Archives of Oral Biology (2006) 51, 996—1005

www.intl.elsevierhealth.com/journals/arob

ADAM28 participates in the regulationof tooth development

Zheng Zhao a,b,c, Ling-Ying Wen a, Ming Jin b,Zhi-Hong Deng d, Yan Jin b,c,*

aDepartment of Paediatric Dentistry, College of Stomatology, Fourth Military Medical University,Xi’an, Shaanxi 710032, ChinabResearch and Development Centre for Tissue Engineering, Fourth Military Medical University,Xi’an, Shaanxi 710032, ChinacDepartment of Oral Histology and Pathology, College of Stomatology,Fourth Military Medical University, Xi’an, Shaanxi 710032, ChinadDepartment of Otolaryngology, Xijing Hospital, Fourth Military Medical University, Xi’an,Shaanxi 710032, China

Accepted 24 May 2006

KEYWORDSADAM28;Polyclonal antibody;Tooth germdevelopment;Western blot;Immunohistochemistry

Summary Disintegrin and metalloprotease (ADAM) proteins are a family of mem-brane-anchored glycoproteins with diverse functions in fertilisation, development,neurogenesis and protein ectodomain shedding. ADAM28 is a newly discoveredmember of the ADAM family in humans and murine with autocatalytic activity.Recently, the authors screened ADAM28 genes from patients with congenital hypo-plasia of tooth root, and studied the relationship between ADAM28 and toothdevelopment. A polyclonal antibody (pAb) against ADAM28 was preparared, andthe expression and localisation of ADAM28 were detected in tooth germ and dentalmesenchymal cells. The results indicated that the prokaryotic expression vectorpGEX-4T-ADAM28 was constructed successfully. Glutathione S-transferase-ADAM28fusion protein was generated after inducement by isopropylthio-b-D-galactosideand isolated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis. Thepurified fusion protein was used as an antigen for production of antibody. Western blotand enzyme-linked immunosorbent assay analyses verified that the antibody had ahigh specificity and titre. Immunohistochemistry and reverse transcriptase-polymer-ase chain reaction showed that ADAM28 was expressed at each stage of tooth germdevelopment at different levels. Moreover, it was expressed in human dental follicle

Abbreviations: ADAM, a disintegrin and metalloprotease; HDFCs, human dental follicle cells; HDPCs, human dental papilla cells;HDPSCs, human dental pulp stem cells; HPDLCs, human periodontal ligament cells; HDCLECs, human dental cervical loop epithelial cells;GST, glutathione S-transferase; SDS, sodium dodecyl sulphate; PAGE, polyacrylamide gel electrophoresis; ELISA, enzyme-linkedimmunosorbent assay; RT, reverse transcriptase; PCR, polymerase chain reaction; kDa, kilodalton; PBS, phosphate-buffered saline* Corresponding author. Tel.: +86 29 8477 6147; fax: 86 29 83218039.E-mail addresses: yanjin@fmmu.edu.cn, zz1700@fmmu.edu.cn (Y. Jin).

0003–9969/$ — see front matter # 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.archoralbio.2006.05.010

ADAM28 participates in the regulation of tooth development 997

cells, human dental papilla cells, human dental pulp stem cells, human periodontalligament cells and human dental cervical loop epithelial cells at transcription level. Inconclusion, it is reasonable to suggest that ADAM28 may participate in tooth devel-opment and the regulation of odontogenic mesenchymal cells through progressivereciprocal inductive interactions between the epithelium and the mesenchyme.# 2006 Elsevier Ltd. All rights reserved.

Introduction

Disintegrin and metalloprotease (ADAM) proteinsare involved in several important biological pro-cesses, including neurogenesis, fertilisation, muscledevelopment and release of membrane-anchoredproteins, such as tumour necrosis factor a, trans-forming growth factor a and L-selectin, from theplasma membrane.1—4 ADAMs consist of an N-term-inal signal sequence, followed by a pro-domain, ametalloprotease domain (MP), a disintegrin-likedomain, a cysteine-rich domain, an epidermalgrowth factor (EGF)-like domain, a transmembranedomain and a cytoplasmic domain.5,6

There are 30 known ADAMs, and ADAM28 isthought to be one of the family members withautocatalytic activity.7 From protein sequence com-parisons, ADAM28 is more closely related to snakevenom metalloproteases (SVMPs) than to otherADAMs, and hence may cleave similar substratesto SVMPs, perhaps including components of theextracellular matrix.8 Furthermore, MP activity isof functional importance for extracellular matrixmodelling and in the ectodomain processing ofmolecules such as heparin-binding EGF-like growthfactor. In addition, the disintegrin-like domains areknown to be involved in cell—cell adhesion eventsthat are crucial for odontogenesis.4 Recent studieshave shown that ADAM28 could promote prolifera-tion of human breast cancer cells by IGF signalling,9

and that the IGF system has a fundamental role inprotecting cells from apoptosis. This suggests thatADAM28 may be closely associated with organogen-esis, cell proliferation and differentiation, and for-mation of the extracellular matrix.

Teeth develop as a result of sequential and reci-procal interactions between the oral ectoderm andthe neural-crest-derived mesenchyme. These inter-actions gradually transform the tooth primordia intomineralised structures with various cell types,among which epithelial-derived ameloblasts andmesenchyme-derived dental papilla cells (odonto-blasts) and dental follicle cells (cementoblasts)synthesise and secrete the organic components ofthe hard tooth tissues (enamel, dentin and cemen-tum, respectively).10—12

The regulatory gene of tooth development con-trols advancing morphogenesis and cell differentia-

tion13 by sequential inductive interactions betweenthe epithelium and the mesenchyme. Many growthfactors, transcription factors, cellmembrane surfacecomponents and extracellular matrix molecules areinvolved in these processes. However, little is knownregarding the role of ADAM28 in tooth development.

Thus, to investigate the correlation betweenADAM28 and tooth development, the authors pre-pared a polyclonal antibody (pAb) against ADAM28and detected its expression in tooth germ andmesenchymal cells at transcription and proteinlevels. These data will provide a basis for furtheranalysis of the functional mechanisms of ADAM28gene in tooth development.

Materials and methods

Animal and tissue preparation

Embryos were obtained by mating male and femaleBalb/c mice (Experimental Animal Centre, FourthMilitary Medical University, China). Embryonic agewas determined according to the day of appearanceof the vaginal plug (E0) and was confirmed by mor-phological criteria, while the day of birth was takenas day 0 of postnatal development (P0).

Pregnant females (E13-19d) and P1-12d femaleswere killed by cervical dislocation. The heads includ-ing the tooth germs or primordia were dissected in0.1 M phosphate-buffered saline (PBS, pH 7.4), fixedin ice-cold 4% paraformaldehyde for 24 h at 4 8C, andthen decalcified for 2 days (E18-P2d) and 7 days (P3-12d) with ethylene diamietetracetic acid (EDTA), de-hydrated through graded alcohols, embedded in par-affin, sectionedat3 mmandbakedat60 8Covernight.The protocol was approved by the Research EthicsCommittee of Fourth Military Medical University.

Two New Zealand rabbits (males, 2—2.5 kg) wereobtained from the Experimental Animal Centre,Fourth Military Medical University.

Cell culture

Human odontogenic mesenchymal cells include den-tal papilla cells (HDPCs), dental follicle cells (HDFCs),dental pulp stem cells (HDPSCs) and periodontalligament cells (HPDLCs). HDPCs, HDFCs and human

998 Z. Zhao et al.

dental cervical loop epithelial cells (HDCLECs) weredissected from the lower first deciduous molar toothgerms of 6-month-old human fetal cadaver tissues.(The patients gave informed consent and donationswere voluntary.) HDPSCs and HPDLCs were isolatedfrom the first bicuspids of the lower jaw, uprooteddue to orthodontic treatment. Primary cells under-went enzyme digestion, and were digested in col-lagenase 625 U/mL, collected by centrifugation andcultured in Dulbecco’s modified eagle medium(Gibco, USA) containing 10% (v/v) fetal bovine serum(Gibco, USA). Fourth passage cells were used for theexperiments. HDPSCs were obtained by magneticbead screening. The protocol was examined andapproved by the Ethics Committee on HumanResearch of Fourth Military Medical University.

RNA extraction, reverse transcription andpolymerase chain reaction

The first deciduous molar tooth germs of the lowerjaw from 6-month-old human fetal cadaver tissueswere collected into trizol reagent (Invitrogen, USA)for RNA extraction with a PicoPure RNA extractionkit (Clontech, USA) according to the manufacturer’sprotocol, with modifications as published pre-viously.14 The RNA concentration was measured bymeans of a spectrophotometer at A260, with A260/A280 > 1.9. Reverse transcription was performedwith Superscript II (Invitrogen, USA) according tothe manufacturer’s instructions. Two micrograms ofRNA were reverse transcribed into 20 mL cDNA witholigo dTas the primer. The protocol was approved bythe Ethics Committee on Human Research of theFourth Military Medical University.

Polymerase chain reaction (PCR) primer seque-nces were as follows: human ADAM28 (GenBankNM_014265) (amino acids 2119—2373, product255 bp).

Upstream primer: 50-CGC GGATCC CCA GAG AAAAGC AGA AGA AA-30 (underline represents BamHI-site, CGC represents protection bases) (amino acids2119—2138).

Downstream primer: 50-CCG CTCGAG ATG CTTTTG GAT TTG AGT CC-30 (underline represents XhoI-site, CCG represents protection bases) (amino acids2373—2354).

The PCR was performed with 2 mL of cDNA, gene-specific primers (100 pmol/L), buffer, dNTP and TaqDNA polymerase in 50 mL vol., at 95 8C for 3 min(95 8C for 30 s, 60 8C for 30 s, 72 8C for 30 s) for30 cycles of amplification and then by extension at72 8C for 6 min. The PCR product was separated by1% agarose gel electrophoresis, visualised with ethi-dium bromide staining, and purified by NucleoTrapGel Extraction Kit (Clontech, USA).

Construction and identification of ADAM28prokaryotic expression vector and fusionexpression in Escherichia coli

The 255-bp PCR product was cloned into the corre-sponding sites of a pMD18-T vector (Takara, Japan) toconstruct pMD18-T-ADAM28. The positive cloneswere sequencedonboth strands after transformationinto Escherichia coli DH5a. DNA homology searcheswere performed on the National Centre for Biotech-nology Information (NCBI) server using the BLAST(Basic Local Alignment Search Tool) 2.0 Program.After digestion with BamHI and XhoI, the ADAM28fragmentwas ligated into the corresponding sites of apGEX-4T-1 vector (Amersham Pharmacia, USA) toconstruct pGEX-4T-ADAM28. PGEX-4T-ADAM28 thenunderwent restriction enzyme digestion, PCR identi-fication and DNA sequencing identification. Afterinducement by isopropylthio-b-D-galactoside in E.coli, glutathione S-transferase (GST) fusion proteinswere produced. GST-ADAM28 fusion proteins werefurther isolated by 12% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) andused as antigens for production of rabbit polyclonalantibodies (pAb) as described previously.15,16

Preparation and identification of pAbagainst ADAM28

Acrylamide gels containing fusion proteins(35.3 kDa) were frozen in liquid nitrogen, fullyground and confluenced with Freund’s incompleteadjuvant, and injected intradermally into theamphi-spine, armpit and inguinal lymph nodes ofeach rabbit at a dose of 300—350 mg once a week for4 weeks. Immune antisera were obtained by punc-turing hearts 1 month later. All the antibodies werepurified by the corresponding proteins using theAminoLink Immobilization kit (Pierce, USA) accord-ing to the manufacturer’s standard instructions.

The specificity and titre of pAb were determinedaccording to Western blotting (Chemicon Interna-tional, Temecula, CA, USA) and enzyme-linkedimmunosorbent assay (ELISA) standard protocols.The pAb was diluted 1:200, 1:500, 1:1000, 1:5000and 1:10,000 and the non-immune rabbit serum wasused as a control.

Expression and localisation of ADAM28 inmurine tooth germ and humanodontogenic mesenchymal cells

Immunohistochemistry (LsABC Immunohistochemis-try Kit, Dako, USA) staining was performed asdescribed previously.17 The pAb against ADAM28was diluted 1:200. Control experiments were per-

ADAM28 participates in the regulation of tooth development 999

formed by substituting the pAb with non-immunerabbit serum.

Fourth passage cells (5 � 107) were harvested,centrifugated, washed twice with PBS and thensubjected to reverse transcriptase-polymerasechain reaction (RT-PCR) procedures as describedabove (product 255 bp).

Image analysis

Usingmicroscopy, cells were enlarged 3.3 � 40 timesusin an HPIAS-1000 high-acuity image analysis system(China). Ten non-overlapping cytoplasm regions ofeach section were selected at random and measuredaccording to the same standard, and the averagegrayscale (x̄� s) of each section was obtained. Theresults were subjected to statistical analysis.

Statistical analysis

The results were expressed as means � standarddeviations. Statistical significance was assessed

Figure 1 The identification of pGEX-4T-ADAM28, induced exantibody (pAb) against ADAM28. (A) The construction and idenchain reaction (PCR) identification of pGEX-4T-ADAM28. Lanesfragments of approximately 5000 and 255 bp were seen after dbp DNA ladder. Lane M2, 2000-bp DNA ladder. (B) Sodium dodecGST-ADAM28 fusion protein. A new protein band appeared neaprotein ladder. Lanes 1—5, the samples induced by isoprorespectively. Lane 6, pGEX-4T-1 sample induced by IPTG fothe specificity and titre for pAb against ADAM28 by Western bl4T-1, while pAb against ADAM28 was diluted 1:1000. Lane 2, pwhile non-immune rabbit serum (1:1000 dilution) replaced pAbexpressing pGEX-4T-ADAM28, ADAM28 pAb with dilutions of 1:2M, protein ladder.

using the independent sample t-test in SPSS Version12.0 (SPSS Inc., Chicago, IL, USA). A P value < 0.05was considered to be statistically significant.

Results

Identification of ADAM28 prokaryoticexpression vector, induced expression offusion protein and identification of pAb

RT-PCR results showed that a 255-bp fragment cor-responding to ADAM28 mRNA (amino acids 2119—2373) was cloned (Fig. 1A). Two fragments ofapproximately 5000 and 255 bp were seen afterpGEX-4T-ADAM28 was digestedwith BamHI and XhoI,and a 255-bp fragment was obtained by PCR identi-fication (Fig. 1A). Basic local alignment searchesagainst sequence databases indicated that the255-bp insert was human ADAM28 (NM_014265).Its open reading frame was complete and no muta-tions were found.

pression of fusion protein and identification of polyclonaltification of pGEX-4T-ADAM28. Lanes 1 and 2, polymerase3 and 4, pGEX-4T-ADAM28 digested by BamHI and XhoI; twoigestion. Lane 5, PCR product of ADAM28. Lane M1, 15,000-yl sulphate-polyacrylamide gel electrophoresis analysis ofr 36 kDa. Five hours of inducement was optimal. Lane M,pylthio-b-D-galactoside (IPTG) for 5, 4, 3, 2 and 1 h,r 5 h. Lane 7, uninduced sample. (C) Determinations ofot. Lane 1, protein sample from bacteria expressing pGEX-rotein sample from bacteria expressing pGEX-4T-ADAM28,against ADAM28. Lanes 3—7, protein sample from bacteria00, 1:500, 1:1000, 1:5000 and 1:10,000, respectively. Lane

1000 Z. Zhao et al.

Figure 2 Expression of ADAM28 in murine tooth germ and human odontogenic mesenchymal cells. oee, outer enamelepithelium; iee, inner enamel epithelium; sr, stellate reticulum; oe, oral epithelium; dpc, dental papilla cells; dfc,dental follicle cells; mc, mesenchymal cells; bm, basement membrane; ab, ameloblasts; em, enamel matrix; ers,epithelial root sheath; ob, odontoblasts; pd, predentins; ek, enamel knot. (A) Expression of ADAM28 in murine toothgerm. (a) At the bud stage (E13d), the enamel organ appeared to be negatively stained, while oral epithelia and themesenchymal cells were positively stained (immunohistochemistry, original magnification 400�). (b) Negative staining.(c) At the cap stage (E16d), strongly positive staining was found in oral epithelia, stellate reticulum cells of the enamelorgan, basement membrane, dental papilla and dental follicle. Weakly positive staining was found in the enamel knot.Negative staining was found in the outer and inner enamel epithelium (original magnification 100�). (d) Negativestaining. (e) At the early bell stage (E18d), strongly positive staining was found in oral epithelia and weakly positivestaining was found in stellate reticulum cells. The other samples stained negatively (original magnification 100�). (f)Negative staining. (g) At the late bell stage (P2d), positive staining was found in ameloblasts, enamel matrix, Hertwig’sepithelial root sheath and dental papilla cells (original magnification 100�). (h) Negative staining (original magnification40�). (i) At the crown development stage (P6d), ADAM28 was positive in ameloblasts, odontoblasts and dental papillacells (original magnification 40�). (j) Negative staining. (k) At the root development stage (P12d), positive staining for

ADAM28 participates in the regulation of tooth development 1001

Figure 2. (Continued ).

Induced expression of fusion protein was dis-played by 12% SDS-PAGE (Fig. 1B). A new proteinband was found near 36 kDa (35.3 kDa). The relativemolecular weight of ADAM28 coding protein was9.3 kDa, and that of GST protein expressing pGEX-4T-1 was 26 kDa. GST-ADAM28 fusion protein wasproduced from recombinant pGEX-4T-ADAM28 andits relative molecular weight was 35.3 kDa.

Fusion protein was expressed after 1 h of induce-ment and appeared to increase progressively withtime; most expression was found after 5 h of indu-cement so 5 h was considered to be the optimaltime. Laser optical density scanning showed thatinduced protein accounted for 38% of the totalprotein. Bacteria were collected after 5 h of indu-

ADAM28 was detected in ameloblasts, odontoblasts, predentindental follicle cells (original magnification 40�). (l) Negativemesenchymal cells. Strongly positive staining was found in thcells (HDPCs), positive staining was found in human dental fo(HPDLCs), andweakly positive staining was found in human den(C) Reverse transcriptase-polymerase chain reaction detectionof 255 bp was seen in HDPCs, HDFCs, HPDLCs, HDPSCs and H

cement. Analyses of the concentration fluid anddeposit by SDS-PAGE indicated that the expressionproducts of GST-ADAM28 fusion protein were mainlyinclusions.

The purified antibodies were raised against GST-ADAM28 fusion protein and the specificity andtitre were determined by Western blot (Fig. 1C)and ELISA analyses (Table 1). The results showedthat the specific hybridisation bands appearednear 36 kDa in 1:200, 1:500, 1:1000, 1:5000 and1:10,000 dilutions. As a control for specificity, nospecific protein band was found in the sampleinduced by pGEX-4T-1. These results verified thatpAb against ADAM28 had a high specificity andtitre.

s, Hertwig’s epithelial root sheath, dental papilla cells andstaining. (B) Expression of ADAM28 in human odontogenice cytoplasm and cytomembrane of human dental papillallicle cells (HDFCs) and human periodontal ligament cellstal pulp stem cells (HDPSCs) (original magnification 100�).of five cell types. The results showed that a specific band

DCLECs at different levels.

1002 Z. Zhao et al.

Table 1 Enzyme-linked immunosorbent assay analysisof polyclonal antibody (pAb) against ADAM28:optical density ¼ x̄� s

Dilution pAb Control n Power

1:100 2.88 � 0.05 0.98 � 0.06 10 2.91:200 2.35 � 0.04 0.60 � 0.03 10 3.91:400 1.70 � 0.08 0.27 � 0.03 10 6.31:800 1.65 � 0.05 0.15 � 0.05 10 11

1:1600 1.38 � 0.02 0.04 � 0.005 10 34.51:3200 0.32 � 0.05 0.02 � 0.003 10 15.81:6400 0.15 � 0.04 0.01 � 0.006 10 15

1:12800 0.07 � 0.01 0.002 � 0.001 10 351:16000 0.05 � 0.01 0.001 � 0.001 10 50

Inthecontrolgroup,ADAM28antibodywassubstitutedwithnon-immune rabbit serum. Power (pAb/control)> 2.1 as positive.

Table 2 The average grayscales of ADAM28 in differ-ent cells: optical density ¼ x̄� s

Cell Experimentalgroup

Control group P

HDPCs 173.6 � 3.15 199.2 � 4.38 <0.01HDFCs 177.2 � 4.37 200.5 � 3.23 <0.01HPDLCs 176.5 � 2.42 202.8 � 4.19 <0.01HDPSCs 182.6 � 3.56 203.6 � 3.85 <0.05

HDFCs, human dental follicle cells; HDPCs, human dentalpapilla cells; HPDLCs, human periodontal ligament cells;HDPSCs, human dental pulp stem cells. In the control group,ADAM28 antibody was substituted with non-immune rabbitserum.

Expression of ADAM28 in murine toothgerm and human odontogenicmesenchymal cells

DNA homology analysis indicated that the sequenceof human ADAM28 mRNA (amino acids 2119—2373)has high homology with that of murine ADAM28.Moreover, it is difficult to obtain all stages of humantooth germ, so ADAM28 expression in murine toothgerm was performed in this study.

The results showed (Fig. 2A) that the enamelorgan was negatively stained at the bud stage(E13d), while the oral epithelia and the mesenchy-mal cells were positively stained (Fig. 2A(a)). Up tothe cap stage (E16d), strongly positive staining wasfound in the oral epithelia, stellate reticulum cellsof the enamel organ, basement membrane, dentalpapilla and dental follicle. Weakly positive stainingwas found in the enamel knot. Negative stainingwas found in the outer and inner enamel epithelium(Fig. 2A(c)). At the early bell stage (E18d), stronglypositive staining was found in the oral epithelia andweakly positive staining was found in the stellatereticulum cells. The other cell types stained nega-tive (Fig. 2A(e)). At the late bell stage (P2d), posi-tive staining was found in ameloblasts, enamelmatrix, Hertwig’s epithelial root sheath (HERS)and dental papilla cells (Fig. 2A(g)). At the crowndevelopment stage (P6d), positive staining forADAM28 was found in ameloblasts, odontoblastsand dental papilla cells (Fig. 2A(i)). At the rootdevelopment stage (P12d), positive staining forADAM28 was detected in ameloblasts, odontoblasts,predentins, HERS, dental papilla cells and dentalfollicle cells (Fig. 2A(k)).

The expression of ADAM28 in human odontogenicmesenchymal cells (Fig. 2B, Table 2) showed thatstrongly positive staining was found in the cytoplasmand cytomembrane of HDPCs, positive staining wasfound in HDFCs and HPDLCs, and weakly positive

staining was found in HDPSCs. RT-PCR results showedthat a specific band of 255 bp was seen in HDPCs,HDFCs, HPDLCs, HDPSCs and HDCLECs at differentlevels (Fig. 2C).

Discussion

The diverse functions performed by ADAM familymembers suggest many potential roles for ADAM28.The human ADAM28 gene is located in chromosome8p21.2 and is mainly expressed in dendrite-likecells, lymphocytes and macrophages.18 However,few data have been reported regarding whetherADAM28 is involved in tooth development.

Teeth arise from progressive reciprocal inductiveinteractions between stomodeal epithelial cells andneural-crest-derived mesenchymal cells.19 Sequen-tial inductive interactions between the epitheliumand the mesenchyme regulate advancing morpho-genesis and cell differentiation,13 and transform thetooth primordia into mineralised structures withvarious cell types.19 Studies of the molecular basisof patterning and morphogenesis of organs haveshown that the developing tooth is an excellentmodel, and future studies can be expected toincrease our understanding of the molecularmechanisms.20

In this study to investigate the localisation andpossible roles of ADAM28 in tooth development,ADAM28 cDNA was cloned successfully from humantooth germ, which initially verified its existence intooth germ. Subsequently, an antibody with highspecificity and titre against human ADAM28 was pre-pared, and expression was detected in tooth germand cells using immunohistochemistry and RT-PCR.

The results revealed that oral epithelia andmesenchymal cells showed positive staining fromthe bud stage (E13d), which is reminiscent of odon-togenesis following sequential inductive interac-tions. This may be explained by the finding thatADAM28 comprises an EGF-like domain, which may

ADAM28 participates in the regulation of tooth development 1003

possess certain functions of EGF such as the stimula-tion or maintenance of undifferentiated cell prolif-eration during embryonic development.21 Thepatterns of EGF-like domain binding in various tissuessuggest that EGFmay play a role in the odontogenesisof epitheliomesenchymal organs by stimulatingepithelial proliferation during initial epithelial budformation and branching morphogenesis.21

Up to the cap stage (E16d), positive staining wasfound in the enamel knot, basement membrane,dental papilla and dental follicle, and negativestaining was found in the outer and inner enamelepithelium. These data suggest that the main ten-dency to tooth development has directed tomesenchymal differentiation, and ADAM28 may beimplicated in growth of the enamel knot and base-ment membrane. This is supported by the fact thatthe primary enamel knot is a signalling centre thatforms at the tip of the epithelial tooth bud. Itbecomes fully developed and morphologically dis-cernible in the cap stage dental epithelium, andexpresses at least 10 different signalling moleculesbelonging to the BMP, FGF, Hh and Wnt families. Thesignals of the enamel knot also have importantroles, together with mesenchymal signals, in regu-lating the patterning of the cusps and hence theshape of the tooth crown. In molar teeth, secondaryenamel knots appear in the enamel epithelium atthe sites of the future cusps. They express severalsignalling molecules, and their formation precedesfolding and growth of the epithelium.22 This evi-dence suggests that ADAM28 may be an importantsignalling molecule in the regulation of tooth crownshape. Moreover, the formation of dental papillastarts at the onset of transition from the bud to thecap stage of tooth morphogenesis, and this is regu-lated by epithelial signals from the primary enamelknot. Dental papilla cells and dental follicle cellshave the capacity to differentiate into odontoblastsand cementoblasts, respectively, and the advancingdifferentiation within the odontoblast cell lineage isregulated by sequential epithelial signals.22 There-fore, it is proposed that ADAM28 may link celldifferentiation to morphogenesis.

Recent data suggest that apart from signallingmolecules such as BMP and FGF, the Notch signallingpathway is also important for proper odontogen-esis10 and controls cell fate during the developmentof a wide range of tissues and organs.23 Further-more, the Notch gene encodes a transmembranereceptor with an extracellular domain carryingmultiple EGF-like repeats and a cytoplasmic domainrequired for signal transduction.19

As mentioned above, ADAM28 protein contains atransmembrane domain, a cytoplasmic domain andan EGF-like domain, which include components of

the extracellular matrix and may play an importantrole in signalling transduction, intracellular proteinmaturation, or localisation to sites of activity.8

In addition, the basement membrane, which ispart of the extracellular matrix that separates theinner enamel epithelium from the dental papilla inthe early stages of tooth development, is believedto participate in epithelial—mesenchymal interac-tions during organogenesis.24 Previous work showedthat basement membranes in both monkey andshark teeth at early stages of development arespecialised for functions such as anchoring and firmbinding, which are essential for the successfulgrowth and differentiation of odontoblasts, andfor better understanding of the mechanism of devel-opment and maintenance of the tooth, and specia-lisations of tooth basement membranes in relationto their roles.25 This evidence reveals that ADAM28may participate actively in tooth morphogenesis,mesenchymal cell differentiation and signal trans-duction at the cap stage by interacting withmultiplesignal molecules such as Notch signalling.

However, weakly positive staining at the earlybell stage (E18d) was only found in stellate reticu-lum cells of the enamel organ. It is postulated thatADAM28 could mainly keep the maturation of theenamel organ and no distinct participation wasfound in others.

At the late bell stage (P2d), positive staining wasfound in ameloblasts, enamel matrix, HERS anddental papilla cells. It is well known that the epithe-lial root sheath is the germinal centre of tooth rootdevelopment. Therefore, ADAM28 may be involvedin root development. In addition, the differentia-tion of dental papilla cells into odontoblastsoccurred following their connection to the HERSthrough the basement membrane. Similarly, thedifferentiation of dental follicle cells into cemento-blasts occurred upon their direct connection toepithelial cells of the HERS.26 Consequently, differ-entiation of dental papilla cells and dental folliclecells was directly associated with development ofHERS. During tooth development, ameloblasts dif-ferentiate from the inner enamel epithelium andprogress through different stages of differentiationas they secrete and mineralise the enamel matrix.27

Ameloblastin and amelogenin are structural pro-teins present in the enamel matrix of developingteeth. Amelogenin has been shown to have signal-ling activity and may play a crucial role in theterminal differentiation of both ameloblastsand odontoblasts.28 A recent study indicated thatameloblastin is likely to participate in the attach-ment of ameloblasts to the enamel surface andmineralisation of enamel.29 These data suggest thatADAM28 may significantly accelerate proliferation

1004 Z. Zhao et al.

and differentiation of ameloblasts and dental papillacells, and further influence the secretion of enamelmatrix.

When entering the hard tissue developmentstages (including crown and root), positive stainingfor ADAM28 was found in ameloblasts, odontoblasts,predentins, HERS, dental papilla cells and dentalfollicle cells. Positive staining in dental follicle cellssuggests that ADAM28 may play a role in maintainingthe structure of the dental follicle and the induce-ment to cervical loop. The enamel matrix and pre-dentin were further subjected to mineralisation,while HERS and dental follicle cells were proceedingdifferentiation into cementoblasts, and later miner-alisation of cementum.

Terminal differentiation of odontoblasts (cytolo-gical polarisation and secretion of predentin/den-tin) starts from the tips of the cusps according to atooth-specific pattern, and involves both tempor-ospatially regulated epigenetic signalling and theexpression of specific competence. Moreover, thebasement membrane is considered to play a majorrole both as a substrate and as a reservoir of para-crine molecules.30 Therefore, it is conceivable thatADAM28, as an active signal molecule expressedin the enamel knot, basement membrane andmesenchymal cells, may induce odontoblast differ-entiation. Furthermore, previous work revealedthat the Notch pathway plays a prominent rolein the cell fate choices leading to the terminaldifferentiation of ameloblasts, dental papillacells (odontoblasts), dental follicle cells (cemento-blasts), enamel and dentin-matrix-synthesisingcells of adult teeth.23

In addition, insulin-like growth factor1 (IGF1) andFGFs may also intervene. IGF1 could stimulate cyto-logical but not functional differentiation of odonto-blasts,30 whereas FGFs could stimulate theproliferation of mesenchymal and epithelial cells,and regulate the growth of cusps.30 Tooth develop-ment appears to require cooperation and networkregulation between multiple signal molecules, andADAM28 may participate in this network regulationto link cell proliferation and differentiation withmatrix synthesis.

Detection of ADAM28 using immunocytochemistryand RT-PCR further confirmed its extensive localisa-tion in mesenchymal and cervical loop epithelialcells at protein and transcription levels.

In conclusion, it is reasonable to speculate thatADAM28 may participate in the regulation of toothdevelopment. Its molecular regulatory mechanismmay be via activation of growth factors (such asFGF, IGF and EGF) and their receptors, promotionof cell proliferation and differentiation, and secre-tion and mineralisation of matrix proteins, while

participating in biological signal (such as Notchsignal) transmission. However, the molecularmechanisms of ADAM28 in terms of the mineralisa-tion of hard tissues require further study.

Acknowledgements

This study was supported by the Nature ScienceFoundation of China (Project No. 30572046) andthe Development of High and New Science andTechnology (863 Project) of China (2002AA205041,2005AA205241).

References

1. Schlondorff J, Blobel C. Metalloprotease-disintegrins: mod-ular proteins capable of promoting cell—cell interactions andtriggering signals by protein ectodomain shedding. J Cell Sci1999;112:3603—17.

2. Blobel CP. Metalloprotease-disintegrins: links to cell adhesionand cleavage of TNFa and Notch. Cell 1997;90:589—92.

3. Black RA, White JM. ADAMs: focus on the protease domain.Curr Opin Cell Biol 1998;10:654—9.

4. Primakoff P, Myles DG. The ADAM gene family: surface pro-teins with an adhesion and protease activity packed into asingle molecule. Trends Genet 2000;16:83—7.

5. Bridges LC, Bowditch RD. ADAM-integrin interactions: poten-tial integrin regulated ectodomain shedding activity. CurrPharm Des 2005;11:837—47.

6. White JM. ADAMs: modulators of cell—cell and cell—matrixinteractions. Curr Opin Cell Biol 2003;15:598—606.

7. Howard L, Zheng Y, Horrocks M, Maciewicz RA, Blobel C.Catalytic activity of ADAM28. FEBS Lett 2001;498:82—6.

8. Howard L, Maciewicz RA, Blobel CP. Cloning and character-ization of ADAM28: evidence for autocatalytic pro-domainremoval and for cell surface localization of mature ADAM28.Biochem J 2000;348:21—7.

9. Mochizuki S, Shimoda M, Shiomi T, Fujii Y, Okada Y. ADAM28 isactivated by MMP-7 (matrilysin-1) and cleaves insulin-likegrowth factor binding protein-3. Biochem Biophys Res Com-mun 2004;315:79—84.

10. Mitsiadis TA, Lardelli M, Lendahl U, Thesleff I. Expression ofNotch 1, 2 and 3 is regulated by epithelial—mesenchymalinteractions and retinoic acid in the developing murine toothand associated with determination of ameloblast cell fate. JCell Biol 1995;130:407—18.

11. Mitsiadis TA, Henrique D, Thesleff I, Lendahl U. Murineserrate-1 (jagged-1): expression in the developing tooth isregulated by epithelial—mesenchymal interactions andfibroblast growth factor-4. Development 1997;124:1473—83.

12. Mitsiadis TA, Hirsinger E, Lendahl U, Goridis C. Delta-Notchsignaling in odontogenesis: correlation with cytodifferentia-tion and evidence for feedback regulation. Dev Biol 1998;204:420—31.

13. Thesleff I, Nieminen P. Tooth morphogenesis and cell differ-entiation. Curr Opin Cell Biol 1996;8:844—50.

14. Yao S, Ring S, Henk WG, Wise GE. In vivo expression of RANKLin the rat dental follicle as determined by laser capturemicrodissection. Arch Oral Biol 2004;49:451—6.

ADAM28 participates in the regulation of tooth development 1005

15. Weskamp G, Kratzschmar J, Reid MS, Blobel CP. MDC-L, awidely expressed cellular disintegrin containing cytoplasmicSH3 ligand domains. J Cell Biol 1996;132:717—26.

16. Lum L, Reid MS, Blobel CP. Intracellular maturation of themurine metalloprotease disintegrin MDC15. J Biol Chem1998;273:26236—47.

17. Bascones C, Gonzalez-Moles MA, Esparza G, et al. Apoptosisand cell cycle arrest in oral lichen planus Hypothesis on theirpossible influence on its malignant transformation. Arch OralBiol 2005;50:873—81.

18. Evans JP. Fertilin beta and other ADAMs as integrin ligands:insights into cell adhesion and fertilization. Bioessays2001;23:628—39.

19. Mitsiadis TA, Regaudiat L, Gridley T. Role of the Notchsignalling pathway in tooth morphogenesis. Arch Oral Biol2005;50:137—40.

20. Thesleff I, Sharpe P. Signalling networks regulating dentaldevelopment. Mech Dev 1997;67:111—23.

21. Partanen AM, Thesleff I. Localization and quantitation of125I-epidermal growth factor binding in murine embryonictooth and other embryonic tissues at different developmen-tal stages. Dev Biol 1987;120:186—97.

22. Thesleff I, Keranen S, Jernvall J. Enamel knots as signalingcenters linking tooth morphogenesis and odontoblast differ-entiation. Adv Dent Res 2001;15:14—8.

23. Mitsiadis TA, Fried K, Goridis C. Reactivation of Delta-Notchsignaling after injury: complementary expression patterns of

ligand and receptor in dental pulp. Exp Cell Res1999;246:312—8.

24. Tabata MJ, Matsumura T, Fujii T, Abe M, Kurisu K. Fibronectinaccelerates the growth and differentiation of ameloblastlineage cells in vitro. J Histochem Cytochem 2003;51:1673—9.

25. Sawada T. Ultrastructure of basement membranes in monkeyand shark teeth at an early stage of development. MedElectron Microsc 2003;36:204—12.

26. Suzuki M, Inoue T, Shimono M, Yamada S. Behavior of epithe-lial root sheath during tooth root formation in porcinemolars: TUNEL, TEM, and immunohistochemical studies.Anat Embryol (Berl) 2002;206:13—20.

27. DenBesten PK, Machule D, Zhang Y, Yan Q, Li W. Character-ization of human primary enamel organ epithelial cells invitro. Arch Oral Biol 2005;50:689—94.

28. Tompkins K, Alvares K, George A, Veis A. Two related lowmolecular mass polypeptide isoforms of amelogenin havedistinct activities in murine tooth germ differentiation invitro. J Bone Miner Res 2005;20:341—9.

29. Torres-Quintana MA, Gaete M, Hernandez M, Farias M,Lobos N. Ameloblastin and amelogenin expression inpostnatal developing murine molars. J Oral Sci 2005;47:27—34.

30. Lesot H, Lisi S, Peterkova R, Peterka M, Mitolo V, Ruch JV.Epigenetic signals during odontoblast differentiation. AdvDent Res 2001;15:8—13.