The Prostate 59:252^259 (2004)

Purificationof theKeratan Sulfate ProteoglycanExpressedin Prostatic SecretoryCells andits

Identification as Lumican

John W. Holland,1,2* Katie L. Meehan,2 Sharon L. Redmond,2

and Hugh J.S. Dawkins3

1Institute forMolecular Bioscience,UniversityofQueensland,Brisbane, Australia2Urological ResearchCentre,Departmentof Surgery,UniversityofWestern Australia,

Queen Elizabeth IIMedical Centre,Nedlands, Australia3Departmentof Surgery,UniversityofWestern Australia, Royal PerthHospital, Perth, Australia

BACKGROUND. Secretory epithelial cells of human prostate contain a keratan sulfateproteoglycan (KSPG) associated with the prostatic secretory granules (PSGs). The pro-teoglycan has not been identified, but like the PSGs, it is lost in the early stages of malignanttransformation.METHODS. Anion exchange and affinity chromatography were used to purify KSPG fromhumanprostate tissue. Enzymatic deglycosylationwasused to removekeratan sulfate (KS). Thecore protein was isolated using 2D gel electrophoresis, digested in-gel with trypsin, andidentified by peptide mass fingerprinting (PMF).RESULTS. The purified proteoglycanwas detected as a broad smear onWestern blots with anapparent molecular weight of 65–95 kDa. The KS moiety was susceptible to digestion withkeratanase II andpeptideN-glycosidase F defining it as highly sulfated andN-linked to the coreprotein. The core protein was identified, following deglycosylation and PMF, as lumican andsubsequently confirmed by Western blotting using an anti-lumican antibody.CONCLUSIONS. The KSPG associated with PSGs in normal prostate epithelium is lumican.While the role of lumican in extracellular matrix is well established, its function in the prostatesecretoryprocess is not known. It’s potential to facilitate packagingof polyamines inPSGs, to actas a tumor suppressor and tomark the early stages ofmalignant transformationwarrant furtherinvestigation. Prostate 59: 252–259, 2004. # 2003 Wiley-Liss, Inc.

KEY WORDS: 2D electrophoresis; peptide mass fingerprinting; tumor marker

INTRODUCTION

Recent study has revealed the presence of a keratansulfate proteoglycan (KSPG) in the luminal epithelialcells of the normal human prostate [1]. Immuno-histochemical analysis suggests that this KSPG isassociated with the prostate secretory granules (PSGs)and hence, along with well-recognized secretoryproducts such as prostate specific antigen (PSA), issecreted in an apocrine manner by the prostate. PSGsare not present in malignant glands indicating that thenormal secretory pathway is not functional in prostatecancer [2]. Not unexpectedly, immunostaining shows

loss of the KSPG in these glands demonstrating itspotential as a marker of malignant transformation inthe prostate [1].

The first evidence for the presence of this proteogly-can came from sugar analysis of secretory products

*Correspondence to: Dr. John W. Holland, Institute for MolecularBioscience, University of Queensland, St. Lucia, Qld, 4072, Australia.E-mail: j.holland@imb.uq.edu.auReceived 6 June 2003; Accepted 28 October 2003DOI 10.1002/pros.20002Published online 22 December 2003 in Wiley InterScience(www.interscience.wiley.com).

� 2003 Wiley-Liss, Inc.

found in the prostatic lumen [3]. The presence of highconcentrations of N-acetylglucosamine and galactosesuggested the presence of glycosaminoglycans, morespecifically keratan sulfate (KS). Subsequently, immu-nostaining of paraffin embedded sections from radicalprostatectomy specimens using amonoclonal antibodyspecific for KS showed positive staining in the apicalportion of the epithelial cells lining the lumen [1].Solubilization of prostate tissue using guanidine HClfollowedbyWestern blottingwith the anti-KS antibodyconfirmed the presence of KSPG, which appeared as abroad smear with apparent molecular weight between70–90 kDa. From the size of the KSPG it appeared be amember of the small leucine rich repeat group ofproteoglycans [4], which typically have a core proteinof approximately 40 kDa.

In this report, we describe the purification of KSPGfrom human prostate and preliminary characterizationof its carbohydrate component. In addition, using 2Dgel electrophoresis and peptide mass fingerprinting(PMF),we identify the core protein ofKSPGas lumican.

MATERIALSANDMETHODS

Tissue Extraction

Prostate tissue, either freshly obtained followingopen prostatectomy or from frozen stocks, was homo-genized in 10 Vol. 4 M guanidine HCl/0.1% CHAPS/50 mM sodium acetate, pH 6.0 (GCA) containing com-plete protease inhibitors (Roche) and incubated over-night at 48C with shaking to solubilize proteoglycans.Insoluble material was collected by centrifugation at1,400g for 30 min at 48C and re-extracted with GCA asbefore. The soluble extracts were diluted with 4 Vol.ethanol and allowed to precipitate overnight at 48C.The precipitates were collected by centrifugation at1,400g for 30 min at 4�C and resuspended in 7 MUrea/0.1% CHAPS/50 mM sodium acetate, pH 6(UCA).

Ion ExchangeChromatography

Soluble extracts in UCA were clarified by centrifu-gation at 1,400g for 30 min at 48C and loaded onto acolumn of Sepharose Q Fast Flow (10� 2.6 cm) whichhad been equilibrated with UCA. The column waswashedwith 200mlUCAat a flowrate of 3ml/min andthen eluted with a step gradient consisting of 200 mleach of UCA containing 0.2, 0.5, and 1 M NaCl.Fractions of 10 ml were collected during the 0.2 and0.5 M steps. Aliquots of the starting material, the flowthrough, the wash and each of the fractions wereanalyzed by SDS–PAGE andWestern blotting to detectKSPG. Fractions corresponding to the peak of immuno-reactive material were pooled and concentrated on a

centrifugal filter fitted with a Biomax 10 membrane(Millipore). Concentrates were diluted with 50 mMsodium acetate, pH 6 and reconcentrated twice toremove urea.

Preparation of Aff|nity Resin

Anti-KS antibody (160 mg) was incubated with10 mM sodium periodate in 100 mM sodium acetate,pH 5.5, for 1 hr at room temperature. Excess periodatewas removed by diafiltration and the oxidized anti-body was incubated overnight at room temperaturewith 1 ml of adipic acid dihydrazide-agarose in a totalvolume of 2 ml of 100 mM sodium acetate, pH 5.5.Unbound antibody was removed by washing theaffinity resin (1 ml packed volume) sequentially with5 ml 50 mM sodium acetate pH 6, 2 ml UCA and 2 mlUCA containing 2 M urea and the affinity resin wasre-equilibrated with acetate buffer.

Aff|nityChromatography

Pooled and desalted fractions (3 ml) from the ionexchange columnwere incubatedwith the affinity resinfor 1 hr at room temperature with gentle mixing. Theaffinity resin was packed into a Poly-Prep column(Bio-Rad) and the unbound material allowed to flowthrough. The column was washed stepwise with 2 mleachof 50mMsodiumacetate pH6andUCAbefore theKSPG was eluted with UCA plus 2 M NaCl. Aliquotsfrom each step were analyzed by SDS–PAGE andWestern blotting and the purified KSPG was concen-trated and dialyzed as before.

SDS^PAGEandWestern Blotting

Samples (diluted with H2O when necessary) weremixed with an equal volume of 2� sample buffer(125 mM TrisHCl, pH 6.8/4% SDS/20% glycerol/200 mM dithiothreitol/0.002% bromophenol blue) andheated at 1008C for 5 min. These samples were loadedonto two 0.75 mm SDS–PAGE gels composed of a10% resolving gel and a 4% stacking gel. They wereelectrophoresed at 15mA/gel for approximately 1 hr oruntil the tracking dye reached the bottom of the gel.One gel was stained with colloidal Coomassie Blue(0.1% Coomassie Brilliant Blue G-250 in 17% ammo-nium sulfate, 3% phosphoric acid, and 34% methanol)to determine protein distribution and the other wastransferred to a nitrocellulose membrane at 200 mA for60 min in 25 mM Tris/192 mM glycine/0.02% SDS/20% methanol. The membrane was blocked by over-night incubation at 48C in 5% non-fat milk powder inTBS (150 mM NaCl/50 mM TrisHCl, pH 7.4). Immu-nodetection of the transferred proteins was performedusing anti-KS antibody (Seikagaku Corp., clone 5-D-4,

Prostatic Lumican 253

1:5,000) diluted in TBS containing 2.5% non-fat milkpowder and 0.1% Tween 20. After a 60 min incubationat room temperature, the membrane was washed3 times for 10 min in TBS containing 0.2% Tween andthen incubated for 60 min with secondary antibody(Silenus anti-mouse IgG-HRP,DAH, 1:5,000) diluted asabove. After three washes the bands were visualizedusing the ECLdetection system (AmershamPharmaciaBiotech.). For detection of lumican, membranes werestripped for 30 min at 508C in 100 mM mercaptoetha-nol, 2% SDS, 62.5 mM TrisHCl, pH 6.8 and reprobedusing a rabbit anti-lumican antibody (gift from Dr. P.J.Roughley) at 1/1,000 and anti-rabbit IgG-HRP con-jugate at 1/4,000. Other procedures were as describedabove for the anti-KS antibody.

Enzymatic Digests

KSPG (2 ml) was incubated at 378C for 60 min in atotal volume of 20 ml with 1mU of each of the followingenzymes using the conditions described by the manu-facturer in their unit definitions: keratanase, 50 mMTrisHCl, pH 7.4; keratanase II, 10 mM sodium acetate,pH 6.5; chondroitinase ABC, 100 mM TrisHCl, pH 8.0;PNGaseF, 100 mM TrisHCl, pH 7.8; O-glycosidase,50 mM sodium phosphate, pH 5.0; neuraminidase,50mM sodium phosphate, pH 5.0. Control incubationscontained no enzyme. Reactions were stopped by theaddition of 20 ml of 2� sample buffer followed byheating at 1008C for 5min. Digests were analyzed bySDS–PAGE and Western blotting as described above.

2DGel Electrophoresis

For 2D gel analysis KSPG (15 ml) was incubated at378C for 6 hr in a total volume of 30 ml of 10mM sodiumacetate, pH 6.5 containing 1 mU of keratanase II, 1 mUof neuraminidase and 1 U of PNGaseF. In controlincubations either the enzymes or the KSPG werereplaced with buffer. Reactions were stopped by theaddition of 100 ml of IEF buffer (8 M urea, 4% CHAPS,100 mM DTT, 0.5% carrier ampholytes, 40 mM TrispH 8.0, and 0.001% orange G). Each incubation mixwas then used to hydrate a 7 cm 3–10 linear IPG strip(Pharmacia) overnight at room temperature. 2D elec-trophoresis was carried out with minor modificationsto our previously described method [5]. Hydratedstrips were focussed in a Bio-Rad Protean IEF Cell for18,000 V-hr using a linear gradient to a maximum of8,000 V and a maximum current of 50 mA/strip. Thefocussed strips were equilibrated for 30min in 125mMTrisHCl, pH 6.8 containing 6 M urea, 2% SDS, 2.5%acrylamide, 20% glycerol, and 100 mM DTT, thenembedded on top of 1.5mm thick 12% SDS–PAGE gelsusing 0.5% agarose. The 2nd dimension gels wereelectrophoresed at 30 mA/gel for 75 min and then

stained overnight with colloidal Coomassie Blue.After destaining with 1% acetic acid the gels werescanned at 300 dpi using a standard flatbed scannerand the images compared using Phoretix 2D gelanalysis software (version 5.01, Phoretix International,Newcastle, UK).

PMF

Relevant protein spots were excised from the gel andsubmitted to the Australian Proteome Analysis Facilityfor PMF. Samples underwent an in gel 16 hr trypticdigest at 378C and the resulting peptideswere extractedfrom the gelwith a 50% (v/v) acetonitrile, 1% (v/v) TFAsolution. A 1 ml aliquot was spotted onto a sample platewith 1 ml of matrix (a-cyano-4-hydroxycinnamic acid,8 mg/ml in the same solvent) and allowed to air dry.Matrix assisted laser desorption ionization (MALDI)mass spectrometry was performed with a MicromassTofSpec 2E Time of Flight Mass Spectrometer. Anitrogen laser (337nm)wasused to irradiate the sample.The spectra were acquired in reflectron mode in themass range 600–3,500 Da. An internal calibration wasapplied using two trypsin auto-digestion peaks at842.51 and 2,211 Da. PMF data were compared to theSwiss-Prot and TrEMBL databases using PeptIdent(http://www.expasy.ch/tools/peptident. html).

RESULTS

Ion ExchangeChromatography

The elution profile of the solubilized prostate tissueextract from the Sepharose Q column is shown inFigure 1A. The bulk of the crude protein did not bind tothe column at pH6. Elutionwith theNaCl step gradientresulted in a major protein peak at 200 mM and minorpeaks at 500 mM and 1 M NaCl. Aliquots of eachfraction were analyzed by SDS–PAGE and Westernblotting using monoclonal antibody 5-D-4 directedagainst KS (Fig. 1B). The antibody detected a broadband with molecular weight ranging from 65 to 95 kDaconsistent with the heterogeneity in size normallyassociated with proteoglycans. Also a narrow band at55 kDawhich eluted in the 200mMwashwas observedin the starting material. The bulk of the immuno-reactive KS eluted in fractions 6–12 which correspondto the small protein peak observed at 500 mM NaCl.

Aff|nityChromatography

Pooled immuno-reactive fractions from the ionexchange column were purified further on an anti-KSaffinity column as described under ‘‘Materials andMethods.’’ Fractions from the affinity column wereanalyzed by SDS–PAGE (Fig. 2A) andWestern blotting(Fig. 2B).Thebulkof the startingprotein (Fig. 2A, lane2)

254 Hollandet al.

failed to bind to the affinity column and was located inthe flow through (Fig. 2A, lane 3) or thewashes (Fig. 2A,lanes 4 and 5). In contrast most of the KSPG in thestarting material (Fig. 2B, lane 2) bound to the columnand was recovered in the UCA plus 2 M NaCl eluate(Fig. 2B, lane 6) resulting in substantial purification.Despite the strong immunostaining in the affinitycolumn eluate, there is no corresponding band on theCoomassie stained gel (Fig. 2A, lane 6) which is notunexpected as sulfated proteoglycans do not stainreadily with routine protein stains. In contrast thebands corresponding to IgG heavy and light chainsare visible on both the Coomassie stained gel (Fig. 2A,lane 1) and the Western blot (Fig. 2B, lane 1). A faintsmear is also visible on theWestern blot between about45–55 kDa. Although not visible in the fractions elutedfrom the ion-exchange column, this smear was com-monly observed in the more concentrated samples andalso tended to increase in intensity upon storage at 48Csuggesting it might be the result of degradation.

Susceptibilityof KSPGtoDigestionbyGlycosidases

Purified KSPGwas subjected to digestion by a panelof glycosidases in order to examine the nature of the

carbohydrate component. Figure 3 shows that at equi-valent activities, only keratanase II was able to digestthe KSPG (lane 3). There was no change in size orintensity of the immuno-signalwithkeratanase (lane 2),chondroitinase ABC (lane 4), PNGaseF (lane 5), O-glycosidase (lane 6) or neuraminidase (lane 7). How-ever, when PNGaseF was used at a 1,000-fold higheractivity than keratanase II there was a substantialreduction in intensity of the KSPG signal (lane 8),suggesting that some, if not all, of the KS is N-linked tothe core protein.

2DGel Electrophoresis of KSPG

Identification of a core protein on SDS–PAGE gelsof digested KSPG was not possible because the

Fig. 1. Sepharose Q ion exchange chromatography of prostateextract.A: Elutionprofile of SepharoseQ column showingprotein(*) and NaCl (�) concentration. The unbound material and theregion where 10 ml fractions were collected are indicated by thebars.B: Immunodetection of keratan sulfate (KS) in the fractionsfrom A. SDS^PAGE and Western blotting were performed asdescribed in ‘‘Materials and Methods.’’ The starting material (L),the unboundmaterial (F/T), the washes and numbered10 ml frac-tions areindicated.

Fig. 2. Affinity purification of keratan sulfate proteoglycan(KSPG).A: Colloidal Coomassie Blue stained gel of fractions fromthe affinity column.Lane 1, monoclonal antibody 5-D-4; lane 2,pooled fractions from the Sepharose Q column; lane 3, unboundmaterial; lane 4, acetatewash; lane 5,UCAwash; lane 6,UCAþ2 MNaCl elution.B: Immunodetection of KS in fractions from theaffinitycolumn.Lanes1^ 8wereasdescribedforA.

Prostatic Lumican 255

commercially available glycosidases contained numer-ous contaminating protein bands that obscured theresults. To overcome this problem we used 2D gelelectrophoresis to resolve KSPG before and afterdigestion with glycosidases. Figure 4 shows theresulting 2D gels. In Figure 4A the large number ofproteins derived from the glycosidases alone areclearly visible. Figure 4B shows a number of proteinsare present in the KSPG sample. The arrowheadindicates a large diffuse spot with the same molecularweight range seen for KSPG in Western blots. Theacidic pI is consistent with the high degree of sulfationof KSPG. Figure 4C shows KSPG after digestion withthe same mix of glycosidases shown in Figure 4A. Thelarge diffuse spot corresponding to KSPG is no longervisible after digestion, but two new spots indicated byarrowheads are generated. These spots represent thecore protein(s) of KSPG.

PMF

The two core protein spots (as indicated in Fig. 4C)were excised from the 2D gels along with the broadspot corresponding to fully glycosylated KSPG(Fig. 4B) and subjected to PMF by MALDI-ToF massspectroscopy. Screening of the Swiss-Prot and TrEMBLdatabases using the peptide mass lists produced apositive result for human lumican. Table I shows thepeptide masses and matching sequences for the largerof the two core protein spots. Thirteen matchingpeptides were obtained covering 52% of the maturelumican sequence. Most of the remaining sequence isaccounted for by peptides that would be either toosmall (<800 Da) or too large (>3,200 Da) to be resolvedunder the conditions used. Nine of the peptides werealso observed in the fully glycosylated KSPG, butnotably peptides 3–6which contain 3 of the 4 proposedN-glycosylation sites were absent. Four of the peptidesalso were observed in the smaller of the core proteinspots suggesting this may be a degraded or truncatedform.

Conf|rmation of the Identityof KSPGas Lumican

To confirm the identity of KSPG as lumican, thepurified proteoglycan was digested with glycosidasesand the digests were separated by SDS–PAGE andanalyzed by Western blotting using both anti-KSand anti-lumican antibodies (Fig. 5). Lane 1 contained

Fig. 3. SensitivityofKSPGtoglycosidases.Lane1,purifiedKSPG;lanes 2^ 8, purifiedKSPGafter incubation at 378C for 60minwith1mUof the indicatedglycosidases.Lane 2, keratanase; lane 3, kera-tanase II; lane 4, chondroitinase ABC; lane 5, PNGaseF; lane 6,O-glycosidase; lane7,neuraminidase; lane8,PNGaseF (1,000�).

Fig. 4. 2D electrophoresis of KSPG after glycosidase digestion.A:Enzymesalone.B:KSPGalone.Thearrowheadindicates thespotcorresponding to themolecularweightandprobablepI ofKSPG.C:KSPGafterdigestion.Thearrowheadsindicate thenewspotsgener-atedbydigestion.

256 Hollandet al.

undigested KSPG, which was detected as a broad bandbetween 66–90 kDa by both the anti-KS (Fig. 5A) andanti-lumican (Fig. 5B) antibodies. After digestion withkeratanase II the KS signal was almost completelyabolished (Fig. 5A, lane 3) but the weak lumican signalwas still observed (Fig. 5B, lane 3). In contrast digestionwith PNGaseF only partially reduced the KS signal(Fig. 5A, lane 5) but resulted in a new lumican signal atapproximately 41 kDa (Fig. 5B, lane 5), consistent withthe size of the larger core protein seen with 2D PAGE(Fig. 4C). While digestion with neuraminidase had novisible effect (lane 7), the combination of keratanase IIand PNGaseF was sufficient to completely abolishthe signal from glycosylated KSPG leaving only thedeglycosylated lumican band at 41 kDa (lanes 8 and 9).

DISCUSSION

Keratan sulfate is a glycosaminoglycan comprised ofa repeating disaccharide unit, N-acetylglucosamine-galactose, with a variable degree of sulfation. It is nor-mally present as a proteoglycan with one or more KSchains bound to a core protein. The presence of KS inthe secretory epithelial cells of normal human prostateand its loss in cancer prompted us to identify thecore protein of this proteoglycan. Using a combinationof anion exchange chromatography at pH 6 to takeadvantage of the acidic nature of the proteoglycan dueto the presence of multiple sulfate groups, and affinitychromatography using the same anti-KS antibodyusedto detect KS in the previous study, we have purified aKS proteoglycan from human prostate tissue. To

TABLE I. PeptideMass Fingerprintingof KSPG

PeptideObserved

massMatchingmass Residues Sequence

1 1,024.38 1,024.59 60–69 SVPMVPPGIK2 1,225.29 1,225.58 75–84 NNQIDHIDEK3 2,611.96 2,612.34 85–106 AFENVTDLQWLILDHNLLENSK4 2,011.08 2,011.11 120–137 KLHINHNNLTESVGPLPK5 1,883.38 1,883.01 121–137 LHINHNNLTESVGPLPK6 2,195.28 2,195.17 152–170 LGSFEGLVNLTFIHLQHNR7 1,668.30 1,668.86 185–198 SLEYLDLSFNQIAR8 1,956.37 1,957.10 199–216 LPSGLPVSLLTLYLDNNK9 1,225.29 1,225.61 217–226 ISNIPDEYFK10 1,180.30 1,180.66 227–235 RFNALQYLR11 1,024.38 1,024.56 228–235 FNALQYLR12 2,535.98 2,536.26 268–288 NIPTVNENLENYYLEVNQLEK13 1,824.11 1,824.85 316–330 ISETSLPPDMYECLR

Peptidemasses from the larger of the two core protein spots thatwerematched to human lumican(SwissProt accession no. P51884) are listed with the corresponding amino acid sequences. N-linked glycosylation sites are indicated by bold underlined text.

Fig. 5. DetectionoflumicancoreproteinafterdeglycosylationofKSPG. KSPG was subjected to SDS^PAGE and Western blottingbeforeandafterdigestionwiththeindicatedglycosidases.A:Detec-tionwithanti-KSantibody.B:Detectionwithanti-lumicanantibody.Lane1, KSPG; lane 2, keratanase II control; lane 3, keratanase IIdigest; lane 4, PNGaseF control; lane 5, PNGaseF digest; lane 6,neuraminidasecontrol;lane7,neuraminidasedigest;lane8,kerata-nase II and PNGaseF digest; lane 9, keratanase II, PNGaseF andneuraminidasedigest.

Prostatic Lumican 257

release the protein core we examined a panel ofglycosidases of which only keratanase II and PNGaseFwere active, indicating that the KS was highly sulfatedand attached to the protein through an N-linkage toasparagine. 2D gel analysis revealed a diffuse spot,with an approximate pI of 3.4 which is consistent withheterogeneity of the KS chains and the high degree ofsulfation indicated by the glycosidase susceptibility.After enzymatic deglycosylation the diffuse spot waslost and two smallerwell-defined spotswere generatedwith apparent molecular weights of 41 and 25 kDa anda less acidic pI of 5.3, consistent with loss of the largeacidic KS group(s). Using PMF, the 41 kDa core proteinwas identified as human lumican which consists of320 amino acids with a theoretical molecular weight of37 kDa. It contains four N-glycosylation sites and thePMF data indicates at least three of these (N88, N127,and N160) are linked to KS as the peptides containingthese residues were not observed in the fully glycosy-lated form. The smaller species was also identified aslumican but the fingerprint contained no peptidesbeyond residue 198. In addition, this species was notobserved in anti-lumican immunoblots that used anantibody raised against a C-terminal peptide. Itappears to be a truncated form missing residues 199–338 which would be consistent with the decrease inmass of 16 kDa. This form may have arisen duringdeglycosylation as a result of a contaminating proteasein one of the glycosidase preparations. However, it isnot possible to rule it out as a naturally occurring formwhich, when glycosylated, is not resolved from thebroad smear of the full length protein onWestern blots.

The human lumican gene has been cloned [6,7] andis widely expressed in human tissues. It belongs to afamily of small leucine-rich proteoglycans (SLRP)normally associated with the extracellular matrix,where it plays a key role in collagen fibril assembly[4]. This is supported by studies of lumican knock-outmice which show increased skin fragility and cornealopacity [8].

Several studies have examined lumican expressionin cancer. In breast, lumican mRNA was expressed infibroblast-like cells in the stroma [9] and modestincreases in both mRNA and protein were observedin invasive carcinomas, but not in the cancer cellsthemselves [10]. In pancreas, expression of mRNA andprotein in islet and stromal cells in normal tissue aswell as in cancer cells and adjacent fibroblasts ofcancerous tissue have been reported [11]. A similarpattern of expression has been observed in cervicalcancer with mRNA and protein present in cancercells and fibroblasts but only weak stromal immunor-eactivity in normal tissue [12]. In contrast, lumicanmRNA expression was not detected in 11 of 15 cancercell lines tested [13].

Most studies of lumican in cancer to date haveaddressed only the presence of mRNA or immuno-reactive core protein. It is important to recognize thatapparent differences in lumican expression in cells ordistribution within tissues can result from changes incoreprotein levels, changes in thenumber and lengthofpolylactosamine chains and/or changes in the degreeof sulfation depending on the detection method used.From the results in Figure 5 it is obvious that the anti-lumican antibody used was much more effective atdetecting the de-glycosylated form of lumican than thefully glycosylated one. Thus, studies based on anti-bodies directed against the core proteinmay fail to givea complete picture of lumican distribution. Similarly,studies basedonantibodies directed againstKSwill notdetect non-glycosylated or poorly sulfated forms oflumican. It is apparent from studies of lumican in carti-lage that changes in both glycosylation and sulfation dooccur and may be regulated by cytokines and othergrowth factors [14]. Clearly, to unravel any role oflumican in cancer, it will be necessary to address notonly the expression of the protein but also the extent ofpost-translational glycosylation and sulfation.

While the lumican protein has not been demon-strated previously in prostate, two reports have shownexpression of lumicanmRNA[13,15].However, the celltypes responsible have not been identified. Previousstudies of KS expression in prostate showed strongimmunoreactivity in secretory epithelial cells withweak to no stromal staining. However, in adenocarci-noma, KS immunoreactivity was mostly absent [1]. Itwas not possible to determine whether loss of expres-sion of the proteoglycan per se or a loss of the highlysulfated carbohydrate epitope was responsible for theobserved differences. Now that the KS proteoglycanhas been identified as lumican, it should be possible toelucidatewhat changes are occurringduringmalignanttransformation. In a preliminary examination of onecase we observed co-localization of KS and lumicanprotein immunoreactivity in secretory epithelial cellsof normal tissue with no staining in cancer (result notshown). Intriguingly, the sub-cellular distribution ofthe two epitopes appeared to differ with KS staining inthe apical portion of the cells while lumican stainingwasmore diffuse throughout the cytoplasm. Clearly anextensive studyofmore cases of differinggradeswill benecessary before any conclusions can be drawn.

While the association of KS with prostatic secretorygranules has been clearly established [1], the role of theproteoglycan has yet to be established. In this regardthe recent demonstration of polyamine staining in thegranules [16] may be significant. Packaging of highconcentrations of polycationic amines in the granulescould be facilitated by the polyanionicKS chains.A rolefor lumican as a suppressor of transformation has also

258 Hollandet al.

been described [13]. Such a role would be consistentwith the loss of KS expression that accompaniesmalignant transformation of prostatic epithelial cells.

ACKNOWLEDGMENTS

We are thankful to Ronnie Cohen of Uropath PtyLtd. for provision of prostate tissue. The anti-lumicanwas a gift from Dr. P.J. Roughley, Genetics Unit,Shriners Hospital, 1529 Cedar Avenue, Montreal,Quebec,H3G1A6Canada (E-mail: proughley@shriners.mcgill.ca). This research has been facilitated by accessto the Australian Proteome Analysis Facility estab-lished under the Australian Government’s MajorNational Research Facilities Program.

REFERENCES

1. Cohen RJ, Holland JW, Redmond SL, McNeal JE, Dawkins HJS.Identification of the glycosaminoglycan keratan sulfate in theprostatic secretory cell. Prostate 2000;44:204–209.

2. Cohen RJ, McNeal JE, Edgar SG, Robertson T, Dawkins HJS.Characterization of cytoplasmic secretory granules (PSG), inprostatic epithelium and their transformation-induced loss indysplasia and adenocarcinoma.HumPathol 1998;29:1488–1494.

3. CohenRJ,McNeal JE, Redmond SL,MeehanK, ThomasR,WilceM, Dawkins HJS. Luminal contents of benign and malignantprostatic glands: Correspondence to altered secretory mechan-isms. Hum Pathol 2000;31:94–100.

4. Iozzo RV. The biology of the small leucine-rich proteoglycans:Functional network of interactive proteins. Crit Rev BiochemMol Biol 1997;32:141–174.

5. Meehan KL, Holland JW, Dawkins HJS. Proteomic analysis ofnormal and malignant prostate tissue to identify novel proteinslost in cancer. Prostate 2002;50:54–63.

6. Chakravarti S, Stallings RL, SundarRaj N, Cornuet PK, HassellJR. Primary structure of human lumican (keratan sulfate

proteoglycan) and localization of the gene (LUM) to chromo-some 12q21.3-q22. Genomics 1995;27:481–488.

7. Grover J, Chen X-N, Korenberg JR, Roughley PJ. The humanlumican gene: Organization, chromosomal location, and expres-sion in articular cartilage. J Biol Chem 1995;270:21942–21949.

8. Chakravarti S, Magnuson T, Lass JH, Jepson KJ, LaMantia C,Carroll H. Lumican regulates collagen fibril assembly: Skinfragility and corneal opacity in the absence of lumican. J Cell Biol1998;141:1277–1286.

9. Leygue E, Snell L, Dotzlaw H, Hole K, Hiller-Hitchcock T,Roughley PJ, Watson PH,Murphy LC. Expression of lumican inhuman breast carcinoma. Cancer Res 1998;58:1348–1352.

10. Leygue E, Snell L, Dotzlaw H, Troup S, Hiller-Hitchcock T,Murphy LC, Roughley PJ,Watson PH. Lumican and decorin aredifferentially expressed in human breast carcinoma. J Pathol2000;192:313–320.

11. LuYP, IshiwataT,AsanoG. Lumican expression in alpha cells ofislets in pancreas and pancreatic cancer cells. J Pathol 2002;196:324–330.

12. Naito Z, Ishiwata T, Kurban G, Teduka K, Kawamoto Y,Kawahara K, Sugisaki Y. Expression and accumulation oflumican protein in uterine cervical cancer cells at the peripheryof cancer nests. Int J Oncol 2002;20:943–948.

13. Yoshioka N, Inoue H, Nakanishi K, Oka K, Yutsudo M,Yamashita A, Hakura A, Nojima H. Isolation of transformationsuppressor genes by cDNA subtraction: Lumican suppressestransformation induced by v-src and v-K-ras. J Virol 2000;74:1008–1013.

14. Melching L, Roughley PJ. Modulation of keratan sulfatesynthesis on lumican by the action of cytokines on humanarticular chondrocytes. Matrix Biol 1999;18:381–390.

15. Luo J, Dunn T, Ewing C, Sauvageot J, Chen Y, Trent J, Isaacs W.Gene expression signature of benign prostatic hyperplasiarevealed by cDNA microarray analysis. Prostate 2002;51:189–200.

16. Cohen RJ, Fujiwara K, Holland JW, McNeal JE. Polyamines inprostatic epithelial cells and adenocarcinoma: The effects ofandrogen blockade. Prostate 2001;49:278–284.

Prostatic Lumican 259