Runx2 Deletion in Hypertrophic Chondrocytes Impairs Osteoclast Mediated Bone Resorption

Abstract

Deletion of Runx2 gene in proliferating chondrocytes results in complete failure of endochondral ossification and perinatal lethality. We reported recently that mice with Runx2 deletion specifically in hypertrophic chondrocytes (HCs) using the Col10a1-Cre transgene survive and exhibit enlarged growth plates due to decreased HC apoptosis and cartilage resorption. Bulk of chondrogenesis occurs postnatally, however, the role of Runx2 in HCs during postnatal chondrogenesis is unknown. Despite limb dwarfism, adult homozygous (Runx2^HC/HC) mice showed a significant increase in length of growth plate and articular cartilage. Consistent with doubling of the hypertrophic zone, collagen type × expression was increased in Runx2^HC/HC mice. In sharp contrast, expression of metalloproteinases and aggrecanases were markedly decreased. Impaired cartilage degradation was evident by the retention of significant amount of safranin-O positive cartilage. Histomorphometry and μCT uncovered increased trabecular bone mass with a significant increase in BV/TV ratio, trabecular number, thickness, and a decrease in trabecular space in Runx2^HC/HC mice. To identify if this is due to increased bone synthesis, expression of osteoblast differentiation markers was evaluated and found to be comparable amongst littermates. Histomorphometry confirmed similar number of osteoblasts in the littermates. Furthermore, dynamic bone synthesis showed no differences in mineral apposition or bone formation rates. Surprisingly, three-point-bending test revealed Runx2^HC/HC bones to be structurally less strong. Interestingly, both the number and surface of osteoclasts were markedly reduced in Runx2^HC/HC littermates. Rankl and IL-17a ligands that promote osteoclast differentiation were markedly reduced in Runx2^HC/HC mice. Bone marrow cultures were performed to independently establish Runx2 and hypertrophic chondrocytes role in osteoclast development. The culture from the Runx2^HC/HC mice formed significantly fewer and smaller osteoclasts. The expression of mature osteoclast markers, Ctsk and Mmp9, were significantly reduced in the cultures from Runx2^HC/HC mice. Thus, Runx2 functions extend beyond embryonic development and chondrocyte hypertrophy by regulating cartilage degradation, osteoclast differentiation, and bone resorption during postnatal endochondral ossification.

Introduction:

Endochondral ossification, a tightly orchestrated skeletal development process, involves formation of a precursory cartilage template that is eventually resorbed and replaced with mineralized bone (1–3). This transient cartilage is produced by the mesenchymal cells that undergo a unidirectional process of chondrocyte differentiation. The progression from resting, proliferative, prehypertrophic, and to the terminally mature hypertrophic chondrocytes is characterized by the expression of unique marker genes (4). The resting chondrocytes express high levels of SRY-box transcription factor 9 (Sox9), parathyroid hormone-related protein (Pthrp) and low levels of type II collagen (Col2a1), and aggrecan (Acan) (3–4). The proliferation of the resting chondrocytes is regulated by the combined actions of Pthrp and Indian hedgehog (Ihh) signaling (5). The daughter cells generated by mitotic division arrange perpendicular to the long axis of the growth plate followed by a 90-degree rotation, resulting in the classic columnar shape of proliferative chondrocytes. The non-canonical frizzled and the Wnt/planer cell polarity pathways play major roles in orienting chondrocytes in the proliferative zone (6–7). The expression of Ihh, Pthrp receptor (Pthrpr), and fibroblast growth factor receptor 3 (Fgfr3) in post proliferative chondrocyte mark the prehypertrophic chondrocytes (3). Prehypertrophic chondrocytes undergo significant volumetric swelling which results in chondrocyte hypertrophy (8).

Hypertrophic chondrocytes are the terminally differentiated chondrocytes that express type × collagen (Col10a1), vascular endothelial growth factor (Vegf), matrix metalloproteinases (Mmps) and a disintegrin and metalloproteinase with thrombospondin motifs (Adamts) (9). Hypertrophic chondrocytes undergo not only apoptosis but also transformation to osteoblasts, osteocytes, adipocytes, and marrow-associated skeletal progenitor cells (10–12). Hypertrophic chondrocytes are essential for calcification and degradation of the hyaline cartilage as well as converting perichondrium to periosteum (3–5). Mmp9 and Mmp13 degrade cartilage matrix by cleaving collagen, as well as proteoglycans at the Asn³⁴¹-Phe³⁴² linkage in aggrecan (13). Members of the Adamts family breakdown proteoglycans by specifically cleaving at Glu³⁷³-Ala³⁷⁴ residues in the aggrecan (14). In addition, IL-1β and Tnfα induce the expression and activity of Mmps in hypertrophic chondrocytes to promote cartilage degradation (15). Thus, hypertrophic chondrocytes are essential in regulating cartilage turnover during endochondral ossification.

Hypertrophic chondrocytes are also critical in vascularization of the hyaline cartilage and the recruitment and maturation of matrix-resorbing chondroclasts/osteoclasts (16–20). Hypertrophic chondrocyte secreted Vegfa is stored in the cartilage extra cellular matrix. Upon matrix degradation, increased bioavailability of Vegfa promotes neovascularization of hypertrophic cartilage and recruitment of circulatory monocytes/macrophages (16–17). Hypertrophic chondrocyte-produced Rankl, IL-17a, and Tnfα promote the maturation and functional activity of matrix resorbing chondroclasts/osteoclasts (21–23). Additionally, hypertrophic chondrocytes regulate formation and remodeling of trabecular bone via β-catenin signaling. Activation of β-catenin in hypertrophic chondrocytes inhibits osteoclastogenesis by suppressing the expression of Rankl and increasing the expression of Opg (24). Thus, hypertrophic chondrocytes regulate cartilage homeostasis by both direct and indirect signaling mechanisms.

Several signaling proteins and DNA binding transcription factors direct the differentiation of resting chondrocytes into the terminally mature hypertrophic chondrocytes (3, 9, 25). Key amongst them is the family of Runt-related (Runx) transcription factors comprising the DNA binding Runx1, Runx2, Runx3, and the non-DNA-binding heteromeric partner Cbfβ (25–26). Interaction of Cbfβ with the Runt domain of the Runx proteins enhances their DNA binding and transcription activity. All three Runx genes and the Cbfβ are expressed in the limb bud mesenchyme (25–28).

Global deletion of Cbfβ gene leads to failure of definitive hematopoiesis resulting in embryonic lethality prior to the onset of skeletogenesis (29). Col2a1-Cre mediated deletion of Cbfβ gene in the chondrocyte lineage from the onset of chondrocyte differentiation in embryos and postnatal mice leads to dwarfism and delayed endochondral ossification (30). The enlarged growth plate cartilage in 1-month-old Cbfβ mutant shows reduced length of proliferative and hypertrophic chondrocyte zones and bone formation during postnatal development (30). Like Cbfβ, global disruption of the Runx1 gene results in midgestational lethality due to blockage of fetal liver hematopoiesis and hemorrhaging in the central nervous system (31). Conditional deletion of Runx1 gene in the chondrocyte lineage using Col2a1-Cre transgene results in reduced length of the hypertrophic zone and trabecular bone in 1-month and 6-month-old mice (32). The global Runx3-null mice present sensory limb ataxia and colon inflammation, but embryonic osteogenesis is normal (33). Homozygous mice where Runx3 gene is inactivated in chondrocyte by Col2a1-Cre transgene show normal endochondral ossification (34). Chondrocyte differentiation, gene expression, and bone development were all comparable amongst Runx3 mutants and WT littermates at birth and 3-months of age (34).

The global deletion of Runx2 results in late embryonic lethality with complete failure of osteogenesis (35–37). Interestingly, only the deletion of Runx2 gene in the chondrocyte lineage from the onset of chondrocyte differentiation in embryos and postnatal mice by Col2a1-Cre transgene leads to perinatal lethality due to the failure of endochondral ossification (38–40). However, combined deletion of the Runx2 and Runx3 genes leads to complete blockage of chondrocyte maturation throughout the embryonic skeleton (41). Thus, Runx proteins have distinct and non-redundant functions in developing chondrocytes. However, neither Runx1 nor Runx3 can compensate for the failure of endochondral ossification caused by the loss of Runx2 in developing chondrocytes.

Runx2 deficient resting chondrocytes do not differentiate to hypertrophic chondrocytes leading to the failure of endochondral ossification (39). To bypass the early blockage in chondrocyte differentiation, we deleted Runx2 gene specifically in hypertrophic chondrocytes using Col10a1-Cre mice (42–43). Loss of Runx2 in hypertrophic chondrocytes does not cause lethality or failed endochondral ossification (43–44). However, a significant decrease in apoptosis of Runx2 deficient hypertrophic chondrocyte leads to the doubling of hypertrophic zone (43). Retention of hypertrophic cartilage in Runx2 mutant mice is linked to decreased expression of matrix degrading enzymes by hypertrophic chondrocytes (43). Therefore, Runx2 critical functions extend well beyond the differentiation of chondrocytes during embryonic endochondral ossification.

It is important to note that majority of the chondrogenesis in mammals takes place after birth (45). However, the role of Runx2 in postnatal chondrogenesis and endochondral ossification is largely unknown. Here, we evaluated the role of Runx2 in hypertrophic chondrocytes during postnatal skeletal development. We find a significant increase in hypertrophic cartilage and trabecular bone mass in adult mice. The synthesis of bone matrix by osteoblasts is comparable but the expression of several cartilage degrading enzymes is markedly decreased. Our data further demonstrate that the high trabecular bone mass is due to impaired maturation and function of osteoclasts in adult mice.

Materials and Methods:

Generation of Runx2 conditional knockout mouse

Generation of the Runx2-floxed (Runx2^F/F) mouse model where exon 8 of the Runx2 gene is floxed has been previously reported (39). To delete Runx2 gene specifically in the hypertrophic chondrocytes, Col10a1-Cre transgenic mice were used (42–43). Heterozygous mice (Runx2^+/HC) were bred to generate wildtype (Runx2^+/+) and homozygous (Runx2^HC/HC) littermates for comparative analyses. All littermates used for analyses were positive for Col10a1-Cre transgene. The genotypes of Runx2 littermates were confirmed using specific primer pairs (forward 5’-atcagttcccaatggtacccg-3’, reverse 5’-gcaagatcatgactagggattg-3’). The Col10a-Cre transgene was confirmed using forward 5’-tttagagcattatttcaaggcagtttc-3’ and reverse 5’-aggcaaattttggtgtacgg-3’ primers, and the tdTomato reporter gene was confirmed with forward 5’-ctgttcctgtacggcatgg-3’ and reverse 5’-ggcattaaagcagcgtatcc-3’ primers. The isolation of genomic DNA for genotyping and PCR conditions were described previously (43). The hypertrophic chondrocyte restricted activity of Col10a1-Cre transgene was confirmed using ROSA26-TdTomato reporter mice (42–43). All mouse lines were maintained on C57BL/6 background. All animal experiments were performed with the approval of the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham and conformed to relevant federal and state guidelines and regulations.

Histologic analyses

Hindlimbs from 10-weeks-old littermates were collected and fixed in 4% paraformaldehyde for 15 hours and processed for paraffin or plastic embedding. Fixed hindlimb tissues were dehydrated in increasing serial concentrations of ethanol followed by multiple xylene changes for tissue clearing. For plastic embedding, tissues were placed in 90% methyl methacrylate (MMA) and 10% dibutyl phthalate (DBP) solution for a week. Solution was changed weekly for a month to ensure tissue infiltration and then embedded in 90% MMA and 10% DBP with 0.25% Perkadox as a polymerization catalyst. Polymerized blocks were cured under UV light overnight. Tissues were serially sectioned at 5μm thickness using tungsten carbide blades and Leica RM 2265 microtome.

For Hematoxylin and Eosin staining, tissue sections were washed with xylene, and then rehydrated with decreasing ethanol concentrations and finally rinsed with water. Slides were stained with hematoxylin two times for three minutes each with intermediate water rinses. Slides were washed in 0.25% acid alcohol and bluing agent followed by Eosin Y for two minutes. Slides were then washed with ethanol and xylene. Tissue sections were stained with Safranin-O and Fast Green to identify cartilage matrix. For this, slides were washed with xylene, decreasing ethanol concentrations, water and then stained with Weigert’s Iron Hematoxylin and 0.05% fast green. Slides were then de-stained using 1% acetic acid and stained with 0.1% Safranin-O. Stained slides were then dehydrated and mounted using xylene-based mounting medium.

Tartrate-resistant acid phosphatase (TRAP) staining was performed to visualize osteoclasts. For this, plastic sections were deplasticized in three changes of cold acetone and then placed in 0.2M TRIS pH 9.4. The slides were then washed in water and placed in 0.2M acetate buffer for 20 minutes. Slides were then incubated at 37°C in naphtol AS-MX phosphate (0.5mg/mL), fast red TR salt (1.1mg/mL) and the acetate buffer solution. Finally, slides were rinsed in water and counterstained with dilute hematoxylin prior to mounting. The Goldner’s trichrome and Von Kossa staining were performed as per our previous article (46). Stained slides were imaged using Olympus BX51 microscope and histomorphometry performed using BIOQUANT Osteo 2022 Imaging Analysis Software V.22.5.60.

Micro-CT analysis

Intact femurs from 10-weeks-old littermates were obtained and stored in 1x PBS. Whole femurs were scanned at 12μm voxel size. The Scanco μCT40 desktop cone-beam μCT scanner was used to produce 3D images of undecalcified femurs (Scanco Medical AG, Brüttisellen, Switzerland using μCT Tomography V6.4-2). The following bone parameters were analyzed: bone volume (BV), total volume (TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular spacing (Tb.Sp), connectivity density (Conn.D), bone surface (BS), and structure model index (SMI). The μCT scans and analyses were performed by the UAB Small Animal Phenotyping Core supported by the NIH Nutrition & Obesity Research Center P30DK056336, Diabetes Research Center P30DK079626, and the UAB Nathan Shock Center P30AG050886A.

Structural analysis of bone

To test the flexural strength and stiffness of femurs, hindlimbs were collected from 10-weeks-old male littermates (n=10 each genotype). The isolated femurs were tested to failure by a three-point bending test using an 858 MiniBionix Materials Testing System (MTS Systems, Eden Prairie, MN, USA) in the UAB Experimental Biomechanics Core. A custom made three-point bending fixture was used with a span width of 5mm between the lower contacts. Crosshead speed of 0.05mm/sec was set to break the femur while the forces were measured using a 100N load cell. Force versus displacement curves were used to find the following parameters: ultimate force, P_max [N], from the maximum height of the curve and stiffness, k [N/mm], from the slope of the linear portion of the curve.

RNA isolation and quantitative real time PCR

Analysis of the gene expression by qPCR was performed as previously published (47). Briefly, total RNA was extracted from the distal femur and proximal tibia of hindlimbs of 10-week-old littermates. Muscle tissue was cleared from the deskinned distal femur and proximal tibiae followed by digestion initially with pronase (Roche Diagnostic, Indianapolis, IN, USA) and collagenase-D (Roche Diagnostic, Indianapolis, IN, USA) for 2-hr and subsequently with collagenase-D for 3-hrs at 37 °C. The epiphysis and metaphysis were harvested under dissecting microscope and flash frozen in liquid nitrogen and homogenized in PBS containing 0.1% DEPC. The growth plate homogenates were sonicated in TRIZOL reagent (Life Technologies, Carlsbad, CA, USA) to extract total RNA. The cDNA was prepared from 1.0 μg of total RNA using a cDNA synthesis kit (BioRad, Hercules, CA, USA). Quantitative real time PCR with four replicates was performed using iQ SYBR Green Supermix (BioRad, Hercules, CA, USA) with specific primer pairs detailed in Table 1. Gene expression values were normalized with Gapdh, used as an internal control. Relative gene expression was obtained using the 2^−ΔCT method.

Table 1.

Primer sequences used for qPCR

Gene Symbol	Forward (5’ to 3’)	Reverse (5’ to 3’)
Acan	ccc tcc ggc aga aga aag at	cgc ttc tgt agc ctg tgc ttg
Col2a1	tcc tct gcg atg aca tta tct	gat cat ctc tgg gtc ctt gtt
Col10a1	gca gca tta cga ccca aag at	ctt gaa gcc tga tcc agg ta
Ihh	acg tgc att gct ctg tca ag	ctc gat gac ctg gaa agc tc
Six1	atg ctg ccg tcg ttt ggt t	cct tga gca cgc tct cgt t
Prg4	act tca gct aaa gag aca cgg agt	gtt cag gtg gtt cct tgg ttg tag taa
Adamts4	caa gca gtc ggg ctc ctt	gat cgt gac cac atc gct gta
Adamts5	cca gtt gta caa aga tta tcg gaa cct	gtt gct cct tca ggg atc ct
Mmp9	gat ccc cag agc gtc att	cca cct tgt tca cct cat tt
Mmp13	tgt ttg cag agc act act tga a	cag tca cct cta agc caa aga aa
Tnfsf11	act tgg gat ttt gat gct ggt t	tgg gcc aag atc tct aac atg a
Tnfrsf11b	atg aac aag tgg ctg tg	cct cac tgt gca gtg ctg tt
Il17a	ctt ggt ggg atg gac tgg	tct ttt gcc ttt ggt gtt cc
Ctsk	gcc tag cga aca gat tct caa	cac tgg gtg tcc agc att t
Tnfrsf11a	cat ctt cgg cgt tta cta cag g	tcc act tag act act gca agc a
C-fos	cgg gtt tca acg ccg act a	ttg gca cta gag acg gac aga
NfatC1	gga gag tcc gag aat cga gat	ttg cag cta gga agt acg tct
Trap	gca gta tct tca gga cga gaa c	tcc ata gtg aaa ccg caa gta g
Traf6	atg cag agg aat cac ttg gca	acg gac gca aag caa ggt t
Dcstamp	ggt ctc cta gac agc atg act	gca agg ccg taa atc cac tg
Alp	cag tgg gag tga gcg cag cc	gca ctg ggt gtg gcg tgg tt
Col1a1	aaa gct tct tct cct ctg ag	gag acc cag gaa gac ctc tg
Spp1	atg agg ctg cag ttc tcc tgg	aaa gct tct tct cct ctg gc tgc c
Ibsp	aca atc cgt gcc act cac t	ttt cat cga gaa agc aca gg
Bglap2	cct agg tag tga aca gac tcc ggc g	ctg gtc tga ata gct cgt cac aaa
Sost	aga aca acc aga cca tga acc g	gca gct gta ctc gga cac g
Runx2	gcc ctc cct gaa ctc tgc ac	tca tgc ctg gct ctt ctt act
Sp7	tct cca tct gcc tga ctc ct	gga ctg gag cca tag tga gc
Gapdh	ccg cct gga gaa acc tgc caa g	gga tag ggc ctc tct tgc tca g

Dynamic Bone Formation Assay

Calcein (20mg/kg body weight) was injected intraperitoneally into 9-weeks old Runx2^+/+ and Runx2^HC/HC littermates. After 7-days, a second calcein injection was administered and littermates were sacrificed 2-days later. Hindlimbs were removed, fixed in 4% paraformaldehyde for 15 hours, and embedded in plastic for sectioning. Central femur sections were imaged. The following dynamic bone formation parameters were quantified using BIOQUANT Osteo 2022 Imaging Analysis Software: mineralized surface (MS), bone surface (BS), mineral apposition rate (MAR), bone formation rate (BFR), bone volume (BV), and total volume (TV).

Bone Marrow Osteoclast Cultures and TRAP Staining

Hindlimbs were harvested from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates. Femurs and tibiae from both hindlimbs were deskinned and cleared of muscles. Proximal and distal ends of bones were cut to flush the bone marrow. Whole bone marrow cells were cleared of red blood cells by resuspending the cell pellet in ACK lysis buffer for 5 minutes. Cells were seeded in 10cm plates and cultured in growth media (αMEM + 10% FBS) for 48 hours at 37°C. Nonadherent cells were collected and seeded (2×10⁵ cells/well) in 12-well plates containing growth media supplemented with M-CSF (40ng/μl) for 3 days. Cells were then cultured in M-CSF (20ng/μl) and RANKL (40ng/μl) for 2 days, fixed and processed for TRAP staining or harvested for RNA isolation. TRAP-stained plates were imaged using Keyence BZ-X810, and osteoclast area (OC area) and perimeter (OC perim) were quantified using BZ-X800 Viewer software (V.01.01.01.03).

Statistical analyses

All statistical analyses were performed using GraphPad Prism (Version 9.3.1.). Statistical significance between Runx2^+/+ and Runx2^HC/HC littermates was determined using an unpaired Student’s t-test. Data with p < 0.05 were considered statistically significant. Data are shown with a box and whisker plot or a line diagram showing maximum to minimum values, mean, and standard deviation of the means.

Results:

Runx2 deficiency in hypertrophic chondrocytes leads to increased growth plate and articular cartilage

Global or resting chondrocyte-specific deletion of Runx2 gene results in failure of chondrocyte hypertrophy and embryonic lethality. In contrast, mice where Runx2 gene is deleted in hypertrophic chondrocytes survive but present with limb dwarfism and impaired cartilage resorption (43). Importantly, the majority of chondrogenesis takes place during early postnatal development. Here, we evaluated postnatal chondrogenesis and endochondral ossification in adult mice by deleting Runx2 gene specifically in hypertrophic chondrocytes (Supplemental Fig. S1A–B). Both wildtype (Runx2^+/+) and homozygous mutant (Runx2^HC/HC) littermates showed a progressive increase in body weight from birth to 10-weeks of age. The non-significant lower body weight noted in the Runx2^HC/HC littermates was sustained through adulthood (Fig. S1C). The overall body length (snout-tail) of Runx2^HC/HC was 6% shorter than wildtype littermates (Fig. S1D, Runx2^+/+: 17.1 ± 1.2 cm Runx2^HC/HC: 16.1 ± 0.9 cm; n=13). Alizarin red and alcian blue stained forelimbs and hindlimbs of Runx2^HC/HC littermates also showed reduced length. The humerus showed 18% reduction (Runx2^+/+: 1.46 ± 0.2 cm Runx2^HC/HC: 1.2 ± 0.1 cm; n=5), but the length of radius and ulna were comparable amongst littermates (Fig. S1E). The femur and tibia showed 12% (Runx2^+/+: 1.6 ± 0.3 cm Runx2^HC/HC: 1.4 ± 0.2 cm; n=5) and 8% (Runx2^+/+: 2.1 ± 0.2 cm Runx2^HC/HC: 1.9 ± 0.1 cm; n=5) reduction, respectively (Fig. S1F). These data indicate that limb dwarfism observed in the newborn Runx2^HC/HC mice persists into adulthood.

Histologic analysis revealed an increase in the length of growth plate cartilage in 10-weeks-old Runx2^HC/HC littermates (Fig. 1A). Quantification revealed a significant 53% increase in growth plate thickness in Runx2^HC/HC mice (Fig. 1B). This increase in Runx2^HC/HC growth plate is mostly due to 3x increase in the length of hypertrophic zone (Fig. 1C). The number of hypertrophic chondrocytes noted in the Runx2^HC/HC growth plate were 5–7 cell layers compared to 2–3 cell layers in the Runx2^+/+ littermates (Fig. 1A). Interestingly, the articular cartilage length was also increased by 38% in the Runx2^HC/HC littermates (Fig. S2A–B). A similar increase in the length of growth plate and articular cartilage was noted in Runx2^HC/HC female littermates (Fig. S2C–D and data not shown). To independently evaluate changes in the growth plate cartilage, we performed Safranin-O Fast Green staining. Consistent with the H&E data, a 43% increase was noted in the growth plate cartilage in Runx2^HC/HC littermates (Fig. 2A–B). Interestingly, safranin-O positive cartilage was present in most of the Runx2^HC/HC bone trabeculae that extended toward the diaphysis of the femurs. Compared to wildtype, 56% of the bone trabeculae were safranin-O positive in the homozygous mutant littermates (Fig. 2C). Additionally, a significant amount of marrow adiposity was consistently noted in the tibiae and femurs of the Runx2^HC/HC mice (Fig. 2A). Together, these observations suggest that during postnatal endochondral ossification Runx2 loss in hypertrophic chondrocytes results in increased amount of growth plate cartilage.

Figure 1. Growth plate cartilage and the hypertrophic zone is increased in adult mice with hypertrophic chondrocyte-specific ablation of the Runx2 gene.

(A) Representative images of the hematoxylin and eosin-stained distal femur and proximal tibia of the Runx2^+/+ and Runx2^HC/HC male littermates are shown at 2x magnification. The boxed regions of the femur growth plate are shown at 10x and 40x magnification. The brackets indicate the length of resting/proliferative (black) and the hypertrophic zone (red). (B) Total length of the growth plate was measured digitally, and the pooled data from four littermates (n=4 each genotype) is presented in the box and whisker graph. (C) The length of hypertrophic chondrocytes zones (HZ) in the femur growth plate was measured and the pooled data from four Runx2^+/+ and Runx2^HC/HC littermates is presented in the box and whisker graph. (***P < 0.001)

Figure 2. Increased amount of cartilage is retained in the bone trabeculae of the adult Runx2^HC/HC mice.

(A) Hind limbs from 10-weeks-old Runx2^+/+ and Runx2^HC/HC male littermates were processed for histologic analysis and double stained with Safranin-O/Fast green. Representative images showing the distal femur and proximal tibia regions are presented at 2x magnification. The boxed regions are shown at 10x magnification. (B) Thickness of the Safranin-O positive growth plate cartilage (red) was measured digitally, and the pooled data from three littermates (n=3 each genotype) is presented in the box and whisker graph. (C) The number of Safranin-O positive bone trabeculae were quantified and the pooled data from three Runx2^+/+ and Runx2^HC/HC littermates is presented in the box and whisker graph. (**P < 0.01, ***P < 0.001)

Expression of the aggrecan and collagen degrading enzymes is decreased in Runx2^HC/HC mice.

For a molecular understanding of the increased cartilage phenotype, we performed gene expression analyses. Total RNA was isolated from the distal femurs and proximal tibiae epiphyseal growth plates of adult Runx2^+/+ and Runx2^HC/HC littermates. There was no significant difference in the expression of type II collagen that is expressed by resting and proliferative chondrocytes (Fig. 3). In contrast, the expression of aggrecan, the most abundant proteoglycan in the growth plate, was significantly decreased in the Runx2^HC/HC mice. The Indian hedgehog (Ihh) expressed by prehypertrophic chondrocytes showed no significant difference amongst littermates. Consistent with the expansion of HZ zone, the mRNA levels of type × collagen (Col10a1), was significantly increased in Runx2^HC/HC mice (Fig. 3). In sharp contrast, the vascular endothelial growth factor A (Vegfa) was markedly decreased in the homozygous mice. We also assessed the expression of SIX homeobox 1 (Six1) transcription factor and proteoglycan 4 (Prg4) that are expressed by articular chondrocytes. Expression of both marker genes was significantly decreased in Runx2^HC/HC mice (Fig. 3). Evaluation of the cartilage-degrading enzymes revealed a significant reduction in the expression of collagenase (Mmp13) and aggrecanases (Adamts4, Adamts5). These results demonstrate that Runx2 activity in hypertrophic chondrocytes is required for the expression of enzymes responsible for the turnover of postnatal cartilage.

Figure 3. Runx2 deficiency impairs expression of matrix degrading enzymes by the hypertrophic chondrocytes.

Total RNA was harvested from the flash frozen cartilaginous portions of femurs and tibiae of 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates (n=6 each genotype). Equal amount of cDNA was used for qPCR reactions performed in four technical replicates. Relative expression of Col2a1, Acan, Ihh, Col10a1, Vegfa, Six1, Prg4, Mmp13, Adamts4, and Adamts5 genes normalized with Gapdh is presented in box and whisker plot. (**P < 0.01, ***P < 0.001)

Loss of the Runx2 gene in hypertrophic chondrocytes leads to a significant increase in trabecular bone mass but does not improve structural integrity

The histologic analysis indicated an increase in the trabecular bone in the adult Runx2^HC/HC mice (Fig. 1, 2). Therefore, we performed μCT analysis to quantify changes in the adult bone mass. Femurs of 10-weeks-old Runx2^HC/HC littermates showed a significant increase in the trabecular bone mass (Fig. 4A). Quantification of the bone parameters from five male littermates showed a significant increase in the bone volume and tissue volume leading to significant increase in the BV/TV ratio in Runx2^HC/HC mice (Fig. 4B). A significant increase in the trabecular number led to a significant decrease in the trabecular space. The increase in trabecular thickness, however, did not reach statistical significance between Runx2^+/+ and Runx2^HC/HC littermates (Fig. 4B). In addition, connectivity density and the bone surface of the trabecular bone also showed a significant increase in the Runx2^HC/HC littermates. Interestingly, the structural model index was reduced significantly in the Runx2^HC/HC mice. The cortical bone was comparable amongst Runx2^+/+ and Runx2^HC/HC littermates (Fig. S3A). Quantification of cortical bone revealed no significant changes in the BV/TV ratio and other parameters (Fig. S3B). The increased trabecular bone and unchanged cortical bone parameters were also observed in 10-weeks-old female Runx2^HC/HC littermates (Fig. S4). These results indicate that increased trabecular bone phenotype in the Runx2^HC/HC mice is not influenced by sex differences.

Figure 4. Adult Runx2^HC/HC mice show increased trabecular bone mass.

(A) Representative 3D reconstructions of the trabecular bone from the femurs of 10-weeks-old male littermates performed at 12 μm voxels. Scale bar: 100μm. (B) Quantification of trabecular bone parameters from a 2.4mm region immediately beneath the growth plate from Runx2^+/+ and Runx2^HC/HC littermates (n=5 each genotype). BV, bone volume; TV, total volume; BV/TV, bone volume fraction; Tb.Th, trabecular thickness; Tb.Sp, trabecular space; Tb.N, trabecular number; Conn.D, connectivity density; BS, bone surface; and SMI, structure model index. Data are presented as mean ± SEM, n=5 per group in a box and whisker plot. (*p < 0.05, **p < 0.01, ***p < 0.001)

For an independent assessment of changes in the adult bone mass, femurs from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates were analyzed for bone mineralization and bone histomorphometry. Von Kossa staining revealed an overall significant increase in the mineralized matrix beneath the growth plate and in the number of mineralized bone trabeculae in Runx2^HC/HC mice (Fig. 5A). Histomorphometric analyses confirmed a significant increase in the BV/TV ratio, and the trabecular number in Runx2^HC/HC littermates (Fig. 5B–C). Trabecular thickness was comparable, but the trabecular space decreased significantly in Runx2^HC/HC mice (Fig. 5C). Thus, multiple lines of evidence demonstrate that Runx2 deficiency in the hypertrophic chondrocytes leads to an increased trabecular bone in adult mice.

Figure 5. Increased trabecular mineralization in the adult Runx2^HC/HC mice.

(A) Femur plastic sections from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates were stained with Von Kossa. Representative images of the femur taken at 4x magnification are shown. The boxed region is shown at 10x magnification. Scale bar: 100μm. (B) Goldner’s trichrome stained plastic section of femurs from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates. Representative images of the femur taken at 4x magnification are shown. Scale bar: 100μm. (C) Histomorphometry analysis from Goldner’s trichrome (n=4 per genotype). BV, bone volume; TV, total volume; BV/TV, bone volume fraction; Tb.N, trabecular number; Tb.Th, trabecular thickness; and Tb.Sp, trabecular space. (*p < 0.05, **p < 0.01)

We next examined if the increased trabecular bone mass contributes to bone strength. Femurs from 10-weeks-old littermates were subjected to three-point-bending test (Fig. 6). To our surprise there was a significant decrease in the force required to fracture the femurs of Runx2^HC/HC mice. However, the ultimate displacement was similar in the Runx2^+/+ and Runx2^HC/HC littermates (Fig. 6). The stiffness of the Runx2^HC/HC femurs was significantly less than Runx2^+/+ littermates (Fig. 6). Together, these data indicate that bone in Runx2^HC/HC mice are structurally less sound when compared to their wildtype littermates.

Figure 6. Runx2 deficiency in hypertrophic chondrocytes reduces structural integrity of adult bones.

Hindlimbs collected from 10-weeks-old Runx2^+/+ and Runx2^HC/HC male littermates were used to obtain intact femurs by joint dislocation. The isolated femurs were cleared of soft tissue and subjected to three-point bending test using a custom-made fixture. Representative force displacement curves for two Runx2^+/+ and Runx2^HC/HC femurs are shown. Pooled data from n=6 Runx2^+/+ and n=7 Runx2^HC/HC femurs is presented in the box and whisker plots. (***P < 0.001)

Increased trabecular bone mass in Runx2^HC/HC mice is not linked to osteoblast function

The consistent increase in the trabecular bone mass prompted the examination of functional activity of the osteoblasts. Total RNA obtained from the adult diaphysis of femurs and tibiae was used for analysis of gene expression in osteoblasts. The expression of early markers of osteoblasts, alkaline phosphatase (Alp), and collagen type 1 (Col1a), was similar in Runx2^+/+ and Runx2^HC/HC littermates (Fig. 7A). The expression levels of non-collagen bone matrix proteins osteopontin (Opn) and bone silo protein (Bsp) was also comparable amongst littermates. Moreover, mature osteoblast and osteocyte marker genes, osteocalcin (Ocn), and sclerostin (Sost), showed similar expression (Fig. 7A). The major transcription factors required for osteoblast differentiation and bone synthesis, Runx2 and Sp7, showed comparable expression in Runx2^+/+ and Runx2^HC/HC littermates (Fig. 7A). These data indicate that Runx2 deficiency in the hypertrophic chondrocytes does not impact gene expression in osteoblasts. We next performed histomorphometric analysis on Goldner’s trichrome stained femurs of adult littermates. The number of osteoblasts per bone surface (N.Ob/BS) showed no significant change amongst Runx2^+/+ and Runx2^HC/HC littermates (Fig. 7B). Additionally, the osteoblast surface (ObS) and osteoblast surface per bone surface (ObS/BS) were found to be similar. These data are consistent with the gene expression analysis showing no alteration in osteoblast differentiation.

Figure 7. Expression of osteoblast markers is not impaired despite increased bone mass.

(A) Total RNA was isolated from the flash frozen diaphyseal region of femurs and tibiae of 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates (n=6 each genotype). Equal amount of cDNA was used for qPCR reactions performed in four technical replicates. Relative gene expression of osteoblast differentiation markers Alp, Col1a, Opn, Bsp, Ocn, Sost, Runx2, and Sp7 genes normalized with Gapdh is presented in box and whisker plot. (B) Histomorphometry from Goldner’s trichome stained plastic sections of 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates (n=4 each genotype). N.Ob/BS, number of osteoblasts per bone surface; ObS, osteoblast surface; ObS/BS (%), osteoblast surface per bone surface.

To test whether the increase in trabecular bone was due to an increase in bone synthesis, we performed double calcein labeling to track the osteoblast activity during dynamic bone synthesis (Fig. 8). Calcein incorporation was apparent on >60% of the bone surface in both Runx2^+/+ and Runx2^HC/HC littermates (Fig. 8A). Interestingly, littermates showed comparable intensity of the calcein signal in the cortical and trabecular bone surfaces (Fig. 8A). Analysis of the double-labeled bone surfaces revealed similar mineral apposition rate (MAR) and comparable bone formation rate (BFR) amongst littermates (Fig. 8B). Thus, postnatal bone synthesis functions of osteoblasts are not affected by the deficiency of Runx2 gene in the hypertrophic chondrocytes.

Figure 8. Dynamic bone synthesis is unchanged in Runx2^HC/HC mice.

(A) Runx2^+/+ and Runx2^HC/HC male littermates were injected with calcein at 9-weeks of age and again at 10-weeks (n=4 each genotype). Representative images of femurs in FITC channel are shown at 2x magnification. Scale bar: 500μm (B) Dynamic bone synthesis data pooled from four independent femurs is presented in the graphs. MS, mineralizing surface; MS/BS, mineralizing surface per bone surface; MAR, mineral apposition rate; BFR/BS, bone formation rate per bone surface; BFR/BV, bone formation rate per bone volume; BFR/TV, bone formation rate per total volume.

Runx2-deficiency in hypertrophic chondrocytes impairs osteoclast differentiation

Lack of changes in the osteoblast markers and bone synthesis in the Runx2^HC/HC mice prompted the evaluation of osteoclast mediated bone resorption. Histomorphometric analysis revealed a significant decrease in both the osteoclast number and osteoclast surface per bone surface in the Runx2^HC/HC littermates (Fig. 9A). The erosion surface was markedly decreased but the quiescent surface showed no significant change in the Runx2^HC/HC mice. For direct assessment of osteoclasts, we performed TRAP staining on femurs of 10-weeks-old littermates (Fig. 9B). A marked reduction in the TRAP signal was evident in the Runx2^HC/HC mice. Quantification of TRAP positive cells showed a significant decrease in the number of osteoclasts in the Runx2^HC/HC mice when compared to Runx2^+/+ littermates (Fig. 9C). Total RNA obtained from metaphysis of distal femur and proximal tibia showed a significant decrease in the expression of Rankl (Fig. 9D). The expression of the decoy receptor Opg was comparable amongst littermates. Consequently, the Rankl/Opg ratio was significantly decreased in Runx2^HC/HC mice (Fig. 9D). A reduced Rankl/Opg ratio has a well-known inhibitory effect on osteoclast differentiation. Thus, the decreased Rankl/Opg ratio is consistent with the reduced number of osteoclasts observed in Runx2^+/+ littermates (Fig. 9A–C). IL-17a secreted by hypertrophic chondrocytes also regulates osteoclastogenesis. Interestingly, expression of IL-17a was significantly decreased in the Runx2^HC/HC mice (Fig. 9D). Together, these data indicate that changes in the gene expression due to Runx2 deficiency in hypertrophic chondrocytes impairs development of osteoclast/chondroclast.

Figure 9. Reduced expression of Rankl and IL-17a is associated with fewer osteoclasts in adult Runx2^HC/HC mice.

(A) Histomorphometric quantification of Goldner’s trichome stained femurs from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates (n=4 each genotype). N.Oc/BS, number of osteoclasts per bone surface; OcS, osteoclast surface; OcS/BS, osteoclast surface per bone surface; ES, erosion surface; QS, quiescent surface. (B) Representative images of TRAP-stained femurs from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates captured at 10x magnification. The boxed regions are shown at 20x magnification. Scale bar: 100μm. (C) Quantification of TRAP+ cells from four independent femurs (n=4), Runx2^+/+ and Runx2^HC/HC littermates. (D) Total RNA was isolated from the flash frozen metaphysis of femurs and tibiae (n=6 each genotype). Expression of Rankl, Opg, and IL-17a genes evaluated by qPCR analysis. Normalized value with Gapdh is presented in box and whisker plot. (*P < 0.05, **P < 0.01, ***P < 0.001).

To better analyze impairment in the differentiation and function of osteoclast, we performed ex vivo culture of bone marrow cells from Runx2^+/+ and Runx2^HC/HC littermates. Briefly, flushed bone marrow was cleared of erythrocytes, and the non-adherent cells were cultured in osteoclast differentiation media using Mcsf and Rankl. Compared with the Runx2^+/+ littermates, fewer multinucleated osteoclasts developed in the Runx2^HC/HC culture (Fig. 10A). Quantification of the cells from three different plates revealed a significant reduction in the number of osteoclasts (Fig. 10B). Moreover, both the area and perimeter of the osteoclasts formed in the Runx2^HC/HC culture were significantly smaller (Fig. 10B). To understand the impairment in osteoclast differentiation, gene expression analyses were performed from RNA collected from the osteoclast cultures. Interestingly, expression of the essential transcription factors for osteoclast development, Cfos and NfatC1, was comparable (Fig. 10C). Similarly, the expression of key osteoclast receptors, RankR and Traf6, showed no significant change between Runx2^+/+ and Runx2^HC/HC cultures (Fig. 10C). Consistent with the small size of osteoclasts, the expression of DC-Stamp that promotes fusion of mono-nuclear cells was reduced significantly in the Runx2^HC/HC cultures (Fig. 10C). The expression levels of Trap, an early marker of osteoclasts, was similar amongst the Runx2^+/+ and Runx2^HC/HC cultures. In sharp contrast, the levels of both Mmp9 and Ctsk, which are primarily expressed by mature osteoclasts, were significantly reduced in the Runx2^HC/HC cultures (Fig. 10C). Taken together these results demonstrate that impairment in osteoclast differentiation contributes to the increased cartilage and trabecular bone mass noted in adult Runx2^HC/HC mice.

Figure 10. Runx2 deficiency in hypertrophic chondrocytes impairs osteoclast differentiation.

(A) Non-adherent bone marrow cells isolated from 10-weeks-old Runx2^+/+ and Runx2^HC/HC littermates (n=3 each genotype) were cultured in osteoclast differentiation media. Representative images of TRAP-stained cells captured at 10x magnification are shown. (B) Number of TRAP+ multi-nucleated osteoclast and their area and perimeter was quantified from three independent cultures and is presented in the bar graphs. (C) Cells at day 5 of osteoclast differentiation were collected to isolate total RNA. Expression of osteoclast marker genes Cfos, NfatC1, RankR, Traf6, DC-Stamp, Trap, Mmp9, and Ctsk was determined by qPCR analysis. Data normalized with Gapdh is presented in box and whisker plot. (*P < 0.05, **P < 0.01, ***P < 0.001).

Discussion:

The unidirectional and sequential process of endochondral ossification involves chondrocyte proliferation, hypertrophic differentiation, cartilage calcification, and turnover. Runx2 deletion in the chondrocyte lineage from the onset of chondrocyte differentiation using the Col2a1-Cre transgene results in perinatal lethality due to failed endochondral ossification (38–40). Runx2 expression increases progressively from the resting to hypertrophic chondrocytes. Deletion of Runx2 specifically in hypertrophic chondrocytes does not cause lethality but leads to impaired apoptosis of hypertrophic chondrocytes and cartilage resorption (43). Bulk of active cartilage growth occurs after birth, however, the role of Runx2 in hypertrophic chondrocytes during postnatal chondrogenesis and endochondral ossification is unknown. Here, we report the loss of Runx2 in hypertrophic chondrocytes results in increased growth plate and articular cartilage in adult mice. Runx2 deficient hypertrophic chondrocytes show markedly reduced expression of enzymes responsible for the degradation of cartilage matrix. The increased trabecular bone mass in Runx2^HC/HC mice is not related to enhanced osteoblast differentiation or bone synthesis. We discovered that Runx2 activity in hypertrophic chondrocytes is critical for the development of osteoclasts and bone resorption. Together, our data demonstrates that Runx2 functions extend beyond chondrocyte hypertrophy by regulating cartilage degradation and bone turnover during postnatal chondrogenesis.

The majority of chondrogenesis occurs within eight weeks of postnatal skeletal development (45). To understand the effects of Runx2 loss in the hypertrophic chondrocytes on postnatal growth, body weights of Runx2^+/+ and Runx2^HC/HC littermates were recorded weekly. Both littermates showed a comparable weight gain through adulthood. Our data is consistent with recent reports that find no changes in the body weight of 20-weeks old Runx2^HC/HC littermates (44). Interestingly, the limb dwarfism observed in Runx2^HC/HC mice at birth is sustained into adulthood. Runx2-deficiency in hypertrophic chondrocytes increases the length of the growth plate cartilage in both male and female littermates, suggesting that these effects are independent of sex steroids. A decrease in apoptosis of Runx2-deficient hypertrophic chondrocytes during embryonic development may account for this increased number of hypertrophic chondrocytes and growth plate cartilage (43). We also noted an increase in articular cartilage and the hypertrophic zone in Runx2^HC/HC mice, suggesting Runx2 involvement in the survival of hypertrophic chondrocytes in the articular cartilage. The increased amount of type × collagen (Col10a1) mRNA in Runx2^HC/HC mice is consistent with these observations.

One interesting finding in the adult mice is the increased amount of growth plate and articular cartilage despite a significant reduction in the expression of key proteoglycans, aggrecan and lubricin. This increased cartilage is linked to a dramatic reduction in the expression of Mmp13 enzyme responsible for degradation of highly abundant type II collagen and aggrecanases Adamts4 and Adamts5 (48). Runx2 regulates the expression of Mmp13 and Adamts5 genes by directly binding to the Mmp13 and Adamts5 promoter (49–50). The reduced expression of both Mmp13 and Adamts5 in the adult Runx2^HC/HC mice is consistent with early observations in the embryonic and newborn Runx2^HC/HC littermates (43). Our data is in sharp contrast to a previous study that reported decreased expression of Mmp13 and Adamts5 in the Runx2 mutant mice but without any changes in cartilage turnover either in newborns or adults (44). Importantly, Runx2 overexpression in chondrocytes induces the expression of both Mmp13 and Adamts5 (51). Deletion of exon 4 of the Runx2 gene in adult mice inhibits expression of Mmp9, Mmp13 and Adamts4, 5, 7 and 12 and slows the degradation of cartilage (52). Thus, the increased amount of cartilage noted in the adult Runx2^HC/HC mice reflects a retention of cartilage matrix due to the impairments in degradation.

Terminally mature secretory hypertrophic chondrocytes produce angiogenic factors such as Vegfa that promote neovascularization (17). The expression of Vegfa was nearly absent in the Runx2^HC/HC mice. Our in vivo findings are consistent with earlier reports showing a direct transcriptional regulation of Vegfa promoter by the Runx2 protein (53). We and others have previously reported that Runx2 regulates Vegfa expression in chondrocytes and poor vascular invasion of hypertrophic cartilage when Runx2 is deleted in the resting chondrocyte using Col2a-Cre transgene (39, 54). Runx2 also stabilizes Hif1α protein and stimulates growth plate angiogenesis (55). Low Vegfa expression in the Runx2^HC/HC littermates suggests a likely defect in the vascularization of hypertrophic cartilage.

The surprised increase in trabecular bone mass in the adult Runx2^HC/HC mice led to the evaluation of osteoblast differentiation and postnatal bone synthesis. Interestingly, the expression levels of both early (Alp, Col1a, Opn) and late (Bsp, Ocn, Sost) markers of osteoblast differentiation showed no significant change amongst littermates. Consistent with these results, histomorphometric analysis revealed comparable numbers of osteoblasts in the Runx2^+/+ and Runx2^HC/HC littermates. Furthermore, double calcein labeling showed no change in the mineral apposition rate or dynamic bone synthesis. Our result of increased trabecular bone mass in adult Runx2^HC/HC mice is consistent with earlier report showing enhanced bone mass in the newborn and 3-days old littermates (43). Thus, osteoblast gene expression and postnatal bone synthesis are not affected by Runx2 deficiency in the hypertrophic chondrocytes.

Evidence is accumulating that, apart from apoptosis, hypertrophic chondrocytes in the growth plate can transform to osteoblasts, adipocytes, pericytes, and marrow-associated skeletal progenitors (10–12). However, the reported proportions of hypertrophic chondrocyte-derived osteoblasts vary widely, ranging from 40%−80% at E18, 15%−73% at 1-month, and 83% at 2-months of age (10–12). A 30% transformation of hypertrophic chondrocytes into osteoblasts was reduced to 1–3% in mice, when exon 4 of the Runx2 gene was deleted by Col10a1-Cre line (44). Surprisingly, no changes were observed either in the expression of osteoblast marker genes, number of osteoblasts, trabecular bone volume, or dynamic bone synthesis in the newborn, 1.5-months, and 5-months old Runx2 mutant mice (44). Using the same Col10a1-Cre line, we also found similar numbers of osteoblasts, gene expression, and dynamic bone synthesis. However, we found a significant increase in the trabecular bone in our model that is linked to marked reduction in cartilage and bone resorption. Therefore, molecular regulators of hypertrophic chondrocyte transformation to osteoblasts and the contribution of hypertrophic chondrocyte-derived osteoblasts to endochondral ossification remains unclear.

The increased trabecular bone mass in Runx2^HC/HC mice, surprisingly, decreases the structural integrity of the bones. Runx2^HC/HC femurs had a significant decrease in stiffness and required less force for fracture. The three-point-bend test primarily assesses the structural integrity of the cortical bone. However, the cortical bone was comparable amongst Runx2^+/+ and Runx2^HC/HC littermates. The increased trabecular bone that extends to the mid-diaphysis was consistently noted in both male and female Runx2^HC/HC mice. We also observed an increased amount of marrow adiposity in the tibiae and femurs of the Runx2^HC/HC mice. Hypertrophic chondrocytes in the growth plate can transdifferentiate to adipocytes (10, 12). Enhanced marrow adiposity is associated with increased fracture risk and bone fragility (56). This increased marrow adiposity may explain the reduced structural strength noted in the bones of adult Runx2^HC/HC mice.

Chondroclasts and osteoclasts are essential monocyte/macrophage-derived cells responsible for cartilage degradation and bone erosion, respectively. Histomorphometric quantification showed fewer TRAP+ cells associated with the cartilage islands and bone surface of Runx2^HC/HC mice. Hypertrophic chondrocytes are not known to transform to monocyte/macrophage lineage. The vascularization of hypertrophic cartilage allows the recruitment of monocytes/macrophages and pre-osteoclasts. One possibility for the low number of osteoclasts in Runx2^HC/HC mice is likely poor vascularization due to the reduced expression of Vegfa. The osteoclast surface/bone surface and the erosion surface were also decreased in the Runx2^HC/HC mice. The poor resorption of the cartilage matrix likely affords the osteoblasts more cartilage template for the deposition of bone matrix. Furthermore, fewer osteoclasts lead to reduced bone resorption. Together, these events likely contributed to increased trabecular bone mass noted in the Runx2^HC/HC littermates.

Several secreted factors such as Mcsf, IL-17a, TNFα, Rankl, and Opg either directly or indirectly required for the commitment and differentiation of monocyte/macrophage into osteoclasts (21–24). These ligands produced by hypertrophic chondrocytes promote the functional maturation of matrix resorbing chondroclasts and osteoclasts. Runx2^HC/HC mice showed a dramatic reduction in Rankl but no change in the Opg expression, leading to a significant decrease in the Rankl/Opg ratio. Interestingly, the expression of IL-17a which stimulate the production of Mcsf and Rankl was significantly reduced in the Runx2^HC/HC littermates (57–58). Local expression of RANKL by chondrocytes plays a major role in chondroclast/osteoclast development (21). Hypertrophic chondrocytes are the main source of cartilaginous RANKL (21). The significant reduction in the availability of Rankl and IL-17a from hypertrophic chondrocytes may explain the decrease in osteoclast development observed in the adult Runx2^HC/HC mice.

We employed ex vivo culture of bone marrow cells to evaluate the effect of Runx2 deficiency in hypertrophic chondrocytes on the osteoclast differentiation. Consistent with the in vivo findings, the number of TRAP+ osteoclasts were significantly reduced in the Runx2^HC/HC culture, when compared to the bone marrow culture from Runx2^+/+ littermates. In addition, both the size and perimeter of osteoclasts formed in the Runx2^HC/HC culture were significantly reduced. While the expression of Cfos and NfatC1 transcription factors that are required for commitment to the osteoclast lineage were similar, the osteoclast maturation was impaired in the Runx2^HC/HC cultures. The expression of transmembrane DC-Stamp protein that is required for fusion of mononuclear cells into multinucleated osteoclasts was significantly reduced. This data is consistent with the fewer multinucleated osteoclast observed in the Runx2^HC/HC cultures. The expression of early marker Trap was similar, but the Ctsk and Mmp9, considered hallmark of mature osteoclast, were significantly reduced in the Runx2^HC/HC cultures. It is likely that the monocyte-lineage cells developed in the bone marrow niche of Runx2^HC/HC mice are not primed with the necessary signals from Runx2-deficient HCs and their derivatives to differentiate to osteoclasts. While consistent with the in vivo findings, however, the reason for reduced osteoclast maturation in the ex vivo cultures are unclear. The exogenous Mcsf and Rankl were insufficient to promote development of non-adherent monocytes to fully mature osteoclasts. Thus, non-cell autonomous factors are also affecting osteoclast maturation in vivo. Future studies are required to identify HCs-derived priming factors that allow the maintenance and rapid differentiation of osteoclast precursor.

We noted significant increase in articular cartilage in 10-weeks-old mice, but the underlying mechanisms remains unknown. The effect of cartilage retention in preventing age-associated cartilage degradation and osteoarthritis is yet to be explored. Similarly, the contributions of high trabecular bone mass in delaying age-related osteoporosis are yet to be analyzed in the Runx2^HC/HC mouse model. Future studies will define if Runx2 deficiency in hypertrophic chondrocytes exerts these protective effects during skeletal aging. In conclusion, Runx2 functions extend beyond hypertrophic maturation of chondrocytes. Runx2 controls postnatal endochondral ossification by regulating the expression of matrix degrading enzymes and differentiation of osteoclasts.

Highlights.

• Runx2 deficiency in hypertrophic chondrocytes leads to increased growth plate and articular cartilage

• Runx2 controls cartilage turnover by regulating the expression of aggrecan and collagen degrading enzymes in hypertrophic chondrocytes

• Loss of Runx2 in hypertrophic chondrocytes leads to increased trabecular bone mass without affecting osteoblast function

• Runx2 deficiency in hypertrophic chondrocytes impairs osteoclast differentiation and bone resorption