β₂-adrenoceptor signaling in chondrocytes regulates long bone growth, bone homeostasis, and fracture healing in mice

Abstract

Catecholamines are known to interact with bone and cartilage cells via binding to adrenergic receptors. Among these, the β₂-adrenoceptor (β₂-AR) plays a key role in mediating the effects of catecholamines on bone. Mice lacking the β₂-AR systemically or specifically in osteoblasts show an increased bone mass. Previous studies further indicated an important influence of catecholamines on transdifferentiation of chondrocytes towards osteoblasts during endochondral ossification. Therefore, in this study, cartilage and bone phenotype as well as fracture healing of mice with a specific knockout of the β₂-AR in chondrocytes were investigated. Tibiae and femora of 6-, 12-, and 40-wk-old male and female mice were analyzed. The knockout resulted in a reduced long bone growth, most likely due to a faster transition of proliferative to hypertrophic chondrocytes. The older knockout mice additionally showed an osteopenic bone phenotype due to a reduced number of osteoblasts. Fracture healing after a standardized femur osteotomy was delayed, showing reduced cartilage area at an intermediate time point during fracture healing. Gene expression analysis in an additional in-vitro-experiment identified pathways like “Wnt-signaling” and “TGFβ-signaling” to be involved. In conclusion, our data showed an important role of the β₂-AR specifically in chondrocytes during long bone growth, bone homeostasis in older animals and fracture healing.

Graphical Abstract

Graphical Abstract.

Introduction

The human body has over 200 bones that stabilize the skeleton, enable the upright gait and protect internal organs mechanically. Besides, bones are a crucial place for calcium and phosphate storage as well as for hematopoiesis.^1,2 Therefore, bones are a highly metabolically active and dynamic tissue. Bone growth of long bones like tibia or femur takes place in the growth plate via proliferation and hypertrophy of chondrocytes.³  Bone homeostasis with the aim of a constant bone mass is maintained by the balanced activity of bone-forming osteoblasts and bone-degrading osteoclasts.⁴  Fracture healing via secondary bone healing takes place in the fracture callus in three phases with fluid transitions: the inflammation phase, the repair phase, and the remodeling phase. During secondary fracture healing and long bone growth, a cartilaginous template is formed which is then transformed into bone via endochondral ossification.⁵ This process is heavily influenced by transdifferentiation of chondrocytes towards osteoblasts. Long bone growth, bone homeostasis, and fracture healing are all highly regulated processes influenced by different factors like biomechanical loading, cytokines, chemokines, and hormones.⁴ Besides these factors, neurotransmitters like catecholamines are also known to have an influence on bone. It has already been shown that catecholamine-producing tumors like pheochromocytoma and paraganglioma lead to a deterioration of bone quality and increase the risk for fractures.⁶ Furthermore, increased local catecholamine signaling caused by chronic psychosocial stress results in a reduced long bone growth and an impaired fracture healing in mice due to an impaired endochondral ossification.^7,8 Also, decreased local catecholamine signaling under physiological conditions leads to a reduced long bone growth, an impaired bone homeostasis, and a delayed fracture healing in mice.⁹  This suggests that a certain level of catecholamines is necessary for the balanced regulation of long bone growth, bone homeostasis, and fracture healing. There are already several studies investigating the question of which receptor mediates the effects of catecholamines on bone. It has already been shown that on a cellular level, different subtypes of adrenoceptors (α₁, α₂, β₁, β₂, β₃) are expressed by bone and cartilage cells like osteoclasts, osteoblasts, osteocytes, and chondrocytes.^9–13 In recent years, it became more and more evident that β-adrenoceptors, especially the β₂-adrenoceptor (β₂-AR), play an important role in mediating the effects of catecholamines on bone. Clinical evidence is given by the fact that β-AR-blockers like propranolol prevent bone mass loss and therefore reduce the risk for osteoporosis and fractures.^14,15 Preclinical studies could show that β₂-AR-signaling in osteoblasts inhibits bone formation and induces RANKL-mediated bone resorption.^14,16–18 In line with this, mice with a systemic or an osteoblast-specific knockout of the β₂-AR show an increased bone mass due to an increased osteoblast and a decreased osteoclast activity.^16,18 However, a more recent study suggested that increased β₂-AR signaling promotes bone regeneration.¹⁹ None of the previous studies shed light on the specific role of the β₂-AR on chondrocytes. Therefore, the aim of this study was to investigate the role of the β₂-AR in chondrocytes during long bone growth, bone homeostasis, and fracture healing. Therefore, a mouse line with a specific knockout of the β₂-AR in chondrocytes was bred and the effects of this knockout were investigated.

Materials and methods

Animals

Male and female Adrb2^flox/flox/Col2a1-Cre⁺ mice with a specific knockout of the β₂-adrenoceptor in chondrocytes (β₂-AR-KO) were analyzed at the age of 6, 12, 40, and 52 wk (C57BL/6 strain; referred to as Adrb2^Col2a1-Cre). They were generated by crossing Adrb2^flox/flox mice (Adrb2tm1Kry; kindly provided by Prof. Karsenty, Department of Genetics & Development, Columbia University Medical Center, New York, USA)²⁰ with Col2a1-Cre mice (B6; SJL-Tg (Col2a1-cre)1Bhr/J) (JAX stock #003554).²¹ The Cre⁻ littermates served as control mice (C57BL/6 strain; referred to as Adrb2^flox/flox). Genotyping of the animals was performed with digested ear punches via PCR using the following primer pairs: Col2a1-Cre (Forward primer sequence (F): 5’-GAGTGATGAGGTTCGCAAGA-3′, Reverse primer sequence (R): 3’-CTACACCAGAGACGG-5′); Adrb2flox (F: 5’-CCAAAGTTGTTGCACGTCAC-3′, R: 3’-GCACACGCCAAGGAGATTAT-5′). All mice were kept under constant standard laboratory conditions (12 h light/dark cycle, 60% humidity, 22 °C room temperature [RT]) in cages with up to five mice. All mice had access to tap water and mouse feed (ssniff, V 1534-000 R/M-H, Ssniff GmbH, Soest, Germany) ad libitum. Each experiment of this study was performed considering international guidelines on the ethical use of animals (ARRIVE guidelines).²² No adverse events were recorded during the study.

Study design

Male and female mice were euthanized at either 6 (Set 1, n = 4), 12 (Sets 2 and 3, n = 6), or 40 wk (Sets 4 and 5, n = 6) of age. Intact femora and tibiae were removed, the bone lengths were measured with an electric caliper, and the lumbar spine was investigated. In addition, the joint cartilage of the femora was examined in 52-wk-old female mice (Set 6, n = 6). To investigate fracture healing, two other sets of 12-wk-old male mice underwent a standardized femur osteotomy and were sacrificed either 10 (Set 7, n = 6) or 21 d (Set 8, n = 6) after fracture. With the intact and fractured femora, micro-computed tomography (μCT) and several histological and immunohistochemical analyses were performed. Besides, flexural rigidity of the fractured femora was quantified using a non-destructive 3-point bending test as described previously.²³ Briefly, fractured femora were loaded with a maximum force of 2 N using a testing machine (Zwick Roell, Ulm, Germany). Then, flexural rigidity was determined with the slope of the recorded load-deflection curve. The study design was approved by the Federal Animal Care and Use Committee of the local government (Regierungspräsidium Tübingen, Germany; licenses 1437, o.135-10, o.135-19).

Femur osteotomy

Twelve-week-old male mice (Sets 7 and 8) underwent a standardized osteotomy of the right femur diaphysis stabilized with a semi-rigid external fixator (RISystem, Davos, Switzerland) as described previously.²³ Anesthesia was maintained with 2% isoflurane. Tramadol (25 mg/L) via drinking water was used as analgesia from one day before until three days after the osteotomy. The postoperative weight loss of all mice was less than 10%.

μCT analysis

Intact femora, fractured femora, and the lumbar spine of the mice (Sets 2-5 and 8) were scanned with a SkyScan 1172 scanning device (Bruker, Kontich, Belgium) using 50 kV, 200 mA and a voxel resolution of 8 μm. The analysis of the trabecular bone in the intact femora was performed in a 280 μm high volume of interest in the femoral metaphysis with a distance of 360 μm from the distal growth plate. The cortical bone was analyzed in an 80 μm high volume of interest located under the third trochanter in the femoral mid-diaphysis. The volume of interest in the fractured femora was defined as the entire fracture callus in the fracture gap (FG). The analysis of the lumbar spine was performed in the L5 vertebral body and the volume of interest included a cylinder with a diameter of 0.8 mm located in the center of the vertebral body. Two phantoms with 250 mg/cm³ and 750 mg/cm³ of hydroxyapatite (HA) were used to calibrate tissue mineral density (TMD). To differentiate between mineralized and non-mineralized tissue, 394 mgHA/cm³ was set as threshold for trabecular and 642 mgHA/cm³ for cortical bone. Bony bridging score of the fractured femora was assessed in two perpendicular planes. One point was given for each bridged cortex, with a maximum achievable score of 4. A score of 3 or 4 signified a successful fracture healing. Micro-computed tomography analysis was performed regarding the guidelines of the American Society for Bone and Mineral Research.²⁴

Histomorphometric analysis

Intact and fractured femora were fixed in 4% formaldehyde for 48 h before embedding in paraffin to obtain decalcified histological sections. Toluidine blue staining was used to determine the growth plate thickness, the growth plate zones, the osteoblast number per bone perimeter (N.Ob/B.Pm) and the osteoblast surface per bone surface (Ob.S/BS) in the distal femoral metaphysis and in the fracture callus. Tartrate-resistant acid phosphatase-(TRAP)-staining was performed to analyze the osteoclast number per bone perimeter (N.Oc/B.Pm) and the osteoclast surface per bone surface (Oc.S/BS) in the same regions of interest. All parameters were measured with the Osteomeasure software (OsteoMetrics, Decatur, GA, USA). The tissue composition of the fracture callus in the FG was determined in Safranin O/Fast Green-stained decalcified histological sections of the fractured femora using Leica Application Suite X software (Leica, Wetzlar, Germany). To assess the severity and histological extent of potential osteoarthritic changes, the Osteoarthritis Research Society International Score (OARSI Score) of the joint cartilage of the femora was analyzed in histological sections stained with 1,9-dimethyl-methylene blue (DMMB, Sigma-Aldrich, Munich, Germany) using Fiji (ImageJ) software (version 1.52p).

Immunohistochemical and immunofluorescence analysis

Decalcified histological sections of the intact femora were used for immunohistochemical staining of Collagen 2⁺ (rabbit anti-mouse Col2, 1:100, #600-401-104-0.1, Rockland), Collagen 10⁺ (rabbit anti-mouse Col10, 1:200, #ABIN1077945, Antibodies online), Runx2⁺ (rabbit anti-mouse Runx2, 1:50, #8486, Cell Signaling Technology), Proliferating Cell Nuclear Antigen (PCNA⁺) (rabbit anti-mouse PCNA, 1:100, #MA5-32051, Thermo Fisher Scientific) and β₂-AR⁺ (rabbit anti-mouse β₂-AR, 1:100, #bs-0947R, Bioss) chondrocytes in the growth plates. Species-specific non-targeting immunoglobulins (rabbit IgG, 011-000-003, Jackson ImmunoResearch) were used as negative controls and goat anti-rabbit IgG-biotin (1:200, B2770, Invitrogen) was used as secondary antibody. First, the histological sections of the bones were deparaffinized, rehydrated, and demasked in 10 mM citrate buffer (Col2, Col10, Runx2, β₂-AR) or tris-EDTA-solution (PCNA). Samples were then blocked with 3% peroxidase and incubated with 5% goat serum (Col2, Col10, Runx2, β₂-AR) or 5% BSA (PCNA) for one hour at RT. Thereafter, the samples were incubated overnight at 4 °C with the primary antibodies and the negative controls. After incubation with the secondary antibodies for 30 min (Col2) or 1 h (Col10, Runx2, PCNA, β₂-AR) at RT, staining was continued using the VECTASTAIN Elite ABC-HRP Kit (#PK-6100, Vector Laboratories) and NovaRED (#SK-4800, Vector Laboratories) as chromogen. Hematoxylin (1:2000, #2C-306, Waldeck) was used for counterstaining of the sections. The quantification of the chondrocytes and the analysis of the different zones of the growth plate was performed with Osteomeasure software (OsteoMetrics, Decatur, GA, USA).

For immunofluorescence double staining of Collagen 2 (goat anti-mouse Col2, 1:50, #1320-01, Southern Biotech) and β₂-AR (rabbit anti-mouse β₂-AR, 1:50, #PA5-77283, Thermo Fisher Scientific), decalcified sections of intact and fractured femora were used. Following deparaffinization, rehydration, demasking, and blocking with 5% BSA as described above, sections were incubated with the primary antibodies overnight at 4 °C. Thereafter, secondary antibodies were applied for 1 h to the sections (donkey anti-rabbit, 1:250, #A21207, Invitrogen; donkey anti-goat, 1:200, #SAB3700288-2, Sigma-Aldrich). Finally, sections were incubated with FITC-streptavidin (1:200, #405202, BioLegend) for 30 min, followed by counterstaining with Hoechst for 1 min.

Decalcified histological sections of the intact femora were also used for immunohistochemical staining of Collagen 2⁺ ([Ab-1] mouse monoclonal antibody [II-4C11, Sigma Aldrich], diluted 1:250 in 1% BSA/TBS) and Collagen 10⁺ (rabbit anti-mouse Col10, 1:100, #MBS2004541, MyBioSource, diluted 1:100 in 1% BSA/TBS) chondrocytes in the femoral joint cartilage. After deparaffinization and rehydration of sections, antigen unmasking was performed using pepsin (0.025% in 0.2 N HCl in 1xTBS for 15 min, 37 °C), hyaluronidase (500 U/mL in 1xTBS for 30 min, 37 °C), and proteinase K (10 μg/mL in 1xTBS, 10 min, 55 °C) for type II collagen detection or pepsin (0,1% in 0.5 M acetic acid, P7125 Sigma Aldrich, 2 h 37 °C) for type X collagen staining. Endogenous peroxidases were quenched with H₂O₂ (3% in 1xTBS, 10 min, 37 °C) and unspecific antibody binding was blocked using 5% goat serum (in 1xTBS for 60 min at 37 °C). The sections were then incubated with the primary anti-collagen type II or anti-collagen X antibody overnight at 4 °C. After incubation with the secondary antibody (ZytoChem plus [HRP] polymer anti-mouse [ZUC050-006, Zytomed systems]) for 60 min at RT, staining was visualized using 3,3′-diaminodbenzidine (0.01 M in 1xPBS, pH 7.2, Sigma Aldrich with 3% H₂O₂) for 2-5 min.

Cell culture, gene expression analysis, and RNA-sequencing

ATDC5 cells were seeded at a density of 2500 cells/cm² and cultured at 6% O₂ for 7 and 10 d, respectively, in chondrogenic medium (DMEM/F12 [1:1] [Gibco]), supplemented with 5% FCS (Merck Millipore), 1% L-glutamine (Gibco), 1% penicillin/streptomycin (Gibco), human transferrin (10 μg/mL, Sigma), sodium selenite (30 nM), human insulin (10 μg/mL, Sigma), and ascorbate 2-phosphate [0.2 mM, Sigma]). Medium was changed two times per week. Adrenaline or Adrenaline in combination with the selective β₂-AR-blocker Butoxamine was added. Respective DMSO concentrations were used as the vehicle control. After RNA isolation with the PureLink RNA Mini Kit (Thermos Fisher Scientific), one-step Real-Time-PCR (RT-PCR) was performed using the SensiFAST SYBR Hi-ROX One-Step Kit (Bioline) and the QuantStudio 3 RT-PCR System (Thermo Fisher Scientific). Samples were pre-screened for Acan, Col2a1, and Col10a1 expression. Based on the results, RNA-Sequencing (RNAseq) was performed with samples harvested on day 10 of culture. RNAseq with Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed by the Novogene Corporation (Cambridge, UK) as described previously.²⁵ Briefly, a sample quality control (Sample QC) was performed to ensure that the samples fulfilled the required criteria. Thereafter, the appropriate library was prepared and tested for its quality (Library QC). The RNA library was built by polyadenylated (polyA) capture and reverse transcription of cDNA. Illumina NovoSeq X Plus was employed for sequencing. Genes with a p-value below 0.05 (p < .05) were considered as differentially regulated genes. Furthermore, RNA from whole flushed bone was isolated as described previously.²⁶ Real-Time-Polymerase-Chain-Reaction was performed for B2m and Adrb2 to confirm the knockout (Figure S4).

Statistical analysis

Statistical analysis was conducted using GraphPad Prism (v 8.4.3, GraphPad Software). Student’s t-test and Mann-Whitney-U test were applied for either normally or non-normally distributed data. Differences were considered statistically significant at p < .05 (^* .05 > p ≥ .01, ^** .01 > p ≥ .001, ^*** .001 > p ≥ .0001, ^****  p < .0001). Data are presented as mean ± SD. Individual values are shown as points. Each group consisted of 4-6 animals.

Results

Verification of the knockout

In this study, we investigated the effects of a specific knockout of the β₂-AR in chondrocytes on long bone growth, bone homeostasis, and fracture healing in male and female mice of different ages. Figure S1A, C, and  D show the β₂-AR knockout in the growth plate and in the fracture callus of the Adrb2^Col2a1-Cre mice compared to the Adrb2^flox/flox control mice. We found that Col2a1-expressing chondrocytes are affected by the knockout, but also a proportion of osteoblasts in the trabecular bone, indicating that these osteoblasts are derived from chondrocytes (Figure S1B). In addition, RT-PCR analysis revealed a significantly decreased relative gene expression of Adrb2 in flushed bones from knockout mice compared with control mice (Figure S4).

Long bone growth phenotype

To investigate the effects of the β₂-AR knockout on long bone growth, we measured the femur and tibia length of 12-wk-old male Adrb2^Col2a1-Cre mice which showed significantly reduced bone lengths compared to the control mice (Figure 1A and B). To further investigate the reduced long bone growth in the knockout mice, the growth plates were analyzed histologically. Although there was no difference in the growth plate thickness (Figure 1C and I), the Adrb2^Col2a1-Cre mice displayed a significantly increased thickness of the hypertrophic zone (Figure 1E), whereas the non-hypertrophic zone did not differ (Figure 1D). Immunohistochemical staining of the growth plates revealed a significantly increased amount of Collagen 10⁺ and Collagen 2⁺ chondrocytes (Figure 1G and I), whereas there was no difference in the amount of Runt-related transcription factor 2 (Runx2)⁺ chondrocytes (Figure 1H and I). While PCNA as marker for proliferative chondrocytes could not be detected in the growth plates of the 12-wk-old Adrb2^Col2a1-Cre mice (data not shown), 6-wk-old male Adrb2^Col2a1-Cre mice presented a reduced number of proliferative chondrocytes (Figure S2). The female 12-wk-old Adrb2^Col2a1-Cre mice also showed significantly reduced femur (Figure 2A) and tibia lengths (Figure 2B), although the difference was less pronounced in the females compared to the males. The growth plate thickness (Figure 2C and I) as well as the thickness of the non-hypertrophic (Figure 2D) and the hypertrophic zone of the growth plate (Figure 2E) did not differ between the female Adrb2^Col2a1-Cre and the Adrb2^flox/flox control mice. Besides, there was no difference in the amount of Collagen 10⁺ (Figure 2F and I), Collagen 2⁺ (Figure 2G and I) and Runx2⁺ chondrocytes (Figure 2H and I) in the growth plates. The long bone growth was also reduced in the 40-wk-old male (Figure 5A and B) and female (Figure 6A and B) Adrb2^Col2a1-Cre mice with less pronounced effects in the females.

Figure 1.

Long bone growth of 12-wk-old male Adrb2^flox/flox and Adrb2^Col2a1-Cre mice. (A) Femur and (B) tibia length. (C) Femoral growth plate thickness. Thickness of the (D) non-hypertrophic and (E) hypertrophic zone of the femoral growth plate. (F) Collagen 10⁺ chondrocytes, (G) collagen 2⁺ chondrocytes and (H) runt-related transcription factor 2 (Runx2)⁺ chondrocytes in the femoral growth plate. (I) Histological (toluidine blue) and immunohistochemical (collagen 10, collagen 2, Runx2) staining of the femoral growth plate. Scale bar = 50 μm. n = 4-6. ^*p < .05, ^**p < .01, ^****p < .0001.

Figure 2.

Long bone growth of 12-wk-old female Adrb2^flox/flox and Adrb2^Col2a1-Cre mice. (A) Femur and (B) tibia length. (C) Femoral growth plate thickness. Thickness of the (D) non-hypertrophic and (E) hypertrophic zone of the femoral growth plate. (F) Collagen 10⁺ chondrocytes, (G) collagen 2⁺ chondrocytes, and (H) runt-related transcription factor 2 (Runx2)⁺ chondrocytes in the femoral growth plate. (I) Histological (toluidine blue) and immunohistochemical (collagen 10, collagen 2, Runx2) staining of the femoral growth plate. Scale bar = 50 μm. n = 4-6. ^*p < .05.

Figure 5.

Long bone growth and bone homeostasis of 40-wk-old male Adrb2^flox/flox and Adrb2^Col2a1-Cre mice. (A) Femur and (B) tibia length. (C) Cortical tissue mineral density (C.TMD), (D) cortical thickness (C.Th), (E) trabecular tissue mineral density (Tb.TMD), (F) trabecular thickness (Tb.Th), (G) trabecular number (Tb.N), (H) trabecular separation (Tb.Sp), (I) bone volume per tissue volume (BV/TV), and (J) bone diameter of the femora. (K) 3D reconstruction of the trabecular region of interest from μCT analysis of the femora. (L) Number of osteoblasts per bone perimeter (N.Ob/B.Pm), (M) osteoblast surface per bone surface (Ob.S/BS), (N) number of osteoclasts per bone perimeter (N.Oc/B.Pm), and (O) osteoclast surface per bone surface (Oc.S/BS) in the femoral metaphysis. n = 5-6. ^*p < .05, ^***p < .001, ^****p < .0001.

Figure 6.

Long bone growth and bone homeostasis of 40-wk-old female Adrb2^flox/flox and Adrb2^Col2a1-Cre mice. (A) Femur and (B) tibia length. (C) Cortical tissue mineral density (C.TMD), (D) cortical thickness (C.Th), (E) trabecular tissue mineral density (Tb.TMD), (F) trabecular thickness (Tb.Th), (G) trabecular number (Tb.N), (H) trabecular separation (Tb.Sp), (I) bone volume per tissue volume (BV/TV), and (J) bone diameter of the femora. (K) 3D reconstruction of the trabecular region of interest from μCT analysis of the femora. (L) Number of osteoblasts per bone perimeter (N.Ob/B.Pm), (M) osteoblast surface per bone surface (Ob.S/BS), (N) number of osteoclasts per bone perimeter (N.Oc/B.Pm), and (O) osteoclast surface per bone surface (Oc.S/BS) in the femoral metaphysis. n = 6. ^*p < .05, ^**p < .01, ^***p < .001.

Due to the changes in the cartilage of the femoral growth plate, the joint cartilage of the femora was also examined. The knockout mice showed a significantly increased cartilage thickness in the lateral compartment of the femoral joint cartilage (Figure S3A). However, this difference was not observed in the medial compartment (Figure S3B). Besides, there were no differences in the expression of Collagen 2 and Collagen 10 (Figure S3E and F). To assess the severity and histological extent of potential osteoarthritic changes, the OARSI Score was recorded. The score did not differ between the Adrb2^Col2a1-Cre and Adrb2^flox/flox mice (Figure S3C and D).

Bone phenotype

To analyze the influence of the β₂-AR knockout on bone homeostasis, μCT, and histomorphometric analyses of the femora were performed. While the 12-wk-old male (Figure 3) and female (Figure 4) Adrb2^Col2a1-Cre mice did not show any effect of the β₂-AR knockout on bone homeostasis, the 40-wk-old Adrb2^Col2a1-Cre mice showed an osteopenic bone phenotype (Figures 5K and 6K). The 40-wk-old male Adrb2^Col2a1-Cre mice displayed a significantly reduced trabecular number (Tb.N, Figure 5G), a significantly reduced bone volume per tissue volume (BV/TV, Figure 5I) and a significantly smaller bone diameter (Figure 5J), while the trabecular separation was significantly increased (Tb.Sp, Figure 5H) compared to Adrb2^flox/flox mice. In line with this, the osteoblast parameters (number and surface) were reduced (Figure 5L and M), while there were no differences in the osteoclasts (Figure 5N and O). Cortical tissue mineral density (C.TMD, Figure 5C), cortical thickness (C.Th, Figure 5D), trabecular tissue mineral density (Tb.TMD, Figure 5E,), and trabecular thickness (Tb.Th, Figure 5F) did not differ between the male Adrb2^Col2a1-Cre and Adrb2^flox/flox mice. Like the males, the 40-wk-old female Adrb2^Col2a1-Cre mice displayed a significantly reduced Tb.N, (Figure 6G) and BV/TV, (Figure 6I), as well as a significantly increased Tb.Sp (Figure 6H). In contrast to the male mice, the female knockout mice showed a significantly reduced Tb.TMD (Figure 6E) and a significantly reduced trabecular thickness (Tb.Th, Figure 6F). Cortical bone parameters (Figure 6C, D, and  J) did not differ between the Adrb2^Col2a1-Cre and the Adrb2^flox/flox mice. As in the males, the osteopenic phenotype of the female knockout mice was also associated with a reduced number (Figure 6L) and surface (Figure 6M) of osteoblasts, whereas the osteoclast parameters showed no differences (Figure 6N and O). We also checked for bone marrow adiposity but could not detect unusual amounts of bone marrow fat in any of the histological sections. The osteopenic bone phenotype could also be detected in the lumbar spine of the 40-wk-old male Adrb2^Col2a1-Cre mice, but not in the females (Table S1).

Figure 3.

Bone homeostasis of 12-wk-old male Adrb2^flox/flox and Adrb2^Col2a1-Cre mice. (A) Trabecular tissue mineral density (Tb.TMD), (B) bone volume per tissue volume (BV/TV), (C) trabecular thickness (Tb.Th), (D) trabecular number (Tb.N), (E) trabecular separation (Tb.Sp), (F) cortical tissue mineral density (C.TMD), (G) cortical thickness (C.Th), and (H) bone diameter of the femora. (I) Number of osteoblasts per bone perimeter (N.Ob/B.Pm), (J) osteoblast surface per bone surface (Ob.S/BS), (K) number of osteoclasts per bone perimeter (N.Oc/B.Pm), and (L) osteoclast surface per bone surface (Oc.S/BS) in the femoral metaphysis. n = 6.

Figure 4.

Bone homeostasis of 12-wk-old female Adrb2^flox/flox and Adrb2^Col2a1-Cre mice. (A) Trabecular tissue mineral density (Tb.TMD), (B) bone volume per tissue volume (BV/TV), (C) trabecular thickness (Tb.Th), (D) trabecular number (Tb.N), (E) trabecular separation (Tb.Sp), (F) cortical tissue mineral density (C.TMD), (G) cortical thickness (C.Th), and (H) bone diameter of the femora. (I) Number of osteoblasts per bone perimeter (N.Ob/B.Pm), (J) osteoblast surface per bone surface (Ob.S/BS), (K) number of osteoclasts per bone perimeter (N.Oc/B.Pm), and (L) osteoclast surface per bone surface (Oc.S/BS) in the femoral metaphysis. n = 6.

Fracture healing

To determine the effects of the β₂-AR knockout on fracture healing, 12-wk-old male Adrb2^Col2a1-Cre mice underwent a standardized femur osteotomy stabilized with an external fixator. We have chosen to analyze only male mice in that part of the study, because the influence of the knockout on bone phenotype was greater in male mice. However, in terms of clinical translation, it would be important to repeat the experiment also in female mice. Fracture healing was analyzed 10 and 21 d after fracture. On day 10, the knockout mice displayed a significantly increased relative soft tissue area in the FG (Figure 7B) while the relative cartilage area was significantly reduced (Figure 7C). Total callus area (Figure 7A) and relative bone area (Figure 7D) in the FG did not differ. Besides, there were no differences concerning the number and surface of osteoblasts (Figure 7E and F) and osteoclasts (Figure 7G and H). On day 21 after fracture, the Adrb2^Col2a1-Cre mice showed a significantly increased relative cartilage area (Figure 8C and M) and a significantly decreased BV/TV in the FG (Figure 8D and N). In line with this, the relative flexural rigidity of the fractured femora of the knockout mice was significantly reduced (Figure 8G). Micro-computed tomography analysis of the fractured femora revealed a reduced TMD (Figure 8E and N) and a decreased Tb.N (Figure 8F and N) in the fracture callus. While 83% of all Adrb2^flox/flox mice showed a successfully healed fracture after 21 d, only 17% of the Adrb2^Col2a1-Cre mice were successfully healed (Figure 8H). All the other parameters, especially the number and surface of osteoblasts (Figure 8I and J) and osteoclasts (Figure 8K and L) did not differ between the groups.

Figure 7.

Fracture healing in male Adrb2^flox/flox and Adrb2^Col2a1-Cre mice on day 10 after fracture. (A) Callus area, (B) relative soft tissue area, (C) relative cartilage area, and (D) relative bone area in the FG. (E) Number of osteoblasts per bone perimeter (N.Ob/B.Pm), (F) osteoblast surface per bone surface (Ob.S/BS), (G) number of osteoclasts per bone perimeter (N.Oc/B.Pm), and (H) osteoclast surface per bone surface (Oc.S/BS) in the fracture callus. n = 5-6. ^*p < .05.

Figure 8.

Fracture healing in male Adrb2^flox/flox and Adrb2^Col2a1-Cre mice on day 21 after fracture. (A) Callus area, (B) relative soft tissue area, (C) relative cartilage area, and (D) relative bone volume (BV/TV) in the FG. (E) TMD and (F) trabecular number (Tb.N) in the FG. (G) Relative flexural rigidity of the fractured femora. (H) Percentage of healed and non-healed fractured femora (bridging score). (I) Number of osteoblasts per bone perimeter (N.Ob/B.Pm), (J) osteoblast surface per bone surface (Ob.S/BS), (K) number of osteoclasts per bone perimeter (N.Oc/B.Pm), and (L) osteoclast surface per bone surface (Oc.S/BS) in the fracture callus. (M) Histological sections (safranin O/fast green) and (N) 3D reconstruction of the region of interest from μCT analysis of fractured femora. Scale bar = 500 μm. n = 6. ^*p < .05, ^**p < .01.

Transcriptomic profiling in vitro

RNA-Sequencing data of ATDC5 cells stimulated with Adrenaline or Adrenaline + the β₂-AR blocker Butoxamine revealed different pathways being involved in mediating the effects of catecholamines on chondrocytes. The GO analysis identified, among others, the differential expression of GO terms such as “ossification” and the “development” of “bone,” “cartilage,” “connective tissue,” and “skeletal system” (Figure S5A). Downstream signaling pathways in the KEGG analysis were, among others, the “Wnt-signaling-pathway” and the “TGFβ-signaling-pathway” (Figure S5B). Genes of chondrogenesis like Col10a1 and Col2a1 were, for example, upregulated upon Adrenaline + Butoxamine treatment vs. Adrenaline only, as well as Wif1 and Frzb as inhibitors of the Wnt-signaling-pathway. Genes like Transforming Growth Factor β3 (Tgfb3), Igf1, and Ccn2 were, for example, downregulated upon Adrenaline + Butoxamine treatment vs. Adrenaline only. Indian hedgehog (Ihh) was upregulated upon Adrenaline + Butoxamine treatment vs. Adrenaline only. The results are shown in Table S2. Additionally, Tables S3 and S4 show up- or downregulation of genes associated with activation or repression of the β₂-AR to proof that the stimulation with Adrenaline and the blocking with Butoxamine worked.

Discussion

Catecholamine signaling is known to influence long bone growth, bone homeostasis, and fracture healing. We recently could show that chronically psychosocially stressed mice displayed an impaired endochondral ossification during long bone growth and fracture healing due to an increased local catecholamine signaling.^7,8 Other studies have also shown that enhanced catecholamine levels lead to a decrease of bone mass.¹⁷ The β₂-AR was identified as one adrenergic receptor mediating the observed effects of catecholamines on bone. Striking evidence for this is given by the fact that mice with a systemic or an osteoblast-specific lack of this receptor have an increased bone mass.^16,18 In this study, the aim was to analyze the influence of a chondrocyte-specific knockout of the β₂-AR on long bone growth, bone homeostasis, and fracture healing in mice.

Long bone growth in Adrb2^Col2a1-Cre mice

Regardless of age and sex, all Adrb2^Col2a1-Cre mice showed a reduced long bone growth as the femur and tibia lengths were significantly reduced in all knockout mice. In the growth plate of the 12-wk-old male Adrb2^Col2a1-Cre mice, the expression of Collagen 10 as marker for hypertrophic chondrocytes was significantly increased. Also, the ATDC5 cells stimulated with Adrenaline + Butoxamine showed an increased expression of Collagen 10 (Col10a1). In line with this, previous studies could show that the stimulation of the β₂-AR with Isoproterenol lead to a decreased expression of Collagen 10 in chondrocytes indicating that catecholamines inhibit the differentiation of chondrocytes via interacting with β-ARs.²⁷ Besides, the thickness of the hypertrophic zone in the growth plate of the 12-wk-old male Adrb2^Col2a1-Cre mice was significantly increased. All observations together led us to the hypothesis that the reduced long bone growth in the knockout mice is most likely due to a faster transition and differentiation of proliferative to hypertrophic chondrocytes in the growth plate. This was proven by the fact that the 6-wk-old male Adrb2^Col2a1-Cre mice displayed a reduced chondrocyte proliferation in the growth plate. In our previous study, psychosocial stress-induced disturbed long bone growth was mediated by a disturbed chondrocyte-to-osteoblast transdifferentiation in the growth plate.⁸ The unchanged amount of Runx2⁺ hypertrophic chondrocytes in this study indicates an unaffected transdifferentiation of chondrocytes to osteoblasts during long bone growth, further supporting that the impaired growth is mediated by an accelerated chondrocyte differentiation. Additionally, our RNA-Sequencing data showed a significantly decreased expression of Insulin-like Growth Factor 1 (Igf1) and a significantly increased expression of Fibroblast Growth Factor Receptor 3 (Fgfr3) in ATDC5 cells stimulated with Adrenaline + Butoxamine. Both genes are associated in previous studies with an important role for the regulation of long bone growth. For example, Igf1 is declared in one article as the major hormone for the regulation of long bone growth, why Igf1 is used as treatment option for skeletal disorders in children.²⁸ Besides, mice with a specific knockout of Igf1 in type IIαI collagen⁺ cells show a significantly decreased postnatal body length.²⁹ In contrast to Igf1, Fgfr3 is associated with a negative impact on long bone growth. Mice with a systemic knockout of Fgfr3 have longer bones, whereas mice with activating mutations of Fgfr3 display a reduced long bone growth with a disorganization of the growth plate.³⁰ Humans with activating Fgfr3 mutations suffer from achondroplasia, a genetic form of dwarfism.³¹ Besides, the 12-wk-old male Adrb2^Col2a1-Cre mice in our study showed an increased amount of Collagen 2⁺ chondrocytes in the growth plates. In line with this, our in vitro experiment revealed a significantly increased expression of Collagen 2 (Col2a1), Aggrecan (Acan), and Collagen 10 (Col10a1) in ATCD5 cells stimulated with Adrenaline + Butoxamine. Collagen 2, Aggrecan, and Collagen 10 are well-known markers for mature chondrocytes and cartilage. It has already been shown that their expression is also influenced by catecholamine signaling as stimulation of the β₂-AR with Isoproterenol leads to a reduced expression of Collagen 2³² and Collagen 10.²⁷ Besides, Isoproterenol was shown to inhibit the expression of Ihh.²⁷ These data indicate that β₂-AR signaling promotes the growth and inhibits the differentiation of chondrocytes, which is consistent with our findings. Interestingly, these observations were not found in the joint cartilage of the femora, which indicate site-specific differences in the regulation of cartilage formation. The less pronounced effects on long bone growth of the female Adrb2^Col2a1-Cre mice and the fact that no differences were detected in the growth plates of the females point towards pronounced sex-specific differences. Although our study was not designed to mechanistically dissect sex differences, several non-mutually exclusive explanations seem plausible. First, longitudinal bone growth and growth plate maturation are strongly controlled by sex steroids, and multiple studies demonstrated that chondrocytes and osteoblast progenitors respond to estrogen in a sex-dependent manner. For example, conditional deletion of estrogen receptor-α in Runx2-expressing osteoblast progenitors and hypertrophic chondrocytes results in opposite bone phenotypes in male and female mice, indicating that the same molecular perturbation can have divergent consequences depending on sex.³³ Second, accumulating evidence in the cardiovascular system shows clear sex/gender- and age-related differences in β-AR expression and signaling that are modulated by estrogens and sympathetic tone,³⁴ suggesting that β-AR signaling is intrinsically sexually dimorphic. Third, recent work on neuro-bone crosstalk indicates sex-specific differences in sympathetic innervation and receptor profiles within neuro-bone regulatory pathways,³⁵ which may translate into different contributions of β₂-AR signaling to endochondral ossification in males and females. To our knowledge, sex-specific regulation of β₂-ARs in growth plate chondrocytes has not been systematically investigated, and our explanations therefore remain speculative. However, the available literature supports the notion that sex hormones and sex-dependent sympathetic/β-AR signaling likely contribute to the attenuated growth phenotype observed in female Adrb2^Col2a1-Cre mice.

Bone homeostasis in Adrb2^Col2a1-Cre mice

The analysis of the influence of the β₂-AR knockout in chondrocytes revealed age- and sex-dependent differences. While the knockout had no effect on bone homeostasis of the 12-wk-old Adrb2^Col2a1-Cre mice, the 40-wk-old knockout mice showed an osteopenic bone phenotype in the femur and in the lumbar spine with less pronounced effects in the females. This loss of bone mass was due to a decreased number and active surface of osteoblasts, while the osteoclast parameters did not differ. Our RNA-Sequencing data gave indications that, among others, the “Wnt-signaling-pathway” and the “TGFβ-signaling-pathway” may be involved in mediating the knockout effects on osteoblasts. Since osteoblasts are partially chondrocyte-derived,^36,37 a proportion of these cells are affected by the β₂-AR knockout as well. The ATDC5 cells stimulated with Adrenaline + Butoxamine displayed a significantly increased expression of Wnt Inhibitory Factor 1 (Wif1), Frizzled-related protein (Frzb), and Frizzled class receptor 9 (Fzd9). These factors are known to inhibit the Wnt-signaling-pathway leading to a decreased osteoblast differentiation.³⁸ Besides, the expression of Runt-related transcription factor 1 (Runx1) was significantly decreased in ATDC5 cells stimulated with Adrenaline + Butoxamine. It has been shown that Runx1 is expressed in osteoblasts and positively influences osteogenesis via activation of the Wnt-signaling-pathway.³⁹ A decreased expression of Runx1 therefore is associated with bone loss. Mice with a specific knockout of Runx1 in chondrocytes displayed an osteoporotic bone phenotype with a reduced bone density.⁴⁰ Additionally, the TGFβ-signaling-pathway seems to play a role for mediating the effects of the β₂-AR knockout on bone. Our in vitro experiment displayed a decreased expression of Tgfb3 in ATDC5 cells stimulated with Adrenaline + Butoxamine. TGFβ-signaling-pathway is known to be important for the regulation of postnatal bone homeostasis.⁴¹ The less pronounced effects of the β₂-AR knockout on bone homeostasis of the female mice and the age-dependent differences could be due to an impact of sex hormones, an altered expression of the β₂-AR or differences in the sympathetic tonus. Previous studies could, for example, show that the sympathetic tonus rises with age and increases the risk for sympathetically induced hypertension.⁴² Further analyses are needed to investigate the reasons for the age- and sex-dependent influence of the knockout on the bone homeostasis. In conclusion, our data indicate that the β₂-AR on chondrocytes, and thereby also on chondrocyte-derived osteoblasts, plays a different role in bone development and metabolism than the β₂-AR on mature osteoblasts, as osteoblast-specific knockout mice rather displayed increased bone mass. This might be due to the heterogeneity of osteoblasts, as it has been shown that osteoblasts from different origins display broadly different expression patterns and therefore differences in signaling pathways.⁴³ However, this needs to be explored regarding β₂-AR signaling in more detail in future studies.

Fracture healing in Adrb2^Col2a1-Cre mice

The analysis of fracture healing in Adrb2^Col2a1-Cre mice revealed a delayed healing process with a reduced relative flexural rigidity and bridging score 21 d after femur osteotomy. This was accompanied by a reduced TMD, Tb.N, and BV/TV. The histomorphometric analysis revealed an increased relative cartilage area in the FG and no differences concerning osteoblasts and osteoclasts. On day 10 after fracture, relative cartilage area was significantly decreased and relative soft tissue was increased. This suggests that cartilage is formed delayed and persists longer in the fracture callus indicating a delayed endochondral ossification and degradation of cartilage. In line with this, our in vitro experiment revealed a significantly decreased expression of Cellular Communication Network Factor 2 (Ccn2) and MMP19 in ATDC5 cells stimulated with Adrenaline + Butoxamine. Ccn2 has been shown to mediate endochondral ossification and to induce cartilage formation during fracture healing. Mice with a specific knockout of Ccn2 in TRAP-positive monocytes (preosteoclasts) displayed a delayed cartilage formation and lower bone volume in the fracture callus leading to non-unions.⁴⁴ Additionally, MMPs have important functions during fracture healing as they mediate callus remodeling and resorption. While MMP19 has not yet been specifically investigated in bone healing models, there are studies showing the important role of other MMPs during fracture healing. For example, MMP13 deficient mice displayed a delayed cartilage remodeling and removal.⁴⁵ Similar results were shown in mice lacking MMP9.⁴⁶ The reduced cartilage formation in the fracture callus could possibly be due to a faster differentiation from proliferative to hypertrophic chondrocytes as observed in the femoral growth plate. Recent data showed that specifically β₂-AR signaling is important for fracture healing, while cell specific effects were not investigated.¹⁹ To conclude, delayed fracture healing in Adrb2^Col2a1-Cre mice might be due to two processes, which are not mutually exclusive: impaired cartilage formation and/or a faster transition from proliferation to hypertrophy in chondrocytes. Therefore, in light with our previous data showing that fracture healing is delayed in chronically stressed mice due to increased catecholamine levels and the data of the current study, we conclude that catecholamine signaling needs to be tightly balanced to allow uneventful fracture healing and that the effects of β₂-AR signaling may depend strongly on the stage of cellular differentiation.

In conclusion, we could show that β₂-AR signaling in chondrocytes is important for the regulation of long bone growth, bone homeostasis, and fracture healing in mice in an age- and sex-dependent manner. Further investigations are needed to clarify the underlying molecular mechanisms to improve our understanding of the interaction between catecholamines, cartilage, and bone.

Author contributions

Melanie Rebecca Kuhn (Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing—original draft), Melanie Haffner-Luntzer (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Project administration, Resources, Writing—review & editing), Sandra Dieterich (Data curation, Investigation, Writing—review & editing), Meike Sachs (Data curation, Formal analysis, Investigation, Writing—review & editing), Gundula Rösch (Data curation, Formal analysis, Investigation, Writing—review & editing), Zsuzsa Jenei-Lanzl (Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Project administration, Resources, Writing—review & editing), Stefan Oskar Reber (Conceptualization, Funding acquisition, Project administration, Resources, Writing—review & editing), Anita Ignatius (Conceptualization, Funding acquisition, Project administration, Resources, Writing—review & editing), and Miriam Eva Angelica Tschaffon-Müller (Conceptualization, Data curation, Formal analysis, Investigation, Project administration, Writing—review & editing)

Funding

This study was funded by the German Research Foundation (CRC1149, DFG, project number 251293561 as well as to ZJ-L [JE 642/4-2, project number 277277765] within the DFG Research Unit FOR2407 ExCarBon) and by the European Union within the Horizon Europe MSCA program under grant agreement No. 101072766 to ZJ-L.

Conflicts of interest

The authors declare no conflicts of interest.

Data availability

All primary data points are displayed in the manuscript.

Approval statement

The study design was approved by the Federal Animal Care and Use Committee of the local government (Regierungspräsidium Tübingen, Germany; licenses 1437, o.135-10, o.135-19).