The Bone-specific Expression of Runx2 Oscillates during the Cell Cycle to Support a G₁-related Antiproliferative Function in Osteoblasts

Abstract

The Runx2 (CBFA1/AML3/PEBP2αA) transcription factor promotes skeletal cell differentiation, but it also has a novel cell growth regulatory activity in osteoblasts. We addressed here whether Runx2 activity is functionally linked to cell cycle-related mechanisms that control normal osteoblast proliferation and differentiation. We found that the levels of Runx2 gene transcription, mRNA and protein, are each up-regulated with cessation of cell growth (i.e. G₀/G₁ transition) in preconfluent MC3T3 osteoblastic cells that do not yet express mature bone phenotypic gene expression. Cell growth regulation of Runx2 is also observed in primary calvarial osteoblasts and other osteoblastic cells with relatively normal cell growth characteristics, but not in osteosarcoma cells (e.g. SAOS-2 and ROS17/2.8). Runx2 levels are cell cycle-regulated in MC3T3 cells with respect to the G₁/S and M/G₁ transitions: expression oscillates from maximal levels during early G₁ to minimal levels during early S phase and mitosis. However, in normal or immortalized (e.g. ATDC5) chondrocytic cells, Runx2 expression is suppressed during quiescence, and Runx2 levels are not regulated during G₁ and S phase in ATDC5 cells. Antisense or small interfering RNA-mediated reduction of the low physiological levels of Runx2 in proliferating MC3T3 cells does not accelerate cell cycle progression. However, forced expression of Runx2 suppresses proliferation of MC3T3 preosteoblasts or C2C12 mesenchymal cells which have osteogenic potential. Forced elevation of Runx2 in synchronized MC3T3 cells causes a delay in G₁. We propose that Runx2 levels and function are biologically linked to a cell growth-related G₁ transition in osteoblastic cells.

Stringent control of proliferation in mesenchymal stem cells, osteoprogenitors, and preosteoblasts is essential for the normal growth and development of bone. Progression of differentiation along the osteogenic lineage requires the complex genetic and biochemical interplay of gene regulatory signaling pathways involving several key transcription factors and distinct growth factor-dependent kinase cascades (1–11). Molecular mechanisms that regulate expression of the osteoblast phenotype have been defined in considerable detail, although the principal signaling cascades that control osteoblast proliferation remain to be comprehensively characterized. Differentiation of osteoblasts is mediated in part by a central pathway that involves the binding of BMP2¹ to its cognate heteromeric type I/type II receptor (12–14), the BMP2-mediated enhancement of Runx2 gene expression (15–19), the interactions of BMP2-responsive Smad proteins with the transcription factor Runx2 (CBFA1/AML3/PEBP2αA) (20–23), and the expression of the bone-related zinc finger protein Osterix/Osx, which acts downstream from Runx2 (24–26). The essential role of Runx2 in osteoblast maturation is reflected by severe bone phenotypes resulting from genetic mutations that abrogate the normal function of Runx2 in mouse models and human disease (27–30). However, recent evidence suggests that Runx2 performs another principal function, which is related to control of osteoblast proliferation (31).

The biological activity of Runx2 must be understood within the biochemical and cellular concepts that Runx2 is a critical nuclear gene regulatory effector that supports transduction of osteogenic signals from the extracellular milieu into the nucleus. The ultimate effect of these signaling cascades is that Runx2, together with a number of coregulatory partner proteins, interacts at specific subnuclear sites to activate or repress Runx2-dependent target genes, many of which represent bone phenotype-related genes (32–35). The essential nature of Runx2 in regulating major biological events is attributable in part to its molecular role in facilitating chromatin remodeling of its cognate genes (36, 37). This biochemical function, which serves to enhance or inhibit target gene transcription depending on the integrated input of physiological signals, is directly linked to its ability to interact with specific cofactors (e.g. groucho/TLE, Smad, YAP) (21, 32, 35) and chromatin-related enzymes capable of modifying histones and/or non-histone proteins (e.g. p300, histone deacetylase 6) (36, 38). To control the potent gene regulatory activity of Runx proteins, both transcriptional and post-transcriptional mechanisms are operative to modulate their expression (39–45). One major question that remains to be resolved is how Runx2 expression and function are interrelated with its cell growth regulatory potential.

Runx2 gene expression is up-regulated during ex vivo differentiation of primary calvarial osteoblasts and in established osteoblastic cell lines (e.g. MC3T3 cells) (17, 31, 46, 47). At the onset of quiescence, a postproliferative stage in which osteoblast markers representative of the later stages of bone cell differentiation are not yet expressed, Runx2 expression is elevated above its basal level in actively dividing MC3T3 cells (31, 47). Furthermore, we have shown that Runx2-deficient calvarial cells exhibit accelerated proliferative potential and that reintroduction of Runx2 restores stringent control of proliferation (31). Thus, enhanced expression of Runx2 upon induction of the postproliferative state in osteoblasts may be functionally coupled to its intrinsic cell growth inhibitory properties and distinct from its role in promoting expression of mature bone phenotypic genes. However, to understand the growth regulatory role of Runx2, it is necessary to define more precisely how the timing of Runx2 expression is related to the onset of quiescence, progression through the cell cycle, and/or arrest at specific cell cycle stages in osteoblasts.

The main findings of this study are that Runx2 deficiency modulates expression of cell cycle regulators and that Runx2 expression is controlled with respect to the G₀/G₁, G₁/S, and M/G₁ transitions in osteoblastic MC3T3 cells and primary calvarial osteoblasts. In contrast, Runx2 levels do not respond to serum deprivation in osteosarcoma cells, whereas Runx2 protein expression is suppressed in quiescent chondrocytic cells. Elevation of Runx2 levels by forced expression significantly decreases the rate of cell proliferation by lengthening G₁ in MC3T3 cells. Thus, Runx2 performs an antiproliferative function by suppression of cell cycle progression through G₁ and supporting the transition from active cell growth to quiescence (G₀/G₁) in osteoblasts. The cell cycle-controlled down-regulation of Runx2 expression may control entry into S phase during the proliferative period of osteoblast differentiation.

EXPERIMENTAL PROCEDURES

Cell Culture and Synchronization

Experiments were performed with osteoblastic cells, which include rat calvarial osteoblasts, temperature-sensitive SV40 T antigen-transformed human osteoblasts (48), mouse MC3T3 cells (49), osteosarcoma cells (rat ROS17/2.8 and human SAOS-2 cells), chondrocytic cells (mouse ATDC5 clonal teratocarcinoma cells) (50), as well as primary human articular chondrocytes and mesenchymal progenitor cells (mouse C2C12 cells) (51). The osteoblastic cell line MC3T3-E1 was used for the majority of our experiments. MC3T3-E1 cells were maintained in αMEM supplemented with 10% fetal bovine serum (FBS). Cells were seeded in either 6-well or 100-mm plates at 0.08 × 10⁶ cells/well or 0.4 × 10⁶ cells/plate, respectively, and grown in a subconfluent state for 72 h until the onset of exponential growth. For serum reduction experiments, cells were washed three times in phosphate-buffered saline, and refed with α MEM plus 10%, 5%, 2.5%, 1% or 0% FBS. For heparin treatment experiments, cells were refed with αMEM plus 10% FBS and treated with 250–1,000 μg/ml heparin. Cells were maintained in culture for 2 days before harvesting.

Cell cycle arrest in G₀/G₁ was achieved by serum deprivation. Cells were washed three times in phosphate-buffered saline and cultured in serum-free αMEM for 1, 2, 3, or 4 days. Cell cultures were arrested in G₁/S by mimosine treatment or in early S by aphidicolin or hydroxyurea treatment or in mitosis by nocodazole treatment. Cells grown in αMEM plus 10% FBS were treated with 400 μM mimosine or with 2.5 μg/ml aphidicolin or with 2 mM hydroxyurea for 24 h or with 100 ng/ml nocodazole for 16 h.

Cells arrested in G₀/G₁ were released from quiescence by serum stimulation. Briefly, cells exponentially growing in αMEM plus 10% FBS were washed three times in phosphate-buffered saline and cultured in serum-free medium for 48 h. Then, the cells were stimulated to progress through the cell cycle by the removal of medium and the addition of αMEM plus 10% FBS. Cells arrested in mitosis (nocodazole) or at the G₁/S boundary (mimosine) were stimulated to progress, respectively, in G₁ or S phase by the addition of αMEM plus 10% FBS without drugs. After serum stimulation, cells were harvested at selected time points for Western blot analysis, Northern blot analysis and fluorescence-activated cell sorting (FACS) analysis. To inhibit cell cycle-specific degradation of proteins, the proteasome inhibitor MG132 (Calbiochem) was added to the cells. MG132 was prepared as a stock solution of 25 mM in dimethyl sulfoxide and administered to the cells after a 1:1,000 dilution into the cell culture medium (25 μM final concentration).

Flow Cytometric Analysis

The distribution of cells at specific cell cycle stages was evaluated by flow cytometry. Cells were trypsinized, washed with phosphate-buffered saline, and fixed in 70% ethanol at −20 °C overnight. Subsequently, cells were stained with propidium iodide and subjected to FACS analysis based on DNA content (52). The samples (1 × 10⁶ cells) were analyzed by cell cycle distribution using the FACStar cell sorter and Consort 30 software (Becton Dickinson). In several experiments, we performed flow cytometric analysis of MC3T3 cells stained with a fluorescein isothiocyanate-labeled annexin V antibody and propidium iodide (Calbiochem) to monitor the fraction of necrotic (propidium iodide-positive and annexin V-negative cells) or apoptotic cells (annexin V-positive).

Western Blot Analysis

Runx2 and cell cycle markers of MC3T3 cell were analyzed by Western blot analysis as described previously (52, 53). Briefly, equal amounts of total cellular protein or nuclear extracts were resolved in 10% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore Corp.). Blots were incubated with a 1:2,000 dilution of each primary antibody for 1 h. Rabbit polyclonal antibodies (p27, Cdk2, Cdc2, cyclin A, cyclin B1, cyclin E, Bcl-2, and Bax), mouse monoclonal antibodies (cyclin D1 and HA probe), and goat polyclonal antibodies (actin) were acquired commercially (Santa Cruz Biotechnology, Inc.). Mouse monoclonal antibodies specific for lamin B1 (Zymed Laboratories, Inc., San Francisco) and Runx2 (22) were also used in these studies. Membranes were then incubated with horseradish peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology, Inc.) for 1 h. Immunoreactive protein bands were visualized by a chemiluminescence detection kit (PerkinElmer Life Sciences), and signal intensities were quantitated by densitometry. Each experiment was repeated at least three times.

Northern Blot Analysis

Total RNA was isolated from MC3T3 cells by using TRIzol reagents (Invitrogen) according to the manufacturer’s specifications. Total RNA (20 μg/lane) was separated in a 1% agarose-formaldehyde gel, transferred onto Hybond Plus membranes (Amersham Biosciences), and hybridized to probes specific for Runx2, histone H4, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Hybridization was performed as described previously (52, 53) in the presence of buffer containing 50% formamide at 42 °C, and the blots were washed extensively in buffer containing 1 × SSC (standard saline citrate) and 0.1% SDS at 55 °C. Data were analyzed after overnight exposure using a Storm 840 PhosphorImager (Molecular Dynamics, Inc.). Ethidium bromide (EtBr) staining of the gels was used to assess equal loading of samples.

Nuclear Run-off Transcription Assays

Nuclear run-off transcription assays were performed as described previously (52), with some modifications. Nuclei were isolated from 1 × 10⁷ cells that were cultured in the presence (i.e. proliferating cells) or absence (i.e. quiescent cells) of serum. Nuclei were resuspended in 150 μl of glycerol storage buffer, and run-off transcription was carried out in the presence of 50 μCi of [³²P]UTP 3000 Ci/mmol, ICN) according to a standard protocol. After nuclei isolation and run-off transcription reactions, RNA was extracted with TRIzol reagent and used for hybridization with 4-μg cDNA probes slot-blotted onto Hybond N+ nylon membranes at 65 °C for 24 h.

Antisense and siRNA Oligonucleotide Treatment

MC3T3 cells were incubated for 4 h with each oligonucleotide (Oligo Etc., Inc.) at 400 nM with Lipofectin (Invitrogen), following the manufacturer’s instructions. We used an antisense oligonucleotide (5′-ACG CCA TAG TCC CTC CTT TT-3′), which is complementary to the translation-initiation region (nucleotide 343–362) of the P1 promoter-derived mRNA, which encodes the Runx2 MASNS/p57/Type II isoform. Scrambled oligonucleotide (5′-TCG TAC CTT ATC CAG TCT CC-3′) was used as a control. After the 4-h incubation with oligonucleotides and Lipofectin, the culture medium was replaced with fresh medium with or without 10% FBS and incubated for 30–42 h. Knock-down experiments were performed with siRNA duplexes specific for Runx2 (Runx2-siRNA1 = r(CUC UGC ACC AAG UCC UUU U)d(TT); Runx2-siRNA2 = r(AAA AGG ACU UGG UGC AGA G)d(TT)) or green fluorescent protein (GFP-22, rhodamine, Qiagen) (r(GCA AGC UGA CCC UGA AGU UCA U)). Transfection of siRNAs was performed with oligofectamin using conditions as recommended by the supplier (Invitrogen).

Plasmid Transfection and Adenoviral Infection

MC3T3 cells were plated in 100-mm plates at a density of 0.8 × 10⁶ cells/plate and were transiently transfected with 5 μg of pHA-Runx2 expression plasmid (33) using 10 μl of Plus reagent and 15 μl of Lipofectamine reagent (Invitrogen) following the manufacturer’s protocol. Adenovirus infection of calvarial osteoblasts was performed with a vector expressing Xpress-tagged mouse RUNX2 under control of the cytomegalovirus CMV)5 promoter (pAd/CMV5/Xpress-Runx2/IRES/GFP) or the corresponding empty vector. Virus was administered at 200 particles/cell in αMEM with 1% FBS and incubated for 1 h at 37 °C. Cells were exposed to virus by rotating dishes three times at 15-min intervals, and the virus-containing medium was removed after 1 h. Cells were washed two times with serum-free medium and then cultured in αMEM containing 10% FBS. The cell number was monitored at daily intervals to determine effects on cell growth.

RESULTS

Parallel Regulation of Runx2 mRNA and Protein Levels during the Cessation of Cell Growth

We have previously observed increased proliferation of Runx2-deficient calvarial cells (31) from mice that are homozygous for either a Runx2-null allele or a mutation that causes production of a functionally defective and C-terminally truncated protein (ΔC) (29). Consistent with the deregulation of cell growth in Runx2-deficient osteoblast progenitor cells (i.e. cells from Runx2^−/− knock-out and Runx2Δ^C/−C knock-in mice) (31), we find there are selective changes in the expression of activators and inhibitors of cyclin-dependent kinases (CDKs) based on gene expression profiling experiments with RNA from wild type and Runx2Δ^C/−C mutant calvaria (data not shown).

Apart from Runx2 genotype-dependent changes in cell proliferation (31) and the expression of several cell cycle regulators (data not shown), Runx2 itself is regulated with respect to cell growth (Fig. 1). Cessation of osteoblast growth by serum reduction or deprivation for at least 48 h correlates with increased Runx2 protein levels in MC3T3 cells (Fig. 1A). We assessed the gene regulatory levels at which Runx2 expression is controlled during cell growth inhibition. First, we analyzed the correlation between Runx2 protein and mRNA expression upon limiting the availability of serum growth factors. Cells were cultured for 2 days in either normal or reduced serum concentrations ranging from 10 to 0%. Growth inhibition of MC3T3 cells is reflected by a decrease in the level of histone H4 mRNA as serum concentrations are reduced (Fig. 1A). Western blot and Northern blot analyses reveal that the levels of Runx2 protein and mRNA increase by ~5-fold in a manner that is inversely related to serum concentrations (Fig. 1A).

Fig. 1. Runx2 protein, mRNA and transcription levels are up-regulated in osteoblastic cells upon induction of quiescence.

Runx2 protein and mRNA levels as well as Runx2 gene transcription rates were determined by, respectively, Western blot, Northern blot, and nuclear run-on analysis in MC3T3 cells subjected to serum deprivation. A, MC3T3 cells were grown in the presence of reduced serum (from 10 to 1%) or in the absence of serum (0%) for 48 h. Shown from top to bottom are the protein levels of Runx2 relative to lamin B, mRNA levels of Runx2 relative to histone H4 and GAPDH, as well as the 28 S/18 S EtBr stain. The graph at the bottom shows a quantitation of Runx2 mRNA (dark bars) and protein (light bars) relative to, respectively, GAPDH and lamin B. B and C, same as in A, but cells were maintained for 2, 3, or 4 days in the absence of serum as indicated, and Runx2 protein levels were normalized relative to actin (B). Cell cycle and apoptotic markers were monitored by Western blot analysis (C). Cell cycle arrest is clearly evidenced by reduced protein levels of CDKs and cyclins, and the absence of apoptosis is reflected by a similar ratio of Bcl-2 and Bax at all days examined. D, cells were analyzed 48 h after incubation with 10% (proliferating cells) or 0% (quiescent cells) FBS by flow cytometry using an annexin V antibody and propidium iodide staining to monitor apoptotic and necrotic cells. Our molecular analyses throughout the manuscript were performed using the adherent cell population after serum removal. The apparent absence of appreciable necrosis or apoptosis in the quiescent cells that we analyzed may be because we eliminated cells that became nonadherent during the serum deprivation procedure. E, Runx2 protein levels were also monitored by Western blot during a short term time course after serum removal (bottom), and FACS analysis (top) was used to confirm cell growth arrest. F, nuclear run-on transcription data were obtained by hybridization of radiolabeled nuclear RNA transcripts isolated from proliferating (P) and quiescent (Q) MC3T3 cells to slot-blot membranes containing cDNAs for the indicated genes. Proliferating and quiescent cells were cultured for 48 h, respectively, in the presence (i.e. P) or absence (i.e. Q) of serum. Transcription rates for the Runx2, histone H4, and GAPDH genes were expressed as a ratio of the values observed in quiescent and proliferating cells (Q/P values in the right column). Cell cycle distributions were monitored in parallel by FACS analysis for all experiments presented in this figure, and these supporting data revealed that greater than 90% of the MC3T3 cells consistently arrest in G₀ within 24 h after serum removal (see E and data not shown).

To determine the temporal relationship between Runx2 protein and mRNA expression during the induction of quiescence, we examined the levels of Runx2 at daily intervals after serum deprivation (Fig. 1, B and C). Maintenance of cells for 2–4 days in 0% serum causes cell growth arrest as reflected by decreased levels of the cell cycle markers cyclin D1 (G₁ phase), cyclin E (G₁/S phases), cyclin A (S/G₂ phases), cyclin B1 (G₂/M phase), and the proliferation-related kinases CDK1 and CDK2, as well as an increase in the level of the CDK inhibitor p27, as determined by Western blot analysis (Fig. 1C). Furthermore, there is a significant reduction in the number of S phase cells as measured by FACS analysis (data not shown, see also Fig. 1E) and as reflected by decreased expression of DNA replication-dependent histone H4 within 2 days of serum withdrawal (Fig. 1B). Serum deprivation of MC3T3 cells does not affect cell survival under our conditions, and the absence of apoptosis and/or necrosis is evidenced by a lack of cells with less than diploid content (data not shown), maintenance of a similar ratio in the levels of the apoptotic markers Bcl-2 and Bax (Fig. 1C), as well as absence of annexin V- or propidium iodide-positive cells (Fig. 1D). Because Runx2 protein and mRNA accumulation are regulated in parallel, it appears that Runx2 mRNA levels are rate-limiting for Runx2 protein expression. Furthermore, the elevation of Runx2 and p27 levels during growth arrest of osteoblasts in the absence of apoptosis suggests that Runx2 function is linked to cell cycle exit in preparation for the onset of differentiation.

Runx2 Expression Is Transcriptionally Activated at the G₀/G₁ Transition upon Induction of Quiescence in MC3T3 Cells

To examine further the temporal coupling between the quiescence-related induction of Runx2 protein and cell growth suppression, we determined the levels of Runx2 protein at multiple hourly intervals after serum deprivation (Fig. 1E). There is a clear reduction of cells in S phase as early as 4–8 h after removal of serum growth factors, and greater than 90% of the cells appear to be arrested in the G₀/G₁ phase of the cell cycle by 16–24 h. Cell growth arrest is corroborated by the reduced levels of cyclins D1, E, A, and B and increased expression of p27 (data not shown). More importantly, the time course shows that Runx2 levels are up-regulated by 16 h (Fig. 1E). Our data indicate that Runx2 protein levels do not respond acutely to removal of serum factors, but rather increase when the majority of the cells gradually exit the cell cycle in G₁ after completion of mitosis.

The observed increase in steady-state mRNA levels of Runx2 upon induction of quiescence (Fig. 1B) could be the result of increased synthesis or stability of Runx2 mRNA. To distinguish between these possibilities, we performed nuclear run-on analysis to measure RNA polymerase II-mediated in situ transcription of the Runx2 locus in intact nuclei isolated from quiescent and proliferating MC3T3 cells (Fig. 1F). Strikingly, we found that Runx2 gene transcription rates are 4-fold higher in serum-deprived quiescent cells than in proliferating cells. For comparison, we also measured transcription of the GAPDH gene, which supports general metabolism, and the proliferation-specific histone H4 gene. Our data show that transcription rates of the GAPDH gene in quiescent and proliferating cells are similar, whereas histone H4 gene transcription is 3–4-fold higher in proliferating cells as expected (see histone H4 mRNA levels in Fig. 1, A and B). Hence, the elevation of Runx2 levels in growth-arrested cells is at least in part mediated by a transcriptional mechanism.

Nuclear Localization of Elevated Levels of Runx2 in Quiescent Osteoblastic Cells

Elevation of Runx2 levels may be inconsequential for gene regulation unless Runx2 is localized to the nucleus. It is well established that Runx2 is a nuclear gene regulatory factor that is tightly associated with subnuclear domains that support transcription (54). To assess whether the postproliferative up-regulation of Runx2 alters its subcellular localization, we performed in situ immunofluorescence analysis of Runx2 in serum-supplemented and serum-deprived MC3T3 cells (data not shown). The results indicated that the large majority of Runx2 is localized within the nucleus in both biological conditions. Hence, the general nuclear localization of Runx2 is retained during the switch from proliferation to quiescence in osteoblastic cells.

Selective Up-regulation of the Osteoblast-related Runx2 Isoform during the Onset of Quiescence

Runx2 expression is regulated by two distinct promoters (P1 and P2) that produce mRNAs with different 5′-untranslated region sequences and distinct coding regions for the initial 5–20 amino acids of the protein (Fig. 2A). Although the P1 promoter is known to be active in mesenchymal progenitor cells, the P1 promoter is up-regulated during osteogenesis and may produce the majority of Runx2-related mRNAs in osteoblasts and osteoblastic cell lines (41–43). We tested directly whether the P1 promoter is responsible for expression of Runx2 in MC3T3 cells used in our studies. First, we applied real time quantitative reverse transcription-PCR analysis using isoform-specific primers to detect the main transcripts from each of the two promoters that encode distinct Runx2 proteins (P1/p57/MASNS/Type II and P2/p56/MRIPV/Type I). The data show that in untreated cells P1 mRNA is readily detected, but P2 mRNA levels are at or below the level of detection in both proliferating and differentiated cells (Fig. 2B and data not shown). We then used an antisense oligonucleotide approach that specifically targets the 5′-untranslated region of the P1-derived mRNA. We found that the P1-related antisense oligonucleotide selectively decreases total Runx2 mRNA and protein expression based on quantitative PCR data (Fig. 2B), as well as Western and Northern blot analyses (Fig. 2C). Because P2 mRNAs are virtually undetectable and inhibition of P1 mRNAs selectively decreases Runx2 protein expression, it appears that most if not all Runx2 mRNA expressed in proliferating MC3T3 cells is derived from the P1 promoter.

Fig. 2. The Runx2 P1-related mRNA, which encodes the p57/til-1/MASNS protein isoform, is specifically expressed in quiescent MC3T3 cells.

A, schematic drawing of the Runx2 gene with its two promoters (P1 and P2) that initiate mRNA transcripts (mRNA cap sites indicated by hooked arrows) with two different 5′-untranslated regions and initial N-terminal coding sequences (respectively, P1-MASNS/p57/Type II and P2-MRIPV/p56/Type I), which are alternatively spliced to common exons (gray boxes with numbers). Primers (horizontal small arrows) and fluorescent probes (small boxes) used for quantitative PCR detection of the P1 (black small arrow and box) and P2 (gray small arrow and box) mRNAs are indicated below the diagram. Sequences for the P1-related antisense (AS) and scrambled (SCR) oligonucleotides are shown in the expansion below exon 1. B, graphic depiction of quantitative real time reverse transcription-PCR analyses performed with normal cells (C) or cells treated with antisense (AS) and scrambled (SCR) oligonucleotides using primers specific for the P1 or P2 mRNA isoform as indicated. The P2 mRNA values are at or below the threshold of specific detection. C, cells incubated in the absence or presence of the antisense or scrambled oligonucleotide for 4 h and then cultured for 30 h in standard growth conditions (i.e. 10% FBS). Samples were analyzed by Western blotting (top) to determine Runx2 protein levels relative to lamin B and by Northern blotting (middle) to assess Runx2 mRNA levels relative to GAPDH. RNA integrity was evaluated by 28 S/18 S EtBr staining (bottom). D, same as in C, but oligonucleotide- or mock-treated cells were then incubated for 42 h in the presence (proliferating cells, P) or absence (quiescent cells, Q) of 10% FBS. Antisense treatment almost quantitatively reduces expression of the P1 mRNA isoform observed in quiescent cells.

We then used antisense inhibition experiments to address whether the transcriptional up-regulation of Runx2 during the induction of quiescence in osteoblastic cells is mediated by the P1 promoter. Northern and Western blot analyses show that quiescent MC3T3 cells, which were incubated for 42 h in the absence of serum, do not exhibit appreciable up-regulation of Runx2 expression in the presence of the P1-related antisense oligonucleotide at either the mRNA or protein level (Fig. 2D). We conclude that the P1-related Runx2 mRNA species is selectively up-regulated in response to serum withdrawal in MC3T3 cells.

Runx2 Is Expressed at Minimal Levels in Late G₁/S Phase and Mitosis

The increased expression of Runx2 in quiescent MC3T3 cells may be the result of cell cycle regulatory mechanisms that sense growth factor status (i.e. regulation with respect to the G₀/G₁ transition) and/or by regulatory decisions that occur during cell cycle stages beyond the restriction point, when proliferation becomes growth factor-independent (i.e. regulation with respect to the G₁/S or G₂/M phase transitions). Therefore, we treated actively proliferating MC3T3 cells for 24 h with a panel of inhibitors that arrest cell proliferation at specific cell cycle stages. Dose-response curves were established for each inhibitor to determine the optimal concentration that results in a cell cycle blockade in these cells (data not shown). Flow cytometry and Western blot analysis of cell cycle-related markers (e.g. p27 and cyclins D, E, A, and B) were used to validate arrest at distinct positions during the cell cycle (Fig. 3A). The levels of the apoptotic markers Bcl-2 and Bax were also monitored. Our results show that Runx2 protein and mRNA levels are decreased by 5–10-fold, relative to untreated and actively proliferating MC3T3 cells, when cell cycle arrest is induced in late G₁ (lovastatin or mimosine), early S phase (aphidicolin or hydroxyurea), or during G₂/M (nocodazole) (Fig. 3,B and C). These findings indicate that Runx2 protein levels are cell cycle-controlled and remain at a minimum in late G₁/S phase and mitosis in MC3T3 cells and thus are likely to be highest in mid G₁.

Fig. 3. Runx2 protein and mRNA levels are down-regulated in MC3T3 cells arrested near the G₁/S and G₂/M transitions.

Treatments that result in cell cycle blockades in late G₁ (i.e. lovastatin or mimosine), early S (i.e. aphidicolin or hydroxyurea), or mitosis (i.e. nocodazole) in each case decrease Runx2 gene expression (A–D). A, MC3T3 cells treated for 1 day with 50 μM lovastatin (Lova), 400 μM mimosine (Mimo), 2.5 μg/ml aphidicolin (APH), 1 mM hydroxyurea (HU), or 500 ng/ml nocodazole (Noco) were analyzed by FACS for cell cycle distribution (top) and by Western blot analysis for expression of cell cycle (middle) and apoptotic (bottom) markers. B, same as in A, but Western (top) and Northern (middle) blot analyses were performed to assess, respectively, Runx2 protein levels relative to actin, and Runx2 mRNA levels relative to histone H4 and GAPDH. C, graphic representation of data presented in B for Runx2 mRNA (dark bars, normalized to GAPDH) and Runx2 protein (light bars, normalized to actin). D, MC3T3 cells were synchronized by serum deprivation (48 h in medium with 0% FBS). After serum stimulation, cells were treated with nocodazole 24 h after release (when cells have progressed into G₂) in the presence (N+MG132) or absence (N) of the proteasome inhibitor MG132, followed by mitotic shake-off 12 h later (i.e. 36 h after release). Levels of Runx2 protein were measured by Western blotting relative to β-actin and a panel of cell cycle markers. Runx2 levels remain low in mitotic cells irrespective of the presence of MG132. Levels of cyclins A and D1 are clearly elevated, indicating that proteasomal degradation is blocked by MG132.

Northern blot data show that Runx2 mRNA levels parallel the protein levels during G₁ or S phase arrest, but not mitotic arrest (Fig. 3, B and C). In nocodazole-treated cells, Runx2 mRNA levels are elevated, but protein levels are low. The difference between Runx2 mRNA and protein accumulation in mitosis is also observed during cell cycle progression in synchronized cells (see Fig. 5). The discordance between these two parameters suggests that Runx2 gene expression at mitosis is regulated at the level of protein stability or translation. To address this question, we treated cells synchronized in mitosis using nocodazole with the proteasomal inhibitor MG132 (Fig. 3D). The data show that MG132 fails to elevate Runx2 levels during mitosis or the levels of other proliferation markers (e.g. CDK1 and CDK2), whereas other cell cycle markers (e.g. cyclins D1 and A) are stabilized in the presence of MG132 (Fig. 3D). This result indicates that the ubiquitin/proteasome pathway does not actively degrade Runx2 in mitotically blocked cells. It appears that Runx2 protein levels are either translationally controlled or regulated by a proteasome-independent post-translational mechanism. Although the mechanisms by which Runx2 mRNA and protein levels are controlled during mitosis remain to be investigated further, it is clear that Runx2 mRNA availability is not rate-limiting at the G₂/M transition (Fig. 3, B and C).

Fig. 5. Runx2 expression is cell cycle-regulated and suppressed in late G₁ in mouse osteoblastic cells.

Runx2 gene expression was assessed during progression through the osteoblast cell cycle to determine the specific transition stages when Runx2 levels are modulated. MC3T3 cells were synchronized by cell growth arrest through serum starvation. Cell growth was then stimulated by the addition of serum, and cells were harvested after 0, 6, 9, 12, 15, 18, and 24 h. A, progression through successive cell cycle phases was monitored by flow cytometry (top) and Western blot analysis (bottom). B, cell cycle-dependent modulations in Runx2 protein and mRNA levels were assessed as described in Fig. 2. C, cells were treated 6 h after serum stimulation with the proteasome inhibitor MG132 (25 μM) and harvested 12 or 15 h after induction of cell growth. Expression of Runx2 and cell cycle markers upon MG132 treatment was evaluated by Western and Northern blot analyses as indicated (see also Fig. 1). Cell cycle progression was assessed by FACS analysis. D, graphic presentation of cell cycle-related changes in Runx2 protein and mRNA levels after serum stimulation in the absence (closed circles and squares) or presence (open circles and squares) of MG132. Levels of H4 mRNA (open triangles and dotted line) and cyclin E (open diamonds and dashed line) are indicated to demarcate progression through S phase. The graph shows that Runx2 levels are down-regulated in late G₁ phase (i.e. 12 h after serum stimulation) and that MG132 treatment inhibits this decline in Runx2 expression. Runx2 protein levels are up-regulated again when osteoblasts progress into the next cell cycle.

The panel of five cell cycle inhibitors (i.e. lovastatin, mimosine, aphidicolin, hydroxyurea, and nocodazole) blocks proliferation in either late G₁ or subsequent cell cycle stages concomitant with down-regulation of Runx2 expression (see Fig. 3, A, B, and C). The combined results indicate that Runx2 is normally at maximal levels at a stage during early to mid G₁. Interestingly, treatment of MC3T3 cells with different concentrations of heparin (which is known to bind osteoblast-related growth factors and to impact on cell surface-related signaling mechanisms) causes a dramatic cell cycle arrest in G₁ concomitant with up-regulation of Runx2 protein (data not shown). This G₁ arrest mimics the induction of quiescence (i.e. G₀) after serum withdrawal as evidenced by FACS analysis and cell growth curves (data not shown). Thus, Runx2 levels are modulated at a specific transition period in the cell cycle close to or coinciding with a G₁ transition when cells monitor growth factor requirements.

Cell Growth Regulation of Runx2 Expression Is Cell Type-dependent

We tested whether the cell growth and cell cycle stage-dependent regulation of Runx2 is observed in cell types other than immortalized mouse MC3T3 osteoblastic cells. Serum deprivation experiments with primary rat calvarial osteoblasts and immortalized human osteoblasts derived from endochondral bone also exhibit elevation of Runx2 levels after induction of quiescence (Fig. 4A). Furthermore, direct comparison of primary rat calvarial osteoblasts and their immortalized counterparts stably transfected with telomerase (i.e. mTERT) yielded identical results (i.e. up-regulation of Runx2 expression in G₀) (data not shown). This finding indicates that cellular immortalization does not influence cell growth control of Runx2 gene expression in osteoblasts. Strikingly, rat ROS17/2.8 osteosarcoma cells, which express mature bone phenotypic markers (e.g. osteocalcin), and human SAOS-2 osteosarcoma cells, which do not express osteocalcin (i.e. resembling immature osteoblasts), do not modulate Runx2 levels. Thus, loss of cell growth control in osteosarcoma cells abrogates serum-dependent regulation of Runx2 expression. Furthermore, mouse ATDC5 chondrocytic progenitor cells, as well as primary human articular chondrocytes (Fig. 4A), respond to serum withdrawal by down-regulating rather than elevating Runx2 levels. Consistent with these data, chondrocytes maintain high levels of Runx2 in late G₁ (mimosine block) and early S (hydroxyurea block) unlike osteoblastic cells (rat calvarial osteoblasts, human osteoblasts, and MC3T3) (Fig. 4B; see also Fig. 3). Taken together, these data indicate that enhancement of Runx2 during the onset of quiescence is related to normal progression of osteoblast growth and differentiation, and this regulatory mechanism is deregulated in osteosarcomas. Furthermore, differences in Runx2 regulation in cells from osteoblastic or chondrogenic lineages clearly indicates that cell growth control of Runx2 levels is highly cell type-dependent.

Fig. 4. Cell type-dependent regulation of Runx2 during growth arrest.

Different cell types were subjected to cell growth arrest by serum deprivation (0% FBS) or cell cycle inhibitors (i.e. mimosine (Mimo), hydroxyurea (HU), and nocodazole (Noco)) and analyzed by Western blot similar to the experiments described in Figs. 1 and 3. Untreated control cells were allowed to proliferate under normal culture conditions (medium supplemented with 10% FBS). Experiments were performed with osteoblastic cells (primary rat calvarial osteoblasts (ROB), immortalized human osteoblasts transformed with temperature-sensitive SV40 T antigen (HOB)), osteosarcoma cells (rat ROS 17/2.8 and human SAOS-2 cells), as well as chondrocytic cells (primary articular chondrocytes from human cartilage (HAC) and immortalized mouse ATDC5 cells).

Runx2 Levels Are Cell Cycle-regulated

The findings obtained by serum deprivation (Figs. 1 and 2) and cell cycle inhibitor experiments (Fig. 3) are consistent with a temporal oscillation of Runx2 gene expression during the MC3T3 cell cycle. Runx2 protein levels may be transiently up-regulated at the beginning of G₁ and then down-regulated prior to the G₁/S phase transition, provided that growth factors are present. Therefore, we examined directly Runx2 gene expression during cell cycle progression in synchronized cells at multiple hourly time points after stimulation of proliferation in serum-deprived cells (Fig. 5).

The results indicate that during serum stimulation, MC3T3 cells enter the cell cycle and have progressed beyond the G₀/G₁ transition within 6 h, which is reflected by a decrease in p27 levels (Fig. 5A). Late G₁ occurs around the 9 h time point reflected by down-regulation of cyclin D1 and the up-regulation of cyclin E by 12 h (Fig. 5A). Progression beyond the G₁/S phase boundary is evident by 12–15 h when expression of DNA replication-dependent histone H4 gene expression is induced (Fig. 5B), and the percentage of cells in S phase is increased markedly (Fig. 5A). By 24 h, cells have progressed into G₂ and subsequent stages (M and G₁) based on FACS analysis, which shows an increased representation of cells in G₂/M and G₁ (Fig. 5A) as well as a decline in cyclin E levels, elevation of cyclin A and CDK1 (Fig. 5A), and a decline in H4 gene expression (Fig. 5B). We find that Runx2 protein is decreased between 9 and 12 h (Fig. 5A), whereas Runx2 mRNA declines between 6 and 9 h after stimulation of cell growth (Fig. 5B). Thus, Runx2 gene expression is down-regulated shortly after cell cycle entry, whereas Runx2 protein is reduced just prior to S phase.

Runx2 mRNA levels begin to increase again after 18–24 h (Fig. 5B) when the majority of cells are in the late stages of S phase (63.6%) (Fig. 5C) and when H4 gene expression is maximal (Fig. 5B). The elevation in Runx2 protein lags behind the accumulation of Runx2 mRNA and occurs between 18 and 24 h at a cell cycle interval between G₂/M and the subsequent progression into G₁ after mitosis (Fig. 5, A and B). The demonstration that Runx2 levels oscillate from maximal levels during early G₁ and minimal levels during early S phase unequivocally establishes that Runx2 gene expression is cell cycle-regulated.

Runx2 Expression Is Modulated in Relation to a Ubiquitin/Proteasome-dependent Cell Cycle Transition in G₁ and Up-regulated after Completion of Mitosis in MC3T3 Cells

One major mechanism for rapidly modulating protein levels of cell growth regulators at specific cell cycle stages involves the ubiquitination-dependent degradation of proteins through proteasomal pathways (55–57). In late G₁ the selective and periodic destabilization of CDK inhibitors by proteasomal degradation is required for progression beyond the G₁/S transition. We used MG132, a membrane-permeable peptidyl-aldehyde that inactivates the 20 S proteasome, to study the influence of proteasomal activity on cell cycle progression of MC3T3 osteoblastic cells. Treatment of synchronized and serum-stimulated MC3T3 cells after cell cycle entry with MG132 causes a G₁ arrest (Fig. 5C). This G₁ arrest is evidenced by the percentage of cells in G₁ (92.3%) versus cells in S phase (3.9%) in MG132-treated MC3T3 cells relative to untreated cells at 15 h after serum stimulation (i.e. G₁ cells = 68.9% and S phase cells = 31.1%) (Fig. 5C). Furthermore, histone H4 gene expression is not up-regulated at 15 h in serum-stimulated cells treated with MG132 (Fig. 5C). The G₁-specific cell cycle blockade is also evident from Western blot analysis, which reveals that cyclin E is not up-regulated at 12 or 15 h after serum stimulation in the presence of MG132 (i.e. 6 or 9 h after MG132 treatment), as normally observed in control cells (Fig. 5, A and C). As expected, cyclin A still remains at or below the level of detection. Taken together, the results indicate that MG132 mediates a cell cycle arrest in serum-stimulated MC3T3 cells which occurs at or before a specific stage in G₁ (presumably the restriction point) which precedes the cell cycle stages that are blocked in cells treated with lovastatin, mimosine, aphidicolin, or hydroxyurea. During this MG132-mediated cell cycle blockade in G₁, Runx2 protein levels are elevated in parallel with p27 levels, whereas the levels of β-actin are comparable at all time points examined (Fig. 5, C and D; see also Fig. 5A). More importantly, because MG132-treated cells exhibit elevated levels of Runx2 mRNA (Fig. 5, C and D), mRNA availability can by itself account for modulations in Runx2 expression during G₁.

To maintain their osteoblastic identity, MC3T3 cells must re-express Runx2 protein after the initial down-regulation of Runx2 in late G₁. Indeed, our time course analysis of synchronized and quiescent MC3T3 cells reveals that Runx2 protein is elevated again by 24–36 h after serum stimulation, coincident with progression through mitosis into the next G₁ (Fig. 6A). We note that the absolute but not the relative timing of this Runx2 up-regulatory event is somewhat variable from experiment to experiment, presumably because of differences in cell density, passage number, and/or serum batches that are known to affect MC3T3 cell proliferation (e.g. compare Figs. 5B and 6A). Treatment of cells at 27 h after serum stimulation with the mitotic inhibitor nocodazole prevents the increase in Runx2 protein levels but causes as expected a mitotic block based on elevated cyclin B1 levels (Fig. 6A). We conclude that the cell cycle-dependent increase in Runx2 protein levels requires progression beyond mitosis.

Fig. 6. Runx2 expression is enhanced in early G₁ in MC3T3 cells.

A, quiescent MC3T3 cells were serum stimulated. Upon progression beyond the G₁/S phase transition (i.e. 24 h after growth induction) cells were incubated in the absence or presence of nocodazole to prevent progression beyond mitosis. Levels of Runx2 protein and the indicated cell cycle markers were analyzed as described in Fig. 1 by Western blotting (right panels). The fraction of cells in G₂/M or S phase as determined by flow cytometry is depicted graphically (left panels). B, MC3T3 cells were synchronized by incubation for 16 h with nocodazole to generate a mitotic block. Mitotic cells were harvested by gentle agitation and then released from mitotic arrest by the addition of culture medium without nocodazole. Expression of Runx2 and cell cycle-related proteins was determined by Western blot analysis. Runx2 levels are elevated coincident with the disappearance of the mitotic marker cyclin B1.

To assess whether Runx2 expression is increased immediately after mitosis, we analyzed Runx2 protein levels upon release from a nocodazole-imposed mitotic block (Fig. 6B). We find that Runx2 is acutely up-regulated (within 1.5 h) together with p27, whereas other markers are down-regulated (e.g. cyclin B and CDK1) or remain relatively constant (e.g. CDK2, CDK4) (Fig. 6B). Taken together, our results suggest that Runx2 regulation is cyclical and is initiated with a transient induction immediately at or after the M/G₁ phase transition that persists until a stage in late G₁ when Runx2 is destabilized as cells commit to the initiation of S phase.

Forced Expression of Runx2 Impedes Proliferation of MC3T3 Osteoblastic Cells

We next assessed whether deliberate perturbations in Runx2 levels can influence cell cycle progression. We first performed experiments with the P1-specific antisense oligonucleotide that selectively eliminates expression of the p57/MASNS/Type II isoform (data not shown; see also Fig. 2). Down-regulation of the P1-related isoform of Runx2 affects neither cell growth nor the cell cycle distribution of proliferating MC3T3 cells that were treated with antisense oligonucleotide at 2-day intervals for 1 week (data not shown). We also tested a regimen that involved successive treatments of MC3T3 cells for 4 h with the P1-related antisense oligonucleotide, for 48 h in the absence of serum, and for up to 30 h in the presence of serum. Quiescent MC3T3 cells exhibit the expected anti-sense-mediated reduction of Runx2 levels, but they do not have overt deficiencies in the potential to enter the cell cycle after serum stimulation (data not shown). In addition, treatment of asynchronous MC3T3 cells with siRNA for 48 or 72 h does not affect cell growth rates as measured by cell counts or cell cycle distribution as determined by flow cytometry. Taken together, forced reduction of Runx2 levels, which are normally maintained at low levels during cell cycle progression through the S, G₂, and M phases, does not accelerate cell growth or cell cycle entry from quiescence in MC3T3 cells.

To address whether elevating Runx2 levels above normal physiological levels in proliferating cells can influence cell growth, we transfected MC3T3 cells with a CMV-driven expression vector that produces an epitope-tagged version of the P1-related isoform of Runx2. The expression of exogenous Runx2 is initially increased and subsequently decreased, presumably because of the loss (or silencing) of the transiently transfected construct. Two days after treatment we observed robust expression of the exogenous protein above the normal endogenous levels and a delay in cell growth (Fig. 7, A and B). Assessment of the cell growth rate, which represents the slope of the growth curve, shows that cell proliferation is only affected on day 2 when total expression of Runx2 is maximal, but not on days 4 and 6 when total Runx2 levels are comparable with control cells (Fig. 7B and data not shown). We ruled out that the difference in the net growth rate of Runx2 overexpressing cells is not the result of necrosis or apoptosis, because annexin V or propidium iodide staining is negligible in transfected cells (data not shown). Thus, forced elevation of Runx2 in MC3T3 preosteoblastic cells is inhibitory for cell cycle progression, consistent with the postulated function of Runx2 as a cell growth suppressor of osteoblasts.

Fig. 7. Runx2 influences cell growth in osteoblastic MC3T3 cells and C2C12 mesenchymal progenitor cells.

A and B, transient expression of Runx2 above the normal minimal levels present in actively proliferating MC3T3 cells delays cell growth. Untransfected MC3T3 cells (labeled C) or cells transfected with 5 μg of CMV-HA-Runx2 vector (labeled RV) or CMV-empty vector (labeled EV), were cultured for 6 days in completed medium (i.e. containing 10% FBS). Cells were harvested on the indicated days for analysis of protein levels (A) and cell growth (B). Western blot analysis was used to determine exogenous Runx2 levels using the HA epitope tag and total Runx2 protein using a Runx2 antibody that recognizes both exogenous and endogenous proteins (A). Actin was used as an internal control. Cell growth was assessed by counting the number of cells at different times during the culturing period (i.e. days 2, 4, and 6) (B). C and D, C2C12 cells were mock infected (labeled C) or infected with adenovirus that either expresses β-galactosidase or Runx2 on day 0. The exogenous expression of Runx2 was monitored during the subsequent days (days 1 and 2) using an Xpress antibody, and endogenous β-actin was detected as an internal control (C). As expected, endogenous levels of Runx2 are below the level of detection in actively proliferating C2C12 cells (data not shown); proliferating C2C12 cells exhibit a premyoblastic phenotype unless treated with BMP2. The cell growth curve of transfected and untransfected cells is based on cell numbers determined on the indicated days (D).

To assess whether Runx2 can suppress cell growth in other mesenchymal cell types, we used an adenoviral vector to express Runx2 in mouse C2C12 cells that have osteoblastic potential (i.e. in the presence of BMP2). At 1 and 2 days after infection, exogenous Runx2 levels were clearly detectable (Fig. 7C). More importantly, Runx2-expressing C2C12 cells exhibit very slow cell growth (Fig. 7D). Thus, Runx2 is capable of regulating cell growth in different mesenchymal cell types with either a preosteoblastic phenotype (i.e. MC3T3 cells) or osteoblastic potential (i.e. C2C12 cells).

To determine how Runx2 functions to delay cell growth at specific cell cycle stages, we examined directly the effect of Runx2 overexpression on cell cycle progression using synchronized cells. MC3T3 cells were transfected with CMV-Runx2 vector and subsequently synchronized in G₁ by mimosine treatment (Fig. 8). As predicted from our previous data (Figs. 3B, 5B, and 6A), control cells exhibit minimal Runx2 protein levels after a mimosine-imposed late G₁ block (Fig. 8A), and progress into S phase (between 6 and 12 h) and the G₂/M phases (between 12 and 18 h) after serum stimulation (Fig. 8, B and C). However, MC3T3 cells in which total Runx2 levels are constitutively elevated by exogenous expression of Runx2 do not progress into S phase by 12 h. This result indicates that high levels of Runx2 affect cell cycle progression prior to the G₁/S phase transition. Between 12 and 18 h a subset of the Runx2-expressing cells enters into G₂/M coinciding with a reduction in the G₀/G₁/S population (Fig. 8C) and the overall expression levels of both exogenous and total Runx2 (Fig. 8A). These data indicate that the Runx2-imposed G₁ cell cycle block is reversible and that cells can bypass this block upon reduction of Runx2 levels (Fig. 8A). The reduction in total Runx2 protein levels is most likely the result of loss of the transfected expression vector and the cell cycle-dependent down-regulation of Runx2 as cells progress from G₁ into S (Figs. 5 and 6). We conclude that maintaining high levels of exogenous Runx2 in MC3T3 cells, which counteracts the normal down-regulation of endogenous Runx2 levels in G₁, suppresses cell growth by delaying cell cycle transit in G₁.

Fig. 8. Forced elevation of Runx2 in late G₁ impedes normal cell cycle progression.

Runx2 was expressed above normal physiological levels of MC3T3 cells synchronized in G₁ resulting in a delay in S phase entry. MC3T3 cells were transiently transfected with 5 μg of CMV-HA-Runx2 vector (R) or CMV-empty vector (Ctrl), cultured for 12 h and arrested in G₁ later by mimosine treatment for 24 h. Upon removal of the inhibitor, cells were released into S phase and harvested after 0, 6, 12, and 18 h for analysis of protein levels (A) and cell cycle markers (B and C). A, Western blot analysis was used to determine exogenous Runx2 levels using the HA epitope tag and total Runx2 protein using a Runx2 antibody that recognizes both exogenous and endogenous proteins. Actin was used as an internal control. B, progression through successive cell cycle phases was monitored by flow cytometry. C, the fraction of cells in G₁/S or G₂/M phase as determined by flow cytometry (see B) is depicted graphically.

DISCUSSION

The key finding of the current study is that the bone-specific expression of the Runx2 gene is tightly regulated during the cell cycle. Our data show that the levels of both Runx2 protein and mRNA are strikingly up-regulated at the onset of quiescence in immature MC3T3 osteoblastic cells in response to serum deprivation. Moreover, Runx2 levels in these cells are elevated in early G₁ and destabilized in late G₁. The down-regulation of Runx2 in G₁ appears to be linked to the restriction point, which integrates growth factor signaling and competence for progression beyond the G₁/S phase transition. Elevation of Runx2 protein levels by forced expression in MC3T3 cells decelerates cell growth and delays progression into S phase. Thus, our data suggest that Runx2 expression is cell cycle-regulated with respect to G₁ to attenuate proliferation of osteoblasts. The findings presented here are consistent with the concept that Runx2 promotes bone cell phenotype development at least in part by regulating genes that are required for progression through and exit from the cell cycle.

We note that although the levels of Runx2 protein are dramatically down-regulated prior to the onset of S phase, immunofluorescence microscopy has revealed at the single cell level that a small amount of Runx2 persists throughout the cell cycle in MC3T3 cells (58). This constitutive minor fraction of Runx2 is tightly associated with both interphase and metaphase chromosomes and may support maintenance of lineage identity during mitotic progression into the next G₁ (58). We propose that the cell cycle-regulated levels of Runx2 during G₁ that we have analyzed in this study may reflect its antiproliferative function.

The enhancement of Runx2 expression in quiescent MC3T3 cells could potentially coincide with the initial induction of Runx2-dependent mature bone differentiation-specific markers. However, we have shown previously that induction of quiescence by serum deprivation in a subconfluent population of nonproliferative MC3T3 cells does not induce expression of late osteoblastic markers (e.g. osteocalcin) (31). Hence, the MC3T3 cell culture model permits experimental distinctions between the normal proliferative stage of MC3T3 cells (days 1–7 in culture) versus a nonproliferative, undifferentiated state of serum-deprived cells that have been subject to short term culture (less than 4 days) and a postproliferative/confluent differentiated state reflected by presence of mineralized extracellular matrix, which typically forms only after several weeks in culture (59).

Cell Type-specific Regulation of Runx2 Levels

Our findings on cell growth control of Runx2 in MC3T3 cells are corroborated by similar findings in multiple osteoblastic cell types, including primary rat calvarial osteoblasts and immortalized human osteoblasts derived from endochondral bone. Although cellular immortalization does not affect cell growth control of Runx2 levels, we find that Runx2 levels are not regulated with respect to serum deprivation and the G₁/S phase transition in osteosarcoma cells (i.e. rat ROS17/2.8 and human SAOS-2 cells). Thus, cell cycle regulation of Runx2 expression is abrogated in osteosarcoma cells that exhibit loss of stringent cell growth control. Furthermore, our findings show that chondrocytic progenitor cells (i.e. ATDC5) and primary human articular chondrocytes respond oppositely to serum deprivation by suppressing Runx2 expression during quiescence. In further contrast, chondrocytic cells also maintain high levels of Runx2 during late G₁ and S, whereas Runx2 levels are minimal after progression beyond the G₁/S transition in osteoblastic cells. Thus, cell growth regulation of Runx2 expression is highly cell type-dependent and may reflect differences in the requirement for Runx2 in mesenchymal cell growth control during skeletal tissue development in vivo.

Tissue-specific Control of Cell Growth by Runx-related Transcription Factors

Our previous data have demonstrated that Runx2-deficient calvarial cells exhibit increased proliferation and provided the initial indication that Runx2 functions to regulate osteoblast growth (31). Similarly, genetic defects or inactivation of the Runx1 and Runx3 genes are related to cell growth abnormalities. Runx1 is a frequent target of chromosomal translocations in acute myelogenous leukemia (60, 61), and Runx3 is associated with gastrointestinal hyperplasias and stomach cancer (62, 63). Thus, it has now become apparent that the entire class of Runx-related transcription factors, which encompasses Runx1/AML1, Runx2/Cbfa1 and Runx3/PEBP2αC, regulates cell growth in a tissue- and cell type-restricted manner (23, 31, 60, 63–67).

Our results show that Runx2 levels are inversely linked to cell growth and S/G₂/M phase progression in osteoblastic cells, but positively correlated with cell proliferation in chondrocytic cells. Interestingly, recent findings by the Friedman laboratory have revealed that Runx1 has a cell growth stimulatory function in 32Dc13 hematopoietic progenitor cells and regulates cyclin D3 levels (68). AML1/Runx1 increases during G₁ to S cell cycle progression independent of cytokine-dependent phosphorylation and induces cyclin D3 gene expression (68). Consistent with a growth stimulatory function, Runx1 protein but not mRNA levels are 2–4-fold enhanced in S/G₂/M phases of the cell cycle. These results are comparable with findings we obtained for Runx2 in chondrocytic cells, but distinct from the regulation of Runx2 in osteoblastic cells. Nevertheless, one common theme that emerges for Runx1 and Runx2 is that both proteins are cell cycle-regulated to support their cell growth suppressive or stimulatory functions.

Transcriptional and Post-transcriptional Regulation of Runx2 Activity during Principal Cell Cycle Transitions

Transcriptional control of Runx2 gene expression during the cell cycle is evident from the nuclear run-on transcription and antisense inhibition data presented in this paper. Removal of serum factors or treatment with heparin induces quiescence in MC3T3 cells and enhances Runx2 gene expression by increasing transcription of the upstream P1 promoter, which generates the p57/MASNS/til-1 isoform of Runx2. Our findings suggest that the P1 promoter is repressed by default in proliferating cells through the action of serum components (e.g. growth factors, steroid hormones, or other ligands) that regulate mitogenic signaling pathways. It is conceivable that serum deprivation may negate the repression of P1 transcription by passive inactivation of this mitogenic signal. Interestingly, a distal region within the P1 promoter encompasses a negative element, which attenuates transcriptional activity (42, 45). This negative element contains an AP1 site, and the binding activity of the cognate AP1 proteins (i.e. Fra2 and JunD) is enhanced by growth factors and steroid hormones (42, 45, 69). Cell cycle inhibitor experiments reveal that Runx2 levels are elevated in G₁ phase and selectively down-regulated prior to the G₁/S phase transition. The minimal levels of Runx2 which persist in later cell cycle stages (e.g. G₂/M) do not appear to be sensitive to the proteasome inhibitor MG132. Rather, our results indicate that maintenance of minimal Runx2 levels beyond the G₁/S phase transition may occur by translation control.

Control of Cell Growth and Differentiation by Lineage-specific Transcription Factors

Immature progenitor cells proliferate during progression of differentiation, and expression of mature phenotype markers normally requires cell cycle arrest. Our data indicate that elevation of Runx2 is sufficient to delay cell cycle progression, but other transcription factors may be required to force cell cycle exit and/or promote osteoblast maturation. For example, members of the C/EBP, Msx, and Dlx classes of transcription factors together with Runx2 are likely candidates to coregulate osteoblast cell growth and differentiation (2). Similarly, the intricate interplay of Runx1, c-Myb, and C/EBP proteins regulates proliferation at both early and late stages of hematopoietic differentiation (65).

The function of Runx2 in the osteogenic lineage may be analogous to the role of the muscle-related regulatory factor MyoD during myogenesis. MyoD expression is also enhanced upon serum deprivation of C2C12 mesenchymal cells prior to the onset of differentiation (70–73). Cell cycle regulation of MyoD involves the CDK2/cyclin E-dependent degradation by the ubiquitin/proteasome pathway in late G₁ phase, and MyoD is stabilized by direct association with p57 (70, 71). Similar to Runx2, MyoD may influence cell cycle progression by transcriptionally regulating the gene for the CDK inhibitor p21. However, MyoD may also inhibit cell growth by influencing the interactions of cyclin A/CDK2 with E2F4, which maintains suppression of E2F-dependent genes (74), or by abrogating the CDK4-dependent phosphorylation of the tumor suppressor pRB (75). Because Runx2 is known to interact directly with pRB (76) and controls the expression of the p21 gene in osteoblastic cells (38), the possibility arises that Runx2 may also execute its antiproliferative functions by directly modulating the growth factor-dependent cyclin/CDK/pRB/E2F cascade, which regulates cell cycle progression in G₁. Consistent with this concept, Thomas and colleagues (77) have shown that Runx2 regulates the p27/KIP1 gene during cell cycle exit and osteoblast differentiation. Additional studies will be required to define further the precise cell growth regulatory pathways that respond to Runx2 in osteoblasts.