Mesenchymal Deletion of Histone Demethylase NO66 in Mice Promotes Bone Formation

Abstract

Our previous studies indicated that the Jumonji C (JmjC)-domain-containing NO66 is a histone demethylase with specificity for methylated histone H3K4 and H3K36. NO66 binds to the transcription factor Osterix (Osx) and inhibits its transcriptional activity in promoter assays. However, the physiological role of NO66 in formation of mammalian bones is unknown. Here, using a genetically engineered mouse model, we show that during early skeletal development, Prx1-Cre dependent mesenchymal deletion of NO66 promotes osteogenesis and formation of both endochondral as well as intramembranous skeletal elements, leading to a larger skeleton and a high bone mass phenotype in adult mice. The excess bone formation in mice where NO66 was deleted in cells of mesenchymal origin is associated with an increase in the number of preosteoblasts and osteoblasts. Further analysis revealed that in the embryonic limbs and adult calvaria of mice with deletion of NO66 in cells of mesenchymal origin, expression of several genes including bone morphogenetic protein 2, insulin-like growth factor 1 and osteoclast inhibitor osteoprotegerin was increased, concurrent with an increase in expression of bone formation markers such as Osx, type I collagen and bone sialoprotein. Taken together, our results provide the first in vivo evidence that NO66 histone demethylase plays an important role in mammalian osteogenesis during early development as well as in adult bone homeostasis. We postulate that NO66 regulates bone formation, at least in part, via regulating the number of bone-forming cells and expression of multiple genes that are critical for these processes.

Introduction

Mammalian bone is formed via two different mechanisms--intramembranous and endochondral ossification. Both mechanisms involve commitment of mesenchymal precursors that differentiate into osteochondroprogenitors, from which osteoprogenitors or preosteoblasts segregate and then further differentiate into mature osteoblasts and later in osteocytes to form bone tissue. This process integrates expression of diverse signaling molecules including bone morphogenetic protein 2 (Bmp2) and insulin-like growth factor 1(Igf1), as well as key transcription factors such as runt-related transcription factor 2 (Runx2), osterix (Osx), and β-catenin. ^(1-6) Expression of Runx2 is known to begin in the notochord on mouse embryonic day 9.5 (E9.5); it is later restricted to prechondrogenic mesenchymal condensation and chondrocytes and then in osteoblast lineage cells. ^{(3, 6)} Osx expression first appears in the perichondrium/periosteum in mouse embryos on E13, and is essential to formation of osteoblast lineage cells. ^(4-6) In Runx2 deficient mice (Runx2^-/-), no osteoblast differentiation occurs and no endochondral and intramembranous skeletal elements form beyond the stage of cartilage anlagens. ^(1-3) Similarly, in Osx^-/- mice, formation of both endochondral and intramembranous bones was completely abolished. ⁽⁴⁾ Expression of Runx2 and Osx is regulated by a broad signaling network including members of BMP, IGF and Wnt families. ^(7-11)

In recent years, histone methylation has been shown to be of great importance in the control of gene expression. Methylation of H3K4 and H3K36 is associated with gene activation. ^{(12, 13)} It has been demonstrated that histone methylation can be eliminated by demethylases. ^{(14, 15)} Members of the Jumonji C (JmjC)-domain-containing protein family encode a large class of histone demethylases. NO66 is a member of the JmjC-domain-only subfamily. ⁽¹⁴⁾ It was previously described as a component of nucleoli in oocytes of Xenopus laevis. ⁽¹⁶⁾ The physiological role of NO66 remains largely unexplored, however. NO66 was shown to be associated with c-Myc, ⁽¹⁷⁾ was also observed to function as a stress-responsive mediator in Caenorhabditis elegans, ⁽¹⁸⁾ and was identified in human optic nerve head lamina cribrosa cells in response to mechanical strain. ⁽¹⁹⁾ We previously demonstrated that NO66 harbors histone demethylase activity that is specific for methylated H3K4 as well as H3K36, and that NO66 binds to Osx and inhibits its activity in reporter assays. ⁽²⁰⁾ We also showed that knockdown of NO66 in MC3T3 preosteoblasts increases expression of Osx-downstream targets, bone sialoprotein (Bsp) and Osteocalcin (Oc). Moreover, during the induced differentiation of MC3T3 preosteoblasts, NO66 occupancy in the chromatin of Bsp and Oc genes decreases, whereas levels of trimethylated histone H3 at lysine 4 (H3K4me3) and lysine 36 (H3K36me3) in the chromatin of those genes increase.⁽²⁰⁾ It was reported that NO66 can be recruited to the Polycomb Repressive Complex 2 (PRC2) during embryonic stem cell differentiation, leading to loss of H3K36me3 and transcriptional silencing of previously active genes, ⁽²¹⁾ highlighting an important role for NO66 in gene regulation.

To study the physiological role of NO66 in mammalian osteogenesis, we generated conditional knockout mice in which NO66 was inactivated in cells of Prx1-Cre expressing mesenchymal lineage. We found that mesenchymal deletion of NO66 accelerated bone development in embryos, leading to high bone mass in mice at adult stages.

Materials and Methods

Generation and Genotyping of NO66flox and Prx1-Cre; NO66^f/f Mice

A 12-kb fragment containing the NO66 promoter and coding region was retrieved from a BAC (BACPAC Resources Center) and inserted in the pBluescript SK vector via recombineering approach. Using bacterial recombination, a PGK-neo cassette flanked by FRTs and two LoxP sites were introduced into the pBluescript SK vector to generate a NO66-targeting vector (Fig. 1A), which was then electroporated into G4 (C57BL/6Ncr × 129S6) mouse embryonic stem (mES) cells. Integration of the targeting vector into the NO66 locus was confirmed via Southern blot hybridization (Fig. 1C) and polymerase chain reaction (PCR) (Fig. 1D). Two positive mES clones were randomly selected and microinjected into C57BL/6J blastocysts to generate NO66flox mice, which were then crossed with Prx1-Cre mice ⁽²²⁾ to generate Prx1-Cre; NO66^f/+ and Prx1-Cre; NO66^f/f mice. Genotyping of mice see Supplemental Material & Methods.

Fig. 1. Generation of mesenchyme-specific NO66-knockout mice.

A: NO66-targeting vector. B: NO66 mutant allele. C: Southern blot. The genomic DNA of mES cells or mice was digested with 5′ EcoR I (E) or 3′ BamH I (B), and blotted with indicated probes. D and E: PCR. PCR was performed using mouse tail DNA and two pairs of primers as indicated. F: QPCR. qPCR was performed using total RNAs from limbs, calvaria and ventral ribs of mice at different developmental stages (n = 3).

RNA Isolation and Quantitative Real-Time PCR (qPCR)

Total RNA of different murine tissues were isolated using TRIzol reagent (Invitrogen; 15596-018) according to the manufacturer's protocol. QPCRs see Supplemental Material & Methods.

Histology

For histological analyses, paraformaldehyde-fixed, paraffin-embedded embryonic undecalcified or adult decalcified mouse tissue sections were stained with Alcian blue, Safranin O, and hematoxylin & eosin following routine experimental procedures. To examine mineral deposition in bones, long bone and calvarial tissue sections were stained using von Kossa's experimental approach and then counterstained with Nuclear Fast Red as described previously. ^{(4, 5)}

Immunofluorescence and Western Blots

For detection of bone matrix proteins, paraffin-slides were deparaffinized and subjected to epitope retrieval using enzymatic digestion with hyaluronidase (2 mg/mL in phosphate buffer, pH 5.5; MP Biomedicals). For detection of Osx, NO66, H3K4me3 and H3K36me3, tissue slides were boiled in 10 mM sodium citrate buffer (pH 6.0) for 15 min using a microwave oven to expose antigen. Primary and secondary antibodies see Supplemental Material & Methods. For Western blots, nuclear lysates of femurs from embryos at embryonic day 18.5 (E18.5) were isolated using NE-PER Nuclear and Cytoplasmic Extraction kit (Pierce, Rockford, IL) following manufactory instruction. Antibodies against H3K4me3, H3K36me3 and histone H3 were obtained from Abcam (Cambridge, MA). The Western blots were performed following experimental procedure described previously. ^{(23, 24)}

BrdU and Calcein Incorporation

For cell proliferation assay, 5-bromo-2′-deoxyuridine (BrdU) (100 μL/10g body weight; Invitrogen) was injected into pregnant female mice. Two hours later, the mice were sacrificed and analyzed using immunofluorescence staining as described previously. ⁽²⁴⁾ For Calcein labeling, mice were given twice intraperitoneal injection with calcein solution (10 mg/ml, diluted in 2% NaHCO₃, PH 7.4; 10 mg calcein per kg body weight). The second injection was performed five days after the first injection, and two days before the injected mice were sacrificed.

μCT and Histomorphometry

Femurs and calvaria of adult male mice were fixed in 4% paraformaldehyde overnight and stored in 70% ethanol for μCT scanning (eXplore GE Locus SP; GE Healthcare). The resulting images were analyzed using the MicroView software program (version 2.2; GE Healthcare). Histomorphometry was performed using an OsteoMeasure histomorphometry system (OsteoMetrics) (see Supplemental Material & Methods).

Statistical Analysis

All statistical results are presented as the mean ± standard deviation (S.D.). The differences between groups were calculated using the two-tailed Student t-test. P values less than 0.05 were considered statistically significant.

Results

Generation of NO66flox and Mesenchyme-Specific NO66-Knockout Mice

Since NO66 is widely expressed, ⁽¹⁹⁾ we first generated NO66flox mice, in which the single exon of NO66 is flanked by two LoxP sites (Fig. 1A). To delete the NO66 gene in cells of mesenchymal lineage, we generated Prx1-Cre; NO66flox/flox (Prx1-Cre; NO66^f/f) mice, where expression of Cre was driven by the Prx1 promoter (Fig. 1B). We genotyped the NO66^f/f, Prx1-Cre; NO66^f/+ and Prx1-Cre; NO66^f/f mice using Southern blot (Fig. 1C) and PCR assays (Fig. 1 D and E). We performed qPCRs to examine a reduction of NO66 expression in mouse skeleton. We observed a marked decrease in NO66 mRNA expression in whole tissues such as limbs, calvaria, and ventral part of ribs in homozygous Prx1-Cre; NO66^f/f mice at different developmental stages (Fig. 1F). To determine whether NO66 protein was completely depleted from cells of mesenchymal lineage in the homozygous Prx1-Cre; NO66^f/f mice, we performed immunostaining using anti-NO66 antibody on limb and head sections of Prx1-Cre; NO66^f/f and NO66^f/f embryos at embryonic day 15.5 (E15.5) as well as newborn mice at postnatal day 3 (P3). We found that in limbs of E15.5 NO66^f/f control embryos NO66 protein was detectable in all cells of limbs including skins, muscles and long bones (S. Fig. 1C, 1E, 1G). NO66 protein was also detectable in all cells of presumptive calvaria of E15.5 NO66^f/f embryos and calvaria of P3 NO66^f/f mice (S. Fig. 2C, 2G). However, in limbs of E15.5 homozygous Prx1-Cre; NO66^f/f embryos, NO66 protein were no longer detectable in chondrocytes and cells in the perichondral/periosteal regions of developing long bones (S. Figs. 1D, 1F, 1H), where the Prx1 promoter was shown to be active. ⁽²²⁾ As a control, the NO66 protein was still present in skin, muscle and bone marrow of those Prx1-Cre; NO66^f/f embryos (S. Figs. 1D, 1F, 1H), in which the Prx1 promoter was demonstrated to be inactive. ⁽²²⁾ Moreover, we found that NO66 protein was still detectable in some cells of presumptive calvaria of E15.5 Prx1-Cre; NO66^f/f embryos (S. Fig. 2D) and endosteal cells in calvaria of P3 Prx1-Cre; NO66^f/f mice (S. Fig. 2H), indicating an incomplete depletion of NO66 in cells of developing mouse calvaria.

In addition, we performed qPCR assay on heterozygous Prx1-Cre; NO66^f/+ mice. We found that NO66 expression was significantly reduced in limbs and calvaria of the heterozygous Prx1-Cre; NO66^f/+ mice when compared to those of NO66^f/f controls (S. Fig. 3; p < 0.05).

Increase in Accumulation of Endochondral and Intramembranous Bones of Mesenchymal NO66-Knockout Mice during Skeletal Development

Adult male and female homozygous Prx1-Cre; NO66^f/f mice were viable and fertile. We observed that the average body weights and lengths of newborn as well as two-month-old Prx1-Cre; NO66^f/f mice were significantly greater than those of NO66^f/f mice (S. Fig. 4 and S. Fig. 5; p < 0.05). By comparison, we did not observe a significant difference in body weights and lengths between heterozygous Prx1-Cre; NO66^f/+ and NO66^f/f mice (S. Fig. 4 and S. Fig. 5; p > 0.05). Skeleton preparation showed that the ossified regions in skulls and limbs of the homozygous Prx1-Cre; NO66^f/f embryos were larger than those of NO66^f/f littermates on embryonic day 18.5 (E18.5) (Fig. 2), suggesting an increase in intramembranous and endochondral bone formation.

Fig. 2. Skeleton preparation.

A, B, D, E: Skeleton preparation of mouse embryos at E18.5. Squares with dashed lines in A and B show the size of ossified areas in frontal calvaria; brackets in D and E indicate the length of ossified regions in humeri. C: Quantification of the size of ossified areas in frontal calvaria shown in A and B (n = 4). F: Quantification of the length of ossified regions in humeri shown in D and E (n = 8).

To confirm the increase of bone formation, we performed Alcian blue and von Kossa staining of long bone and calvarial sections from mice at different developmental stages. Alcian blue staining showed that in femurs of E14.5 and E15.5 homozygous Prx1-Cre; NO66^f/f embryos, the hypertrophic zone (Fig. 3C) and primary ossification center (Fig. 3G) were larger than those in femurs of NO66^f/f controls (Fig. 3 A and E; S. Fig. 6), implicating an increase in hypertrophic chondrocyte differentiation and primary ossification. von Kossa staining revealed more mineralized tissues in periostea as well as primary ossification center of femurs of homozygous Prx1-Cre; NO66^f/f embryos (Fig. 3 D and H) than in those of NO66^f/f controls (Fig. 3 B and F), suggesting accelerated endochondral bone formation. Consistent with these observations, immunostaining of type I collagen (COL1), a marker of bone tissue, showed more COL1 in the periostea and primary ossification center of femurs of the E14.5 and E15.5 homozygous Prx1-Cre; NO66^f/f embryos (Fig. 3 K, L, O, and P) than in the femurs of NO66^f/f controls (Fig. 3 I, J, M, and N). Similarly, immunostaining of type 10 collagen (COL10), a marker of hypertrophic chondrocytes, showed more COL10 in the hypertrophic zone of femurs in those E14.5 homozygous Prx1-Cre; NO66^f/f embryos (S. Fig. 7B) than in the femurs of NO66^f/f controls (S. Fig. 7A). In addition, von Kossa and immunostaining revealed more mineralized tissues, more COL1 and BSP, another marker of bone formation, in calvaria of newborn homozygous Prx1-Cre; NO66^f/f mice (Fig. 3 T, W and X) than in calvaria of NO66^f/f controls (Fig. 3 R, U and V). By comparison, we did not observe overt differences in Alcian blue, von Kossa and COL1 staining of femurs of the E15.5 heterozygous Prx1-Cre; NO66^f/+ embryos when compared to those of NO66^f/f controls (S. Fig. 8). We therefore selected the homozygous Prx1-Cre; NO66^f/f mice and NO66^f/f controls for further analyses.

Fig. 3. Examination of bone formation in mesenchymal NO66-knockout mice during skeletal development.

A-P: Alcian blue (A, C, E, G), von Kossa (B, D, F, H) and type I collagen (COL1) staining (green, COL1; blue, DAPI) (I-P) of femoral sections of E14.5 and E15.5 NO66^f/f and Prx1-Cre; NO66^f/f embryos. The red squares in A and C show hypertrophic zones; the oval in E and rounded square in G show primary ossification centers; the arrows in B, D, F, and H indicate mineralized tissues; and the arrows in I-P indicate COL1 in periostea. The images in J, L, N, and P are higher magnification of the images in I, K, M, and O, respectively. Q-X: von Kossa (Q-T), COL1 (U and W) and BSP (V and X) antibody staining of frontal calvaria of P1 NO66^f/f and Prx1-Cre; NO66^f/f mice. The staining signals are indicated by arrows. Q and S are sagittal sections of mouse heads in frontal area. R and T are higher magnification of the boxed regions in Q and S. R, U, V and T, W, X are serial sections from the same tissue block, respectively.

A High-Bone-Mass Phenotype in Adult Mesenchymal N066-Knockout Mice

To determine whether the mesenchymal deletion of NO66 continued to lead to the presence of more bones and affect bone homeostasis in adult mice, we used μCT scan to analyze the long bones and calvaria of NO66^f/f and Prx1-Cre; NO66^f/f mice at two months of age. The results revealed greater bone mineral density, percentage of bone volume, and cortical bone thickness in distal femurs of the Prx1-Cre; NO66^f/f mice than in those of controls (Fig. 4 A-D; p < 0.01). Moreover, the calvaria of the Prx1-Cre; NO66^f/f mice were significantly thicker than those of the controls (Fig. 4 E and F; p < 0.01), indicating a high-bone-mass phenotype. Von Kossa and Safranin O staining showed more mineralized tissues or trabeculae in the distal femurs of those Prx1-Cre; NO66^f/f mice than in those of the controls (Fig. 5 A-D). Furthermore, bone histomorphometry indicated that in the Prx1-Cre; NO66^f/f mice, the number of osteoblasts was increased, whereas the number of osteoclasts was decreased (Fig. 5 I and J; Tables S1 and S2; S. Fig. 9), resulting in an increased bone-formation rate (Fig. 5 E-H and K; Table S1). These results suggested that ablation of NO66 in Prx1-Cre expressing cells continued to affect adult bone homeostasis, leading to a high bone mass phenotype.

Fig. 4. Micro-computed tomography (μCT).

A: μCT images of femurs in male mice at 2 months of age (2m). B-D: Quantification of the μCT images shown in A. The bone mineral density (BMD), percentage of bone volume (BV) versus total volume (TV) and thickness of cortical bones are shown. E: μCT images of the frontal calvaria of 2m mice. F: Quantification of the μCT images shown in E. (n = 5)

Fig. 5. Bone histological and histomorphometry in adult mesenchymal NO66-knockout mice.

A-D: von Kossa (A and B) and Safranin O (C and D) staining of distal femur sections of 2m male NO66^f/f and Prx1-Cre; NO66^f/f mice. E-H: Double fluorochrome calcein lines in femurs of 2m NO66^f/f and Prx1-Cre; NO66^f/f mice. I-K: Histomorphometry. (I) Number of osteoblasts (N.Ob) per total area (T.Ar) analyzed; (J) Number of osteoclasts (N.Oc) per total area (T.Ar) analyzed; (K) Bone formation rate (BFR) per bone surface (BS) analyzed per year. (n = 4)

Increase in Number of Osx-Expressing Preosteoblasts and Osteoblasts in Endochondral and Membranous Bones of Mesenchymal NO66-Knockout Mice

In our previous in vitro cell culture study, we showed that NO66 knockdown in MC3T3 preosteoblasts accelerated osteoblast differentiation. ⁽²⁰⁾ Because the adult Prx1-Cre; NO66^f/f mice had increased numbers of osteoblasts, we wanted to know whether it was due to an increase in the number of preosteoblasts, or an increase in the differentiation of preosteoblasts into mature osteoblasts. During early skeletal development in mouse embryos, generation of osteoprogenitors or preosteoblasts first occurs in the perichondrium and periosteum in long bones and presumptive calvaria on about E13. Since Osx is highly expressed in cells of osteoblast-lineage, we performed immunostaining of Osx to mark the preosteoblasts and osteoblasts in periosteal regions of mouse long bones and calvaria. The staining results demonstrated that the numbers of Osx-positive cells in periosteal regions of femurs in Prx1-Cre; NO66^f/f embryos were significantly higher than those in these regions in their NO66^f/f littermates at E14.5, E15.5, and E18.5 (Fig. 6 A-L; Table 1; p < 0.01), suggesting an increase in the number of preosteoblasts and/or osteoblasts. We observed similar results in calvaria of newborn Prx1-Cre; NO66^f/f and NO66^f/f mice (Fig. 6 M-P; Table 1).

Fig. 6. Immunostaining for Osx.

A-L: Femur sections of NO66^f/f (A, B, E, F, I, J) and Prx1-Cre; NO66^f/f (C, D, G, H, K, L) embryos on E14.5, E15.5, and E18.5. M-P: Frontal calvarial sections of NO66^f/f (M, N) and Prx1-Cre; NO66^f/f (O, P) mice at postnatal day 3 (P3). B, D, F, H, J, L, N and P are higher magnification images of the boxed areas in A, C, E, G, I, K, M and O, respectively. The yellow and white arrows indicate Osx-positive cells in periosteal regions of femurs and calvaria. (Red, Osx; Blue, DAPI)

Table 1. Percentage of Osx-Positive Cells per Area Analyzed.

	Mean percentage ± S.D.
Mice	E14.5	E15.5	E18.5	P3 (calvaria)
NO66f/f	18.0 ± 2.5	30.1 ± 2.7	40.5 ± 3.1	12.9 ± 1.9
Prx1-Cre; NO66f/f	38.3 ± 4.5*	48.9 ± 3.9*	60.6 ± 4.1*	21.8 ± 2.6*

^*p < 0.01; n = 3

Preosteoblasts are proliferating cells and predominantly locate in the periosteal regions of long bones and calvaria. To determine whether the number of proliferating cells were increased in the periosteal regions of long bones and calvaria of E15.5 Prx1-Cre; NO66^f/f embryos, we performed BrdU incorporation. We found more BrdU-positive cells in the perichondrium and periosteum in femurs of E15.5 Prx1-Cre; NO66^f/f embryos than in NO66^f/f controls (Fig. 7 A, B and E; p < 0.05). We observed similar results in the presumptive calvaria of those Prx1-Cre; NO66^f/f and NO66^f/f embryos (Fig. 7 C-E; p < 0.05). These results indicated that deletion of NO66 in Prx1-Cre expressing mesenchymal cells increased the number of proliferating cells in the periosteal region and in calvaria. This increase is also consistent with the observed increase in the number of preosteoblasts.

Fig. 7. BrdU Incorporation and qPCR Assays.

A-D: BrdU Incorporation. The femur (A, B) and head (C, D) sections of E15.5 NO66^f/f and Prx1-Cre; NO66^f/f embryos were stained using anti-BrdU antibody (red, BrdU; blue, DAPI). The yellow and white arrows indicate the BrdU-positive cells in the boxed periosteal areas of femurs and presumptive frontal calvaria. E: Percentage of BrdU-positive cells in the boxed areas shown in A-D. F: Results of qPCRs using total RNAs from limbs of E13.5 NO66^f/f and Prx1-Cre; NO66^f/f embryos (* p < 0.05; ** p < 0.01; n = 3). G: Results of qPCRs using total RNAs from calvaria of 1m male NO66^f/f and Prx1-Cre; NO66^f/f mice (* p < 0.05; ** p < 0.01; n = 3).

Increase in Expression of Genes in Embryonic Limbs and Adult Calvaria of Mesenchymal NO66-Knockout Mice

Formation of normal bones depends upon a balance between osteoblast-involved bone formation and osteoclast-mediated bone resorption. The generation and function of osteoblasts are regulated by diverse signals, including BMP, IGF and Wnt signaling pathways. ⁽²⁵⁾ At the same time, osteoblasts produce osteoprotegerin (Opg); a decoy inhibitor of receptor activator of nuclear factor-κB ligand (Rankl), which itself is an activator of osteoclastogenesis and in bone resorption. ⁽²⁶⁾ To determine whether an acceleration of primary ossification in the limbs of NO66 mutant embryos could be associated with changes in expression of marker genes and their upstream signaling molecules, we performed qPCRs using total RNA obtained from limbs of E13.5 NO66^f/f and Prx1-Cre; NO66^f/f embryos (Fig. 7 F). We found that expression of Bmp2, Bmpr1b, and Igf1 mRNA was upregulated concurrently with elevated expression of osteoblast and hypertrophic chondrocyte differentiation markers, such as Runx2, Osx, ALP, Col1 (Col1a1), and Col10, in limbs of E13.5 Prx1-Cre; NO66^f/f embryos when compared to those in their controls. By comparison, expression of Tcf, a downstream effector of Wnt signaling, and Ihh was not significantly changed (Fig. 7 F; p > 0.05).

To further confirm whether expression of the above genes was increased in osteoblast-lineage-enriched calvaria of the adult NO66 mutant mice, we performed qPCRs using total RNA obtained from calvaria of one-month-old NO66^f/f and Prx1-Cre; NO66^f/f mice (Fig. 7 G). We again observed an upregulaton of expression of Bmp2, Igf1 and Osx, but not Bmpr1b, Igf1r and Runx2 (Fig. 7 G; p > 0.05); this upregulation was accompanied by an increase in the expression of bone formation markers, including ALP, Cola1 and Bsp, in calvaria of the Prx1-Cre; NO66^f/f mice (Fig. 7 G; p < 0.05). Moreover, expression of Opg was markedly increased (Fig. 7 G; p < 0.01), whereas expression of Rankl did not change significantly in calvaria of these Prx1-Cre; NO66^f/f mice (Fig. 7 G; p > 0.05). By comparison, expression of TRAP (Tartrate-resistant acid phosphatase), a marker of osteoclast activity, was downregulated in calvaria of the Prx1-Cre; NO66^f/f mice (Fig. 7 G; p < 0.05). These results suggested that NO66 might negatively regulate expression of Bmp2, Igf1 and Opg to control bone formation and bone resorption.

Increase in Levels of H3K4me3 and H3K36me3 in Long Bones of Mesenchymal NO66-Knockout Mice

We previously showed that NO66 has histone demethylase activity for H3K4me3 and H3K36me3. ⁽²⁰⁾ To determine whether mesenchymal deletion of NO66 altered cellular levels of these chromatin marks, we performed immunostaining using an antibody against H3K4me3 or H3K36me3 on femur sections obtained from NO66^f/f and Prx1-Cre; NO66^f/f mouse embryos at E15.5 and E18.5. We observed increased staining of H3K4me3 and H3K36me3 in proliferating and prehypertrophic chondrocytes as well as cells in perichondral/periosteal regions of femurs from E15.5 or E18.5 Prx1-Cre; NO66^f/f embryos, when compared to those from NO66^f/f controls (S. Fig. 10). To further confirm this observation, we performed Western blots to measure levels of H3K4me3 and H3K36me3 in femurs of E18.5 NO66^f/f and Prx1-Cre; NO66^f/f embryos. We observed the increased levels of H3K4me3 and H3K36me3 in femurs of E18.5 Prx1-Cre; NO66^f/f embryos (S. Fig. 11), consistent with the immunostaining results.

Discussion

In this study, we used a genetically engineered mouse model to study the in vivo function of NO66 in bone formation. Our results showed first that deletion of NO66 in Prx1-Cre-expressing mesenchymal cells leads to an increase in endochondral bone development. This included earlier hypertrophic chondrocyte differentiation and subsequent bone formation, suggesting that the entire process of endochondral bone development might be accelerated. At E18.5, membranous bone ossification was also markedly increased which could be consistent with the hypothesis that this process too was accelerated. The increased number of preosteoblasts and osteoblasts in the NO66 mutant embryos at defined time points could be a consequence of the acceleration of endochondral bone development. We hypothesize that the persistence of increased numbers of osteoblast lineage cells in adult animals leads to a high-bone-mass phenotype. In the NO66 mutant embryos, there was also a higher level expression of several genes, including Bmp2, Igf1, Osx and Opg. These in vivo observations suggested that NO66 plays an important physiological role in bone formation probably by controlling the timing of bone development as well as the number of osteoblast lineage cells, and also by regulating the expression of a number of genes that are critical for these processes.

Our previous in vitro study reported that knockdown of NO66 in MC3T3 preosteoblasts increased osteoblast mineralization, suggesting a negative regulator of NO66 in preosteoblast differentiation. ⁽²⁰⁾ What distinguishes the present in vivo findings from these in vitro results is that in vivo the developmental process of endochondral bone formation and maybe that of the membranous skeleton are accelerated, and as a possible result of this acceleration, the number of Osx-positive cells is increased. We therefore postulate that NO66 not only regulates preosteoblast differentiation but also plays an important role in the in vivo commitment or differentiation of osteochondroprogenitors. ⁽⁶⁾ We also hypothesize that in Prx1-Cre; NO66^f/f embryos, the increase in the number of Osx-expressing cells, which represent preosteoblasts and osteoblasts in the periostea and calvaria, is likely responsible for the increased bone formation in these embryos. Moreover, the increased number of osteoblasts measured by histomorphometry in adult Prx1-Cre; NO66^f/f mice was also consistent with the increased numbers of Osx-positive cells in bones during embryonic development. However, we note that ex vivo cultured osteoblasts from NO66^f/f and Prx1-Cre; NO66^f/f mice did not reveal significant differences in differentiation of the cells (data not shown), which was attributed to a mixed population of cells due to incomplete deletions of NO66. Therefore, whether or not the effects of NO66 observed in the Prx1-Cre; NO66^f/f mice are cell-autonomous remains to be determined.

Although we do not yet know which specific molecular mechanisms control the acceleration of endochondral bone development as well as the increased number of preosteoblasts/osteoblasts in bones of Prx1-Cre; NO66^f/f embryos and adult mice, several signaling molecules may have a role in these controls. We measured the mRNA expression for only a limited number of genes in limbs and calvaria of NO66^f/f and Prx1-Cre; NO66^f/f mice in this study, and observed a consistent increase in Bmp2, Igf1 and Osx mRNA expression in the Prx1-Cre; NO66^f/f mice. Because BMP2 and IGF1 are important signaling molecules that can induce expression of Osx,^{(7, 8, 27)} we speculate that upregulation of Bmp2 and Igf1 expression in bones of the Prx1-Cre; NO66^f/f mice may contribute, at least in part, to the elevated expression of Osx and an increase in the number of Osx-expressing preosteoblasts/osteoblasts. Nevertheless, other signaling molecules may have similar roles in the regulation of Osx expression and the number of Osx-expressing cells. We previously found that expression of Osx mRNA was not affected by shRNA knockdown of NO66 in MC3T3 preosteoblasts. ⁽²⁰⁾ This led us to hypothesize that the in vivo increased expression of Bmp2 and Igf1 might be responsible for the concurrent elevation in Osx expression in the NO66 mutant mice. The increase in Osx mRNA should also amplify the effects of the larger number of Osx-expressing cells in bone formation. Whether NO66 directly regulates expression of Bmp2 and Igf1 remains to be defined.

It has been known that the osteoblasts produce both OPG and Rankl. ⁽²⁶⁾ Since expression of Opg, but not Rankl mRNA, was increased in calvaria of the Prx1-Cre; NO66^f/f mice, it implied that NO66 negatively regulates expression of Opg gene. Moreover, because OPG is an inhibitor of osteoclastogenesis, ⁽²⁶⁾ the finding that Opg mRNA was increased in NO66 mutant calvaria, but not that of Rankl, may explain the decreased osteoclast numbers in those mice. The lower osteoclast numbers likely contributed to the increased bone mass in Prx1-Cre; NO66^f/f mice. Given that Opg expression is not directly regulated by Osx, ⁽²⁸⁾ upregulation of Opg mRNA expression observed in calvaria of the Prx1-Cre; NO66^f/f mice is less likely secondary to the increased Osx activity. Because TRAP is an important marker of osteoclasts, we speculate that the lower number of osteoclasts is likely responsible for the moderate decrease in TRAP mRNA expression in bones of the Prx1-Cre; NO66^f/f mice.

We previously reported that NO66 has histone demethylase activity for H3K4me3 and H3K36me3. ⁽²⁰⁾ The increased levels of H3K4me3 and H3K36me3 in bones of the Prx1-Cre; NO66^f/f embryos are consistent with this previous report. Since H3K4me3 and H3K36me3 are regarded as two marks of transcriptionally active chromatin, it is likely that NO66 controls expression of genes, such as Bmp2, Igf1 and/or Opg, via a change in histone methylation state in the chromatin of these genes. However, at the moment we do not yet know whether the increased levels of H3K4me3 and H3K36me3 in the Prx1-Cre; NO66^f/f embryos were solely or directly caused by depletion of NO66. Given that the methylation state of histone H3 at lysine 4 and lysine 36 can be affected by many other histone demethylases, it is possible that deletion of NO66 interferes with the activity of other histone demethylase(s) which may then contribute to the increased levels of H3K4me3 and H3K36me3 in those embryos. Nevertheless, whether NO66 affects other histone demethylase(s) and whether NO66-deletion-mediated increases in H3K4me3 and H3K36me3 are cell-autonomous still remain to be determined.

Taken together, our data provide evidence that NO66 plays an important role in osteogenesis during mouse development as well as bone homeostasis in adult mice. NO66 controls osteogenesis by regulating the process of endochondral and membranous skeletal development, by controlling the number of preosteoblasts/osteoblasts, and the expression of a number of genes which are important for bone formation. Identification of the NO66 downstream targets will be a main focus in a future study.

Supplementary Material

Supplemental

S. Fig.1. Immunostaining of NO66 in long bones. Immunostaining using anti-NO66 antibody on limb sections of NO66^f/f (A, C, E, G) and Prx1-Cre; NO66^f/f (B, D, F, H) embryos at embryonic day 15.5 (E15.5). A and B, bright field images of paraffin sections of tibias. C – H, higher magnification staining images of the different parts in the boxed regions in A and B. Blue, DAPI; Pink, NO66. White dashed lines in C and D indicate the edges of tibia heads. Yellow arrow head in H points to NO66 positive cells in primary ossification center of tibia; white arrow head indicates NO66 positive cells in adjacent muscle tissue. Rectangles in G and H show periosteal regions of tibias. Sk, skin; Kj, knee joint; Tb, tibia; Ms, muscle. Epi, epiphysis; Met, metaphysis; Dia, diaphysis.

S. Fig.2. Immunostaining of NO66 in calvaria. Immunostaining using anti-NO66 antibody on head sections of NO66^f/f (A, C, E, G) and Prx1-Cre; NO66^f/f (B, D, F, H) embryos at E15.5 (A -D) and newborn mice at postnatal day 3 (P3) (E - H). A, B and E, F, paraffin sections of heads of embryos and newborn mice. C, D and G, H are higher magnification staining images of the boxed areas in A, B and E, F (Blue, DAPI; Pink, NO66). Yellow dashed lines in C and D indicate the boundary between skin tissues and presumptive calvaria. White arrow in D points to the NO66-positive cells in the presumptive calvaria.

S. Fig.3. QPCR assay. Comparison of mRNA expression of NO66 in limbs of E15.5 NO66^f/f and Prx1-Cre; NO66^f/+ embryos, as well as calvaria of newborn mice at postnatal day 1 (P1). (n = 3)

S. Fig.4. Changes in body weights and limb lengths of newborn mice. The body weights (A) and hindlimb lengths (B) of newborn NO66^f/f (n = 14), Prx1-Cre; NO66^f/+ (n = 12) and Prx1-Cre; NO66^f/f (n = 11) at postnatal day 3 (P3) were measured (both males and females were included in each group).

S. Fig.5. Changes in body weights and lengths of adult mice. The body weight (A) and length (B) of male NO66^f/f (n = 10), male Prx1-Cre; NO66^f/+ (n = 9) and male Prx1-Cre; NO66^f/f (n = 9) mice at two months old were measured.

S. Fig.6. Quantification of the lengths of hypertrophic zones in femurs of mouse embryos. The lengths of hypertrophic zones in femur sections of NO66^f/f (n = 3) and Prx1-Cre; NO66^f/f (n = 3) embryos at E14.5 and E15.5 were measured under microscope.

S. Fig.7. COL10 staining. The immunostaining of type 10 collagen (COL10) was performed on femur sections of NO66^f/f (A) and Prx1-Cre; NO66^f/f (B) embryos at embryonic day 14.5 (E14.5). Blue, DAPI; Green, COL10.

S. Fig.8. Alcian blue (A, B), von Kossa (C, D) and immunostaining of type I collagen (COL1) (E, F) on femoral sections of E15.5 NO66^f/f (A, C, E) and Prx1-Cre; NO66^f/+ (B, D, F) embryos.

S. Fig.9. Double staining of TRAP and von Kossa. The MMA-embedded femur slides of adult NO66^f/f (A) and Prx1-Cre; NO66^f/f (B) mice were stained with von Kossa, followed by TRAP staining to visualize trabecular bones (in black) and lay down osteoclasts (in red, indicated by yellow arrows).

S. Fig.10. Immunostaining of H3K4me3 and H3K36me3. The immunostaining using anti-H3K4me3 or anti-H3K36me3 antibody was performed on femur sections of NO66^f/f (A, C) and Prx1-Cre; NO66^f/f embryos (B, D) at E15.5 (A, B) and E18.5 (C, D). Blue, DAPI; Pink, anti-H3K4me3 or H3K36me3. Prolif, proliferating zone of chondrocytes; PreH, prehypertrophic zone of chondrocytes. White arrows indicate cells in perichondrium/periosteum of femurs. (E) Percentage of strong positive cells for H3K4me3 or H3K36me3 versus total cell nuclei in each selected area as indicated in the figure. (n = 3)

S. Fig.11. Western blots. (A) The cell nuclear lysates were isolated from femurs of the E18.5 NO66^f/f or Prx1-Cre; NO66^f/f embryos (n = 3), and blotted with antibody against H3K4me3, H3K36me3 or histone H3. (B) The blotting images were quantified using NIH Image J software. The relative ratio corresponds to the measured intensity of H3K4me3 or H3K36me3 versus that of histone H3 shown in A.

Table S1. Histomorphometry analysis of bone formation parameters 

Table S2. Histomorphometry analysis of osteoclasts