Presented at the 23rd annual meeting of the American Society for Bone and Mineral Research, Phoenix, Arizona, 2001, for which P.J.E. was a 2001 ASBMR Young Investigator Award recipient.
The authors have no conflict of interest.
The presence of estrogen receptor α (ERα) in osteocytes was identified immunocytochemically in transverse sections from 560 to 860 μm distal to the midshaft of normal neonatal and adult male and female rat ulnas (n = 3 of each) and from adult male rat ulnas that had been exposed to 10 days of in vivo daily 10-minute periods of cyclic loading producing peak strains of either −3000 (n = 3) or −4000 microstrain (n = 5). Each animal ambulated normally between loading periods, and its contralateral ulna was used as a control. In animals in which limbs were subject to normal locomotor loading alone, 14 ±1.2% SEM of all osteocytes in each bone section were ERα positive. There was no influence of either gender (p = 0.725) or age (p = 0.577) and no interaction between them (p = 0.658). In bones in which normal locomotion was supplemented by short periods of artificial loading, fewer osteocytes expressed ERα (7.5 ± 0.91% SEM) than in contralateral control limbs, which received locomotor loading alone (14 ± 1.68% SEM; p = 0.01; median difference, 6.43; 95% CI, 2.60, 10.25). The distribution of osteocytes expressing ERα was uniform across all sections and thus did not reflect local peak strain magnitude. This suggests that osteocytes respond to strain as a population, rather than as individual strain-responsive cells. These data are consistent with the hypothesis that ERα is involved in bone cells' responses to mechanical strain. High strains appear to decrease ERα expression. In osteoporotic bone, the high strains assumed to accompany postmenopausal bone loss may reduce ERα levels and therefore impair the capacity for appropriate adaptive remodeling.
Although physiological strains act effectively as stimuli to maintain the structural competence of bone before menopause, this cellular response appears to become less effective with age(1–3) and estrogen loss.(4,5) The involvement of estrogen in the prevention of osteoporosis and the maintenance of appropriate bone turnover in both men and women has been well established.(6–10) In postmenopausal women with low circulating serum estradiol there is an increased risk of osteoporotic fracture.(11–13) In early postmenopausal women and ovariectomized rats, estrogen withdrawal is associated with increased bone turnover, in which resorption predominates over bone formation.(14–16) When estrogen is replaced, bone loss and fracture risk are reduced, but unless estrogen loss is transient, original bone mass usually is not reestablished.(7,17–20)
Estrogen has many effects on bone cells including modulation of cytokine and growth factor production,(6,21–23) prevention of osteoclastogenesis and osteoclast activation,(24) promotion of osteoclast apoptosis,(25,26) and maintenance of the viability of osteocyte populations.(27,28) Estrogen influences both bone resorption and bone formation in women and bone resorption in men.(8) How (or whether) these various effects are coordinated and how a circulating hormone such as estrogen could target remodeling activity to maintain bone architecture appropriate for different local loading conditions has never been established. Most studies on the effect of estrogen have not reported the prevalence of the estrogen receptor (ER).
The adaptive modeling and remodeling response that establishes and maintains appropriate bone architecture is thought to be controlled at each location by the environment of mechanical strain engendered by functional loading.(29–33) Therefore, continual exposure to mechanical loading is vital to the maintenance of appropriate bone strength as evidenced by the rapid bone loss that accompanies disuse.(34–36) Appropriate loading exercise both inhibits bone resorption and enhances bone formation.(37–39) Such exercise is most effective in increasing bone mass in growing animals and adolescent boys and girls.(40–43) In skeletally mature rats, exercise increases periosteal bone formation.(44–46) Treadmill exercise(47) and feeding from an erect bipedal stance(48) averts orchidectomy-induced bone loss in the rat femur and lumbar vertebrae, respectively. This suggests that appropriately osteogenic loading can counter the effects of sex hormone deprivation.
It has been hypothesized that strain within the bone tissue influences modeling and remodeling through a feedback mechanism by which bone cells relate the current bone architecture to its loading requirements.(29,49,50) Frost has termed this relationship the “mechanostat.”(51,52) The bone loss that accompanies estrogen withdrawal could be caused by a reduction in the effectiveness of the loading-related stimulus after estrogen withdrawal.(53) This type of bone loss may occur preferentially in anatomical sites of low mechanical strain.(54,55) The apparent interaction between estrogen, mechanical strain, and an effective loading response has led several authors to propose that strain and estrogen share a common regulatory pathway.(4,54,56,57)
Both mechanical strain and estrogen elicit a proliferative response in osteoblast and ROS 17/2.8 cells. In both instances, this response can be inhibited by selective ER modulators (SERMs), suggesting that ERα may be a common component of bone's adaptive response to both estrogen and strain.(23,58,59) In cells of the ROS 17/2.8 cell line stably transfected with additional numbers of ERα (ROS.SMER 14 cells) the cells' proliferative response to both strain and estrogen is enhanced.(60) Both estrogen and strain in these cells also increase the activity of estrogen response elements (EREs).(60) ROS 17/2.8 cells' early responses to strain and estrogen also stimulate extracellular signal-regulated kinase (ERK)-mediated phosphorylation of ser122 within the amino terminus of ERα.(61) These data are consistent with ERα providing a common pathway through which both estrogen and strain influence bone-remodeling activity.(62) If this is the case, a reduction in the number of ERα positive osteocytes could explain bones' limited ability for adequate mechanically adaptive remodeling after menopause.
The purpose of this study was to determine whether osteocytes' strain environment influences the level of their ERα expression. To do this, we used fluorescently labeled antibody to ERα to establish the percentage of osteocytes expressing ERα in normal neonatal and adult male and female rat ulna cortical bone. Using this technique, we first established the relationship between the location of ERα positive osteocytes and in vivo ulnar strain distributions during normal locomotion.(63) We then documented the effects of short daily periods of artificial loading on ERα expression in osteocytes.
MATERIALS AND METHODS
Immunofluorescent expression of ERα in normal rat ulna
Ulnas from three male (8 g ± 1 g, 6 days old) and three female (8 ± 1 g, 6 days old) neonatal rats and three male (330 ± 9 g, 67 days old) and three female (283 ± 9 g, 54 days old) adult rats were freshly dissected, coated in 5% polyvinyl alcohol (PVA), and snap-frozen in hexane at −70°C. For consistency with samples analyzed after loading, comparable areas of bone between the middle and distal third of the bone were sampled. Samples were cut transversely to produce 8-μm-thick cryostat sections and transferred to glass slides (Menzel Superfrost Plus Gold; Merck & Co., Poole, UK). Cryostat sections were fixed in 4% (wt/vol) paraformaldehyde (Sigma, Poole, UK) in PBS for 30 minutes at room temperature, washed in 10 mM of glycine (Sigma) in PBS (Sigma), washed in PBS, and stored at 4°C.
Two cryostat sections from each animal were examined. Sections were decalcified (0.25 M of EDTA in 50 mM of Tris-HCl, pH 7.4) for 10 minutes followed by permeabilization with 0.5% (vol/vol) NP-40 (Roche Diagnostics, Ltd., Lewes, UK) in PBS for 10 minutes. Sections were incubated with 4% (wt/vol) bovine serum albumin (BSA) powder (Sigma) in PBS for 1 h at room temperature. Sections then were incubated for 18 h at 4°C with an antibody raised against the ligand binding domain of ERα (MC-20, rabbit polyclonal; Santa Cruz Biotechnology, Inc. Santa Cruz, CA, USA) diluted 1:500 in 4% (wt/vol) BSA in PBS solution. All steps of the procedure were preceded and followed by thorough washing with PBS.
Subsequently, the bone sections were incubated in a 1:200 dilution of secondary biotinylated anti-rabbit antibody in normal 1.5% (vol/vol) goat serum in PBS. (Vectastain Elite ABC Kit; Vector Laboratories, Burlingame, CA, USA) for 30 minutes at room temperature. After a thorough washing with PBS, sections were incubated in Fluorescein Avidin D (10 μg/ml; Vector Laboratories) for 45 minutes at room temperature. Next, sections were incubated in propidium iodide (10 ng/ml of H2O; Sigma) for 4 minutes at room temperature, washed in PBS, dried, and mounted using Dako fluorescent mounting medium (Dako Corp., Ely, UK). For gender and age comparisons, eight microscopic fields (∼450 μm × 450 μm) from each section were visualized using a confocal microscope (Zeiss Corp., Welwyn Garden City, UK). The number of osteocytes and their location was ascertained by means of the propidium iodide fluorescent staining of each osteocyte nucleus. The use of a secondary biotinylated antibody and Fluorescein Avidin D exaggerated the volume occupied by the antigen and thus did not allow definitive determination of whether ERα immunofluorescence was nuclear or cytoplasmic. The presence or absence of ERα immunofluorescence was determined for each osteocyte and used to calculate the proportion of ERα positive osteocytes both manually and with the assistance of Scion Image Analysis software (Scion Image Beta 4.02; Scion Corp., Frederick, MD, USA). One-way ANOVA statistical analysis was used to examine whether ERα expression changed with age or gender (SPSS for Windows, Release 10.0.0; SPSS, Inc., Chicago, IL, USA).
Distribution of ERα in bones experiencing only normal locomotor loading
Two transverse bone sections from the right ulna of eight male adult Sprague-Dawley rats (330 ± 5 g, 67 days old) were examined to determine the strain-related distribution of ERα in cortical bone osteocytes during normal locomotion. A previous study of the osteogenic response to axial loading along the length of the diaphysis had found that the area between the middle and distal third of the bone has a larger osteogenic response to strain than bone located more proximally or distally.(63–66) Therefore, we chose this region as our “area of interest.” Bone sections were taken from an area 560–860 μm distal to the midshaft and examined for immunofluorescent expression of ERα as discussed previously. The location of each osteocyte was mapped across 26–30 confocal microscopic fields (each field ∼450 μm × 450 μm; Fig. 1). To establish the presence and distribution of ERα immunofluorescence, analysis was performed on a Macintosh computer using the public domain National Institutes of Health (NIH) Image program (developed at the U.S. NIH and available on the Internet at http://rsb.info.nih.gov/nih-image/). The distance of each ERα positive osteocyte from the neutral strain axis was used to calculate the peak strain magnitude during locomotion at the location of each ERα positive osteocyte (Fig. 1). The position of the neutral plane had been calculated previously from strain gauge data recorded from the caudal, medial, and lateral surface of the ulna during artificial axial loading.(63) Each ERα positive osteocyte was classified according to this strain magnitude and divided into groups of 500 microstrain each for ease of presentation and interpretation.
Comparison of ERα distribution in mechanically loaded and control rat ulna
In the same rats used to determine the relationship between ERα and peak strain magnitude during normal locomotion, the left ulna had been subjected to short (10 minutes) daily periods of dynamic loading. The loading was performed over two periods of 5 consecutive days with an intervening 2-day rest period. Calcein (10 mg/kg; Sigma) was administered by intraperitoneal (ip) injection 3 days before commencement of loading and again on the final day of loading. The calcein fluorescence in the bone sections allowed us to document new bone formation over the 2-week course of the experiment. Axial loading was achieved noninvasively, in vivo, via the flexed carpus and olecranon as described previously.(63,65) Animals were anesthetized (11 mg/kg of xylazine ip [Rompun; Bayer, Bury St. Edmunds, UK] and 55 mg/kg of ketamine ip [Vetalar, Parke, Davis, and Co., Pontypool, UK]), to allow loading without discomfort or movement, and placed in dorsal recumbency on an insulated mat. Corneas were protected by the application of a lubricating ophthalmic ointment (Lacrilube; Allergan, High Wycombe, UK). The flexed carpus and olecranon of the left forearm were held in place with padded cups, and a controlled cyclic load was applied to the left ulna via the flexed carpus using a closed loop servohydraulic materials testing machine (Dartec; Zwick Testing Machines Limited, Herefordshire, UK; Fig. 2). The load was applied cyclically (1200 cycles, 2 Hz) between 1 and 20 N in a trapezoidal pattern to generate compressive peak strains of either −0.0030 (−3000 microstrain, n = 3) or −0.004 (−4000 microstrain, n = 5) at the medial midshaft.(65) The loading/unloading rate was adjusted to engender a uniform maximum strain rate of 0.03 s−1. After loading, the animals were allowed to recover in an insulated box. The animals ambulated normally between loading periods, creating a model in which short periods of novel loading were inserted with longer periods of normal activity.(63)
The left ulnas were harvested, cut, and examined for ERα as discussed previously, and then data were compared with paired sections from the control right ulna to document the relative changes in ERα expression with loading in any given individual. ANOVA was used to determine whether there was any difference in expression of ERα in ulnar specimens experiencing microstrains of 1500 (normal ambulation) versus 3000 or 4000 (artificially loaded). The Wilcoxon signed rank test (release 10.0.0; SPSS, Inc.) was used to determine the statistical significance of variations in ERα expression with strain.
To confirm the specificity of the anti-ERα antibody, protein extracts from both adult and neonatal male rat primary bone were subjected to Western blot analysis using the ERα antibody (MC-20; Santa Cruz Biotechnology, Inc.). Individual bones were ground to a powder under liquid nitrogen and the powder was scraped into 1 ml of lysis buffer. The protein content of lysates, before 10% SDS-PAGE, were determined using BCA protein assay kits (Pierce & Warriner, Chester, Cheshire, UK). The assays were carried out using 40-μl aliquots of lysate following the manufacturer's instructions and read on a Dynex MRX microplate reader (Dynex Technologies, Billingshurst, UK) using Revelation 3.04 software (Microsoft Corp., Redmond, WA, USA).
Immunoblotting for ERα protein was undertaken using a modification of the methodology of Wheeler-Jones.(67) Briefly, aliquots of lysates (100 μg) were resolved by SDS-PAGE (10%) and transferred onto polyvinylidene difluoride (PVDF) membrane (Immobilon-P; Millipore, Bedford, MA, USA). Membranes were blocked with 0.2% (wt/vol) I-block (Applied Biosystems, Foster City, CA, USA) in TBST (50 mM of Tris, 150 mM of NaCl, and 0.02% [vol/vol] Tween 20 [Sigma], pH 7.4) for 3 h at room temperature. Membranes were probed overnight using a 1:5000 dilution of anti-ERα antibody MC-20 (Santa Cruz Biotechnology, Inc.) in 0.2% (wt/vol) I-block at room temperature with agitation. The blots were washed using 0.2% (wt/vol) I-block (6 × 10 minutes) and incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin G (IgG; Pierce & Warriner) diluted 1:5000 in 0.2% (wt/vol) I-block for 45 minutes at room temperature. After further washing with 0.2% (wt/vol) I-block (6 × 10 minutes), immunoreactive bands were visualized using luminol reagent (Santa Cruz Biotechnology, Inc.) and Hyperfilm-ECL film (Amersham International, Amersham, Bucks., UK) according to the manufacturer's instructions. This analysis confirmed that the antibody used identified a 65-kDa band characteristic of ERα.
Number of osteocytes immunopositive for ERα in normal adult and neonatal rats
An average of 14 ± 1.25% SEM of all osteocytes in the cortical cross-section were positive for ERα (Fig. 3). There was no influence of either gender (p = 0.725) or age (p = 0.577) on the proportion of osteocytes displaying ERα immunofluorescence and no interaction between age and gender (p = 0.658; adult female rat, 17 ± 3.5% SEM; adult male rat, 13 ± 2.6% SEM; neonatal female rat, 13.5 ± 1.7% SEM; neonatal male rat, 12.5 ± 1.7% SEM). ERα immunofluorescence was apparent in a large proportion of osteoblastic cells (and/or lining cells) on both the endosteal and the periosteal surfaces. There was some variation in the intensity and distribution of ERα immunostaining around the perimeter of the bone attributed to overlapping of periosteal osteoblasts and surrounding muscle tissue because of the sectioning process. All endothelial cells in cortical blood vessels appeared to be highly positive (Fig. 1).
Distribution of ERα positive osteocytes in ulnas experiencing only normal loading
In transverse sections from the midshaft of control bones, which had experienced only normal loading, an average of 14% of osteocytes were positive for ERα immunofluorescence. The distribution of ERα positive osteocytes was uniform and did not vary with local functional (locomotor) strain magnitude across the bone section (Fig. 4). Osteocytes experiencing low strain levels due to their location near the neutral strain axis had the same incidence of ERα expression as those farther from it, which experienced higher local strains (Fig. 4). All newly formed osteocytes in the bone between calcein labels at the periosteal surface were positive for ERα.
ERα expression in artificially loaded male rat ulnas
There was no significant difference in the percentage of total osteocytes expressing ERα between adult male rat ulnas loaded at −3000 or −4000 microstrain (7.5 ± 69% SEM and 7.8 ± 0.87% SEM, respectively). Likewise, there was no significant difference between the percentage of total osteocytes expressing ERα in the control limbs for rats in which contralateral ulnas were loaded to produce either −3000 or −4000 microstrain (12 ± 1.9% SEM and 16 ± 2.4% SEM, respectively). Consequently, these data were combined to create one group of artificially loaded bones and one group of contralateral controls.
Comparison of the percentage of ERα positive osteocytes in artificially loaded limbs to that in their contralateral control limbs showed that the proportion of ERα positive osteocytes (7.5 ± 0.91% SEM) in artificially loaded bones was significantly lower than in their contralateral controls (14 ± 1.68% SEM; p = 0.01; median difference, 6.4; 95% CI, 2.60, 10.25). This decrease in osteocyte ERα expression was uniform across the section and unrelated to local microstrain magnitude. When compared with the contralateral control limb, the artificially loaded limb had significantly fewer osteocytes expressing ERα immunofluorescence at all comparable strain magnitudes (p = 0.001; Fig. 4).
These data confirm the presence of ERα in osteocytes of the cortical bone of the rat ulna midshaft. The number of osteocytes expressing ERα in this location is very similar to published values for human osteocytes from bone sections of normal men.(68) However, in this rat study, in which the strain distribution throughout the bone section was known, it was possible to show that the proportion of osteocytes expressing ERα was uniform and unrelated to local strain magnitude. Exposure to short daily periods of artificial loading engendering high physiological strain levels (superimposed on normal loading) diminished the number of osteocytes expressing ERα by 46%. This reduction was uniform across the section. Because osteocytes appeared to respond in concert to the strain-related stimulus, this suggests that they communicate and act together as a population.
ERα mRNA and protein were first documented in skeletal tissues in 1988.(69,70) ERα expression subsequently has been identified in human osteoblasts and osteocytes in vivo.(71–73) Others have identified ERα in both active osteoblasts and lining cells on trabecular surfaces of human and rabbit specimens but have not detected ERα in osteocytes or osteoclasts of either species.(74) Preosteoclasts have been shown to express ERα, but this expression is lost with osteoclast maturation and bone resorption.(75) The presence of ERα protein has been reported in growth plate sections from male and female rats using both immunohistochemical and immunofluorescent antibody staining.(76,77) This expression of ERα at the growth plate is lost with sexual maturity in rats but not rabbits.(77) ER also has been shown in high concentrations in bone endothelial cells.(78) To our knowledge, this report is the first to document the presence of ERα protein in rat cortical bone osteocytes and to show in vivo responses of ERα to mechanical stimulation.
The data presented here show that the mean percentage of osteocytes expressing ERα immunofluorescence in normal rat cortical bone (14% [±1.2% SEM]) is remarkably similar to that reported in male and female human osteocytes.(68) However, in men with pathophysiological conditions such as idiopathic osteoporosis, ERα is down-regulated from 14% (±2%) to 1% (±0.5%).(73,79,80) Likewise, in women with ovarian steroid deficiency, the proportion of osteocytes with immunodetectable ERα also decreases from 25% (±3%) to 12% (±4%).(73) Consequently, it has been suggested that estrogen responses in bone may become defective when associated with impaired ERα expression.(81) Experiments with ERα and ERβ double knockout rats have indicated that the presence of functional ER is required for estrogen's ability to block ovariectomy-induced bone loss in cortical as well as trabecular bone.(82) Genetic polymorphism of the ER may further direct and refine estrogen's effect on bone mass.(83–89) In young men, for example, ERα polymorphism is related to bone density, but estradiol levels are not.(90) These studies suggest that both the genotype and level of ERα are important to the safeguarding of appropriate bone density.
The data presented here suggest that loading down-regulates the number of osteocytes that express ERα. It is possible, but we consider, unlikely, that this effect is caused by a strain-induced conformational change in ERα, which has affected the detectable expression of immunofluorescence rather than indicating a reduction in the presence of ERα protein. Any such conformational changes would have to be long lasting because the exposure to loading is transient. SERMs have been shown to generate unique conformational changes in ERα that affect in vitro receptor stability(91) and cause a dose-dependent reduction in ERα protein level, but we are not aware that this outlasts the presence of the SERM.(92) Estrogen itself also plays an important feedback role in the control of ERα expression and is a source of tissue-specific ERα down-regulation.(93–96) However, the examination of male rats in this study ensured little cyclical fluctuation of estrogen levels and, because each animal served as its own control, loaded and control limbs had equivalent exposure to the influence of estrogen.
Osteocyte ERα expression was uniform across sections of bone subject to both locomotor loading and artificial loading (Fig. 4). Hence, the same percentage of osteocytes expressed ERα in the low strain region of the bones' neutral strain axis as osteocytes experiencing higher strain toward the outer cortex. ERα mRNA shows increased expression in conjunction with osteoblast differentiation,(97,98) and newly formed osteocytes had a higher expression of ERα. However, the expression of ERα in the mature osteocyte population was stable and only changed (decreased) after mechanical loading. The finding that a short period of loading in addition to that experienced during locomotion could depress ERα expression by 46% indicates that some consequence of loading has a direct causal effect on ERα expression in osteocytes.
Previous examination of the effects of estrogen and exercise on the maintenance of bone mass have led some researchers to conclude that strain-related information directs appropriate bone remodeling that is maintained under the influence of estrogen.(8) Others have suggested that estrogen may reduce the mechanosensory (mechanostat) set point or the level of strain-related stimulation at which osteogenic adaptation occurs.(4) Our results suggest that prevailing strain levels within an osteocyte population may influence the amount of ERα available to respond to loading as well as estrogen.
Both estrogen replacement and exercise have been shown to decrease bone turnover and bone resorption in postmenopausal women and ovariectomized rats.(54,99–103) It has been hypothesized that the activation frequency of bone remodeling is kept at a relatively low level because of an inhibitory signal produced as a response to physiological strain by osteocytes.(104–108) Accumulating evidence supports the role of ERα as a common component of the pathway by which estrogen and mechanical strain influence modeling and remodeling activity. In this situation, a decrease in the number of osteocytes expressing ERα could restrict the osteocytes' potential response to circulating estrogen and local strain and thus modulate the potential effect of both influences on inhibition of bone resorption.
Because of its inadequate mass and inappropriate architecture, it is likely that osteoporotic bone experiences higher strains than normal during customary daily activity. Our current findings suggest that such higher strains would down-regulate osteocyte ERα levels, compounding any reduction of ERα expression associated with estrogen deficiency.(73) This would further impair the capacity for appropriate adaptive (re)modeling in postmenopausal women.
In conclusion, the data presented here show that local mechanical loading influences osteocyte ERα expression with exposure to short periods of high physiological strain levels depressing the number of osteocytes expressing ERα. ERα regulation by mechanical strain may provide a potential means for localized direction and refinement of the effect of both strain and estrogen on bone remodeling. A strain-related reduction in ERα expression could contribute to the reduced ability of bone to remodel adaptively after menopause.
The authors acknowledge support from The Wellcome Trust.