CXCR4 antagonism attenuates load-induced periosteal bone formation in mice

Authors

  • Philipp Leucht,

    1. Department of Orthopaedic Surgery, Stanford University School of Medicine, Stanford, California
    Search for more papers by this author
  • Sara Temiyasathit,

    1. Rehabilitation R&D Center for Tissue Regeneration, Repair, and Restoration, Veterans Affairs Palo Alto Health Care System, Palo Alto, California
    Search for more papers by this author
  • Ashley Russell,

    1. Rehabilitation R&D Center for Tissue Regeneration, Repair, and Restoration, Veterans Affairs Palo Alto Health Care System, Palo Alto, California
    Search for more papers by this author
  • Juan F. Arguello,

    1. Rehabilitation R&D Center for Tissue Regeneration, Repair, and Restoration, Veterans Affairs Palo Alto Health Care System, Palo Alto, California
    Search for more papers by this author
  • Christopher R. Jacobs,

    1. Department of Biomedical Engineering, Columbia University, New York, New York
    Search for more papers by this author
  • Jill A. Helms,

    1. Division of Plastic and Reconstructive Surgery, Department of Surgery, Stanford University School of Medicine, Stanford, California
    Search for more papers by this author
  • Alesha B. Castillo

    Corresponding author
    1. Division of Plastic and Reconstructive Surgery, Department of Surgery, Stanford University School of Medicine, Stanford, California
    • Rehabilitation R&D Center for Tissue Regeneration, Repair, and Restoration, Veterans Affairs Palo Alto Health Care System, Palo Alto, California
    Search for more papers by this author

  • The authors have no conflicts of interest.

Correspondence to: Alesha B. Castillo (T: 650-493-5000 ext. 66252; F: 650-493-4919; E-mail: alesha.castillo@stanford.edu)

ABSTRACT

Mechanical loading is a key anabolic regulator of bone mass. Stromal cell-derived factor-1 (SDF-1) is a stem cell homing factor that is important in hematopoiesis, angiogenesis, and fracture healing, though its involvement in skeletal mechanoadaptation is virtually unknown. The objective of this study was to characterize skeletal expression patterns of SDF-1 and CXCR4, the receptor for SDF-1, and to determine the role of SDF-1 signaling in load-induced periosteal bone formation. Sixteen-week-old C57BL/6 mice were treated with PBS or AMD3100, an antagonist against CXCR4, and exposed to in vivo ulnar loading (2.8 N peak-to-peak, 2 Hz, 120 cycles). SDF-1 was expressed in cortical and trabecular osteocytes and marrow cells, and CXCR4 was primarily expressed in marrow cells. SDF-1 and CXCR4 expression was enhanced in response to mechanical stimulation. The CXCR4 receptor antagonist AMD3100 significantly attenuated load-induced bone formation and led to smaller adaptive changes in cortical geometric properties as determined by histomorphometric analysis. Our data suggest that SDF-1/CXCR4 signaling plays a critical role in skeletal mechanoadaptation, and may represent a unique therapeutic target for prevention and treatment of age-related and disuse bone loss. © 2013 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 31:1828–1838, 2013

Mechanical loading is a key regulator of bone mass[1]; however, aging bone is less responsive to mechanical stimuli[2] resulting in age-related bone loss and increased fracture risk.[3] One in two women and one in four men over the age of 50 will suffer an osteoporosis- or low bone mass-related fracture, with estimated total costs of treatment at more than $17 billion per year.[4] In addition, fracture treatment often requires long immobilization periods, which further exacerbates bone loss in this patient population.[5]

Exercise-induced bone accrual in long bones occurs at the periosteal and endosteal surfaces, and within the trabecular envelope.[6] Mechanical stimulation activates bone-lining cells to divide and differentiate into bone-forming osteoblasts,[7] and compelling evidence suggests that osteoprogenitors are recruited to bone surfaces from the marrow proper in response to loading.[8] More recent data demonstrate that even progenitors from distant sites are recruited to the loaded bone via the circulation.[9] There are several well-described mediators of osteoblast proliferation, differentiation, and activity, but relatively little is known about underlying molecular mechanisms regulating osteoprogenitor mobilization in the context of mechanoadaptation. A more thorough understanding of these mechanisms will likely uncover novel therapeutic targets for osteoporosis and disuse bone loss.

An emerging player in skeletal progenitor cell activation and recruitment is stromal cell-derived factor-1 (SDF-1), a potent chemokine that is expressed in various cell types including endothelial cells, pericytes,[10] bone marrow stromal cells,[11] fibroblasts,[12] endomysial cells,[13] and osteoblasts.[14] SDF-1 is upregulated in injured tissue, and has been shown to play an important role in progenitor homing, hematopoiesis, angiogenesis, and wound healing.[7, 15] SDF-1 has two known receptors, CXCR4 and CXCR7. CXCR4 is a seven-pass G protein-coupled transmembrane receptor and is expressed in a variety of cell types including stromal, endothelial, hematopoietic, neuronal, and mesenchymal stem cells.[4, 16, 17] Upon binding, SDF-1α activates multiple pathways involved in adhesion, migration, and cell survival including focal adhesion kinase (FAK), PI3K, MEK, and Jak/Tyk.[17] CXCR7 signaling does not result in G-protein dependent signaling[18] but instead is thought to internalize SDF-1 and act as a scavenger receptor for SDF-1[19]; thus, potentially playing an important role in controlling SDF-1 gradients. SDF-1 and CXCR4 global knockouts exhibit severe developmental defects in the heart and nervous system and die perinatally due to impaired lymphopoiesis and myelopoiesis,[7] demonstrating a critical role for SDF-1/CXCR4 signaling in development.

With regard to bone, SDF-1 is necessary for BMP-2-induced osteogenic differentiation.[20] In vivo studies show that anabolic PTH treatment enhances SDF-1 expression in the tibial growth plate and metaphysis,[21] CXCR4-expressing osteoprogenitors are recruited to sites rich in SDF-1, and SDF-1 promotes revascularization of injured tissue by mobilizing CXCR4-positive proangiogenic cells.[22] Furthermore, inhibition of SDF-1/CXCR4 signaling has been shown to attenuate fracture healing.[12, 23, 24] Taken together, these results indicate a critical role for SDF-1 in stem cell recruitment and osteoblast differentiation; however, its involvement in skeletal mechanoadaptation has not been characterized. If we could orchestrate SDF-1 expression in bone through application of mechanical stress, then theoretically, we could selectively stimulate MSC homing and differentiation, which would result in a targeted augmentation of osteoporotic bone.[25]

The objective of this study was to characterize the function of SDF-1 signaling in mechanically induced osteogenesis, with the idea that the SDF/CXCR4 axis may represent a novel therapeutic target in the prevention and treatment of osteoporosis and disuse bone loss. With this objective in mind, we posed several questions. First, which skeletal cells express SDF-1 and CXCR4? Are expression levels of SDF-1 and CXCR4 modulated by mechanical stimulation? Does altered SDF-1/CXCR4 signaling affect load-induced bone formation? If SDF-1 signaling does affect load-induced bone formation, does it do so by regulating proliferation and/or osteogenic differentiation of progenitors? To address these questions, we first characterized spatial expression of SDF-1 and CXCR4 protein in the mouse tibia. We next determined whether SDF-1 and CXCR4 expression is modulated by a mechanical stimulus using both in vitro and in vivo approaches. To determine whether SDF-1/CXCR4 signaling can regulate load-induced bone formation, we performed in vivo ulnar loading experiments in C57BL/6 mice that were treated with AMD3100, a CXCR4 antagonist, and assessed periosteal bone formation rates. We then assessed whether SDF-1α can regulate proliferation and osteogenic differentiation of multipotent progenitors in vitro.

METHODS

Animals

The Palo Alto Veterans Affairs Medical Center Institutional Animals Care and Use Committee approved all experimental procedures. C57BL/6 mice were obtained from The Jackson Laboratory. SDF-1-monomeric red fluorescent protein (mRFP)::CXCR4-enhanced green fluorescent protein (eGFP) (SRXG) mice (gift from Dr. Richard Miller, Northwestern University) express SDF-1α-mRFP1 in cells where any of the SDF-1 isoforms (α, β, γ) are expressed, and CXCR4-eGFP in cells where CXCR4 is expressed.[26] SRXG mice were used to assess basal protein expression of SDF-1 and CXCR4 in long bones (see Fluorescence microscopy). Siblings were housed in groups of up to five animals and they had ad libitum access to standard mouse chow and water. Animals were weighed at 16 weeks of age and just prior to euthanasia.

In Situ Ulnar Strain Analysis

The murine axial compression ulna loading model[27, 28] was used in our studies to activate new periosteal bone formation (Fig. 1). Mechanical strains achieved on the medial surface of the ulna during in vivo loading were estimated using a load-strain calibration procedure as described previously.[28] Briefly, immediately following anesthesia and euthanasia by cervical dislocation, the forelimbs were dissected free at the shoulder, keeping the soft tissue surrounding the forearm intact. The musculature was retracted exposing only the medial diaphysis of the ulna, and a 120 Ω single-element strain gage (EA-06-015DJ-120, Vishay Measurements Group, Raleigh, NC) was bonded to the surface with cyanoacrylate (M-Bond 200, Vishay Measurements Group) centered approximately 3.5 mm distal to the insertion point of the brachialis muscle. Previous studies showed that this was the most reliable method of positioning the gage at the ulnar midshaft.[29] Five 16-week-old mice from each experimental group were used for strain analyses. Each gage was conditioned with a 0.8 V bridge excitation voltage and amplified with a gain of 300× using a signal conditioner (Model 2210, Vishay Measurements Group). The amplified analog gage signals were digitized using an AD-DA board (aISA-A57, Adtek-System Science, Kanagawa, Japan) and fed into an oscilloscope (Agilent Infiniium 54830B DSO, Santa Clara, CA). With the strain gage voltage zeroed, each forearm was axially loaded using a mechanical loading system (Bose ElectroForce 3200) with increasing load levels (1N, 1.5N, 2.0N, 2.5N, and 3.0N). The average peak-to-peak voltage was observed on the oscilloscope and recorded. Voltage data for each of the loading waveforms were converted to strain values using a supplied conversion factor (1 V = 1,000 µϵ) confirmed by electronic shunt calibration of the measuring hardware and by calculated strains using an aluminum cantilever.

Figure 1.

In vivo mouse ulnar loading model. The ulna-radius complex is axially loaded across the flexed carpus and olecranon. Due to the natural curvature of the ulna in the medial direction, bending about the craniocaudal axis creates compressive and tensile bending strains on the medial and lateral surfaces, respectively. Most new bone formation occurs on the medial and lateral surfaces where strains are highest.

Based on a procedure previously described,[30] peak strains induced on the medial aspect of the periosteal surface during loading on the plane on which bone formation rates are determined (see In Vivo Ulnar Loading Section) were estimated using applied load, moment arm, distance to the neutral axis, flexural modulus, and second moment of area.

In Vivo Ulnar Loading

Two in vivo loading experiments were performed (Studies 1 and 2). The right mouse forearm was secured between two metal cups for in vivo compressive cyclic loading controlled via load feedback using a 50 lb load cell (Honeywell Sensotec, Columbus, OH). The load was applied (2.8 N peak-to-peak, 2 Hz, 120 cycles) axially across the olecranon and flexed carpus while the animal was under isoflurane anesthesia (Forane, Baxter International). Due to the natural curvature of the ulna with lateral convexity, loading in this manner causes bending about the craniocaudal axis creating compressive and tensile bending strains on the medial and lateral surfaces, respectively.[31] Most new bone formation occurs on the medial and lateral surfaces where bending strains are highest.

In Situ SDF-1 mRNA Expression in Response to Mechanical Loading

The right forearm of 16-week-old C57BL/6 mice (n = 2) was subjected to one bout of loading (2.8 N peak-to-peak, 2 Hz, 120 cycles), while the left ulna served as an internal control. Animals were euthanized at 24 h post-loading by CO2 asphyxiation, cervical dislocation, and cardiac perfusion with 0.4% paraformaldehyde (PFA). Ulnas were dissected and stored in 0.4% PFA for 24 h and embedded in paraffin for analysis of SDF-1 mRNA expression (see In Situ Hybridization Section).

Antagonism of CXCR4 Signaling and Effects on Load-Induced Periosteal Bone Formation

Animals were divided into two groups. Animals in the treatment group (n = 6) were treated with AMD3100 (5 mg/kg, intraperitoneal [IP], daily), a CXCR4 antagonist, beginning 1 day prior to the first day of loading and ending on the day of euthanasia. Animals in the control group (n = 5) were treated with vehicle (PBS, IP, daily). The right forearm was loaded on three consecutive days (2.8 N peak load, 120 cycles, 2 Hz), while the left ulna served as an internal nonloaded control. All animals were given in vivo sequential bone labels at 7 (calcein, 30 mg/kg, IP) and 14 (alizarin, 50 mg/kg, IP) days after the first day of loading. Animals were sacrificed 21 days after the first day of loading, and the right and left ulnas were harvested, fixed in 10% neutral buffered formalin for 48 h and stored in 70% ETOH at 4°C until processed for histomorphometric analysis (see Histomorphometry).

In Situ Hybridization

Loaded and nonloaded ulnas of C57BL/6 mice (Study 1) were decalcified in 19% ethylenediaminetetraacetic acid (EDTA) at 4°C for 11–14 days. The solution was changed every 48 h. Ulnas were then dehydrated in a graded alcohol series, and embedded in paraffin as described previously.[32] Ulnas were serially sectioned longitudinally into 10-micron-thick sections. DIG-labeled sense and antisense RNA probes were prepared from cDNA templates for SDF-1 and CXCR4[33] and in situ hybridization was performed.[32] For each section probed with antisense RNA probe, an adjacent serial section was probed with sense as a background control. Ulnar cortical bone sections were qualitatively evaluated for SDF-1 expression. In addition, five regions of interest (ROIs) with an area of 0.3 mm2 per bone were evaluated in loaded and nonloaded ulnas in longitudinal section.

Fluorescence Microscopy

Long bones from SRXG mice were examined for basal protein expression of SDF-1 and CXCR4. Bones were decalcified in 19% EDTA, washed in PBS and transferred to a 30% sucrose solution in PBS and stored at 4°C for 24 h. Bones were then transferred to a 1:1 mixture of optimal cutting temperature (OCT) media and 30% sucrose in PBS and placed on a rocking platform for 30 min. The bones were then oriented in tissue embedding molds containing OCT media and placed on liquid nitrogen. Ulnas were serially sectioned transversely into 10-micron-thick sections using a cryostat (Leica CM-3050-S), and SDF-1-mRFP and CXCR4-eGFP were visualized using a fluorescent microscope (Nikon Eclipse 80i, Melville, NY).

Histomorphometric Analysis

Loaded and nonloaded ulnas of C67BL/6 mice were dehydrated in sequential ascending concentrations of ethanol (70%, 80%, 90%, and 100%) and xylene and embedded undecalcified in methylmethacrylate (MMA, K-Plast, Delaware Diamond Knives). Three 90-µm thick sequential transverse sections were cut at the midshaft using an Isomet Precision Saw (Buehler Ltd., Lake Bluff, IL) The sections were ground to a final thickness of 50 µm and then mounted unstained on standard microscope slides. One section per ulna was analyzed at a magnification of 10× using a Nikon TE-2000/C1 confocal microscope (Nikon, Inc.). Static histomorphometric variables on the periosteal surface were obtained using Image J (National Institutes of Health), and dynamic bone formation indices calculated. The static variables measured included total bone perimeter (B.Pm,mm), single label perimeter (sL.Pm,mm), double label perimeter measured along the innermost label (dL.Pm,mm) and the interlabel width (Ir.L.Wi, µm), which is the distance between the first and second label. When only single labels were present on a bone section, mineral apposition rate was estimated as the minimum value observed in that particular experimental group.[28] Dynamic variables calculated were mineralizing surface (MS/BS = 100 × [0.5 × sL. Pm + dL.Pm]/B.Pm, %), mineral apposition rate (MAR = dL.Ar/dL.Pm/days between labels, µm/day) and bone formation rate (BFR/BS = MAR × [MS/BS] × 3.65 µm3/µm2/year). Relative values (rMS/BS, rMAR, and rBFR/BS) for each animal were calculated by subtracting nonloaded (left) from loaded (right) values to control for individual differences between animals.

Structural Properties of the Ulna

Load-induced changes in cortical cross sectional geometry were analyzed in animals from Study 2. Changes in geometry are represented by changes in maximum and minimum second moments of area (Imax, Imin), measures of resistance to bending, about two perpendicular principal axes in the plane of the bone section. One section from the loaded (right) ulnar midshaft was imaged at a magnification of 10× using a Nikon TE-2000/C1 confocal microscope (Nikon, Inc.). Initial and final cortical bone geometry was established by observing bone labels administered at the beginning and end of loading period as previously described.[28] Images of initial and final cross-sections were imported into Image J (Scion Corporation, Bethesda, MD), and total area encompassed by the periosteal surface (mm2), cortical area (mm2), medullary area (mm2), Imin (mm4), and Imax (mm4) were calculated for each image using a customized macro. Percent changes in all variables before and after loading were calculated.

Oscillatory Fluid Flow-Induced SDF-1 and CXCR4 Expression in MLO-Y4 Osteocyte-Like Cells and Multipotent C3H10T1/2 Cells

All fluid flow experiments were performed under sterile conditions. In the first experiment, MLO-Y4 osteocyte-like cells (gift from Dr. Lynda Bonewald, UMKC) were plated at 2,500/cm2 onto glass slides coated with rat type I collagen and grown to 80% confluence in growth media (αMEM + 5% FBS + 5% FCS + 4 mM L-glutamine + 1% P/S). Twenty-four hours before flow, cells were cultured in serum-free media (αMEM + 0% serum + 4 mM L-glutamine + 1% P/S). On the day of flow, slides were maintained in static culture or loaded individually into parallel plate flow chambers[34] charged with low-serum media. Chambers were connected in series to rigid-walled tubing and a Hamilton glass syringe and plunger, which was driven by a custom-built oscillatory loading device. Cells were exposed to 1 h of oscillatory fluid flow[35] (10 dyn/cm2, 1 Hz, 37°C) or no flow conditions.

For SDF-1 mRNA expression levels, cells subjected to flow and no flow controls were returned to the tissue culture incubator. Cells were lysed at 2, 4, and 24 h. Total RNA was isolated using TRI Reagent® (Sigma–Aldrich, St. Louis, MO) per the manufacturer's instructions, and expression of SDF-1 was assessed by qPCR (see Quantitative real-time PCR Section).

For SDF-1 protein expression levels, slides were then placed in dry 100 mm dishes, and cells were covered with 2 ml fresh low-serum (αMEM + 0.25% FBS + 0.25% FCS + 4 mM L-glutamine + 1% P/S) media and incubated at 37°C for 7 days. One milliliter of fresh low-serum media was added at day 4 to prevent sample from drying out. At the end of 7 days, media was collected and then cells were washed with PBS and treated with 0.5% Triton-X for 30 min. Cells were scraped and collected. Secreted (media) and cellular (lysate) SDF-1α protein fractions were determined by ELISA (see SDF-1α Protein Expression by ELISA Section).

C3H10T1/2, Clone 8 cells (CCL-226, ATCC) were plated at 2,200/cm2 onto glass slides coated with rat type I collagen and grown to 80% confluence in growth media (αMEM + 10% FBS + 4 mM L-glutamine + 1% P/S). Twenty-four hours before flow, cells were cultured in serum-free media (αMEM + 0% serum + 4 mM L-glutamine + 1% P/S). On the day of flow, slides were maintained in static culture (n = 7) or loaded individually into parallel plate flow chambers (n = 7) charged with low-serum media, and exposed to 1 h of oscillatory fluid flow as described above. Total RNA was collected from cells after a 3-h post-flow incubation using TRI Reagent® (Sigma–Aldrich) per the manufacturer's instructions, and expression of CXCR4 was assessed by qPCR (see Quantitative real-time PCR Section).

Proliferation and Differentiation of hMSCs in Response to SDF-1α Treatment

Human mesenchymal stem cells (hMSCs) were obtained from a commercial source (Lonza, Inc., Allendale, NJ) and cultured in proliferation media (PM, low glucose DMEM + 10% FBS + 1% Pen Strep). Passages 1–5 were used for all experiments. At 80% confluence, cells were cultured in low serum (0.5% FBS) media for 48 h. hMSCs were then treated with recombinant human SDF-1α (R&D Systems, Minneapolis, MN) at 200 ng/ml in 0.1% BSA or AMD3100, a CXCR4 antagonist, at 200 ng/ml for 48 h, and cells were fixed with 4% PFA and probed for Ki-67 to assess proliferation. Ki-67 is a nuclear protein that is expressed during all phases of the cell cycle except the resting phase.[36] At 75% confluence, cells were rinsed twice in cold PBS, fixed in 4% PFA, and permeabilized in 0.1% Triton-X (Sigma). Cells were incubated in blocking solution containing 1% BSA in PBS for 1 h at room temperature and probed using a primary antibody against Ki-67 (Abcam, Cambridge, MA ab16667) at 1:1,000 and a donkey anti-rabbit-Alexa Fluor 488 (Molecular Probes, Carlsbad, CA, A21206) secondary antibody at 10 μg/ml. Cells were counterstained with DAPI to visualize the nucleus. Images were captured at 10× magnification and analyzed using ImageJ. The number of Ki-67-GFP+ positive cells per treatment group was determined and normalized by the total number of DAPI-stained cells. In a separate experiment, hMSCs were treated for 24 h with PBS, rhSDF-1α, AMD3100, and rhSDF-1 + AMD3100 (n = 6 per treatment group) at 200 ng/ml, and total RNA was isolated using TRI Reagent® per the manufacturer's instructions. Expression of osteogenic genes runx2, osterix, and osteopontin were determined by quantitative real-time PCR using the standard curve method with 18s serving as an endogenous control (see Quantitative real-time PCR Section).

Quantitative Real-Time PCR

For cell culture gene expression experiments, cDNA was synthesized from total RNA using the High Capacity cDNA Reverse Transcription kit (ABI). cDNA samples were then amplified by real time PCR (ABI Prism 7900 Sequence Detection System). Commercially available TaqMan gene expression assays containing unlabeled PCR primers and TaqMan probes for SDF-1 (Mm00445553_m1), CXCR4 (Mm01996749), runx2 (Hs00231692), osterix (Hs01866874), osteopontin (Hs00959010), 18s (4333760-0707022), and β-actin (Mm00517812_m1) were used (ABI). Amplification curves for control and genes of interest were recorded and relative expression levels between groups were quantified using the relative standard curve method (ABI Prism 7700 User Bulletin #2). All samples were normalized to endogenous controls 18s ribosomal RNA levels or β-actin levels. All standards and samples were run in replicates of three to nine.

SDF-1α Protein Expression by ELISA

MLO-Y4 osteocyte-like cells (gift from Dr. Lynda Bonewald, UMKC) were grown in aMEM + 5% FBS + 5% FCS and seeded onto collagen-coated glass slides at a density of 160 k cells per slide 48 h prior to experimentation. At 80% confluence cells were exposed to oscillatory fluid flow (10 dyn/cm2, 1 Hz) or no flow conditions at 37°C for 1 h.[37] Cells were then covered with 2 ml of media and incubated at 37°C for 7 days; 1 ml media was added after 3 days of incubation. After 7 days of incubation, media was removed and frozen at −20°C until ELISA. Cells were then lysed in 0.5% Triton X 100 in PBS for 10 min and cell lysate was frozen at −20°C until ELISA. Total SDF-1α, released and cellular forms, was determined by ELISA (R&D Systems) per the manufacturers instructions and normalized by the total cell protein determined by the Bicinchoninic Acid (BCA) total protein assay (Thermo Scientific Pierce, Waltham, MA).

Data Analysis

Data were checked for normality and constancy of variance. SPSS® Base 16.0 Statistical Software (SPSS, Inc., Armonk, NY) was used for all analyses. Relative bone formation rates, ELISA, and qPCR data were analyzed using an unpaired Student's t-test. Changes in cortical geometric properties at pre- and post-loading were analyzed using a repeated measures two-way ANOVA with treatment and loading as the main factors. An interaction effect (treatment × loading) was also calculated. Differences in average percent change in cortical geometric properties between vehicle- and AMD100-treated groups were analyzed by an unpaired Student's t-test. Differences in the number of SDF-1 positive osteocytes in loaded and nonloaded ulnas were also analyzed by an unpaired Student's t-test. Statistical significance was assumed for p < 0.05. Data are presented as mean ± standard error of the mean (SE) unless noted otherwise.

RESULTS

Basal Expression of SDF-1 and CXCR4 in Bone and Marrow Cells

SDF-1-mRFP protein is expressed in cortical and trabecular osteocytes, in cells within the marrow cavity, and in cells adjacent to the endosteal/periosteal surface (Fig. 2A). SDF-1 expression was observed in cortical osteocytes and a subset of marrow cells. CXCR4-eGFP was expressed in cells in the marrow in both the cortical and trabecular compartments (Fig. 2A,B). SDF-1 and CXCR4 appeared to be co-expressed in some cells in the marrow (Fig. 2B). CXCR4 mRNA was expressed in cells on the endosteal surface and in the marrow, as well as in some cortical osteocytes and cells on the periosteal surface (Fig. 2C). An H&E stained section is included as a reference for cortical and trabecular sites shown in Panels A and B (Fig. 2D). Primary marrow cells grown in culture showed abundant CXCR4-eGFP cytoplasmic expression (Fig. 2E) as well as SDF-1 expression (Fig. 2F).

Figure 2.

Endogenous expression of SDF-1 and CXCR4 in SRXG mouse long bones (A–C), and primary adherent bone marrow cells (BMC) in culture (E,F). Transgenic SRXG mice express SDF-1-mRFP1 (red) in cells where SDF-1 is expressed, and CXCR4-eGFP (green) in cells where CXCR4 is expressed. (A) SDF-1 is expressed in cortical (Ct) bone osteocytes and in cells in the marrow (M) (white arrows). SDF-1 expression was not observed on the periosteal (Ps) surface. CXCR4 expression was observed primarily in cells within the marrow. (B) SDF-1 expression was observed in trabecular (Tb) bone (white arrows) and in some cells in the marrow (M) depicted in yellow (overlay with green). CXCR4 expression was observed in the marrow. GP, growth plate. (C) CXCR4 mRNA is expressed in cells on the endosteal surface (black arrows 1, 2) and in some cells at the periosteal surface. (D) H&E stain showing black boxes positioned over cortical and trabecular compartments of interest shown in Panels A and B. (E) Primary adherent marrow cells from adult mice exhibit CXCR4 expression in the cytoplasm. (F) These primary adherent marrow cells also show cytoplasmic expression of SDF-1. Scale bar = 100 µm (A–D). Scale bar = 10 µm (E). Scale bar = 20 µm (F).

Oscillatory Fluid Flow and SDF-1α mRNA and Protein Expression in Osteocytes

We next determined the effects of mechanical stimulation in the form of oscillatory fluid flow (OFF) on SDF-1α mRNA and protein expression in the osteocyte-like cell line MLO-Y4. MLO-Y4 osteocyte-like cells subjected to flow expressed significantly greater levels of SDF-1 at 2 and 4 h post-flow compared to controls at each time point (*p < 0.01). Levels were not significantly different at 24 h post-flow (Fig. 3A). MLO-Y4 cells express SDF-1a protein in both the cellular and released fractions when held in no flow (static) culture for 7 days (Fig. 3B). In response to loading, both fractions exhibited increased amounts of SDF-1α normalized to total cell protein with the greatest difference observed in the cellular fraction. This difference was not significant, but represented a trend.

Figure 3.

SDF-1α and CXCR4 expression in response to mechanical stimulation. (A) MLO-Y4 osteocyte-like cells subjected to flow expressed significantly greater levels of SFD-1 at 2 and 4 h post-flow compared to controls at each time point (*p < 0.01). Levels were not significantly different at 24 h post-flow. (B) MLO-Y4 osteocyte-like cells exhibit increased SDF-1α release in response to oscillatory fluid flow (10 dynes/cm2, 1 h, 1 Hz, 7 day post-flow incubation) compared to cells exposed to no flow conditions. Chart displays mean of total SDF-1 (released + cellular fraction) + SE. Differences between total SDF-1 for no flow and flow groups were not significant, p = 0.341. Mean ± SE values for SDF-1 by source are: No flow (NF), released = 0.057 ± 0.029 ng/mg; NF, cellular fraction = 0.012 ± 0.002 ng/mg; Flow, released = 0.087 ± 0.036 ng/mg; Flow, cellular fraction = 0.074 ± 0.039 ng/mg. (C) Oscillatory fluid flow (OFF) (10 dynes/cm2, 1 h, 1 Hz, 24 h post-flow incubation) enhanced CXCR4 expression in C3H10T1/2 multipotent cells. *p = 0.041.

Oscillatory Fluid Flow Enhances CXCR4 mRNA Expression in C3H10T1/2 Multipotent Cells

As CXCR4 was prominently expressed in marrow cells in vivo, we next assessed whether oscillatory fluid flow (OFF) modulates CXCR4 expression in the multipotent C3H10T1/2 cell line with the idea that mechanical stimulation might “prime” progenitors for homing by upregulating CXCR4. Cells exposed to 1 h of OFF exhibited increased CXCR4 expression after a 3-h post-flow incubation compared to cells kept under no flow (static) conditions (Fig. 3C).

Peak-to-Peak Periosteal Strains Increase Linearly as a Function of Applied Load Level

To establish a relationship between applied load and resulting strains on the concave, medial periosteal surface of the ulna, a single-element strain gauge was used to measure peak-to-peak strain during cyclic compressive axial loading as a function of increasing load levels. Average peak-to-peak microstrain increased as load magnitude increased (Fig. 4) according to the relationship: microstrain (µϵ) = 1,167 × Newtons (N), R2 = 0.972.

Figure 4.

Average peak-to-peak strain measured on the medial surface of the ulnar midshaft. Strain increased as a function of load. Peak estimated strain reached during in vivo ulnar loading is 3,200 microstrain. The least-squares regression line was forced through the origin. Error bars represent standard deviation of groups at each load level.

SDF-1 mRNA Expression Is Enhanced in Osteocytes and Periosteal Cells in Response to In Vivo Mechanical Stimulation

Estimated average peak strain in C57BL/6 mice with an applied 2.8N load was 3,550 µϵ on the medial periosteal surface at midshaft, where load-induced strains are estimated to be highest.[30] In normally ambulating mice, SDF-1 mRNA was expressed in some cortical osteocytes in the ulna with very little expression on the periosteal surface (Fig. 5), similar to SDF-1 protein expression in the SRXG mouse. In response to loading, SDF-1 expression was upregulated in both cortical osteocytes and on the periosteal surface. Nonloaded and loaded ulnar cortical bone showed 216 ± 96 and 701 ± 323 (mean ± SD) SDF-1 positive osteocytes per mm2, respectively. The difference between groups was significant with a p-value of 0.0128 at α = 0.05.

Figure 5.

SDF-1 mRNA expression in mouse cortical bone subjected to mechanical stimulation. (A) In normally loaded control bones, SDF-1 mRNA (purple) is expressed in some cortical (Ct) osteocytes (large black arrows) in the ulna. Very little expression was observed at the periosteal surface (small black arrows). (B) In response to mechanical loading, SDF-1 is more broadly expressed in cortical osteocytes and periosteal cells (Ps; bottom). Scale bar = 50 µm.

Antagonism of CXCR4 Signaling With AMD3100 Attenuates Load-Induced Periosteal Bone Formation and Changes in Cortical Structural Properties

To ascertain whether SDF-1/CXCR4 signaling plays a role in skeletal mechanoadaptation, a CXCR4 antagonist was administered to wild-type mice undergoing an in vivo ulnar loading protocol. Sequential fluorochrome bone labels showed enhanced periosteal bone formation in vehicle-treated mice in response to mechanical loading (Fig. 6A). Most new bone formation occurred on the medial and lateral aspects of the ulna as depicted by double calcein labels, with new bone deposition occurring between the two labels. There was also new bone formation on the cranial and caudal periosteal surfaces, though less of it, as shown by single labels. AMD3100-treated mice also exhibited load-induced bone formation on the medial and lateral periosteal surfaces, but to a lesser degree as indicated by the smaller surface area between the two labels, suggesting that antagonism of the CXCR4 receptor attenuates new periosteal bone formation. AMD3100-treated mice exhibited significantly lower relative mineralizing surface (rMS/BS) and bone formation rates (rBFR/BS), and unchanged mineral apposition rate (rMAR; Fig. 6B). All other things being equal, bone formation rates have been shown to increase in a linear fashion with increasing peak strain levels. As a means of normalizing bone formation rates to load-induced peak strain, we estimated the average peak-to-peak strains during loading. These peak-to-peak strains were not significantly different between experimental groups (Fig. 6C); thus, differences in bone formation rates between AMD3100-treated mice and controls were a result of treatment rather than differences in peak-to-peak strain levels.

Figure 6.

(A) Calcein labeled bone formation in AMD3100-treated and control mouse ulnas. Representative images. (Top left) Non-loaded, vehicle-treated ulnas exhibited new bone formation on the periosteal and endosteal surfaces as indicated by single calcein labels. (Top right) Loaded, vehicle-treated ulnas showed load-induced bone formation on the medial and lateral periosteal surfaces, with most new bone formation occurring on the medial surface as indicated by double parallel calcein labels (white arrows). (Bottom left) Non-loaded, AMD3100-treated mouse ulnas exhibited new bone formation on the periosteal and endosteal surfaces. (Bottom right) Loaded, AMD3100-treated mouse ulnas showed load-induced bone formation on both the medial and lateral periosteal surfaces as indicated by double calcein labels (white arrows). Double parallel labels are closer together in the AMD3100-treated mouse ulnas compared to vehicle-treated controls indicating attenuated load-induced bone formation with AMD3100 treatment. (B) Load-induced periosteal bone formation rates grouped by treatment (AMD3100 or vehicle). Treatment with AMD3100, a CXCR4 antagonist, results in significantly attenuated relative (r) MS/BS and BFR/BS. Vehicle n = 5; AMD3100 n = 6. Relative values were determined by subtracting non-loaded values from loaded values.[28] rMS/BS, relative mineralizing surface; rMAR, relative mineral apposition rate; rBFR/BS, relative bone formation rate. *Significant versus vehicle-treated group at p < 0.002. **Significant versus vehicle-treated group at p < 0.001. (C) Loaded-induced periosteal bone formation rates versus peak-to-peak strain. Estimated load-induced peak-to-peak strains on the periosteal surface were not significantly different in AMD3100-treated and vehicle-treated groups. rMS/BS and rBFR/BS were significantly lower in AMD3100-treated mice compared to controls at similar strain levels. There was no significant difference in rMAR between groups. Scale bar = 250 μm.

We were also interested in the role SDF-1/CXCR4 signaling may play in regulating mechanically modulated bending properties of bone. Thus, surrogates for bone strength (cross-sectional cortical area (Ct.Ar), minimum and maximum second moments of area (Imin, Imax)) were quantified in loaded ulnas from vehicle- and AMD3100-treated mice at pre- and post-loading times (Table 1). Total area within the periosteal envelope (Tt.Ar) and medullary area (Me.Ar) were also calculated. Pre-loaded values for cortical geometry were not significantly different between treatment groups. Loading had a significant effect on all variables except medullary area. There was a significant interaction effect on cortical area and Imin, whereas treatment had no significant effect on any of the variables. When comparing pre-load versus post-load values, cortical area, Imin, and Imax increased significantly in response to loading in both treatment groups. Pre- versus post-vehicle-treated: Ct.Ar, 11.2%, p = 0.0002; Imin, 28.7%, p = 0.007; and Imax, 14.2%, p = 0.0004. Pre- versus post-AMD3100-treated: Ct.Ar, 7.6%, p = 0.0068; Imin, 18.7%, p = 0.0027; and Imax, 6.8%, p = 0.0014; however, AMD3100-treated animals exhibited significantly lower average percent increases (%Δ) in Ct.Ar (p = 0.020), Imin (p = 0.009), and Imax (p = 0.010).

Table 1. Ulnar Midshaft Geometric Properties in Vehicle- and AMD3100-Treated Mice at Pre- and Post-Loading
 VehicleAMD3100p-Value
 Pre-LoadPost-LoadPre-LoadPost-LoadTreatmentLoadingInt
  1. Data are presented as mean (SD). %Δ, average percent difference in geometric properties pre- and post-loading within experimental groups; Ct.Ar, cortical area; Imin, minimum second moment of area; Imax, maximum second moment of area.
Ct.Ar0.226 (0.019)0.251 (0.025)↑ 11.2 (2.3)0.236 (0.030)0.254 (0.030)↑ 7.6 (1.7)0.680<0.00010.041
Tt.Ar0.263 (0.035)0.286 (0.039)↑ 9.2 (2.0)0.279 (0.037)0.297 (0.037)↑ 6.5 (1.4)0.569<0.00010.071
Me.Ar0.037 (0.017)0.036 (0.015)↓ 2.0 (5.1)0.043 (0.011)0.043 (0.011)↑ 0.5 (1.2)0.4550.4660.084
Imin0.0027 (0.0009)0.0035 (0.0012)↑ 28.7 (4.7)0.0037 (0.0014)0.0044 (0.0014)↑ 18.7 (6.3)0.231<0.00010.031
Imax0.0128 (0.0031)0.0146 (0.0039)↑ 14.2 (4.0)0.0136 (0.0049)0.0145 (0.0049)↑ 6.8 (3.3)0.887<0.00010.351

Proliferation and Differentiation of hMSCs in Response to SDF-1α Treatment

To assess whether SDF-1 has an effect on progenitor proliferation, hMSCs in culture were treated with either SDF-1 or PBS. Proliferation as determined by the percentage of Ki-67 positive cells relative to total number of cells was not significantly different between SDF-1α-treated (69.3 ± 9.24; 4.8 k cells counted) and PBS-treated (73.4 ± 4.55; 5.5k cells counted) groups (p = 0.53). SDF-1α did not significantly affect expression of osteogenic genes as assessed by real-time qPCR (data not shown).

DISCUSSION

Relatively little is known about molecular mechanisms regulating osteoprogenitor activation, mobilization, and recruitment in the context of skeletal mechanoadaptation. SDF-1 is a stem cell homing factor that recruits CXCR4-expressing progenitors to sites of tissue regeneration and healing, and is essential to BMP-2-induced osteogenic differentiation[20] and fracture healing.[24] In this study, we asked whether SDF-1 plays a role in mechanically induced osteogenesis as a first step towards identifying unique targets for the prevention and treatment of osteoporosis and immobilization-induced bone loss.

To address this question, we first characterized spatial location of SDF-1-positive and CXCR4-positive cells in long bones to establish which cells might initiate communication through release of SDF-1 and which cells are capable of responding to the stimulus. Using the SRXG transgenic mouse, we showed that SDF-1 is expressed in cortical and trabecular osteocytes, endosteal cells, and in a subset of bone marrow cells (Fig. 2). These findings are consistent with studies describing SDF-1 expression in osteoblasts[38] and marrow cells.[39] CXCR4 expression in long bone is observed on periosteal and endosteal surfaces (Fig. 2D) and in marrow cells (Fig. 2C). These expression patterns suggest that osteocytes and cells in the marrow can communicate with CXCR4-expressing progenitors in the marrow, as well as with cells on the periosteal and endosteal surfaces.

The distinct expression of SDF-1 in osteocytes (Fig. 2) suggests that osteocytes are potentially an important source of SDF-1 in the bone microenvironment. In fact, osteocytes are already known to express and release several factors that influence bone metabolism including SOST,[40] DMP-1,[41] MEPE,[42] PHEX, and FGF-23,[43] and whose levels are regulated, in part, by mechanical stimulation.[44] Furthermore, load-induced interstitial fluid flow can direct movement of these factors from osteocytes through the lacunocanalicular network to effector cells on bone surfaces.[45] Our data showing that mechanical stimulation upregulates SDF-1α mRNA and protein expression and release in MLO-Y4 osteocyte-like cells in vitro (Fig. 3A), and SDF-1 mRNA expression in osteocytes and periosteal cells in vivo (Fig. 5), coupled with the fact that SDF-1α is small enough (8 kDa) for travel through the lacunocanalicular space,[46] suggests that osteocyte-specific SDF-1 may serve as an important regulator of skeletal mechanoadaptation. Furthermore, our data showing that CXCR4 expression is enhanced in multipotent cells in response to mechanical stimulation (Fig. 3B) suggests that loading might prime progenitors to be more responsive to SDF-1. Further data are needed to confirm this hypothesis.

In this study, we demonstrated that systemic treatment with the CXCR4 antagonist AMD3100 significantly attenuated load-induced periosteal bone formation (Fig. 6A,B). AMD3100 treatment led to significantly smaller increases in ulnar cortical area, Imin, and Imax (Table 1) in response to loading, indicating that SDF-1/CXCR4 signaling is an important regulator of skeletal mechanoadaptation, though whether primary effects are on progenitor homing, proliferation, or differentiation remains to be elucidated. Our findings are consistent with other reports showing a role for SDF-1 signaling in osteogenesis. Inactivation of the CXCR4 receptor in osteoblasts in mice leads to lower mineral apposition rate, reduced trabecular and cortical bone mass, and a disorganized epiphyseal growth Plate.[47] In vivo homing data show that SDF-1 recruits CXCR4-expressing progenitors in several contexts including ischemia,[48] ectopic bone formation,[49] and fracture healing.[23, 50]

Mice in the AMD3100-treated group had slightly larger pre-loading cortical geometric properties compared to control mice, and we recognize that these small differences could account, in part, for load-induced changes in bone formation and geometry. However, these pre-loading differences as well as estimated periosteal strain at all load levels were not significant between experimental groups, and thus we believe they had only minor effects on outcome measures.

We elected to report load-induced increases as a %change because it inherently normalizes for the nonloaded limb in the same animal ((loaded−nonloaded)/nonloaded). If we separately compare magnitudes (loaded−nonloaded), the differences between groups for Ct.Ar and Imax are significant, further supporting our findings that normal load-induced bone formation and mechanical adaptation rely on SDF-1/CXCR4 signaling.

Our bone formation data (Fig. 6B) show that AMD3100 treatment had the greatest effect on mineralizing surface (MS/BS), which suggests that number of active osteoblasts, rather than the bone-forming capacity of individual osteoblasts, was attenuated with treatment. Accordingly, we expected SDF-1α treatment to affect osteogenic differentiation and/or proliferation, but our in vitro results did not support this hypothesis. That is, treatment of hMSCs with SDF-1α did not result in significantly enhanced proliferation or osteogenic differentiation, though there was a trend of increased osteogenic gene expression with SDF-1α treatment.

Published data on osteogenic activity of SDF-1 shows enhanced differentiation only when other factors are present in combination with SDF-1.[11] Primary osteoblasts derived from CXCR4fl/fl::Osx-Cre mice showed decreased proliferation and impaired osteoblast differentiation in response to BMP2 and BMP6 stimulation. SDF-1α and β splice variants have been shown to regulate BMP-2-induced osteogenic differentiation of primary bone marrow stromal cells.[20, 51] In addition, AMD3100 treatment of osteoblast-like MC3T3-E1 cells grown in osteogenic media resulted in down-regulation of osteogenic genes.[52] Thus, in the context of load-induced periosteal bone formation, effects on bone formation appear to be indirect and other factors are likely required for a robust bone formation response to SDF-1 in vivo.

One aspect that we did not address in this study was whether AMD3100 treatment affected progenitor recruitment in response to loading. While there are published data suggesting that mechanical loading results in recruitment of new osteoblasts to the endosteal surface from the marrow,[8] there is still relatively little data showing significant load-induced recruitment of progenitors, endogenous or transplanted, to the periosteal surface. Enhanced stem cell homing to bone surfaces may be an important strategy in the prevention and treatment of osteoporosis. A recent study showed that a synthetic ligand for α4β1 integrin conjugated to alendronate (LLP2A-Ale) significantly enhanced homing of transplanted hMSCs to bone surfaces resulting in significant increases in bone formation on periosteal, endosteal, and trabecular surfaces.[25] Further data are needed to determine whether SDF-1 serves as a homing factor in bone in the context of mechanoadaptation or whether it is specific to a healing response.

In conclusion, we have shown that SDF-1/CXCR4 signaling plays an important role in bone mechanoadaptation. Whether in vivo effects are primarily on progenitor recruitment, proliferation, and/or osteoblast differentiation remains to be determined. SDF-1 signaling has been shown to be important in several other contexts including hematopoiesis, angiogenesis, and fracture healing, and the SDF-1/CXCR4 axis may represent a unique target in the prevention and treatment of osteoporosis and disuse bone loss.

ACKNOWLEDGMENTS

The authors thank Vasavi Ramachandran for assistance with animal husbandry, and Kris Morrow for original artwork. This work was supported by a Stanford Center on Longevity Fellowship (A.B.C.) and Veterans Affairs Career Development Award-2 A6824W (A.B.C.).

Ancillary