Loss of Runx2 in Committed Osteoblasts Impairs Postnatal Skeletogenesis

Abstract

The Runx2 transcription factor is critical for commitment to the osteoblast lineage. However, its role in committed osteoblasts and its functions during postnatal skeletogenesis remain unclear. We established a Runx2-floxed line with insertion of loxP sites around exon 8 of the Runx2 gene. Runx2 protein lacking the region encoded by exon 8 is imported into the nucleus and binds target DNA, but exhibits diminished transcriptional activity. We specifically deleted the Runx2 gene in committed osteoblasts using 2.3kb col1a-Cre transgenic mice. Surprisingly, the homozygous Runx2 mutant mice were born alive. The Runx2 heterozygous and homozygous null were grossly indistinguishable from wild-type littermates at birth. Runx2 deficiency did not alter proliferative capacity of osteoblasts during embryonic development (E18). Chondrocyte differentiation and cartilage growth in mutants was similar to wild-type mice from birth to 3 months of age. Analysis of the embryonic skeleton revealed poor calcification in homozygous mutants, which was more evident in bones formed by intramembranous ossification. Runx2 mutants showed progressive retardation in postnatal growth and exhibited significantly low bone mass by 1 month of age. Decreased bone formation was associated with decreased gene expression of osteoblast markers and impaired collagen assembly in the extracellular matrix. Consequently, Runx2 mutant bones exhibited decreased stiffness and structural integrity. By 3 months of age, bone acquisition in mutant mice was roughly half that of wild-type littermates. In addition to impaired osteoblast function, mutant mice showed markedly decreased osteoclast number and postnatal bone resorption. Taken together, functional deficiency of Runx2 in osteoblasts does not result in failed embryonic skeletogenesis, but disrupts postnatal bone formation.

Introduction

Runt-related transcription factor-2 (Runx2) is a key regulator of osteoblast differentiation and bone development^(1–3). Overexpression of Runx2 in non-osseous mesenchymal cells promotes their commitment to the osteoblast lineage⁽³⁾. Mice with global deletion of Runx2 exhibit a complete lack of osteoblasts, failed ossification and early lethality^(3–6). The zinc finger transcription factor, Sp7, is also essential for osteoblast differentiation and bone formation^(7,8). Sp7 gene deletion results in failed development of mineralized tissue^(7,8). However, Sp7 is a downstream target of Runx2 and Sp7 mRNA is absent in Runx2 deficient cells^(7,9). Thus, Runx2 remains the master transcription factor necessary and sufficient for mesenchymal cell commitment to the osteoblast lineage.

Several lines of evidence suggest that physiologic levels of Runx2 promote osteoblast function after lineage commitment. For example, in both humans and rodents, expression of endogenous Runx2 progressively increases from committed to mature osteoblasts^(10,11). Additionally, Runx2 is well documented to induce the expression of stage-specific osteoblast marker genes^(1,2,12). These include markers of committed (ALP, Col1a1), differentiating (BSP, OPN) and mature osteoblasts (OC, MMP13), as well as osteocyte markers (DMP1, SOST). In addition to ECM proteins, Runx2 is required for expression of several signaling molecules produced by osteoblasts at various stages of maturation, such as RANKL, OPG, DKK1, Wnt10, TGFβ1 and BMP4^(2,13,14). Finally, Runx2 expression increases with progressive osteogenesis during embryonic development⁽³⁾. Taken together, these studies strongly support a positive role for endogenous Runx2 protein during osteoblast differentiation and skeletal development. However, it remains unknown whether Runx2 function is required after commitment to the osteoblast lineage and/or during postnatal bone development.

To understand the cell specific function of Runx2 in vivo, a number of transgenic models that overexpress both full-length and dominant-negative forms of Runx2 have been analyzed^(15–18). The truncated dominant-negative form of Runx2 (dn-Runx2) primarily consists of the N-terminal DNA-binding Runt homology domain and lacks 298 amino acids at the C-terminus. The dn-Runx2 (Δ230) protein lacks transcriptional activity, but has a higher binding affinity for target DNA⁽¹⁹⁾. Therefore, it is anticipated that in these models, dn-Runx2 blocks the activity of the endogenous Runx2 protein. Transgenic mice overexpressing dn-Runx2 in committed osteoblasts by the 2.3kb α1(I)-collagen promoter were born alive with a normal skeleton⁽¹⁵⁾. Interestingly, by 4 weeks of age, α1(I)-collagen transgenic mice showed ~25% fewer osteoblasts and a decrease in bone formation rate⁽¹⁵⁾. However, expression of dn-Runx2 in mature osteoblasts using the Osteocalcin promoter resulted in a ~75% decreased bone formation rate at 3 weeks of age without affecting osteoblast number⁽¹⁶⁾. Thus, overexpression of dn-Runx2 in either committed or mature osteoblasts results in distinct skeletal phenotypes and suggests that Runx2 exerts stage-specific functions during skeletogenesis.

In sharp contrast, overexpression of full-length Runx2 in committed osteoblasts resulted in osteopenia during early postnatal skeletogenesis^(17–18). The low bone mass observed in these transgenic models led to the current notion that Runx2 is a negative regulator of bone synthesis and osteoblast differentiation. It is important to note that despite similar skeletal phenotypes, osteoblast function was differentially altered by overexpression of the full-length Runx2 protein^(17–18). For example, in one model, osteopenia was due to a 50–80% decrease in the rate of bone formation⁽¹⁷⁾. Bone in these mice contained fewer functional osteoblasts, suggesting that Runx2 inhibits osteoblast maturation. A separate transgenic model exhibited a ~50% increase in bone formation rate in the periosteal surfaces at 4 months of age⁽¹⁸⁾. While the number of mature osteoblasts was not affected, osteoclast number and activity were increased, suggesting that osteopenia in this transgenic model was not due to poor osteoblast function, but rather increased bone resorption⁽¹⁸⁾. The opposing phenotypes in overexpression models cannot be explained by stage-specific functions of Runx2, since these models overexpress the full-length Runx2 protein by the same 2.3kb α 1(I)-collagen promoter^(17–18). Thus, due to contradictory phenotypes, it remains unclear if after commitment to the osteoblast lineage Runx2 is a positive or negative regulator of bone synthesis.

To circumvent issues associated with the lack of spatiotemporal regulation and non-physiological levels in overexpression models, we established a Runx2-floxed model. The goal of this study was to determine if cells that are already committed to the osteoblast lineage require Runx2 for osteoblast function and postnatal bone acquisition. We employed the 2.3kb α1(I)-collagen-Cre line to conditionally ablate Runx2 from committed osteoblasts⁽²⁰⁾. Our results demonstrate that Runx2 exerts a positive function after commitment to the osteoblast lineage and that Runx2 deficiency in osteoblasts disrupts postnatal bone synthesis and skeletal growth.

Materials and Methods

Generation of Runx2 conditional knockout mice

Generation of Runx2-floxed (Runx2^F/F) mice containing directional loxP sites around exon 8 is described elsewhere^(21,22). To delete Runx2 specifically in osteoblasts, Runx2^F/F mice were crossed with 2.3kb α1(I)-collagen-Cre mice⁽²⁰⁾. The resulting Runx2^+/ΔE8 mice were intercrossed to obtain Runx2^ΔE8/ΔE8, Runx2^+/ΔE8, and Runx2^+/+ mice. All genotypes were determined using DNA from tail biopsy with Direct PCR lysis reagent (Viagen biotech; Los Angelos, CA). The Runx2 wild type and floxed alleles and the Cre transgene were detected by PCR using the specific primers listed in the Table S1. We maintained all mice under 12hr light:dark cycle with ad libitum access to regular chow and water. All animal experiments were performed with the approval from the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham and conformed to relevant federal and state guidelines and regulations.

Assessment of osteoblast specific activity of Cre recombinase

Tissue and cell-specific expression of the 2.3kb α1(I)-collagen-Cre transgene was confirmed with Td-Tomato reporter mice⁽²³⁾. Littermates with and without the Cre transgene were harvested at birth and whole embryos were fixed for 30min in ice-cold 4% paraformaldehyde (Sigma-Aldrich; St. Louis, MO). Skeletal elements were then dissected and embedded in O.C.T. compound embedding medium (Tissue-Tek; Torrence, CA.). Tissues were sectioned at 7µm thickness by cryostat and slides were stained with 1µg/mL DAPI for 5min. Slides were rinsed in PBS prior to mounting with ProLong Gold anti-fade medium (Life Technologies; Grand Island, NY). Sections were analyzed under brightfield and fluorescence light with Tetramethyl Rhodamine Iso-Thiocyanate (TRITC) and DAPI filters by a Nikon Eclipse 80i microscope.

Skeletal preparation and staining

For whole mount skeletal staining, littermates of all genotypes were collected at birth. Mice were de-skinned, eviscerated, and their muscles were partially removed. Pups were fixed for 2 days in 95% ethanol (EtOH) and then placed in acetone for 2 days to remove fat tissues. Skeletons were then placed in Alizarin red and Alcian blue staining solution 0.015% Alcian blue (Sigma-Aldrich), 0.005% Alizarin red (MP Biomedicals; Solon, OH) and 5% Acetic acid in 70% ethanol for 3 days with gentle shaking. The soft tissues from the stained skeleton were cleared with 1% KOH for 1–2 days. Further clearing was carried out with a 1:4 ratio of 1% KOH and sequential concentrations of glycerol (20%, 50%, and 80%) for 12hrs each. Skeletal preparations were stored in 100% glycerol until imaging. Whole skeleton images were captured with Nikon Cool PIX S10 digital camera. Limbs, spine and skull were dissected and imaged with a Nikon dissecting microscope. Level of calcification for individual bones was determined as the mean saturation of Alizarin red staining by NIS-Elements software. Regions of interest included the diaphysis of the limbs, vertebral bodies of the lumbar spine, and frontal, parietal and occipital skull bones. Calculations for each skeletal element were performed in triplicate.

RNA isolation and real time PCR analysis

Total RNA was extracted from hindlimbs of 1-month old mice using Trizol reagent (Life Technologies). Total RNA was isolated from the diaphysis for analysis of osteoblast and osteoclast marker genes, and from the epiphysis for analysis of chondroblast marker genes. The cDNA was prepared from 2µg of total RNA using cDNA synthesis kit (BioRad; Hercules, CA). Quantitative real time PCR was performed using iQ SYBR Green Supermix (BioRad) with specific primer pairs detailed in Table S1. Gene expression was normalized to beta-actin expression levels, and relative gene expression was quantified using the ΔΔCT method.

Histochemistry, BrdU and double calcein labeling

Mouse skulls were harvested at birth and 1 month, and limbs at birth, 1 and 3 months of age. For standard histological analysis, 1 and 3 month old limbs were fixed and decalcified in 10% EDTA for 2 weeks. Tissues were dehydrated in increasing concentrations of EtOH and cleared with xylene prior to embedding. For cell proliferation assay, pregnant females at E18 were injected with 0.1mg BrdU per g of body weight 3hrs prior to embryo harvest. To evaluate bone formation and mineral apposition, 3-month old male mice were injected intraperitoneally with 20mg of calcein per kg of body weight (Sigma-Aldrich) in a 2% sodium bicarbonate solution. All tissues were sectioned at 5µm thickness and detailed histochemistry procedures are described in supplemental methods.

Histomorphometry and µ-CT analysis

For histomorphometric analysis, 5µm lateral sections of undecalcified tissues from the central region of the tibia and femur of 3-month old mice were processed as noted above and stained with Goldner’s trichrome. Analysis was performed using Bioquant Osteo semiatuomated system for skeletal phenotyping on lateral sections midway through the tibia and femur. Nomenclature, symbols, and units used are those recommended by the Nomenclature Committee of the American Society for Bone and Mineral Research⁽²⁴⁾. Bone structure and mineral density were analyzed in 1-month old mice by µCT analysis. Detailed description for both analyses is in supplemental methods.

3-Point Bend Test

The bones were tested to failure by three-point bending to measure flexural strength and stiffness of the bones. Mechanical testing was performed using a 858 MiniBionix Materials Testing System (MTS Systems, Eden Prairie, MN, USA) and a 100N load cell. A custom made three-point bending fixture was used with a span width of 5mm between the lower contacts. Cross head speed of 0.05mm/sec was set to break the femur. Measurements were made using force-vs-displacement curves including: ultimate force, Pmax [N] = height of the curve, and stiffness, k [N/mm] = slope of linear portion of curve.

Western blot analysis

Nuclear extracts were isolated from whole limbs of wild type and Runx2 mutant mice. Tissues were flash frozen and pulverized with a dounce homogenizer in ice-cold PBS and 1× complete protease inhibitor (Roche Applied Sciences; Indianapolis, IN) and 25µM of MG132 on ice. Nuclear pellets were lysed in ice-cold NP-40 lysis buffer (10mM Tris, 3mM MgCl2, 10mM NaCl, 1% NP-40, protease inhibitor, 25µM MG132). Equal amount of protein samples were separated on 10% SDS-polyacrylamide gels. Blots were probed with monoclonal Runx2 antibody (MBL International, Woburn, MA) at a dilution of 1:1000, washed four times with PBST and then incubated with anti-mouse IRDye-700 labeled secondary antibody for 1hr. Blot was scanned with odyssey imaging system. Stripped blots were then reprobed with goat polyclonal Lamin-B1 antibody (Santa cruz, Sandiago, CA), used as a loading control.

Cell culture, transfection and Luciferase assays

HEK293T cells were cultured and maintained in DMEM medium supplemented with 10% fetal bovine serum, 1% penicillin G (100 units/ml), and streptomycin (100µg/ml). Cells were maintained at 37°C in a humidified atmosphere of 5% CO₂ in air. Transient transfections were performed using PolyJet™ transfection reagent (SigmaGen Laboratories Rockville, MD). For promoter reporter assays, cells plated in 12-well dishes were co-transfected with 100ng and 200ng of WT and Δ369 Runx2 expression plasmid, and 25ng of either Osteocalcin or Sclerostin promoter reporter plasmids. Transfection was carried as per manufacturer’s protocol. Luciferase activity was determined using a Luciferase Reporter Assay kit (Promega; Madison, WI). Data were obtained from at least three independent experiments with four replicates each (n=12).

In situ immunofluorescence

Rat osteoblastic (ROS17/2.8) cells were plated on gelatin-coated coverslips in 6-well dishes and fed with F12 medium with 5% FBS. Cells were transfected with 2µg of flag-epitoped WT and Δ369 Runx2 expression plasmid. Cells were processed for in situ immunofluorescence analysis 20hrs post transfection. Cells were fixed with whole cell fixative (3.7% formaldehyde in PBS) for 10min on ice, washed with PBS and incubated in permeabilizing solution (0.25% Triton X-100 in filtered PBS) for 10min on ice. Washed cells were immuno-stained with 1:250 dilution of flag or Runx2 antibody for 1hr at 37°C in a humidified chamber. Cells were washed four times with ice-cold PBSA and stained with Alexa-488 secondary antibody for 1hr at 37°C. Cells were washed four times with PBSA, then stained with DAPI for 5min on ice and washed twice with 0.1% Triton X-100 in PBSA and one time with PBS. Cover slips were immediately mounted in ProLong Gold anti-fade medium, sealed and stored at −20°C. Digital Z-series images were acquired with Nikon Eclipse 80i inverted fluorescence microscope at 60× magnifications.

Results

Runx2 deficiency in committed osteoblasts impairs embryonic bone synthesis

Global or chondrocyte-specific deletion of Runx2 results in failed ossification and perinatal lethality^{(3–6,21,22,)}. We have previously established that the exon 8 encoded region of the Runx2 protein is essential for its biological function^(21,22,25). To understand the regulatory requirement of Runx2 in committed osteoblasts during skeletogenesis, we deleted exon 8 of the Runx2 gene in cells expressing the 2.3kb α1(I)-collagen-Cre transgene (Fig. 1A, S1A). Western blot analysis confirmed the generation of the mutant Runx2 protein (Δ369), which lacks the region encoded by exon 8, in homozygous mice (Fig. 1B). The presence of full-length Runx2 protein in mutant hindlimbs either reflects residual Runx2 protein prior to deletion in osteoblasts, or Runx2 protein expressed in skeletal cells not targeted by Cre recombinase in the entire limb. It is important to note that the N-terminal Runt Homology Domain and NLS are intact in the Runx2 mutant protein (Fig. 1A). Thus, wild-type and mutant Δ369 Runx2 proteins showed similar patterns of distribution, as both were excluded from nucleoli of the osteoblasts (Fig. 1C). The Osteocalcin and Sclerostin promoter fragments that contain Runx regulatory motifs were used to assess the transcriptional activity of the Δ369 protein (Fig. 1D). Wild-type Runx2 induced both promoters in a dose-dependent manner (Fig. 1D). However, transactivation of both promoters by mutant Δ369 Runx2 was less than half of the wild-type protein. To confirm osteoblast specificity of 2.3kb α1(I)-collagen-Cre transgene, we utilized Td-Tomato reporter mice (Fig. S1). The intact hindlimbs from newborn control and Cre positive littermates were viewed under TRITC fluorescence (Fig. S1B). Fluorescent signal was absent in cartilaginous and non-skeletal tissues. Cre activity was restricted to osteoblasts in the cortical and trabecular bone within the diaphysis (Fig. S1C, D). Importantly, we did not detect Cre activity in chondrocytes (Fig. S1C). Together, these data demonstrate that recombination of the Runx2-floxed allele by 2.3kb α1(I)-collagen-Cre generates a mutant Runx2 protein specifically in osteoblasts, which lacks transcriptional activity of the full-length Runx2 protein.

Figure 1. Recombination of Runx2-floxed allele in osteoblasts generates a mutant protein that lacks transcriptional activity.

(A) Schematic illustration of Runx2-floxed allele and Runx2 recombined allele (Runx2 ΔE8). Eight exons derived from P1 promoter are indicated; black region represents exons that encode Runt Homology Domain. Light grey boxes represent untranslated regions. Black arrows represent loxP sites. (B) Nuclear extracts were isolated from hindlimbs of 1-month old Runx2^+/+ and Runx2^ΔE8/ΔE8 mice. Blots were probed with Runx2 antibody, stripped, and re-probed for Lamin B antigen, used as an internal control. Full length Runx2 (WT) and mutant Δ369 proteins are indicated. (C) Rat osteoblastic (ROS17/2.8) cells were transfected with WT and Δ369 Runx2 expression plasmids. Cells were processed for in situ immunofluorescence and imaged at 60× magnification. Representative images of nuclei stained with Runx2 antibody and DAPI are shown. Inset shows respective phase images. (D) Schematic illustration of −0.2kb OC and −2.0kb SOST promoters. Gray lines indicate position of Runx binding motifs in each promoter. HEK293T cells were co-transfected with OC or SOST promoter luciferase plasmid and indicated Runx2 expression vectors. Luciferase activity was determined 24 hours later and pooled data from 3 independent experiments with 4 replicates each is shown in graph.

Homozygous mutants (Runx2^ΔE8/ΔE8) were born alive with comparable size and weight to wild-type littermates (Fig. 2). Skeletons from newborn wild-type, heterozygous and homozygous mice from 4 independent litters were double-stained (Fig. S1A, 2A). Alizarin red and Alcian blue signal showed normal skeletal patterning and similar length of the ossified regions in all the long bones (Fig. 2A–C). Consistent with this observation, the number of chondrocytes in the resting, proliferating and hypertrophic zones were similar in wild-type and Runx2^ΔE8/ΔE8 mice (Fig. S2A, B). Despite no change in the length of bones, the degree of calcification was decreased in the femur (36%), tibia (25%), digits (11%), and vertebral bodies (41%) of Runx2^ΔE8/ΔE8 compared with wild-type mice (Fig. 2A–C). Similarly, forelimbs in mutant mice showed a decrease (25–45%) in calcification from scapula to radius and ulna (Fig. 2A). Masson’s trichrome staining indicated less collagen in the trabecular bone of homozygous mutants (Fig. 2D). To assess if decreased ossification is linked with low proliferative ability and number of osteoblasts, BrdU assay was performed at E18 (Fig. 2E). The number of BrdU positive cells lining the trabecular and cortical bone surfaces was similar in both wild-type and homozygous littermates. Taken together, these data suggest that Runx2 deficiency in osteoblasts does not alter their proliferation, but appears to affect osteoblast function.

Figure 2. Runx2 function in committed osteoblasts is not essential for embryonic skeletogenesis.

(A) Skeletons from Runx2^+/+, Runx2^+/ΔE8 and Runx2^ΔE8/ΔE8 newborn mice were double-stained with Alizarin red and Alcian blue. A representative picture of skeletons stained and imaged together is shown. (B) Hindlimbs were disjointed and skeletal elements were imaged with a Nikon dissecting microscope. The length of Alizarin red stained region between proximal and distal epiphyseal cartilages in each skeletal element was measured using NIS-Elements software. Graph represents pooled value from 4 independent litters. (C) Femur, lumbar spine and feet were dissected from WT and homozygous littermates at birth and imaged simultaneously. The degree of calcification was quantified by Alizarin red signal using NIS-Elements software. (D) Hindlimbs from newborn WT and homozygous littermates were sectioned laterally and processed together for Masson’s trichrome staining. Representative images of femurs captured at 10× and 40× (boxed region) magnification are shown. Collagen stained in blue showed a decreased signal in Runx2^ΔE8/ΔE8 mutants. (E) Pregnant mothers at E18 were injected with BrdU, sacrificed 3 hours later and processed for immunohistochemistry. BrdU and DAPI stained femurs from hypertrophic zone to mid-diaphyseal region are shown. BrdU positive osteoblasts were counted from 6 randomly selected 100µm² regions containing both cortical and trabecular bone. Pooled data from 3 independent litters with SD is shown. HC; hypertrophic chondrocytes, TB; trabecular bone, BM; bone marrow. Scale bars are 100 µm.

We also analyzed skeletal elements formed by intramembranous ossification in Runx2^ΔE8/ΔE8 mice (Fig. 3). Compared to wild-type, fontanels were wide open in mutants (Fig. 3A). Parietal and occipital bone tissues were poorly developed, indicating impaired cranial development. We quantified Alizarin red staining and detected a significant decrease in calcification of frontal (22%), parietal (26%) and occipital (43%) bones in Runx2^ΔE8/ΔE8 mice (Fig. 3A). Histological analysis on the frontal bones revealed spicules of bone tissue lined with osteoid in wild-type mice. In contrast, frontal bones in Runx2 mutants were thin and consisted of osteoid layers that did not develop into bone (Fig. 3B). Measurement along the coronal plane showed a dramatic decrease in thickness of frontal bones in Runx2^ΔE8/ΔE8 mice (Fig. 3C). Together, these data show that osteoblast-specific deficiency of Runx2 impairs embryonic skeletal development.

Figure 3. Intramembranous ossification is impaired in Runx2 deficient mice.

(A) Dorsal view of Alizarin red/Alcian blue stained skulls from newborn littermates is shown. Dotted lines mark the borders of calcification in occipital bones and open fontanels at the sagittal plane. (B) Coronal sections from rostral part of newborn calvaria were stained with H&E. Frontal bones are shown at 10× magnification. Double arrows indicate thickness of the frontal bone in the boxed region. (C) Calvarial thickness was determined at 6 randomly selected areas along the frontal bone. Pooled data from 3 litters is shown in graph. Statistical significance was calculated by Student’s t test. Asterisks represent p < 0.05. Scale bars are 100 µm.

Runx2 function in osteoblasts is required for postnatal growth and skeletogenesis

The unexpected survival of mutant mice provided the opportunity to elucidate the role of Runx2 in committed osteoblasts during postnatal skeletal development. After birth, rodents undergo rapid growth in size and weight. We monitored postnatal growth of sex-matched littermates through 3 months of age (Fig. 4). Runx2 mutants weighed 20–25% less than wild-type littermates at 1 month of age (Fig. 4A). Growth failure was observed in both male and female mutants, which were noticeably smaller than wild-type mice (Fig. 4A, S2C). Growth plate chondrocytes are central for postnatal bone enlargement and skeletal growth. Histological analysis revealed no alterations in various zones of the growth plate at 1 and 3 months of age (Fig. S3A, B). Moreover, expression levels of Sox9, PTHrP and IHH in mutants were similar to wild-type at 1 month of age (Fig. S3C). These findings indicate that the smaller body mass did not result from impaired chondrogenesis. Consistent with this observation, the length of postnatal limbs was similar between wild-type and Runx2^ΔE8/ΔE8 mice at 1 month of age (Fig. 4B). Despite normal limb length, mutant mice exhibited poor osteoid deposition immediately beneath the growth plate, indicating impaired osteoblast function (Fig. S4A). At 1 month of age, the orientation of collagen fibers, noted by polarized light, reflected that of mature bone matrix in wild-type mice (Fig. 4C). In sharp contrast, collagen fibers were sparse and lacked orientation in postnatal bones of Runx2 mutant mice. We next determined the structural integrity of Runx2 mutant bones at 1 month of age by 3-point bending test (Fig. 4D). Compared to wild-type littermates, both male and female Runx2^ΔE8/ΔE8 mice exhibited a significant decrease in bone stiffness and flexural strength. Together, these data suggest that Runx2 deficiency in osteoblasts impairs the structural and functional integrity of postnatal bones.

Figure 4. Runx2 deficiency in committed osteoblasts disrupts postnatal growth and structural integrity of bones.

(A) Body weights of wild-type and mutant littermates of both genders were monitored weekly from birth to 3 months of age. Graph represents average weight of 4 pairs of male mice. (B) Femurs from 1-month old wild-type and homozygous littermates were scanned for 3D µ-CT imaging. Longitudinal section through center of femur is shown. Scale bar; 1.0mm. (C) Femur of 1-month old littermates were embedded in plastic, sectioned along the frontal plane and stained with H&E. Images were captured immediately beneath the growth plate with bright and polarized light to assess orientation of collagen in bone. Magnification 20×. Scale bars are 200 µm. (D) 3-point bend test was performed on femurs of 1-month old wild-type and homozygous littermates. The line graph shows data from 2 wild-type and 2 homozygous mice tested in the same group. Bar graphs show the average of 3 wild-type and 3 homozygous mice. Runx2 mutant mice exhibited a decrease in stiffness as well as reduced ability to withstand mechanical loading. Asterisks represent p < 0.05.

We next assessed postnatal skeletogenesis by µ-CT analysis. Runx2^ΔE8/ΔE8 femur from 1-month old mice showed a striking decrease in bone width (Fig. 4B, 5A). Quantification of cortical bone parameters showed a significant decrease in BV (39%), TV (30%) and BV/TV ratio (38%) in mutant mice (Fig. 5A). Runx2 deficiency in osteoblasts also resulted in a significant decrease in trabecular BV (42%) and BV/TV (55%) (Fig. 5B). Moreover, mutant bones exhibited a significant decrease in trabecular number (40%), thickness (40%), and a concomitant increase in trabecular space (65%). This data demonstrates that Runx2 function in osteoblasts is required for appositional growth and development of the postnatal skeleton.

Figure 5. Runx2 deletion results in impaired osteoblast function and low postnatal bone mass.

(A) Representative µ-CT images of 1-month old femurs taken immediately beneath the growth plate are shown in cross-section (top). The 3D reconstructions of cortical bone at the mid-diaphyis (bottom) and (B) trabecular bone beneath the growth plate are shown. Pooled data from 4 independent litters were utilized for analyses of indicated bone parameters for cortical bone and trabecular bone. BV; bone volume, TV; total volume, Tb.N; trabecular number, Tb.Th; trabecular thickness and Tb.Sp; trabecular space. Statistical significance was calculated by Student’s t test. Asterisks represent p < 0.05. Scale bar; 100µm. (C) Total RNA was isolated from the diaphyseal region of hindlimbs of 1 month old mice. The CT values from 6 replicates were normalized to β-actin and relative mRNA levels are shown for indicated genes. Nuclear extracts were obtained from 1-month old whole limbs for western blot analysis. Blots were probed with rabbit polyclonal Sp7 antibody. Lamin B is shown as loading control.

The disruption in cortical and trabecular bone prompted us to examine the status of osteoblast function in homozygous mice. The expression level of early and late osteoblast markers was determined in hindlimbs of 1-month old mice. Interestingly, mRNA levels of early marker, ALP, were similar between wild-type and homozygous mice (Fig. 5C). However, Runx2^ΔE8/ΔE8 mice exhibited a 50–60% decrease in expression of differentiated and mature osteoblast markers, BSP and OC. These results are consistent with reduced transcriptional activation of OC promoter by the mutant protein (Fig. 1D). We also observed a robust decrease in expression of Sp7, a downstream target transcription factor of Runx2, at both the mRNA and protein level (Fig. 5C). Taken together, these data demonstrate that postnatal bone synthesis requires functional Runx2 protein in osteoblasts.

Deletion of Runx2 in osteoblasts disrupts postnatal bone acquisition and remodeling

Runx2 deficient mice continued to show low body mass, weighing 30% less than wild-type littermates by 3 months of age (Fig. 4A). At this time, wild-type mice achieve skeletal maturity and show osteoclast-mediated skeletal remodeling. To assess whether Runx2 is involved in osteoblast and osteoclast function during skeletal remodeling, we performed histomorphometric analysis on femurs from 3-month old mice (Fig. 6). We noted a dramatic decrease in the thickness of secreted osteoid matrix on bone surfaces of adult mutant mice (Fig. 6A). Moreover, Von kossa staining showed an overall decrease in the mineralized matrix beneath the growth plate (Fig. S4B). Runx2 deficiency in committed osteoblasts led to a 53% decrease in BV/TV (Fig. 6B). Similarly, thickness of trabeculae was decreased by 47%, resulting in an increase in trabecular space (Fig. 6B and data not shown). Interestingly, the number of osteoblasts was comparable, but the osteoblast surface to bone surface ratio was decreased by 24% in Runx2^ΔE8/ΔE8 mice. No change in the osteoblast number is consistent with comparable BrdU labeling at birth (Fig. 2E). To assess if the decrease in osteoblast size reflects a loss of function in mutant mice, double-calcein labeling was performed (Fig. 6C, D). Calcein incorporation was apparent on 55% of the bone surface in wild-type mice (Fig. 6D). However, only 34% of the bone surface in homozygous skeleton was labeled. Moreover, calcein signal was noticeably weak in both cortical and trabecular bone surfaces of mutant mice (Fig. 6C). Analysis of the double-labeled bone surfaces revealed a significant decrease (38%) in the mineral apposition rate, resulting in a larger decrease (60%) in bone formation rate within Runx2ΔE8/ΔE8 femurs. Thus, osteoblast function depends upon the integrity of the Runx2 protein.

Figure 6. Runx2 function in osteoblasts is required for postnatal bone acquisition.

(A) Undecalcified hindlimbs from 3 month old littermates were embedded in plastic, sectioned laterally and stained with Goldner’s trichrome. Representative images of the endosteal surfaces of cortical bone are shown at 40× magnification. (B) Histomorphometric analysis was performed on 3 independent stained sections and pooled data for bone parameters is shown in graphs. N.Ob/BS; number of osteoblasts per bone surface, Ob.S./BS; osteoblast surface per bone surface. (C) Male littermates at 3 months of age received 2 intraperitoneal injections of calcein 7 days apart. Undecalcified hindlimbs were embedded in plastic and sectioned laterally. Images show areas of double-labeled bone surfaces in WT and Runx2 mutant femurs. Representative phase and FITC fluorescent images taken at the same exposure time are shown. Boxed region of endosteal surfaces is shown at 40× magnification. (D) Dynamic histomorphometry was performed on double-calcein labeled bones from 3 independent litters. Pooled data from 3 independent WT and homozygous femurs show mineralizing parameters in graphs. MS/BS; mineralizing surface to bone surface, BFR; bone formation rate, MAR; mineral apposition rate. Scale bars are 200 µm.

We next assessed osteoclast activity to clarify if the decrease in postnatal bone mass in homozygous mice was also due to increased bone resorption (Fig. 7). We observed that the number of TRAP positive cells was markedly decreased in both cortical and trabecular bone of Runx2^ΔE8/ΔE8 mice at 1 month of age (Fig. 7A). We also noted a significant decrease (50%) in RANKL and a modest increase (27%) in levels of OPG mRNA (Fig. 7B). The differential expression increased the OPG/RANKL ratio nearly 2-fold, which would likely inhibit osteoclastogenesis. Osteoclast activity is an integral component of bone remodeling in the mature skeleton. In 3-month old homozygous mice, we detected a 65% decrease in osteoclast number and a 53% decrease in osteoclast surface to bone surface (Fig. 7C). Similarly, erosion surface was decreased by 51%, while an increase of 33% in quiescent surface was noted in mutant bones. Thus, osteoclast development requires Runx2 activity in post-committed osteoblasts during postnatal skeletogenesis.

Figure 7. Mice with osteoblast-specific deficiency of Runx2 exhibit low osteoclast activity.

(A) Frontal sections of 1-month old WT and homozygous femurs were stained simultaneously for TRAP activity. Images of metaphyseal region captured with the same exposure time are shown at 4× and 10× (boxed region) magnifications. Scale bars are 200 µm. (B) Total RNA was isolated from the diaphyseal region of hindlimbs of 1-month old mice. The CT values from two independent WT and homozygous mice with 4 replicates each were normalized to β-actin. Relative mRNA levels are shown for RANKL and OPG. (C) Undecalcified hindlimbs from 3 month old littermates were embedded in plastic, sectioned laterally and stained with Goldner’s trichrome. Histomorphometric analyses of osteoclast and bone resorption parameters were measured from 3 independent stained sections and pooled data for bone parameters is shown in graphs. N.Oc/BS; number of osteoclasts per bone surface, Oc.S/BS; osteoclast surface per bone surface, ES/BS; erosion surface per bone surface, QS/BS; quiescent surface to bone surface. Asterisks represent p < 0.05.

In summary, these data demonstrate that postnatal skeletogenesis and remodeling depends upon the continuous positive regulatory role of Runx2 in committed osteoblasts.

Discussion

In this study, we investigated the in vivo requirement of Runx2 specifically in osteoblasts. Even with the absence of Runx2 in osteoblasts, embryonic skeletogenesis proceeded, and survival of mutant mice was not affected. Despite normal development, impaired ossification was observed in the newborn skeleton. Runx2 deficiency in committed osteoblasts resulted in a progressive disruption of postnatal growth and skeletogenesis. Marked reduction in bone mass was due to poor osteoblast function, noted by decreased gene expression and bone synthesis. Alteration in OPG/RANKL expression by osteoblasts was coupled with impaired osteoclast activity in Runx2 mutant mice. In adult mice, lack of Runx2 in osteoblasts led to poor bone mineral apposition and bone formation rate. Our results provide the first evidence for the physiologic requirement of Runx2 in committed osteoblasts during early postnatal bone acquisition and remodeling.

Osteoblasts and chondrocytes are obligatory cell types for synthesis of the extracellular matrix in bone. Global knockout of Runx2 indicates its essential requirement for functional maturation of both cell types^(3–6). Specific deletion of Runx2 in chondrocytes results in failed endochondral ossification and perinatal lethality due to respiratory failure^(21,22,26). However, mice with osteoblast-specific Runx2 deficiency are viable with grossly normal development at birth^(26,27). These results suggest that Runx2 exerts different regulatory functions in osteoblasts and chondrocytes during embryonic skeletogenesis. Several pathways not only provide initial inductive signals for lineage commitment, but also regulate progressive stages of chondrocyte and osteoblast differentiation. These include FGF, BMP2/TGFβ, Wnt/β-catenin, Ihh, PTH/PTHrP, and notch signaling^(28–36). Runx2 is both a transcriptional and post-translational target of these pathways^{(12, 28–36)}. In addition, during skeletogenesis, Runx2 is a central player in the execution of the osteo-chondrogenic signals of these pathways^{(1,12, 30–36)}. However, the specific regulatory crosstalk between Runx2 and these signaling pathways during commitment and progressive differentiation of chondrocytes and osteoblasts is yet to be fully understood. Indeed, unique Runx2 regulatory complexes formed in each cell type elicit differential responses^{(1,12,28–36)}. These may explain the failure of endochondral ossification and lethality in chondrocyte-specific Runx2-null mice in comparison to the grossly normal embryonic skeletogenesis and viability of osteoblast-specific Runx2-null mice.

Runx2 protein is a multi-partite transcription factor with distinct functional domains. Two models reporting conditional ablation of Runx2 have targeted different regions of the gene. We reported the first model where exon 8 of the Runx2 gene was floxed^(21,22,25). Deletion of exon 8 results in a mutant Runx2 protein (Δ369) that lacks 159 amino acids at the C-terminus. Notably, the Δ369 mutant does not act as a dominant negative Runx2 protein, and yields a null phenotype^{(6,21,22,25,27)}. Most of the N-terminal domains, including the Q/A (poly-glutamic-alanine), RHD (DNA binding domain), the NLS (nuclear localization signal), and part of the transcriptional activation/suppression domain are preserved in the Δ369 protein. However, this protein lacks both transcriptional activity and several key protein-protein interactions occurring in the C-terminus of Runx2. These include interactions with co-activators (ATF4, C/EBP, TAZ, pRb, SMAD), co-repressors (Groucho/TLE, HES, HEY, YAP), histone and chromatin modifiers (MOZ, MORF, P300, HDACs, SWI/SNF, Ubiquitin ligases) as well as transducers of Vitamin D, BMP/TGFβ, hedgehog, notch, and Wnt signaling pathways. Disruption of any of these pathways in mouse models results in a skeletal phenotype. Importantly, osteogenic signals from these pathways are not integrated into the osteoblast cell lineage without Runx2 C-terminus. The phenotype observed in Runx2^ΔE8/ΔE8 mice is likely caused by the inability of the Runx2 mutant protein to incorporate any number of these critical pathways. Thus, C-terminus dependent activity of Runx2 protein is essential for endochondral and intramembranous bone formation. Future studies will determine the precise mechanism underlying impaired bone synthesis in Runx2 mutant mice.

In the second model reported recently, exon 4 of the Runx2 gene was floxed⁽²⁶⁾. Deletion of exon 4 results in disruption of the RHD and the NLS of the Runx2 protein. Most likely, this mutant Runx2 protein is not imported to the nucleus, and is targeted for degradation⁽²⁶⁾. The short report utilized 2.3kb α1(I)-collagen-Cre transgenic line to delete exon 4 of the Runx2 gene in osteoblasts⁽²⁶⁾. Surprisingly, authors stated that “no apparent phenotypes were observed from E15, newborn, and 3-week old to 6-week old in α1(I)-Cre-Runx2flox/flox mice”. Using the same Cre line with our Runx2-floxed model, we also observed live birth of homozygous mutant mice with expected Mendelian frequency, and Runx2^ΔE8/ΔE8 mice were similar in size and weight to wild-type littermates. However, careful analysis and characterization of our mutant mice at the newborn stage revealed decreased calcification, poorly developed calvarial bones and wide-open fontanels. No changes were noted in femoral bone mass at 6 weeks of age in exon 4 Runx2 mutant mice⁽²⁶⁾. In sharp contrast, in our mice low bone mass was consistently noted in homozygous limbs of both genders at 4 and 12 week of age. We observed a significant decrease in both trabecular (55%) and cortical (32%) BV/TV in the femur of 4-week old Runx2^ΔE8/ΔE8 mice by µ-CT analysis. Histomorphometric analysis of Runx2^ΔE8/ΔE8 femurs showed a 53% decrease in BV/TV at 12 weeks of age. Additionally, we employed double-calcein labeling at 12 and 13 weeks of age, to understand dynamic bone parameters. We detected a striking decrease in mineral apposition rate and mineralizing surface to bone surface, and a 60% decrease in bone formation rate. Like our study, Takarada et al employed histomorphometric analyses on von Kossa stained vertebrae; however, no change in bone formation rate was reported at 3 and 6 weeks of age⁽²⁶⁾.

Earlier studies using global null and transgenic models demonstrate that Runx2 function in osteoblasts promotes osteoclastogenesis. Consistent with this notion, when we deleted Runx2 in committed osteoblasts, fewer TRAP positive cells were noted in both cortical and trabecular bone at birth and 4-weeks of age. Indeed, we confirmed that the OPG/RANKL ratio was increased due to Runx2 loss in osteoblasts, perhaps contributing to inhibition of osteoclastogenesis. Histomorphometric analysis also showed a 65% decrease in osteoclast number and a 53% decrease in osteoclast surface to bone surface in mutant mice at 12 weeks of age. Again, in contrast with our findings, Takarada et al showed a marked increase in osteoclast surface to bone surface, which did not affect BV/TV at 3 or 6 weeks of age. The molecular reasons for the discrepancy in the Takarada model were not revealed. The basis for contrasting phenotypes between Takarada and our osteoblast-specific Runx2 knockout models is yet to be fully investigated. However, we note that in both Runx2-floxed models, chondrocyte specific deletion results in a similar phenotype, suggesting that other variations in the model system may have contributed to the difference in osteoblast-specific Runx2 knockout models. For example, the chimeras of our exon 8 floxed line were outbred for 10 consecutive generations to establish a floxed allele on a pure C57BL/6 background. Significant variability in the skeletal mass and remodeling are known among different strains⁽³⁷⁾. To avoid any confounding factors associated with strain variation, we also outbred α1(I)-collagen-Cre transgenic mice from the SV129 onto the C57BL/6 background by 12 successive rounds of breeding. At this point, it remains unclear if the lack of phenotype in Takarada et al model is due to the utilization of mice with a mixed background.

One interesting finding in our model was the role of Runx2 in postnatal growth. At birth, mutant mice were similar to wild-type in weight, but exhibited a progressive reduction in growth rate, weighing 30% less than wild-type at 3 months of age. Low body mass was not associated with poor food intake, as Runx2 deficient mice consumed a similar amount of chow as wild-type littermates. Moreover, the number and location of teeth was normal in Runx2^ΔE8/ΔE8 mice. To our surprise, the length of limbs was similar between wild-type and mutant mice from birth to 3 months of age. Postnatal interstitial growth of the long bones is driven by chondrocytes. We confirmed that the level of Sox9, IHH and PTHrP, and the number of chondrocytes in resting, proliferating and hypertrophic zones, was similar between wild-type and homozygous mice. Moreover, we noted age-appropriate and comparable exhaustion of the growth plate between wild-type and mutant mice. Thus, the overall small stature of Runx2 mutant mice was not associated with impaired limb lengthening. Interestingly, we observed less visceral adipose tissue and skeletal muscle mass in homozygousmice. These results suggest that growth deficiencies in Runx2^ΔE8/ΔE8 mice may represent a systemic response to disrupted skeletal development. This notion is being further investigated.