Glucocorticoid-Induced Bone Fragility Is Prevented in Female Mice by Blocking Pyk2/Anoikis Signaling

Abstract

Excess of glucocorticoids (GCs) is a leading cause of bone fragility, and therapeutic targets are sorely needed. We report that genetic deletion or pharmacological inhibition of proline-rich tyrosine kinase 2 (Pyk2) prevents GC-induced bone loss by overriding GC effects of detachment-induced bone cell apoptosis (anoikis). In wild-type or vehicle-treated mice, GCs either prevented osteoclast apoptosis or promoted osteoblast/osteocyte apoptosis. In contrast, mice lacking Pyk2 [knockout (KO)] or treated with Pyk2 kinase inhibitor PF-431396 (PF) were protected. KO or PF-treated mice were also protected from GC-induced bone resorption, microarchitecture deterioration, and weakening of biomechanical properties. In KO and PF-treated mice, GC increased osteoclasts in bone and circulating tartrate-resistant acid phosphatase form 5b, an index of osteoclast number. However, bone surfaces covered by osteoclasts and circulating C-terminal telopeptides of type I collagen, an index of osteoclast function, were not increased. The mismatch between osteoclast number vs function induced by Pyk2 deficiency/inhibition was due to osteoclast detachment and anoikis. Further, GC prolongation of osteoclast lifespan was absent in KO and PF-treated osteoclasts, demonstrating Pyk2 as an intrinsic osteoclast-survival regulator. Circumventing Pyk2 activation preserves skeletal integrity by preventing GC effects on bone cell survival (proapoptotic for osteoblasts/osteocytes, antiapoptotic for osteoclasts) and GC-induced bone resorption. Thus, Pyk2/anoikis signaling as a therapeutic target for GC-induced osteoporosis.

Glucocorticoids (GCs), produced and released by the adrenal glands, regulate numerous physiological processes in a wide range of tissues (1, 2). Because of their profound immunosuppressive and anti-inflammatory actions, these agents are widely prescribed to treat an extensive array of pathological conditions, including rheumatoid arthritis, asthma, inflammatory bowel disease, chronic lung, and liver and skin diseases, as well as for the management of organ transplantation and as components of chemotherapy regimens for cancers (3). However, pharmacological administration of GCs, similar to pathologic endogenous elevation, is associated with severe adverse side effects manifested in several tissues and organs, in particular the skeleton (4). Prolonged GC administration leads to a prominent loss of bone mass and strength and increased risk for atraumatic fractures in ∼30% to 50% of patients (4, 5). The initial rapid bone loss induced by GCs is due to exaggerated bone resorption and is accompanied by suppressed bone formation. Excess GC also causes muscle weakness with the consequent loss of body balance and increased propensity to fall, which contribute to the risk of bone fractures (6–8).

The current standard of care, bisphosphonates (BPs) (4), as well as the anti-RANKL antibody recently approved by the US Food and Drug Administration to treat GC-induced osteoporosis (9), effectively prevent bone loss by inhibiting resorption. However, these agents lead to further reduction in bone formation compared with GCs alone, thus preventing rebuilding the bone that was lost (10–12). In addition, profound reduction in bone turnover is not desirable because it leads to accumulation of microdamage and advanced glycation end products, which are associated with osteonecrosis of the jaw with long-term treatments (13, 14). Severe suppression of bone turnover can also reduce toughness, the energy that bone tissue absorbs before failure, with potential increased risk of low-energy atypical fractures (13). Thus, there is an unmet need for therapeutic interventions that prevent GC-induced bone disease that lack these skeletal complications.

Deterioration of bone tissue is due to direct effects of GC on bone cells (15, 16). Rapid increased bone resorption is caused by accumulation of osteoclasts on bone surfaces, resulting from stimulation of osteoclast generation combined with prolongation of the lifespan of preexisting osteoclasts (17). In addition, GCs drastically inhibit bone formation, by suppressing the synthetic capacity of osteoblasts (inhibiting osteocalcin and collagen 1 synthesis and stimulating Wnt antagonist production) and by promoting osteoblast and osteocyte apoptosis (18). A remarkable feature of GC effects on bone is the opposing regulation of apoptosis depending on the bone cell lineage (i.e., antiapoptotic for osteoclasts and proapoptotic for osteoblasts/osteocytes). However, the underlying mechanism of this hallmark has remained elusive.

All osteoclasts die by apoptosis on completing resorption (19). Failure to timely undergo apoptosis is associated with increased depth of resorption pits, which leads to cortical and trabecular bone thinning, and to trabeculae perforation and loss of connectivity in cancellous bone (20). Most osteoblasts (∼70% in human bone) also undergo apoptosis after producing bone matrix proteins; and osteocytes, although long-lived cells can prematurely undergo apoptosis because of changes in hormonal as well as mechanical cues (21). The survival of bone cells is critically dependent of their attachment to the surrounding bone matrix through integrins. Signaling mediated by integrins is bidirectional. Extracellular matrix proteins bind to integrins and activate intracellular pathways, referred to as “outside-in” signaling. In turn, activation of intracellular pathways alters the interaction of integrins with the surrounding matrix, referred to as “inside-out” signaling. Disrupted interactions between integrins and the extracellular matrix lead to detachment and apoptosis (termed anoikis, the Greek word for homelessness) of both osteoclasts and osteoblastic cells (22, 23). In particular, GCs activate in cultured osteocytic cells the proline-rich tyrosine kinase 2 (Pyk2), a member of the focal adhesion kinase family of nonreceptor tyrosine kinases, which in turn induces osteocytic cell detachment and anoikis (23). However, whether this pathway is involved in the opposing effects of GCs on osteoblast/osteocyte vs osteoclast apoptosis in vivo had remained unknown.

We report here that Pyk2 is an essential mediator of anoikis regulated by GC in vivo in bone cells of both lineages: osteoclasts and osteocytes/osteoblasts. Pyk2 activation is required for GC-induced prolongation of osteoclast lifespan and, conversely, for promotion of osteoblast and osteocyte apoptosis. Moreover, circumventing Pyk2 activation by genetic or pharmacological means prevents GC-induced bone loss and increased skeletal fragility, by stimulating anoikis of osteoclasts while preventing anoikis of osteoblasts and osteocytes. Furthermore, overriding GC effects on bone cell survival by Pyk2 inhibition is achieved without altering bone renewal rate or bone biomechanical material properties. Hence, targeting the Pyk2/anoikis pathway represents a mechanistic approach to preserve skeletal integrity with GC excess, devoid of undesirable skeletal complications.

Materials and Methods

Mice and tissue procurement

Skeletally mature 4-month-old female mice with global deletion of Pyk2 [knockout (KO)] and wild-type (WT) littermate controls of C57BL/6 background were generated from breeders provided by Dr. Charles Turner, Indiana University, originally described in Okigaki et al. (24). Genotyping was performed by extracting genomic DNA from tissue samples, followed by PCR reaction using the following primers Pyk2 reverse (CCTGCTGGCAGCCTAACCACAT), Pyk2 WT forward (GGAGGTCTATGAAGGTGTCTACACGAAC), and Pyk2 mutated forward (GCCAGCTCATTCCTCCCACTCAT). PCR products were run on an agarose gel with electrophoresis to distinguish the WT and Pyk2 KO bands. Mice (five per cage) were fed with a regular diet (Teklad Global 18% Protein Extruded Rodent Diet Sterilizable, catalog no. 2018SX, Harlan/ENVIGO, Indianapolis, IN), received water ad libitum, and were maintained on a 12-hour light/dark cycle in polycarbonate cages at the Laboratory Animal Research Center at Indiana University School of Medicine. Mice were implanted with slow-release pellets delivering placebo, 1.4 mg/kg/d (GC1), or 2.1 mg/kg/d (GC2) prednisolone (Innovative Research of America, Sarasota, FL) for 4 weeks (25). Previous studies showed that these doses reproduce in the mouse the hallmarks of GC‐induced osteoporosis and are equivalent to medium and high therapeutic glucocorticoid doses in humans (25, 26). C57BL/6 4-month-old female mice were injected with vehicle or 10 mg/kg of the Pyk2/focal adhesion kinase inhibitor PF-431396 (PF) (Tocris Bioscience, Bristol, UK) five times per week, as previously published (27), beginning 3 days before placebo or GC2 pellet implantation for 2 weeks. To measure the dynamic histomorphometric indexes mineralizing surface to bone surface (MS/BS), mineral apposition rate (MAR), and bone formation rate normalized to bone surface (BFR/BS), KO and WT mice were injected 10 and 3 days before euthanization with calcein (30 mg/kg; Sigma-Aldrich, St. Louis, MO) and alizarin red (50 mg/kg; Sigma-Aldrich) solutions, respectively. The use of two different colors of fluorochromes permits the unequivocal identification of first and second labels and increases accuracy in the measurement of the distance between the two labels required for MAR calculation. Mice were euthanized by, first, sedation with 2% isoflurane (Abbott Laboratories, Chicago, IL) administered by a Drager 19.1 Anesthetic Vaporizer, and then by cervical dislocation. Bones were dissected and processed as indicated in the following section, and muscles were also dissected and weighed. Analysis was performed in a blinded fashion. All animal procedures were approved by the Institutional Animal Care and Use Committee of Indiana University School of Medicine, and animal care was carried out in accordance with institutional guidelines.

Bone mineral density measurement and micro-CT analysis

Lean body mass and bone mineral density (BMD) of the total body excluding the head and the tail, the lumbar spine (L1-L6), and the femur were measured by dual-energy X-ray absorptiometry (DXA) scanning by using a PIXImus II densitometer (GE Medical Systems, Lunar Division, Madison, WI). DXA measurements were performed 2 to 4 days before (initial) and 14 or 28 days (final) after pellet implantation (25, 28). Mice were randomized to the experimental groups based on initial spine BMD. Briefly, mice were sorted by spinal BMD starting from the highest BMD, and then randomly distributed to the treatment groups. After randomization, no statistical differences on spinal BMD were found between experimental groups.

For micro-CT analysis, L6 vertebral and femoral bones were cleaned of soft tissue, fixed in 10% buffered formalin, and stored in 70% ethanol until scanned. Vertebrae were scanned at 6-μm resolution (Skyscan 1172; SkyScan, Aartselaar, Belgium), femoral bones were scanned at 10-μm resolution (SCANCO 35; SCANCO Medical, Brüttisellen, Switzerland), and cancellous bone measurements were done 60 µm away from the growth plates as previously described (29). Cortical bone of the midshaft femur was performed for a 0.2-mm region located at the calculated femoral midpoint. The nomenclature for micro-CT indexes follows the guidelines previously recommended (30).

Mechanical testing

The mechanical properties of L6 vertebrae were determined by axial compression after removal of vertebral processes and the cranial and caudal endplates. Vertebral bodies were loaded at a rate of 0.5 mm/min until failure (100P225 Modular Test Machine; TestResources, Shakopee, MN), as previously described (25). The mechanical properties of femoral middiaphysis were assessed by three-point bending using standard methods (31). Femurs were placed with the posterior side down on the bottom support (9-mm wide) with the descending probe contacted with the central anterior surface and loaded at a rate of 2 mm/min until failure (100P225 Modular Test Machine), as previously described (32). Structural or extrinsic properties (energy to ultimate force, ultimate force, and stiffness) of the vertebra or femur were derived from the load/displacement curves obtained during the vertebral compression or three-point bending tests, respectively. The material or intrinsic properties (toughness, ultimate stress, and the Young modulus or intrinsic stiffness) were derived from the structural properties by taking into account the geometry and volume of the samples quantified by micro-CT, following the calculations previously published (31, 33).

Apoptosis analysis

Apoptosis of osteoblasts and osteocytes was detected by the TUNEL reaction (GeneCopoeia, Rockville, MD) in undecalcified vertebral and femoral bone sections counterstained with 2% methyl green as previously described (34). Apoptosis of osteoclasts and marrow osteoclasts was detected in double-stained tartrate-resistant acid phosphatase (TRAPase), to identify osteoclasts, and TUNEL to identify apoptotic cells in the cancellous bone region starting 300 µm from the distal growth plate and ending 1100 µm proximal to the distal growth plate of paraffin-embedded femora. The total number of TRAPase-positive cells (TRAP⁺ cells) was also quantified including multinucleated cells both on bone surfaces and in the marrow as well as TUNEL⁺ and TUNEL⁻, which was then normalized by both bone surface and marrow area. The number of live and TUNEL⁺ marrow osteoclasts was normalized to the total marrow area. The prevalence of apoptotic osteoblasts, osteocytes, osteoclasts, and marrow osteoclasts was calculated by enumerating the total number and TUNEL⁺ cells exhibiting condense chromatin, nuclear fragmentation, or cell shrinkage.

Serum biochemistry

Blood was collected 2 and 4 weeks after pellet implantation from the facial vein of 3-hour fasted mice. N-terminal propeptide of type I procollagen (P1NP) (35), C-terminal telopeptides of type I collagen (CTX) (36), and TRAPase form 5b (TRAP5b) (37) were measured using enzyme-linked immunosorbent assays (Immunodiagnostic Systems Inc., Gaithersburg, MD) (25, 28). Osteocalcin (OCN) (38) was measured using the Mouse Osteocalcin KIA Kit (Alfa Aesar, Tewksbury, MA).

Bone histomorphometry

A subset of three to six samples was studied. L1-L3 vertebrae and femoral bones were fixed in 10% neutral buffered formalin and embedded undecalcified in methyl methacrylate, as previously described (34). Dynamic histomorphometry measurements were performed 4-μm unstained bone sections under epifluorescence microscopy. Histomorphometric analysis was performed using OsteoMeasure High Resolution Digital Video System (OsteoMetrics, Decatur, GA) interfaced to an Olympus BX51 fluorescence microscope (Olympus America Inc., Center Valley, PA) (28). Osteoclasts were quantified on L2 lumbar vertebral by analyzing the whole vertebral body region and distal femoral by analyzing the femoral cancellous bone region (starting 300 µm from the distal growth plate and ending 1100 µm proximal to the distal growth plate) of thin sections stained for TRAPase and counterstained with Toluidine Blue, as previously published (25, 28). Osteoclasts were defined as TRAPase⁺ cells attached to bone with two or more nuclei, following standardized histomorphometry guidelines by the American Society for Bone and Mineral Research (39) and recent publications (40–42). Osteoclast number was calculated by dividing the total number of osteoclasts attached to bone surfaces by the total bone surface (number/mm); and osteoclast surface was calculated by dividing the bone surface covered by osteoclasts by total bone surface, and multiplied by 100 (%). A marrow osteoclast was defined as a multinucleated TRAPase⁺ cell with two or more nuclei separated from the bone surface by at least one other cell. Marrow osteoclast number was calculated by dividing the total number of marrow osteoclasts by the marrow area (number/mm²). Longitudinal sections of the distal femurs were stained for von Kossa and counterstained with McNeal, as previously described (28).

Quantitative PCR

Total RNA was extracted from vertebral lumbar bones (L4) and quantitative PCR was performed as previously described (25). RNA was extracted with TRIzol (Invitrogen, Carlsbad, CA), and cDNA was synthesized by using high-capacity cDNA reverse transcription kit (Applied Biosystems Inc., Foster City, CA). Primer and probe sets were from Applied Biosystems or from Roche Applied Science (Indianapolis, IN). Relative mRNA expression was normalized to the housekeeping gene ribosomal protein, large P2 (Rplp2) using the ΔCt method. Ratios between genes of interest and housekeeping gene are expressed as fold change compared with WT mice receiving placebo.

Mineralization assay

Primary osteoblastic cells were isolated from the neonatal calvarial bones of C57BL/6 mice, KO, or WT littermate control mice, and plated at 5000 cells/cm² density in MEM Alpha medium with 10% fetal bovine serum and 1% penicillin/streptomycin, as previously described (25, 43). Osteogenic medium was used after cultures reached confluence consisting of 50 μg/mL ascorbic acid and 10 mM β-glycerophosphate and treated with 1 μM dexamethasone or the corresponding vehicle (ethanol), and Pyk2 inhibitor (PF-431396) or the corresponding vehicle (dimethyl sulfoxide). Medium was replaced every 2 to 3 days, and mineralization was visualized using the OsteoImage Mineralization Assay Kit (Lonza Walkersville Inc., Walkersville, MD) or Alizarin Red S (Sigma-Aldrich) staining (43). Mineralization was quantified using a microplate reader for Lonza staining (492/520 nm excitation/emission fluorescence) or for Alizarin Red S staining (405 nm absorbance) (25, 43).

Osteoclast apoptosis assay in vitro

Bone marrow cells were flushed with sterile PBS from tibiae and femora of C57BL/6 mice or WT and KO mice and cultured in MEM Alpha medium with 10% fetal bovine serum and 1% penicillin/streptomycin for 48 hours, as previously described (17). Nonadherent cells were collected, replated at 300,000 cells/cm² density and differentiated with 80 ng/mL recombinant murine soluble RANKL (PeproTech, Rocky Hill, NJ) and 20 ng/mL recombinant murine M-CSF (PeproTech) for 4 to 6 days. Medium was replaced every 2 days and cells were treated with 10⁻⁵M alendronate or corresponding vehicle (PBS) 1 hour before 1 μM dexamethasone or the corresponding vehicle (EtOH) addition. After 24 hours, cells were fixed with 10% buffered formalin, rinsed with dH₂O, and stained for TRAPase (Sigma-Aldrich) with hematoxylin counterstaining. Apoptotic osteoclasts were identified by the presence of morphologic characteristics such as the loss of cell membrane integrity, nuclear condensation, and fragmentation.

Statistical analysis

Data are expressed as means ± SDs. Sample differences were evaluated using SigmaPlot 12.0 (Systat Software, Inc., San Jose, CA). Statistically identified outliers for particular end points (value ± 2 times SD from the mean) were removed. Data of the in vivo and in vitro experiments were analyzed by two-way ANOVA using genotype (WT vs KO) or intervention (vehicle vs PF-431396) and treatment (placebo vs glucocorticoid) as independent variables. For apoptotic osteoclasts in vitro assays, a three-way ANOVA was used for genotype (WT vs KO) or intervention (vehicle vs PF-431396), treatment (vehicle vs dexamethasone), and apoptotic agent (vehicle vs alendronate). When ANOVA detected a significant interaction between the variables or a significant main effect, the post hoc test Tukey was used to determine the significance of the effect of the treatment within each genotype or intervention condition.

Study approval

All animal procedures were approved by the Institutional Animal Care and Use Committee of Indiana University School of Medicine, and animal care was carried out in accordance with institutional guidelines.

Results

Genetic deletion of Pyk2 protects from GC-induced loss of bone mass and strength, and osteoblast/osteocyte apoptosis, but not from the suppression of bone formation

Mice lacking Pyk2 (KO) display higher bone mass at all sites, measured by DXA (Fig. 1A); and increased vertebral (L6) cancellous bone (BV/TV), trabecular thickness, trabecular number, and decreased trabecular separation, quantified by micro-CT (Fig. 1B and 1C), consistent with previous reports (27, 44). In WT mice, administration of two different doses of the GC prednisolone (GC1 = 1.4 or GC2 = 2.1 mg/kg/d) decreased total, spine, and femur BMD (Fig. 1A) and trabecular thickness (Fig. 1B and 1C). The higher GC dose also decreased cancellous BV/TV in WT mice (Fig. 1C, left). In contrast, KO mice were resistant to GC effects on bone mass and volume (Fig. 1A–1C). WT and KO littermate mice exhibit similar body weight that was not altered by GC (not shown), and similar lean body mass that was reduced by GC similarly in both genotypes (Fig. 2A). In addition, the weight of tibialis anterior and gastrocnemius muscles was decreased by GC to a similar extent in WT and KO mice (Fig. 2B).

Figure 1.

Life-long genetic deletion of Pyk2 protects from GC-induced loss of bone mass and strength, and osteoblast/osteocyte apoptosis. (A) BMD for WT or KO mice treated with placebo, 1.4 mg/kg/d prednisolone (GC1), or 2.1 mg/kg/d prednisolone (GC2) for 28 days, measured by DXA. WT, n = 10/10/9, placebo/GC1/GC2; KO, n = 10/9/7, placebo/GC1/GC2. (B, C) Bone volume fraction of BV/TV, Tb.Th, Tb.N, and Tb.Sp of WT or KO lumbar vertebrae L6 are shown. WT, n = 10/10/9, placebo/GC1/GC2; KO, n = 10/9/7, placebo/GC1/GC2. (D) Biomechanical properties were measured in vertebral bone (L6) by axial compression testing; WT, n = 10/10/9, placebo/GC1/GC2; KO, n = 10/9/7, placebo/GC1/GC2. (E) Apoptosis of Ob and Ot was quantified in cancellous and cortical bone in longitudinal sections of lumbar vertebrae (L1–L3) of WT or KO mice stained for TUNEL. WT, n = 4/3/4, placebo/GC1/GC2; KO, n = 4/3/3, placebo/GC1/GC2. (F, G) P1NP, WT, n = 10/10/10, placebo/GC1/GC2; KO, n = 8/10/10, placebo/GC1/GC2, and OCN, WT, n = 7/7/6, placebo/GC1/GC2; KO, n = 7/6/6, placebo/GC1/GC2, were measured in blood samples collected 2 and 4 weeks after pellet implantation for WT or KO mice. (H) MS/BS, MAR, and BFR/BS were quantified in longitudinal sections of lumbar vertebra (L1–L3). WT, n = 5/4/4, placebo/GC1/GC2; KO, n = 5/4/4, placebo/GC1/GC2. (I) Expression of indicated genes in L4 bones is shown, WT, n = 8/8/8, placebo/GC1/GC2; KO, n = 8/8/8, placebo/GC1/GC2. Bars represent means ± SD. *P < 0.05 vs corresponding placebo-treated mice and ^#P < 0.05 vs placebo-treated WT mice, by two-way ANOVA, Tukey post hoc test. Ob, osteoblast; Ot, osteocyte; Tb.N, trabecular number; Tb.Sp, trabecular separation; Tb.Th, trabecular thickness.

Figure 2.

Pyk2 deficiency or inhibition does not prevent glucocorticoid induced muscle wasting. (A) Lean body mass and (B) mass of isolated muscles normalized by body weight are shown. WT, n = 10/10/9, placebo/GC1/GC2; KO n = 10/9/7, placebo/GC1/GC2; veh, n = 9/9, placebo/GC2; PF, n = 10/6, placebo/GC2. *P < 0.05 vs corresponding placebo-treated mice and ^#P < 0.05 vs corresponding GC-treated WT mice by two-way ANOVA, Tukey post hoc test. veh, vehicle.

KO mice also exhibited enhanced structural (or extrinsic) mechanical properties of ultimate force, energy to ultimate load, and stiffness, measured in vertebral bone compared with WT mice (Table 1). However, no differences in material (or intrinsic) mechanical properties (toughness, ultimate stress, and Young modulus) were detected between the two genotypes. These findings suggest that the gain in bone strength in KO mice is due to increased bone mass and not to alterations at the tissue level (i.e., mineralization or collagen structure). GC treatment decreased the structural properties of ultimate force and energy to ultimate load, as well as the material property of toughness in WT mice (Fig. 1D). In contrast, KO mice were protected from any changes in structural or material mechanical properties induced by GC.

Table 1.

Genetic Deletion of Pyk2 Protects From GC-Induced Loss of Bone Strength

L6 Axial Compression (Cancellous Bone)	WT Placebo	WT GC1	WT GC2	KO Placebo	KO GC1	KO GC2
Structural (or extrinsic) mechanical properties
Energy to ultimate force, mJ	4.1 ± 0.8a	2.1 ± 1.5a	2.0 ± 1.4a	6.3 ± 1.8b	6.8 ± 1.2	5.9 ± 1.4
Ultimate force, N	16.8 ± 2.7	11.8 ± 4.1	9.5 ± 3.3a	24.8 ± 3.3b	22.7 ± 6.3	25.2 ± 6.3
Stiffness, N/mm	71.7 ± 14.9	83.0 ± 19.4	61.0 ± 23.1	105.6 ± 21.0b	113.1 ± 30.0	103.2 ± 27.5
Material (or intrinsic) mechanical properties
Toughness, mJ/mm3	14.9 ± 3.7	9.1 ± 4.7a	6.2 ± 4.7a	15.5 ± 5.6	14.0 ± 1.3	13.5 ± 1.5
Ultimate stress, MPa	1.8 ± 0.4	1.3 ± 0.3	1.4 ± 0.5	1.8 ± 0.3	1.6 ± 0.6	1.7 ± 0.4
Young modulus, MPa	728.7 ± 276.9	717.8 ± 187.3	566.5 ± 276.1	739.8 ± 285.6	763.3 ± 282.8	712.8 ± 222.8

Biomechanical properties were measured in vertebral bone (L6) by axial compression testing. WT, n = 10/10/9, placebo/GC1/GC2; KO, n = 10/9/7, placebo/GC1/GC2.

^a P < 0.05 vs corresponding placebo-treated mice, by two-way ANOVA, Tukey post hoc test.

^b P < 0.05 vs placebo-treated WT mice, by two-way ANOVA, Tukey post hoc test.

We next examined the underlying mechanism of the protection from GC-induced bone loss by Pyk2 deletion. Overall, deletion of Pyk2 under basal conditions did not alter osteoblast or osteocyte apoptosis, circulating levels of formation and remodeling markers (P1NP or OCN, respectively), or dynamic histomorphometric parameters (MS/BS, MAR, and BFR/BS), except for a decrease in P1NP exhibited by KO mice detected at 4 weeks but not at 2 weeks after pellet implantation (Fig. 1E–1H). GC induced the expected increase in osteoblast and osteocyte apoptosis in cancellous and cortical bone in WT mice, whereas KO mice were protected (Fig. 1E). This finding is consistent with the previous demonstration that Pyk2 kinase expression and activity is required for GC-induced apoptosis of osteocytic cells in vitro (23). However, KO mice were not protected from the reduction of bone formation induced by GC. Thus, GC decreased serum P1NP and OCN in both WT and KO mice (Fig. 1F and 1G) and decreased MS/BS, MAR, and BFR/BS to the same extent in WT and KO mice (Fig. 1H). GC reduced the mRNA expression of osteocalcin, collagen 1a1, and alkaline phosphatase in bones from both WT and Pyk2 KO mice (Fig. 1I). This persistent inhibitory effect of GC on the biosynthetic activity of osteoblasts in the KO mice explains the inability of Pyk2 inhibition to prevent the decreased bone formation induced by the hormones.

Short-term pharmacologic Pyk2 inhibition prevents GC-induced bone loss, fragility, and osteoblast/osteocyte apoptosis, but not the suppression of bone formation

Consistent with the musculoskeletal response of KO mice, pharmacologic inhibition of Pyk2 with the kinase inhibitor PF prevented the loss of BMD induced by short-term (2 weeks) treatment with GC (Fig. 3A). In contrast, and similar to the genetic deletion of Pyk2, PF did not prevent the loss of lean body mass or the mass of individual muscles induced by GC (Fig. 2A and 2B). PF-treated mice were also protected from GC-induced deterioration of femoral bone microarchitecture, specifically from trabecular and cortical thinning (Fig. 3B and 3C). PF prevented the reduction induced by GC in the cortical area fraction, which was mainly driven by the loss of cortical bone area from the endosteal bone surface because no reductions in the cortical bone total cross-sectional area were detected. PF also prevented the decreases in ultimate force, energy to ultimate load, and toughness induced by GC (Fig. 3D; Table 2).

Figure 3.

Short-term pharmacologic Pyk2 inhibition prevents GC-induced bone loss, fragility, and osteoblast/osteocyte apoptosis. (A) BMD for veh or PF-treated C57BL/6 mice implanted with placebo or 2.1 mg/kg/d prednisolone (GC2) for 14 days, measured by DXA. veh, n = 9/9, placebo/GC2; PF n = 10/6, placebo/GC2. (B) BV/TV, Tb.Th, Tb.N, and Tb.Sp of the distal femur of veh or PF-treated mice measured by micro-CT. veh, n = 9/9, placebo/GC2; PF, n = 10/6, placebo/GC2. (C) Ct.Ar/Tt.Ar, Ct.Ar, Tt.Ar, and Ct.Th of veh or PF-treated femoral midshafts are shown. veh, n = 9/9, placebo/GC2; PF, n = 10/6, placebo/GC2. (D) Biomechanical properties were measured in the femoral middiaphysis by three-point bending. veh, n = 9/9, placebo/GC2; PF, n = 10/6, placebo/GC2. (E) Ob and Ot apoptosis was quantified in cancellous and cortical bone in TUNEL-stained longitudinal lumbar vertebrae (L1-L3) sections of veh or PF-treated mice. veh, n = 5/5, placebo/GC2; PF, n = 5/3, placebo/GC2, n = 3-5. (F) P1NP, veh, n = 9/9, placebo/GC2; PF n = 10/6, placebo/GC2, and OCN, veh, n = 7/7, placebo/GC2; PF, n = 7/6, placebo/GC2, were measured in veh or PF-treated mice. (G) Ob.N/BS and Ob.S/BS were measured in von Kossa/McNeal-stained longitudinal distal femur sections. veh, n = 3/3, placebo/GC2; PF, n = 3/4, placebo/GC2. (H) Quantification of mineralization in cultures of calvaria-derived primary osteoblastic cells from WT or KO mice treated with veh or dex and stained using the OsteoImage Mineralization Assay Kit. Scale bars, 400 μm. WT, n = 8/7, veh/dex; KO, n = 8/8, veh/dex. (I) Mineralization quantification of calvaria-derived primary osteoblastic cell cultures from C57BL/6 mice treated with or without PF-431396 with veh or dex for 10 days stained using Alizarin Red S (405 nm absorbance). veh, n = 4/4, veh/dex; PF, n = 4/4; veh/dex, measured in triplicate. Bars represent means ± SD (A–G) *P < 0.05 vs corresponding placebo-treated mice and ^#P < 0.05 vs placebo-treated WT mice by two-way ANOVA, Tukey post hoc test. (H, I) *P < 0.05 vs corresponding veh-treated cells and ^#P < 0.05 vs veh-treated cells without GC by two-way ANOVA, Tukey post hoc test. Ct.Ar, cortical bone area; Ct.Th, cortical thickness; dex, dexamethasone; OB, osteoblast; Ob.N/BS, osteoblast number normalized to bone surface; Ob.S/BS, surface normalized to bone surface; Ot, osteocyte; Tb.Th, trabecular thickness; Tt.Ar, total cross-sectional area; veh, vehicle.

Table 2.

Pyk2 Pharmacologic Inhibition Prevents GC-Induced Bone Fragility

3-Point Bending (Cortical Bone)	Vehicle Placebo	Vehicle GC2	PF-431396 Placebo	PF-431396 GC2
Structural (or extrinsic) mechanical properties
Energy to ultimate force, mJ	2.9 ± 0.4	2.0 ± 0.5a	2.8 ± 0.8	2.5 ± 0.7
Ultimate force, N	12.6 ± 0.9	10.9 ± 0.6a	12.5 ± 1.4	11.7 ± 1.9
Stiffness, N/mm	59.0 ± 4.9	54.1 ± 6.0	57.0 ± 5.6	56.9 ± 7.6
Material (or intrinsic) mechanical properties
Toughness, mJ/mm3	11.8 ± 1.9	8.4 ± 1.7a	11.4 ± 1.5	11.6 ± 1.4
Ultimate stress, MPa	174.9 ± 12.9	168.1 ± 12.5	170.7 ± 15.8	174.5 ± 18.7
Young modulus, MPa	2927.6 ± 253.7	2892.7 ± 397.5	2778.1 ± 365.0	2966.2 ± 112.5

Biomechanical properties were measured in the femoral middiaphysis by three-point bending; veh, n = 9/9, placebo/GC2; PF, n = 10/6, placebo/GC2.

^a P < 0.05 vs corresponding vehicle-treated mice, by two-way ANOVA, Tukey post hoc test.

The 2-week treatment with GC induced the expected increase in osteoblast and osteocyte apoptosis in cancellous and cortical bone in vehicle-treated mice, but not in PF-treated mice (Fig. 3E). PF alone did not affect osteoblast/osteocyte apoptosis, but increased circulating P1NP and osteoblast number and surface compared with vehicle-treated mice (Fig. 3F and 3G). However, all GC-treated mice exhibited similar reductions in P1NP, OCN, osteoblast number, and osteoblast surface regardless of PF administration when compared with respective placebo controls, demonstrating that GC affect osteoblast number and function independently of Pyk2 activity (similar to genetic deletion of Pyk2 shown in Fig. 1).

Consistent with these in vivo effects of GC, addition of the synthetic GC dexamethasone decreased hydroxyapatite deposition to a similar extent in primary cultures of calvarial osteoblasts derived from WT or KO mice (Fig. 3H); and decreased calcium deposition in the absence or presence of PF (Fig. 3I). Further, mineral deposition was increased in cultures of KO-derived osteoblasts or osteoblasts treated with PF alone compared with the respective controls (Fig. 3H and 3I), mimicking the increase in P1NP (Fig. 3F) and in osteoblast number and surface (Fig. 3G) exhibited by PF-treated mice. These findings are consistent with the increase in BFR/BS, MAR, and MS/BS induced by PF in ovariectomized rats previously reported (27). These findings demonstrate that Pyk2 deletion/inhibition increases osteoblast activity under basal conditions but cannot overcome GC-induced decrease in matrix mineral production or bone formation.

Loss of Pyk2 signaling by genetic or pharmacologic means protects from GC-induced bone resorption by decreasing activity of individual osteoclasts and promoting osteoclast detachment from bone

We next examine the effects of Pyk2 deletion/inhibition on GC-induced bone resorption. Circulating levels of the resorption marker CTX were not affected by genetic deletion of Pyk2 (Fig. 4A). However, KO mice exhibited increased circulating levels of TRAP5b, an index of osteoclast number (Fig. 4B). As a consequence, KO mice exhibited lower CTX/TRAP5b ratio, which indicates decreased activity of individual osteoclasts (Fig. 4C). Consistent with normal CTX levels in KO mice, the number of osteoclasts or the bone surface covered by osteoclasts quantified by histomorphometry was not different in KO compared with WT mice (Fig. 4D, left and middle). However, KO mice exhibited approximately a threefold increase in the number of osteoclasts present in the bone marrow compared with WT, consistent with the increased TRAP5b in the circulation. Therefore, KO mice exhibited increased number of inactive osteoclasts detached from the bone surface explaining the mismatch between the low CTX/TRAP5b ratio.

Figure 4.

Loss of Pyk2 signaling by genetic or pharmacologic means protects from GC-induced bone resorption by decreasing activity of individual osteoclasts and promoting osteoclast detachment from bone. (A–C, H) CTX, WT, n = 7/7/7, placebo/GC1/GC2; KO, n = 7/6/6, placebo/GC1/GC2, TRAP5b, WT, n = 7/6/6, placebo/GC1/GC2; KO, n = 7/6/6, placebo/GC1/GC2, CTX/TRAP5b, WT, n = 7/5/6, placebo/GC1/GC2; KO, n = 7/6/6, placebo/GC1/GC2, were measured in blood collected 4 weeks after pellet implantation for WT or KO mice and CTX, veh, n =9/9, placebo/GC2; PF, n = 10/6, placebo/GC2, and TRAP5b, veh, n = 9/9, placebo/GC2; PF, n = 10/6, placebo/GC2 were measured after 2 weeks for veh and PF-treated mice. (D, I) N.Oc/BS and Oc.S/BS and marrow osteoclast number normalized to total marrow cavity area were determined in longitudinal sections of lumbar vertebra (L1–L3) for WT, n = 6/6/6, placebo/GC1/GC2; KO, n = 6/6/6, placebo/GC1/GC2 mice, or of longitudinal distal femur sections, veh, n = 6/8, placebo/GC2; PF, n = 8/5, placebo/GC2, for veh or PF-treated mice. (E) Expression of indicated genes in L4 bones is shown, WT, n = 8/8/8, placebo/GC1/GC2; KO, n = 8/8/8, placebo/GC1/GC2. (F) Representative microscopy images of osteoclasts on cancellous bone surface or in the marrow of lumbar vertebra (L2) stained for TRAPase. Arrows point to osteoclasts fully attached to bone on top images from WT mice and osteoclasts partially attached on bottom images from KO mice. Scale bars, 25 μm. (G) Cartoons and representative high-magnification images of WT and KO osteoclasts. (A–E) Bars represent means ± SD. *P < 0.05 vs corresponding placebo-treated mice and ^#P < 0.05 vs placebo-treated WT mice, by two-way ANOVA, Tukey post hoc test. (H, I) *P < 0.05 vs corresponding vehicle-treated mice and ^#P < 0.05 vs placebo and vehicle-treated mice by two-way ANOVA, Tukey post hoc test. N.Oc/BS, osteoclast number normalized to bone surface; Oc.S/BS, osteoclast surface normalized to bone surface; veh, vehicle.

Both doses of GC induced the expected increase in CTX and TRAP5b in WT mice (Fig. 4A and 4B). However, GC did not affect CTX or further increase the already elevated circulating TRAP5b levels in KO mice. Moreover, GC did not change the CTX/TRAP5b ratio in either WT or KO mice (Fig. 4C), suggesting that increased resorption induced by GC in WT mice is due to an increase in osteoclast number but not in the activity of individual osteoclasts. Indeed, GC increased osteoclast number in bone in both WT and KO mice, but GC increased osteoclast surface in WT mice only (Fig. 4D, left and middle). Consistent with increased osteoclast number in the marrow (Fig. 4D, right), mRNA expression of TRAPase as well as cathepsin K is increased in vertebral bone/bone marrow preparations from KO mice (Fig. 4E). This finding demonstrates that osteoclasts in the marrow also express cathepsin K because the majority of osteoclasts in the KO mice are present in the marrow. Close examination of bone sections showed that osteoclasts were fully attached to bone in GC-treated WT mice (Fig. 4F, top). In contrast, osteoclasts were only partially attached in GC-treated KO mice (Fig. 4F, bottom). Cartoons and representative images of WT and KO osteoclasts are shown in Fig. 4G. The difference in osteoclast attachment to bone surfaces explains the apparent discrepancy between osteoclast number and surface in response to GC in WT vs KO mice.

The protection from bone loss induced by GC by genetic deletion of Pyk2 was reproduced by short-term blockade of Pyk2 kinase activity with the PF. Two weeks of GC treatment increased CTX, TRAP5b, and osteoclast number and surface in vehicle-treated mice (Fig. 4H and 4I). PF did not alter any of these parameters in placebo treated mice; however, it inhibited the increase in CTX and osteoclast number/surface induced by GC. In contrast, PF did not inhibit the increase in TRAP5b induced by GC. Moreover, similar to genetic deletion of Pyk2, treatment with PF increased the number of osteoclasts in the marrow, which were further elevated in mice treated with both PF and GC.

Decreased GC-induced resorption by genetic or pharmacologic inhibition of Pyk2 is due to osteoclast apoptosis, which overrides the prosurvival effects of GC on osteoclasts

The impaired attachment to bone of osteoclasts from KO or PF-treated mice was due to premature death by apoptosis, as evidenced by increased prevalence of apoptotic osteoclasts (TUNEL and TRAPase positive) on bone surfaces as well as in the bone marrow in bones from KO or from PF-treated mice compared with respective controls (Fig. 5A–5B). These findings strongly suggest that the reduction in osteoclast function by inhibition of Pyk2 signaling in vivo is due to detachment-induced apoptosis (i.e., anoikis).

Figure 5.

Decreased GC-induced resorption by genetic or pharmacologic inhibition of Pyk2 is due to osteoclast apoptosis, which overrides the prosurvival effects of GC on osteoclasts. (A, B) Apoptotic osteoclasts on bone (left) and in the marrow (right) were quantified in double-stained TRAPase and TUNEL longitudinal sections of the distal femur. WT, n = 6/6/6, placebo/GC1/GC2; KO, n = 6/6/6, placebo/GC1/GC2 or veh, n = 5/6, placebo/GC2; PF, n = 6/6, placebo/GC2. (A, right) Representative microscopy images of TUNEL-negative osteoclasts fully attached to bone in WT mice and TUNEL-positive osteoclasts partially attached to bone in KO mice. Scale bars, 20 μm. (C–F) Apoptosis of differentiated primary osteoclasts was quantified by evaluating loss of cell membrane integrity, nuclear condensation, and nuclear fragmentation. Osteoclasts derived from (C, D) KO and WT mice or from (E, F) C57BL/6 mice treated with or without PF, were treated in vitro with alendronate, dex, or respective veh, for 24 hours. TRAPase and hematoxylin staining were subsequently performed. (D, F) Green star and arrow symbols indicate alive and apoptotic osteoclasts, respectively. Scale bars, 20 μm. n = 6. (A, B) Bars represent means ± SD. *P < 0.05 vs corresponding placebo-treated mice and ^#P < 0.05 vs placebo-treated mice, by two-way ANOVA, Tukey post hoc test. (C, E) *P < 0.05 vs corresponding vehicle-treated cells and ^#P < 0.05 vs alendronate treated cells from WT or C57BL/6 mice by three-way ANOVA, Tukey post hoc test. veh, vehicle

GC did not alter the increased osteoclast apoptosis in bone from either KO or PF-treated mice. However, mice treated with GC and the PF inhibitor exhibited a higher prevalence of apoptotic osteoclasts in the marrow compared with mice treated with PF alone. This effect paralleled the increased TRAP5b induced by GC (Fig. 4H), suggesting that the increased osteoclasts generated by GC undergo premature apoptosis induced by PF and are unable to reach the bone surfaces, thus accumulating in the marrow.

Increased apoptosis is an intrinsic feature of osteoclasts from KO mice, as it was evidenced in vitro in osteoclast precursors derived from KO mice cultured in the presence of RANKL and M-CSF (Fig. 5C). Osteoclasts in these cultures exhibited abnormal morphology, impaired spreading, and absence of organized podosomes (Fig. 5D), consistent with previous reports using immunofluorescence microscopy to detect F-actin and tubulin and depicting morphologic defects of Pyk2 deficient osteoclasts in vitro (44). Moreover, KO osteoclast cultures exhibited a reduction in the total number of TRAPase⁺ cells with three or more nuclei (898 ± 238 vs 309 ± 249, WT vs KO respectively), suggesting that impaired attachment leads to premature apoptosis (i.e., anoikis). This phenomenon could explain the decreased activity of individual osteoclasts in vivo evidenced by a lower CTX/TRAP5b ratio (Fig. 4C).

Earlier findings showed that prolongation of the lifespan of preexisting osteoclasts explains part of the increase in bone resorption induced by GC (17, 18). This antiapoptotic effect of GC was readily demonstrated in cultures of osteoclasts derived from WT mice treated with the BP alendronate (^# in Fig. 5C and images shown in Fig. 5D). Thus, BP treatment increased apoptosis of TRAPase⁺ cells either with three or more nuclei or less than three nuclei, which was prevented by dexamethasone by 60% and 46%, respectively. However, the survival effect of dexamethasone was absent in osteoclasts derived from KO mice. Further, whereas dexamethasone prevented BP-induced apoptosis in WT osteoclast cultures treated with vehicle by 51%, it was ineffective in cultures treated with the Pyk2 inhibitor PF (Fig. 5E and 5F), although PF addition did not significantly altered total number of TRAPase⁺ cells (1101 ± 280 vs 907 ± 144, vehicle vs PF, respectively). Taken together, these pieces of evidence demonstrate that prolongation of osteoclast lifespan by GC is negated in the absence of Pyk2 expression/activity and, thus, the inherent osteoclast dysfunction and increased osteoclast apoptosis conferred by lack of Pyk2 overrides GC-induced resorption.

Discussion

We identified in the current studies the Pyk2/anoikis signaling pathway as a therapeutic target that opposes GC effects on bone cells. Pyk2 activation is required for GC to induce apoptosis of osteoblasts and osteocytes and to prolong the lifespan of osteoclasts, leading to increased resorption. Either genetic deletion or short-term pharmacologic inhibition of Pyk2 overrides GC regulation of bone cell anoikis, maintaining osteoblast/osteocyte viability while shortening osteoclast lifespan. Remarkably, Pyk2 inhibition prevents bone loss primarily by inhibiting resorption and interfering with osteoclast survival induced by GC because bone formation is still suppressed even when osteoblast apoptosis induced by GC is prevented by Pyk2 inhibition. Importantly, inhibiting Pyk2 signaling bypasses the negative effects of the current standard of care, thus preserving bone renewal rate and the biomechanical material properties of bone. We conclude that targeting the Pyk2/anoikis pathway represents a mechanistic approach to maintain skeletal integrity with GC excess (Fig. 6).

Figure 6.

Inhibition of the Pyk2/anoikis pathway represents a mechanistic approach to preserve skeletal integrity with GC excess. Activation of proline-rich tyrosine kinase 2 (Pyk2) is required for the opposing GC regulation of bone cell anoikis, depending on bone cell lineage, with antiapoptotic effects for osteoclasts but proapoptotic effects for osteoblasts/osteocytes. Inhibiting Pyk2 activation prevents GC-induced bone loss and fragility by stimulating osteoclasts anoikis and preventing osteoblasts/osteocytes anoikis, without further suppressing bone renewal rate or altering biomechanical material properties. Targeting the Pyk2/anoikis pathway is a mechanistic approach to preserve skeletal integrity with GC, without undesirable skeletal complications.

Our studies reveal that the Pyk2 kinase is at the core of the mechanism underlying the opposing regulation of apoptosis by GC depending on the bone cell lineage, a hallmark of GC action. Whereas Pyk2 phosphorylation induced by GC promotes apoptosis of osteoblasts and osteocytes, it stimulates podosome formation and induces attachment of osteoclasts to bone surfaces, prolonging osteoclast lifespan. Consequently, genetic absence of Pyk2 or treatment with a Pyk2 inhibitor prevented GC effects on bone cell anoikis maintaining osteoblast/osteocyte viability and at the same time promoting osteoclast apoptosis.

A remarkable feature of the Pyk2/anoikis pathway is that its pharmacologic inhibition rapidly inhibits bone resorption stimulated by GC by inducing osteoclast detachment from bone and apoptosis of osteoclasts, which accumulate in the marrow. This mechanism could potentially be exploited to block bone loss in other pathological conditions in which osteoclast production is exaggerated. In particular, inhibitors of Pyk2/anoikis signaling could also potentially block the bone loss associated with aging, which is another leading cause of osteoporosis world-wide. GC activity is regulated by two isoforms of 11β-hydroxysteroid dehydrogenase (11β-HSD): 11β-HSD1 converts inactive (cortisone) to active (cortisol) metabolites and 11β-HSD2 catalyzes the opposite reaction (45). The level of expression of 11β-HSD1 predicts the severity of the skeletal response to GC action (46) and also increases with age (47). Thus, exaggerated endogenous GC activity likely contributes to the mechanism underlying age-related bone loss, and interventions that prevent iatrogenic GC-induced bone disease could also ameliorate the skeletal fragility resulting from aging.

Unlike bisphosphonates and denosumab (RANKL antibody), also used to manage GC-induced osteoporosis (9–12, 48), targeting the Pyk2/anoikis pathway decreased bone resorption to placebo levels without severely reducing bone turnover and did not further suppress bone formation below the already low rate induced by GC. Thus, inhibition of the Pyk2/anoikis pathway represents an alternative approach to effectively reduce GC-induced resorption while circumventing the side effects associated with potent antiresorptive agents.

Apoptosis of osteoblasts induced by GC is prevented by Pyk2 inhibition. However, bone formation is still suppressed in the KO mice or in mice treated with the Pyk2 inhibitor. These findings suggest minimal, if any, contribution of apoptosis to osteoblast function in the frame of GC excess. This outcome is not surprising, considering the pleiotropic effects of the hormones, which affect osteoblast function by several mechanisms. Indeed, in mice with Pyk2 genetic deletion, GC still inhibited the synthetic activity of osteoblasts as evidenced by the reduction in osteocalcin, collagen 1a1, and alkaline phosphatase expression. These findings demonstrate that Pyk2 inhibition specifically interferes with the regulation of osteoblast anoikis by GC, whereas GC inhibition of bone formation and mineral matrix production in osteoblasts occurs independently of Pyk2. Future investigations are required to examine the efficacy of Pyk2 inhibition in long-term experiments in which bone loss induced by GC is primarily driven by the reduction in bone formation induced by the hormones.

Maintenance of osteocyte viability with Pyk2 inhibition could contribute to preservation of bone strength in our studies because previous reports demonstrated that prevention of osteocyte apoptosis by genetic means preserves at least part of bone strength even when GCs induce loss of bone mineral (15). Future studies are warranted to examine whether long-term inhibition of Pyk2 activity prevents the development of osteonecrosis, a process that has been linked to GC-induced osteocyte apoptosis (49).

Because bone and muscle act as a mechanical unit, it could be expected that changes in one tissue might affect the other. However, in contrast to the observed bone protection, KO or PF-treated mice still exhibited muscle wasting induced by GC. These findings show a lack of crosstalk between these two tissues in the context of GC excess and also demonstrate that the effects of the GC on skeletal muscle are independent of Pyk2 expression or activity. One mechanism by which bone might influence muscle is through growth factors stored in the bone matrix. In particular, TGF-β released into the circulation by the process of bone resorption has been implicated in the oxidation and nitrosylation of the ryanodine receptor and calcium release channel in skeletal muscle leading to muscle weakness in cancer (50). Our findings in which Pyk2 inhibition effectively blocks resorption but does not alter muscle atrophy argue against this mechanism in the context of GC excess. Similarly, blocking GC-induced resorption by genetic deletion of Sost/sclerostin does not protect muscle (25). Taken together, this evidence highlights the need for therapeutic interventions that simultaneously interfere with the damaging actions of GC in both bone and muscle.

In conclusion, targeting the Pyk2/anoikis pathway represents a mechanistic approach to preserve skeletal integrity in glucocorticoid excess, devoid of undesirable skeletal complications.

Glossary

Abbreviations:

11β-HSD

11β-hydroxysteroid dehydrogenase

BP

bisphosphonate

BFR/BS

bone formation rate normalized to bone surface

BMD

bone mineral density

BV/TV

bone volume/tissue volume

CTX

C-terminal telopeptides of type I collagen

DXA

dual-energy X-ray absorptiometry

GC

glucocorticoid

KO

knockout

MAR

mineral apposition rate

MS/BS

mineralizing surface to bone surface

OCN

osteocalcin

P1NP

N-terminal propeptide of type I procollagen

PF

Pyk2 kinase inhibitor PF-431396

TRAP5b

tartrate-resistant acid phosphatase 5b

TRAPase

tartrate-resistant acid phosphatase

WT

wild-type