Boneand Mineral, 18 (1992) 15-30
Elsevier BAM 00443
iginio Arzate, Steven
path Narayanan
Department of Pathology SM30 and Research Center in Oral Biology University of Washington School of Medicine, Seattle, WA, USA
(Received 4 October 1991) (Accepted 9 January 1992)
We have cultured cells from explants of a human cementum tumor. The cells obtained were multipolar, they formed network-like structures and they were alkaline-phosphatase positive. Immunostaining and Western blots using specific antibodies revealed that these cells produced bone sialoprotein and collagen types I and V, and they also mineralized in vitro. Conditioned medium was mitogenic to fibroblasts and mitogens present were separated by heparin-affinity chromatography. Based on affinity to heparin and antibody-inhibition studies, the heparin fractions were shown to contain cementurn-derived growth factor, platelet-derived growth factor and fibroblast growth factors. The cementum tumor cells, but not gingival fibroblasts, were stained positively by an antibody to cementum-derived attachment protein. ‘The attachment protein was separated by immunoaffinity chromatography, and Western blots revealed that the preparation contained 56-kDa and 43-kDa proteins as major bands. Cells pulse-labeled with radioactive amino acids contained a 43-kDa protein as the major component; however, this protein was absent after a cold chase in the presenceof cycloheximide,but 56.kDa, 39-kDa and 26.kDa species became prominent. These data indicated that the 5dkDa cementum attachment protein is derived from a 43. kDa precursor. Our data show that the celis cultured from the cementum tumor represent cementum ceils capable of synthesizing and secreting cementum proteins in culture.
Key words: Cementurn; Cementurn tumor; Cementoblast; Growth factor; Attachment protein
Abbrevialions: Blotto, nonfat dry milk; BSP, bone sialoprotein; CAP, cementum-derived fibroblast at-
tachment protein; CGF, cementum-derived growth factor; DV, Dulbecco-Vogt; FBS, fetal bovine serum; HERS, Hertwig’s epithelial root sheath; PBS, phosphate-buffered saline
Correspondence to: A.S. Narayanan, Department of Pathology SM 30, University of Washington School of Medicine, Seattle, WA 98195, USA. 0169~6009/92/$05 .OO@ 1992 Elsevier Science Publishers B.V. All rights reserved.
16 Introduction
Cementum is the mineralized connective tissue which covers the tooth root dentin and serves as the attachment site for collagen fibers of the adjacent periodontal ligament and gingival connective tissues. It resembles bone in composition and structure, however, it lacks a direct blood supply, lymph drainage and innervation. The organic matrix of cementum consists of collagens, primarily the type I species, noncollagenous proteins, phosphoproteins and proteoglycans [l-7]. While these are predominantly structural macromolecules, recent studies have indicated that cementurn contains additional moIecuIeswhich regulate the migration, growth and attachment of connective tissue cells [g-12]. Some of these molecules appear to be localized in cementurn, and they have been partially identified and eharacterized 113-15, Yonemura et al., manuscript submitted]. Two types of cells, cementoblasts and cementocytes, are associated with the cementum. The cementoblasts, which are believed to originate from undifferentiated mesenchymal cells localized in endosteal spaces of alveolar bone and paravascular areas of periodontal ligament [l&-18], produce the cementum matrix. The cementocytes are cementoblasts whichbecome entrapped by the matrix they secrete; they are relatively inactive and eventually degenerate. So far cells from cementurn have not been obtained and studied in culture. Therefore, it has not been possible to investigate regulation of cementum formation or the processes that affect the recruitment and differentiation of cementum-forming cells. The identity of cells that synthesize and secrete the cemcnaum-derivedgrowth factor, attachment proteins and other biologically active cementum components and regulation of production of these proteins have also not been examined. This lack of knowledge has hampered efforts to understand the mechanisms involved in new cementum formation and restoration of soft tissue attachment to tooth toot surfaces from which attachment has been lost due to inflammatory periodontitis. In this manuscript we describe the isolation of a cell strain which appears to manifest the cementum phenotype in culture. These cells wete cultured from a human cementurn tumor and propagated in subculture, and they were shown to synthesize at least two cementum proteins, the cementum-derived growth factor (CGF)‘and a 56-kDa attachment protein (CAP).
Materials
Antibody to human bone sialoprotein (BSP) LF-6 [19]was a generous gift from Dr. Larry Fisher, NIDR, NIH. Rat anti-human type 1 and V collagen antibodies [20] and anti-CGF monoclonal antibody, 2SN232,were raised in our laboratory. Monoclonal antibody HHF-35 was obtained from Dr. Allen Gown, Department of Pathology, Universityof Washington.This antibody, which recognizes42-kDa smooth muscle actin, immunostains smooth muscle cells in tissues [21]. FITC-conjugated secondary antibodies and histochemical staining kit for alkaline phosphatase were purchasedfrom SigmaChemicalCo., St. Louis, MO. Immobilon-PPVDF membrane
17
was obtained from illipore Corp., Bedford Amersham Life SC ce Products Inc., Arlin
nd radioactive amino acids from eights, IL.
Cells Surgical specimens were obtained from the mandible of a human patient who had a long history of diagnosed osteogenesis imperfecta as well as concomitant cementifying fibroma. The tumor was minced and placed in Dulbecco-Vogt medium (DV. medium) containing 10% fetal bovine serum and antibiotics, and cells were allowed conventional explant technique [22]. The cells obtained, referred hereentum cells or tumor cells for s~s~~licity, were maintained in DV-mediurn containing 15% S and 3% bovine pituitary extract and subcultured every three weeks (one to t split). Cells between 5 to IO transfer numbers were used for the experiments described here; at this stage the cells ma~ifeste osteoblast-like features (see later, Pig. lb-d). Human gingival fibroblasts and periodontal ligament cells were obtained as described previously 1221. Immunostaining ately 500-1000 cells/well were plated in g-well Lab-Tek chamber slides. After 72 h they were fixed with 3.5% paraformaldehyde for 5 min at room temperature and permeabilized with ice-cold acetone at -20°C for 2 min [23]. They were incubated with 1:lOO diluted primary antibodies in phosphate-buffered s containing 1 mg/ml bovine serum albumin for 1 h at 4”C, then with 1:lO jugated secondary antibodies (3 mg/ml t-anti-mouse IgG, goat-anti-rabbit IgG and rabbit-anti-rat-IgG for anti-CAP, and collagen antibodies, respectively) for 1 h at 4”C, and staini rg was visualized by indirect immunofluorescence [20]. Detectionof alkalinephosphatase Cells were plated at a density of 500-1000 cells/well in 4-well Lab-Tek chamber slides and incubated for 72 h in the growth medium. Then they were stained histochemically for the enzyme according to the instructions provided by the manufacturer. Briefly, the cells were fixed for 30 s and permeabilized as described above, stained at room temperature with a solution of alkaline-dye mixture for 15 min at room temperature and counterstained with Neutral Red for 2 min. Staining was evaluated by light microscopy. Isolationand separationof mitogenicfactors Conditioned medium was obtained from confluent cultures in 75-cm* flasks. The cells were washed 3x with PBS and then incubated with 10 ml PBS for 10 min at The PBS was discarded and then cells incubated in fresh serum-free medium. ditioned medium was collected membranes. It was dialyzed vs. 50
18 each of phenylmethane sulfonylfluoride and N-ethylmaleimide, 25 mM EDTA and 1pg!ml each of pepstatin and leupeptin, and loaded on a heparinSepharose column (0.9 x 10.0 cm) maintained at 4°C. The column was washed with 5-10 column volumes of buffer to remove unbound proteins and then eluted stepwise with 0.5 M, 1.0 M and 2.0 M NaCl as described elsewhere [13]. The eluates were concentrated and dialyzed vs. serum-free medium prior to mitogenic assays. Assay of mitogenic activity
Human gingival fibroblasts between the fourth and twelfth transfer numbers were used as target cells. Approximately 20 000 cells were plated in 24-wellplates, incubated overnight for the cells to attach and made quiescent by incubation with serum-free DV-medium for 48 h. The medium was replaced with fresh serum-free medium containing indicated mitogens. After 22 h fresh serum-free medium containing S.O@Yml [3H]thymidinewas added. Six h later the cells were washed with ice-cold PBS, taken in 1.ON NaOH and radioactivity counted [8]. Immunoaffinity purification of attachment protein
Conditioned medium obtained as described above was dialyzed against PBS and. loaded on a column containing 3.0 ml of CNBr-activated Sepharose-4B coupled to a monoclonal antibody raised against 56kDa cementum attachment protein (CAP) [15, Arzate et al., manuscript in preparation]. The column was washed with 20-30 column volumes of PBS to remove unbound proteins (the wash contained no material absorbingat 228 nm) and then eluted with 0.1 M glycine,pH 2.5 to recover bound proteins. The eluate was immediately neutralized to pH 7.0-7.4 with 1 N NaOH, concentrated and stored at -20°C. Fibroblast attachment assay
Confluent cultures of human gingival fibroblasts were labeled with 10 ~Cilml L[ssS]-methioninein methionine-free RPM1 medium overnight; the cells were trypsimzed aud ce!l attachment measured as described by McAllister et al. [ 121.Briefly, 48-wellCostar plastic plates not treated for tissue culture were added with different dilutions of attachment protein in PBS, incubated overnight at 4”C, washed and treated with 2 mg/ml of bovine serum albumin to block non-specific binding. The plates were incubatedfor 1 h with 1.5 x lo4 labeled cells and unbound cells removed by washing with PBS. Attached cells were solubilized in 1% NaDodSO, and radioactivity counted in a Packard 1500liquid scintillation counter. Plates coated with 5 pg/ml type I collagen served as positive controls, while those coated with serumfree medium were negative controls [151. Polyacrylamide gel electrophoresis
NaDodSO,-polyacrylamide gel electrophoresis was performed in 12.5% slab gels using a HoeRer SE 250 minigel apparatus, as described by Nakae et al. [ 131. tern
blot
After electrophoresis proteins were electroblotted on to a PVDF membrane using a
19
Bio-Rad transblot cell and proteins vistuaiized by staining with Ponceau branes were washed, blocked with 5% nonfat dry milk (blotto) for 1 h and then incubated for 1 h at room temperature with 1:lOOOdiluted primary antibodies (antiCAP monoclonal antibody or anti-BSP polyclonal antibody) in blotto. The membranes were washed for 30 min in blotto, and then incubated with 1:lOOOdiluted horseradish-peroxidase-conjugated secondary antibody (goat-anti-mouse or goatanti-rabbit IgG, respectively) for 1 h at room temperature. The membranes were washed with blotto and developed with diamino-benzidine. Labeling of cells Cultures in 75-cm2 flasks were preincubated in serum-free IN-medium leucine, glycine and proline for 1 h and then p yCi/ml of each of G[4,5-3 ter the labeling, cells were subjected to immunoaffinity chromatography. placed with fresh medium containin nonradioactive leucine, glycine and proline.
without
Protein Protein concentrations were determined with a Bio-Rad protein assay kit by the Bradford assay using bovine serum albumin as standard [24].
es&s
Cultures were established from cells obtained from explants of the cementum tumor in medium containing 10% FM. These cells initially manifested fibroblastic morphology (Fig. la). owever, after 4 to 5 passages (1:2 split) they assumed a distinctly different morphology. They were broad, flattened, compact and multipolar, especially when they reached confluency (Pig. lb). After 6 to 8 weeks in culture, *..* . _. -~--a;~= 1:l-n r4r~anhrre.2 rsrpfp p&-&an) (Fig. they became stratrtrea (rig. ic), and ~.WLUIPI~~~ Jclu’ccu.~v ..__.. _,______ Id). At this time their growth rate slackened, therefore the cells were subsequently maintained in the presence of 15% FBS and 3% bovine pituitary extract. The cells are being grown and maintained currently in this medium and they have also been stored frozen. Because these cells were obtained from a cementifying tumor, we first examined whether they manifest any osteoblast-like characteristics. Three experiments were done and gingival and periodontal ligament fibroblasts were compared as controls. In the first experiment, cultures were examined histochemically for the production of alkaline phosphatase. The results showed that all the cementum tumor cells exhibited intense alkaline phosphatase activity, and staining was present around the cell membrane and nuclei (Pig. 2a). In contrast, gingival fibroblasts did not stain for this enzyme (Fig. 2b). Periodontal ligament cells were also positive, but staining was not as intense as the cementum tumor cells (not shown). In the second experiment, the cells were immunostained with an antibody to BSP. The cementum cells
. 1. Morphology of cultured human cementum tumor cells. At early transfers the cells had a fibroblastic morphology (a). They became multipolar at transfer 4-5 (b) and stratlfied when confluencywas reached (c). Network-likestructures were also present (d). Bar, 20lm.
Fig. 2. Histological staining for alkaline phosphatase of (a) cementum cells and (b) human gingival fibroblasts.Bar, 20pm.
Fig. 3. Immunostaining of cementum cells (a) and gingival fibroblasts (b) with an antibody to bone sialoprotein. Bar, 20ym.
manifested intense staining by this antibody and staining was detected over the cell owev2r, the gingival fibroblasts were negative (Fig. 3b). ern blots revealed that conditioned mediu of cementum cells, but not gingival and periodontal fibroblasts, contained sever cross-reacting bands with -110-120, 70-75, 45 and 31 kDa protein bands cross-reacting with anti-B!?@antibody (data not shown). The cells were also incubated with /3-glycero dexamethasone and ascorbic acid for 30 days [25]; on staining with A , mineralization was detected in cultures of cementum cells, but not gingival or periodontal hgament fibroblasts (Fig. 4). The cementum cell cultures were also immunostained with a battery of antibodwever, no significant ies to determine whether they contain other cell types. munostaining was detectable for antibody to factor V which is specific for dothelial cells [26], or for epithelial-cell-specific anti-cytokeratin antibodies [27]. They stained positively with smooth-muscle-cell-specific antibody F-35 [21].
Fig. 4. Mineralization of cultured cementum cells and gingival and periodontal ligament fibroblasts. Cells in 35mm plates were incubated for 30 days with 10 mM p-glycerophosphate, lo-’ M dexamethasone and 35 mM ascorbic acid and then stained with Alizarin Red as described in [25]. (a) cementum cells; (b) periodontal ligament fibroblasts;(c) gingival tibroblasts. Bar, 0.5 cm.
22
Table 1 Mitogenic activity of heparin fractions of conditioned medium obtained from cementum tumor cells Fraction
Unbound 0.5 M NaCl 1.OM NaCl 2,OMN&l
%b
Mitogenic activity (corn x 10-3)a
(total)
13.4 + 4.6 13.7 Lb3.6 11.7 f 4.1 15.7 14.9
25.3 26.0 20.9 31.0
24 (19 f 8) 26 (29 !I 3) 20 (21 I 6) 30 (34 + 14)
n [3Hj-thymidincuptake x 1W3;cpm for serum-freemedium (negative control) was 3.3 a: 1,l x 10m3. b Data in parenthesis are mean $ SD for four different separations.
They were also strongly positive when stained with anti-type I and type V collagen antibodies (not shown). Production of cementurn-derived growth factor (CGF)
Cementurn has been shown to contain a 23”kDa growth factor; this growth factor, CGF, appears to be different from other factors and it is not present in dentin or soft periodontal tissues [8,13]. We examined whether cementum tumor cells produce and secrete CGF. Conditioned medium from cells was mitogenic to human gingival fibroblasts and activity was partially inhibited by 2SN232, a monoclonal antibody raised against the CGF. This indicated the presence of CGF in the medium (data not shown); therefore, the medium was subjected to heparin-affinity chromatography to separate CGF and other mitogens [ 13,281.The results showed that 29 + 3% of loaded activity was present in the 0.5 M fraction and 21 + 6 and 34 f 14% in 1.0 and 2.0 M eluates, respectively(n = 4 preparations).The remainingactivity(19 f ) was presentin the unboundfraction(Table 1). The 1.0 M and 2.0 M NaCl fractions were shown to represent acidic and basic fibroblast growth factors (a- and bFGF), respectively [13,28], therefore they were not processed further. Because
Table 2 Action of anti-PDGF and anti-CGF antibodies on mitogenic activity of 0.5 M NaCl heparin fraction Mitogen None 0.5 M fraction 0.5 M fraction 0.5 M fraction 10% FBS a Mean f SD of triplicates. b 2SN232.
Antibody
anti-PDGF anti-CGFb
cpm X 1O-3a
%
3.4 * 0.2 33.0 k 4.4 22.5 f 2.7 21.5 AI4.7 49.8 zk26.5
100 65 61
23
Fig. 5. Dot-blot assay of heparin fractionsof conditioned medium obtained from cementum tumor cells for CGF. Proteins on nitrocellulose were first incubated with monoclonal antibody 2SN232 and then with 12SI-rabbit antimouse IgM+IgG and visualized by fluorography. (a): I, unbound; 20.2 M NaCl eluate; 3,0.6 M eluatc; 4,2.0 M eluate. (b) Increasingconcentrations of authentic CGF. (c): 1, P 5O~llO%~FBS; 3,200 ng PDGF; 4,200 ng EGF.
are eluted by Cl.5 NaCl under the conditions used [13,28], this fraction was examined for the presence of these growth factors by antibody inhibition of mitogenic activity. Results showed that both 2SN232and anti-P tibodies removed 39 and 35% of mitogenic activity, indicating both CGF a were present (Table 2). To confirm the presence of CGF, the fraction was blotted on a nitrocellulose membra treated with 2SN232 antibody and then with 1251-labeled rabbit-anti-mouse (I ), Autoradiography revealed that the 0.5 NaCl fraction, but none of the otiier fractions, contained CGF (Fig. 5). Production of attachment protein
Conditioned medium was also examined for production of a fibroblast attachment protein which has been identified in cementum. This protein (CAP) is a 56-k molecule [15] and a monoclonal antibody raised agai turn, but not other tissues [Arzate et al., manuscript su were done. In the first experiment immunostaining by examined. The cementum cells stained positively and staining was present principally on the cell surface (Pig. 6a). In contrast, gin ival fibroblasts (and periodontal
Flg. 6. Immunostaining of cementum cells (a) and human gingival fibroblasts (b) with antibody raised against CAP. Bar, 20/cm.
24
Pig, 7, Fibroblast attachment by immunoaffinity-purified conditioned medium from cementum tumor cells using a column of a monoclonal antibody raised against CAP. 100 ml medium was used for purification and concentrated to 50@. Activity at different dilutions of the concentrate is shown.
ligament fibroblasts) used as controls were negative (Fig. 6b). In the second experiment, conditioned medium obtained from cementum tumor cells was subjected to immunoaffinity chromatography using the antibody coupled to Sepharose 4B, as described in the Methods. The bound proteins promoted the attachment of fibroblasts in a dose-dependent manner (Fig. 7). NaDodSQ-polyacrylamide gel electrophoresis revealed that the bound fraction contained very little material other
. 8. NaDodSO,-polyacrylamide gel electrophoresis of immunoaffinity-purified CAP fraction. (a): CAP fraction, Coomassie Blue stained. (b): Western blots; lane 1: CAP fraction, anti-CAP antibody; lane 2: CAP fraction, anti-BSP antibody; lane 3: human serum; anti-CAP antibody. The arrows from top to bottom indicate the migration of 68-kDa, 44-kDa, 22-kDa and 145kDa protein standards.
Fig. 9. Fluorography of metabolically labeled proteins obtained from cementum cells. (a) After 2-h pulse.Confluentculturesin X-cm* CPasks were pulse-labeled with [3 ]-leucine, glycine and -proline as described in the Methods. (b) After cold chase in the presence of cycloheximide and unlabeled amino acids for 1 h. Both cells and medium were subjected to immunoaffinitychromatography and NaDodSQ,polyacrylamide gel electraphsresis as described in the Methods.
than a 56-kDa protein component (Pig. 8a). stern blots revealed that the fraction contained 56-kDa and 43-kDa bands as major components lane 1). Additional bands with 39 k a were also visible, altho nor species. Human serum did not ss-react wtth the antibody (lane 3)) and conditioned medium fr fibroblasts processe through the antibody colu had no cross-reacting bands (data not sh . To rule out whether t resent BSP, the fraction was subjected to estern blots using anti however, it was negative (lane 2). E riments were performed to determine whether the protein bands detected by the estern blots are related. Cells were beled with radioactive amino acids as described in the
Table 3
Comparison of properties of cementum tumor cells with gingival and periodontal ligament fibroblasts Property -Morphology BSP Collagensb Cytokeratins Factor VIII SM-actinc CGF CAP
Cementum cell
Gingival fibroblast
Periodontal fibroblast
bones + +
fibroblast +
fibroblast + +
ND
ND ND ND
+ + +
a Fibroblasticduring early passages. b Types I and V. EBy immunostainingwith antibody HHF-35. ND: not done.
26 separated by immunoaffinity chromatography and NaDodSQ,-polyacrylamide gel electrophoresis. Fluorography of gels revealed that in pulse-labeled cells a band migrating near the origin and another 43-kDa protein were major components, and in addition 56-kDa, 31-kDa and 26-kDa bands were present as minor components (Fig. 9, lane a). However, after cold chase in the presence of cycloheximide, the 43kDa protein was absent and the 56kDa species became prominent. The 26-kDa band became more intense and a new 39-kDa protein appeared. The protein near the origin and the 31-kDa species remained unaffected (Fig. 9, lane b). The former appeared to have been retained nonspecifically by the antibody column because immunoaffinity-purified cementum extracts contained a protein band near the origin which cross-reacted with non-immune mouse serum (not shown). The properties of cementum tumor cells and gingival and periodontal ligament fibrobladts are summarized in Table 3.
iscussion Cementurn is associated with two cell types, the cementoblasts and cementocytes. Of these, only the cementoblasts are metabolically active; they are responsible for synthesizing cementoid matrix, while the cementocytes are relatively inactive. The cementobiasts are believed to originate from precursor cells present in the endosteal spaces of alveolar bone; these cells appear to migrate toward the dentin matrix and differentiate into cementoblasts once they contact the latter [16-18,291. Even though cells from other bones have been shown to form cementum-like structure [ 16,181,so far it has not been possible to culture and study cementoblasts in culture. We have cultured and propagated cells obtained from a human cementum tumor. These cells are not epithelial or endothelial cells based on immunostaining by celltype-specific antibodies. Although they stained positively with HHF-35, an antibody specificfor smooth muscle cells, these may not be smooth muscle cells because non-smooth muscle cells including fibroblasts are induced to express muscle actin when grown in culture [21, AM Gown, personal communication]. However, we did not immunostain cementum tissues to examine whether HHF-35 positive cells are present. The cells manifested flattened and stratified morphology with networklike structures typical of human bone cells [30]. They produced type I and V collagens similar to fibroblasts, however their other properties indicate that they are not fibroblasts. For example, unlike the latter they are alkaline-phosphatase positive, they synthesize BSP and mineralize in vitro, and these properties together are strongly suggestive of the osteoblast-like nature of the cementum cells. Even though cultures of periodontal ligament fibroblasts also contain alkaline-phosphatase-positive cells [31,32], these cells, unlike the tumor cells, are typically fibroblastic in morphology and other properties. While data strongly indicate that the cementum cells manifest osteoblast-like features, we did not characterize them further as osteoblasts because our primary objective was to determine whether they express cementum phenotype. Interestingly, during early passages cultures of these cells manifested fibroblastic morphology; whether the change in morphology
27 is due to a transformation of cellular phenotype during passage or due to selective growth of osteoblast-like cells is not clear. More importantly, the cultured cementum cells produce two cementum proteins, CGF and CAP. The CGF is a 23-kDa protein which is present in cementurn and it is not detectable in gingiva or periodontal ligament [8,13]. The CGF produced by the tumor cells was characterized by chromatographic properties and by cross-reactivity with a monoclonal antibody to bovine CGF [13, Yonemura et al., manuscript submitted]. In addition to the CGF, the cementum cells also produce PDGF, FGF and other growth factors, and their proportion resembles alveolar bone rather than cementum [13,28]. For example, PDGF is not detectable and Ps are minor comnents in alveolar F and FGFs are major co ponents in cementurn, while PD and other bones [13,28, Yonemura et al., manuscript submitted], duction of significant quantities of CGF indicates that the cultured display both bone and cementum phenotypes. Several possibilities can give rise to these results. For example, these cells were derived from a tumor and tumor cells are known to produce several growth factors. The culture may contain a mixture of cells representing cementum as well as other phenotypes. Alternatively, these cells may manifest features of cementoblast progenitor cells which are reside in the alveolar bone. Nevertheless, synthesis and secretion of CG protein which appears to be localized in cementum but not in other tissues [ 15, Arzate et al., manuscript in preparation], indicates that we have cultured putative cementum cells capable of synthesizing cementum proteins. Western blots revealed several protein bands cross-reacting with the anti-CAP antibody. The 56-kDa and 43-kDa components were the major species, while additional bands with 39-kDa and 26-kDa were also present, although barely visible. Based on cross-reactivity with antibodies, we have previously observed that these proteins do not cross-react with antibadies to rat osteopontin or human [ 12,151. Immunoaffinity-purified protein obtained from human cementum extracts were also negative in Westerns using polyclonal anti-porcine osteopontin which crossreacts with human species (J Sodek, personal communication). Pulse-chase studies indicate that these proteins are possibly derived from a common precursor of 43kDa, presumably by one or more posttranslational processing mechanisms such as glycosylation, sulfation, phosphorylation or proteolysis. Although osteo also synthesized in 44 kDa, 55 kDa and possibly other sizes [33,34], the 43tein produced by cementum tumor cells does not appear to be osteopontin because thrombin does not generate 26-kDa peptides from 43- and 56.kDa proteins (data not shown), and while osteopontin is present in most calcified structures, CAP is restricted to cementum [Arzate et al., manu ipt submitted]. Interestingly, 43- and 56-kDa bands are major species in estern blots, additional band by fluorography were either minor species or not detectable; the reasons for this are and osteopontin are different pronot known. While these data indicate that C AP may be a variant of osteopontin teins, they do not rule out the possibili formed by differential splicing of its m A and/or differences in posttranslational processing such as glycosylation, phosphorylation and sulphation. Irrespective of their origin and phenotype, we believe that the cultured cemen-
28 turn tumor cells will be useful for at least three purposes. First, they could be used to obtain sufficient quantities of biologically active cementum components as this has been difficult so far due to the paucity of cementum in human teeth and laborious dissection procedures necessary. Second, the cells could be used to study the biosynthesis and processing of CAP and other cementum proteins. Third, early cementogenesis is believed to involve deposition of extracelhtlar matrix produced by Hertwig’sepithelial root sheath (HERS) cells on dentin, disruption of the HERS, migration and attachment of ectomesenchymal cells from dental follicle on to the matrix, their differentiation into cemeutoblasts and laying down of cementum [35]. Recent S cells may induce cementowork has demonstrated that gene products of genesis, and that these cells synthesize 72-kDa and 26-kDa intermediate cementum proteins and possibly osteopontin [35,36]. In adult cementurn, maturation of cementoblasts involves the directed migration of cementum progenitor cells, their adhesion to matrix and subsequent differentiation [18,29]. The cementum cells described here can serve as an in vitro model to examine these processes and to study how cementum formation and differentiation are regulated.
Acknowledgement
This work was supported by NIH grant DE-08229. We thank Drs. Larry Fisher and Allen Gown for their generous gift of antibodies. The assistance of Xidong Li in harvesting cemeutum is highly appreciated.
References 1 Birkedal-Hansen H, Butler WT, Taylor RC. Proteins of the pcriodontium. Characterization of the insoluble collagens of bovine dental cementum. Calcif Tissue Res 1977;23:39-44. 2 Christner P, Robinson P, Clark CC. A preliminary characterization of human cementum collagen. Calcif Tissue Res 1977;23:147- 150. 3 Smith AJ, Leaver AG, Smith G. The amino acid composition of the non-collagenous organic matrix of human cementum. Arch Oral Bioll983;28: 1047- 1054. 4 Glimcher MJ, Lefteriou B. Soluble glycosylated phosphoproteins of cementurn. Calcif Tissue Int 1989;45:165- 172. 5 McCurdy SP, Clarkson BH, Speirs RL, Feagin FF. Phosphoprotein extraction from the dentine/cementum complex of human tooth roots. Arch Oral Bioll990;35:347-357. 6 Bartold PM, Reinboth B, Nakae H, Narayanan AS, Page RC. Proteoglycans of bovine cementurn: isolation and characterization. Matrix 199O;lO:lo- 19. 7 Bartold PM, Miki Y, McAllister B, Narayanan AS, Page RC. Glycosaminoglycans of human cemenCum.J Periodont Hes 1988;23:13- 17. 8 Miki Y, Narayanan, AS, Page RC. Mitofwic activity of cementum components to gingival fibroblasts. J Dent Res 1987;66:1399- 1403. 9 Somerman MJ, Archer SY, ShtGyer A, Foster RA. Protein production by human gingival fibroblasts is enhanced by guanidine EDTA extracts of cementurn. J Peridont Res 1987;22:75-77. 10 Nishimura K, Hayashi M, Matsuda K, Shigeyama Y, Yamasaki A, Yamaoka AJ. The chemoattractive potency of periodontal ligament, cementum and dentin for human gingival fibroblasts. J Peridont Res 1989;24:146-148.
29 11 Somerman MJ, Foster RA, Imm GM, Sauk JJ, Archer SY. Periodontal ligament ceIIs and gingival fihblasts respond differently to attachment factors in vitro. J Periodonto] lg8g;6&73_77. 12 McAllister B, Narayanan AS, Miki Y, Page RC. Isolation of a fibroblast attachment protein from cementum, J Periodont Res 1990;25:99-105. 13 Nakae H, Narayanan AS, Raines E, Page RC. Isolation and partial characterization of mitogenic factors from cementum. Biochemistry 1991;30:7047-7052. 14 Somerman MJ, Sauk JJ, Foster RA, Dickerson K, Norris K, Argraves WS. Cell attachment activity of cementum: bone sialoprotein II identified in cementum. J Periodont Res 1991;26:10-16. 15 Olson S, Arzate H, Narayanan AS, Page RC. Cell attachment activity of cementum proteins and mechanism of endotoxin inhibition. J Dent Res 1991;10:1272-1277. 16 Melcher AH, Cheong T, Cox J, Nemeth E, Shiga A. Synthesis of cementum-like tissue in vitro by cells cultured from bone: a light and electron microscope study. J Periodont Res 1986;21:592-612. 17 McCulloch CAG, Nemeth HE, Lowenberg, B, Melcher AH. Paravascular cells in endosteal spaces of alveoidr bone contribute to periodontal ligament cell populations, Anat Ret 1987;219:233-242. 18 Melcher AH. Does the developmental origin of cementum, periodontal ligament and bone predetermine their behavior in adults? In: Guggenheim B, ed. Periodontology Today. Basel: Krager AG, 1988;6-14. 19 Fisher LW, Hawkins GR, Tuross N, Termine JD. Purification and partial characterization of small proteoglycans I and 11, bone sialoproteins I and II, and osteocalcin from the mineral compartment of developing human bone. J Biol Chem 1987;262:9702-9708. 20 Narayanan AS, Clagett JA, Page RC. Effect of inflammation on the distribution of collagen types, I, III, IV, and V and type I trimer and fibronectin in human gingivae. J Dent Res 1985;64:111l-l 116. 21 Tsukada T, Tippens D, Gordon D, Ross R, Gown AM. HHF35, a muscle-actin specific mouoclonal antibody. I. Immunocytochemical and biochemical characterization. Am J PathoI1987;126:51-60. 22 Narayanan AS, Page RC. Biochemical characterization of collagens synthesized by fibroblasts derived from normal and diseased human gingivha. J BioI Chem 1976;251:5464-5471. 23 Narayanan AS, Page RC. Synthesis of type V collagen by fibroblasts derived from normal. inflamed and hyperplastic human connective tissues. Collagen Rel Res 1985;5:297-304. 24 Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 1976;72:248-254. 25 Noff D, Pitaru S, Savion N. Basic fibroblast growth factor enhances the capacity of bone marrow cells to form bone-like nodules in vitro. FEBS Letters 1989;250:619-621. 26 Jaffe EAL, Hoyer W, Nachman RL. Synthesis of anti-hemophilic factor antigen by cultured human endothelial cells. J Clin Invest 1973;52:2757-2764. 27 Weiss RA, Guillet GYA, Freedberg IM, Farmer ER, Small EA. Weiss MM, Sun T-T. The use of monoclonal antibody to keratin in human epidermal disease: alterati. QS in immunocytochemical staining pattern. J Invest Dermatol1983;81:224-230. 28 Hauschka PV. Mavrakos AE, Iafrati MD, Doleman SE, Klagsbrun M. Growth factors in bone matrix. Isolation of multiple types by affinity chromatography on heparin-sepharose. J Biol Chem 1986;261:12665-12674. 29 Cho, M-I, Garant PR. Radioautographic study of [“Hlmannose utilization during cementoblast differentiation, formation of acellular cementum, and development of periodontal ligament principal fibers. Anat Ret 1989;223:209-222. 30 AufmkoIk B, Hauschka PV, Schwartz ER. Characterization of human bone cells in culture. Cakif Tissue Int 1985;37:228-235. 31 Nojima N, &hay&i M, Shionome M, Takahashi, N, Suda T, Hasegawa K. Fibroblastic cells derived from bovine periodontal ligaments have the phenotypes of osteoblasts. J Periodont Res 1990;25:179-185. 32 Somerman MJ, Archer SY, Imm GR, Foster RA. A comparative study of human periodontal ligament cells and gingival fibroblasts in vitro. J Dent Res 198+6’/. $6-70. 33 Nagata ‘I’, Bellows CG, Kasugai S, Butler WT. Sodek J. Biosynthesis of bone Proteins ISPP-1 (secreted phosphoprotein-1 , osteopontin), BSP (bone sialoprotein) and SPARC (osteonectin)] in association with mineralized-tissue formation by fetal-rat calvarial cells in culture. Biochem J 1991;274:513-520.
34 Kasugai S, Zhang Q, Overall CM, Wrana JL, Butler WT, Sodek, J. Differential regulation of the 55 and 44 kDa forms of secreted phosphoprotein 1 (SPP-1, osteopontin) in normal and transformed rat bone cells by osteotwpic hormones, growth factors and a tumor promoter. Bone Miner 1991;13:235-250. 35 Slavkin HC, Bringas P Jr, Bessem C, Santos V, Nakamura M, Hsu My, Snead ML, Zcichner-David M, Fin&am AG. Hertwig’s epithelial root sheath differentiation and initial cementum and bone formation during long-term organ culture of mouse mandibular first molars using sepJmless, chemically-defined medium. J Periodont Res 1988;23:28-40. 36 Somerman MJ, Shroff B, Agraves WS, Morrison G, Craig AM, Denhardt DT, Foster RA, Sauk JJ. Expression of attachment proteins during cementogenesis. J Biol Buccale 1990;18:2fi7-214.
Elsevier BAM 00443
iginio Arzate, Steven
path Narayanan
Department of Pathology SM30 and Research Center in Oral Biology University of Washington School of Medicine, Seattle, WA, USA
(Received 4 October 1991) (Accepted 9 January 1992)
We have cultured cells from explants of a human cementum tumor. The cells obtained were multipolar, they formed network-like structures and they were alkaline-phosphatase positive. Immunostaining and Western blots using specific antibodies revealed that these cells produced bone sialoprotein and collagen types I and V, and they also mineralized in vitro. Conditioned medium was mitogenic to fibroblasts and mitogens present were separated by heparin-affinity chromatography. Based on affinity to heparin and antibody-inhibition studies, the heparin fractions were shown to contain cementurn-derived growth factor, platelet-derived growth factor and fibroblast growth factors. The cementum tumor cells, but not gingival fibroblasts, were stained positively by an antibody to cementum-derived attachment protein. ‘The attachment protein was separated by immunoaffinity chromatography, and Western blots revealed that the preparation contained 56-kDa and 43-kDa proteins as major bands. Cells pulse-labeled with radioactive amino acids contained a 43-kDa protein as the major component; however, this protein was absent after a cold chase in the presenceof cycloheximide,but 56.kDa, 39-kDa and 26.kDa species became prominent. These data indicated that the 5dkDa cementum attachment protein is derived from a 43. kDa precursor. Our data show that the celis cultured from the cementum tumor represent cementum ceils capable of synthesizing and secreting cementum proteins in culture.
Key words: Cementurn; Cementurn tumor; Cementoblast; Growth factor; Attachment protein
Abbrevialions: Blotto, nonfat dry milk; BSP, bone sialoprotein; CAP, cementum-derived fibroblast at-
tachment protein; CGF, cementum-derived growth factor; DV, Dulbecco-Vogt; FBS, fetal bovine serum; HERS, Hertwig’s epithelial root sheath; PBS, phosphate-buffered saline
Correspondence to: A.S. Narayanan, Department of Pathology SM 30, University of Washington School of Medicine, Seattle, WA 98195, USA. 0169~6009/92/$05 .OO@ 1992 Elsevier Science Publishers B.V. All rights reserved.
16 Introduction
Cementum is the mineralized connective tissue which covers the tooth root dentin and serves as the attachment site for collagen fibers of the adjacent periodontal ligament and gingival connective tissues. It resembles bone in composition and structure, however, it lacks a direct blood supply, lymph drainage and innervation. The organic matrix of cementum consists of collagens, primarily the type I species, noncollagenous proteins, phosphoproteins and proteoglycans [l-7]. While these are predominantly structural macromolecules, recent studies have indicated that cementurn contains additional moIecuIeswhich regulate the migration, growth and attachment of connective tissue cells [g-12]. Some of these molecules appear to be localized in cementurn, and they have been partially identified and eharacterized 113-15, Yonemura et al., manuscript submitted]. Two types of cells, cementoblasts and cementocytes, are associated with the cementum. The cementoblasts, which are believed to originate from undifferentiated mesenchymal cells localized in endosteal spaces of alveolar bone and paravascular areas of periodontal ligament [l&-18], produce the cementum matrix. The cementocytes are cementoblasts whichbecome entrapped by the matrix they secrete; they are relatively inactive and eventually degenerate. So far cells from cementurn have not been obtained and studied in culture. Therefore, it has not been possible to investigate regulation of cementum formation or the processes that affect the recruitment and differentiation of cementum-forming cells. The identity of cells that synthesize and secrete the cemcnaum-derivedgrowth factor, attachment proteins and other biologically active cementum components and regulation of production of these proteins have also not been examined. This lack of knowledge has hampered efforts to understand the mechanisms involved in new cementum formation and restoration of soft tissue attachment to tooth toot surfaces from which attachment has been lost due to inflammatory periodontitis. In this manuscript we describe the isolation of a cell strain which appears to manifest the cementum phenotype in culture. These cells wete cultured from a human cementurn tumor and propagated in subculture, and they were shown to synthesize at least two cementum proteins, the cementum-derived growth factor (CGF)‘and a 56-kDa attachment protein (CAP).
Materials
Antibody to human bone sialoprotein (BSP) LF-6 [19]was a generous gift from Dr. Larry Fisher, NIDR, NIH. Rat anti-human type 1 and V collagen antibodies [20] and anti-CGF monoclonal antibody, 2SN232,were raised in our laboratory. Monoclonal antibody HHF-35 was obtained from Dr. Allen Gown, Department of Pathology, Universityof Washington.This antibody, which recognizes42-kDa smooth muscle actin, immunostains smooth muscle cells in tissues [21]. FITC-conjugated secondary antibodies and histochemical staining kit for alkaline phosphatase were purchasedfrom SigmaChemicalCo., St. Louis, MO. Immobilon-PPVDF membrane
17
was obtained from illipore Corp., Bedford Amersham Life SC ce Products Inc., Arlin
nd radioactive amino acids from eights, IL.
Cells Surgical specimens were obtained from the mandible of a human patient who had a long history of diagnosed osteogenesis imperfecta as well as concomitant cementifying fibroma. The tumor was minced and placed in Dulbecco-Vogt medium (DV. medium) containing 10% fetal bovine serum and antibiotics, and cells were allowed conventional explant technique [22]. The cells obtained, referred hereentum cells or tumor cells for s~s~~licity, were maintained in DV-mediurn containing 15% S and 3% bovine pituitary extract and subcultured every three weeks (one to t split). Cells between 5 to IO transfer numbers were used for the experiments described here; at this stage the cells ma~ifeste osteoblast-like features (see later, Pig. lb-d). Human gingival fibroblasts and periodontal ligament cells were obtained as described previously 1221. Immunostaining ately 500-1000 cells/well were plated in g-well Lab-Tek chamber slides. After 72 h they were fixed with 3.5% paraformaldehyde for 5 min at room temperature and permeabilized with ice-cold acetone at -20°C for 2 min [23]. They were incubated with 1:lOO diluted primary antibodies in phosphate-buffered s containing 1 mg/ml bovine serum albumin for 1 h at 4”C, then with 1:lO jugated secondary antibodies (3 mg/ml t-anti-mouse IgG, goat-anti-rabbit IgG and rabbit-anti-rat-IgG for anti-CAP, and collagen antibodies, respectively) for 1 h at 4”C, and staini rg was visualized by indirect immunofluorescence [20]. Detectionof alkalinephosphatase Cells were plated at a density of 500-1000 cells/well in 4-well Lab-Tek chamber slides and incubated for 72 h in the growth medium. Then they were stained histochemically for the enzyme according to the instructions provided by the manufacturer. Briefly, the cells were fixed for 30 s and permeabilized as described above, stained at room temperature with a solution of alkaline-dye mixture for 15 min at room temperature and counterstained with Neutral Red for 2 min. Staining was evaluated by light microscopy. Isolationand separationof mitogenicfactors Conditioned medium was obtained from confluent cultures in 75-cm* flasks. The cells were washed 3x with PBS and then incubated with 10 ml PBS for 10 min at The PBS was discarded and then cells incubated in fresh serum-free medium. ditioned medium was collected membranes. It was dialyzed vs. 50
18 each of phenylmethane sulfonylfluoride and N-ethylmaleimide, 25 mM EDTA and 1pg!ml each of pepstatin and leupeptin, and loaded on a heparinSepharose column (0.9 x 10.0 cm) maintained at 4°C. The column was washed with 5-10 column volumes of buffer to remove unbound proteins and then eluted stepwise with 0.5 M, 1.0 M and 2.0 M NaCl as described elsewhere [13]. The eluates were concentrated and dialyzed vs. serum-free medium prior to mitogenic assays. Assay of mitogenic activity
Human gingival fibroblasts between the fourth and twelfth transfer numbers were used as target cells. Approximately 20 000 cells were plated in 24-wellplates, incubated overnight for the cells to attach and made quiescent by incubation with serum-free DV-medium for 48 h. The medium was replaced with fresh serum-free medium containing indicated mitogens. After 22 h fresh serum-free medium containing S.O@Yml [3H]thymidinewas added. Six h later the cells were washed with ice-cold PBS, taken in 1.ON NaOH and radioactivity counted [8]. Immunoaffinity purification of attachment protein
Conditioned medium obtained as described above was dialyzed against PBS and. loaded on a column containing 3.0 ml of CNBr-activated Sepharose-4B coupled to a monoclonal antibody raised against 56kDa cementum attachment protein (CAP) [15, Arzate et al., manuscript in preparation]. The column was washed with 20-30 column volumes of PBS to remove unbound proteins (the wash contained no material absorbingat 228 nm) and then eluted with 0.1 M glycine,pH 2.5 to recover bound proteins. The eluate was immediately neutralized to pH 7.0-7.4 with 1 N NaOH, concentrated and stored at -20°C. Fibroblast attachment assay
Confluent cultures of human gingival fibroblasts were labeled with 10 ~Cilml L[ssS]-methioninein methionine-free RPM1 medium overnight; the cells were trypsimzed aud ce!l attachment measured as described by McAllister et al. [ 121.Briefly, 48-wellCostar plastic plates not treated for tissue culture were added with different dilutions of attachment protein in PBS, incubated overnight at 4”C, washed and treated with 2 mg/ml of bovine serum albumin to block non-specific binding. The plates were incubatedfor 1 h with 1.5 x lo4 labeled cells and unbound cells removed by washing with PBS. Attached cells were solubilized in 1% NaDodSO, and radioactivity counted in a Packard 1500liquid scintillation counter. Plates coated with 5 pg/ml type I collagen served as positive controls, while those coated with serumfree medium were negative controls [151. Polyacrylamide gel electrophoresis
NaDodSO,-polyacrylamide gel electrophoresis was performed in 12.5% slab gels using a HoeRer SE 250 minigel apparatus, as described by Nakae et al. [ 131. tern
blot
After electrophoresis proteins were electroblotted on to a PVDF membrane using a
19
Bio-Rad transblot cell and proteins vistuaiized by staining with Ponceau branes were washed, blocked with 5% nonfat dry milk (blotto) for 1 h and then incubated for 1 h at room temperature with 1:lOOOdiluted primary antibodies (antiCAP monoclonal antibody or anti-BSP polyclonal antibody) in blotto. The membranes were washed for 30 min in blotto, and then incubated with 1:lOOOdiluted horseradish-peroxidase-conjugated secondary antibody (goat-anti-mouse or goatanti-rabbit IgG, respectively) for 1 h at room temperature. The membranes were washed with blotto and developed with diamino-benzidine. Labeling of cells Cultures in 75-cm2 flasks were preincubated in serum-free IN-medium leucine, glycine and proline for 1 h and then p yCi/ml of each of G[4,5-3 ter the labeling, cells were subjected to immunoaffinity chromatography. placed with fresh medium containin nonradioactive leucine, glycine and proline.
without
Protein Protein concentrations were determined with a Bio-Rad protein assay kit by the Bradford assay using bovine serum albumin as standard [24].
es&s
Cultures were established from cells obtained from explants of the cementum tumor in medium containing 10% FM. These cells initially manifested fibroblastic morphology (Fig. la). owever, after 4 to 5 passages (1:2 split) they assumed a distinctly different morphology. They were broad, flattened, compact and multipolar, especially when they reached confluency (Pig. lb). After 6 to 8 weeks in culture, *..* . _. -~--a;~= 1:l-n r4r~anhrre.2 rsrpfp p&-&an) (Fig. they became stratrtrea (rig. ic), and ~.WLUIPI~~~ Jclu’ccu.~v ..__.. _,______ Id). At this time their growth rate slackened, therefore the cells were subsequently maintained in the presence of 15% FBS and 3% bovine pituitary extract. The cells are being grown and maintained currently in this medium and they have also been stored frozen. Because these cells were obtained from a cementifying tumor, we first examined whether they manifest any osteoblast-like characteristics. Three experiments were done and gingival and periodontal ligament fibroblasts were compared as controls. In the first experiment, cultures were examined histochemically for the production of alkaline phosphatase. The results showed that all the cementum tumor cells exhibited intense alkaline phosphatase activity, and staining was present around the cell membrane and nuclei (Pig. 2a). In contrast, gingival fibroblasts did not stain for this enzyme (Fig. 2b). Periodontal ligament cells were also positive, but staining was not as intense as the cementum tumor cells (not shown). In the second experiment, the cells were immunostained with an antibody to BSP. The cementum cells
. 1. Morphology of cultured human cementum tumor cells. At early transfers the cells had a fibroblastic morphology (a). They became multipolar at transfer 4-5 (b) and stratlfied when confluencywas reached (c). Network-likestructures were also present (d). Bar, 20lm.
Fig. 2. Histological staining for alkaline phosphatase of (a) cementum cells and (b) human gingival fibroblasts.Bar, 20pm.
Fig. 3. Immunostaining of cementum cells (a) and gingival fibroblasts (b) with an antibody to bone sialoprotein. Bar, 20ym.
manifested intense staining by this antibody and staining was detected over the cell owev2r, the gingival fibroblasts were negative (Fig. 3b). ern blots revealed that conditioned mediu of cementum cells, but not gingival and periodontal fibroblasts, contained sever cross-reacting bands with -110-120, 70-75, 45 and 31 kDa protein bands cross-reacting with anti-B!?@antibody (data not shown). The cells were also incubated with /3-glycero dexamethasone and ascorbic acid for 30 days [25]; on staining with A , mineralization was detected in cultures of cementum cells, but not gingival or periodontal hgament fibroblasts (Fig. 4). The cementum cell cultures were also immunostained with a battery of antibodwever, no significant ies to determine whether they contain other cell types. munostaining was detectable for antibody to factor V which is specific for dothelial cells [26], or for epithelial-cell-specific anti-cytokeratin antibodies [27]. They stained positively with smooth-muscle-cell-specific antibody F-35 [21].
Fig. 4. Mineralization of cultured cementum cells and gingival and periodontal ligament fibroblasts. Cells in 35mm plates were incubated for 30 days with 10 mM p-glycerophosphate, lo-’ M dexamethasone and 35 mM ascorbic acid and then stained with Alizarin Red as described in [25]. (a) cementum cells; (b) periodontal ligament fibroblasts;(c) gingival tibroblasts. Bar, 0.5 cm.
22
Table 1 Mitogenic activity of heparin fractions of conditioned medium obtained from cementum tumor cells Fraction
Unbound 0.5 M NaCl 1.OM NaCl 2,OMN&l
%b
Mitogenic activity (corn x 10-3)a
(total)
13.4 + 4.6 13.7 Lb3.6 11.7 f 4.1 15.7 14.9
25.3 26.0 20.9 31.0
24 (19 f 8) 26 (29 !I 3) 20 (21 I 6) 30 (34 + 14)
n [3Hj-thymidincuptake x 1W3;cpm for serum-freemedium (negative control) was 3.3 a: 1,l x 10m3. b Data in parenthesis are mean $ SD for four different separations.
They were also strongly positive when stained with anti-type I and type V collagen antibodies (not shown). Production of cementurn-derived growth factor (CGF)
Cementurn has been shown to contain a 23”kDa growth factor; this growth factor, CGF, appears to be different from other factors and it is not present in dentin or soft periodontal tissues [8,13]. We examined whether cementum tumor cells produce and secrete CGF. Conditioned medium from cells was mitogenic to human gingival fibroblasts and activity was partially inhibited by 2SN232, a monoclonal antibody raised against the CGF. This indicated the presence of CGF in the medium (data not shown); therefore, the medium was subjected to heparin-affinity chromatography to separate CGF and other mitogens [ 13,281.The results showed that 29 + 3% of loaded activity was present in the 0.5 M fraction and 21 + 6 and 34 f 14% in 1.0 and 2.0 M eluates, respectively(n = 4 preparations).The remainingactivity(19 f ) was presentin the unboundfraction(Table 1). The 1.0 M and 2.0 M NaCl fractions were shown to represent acidic and basic fibroblast growth factors (a- and bFGF), respectively [13,28], therefore they were not processed further. Because
Table 2 Action of anti-PDGF and anti-CGF antibodies on mitogenic activity of 0.5 M NaCl heparin fraction Mitogen None 0.5 M fraction 0.5 M fraction 0.5 M fraction 10% FBS a Mean f SD of triplicates. b 2SN232.
Antibody
anti-PDGF anti-CGFb
cpm X 1O-3a
%
3.4 * 0.2 33.0 k 4.4 22.5 f 2.7 21.5 AI4.7 49.8 zk26.5
100 65 61
23
Fig. 5. Dot-blot assay of heparin fractionsof conditioned medium obtained from cementum tumor cells for CGF. Proteins on nitrocellulose were first incubated with monoclonal antibody 2SN232 and then with 12SI-rabbit antimouse IgM+IgG and visualized by fluorography. (a): I, unbound; 20.2 M NaCl eluate; 3,0.6 M eluatc; 4,2.0 M eluate. (b) Increasingconcentrations of authentic CGF. (c): 1, P 5O~llO%~FBS; 3,200 ng PDGF; 4,200 ng EGF.
are eluted by Cl.5 NaCl under the conditions used [13,28], this fraction was examined for the presence of these growth factors by antibody inhibition of mitogenic activity. Results showed that both 2SN232and anti-P tibodies removed 39 and 35% of mitogenic activity, indicating both CGF a were present (Table 2). To confirm the presence of CGF, the fraction was blotted on a nitrocellulose membra treated with 2SN232 antibody and then with 1251-labeled rabbit-anti-mouse (I ), Autoradiography revealed that the 0.5 NaCl fraction, but none of the otiier fractions, contained CGF (Fig. 5). Production of attachment protein
Conditioned medium was also examined for production of a fibroblast attachment protein which has been identified in cementum. This protein (CAP) is a 56-k molecule [15] and a monoclonal antibody raised agai turn, but not other tissues [Arzate et al., manuscript su were done. In the first experiment immunostaining by examined. The cementum cells stained positively and staining was present principally on the cell surface (Pig. 6a). In contrast, gin ival fibroblasts (and periodontal
Flg. 6. Immunostaining of cementum cells (a) and human gingival fibroblasts (b) with antibody raised against CAP. Bar, 20/cm.
24
Pig, 7, Fibroblast attachment by immunoaffinity-purified conditioned medium from cementum tumor cells using a column of a monoclonal antibody raised against CAP. 100 ml medium was used for purification and concentrated to 50@. Activity at different dilutions of the concentrate is shown.
ligament fibroblasts) used as controls were negative (Fig. 6b). In the second experiment, conditioned medium obtained from cementum tumor cells was subjected to immunoaffinity chromatography using the antibody coupled to Sepharose 4B, as described in the Methods. The bound proteins promoted the attachment of fibroblasts in a dose-dependent manner (Fig. 7). NaDodSQ-polyacrylamide gel electrophoresis revealed that the bound fraction contained very little material other
. 8. NaDodSO,-polyacrylamide gel electrophoresis of immunoaffinity-purified CAP fraction. (a): CAP fraction, Coomassie Blue stained. (b): Western blots; lane 1: CAP fraction, anti-CAP antibody; lane 2: CAP fraction, anti-BSP antibody; lane 3: human serum; anti-CAP antibody. The arrows from top to bottom indicate the migration of 68-kDa, 44-kDa, 22-kDa and 145kDa protein standards.
Fig. 9. Fluorography of metabolically labeled proteins obtained from cementum cells. (a) After 2-h pulse.Confluentculturesin X-cm* CPasks were pulse-labeled with [3 ]-leucine, glycine and -proline as described in the Methods. (b) After cold chase in the presence of cycloheximide and unlabeled amino acids for 1 h. Both cells and medium were subjected to immunoaffinitychromatography and NaDodSQ,polyacrylamide gel electraphsresis as described in the Methods.
than a 56-kDa protein component (Pig. 8a). stern blots revealed that the fraction contained 56-kDa and 43-kDa bands as major components lane 1). Additional bands with 39 k a were also visible, altho nor species. Human serum did not ss-react wtth the antibody (lane 3)) and conditioned medium fr fibroblasts processe through the antibody colu had no cross-reacting bands (data not sh . To rule out whether t resent BSP, the fraction was subjected to estern blots using anti however, it was negative (lane 2). E riments were performed to determine whether the protein bands detected by the estern blots are related. Cells were beled with radioactive amino acids as described in the
Table 3
Comparison of properties of cementum tumor cells with gingival and periodontal ligament fibroblasts Property -Morphology BSP Collagensb Cytokeratins Factor VIII SM-actinc CGF CAP
Cementum cell
Gingival fibroblast
Periodontal fibroblast
bones + +
fibroblast +
fibroblast + +
ND
ND ND ND
+ + +
a Fibroblasticduring early passages. b Types I and V. EBy immunostainingwith antibody HHF-35. ND: not done.
26 separated by immunoaffinity chromatography and NaDodSQ,-polyacrylamide gel electrophoresis. Fluorography of gels revealed that in pulse-labeled cells a band migrating near the origin and another 43-kDa protein were major components, and in addition 56-kDa, 31-kDa and 26-kDa bands were present as minor components (Fig. 9, lane a). However, after cold chase in the presence of cycloheximide, the 43kDa protein was absent and the 56kDa species became prominent. The 26-kDa band became more intense and a new 39-kDa protein appeared. The protein near the origin and the 31-kDa species remained unaffected (Fig. 9, lane b). The former appeared to have been retained nonspecifically by the antibody column because immunoaffinity-purified cementum extracts contained a protein band near the origin which cross-reacted with non-immune mouse serum (not shown). The properties of cementum tumor cells and gingival and periodontal ligament fibrobladts are summarized in Table 3.
iscussion Cementurn is associated with two cell types, the cementoblasts and cementocytes. Of these, only the cementoblasts are metabolically active; they are responsible for synthesizing cementoid matrix, while the cementocytes are relatively inactive. The cementobiasts are believed to originate from precursor cells present in the endosteal spaces of alveolar bone; these cells appear to migrate toward the dentin matrix and differentiate into cementoblasts once they contact the latter [16-18,291. Even though cells from other bones have been shown to form cementum-like structure [ 16,181,so far it has not been possible to culture and study cementoblasts in culture. We have cultured and propagated cells obtained from a human cementum tumor. These cells are not epithelial or endothelial cells based on immunostaining by celltype-specific antibodies. Although they stained positively with HHF-35, an antibody specificfor smooth muscle cells, these may not be smooth muscle cells because non-smooth muscle cells including fibroblasts are induced to express muscle actin when grown in culture [21, AM Gown, personal communication]. However, we did not immunostain cementum tissues to examine whether HHF-35 positive cells are present. The cells manifested flattened and stratified morphology with networklike structures typical of human bone cells [30]. They produced type I and V collagens similar to fibroblasts, however their other properties indicate that they are not fibroblasts. For example, unlike the latter they are alkaline-phosphatase positive, they synthesize BSP and mineralize in vitro, and these properties together are strongly suggestive of the osteoblast-like nature of the cementum cells. Even though cultures of periodontal ligament fibroblasts also contain alkaline-phosphatase-positive cells [31,32], these cells, unlike the tumor cells, are typically fibroblastic in morphology and other properties. While data strongly indicate that the cementum cells manifest osteoblast-like features, we did not characterize them further as osteoblasts because our primary objective was to determine whether they express cementum phenotype. Interestingly, during early passages cultures of these cells manifested fibroblastic morphology; whether the change in morphology
27 is due to a transformation of cellular phenotype during passage or due to selective growth of osteoblast-like cells is not clear. More importantly, the cultured cementum cells produce two cementum proteins, CGF and CAP. The CGF is a 23-kDa protein which is present in cementurn and it is not detectable in gingiva or periodontal ligament [8,13]. The CGF produced by the tumor cells was characterized by chromatographic properties and by cross-reactivity with a monoclonal antibody to bovine CGF [13, Yonemura et al., manuscript submitted]. In addition to the CGF, the cementum cells also produce PDGF, FGF and other growth factors, and their proportion resembles alveolar bone rather than cementum [13,28]. For example, PDGF is not detectable and Ps are minor comnents in alveolar F and FGFs are major co ponents in cementurn, while PD and other bones [13,28, Yonemura et al., manuscript submitted], duction of significant quantities of CGF indicates that the cultured display both bone and cementum phenotypes. Several possibilities can give rise to these results. For example, these cells were derived from a tumor and tumor cells are known to produce several growth factors. The culture may contain a mixture of cells representing cementum as well as other phenotypes. Alternatively, these cells may manifest features of cementoblast progenitor cells which are reside in the alveolar bone. Nevertheless, synthesis and secretion of CG protein which appears to be localized in cementum but not in other tissues [ 15, Arzate et al., manuscript in preparation], indicates that we have cultured putative cementum cells capable of synthesizing cementum proteins. Western blots revealed several protein bands cross-reacting with the anti-CAP antibody. The 56-kDa and 43-kDa components were the major species, while additional bands with 39-kDa and 26-kDa were also present, although barely visible. Based on cross-reactivity with antibodies, we have previously observed that these proteins do not cross-react with antibadies to rat osteopontin or human [ 12,151. Immunoaffinity-purified protein obtained from human cementum extracts were also negative in Westerns using polyclonal anti-porcine osteopontin which crossreacts with human species (J Sodek, personal communication). Pulse-chase studies indicate that these proteins are possibly derived from a common precursor of 43kDa, presumably by one or more posttranslational processing mechanisms such as glycosylation, sulfation, phosphorylation or proteolysis. Although osteo also synthesized in 44 kDa, 55 kDa and possibly other sizes [33,34], the 43tein produced by cementum tumor cells does not appear to be osteopontin because thrombin does not generate 26-kDa peptides from 43- and 56.kDa proteins (data not shown), and while osteopontin is present in most calcified structures, CAP is restricted to cementum [Arzate et al., manu ipt submitted]. Interestingly, 43- and 56-kDa bands are major species in estern blots, additional band by fluorography were either minor species or not detectable; the reasons for this are and osteopontin are different pronot known. While these data indicate that C AP may be a variant of osteopontin teins, they do not rule out the possibili formed by differential splicing of its m A and/or differences in posttranslational processing such as glycosylation, phosphorylation and sulphation. Irrespective of their origin and phenotype, we believe that the cultured cemen-
28 turn tumor cells will be useful for at least three purposes. First, they could be used to obtain sufficient quantities of biologically active cementum components as this has been difficult so far due to the paucity of cementum in human teeth and laborious dissection procedures necessary. Second, the cells could be used to study the biosynthesis and processing of CAP and other cementum proteins. Third, early cementogenesis is believed to involve deposition of extracelhtlar matrix produced by Hertwig’sepithelial root sheath (HERS) cells on dentin, disruption of the HERS, migration and attachment of ectomesenchymal cells from dental follicle on to the matrix, their differentiation into cemeutoblasts and laying down of cementum [35]. Recent S cells may induce cementowork has demonstrated that gene products of genesis, and that these cells synthesize 72-kDa and 26-kDa intermediate cementum proteins and possibly osteopontin [35,36]. In adult cementurn, maturation of cementoblasts involves the directed migration of cementum progenitor cells, their adhesion to matrix and subsequent differentiation [18,29]. The cementum cells described here can serve as an in vitro model to examine these processes and to study how cementum formation and differentiation are regulated.
Acknowledgement
This work was supported by NIH grant DE-08229. We thank Drs. Larry Fisher and Allen Gown for their generous gift of antibodies. The assistance of Xidong Li in harvesting cemeutum is highly appreciated.
References 1 Birkedal-Hansen H, Butler WT, Taylor RC. Proteins of the pcriodontium. Characterization of the insoluble collagens of bovine dental cementum. Calcif Tissue Res 1977;23:39-44. 2 Christner P, Robinson P, Clark CC. A preliminary characterization of human cementum collagen. Calcif Tissue Res 1977;23:147- 150. 3 Smith AJ, Leaver AG, Smith G. The amino acid composition of the non-collagenous organic matrix of human cementum. Arch Oral Bioll983;28: 1047- 1054. 4 Glimcher MJ, Lefteriou B. Soluble glycosylated phosphoproteins of cementurn. Calcif Tissue Int 1989;45:165- 172. 5 McCurdy SP, Clarkson BH, Speirs RL, Feagin FF. Phosphoprotein extraction from the dentine/cementum complex of human tooth roots. Arch Oral Bioll990;35:347-357. 6 Bartold PM, Reinboth B, Nakae H, Narayanan AS, Page RC. Proteoglycans of bovine cementurn: isolation and characterization. Matrix 199O;lO:lo- 19. 7 Bartold PM, Miki Y, McAllister B, Narayanan AS, Page RC. Glycosaminoglycans of human cemenCum.J Periodont Hes 1988;23:13- 17. 8 Miki Y, Narayanan, AS, Page RC. Mitofwic activity of cementum components to gingival fibroblasts. J Dent Res 1987;66:1399- 1403. 9 Somerman MJ, Archer SY, ShtGyer A, Foster RA. Protein production by human gingival fibroblasts is enhanced by guanidine EDTA extracts of cementurn. J Peridont Res 1987;22:75-77. 10 Nishimura K, Hayashi M, Matsuda K, Shigeyama Y, Yamasaki A, Yamaoka AJ. The chemoattractive potency of periodontal ligament, cementum and dentin for human gingival fibroblasts. J Peridont Res 1989;24:146-148.
29 11 Somerman MJ, Foster RA, Imm GM, Sauk JJ, Archer SY. Periodontal ligament ceIIs and gingival fihblasts respond differently to attachment factors in vitro. J Periodonto] lg8g;6&73_77. 12 McAllister B, Narayanan AS, Miki Y, Page RC. Isolation of a fibroblast attachment protein from cementum, J Periodont Res 1990;25:99-105. 13 Nakae H, Narayanan AS, Raines E, Page RC. Isolation and partial characterization of mitogenic factors from cementum. Biochemistry 1991;30:7047-7052. 14 Somerman MJ, Sauk JJ, Foster RA, Dickerson K, Norris K, Argraves WS. Cell attachment activity of cementum: bone sialoprotein II identified in cementum. J Periodont Res 1991;26:10-16. 15 Olson S, Arzate H, Narayanan AS, Page RC. Cell attachment activity of cementum proteins and mechanism of endotoxin inhibition. J Dent Res 1991;10:1272-1277. 16 Melcher AH, Cheong T, Cox J, Nemeth E, Shiga A. Synthesis of cementum-like tissue in vitro by cells cultured from bone: a light and electron microscope study. J Periodont Res 1986;21:592-612. 17 McCulloch CAG, Nemeth HE, Lowenberg, B, Melcher AH. Paravascular cells in endosteal spaces of alveoidr bone contribute to periodontal ligament cell populations, Anat Ret 1987;219:233-242. 18 Melcher AH. Does the developmental origin of cementum, periodontal ligament and bone predetermine their behavior in adults? In: Guggenheim B, ed. Periodontology Today. Basel: Krager AG, 1988;6-14. 19 Fisher LW, Hawkins GR, Tuross N, Termine JD. Purification and partial characterization of small proteoglycans I and 11, bone sialoproteins I and II, and osteocalcin from the mineral compartment of developing human bone. J Biol Chem 1987;262:9702-9708. 20 Narayanan AS, Clagett JA, Page RC. Effect of inflammation on the distribution of collagen types, I, III, IV, and V and type I trimer and fibronectin in human gingivae. J Dent Res 1985;64:111l-l 116. 21 Tsukada T, Tippens D, Gordon D, Ross R, Gown AM. HHF35, a muscle-actin specific mouoclonal antibody. I. Immunocytochemical and biochemical characterization. Am J PathoI1987;126:51-60. 22 Narayanan AS, Page RC. Biochemical characterization of collagens synthesized by fibroblasts derived from normal and diseased human gingivha. J BioI Chem 1976;251:5464-5471. 23 Narayanan AS, Page RC. Synthesis of type V collagen by fibroblasts derived from normal. inflamed and hyperplastic human connective tissues. Collagen Rel Res 1985;5:297-304. 24 Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 1976;72:248-254. 25 Noff D, Pitaru S, Savion N. Basic fibroblast growth factor enhances the capacity of bone marrow cells to form bone-like nodules in vitro. FEBS Letters 1989;250:619-621. 26 Jaffe EAL, Hoyer W, Nachman RL. Synthesis of anti-hemophilic factor antigen by cultured human endothelial cells. J Clin Invest 1973;52:2757-2764. 27 Weiss RA, Guillet GYA, Freedberg IM, Farmer ER, Small EA. Weiss MM, Sun T-T. The use of monoclonal antibody to keratin in human epidermal disease: alterati. QS in immunocytochemical staining pattern. J Invest Dermatol1983;81:224-230. 28 Hauschka PV. Mavrakos AE, Iafrati MD, Doleman SE, Klagsbrun M. Growth factors in bone matrix. Isolation of multiple types by affinity chromatography on heparin-sepharose. J Biol Chem 1986;261:12665-12674. 29 Cho, M-I, Garant PR. Radioautographic study of [“Hlmannose utilization during cementoblast differentiation, formation of acellular cementum, and development of periodontal ligament principal fibers. Anat Ret 1989;223:209-222. 30 AufmkoIk B, Hauschka PV, Schwartz ER. Characterization of human bone cells in culture. Cakif Tissue Int 1985;37:228-235. 31 Nojima N, &hay&i M, Shionome M, Takahashi, N, Suda T, Hasegawa K. Fibroblastic cells derived from bovine periodontal ligaments have the phenotypes of osteoblasts. J Periodont Res 1990;25:179-185. 32 Somerman MJ, Archer SY, Imm GR, Foster RA. A comparative study of human periodontal ligament cells and gingival fibroblasts in vitro. J Dent Res 198+6’/. $6-70. 33 Nagata ‘I’, Bellows CG, Kasugai S, Butler WT. Sodek J. Biosynthesis of bone Proteins ISPP-1 (secreted phosphoprotein-1 , osteopontin), BSP (bone sialoprotein) and SPARC (osteonectin)] in association with mineralized-tissue formation by fetal-rat calvarial cells in culture. Biochem J 1991;274:513-520.
34 Kasugai S, Zhang Q, Overall CM, Wrana JL, Butler WT, Sodek, J. Differential regulation of the 55 and 44 kDa forms of secreted phosphoprotein 1 (SPP-1, osteopontin) in normal and transformed rat bone cells by osteotwpic hormones, growth factors and a tumor promoter. Bone Miner 1991;13:235-250. 35 Slavkin HC, Bringas P Jr, Bessem C, Santos V, Nakamura M, Hsu My, Snead ML, Zcichner-David M, Fin&am AG. Hertwig’s epithelial root sheath differentiation and initial cementum and bone formation during long-term organ culture of mouse mandibular first molars using sepJmless, chemically-defined medium. J Periodont Res 1988;23:28-40. 36 Somerman MJ, Shroff B, Agraves WS, Morrison G, Craig AM, Denhardt DT, Foster RA, Sauk JJ. Expression of attachment proteins during cementogenesis. J Biol Buccale 1990;18:2fi7-214.
