Immortalization of Cementoblast Progenitor Cells With Bmi-1 and TERT


  • Masahiro Saito,

    Corresponding author
    1. Department of Operative Dentistry and Endodontics, Kanagawa Dental College, Yokosuka, Japan
    2. Research Center of Advanced Technology for Craniomandibular Function, Kanagawa Dental College, Yokosuka, Japan
    3. These authors contributed equally to this work
    • Masahiro Saito, DDS, PhD, Department of Operative Dentistry and Endodontics, Kanagawa Dental College, 82 Inaoka-cho, Yokosuka-city, Kanagawa 238–8580, Japan
    Search for more papers by this author
  • Keisuke Handa,

    1. Department of Operative Dentistry and Endodontics, Kanagawa Dental College, Yokosuka, Japan
    2. These authors contributed equally to this work
    3. Virology Division, National Cancer Center Research Institute, Tokyo, Japan
    Search for more papers by this author
  • Tohru Kiyono,

    1. Virology Division, National Cancer Center Research Institute, Tokyo, Japan
    Search for more papers by this author
  • Shintaro Hattori,

    1. Department of Prostodontics, Kanagawa Dental College, Yokosuka, Japan
    Search for more papers by this author
  • Takamasa Yokoi,

    1. Department of Operative Dentistry and Endodontics, Kanagawa Dental College, Yokosuka, Japan
    2. Department of Periodontology, School of Dentistry, Aichi-Gakuin University, Nagoya, Japan
    Search for more papers by this author
  • Takanori Tsubakimoto,

    1. Department of Operative Dentistry and Endodontics, Kanagawa Dental College, Yokosuka, Japan
    Search for more papers by this author
  • Hidemitsu Harada,

    1. Osaka University Graduate School of Dentistry, Department of Oral Anatomy and Developmental Biology, Osaka, Japan
    Search for more papers by this author
  • Toshihide Noguchi,

    1. Department of Periodontology, School of Dentistry, Aichi-Gakuin University, Nagoya, Japan
    Search for more papers by this author
  • Minoru Toyoda,

    1. Department of Prostodontics, Kanagawa Dental College, Yokosuka, Japan
    Search for more papers by this author
  • Sadao Sato,

    1. Research Center of Advanced Technology for Craniomandibular Function, Kanagawa Dental College, Yokosuka, Japan
    2. Department of Orthodontics, Kanagawa Dental College, Yokosuka, Japan
    Search for more papers by this author
  • Toshio Teranaka

    1. Department of Operative Dentistry and Endodontics, Kanagawa Dental College, Yokosuka, Japan
    2. Research Center of Advanced Technology for Craniomandibular Function, Kanagawa Dental College, Yokosuka, Japan
    Search for more papers by this author

  • The authors have no conflict of interest.


A cementoblast progenitor cell line designated BCPb8 was successfully isolated from dental follicle cells immortalized with Bmi-1 and hTERT. BCPb8 showed the potential to differentiate into cementoblasts on implantation into immunodeficient mice. BCPb8 was confirmed to be the first established cementoblast progenitor cell line and will provide a useful model for investigating cementogenesis.

Introduction: The dental follicle is the mesenchymal tissue surrounding the developing tooth germ. During tooth root development, progenitor cells present in the dental follicle are believed to play a central role in the formation of periodontal components (cementum, periodontal ligament, and alveolar bone). However, little more is known about the biology of these progenitors. Previously, we observed that cultured bovine dental follicle cells (BDFCs) contained putative cementoblast progenitors. To further analyze the biology of these cells, we attempted to isolate cementoblast progenitors from immortalized BDFC through expression of the polycomb group protein, Bmi-1, and human telomerase reverse transcriptase (hTERT).

Materials and Methods: BDFCs were transduced with replication-deficient retroviruses carrying human Bmi-1(LXSN-Bmi-1), and hTERT (LXSH-hTERT) for immortalization. Single cell clones were established from immortalized BDFC, and differentiation into cementoblasts was assessed by implantation into immunodeficient mice.

Results and Conclusion: BDFCs expressing Bmi-1 and hTERT showed an extended life span - 90 population doublings more than normal BDFCs - and still contained cells with the potential to differentiate into cementoblasts on implantation into immunodeficient mice. From these cells, we established a clonal cell line, designated BCPb8, which formed cementum-like tissue that was reactive to the anti-cementum-specific monoclonal antibody 3G9 and expressed mRNA for bone sialoprotein, osteocalcin, osteopontin, and type I collagen on implantation. Thus, by using Bmi-1 and hTERT, we succeeded in immortalizing cementoblast progenitor cells from BDFC without affecting differentiation potential. The BCPb8 cell line is the first immortalized clonal cell line of cementoblast progenitors and could be a useful tool not only to study cementogenesis but also to develop regeneration therapy for patients with periodontitis.


THE DENTAL FOLLICLE is formed at the cap stage of tooth germ development by an ectomesenchymal progenitor cell population originating from cranial neural crest cells.(1,2) The critical role that the progenitor cell population seems to play in the development and regeneration of the periodontium has resulted in considerable interest in the biology of these cells. Progenitors present in the dental follicle are thought to contribute to the formation of periodontal tissues, including cementum, periodontal ligaments, and osteoblasts.(3) During development of the cementum, differentiation of cementoblast progenitors within the dental follicle seems to be regulated by epithelial-mesenchymal interactions that generate specific signals for cellular differentiation.(4-6) After the formation of root dentin in the tooth germ, epithelial matrices are deposited on the newly formed root dentin surface by overlying the cells of the enamel epithelium cells, also known as Hertwig's epithelial root sheath (HRS). After fragmentation of the HRS, cementoblast progenitors seem to migrate and attach to the matrix coating the root surface, differentiate into cementoblasts, and subsequently form cementum.(3) Recent reports have revealed that the dental follicle contains cementoblast progenitors.(7,8) Moreover, Zhao et al. found that bone morphogenic protein-2 promotes differentiation of the dental follicle cells into the cementoblast/osteoblast lineage,(9) and we have previously shown that bovine dental follicle cells (BDFCs) isolated from developing tooth germ are able to form cementum matrix when transplanted into immunodeficient mice.(10) In a previous study, the cementum matrix formed by BDFCs was heavily stained after addition of anti-cementum-specific monoclonal antibody (mAb 3G9). This antibody was raised against partially purified bovine cementum matrix containing cementum-derived attachment protein and specifically recognizes cementum matrix and cementoblasts but not tissues such as bone matrix.(11) Although these results support the presence of cementoblast progenitors in the dental follicle, details of the progenitor biology and their differentiation potential remain largely unknown.

Normal somatic cells are not capable of indefinite expansion because their life span is limited by cellular senescence.(12) This limitation has hampered progress in isolation and expansion of tissue-specific progenitor or stem cells from each tissue. The limited life span of normal cells seems to be controlled, at least in part, by telomere shortening, which occurs on each cell division in cells lacking telomerase activity.(13) However, telomere-independent mechanism(s) involving the expression of cyclin-dependent kinase inhibitors, such as p16Ink4a and p21WAF1 seem to induce premature senescence in human epithelial cells.(14) Indeed, inactivation of the p16Ink4a/retinoblastoma protein (Rb) pathway in addition to the expression of TERT is required for preventing cellular senescence of some cell types in culture.(15) The polycomb group gene bmi-1 represses the transcription of p16Ink4a and p19Arf, and overexpression of Bmi-1 can immortalize mouse embryonic fibroblasts but not human fibroblasts.(16,17) Therefore, we hypothesized that Bmi-1 and TERT might immortalize BDFCs without affecting their differentiation potential and help to successfully establish a cementoblast progenitor cell line from the immortalized cells.


Tissue culture

Isolation of BDFCs was described previously.(10) Briefly, bovine dental follicle tissue was removed from the root dentin of developing tooth germs with a scalpel and digested with 3 mg/ml bacterial collagenase (Boehringer Mannheim, Mannheim, Germany) in Krebs buffer (111.2 mM NaCl, 21.3 mM Tris, 13.0 mM glucose, 5.4 mM KCl, 1.3 mM MgCl2 and 1.5 mM ZnCl2). The released cells were incubated with α-MEM (Life Technologies) containing 10% FBS (BioWhittaker, Walkersville, MD, USA), 50 μg/ml of ascorbic acid, and 100 units/ml of streptomycin and penicillin in a humidified atmosphere of 5% CO2 at 37°C. When the cells reached ∼80% confluence, they were passaged with 0.25% trypsin/1 mM EDTA and maintained as BDFCs. These cells were plated into six wells at a density of 3 × 104 cells/ml, and the medium was changed every 3 days.

Infection of retrovirus constructs and establishment of cell line

The full-length human bmi-1 cDNA was cloned by RT-PCR using RNA extracted from K562 cells. Thermoscript RT (Invitrogen) and KOD polymerase (TOYOBO) were used for the RT and PCR reactions, respectively. The forward primer, 5′-ACGCGTCGACCGCCATGCATCGAACAACGAGAAT-3′, and reverse primer, 5′-CGGATCCTCAACCAGAAGAAGTTGCTG-3′, were designed to obtain the coding sequence of human bmi-1 flanked by a SalI site (underlined), a Kozak consensus sequence at the 5′-end, and a BamHI site (underlined) at the 3′-end. The SalI-BamHI segment of the PCR product was cloned between the XhoI and BglII sites of pCLXSN to generate pCLXSN-Bmi-1. The coding sequence of the cDNA was confirmed as identical to the published sequence (NCBI ACC# NM_005180.4). Production of LXSN-Bmi-1 and LXSH-hTERT retroviruses was performed as described previously.(18) One milliliter of producer cell culture fluid was added to BDFCs (passage 3) in the presence of polybrene (8 μg/ml). For combination retroviral infection, cells were sequentially transduced with LXSN-Bmi-1 and then with LXSH-hTERT, and subsequently selected in the presence of G418 (100 μg/ml) and hygromysin B (50 μg/ml). Stably transduced cells were maintained in the medium described above. For single cell cloning, BDFCs with an extended life span were seeded in a 96-well plate (Falcon, Bedford, MA, USA) at a density of 0.3 cell/well. Single cell-derived clones were established after 10 days of culture and further analyzed. For comparison, BDFCs were sequentially transduced with LXSN-E6E7 and with LXSH-hTERT using the same procedure as with LXSN-Bmi-1 and LXSH-hTERT, and the cells were used as controls for immortalization.

In vivo differentiation assay

The potential of individual BDFC clones to differentiate into cementoblasts on transplantation into immunodeficient mice was assessed as described previously.(10) Briefly, cell clones were inoculated subcutaneously into 5-week-old female CB-17 scid/scid (SCID; severe combined immunodeficiency) mice (Nihoncrea, Tokyo, Japan) after incubating 1.5 × 106 cells in a mixture of 40 mg of hydroxyapatite powder (Apaseram, Pentax, Tokyo, Japan) and fibrin clot (mixture of mouse fibrinogen and thrombin: both from Sigma, St Louis, MO, USA). Mice were killed after 4 weeks, and implants were cut in half for histochemical analysis and analysis of mRNA expression of cementum matrix components.

Osteogenic differentiation

Cells were plated into six wells at a density of 3 × 104 cells/ml and cultured in the medium described above supplemented with 100 nM dexamethasone, 50 μg/ml of ascorbic acid, and 10 mM β-glycerophosphate. The culture medium was replaced every 3 days, and the cells were maintained for 3 weeks.

Histochemical analysis

The transplants were fixed in 4% paraformaldehyde for 1 day, decalcified with 10% formic acid for 3 days, and embedded in paraffin. To avoid nonspecific staining of mouse monoclonal antibodies, sections were blocked using the M.O.M kit (Vector Burlingame) as described previously.(10) The sections were incubated with mAb 3G9 (5 μg/ml) for 1 h and treated with biotinylated secondary antibody and avidin-peroxidase conjugate. Dilutions were made with PBS containing 1 mg/ml bovine serum albumin. Sections were incubated with nonimmune mouse IgG or anti-vimentin monoclonal antibody (V9, Dako, Carpinteria, CA, USA), which served as a control. The color reaction was developed with diaminobenzidine.

RNA preparation and RT-PCR analysis

Total RNA was isolated from cells and transplants using ISOGEN (Nippon Gene, Tokyo, Japan) as described previously.(10) cDNA was synthesized from 1 μg of total RNA in a 20 μl reaction containing 10× reaction buffer, 1 mM dNTP mixture, 1 U/μl RNase inhibitor, 0.25 U/μl RT (Moloney murine leukemia virus [M-MLV] RT, Invitrogen), and 0.125 μM random 9-mers (Takara, Tokyo, Japan). Amplification was performed in a PCR Thermal Cycler SP (Takara) for 25 cycles according to the following reaction profile: 94°C for 1 minutes, 60°C for 30 s, and 72°C for 30 s. Synthesized cDNA served as a template for subsequent PCR amplification using specific primers for bone sialoprotein, osteocalcin, osteopontin, type I collagen, and GAPDH as previously described.(10)

SDS-PAGE and immunoblotting

Cells were lysed in a buffer containing 50 mM Tris-HCl (pH 7.4), 125 mM NaCl, 0.1% (v/v) Nonident P-40 (NP-40; Sigma), and 1 mM each of EDTA and phenylmethylsulfonyl fluoride, followed by disruption by ultrasonication. Cell debris was removed by centrifugation at 14,000g for 15 minutes, and the protein concentration was determined using the Bradford protein assay (Bio-Rad Laboratories, Richmond, CA, USA). Twenty micrograms of protein was separated by 7% or 12% SDS-PAGE under reducing conditions, followed by electrophoretic transfer to a polyvinylidene difluoride (PVDF) membrane (Amersham Pharmacia Biotech, Buckinghamshire, UK). Membranes were blocked with 5% nonfat dry milk in PBS containing 0.1% Tween20 for 1 h. They were then probed with primary antibody with horseradish peroxidase-linked secondary antibody (Cell Signaling Technology, Beverly, MA, USA) and subsequently visualized by chemiluminescence (Lumi-LightPLUS, Alameda, CA, USA) as described elsewhere.(15) AntiBmi-1 monoclonal antibody (5G4) raised against the C-terminal 99 amino acids of human Bmi-1, anti-Rb monoclonal antibody (G3-245; BD Pharmingen, San Diego, CA, USA), anti-p16Ink4a monoclonal antibody (F-12; Santa Cruz Biotechnology, Santa Cruz, CA, USA), or anti-β-actin monoclonal antibody (AC-15; Sigma) was used as the primary antibody.

Telomerase repeat amplification protocol assay

Telomerase activity was determined using the telomerase repeat amplification protocol assay performed with the TRAPeze Telomerase Detection Kit (Intergen), according to the manufacturer's instructions. Cells were grown to 70-80% confluence, and lysed with CHAPS buffer. A lysate volume equal to 1000 cells was used for each reaction. Electrophoresis was performed on a 12% nondenaturing acrylamide gel, stained with SYBR green (SYBR Green; Molecular Probes, Eugene, OR, USA), and bands were visualized using a chemiluminescence image analyzer (LumiVision Pro; AISIN TAITEC).

Senescence-associated β-galactosidase assay

Cells were washed in PBS and fixed with 3% paraformaldehyde for 3-5 min. Senescence-associated β-galactosidase (SA-β-gal) staining was performed as described elsewhere.(19)


Immortalization of BDFC

We attempted to extend the life span of BDFCs, first by using a retrovirus expressing bmi-1 and then with a retrovirus expressing hTERT.

After retroviral transfer, telomerase activity was examined using the telomerase repeat amplification protocol assay. No telomerase activity was detected in either normal BDFCs or BDFCs transduced with bmi-1 alone. As expected, hTERT-transduced cells showed strong telomerase activity Fig. 1), which was comparable to that of Hela cells (data not shown), indicating that hTERT was competent in bovine cells. However, we failed to extend the life span of BDFCs by introduction of hTERT alone, and thus we further analyzed BDFC expressing both bmi-1 and hTERT (BDFCsbmi-1+hTERT). BDFCsbmi-1+hTERT bypassed senescence around population doubling (PD)30 and grew to over PD200 without significant growth retardation.

Figure FIG. 1.

Telomerase activity of immortalized BDFCs. BDFCs were infected with retroviruses, and telomerase activity was measured using telomerase repeat amplification protocol assay. Cell lysates of BDFC at an early stage (PD9) were used as controls.20

Expression of Bmi-1 was confirmed by immunoblot analysis (Fig. 2A). Strong expression of Bmi-1 was detected in BDFCsbmi-1+hTERT (Fig. 2A), although lower levels of possibly endogenous bovine Bmi-1 were detected during early passages (PD15) of BDFCs, and much lower levels were observed in late passages (PD30) of BDFCs (Fig. 2A). Normal BDFCs were able to propagate until PD30, at which point they practically discontinued division, exhibited the flattened morphology characteristic of senescence, and showed strong SA-β-Gal, a biomarker associated with cellular aging (Fig. 2Bb), but these characteristics were not observed at PD15 (Fig. 2Ba). In contrast, BDFCsbmi-1+hTERT did not show SA-β-Gal activity and retained their original morphology at PD30, and cell proliferation continued, even when the cells were cultured beyond PD200 (Fig. 2Bc).

Figure FIG. 2.

Inactivation of the p16Ink4a-Rb pathway and SA-β-gal activity in BDFC immortalized with Bmi-1. Cell lysates were prepared from normal BDFCs at an early stage (PD15), presenescence stage (PD30), or from those tranduced with indicated retrovirus. (A) Western blot analysis for Bmi-1, Rb, p16ink4a, and β-actin are shown. Arrows indicate the unphosphorylated (bottom) and phosphorylated (top) forms of Rb. (B) β-Gal staining in (a) PD15 BDFC, (b) PD30 BDFC, or (c) BDFCbmi-1+hTERT are shown. Arrows indicate cells positive for SA-β-gal activity. Original magnification, ×40.20

Bmi-1 has been shown to extend the life span of human fibroblasts because of the downregulation of p16Ink4a(16,20) Downregulation of p16Ink4a suppresses Rb function by hyperphosphorylation, leading to S phase commitment at the G1 phase of the cell cycle.(21) We therefore examined the expression of p16Ink4a and the phosphorylation status of Rb by immunoblotting. p16Ink4a was expressed at high levels in PD30 BDFCs compared with BDFCsbmi-1+hTERT or PD15 BDFCs. Although large amounts of hyperphosphorylated Rb were observed in PD15 BDFCs, a senescence-associated decline in hyperphosphorylated Rb was apparent in PD30 (Fig. 2A). Furthermore, hyperphosphorylated Rb was observed in BDFCsbmi-1+hTERT with an expanded life span (Fig. 2A). In the control experiment, BDFCs transduced with E6E7 and hTERT (BDFCsE6E7+hTERT) also overcame replicative senescence; they did not exhibit SA-β-Gal activity and maintained their original morphology and cell proliferation activity, even when the cells were cultured beyond PD200 (data not shown).

Differentiation potential of immortalized BDFCs

To investigate the differentiation potential of immortalized BDFCs, the cells were implanted into immunodeficient mice as described previously.(10) Four weeks after implantation, normal BDFCs generated cementum-like tissue on the border of hydroxyapatite beads (Fig. 3A, arrows), and it resembled BDFCbmi-1+hTERT transplants (Fig. 3B, arrows). RT-PCR analysis was performed using bovine-specific primers for cementum components such as bone sialoprotein (BSP), osteocalcin (OC), osteopontin (OP), and type I collagen (ColI).(22,23) BDFCs expressed OP, ColI, and OC mRNA, but not BSP mRNA (Fig. 4A). The expression pattern of BDFCsbmi-1+hTERT was similar to BDFC except that there was no OC expression (Fig. 4A). After the in vivo differentiation assay, BDFCs expressed bovine-specific mRNA for BSP, OC, OP, and ColI (Fig. 4B). BDFCbmi-1+hTERT transplants also expressed the mRNA, whereas no mRNA expression was observed in transplants without bovine cells (data not shown). In the case of BDFCsE6E7+hTERT, although they exhibited an undifferentiated phenotype expressing OP and COLI Fig. 4A) in vitro, neither cementum formation nor expression of BSP or OC mRNA was observed in vivo (Figs. 3C and 4B).

Figure FIG. 3.

Differentiation potential of immortalized BDFCs. (B and C) BDFCs immortalized with retroviruses were transplanted into SCID mice for 4 weeks. (A) Untransfected BDFCs were used as positive controls for the transplantation assay. Transplants were analyzed by hematoxylin and eosin staining. Arrows indicate cementum-like tissue deposited in the transplants. Bar = 50 μm.20

Figure FIG. 4.

RT-PCR analysis for BSP, OC, OP, ColI, and GAPDH in vitro and in vivo. mRNA expression of BSP, OC, OP, and ColI was detected by RT-PCR analysis. Untransfected BDFCs were used as controls. (A) Total RNA isolated from BDFCs (left), BDFCsbmi-1+hTERT (middle), or BDFCsE6E7+hTERT (right). Transcripts for OC, OP, ColI, and GAPDH were prominent in all three cell types, but no OC expression was observed in BDFCsbmi-1+hTERT or BDFCsE6E7+hTERT (arrow). (B) Total RNA isolated from transplants of BDFCs (left), BDFCsbmi-1+hTERT (middle), or BDFCsE6E7+hTERT (right). Intense expression of BSP and OC mRNA was induced in transplants of BDFCs and BDFCsbmi-1+hTERT (arrowheads). However, no changes were observed in BDFCE6E7+hTERT transplants.20

Isolation and characterization of bovine cementoblast progenitors

To isolate cementoblast progenitors, BDFCsbmi-1+hTERT were seeded at a density of 0.3 cells/well in 96-well plates. After 10 days, 99 randomly isolated clones were expanded, and the differentiation potential of each was examined using the in vivo differentiation assay. Of 99 clones, only 3 were able to form cementum-like tissue, and 1 was designated BCPb8. As seen in normal BDFCs and BDFCsbmi-1+hTERT, BCPb8 transplants formed cementum-like tissue that stained strongly positive with anti-cementum-specific mAb 3G9 (Figs. 5Aa5Ac). Furthermore, mAb 3G9+ cementoblasts within cementum-like tissue and surrounding fibrous tissue were stained withanti-vimentin antibody, which recognized bovine but not mouse cells (Figs. 5Ae5Ag). In control sites without bovine cells, no cementum-like tissue was observed, and no regions reacted with mAb 3G9 or anti-vimentin antibody (Figs. 5Ad and 5Ah). Notably, cells attached to hydroxyapatite were able to form cementum-like tissue in BCPb8 transplants, whereas cells that were not attached to hydroxyapatite formed fibrous tissue. Azan staining revealed a characteristic structure in BCPb8 transplants (Fig. 5B). Fiber-like structures were observed at the interface between cementum-like tissue and fibrous tissue in BCPb8 and BDFC transplants (Figs. 5Ba and 5Bb), but not around the bone-like tissue formed by bovine osteoblast transplants (Fig. 5Bc). The fiber-like structures were reminiscent of Sharpey's fibers present at the interface between the periodontal ligament and cementum on the tooth root surface. RT-PCR analysis showed that BCPb8 expressed OP and ColI in vitro, but also expressed BSP and OC in the transplants, similar to the results with normal BDFCs (Fig. 6A). To investigate whether BCPb8s are able to differentiate cementoblasts in vitro, they were treated with osteogenic differentiation medium for 3 weeks.(24) RT-PCR showed that intense expression of OC but not BSP was induced in BCPb8s (Fig. 6B). In addition, neither mineralized nodules nor mAb 3G9+ cementum matrices were formed in this culture condition (data not shown).

Figure FIG. 5.

Differentiation potential of BCPb8 in vivo. (A) Immunohistochemical staining with (a-d) 3G9 or (e-h) antivimentin monoclonal antibody in the (a and e) BDFC transplants, (b and f) BDFCsbmi-1+hTERT, or (c and g) BCPb8s are shown. (d and h) Transplants without cells (HAP) are also shown. (a-c) Specific staining with mAb 3G9 was observed in the cementum-like tissue (arrow) formed by all of the transplants. (e-f) Staining with antivimentin monoclonal antibody is evident in cells inside the cementum-like tissue and surrounding matrix, respectively. (d and h) However, HAP transplants did not exhibit reactivity with all of the antibodies. (B) Azan staining of transplants of (a) BCPb8s, (b) BDFCs, and (c) bovine osteoblasts are shown. Arrows indicate fiber-like structures formed in the transplants of BDFCs and BCPb8s. Bar = 50 μm.20

Figure FIG. 6.

Differentiation potential of BCPb8 in vitro and in vivo. mRNA expression of BSP, OC, OP, and ColI was detected by RT-PCR analysis. (A) Total RNA isolated from BCPb8 (left lane) or BCPb8 transplants (right lane). Transcription of BSP and OC was induced in BCPb8 transplants (arrows). (B) Total RNA isolated from BCPb8 incubated with osteogenic differentiation medium for the indicated period. After 3 weeks of incubation, expression of OC was induced by osteogenic differentiation medium (arrowhead). However, no expression of BSP was observed.20


The dental follicle has been suggested as a source of cementoblast progenitors. Nevertheless, the origin and nature of the progenitors have yet to be characterized. This study outlined a method for immortalization, isolation, and characterization of cementoblast progenitors. We showed the existence of cementoblast progenitors in cultured dental follicle cells (BDFCs) by isolating cementoblast progenitors from immortalized cells (BDFCsbmi-1+hTERT). The proportion of cementoblast progenitors present among BDFCsbmi-1+hTERT was about 3% of primary BDFCs. The immortalized cell line, designated BCPb8, had the capacity to differentiate into cementoblasts on transplantation when mixed with hydroxyapatite, because they formed cementum-like tissue that was positive for the cementum-specific antibody mAb 3G9, and expressed cementum matrix components including BSP, OC, OP, and ColI. Histologically, cementum-like tissue was observed only around the hydroxyapatite beads, suggesting that the differentiation of cementoblast progenitors required contact with hydroxyapatite, which presumably mimicked root dentin.

Identification of cementoblast progenitor cells required amplification of a single cell lineage of BDFCs. However, BDFCs proliferate in culture for a finite number of PD, which varies with cell type, referred to as cellular senescence. Therefore, attempts were made to immortalize BDFCs to prevent cellular senescence during isolation and amplification of cementoblast progenitors. Cellular senescence of some cell types, such as human foreskin fibroblasts (BJ fibroblasts), seems to be induced only by telomere shortening, because overexpression of hTERT alone can allow cells to become immortalized.(25) However, other cell types, such as human mammary epithelial cells, have a telomere-independent senescence pathway and require inactivation of p16Ink4a/Rb pathway in addition to activation of telomerase for immortalization.(26-28) Although SV40 LT or HPV16 E6E7 can extend the life span of most cell types by inactivating both p16Ink4a/Rb and telomere/p53 pathways, cells without induced telomerase activity eventually enter “crisis” because of the progressive loss of the telomere.(29) Previous studies have been reported that E6 and bmi-1 activate telomerase activity in some cell types.(30,31) In this study, however, BDFCs expressing Bmi-1 did not exhibit telomerase activity. BDFCs transduced with hTERT alone did not exhibit an extended life span, whereas cells expressing hTERT and bmi-1 overcame replicative senescence as shown by SA-β-Gal activity in Fig. 2B. These results suggest that both inactivation of the p16Ink4a/Rb pathway and activation of telomerase are required for immortalization of BDFCs. In this study, we showed that bmi-1 extended the life span of BDFCs by inactivation of p16Ink4a/Rb pathway as evidenced by suppression of p16Ink4a and the presence of hyperphosphorylated Rb. Although little is known about the immortalization of bovine cells, these results parallel those observed in bmi-1-transduced human and mouse cells.

Although immortalization of BDFCsbmi-1+hTERT was achieved by inactivation of the p16Ink4a/Rb pathway and telomerase activity, it is possible that other cellular functions, such as cell differentiation, may have been affected. However, the results of clonal analysis indicated that BCPb8 had in vivo differentiation potential. Results of immunohistochemical analysis of cementum-like tissue by using mAb 3G9 and the expression pattern of mRNA for cementum matrix components in the transplants were similar to those in normal BDFC. Bmi-1 was originally identified as a c-myc cooperating oncogene involved in lymphogenesis.(32) However, it also belongs to the polycomb group genes that are crucial for maintaining proper gene expression pattern during development.(33) Moreover, Bmi-1 has been found to regulate the proliferation activity of normal stem cells and progenitors.(17,34) Our data confirmed that no tumor formation occurred in BDFCbmi-1+hTERT transplants, thus suggesting that Bmi-1 has minimal effect on the differentiation potential of BDFCs. In contrast, BDFCsE6E7+hTERT did not differentiate in vivo. The loss of differentiation activity in this case is likely to be because of the influence of the viral oncoprotein E6E7,(35,36) although E6E7 has been shown to immortalize some cell types without loss of differentiation potential.(18,37)

We have previously shown that cementoblast progenitors are present in BDFCs and that they are phenotypically different from osteoblasts.(10) In this study, we successfully isolated a cementoblast cell line, BCPb8, which seems to behave similarly to the cementum progenitors present in BDFCs, both in vitro and in vivo. BCPB8s exhibited low alkaline phosphatase activity in vitro (data not shown), and the activity was similar to that of BDFCs.(10) BCPb8s seem to maintain the undifferentiated dental follicle phenotype in vitro as indicated by the lack of mAb 3G9 binding and BSP expression. However, they are capable of differentiation into cementoblasts on hydroxyapatite beads in vivo. Furthermore, BCPb8 transplants created Sharpey fiber-like tissue at the interface of cementum-like tissue as shown by azan staining in Fig. 5B. No such morphology was observed in bovine osteoblast transplants. These findings are consistent with those of a previous report showing that the organization of collagen bundles in human cementoblast transplants is apparently different from that of human bone marrow stromal cells transplants.(38) The cementum-like tissue formed by BCPb8 transplants was strongly positive for mAb 3G9 staining, whereas bone-like tissues formed by bovine osteoblast transplants was negative for mAb 3G9 staining (data not shown). These data indicate that the BCPb8 belong to cementoblast progenitors derived from dental follicle cells and are phenotypically different from osteoblasts.

We have previously proposed that attachment to extracellular matrix on the surface of hydroxyapatite might be critical for cementoblast differentiation and that those cells that fail to attach to the matrix become committed periodontal ligament cells or remain undifferentiated.(10) Staining of BCPb8 transplants using bovine-specific antivimentin monoclonal antibody showed that only cells that attached to hydroxyapatite differentiated into cementoblasts. However, co-culture of BCPB8 and hydroxyapatite did not promote differentiation of cementoblasts in vitro (data not shown). These data strongly support our hypothesis and suggest that extracellular matrix localized on hydroxyapatite is critical for cementoblast differentiation in vivo. Alternatively, cementum-like structures present in BCPb8 transplants were smaller than those of normal BDFCs, suggesting that cell-cell interaction or paracrine factors also influence cementoblast differentiation. During root development, extracellular matrix that contains a cell binding motif, such as OP, BSP, or ameloblastin, seems to play an important role in the differentiation of cementoblasts from cementoblast progenitors.(3,39) Therefore, specific integrin-matrix interactions may regulate recruitment and differentiation of cementoblast progenitors.(40-42) As a result, a microenvironment such as that found in the extracellular matrix on the root dentin surface and anchorage-dependent signals may promote differentiation of cementoblast progenitors into cementoblasts.(41)

Progress in understanding of the biology of cementoblast progenitors has been slow because of challenges of studying dental follicle cells. BDFCs immortalized by a combination of Bmi-1 and hTERT permitted us to identify cementoblast progenitors. BCPb8 cells could facilitate better understanding of the mechanisms of cementogenesis, including those pertaining to cementoblast differentiation. They also provide a powerful tool for developing therapeutic strategies, such as cell therapy, and treatment of periodontitis using manipulated cementoblast progenitors.


We thank Drs Sampath Narayanan, Masato Yamauchi, and Ken-ichi Kozaki for advice and discussions during the course of this work. We also thank Dr Hayato Ohshima for advice regarding histological work. This work was primarily performed at the Research Center of Advanced Technology for Craniomandibular Function at Kanagawa Dental College and was supported in part by Grants-in-Aid for Bioventure Research, the AUG High-Tech Research Center Project, the 2003-Multidisciplinary Research Project from MEXT, and grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan.