A Role for N-Cadherin in the Development of the Differentiated Osteoblastic Phenotype


  • Serge L. Ferrari,

    1. Division of Bone and Mineral Metabolism, Charles A. Dana and Thorndike Laboratories, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Kathy Traianedes,

    1. Division of Endocrinology/Medicine, University of Texas Health Sciences Center, San Antonio, Texas, U.S.A.
    Search for more papers by this author
  • Marielle Thorne,

    1. Division of Bone and Mineral Metabolism, Charles A. Dana and Thorndike Laboratories, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Marie-Helene Lafage-Proust,

    1. Faculté de Médecine, Laboratorie de Biologie des Tissus Osseux, St-Etienne, France
    Search for more papers by this author
  • Paul Genever,

    1. Bone and Joint Research, Department of Biology, University of York, York, United Kingdom
    Search for more papers by this author
  • Marco G. Cecchini,

    1. Gene Therapy Laboratory, University of Berne, Berne, Switzerland
    Search for more papers by this author
  • Vered Behar,

    1. Division of Bone and Mineral Metabolism, Charles A. Dana and Thorndike Laboratories, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Alessandro Bisello,

    1. Division of Bone and Mineral Metabolism, Charles A. Dana and Thorndike Laboratories, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Michael Chorev,

    1. Division of Bone and Mineral Metabolism, Charles A. Dana and Thorndike Laboratories, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Michael Rosenblatt,

    1. Division of Bone and Mineral Metabolism, Charles A. Dana and Thorndike Laboratories, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, U.S.A.
    Search for more papers by this author
  • Larry J. Suva

    Corresponding author
    1. Department of Bone and Cartilage Biology, SmithKline Beecham Pharmaceuticals, King of Prussia, Pennsylvania, U.S.A.
    • L. J. Suva Bone and Cartilage Biology SmithKline Beecham Pharmaceuticals 709 Swedeland Road King of Prussia, PA 19406, U.S.A.
    Search for more papers by this author

  • This work was partly presented in abstract form at the American Society for Bone and Mineral Research–International Bone and Mineral Society, Second Joint Meeting, San Francisco, California, U.S.A., December 1998.


Cadherins are a family of cell surface adhesion molecules that play an important role in tissue differentiation. A limited repertoire of cadherins has been identified in osteoblasts, and the role of these molecules in osteoblast function remains to be elucidated. We recently cloned an osteoblast-derived N-cadherin gene from a rat osteoblast complementary DNA library. After in situ hybridization of rat bone and immunohistochemistry of human osteophytes, N-cadherin expression was localized prominently in well-differentiated (lining) osteoblasts. Northern blot hybridization in primary cultures of fetal rat calvaria and in human SaOS-2 and rat ROS osteoblast-like cells showed a relationship between N-cadherin messenger RNA expression and cell-to-cell adhesion, morphological differentiation, and alkaline phosphatase and osteocalcin gene expression. Treatment with a synthetic peptide containing the His-Ala-Val (HAV) adhesion motif of N-cadherin significantly decreased bone nodule formation in primary cultures of fetal rat calvaria and inhibited cell-to-cell contact in rat osteoblastic TRAB-11 cells. HAV peptide also regulated the expression of specific genes such as alkaline phosphatase and the immediate early gene zif268 in SaOS-2 cells. Transient transfection of SaOS-2 cells with a dominant-negative N-cadherin mutant (NCADΔC) significantly inhibited their morphological differentiation. In addition, aggregation of NCTC cells derived from mouse connective tissue stably transfected with osteoblast-derived N-cadherin was inhibited by either treatment with HAV or transfection with NCADΔC. Together, these results strongly support a role for N-cadherin, in concert with other previously identified osteoblast cadherins, in the late stages of osteoblast differentiation. (J Bone Miner Res 2000;15:198–208)


Osteoblasts are highly differentiated skeletal cells responsible for bone formation. Their functions include regulated synthesis and mineralization of the bone matrix as well as bone remodeling through the activation of osteoclasts.(1) Osteoblasts are derived from pluripotent mesenchymal stem cells in the bone marrow, which are morphologically similar to fibroblasts but differentiate through prominent changes of cell morphology and tissue organization. The differentiation pathway is further characterized by sequential expression of specific genes that code for collagenous and noncollagenous bone proteins, cytokines, cytokine receptors, and hormone receptors.(2,3) Ultimately, bone-forming osteoblasts appear as cuboidal cells lining the bone surface as an epithelium-like monolayer cell sheet. Cell-to-cell adhesion, therefore, is a hallmark of late osteoblastic differentiation. Gap junctions have been observed between adjacent osteoblasts, and the expression of gap junction-associated molecules, such as connexin 43, is regulated by osteotropic factors such as parathyroid hormone (PTH).(4–6) In turn, osteoblasts form networks of functionally coupled cells in vitro.(5–8) Interestingly, among the variety of osteoblast-like cell cultures available, notable differences in cell coupling have been found: Rat osteosarcoma (ROS) cells are coupled most efficiently, possibly reflecting a more advanced stage of differentiation.(5–8)

Neural (N-), epithelial (E-), and placental (P-) cadherins play a pivotal role in Ca2+-dependent cell-to-cell adhesion.(9) A striking feature of the type I (classic) cadherins is a highly conserved His-Ala-Val (HAV) motif in the first extracellular domain that mediates homophilic interactions between cadherin molecules expressed on adjacent cells.(9) Hence, cadherins act as both ligands and receptors on the cell surface. On the other side of the cell membrane, the cadherin intracellular domain forms a protein complex with catenin molecules, which anchor the cadherins to the cytoskeleton and provide a unique signal transduction pathway for cadherin interactions.(10) Such interactions are crucial for embryonic development as well as the differentiation and maintenance of various tissues.(9)

A limited repertoire of cadherins has been identified in osteoblasts.(11) The first “osteoblast-cadherin,” cloned from a mouse osteoblastic cell line (MC3T3-E1),(12) was later recognized to be homologous to human Cad-11,(11,13) a type II cadherin abundantly expressed in mesenchymal cells.(9) Its expression in osteoblasts presumably reflects their stromal cell origin. Cadherin-4, a type I cadherin, has recently been identified in both human trabecular bone osteoblasts and bone marrow stromal cells,(13) whereas the type II cadherin-6 appears to be expressed in bone marrow stromal cells/osteoblast precursors as well as in osteoclast precursors, suggesting its involvement in mediating heterotypic interactions (between different cell types).(14) In addition, monoclonal antibodies against E-cadherin have been reported to interfere with osteoblast cell-to-cell adhesion;(15) however, expression of E-cadherin mRNA has not been demonstrated in various osteoblast cell types.(13)

Using a consensus sequence derived from the conserved cadherin cytoplasmic domain, we identified two N-cadherin-related transcripts expressed in rat preosteoblastic cells (RCT-1) and in rat osteosarcoma cells (ROS 17/2.8). Subsequently, the corresponding full-length complementary DNA was isolated from both a ROS cell and a rat tibia cDNA library.(16) Sequencing and tissue distribution of this osteo-blast-derived rat N-cadherin gene confirmed its virtual identity to human and murine N-cadherin genes.(17)

Despite growing evidence for the expression of various cadherins in osteoblasts, their role in osteoblast differentiation remains largely unknown. This prompted us to investigate N-cadherin mRNA expression in bone in situ, in primary cultures of fetal rat calvaria (FRC) and in SaOS-2 and ROS cells. We hypothesized that the observed pheno-typic differences between the two latter well-differentiated osteoblast-like cell models may be related to the expression of N-cadherin. Furthermore, we examined the influence of N-cadherin interactions on osteoblast cell-to-cell adhesion and differentiation by using a synthetic peptide containing the HAV adhesion motif of type I cadherins and by transfecting osteoblast-like cells with a dominant-negative N-cadherin mutant.(18)


Osteoblast-like cell culture and morphology

Human SaOS-2/B10(19) and rat ROS 17/2.8 osteosarcoma cells(20) (kindly provided by G. A. Rodan, Merck Research Laboratories, West Point, PA, U.S.A.), were maintained in RPMI 1640 supplemented with 10% fetal bovine serum (FBS) and 2 mM glutamine and in Ham's F-12 medium supplemented with 10% FBS (all reagents from GIBCO BRL, Gaithersburg, MD, U.S.A.), respectively, at 37°C in a humidified atmosphere of 95% air/5% CO2. The medium was changed every other day, and cells were subcultured every week. For most experiments, cells were plated at a density of 30–35 × 103/cm2 on 35-mm-diameter plastic dishes (Costar; Corning, Inc., Corning, NY, U.S.A.) and grown as described above for 1–8 days. Cell numbers in various experimental conditions were calculated by counting trypsinized cells resuspended in isotonic saline by using a Coulter Counter (Coulter Electronics, Inc., Hielade, FL, U.S.A.). Results represent means from at least triplicate determinations and were repeated in more than three separate experiments.

Cell morphology was evaluated by phase-contrast microscopy (Nikon Diaphot 300) performed on living cells cultured on plastic dishes after replacement of the culture medium with 2 ml of Dulbecco's modified phosphate-buffered saline (PBS). Micrographs were taken with a Sensys charge-coupled device (CCD) digital camera (Photometrics, Tucson, AZ, U.S.A.) and Image-Pro Plus software (Media Cybernetics, Silver Spring, MD, U.S.A.).

TRAB-11 cells derived from trabecular rat bone were grown in RPMI 1640 medium supplemented with 10% FBS.(21) At the time of the experiment, TRAB-11 cells were trypsinized and plated at 5000 cells/well in RPMI 1640 supplemented with 1% bovine serum albumin (BSA), into 96-well tissue culture plates precoated with 90 μl of growth factor-reduced Matrigel (Collaborative Research, Bedford, MA, U.S.A.). Network formation was monitored hourly for 14 h.

Cultures of FRC and bone nodule formation

Primary osteoblast cells were prepared from FRC as described previously.(22,23) Briefly, cells were harvested by sequential collagenase/trypsin (Sigma Chemical Co., St. Louis, MO, U.S.A.) digestions of calvaria from 19-day fetuses. Cells from the third to fifth digestions were pooled and plated at 25,000 cells/cm2 in T-75 tissue culture flasks with a modified essential medium (α-MEM) supplemented with 10% FBS, vitamin C (100 μg/ml), 100 U/ml penicillin, and 100 μg/ml streptomycin sulfate (all reagents from GIBCO) and grown at 37°C with 5% CO2 for 3–5 days (or until confluence). After the third passage, confluent FRC cells in 12-well plates were treated with α-MEM supplemented with 5% FBS, 100 μg/ml ascorbic acid, and 5 mM β-glycerophosphate (both from Sigma) in the presence of HAV peptide or reverse control peptide (see below). Progression of nodule formation was monitored for up to 28 days. Osteocalcin (OC) levels in the culture medium were measured using a standard OC assay according to the manufacturer's instructions (Biomedical Technologies, Inc., Stoughton, MA, U.S.A.).

Northern blot analysis

Cells from three 35-mm-diameter dishes were trypsinized and pooled before centrifugation at 2000g for 10 min. Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Santa Clarita, CA, U.S.A.) according to the manufacturer's protocol. Total RNA was suspended in 60 μl of RNAse-free water, then precipitated again with 100% ethanol and 200 mM sodium acetate, pH 4. The RNA pellet was washed once in 70% ethanol and resuspended in an appropriate volume of RNAse-free water. Total RNA was determined by measuring absorbance at 260 nm.

RNA (20 μg/lane) was separated on denaturing formaldehyde/1% agarose gels by electrophoresis, and equivalent RNA loading was verified by staining with ethidium bromide. RNA gels were blotted overnight onto nylon membranes (Amersham Life Science, Inc., Arlington, IL, U.S.A.) using Turboblotter (Schleicher & Schuell, Keene, NH, U.S.A.) and UV cross-linked. Specific cDNA probes were labeled with [α-32P]dCTP by random priming (Oligolabelling Kit; Pharmacia, Piscataway, NJ, U.S.A.) and purified on NucTrap columns (Stratagene, La Jolla, CA, U.S.A.). The membranes were hybridized at 68°C for 1 h in Express Hyb solution (Clontech, Palo Alto, CA, U.S.A.) with 32P-labeled riboprobes at 106 cpm/ml, washed twice in 2× standard saline citrate (SSC) with 0.05% sodium dodecyl sulfate (SDS) followed by 0.1× SSC with 0.1% SDS, then exposed at −70°C.

cDNA probes used in these studies were full-length rat osteoblast N-cadherin;(17) full-length rat alkaline phosphatase (ALP);(24) a 1.4-kilobase polymerase chain reaction (PCR)-derived PTH1RCc fragment;(25) and full-length zif268 and c-jun.(26) Inserts were purified from agarose gels by QiaexII (Qiagen) before radiolabeling.

Autoradiographs were scanned (Scan Jet 4C; Hewlett Packard, Palo Alto, CA, U.S.A.), and the intensity of specific hybridization signals was evaluated by densitometry (NIH Image, Version 1.61) and further normalized to the background of each respective lane. Statistical analysis for multiple comparisons was performed with ANOVA (Stat-view 4.0; SAS Institute, Inc., Cary, NC, U.S.A.). Northern blot data are representative of three to five separate experiments.


Human osteophytic bone, removed from femoral heads obtained from routine hip replacement surgery, was dipped in 10% polyvinyl alcohol (PVA) (Sigma), immediately frozen in chilled hexane (−70°C), and mounted in 10% PVA on brass microtome chucks. Sections (5- to 7-μm thickness) were prepared, fixed in 4% paraformaldehyde, depleted from endogenous peroxidase activity, and further preincubated with 10% normal horse serum (Vector Laboratories, UK) to block nonspecific antibody binding, as previously described.(27) Sections were then incubated for 30 minutes with anti-N-cadherin monoclonal antibody (1:200 dilution; Sigma), which identifies N-cadherin, followed by biotinylated horse anti-mouse secondary antibody (1:200 dilution; Vector Laboratories) for 15 minutes and avidin-biotinylated-peroxidase reagent (1:50 dilution; ABC Elite; Vector Laboratories) for 20 minutes. Peroxidase activity was visualized with 0.5 mg/ml 3,3′-diaminobenzidine (Sigma) and 0.3% hydrogen peroxide as substrate. All dilutions were made up in PBS, pH 7.4, and incubations were performed at room temperature with three PBS washes between each incubation. Negative controls received the same concentration of normal mouse immunoglobulin G (IgG) (Vector Laboratories) in place of secondary antibody. Sections were counterstained with hematoxylin before mounting in glycerol-PBS.

In situ hybridization

Rats were anesthesized with ketamine–xylazine and killed by CO2 inhalation; the lower limbs were perfused with a solution of 4% paraformaldehyde. The fixed tissues were then decalcified over a 4-week period with daily changes of 10% EDTA. The fixed rat tibia samples were decalcified, embedded in paraffin, and sectioned as described earlier.(27) Serial sections were demineralized with citric acid, treated with 1 μg/ml proteinase K (Sigma), and acetylated with acetic anhydride. The sections were rinsed in water treated with diethylpyrocarbonate (DEPC) and dehydrated through graded ethanol (30%, 60%, 80%, 99% ethanol mixed with DEPC-treated water) and air dried. Sections were hybridized overnight with labeled riboprobes prepared with [33P]UTP (Amersham) at 55°C in a humid chamber. Hybridization was carried out in a buffer containing 50% formamide, 2× SSC, 300 mM NaCl, 20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 10% dextran sulfate, 100 mM dithiothreitol (DTT), transfer RNA (tRNA), salmon sperm DNA, and 1 × Denhardt's solution. Posthybridization washes were optimized by using 50% formamide in 1 mM EDTA. After hybridization, the slides were dehydrated, coated with photographic emulsion (Kodak, NTB-2; Rochester, NY, U.S.A.), and exposed for 10–14 days before developing and counterstaining with hematoxylin and eosin.

Preparation of a synthetic HAV-containing peptide

The HAV decapeptide LRAHAVDING(-NH2) containing the His-Ala-Val sequence from the first extracellular domain of type I cadherins(9) and its inactive reverse isomer (R-HAV), GNIDVAHARL(-NH2), were synthesized by using a 430A automatic peptide synthesizer (Applied Biosystems, Inc., Foster City, CA, U.S A.) and purified by reversephase high-performance liquid chromatography (HPLC) following methods previously described by our laboratory.(28) In selected experiments, calvarial osteoblasts or osteoblast-like cells were treated daily with HAV or R-HAV at various concentrations (0.01–1.0 mg/ml).

Transient transfection with a dominant-negative mutant

The N-cadherin dominant-negative mutant (NCADΔC) (a generous gift of Dr. C. Kintner, CA, U.S.A.) had been produced by the deletion of the signal, pre- and extracellular domains 1–4 of Xenopus N-cadherin and substitution of the remaining cysteins in extracellular domain 5 with serines.(18) The NCADΔC construct was regenerated by PCR to incorporate the polyadenylation motif “TTTATT” at the 3′-end and the product subcloned into pcDNA3 (Invitrogen, Carlsbad, CA, U.S.A.). The integrity and orientation of the final construct were verified by DNA sequencing before transfection.

SaOS-2 cells were plated at 16 × 103 /cm2 on 35-mm-diameter dishes and left to adhere overnight before transient transfection with 2.2 μg/well pcDNA3-NCADΔC or equivalent amounts of the empty vector (pcDNA3) using 4 μl/well Lipofectamine Plus (GIBCO). Transfection medium was removed after 4–6 h, and cells were grown with fresh culture medium for 3 days before microscopic evaluation. Assessment of transfection efficiency and cotransfection experiments were performed by using a pSv-ß-Galactosidase vector (Promega, Madison, WI, U.S.A.). Positive transfectants were identified by fixing the cells briefly in 4% paraformaldehyde followed by 2-h incubation with 1 mg/ml X-gal (5-bromo-4-chloro-3-indolyl-ß-D-galactoside; GIBCO) at 37°C. Four separate transient transfection experiments were performed in triplicate. Data represent reproducible observations within and between experiments, as assessed independently by two of the investigators (SF and MT).

The effects of a NCADΔC on SaOS-2 cell morphology were further evaluated by calculating a morphology index, that is, the ratio of cell long-axis length to cell short-axis length.(29) For this purpose, four distinct cell culture areas were randomly selected on digitalized micrographs, and the axis lengths of 20 adjacent cells in each area were measured by using Image-Pro Plus software. Results represent the mean morphology index of eight distinct micrographic areas from two separate experiments, equivalent to 160 cells per experimental condition.

Aggregation assay in cells stably transfected with osteoblast N-cadherin

NCTC cells derived from mouse connective tissue, which lack endogenous cadherin expression,(30) were transfected with the full-length osteoblast N-cadherin cDNA subcloned into pCDNA3, and stable transfectants (NCTC-4) selected with G418 (800 μg/ml). Furthermore, NCTC-4 were transiently transfected with NCADΔC or control vector as explained earlier.

Calcium-dependent aggregation assays were performed in NCTC-4 cells as described previously.(31,32) Briefly, cells were detached with 0.001% trypsin in the presence of 0.04 mM Ca2+ (to protect cadherins), then suspended in medium containing soybean trypsin inhibitor and allowed to aggregate in BSA-precoated 24-well dishes by shaking in the presence of 1.25 mM Ca2+ for 10–60 minutes. In experiments where the effects of HAV peptide were evaluated, the peptide was added during this period. The number of cells was counted by using a Coulter counter at the beginning (N0) and at the end (N10–60) of the assay. The N10–60-to-N0 ratio is the aggregation index. A value of 1 indicates no aggregation, whereas lower values indicate aggregation, that is, an apparent decrease in the number of isolated cells.


N-Cadherin expression in rat and human bone in vivo

N-Cadherin mRNA expression in bone was assessed by performing in situ hybridization of longitudinal rat tibial sections with an osteoblast-derived N-cadherin-specific antisense cDNA. The signal was predominantly and specifically localized to the cells lining the periosteum, presumably functional osteoblasts (Fig. 1A). Similar sections hybridized with a sense control probe showed no significant hybridization (Fig. 1B). In other sections, specific N-cadherin hybridization was observed in differentiating osteoblasts located in the primary spongiosa and in some marrow stromal cells (data not shown).

We further identified N-cadherin protein expression in sections of human osteophytes by immunohistochemistry using a specific monoclonal antibody that confirmed the preferential expression of N-cadherin on lining osteoblasts (Figs. 1C and 1D). We also observed α-, β-, and γ-catenin expression in sections of human osteophytes (data not shown).

N-Cadherin mRNA expression in osteoblast-like cells

The expression of N-cadherin mRNA was further investigated in two cell culture models representing well-differentiated osteoblast-like cells but with different pheno-types, namely, human SaOS-2 and rat ROS cells. SaOS-2 cells (clone B10) showed prominent morphological changes during the rapid proliferation phase. They appeared at first as predominantly isolated, spindle-shaped cells that spread on the plastic surface, then changed to cuboidal and eventually contracted cells as cell-to-cell adhesion occurred (Fig. 2). In contrast, ROS cells were characterized by well-delineated cell borders and formed aggregates of small cuboidal cells within a few hours after plating, without further changes thereafter (Fig. 2).

N-Cadherin mRNA expression was detectable in both cell lines 24 h after plating (Fig. 3A). In SaOS-2 cells, N-cadherin mRNA was seen as two distinct hybridization signals (∼4.4 and 5.6 kb, respectively), presumably representing two alternatively spliced mRNA transcripts. In ROS cells, N-cadherin mRNA was expressed as a single, strong broad signal of ∼5.6 kb. A significant time-dependent increase of N-cadherin mRNA levels subsequently occurred in SaOS-2 cells (Figs. 3A and 3B) in relation to cell-to-cell adhesion and morphological differentiation (Fig. 2). In ROS cells, N-cadherin mRNA levels remained remarkably stable overtime (Figs. 3A and 3B).

Figure Fig. 1..

Expression of N-cadherin in rat and human bone in vivo. (A) In situ hybridization analysis using a radiolabeled N-cadherin antisense cRNA probe viewed by light field in rat tibia. N-Cadherin mRNA expression is localized to osteoblasts (arrows) and, less markedly, to osteocytes (arrowheads). (B) Control rat tibia sections hybridized with sense cRNA probe, showing no significant hybridization. (C) Immunolocalization of N-cadherin in human osteophytic bone. Human osteophytes incubated with mouse anti-N-cadherin monoclonal antibody were visualized by secondary antibody-coupled peroxidase reaction and demonstrate pericellular N-cadherin immunoreactivity on osteoblasts lining bone surfaces (arrows). Staining was absent in osteocytes (arrowheads). (D) Control mouse IgG. Hematoxylin counterstain. (magnification ×100)

We next examined whether N-cadherin mRNA levels were associated with the expression of specific osteoblastic genes, such as ALP, OC, or the PTH1 receptor (PTH1RCc). At the time of maximal N-cadherin mRNA expression in SaOS-2 cells (i.e., after 8 days in culture), ALP and OC gene expression was markedly decreased and increased, respectively, whereas PTH1RCc mRNA levels remained almost unchanged (Fig. 4). Moreover, 6-h treatment with PTH further induced both N-cadherin and OC mRNA expression (data not shown). In contrast, in ROS cells, the specific time-dependent changes of N-cadherin, ALP, and OC mRNA expression were not observed (Fig. 4).

Effects of a cadherin adhesive domain-containing peptide (HAV) on osteoblasts

To gain direct insight into the function of endogenously expressed cadherins in osteoblasts, various osteoblast-like cells were exposed to a synthetic peptide (HAV) containing the His-Ala-Val adhesion motif from the first extracellular domain of type I classic cadherins(9) and compared with controls treated with a reverse, inactive peptide (R-HAV).

In primary cultures of FRC, cells mature from a proliferative state to an advanced stage of osteoblastic differentiation, as shown by a marked time-dependent increase in OC levels measured in the conditioned medium (Fig. 5B). Northern blot hybridization showed that these changes were paralleled by a time-dependent increase of N-cadherin mRNA expression (Fig. 5A). In primary cultures of FRC continuously treated with HAV peptide during 3 weeks, there was a significant concentration-dependent decrease in nodule formation compared with cultures treated with R-HAV (−20 ± 5% and −62 ± 8% [p < 0.05] at 0.1 and 1.0 mg/ml HAV peptide, respectively) (Fig. 5C). Moreover, OC levels remained low (<50 ng/ml) in the conditioned medium of HAV-treated cells after 3 weeks.

When plated on an extracellular matrix, rat TRAB-11 cells differentiate as osteoblasts and develop long processes connecting individual cell bodies within 4–5 h after plating (Fig. 6).(21) A concentration-dependent inhibition of the interconnecting cytoplasmic processes was observed in cells cultured in the presence of HAV peptide, with a complete absence of cell-to-cell contacts at 1.0 mg/ml HAV, whereas cells treated with control R-HAV peptide developed normally (Fig. 6). Removal of the HAV peptide and replacement of the cell growth medium completely reversed the inhibition of TRAB-11 network formation (data not shown).

Figure Fig. 2..

Morphological differentiation of SaOS-2 and ROS cells. Photomicrographs show SaOS-2 cells cultured on plastic dishes after 1 (a), 3 (b), and 4 (c) days in complete medium. ROS cells are shown after 1 (d) and 4 (e) days. Poorly differentiated SaOS-2 cells are characterized by a spindle-shaped morphology (arrows) or mature, small cuboidal cells (arrowheads). (magnification ×200)

SaOS-2 cells treated with HAV peptide (1.0 mg/ml) for 48 h proliferated normally, and no remarkable differences in morphological differentiation were observed (data not shown). However, this treatment inhibited ALP mRNA expression, without affecting endogenous N-cadherin mRNA levels (Fig. 7A). Moreover, short exposure to HAV peptide (1.0 mg/ml) resulted in a rapid and transient increase in the levels of the immediate early gene zif268, which was maximally induced within 30 minutes of treatment (Fig. 7B). The effects of HAV peptide on early gene expression appeared to be specific for zif268, because c-jun did not show the same regulation (Fig. 7B).

Effects of NCADΔC on SaOS-2 cell adhesion and morphological differentiation

To disrupt functional interactions between endogenous cadherins, SaOS-2 cells were transiently transfected with a NCADΔC that lacked the extracellular cadherin domain. This mutant, although defective in its contact domains, can bind and sequester intracellular catenins, demonstrating intact function of its cytoplasmic domain.(18)

Transfection with pcDNA3-NCADΔC did not significantly alter the cell culture density after 3 days versus controls (94.6 ± 7.6, 102.3 ± 8.0, and 112.9 ± 10.5 × 103 cells/cm2 in NCADΔC-transfected, empty vector-transfected, and untransfected cells, respectively). N-Cadherin mRNA also was normally expressed in these cells (data not shown). The most dramatic effects, however, were on cell-to-cell adhesion and morphological differentiation, which both were markedly reduced in NCADΔC-transfected cells; most of the cells maintained a distinct spindle-shaped morphology (Fig. 8). Control cells transfected with the empty vector matured into typical aggregates of small cuboidal cells, as did untransfected cells (Fig. 8). As a result, cultures of NCADΔC-transfected cells appeared to be less confluent than cultures of empty vector–transfected cells, despite similar cell densities (see earlier). A quantitative analysis of the cell morphology further indicated a significantly higher morphology index (i.e., a more elongated shape) in NCADΔC-transfected cells than in empty vector–transfected or untransfected controls (Table 1).

Figure Fig. 3..

Time-dependent changes of N-cadherin mRNA expression in SaOS-2 and ROS cells. (A) Northern blots showing hybridization of N-cadherin RNA from SaOS-2 and ROS cells after 1–4 days in culture. Positions of the 28S ribosomal band and of molecular weight markers are shown for size reference. Ethidium bromide-stained 28S mRNA is also shown (bottom) as control for the amount of total RNA loaded (20 μg/lane). (B) Quantitative analysis of N-cadherin mRNA levels in SaOS-2 (hatched bars) and ROS cells (open bars). The intensity of the specific hybridization signals was determined by densitometry and normalized to the background of the respective lanes. Results are means ± SEM of three separate experiments and are expressed as fold increase over basal (day 1). *p < 0.05 versus baseline.

Cotransfection experiments with a pSv-LacZ vector and pcDNA3-NCADΔC directly identified cells transfected with the dominant-negative N-cadherin mutant. As illustrated in Fig. 8, adhesion to adjacent cells was markedly impaired, disrupting the normal osteoblast-like cell network. Importantly, in this case, cell-to-cell contacts were impaired not only between NCADΔC-transfected cells but also with untransfected cells. It appeared that the osteoblast-like cell network did not form unless <30% of the cells were transfected with NCADΔC. In contrast, control cells transfected with pSv-LacZ alone or cotransfected with pSv-LacZ and the empty vector were adherent to several adjacent cells and formed typical osteoblast-like cell networks (Fig. 8).

Figure Fig. 4..

Relationship between N-cadherin mRNA expression and expression of specific osteoblastic genes in SaOS-2 and ROS cells. Total RNA (20 μg/lane) from SaOS-2 and ROS cells cultured 1 and 8 days was hybridized with specific cDNA probes for N-cadherin (N-Cad), PTH type 1 Receptor (PTH1Rc), ALP, and OC. Hybridizations were performed simultaneously for both cell lines, and exposure times varied from 18 h (N-Cad and PTH1Rc) to 36 and 72 h (ALP and OC, respectively). Positions of the 28S and 18S ribosomal bands as well as molecular weight markers are indicated for size reference. Ethidium bromide–stained 28S mRNA is also shown (bottom) as control for the amount of total mRNA loaded.

Effects of NCADΔC on aggregation of mouse connective tissue cells stably transfected with osteoblast-derived N-cadherin

To gain insight into the specific influence of the osteoblast-derived N-cadherin on cell-to-cell adhesion, NCTC cells derived from mouse connective tissue, which lack endogenous cadherin expression,(30–32) were stably transfected with the osteoblast N-cadherin cDNA construct (NCTC-4 cells), and their capacity to aggregate was further evaluated by a specific aggregation assay. Compared with parental NCTC cells, NCTC-4 cells aggregated rapidly in the presence of calcium (Fig. 9), which is characteristic of cadherin-promoted cell-to-cell adhesion. Furthermore, aggregation in NCTC-4 cells transiently transfected with NCADΔC was inhibited almost back to the level of the parental cells (Fig. 9). Aggregation of both NCTC-4 cells and SaOS-2 cells, as evaluated by this assay, was also significantly inhibited in the presence HAV peptide (1 mg/ml; data not shown).

Figure Fig. 5..

N-Cadherin mRNA expression in FRC primary cultures. (A) Total RNA (20 μg/lane) was isolated from FRC cultures at the indicated days and hybridized with an N-Cad-specific cDNA probe. Exposure time was 18 h. Arrows show positions of the N-cadherin mRNA transcripts. Loading was controlled by the steady-state screening with a β-actin cDNA probe (bottom). (B) Time-dependent increase of OC production in these cultures is shown as a marker of osteoblastic differentiation. (C) In primary osteoblastic cell cultures treated for 3 weeks with a synthetic peptide containing the His-Ala-Val adhesion motif of N-cadherin first extracellular domain (HAV; open bars), there was a concentration-dependent decrease of bone nodule area versus controls treated with a reverse, inactive peptide (R-HAV, 1 mg/ml; hatched bar).


Our observations confirm the expression of osteoblast N-cadherin mRNA in rat ROS and human SaOS-2 osteoblast-like cells, in primary cultures of FRC, and in rat and human bone in vivo. The presence of two N-cadherin mRNA transcripts of different sizes is presumably the result of post-transcriptional modification or alternate splicing, as recently shown for cadherin-6 gene expression in osteoclasts.(14) The N-cadherin gene is thus one of the limited repertoire of cadherin genes expressed in osteoblasts.(11–15)

N-Cadherin mRNA expression increases concomitantly with osteoblastic differentiation, as evaluated by changes of cell morphology and specific gene expression. Furthermore, in situ hybridization shows that N-cadherin is predominantly expressed in mature osteoblasts lining the periosteal surface (Fig. 1). In addition, PTH, which stimulates osteoblastic differentiation and bone formation in vivo, increases both N-cadherin and OC mRNA expression in osteoblast-like cells.(17) PTH also stimulates cell coupling between adjacent osteoblasts,(4–8) whereas the degree of cell coupling and stage of differentiation modulates the PTH-responsiveness of osteoblasts.(7,33) Together, these data suggest a role for N-cadherin in osteoblast cell-to-cell adhesion and late stage(s) of differentiation.

Figure Fig. 6..

Effects of HAV peptide on the morphological differentiation of osteoblast-like cells. Photomicrographs show rat osteoblastic TRAB-11 cells cultured on Matrigel matrix with an inactive, control peptide (R-HAV, 1.0 mg/ml) (a) or increasing concentrations of HAV peptide: 0.01 mg/ml (b), 0.1 mg/ml (c), and 1.0 mg/ml (d). HAV inhibited the development of interconnecting cytoplasmic processes within a few hours after plating. (magnification ×100)

To gain direct insight into the role played by osteoblast cadherins in differentiation, we used both a pharmacological and a molecular approach to interfere with endogenous cadherin function. Treatment of rat trabecular bone-derived TRAB-11 cells(21) with a synthetic peptide containing the HAV motif from the first extracellular domain of type I cadherins, which is the major site of homophilic molecular interactions(9), inhibits the formation of cytoplasmic network processes and the establishment of cell-to-cell contacts (Fig. 6). Treatment with HAV peptide also significantly decreases, but does not abolish, bone nodule formation in primary cultures of FRC. In contrast, HAV peptide has no detectable effects on morphological differentiation in SaOS-2 or ROS cells. However, it down-regulates ALP gene expression and specifically induces the expression of the immediate early gene zif268 in SaOS-2 cells. Similar HAV effects on ALP protein synthesis have been reported in bone marrow osteoprogenitor stromal cells cultured for 7 days.(13) As for zif268, it is known to be regulated during preosteoblastic differentiation.(26) These observations suggest that N-cadherin and other osteoblast cadherins may be directly involved in osteoblast differentiation.

The stimulation of cadherin signaling by short-term treatment with HAV peptide has rapid and direct effects on target gene expression, whereas the regulation of more complex processes, such as morphological changes, may require long-term homotypic (cell-to-cell) interactions and coupling. In this regard, the failure of HAV to completely inhibit cadherin-mediated cell-to-cell adhesion in ROS and SaOS-2 cells appears particularly interesting. This observation may be explained by the inability of the HAV peptide to fully compete for the relatively high number of cadherin adhesion sites on the cell surface. In fact, the down-regulation of ALP gene expression in early SaOS-2 cell cultures treated with HAV peptide is similar to the pattern observed in late SaOS-2 cell cultures endogenously expressing high levels of N-cadherin mRNA. This observation suggests that HAV may interact with cadherins as an agonist rather than an antagonist in these cells. However, the failure to inhibit cell-to-cell adhesion and morphological differentiation in SaOS-2 cells suggests that other cell adhesion molecules, such as integrins, are required to support osteoblast cell-to-cell adhesion and differentiation.

Figure Fig. 7..

Effects of HAV peptide on specific osteoblast-like cell gene expression. (A) Northern blots of SaOS-2 cells continuously treated with HAV (1 mg/ml) during 48 h show decreased ALP mRNA but unchanged N-Cad mRNA expression versus control cells treated with a reverse, inactive peptide (HAV-, 1 mg/ml). Positions of 2.5-, 4.4-, and 5.6-kb markers are shown for size reference. Ethidium bromide-stained 28S mRNA is also shown (bottom) as control for the amount of total mRNA loaded (20 μg/lane). (B) Northern blot of the immediate early gene zif268 and c-jun mRNA expression in SaOS-2 cells treated with HAV. SaOS-2 cells were treated with HAV (1 mg/ml) for the times indicated, and total RNA was isolated. The blot was screened with a zif268 cDNA or a c-jun cDNA probe, showing a specific time-dependent, increase of zif268 expression by HAV. Both 28S and 18S ribosomal RNA markers are shown. Ethidium bromide staining of the 28S mRNA on the original Northern blot is shown (bottom).

To selectively antagonize endogenous cadherin function, osteoblast-like cells were further transfected with NCADΔC. Transfected cells are unable to support homophilic interactions, yet maintain their capacity to interact with and sequester intracellular catenins, which are the predominant signaling molecules for cadherins.(18,34) This strategy has recently proved to be effective in disrupting morphogenesis in the developing embryo,(9,18) cardiac tissue differentiation,(35) and osteoclastogenesis.(14) In the present study, both cell-to-cell adhesion and morphological differentiation were prevented in SaOS-2 cells transiently transfected with NCADΔC. Interestingly, these effects were not observed in ROS cells, which died after transfection with NCADΔC (data not shown). In this case, the apparent failure to transfect ROS cells with NCADΔC also could reflect the role of cadherins in supporting mature osteoblast differentiation.

Figure Fig. 8..

Effects of NCADΔC on the morphology and cell-to-cell adhesion of SaOS-2 cells. (a) Photomicrographs of SaOS-2 cells 3 days after transient transfection with pcDNA3-NCADΔC show predominantly a spindle-shaped cell morphology. Cells transfected with an empty vector (pcDNA3) (b) and untransfected cells (c) demonstrate cell-to-cell adhesion and morphological maturation into small cuboidal cells. In the other representative experiment, SaOS-2 cells transiently transfected with pSv-LacZ (d), cotransfected with pSv-LacZ and pcDNA3 (e) or with pSv-LacZ and pcDNA3-NCADΔC (f) are identified by blue staining after incubation with X-gal. Disruption of cell-to-cell adhesion and cell network organization is present in NCADΔC-transfected cells. Results are representative of four separate experiments. (magnification ×100 [a–d], ×200 [e, f])

Table Table 1.. Influence of a Dominant-Negative N-Cadherin Mutant on SaOS-2 Cells Morphology Index
  1. Cells were Lipofectamine®-transfected with 2.2 μg/well of an empty pcDNA3 vector (Vector) or a pcDNA3-dominant-negative N-cadherin mutant construct (NCADΔC). Cell morphology was quantitated using digitalized micrographs of subconfluent cell cultures, i.e. three days after transfection. A morphology index (cell long axis/short axis length) (21) close to 1 indicates a cuboidal shape, whereas a higher index indicates a more elongated shape. Results represent the mean and range of morphology indices calculated in 8 distinct cell culture areas from 2 separate experiments.

  2. **p < 0.0001 as compared to either untransfected or empty vector-transfected cells (ANOVA).

Mean(±sem)2.49 (±0.16)2.01 (±0.12)4.18 (±0.14)**

Although our results support a direct role of N-cadherin in osteoblastic cell-to-cell adhesion and differentiation, they should not be interpreted as evidence for a unique role of N-cadherin in these complex processes. Indeed, not only are multiple cadherins subtypes abundantly expressed in various osteoblast-like cell models,(11–17) but also antibodies against E-cadherin, another type I classic cadherin, inhibit cell-to-cell adhesion in UMR 106 and MC3T3-E1 osteoblast-like cells.(15) Besides, dominant-negative cadherin constructs such as NCADΔC are not specific for one cadherin sub-type (9,14,18) Unfortunately, the N-cadherin null mutation model is lethal early in the course of embryogenesis and does not provide any valuable information on the importance of N-cadherin in the late stage of differentiation of bone-forming cells.(36) Nevertheless, aggregation assays in mouse connective tissue-derived NCTC cells stably transfected with an osteoblast-derived N-cadherin (Fig. 9) support a specific role for this molecule in cell-to-cell adhesion. Taken together, the data presented here indicate that N-cadherin participates to the organization and function of osteoblasts, with other previously identified osteoblast cadherins. As such, the cadherin family may represent novel targets for the development of new anabolic agents in the treatment of osteoporosis.


We are indebted to S. Steuckle for dedication and technical expertise and to J. Lian for assistance with bone nodule formation assays.

S. L. Ferrari is supported by a postdoctoral fellowship grant from the Swiss National Foundation for Scientific Research and the Fondation Suisse des Bourses en Médecine et Biologie.

Figure Fig. 9..

Effects of NCADΔC on aggregation of NCTC cells derived from mouse connective tissue stably transfected with osteoblast N-cadherin. NCTC cells stably transfected with an osteoblast-derived N-cadherin (NCTC-4; ▪), untransfected (parental) cells (□), and NCTC-4 cells transiently transfected with NCADΔC (○) were trypsinized and allowed to adhere on BSA-coated dishes by shaking in the presence of 1.25 mM Ca2+ for 10–60 minutes. Aggregation index is a ratio of the cell number counted at the end of this period to the cell number at the beginning of the assay. A value of 1 indicates no aggregation, whereas lower values indicate aggregation. Results show that rapid aggregation occurred in NCTC-4 cells and was competed by NCADΔC.