment with decorin, which would suggest the potentialutility of decorin against excess fibrosis.

Overall, we have shown that syndecan-2 and FGF-2 arenot only overexpressed in keloid tissues but may also interactwith each other, resulting in the shedding of syndecan-2, whichin turn activates a whole cascade of events that result in keloi-dogenesis. In addition, decorin seems to be downregulated inkeloid tissues, and this protein with strong antifibrotic effects haspotential to be used as a therapeutic agent for keloids.

REFERENCES1. Mutalik S. Treatment of keloids and hypertrophic scars. Indian J

Dermatol Venereol Leprol. 2005;71:3–8.2. Derynck R, Zhang YE. Smad-dependent and Smad-independent path-

ways in TGF-� family signaling. Nature. 2003;425:577–584.3. Kim WJH, Levinson H, Gittes GK, Longaker MT. Molecular mecha-

nisms in keloid biology. In: Garg HG, Longaker MT, eds. ScarlessWound Healing. New York, NY: Marcell Dekker, Inc; 2000:161–171.

4. Tuan TL, Nitcher LS. The molecular basis of keloid and hypertrophicscar formation. Mol Med Today. 1998;4:19–24.

5. Carey DJ. Syndecans: multifunctional cell—surface co-receptors. Bio-chem J. 1997;327:1–16.

6. Lories V, Cassiman JJ, Berghe HVV, David G. Differential expressionof cell surface heparan sulphate proteoglycans in human mammaryepithelial cells and lung fibroblasts. J Biol Chem. 1992;267:1116–1122.

7. Zimmermann P, David G. The syndecans, tuners of transmembranesignaling. FASEB J. 1999;13:S91–S100.

8. Bernfield M, Hinkes MT, Gallo RL. Developmental expression of thesyndecans: possible function and regulation. Dev Suppl. 1993;205–212.

9. Gallo RL, Bernfield M, Kainulainen V. Syndecan biology in woundrepair. In: Garg HG, Longaker MT, eds. Scarless Wound Healing. NewYork, NY: Marcell Dekker, Inc; 2000:37–61.

10. Essner JJ, Chen E, Ekker SC. Syndecan-2. Int J Biochem Cell Biol.2006;38:152–156.

11. Chen L, Klass C, Woods A. Syndecan-2 regulates transforming growthfactor-beta signaling. J Biol Chem. 2004;279:15715–15718.

12. Burgess WH, Maciag T. The heparin-binding (fibroblast) growth factorfamily of proteins. Annu Rev Biochem. 1989;58:575–606.

13. Leung DW, Cachianes G, Kuang WJ, Goeddel DV, Ferrara N. Vascularendothelial growth factor is a secreted angiogenic mitogen. Science.1989;246:1306–1309.

14. Clasper S, Vekemans S, Fiore M, et al. Inducible expression of the cellsurface heparan sulfate proteoglycan syndecan-2 (fibroglycan) on humanactivated macrophages can regulate fibroblast growth factor action.J Biol Chem. 1999;274:24113–24123.

15. Gulyas M, Hjerpe A. Proteoglycans and WT1 as markers for distin-guishing adenocarcinoma, epithelioid mesothelioma, and benign me-sothelium. J Pathol. 2004;199:479–487.

16. Lozzo RV. Matrix proteoglycans: from molecular design to cellularfunction. Annu Rev Biochem. 1998;67:609–652.

17. Heino J, Kahari VM, Mauviel A, Krusius T. Human recombinantinterleukin 1 regulates cellular mRNA levels of dermatan sulphateproteoglycan core protein. Biochem J. 1988;252:309–312.

18. Kahari VM, Hakkinen L, Westermarck J, Larjava H. Differential regu-lation of decorin and biglycan gene expression by dexamethasone andretinoic acid in cultured human skin fibroblasts. J Invest Dermatol.1995;104:503–508.

19. Kahari VM, Larjava H, Uitto J. Differential regulation of extracellularmatrix proteoglycan (PG) gene expression. Transforming growth fac-tor-�1 up-regulates biglycan (PG1), and versican (large fibroblast PG)but down regulates decorin (PGII) mRNA levels in human fibroblasts inculture. J Biol Chem. 1991;266:10608–10615.

20. Hilderbrand A, Romanis M, Rasmussen L, et al. Interaction of the smallinterstitial proteoglycans biglycan, decorin and fibromodulin with trans-forming growth factor beta. Biochem J. 1994;302:527–534.

21. Border WA, Noble NA, Yatmatmoto T, et al. Natural inhibitor of TGF-�protects against scarring in experimental kidney disease. Nature. 1992;360:361–364.

22. Giri SN, Hyde DM, Braun RK, Gaarde W, Harper JR, PierschbacherMD. Antifibrotic effect of decorin in a bleomycin hamster model of lungfibrosis. Biochem Pharmacol. 1997;54:1205–1217.

23. Fukushima K, Badlani N, Usas A, Riano F, Fu F, Huard J. The use of anantifibrosis agent to improve muscle recovery after laceration. Am JSports Med. 2001;29:394–402.

24. Iyer VR, Eisen MB, Ross DT, et al. The transcriptional program in theresponse of human fibroblasts to serum. Science. 1999;283:83–87.

25. Ruoslahti E, Yamaguchi Y. Proteoglycans as modulators of growthfactor activities. Cell. 1991;64:867–869.

26. Pratibha V, Gupta VD. Cutaneous wound healing: significance ofproteoglycans in scar formation. Curr Sci. 2000;78:1–5.

27. Worapamorn W, Haase HR, Li H, Bartold PM. Growth factors andcytokines modulate gene expression of cell-surface proteoglycans inhuman periodontal ligament cells. J Cell Physiol. 2001;186:448–456.

28. Lim IJ, Phan TT, Song C, Tan WT, Longaker MT. Investigation of theinfluence of keloid-derived keratinocytes on fibroblast growth and pro-liferation in vitro. Plast Reconstr Surg. 2001;107:797–808.

29. Lim IJ, Phan TT, Bay BH, et al. Fibroblasts cocultured with keloidkeratinocytes: normal fibroblasts secrete collagen in keloid-like manner.Am J Physiol Cell Physiol. 2002;283:C212–C222.

30. Fitzgerald ML, Wang Z, Park PW, Murphy G, Bernfield M. Shedding ofsyndecan-1 and -4 ectodomains is regulated by multiple signalingpathways and mediated by a TIMP-3 sensitive cell surface metallopro-teinase. J Cell Biol. 2000;148:811–824.

31. Subramanian SV, Fitzgerald ML, Bernfield M. Regulated shedding ofsyndecan-1 and -4 ectodomains by thrombin and growth factor receptoractivation. J Biol Chem. 1997;272:14713–14720.

32. Fears CY, Gladson CL, Woods A. Syndecan-2 are expressed in themicrovasculature of gliomas and regulate angiogenic processes in mi-crovascular endothelial cells. J Biol Chem. 2006;281:14533–14536.

33. Bodo M, Lilli C, Bellucci C, et al. Basic fibroblast growth factorautocrine loop controls human osteosarcoma phenotyping and differen-tiation. Mol Med. 2002;8:393–404.

34. Fujiwara M, Muragaki Y, Ooshima A. Keloid-derived fibroblasts showincreased secretion of factors involved in collagen turnover and depend onmatrix metalloproteinase for migration. Br J Dermatol. 2005;153:295.

35. Imai K, Hiramatsu A, Fukushima D, Pierschbacher MD, Okada Y.Degradation of decorin by matrix metalloproteinases: identification ofthe cleavage sites, kinetic analyses and transforming growth factor-�1release. Biochem J. 1997;322:809–814.

Mukhopadhyay et al. The Journal of TRAUMA® Injury, Infection, and Critical Care • Volume 68, Number 4, April 2010

© 2010 Lippincott Williams & Wilkins1008