extraction and purification of decorin from corneal stroma retain structure and biological activity

Extraction and purification of decorin from corneal stromaretain structure and biological activityq

Christopher T. Brown,a,b P. Lin,a Mary T. Walsh,a,c Donald Gantz,c

Matthew A. Nugent,a,b and V. Trinkaus-Randalla,b,*

a Department of Biochemistry, Boston University School of Medicine, Boston, MA 02118, USAb Department of Ophthalmology, Boston University School of Medicine, Boston, MA 02118, USA

c Department of Biophysics and Physiology, Boston University School of Medicine, Boston, MA 02118, USA

Received 21 September 2001, and in revised form 21 February 2002

Abstract

We developed a method to purify decorin core protein from tissue with the goal of preserving its native structure and biological

function. Currently, most procedures rely on the use of denaturing reagents potentially altering the biological activity. Decorin was

purified from corneal stromas without the use of detergents or chaotropic reagents. Proteoglycans isolated using anion exchange

chromatography on Q-Sepharose were treated with chondroitinase ABC. Decorin was isolated by a second Q-Sepharose chro-

matography with affinity chromatographies on heparin–Sepharose and concanavalin A–Sepharose. SDS–PAGE revealed a 98.4%

pure 44 kDa protein identified as decorin with a yield of 35mg per 100 bovine corneas. Identification was confirmed by NanoESI

and MALDI qTOF. The novel inclusion of 20% propylene glycol in extraction and column buffers resulted in recoveries of pro-

teoglycans comparable with those observed with detergents and urea. Purified decorin did alter the rate of fibrillogenesis of type I

collagen and inhibited the lateral fusion of collagen fibrils. It also bound to [125I]TGF-b1 with an apparent Kd of 40 nM. Circulardichroism spectroscopy of decorin displayed the spectra of a-helices and b-pleated sheets consistent with those obtained fromrecombinant decorin. Urea-induced unfolding was cooperative and reversible while thermal denaturation caused irreversible

unfolding. Native decorin can be purified from tissue in quantity and quality for biophysical, biochemical, and biological

assays. � 2002 Elsevier Science (USA). All rights reserved.

Keywords: Purification; Decorin; Fibrillogenesis; TGF-b

The extracellular matrix (ECM)1 has a striking in-fluence on cell behavior, influencing shape, polarity,movement, metabolism, growth, and development. Overthe past two decades, it has been shown that the ECM isa dynamic entity, which is capable of modulating cellgrowth and differentiation as well as influencing themetabolism of a number of different ECM components[1,2]. Many of these effects are regulated either directlythrough interaction with receptors on the cell surface orindirectly by modulating growth factor activity in thelocal microenvironment, ultimately leading to changesin gene expression [1–3].All mammalian extracellular matrices possess proteo-

glycans. Presently, extracellular proteoglycans can bedivided into two groups, the small leucine-rich proteo-glycans (SLRPs) and the large modular proteoglycans[4]. The SLRPs are typically compact proteins with

Protein Expression and Purification 25 (2002) 389–399

www.academicpress.com

qThis work was supported by NEI Grant EY11000-4 (to M.N.) and

by departmental grants from the Massachusetts Lions Eye Research

Fund, Research to Prevent Blindness, Inc. and from the New England

Corneal Transplant Fund.*Corresponding author. Fax: 1-617-638-5337.

E-mail address: [email protected] (V. Trinkaus-

Randall).1 Abbreviations used: SDS–PAGE, sodium dodecyl sulfate–poly-

acrylamide gel electrophoresis; TGF-b, transforming growth factor b;ECM, extracellular matrix; SLRP, small leucine-rich proteoglycan;

EGF, epidermal growth factor; MMP, metalloproteinase; DMB,

dimethylmethylene blue; NaCl, sodium chloride; EDTA, ethylenedi-

aminetetraacetic acid; UV, ultraviolet; PMSF, polymethylsulfonyl

fluoride; GAG, glycosaminoglycan; BSA, bovine serum albumin;

MgCl2, magnesium chloride; MnCl2, manganese chloride; CSPG,

chondroitin sulfate proteoglycan; KSPG, keratan sulfate proteoglycan;

TEM, transmission electron microscopy.

1046-5928/02/$ - see front matter � 2002 Elsevier Science (USA). All rights reserved.PII: S1046 -5928 (02 )00025-6

alternating hydrophilic and hydrophobic amino acidresidues that have multiple amino acid repeats withconserved leucine residues. The SLRPs include at leasteight structurally related but genetically distinct mem-bers: decorin, biglycan, fibromodulin, lumican, kerato-can, osteoadherin, osteoglycin, and epiphycan [5].Decorin represents nearly 40% of the total proteo-

glycan in the cornea and has been proposed to playimportant roles in fibrillogenesis, tissue repair, and theregulation of transforming growth factor-b (TGF-b)[6,7]. It has a single chondroitin/dermatan sulfate chainnear the N-terminus and is subject to N-linked gly-cosylation. Decorin can bind to collagens, TGF-b, epi-dermal growth factor (EGF) receptors, and fibronectin[4,5]. In decorin-null mice, the collagen network in theskin is loosely packed with irregular contours and lateralfusion of collagen fibrils is evident [8]. While the corneamaintains its integrity in uncompromised conditions, itis not known what happens in response to injury. De-corin’s function may be related to its interactions withother proteins, such as TGF-b, that play crucial rolesduring wound repair [9]. Thus, binding of decorin toTGF-b has been suggested to mediate the biologicalactivity of TGF-b. The growth of Chinese hamsterovary cells is suppressed by ectopic expression of deco-rin through disruption of the TGF-b autocrine loop [10].These in vitro observations have been extended to ani-mal models of glomerulonephritis where blocking anti-TGF-b antibodies or addition of decorin preventsfibrosis [11]. Although this study implicates decorin inblocking the action of TGF-b, there is also evidencesuggesting that decorin can enhance TGF-b binding toits receptors [12]. Furthermore, recent studies havedemonstrated that distinct domains of the decorin coreprotein bind TGF-b [12] and that members of the matrixmetalloproteinase (MMP) family of enzymes can gen-erate specific fragments of decorin [14]. These resultssuggest that the ability of decorin to either enhance orinhibit TGF-b action may depend on its structure andpericellular location. In addition, a decorin ‘‘receptor’’has been identified that binds and internalizes decorin,suggesting additional complexities [15].Decorin has been shown to exert negative control on

cell proliferation by enhancing the expression of cyclin-dependent kinase inhibitor, p21 mRNA, and protein[16]. Control is thought to be mediated by the binding ofdecorin to the EGF receptor, which would then triggerthe onset of an intracellular signalling pathway [17]. Itwas recently demonstrated that ectopic expression ofdecorin can block the growth of tumor xenografts inboth autocrine and paracrine manner [18]. These studiessuggest that decorin is a regulator of cell growth andplays a crucial function in maintaining homeostasis.We developed a method to purify decorin core pro-

tein from tissue that preserves native structure andfunction. Unlike most procedures, the method described

here does not rely on the use of strong denaturing re-agents that may compromise biological activity. Theability of decorin to influence the rate of collagen fib-rillogenesis and bind to TGF-b demonstrates the bio-chemical integrity of the decorin core protein.Furthermore, the stability and native conformation ofdecorin were studied, demonstrating that urea-inducedfolding was reversible while heat induced an irreversibleunfolding at 42 �C.

Materials and methods

Materials. Chondroitinase ABC (protease-free) anddermatan sulfate (chondroitin sulfate B superspecialgrade) were purchased from Seikagaku America(Ijamsville, MD). Chondroitin sulfate A and dimeth-ylmethylene blue were purchased from Sigma–Aldrich(St. Louis, MO). Q-Sepharose, heparin–Sepharose, andCon A–Sepharose were from Pharmacia Biotech (Upp-sala, Sweden). BCA protein assay reagent and Sulfo-NHS biotin were obtained from Pierce (Rockford, IL).All common laboratory reagents were obtained fromAmerican Bioanalytical (Natick, MA). TGF-b1 wasobtained from R&D Systems (Minneapolis, MN). TheMini-Pellicon Biomax 30 tangential flow filter and theCentricon-plus 20 ultrafiltration device were obtainedfrom Millipore (Bedford, MA). Bovine eyes were pur-chased from Pel Freeze (Rogers, AR). The polyclonalsheep antiserum directed against rabbit corneal decorinwas a gift from Dr. Charles Cintron (Schepens EyeResearch Institute, Boston, MA).Purification of decorin. Corneas were excised from

fresh bovine eyes. The corneas were frozen, pulverized inthe presence of liquid nitrogen with a Waring blender,and homogenized using a Polytron (Brinkman Inst.,Westbury, NY). We found that the novel inclusion of20% propylene glycol in the Q1 buffer (50mM sodiumacetate, 300mMNaCl, 10mMEDTA, and 1mM PMSF,pH 6.0) allowed for the extraction of proteoglycanswithout the use of detergents or chaotropic reagents, yetretaining the biological activity. Proteoglycans were ex-tracted for 18 h at 4 �C (in excess of 100 v/w) in Q1 buffer.The extracts were centrifuged for 30min at 40,000g andthe glycosaminoglycans (GAGs) in the supernatant werequantitated using the dimethylmethylene blue assay(DMB) that was developed by Farndale et al. [19]. Thepellet was re-extracted until the quantity of GAG in thesupernatants was 10-fold lower than the amount of GAGin the first extract. Supernatants were pooled and filteredthrough a Polycap 75 HD 1.0 lm filter (Whatman,Clifton, NJ). The conductivity and UV absorbance at280 nm ðA280Þ were monitored during all chromatogra-phies and 1.4ml aliquots of all pooled column fractionswere kept for analysis. Extracts were loaded onto a200ml Q-Sepharose column equilibrated with Q1 buffer

390 C.T. Brown et al. / Protein Expression and Purification 25 (2002) 389–399

and the column was washed with Q1 buffer until the A280decreased to baseline. The column was washed with fivecolumn volumes of buffer containing 50mM sodiumacetate and 300mM NaCl, pH 6.0. Proteoglycans wereeluted with 50mM sodium acetate and 1.5M NaCl, pH6.0, and fractions were collected. Fractions were ana-lyzed using the DMB assay. GAG containing fractionswere pooled and exhaustively dialyzed against 10mMTris, pH 7.8.Isolated proteoglycans were digested with chondro-

itinase ABC (4.0mU/mg total GAG) for 6 h at 37 �C toremove the chondroitin sulfate chains from decorin. Thedigest was dialyzed against Q2 buffer (50mM sodiumacetate, 250mM NaCl, and 20% propylene glycol, pH6.0) and loaded onto a 200ml Q-Sepharose column. Thecolumn was washed with Q2 buffer until the A280 de-creased to baseline and the decorin core protein wasrecovered in the flowthrough fraction. Proteoglycansresistant to enzymatic deglycanation were recoveredwith 1.5M NaCl, pH 6.0. The flowthrough fraction wasconcentrated (10-fold) and the buffer was exchanged toHS Buffer (50mM sodium acetate, 125mM NaCl, and20% propylene glycol, pH 6.0) by tangential flow filt-ration using a Mini-Pellicon Biomax 30 (Millipore,Bedford, MA) according to manufacturer’s instructions.The decorin preparation was loaded onto a 5.0ml hep-arin–Sepharose column equilibrated with HS buffer. Thecolumn was washed with five column volumes of HSbuffer and decorin was eluted with 50mM sodium ace-tate and 300mM NaCl, pH 6.0. Contaminants wererecovered with 1.5M NaCl, pH 6.0. The ionic compo-sition and pH of the heparin–Sepharose eluent wereadjusted to those of Con A buffer (20mM Tris and500mM NaCl, pH 7.2) by adding 5.0M NaCl and 1.0MTris base while monitoring pH and conductivity.The decorin preparation was loaded onto a Con A–Sepharose column equilibrated with Con A buffer. Thecolumn was washed with five column volumes of equ-ilibration buffer and eluted with 10% methyl a-D-mannopyranoside. The eluent was dialyzed to 10mMHEPES, 150mM NaCl, and 10% propylene glycol, pH7.2, and concentrated to approximately 2mg/ml totalprotein using a Centricon-plus 20. The decorin was fil-tered through a 0.22 lm PES syringe filter and stored at)20 �C.Quantitation of total GAG and total protein. Total

sulfated GAGs were quantitated using the DMB assay[19]. Briefly, samples were mixed with DMB reagent(1:25) and absorbance (525 nm) was read immediately.Concentrations were determined from standard curvesof highly purified dermatan sulfate. Total protein wasdetermined using the BCA assay according to manu-facturer’s instructions.Assay for chondroitinase ABC. Samples and standards

were subjected to seven replicate serial 1:2 dilutions (1:4to 1:256) in reaction buffer (25mM Tris–acetate, 100 lg/

ml BSA, 0.5mM MgCl2, and 0.5 MnCl2, pH 8.0).Diluted samples were combined with an equal volumeof 0.3mg/ml purified chondroitin sulfate A dissolved inreaction buffer and incubated for 3 h at 37 �C with pu-rified chondroitin sulfate A for 3 h at 37 �C. The reactionwas stopped with the addition of two volumes of 60mMglycine and 5mM EDTA, pH 3.0. Samples and stan-dards were analyzed using the DMB assay and con-centrations of chondroitinase ABC in the sample wereread from standard curves of known concentrations ofchondroitinase ABC.SDS–PAGE and Western blot analyses. Aliquots of

pooled column fractions were analyzed using SDS–PAGE and/or Western blotting. Aliquots of columnfractions were run on 5–15% gradient SDS–PAGE gelsunder reducing conditions and loading was normalizedto the total protein present in fractions [20]. Gels wereeither stained with silver or electrophoretically trans-ferred to polyvinylidene difluoride membrane using asemi-dry transfer apparatus in 25mM Tris, 192mMglycine, and 20% methanol. The membranes wereblocked in 5% BSA in TBS-T buffer (10mM Tris,100mM NaCl, and 0.1% Tween 20, pH 7.2) and incu-bated with polyclonal sheep antiserum directed againstrabbit corneal decorin (1:9000) in 5% BSA in TBS-T atroom temperature for 1 h. Blots were washed with 1%BSA in TBS-T and incubated with appropriate sec-ondary antibodies coupled to horseradish peroxidase(1:3000) for 1 h at room temperature. Proteins were vis-ualized using chemiluminescence.Decorin core protein was quantified using selective

digestion with chondroitinase ABC in conjunction withSDS–PAGE. A minimum of three serial dilutions ofaliquots of pooled column fractions were run on 5–12%gradient SDS–PAGE gels under reducing conditionsand compared to purified decorin. Gels were stainedwith 0.25% Coomassie blue R25 in 50% methanol and10% acetic acid and destained with 50% methanol and10% acetic acid. Decorin core proteins were identified bythe electrophoretic shift that was observed after chond-rotinase ABC treatment. Gels were analyzed by den-sitometry and concentrations were determined fromstandard curves of purified decorin (0.25–1.0 lg/lane).Collagen fibrillogenesis–electron microscopy. The

structure of collagen fibrils during fibrillogenesis wasmonitored using the negative staining drop technique[21]. Decorin (final concentration 100 lg/ml) was addedto a solution of acid solubilized rat tail tendon collagen(300 lg/ml in 100mM Tris and 150mM NaCl, pH 7.8)and incubated at 37 �C for 0, 9, and 30min. The collagenand decorin were prepared in an identical manner asthey were for the plate reader assay described below.Collagen alone in the same buffer served as control ateach time point. Aliquots of 5 ll were removed at eachtime point and added to 25 ll of 100mM Tris buffer, pH7.8, to reduce the salt concentration, prior to staining.

C.T. Brown et al. / Protein Expression and Purification 25 (2002) 389–399 391

A 4 ll aliquot of each diluted sample was placed for 30 son a carbon, formvar-coated 300 mesh grid (ElectronMicroscopy Sciences, Fort Washington, PA) that wasmade freshly hydrophilic by glow discharging (BalzersUnion CTA 010; Bal-Tec Products, Middlebury, CT).The grid was blotted, immediately stained with 1% so-dium phosphotungstate, pH 7.8, blotted, and air dried.Grids were viewed in a CM12 transmission electronmicroscope (Philips Electron Optics, Eindhoven, TheNetherlands). Images were recorded on S0163 film(Eastman Kodak, Rochester, NY), developed in D19,scanned on an EverSmart Supreme Scanner (CreoScitexCorp., Vancouver, BC, Canada), and cropped andprocessed in Photoshop 6.0 (Adobe Systems, San Jose,CA).Absorbance assay. Acid solubilized rat tail tendon

collagen (0.6mg/ml dissolved in 2mM HCl) was com-bined with an equal volume of 100mM Tris and 300mMNaCl, pH 7.8, containing various concentrations of de-corin in a 96-well plate. The plate was sealed with trans-parent adhesive cellophane and the temperature wasincreased to 37 �C. Fibrillogenesis was monitored in anOptimax 96-well plate reader (Molecular Devices, Sun-nyvale, CA) and the absorbance at 340 nm was recordedevery 3min for 2 h. A first-order kinetic model wasapplied to determine the apparent half times for fibrillo-genesis. Data were fit to the following equation:lnð½A340�t=½A430�maxÞ ¼ kt, where k is the observedfirst-order rate constant of fibrillogenesis and t1=2 ¼ 0:693

k .Decorin–TGF-b binding. Purified corneal decorin was

biotinylated with Sulfo-NHS biotin according to man-ufacturer’s instructions (Pierce). Biotinylated decorinwas immobilized on neutravidin–agarose beads (100 lgdecorin/ml bead). Twenty microliters of decorin beadswas incubated for 1 h at 4 �C with the indicated con-centrations of [125I]TGF-b1 dissolved in PBS containing20mg/ml BSA. Bound TGF-b1 was separated from freepeptide using micro-biospin columns (BioRad). Afterwashing, bound and free TGF-b1 were determined usinga c-counter. Binding data were analyzed by non-linearleast-squares curve fitting (Kaliedagraph) and Scatchardplot. Data points were the average of triplicatedeterminations.Circular dichroism spectroscopy. Far-UV CD spectra

of purified corneal decorin were recorded on an AVIV62DS CD spectropolarimeter using a 0.5mm path-length quartz cell at 25 �C in 10mM sodium phosphatebuffer, pH 7.2 (native and renatured decorin) or inphosphate buffer containing 8.0M urea (denatured de-corin). The spectra were the average of 10 scans andwere corrected for protein concentration and buffercomposition. The CD signal (mdeg) was converted tomolar ellipticity (h) by the following equation:

½h�c ðdeg cm2=dmolÞ ¼ h �MRW10� l� c

; ð1Þ

where h is the measured ellipticity, MRW is the meanresidue weight of the amino acids in decorin, l is thepathlength of the cuvette in cm, and c is the concen-tration of decorin in g/ml.Thermal unfolding studies were performed utilizing a

Peltier thermoelectric temperature controller in 0.5mmcuvettes at 217 (b-sheet) and 222 nm (a-helix). Individ-ual melts were performed by monitoring 217 and 222 nm(N ¼ 3). Ellipticity at each wavelength was monitoredfrom 5 to 95 �C at 0.2� increments for 60 s per temper-ature increment.

Results and discussion

Decorin appears to be distributed in all the connec-tive tissues and is a major component of the cornealstroma comprising more than 99% of CSPGs and ap-proximately 1% of the total corneal protein, making thetissue an ideal source [7,25–27].Purification. Standard procedures for isolation of

proteoglycans from tissue depend upon strong de-naturing reagents (i.e., chaotropic salts and detergents) tosolubilize ECM components. These treatments disruptthe secondary structure of the core proteins and mayirreversibly compromise the biological activity of themolecule. For many proteoglycans, the biological ac-tivity of the GAG chain can be studied, even when thecore protein is not in its native state. However, the coreprotein of decorin appears to play significant roles, in-dependent of its CS chains [4–6]. One example of thiscomes from previous proliferation studies we have per-formed, where decorin isolated under native conditionsdoes not promote proliferation, while decorin isolatedunder denaturing conditions can induce proliferation(data not shown). To circumvent the need for denatur-ing conditions, several groups have expressed and pu-rified recombinant decorin using mammalian culturesystems [22–24]. However, prior to this study, largeamounts of native core were not available for study.This new technology will enable investigators to use thenative decorin and/or compare its biological functionsand secondary structures with recombinant forms. Thislack of suitable quantities of intact decorin from tissuemotivated us to develop a purification method to isolatedecorin under native conditions and perform analyses toevaluate its biological structure and function.Exhaustive extraction of 100 pulverized bovine cor-

neal stromas (99.7 g wet) yielded 225mg total GAG,2.45 g total protein, and 110.8mg total decorin (Table1). This extract consisted of a heterogeneous mixture ofproteins and proteoglycans (CSPGs and KSPGs) (Fig.1B, lane 1). Anion-exchange chromatography has tra-ditionally been used to isolate the negatively chargedproteoglycans [6,7,25–29]. Q-Sepharose was used ini-tially to capture corneal proteoglycans from the crude


Fig. 1. Q-Sepharose chromatographies of corneal proteoglycans. (A) Approximately 3.5 liters of corneal extract was applied to a 5� 11 cm Q-Sepharose column at 110 cm/h in the presence of 0.3M NaCl. After washing, corneal proteoglycans were eluted in a single step with 1.5M NaCl. (B)

SDS–PAGE analysis of crude extract (lane 1), the flowthrough fraction (lane 2), and 1.5M NaCl elution (lane 3) from the first Q-Sepharose

chromatography. (C) The product of the first Q-Sepharose chromatography was treated with chondroitinase ABC and loaded onto the same Q-

Sepharose column at 60 cm/h in the presence of 0.25M NaCl. Decorin was recovered in the flowthrough fraction. The molecules that bound to the

column were recovered with 1.5M NaCl. (D) SDS–PAGE analysis of chondroitinase ABC treated corneal proteoglycans (lane 1), the flowthrough

fraction (lane 2), and 1.5M NaCl elution (lane 3) from the second Q-Sepharose chromatography. All gels were stained with silver and loading was

normalized to 0.5 ng total protein per lane.

Table 1

Purification of decorin from bovine corneal stromasa

Step Volume (ml) Total GAG (mg) Total protein (mg) Decorin (mg) Recovery (%) Purification (fold)

Corneal extract 3490 224:6� 3:9 2460� 28 110:8� 10:2 100.0 1.0

First Q-Sepharose 336 221:3� 1:7 335� 13 86:6� 10:7 78.2 5.7

Chondroitinase ABC 250 160� 1:2 337� 10 80:3� 4:4 72.5 5.3

Second Q-Sepharose 760 NDb 90� 5 82:3� 7:4 74.3 20.4

Ultrafiltration 44 ND 44� 2 37:2� 0:6 33.5 18.9

Heparin–Sepharose 45 ND 45� 4 37:2� 0:4 33.5 18.3

Con A–Sepharose 46 ND 33� 2 33:4� 1:4 30.2 22.2

aCorneal decorin was extracted from 100 bovine corneas and purified as described under ‘‘Materials and method.’’ Aliquots of pooled fractions

were collected for analysis during all phases of the purification. Total GAG was determined using the DMB assay. Total protein was determined

using the BCA assay. Decorin was determined by chondroitinase ABC digestion in conjunction with SDS–PAGE and densitometry. Measurements

presented are means of at least three data points� standard error.bND, not detected.


extract. Since we have previously shown that cornealproteoglycans elute from Q-Sepharose between 0.3 and1.2M NaCl with a linear salt gradient, the column wasloaded and washed with 0.3M NaCl, and eluted with1.5M NaCl (Fig. 1A) [29]. The flowthrough fractioncontained approximately 85% of the proteins loadedand did not contain detectable amounts of GAG. Elu-tion of the proteoglycans with 1.5M NaCl resultedin 98% recovery of GAGs and 95% recovery of thedecorin.The first Q-Sepharose chromatography yielded a

mixture that is primarily composed of corneal KSPGsand CSPGs, which migrated as a smear betweenMr¼ 97,000 and 160,000 on SDS–PAGE gels (Fig. 1B,lane 3). Chondroitin sulfate side chains were selectivelyremoved from decorin with chondroitinase ABC. Sincedecorin comprises more than 99% of corneal CSPGs,digestion resulted in the release of a single 45 kDa pro-tein, consistent with that of deglycanated decorin (Fig.1D, lane 1). The decorin core protein was isolated by asecond Q-Sepharose chromatography using a reducedNaCl concentration (0.25M) during loading to optimizepurity. Without the highly charged CS chain, the deco-rin core protein flows through the column (Figs. 1C andD, lane 2). The remaining molecules, consisting pri-marily of KSPGs, bound to the column and were elutedwith 1.5M NaCl (Fig. 1D, lane 3). Decorin core proteinpresent in the flowthrough fraction was identified byWestern blot analysis (Fig. 1E). Collectively, both Q-Sepharose chromatographies resulted in 74.3% recoveryand a 20.4-fold purification of the corneal decorin coreprotein.The products from the second Q-Sepharose chroma-

tography consisted primarily of decorin core, minorprotein contaminants, and chondroitinase ABC. Theflowthrough fraction from the second Q-Sepharosechromatography was concentrated by ultrafiltration us-ing a tangential flow system. This procedure resulted in asubstantial loss of total protein and decorin and visibleprecipitates were observed after the process. Affinitychromatography on heparin–Sepharose was chosen asthe third step because of the reported interactions betweenchondroitinase ABC and heparin [30]. When the con-centrated material was loaded onto a heparin–Sepharosecolumn in the presence of 0.125M NaCl, only traceamounts of protein flowed through the column. Afterwashing, decorin was eluted with 0.3M NaCl (Fig. 2A).The small peak that eluted with 1.5M NaCl contained50% of the chondroitinase ABC that was loaded onto thecolumn (Fig. 2B). As an added benefit, a significantnumber of protein contaminants remain bound to thecolumn that eluted with 1.5M NaCl (Fig. 2C, lane 3).Affinity chromatography on Con A–Sepharose was

used as a final step to eliminate minor contaminants[4–6]. In addition to a single CS chain, decorin alsopossesses one or two N-linked oligosaccharides, which

Fig. 2. Affinity chromatographies of corneal decorin. (A) The con-

centrated flowthrough fraction from the second Q-Sepharose chro-

matography was loaded onto a 1:6� 2:5 cm heparin–Sepharose

column at 30 cm/h in the presence of 0.125M NaCl. Negligible

quantities of protein flowed through the column. Decorin eluted as a

major peak with 0.3M NaCl. A small peak eluted with 1.5M NaCl.

(B) Aliquots of pooled column fraction from the heparin–Sepharose

chromatography were analyzed for chondroitinase ABC activity as

described in ‘‘Materials and methods.’’ (C) Aliquots of pooled column

fractions from the heparin–Sepharose and Con A chromatographies

were analyzed by SDS–PAGE and silver staining. (D) The 0.3M NaCl

fraction from the heparin–Sepharose chromatography was applied to a

1:0� 11:5 cm Con A–Sepharose column at 45 cm/h. Decorin waseluted with 500mM a-methyl-mannopyranoside.


contain mannosyl and glucosyl residues. When theproduct of the HS chromatography was loaded onto theCon A column, negligible amounts of protein werepresent in the flowthrough fraction (Fig. 2C). Decorineluted as a single peak with the mannose analog methyla-D-mannopyrannoside (Fig. 2B, lane 5, Fig. 2D). Thefinal product in Figs. 2B and D has been shown to bedecorin by mass spectrometry using the ZipTip C18technique. This resulting peak was extracted and ana-lyzed by NanoESI on a Quattro II and by MALDIqTOF on a AB Qstar (Mass Spectrometry Resource,Boston University School of Medicine).To enhance reproducibility, all chromatographic

steps utilized single concentration elutions and did notemploy gradients. It was observed that the inclusion of20% propylene glycol in the extraction buffer, columnloading buffers, and column washing buffers resulted inrecoveries of proteoglycans comparable with those ob-served with detergents and urea (data not shown). Theuse of propylene glycol is novel to extraction of prote-oglycans but has been employed in protein storage orchromatography of blood cells [31].The purification yielded 35mg decorin core protein

per 100 corneas. Comparable yields have been reportedfor recombinant decorin using various expression sys-tems. In a recent study, large quantities of decorin wereexpressed in E. coli as a maltose-binding fusion proteinthat formed insoluble aggregates in bacterial inclusionbodies requiring the use of guanidine–HCl [22]. Gu et al.[23] reported the purification of non-denatured decorin(1.5mg/3� 108 cells) using a baculovirus system. How-ever, this protein lacked mammalian post-translationalmodifications. Ramamurthy et al. [24] reported the ex-pression and purification of native recombinant decorin(3.3mg /109 cells) from the supernatant of HT-1080cells. This system produced a form of decorin thatpossessed both chondroitin sulfate chains and N-linkedoligosaccharides, yet requiring a large-scale cell culture.Our process yielded a comparable quantity of decorincore from a small amount of tissue and is potentiallyscalable to larger quantities.Role of decorin on collagen fibrillogenesis. Collagen

fibrillogenesis assays were used to evaluate the biologicalintegrity of purified corneal decorin in vitro. Birk andLande [32] demonstrated that acid solubilized collagenspontaneously formed fibrils at neutral pH and 37 �C.The addition of decorin isolated under denaturing con-ditions resulted in reduced turbidity, suggesting achange in the fibrillar ultrastructure [32–34]. Decreasedfibrillogenesis was first reported to be caused by a de-crease in the rate of fibril formation and, more recently,by a decrease in the initial fibril formation [32–34].Differences in rate were reported to depend on tissuesource of the proteoglycan extract [32,35]. The antifib-rillogenic activity was localized to the core protein andwas not dependent upon the chondroitin sulfate chain

[35,36]. Recently, Danielson et al. [8] demonstrated in-teresting findings on decorin-null mice where the colla-gen fibrils possessed lateral fusion.To evaluate the role of decorin isolated under native

conditions on fibrillogenesis, we used two independentmethods of analysis. In the first, we used a modificationof an in vitro collagen fibrillogenesis assay developed byBirk and collaborators [32,33]. Rat tail tendon collagen(300 lg/ml) was incubated with 3–100 lg/ml purifiedcorneal decorin and the rate and extent of fibrillogenesiswere monitored by measuring turbidity (absorbance at340 nm) over a two hour period (Fig. 3A). The presence

Fig. 3. Collagen fibrillogenesis absorbance assay. (A) Purified corneal

decorin (0–100lg/ml) was incubated in the presence of 300lg/ml rattail tendon collagen and fibrillogenesis was followed over a 2 h period.

(B) Corneal decorin reduces the rate of collagen fibrillogenesis in a

concentration-dependent manner. (C) Collagen gels formed in the

presence of corneal decorin exhibited reduced turbidity.


of decorin resulted in a concentration-dependent de-crease in the rate of absorbance (40% reduction with100 lg/ml corneal decorin) (Fig. 3B). Furthermore, ad-dition of corneal decorin resulted in a concentration-dependent reduction in maximal gel turbidity, indicatingalterations in the fibril ultrastructure consistent withprevious reports (Fig. 3C) [32–36].We then addressed whether there were alterations in

the fibril ultrastructure using electron microscopy ofsolutions of decorin and collagen or collagen alone. Theprotein solutions were incubated at 37 �C for three timeperiods, which were chosen to examine three points onthe curve in Fig. 3A (prior to fibrillogenesis, initiation,and a point where the curve had plateaued in the ab-sence of decorin). Decorin was applied to grids andpossessed a structure typical of a protein of its size (datanot shown). Collagen filaments were detected across thegrids at the 0 time point (Figs. 4A and B) and there wasno difference in amount or distribution. By 9min, therewas a clear distinction between the two conditions (Figs.4C and D) with fibrils formed under both conditions butlateral fusion of fibrils was detected when decorin wasabsent. In addition, filaments were apparent under bothconditions but were more prevalent in the presence ofdecorin, indicating that decorin modified the rate offibrillogenesis. By 30min when decorin was absent, nocollagen filaments were detected and lateral fusion ofmany fibrils was detected (Fig. 4E). In contrast, whendecorin was present the fibrils retained the typicalbanding pattern of collagen and did not possess lateralfusion when filaments were still present (Fig. 4F), indi-cating that the kinetics of fibrillogenesis was shifted tothe right. Attempts to decrease the salt concentration toavoid precipitates were hindered by requirement of saltfor fibrillogenesis. The characteristics of fibrillogenesiscorrelate with those found by Danielson et al. [8] indecorin-null mice where the mice showed fibrils thatwere more ‘‘loosely woven’’ or possessed lateral fusion.Our observations are similar to those of Rada et al. [36]where they describe thinner fibrils formed in the pres-ence of lumican. However, they did not show any TEMdata with decorin. We predict that the increase in ab-sorbance detected at 340 nm in collagen alone reflectedthe lateral fusion present in (Figs. 4C and E).Decorin binds TGF-b. It is well accepted that TGF-b

binds to decorin [10,12,13]. Takeuchi et al. [12] firstdemonstrated that TGF-b binding was localized to thecore protein and was not dependent upon the chond-roitin sulfate chain. In addition, decorin was shown tocontain more than one binding site with high and lowaffinities [12,13]. Schonherr et al. [13] localized the highaffinity binding site for TGF-b to Leu155–Val260;however, it was a weak competitor of wild-type decorinto type I collagen fibrils. These data indicate that inde-pendent binding sites of decorin for TGF-b and type Icollagen exist.

Decorin was biotinylated with Sulfo-NHS biotin us-ing standard procedures [34]. Biotinylated corneal de-corin was immobilized on neutravidin–agarose beadsand incubated with a range of [125I]TGF-b1 concentra-tions [37]. Bound TGF-b1 was separated from freepeptide by rapid chromatography on neutravidin–Sepharose microcolumns. Binding data were analyzed bynon-linear least-squares curve fitting (Kaliedagraph) andScatchard plot as previously described [6]. An apparentKd of 40 nM was calculated (Fig. 5A) [38]. Binding ofnative decorin was similar to that of decorin isolatedunder denaturing conditions [12,13]. Preliminary datashowing controlled digestion of decorin retain binding aslong as the binding region is intact (data not shown).

Fig. 4. Collagen fibrillogenesis is influenced by decorin. Purified corneal

decorin (100lg/ml) was incubated in the presence of 300lg/ml rat tailtendon collagen and fibrillogenesis was monitored at 0, 9, and 30min

using negative stain electron microscopy. Collagen alone: (A), (C), and

(E). Collagen and decorin: (B), (D), and (F). Arrows indicate collagen

filaments. At 0min of incubation, collagen filaments are present dis-

tributed evenly over the grid in the absence (A) or presence (B) of de-

corin.At 9min, fibrillogenesis is evident in both (C) and (D)with a larger

number of filaments in the presence (D) of decorin. Lateral fusion is

present inC (collagenalone).At 30min, fibrillogenesis is detected in both

(E) and (F) with collagen filaments still detected in the presence (F) of

decorin. All micrographs are magnified 35,000�. Scale bar¼ 100 nm.


Circular dichroism spectroscopy. Far-UV CD spectrawere measured for native corneal decorin, denaturedcorneal decorin, and renatured corneal decorin. Nativecorneal decorin exhibited a broad far-UV CD spectrum(Fig. 6A, solid line) with a minimum from 221 to 217 nmand a maximum at 204 nm. In the presence of 8M urea,decorin was completely denatured (dashed line) and itwas reversibly renatured when urea was removed viadialysis (Fig. 6A). Decorin isolated by the currentmethod displays a similar far-UV CD spectrum andreversibility to urea unfolding as decorin was expressedand purified from eukaryotic cells described by Krish-nan et al. [39].Thermal denaturation was monitored continuously

from 5 to 95 �C at 217 and 222 nm to examine the sta-bility and reversibility of denaturation of the b-sheet anda-helical regions. Spectra of corneal decorin recorded at25 �C (Fig. 6B) before and after thermal denaturationstudies (Fig. 6C) show that heat-induced denaturationof decorin was irreversible. Thermal denaturation re-sulted in unfolding of decorin at 42 �C from a predom-inantly b-sheet–a-helical conformation to a random coilor unstructured.Prior to the present study, many interactions of de-

corin with extracellular matrix molecules and cytokineshave not been evaluated using decorin from a naturalsource. Most of the current knowledge of decorin–TGF-b binding and decorin–EGF receptor interactions hasbeen deduced using recombinant decorin expressed in

Fig. 6. Circular dichroism of corneal decorin. Far-UV CD spectra (A)

and (B) were recorded at 25 �C. All CD spectra and thermal dena-turation studies were performed on an Aviv 62DS CD Spectropolar-

imeter in 0.05 cm cuvettes with 0.2mg/ml decorin. Temperature was

maintained by use of a thermoelectric temperature controller. All

spectra were corrected for contribution from buffer and normalized to

molar ellipticity. (A) Solid line, native corneal decorin; dashed line,

decorin equilibrated in buffer containing 8M urea; dotted line, decorin

denatured in 8M urea dialyzed exhaustively against urea-free buffer.

(B) Solid line, native corneal decorin; dashed line, heat denatured

corneal decorin (after performing one of the studies presented in (C)).

(C) Heat-induced denaturation of corneal decorin. Samples were he-

ated continuously from 5 to 95 �C and ellipticity at 217 or 222 nm wasmonitored.

Fig. 5. Corneal decorin binds TGF-b. Purified corneal decorin wasbiotinylated and decorin beads were prepared as described in ‘‘Mate-

rials and methods.’’ Twenty microliters of decorin beads was incubated

with the indicated concentrations of [125I]TGF-b1 for 1 h at 4 �C. Afterwashing, bound and free TGF-b1 were determined using a c-counter.Data presented represent the amount of TGF-b bound to decorin as afunction of free TGF-b concentration. Inset shows the Scatchard plotof binding data. Kd was determined by least-squares curve fitting. Datapoints are the average of triplicate determinations.


mammalian systems. This purification regime permittedthe purification of milligram quantities of decorin fromtissue under native conditions without the use of strongdenaturing reagents. The biological activity of decorinwas confirmed by collagen fibrillogenesis assays andTGF-b binding. These activities presumably dependupon proper protein folding. Furthermore, we havedemonstrated that the secondary structure and stabilityof the decorin core protein were remarkably similar tothose of decorin produced in recombinant systems usingCD spectroscopy [39].

Acknowledgments

We thank Dr. Phil Stone for amino acid analysis andDr. Catherine Costello for mass spectrometry confir-mation of decorin and for crucial discussions.

References

[1] E. Hay, in: E. Hay (Ed.), Cell Biology of Extracellular Matrix,

Plenum Press, New York, NY, 1997, pp. 1–6 (Chapter 12).

[2] J. Folkman, M. Klagsbrun, Angiogenic factors, Science 235 (1987)

442–447.

[3] J.C. Adams, F.M. Watt, Regulation of development and differ-

entiation by the extracellular matrix, Development 117 (1993)

1183–1198.

[4] I.R. Iozzo, A.D. Murdoch, Proteoglycans of the extracellular

environment: clues from the gene and protein side offer novel

perspectives in molecular diversity and function, FASEB J. 10

(1996) 598–614.

[5] R.V. Iozzo, The biology of the small leucine-rich proteoglycans.

Functional network of interactive proteins, J. Biol. Chem. 274

(1999) 18843–18846.

[6] R.V. Iozzo, Matrix proteoglycans: from molecular design to

cellular function, Annu. Rev. Biochem. 67 (1998) 609–652.

[7] J.D. Gregory, L. Coster, S.P. Damle, Proteoglycans of rabbit

corneal stroma: isolation and partial characterization, J. Biol.

Chem. 257 (1982) 6965–6970.

[8] K.G. Danielson, H. Baribault, D.F. Holmes, H. Graham, K.E.

Kadler, R.V. Iozzo, Targeted disruption of decorin leads to

abnormal collagen fibril morphology and skin fragility, J. Cell.

Biol. 136 (1997) 729–743.

[9] J. Massague, TGF-beta signal transduction, Annu. Rev. Biochem.

67 (1998) 753–791.

[10] Y. Yamaguchi, E. Ruoslahti, Expression of human proteoglycan

in Chinese hamster ovary cells inhibits cell proliferation, Nature

336 (1988) 244–246.

[11] W.A. Border, N.A. Noble, T. Yamamoto, J.R. Harper, Y.

Yamaguchi, M.D. Pierschbacher, E. Ruoslahti, Natural inhibitor

of transforming growth factor-beta protects against scarring in

experimental kidney disease, Nature 360 (1992) 361–364.

[12] Y. Takeuchi, Y. Kodama, T. Matsumoto, Bone matrix decorin

binds transforming growth factor-beta and enhances its bioacti-

vity, J. Biol. Chem. 269 (1994) 32634–32638.

[13] E. Schonherr, M. Broszat, E. Brandan, P. Bruckner, H. Kresse,

Decorin core protein fragment Leu155–Val260 interacts with

TGF-beta but does not compete for decorin binding to type I

collagen, Arch. Biochem. Biophys. 355 (1998) 241–248.

[14] K. Imai, A. Hiramatsu, D. Fukushima, M.D. Pierscbacher, Y.

Okada, Degradation of decorin by matrix metalloproteinases:

identification of the cleavage sites, kinetic analyses and trans-

forming growth factor-beta1 release, Biochem. J. 322 (1997)

809–814.

[15] H. Hausser, H. Kresse, Decorin endocytosis: structural features of

heparin and heparan sulphate oligosaccharides interfering with

receptor binding and endocytosis, Biochem. J. 344 (1999) 827–835.

[16] D.K. Moscatello, M. Santra, D.M. Mann, D.J. McQuillan, A.J.

Wong, R.V. Iozzo, Decorin suppresses tumor cell growth by

activating the epidermal growth factor receptor, J. Clin. Invest.

101 (1998) 406–412.

[17] A. De Luca, M. Santra, A. Baldi, A. Giordano, R.V. Iozzo,

Decorin-induced growth suppression is associated with up-regu-

lation of p21, an inhibitor of cyclin-dependent kinases, J. Biol.

Chem. 271 (1996) 18961–18965.

[18] G. Csordas, M. Santra, C.C. Reed, I. Eichstetter, D.J. McQuillan,

D. Gross, M.A. Nugent, G. Hajnoczky, R.V. Iozzo, Sustained

down-regulation of the epidermal growth factor receptor by

decorin. A mechanism for controlling tumor growth in vivo,

J. Biol. Chem. 42 (2000) 32879–32887.

[19] R.W. Farndale, D.J. Buttle, A.J. Barrett, Improved quantitation

and discrimination of sulphated glycosaminoglycans by use

of dimethylmethylene blue, Biochim. Biophys. Acta 883 (1986)

173–177.

[20] U.K. Laemmli, Cleavage of structural proteins during the

assembly of the head of bacteriophage T4, Nature 227 (1970)

680–685.

[21] H.E. Huxley, G. Zubay, Electron microscope observations on the

structure of microscomal particles of E. coli, J. Mol. Biol. 2 (1960)

10–18.

[22] T.M. Hering, J. Kollar, T.D. Huynh, J.B. Varelas, Purification

and characterization of decorin core protein expressed in Esche-

richia coli as a maltose-binding protein fusion, Anal. Biochem. 240

(1996) 98–108.

[23] J. Gu, Y. Nakayama, K. Nagai, Y. Wada, The production and

purification of functional decorin in a baculovirus system,

Biochem. Biophys. Res. Commun. 232 (1997) 91–95.

[24] P. Ramamurthy, A.M. Hocking, D.J. McQuillan, Recombinant

decorin glycoforms. Purification and structure, J. Biol. Chem. 271

(1996) 19578–19584.

[25] J.L. Funderburgh, G.W. Conrad, Isoforms of corneal keratan

sulfate proteoglycan, J. Biol. Chem. 265 (1990) 8297–8303.

[26] I. Axelsson, D. Heinegard, Characterization of the keratan

sulphate proteoglycans from bovine corneal stroma, Biochem.

J. 169 (1978) 517–530.

[27] J.R. Hassell, D.A. Newsome, V.C. Hascall, Characterization and

biosynthesis of proteoglycans of corneal stroma from rhesus

monkey, J. Biol. Chem. 254 (1979) 12346–12354.

[28] C.T. Brown, M. Vural, M. Johnson, V. Trinkaus-Randall, Age-

related changes of scleral hydration and sulfated glycosaminogly-

cans, Mech. Ageing Dev. 77 (1994) 97–107.

[29] C.T. Brown, E. Applebaum, R. Banwatt, V. Trinkaus-Randall,

Synthesis of stromal glycosaminoglycans in response to injury,

J. Cell. Biochem. 59 (1995) 57–68.

[30] S. Ernst, R. Langer, C.L. Cooney, R. Sasisekharan, Enzymatic

degradation of glycosaminoglycans, Crit. Rev. Biochem. Mol.

Biol. 30 (1995) 387–444.

[31] Y. Shibusawa, Separation and retention of human blood cells by

surface-affinity chromatography, J. Biochem. Biophys. Methods

49 (2001) 683–703.

[32] D.E. Birk, M.A. Lande, Corneal and scleral collagen fiber

formation in vitro, Biochem. Biophys. Acta 670 (1981) 362–369.

[33] D.E. Birk, J.M. Fitch, J.P. Babiarz, K.J. Doane, T.F. Linsenma-

yer, Collagen fibrillogenesis in vitro: interaction of types I and V

collagen regulates fibril diameter, J. Cell. Sci. 95 (1990) 649–657.

[34] P.G. Neame, C.J. Kay, D.J. McQuillan, M.P. Beales, J.R. Hassell,

Independent modulation of collagen fibrillogenesis by decorin and

lumican, Cell Mol. Life Sci. 57 (2000) 859–863.


[35] J.M. Kuc, P.G. Scott, Increased diameters of collagen fibrils

precipitated in vitro in the presence of decorin from various

connective tissues, Connect Tiss. Res. 36 (1997) 287–296.

[36] J.A. Rada, P.K. Cornuet, J.R. Hassell, Regulation of corneal

collagen fibrillogenesis in vitro by corneal proteoglycan (lumican

and decorin) core proteins, Exp. Eye Res. 56 (1993) 635–648.

[37] J.G. Altin, E.B. Pagler, A one-step procedure for biotinylation

and chemical cross-linking of lymphocyte surface and intracellular

membrane-associated molecules, Anal. Biochem. 224 (1995) 382–

389.

[38] G. Scatchard, The attraction of proteins for small molecules and

ions, Ann. NY Acad. Sci. 51 (1949) 660–672.

[39] P. Krishnan, A. Hocking, J.M. Scholtz, N. Pace, K.K. Holik, D.J.

McQuillan, Distinct secondary domains of the leucine-rich repeat

proteoglycans decorin and biglycan, J. Biol. Chem. 274 (1999)

10945–10950.


extraction and purification of decorin from corneal stroma retain structure and biological activity

Documents