J Clin Periodontol 1997; 24: 685-692Printed in Denmark . All rights reserved

Copyright © Munksgaard 1997

S."! 03U3-697SI

In vitro studies on periodontalligament cells and enamelmatrix derivative

Stina Gestrelius\Christer Andersson'',Dagny Lidstr6m^Lars Hammarstrom^ andMartha Somerman^'BIORA AB. Malmo, Sweden, Center for OralBiology, Karolinska institute, Stocktiolm, Sweden,'Department of Periodontlcs/Prevention/Geriatrics and Department of Ptiarmacoiogy,University of Michigan, Ann Artjor, Ml, USA

Gestrelius S. Andersson C, Lidsirom D, Hammarstrom L, Somerman M: In vitrostudies on periodontal ligament cells and enamel matrix derivative, J Clin Periodontol1997: 24: 685-692. © Munksgaard, 1997,

Abstract. The recognition that periodontal regeneration can be achieved has re-sulted in increased efforts focused on understanding the mechanisms and fac-tors required for restoring periodontal tissues so that clinical outcomes of suchtherapies are more predictable than those currently being used. In vitro modelsprovide an excellent procedure for providing clues as to the mechanisms that maybe required for regeneration of tissues. The investigations here were targeted atdetermining the ability of enamel matrix derivative (EMD) to influence specificproperties of periodontal ligament cells in vitro. Properties of cells examinedincluded migration, attachment, proliferation, biosynthetic activity and mineralnodule formation, Immunoassays were done to determine whether or not EMDretained known polypeptide factors. Results demonstrated that EMD tinder invitro conditions formed protein aggregates, thereby providing a unique environ-ment for cell-matrix interaction. Under these conditions, BMD: (a) enhancedproliferation of PDL cells, but not of epithelial cells; (b) increased total proteinproduction by PDL cells; (c) promoted mineral nodule formation of PDL cells,as assayed by von Kossa staining; (d) had no significant effect on migration orattachment and spreading of cells within the limits of the assay systems used here.Next, EMD was screened for possible presence of specific molecules including:GM-CSF, calbindin D. EGF, fibronectm, bEGF. y-interferon. IL-1/?, 2. 3. 6: IGF-1.2; NGF, PDGF, TNF, TGF^, With immunoassays used, none of these moleculeswere identified in EMD, These in vitro studies support the concept that EMDcan act as a positive matrix for cells at a regenerative site.

Key words: enamel proteins: immunoassay:proliferation, attactiment, btosynttiesis, mineral

Accepted for publication 25 March 1997

Periodontal diseases are marked by in-flammation and subsequent loss and/ordamage to tooth-supporting tissues, in-cluding cementum. periodontal liga-ment (PDL) and bone, Increasmg evi-dence that tissues of the periodontiumharbour cells with the capacity to regen-erate the periodontium has fostered aninterest in developing clinical pro-cedures to restore periodontal supportby promoting such cells to function ina regenerative capacity, i,e,, formationof new cementum, periodontal ligamentand new bone (Hammarstrom 1996,MacNeil et al, 1995, Pitaru et al, 1994,McCulIoch 1993, Aukhil et al, 1990,

Nanci et al, 1996), Logical candidatesfor stimulating cells at the healing siteto function in a regenerative capacityare proteins in their local environment,in particular extracellular matrix pro-teins.

An approach used to determinewhich proteins may be involved in re-generation of specific tissues is to firstexamine the tissues of interest duringdevelopmental stages. While it is recog-nized that regeneratioti does not followexactly the pattern required during de-velopment of a given tissue, there aremany similarities between the 2 pro-cesses. Thus, using histology, immuno-

histochemistry, in situ hybridizationand tissue recombination experimentsone can determine the temporal andspatial distribution of specific cells andtheir associated proteins during forma-tion of a given tissue. Information ob-tained by such investigations providesclues as to which proteins may be re-quired for altering cellular behaviour,e,g,, migration, attachment, prolifer-ation and cell differentiation, as re-quired for inducing regeneration oftissues. In particular, with regard tooral tissues, it is now recognized thatepithelial-mesenchymal interactions arerequired for formation of enamel

686 Gestrelius et al.

(ameloblasts) and dentin (odontobJasts)(TheslefT et al. 1996). These inductiveinteractions can result from cell-cell in-teractions, from interactions betweenextracellular matrix molecules and cellsurface receptors and. or from diffusiblesignals such as growth factors duringcrown development (Vainio et al. 1993).

With regard to cementum. Slavkin &Boyde (1975) and Slavkin (1976) pro-posed that similar interactions occurbetween dental ectomesenchymal cells(follicle cells) and cells of the Hertwigepithelial root sheath. In this situationthey suggested that enamel related pro-teins from the root sheath initiate for-mation of cementum. Studies over thepast twenty years support this hypo-thesis (for review; see Hammarstrdm(1997)). and studies by Hammarstrom(1997) and Hammarstrom et al. (1997)demonstrated that enamel matrix is in-volved in the formation of acelluiar ce-mentum during tooth development, andthat this matrix has the potential to in-duce regeneration of acellular ce-mentum in experimental periodontaldefects in monkeys.

To further explore the role for en-amel matrix proteins in periodontal re-generation, the effect of enamel matrixderivative (EMD) on PDL cells in vitrowas examined. EMD contains purifiedhydrophobic amelogenins. which forminsoluble aggregates ("matrix"! underphysiological conditions. Specific cellu-lar behaviours examined were ceil mi-gration, attachment, proliferation andbiosynthetic activity and included for-mation of mineralized nodules. Allthese are events known to be requiredfor promoting development as well asregeneration of periodontal tissues. Inaddition, since extracellular matricesmay have the capacity to bind and re-tain polypeptides. it was important todetermine whether or not EMD con-tained any serum associated factors,such as growth factors and adhesionmolecules, that may contribute to theobserved biological events.

Material and MethodsProteins and polypeptides

Enamel matrix derivativeEnamel matrix derivative (EMDO-GAIN*), was obtained from BIORAAB, Malmo. Sweden. For in vitro stu-dies, lyophilized protein was added tothe medium as an aqueous solution (pH5-6). Alternatively, wells or dishes werefirst coated by allowing an alkaline

EMD solution (carbonate buffer pH10) to incubate overnight (attachmentassays and mineralizations assays).

Reference polypeptidesla) Radiotabelted (^--1)bFGF (human. IM 243). spec, activity

lOOO/iCi'nmol. Amersham, UKEGF (human, IM 196). spec, activity

1300//Ci/nmol. Amersham. UKIGF-i (human. IM 172). spec, activity

1900 /iCi/runol. Amersham. UKIGF-2 (human. IM 238). spec, activity

2000 /;Ci/nmol. Amersham. UKPDGE BB homodimer (human. IM

213). spec, activity 1000 //Ci/nmol.Amersham. UK

IL-3 (human. IM 220). spec, activity600 /(Ci/nmol. Amersham. UK

GM-C5f (human. IM 224). spec, activi-ty 1200/(Ci/nmoL Amersham. UK

IFN-gamtria (htraian. IM 202). spec, ac-tivity 600 fjCi/nmoL Amersham. UK

NGF (murine. IM 207). spec, activity1500 fiCh'nmo], Amersham. UK

TGFpHhuman, NEX 267). spec, activi-ty 3000 /(CL nmol. DuPont. Germany

TGFfil (human. IM 246). spec, activity2284 /iCi/nmol. Amersham. UK

(h) UnlahetledbFGF (bovine. BDP 13). British Bio-

tech. UKTGF(il.2 (porcine platelets, lot R811).

R&D Systems. UKrC/"/i/(human. TGOIO). Chemicon.

USATGFjil (human), from Quantikine ELI-

SA kit (DB IOO). R&D Systetns. UK

Antibodies

.Anti-EMDPolycional anti-EMD was raised in rab-bits by repeated injections of EMD inFreund's adjuvant. The rabbits were im-munized and blood was collected du-ring a period of 4-6 months. Pooledwhole serum was also purified by affi-nity chromatography using EMDbound (6 mg EMD/ml gel) to Sepharo-se 4B (Pharmacia. Uppsala. Sweden) inphosphate buffered saline (PBS), andstepwise eluted with I M NaCl and 0.2M 2-aminobutane pH 11. Eluted anti-body fraction was desalted using Se-phadex G-25sf (Pharmacia, Uppsala.Sweden) and diluted to 25 ;/g/ml. asanalysed with HPLC (Superdex 75) at215 nm with human IgG used for cali-bration. The whole serum and the affi-nity purified antibody recognized EMDin immunoradiometric assays at a dilu-tion of 1:10* and 1:10'. respectively.

Antibodies to polypeptide factors andproteinsIGF-I RlA-kit IM.I72I. Amersham.

UKPDFG RIA-kit RPA.536. Amersham.

UKTNF RIA-kit RPA.532. Amersham,

UKJL-10 RIA-kit RPA.533. Amersham.

UK!L-2 ELISA kit. Genzyme Corpora-

tion. USAIL-6 ELISA kit. Genzyme Corpora-

tion. USAFGF basic ELISA kit RPN 2158.

Amersham. UKTGF-fit. Quantikine ELISA kit (DB

100). R&D Systems. UKTGF-/i2. ELISA kit RPN 2163. Amers-

ham. UK.4nti-FGF basic (bovine. BD.A4). pure

IgG. British Biotech. UKAnti-TGFP (porcine. BDAl). British

Biotech. UKAntl-fibronectin (human. MAB042).

Chemicon. USAAnti-calbindirt D (chicken, C-8666). Sig-

ma. USA

Cell cultures and media

Human PDL cellsHuman periodontal ligament cells(PDL cells) were obtained from healthyhuman periodontal tissues of indivi-duals undergoing extractions of premo-lars for orthodontic reasons. Details forgrowing such cells have been published(Somerman et al. 1988a). Cells wereused between passage 4 and 10.

In vitro assays with human PDL cells

Attachment assaysThe attachment assay used was a modi-fication of Klebe (1974) (Somerman etal. 1989). The dishes were either precoa-ted with the protein or the dishes werepreincubated with DMEM +0.1% BSAand EMD for Ih prior to the additionof the cells. For these assays attachmentactivity was evaluated at 1.5 h to 6 h.Photographs were taken to evaluatespreading and then cells were removedenzymatically and counted electronical-ly using a Coulter Counter.

Chemotaxis (migration) assayIn order to determine the ability of cellsto move directionally toward extracts ofEMD in vitro, a Boyden Chamber sy-stem was used (Somerman et al. 1989.1982 as modified from Postlethwaite et

PDC cells and EMD 687

al. 1976). For these studies both PDLcells and CRL 1475 foreskin fibrobiastswere used. CRL 1475 are known tohave a marked chemotactic response tofibronectin (Somerman el al. 1989)whereas PDL cells have less of an abili-ty to migrate to a chemoattractant (un-published data). EMD in DMEM/1%BSA was added to the bottom chamberat 5. 25. 50 and 100 n^JmX. a milliporepolycarbonate filter (8 ^m pore size-precoated with gelatin) was placed overthis solution and then ceils in DMEMcontaining 1 mg/ml BSA were placed inthe upper chamber. Chambers were in-cubated at 37°C for 4 h. then filterswere removed, stained wilh Hemacolor(Diagnostic Systems Inc. Gibbstown.NJ. USA) and transferred to slides.Cells attached to the upper surface wereremoved and cells migrating throughthe filter determined by counting a 1mm' microscopic field per filter.

Proliferation assay (cell toxicity assayjProliferation rate was carried out overa 10-day period, so as lo proxide infor-mation on both cell proliferation andcell loxicity. About 5000 cells were see-ded into test wells m DMEM/10% FBSwith antibiotics and allowed to attachfor 24 h. After cells attach, medium wasremoved and replaced with DMEM/10"/ti FBS (positive control); 2*'/.. FBS/DMEM (negative control) or 2% FBS/DMEM plus FMD at 25. 50 or 100 /;gml. Under standard protocol medium ischanged on days 3. 6 and 9 (Somermanet al. 1988b); however, in order to con-trol the amount of EMD aggregate inthe wells, media were not changed. Tri-plicate wells were harvested on days 1.2. 4. 7 and 10 and cell counting perfor-med by Coulter Counter. Statisticaltests were performed with analysis ofvariance (ANOVA) and Duncan's mul-tiple range test.

BicLfynthesis a.'isaysTotal protein/collagen. Cells in DMEM/107(1 FCS and antibiotics were exposedto EMD (25. 50. 100 //g/nil) for 48 hand labelled with L[5-'H]-proline (10fiCi/ml) to determine total protein andcollagen (Somerman et al. 1988). Cellsand media were pooled, dialysed exten-sively against buffer and samples coun-ted by liquid scintillation spectrometer.These counts were used to measure to-tal protein production. Concurrent withassays for protein production, cultureswere prepared to determine cell prolife-ration and thus results were normalized

to cell number. To estimate collagenproduction, an aliquot of the dialysedmaterial was reacted for 6 h with bacte-rial coilagenase (Peterkofsky & Diegel-mann 1971). Resulting proteins wereprecipitated with TCA-tannic acid, andthe supernatant and a control superna-tant were counted by scintillation spec-trometer. Thus, counts from collagena-se-treated sample minus control samplerepresented collagen production. Stati-stical tests were performed withANOVA and Duncan's multiple rangetest.

Electrophore.sis. Pooled media andcell samples were dialysed against waterand lyophilized. then dissolved in SDS-PAGE buffer, where concentration ofsample used was corrected for cell num-ber. Thus any variation in protein pro-files found is based on increased proteinproduction per cell when evaluated bygel electrophoresis. After reductionwith mercaptoethanol the protein sam-ples were analysed by electrophoresis onSDS 4-20% polyacrylamide gradientslab gels according to the method ofLaemmli (1970). The stacking gel wasA"/« acrylamide. The gels were fixed,dried and autoradiographed by the me-thod of Bonner & Laskey (!974).

Mineralization as.savsPDL cells were cultured in 175 cm-flasks (Nunclon. Nunc. Denmark) in a-ME.VI supplemented with 10"™ calf se-rum, antibiotics (Peniciilin-Strepiomy-cin). L-glutamine and ascorbic acid (50/ig/mi). After 10 days, the cells weretrypsinized. counted and plated onEMD-coated Petri dishes. Attempts todecrease the serum concentration from10 to 5"ii resulted in very low cloningefficiencies and thus cells were main-tained in 10"/i. FCS. Appropriate con-trols included cells seeded on FMDwith mineralization enhancers, dexa-methasone (10 nM) and /i(-g!ycero-phosphate (10 mM). added (Tenen-baum & Heersche 1982). and cells notcultured on EMD matrix plus or minusdexamethaxone and ;S-glycerophospha-te. PDL cells (50. 100 or 200 in 1 mlsolution), treated as described above,were plated and incubated at 37°C in ahumidified atmosphere of 95'Xi air and5% CO, for 7 days before the first chan-ge of medium. Fresh medium was thensupplied twice weekly Von Kossa assay(Von Kossa. 1901. as described by-Bancroft & Stevens. 1990) was used todetermine presence of mineral nodulesand was performed 15-30 days after

cultivation, most typically after 21-22days. Mineral nodules were examinedmicroscopically and the number of stai-ned clones, as well as the total numberof clones, were counted. Classificationwas performed according to the follo-wing system.

Type I. very dense clones; Type II.dense clones but without merged fila-ments; and Type III. diffuse clones.

Statistical tests on impact of differentconcentrations and additives were per-formed with one-way analysis o{ var-iance using SPSS/PC+, version 4.01.

Immunoassays

Itnmunoradiometric assay (IRMA)Coated substances were detected bydouble antibody technique. The second(labelled) antibody was either goat anti-rabbit IgG (for EMD. bGFG and TGF-/?) or goat anti-mouse IgG (for cal-bindin and fibronectin). and radioacti-vity was measured using a gammacounter (LKB Wallac 1275. Stockholm.Sweden).

Radioimmunoa.%say (RIA)Radiolabelled polypeptides were detec-ted by a double antibody techniqueusing anti-EMD plus Pharmacia De-canting System 3 (sheep anti-rabbitantibody). About 100 fmoi of radiola-belled polypeptide was added per tube,resulting in a detection limit of 2-10fmol ('=5"f> bound tracer).

Commercial kits (RIA or ELISA)Commercial kits were used according tothe Package Insert. ELISA microtiterplates were analysed using Milenia equ-ipment (Diagnostic Products Corp.,Los Angeles. CA. USA). The Multicalcevaluation program (version 2.0. Wal-lac. Stockholm. Sweden) was used forcalculating the standard curves and re-sults. The detection limits were set atten times the coefficient of variation ofa buffer blank.

Four different TGFy? standards werefractionated on a HPLC column(Superdex 75. Pharmacia. Sweden) equ-ilibrated in 3% acetic acid, 30% aceto-nitrile and 0.9"/., NaCl. All fractionswere concentrated by vacuum centrifu-gation. and PBS with 0.1% BSA and0.1% Tween was added to the originalvolume. pH was adjusted to 5.6 withNaOH. Each fraction was then testedwith the Quantikine test for activated1G¥P (analytical range 31-2000 pg/ml). In parallel, EMD alone or FMD

68B Gestrelius et al.

I

tron/. Enamel matrix derivative (EMD) precipitated from aqueous solution. Scanning elec-tnicrograph, Bar= 100 fim.

100000

Jt

sZ 10000«

1000

Negative control

EMD 25 us/ml

EMD 50ug/ml

EMD 100ug/ml

Positive control

Days

Fig. 2. Proliferation of PDL cells exposed to EMD, PDL cells were allowed to attach overnightprior to addition of EMD at 25 /jg,'ml (A) 50 ^g/ml ( + ) and lOO/j&'ml (•) , Cells exposed to2% FCS (x) served as a negative control, while cells exposed to 10% FCS ( • ) served as apositive control. At days 2, 4, 7 and 10, using triplicate welis, cells were trypsinized andcounted by Coulter counter. Shown here is a representative experiment where results werereproduced in three separate experiments. Data evaluated by ANOVA and Duncan's multiplerange test indicated that EMD enhanced cell proliferation as follows:

Day 2: 2'% FCS, EMDloy. FCS

Day 4: 2% FCSEMD 25 fis/m\EMD 50 f!^ml \00 fig/ml10% FCS

- subset 1- subset II- subset I- subset ]I- subset III- subset IV

Days 7, 10: 2%i FCSEMD 25 fig/mlEMD 50 ftg/m\EMD 100 /ig/ml10%, FCC

- subset I- subset II- subset in- subset IV- subset V

mixed with TGF/? standard was fractio-nated using the same procedure and te-sted with the same kit.

ResultsIn vitro assays with human PDL cells

.iitaihment artd chemoiactic acliritrFor adhesion assay uncoated wells werepre-incubated for 1 h with DMEM plusEMD or fibronectin (positive control) oruncoated wells were pre-coated withthese agents, Celis were added for 1,5 hto 6 h. Under both assay conditions,EMD enhanced cell attachment to aminimal degree at 20/;g/ml, but did notpromote cell spreading during this timeperiod, i,e,, the ce!ls remained roundedbut viable. In contrast, cells exposed tofibronectin were attached and spread by1 h.

Next, the ability of PDL cells tomigrate to EMD was tested using aBoyden chamber assay, LJnder the con-ditions and doses used EMD producedno enhancement of cell migration andthis was repeated in three separateexperiments (data not shown).

ProliferationWhen an acidic solution of EMD wasadded to the cell culture medium a pro-tein aggregate was formed. As seen inFig, 1. this appeared as an accumulationof microspheres with a diameter of about1 /im. In order not to alter the quantity ofEMD these assays were performed with-out change of culture media, EMD,added at baseline, influenced cell pro-liferation, with a small increase after 4days and a marked enhancement ,seenafter 7 or 10 days. Fig, 2,

Next, to establish whether or not thisenhancement was selective to PDL cells,the effect of EMD on proliferative ac-tivity of epithelial cells was investigated.Using the same protocol as for PDLcells but with rat tongue epithelium, de-rived from explant cultures of rat ven-tral mucosa, no difference vvas observedin proliferation between celis exposed toEMD and control cells over the 10 dayperiod. Fig, ,1, Similar results werenoted for HT 1080 cells, a cell line thatexhibits epithelial-like characteristicsincluding expression of keratins (datanot shown),

Biosynthelic activityWhen preattached cells were exposed toEMD and 'H-prolme for 48 h, the pro-duction of total protein and collagenwas enhanced up to the level observed

PDC cells and EMD 689

100000 -I

"t 10000

EZ

1000

DMEM* 2%FBS- DMEH + 2% FBS + lODufl/mlEMD- DMEM • 10% FBS

Days

Fig. 3. Proliferation of rat tongue epithelial cells exposed to EM. Cells were allowed to attachovernighl prior lo adduion of EMD, 100 //g.ml {A|, Cells exposed lo 2".. FCS (x) served asnegative control, while 10"-i> ECS ( • ) served as a positive control. Cells in triplicate wells, atdays 2, 4, 7 and 10, were trypsinized and counted by Coulter counter. Data analysis usingANOVA and Duncan's multiple range lest, show that EMD fits in the same subset as 2';-uFCS, indicating that EMD had no effec! beyond control ai any time point.

Tahle I. Relative protein production by PDL cells exposed to EMD in the presence of 'H-proline for 4S h

Treatment

negative control (no additive)

EMD25 /Jg nil50 >(g/nil

100 figlmlpositive control '(dentin 100;;gml)

Protein(ratio treated control

1,00=0.03

2.06 = 0.20*2,54=0,24*2.60 = 0,15*2.48=0.18*

Collagen{ratio treated control)

1,00=0,09

1,8,̂ = 0,2,1*1.95=0.29*2.44=0.15*2,,12=0,08*

* Dentin is a guanidine.EDTA extract of boxine dentin.* Triplicate results were significant connpared to the control as demonstrated by the ANOVA.Duncan multiple range test:Protein:No additon- .subset 1,25 /(g.'ml EMD- subset II,50/(g'ml EMD, deniin extract - ,subset III100 ;<g.'ml EMD, dentin extract - subset IV

Collagen:No addition- subset 1:25 ^g'ml, 50 //gml EMD - subset II100 /;g'ml EMD. dentin extract - subset III

for the positive control (100 /ig/mlguanidine/EDTA extract of dentin), a,sis seen in Table I. From the results itdoes not appear that EMD enhancedcollagen prodttctioti tnore than othertypes of proteins, i,e,, there was an in-crease in total protein production,Autoradiographic examination of theprotett: profile for cells exposed toEMD vs, dentin extract (O/E-D) re-vealed several bands in cottimoti (Fig,4), For example, intense banding notedin the 100-200 kDa region with both

treatments probably reflects productionof type I collagen and fibronectin. Inaddition, less intense but similar pro-files for EMD and G/E-D were notedbetween 69-97 kDa and 2 1 ^ 6 kDa,However, there was one region between46-69 kDa, where the banding patternappeared different between these twoagents (Fig, 4: bracket region|, In ad-dition, cells exposed to G/E-D ex-hibited an intense banding in the regionaround 100 kDa (Fig, 4., arrow), as re-ported previously (Somerman et al.

)9S7a) and this band was not apparentin cells exposed to EMD,

In vitro mineralization modelCloning efficiencies, reflected as the totalnumber of clones in medium with J0%serum, were around 30% both in thepresence and absence of additives suchas EMD or dexamethasone plus ^-gly-cerophosphate (Table 2), Examples ofthe different types of clones are shown inFig, 5, The morpho!og\ varied with theadditives, showing the highest number ofvery dense clones. Type I, in the presenceof ^-giycerophosphate and dexametha-sone. An increased number of Type / /clones were found in the presence ofEMD, The most striking differencenoted from treatment with EMD, how-ever, was seen as the frequency of clonesthat were von Kossa stained. This in-creased fivefold from the control to a fre-quency of about 25'y'o in the presence ofEMD, .Addition of yS-glycerophosphateand dexamethasone to the cotttrol in-creased the frequency of von Kossastained clones to about 15"/.), As a con-trast, the mineralization enhancers otilyadded marginally to the establishmentof nodules, m the presence of EMD,Thus, the presence of EMD, even in theabsence of mineralization enhancers, re-sulted m a significantly higher number ofvon Kossa stained clones compared tothe negative control as well as comparedto the positive control,

EMD was used at an amount of iOfig/'ml and the effect on von Kossa-stainedclones was found to increase both in thepresence and in the absence of dexa-methasone. The addition of dexametha-sone appeared to induce mineralizationat an earlier time point, as seen from re-sults on days 16-17 compared with thoseon days 21-22, but did not increase thefinal ratio of von Kossa-stained clones.Precipitation of EMD aggregate was no-ticed on the dishes and the fibroblastsappeared to grou' in close associationwith the protein aggregate. While inthese studies we did not examine the as-sociation of mineral tnodules with colla-gen fibrils, it has been shown pre\iouslythat PDL cells, isolated as those usedhere, will form mineral-collagen associ-ated nodules (Arceoet al, 1991),

Immunoanalytical tests

For this study ?• sensitive methods wereselected and a number of analyses per-formed.

In method 1, commercial antibodies

690 Gestrelius et al.

EMD fig/ml G/E-D fig/ml

Control 25 SO 100 100

i I200 kDa —

97kDa —

69 kDa —

46 kDa —

21 kDa —14 kDa —

Fig. 4. Autoradiograph of SDS-PAGE from 'H-proline labelled PDL cells exposed to EMDor dentin extract. Ceils were either untreated (control) or treated with EMD at 25. 50 or IOO/jgml or guanidine EDTA extract of dentin (G E-Dl at 100 /igml for 48h as described inmethods section. Each pooled mediutn and cell sample having equal cell numbers was dialyzedagainst water, lyophilized and dissolved in gel buffer and loaded al 50 /iglane. Thus, anyvariation in protein profiles found was based on increased protein production per cell.<— points to intense band. 100 kDa region noted in G E-D treated cells but not in controi orEMD treated cells. I] demonstrates area where protein profile differences are suggested be-tween cells exposed to EMD versus G E-D and where banding reflects an EMD effect onprotein production.

to polypeptide factors and proteinswere exposed to EMD solution. The de-tection limits were found to be 10-300pgiwl by testing with reference pep-tides. None of the antibodies tested re-acted with EMD (Table 3).

Iti method l\, anti-EMD serum pro-duced by Biora was exposed to commer-cial polypeptides. This antiserum wouldcontain antibodies to growth factors, ifany were present in EMD. A reaction be-

tween anti-EMD and a factor would in-dicate the presence of the factor in EMD.The detection limits varied from 100 pg.'mi up to 1-5 ng/ml. By radioimmuno-assay (RIA) it was shown that anti-EMDdid not react with any ofthe 12 factors/proteins tested (Table 3).

In method III, EMD was tested forinteraction with an immobilized TGF/i;receptor (Quantikine test) and a posi-tive reaction was found, even without

Fig. 5. Human PDL cell clones that havebeen von Kossa stained after 22 days (cf.Table 2). (Al shows a negative control, with-out additives; (B) has been cultured on EMD(30^g/ml) added from start and at change oimedia; (C) is a positive controi with additionof iO nM dexamethasone plus 10 /;M /J-gly-cerophosphate from start and with newmedia.

Table 2. Formation of mineral nodules by PDL cells exposed to EMD; frequencies of Von Kossa stained clones and dense clones (Types I and II)

16-17 days 21-22 davs

VonKossa

clones% of total

Treatment

totalclones(I-IIl) n

VonKossa

clones'/., of total

totalclones(I-III)

negativeEMD 30 /ig/m!positive control"EMD+

(Dex/^GP)

-1

10*4

!9»

0039*

79

1116*

38343535

5664

523**14*25**

< l15*4*

1326*1526*

31353129

1518154

" Positive control is 10 nM dexamethasone and 10 mM /i-g)ycerophosphate.* Significantly different from the negative control demonstrated by ANOVA.

** Significantly different from the positive control demonstrated by ANOVA.

PDC ceils and EMD 691

Table 3.

Analysis

lmmunoassays for

MW (kDa)of peptideor protein

polypeptides in

Detectionlimit

(ng/ml)

Enamel Matrix

Method I(anti-peptideversus EMD)

Derivative (EMDOGAIN*)

Method II(anti-EMD

versuspolypeptide) Test

GM-CSFcalbindin DEGFfibronectin

bFGFy-lnterferonlL-1^IL-2IL-3IL-6lGF-lIGF-2NGFPDGFTNFTGF-^TGF-/J2

2228

6-500

18401715.52026

7.77.7

3030172525

0.510.11

1.0.5,0.0110.10.30.50.3

5:0.10.10.5

2:0.50.2

1:0.10.02: 0.1

RIAIRMARIA

IRMAIRMA. RIA.

ELISARIARIA

ELISARIA

ELISARIARIARIARIARIA

IRMA. RIAELIS.^. RIA

< denotes that the result is below the detection limit.Abbreviations used: GM-CSF macrophage colony stimulating factor: EGF epidermal growthfactor: bFGF basic fibroblast growth factor: IL interleukin: IGF insulin like growth factor:NGF nerve growth factor: PDGF platelet derived growth factor: TNF tumour necrosis factor,TGF transforming growth factor.

100

75

50

25 .EMD

10 15 20 25 30 35 40

Fraction No.Fig. 6. Tes! for activated TGF/ZI (Quantikine) on fractions from size exclusion chromatogra-pby (Superdex 75) of EMD (55 /(g) or TGF/J reference (500 pg). After separation the fractionswere concentrated and dissolved as described in the Methods section.

preactivation. This reaction was ques-tioned since EMD had visibly precipi-tated in all wells, as expected ai thephysiological cotiditions used, and sincenone of the immunoassays using

methods ! or 2 had indicated a presenceof TGF,ff. Therefore. EMD was frac-tionated oti a size exclusioti column,and the separate fractions were tested.In parallel, commercial TG¥fi stan-

dards were fractionated on the samecolumn and again all fractions weretested. As seen from Fig. 6. all fraction-ated TGF^ standards were detected butnone of the EMD fractions unlessTGF/y had been added, indicating thata false positive result had been observeddue to precipitation of EMD.

Discussion

The results of the in vitro studiesfurther contributed to the understand-ing of the mechanisms by which EMDmay play a role in promoting cementumformation and periodontal regenera-tion. The stimulation of cellular pro-liferation and protein and collagen pro-duction and formation of mineralizednoduies in PDL ceil cultures exposed toEMD are compatible with the regenera-tion results obtained in vivo (Hammar-strom et al. 1997). When interpretingthe results obtained in vitro it should bekept in mind that EMD is insoluble atihe pH and temperature at which thestudies were performed (Gestrelius etal. 1997). A visible protein aggregatewas noted when EMD was added to cellculture media at concentrations of 5 upml or more. This motivated the designof the proliferation and protein syn-thesis studies, i.e.. no change of culturemedium was done. Thus, the resultscould be explained by EMD being a vi-able matrix for cells and.or the resuh ofa cell-matrix interaction. This is anal-ogous to the normal m vivo environ-ment where most cells require an extra-cellular matrix in order to function(Clark, 1995). In addition, while nopolypeptide factors were detected inEMD, there is a possibility that growthfactors from the medium itself or fromthe medium conditioned by the cellswere taken up by the EMD aggregatesand thereby enhanced the bioactivity ofthe ceils.

The studies on cell attachment indi-cated that dishes coated with FMD werewell accepted by the PDL cells, but thismatrix had a. minima! effect on eel!attachment or cell spreading. Whencompared with the effects of fibronectin,one of the known attachment proteins,the effect of EMD was markedly less andvery much slower in development. EMDcontains no known attachment se-quences (BIORA AB data on file) andthe slight increased attachment is morelikely caused by added media or sub-stances produced by the PDL cells ratherthan a direct effect of EMD.

692 Gestrelius et al.

The Boyden chamber experiments.which measure the response of cells to asoluble chemical gradient, showed no re-sponse to EMD. Future studies on cellmigration to surface-bound adhesiongradient (heptataxis) are required tofurther explore the migratory behaviourof cells in response to EMD.

As discussed above, the increase iti cellproliferation and biosynthetic activity inPDL cells exposed to EMD could be ex-plained by the suitable three dimensionalsubstratum provided. In addition, theassays were carried out with a single ad-dition of EMD and cells were main-tained in the same medium for up to 10days, thus the enhancement of prolifer-ation, protein production and mineral-ization seen over time may well havebeen caused by endogenous cellularproducts.

Interestingly, the response of epi-thelial cells to the addition of E.MD dif-fered from that of the PDL cells. Thelack of enhanced proliferation of theepithelial cells is in agreement with theresults obtained in the in vivo dehis-cence experiments in monkeys, in whichepithelial down-growth was very limitedwhen EMD was applied on the rootsurface (Hammarstrom et al. 1997).The mechanism(s) behind the differentresponses by epithelial and mesenchy-mal cells is presently not known.

In conclusion, the addition of EMDto cell culture media resulted in a num-ber of changes, which included en-hanced proliferation ofthe PDL cells aswell as increased protein and collagenproduction and promotion of mineral-ization. In contrast, addition of EMDhad no significant effect on epithelialcell proliferation in vitro. No growthfactors could be detected in EMD. Thecurrent data thus support the hypo-thesis that EMD can act as a matrix forcells at a periodontal regenerative site.Specifically, the results suggest that ag-gregates formed by EMD at physiologi-cal pH and temperature are responsiblefor creating a positive environment forcells to proliferate and differentiate, asrequired for regeneration of the peri-odontium. Clearly, future studies are re-quired in order to determine the exactmechanisms by which EMD may influ-ence cell function, in vitro and in vivo.

References

Arceo. N.. Sauk. J. J., Moehring. J.. Foster.R. A. & Somerman, M. J. (1991) Htunanperiodontai cells initiate mineral-like nod-

ules in vitro. Journal of Periodontology 62,499-503.

.Aukhil. L. Nishimura. K. & Fernyhough. W.(1990) Experimental regeneration of theperiodontium. Cntiial Reviews in Oral Bi-ology and .Medii'ine 1, 101-105.

Bancroft. J. D. & Stevens. A. (1990) Theoryand practice of histological technique.';. 3rdedition. Churchill Livingstone, pp 324.333-334.

Bonner. W. M. & Laske>. R. A. (1974) A filmdetection method for tritium labelled pro-teins and nucleic acids in polyacrylamidegels. European BiochemicaUotirnal 46, 83-88.

Clark. A. F. 11995) Molecular and cellular bi-ology of wound healing. 2nd edition.

Gestrelius. S.. Persson. E.. Johansson. A. C .Andersson. C . Brodin. A.. Rydhag. L. &Hammarstrom. L. (1997). Formulation ofenamel matrix derivative for surface coat-ing. Kinetics and cell colonization. Journalof Clinical Periodontology 24. 678-684,

Hammarstrom. L. (1996) The role of enamelmatrix proteins in the development of ce-mentum and periodontal tissues. In 1997Dental enamel Wiley. Chichester (CibaFoundation Sympositim 205). p. 246-260.

Hammarstrom. L. 11997). Enamel matrix, ce-mentum development and regeneration.Journal of Clinical Periodontology 24, 658-668.

Hammarstrom. L.. Heijl. L. & Gestrelius. S.(1997). Periodontal regeneration m a buc-cal dehiscence model in monkeys after ap-plication of enamel matrix proteins.Journal of Clinical Periodonlology 24, 669-677.

Kossa. J. von (1901). Nachweis von Kalk.Beitrage zur pathologischen Anatomie undzur allgemeinen Pathologie 29, 163-

Klebe. R. (1974) Isolation of a collagen-de-pendent cell attachment factor. Nature2iO. 248-253

Laemmli. V (1970) Cleavage of structuralproteins during assembly of the head ofbacteriophage T4. Nature 227, 680-685.

McCulloch. C. A. G. (1993) Basic consider-ations in periodontal wound healing toachieve regeneration Periodoniohgy 20001, 16-25.

MacNeil. R. L.. Berry. J.. D'Errico. J.. Stray-horn. C. & Somerman. M. (1995). Localiz-ation and expression of osteopontin inmineralized and nonmineralized tissues ofthe periodontium. In; Denhardt. D. X.Butler. W. T. Chambers. A. F & Singer.D. R. (eds); ANYAS Osteopontin: Role incell signaling and adhesion 760. 16fr-176.

Nanci. A.. McKee. M. D.. Smith. C. E.. (eds)The Anatomical Record 245: 1996 SpecialIssue: The biology of dental ti.ysues.

Peterkofsky. V & Diegelmann. R. (1971) Useof a mixture of proteinase-free coUagen-ases for the specific assay of radioactivecollagen in the presence of other proteins.Biochemistry 10, 988-994.

Pitaru. S.. McCulloch. C. A. G. & Narayan-an. S. A. (1994) Cellular origins and differ-entiation controi mechanisms during peri-

odontal development and wound healing.Journal of Periodontal Research 29, 81-94.

Postlethwaite. A. E.. Snyderman. R. & Kang.A. H. (1976) Chemotactic attraction ofhuman fibroblasts to a lymphocyte-de-rived factor. Journal of ExperimentalMedicine •iii, 1188-1203.

Slavkin. H. C & Boyde. A. (1975) Ce-mentum; An epithelial secretory product?Journal of Dental Research S3, 157 (abstr#409).

Slavkin. H. C. (1976) Towards a cellular andmolecular understanding of periodontics;Cementogenesis revisited. Journal of Perl-odontology 47, 249-255.

Somerman. M. J.. Schiffmann. E.. Reddi. A.H. & Termine. J. (1982) Regulation of theattachment and migration of bone cells, invitro. Journal of Periodontal Research 17.527-529.

Somerman. M. J.. Archer. S. Y.. Hassel. TM. & Foster. R. A. (1987a) Enhancementby extracts of mineralized tissues of pro-tein productivity by human gingivalfibroblasts. in vitro. Archives of Oral Bi-ology 32, 879-883.

Somerman. M. J.. Archer. S. Y. Sheyeter.A. & Foster. R. A. (1987b) Protein pro-duction b> human gingival fibroblasts isenhanced by guanidine EDTA extracts ofcementum. in vitro. Journal of Periodonta]Research 22, 75-77.

Somerman. M. J.. Archer. S. Y.. Imm. G.R. & Foster. R. A. (1988a) A comparativestudy of human periodontal ligament cellsand gingival fibroblasts in vitro. Journal ofDenial Re.nearch 67. 66-70.

Somerman. M. J.. Foster. R. A.. Vorsteg. G..Progebin. K. & Wynn. R. L. (1988b) Ef-fects of minocydine on fibrobiast attach-ment and spreading. Journul of Pin-odontal Re.warch 23, 154-159

Somermar. M. J.. Prince. C. W., Butler. W.T . Foster. R. A.. Moehring. J. M. & Sauk.J. J. 1989) Cell attachment activity of the44 kilodalton bone phosphoprotein is notrestricted to bone cells. Matrix 9, 49-54

Tenenbaum. H. C. & Heersche. J. N. M.(1982) Differentiation of osteoblasts andformation of mineralized bone in vitro.Calcified Tissues International 34. 76-79.

Thesleff. I. (1996) Molecular mechanisms ofcell and tissue interactions during earlytooth development. The Anatomical Rec-.ord245. 151-161.

Vainio. S.. Karavunova. L. Jowett. A. &Thesleff. I. (1993) Identification of BMP-4as a signal mediating secondary inductionbetween epithelial and mesenchymaltissues during tooth development. Cell 75,45-58.

Address;

Stina Ge.'itreliusBIORA ABIdeon/MalmoS-205 12 Malmo. Sweden

Tel.; +46 40 32 13 30Fax; +46 40 32 13 55e-mail; stina.gestrelius(aibiora.se