Matrix Biology 23 (2004) 153–161

0945-053X/04/$30.00� 2004 Elsevier B.V./International Society of Matrix Biology. All rights reserved.doi:10.1016/j.matbio.2004.04.003

Expression of collagen XVIII and MMP-20 in developing teeth andodontogenic tumors

Anu Vaananen , Leo Tjaderhane , Lauri Eklund , Ritva Heljasvaara , Taina Pihlajaniemi ,a b c c c¨ ¨ ¨ ¨Riitta Herva , Yanli Ding , John D. Bartlett , Tuula Salo *d e e a,

Department of Diagnostics and Oral Medicine, Institute of Dentistry, University of Oulu, P.O. Box 5281, 90014 Oulu, Finlanda

Institute of Dentistry, University of Helsinki; Department of Oral and Maxillofacial Diseases, Helsinki University Central Hospital, Helsinki,b

FinlandCollagen Research Unit, Biocenter and Department of Medical Biochemistry and Molecular Biology, University of Oulu, Oulu, Finlandc

Department of Pathology, University of Oulu, Oulu, Finlandd

Department of Cytokine Biology, Forsyth Institute, and Department of Oral and Developmental Biology, Harvard Medical School, Boston, MA,e

USA

Received 7 January 2004; received in revised form 27 April 2004; accepted 30 April 2004

Abstract

Collagen XVIII is a basement membrane(BM) component, whereas MMP-20(enamelysin) is a matrix metalloproteinasepredominantly expressed in teeth. Since MMP-20 was found to degrade collagen XVIII, we studied the co-expression of theseproteins in dental tissues. Collagen XVIII surrounded the developing tooth during early and late bell stages and was also presentin developing enamel. Western blotting indicated that developing enamel contains collagen XVIII N-terminal fragments of thefrizzled variant. Enamelysin was co-localized with collagen XVIII in the developing enamel matrix and stratum intermedium.Electron microscope analysis showed that total mineral, calcium and phosphorus contents of enamel were slightly increased incollagen XVIII null mice but the analysis revealed no visible defects in the enamel or dentin structures. In odontogenic tumorsMMP-20 and collagen XVIII were co-localized in the enamel-like tumor matrix. Our results show that collagen XVIII is presentin developing teeth, but its absence seems not to be critical for the development of the teeth.� 2004 Elsevier B.V./International Society of Matrix Biology. All rights reserved.

Keywords: Collagen XVIII; Enamelysin; Matrix metalloproteinases; Tooth mineralization

1. Introduction

The collagens are a family of 25 structural proteinsof the connective tissue(Myllyharju and Kivirikko,2001). Collagen I is the most prominent collagen indentin and collagen III expression has been detected inthe odontoblasts and in the early predentin(Lukinmaaet al., 1993; Palosaari et al., 2001). Collagen X proteinhas been located in the human tooth germs duringenamel maturation(Felszeghy et al., 2000). CollagenXVIII is a non-fibrillar heparan sulfate proteoglycanthat is found in the epithelial and vascular basementmembranes(BMs) throughout the body(Muragaki et

*Corresponding author. Tel.:q358-8-537-5592; fax:q358-8-537-5560.

E-mail address: Tuula.Salo@oulu.fi(T. Salo).

al., 1995). The molecule appears as three N-terminalvariants, which arise from the use of two alternativepromoters and alternative splicing. The longest variantcontains a cysteine-rich frizzled motif, which is homol-ogous to the ligand-binding domain of wingless(Wnt)receptors(Rehn and Pihlajaniemi, 1995).The matrix metalloproteinases(MMPs) are a group

of zinc-dependent endopeptidases, which degrade mac-romolecules in the extracellular matrix and BMs(Naga-se and Woessner, 1999). Several MMPs are expressedin forming dental tissues and may play a role in thebiomineralization of dentin and enamel(Bartlett andSimmer, 1999). Porcine MT1-MMP is expressed byameloblasts and odontoblasts during secretory and mat-uration stages(Caron et al., 1998). MMP-2 expressionhas been detected in the dental papilla, secretory odon-toblasts, and in the inner enamel epithelium of the late

154 A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

Fig. 1. Expression of collagen XVIII during human tooth development. Immunohistochemical analysis with anti-all huXVIII antibody detectedcollagen XVIII protein in the BM zone surrounding the tooth germ at the cap stage of development(a, arrowhead). At the late bell stage, theprotein was also detected in the small blood vessels(b, arrow) and in the BM surrounding the germ(b, arrowhead). Staining was also seen inthe ameloblasts and developing enamel(c, arrowhead), and in the stratum intermedium cells immediately below the ameloblasts. Pepsin digestionintensified collagen XVIII immunostaining in the developing enamel(d, arrowhead). In situ hybridization showed collagen XVIII mRNA expres-sion during the early bell stage in the ameloblasts and stratum intermedium cells(e, arrow) and odontoblasts(e, arrowhead) and in the cells ofthe Hertwig’s root sheath(f, arrowhead). Collagen XVIII mRNA expression was also seen in the mesenchymal preosteoblasts(g, arrowhead).No staining was seen in the in situ sense control(h) or in the non-immunoserum negative control(i). Bars (a), (b), (d)–(i) 100 mm, (c) 50mm. d: dentin, e: enamel, p: pulp.

cap and early bell-stage human teeth(Heikinheimo andSalo, 1995).MMP-20 (enamelysin) was isolated from a porcine

enamel organ cDNA library(Bartlett et al., 1996).Enamelysin transcripts are produced during murine den-tal development at the onset of predentin secretion andafter mineralization in functional odontoblasts(Begue-´Kirn et al., 1998). Enamelysin is also produced in theporcine ameloblasts, which transport enamelysin insecretory vesicles(Fukae et al., 1998). Human ename-lysin has been detected in the developing enamel(Takataet al., 2000) and it has been shown to degrade enamelmatrix protein amelogenin(Llano et al., 1997) and typeIV collagen, but not type I and II collagen(Vaananen¨¨ ¨et al., 2001).In this study MMP-20 is shown to effectively degrade

collagen XVIII. Because MMP-20 expression is almostexclusively restricted to tooth forming cells and has anessential role in normal enamel development(Caterinaet al., 2002; DenBesten et al., 2002), a hypothesis wasset that collagen XVIII and MMP-20 could be co-

expressed in developing human and murine teeth, andin odontogenic tumors.

2. Results

2.1. Collagen XVIII and enamelysin in the developinghuman jaw

In developing human teeth collagen XVIII proteinwas detected weakly in the BM zone surrounding thetooth bud in the cap stage(Fig. 1a) and stronger stainingwas seen in the bell stage(Fig. 1b). At the onset ofmineralization (late bell stage) the protein was alsodetected in the small blood vessels of the pulp(Fig. 1b)and in the developing enamel(Fig. 1c). Pepsin digestionintensified the immunohistochemical staining with anti-all huXVIII in the developing enamel(Fig. 1d), whereasno protein was observed in the forming dentin. In situhybridization detected mRNA expression in ameloblastsand stratum intermedium cells, faint expression in odon-toblasts(Fig. 1e) and dental laminal cells, as well as in

155A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

Fig. 2. Expression of MMP-20 during human tooth development and processing of collagen XVIII by MMP-20. No MMP-20 mRNA expressionwas detected by in situ hybridization during the cap stage(a), but mRNA expression was detected during the late bell stage in odontoblasts(b,arrowhead) and ameloblasts(b, arrow). Higher magnification shows ameloblasts and stratum intermedium cells(b9) and odontoblasts(b0). Inimmunohistochemistry the polyclonal antibody detected the MMP-20 protein in the odontoblasts(c, arrow) and ameloblasts(c, arrowhead). Themonoclonal antibody detected the MMP-20 protein in the developing enamel(d, arrowhead) and dentin(d, arrow). No staining was observedwith the in situ hybridization sense control probes(e), and no staining for protein was detected in the monoclonal non-immune IgG control(f).Western blot analysis revealed that incubation of purified human collagen XVIII with human recombinant MMP-20(1:3 enzyme:substrate molarratio) degraded efficiently the intact molecule into several fragments sized 120 kDa and below(lane 20,(g)); collagen XVIII control wasincubated without MMP-20(lane c,(g)). Bars:(a)–(f) 100mm, (b9) and(b0) 40 mm. a: ameloblasts, cm: condensed mesenchyme, d: dentin, e:enamel, iee: inner enamel epithelium, o: odontoblasts, oee: outer enamel epithelium, p: pulp.

the inner enamel epithelium and in Hertwig’s root sheath(Fig. 1f). In situ hybridization also revealed collagenXVIII expression in the endothelial cells, epidermalkeratinocytes(not shown), and mesenchymal preosteob-lasts (Fig. 1g). Faint expression was detected in theouter enamel epithelium and muscle cells(not shown).No staining was observed in the negative controls usingin situ sense control or non-immune serum(Fig. 1h,i).Serial sections were used to analyze MMP-20 expres-

sion, which was not detected during cap stage or earlybell stage by immunohistochemical staining(not shown)or by in situ hybridization(Fig. 2a). Late bell and latermineralization stages demonstrated MMP-20 mRNAexpression in odontoblasts, ameloblasts and stratumintermedium cells(Fig. 2b). The polyclonal antibodydetected MMP-20 in the odontoblasts and ameloblasts(Fig. 2c), whereas the monoclonal antibody detectedMMP-20 mostly in the developing enamel and in thedentin matrix near the odontoblasts, but not in theodontoblasts or ameloblasts(Fig. 2d). No staining wasobserved in controls that were treated with in situ senseprobe (Fig. 2e) or in the negative controls using non-immune serum for the polyclonal antibody(not shown)or non-immune IgGs for the monoclonal antibody(Fig.2f). Recombinant human MMP-20 was shown todegrade human collagen XVIII effectively into severalfragments with the size of 120 kDa and below(Fig.2g).

2.2. Collagen XVIII in wild type and Col18a1 miceyyy

In four-day old mice jaws, collagen XVIII was presentin the BMs separating the inner enamel epithelium anddental papilla and in the small blood vessels of the pulp(Fig. 3a). With anti-long moXVIII antibody the proteinwas also detected in the ameloblasts and in the devel-oping enamel(Fig. 3b), but not in dentin. Similar resultswere observed with FZ2 and anti-all moXVIII antibodies(not shown). No specific staining was observed in theCol18a1 mice (Fig. 3c,d). Similar collagen XVIIIyyy

expression to wild type mice was detected in MMP-20null mice (not shown). MMP-20 expression was alsoanalyzed in wild type andCol18a1 mice. Theyyy

expression was similar to human teeth and no differenceswere seen in the expression between wild type and nullteeth (not shown). Western blotting performed oncrushed murine enamel showed distinct collagen XVIIIbands at 120, 80 and 56 kDa, and weaker intermediatedegradation product bands between 80 and 56 kDabands, all reactive to both anti-long moXVIII(notshown) and FZ2 antibodies, and a weaker band at 32kDa reactive to FZ2(Fig. 3e).In field emission scanning electron microscope

(FESEM) images, no distinct differences were seenbetween wild type andCol18a1 mice in molar oryyy

incisor enamel or dentin or in molar or incisor crownmorphology (not shown). A slight increase was

156 A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

Fig. 3. Collagen XVIII in wild type andCol18a1 mouse teeth, studied with immunohistochemical staining. The collagen XVIII protein wasyyy

detected by an anti-long moXVIII antibody in wild type mice in the basal membrane separating inner enamel epithelium and dental papilla(a,arrow), as well as surrounding the capillaries in the pulp(a, arrowhead), and in the ameloblasts plus forming enamel(b), but not in the collagenXVIII deficient teeth (c, d). Western blotting with the anti-frizzled moXVIII antibody FZ2 antibody detected 120, 80, 56 and 32 kDa collagenXVIII bands in the extracted enamel of wild type mice(lane e,(e)). The weak bands between 80 and 56 kDa bands most likely represent collagenXVIII intermediate degradation products(e). Bars(a)–(d) 100mm. a: ameloblasts, d: dentin, m: enamel organ, p: pulp.

Fig. 4. The mineral element contents(weight percentage) of lower incisor enamel(a) and dentin(b) in the wild type andCol18a1 mice.yyy

The box blot lines indicate the highest observed value, 75% percentile, median, 25% percentile and the lowest observed value. *indicatesstatistically significant difference from the wild type values(P-0.05; Mann–Whitney test). WT ns7 (4 monthss2, 1 years5); Col18a1yyy

ns6 (4 monthss2, 1 years4).

157A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

Fig. 5. Collagen XVIII and MMP-20 expression in odontomas. Immu-nohistochemical staining with a MMP-20-specific monoclonal anti-body demonstrated the presence of MMP-20 in enamel-like fragmentsin tumor tissue(a, arrowhead). A polyclonal antibody demonstratedthe presence of MMP-20 in ameloblast-like tumor cells(b, arrow-head). With the anti-all huXVIII antibody, the collagen XVIII proteinwas detected in the BM of odontogenic epithelial islands(c, arrow-head) and of small blood vessels(d, arrowhead). Pepsin digestionintensified the collagen XVIII immunostaining in the enamel(e,arrowhead). Polyclonal non-immuneserum control was negative forstaining(f). Bars:(a) 50 mm, (b)–(f) 100mm. e: enamel.

observed in the weight percentage of total mineralcontent, calcium and phosphorus in theCol18a1yyy

mice enamel(Fig. 4a). In dentin, the same phenomenonof lesser magnitude was observed, only the phosphoruscontent demonstrating statistically significant difference(Fig. 4b).

2.3. Collagen XVIII and enamelysin in odontogenictumors

Immunohistochemical analysis of odontomas with theenamelysin-specific monoclonal antibody revealed thepresence of enamelysin only in enamel fragments(Fig.5a). In contrast, with the polyclonal antibody, the enamelfragments were negative and staining was observed inthe tumor cells and ameloblast-like cells near the enamelmatrix (Fig. 5b). There was no staining in other min-eralized tissues or in the extracellular matrix of thetumors assayed. Collagen XVIII was detected in the BMzones surrounding tumor cell islands(Fig. 5c), capillar-ies (Fig. 5d) as well as in the same specimens asenamelysin in the enamel fragments after pepsin diges-tion (Fig. 5e). Staining was also seen in the epithelial

BM (not shown). No staining was observed in thenegative control using non-immune serum(Fig. 5f).

3. Discussion

Enamel is the only mineralized tissue among verte-brates in which collagens are not the major constituent.Recently, collagen X was detected in human enamel andsecretory ameloblasts during enamel maturation, sug-gesting that collagen X may have a role in the regulationof enamel mineralization(Felszeghy et al., 2000).MMP-2, which degrades collagen X, is also expressedby porcine ameloblasts(Birkedal-Hansen, 1995; Caronet al., 2001). Since this study found that MMP-20digests collagen XVIII effectively in vitro, and sinceMMP-20 is crucially important for the normal enameldevelopment(Caterina et al., 2002; DenBesten et al.,2002), we set out to analyze collagen XVIII and MMP-20 expression in the developing teeth and odontogenictumors.Collagen XVIII is expressed as tissue-specific variants

in fetal blood vessels and several BMs(Muragaki et al.,1995; Saarela et al., 1998). It is also expressed incultured human dental pulp stem cells(Shi et al., 2001).Our study demonstrated collagen XVIII expression indeveloping human and mouse tooth germs. It alsodemonstrated that the longest collagen XVIII variantcontaining the frizzled domain is expressed in the BMzone. The findings indicate that collagen XVIII mayhave a role in the pulp capillary vessel development aswell as in the BM zone in the roots. Perivascularcollagen XVIII expression may also indicate the identityof stem cells around the pulp microvasculature(Shi andGronthos, 2003).Odontogenic tumors are mostly benign neoplasms

originating from dental tissues(epithelial, epithelial-mesenchymal and mesenchymal) and matrix proteinsobserved in developing teeth are commonly expressedin these tumors. In odontomas collagen XVIII proteinwas seen in BM structures as well as in enamel-likematrix indicating that the distribution of collagen XVIIIin odontomas was similar to its distribution in thedeveloping human tooth. The absence of expression inameloblast-like tumor cells, contrary to the developingteeth ameloblasts, may reflect the differential degrees ofdifferentiation between the teeth developmental amelob-lasts and odontoma cells.Collagen XVIII protein was also detected in the BMs

surrounding developing teeth and in the ameloblasts ofwild type mice, while no staining was observed incollagen XVIII deficient mice. A slight increase in thetotal mineral content was observed in the enamel ofcollagen XVIII deficient mice, demonstrating that col-lagen XVIII has a functional role in enamel minerali-zation. However, the collagen XVIII deficient mice hadno visible defects in the general dental or in the fine

158 A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

enamel structures, suggesting that collagen XVIII maynot be a prerequisite for murine dental development orthat another molecule effectively substitutes for itsfunction. Since the effect of absence of collagen XVIIIin teeth seems to be limited, the potential clinical effectscannot be speculated at this point. Western blotting ofcrushed mouse enamel with anti-long and anti-frizzledantibodies showed three distinct bands, which representthe fragments of the longest frizzled variant of collagenXVIII molecule. In vitro degradation of recombinanthuman collagen XVIII revealed similar sized bands. Theintact form of the molecule was not detected in theenamel extract, suggesting that collagen XVIII is degrad-ed during enamel mineralization. The frizzled modulesconstitute the extracellular ligand-binding region of friz-zled receptors and mediate signals of Wnt family mem-bers through these receptors. Interestingly, collagenXVIII had similar expression pattern as Wnt-6, a sig-naling protein expressed strongly throughout the budepithelium and in the external and internal enamelepithelium during cap and early bell stages(Sarkar andSharpe, 1999).The onset of MMP-20 expression in developing

human teeth is consistent with the mRNA and proteinexpression patterns determined for porcine and murineenamelysins(Begue-Kirn et al., 1998; Fukae et al.,´1998). However, another study utilizing a monoclonalantibody detected enamelysin only in the enamel duringhuman tooth development but, surprisingly, not in theameloblasts or odontoblasts(Takata et al., 2000). Thesame study also demonstrated MMP-20 expression inthe immature enamel matrix of odontomas, in ghostcells and in globular amyloid masses. The results of ourstudy were similar with a monoclonal antibody, whereasa polyclonal antibody detected MMP-20 expression onlyin the ameloblast-like tumor cells in odontomas, but notin enamel or amyloid masses. These results may beexplained by differences in the tertiary structure ofenamelysin between pre- and post-secreted forms thatallow one or the other antibody to bind.Collagen XVIII mRNA and protein expression were

first observed during cap and bell stages, while MMP-20 expression was seen later, in the late bell stages andafter the onset of mineralization. The temporal relation-ship between the expression of MMP-20 and its substratecollagen XVIII suggests that MMP-20 could be requiredto degrade collagen XVIII prior to mineralization ofenamel. However, lack of MMP-20 in the MMP-20 nullmice did not affect the expression of collagen XVIII,nor did the absence of collagen XVIII affect the expres-sion of MMP-20. Collagen XVIII may thus participatein the organization of enamel before or during mineral-ization, as suggested for collagen X(Felszeghy et al.,2000).Our results show, for the first time, that collagen

XVIII is expressed in the developing human and murine

tooth, specifically in the BMs and in the developingenamel, in which the molecule was co-localized withMMP-20. The findings indicate that collagen XVIII mayparticipate in the enamel matrix maturation or mineral-ization, and they support the role of MMP-20 in collagenXVIII degradation. While it seems that collagen XVIIIis not absolutely necessary for enamel formation, thisstudy does not exclude the possibility that the dentalstructures in collagen XVIII deficient mice might stillbe defective. At least analysis of susceptibility to cariesand tooth wear should be performed before the finalconclusions can be drawn.

4. Experimental procedures

4.1. Tissue samples

The use of human tissues was approved by the EthicalCommittee for the Faculty of Medicine, University ofOulu. The fetal tissue(ns4) was obtained from legalabortions performed at 14, 21 and 23 weeks of gestationand from a stillborn fetus of 30 weeks of gestation. Theodontomas(seven complex and three compound forms)were surgically removed because of the clinical treat-ment indications. The use of mice was approved by theEthical Committee on Animal Experiments of the Uni-versity of Oulu. The type XVIII collagen null mice(Col18a1 , Fukai et al., 2002) were a kind gift fromyyy

Dr Naomi Fukai and Dr Bjorn R. Olsen(HarvardMedical School, Boston, MA, USA). All mice usedwere C57BLy6J inbred. The mouse jaws(4 days old,ns4 wild type, ns4 Col18a1 ) were embedded inyyy

Jung Tissue freezing medium(Leica Instruments GmbH,Nussloch, Germany) and frozen. The sections were fixedat y20 8C with acetone. MMP-20 null mice(4 daysold, ns4 wild type, ns4 MMP-20 null) have beendescribed earlier(Caterina et al., 2002). The lowerhuman fetal jaws were fixed in 4% paraformaldehyde(PFA) in PBS(pH 7.2) for 2–4 days, demineralized in0.5 M EDTA (pH 7.4) for 7–10 days, and embeddedin paraffin wax. The odontomas and wild type andMMP-20 null mice were fixed in 4% PFA and embeddedin paraffin. The sections were cut 6mm thick, placedon glass slides, and kept overnight at 378C.

4.2. Antibodies

The polyclonal antibody for MMP-20 was porcinerecombinant IgGs(Vaananen et al., 2001). The mono-¨ ¨ ¨clonal antibody C7 for MMP-20 was from Fuji ChemicalIndustries(Toyama, Japan). The collagen XVIII humanpolyclonal antibody anti-all huXVIII has been reportedearlier (Saarela et al., 1998). MMP-20 hinge regionantibody was from Sigma(St. Louis, MO, USA).A polyclonal collagen XVIII antibody anti-all mo-

XVIII (ELQ) was generated against PCR-amplified

159A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

fragment corresponding to the common region of theN-terminal NC1 domain of all mouse collagen XVIIIvariants and a polyclonal antibody anti-long moXVIII(Q36.4) was generated against PCR-amplified fragmentfrom the common region of the long NC1 domain ofmouse collagen XVIII(Saarela et al., unpublished obser-vations). The PCR products were cloned into the pQE-41 vector (Qiagen, Inc., Valencia, CA, USA), fromwhich one can express in-frame dihydrofolate reductase(DHFR) fusion proteins with an amino-terminal histi-dine tag. Fusion proteins were expressed according tothe protocol suggested by Qiagen, and purified byimmobilized metal affinity chromatography(ProBond,Invitrogen, Carlsbad, CA, USA). Antibodies to thefusion proteins were raised by the conventional method.For immunization, 300mg of the purified fusion proteinwas injected into rabbits subcutaneously with completeFreund’s adjuvant followed by booster injections withincomplete Freund’s adjuvant at intervals of 14 days.Type XVIII collagen-specific antibodies were affinity-purified on columns with the recombinant NC1 poly-peptides, which were expressed from the pQE-31 vector(Qiagen) without the DHFR leader and then coupled toCNBr-activated Sepharose 4B(Amersham PharmaciaBiotech, Little Chalfont, Buckinghamshire, England) asdescribed in detail elsewhere(Saarela et al., 1998).The anti-frizzled moXVIII antiserum(FZ2) was

raised against the synthetic peptide ARSSQQHRH-SDVHSD(Innovagen, Lund, Sweden), derived from thelongest N-terminal frizzle variant of mouse collagenXVIII. The antigen was prepared by coupling the peptideto the keyhole limpet hemocyanin(Sigma, St. Louis,MO, USA) by a standard procedure(Harlow and Lane,1988), after which it was used to raise polyclonalantibodies in rabbits as described above. The samepeptide that was used as an antigen was used to preparean affinity-column for the purification of the antibodyas described earlier(Saarela et al., 1998).The specificities of anti-all moXVIII and anti-long

moXVIII antibodies were tested by Western blottingagainst recombinant proteins ELQ and Q36.4 withoutDHFR, and against adult mouse liver and kidney tissueextracts. The specificity of anti-frizzled moXVIII anti-serum was verified by Western blotting against thefrizzled variant of mouse type XVIII collagen expressedin EBNA 293 cells from vector pREP-7(Invitrogen,Carlsbad, CA, USA).

4.3. Immunohistochemical staining

All washes were performed twice in PBS. Paraffinsections were treated with xylene three times for 5 minand rehydrated with ethanol. The tissue sections wereincubated in 0.3% H O in methanol for 3 h(30 min2 2

for frozen sections) to block the endogenous peroxidaseactivity. The paraffin sections were then microwaved in

0.01 M EDTA (MMP-20), in 0.01 M citrate(anti-allhuXVIII ) or alternatively digested for 1 h in 0.8%pepsin in H O(anti-all huXVIII) at 37 8C and washed.2

Non-specific binding was blocked with 2% normal goatserum(polyclonal) or horse serum(monoclonal) for 30min at room temperature(RT). Thereafter the sectionswere incubated with the primary antibody(diluted inPBS with 1% BSA) overnight at 4 8C in a humidchamber. Sections incubated with non-immune rabbit(polyclonal) or non-immune mouse(monoclonal) IgGsinstead of primary antibodies were used as negativecontrols. The sections were incubated with biotinylatedsecondary antibody solution(Anti-Rabbit IgGs for poly-clonal and Anti-Mouse IgGs for monoclonal, VectorLaboratories, Burlingame, CA, USA) for 1 h at RT.After washes the sections were treated with VectastainElite ABC reagent(Vector Laboratories) for 30 min atRT. The tissue sections were stained with AEC(Zymed,San Francisco, CA, USA) for 10 min according to thekit instructions. The sections were counterstained withMayer’s haematoxylin(Histolab Products AB, Gote-¨borg, Sweden). Tissues derived fromCol18a1 miceyyy

were used as negative controls for mouse tissues.

4.4. In situ hybridization

A 374-bp cDNA fragment encoding a portion ofhuman enamelysin(1238–1613 bp) was cloned intopBluescript-vector and linearized for probe production(Vaananen et al., 2001). A 272 bp-fragment detecting¨ ¨ ¨all variants of collagen XVIII was subcloned to plasmidBluescript SK and linearized(Saarela et al., 1998). Thedigoxigenin-11-UTP-labeled(Boehringer Mannheim,Mannheim, Germany) antisense and sense RNA probeswere prepared with RNA polymerases.In situ hybridization for paraffin sections was per-

formed as described by Parikka et al.(2001). Thehybridization took place at 548C overnight and afterwashes the hybridized DIG-labeled probe was detectedwith the antibody(100 mM Tris–HCl, 150 mM NaCl,pH 7.5, with 0.1% Triton X-100, 1% normal goat serum,1:200 anti-digoxigenin-alkalic phosphatase Fab frag-ments, Roche Diagnostics, Mannheim, Germany). TheAP-conjugated antibody was detected by Fast Red colorsubstrate(Boehringer Mannheim, Mannheim, Germany)and the sections were counterstained with Mayer’shaematoxylin(Histolab Products AB).

4.5. Western blotting

First mandibular molars from 4–5-day-old mice weredissected and soft tissue was removed. The mineralizedmaterial was placed in gel loading buffer containingDTT and proteins were allowed to elute overnight at 48C. The samples were heated at 95–1008C for 5 minand then loaded onto a 10% SDS-PAGE gel. After

160 A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

electrophoresis, the proteins were transferred to Immo-bilon P membrane(Millipore, Bedford, MA, USA). Themembrane was treated with 5% non-fat milk for 1 h,and incubated with the primary antibody at RT over-night. The membrane was washed and incubated withanti-rabbit secondary antibody(1:1000, DAKO AyS,Glostrup, Denmark) for 1 h at RT. After washes themembrane was incubated with ABComplexyHRP(1:1000, DAKO AyS, Glostrup, Denmark) for 1 h. Themembrane was treated with ECL Western blotting detec-tion reagent for 1 min and then exposed to Hyperfilm-ECL (Amersham Pharmacia Biotech).

4.6. Processing of human collagen XVIII by MMP-20

Human hepatoblastoma cell line HepG2(HB8065)was purchased from the American Type Culture Collec-tion (ATCC). The cells were maintained in Dulbecco’smodified Eagles’s medium(Gibco BRL, Life Technol-ogies Inc., Grand Island, NY, USA) supplemented with10% fetal bovine serum, 2 mML-glutamine and 1 mMsodium pyruvate. The 90% confluent cells were washedand then cultured for 72 h in serum-free culture medium.Collagen XVIII was purified from the medium withheparin Sepharose column(Amersham Pharmacia Bio-tech) and concentrated by ultracentrifugation(Ultrafree,MWCO 30 kDa, Millipore). The buffer was changed to50 mM Tris–HCl, pH 7.8, 0.2 M NaCl, 1 mM CaCl2

with a desalting chromatography column(Bio-Rad,Richmond, CA, USA). The total protein concentrationwas measured with the Roti-Quant protein assay(CarlRoth GmbHqCo, Karlsruhe, Germany). Complete�EDTA-free protease inhibitor cocktail(BoehringerMannheim) was used during the purification to preventproteolysis.The human collagen XVIII(total protein 8mg) was

incubated with active human recombinant MMP-20(Llano et al., 1997) (molar enzymeysubstrate ratio 1:3)for 24 h at 378C in 50 mM Tris–HCl, pH 7.8, 0.2 MNaCl, 1 mM CaCl . The control collagen XVIII sample2

with no MMP-20 was incubated in the same buffer forthe same time. SDS sample buffer containingb-mercap-toethanol was added to terminate the digestion. Thecleavage products were separate on a SDS-PAGE geland transferred to a nitrocellulose membrane(Protran,Schleicher and Scuell, Keene, NH, USA). Themembrane was probed with human collagen XVIIIantibody anti-all huXVIII, washed and then treated witha horseradish peroxidase-conjugated goat anti-rabbitantibody(Bio-Rad). The immunoreactive proteins werevisualized with ECL Western blotting detection reagent.

4.7. Electron microscopy

Field emission scanning electron microscope(FESEM) (Jeol JSM-6400, Jeol, Tokyo, Japan)

equipped with energy dispersive(EDS) analyzer INCA(Oxford Instruments, Oxfordshire, UK) for the sampleelement analysis, was used for the electron microscopyat the Institute of Electron Optics(University of Oulu,Finland). Accelerating voltage was 15 kV.Lower jaws of 4 month and 1-year-oldCol18a1yyy

mice and wild type mice(ns4 in all groups) wereextracted, defleshed and dehydrated in ethanol(50,y70, y94 andy100%, each for a day). The jawswere attached to a base with carbon glue, polished andcovered with Pt-palladium to evaluate the crown mor-phology with FESEM. Also, the lower jaws(ns4 inall groups) were cut vertically in bucco-lingual directionimmediately in front of molar teeth and molars were cuttransversally, to expose the incisor enamel, dentin andpulp tissue cross-section to analyze the enamel anddentin structures. The sections were embedded in EpofixResin(Epofix, Struers, Denmark), polished and coveredwith Pt-palladium for FESEM analysis. Incisor sectionsof 4 month and 1-year-old mice(ns7 in wild type (4monthss2, 1 years5) and ns6 in Col18a1 miceyyy

(4 monthss2, 1 years4)) were also used for themineral element analysis with EDS using the sameprotocol except that the samples were coated withcarbon. Five spots representing the body of enamel ordentin were used for the measurements from each tooth.

4.8. Statistical analysis

The statistical analysis for the differences in mineralelements and total mineral content was done with SPSSfor Windows Release 11.5.1(SPSS Inc., Chicago, IL,USA). Since the data did not meet the assumption ofhomogeneity of variance, non-parametric Mann–Whit-ney test was used for the analysis. The level of thestatistical significance was set atP-0.05.

Acknowledgments

We thank Ms Merja Tyynismaa, Ms Eeva-MaijaKiljander and Ms Maritta Harjapaa for their expert¨ ¨technical help. This work was supported by grants fromthe Academy of Finland(including grant�104 338),the Finnish Dental Society Apollonia, the Cancer Soci-ety of Northern Finland, and the Oulu University Hos-pital EVO-grants. J.D. Bartlett and Y. Ding weresupported by NIDCR grant�DE13237. L. Eklund, R.Heljasvaara and T. Pihlajaniemi were supported by theEuropean Commission(QLK3-2000-00084) and theFinnish Centre of Excellence Programme(2000–2005)of the Academy of Finland.

References

Bartlett, J.D., Simmer, J.P., 1999. Proteinases in developing dentalenamel. Crit. Rev. Oral Biol. Med. 10, 425–441.

161A. Vaananen et al. / Matrix Biology 23 (2004) 153–161¨ ¨ ¨

Bartlett, J.D., Simmer, J.P., Xue, J., Margolis, H.C., Moreno, E.C.,1996. Molecular cloning and mRNA tissue distribution of a novelmatrix metalloproteinase isolated from porcine enamel organ. Gene183, 123–128.

Begue-Kirn, C., Krebsbach, P.H., Bartlett, J.D., Butler, W.T., 1998.´Dentin sialoprotein, dentin phosphoprotein, enamelysin and ame-loblastin: tooth-specific molecules that are distinctively expressedduring murine dental differentiation. Eur. J. Oral Sci. 106, 963–970.

Birkedal-Hansen, H., 1995. Proteolytic remodeling of extracellularmatrix. Curr. Opin. Cell Biol. 7, 728–735.

Caron, C., Xue, J., Bartlett, J.D., 1998. Expression and localizationof membrane type 1 matrix metalloproteinase in tooth tissues.Matrix Biol. 17, 501–511.

Caron, C., Xue, J., Sun, X., Simmer, J.P., Bartlett, J.D., 2001.Gelatinase A(MMP-2) in developing tooth tissues and amelogeninhydrolysis. J. Dent. Res. 80, 1660–1664.

Caterina, J.J., Skobe, Z., Shi, J., Ding, Y., Simmer, J.P., Birkedal-Hansen, H., et al., 2002. Enamelysin(matrix metalloproteinase20)-deficient mice display an amelogenesis imperfecta phenotype.J. Biol. Chem. 277, 49 598–49 604.

DenBesten, P.K., Yan, Y., Featherstone, J.D.B., Hilton, J.F., Smith,C.E., Li, W., 2002. Effects of fluoride on rat dental enamel matrixproteinases. Arch. Oral Biol. 47, 763–770.

Felszeghy, S., Hollo, K., Modis, L., Lammi, M.J., 2000. Type Xcollagen in human enamel development: a possible role in miner-alization. Acta Odontol. Scand. 58, 171–176.

Fukae, M., Tanabe, T., Uchida, T., Lee, S.K., Ryu, O.H., Murakami,C., et al., 1998. Enamelysin(matrix metalloproteinase-20): locali-zation in the developing tooth and effects of pH and calcium onamelogenin hydrolysis. J. Dent. Res. 77, 1580–1588.

Fukai, N., Eklund, L., Marneros, A.G., Oh, S.P., Keene, D.R.,Tamarkin, L., et al., 2002. Lack of collagen XVIIIyendostatinresults in eye abnormalities. EMBO J. 21, 1535–1544.

Harlow, E., Lane, D.P., 1988. Antibodies: a Laboratory Manual. ColdSpring Harbor Laboratory, Cold Spring Harbor, NY.

Heikinheimo, K., Salo, T., 1995. Expression of basement membranetype IV collagen and type IV collagenases(MMP-2 and MMP-9)in human fetal teeth. J. Dent. Res. 74, 1226–1234.

Llano, E., Pendas, A.M., Knauper, V., Sorsa, T., Salo, T., Salido, E.,¨et al., 1997. Identification and structural and functional character-ization of human enamelysin(MMP-20). Biochemistry 36,15 101–15 108.

Lukinmaa, P.L., Vaahtokari, A., Vainio, S., Sandberg, M., Waltimo,J., Thesleff, I., 1993. Transient expression of type III collagen by

odontoblasts: developmental changes in the distribution of pro-alpha 1(III ) and pro-alpha 1(I) collagen mRNAs in dental tissues.Matrix 13, 503–515.

Muragaki, Y., Timmons, S., Griffith, C.M., Oh, S.P., Fadel, B.,Quertermous, T., et al., 1995. Mouse Col18a1 is expressed in atissue-specific manner as three alternative variants and is localizedin basement membrane zones. Proc. Natl. Acad. Sci. USA 92,8763–8767.

Myllyharju, J., Kivirikko, K.I., 2001. Collagens and collagen-relateddiseases. Ann. Med. 33, 7–21.

Nagase, H., Woessner Jr, J.F., 1999. Matrix metalloproteinases. J.Biol. Chem. 274, 21 491–21 494.

Palosaari, H., Tasanen, K., Risteli, J., Larmas, M., Salo, T., Tjader-hane, L., 2001. Baseline expression and effect of TGF-beta 1 ontype I and III collagen mRNA and protein synthesis in humanodontoblasts and pulp cells in vitro. Calcif. Tissue Int. 68, 122–129.

Parikka, M., Kainulainen, T., Tasanen, K., Bruckner-Tuderman, L.,Salo, T., 2001. Altered expression of collagen XVII in ameloblas-tomas and basal cell carcinomas. J. Oral Pathol. Med. 30, 589–595.

Rehn, M., Pihlajaniemi, T., 1995. Identification of three N-terminalends of type XVIII collagen chains and tissue-specific differencesin the expression of the corresponding transcripts. The longestform contains a novel motif homologous to rat and Drosophilafrizzled proteins. J. Biol. Chem. 270, 4705–4711.

Saarela, J., Rehn, M., Oikarinen, A., Autio-Harmainen, H., Pihlaja-niemi, T., 1998. The short and long forms of type XVIII collagenshow clear tissue specificities in their expression and location inbasement membrane zones in humans. Am. J. Pathol. 153,611–626.

Sarkar, L., Sharpe, P.T., 1999. Expression of Wnt signalling pathwaygenes during tooth development. Mech. Dev. 85, 197–200.

Shi, S., Gronthos, S., 2003. Perivascular niche of postnatal mesen-chymal stem cells in human bone marrow and dental pulp. J. BoneMiner. Res. 18, 696–704.

Shi, S., Robey, P.G., Gronthos, S., 2001. Comparison of human dentalpulp and bone marrow stromal stem cells by cDNA microarrayanalysis. Bone 29, 532–539.

Takata, T., Zhao, M., Uchida, T., Wang, T., Aoki, T., Bartlett, J.D.,et al., 2000. Immunohistochemical detection and distribution ofenamelysin(MMP-20) in human odontogenic tumors. J. Dent.Res. 79, 1608–1613.

Vaananen, A., Srinivas, R., Parikka, M., Palosaari, H., Bartlett, J.D.,¨ ¨ ¨Iwata, K., et al., 2001. Expression and regulation of MMP-20 inhuman tongue carcinoma cells. J. Dent. Res. 80, 1884–1889.