Mechanical input is known to regulate skeletal mass. In vitro, application of strain inhibits osteoclast formation by decreasing expression of the ligand RANKL in bone stromal cells, but the mechanism responsible for this down-regulation is unknown. In experiments here, application of 1.8% equibiaxial strain for 6 h reduced vitamin D-stimulated RANKL mRNA expression by nearly one-half in primary bone stromal cells. Application of strain caused a rapid activation of ERK1/2, which returned to baseline by 60 minutes. Adding the ERK1/2 inhibitor PD98059 30 minutes before strain delivery prevented the strain effect on RANKL mRNA expression, suggesting that activation of ERK1/2 was required for transduction of the mechanical force. Mechanical strain also activated N-terminal Jun kinase (JNK) that, in contrast, did not return to baseline during 24 h of continuous strain. This suggests that JNK may represent an accessory pathway for mechanical transduction in bone cells. Our data indicate that strain modulation of RANKL expression involves activation of MAPK pathways.
THE NORMAL adult skeleton undergoes bone remodeling to maintain the functionality. Biomechanical signals involved in maintenance of skeletal structure are generated during dynamic loading. In states in which skeletal loading is decreased such as prolonged bed rest(1) or in paraplegics,(2) bone resorption is initiated resulting in local recruitment of osteoclasts. This process of adaptation to new loading conditions also involves decreased bone formation and culminates in decreased bone mass with increased susceptibility to fracture. We have been interested in the mechanical signals generated during normal loading that can prevent the process of osteoclastogenesis. In vitro application of dynamic mechanical strain to murine marrow cultures robustly inhibits osteoclastogenesis.(3) The bone stromal cell is targeted directly and responds with decreased expression of the dominant molecule controlling osteoclast formation, RANKL (also termed osteoprotegerin ligand [OPGL], osteoclast differentiating factor [ODF], and TNF-related activation-induced cytokine [TRANCE]).(4) The signaling mechanisms activated by mechanically straining the stromal cell that are involved in down-regulating expression of endogenous RANKL expression are the subject of this work.
Application of mechanical force is known to stimulate multiple transduction cascades in many cell types.(5) During the past several years, phosphorylation cascades involved during activation MAPKs have been shown to be activated by both mechanical strain and shear. Fluid shear of endothelial cells causes activation of three members of the MAPK family, N-terminal c-Jun kinase (JNK),(6) extracellular signal-regulated kinases 1 and 2 (ERK1/2),(7) and big MAPK (BMK-1).(8) Applying substrate strain to attached cells also activates members of the MAPK superfamily. In vascular smooth muscle cells, strain rapidly activates both ERK1/2 and JNK.(9) Because the cellular components in bone respond to both shear(10) and strain forces,(4) it is likely that MAPK activation is used during transduction of mechanical force in the skeleton.
Stromal cells from the bone marrow harbor the progenitors for osteoblasts, which form and maintain the skeleton.(11) Early in their development as well as in their most differentiated state, these stromal cells can be stimulated to display RANKL,(12,13) in essence controlling the bone-remodeling cycle by regulating the appearance of osteoclasts. In this work, we show that strain, which we have shown previously inhibits expression of RANKL by stromal cells,(4) causes a rapid and transient activation of ERK1/2. ERK1/2 activation in turn mediates strain-induced inhibition of RANKL; inhibition of ERK1/2 activation prevents strain alterations in RANKL mRNA. The magnitude of the strain required to activate ERK1/2 in bone stromal cells is significantly lower than that reported for cells of the vasculature, suggesting that the small strains necessary for maintaining the skeleton(14) may indeed be processed through this intracellular signaling pathway.
MATERIALS AND METHODS
Materials and reagents
Antibodies to total ERK1/2, p38MAPK, phosphorylated JNK, and BMK-1 were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA), and antibodies to phosphorylated ERK1/2, p38MAPK, and BMK-1 were from New England Biolabs (Beverly, MA, USA). ERK1/2 inhibitor PD98059 and p38MAPK inhibitor SB202190 were purchased from Calbiochem/Novabiochem Corp. (La Jolla, CA, USA). FBS was from Atlanta Biologicals (Atlanta, GA, USA) and other chemicals and supplies were from Sigma (St. Louis, MO, USA).
Bone marrow stromal cell culture
To generate primary stromal cell cultures, murine marrow cells collected from the tibias and femurs of 3- to 5-week-old male C57BL/6 mice were plated in 6-well plates. After 30 minutes, nonadherent cells containing the stromal elements were collected and plated at 1.4 million cells/cm2 in Bioflex collagen I-coated plates (Flexcell Corp., McKeesport, PA, USA). Twenty-four hours later, all nonadherent cells were washed off and discarded and the remaining stromal cells were cultured for 1 week until nearly confluent in α-modified essential medium (α-MEM/10% FBS. These stromal cells represent <5% of the initial marrow collection,(15) have osteoblastogenic capacity, and can form mineralized nodules under appropriate conditions.(16)
For experiments, stromal cells were grown for 5 days, after which 10 nM of 1,25-dihydroxyvitamin D was added to stimulate RANKL expression. On day 6 or day 7, strain regimens were applied. For experiments in which lysates were made for Western analyses, inhibitors were added 30 minutes before strain and the experiment was stopped as indicated. When the endpoint was to measure RANKL mRNA, strain was delivered as indicated and total mRNA was made 24 h after strain induction. In experiments in which MAPK inhibitors were added, all cultures had media (+vitamin D) exchanged after 6 h to remove all agents present.
Mechanical strain device
To generate mechanical strain in primary marrow cultures, a Flexcell bioflex instrument (Flexcell Corp., Hillsborough, NC, USA) was used. The unit was placed inside a 37°C, 5% CO2 incubator and negative pressure was applied cyclically to stretch the substrate over the edge of a loading post. This induces a uniform and biaxial 1.8% strain at 10 cycles/minute across the plate surface.(3) Controls were plated on similar membranes and kept in the same incubator but not subjected to strain.
Western blotting for MAPKs
For the MAPK assay, dishes were washed on ice with cold PBS and cells were lysed in 75 μl of lysis buffer (50 mM of HEPES, 5 mM of EDTA, 50 mM of NaCl, and 0.1% Triton, pH 7.4) containing 50 mM of NaF, 1 mM of Na3VO4, and 10 mM of NaP2O7. Aprotinin, leupeptin, phenylmethylsulfonyl fluoride (PMSF), and pepstatin were added fresh before each lysis. Cells were lysed and transferred to Eppendorf tubes and centrifuged. The supernatant was concentrated by Centricon filter with a 10,000 molecular weight (MW) cut-off before measuring protein. Ten micrograms of protein was loaded onto a 12% polyacrylamide gel for chromatography before transfer to polyvinylidene difluoride (PVDF) membrane. After blocking with 5% milk and 0.05% Tween 20, incubation with primary antibodies was performed for 60 minutes followed by 3 × 5 minutes washes in 0.2% milk and 0.2% Tween 20. Incubation with secondary antibodies occurred over 60 minutes before exposing membranes to an enhanced chemiluminescence (ECL) solution (ECL-plus; Amersham Life Sciences, Arlington Heights, IL, USA) for 5 minutes before radiography, revealing ERK bands at 42 kDa and 44 kDa. For identification of JNK components, 15 μg of cell lysate was chromatographed on 10% SDS-PAGE and transferred to PVDF membrane for incubation with anti-pJNK antibody (Ab).(17) The bound primary pJNK antibody was detected by using a donkey anti-rabbit immunoglobulin G (IgG) F(ab)2 horseradish peroxidase (HRP)-conjugated (Amersham Life Sciences) and viewed by ECL showing two species at 55 kDa (JNK2) and 46 kDa (JNK1).
Assay of JNK activity
After specific time points, stromal cells from control and strained cells were lysed in cold Triton lysis buffer (TLB) containing 20 mM of Tris-Cl (pH 7.4), 137mM of NaCl, 25 mM of sodium β-glycerophosphate, 2 mM of sodium pyrophosphate, 2 mM of EDTA, 1 mM of sodium orthovanadate, 10% glycerol, 1% Triton X-100, 1 mM of PMSF, 5 μg/ml of leupeptin, and 5 μg/ml of aprotinin.(17) The lysates were centrifuged at 12,500g for 10 minutes at 4°C to remove insoluble material. The clarified lysates (100 μg protein) were incubated with 2.5 μg of agarose-bound glutathione S-transferase (GST)-c-jun (1-79) (Calbiochem, San Diego, CA, USA) in a total volume of 400 μl of TLB overnight at 4°C on a rotating platform. After washing beads several times with TLB and kinase assay buffer (KB; 25 mM of HEPES, pH7.4, 25 mM of sodium β-glycerophosphate, 25 mM of MgCl2, 0.1 mM of sodium orthovanadate, and 0.5 mM of dithiothreitol [DTT]), 10 μl of KB with 100 μM ATP, and 5 μCi of [γ-32P]ATP were added to beads. The reactions were carried out at 30°C for 30 minutes and terminated by adding 10 μl of 2× SDS-PAGE sample buffer. The phosphoproteins were separated by SDS-PAGE for autoradiography. The phosphorylated GST-c-jun (1-79) showed a predominant band at ∼37 kDa.
Assessment of mRNA species
Semiquantitative reverse-transcription polymerase chain reaction (RT-PCR) was performed as previously indicated.(4) Briefly, after lysis in Trizol (Gibco BRL, Bethesda, MD, USA), total RNA was probed for RANKL and 18S transcripts. For RANKL, the forward primer was 5′-GCTAT TATGG AAGGC TCATG and the reverse primer was 5′-CACCA TCAGC TGAAG ATAGT. For quantification of all PCR products, 0.1 μl of [32P]deoxycytidine triphosphate (dCTP) was added to each reaction. PCR was performed on a Perkin-Elmer thermocycler 9700 (Perkin-Elmer, Norwalk, CT, USA) for RANKL at 27 cycles (within the linear range for this product) at 30 s each (94°C, 56°C, and 72°C), and 18 cycles for 18S at 30 s each (94°C, 56°C, and 72°C). PCR products were separated on a 15% polyacrylamide gel, and densitometry of [32P] signals were captured by a Molecular Dynamics PhosphorImager (Molecular Dynamics, Piscataway, NJ, USA). Samples were normalized by concurrent RT-PCR for 18S as previously mentioned.(4,18) RT-PCR for RANKL has been shown previously to reflect data obtained via Northern analysis.(4)
Real-time PCR for RANKL and 18S was performed using the iCycler (Bio-Rad Laboratories, Hercules, CA, USA). RT of 0.5 μg total RNA was performed with random decamers (Ambion, Inc. Austin, TX, USA) and superscript reverse transcriptase (Gibco BRL). For real time, PCR amplification reactions were performed in 25-μl-containing primers at 0.5 μM and deoxynucleoside triphosphates (dNTPs; 0.2 mM each) in PCR buffer and 0.03 U Taq polymerase (Gibco BRL) along with Sybr-green (Molecular Probes, Eugene, OR, USA) at 1:150,000. Aliquots of cDNA from reverse-transcribed control RNA were diluted 4- to 256-fold to generate relative standard curves to which sample cDNA was compared.(19) For RANKL, forward and reverse primers were 5′-CAC CAT CAG CTG AAG ATA GT and 5′- CCA AGA TCT CTA ACA TGA CG, respectively, creating an amplicon of 150 bp. For 18S, an amplicon of 345 was generated with forward primer 5′-GAA CGT CTG CCC TAT CAA CT and reverse 5′-CCA AGA TCC AAC TAC GAG CT. Standards and samples were run in triplicate. RANKL was normalized for amount of 18S in the RT sample, which was standardized also on a dilution curve from a control RT sample, as performed by Johnson et al.(19) The efficiencies of reactions for both RANKL and 18S were >90% (slope for RANKL = −3.29 and slope for 18S = −3.39, in those shown in Fig. 5).
Results are represented as mean ± SEM. Groups were analyzed for significance using PRISM software to apply ANOVA statistics. Statistical significance was achieved as a difference between groups of p < 0.05 by the Bonferroni method.
Strain stimulates ERK1/2 activity
To explore the possibility that MAPKs are activated in primary bone stromal cells by levels of strain shown to inhibit both RANKL expression and osteoclast formation, ERK1/2, p38MAPK, and BMK-1 activation were studied. As shown in Fig. 1, application of 1.8% strain at 10 cycles/minute resulted in significant increases in ERK1/2 phosphorylation as assessed by Western blot analysis using a phosphospecific ERK1/2 antibody. Total ERK1/2 was unchanged. Phosphorylation was inhibited by 30 minutes of preincubation of stromal cells with the specific ERK1/2 inhibitor PD98059, which prevents the ability of the specific MEKK (MAP-ERK kinase) to phosphorylate ERK1/2. In contrast, strain did not activate p38MAPK as measured on blots probed with an antibody against phospho-p38MAPK (data not shown). We also evaluated whether strain could activate BMK-1 in bone stromal cells, which represents a later response in endothelial cells.(8) BMK-1 was not stimulated by mechanical strain of 1.8% up to 60 minutes in our system, and BMK-1 was not increased with a higher strain magnitude of 5% (data not shown).
Kinetic analysis showed that the activation of the ERK1/2 was rapid, reaching a peak by 5 minutes. As shown in Fig. 2, the activation was not sustained, dropping by 15 minutes and returning toward baseline by 60 minutes.
Strain effects on RANKL expression occur after 6 h and are reversible
We had previously shown that 1.8% strain delivered continuously over 24 h decreased RANKL expression in direct proportion to its inhibitory effect on osteoclastogenesis.(4) As shown in Fig. 3A, strain for as little as 6 h caused reductions in RANKL mRNA expression similar to the 24-h strain application. This inhibition was reversible as shown in Fig. 3B; when cultures were continued for a subsequent 24 h with strain removed, RANKL mRNA levels return to control showing that strain was not toxic to the cells. Applications of strain for <6 h had no significant effect on RANKL levels measured at 24 h.
ERK1/2 inhibitor abrogates the strain effect
Next, we investigated the role of strain-induced ERK1/2 activation in regulating RANKL mRNA expression. PD98059 was applied for 30 minutes before initiating the strain regimen, as previously mentioned, and the cultures then were strained for 6 h. Twenty-four hours after the start of the strain regimen, cultures were examined for RANKL expression. As shown in Fig. 4 (compiled in Fig 4A and representative blot in Fig. 4B), strain caused the expected inhibition of RANKL mRNA expression to 50% of control expression in this series of experiments. PD98059 alone had no reproducible effect on RANKL expression when dosed for 6-10 h, although longer exposure tended to decrease both RANKL and 18S mRNA. When given during the strain application, PD98059 inhibited the strain effect such that strained cells expressed RANKL at levels equivalent or above control.
To quantitate the degree to which PD98059 could inhibit strain effect, we used real-time PCR analysis. Total RNA was reverse-transcribed with random decamers and cDNA dilution curves were made for both RANKL and 18S from a single control sample. All other samples were compared with this dilution curve and normalized to 18S content.(19) As shown in Fig. 5, which was compiled from four separate experiments containing control and strained cells with and without PD98059, strained cells had 57 ± 8% of RANKL mRNA after correcting for 18S. Exposure to PD98059 alone had a somewhat variable effect on the RANKL/18S ratio at 106 ± 26% but was not significantly different from control. When PD98059 was added before the strain regimen, it completely abrogated the strain effect, with the values for RANKL/18S at 133 ± 11%, slightly above control levels.
To assess whether another MAPK inhibitor might have a similar effect to inhibition of ERK1/2, control and strained cells were exposed to a p38MAPK inhibitor (SB202190, 100 μM) during a 6-h strain application. In three consecutive experiments (data not shown) strain caused a reduction in RANKL to 50% that of control cells (as assessed by RT-PCR); the p38MAPK inhibitor did not significantly alter the strain effect.
Activation of ERK1/2 occurs at a low strain magnitude in bone cells
To assess the level of strain necessary to activate ERK1/2 in bone stromal cells, different strain magnitudes were applied for 5 minutes. Again, as shown in Fig. 6, unstrained primary stromal cells have low levels of phosphorylated ERK1/2. This level rose after the application of 0.5% substrate strain. Increasing strain magnitudes to as much as 8.5% did not increase further ERK1/2 activation. Because ERK1/2 was stimulated by strain magnitudes lower than 1.8%, we tested the effect of 0.5% and 1% strain on RANKL mRNA. Although we were not able to show significant effects of 0.5% strain on RANKL mRNA, application of 1% strain for 24 h was effective, decreasing RANKL mRNA to 70% of control (Fig. 7).
Strain stimulates JNK2 activity
Although results with the ERK1/2 inhibitor indicate that ERK1/2 is necessary for strain to reduce RANKL mRNA, the requirement of at least 6 h of strain suggests that ERK1/2 activation alone is not sufficient. Because JNK is activated by strain and shear in vascular cells but may have different kinetics than the activation of ERK1/2,(20) it was possible that this stress-activated kinase also might contribute to strain effects in bone stromal cells. As shown in Fig. 8, strain indeed activated JNK in these bone stromal cells. This was shown both using a JNK assay, in which total phosphorylation of c-jun substrate was examined (as shown in Fig. 8A, top panel), as well as a phosphorylated Western blot. Increases in JNK2 paralleled the increased activity seen in the kinase assay (as seen in Fig. 8A, middle panel). JNK1 blots were examined also after shorter chemiluminescent exposures and showed no increased phosphorylation. In contrast to the transience of strain's activation of ERK1/2, continued application of strain had a persistent effect on the activation of JNK. Increases in both JNK activity and phosphorylated JNK2 were maintained at 24 h, although in a slightly reduced fashion compared with the maximal effect that was seen between 15 and 45 minutes. In Fig. 8B, the activation of JNK, as shown by the kinase assay, was similar to ERK1/2 activation in terms of a maximal response to the 0.5% strain magnitude without further increase as the strain was raised to 5%.
Physical input is known to modulate the function of cells involved in skeletal remodeling. The mechanisms by which bone cells respond to signals arising during mechanical events are only just beginning to be understood. Previously, we have shown that bone stromal cells subjected to substrate strain, which models events occurring during loading, will depress the level of RANKL displayed on the cell surface.(4) Reduced RANKL expression will result in decreased osteoclast recruitment and bone resorption.(21) In this work, we show that strain activates both ERK1/2 and JNK in bone stromal cells.
MAPKs have been described in bone cells. The effect of the skeletal growth factor insulin-like growth factor (IGF) 1 to inhibit expression of parathyroid hormone (PTH) receptors by UMR106 osteoblast-like cells is prevented by the ERK1/2 inhibitor PD98059.(22) MAPKs have been implicated in the action of other skeletally active hormones in bone cell cultures, including estrogen(23) and dexamethasone.(24) As well, hypoxia and mechanical stress have been shown to activate differentially MAPKs in human osteoblasts taken from periodontal ligaments(25); cyclic stretch at 90,000 microstrains stimulated JNK but not ERK1/2. The slowly growing primary stromal cells studied here had very low levels of activated ERK1/2, which were reliably activated on strain delivery. The low levels expressed at baseline likely contributed to our ability to measure strain induction of ERK1/2 in the absence of serum starvation, a common manipulation in studies of MAPK activation.(26–28)
The ability of substrate strain to activate ERK1/2 is consistent with studies of strain in other mechanically sensitive cells.(26,28,29) One difference between the widely studied cells of the vasculature and the bone cells studied here is the magnitude of strain to which the cells responded. As might be expected in vascular endothelial cells or contractile muscle cells of the heart, elongations of >5% are required to stimulate MAPKs and show dose dependence as the strain magnitude is raised to >20%.(9,30) Although the actual strains to which cells present on the cortical and trabecular surfaces of bone are not known,(3) very small perturbations of the hard tissue result in significant changes in adaptive skeletal remodeling.(14,31) Therefore, it is expected that bone cells might respond to lower levels of strain. We measured maximal ERK1/2 activation at the lowest level of strain applied, 0.5%. These data suggest that bone cells sense and respond to strain at levels an order of magnitude less than muscle or endothelial cells.
A connection between mechanically induced activation of ERK1/2 and strain inhibition of RANKL expression was explored with the use of the ERK1/2 inhibitor PD98059. Inhibition of ERK1/2 during the strain period prevented the reduction in RANKL as assessed by both standard and real-time RT-PCR as well as by Northern analysis. Inhibiting activation of p38MAPK did not prevent the strain effect. These results suggest that mechanical inhibition of RANKL expression requires activation of ERK1/2. However, 6 h of strain were required to reduce RANKL mRNA, suggesting either that a short burst of strain may not have a lasting effect on RANKL or that ERK1/2 activation alone might not be sufficient to generate the full effect.
In accordance with the idea that ERK1/2 may not be sufficient to convey the full inhibitory effect of strain on RANKL expression are our results showing 0.5% strain, which caused maximal ERK1/2 activation, did not significantly inhibit RANKL, and, indeed, that strains of 1% had to be continued for 24 h to down-regulate RANKL. This might imply that ERK1/2 activation, although necessary to the strain effect, may not be sufficient to transmit the full inhibitory effects of strain on RANKL mRNA. Alternatively, the cell population may have been able to recover from the effects of a lower magnitude strain during the subsequent time before RNA collection. To explore other kinases that might have a different time or magnitude response to strain, we measured BMK-1, p38MAPK, and JNK. BMK-1 is stimulated by shear in endothelial cells but has a later peak effect after shear is begun.(32) Strains of even 5% in stromal cells did not activate either p38MAPK or BMK-1 at 5 minutes or 60 minutes.
We did find an effect of low strain magnitude to activate JNK2. JNK2 activation rose slightly later than that of ERK1/2 and continued above the levels seen in unstimulated cells during 24 h of continuous strain. The persistence of JNK activation, which is similar to that induced by shear in vascular cells,(20) suggests that JNK might have a role in conveying the full effects of strain. The complex interrelationships between the MAPKs should be of interest in understanding strain-induced effects in bone cells in the future.
MAPKs are known to affect gene transcription, for instance, increasing expression of immediate early genes such as c-jun and NF-κB.(33) Are the MAPKs involved in RANKL gene expression? Regulation of RANKL expression is not, as yet, well understood. Nearly 1000 bases of the murine promoter have been studied but do not appear to be enough to confer either tissue specificity, or a significant 1,25-dihydroxyvitamin D effect through a putative vitamin D response element.(34) There are at least three AP-1 consensus sites between 1000 and 2000 nucleotides from the transcription start site in the Celera murine database; AP-1 sequences have been related previously to transcriptional control through biomechanical input.(35)
Studies suggest that OPG is regulated reciprocally along with RANKL by many factors such as PTH(36) and glucocorticoid,(37) in sum suggesting that the RANKL/OPG ratio determines local osteoclast recruitment. We have not examined OPG, the soluble inhibitory receptor for RANKL, in these studies. Our initial findings suggested that the ability of strain to inhibit osteoclastogenesis was not caused by a secreted factor such as OPG.(3) In data not shown here, we have not been able to show an effect of strain to increase OPG expression in primary mouse stromal cells. Thus, although it is possible that OPG also may be regulated by mechanical factors, at this point we only have data showing that mechanical strain down-regulates RANKL in proportion to its ability to decrease osteoclast formation.(4)
In the past several years, investigations have not only pointed to the importance of the mechanical environment for cell survival(38,39) but also have provided information as to how the cytoskeleton might transmit physical signals from the outside in. During cell attachment, integrin-dependent pathways are activated by deformation of the cell membrane.(40) One transduction mechanism stimulated through mechanical factors is the focal adhesion kinase (FAK) activation and subsequent association with the son-of-Sam (SOS) guanine nucleotide exchange factor along with Grb-2, leading to activation of Raf and ensuing MAPK stimulation.(41) This suggests that the matrix on which the cells exist may have significant contributions to how the mechanically sensitive cell perceives its environment. Because integrin attachment to the substrate would be altered during cell deformation, an integrin association with MAPKs has been posited.(6) Indeed, in cells involved in vascular reactivity such as smooth muscle or endothelial cells, the activation of MAPK by shear or strain parallels that achieved by integrin activation.(6) In agreement with the idea that substrate strain might activate MAPK through integrin attachment sites is work showing that straining smooth muscle cells differentially affect ERK1/2 and JNK activation depending on whether the cells have been plated on laminin or pronectin.(28) With regard to cells of the skeleton, FAKs associated with integrin-dependent pathways can be stimulated by strain within physiological parameters in bone osteoblasts.(42) The matrix secreted by these cells, critical to their functional ability, will likely turn out to have important regulatory constraints via mechanical input.
In conclusion, we have shown that the bone stromal cell can respond to its mechanical environment by regulating expression of a key molecule RANKL. Mechanical strain stimulates both ERK1/2 and JNK in these cells. Furthermore, inhibition of ERK1/2 activation inhibits the effect of strain to regulate RANKL expression. Therefore, MAPKs appear be important signal pathways by which mechanical input can affect cellular skeletal remodeling.
This work was funded by NIAMS AR42360 (to J.R.), NHLBI HL58000 (to W.R.T.), and the Veterans Administration (to J.R.).