Differential Effect of BMP4 on NIH/3T3 and C2C12 Cells: Implications for Endochondral Bone Formation

Transcription

1 JOURNAL OF BONE AND MINERAL RESEARCH Volume 20, Number 9, 2005 Published online on May 23, 2005; doi: /JBMR American Society for Bone and Mineral Research Differential Effect of BMP4 on NIH/3T3 and C2C12 Cells: Implications for Endochondral Bone Formation Guangheng Li, 1 Hairong Peng, 1 Karin Corsi, 1,2 Arvydas Usas, 1 Anne Olshanski, 1 and Johnny Huard 1,2,3 ABSTRACT: After intramuscular implantation, BMP4-expressing NIH/3T3 fibroblasts and BMP4-expressing C2C12 myoblasts can promote ectopic cartilage and bone formation. Fibroblasts tend to undergo chondrogenesis, whereas myoblasts primarily undergo osteogenesis. These results suggest that endochondral bone formation may involve different cell types, a finding that could have major implications for the tissue engineering of bone and cartilage. Introduction: The delivery of BMP4 through cell-based gene therapy can trigger ectopic endochondral bone formation in skeletal muscle. We hypothesized that, when stimulated with or transduced to express BMP4, different types of cells residing within skeletal muscle might participate in different stages of endochondral bone formation. Materials and Methods: We compared the responses of a fibroblast cell line (NIH/3T3), a myoblast cell line (C2C12), primary fibroblasts, and primary myoblasts to BMP4 stimulation in vitro. We then transduced the four cell populations to express BMP4 and compared their ability to promote ectopic endochondral bone formation in skeletal muscle. Results: Under the influence of BMP4 in vitro and in vivo, NIH/3T3 cells differentiated toward both chondrogenic and osteogenic lineages, whereas most C2C12 cells underwent primarily osteogenic differentiation. NIH/3T3 cells genetically modified to express BMP4 induced delayed but more robust cartilage formation than did genetically modified C2C12 cells, which promoted rapid ossification. These differences in terms of the timing and amount of cartilage and bone formation persisted even after we introduced a retrovirus encoding dominant negative Runx2 (DNRunx2) into the C2C12 cells, which interferes with the function of Runx2. Superior osteogenic potential was also displayed by the primary myoblasts in vitro and in vivo compared with the primary fibroblasts. The different proliferation abilities and differentiation potentials exhibited by these cells when influenced by BMP4 may at least partially explain the differing roles that BMP4-expressing myogenic cells and BMP4-expressing fibroblastic cells play in endochondral bone formation. Conclusions: Our findings suggest that the process of endochondral bone formation in skeletal muscle after delivery of BMP4 involves different cell types, including fibroblastic cells, which are more involved in the chondrogenic phases, and myoblastic cells, which are primarily involved in osteogenesis. These findings could have important implications for the development of tissue engineering applications focused on bone and cartilage repair. J Bone Miner Res 2005;20: Published online on May 23, 2005; doi: /JBMR Key words: BMP4, skeletal muscle, osteoblast, chondrocyte, fibroblasts, myoblasts, endochondral bone formation The authors have no conflict of interest. INTRODUCTION SKELETAL DEVELOPMENT IS a multistep process that involves patterning of skeletal elements, commitment of mesenchymal cells to chondrogenic and osteogenic lineages, and terminal differentiation of precursor cells into two specialized cell types: the chondrocytes present in cartilage and the osteoblasts present in bone. Intramembranous bone is derived from the direct differentiation of mesenchymal cells into osteoblasts, whereas endochondral bone first appears as a cartilaginous intermediate that later is replaced with bone. (1,2) This process of endochondral ossification that occurs during skeletal development is similar to that observed during ectopic endochondral bone formation triggered by the implantation of BMPs into skeletal muscle. Ectopic bone formation begins with the differentiation of mesenchymal stem cells into chondrocytes 5 7 days after the administration of BMPs; these chondrocytes 1 Growth and Development Laboratory, Children s Hospital of Pittsburgh, Department of Orthopaedic Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA; 2 Department of Bioengineering, University of Pittsburgh, Pittsburgh, Pennsylvania, USA; 3 Department of Molecular Genetics and Biochemistry, University of Pittsburgh, Pittsburgh, Pennsylvania, USA. 1611

2 1612 become hypertrophic chondrocytes sometime between days 10 and 14, and extracellular matrix calcification follows. Osteoblasts attach to the calcified matrix and form bone ossicles days after the administration of BMPs. Meanwhile, osteoclasts digest the matrix formed by the chondrocytes and remodel the newly formed bone tissue. Because they induce ectopic cartilage and bone formation, BMPs are good candidates for applications designed to promote cartilage and bone repair. For the same reason, BMPs are useful in efforts to investigate the complicated process of endochondral bone formation. The administration of BMPs to induce ectopic endochondral bone formation in skeletal muscle can also be accomplished through the use of ex vivo gene therapy, a process that depends on the contributions of donor and host cells. First, donor cells are genetically modified to secrete BMPs into the local area. Host and donor cells respond to the presence of the protein by differentiating into chondrocytes and osteoblasts. The synergistic interactions of the donor and host cells initiate and promote ectopic endochondral bone formation in skeletal muscle. Despite the completion of numerous studies on the topic, the identity of the donor and host cells that become chondrocytes and osteoblasts in vivo and their relation to the overall process of endochondral bone formation remain unclear. To better understand this process, we studied the delivery of BMP4 to skeletal muscle through ex vivo gene transfer based on cell populations that reside in skeletal muscle (i.e., fibroblasts and myoblasts). NIH/3T3 is a cell line of highly contact-inhibited cells established from cultures of NIH Swiss mouse embryos in the same manner as the original randomly bred 3T3 and inbred BALB/c 3T3. (3) NIH/3T3 is a well-known fibroblast cell line often used as a negative control for bone and cartilage studies because of these cells unresponsiveness to BMPs and other growth factors. However, a recent study has shown that NIH/3T3 cells treated with 1,25- dihydroxyvitamin D 3 and dexamethasone can express alkaline phosphatase (ALP) and the transcription factor Corebinding factor 1 (Cbfa-1), now known as runt-related protein 2 (Runx2). (4,5) ALP is an early marker of osteogenic differentiation, whereas Runx2 is a key regulator of osteoblastic differentiation. Unfortunately, researchers still know very little about the biology and differentiation behavior of NIH/3T3 cells influenced by BMP4. C2C12 is a subclone (6) of the C2 mouse myoblast cell line established by Yaffe and Saxel. (7) This cell line differentiates rapidly, and the cells form myotubes and produce differentiated muscle proteins. The in vitro treatment of C2C12 cells with BMP2 causes a shift in the differentiation pathway from myoblastic to osteoblastic. (8,9) C2C12 cells transduced with either an adenovirus containing a BMP4 gene (Ad-BMP4) or an adeno-associated virus containing a BMP2 gene (AAV-BMP2) can produce functional BMP4 and BMP2 proteins and differentiate toward the osteogenic lineage in vitro. (10,11) However, it remains unclear whether C2C12 or NIH/3T3 cells treated with BMP4 will differentiate toward both chondrogenic and osteogenic lineages in vitro and whether they will participate in endochondral bone formation in vivo. To ensure that our findings do not pertain solely to these particular myoblast and fibroblast cell lines, we used a similar experimental design to also test primary fibroblasts and primary myoblasts. MATERIALS AND METHODS Cell culture and isolation NIH/3T3 cells and C2C12 cells (purchased from ATCC) were grown in T-75-cm 2 flasks in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin and were incubated at 37 C in humidified air mixed with 5% CO 2. The cultures were never allowed to become confluent. The cells were trypsinized and passaged every 3 4 days. Primary fibroblasts were isolated from skeletal muscle fascia, and myoblasts were isolated from skeletal muscle of Fischer 344 rats through serial enzymatic digestion with 0.2% collagenase XI, dispase, and 0.1% trypsin. (12) The cell culture protocol described above also was used to culture the primary cells used in these experiments. Immunocytochemical staining for vimentin and desmin NIH/3T3 cells were cultured in 6-well plates and were fixed in 10% formalin at room temperature for 15 minutes. The cells were washed in PBS and blocked with 5% horse serum at room temperature for 30 minutes. For vimentin staining, cells were incubated at room temperature for 45 minutes with vimentin antibody labeled with Cy3 (1:200 in PBS; C9080; Sigma). For desmin staining, rabbit anti-mouse desmin antibody (1:200 in 5% horse serum; Sigma) was added to the cells for 2 h at room temperature. The cells were washed in PBS three times for 5 minutes. A secondary antibody, anti-rabbit IgG with Cy3 conjugate (1:200; Sigma), was added to the cells for 1 h at room temperature. Cells were washed in PBS three times and examined under a fluorescence microscope. The same staining techniques were applied to the primary fibroblasts and myoblasts. Flow cytometry analysis of cells LI ET AL. Flow cytometry analysis was used to determine the percentages of CD34- and stem cell antigen-1 (Sca-1)-positive cells in the NIH/3T3 cultures, C2C12 cultures, primary fibroblast cultures, and primary myoblast cultures. Cultured cells were trypsinized, spun, washed in cold PBS containing 0.5% BSA (ICN Biomedicals), and counted. The cells were divided into equal aliquots, spun, and resuspended in mouse serum (1:10; Sigma-Aldrich) diluted with PBS. Cell suspensions were incubated for 10 minutes on ice. Predetermined optimal amounts of both direct and biotinconjugated monoclonal antibodies (CD34 and Sca-1; BD Pharmingen) were added to the cells for 30 minutes. Streptavidin-allophycocyanin conjugate was added for 20 minutes to tubes containing cells labeled with biotinylated antibodies. Immediately before analysis, 7-amino-actinomycin D (Via-Probe; BD Pharmingen) was added to each tube for dead cell exclusion. Live cells were collected and analyzed on a FACSCalibur (Becton Dickinson) flow cytometer equipped with Cell Quest software.

3 IMPLICATIONS FOR ENDOCHONDRAL BONE FORMATION 1613 Pellet culture NIH/3T3 cells cultured in DMEM supplemented with 10% FBS were released through treatment with 0.05% trypsin in 0.01% EDTA, counted, and used to generate pellet cultures to evaluate chondrogenesis in vitro, as previously described. (13) Cells ( ) were centrifuged at 500g in 15-ml polypropylene conical tubes, and the resulting pellets were cultured for up to 3 weeks in chondrogenic medium (Cambrex) supplemented with BMP4 (500 ng/ml; R&D Systems); medium was changed every 2 3 days. Samples were harvested on days 14 and 21 and embedded in a paraffin block. At each time-point, several pellets were subjected to enzymatic digestion (0.1% trypsin digestion for 90 minutes) to determine the number of cells present. Pellet sections were stained with Alcian blue/nuclear fast red and with type II collagen (Chemicon). In all cases, C2C12 cells were used as the control group. Effect of BMP4 dose and exposure time on the cells NIH/3T3 or C2C12 cells were added to a 6-well plate (2000 cells/cm 2 ) and cultured for 5 days in DMEM containing 10% FBS supplemented with various concentrations of BMP4 (100, 200, 500, or 1000 ng/ml). The cells were fixed and stained with an ALP staining kit as detailed in the manufacturer s protocol (Sigma). The percentage of ALP + cells was calculated from images of five fields obtained with a light microscope. DMEM supplemented with 10% FBS and BMP4 (500 ng/ml) was used to test the effect of treatment time on the differentiation of NIH/3T3 and C2C12 cells. The cells were fixed and were stained for ALP 3, 6, 9, and 12 days after treatment with BMP4. The percentages of ALP + cells were calculated as described previously. Primary fibroblasts and myoblasts also were treated with various doses of BMP4 (25, 50, or 100 ng/ml) for 2 days and were stained for ALP as described above. RT-PCR analysis A mini RNA kit (QIAGEN) was used to extract RNA from NIH/3T3 cells stimulated with BMP4 (500 ng/ml) for various lengths of time. Two pairs of primers were designed on the basis of the mrna sequence of -actin and osteocalcin. Sequences of the primers were as follows: -actin forward, 5 -GTGGGCCGCTCTAGGCACCAA-3 ; -actin reverse, 5 -CTCTTTGATGTCACGCACGATTTC-3 ; osteocalcin forward, 5 -TTGGTGCACACCTAGCAGA- CACCATG-3 ; and osteocalcin reverse, 5 -CTAAATAG- TGATACCGTAGATGCGTT-3. The one-step RT-PCR kit (Invitrogen) was used to perform RT-PCR. Conditions of RT-PCR included denaturation at 94 C for 5 minutes followed by 35 cycles of denaturation at 94 C for 15 s, annealing at 55 C for 30 s, extension at 72 C for 60 s, and extension at 72 C for another 10 minutes. Products were analyzed in 1% agarose gel. Retroviral transduction of cells Retro-BMP4 is an murine leukemia virus (MLV)-based retroviral vector containing a human BMP4 gene plasmid, pclbmp4. The BMP4 virus was produced by transient transfection with three plasmids: one encoding viral structural and enzymic proteins, a second encoding VSV-G (stomatitis vesicular virus G protein), and a third encoding the transfer vector containing the BMP4 construct, pclbmp4. These three plasmids were co-transfected into 293T cells by calcium-phosphate precipitation. Analysis 24 h after transfection revealed viral particles in the conditioned medium. (14) A retrovirus carrying genes for the nuclear localizing signal (nls) and -galactosidase was produced from a stably transfected cell line known as Tel-6. That cell line was established by transfecting the cell line TE671 (human fibrosarcoma), which carries the MFGnls LacZ plasmid and the plasmid CeB-, a plasmid carrying MoMLV gag-pol (a retroviral gene necessary for viral coat and reverse transcriptase production) and bsr (a gene for blasticidine S resistance). (15) NIH/3T3 cells were seeded at a density of cells per 75-cm 2 flask and were grown to 50% confluence. On the day of transduction, cells were washed twice with sterile PBS and incubated with a mixture of 2 ml of retro-lacz viral suspension ( cfu/ml), 18 ml of DMEM supplemented with 10% FBS, and polybrene (8 g/ml). Transduction was carried out at 37 Cin5%CO 2 for a total of 48 h with viral suspension replaced at 16 and 32 h. The percentage of LacZ + cells after transduction was determined by LacZ staining. A similar technique also was used to transduce NIH/3T3- LacZ cells with a retro-bmp4 virus. Medium containing 10 ml of retro-bmp4 viral suspension ( cfu/ml), 10 ml of DMEM with 10% FBS, and polybrene (8 g/ml) was added to the flasks. NIH/3T3-LacZ cells were cultured in this medium for a total of 48 h, with medium being changed every 16 h, as described above. After double transduction, the transduced NIH/3T3 cells were stained for LacZ and ALP to determine the efficiency of transduction. The culture medium from NIH/3T3-LacZ-BMP4 (NIH/3T3-L-B) cells was collected after 48 h, centrifuged at 2000 rpm at 4 C for 5 minutes to remove cellular debris, and used to perform a BMP4 bioassay. C2C12 cells, primary fibroblasts, and primary myoblasts were transduced and analyzed using the same methods. Construction, production, and use of retro-dnrunx2 virus A PSG5 vector containing DNRunx2 (673-bp BamHI- HindIII fragment) was given to us by Dr Toshihisa Komori (Nagasaki University Medical School, Nagasaki, Japan). The DNRunx2 fragment was cut by EcoRI and blunt ended by the klenow enzyme to enable introduction of a BglII site. The retroviral vector was manipulated in the same manner. The insert and vector were ligated and were transfected into Phoenix packaging cells (ATCC) using calciumphosphate precipitation. Medium was collected after 24 h of transfection and was stored at 80 C. The transduction of C2C12-LacZ-BMP4 (C2C12-L-B) cells with retro- DNRunx2 was performed as described above. Bioassay of BMP4 production by transduced cells A previously described BMP4 bioassay (16) was used to determine the level of functional BMP4 secreted by the

4 1614 transduced cells. This bioassay uses C2C12 cells to determine the biological activity of BMP4, as reflected by the results of histochemical staining for ALP. For this assay, 100- l aliquots of serial 2-fold diluted samples (collected as described in the methods for retro-lacz and retro-bmp4 transduction) were added in triplicate to the C2C12 cells, which were seeded at a density of cells/well in 96- well plates. Serial dilutions of rhbmp4 were used in each assay to construct a standard curve for comparison. ELISA (Human BMP4 ELISA Kit; R&D Systems) was used to verify the results obtained from the bioassay. Preparation of gelatin sponge implants A 100- l cell suspension containing cells was seeded on the surface of a 6 6-mm piece of sterile gelatin sponge (Gelfoam; Pharmacia & Upjohn). After the Gelfoam absorbed all the cell suspension, 3 ml of DMEM supplemented with 10% FBS was added to each well, and the wells were placed in a cell incubator overnight. Animal surgery was performed the following day. Samples were implanted into the skeletal muscle pouch of the gluteofemoral muscle of each SCID (immunodeficient) mouse. The mice were killed at different time-points after cell implantation. Intramuscular implantation of NIH/3T3-L-B and C2C12-L-B supernatants NIH/3T3-L-B and C2C12-L-B cells were cultured in T-175 flasks at a density of 2000 cells/cm 2. When cells became confluent, the medium (10% FBS in DMEM) was removed, and 25 ml of CD293 medium (GIBCO) was added to the flasks. The culture supernatants were collected after 2 days, and BMP4 activity was measured by ELISA (Human BMP4 ELISA Kit; R&D Systems). The cells were counted, and the supernatants were dialyzed by distilled water for 2 days at 4 C and lyophilized. The lyophilized powder was dissolved in 300 l of PBS, and aliquots of the solution were allowed to soak into 6 6 pieces of Gelfoam (100 l/sponge). These impregnated Gelfoam sponges were implanted into muscle pouches created in SCID mice. Histological tests were conducted 7, 20, and 28 days after implantation. Cell proliferation assay C2C12 cells, NIH/3T3 cells, primary fibroblasts, and primary myoblasts were plated in 96-well plates (5000 cells/ well) and were cultured with various doses of BMP4 for 2 days. After 2 days, 20 l of CellTiter 96 AQ UEOUS One Reagent was added to each well. The plate was incubated in 5% CO 2 at 37 C for 2 h. A 96-well plate reader was used to measure the absorbance at 490 nm. Radiographic analysis Ectopic bone formation was monitored by X-ray examination of the mice at different time-points after implantation (Model MX-20; Faxitron X-ray Corp.). Northern Eclipse imaging software (Empix Imaging) was used to measure and analyze the relative bone area. Western blot analysis for expression of Runx2 Cell lysates were collected, incubated in boiling water for 5 minutes, centrifuged, and stored at 4 C. Samples were resolved on 10% SDS-PAGE and were transferred to pure nitrocellulose membranes (Bio-Rad Laboratories). The membranes were treated as detailed in the manufacturer s protocol (Vectastain ABC-AmP; Vector Laboratories). Runx2 antibody (S-19; 1:500; Santa Cruz Biotechnology) was used for this experiment. Histological analysis The tissue samples were treated with CRYO-GEL Embedding Medium (Cancer Diagnostics), rapidly frozen in precooled 2-methylbutane (Sigma), and stored at 80 C. Frozen sections were stained as detailed below. Alcian blue/eosin staining: A 1% Alcian blue solution was made with 3% acetic acid. Frozen sections were fixed in 10% formalin and rinsed in distilled water. Slides were placed in 3% acetic acid for 3 minutes and were transferred into Alcian blue solution for 30 minutes. Slides were rinsed in running tap water for 1 minute and counterstained with eosin. LacZ/osteocalcin/eosin staining: Frozen sections were fixed in 1% glutaraldehyde for 1 minute and were washed in PBS three times. The sections were stained in LacZ solution until the blue stain developed appropriately. After three washes in PBS, the sections were fixed in acetone for 5 minutes and were processed for osteocalcin immunostaining according to the manufacturer s protocol (Vectastain Elite ABC Kit; Vector Laboratories). Osteocalcin antibody (1:200; Santa Cruz Biotechnology) was used for immunostaining. These sections were counterstained with eosin. Von Kossa/hematoxylin and eosin staining: Frozen sections were fixed in 10% formalin for 10 minutes and were rinsed in distilled water three times. The slides were stained in 2% silver nitrate solution in the dark for 15 minutes, rinsed three times in distilled water, and exposed to light for minutes until appropriate stain development. These sections were counterstained with hematoxylin and eosin. Human BMP4 immunostaining/hematoxylin staining: Frozen sections were fixed in 10% formalin for 10 minutes and were rinsed three times in PBS. The sections were processed for human BMP4 immunostaining as suggested in the manufacturer s protocol (Vectastain Elite ABC Kit, Dab Substrate Kit for Peroxidase; Vector Laboratories). Human BMP4 antibody (1:100 dilution, AF757; R&D Systems) was used for the immunostaining. These sections were counterstained with hematoxylin. Quantitative measurement of newly formed cartilage, bone, and the density of human BMP4 staining LI ET AL. Full views of histological sections were obtained by light microscopy, and various measurements were obtained with Northern Eclipse imaging software. A measurement of the cartilage formation was obtained by analyzing the blue area, which indicated a positive reaction with Alcian blue. A measurement of the bone formation was obtained by analyzing the black area, which indicated a positive reaction

5 IMPLICATIONS FOR ENDOCHONDRAL BONE FORMATION 1615 FIG. 1. Characteristics of NIH/3T3 and C2C12 cells. (A) Immunostaining for desmin expression by NIH/3T3 cells was negative. (B and C) We used immunostaining to monitor vimentin expression by NIH/3T3 cells stimulated with BMP4 (500 ng/ml). NIH/3T3 cells expressed vimentin on day 0 but exhibited decreased expression of vimentin over time when stimulated with BMP4. (D and E) Flow cytometry results showed that large percentages of NIH/3T3 and C2C12 cells coexpressed CD34 and Sca-1. to von Kossa staining. A measurement of the density of the human BMP4 staining was obtained by using Northern Eclipse software to analyze the level of the brown color, which indicated the positive expression of human BMP4 (i.e., a positive result after treatment with the Dab Substrate Kit for Peroxidase). A Student s t-test run with Microsoft Excel was used to analyze statistical differences between groups. RESULTS Expression of myogenic, fibroblastic, and stem cell markers (CD34 and Sca-1) in cells Immunocytochemistry revealed that <1% of the NIH/ 3T3 cells were desmin positive (Fig. 1A, day 0), whereas >99% were vimentin positive (Fig. 1B, day 0). These results confirm the fibroblastic nature of the NIH/3T3 cells. After stimulating the NIH/3T3 cells with BMP4 (500 ng/ml), we observed a gradual decrease in vimentin expression over time, with an almost total loss of vimentin expression after 9 days of stimulation (Figs. 1B and 1C). Using flow cytometry, we also studied the cells expression of the cell surface markers CD34 and Sca-1. The CD34 antigen is a transmembrane cell surface glycoprotein that is commonly used as a stem cell marker in different species, including mice. (17,18) Sca-1 is a protein that has been used to define murine hematopoietic stem cells. (19) These two stem cell markers also have been used to characterize muscle progenitor cells, such as muscle-derived stem cells (MDSCs). (12) Flow cytometry analysis revealed that 94.3% of the NIH/3T3 cells expressed Sca-1, whereas 98.34% expressed CD34. Furthermore, 93% of the NIH/3T3 cells were positive for both Sca-1 and CD34 (Fig. 1D). Of the C2C12 cells, 93.94% expressed Sca-1, 44.52% expressed CD34, and 41.92% co-expressed Sca-1 and CD34 (Fig. 1E). Both the primary fibroblast population and the primary myoblast population tested negative for CD34 and Sca-1 expression (data not shown). Differentiation of NIH/3T3 and C2C12 cells toward the chondrogenic and the osteogenic lineages after stimulation with BMP4 in vitro Alcian blue stains the acid mucopolysaccharides produced by chondrocytes. (20) We observed positive staining of NIH/3T3 cell pellets cultured for 14 or 21 days (Fig. 2A). Immunostaining for type II collagen, an indicator of chondrogenic differentiation, revealed positive results in the pellets made with NIH/3T3 cells and stimulated with BMP4 for 14 days; however, we observed a more intense signal in the NIH/3T3 pellets stimulated with BMP4 for 21 days (Fig. 2B). Assessment of C2C12 pellets at both tested timepoints revealed a lack of both Alcian blue staining and type II collagen immunostaining (Figs. 2C and 2D). The numbers of NIH/3T3 and C2C12 cells within the pellets were (2.7 ± 0.3) 10 4 and (1.02 ± 0.1) 10 5, respectively, on day 14 and (2.75 ± 0.4) 10 4 and (2.1 ± 0.1) 10 4, respectively,

6 1616 LI ET AL. FIG. 2. Chondrogenic differentiation of NIH/3T3 and C2C12 cells in pellet cultures treated with BMP4 (500 ng/ml) for 14 and 21 days. (A) NIH/3T3 cells were Alcian blue positive on days 14 and 21. (B) NIH/3T3 pellet cultures exhibited type II collagen expression on day 14, with a more intense brown signal on day 21. (C) C2C12 pellet cultures were Alcian blue negative and (D) did not express type II collagen. on day 21 (data not shown). Thus, the C2C12 cell pellets contained significantly more cells than the NIH/3T3 cell pellets on day 14 (p < 0.01), but there was no significant difference between the numbers of cells within the two types of pellets on day 21 (p 0.08). After stimulation with BMP4 at a concentration of 100 ng/ml, only 1% of the NIH/3T3 cells were ALP +. However, the percentage of ALP + cells increased to 86% when the BMP4 concentration was 1000 ng/ml. This finding indicates a dose-dependent response of NIH/3T3 cells to BMP4. However, C2C12 cells began to express ALP more rapidly than did NIH/3T3 cells (Fig. 3A). It also led us to select a BMP4 concentration of 500 ng/ml to study the time required for NIH/3T3 cells to undergo osteogenic differentiation. We observed significant differences between the responses of NIH/3T3 cells and C2C12 cells exposed for equal amounts of time to equivalent doses of BMP4 (Fig. 3B). NIH/3T3 cells stimulated with BMP4 at a concentration of 500 ng/ml for 9 days expressed ALP in vitro (Fig. 3C). RT-PCR analysis revealed that, after 6 days of stimulation, the NIH/3T3 cells began to express osteocalcin, and that this expression of osteocalcin continued over time (Fig. 3D). These results, along with the results of the cell pellet cultures, indicate that BMP4 stimulation induces NIH/3T3 cells to differentiate toward the chondrogenic and the osteogenic lineages, whereas most C2C12 cells stimulated with BMP4 under the same conditions primarily undergo a rapid osteogenic differentiation. LacZ transduction and human BMP4 bioassay In terms of LacZ transduction efficiency, 60% of the transduced NIH/3T3 cells and 64% of the transduced C2C12 cells were LacZ + (Figs. 4A and 4B). The BMP4 bioassay showed that transduced C2C12 and transduced NIH/3T3 cells secreted biologically active BMP4 at levels of 96 ± 10 and 80 ± 15 ng/10 6 cells/24 h, respectively. C2C12- L-B cells transduced with retro-dnrunx2 secreted BMP4 at a level of 100 ± 12 ng/10 6 cells/24 h (Fig. 4C). Differences between endochondral bone formation promoted by NIH/3T3 cells and that promoted by C2C12 cells LacZ/Eosin staining of all specimens at different timepoints after implantation enabled identification of the donor cells on the basis of their blue nuclei. We identified the donor cells as either osteoblasts or chondrocytes by examining morphology and co-localization with osteocalcin. These findings suggest that the implanted NIH/3T3-L-B cells differentiated into chondrocytes and osteoblasts in vivo (Figs. 4D, 4F, 4H, and 4J). In comparison, only a small fraction of C2C12-L-B cells underwent chondrogenic differentiation in vivo; the vast majority of these cells colocalized with osteocalcin, which suggests that they primarily underwent osteogenic differentiation (Figs. 4E, 4G, 4I, and 4J). Our observations of the NIH/3T3-L-B group on day 28 revealed that the implanted regions contained primarily cartilage and that very few areas contained newly formed bone (Figs. 5A and 5C). At the same time-point, the C2C12-L-B cell group contained mostly newly formed bone, which suggests that these cells initiated more rapid endochondral bone formation than did the NIH/3T3-L-B cells (Figs. 5B and 5D). Western blot analysis showed that NIH/3T3 cells and NIH/3T3-L-B cells expressed very low levels of the Runx2

7 IMPLICATIONS FOR ENDOCHONDRAL BONE FORMATION 1617 FIG. 3. ALP staining reveals that NIH/3T3 cells treated with BMP4 underwent osteogenic differentiation in a dose- and timedependent manner. (A) Percentages of ALP + cells in C2C12 and NIH/3T3 cell cultures incubated with various doses of BMP4 for 5 days. (B) Comparison of the percentages of ALP + cells in NIH/ 3T3 and C2C12 cell lines cultured in the presence of BMP4 (500 ng/ml) for various periods of time. (C) NIH/3T3 cells became ALP + after treatment with BMP4 (500 ng/ml) for 9 days. (D) We used RT-PCR to visualize the time sequence of osteocalcin gene expression in NIH/3T3 cells treated with BMP4 (500 ng/ml). A time-dependent increase in osteocalcin expression was observed in NIH/3T3 cells after BMP4 stimulation. protein 1 day after transduction of the NIH/3T3 cells with the retrovirus containing a BMP4 gene. This finding parallels the results of previous studies. (21) In contrast, C2C12 and C2C12-L-B cells expressed high levels of the Runx2 protein at the same time-point (Fig. 5E). Retroviral transduction to induce DNRunx2 expression does not alter the differences in endochondral bone formation exhibited by NIH/3T3 versus C2C12 cells Western blot analysis showed successful transduction of the C2C12-L-B cells with a retrovirus encoding DNRunx2 (Fig. 6A). Runx2 expression by NIH/3T3 cells expressing BMP4 increased with longer culture time; Runx2 expression was clearly visible 14 days after transduction with the retrovirus containing BMP4 (Fig. 6A) but was not detected 1 day after transduction (Fig. 5E). Fourteen days after transduction, we implanted these NIH/3T3-L-B cells, which also expressed a high level of Runx2, in an effort to determine the role of Runx2 in the differential endochondral bone formation by NIH/3T3 and C2C12 cells. We implanted C2C12-L-B, C2C12-LacZ-BMP4- DNRunx2 (C2C12-L-B-D), and NIH/3T3-L-B cells intramuscularly in SCID mice to study whether the expression of Runx2 would influence the induction of ectopic endochondral bone formation. We observed less bone formation in the mice that received C2C12-L-B-D cells than in the mice that received C2C12-L-B cells. However, radiographic and histological evaluation revealed that the former group still displayed rapid bone formation that was more robust than that observed in the NIH/3T3-L-B group (Figs. 6B and 6C). Human BMP4 staining results in the injected muscle revealed similar levels of human BMP4 expression in the C2C12-L-B cell group and the C2C12-L-B-D cell group at various time-points after implantation, and both C2C12 cell groups expressed higher levels of human BMP4 than observed in the NIH/3T3-L-B cell group (Fig. 6D). Histological analysis also showed less cartilage in the C2C12-L-B-D group than in the C2C12-L-B group. In comparison, the NIH/3T3-L-B group again exhibited delayed but visible cartilage formation by day 14, and the cartilaginous intermediate lasted for an additional 2 3 weeks (Figs. 7A and 7B). The amount of cartilage in the NIH/3T3-L-B group was much greater than that seen in the mice that received C2C12-L-B cells either expressing Runx2 or expressing DNRunx2 (Fig. 7B). Evaluation of bone formation by von Kossa staining showed earlier and more bone formation in the mice injected with C2C12-L-B cells than in the mice injected with C2C12-L-B-D cells or NIH/3T3-L-B cells. The NIH/3T3-L-B mice exhibited the slowest and the least amount of bone formation (Figs. 8A and 8B). These findings show that a reduction of Runx2 activity, through the use of the DNRunx2, did not compensate for the difference in endochondral bone formation induced by NIH/ 3T3 cells versus that induced by C2C12 cells. Implantation of culture supernatant from NIH/3T3-L-B and C2C12-L-B cells Results of the BMP4 bioassay showed that the supernatants from cultured NIH/3T3-L-B cells and C2C12-L-B cells contained 333 and 600 ng of bioactive BMP4, respectively. Histological results obtained after implantation of the supernatants from both groups showed similar patterns of cartilage formation. The cartilage formation began on day 7, but no cartilage was visible on day 20 or 28. We did not observe bone formation in either group (data not shown). Proliferation assay When treated with various doses of BMP4, C2C12 cells displayed significantly higher rates of proliferation than BMP4-stimulated NIH/3T3 cells (p < 0.01). However, we observed no significant difference between the cell proliferation rates of primary fibroblasts stimulated with BMP4 and primary myoblasts simulated with BMP4 (p 0.95; Fig. 9C). Ex vivo gene transfer based on primary fibroblasts and myoblasts Immunostaining for vimentin and desmin showed that <1% of the primary fibroblasts isolated from the fascia

8 1618 LI ET AL. FIG. 4. NIH/3T3 and C2C12 cells display differential chondrogenic and osteogenic potentials in vivo. (A) NIH/3T3-L-B cells and (B) C2C12-L-B cells were transduced efficiently by retro-lacz viruses. (C) Results of the BMP4 bioassay show the amount of BMP4 produced by the C2C12-L-B, C2C12-L-B-D, and NIH/3T3-L-B cells. Staining for LacZ and eosin in both the (D) NIH/3T3-L-B and the (E) C2C12-L-B groups reveals LacZ + cells in the newly formed bone (arrows). Double staining for LacZ and osteocalcin shows LacZ + osteoblasts in the (F) NIH/3T3-L-B cell implantation group and the (G) C2C12-L-B cell implantation group. LacZ + cells appear blue, and osteocalcin-positive cells appear dark brown. (H) NIH/3T3-L-B cells differentiated into LacZ + chondrocytes (arrow). (I) We observed no LacZ + chondrocytes in the C2C12-L-B cell group. (J) In vivo comparison of the ability of transduced NIH/3T3 and transduced C2C12 cell lines to differentiate toward the osteogenic or chondrogenic lineages. Osteoblasts in each cell group were identified by co-localizing the LacZ + cells with osteocalcin. We determined the percentages of LacZ + cells that became chondrocytes by counting the cells exhibiting the specific round morphology that is characteristic of chondrocytes. Data obtained from the NIH/3T3 cell group were based on samples harvested on day 28 after implantation, whereas data from the C2C12 cell group were based on samples harvested on day 14. C2C12-L-B, C2C12 cells co-transduced with retro-lacz and retro-bmp4 viruses; NIH/3T3-L-B, NIH/3T3 cells co-transduced with retro-lacz and retro-bmp4 viruses; C2C12-L-B-D, C2C12-L-B cells transduced with a retro-dnrunx2 virus. were desmin positive, whereas >99% were vimentin positive. Of the primary myoblasts, 65% were desmin positive and 99% were vimentin positive (Fig. 9A). Treatment with different doses of BMP4 revealed that, of the primary fibroblasts and primary myoblasts, the primary myoblasts showed the greater capacity to undergo osteogenesis in vitro (Fig. 9B). When co-transduced with retroviruses containing LacZ and BMP4, primary fibroblasts and primary myoblasts secreted biologically active BMP4 at levels of 60 ± 5 and 125 ± 25 ng/10 6 cells/24 h, respectively (Fig. 9D). Histological analysis of the skeletal muscle of SCID mice implanted with the two cell types revealed that primary fibroblasts produced less bone than primary myoblasts and that the pattern of cartilage formation by the two cell types was similar at all time points (Figs. 9E 9G). DISCUSSION In this study, fibroblast and myoblast cells, obtained from both cell lines and primary cultures, exhibited different responses to treatment with BMP4. First, NIH/3T3 cells differentiated toward both the chondrogenic and the osteogenic lineages, whereas most C2C12 cells differentiated toward the osteogenic lineage. Second, endochondral bone formation by the NIH/3T3 cells involved the delayed development of a cartilage intermediate that remained present for a prolonged period of time; in contrast, C2C12 cells underwent rapid differentiation to form bone. Primary fibroblasts and primary myoblasts exhibited differentiation capacity and bone formation trends similar to those observed in the cell lines. These findings indicate that the BMP4-stimulated endochondral bone formation that occurs in skeletal muscle involves different cell types, including fibroblastic cells, which contribute more to the chondrogenic phase, and myoblastic cells, which contribute more to the osteogenic phase. These results could have important implications for the development of tissue engineering applications focused on bone and cartilage repair. Runx2 is a member of the runt-domain family of transcription factors that are expressed in osteoblasts and in hypertrophic chondrocytes. (22,23) During endochondral bone formation, Runx2 promotes the differentiation of mesenchymal progenitors into osteoblasts and induces the differentiation of hypertrophic chondrocytes. Researchers have detected Runx2 expression in all osteoblastic cell lines but not in the fibroblastic NIH/3T3 line. (21) These prior findings led us to hypothesize that the differential effect of BMP4 on NIH/3T3 and C2C12 cells in vivo might be caused by their different baseline levels of Runx2 expression. We studied this hypothesis by evaluating the expression of Runx2 in the two cell lines before and after transduction with BMP4 (day 1). NIH/3T3 cells expressed low levels of Runx2 before and after transduction, whereas C2C12 cells expressed high levels of this transcription factor at the same time-points. We attempted to interfere with the function of Runx2 by transducing C2C12-L-B cells with a retrovirus

9 IMPLICATIONS FOR ENDOCHONDRAL BONE FORMATION 1619 FIG. 5. Histological analysis of tissues harvested from SCID mice 28 days after intramuscular implantation of C2C12-L-B or NIH/3T3-L-B cells. Von Kossa and eosin staining shows (A) a small amount of mineralized bone formation in the NIH/3T3-L-B cell group and (B) a large amount of mineralization in the C2C12- L-B cell group. (C) Alcian blue and eosin staining reveals extensive cartilage formation in the NIH/3T3-L-B cell group; those results contrast with the results observed in the (D) C2C12-L-B cell group. (E) Western blot for the Runx2 protein shows that Runx2 was present at a very low level in the NIH/3T3 and NIH/ 3T3-L-B cells and at high levels in the C2C12 and C2C12-L-B cells 1 day after transduction with the retrovirus containing the BMP4 gene. containing a DNRunx2 gene (psg5-dnrunx2 plasmid). We also found that NIH/3T3-L-B cells cultured for 14 days in vitro began to express a much higher level of Runx2 than observed on day 1 after transduction of the cells with BMP4 (Figs. 5E and 6A). We posited that this Runx2 expression may have resulted from continuous BMP4 stimulation and could be responsible for the cells differentiation into osteoblasts in vitro. However, the results of our radiographic and histological analyses of muscle sections implanted with the three cell types (C2C12-L-B, C2C12-L-B-D, and NIH/ 3T3-L-B) do not support the hypothesis that different levels of endogenous Runx2 expression by C2C12 cells and NIH/ 3T3 cells are responsible for the cells differing capacities to promote endochondral bone formation. C2C12-L-B-D cells still formed more bone more rapidly than did NIH/3T3-L-B cells, even though Western blot analysis (before cell implantation) showed the expression of DNRunx2 in C2C12- L-B-D cells and a high level of Runx2 expression by the NIH/3T3 cells (Figs. 6A, 8A, and 8B). In addition, the implanted NIH/3T3-L-B cells expressing Runx2 still showed a prolonged and more pronounced intermediate cartilaginous phase. These findings suggest that the process of endochondral bone formation may be cell type dependent and that Runx2 expression is not a major determinant in the process. Results from this portion of the study also revealed less production of bone and delayed endochondral bone formation in the mice receiving C2C12-L-B-D cells than in the mice receiving C2C12-L-B cells. We stained all the sections to assess human BMP4 expression and observed similar staining intensities between the two groups (Fig. 6D). Therefore, we believe that the reduction of bone formation observed in the C2C12-L-B-D cell group was caused by a deficit in the osteogenic maturation of donor C2C12-L-B cells, a deficit likely caused by the interference of DNRunx2. The cellular mechanism behind the differences reported in this study could have to do with the differential cell proliferation abilities and differentiation capacities of C2C12 and NIH/3T3 cells. In vitro, C2C12 cells treated with various doses of BMP4 grow more rapidly than NIH/3T3 cells treated with those same doses of BMP4 (Fig. 9C). The results of immunohistochemical staining of human BMP4 in muscle samples (in vivo) seem to support this in vitro finding. At different time-points after implantation of the same number of cells that expressed similar amounts of BMP4 in vitro, we observed denser human BMP4 staining in the sections obtained from the SCID mice that received C2C12- L-B cells than in those from the SCID mice that received NIH/3T3-L-B cells (Fig. 6D). Although the C2C12-L-B and NIH/3T3-L-B cell populations showed similar levels of Lac-Z transduction in vitro (Figs. 4A and 4B), we observed more LacZ + cells in sections from mice in the C2C12-L-B cell group than in those from mice in the NIH/3T3-L-B cell group (data not shown). These two observations indicate that C2C12-L-B cells have a greater proliferation capacity than NIH/3T3-L-B cells in vivo. After implantation of the same quantities of both cells in muscle pockets, C2C12-L-B cells proliferated at a higher rate and secreted more BMP4 in the local area than did NIH/3T3-L-B cells at the same time-points. According to this line of reasoning, it should take longer for the NIH/ 3T3-L-B cells to secrete the same amount of BMP4 as secreted by the C2C12-L-B cells in the local area. Based on our staining results, the concentration of human BMP4 in the targeted site of mice in the NIH/3T3-L-B group observed on day 14 approximately equaled the concentration of human BMP4 in the targeted site of mice in the C2C12- L-B group observed on day 7, and the concentration observed on day 27 in the NIH/3T3 group equaled the concentration observed on day 14 in the C2C12 group. If this explanation is correct, host stem cells in the NIH/3T3-L-B group would initiate endochondral bone formation in response to the local BMP4 stimulation later than would stem cells in the C2C12-L-B group. Such a mechanism could at least partially explain why the cartilage phase began later in the NIH/3T3-L-B cell group than in the C2C12-L-B cell group (Figs. 7A and 7B). We also hypothesized that the different differentiation

10 1620 LI ET AL. FIG. 6. Transduction of C2C12 cells with a retrovirus encoding DNRunx2 impaired endochondral bone formation. (A) Western blot analysis confirmed the expression of DNRunx2 in the C2C12 cells transduced with retro-dnrunx2. NIH/3T3 cells transduced to express BMP4 and cultured for 14 days began to express increased levels of Runx2 protein. (B) X-ray analysis revealed less bone and delayed bone formation at different time-points in the mice that received C2C12-L-B-D cells than in the mice that received C2C12-L-B cells. However, the former cells still displayed more rapid and more robust bone formation than did NIH/3T3-L-B cells. (C) Twenty days after implantation, the relative area of the newly formed bone (data based on X-ray analysis) was significantly smaller (p < 0.01) in the group that received C2C12-L-B cells transduced with the DNRunx2 retrovirus (C2C12-L-B-D) than in the group that received C2C12-L-B cells. However, the C2C12-L-B-D cell group still displayed significantly more bone formation than the NIH/3T3-L-B cell group (p < 0.01). (D) Stainings of in vivo sections showed that the C2C12-L-B cell group exhibited human BMP4 expression at levels similar to those of the C2C12-L-B-D cell group at the various time-points and that both C2C12 cell groups showed stronger expression of human BMP4 than observed in the NIH/3T3-L-B cell group. capacities of implanted donor cells play an important role in the cells promotion of endochondral bone formation. We showed that NIH/3T3 cells treated with BMP4 have a higher chondrogenic potential, whereas similarly treated C2C12 cells have a higher osteogenic potential (Figs. 2A 2D, 3A, and 3B). When implanted in vivo in the presence of equal amounts of local BMP4, C2C12 cells would be expected to differentiate directly into osteoblasts and initiate bone formation, whereas NIH/3T3 cells would be expected to differentiate into chondrocytes and make cartilage. These expected results were confirmed by the osteocalcin/ LacZ staining and human BMP4 staining of muscle sections after implantation of C2C12-L-B and NIH/3T3-L-B cells (Figs. 4D 4J, 5A 5D, and 6D). It also is feasible that NIH/3T3-L-B cells might secrete a chondrogenic growth factor that interacts with BMP4 to promote differentiation of host muscle cells into chondrocytes and osteoblasts, a process that also could explain the large amount of cartilage observed in the NIH/3T3-L-B cell group. To study this possibility, we collected, dialyzed, and lyophilized the supernatants from C2C12-L-B and NIH/ 3T3-L-B cell cultures and implanted them (with a Gelfoam matrix) in skeletal muscle pockets created in SCID mice. We posited that if the NIH/3T3-L-B cells secreted a chondrogenic growth factor that interacted with the BMP4 to promote differentiation of muscle cells into chondrocytes and osteoblasts, the NIH/3T3-L-B supernatant group should have displayed a pattern of cartilage formation similar to that displayed by the NIH/3T3-L-B cell implantation group. Interestingly, the histological results at different time-points revealed similar patterns of cartilage formation in the C2C12-L-B and NIH/3T3-L-B supernatant groups. The cartilage formation began on day 7, but no cartilage was visible on day 20 or 28. These findings show that NIH/ 3T3 cells themselves play a more important role in the cartilage formation during endochondral bone formation than does the paracrine effect of secreted growth factors. We did not observe bone formation in either group, perhaps because insufficient amounts of BMP4 were present at the site of implantation. To further validate the findings generated by the experiments involving the cell lines (C2C12 and NIH/3T3), we isolated primary fibroblasts and primary myoblasts and used them to conduct a set of experiments similar to those

11 IMPLICATIONS FOR ENDOCHONDRAL BONE FORMATION FIG. 7. Different characteristics of endochondral bone formation initiated by the various cell types were revealed by staining cartilage with Alcian blue and staining newly formed bone with eosin (red). (A) The C2C12-L-B cell group exhibited cartilage formation by day 7 and bone formation by day 14. We observed no apparent cartilage formation at early time-points in the mice that received C2C12-L-B cells transduced with the DNRunx2 retrovirus. Mice that received NIH/3T3-L-B cells exhibited cartilage formation that began on day 14, despite the fact that the donor cells expressed detectable levels of Runx2 before implantation. The cartilage formation in these mice continued until day 27 but decreased by day 35. (B) Measurement of the area of the newly formed cartilage (in pixels) revealed differences among the three groups. Compared with the C2C12-L-B cell group, the C2C12-LB-D cell group developed less cartilage. The implantation of NIH/ 3T3-L-B cells resulted in prolonged cartilage formation that resulted in a greater amount of cartilage than observed in either of the two C2C12 cell groups. performed with the C2C12 and NIH/3T3 cells. In comparison with the primary fibroblasts, the primary myoblasts differentiated toward the osteogenic lineage more readily in vitro and produced new bone more effectively in vivo (Figs. 9B, 9E, and 9F). These results are similar to those observed in the experiments involving C2C12 and NIH/3T3 cells. Interestingly, the two types of primary cells exhibited similar patterns of cartilage formation (Figs. 9E and 9G). In contrast, NIH/3T3 cells produced much more cartilage than did C2C12 cells (Fig. 7) FIG. 8. Von Kossa staining reveals differences in mineralized bone formation by the cell groups. (A) Implantation of C2C12L-B cells led to the fastest and greatest amount of bone formation. The NIH/3T3-L-B cells formed bone nearly as quickly as the C2C12-L-B-D cells. However, the NIH/3T3-L-B cells formed less bone than did the C2C12-L-B-D cells at different time-points. (B) A measurement of the area of the newly formed bone (in pixels) showed differences among the three groups. C2C12-L-B-D cells induced less bone formation than did C2C12-L-B cells. However, the C2C12-L-B-D cells still induced faster and much more bone formation than the NIH/3T3-L-B cells. We believe that this difference in performance could be attributed to the differing characteristics of NIH/3T3 cells and primary fibroblasts. Although NIH/3T3 cells are widely regarded as a fibroblast cell line, these cells expressed high levels of the stem cell markers CD34 and Sca-1; primary fibroblasts do not express these markers. As an immortalized cell line, NIH/3T3 shows unlimited proliferation potential when cultured in vitro. In contrast, primary fibroblasts likely proliferate only through a limited number of passages. When implanted in vivo, therefore, NIH/3T3 cells should exhibit unlimited proliferation and secrete more BMP4 in the targeted area. Furthermore, we found the chondrogenic potential of the NIH/3T3 cells to be superior to that of the primary fibroblasts (data not shown). Al-

12 1622 LI ET AL. FIG. 9. Results obtained from experiments based on primary fibroblasts and myoblasts. (A) Differential expression of both desmin and vimentin by primary fibroblasts and myoblasts. (B) Results of ALP staining reveal more ALP + cells in the primary myoblast group than in the primary fibroblast group after treatment of both cell groups with various doses of BMP4 for 2 days. (C) Results of the cell proliferation assay show that C2C12 cells proliferate at significantly higher rates than NIH/3T3 cells after stimulation of both groups with various doses of BMP4 (p < 0.01). There was no significant difference between the cell proliferation rates of primary fibroblasts and primary myoblasts after treatment with various doses of BMP4 (p 0.95). Overall, the proliferation potential of the primary cells was lower than that of the cell lines. (D) Results of the BMP4 bioassay show different amounts of BMP4 produced by the two groups. (E) Histological results obtained at different time-points after implantation show that PM-L-B cells induced more rapid endochondral bone formation that yielded more bone than did PF-L-B cells. (F and G) Quantitative measurement of the areas of bone and cartilage (in pixels) show that PM-L-B cells produced more bone than PF-L-B cells; there was no difference in the amount of cartilage formed by the two groups at different time-points after cell implantation. PF, primary fibroblasts; PM, primary myoblasts; PF-L-B, primary fibroblasts co-transduced with retro-lacz and retro-bmp4 viruses; PM-L-B, primary myoblasts co-transduced with retro-lacz and retro-bmp4 viruses. though NIH/3T3 cells are thought to be a fibroblast cell line, their unique characteristics appear to distinguish them from other fibroblast cell lines and primary fibroblasts. Although primary fibroblasts did not show the same pattern of endochondral bone formation mediated by NIH/3T3- L-B cells, they might yield better results when used for cartilage repair because of their limited ability to undergo osteogenesis and generate new bone. In summary, the different proliferation and differentiation potentials of NIH/3T3 cells, C2C12 cells, primary myoblasts, and primary fibroblasts after treatment with BMP4 seem to play an important role in regulating endochondral bone formation. Our results also lend additional support to a prior report indicating that cells involved in cartilage formation may be different from the cells involved in bone formation during endochondral bone formation. (24) Skeletal muscle is complex and contains many types of cells, including myoblasts, fibroblasts, endothelial cells, nerve cells, blood cells, etc. Our findings suggest that endochondral bone formation in skeletal muscle may involve different cell types, including myogenic cells (which tend to promote osteogenesis) and fibroblastic cells (which tend to promote chondrogenesis). These results may facilitate the development of improved techniques by which to use tissue engineering to promote bone and cartilage repair. ACKNOWLEDGMENTS The authors thank Marcelle Pellerin, Ying Tang, and Jing Zhou for technical assistance, Ryan Sauder for editorial assistance during manuscript preparation, Dr Toshihisa Komori for the psg5-dn Runx2 plasmid, and Dr Paul Robbins for the Tel-6 virus. This work was supported in part by NIH Grant 1 R01 DE to JH. This work was also supported by the William F. and Jean W. Donaldson Chair at the Children s Hospital of Pittsburgh and by the Henry J. Mankin Endowed Chair in Orthopaedic Surgery at the University of Pittsburgh. This study was conducted in a facility