Journal of

www.elsevier.com/locate/yjsbi

Journal of Structural Biology 151 (2005) 69–78

StructuralBiology

The cementum–dentin junction also contains glycosaminoglycansand collagen fibrils

Sunita P. Ho a, Rosalyn M. Sulyanto b, Sally J. Marshall a, Grayson W. Marshall a,*

a Division of Biomaterials and Bioengineering, Department of Preventive and Restorative Dental Sciences, 707 Parnassus Avenue,

University of California San Francisco, San Francisco, CA 94143-0758, USAb Department of Bioengineering, 467 Evans Hall #1762, University of California Berkeley, Berkeley, CA 94720-1762, USA

Received 27 January 2005, and in revised form 29 April 2005Available online 31 May 2005

Abstract

The presence of glycosaminoglycans (GAGs) and their contribution to mechanical properties of the cementum–dentin junction(CDJ) were investigated using nanometer scale characterization techniques. Five to two millimeter thick transverse sections from theapical ends of human molars were ultrasectioned at room temperature under wet conditions using a diamond knife and an ultra-microtome. The structure of the CDJ under dry and wet conditions before and after digestion of GAGs and collagen fibrils wasstudied using an atomic force microscope (AFM). The mechanical properties of the untreated and enzyme treated CDJ underwet conditions were studied using an AFM-based nanoindenter. GAG digestion was performed for 1, 3, and 5 h at 37 �C usingchondroitinase-ABC. Collagen fibril digestion was performed for 24 and 48 h at 37 �C using collagenase. As reported previously,AFM scans of dry untreated CDJ (control) revealed a valley, which transformed into a peak under wet conditions. The height dif-ferences relative to cementum and dentin of untreated and treated CDJ were determined by measuring the CDJ profile under dryand wet conditions. The depth of the valley of GAG and collagen-digested CDJ was greater than that of undigested CDJ under dryconditions. The height of the peak of GAG-digested CDJ was significantly higher than that of the undigested CDJ under wet con-ditions. The collagen-digested CDJ under wet conditions is assumed to form a valley because of the removal of collagen fibrils fromthe CDJ. However, the depth of the valley was lower compared to the depth under dry conditions. Wet AFM-based nanoindenta-tion showed that the elastic modulus and hardness of control (3.3 ± 1.2 and 0.08 ± 0.03 GPa) were significantly higher (ANOVA &SNK, P < 0.05) than chondroitinase-ABC treated CDJ (0.9 ± 0.4 and 0.02 ± 0.004 GPa) and collagenase treated CDJ (1.5 ± 0.6and 0.04 ± 0.01 GPa). No significant difference in mechanical properties between chondroitinase-ABC and collagenase treatedCDJ was observed. Based on the results it was concluded that the 10–50 lm wide CDJ is a composite that includes, chondroi-tin-4-sulfate, chondroitin-6-sulfate, and possibly dermatan sulfate, and collagen fibrils. The association of GAGs with the collagenfibrils provides the observed controlled hydration and partially contributes toward the stiffness of the CDJ under wet conditions.� 2005 Elsevier Inc. All rights reserved.

Keywords: Proteoglycans; Cementum–dentin junction; Atomic force microscopy; Structure; Mechanical properties

1. Introduction

Tissue biomechanics are often considered at threehierarchical length scales (macroscale, mesoscale, andnanoscale) (Weiner et al., 1999). The intra- and intermo-

1047-8477/$ - see front matter � 2005 Elsevier Inc. All rights reserved.

doi:10.1016/j.jsb.2005.05.003

* Corresponding author. Fax: +1 415 476 0858E-mail address: graymar@itsa.ucsf.edu (G.W. Marshall).

lecular forces between the nanocomponents within theextracellular matrix (ECM) define the shape and bio-mechanics (Scott and Thomlinson, 1998) of microscaleregions within macroscale tissues (Fig. 1). Two suchnanocomponents that contribute significantly to the bio-mechanics of soft tissues are the proteoglycans (PGs)and collagen fibrils. For hard tissues, a third componentin the form of extra- and intrafibrillar apatite crystalscould contribute toward resisting mechanical loads.

Fig. 1. (A) Macroscale human molar; (B) microscale cementum–dentin interface of interest in this study; light microscopy (B1) and atomic forcemicroscopy (B2) micrographs of microscale tubules in dentin, lamellae in cementum, and the interfacial zone (IZ in B1) containing the 10–50 lmhydrophilic cementum–dentin junction (CDJ in B2); (C) key nanoscale components of the hydrophilic CDJ extracellular matrix are collagen fibrilsand proteoglycans (PGs). Note. Figure not drawn to scale.

70 S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78

The PGs within the ECM of hard and soft tissuesform a network with collagen fibrils, resulting in asupramolecular structure that allows the distributionof functional tensile and compressive loads (Bartoldet al., 1988; Murray et al., 1993). This study focuseson understanding the contribution of the PGs to thestructure and mechanical properties of the cementum–dentin junction (CDJ) that bonds the cementum androot dentin in a tooth.

Over the last eight decades, the union between dissim-ilar mineralized tissues, such as cementum to root dentinhas been a subject of interest (Bodecker, 1957; Benez,1927; Bosshardt and Schroeder, 1996; El Mostehy andStallard, 1968; Hopewell-Smith, 1920; Yamamoto etal., 1999, 2000a). The interests were twofold: (1) biolog-ical; involving the union of two developing dissimilarmineralized tissues and (2) clinical; involving buildingload-resistant unions between newly formed cementumand diseased root. Some investigators reported thatthe union is based on Sharpey�s fibers with a calcifiedcore extending from the root surface of cementum andattaching to root dentin (Strocchi et al., 1999). Accord-ing to other investigators, in rodent molars the union be-tween cementum and dentin is formed during toothdevelopment through the attachment of periodontal fi-bers to the collagen fibers within the CDJ (Bosshardtand Schroeder, 1996; Ten Cate, 1998) thus anchoringthe tooth. However, there remains a large void in corre-

lating the structure, chemical composition, and mechan-ical properties of the CDJ in a human tooth.

Recent microscale and nanoscale mechanical andchemical characterization techniques by our group indi-cated the presence of mineral within an 80–200 lm wideinterface between cementum and dentin (Ho et al.,2004a,b,c). This interface was reported to be a combina-tion of 70–150 lm wide root mantle dentin and a10–50 lm wide hydrophilic cementum–dentin junction(CDJ) (Ho et al., 2004a,b,c). Previous studies concludedthat the structure of the CDJ was contributed by colla-gen fibrils exclusively; named as a collagen hiatus (ElMostehy and Stallard, 1968). However, in this study,the presence of proteoglycans (PGs) in addition to thecollagen fibrils within the CDJ is considered.

The PGs in hard tissues such as cementum, dentinand bone, and various soft tissues act as hydrophilicmonomers in association with other more or less hydro-philic polymers. The basic building block of a PG is ananionic glycosaminoglycan (GAG), an important com-ponent of the ECM of vertebrate tissues. The long un-branched polysaccharide chains of the GAGs, whichcould contain many acidic, carboxylate and/or sulfategroups, make the GAG polymeric chains polyanionicwith a high affinity for water molecules. GAGs are cova-lently attached to a core protein via a link protein form-ing, nanometer size bottle-brush PG (Murray et al.,1993).

S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78 71

Table 1 enumerates the types of PGs and GAGsidentified within dental tissues by numerous investiga-tors. The presence of chondroitin sulfate rich PGs suchas versican, biglycan, and decorin within bovine bulkcementum was illustrated by Bartold et al. (1990),and Cheng et al. (1999). Bartold et al. (1990), showedthat these molecules have a larger distribution by dem-onstrating chondroitin sulfate PGs in the pericellularspaces within the lacunae and in the extracellular spac-es of cementum. Localization of lumnican and fibro-modulin PGs in pre-cementum walls of the lacunaeand cementocytes has recently been shown in bovinebulk cementum by Cheng et al. (1999). In contrast,Yamamoto et al. (1999, 2000a), suggested that PGswere localized within a narrow region 1–3 lm, whichthey termed the CDJ.

The existence of various PGs within cementum isconvincing, but the existence of PGs within the CDJneeds more investigation. Ho et al. (2004a,b,c), suggest-ed that a 10–50 lm wide lower mineral region sensitiveto hydration, may represent a broader region of PGlocalized CDJ. Hence, it was hypothesized that thehydration effects observed as controlled swelling aredue to the interaction of PGs and collagen fibrils withinthe CDJ and that the mechanical integrity of the CDJ isprovided by the PG and collagen interactions. To testthis hypothesis, the following objectives were definedand evaluated: (1) determination of changes in hydra-

Table 1List of recently published works (past 15 years) illustrating the presence ofligament of teeth from different mammals

GAGs; PGs; characterization techniques Tissue and loca

GAGs: C4S, C6S, KS, DS, immunohistochemistry; PGs;Versican, Decorin, Biglycan, Fibromodulin,,and Lumican; immunohistochemical techniques

Human cement

PGs and hyaluronate, agarose–acrylamide gelelectrophoresis

Bovine cementu

PGs: Versican, Decorin, Biglycan; GAG: C6S and C4S,immunoblot assays, immunohistochemistry

Cementum of b

GAGs: KS, C4S Cementifying fiPGs, cuprolinic blue, TEM Cementum, preGAG, cuprolinic blue at various magnesium chlorideconcentrations, TEM

Predentin and d

PG: Lumican, GAG: KS, Immunohistochemistry Human predentGAGs: C4S and C6S; Immunohistochemical techniques Periodontal ligaPG: Fibromodulin; Immunohistochemistry Periodontium o

formationPGs: Decorin, Biglycan, mRNA expressions Mouse toothGAGs: C4S, DS; Immunohistochemical, TEM techniques Predentin of raPGs: Versican, Histochemical and Immunohistochemicalmethods

Stress induced mligament of rats

PGs; cuprolinic blue, TEM techniques UnmineralizedHA, GAGs: C4S, C6S, DS, Immunoblot assays,Gel exclusion chromatography

Deciduous teeth

PG: Versican; Immunoblot assays,Immunohistochemical techniques

Bovine dental p

PGs, NaOH-maceration techniquesHistochemical techniques

Cementun–denthuman teeth

tion effects before and after digestion of the CDJ usingspecific enzymes and an atomic force microscope(AFM), and (2) evaluation of the elastic modulus andhardness before and after digestion of the CDJ usingspecific enzymes and an AFM-based nanoindenter.

2. Materials and methods

2.1. Specimen preparation

Mandibular molars (N = 5) of males with ages rang-ing from 65 to 80 years were sterilized using 0.26 Mradof c-radiation (White et al., 1994). The processes ofcoarse and ultrasectioning the teeth has been describedpreviously (Ho et al., 2004a,b,c). In brief, the teethwas sectioned using a diamond-wafering blade and alow speed saw (Isomet, Buehler, Lake Bluff, IL) underwet conditions and trimmed using a glass knife (Elec-tron Microscopy Laboratory—UCSF, San Francisco,CA) and an ultramicrotome (Ultracut E, Reichert-Jung,Vienna, Austria). Final trimming of the specimens wasperformed using a diamond knife (Micro Star Technol-ogies, Huntsville, TX) by removing 300 nm thick ultra-sections. The ultrasectioned surface of the remainingspecimen block was characterized before and after en-zyme treatment under dry and wet conditions using anAFM and an AFM-based nanoindenter.

PGs and GAGs within predentin, dentin, cementum, and periodontal

tion Investigators

um and human molars Ababneh et al. (1998, 1999)

m Bartold et al. (1990)

ovine molars Cheng et al. (1999)

broma, fibro-osseous lesions Endo et al. (2003)dentin, young mouse incisors Everts et al. (1995)entin of rat incisor Goldberg and Septier (1992)

in and dentin Hall et al. (1997)ment of rat molars Kagayama et al. (1996)f a rat, cementogenesis and root Matias et al. (2003)

Matsuura et al. (2001)t incisor Septier et al. (1998)easurements, periodontal Sato et al. (2002)

dentin of rat tooth germs Tenorio et al. (1990)of calves Waddington et al. (2003)

ulp Yamauchi et al. (1997)

in junction (CDJ) of rat molars, Yamamoto et al. (1999, 2000a,b)

72 S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78

2.2. Digestion of GAGs and collagen fibrils

The specimens were digested with protease free chon-droitinase-ABC (C-ABC) from Proteus vulgaris (100332,Seikagaku Corporation, Tokyo, Japan) based upon theassumption that GAGs (chondroitin-4-sulfate �C4S,�chondroitin-6-sulfate �C6S,� and dermatan sulfate �DS�)reported in cementum and dentin (Table 1) would existin the CDJ. This enzyme digests the GAG moieties fromthe PG molecule. The ultrasectioned specimens were im-mersed in 0.1 IU/ml C-ABC in 0.1% bovine serum albu-min (BSA) (BP675-1, Fisher Scientific, Fair Lawn, NJ)pH 7.2 at 37 �C for 1, 3, and 5 h to digest the GAGs.The specimens were subsequently rinsed with PBS(BP665-1, Fisher Scientific, Fair Lawn, NJ) and Tween20 (0.01% Tween vol/vol) (BP337-100, Fisher Scientific,Fair Lawn, NJ) for 15 min to remove digested GAGfragments (Ababneh et al., 1999). The change in struc-ture of the CDJ after GAG-digestion was estimated bymeasuring the changes in depths of valleys under dryconditions and heights of the peaks under wet condi-tions using an AFM. The mechanical properties of theGAG-digested specimens were studied using an AFM-based nanoindenter. Undigested ultrasectioned speci-mens were used as controls.

The collagen within the CDJ was digested using themethods described by Lin et al. (1993), and Mackie etal. (1989). The specimens were digested with 2100 U ofcollagenase from Clostridium histolyticum (Type II-S,C1764, Sigma–Aldrich, St. Louis, MO) in 0.05 M Trisand 0.01 M calcium acetate-buffered saline at a concen-tration of 0.069 mg/ml, pH 7.2. The specimens weredigested for 1, 3, 5, 24, and 48 h at 37 �C. The mechan-ical properties and structure of the collagenase-digestedspecimens were studied following the same methods de-scribed for the GAG-digested specimens.

2.3. AFM and AFM-based nanoindentation of theultrasectioned specimen surfaces

The structural changes of the digested ultrasectionedspecimens under dry and wet conditions were studiedusing a contact mode AFM (Nanoscope III, Multimode,DI-Veeco Instruments, Santa Barbara, CA). A Si3N4 tipattached to a �V-shaped� type �D� microlever with a nom-inal spring constant of 0.03 N/m (Thermo Microscopes,CA) at a scanning frequency of 0.5 Hz was used to scanthe surfaces of the specimens. The nominal radius ofcurvature of the tip was less than 50 nm (Ho et al.,2004a,b,c). Three different locations for each of threespecimens were scanned. Areas as large as100 lm · 100 lm were evaluated using a �J� type piezoscanner. Specimens digested with C-ABC and collage-nase were scanned under dry and wet conditions. Thestructural changes of the CDJ under dry and wet condi-tions for respective groups were determined using the

scanned areas, Nanoscope III Version 4.43r8 Software(Nanoscope III, Multimode, DI-Veeco Instruments,Santa Barbara, CA), and WSxM 4.0 Develop 7.5 Scan-ning Probe Microscopy Software (Nanotech ElectronicaS.L., Madrid, SPAIN). Additionally, MATLAB & SIM-ULINK (Student Version 7.0.1.15(R14) Service Pack 1,The Math Works, Natick, MA) was used to construct3D images for five section profiles taken across eachCDJ under dry and wet conditions for each respectivegroup.

Nanomechanical tests on dry and wet specimens(N = 3) were performed using an AFM, to which aload-displacement transducer (Triboscope Microme-chanical Test Instrument, Hysitron Incorporated, Min-neapolis, MN) was attached. A sharp diamondBerkovich indenter with a radius of curvature less than100 nm (Triboscope Micromechanical Test Instrument,Hysitron Incorporated, MN) was fitted to the transduc-er. The scan speed, scan area, and load-displacement ofthe indenter on the specimen were controlled by a com-puter (Ho et al., 2004a,b,c). After indentation, the AFMpiezo was used to scan the indented area. Immersion ofthe specimen and indenter in distilled water allowedmeasuring the mechanical properties closer to in vivoconditions. A minimum of 25 indents were made on50 lm · 50 lm areas. The maximum load was1000 lN, with load, hold, and unload for 3 s each whenmeasuring the reduced elastic modulus �Er� and hardness�H.� Fused silica was used as a calibration standard forthe AFM-nanoindenter.

3. Results

3.1. AFM on GAG and collagen-digested CDJ

Representative 75 lm · 75 lm AFM scans of thecontrol, GAG-digested, and collagen-digested speci-mens under dry and wet conditions are shown in Fig.2. The width of the CDJ in this study was approximately20 lm (Figs. 2 and 3). The CDJ in this study spannedapproximately from 20 to 40 lm (Fig. 3). As reportedin our previous studies, the untreated CDJ (control)under dry conditions was a valley (Figs. 2a and 3a)(Ho et al., 2004a,b,c). Under wet conditions, the valleyof the CDJ transformed into a peak (Figs. 2b and 3b),which was an indication of swelling due to hydration(Ho et al., 2004a,b,c). The 1 h GAG specimens showedthe same behavior as the control specimens (Figs. 2aand b). The depth and height of the 3 and 5 h GAG-di-gested CDJ under dry (Figs. 2c and 3c) and wet (Figs. 2dand 3d) conditions increased by twofold when comparedto the undigested CDJ (control) (Figs. 2a and b).

The 1, 3, and 5 h collagen-digested specimens showedthe same behavior as the control specimens under dry(Fig. 3a) and wet (Fig. 3b) conditions. The 24 and

Fig. 2. 75 lm · 75 lm representative AFM micrographs illustrating CDJ for the following groups: (a) control dry; (b) control wet; (c) 3 h GAG-digested dry; (d) 3 h GAG-digested wet; (e) 24 h collagen-digested dry; and (f) 24 h collagen-digested wet. The corresponding section plots (sectionstaken across the CDJ) illustrate the depth of the valley and the height of the peak under dry and wet conditions for the respective groups. Note. C,cementum and D, dentin. Data for 1 h GAG-digested CDJ were not shown because they were similar to the control. Data for 5 h were similar to 3 hGAG-digested specimens. Data for 1, 3, and 5 h collagen-digested CDJ were not shown because they were similar to the control. Data for 48 h weresimilar to 24 h collagen-digested specimens.

S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78 73

48 h collagen-digested specimens showed the samebehavior; a significant increase in depth under dry con-ditions (Fig. 3e) compared to the control. Under wet

conditions, the CDJ was a valley (Fig. 3f), with thedepth smaller than that observed under dry conditions(Fig. 3e).

Fig. 3. MATLAB reconstructured 3D section plots across CDJ subsequently observed under dry and wet conditions. (a) control dry; (b) control wet;(c) 3 h GAG-digested dry; (d) 3 h GAG-digested wet; (e) 24 h collagen-digested dry; and (f) 24 h collagen-digested wet. Note. C, cementum and D,dentin. Data for 1 h GAG-digested CDJ were not shown because they were similar to the control. Data for 5 h were similar to 3 h GAG-digestedspecimens. Data for 1, 3, and 5 h collagen-digested CDJ were not shown because they were similar to the control. Data for 48 h were similar to 24 hcollagen-digested specimens. X, across CDJ (lm); Y, profile number (total 5); and Z, height differences (nm).

74 S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78

3.2. AFM-based nanoindentation on GAG and collagen-

digested CDJ

The elastic modulus and hardness of the undigestedCDJ in this study were the same as those reported inour previous studies (Ho et al., 2004a,b,c). Additionally,the elastic modulus and hardness of the 1 h GAG digest-ed cementum, dentin, and CDJ were the same as thecontrol. However, the elastic modulus (Fig. 4A) andhardness (Fig. 4B) of the 3 and 5 h GAG-digested

CDJ and dentin were significantly lower (Student�s t

test, P < 0.05) compared to the control. No significantdifference in elastic modulus and hardness between the3 and 5 h GAG digested, and undigested cementumwere observed (Student�s t test, P > 0.05).

The elastic modulus (Fig. 5A) and hardness (Fig. 5B)of the undigested CDJ (3.3 ± 1.2 and 0.08 ± 0.03 GPa)were significantly higher (ANOVA, SNK, P > 0.05)than the 3 and 5 h GAG (0.9 ± 0.4 and 0.02 ± 0.004GPa) and 24 and 48 h collagen-digested wet CDJ

Fig. 4. Elastic modulus (A) and hardness (B) of GAG digested cementum, dentin, and the CDJ under wet conditions. These measurements wereperformed only under wet conditions to observe the effects of polyanionic molecules on the mechanical properties of the CDJ.

Fig. 5. Comparison of elastic modulus (A) and hardness (B) of GAG (GAGDIG) and collagen-digested CDJ (COLDIG) with undigested CDJ(UNDIG) under wet conditions.

S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78 75

(1.5 ± 0.6 and 0.04 ± 0.01 GPa). No significant differ-ences in mechanical properties between the 3 and 5 hGAG digested or 24 and 48 h collagen-digested CDJwere observed (Fig. 5).

4. Discussion

Treatment with C-ABC or collagenase caused a sig-nificant reduction in mechanical properties and loss ofstructure (illustrated by differences in depths and heightsof valleys and peaks of the CDJ under dry and wet con-ditions) associated with the CDJ, suggesting that the 10–50 lm wide CDJ contains collagen and GAGs. It hasbeen suggested that the density of the material and theratio of organic and inorganic within the CDJ is lowercompared to its more mineralized neighboring materials,cementum, and dentin (Ho et al., 2004a,b,c). Thus, it ispossible that the collagenous matrix could collapse,forming a valley under dry conditions. Under wet condi-tions, the attraction of water molecules by the GAGswithin the collagen-PG network facilitates swelling,transforming the valley to a peak (Figs. 2b, 3b, and 6).

Under dry conditions, it has been estimated that thePGs are restrained to one-fifth of their volume in freesolution (Myers and Mow, 1983). Thus, the chargedgroups fixed along the GAG chains of the solid matrixare very close to each other, causing large charge–chargerepulsive forces to be exerted against each other. Usingan AFM and controlled environmental conditions, Seog

et al. (2002) suggested that the intermolecular GAG–GAG interactions involve electrostatic repulsive forces:between opposing GAGs, due to interdigitation ofGAGs, between neighboring GAGs, and due to sterichindrance between opposing GAGs. These intrinsicforces between PGs and within the same PG could con-tribute to the biomechanical resistance and the intrinsicexpandability of ECM of the CDJ. However, the elec-trostatic repulsive forces within the CDJ require count-er-ions to maintain electroneutrality within the ECM.The ionic imbalance between the tissue and the bathingsolution (in this study-water) gives rise to a Donnanosmotic pressure that causes an intake of fluid into theCDJ relative to cementum and dentin (Figs. 2b, 3b,and 6). At neutral pH, the extrafibrillar water interac-tion with PGs and intrafibrillar water interactions withinthe collagen fibril, collectively could contribute to theECM swelling pressure (Maroudas, 1976; Maroudas etal., 1975,1980,1991). This internal swelling pressure orhydrostatic pressure creates a tension in the collagen-PG network that provides necessary mechanical integri-ty to the hydrophilic region, as observed from the nan-oindentation results under undigested conditions (Figs.4 and 5).

In this study, specific bacterial mucosopolysaccharid-ases have been used to identify GAGs by comparing thestructure and mechanical properties of the CDJ. Theprotease free C-ABC degrades DS in addition to chon-droitin sulfates-A (C4S) and chondroitin sulfates-C(C6S), and is devoid of any other proteolytic activity

Fig. 6. A two-dimensional schematic of undigested, C-ABC digested, and collagen-digested CDJ under dry (A) and wet (B) conditions. Note. Figurenot drawn to scale.

76 S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78

(Junqueira and Montes, 1983). The significant loss ofstructure and mechanical properties of the CDJ ob-served using an AFM and AFM-based nanoindentationis consistent with the existence of C4S and C6S-GAGs.However, the presence of the DS has to be further inves-tigated by digesting the CDJ with chondroitinase B,which degrades DS only.

The recent evidence of PO3�4 peak within the cemen-

tum–dentin interface (Ho et al., 2004a,b,c) could indi-cate the presence of mineral; a partial contributor tothe measured stiffness of the CDJ. The other stiffnesscontributing factors could be intrafibrillar and interfi-brillar crosslink density, collagen-PG, PG–PG, andGAG–water inter- and intramolecular interactions.The use of C-ABC removed GAGs from the CDJ result-ing in breakdown of the collagen-PG network. Thiscaused a collapse of the collagen fibers observed as a sig-nificant increase in depth of the CDJ (Figs. 2, 3c, and 6)under dry conditions. Under wet conditions, the loss ofstructure resulted in an increase in height of the CDJ(Figs. 2, 3d, and 6).

Collagen fibrils interact through crosslinks to formfibrillar networks (Maroudas, 1975). Collagen isextremely stable and resists the action of non-specificproteases. Bacterial collagenases as used in this studycut collagen a-chains at several sites, forming smallsegments. The use of collagenase may have facilitatedremoval of collagen fibrils, observed as a significant

increase in depth under dry (Figs. 2e, 3e, and 6) andwet (Figs. 2, 3f, and 6) conditions. The significant de-crease in �Er� and �H� (Fig. 5) of the collagen-digestedCDJ could be related to a decrease in intrafibrillar andinterfibrillar crosslink density. Similar observationswere made by Pek et al. (2004), on collagen-chondroi-tin-6-sulfate tissue engineered scaffolds.

The ECM of the CDJ could be modeled as a two-el-ement load-resisting composite material, where the totalload on the matrix is distributed between the tensileload-resisting fibrils, which may be partially mineralizedand compressive load-resisting gel-like PGs (Scott,2001). However, the evidence of mineral within the col-lagen fibrils of the CDJ in particular, should be furthersubstantiated using other characterization techniques.Because of the large slenderness ratio (length/thickness),one of the governing criteria for buckling, the collagenfibrils are more capable of resisting tensile loads com-pared to compressive loads (Mackie et al., 1989). In thisstudy, the observed hydration effects and mechanicalproperties after enzyme treatment of the 10–50 lmCDJ suggest that the localization of PGs is not limitedto a narrow region of 1–3 lm. Under in vivo conditions,the presence of the external bathing solution such aswater in saliva, creates hydrostatic pressure within thetissue that would allow resilience and elasticity, henceallowing resistance to the everyday external cyclic andimpact occlusal loads.

S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78 77

5. Conclusions

This investigation showed that the GAGs are essen-tial components of the 10–50 lm wide CDJ. The hydro-philic CDJ is a composite of mineral and organiccomponents such as chondroitin sulfated GAGs andcollagen fibrils. The presence of GAGs increases elasticmodulus and hardness. Additionally, the GAG-collagennetwork maintains the ultrastructure of the region,which is one of the main contributors to mechanicalintegrity of a tissue.

Acknowledgments

This work was supported by the National Institute ofHealth and National Institute of Dental and Craniofa-cial Research (NIH/NIDCR) Grants T32 DE07306(S.P.H.) and P01 DE09859. The authors thank Dr. JohnGreenspan for the use of the ultramicrotome.

References

Ababneh, K.T., Hall, R.C., Embery, G., 1998. Immunolocalization ofglycosaminoglycans in ageing, healthy and periodontally diseasedhuman cementum. Arch. Oral Biol. 43, 235–246.

Ababneh, K.T., Hall, R.C., Embery, G., 1999. The proteoglycans ofhuman cementum: immunohistochemical localization in healthy,periodontally involved and ageing teeth. J. Periodontal Res. 34, 87–96.

Bartold, P.M., Miki, Y., McAllister, B., Narayanan, A.S., Page, R.C.,1988. Glycosaminoglyacans of human cementum. J. PeriodontalRes. 23, 13–17.

Bartold, P.M., Reinboth, B., Hideaki, N., Narayan, S.A., Page, R.C.,1990. Proteoglycans of bovine cementum: isolation and character-ization. Matrix 10, 10–19.

Bodecker, C.F.W., 1957. The distribution of living matter in humandentine, cement, and enamel. Dent. Cosmos 20, 582–590.

Benez, L., 1927. Befunde an der Dentinzement Grenze. Z. Stomatol. 5,877–896.

Bosshardt, D.D., Schroeder, H., 1996. Cemtogenesis reviewed: acomparison between human premolars and rodent molars. Anat.Rec. 245, 267–292.

Cheng, H., Caterson, B., Yamauchi, M., 1999. Identification andimmunolocalization of chondroitin sulfate proteoglycans in toothcementum. Connect. Tissue Res. 40 (1), 37–47.

El Mostehy, M.R., Stallard, R.E., 1968. Intermediate cementum. J.Periodontal Res. 3, 24–29.

Endo, Y., Uzawa, K., Mochida, Y., Nakatsuru, M., Shiiba, M.,Yokoe, H., Yamauchi, M., Tanzawa, H., 2003. Differentialdistribution of glycosaminoglycans in human cementifying fibromaand fibro-osseous lesions. Oral Dis. 9, 73–76.

Everts, V., Schutter, M., Niehof, A., 1995. Proteoglycans in cemen-tum- and enamel-related predentin of young mouse incisors asvisualized by cuprolinic blue. Tissue Cell 27 (1), 55–60.

Goldberg, M., Septier, D.S., 1992. Differential staining of glycosami-noglycans in the predentine and dentine of rat incisor usingcuprolinic blue at various magnesium chloride concentrations.Histochem. J. 24 (9), 648–654.

Hall, R.C., Embery, G., Lloyd, D., 1997. Immunohistochemicallocalization of the small leucine-rich proteoglycan lumican inhuman predentin and dentin. Arch. Oral Biol. 42 (10–11), 783–786.

Hopewell-Smith, A., 1920. Concerning human cementum. J. Dent.Res. 2, 59–76.

Ho, S.P., Balooch, M., Goodis, H.E., Marshall, G.W., Marshall, S.J.,2004a. Ultrastructure and nanomechanical properties of cemen-tum–dentin junction. J. Biomed. Mater. Res. 68A (2), 343–351.

Ho, S.P., Goodis, H.E., Balooch, M., Nonomura, G., Marshall, S.J.,Marshall, G.W., 2004b. The effect of sample preparation techniqueon determination of structure and nanomechanical properties ofhuman cementum hard tissue. Biomaterials 25 (19), 4847–4857.

Ho, S.P., Balooch, M., Marshall, S.J., Marshall, G.W., 2004c. Localproperties of a functionally graded interface between bulk cemen-tum and dentin hard tissues. J. Biomed. Mater. Res. 70A, 480–489.

Junqueira, L.C.U., Montes, G.S., 1983. Biology of collagen–proteo-glycan interaction. Arch. Histol. Jpn. 46 (5), 589–629.

Kagayama, M., Sasano, Y., Mizoguchi, I., Kamo, N., Takahashi, I.,Mitani, H., 1996. Localization of glycosaminoglycans in periodon-tal ligament during physiological and experimental tooth move-ment. J. Periodontal Res. 31, 229–234.

Lin, C.P., Douglas, W.H., Erlandsen, S.L., 1993. Scanning electronmicroscopy of Type I collagen at the dentin–enamel junction ofhuman teeth. J. Histochem. Cytochem. 41 (3), 381–388.

Maroudas, A., 1975. Biophysical chemistry of cartilaginous tissueswith special reference to solute and fluid transport. Biorheology 12,233–248.

Maroudas, A., 1976. Balance between swelling pressure and collagentension in normal and degenerate cartilage. Nature 260, 808–809.

Maroudas, A., Bayliss, M., Venn, M., 1980. Further studies on thecomposition of human femoral head cartilage. Ann. Rheum. Dis.39, 514–523.

Maroudas, A., Wachtel, E., Grushko, G., Katz, E.P., Weinberg, P.,1991. The effect of osmotic and mechanical pressures on waterpartitioning in articular cartilage. Biochim. Biophys. Acta 1073,285–294.

Mackie, I.G., Green, M., Clarke, H., Isaac, D.H., 1989. Osteoporoticbone microstructure by collagenase etching. Ann. Rheum. Dis. 48,464–469.

Myers, E.R., Mow, V.C., 1983. Biomechanics of cartilage and itsresponse to biomechanical stimuli. In: Hall, B.K. (Ed.), CartilageVolume I, Structure, Function and Biochemistry. Academic Press,New York, pp. 323–325.

Murray, R.K., Granner, D.K., Mayes, P.A., Rodwell, V.W., 1993.Harper�s Biochemistry, twentythirdth ed. Appleton & Lange,Norwalk, Connecticut.

Matias, M.A., Li, H., Young, W.G., Bartold, P.M., 2003. Immuno-histochemical localization of fibromodulin in the periodontiumduring cementogenesis and root formation in the rat molar. J.Periodontal Res. 38, 502–507.

Matsuura, T., Duarte, W.R., Cheng, H., Uzawa, K., Yamauchi, M.,2001. Differential expression of decorin and biglycan genes duringmouse tooth development. Matrix Biol. 20, 367–373.

Pek, Y.S., Spector, M., Yannas, I.V., Gibson, L.J., 2004. Degradationof a collagen-chondroitin-6-sulfate matrix by collagenase and bychondroitinase. Biomaterials 25, 473–482.

Scott, J.E., Thomlinson, A.M., 1998. The structure of interfibrillarproteoglycan bridges (�shape modules�) in extracellular matrix offibrous connective tissues and their stability in various chemicalenvironments. J. Anat. 192, 391–405.

Strocchi, R., Raspanti, M., Ruggeri, A., Franchi, M., PasqualeViviana De, Strings, L., Ruggeri, A., 1999. Intertwined sharpeyfibers in human acellular cementum. Embryology 104 (4), 175–183.

Scott, J.E., 2001. Structure and function in extracellular matricesdepend on interactions between anionic glycosaminoglycans.Pathol. Biol. 49, 284–289.

Septier, D., Hall, R.C., Deanne, L., Embery, G., Goldberg, M., 1998.Quantitative immunohistochemical evidence of a functional gradientof chondroitin 4-sulphate/dermatan sulphate, developmentally reg-ulated in the predentine of rat incisor. Histochem. J. 30, 275–284.

78 S.P. Ho et al. / Journal of Structural Biology 151 (2005) 69–78

Sato, R., Yamamoto, H., Kasai, K., Yamauchi, M., 2002. Distributionpattern of versican, link protein and hyaluronic acid in the ratperiodontal ligament during experimental tooth movement. J.Periodontal Res. 37, 15–22.

Seog, J., Dean, D., Plass, A.H.K., Wong-Palms, S., Grodzinsky, A.J.,Ortiz, C., 2002. Direct measurement of glycosaminoglycan inter-molecular interactions via high-resolution force spectroscopy.Macromolecules 35, 5601–5615.

Tenorio, D., Reid, A.R., Katchburian, E., 1990. Ultrastructuralvisualization of proteoglycans in early unmineralized dentine ofrat tooth germs stained with cuprolinic blue. J. Anat. 169, 257–264.

Ten Cate, A.R., 1998. Oral histology, development, structure andfunction, fifth ed. Mosby—Year Book, Inc., St. Lious, pp. 239

Weiner, S., Traub, W., Wagner, H.D., 1999. Lamellar bone: structure–function relations. J. Struct. Biol. 126 (3), 241–255.

Waddington, R.J., Hall, R.C., Embery, G., Lloyd, D.M., 2003.Changing profiles of proteoglycans in the transition of predentineto dentine. Matrix Biol. 22, 153–161.

White, J.M., Goodis, H.E., Marshall, S.J., Marshall, G.W., 1994.Sterilization of teeth by gamma-radiation. J. Dent. Res. 73 (9),1560–1567.

Yamauchi, S., Cheng, H., Neame, P., Caterson, B., Yamauchi, M.,1997. Identification, partial characterization, and distribution ofversican and link protein in bovine dental pulp. J. Dent. Res. 76(11), 1730–1736.

Yamamoto, T., Domon, T., Takahashi, S., Islam, N., Suzuki, R.,Wakita, M., 1999. The structure and function of the cemento–dentinal junction in human teeth. J. Periodontal Res. 34, 261–268.

Yamamoto, T., Domon, T., Takahashi, S., Islam, N.M., Suzuki, R.,Wakita, M., 2000a. The structure of the cemento–dentinal junctionin rat molars. Ann. Anat. 182, 185–190.

Yamamoto, T., Domon, T., Takahashi, S., Islam, M.N., Suzuki, R.,2000b. The fibrous structure of the cemento–dentinal junction inhuman molars shown by scanning electron microscopy combinedwith NaOH-maceration. J. Periodontal Res. 35, 59–64.