Mechanical strain, testosterone, and estrogen all stimulate proliferation of primary cultures of male rat long bone (LOB)-derived osteoblast-like cells as determined by [3H]thymidine incorporation. The maximum proliferative effect of a single period of mechanical strain (3400 με, 1 Hz, and 600 cycles) is additional to that of testosterone (10−8 M) or estrogen (10−8 M). The cells' proliferative response to strain is abolished both by concentrations of tamoxifen that cause proliferation (10−8 M) and by those that have no effect (10−6 M). Strain-related proliferation also is reduced by the estrogen antagonist ICI 182,780 (10−8 M) but is unaffected by the androgen receptor antagonist hydroxyflutamide (10−7 M). Tamoxifen, ICI 182,780, and the aromatase inhibitor 4-dihydroandrostenedione, at concentrations that have no effect on basal proliferation, significantly reduce the proliferative effect of the aromatizable androgen testosterone but not that of the nonaromatizable androgen 5α-dihydrotestosterone. Hydroxyflutamide, at a concentration that has no effect on basal proliferation (10−7 M), eliminates the proliferative effect of 5α-dihydro-testosterone but had no significant effect on that caused by testosterone. Proliferation associated with strain is blocked by neutralizing antibody to insulin-like growth factor II (IGF-II) but not by antibody to IGF-I. Proliferation associated with testosterone is blocked by neutralizing antibody to IGF-I but is unaffected by antibody to IGF-II. These data suggest that in rat osteoblast-like cells from males, as from females, strain-related proliferation is mediated through the estrogen receptor (ER) in a manner that does not compete with estrogen but that can be blocked by ER modulators. Proliferation associated with testosterone appears to follow its aromatization to estrogen and is mediated through the ER, whereas proliferation associated with 5α-dihydrotestosterone is mediated by the androgen receptor. Strain-related proliferation in males, as in females, is mediated by IGF-II, whereas proliferation associated with estrogen and testosterone is mediated by IGF-I.
THE FINDING that the estrogen receptor (ER) modulator ICI 182,780 reduces, and tamoxifen blocks, the proliferative response to mechanical strain of primary osteoblast-like cells derived from the long bones (LOBS) of female rats(1) suggests that female bone cells' early adaptive response to mechanical strain involves the ER. This interpretation is supported by evidence that strain as well as estrogen up-regulates the activity of estrogen response elements (EREs) in ROS.SMER 14 cells. These are ROS 17/2.8 cells stably transfected with the α-form of the ER and then transiently transfected with EREs.(2)
In ROS 17/2.8 cells it has been shown that proliferation stimulated by mechanical strain is blocked by a neutralizing antibody against insulin-like growth factor II (IGF-II) but not that against IGF-I.(3) In contrast, proliferation stimulated by estrogen is not blocked by neutralizing antibody to IGF-II but is blocked both by neutralizing antibody to IGF-I and by blocking antibody to the IGF-I receptor. This suggests that although estrogen and strain may both use the ER and both stimulate transcription of EREs, their pathways subsequently diverge stimulating transcription and production of separate IGFs. Neither IGF-I nor IGF-II has an ERE in its promoter region so there must be a number of intervening steps at which this divergence can occur.
These findings pose the question “does the response to strain of osteoblast-like cells in males similarly involve the ER and lead to IGF-II-mediated proliferation, or is the androgen receptor and some other growth factor involved?”
In the experiments reported here we addressed this question by investigating the mechanisms involved in the proliferative responses of primary cultures of male rat LOB-derived osteoblast-like cells to mechanical strain, estrogen, the aromatizable androgen testosterone, and the nonaromatizable androgen 5α-dihydrotestosterone. We investigated the extent to which these proliferative responses were modified by the selective ER modulators (SERMs) ICI 182,780 and tamoxifen, the androgen receptor antagonist hydroxyflutamide, and the aromatase inhibitor 4-hydroxyandrostenedione. We also used neutralizing antibodies to investigate the role of IGF-I and IGF-II in these strain-related and hormone-related responses.
MATERIALS AND METHODS
Primary cultures of osteoblast-like cells derived from LOBS of 4- to 5-week-old male Sprague-Dawley rats (100 ± 5 g; Charles River, Margate, Kent, U.K.) were cultured and characterized as described previously.(4) The phenotype of these cells was confirmed to be osteoblastic from their high levels of alkaline phosphatase activity and their ability to form mineralized nodules in long-term culture in the presence of 50 μg/ml ascorbic acid and 10 mM β-glycerophosphate.
The cells were grown in phenol red-free Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine, 100 U/ml penicillin, and 50 μg/ml streptomycin at 37°C in a humidified atmosphere of 95% air and 5% CO2. Subconfluent cells from a single extraction were used at their first passage.
Tissue culture solutions were from Life Technologies (Paisley, Scotland, U.K.). Hormones and all other biochemicals were from Sigma (Poole, Dorset, U.K.) unless otherwise stated.
Proliferative responses to testosterone, dihydrotestosterone, antiestrogens, antiandrogen, and aromatase inhibitor
Male-derived LOBS were passaged at subconfluency by treatment with trypsin-EDTA and seeded onto 24-well dishes at a density of 2.5 × 104 cells per well. They were then maintained in culture for 48 h in 10% FBS medium before being grown in reduced concentrations of serum for 24 h (2% charcoal-dextran-treated FBS [FBS-CD]). For the final 24 h of the experiment, cells were incubated in medium supplemented with 2% FBS-CD containing 1.0 μCi/ml [3H]thymidine and the respective test factors. These included testosterone, 5α-dihydrotestosterone, the pure ER antagonist ICI 182,780 (ICI; gift from Dr. Alan Wakeling, Zeneca Ltd., Macclesfield, U.K.), the ER modulator tamoxifen, the antiandrogen hydroxyflutamide (a gift from Dr. R.O. Neri, Shering-plough, NJ, U.S.A.), and the aromatase inhibitor 4-hydroxyandrostenedione. These were dissolved in absolute ethanol and stored at −20°C. The final concentration of ethanol in test wells never exceeded 0.01% and control wells contained equivalent amounts of ethanol. Dose-responses were established for each factor (10−6−10−10 M).
In experiments designed to establish the combined effect of the receptor blockers or aromatase inhibitor on the cells' response to androgens, cultures were pretreated with the respective inhibitors for 6 h before incubation with testosterone (10−8 M) or 5α-dihydrotestosterone (10−8 M) for 24 h. Concentrations of receptor blockers that had no effect on basal proliferation were used. These were hydroxyflutamide (10−7 M), 4-hydroxyandrostenedione (10−8 M), tamoxifen (10−6 M), and ICI 182,780 (10−8 M). Tamoxifen also was used at a concentration that itself caused proliferation (10−8 M).
Effect of testosterone, ICI, tamoxifen, 4-hydroxyandrostenedione or hydroxyflutamide, and mechanical strain
The monolayer cultures of LOBS were subjected to mechanical strain by loading the strips onto which they were seeded in four-point bending as described previously.(1) The flexible plastic strips (66 mm × 22 mm) were custom cut from 4-well tissue culture-treated dishes, washed and sterilized and had the cells seeded onto them at a density of 1.5 × 105 cells per strip. The strain field across the strips was verified to be uniform and linear from strain gauge recordings on identical sample strips. For each treatment group five strips were incubated together in 150 cm2 plastic sterile nontissue culture-coated dishes for 24 h in medium supplemented with 10% FBS. They were then cultured for 24 h in medium containing 2% FBS-CD, before mechanical stimulation. Where appropriate, cells were pretreated for 6 h with ICI 182,780 (10−8 M), tamoxifen (10−6 M and 10−8 M), hydroxyflutamide (10−7 M), or 4-hydroxyandrostenedione (10−8 M).
Five strips placed in the lower wells of the loading jig with 8 ml of 2% FBS-CD/well containing the test factors were exposed to a single period of dynamic peak strain of 3400 με for 10 minutes at 1 Hz, 600 cycles (maximum strain rate, 23,000 με/s), and at 37°C in a humidified atmosphere of 95% air and 5% CO2. Five identical plastic strips were placed in the top wells of the loading apparatus with 8 ml of 2% FBS-CD medium/well containing the test factors. Here, they were displaced through the medium and experienced similar disturbances of fluid caused by rocking the same amount of fluid at the same frequency and for the same duration as the strained strips subjected to four-point bending. These cells served as the “rocking controls” for those subjected to strain.
Immediately after mechanical stimulation, the strips were transferred to the original 150-cm2 dishes and culture continued for a further 24 h in 2% FBS-CD medium containing 1 μCi/ml [3H]thymidine, the incorporation of which was determined as described in the following. The respective test factors were present during the entire straining and labeling periods.
Effects of IGF-I, IGF-II, and their neutralizing antibodies on basal proliferation and the effect of these antibodies on proliferation associated with testosterone or mechanical strain
For dose-response studies, LOBS were cultured onto 24-well plates as described previously. The truncated forms of IGF-I (tIGF-I; des 1–3) and IGF-II (tIGF-II; des 1–6) (GroPep, Adelaide, Australia) were used to avoid the complicating effects of IGF binding proteins. For each tIGF, [3H]thymidine incorporation was measured over the range of 10−6−10−10 M.
Dose-response studies were then undertaken for neutralizing monoclonal antibodies against IGF-I and -II(5,6) (Upstate Biotechnology, Lake Placid, NY, U.S.A.) over the concentration range of 1–5 μg/ml. In experiments designed to study the effects of the respective antibodies on proliferative responses to strain and testosterone, cells were seeded onto strips and then cultured as described previously. Three micrograms per milliliter of the antibody were added to the culture 1 h before loading or before the addition of testosterone (10−8 M).
Cell proliferation assay
The cell proliferation assay was performed as previously described.(1) In brief, the level of DNA synthesis was quantified by determining the amount of radiolabeled thymidine ([methyl-3H]thymidine, specific activity 16.4 Ci/mmol; Amersham, Bucks, U.K.) incorporation. The acid insoluble radioactivity in duplicate aliquots was determined by scintillation counting and was verified by cell counting using a hemocytometer. To normalize data; results from all experiments are expressed as percentage vehicle-treated control of [3H]thymidine incorporation. Experiments were repeated at least three times to confirm the findings.
Although the data are expressed and illustrated as percentage mean counts per minute ± SEM, all statistical differences between the test conditions and respective controls were analyzed by analysis of variance (ANOVA) using the original numerical data. Significance was assessed for multiple comparisons of means by a post hoc examination with the Scheffe F test using a Statview 512 software package (Abascus Concepts, Inc., Berkely, CA, U.S.A.) on a Macintosh computer. Statistical differences of p < 0.05 (1 factor ANOVA, significance at 95%) were considered significantly different.
The effects on [3H]thymidine incorporation of estrogen, testosterone, 5α-dihydrotestosterone, and mechanical strain independently and in combination
17β-Estradiol, testosterone, and 5α-dihydrotestosterone all stimulated increased incorporation of [3H]thymidine into the primary cultures of LOBS cultured in 24-multiwell plates. Across the range of 10−6−10−10 M, the peak response occurred for 17β-estradiol at 10−8 M (348 ± 28%; p < 0.001), for testosterone at 10−8 M (200 ± 5%; p < 0.001), and for dihydrotestosterone at 10−9 M (283 ± 17%; p < 0.001; Table 1).
Table Table 1.. Dose-Responses of 17β-Estradiol, Testosterone, Dihydrotestosterone, 4-Hydroxyandrosteindione, Hydroxyflutamide, ICI 182,780, Tamoxifen, tIGF-I, and tIGF-II
A single period of dynamic mechanical strain change (600 cycles at 1 Hz) engendering peak strain of 3400 με produced a similar maximum (142 ± 7%) increase in [3H]thymidine incorporation in these male cells as previously reported in females.(1) The level of [3H]thymidine incorporation in cells strained in the presence of 10−8 M testosterone and 10−8 M estrogen was equivalent to the addition of the maximal responses seen with strain or hormone alone (Figs. 1A and 1B).
The effect of aromatase inhibitor on [3H]thymidine incorporation associated with testosterone, dihydrotestosterone, and strain
The aromatase inhibitor 4-hydroxyandrostenedione over the concentration range of 10−8−10−10 M had no significant effect on basal [3H]thymidine incorporation. 4-Hydroxyandrostenedione reduced proliferation slightly but not statistically significantly at 10−7 M and significantly at 10−6 M (Table 1).
Testosterone alone (10−8 M) increased [3H]thymidine incorporation to 200 ± 5% of control values in cells cultured in multiwell plates. This was reduced to 30% (p < 0.01) above controls when the aromatase inhibitor 4-hydroxy androstenedione (10−8 M) was present (Fig. 2A). 4-Hydroxyandrostenedione appeared to reduce the proliferative effect of mechanical strain (Fig. 2B); however, this reduction did not achieve statistical significance either in any single experiment or when data were grouped. 4-Hydroxyandrostenedione had no effect on the level of proliferation produced by 5α-dihydrotestosterone (data not shown).
The effect of SERMs on [3H]thymidine incorporation associated with 5α-dihydrotestosterone, testosterone, and mechanical strain
The SERM ICI 182,780 had no effect on basal [3H]thymidine incorporation between the concentrations of 10−8 and 10−10 M but caused a statistically significant reduction at higher doses (Table 1).
The increase in [3H]thymidine incorporation stimulated by 10−8 M testosterone (200 ± 5%; p < 0.001) was reduced to 157 ± 5% in the presence of 10−8 M ICI 182,780 (Fig. 3A). This concentration of ICI had no effect on the increased [3H]thymidine incorporation stimulated by 5α-dihydrotestosterone (10−8 M; data not shown). The proliferation stimulated by a single period of mechanical strain (peak strain, 3400 με; maximum strain rate, 23,000 με/s, 1 Hz) was inhibited by ICI 182,780 at 10−8 M (p < 0.01; Fig. 3B). ICI, at 10−6 M, eliminated the increases in proliferation associated with testosterone, strain, and 5α-dihydrotestosterone, but at this concentration it also reduced the basal level of [3H]thymidine incorporation (54 ± 3%; p < 0.01).
Tamoxifen, at concentrations of 10−7−10−10 M, stimulated [3H]thymidine incorporation significantly (Table 1). The maximum effect (182 ± 15%; p < 0.01) was in 24-well tissue culture plates at a concentration of 10−9 M. At a concentration of 10−6 M, tamoxifen had no effect either on cells cultured on 24-well plates or on the washed and sterilized strips used in straining experiments. At this concentration, tamoxifen completely blocked the proliferative effect of 10−8 M testosterone (Fig. 4A) but had no significant effect on the proliferation caused by (10−8 M) dihydrotestosterone (data not shown). The increased [3H]thymidine incorporation associated with exposure to mechanical strain was eliminated by tamoxifen at concentrations of 10−8 M, which stimulated proliferation, and 10−6 M, which did not (Fig. 4B).
The effect of the androgen receptor antagonist hydroxyflutamide on [3H]thymidine incorporation associated with 5α-dihydrotestosterone, testosterone, and mechanical strain
The androgen receptor antagonist hydroxyflutamide, at concentrations of 10−9 M and 10−8 M, had a slight stimulatory effect on the basal level of [3H]thymidine incorporation, but this was statistically nonsignificant. Hydroxyflutamide, at 10−7 M, had no effect on the basal cell proliferation, but at 10−6 M, it was associated with a reduction of basal [3H]thymidine incorporation to 72 ± 5% (p < 0.05) of controls (Table 1).
Hydroxyflutamide, at a concentration of 10−7 M, inhibited the proliferative effect of dihydrotestosterone (10−8 M) but not that associated with 10−8 M testosterone or mechanical strain (data not shown).
The effects of IGF-I and -II and their neutralizing antibodies on [3H]thymidine incorporation stimulated by testosterone and mechanical strain
Exposure of LOBS to the truncated forms of IGF-I and IGF-II showed a dose-dependent increase in [3H]thymidine incorporation between 10−7 and 10−10 M. The peak response was at 10−7 M for both IGFs (IGF-I, 173 ± 6% and p < 0.001; IGF-II, 195 ± 12% and p < 0.001 above that of controls; Table 1).
Neutralizing antibody to IGF-II at concentrations from 3 to 5 μg/ml, had no effect on basal proliferation (data not shown) but at a concentration of 3 μg/ml it obliterated the increased [3H]thymidine incorporation associated with strain (Fig. 5). This concentration of IGF-II antibody showed no effect on proliferation associated with 10−8 M testosterone (Fig. 6).
Dose-response studies with the neutralizing antibody to IGF-I showed no effect on basal proliferation between 3 and 5 μg/ml (data not shown). However, at a concentration of 3 μg/ml, this antibody obliterated the increased [3H]thymidine incorporation seen with 10−8 M testosterone (Fig. 6) but had no effect on that associated with mechanical strain (Fig. 5).
The data presented here show that primary cultures of male rat LOB-derived osteoblast-like cells proliferate in response to mechanical strain and estrogen by using the ER in essentially the same way as cells similarly derived from females. In cells from both sexes the proliferative response to estrogen or a single short period of mechanical strain can be blocked by the SERMs ICI 182,780 and tamoxifen.
Blocking of the proliferative effect of strain by an ER modulator can be explained by these compounds altering the ER's conformation so that it no longer facilitates this response to strain.(7–9) Because the maximum proliferative responses to strain and estrogen are additive, processing by the ER is either not the rate-limiting step in these pathways, or these pathways do not compete for the same domain of the receptor. It is noteworthy that tamoxifen, which, from our data and that of others,(10,11) acts as an ER agonist in osteoblast-like cells, was a more effective blocker of the effects of strain than ICI 182,780, which in these cells has no such agonistic effect. This could be explained by the different conformational changes that the two SERMs impose on the ER.
In the primary cultures used in this study, the effects of testosterone appear to be exerted after its aromatization to estrogen. This is consistent with a substantial body of evidence that the skeleton is a site for aromatase activity. Osteoblasts possess aromatase and the other enzymes necessary for the biosynthesis of estrogen.(12–21) Once testosterone is converted to estradiol, it then acts through the cells' ERs.(22–23) In postmenopausal women estrogens are formed almost exclusively by peripheral conversion of the sex steroid precursor androstenedione derived from the adrenals, a reaction mediated by aromatase enzyme complex.(13,24) Dysfunction of the aromatase gene leads to osteopenia and delayed epiphyseal fusion.(25)
Although our present data from cultured rat osteoblasts indicate the presence of an aromatase pathway in bone cells, they provide no evidence of the extent to which testosterone is normally aromatized to estrogen in vivo in rats and humans. Plainly, osteoblasts contain androgen receptors by which androgens can influence their behavior.(26–31) However, it is clear from in vivo data that the effects of androgens on the skeleton are complex and probably site specific. In growing male rats androgen receptor-mediated effects are important because both testosterone and nonaromatizable androgen dihydrotestosterone can independently prevent the osteopenia in cancellous bone, which follows orchidectomy.(32) However, complete compensation of one for the other is not always possible because aromatase inhibition in growing male rats impairs modeling and results in decreased bone density.(15)
It is noteworthy that the osteoblast-like cells used in our study were derived from the periosteal surface of LOBS. The fact that androgens could have their major effect on cells from this bone surface after their aromatization to estrogen is consistent with reports that periosteal cells from rat tibia are unable to convert testosterone to dihydrotestosterone.(33) The observation that aromatase inhibition decreases periosteal expansion in male rats(15) also is consistent with our data that osteoblasts of periosteal origin are responsive to estrogen. It is perhaps significant that in our present study estrogen was more potent at stimulating proliferation of these cells than either testosterone or dihydrotestosterone. Nevertheless, the female type of bone structure seen in androgen-resistant male rats suggests that the picture is not a simple one and that androgens are involved in subperiosteal expansion.(34)
Bone cells' responses to sex steroids are plainly complex. However, our data are consistent with the view expressed by Vanderschueren et al.(35) that “the intrinsic effect of estrogen on bone mass and bone turnover seems stronger than the intrinsic effect of androgen even in male rats; therefore local aromatization of androgens into estrogen even in modest concentrations may be of importance in vivo.”
Regardless of the effect of androgens on growth and remodeling, the data from our present study suggest that the androgen receptor is not involved in male osteoblast-like cells' proliferative response to strain, at least in rats. This inference is based on the lack of any blockade of the proliferative effects of strain by the androgen receptor antagonist hydroxyflutamide, although it abrogates proliferation stimulated by 5α-dihydrotesterone.(29,36–38)
It may be significant that in our present study dihydrotestosterone appears to be more potent than testosterone, showing a maximum stimulatory effect on proliferation at 10−9 M compared with 10−8 M. This is consistent with data from previous studies by others showing stimulation of osteoblast proliferation by 5α-dihydrotestosterone that can be inhibited by androgen receptor blockers.(19,39,40) These data and identification of the enzyme 5α-reductase in osteoblasts(40,41) indicate that these cells can convert testosterone to dihydrotestosterone. However, the 5α-reductase pathway may be of little importance in vivo because the 5α-reductase inhibitor finasteride affects neither bone density nor metabolism in normal male rats(42) or in men with prostatic hyperplasia.(43)
In our present study the aromatase inhibitor 4-hydroxyandrostenedione reduced the proliferative response to strain consistently in each experiment but not sufficiently for this reduction to achieve statistical significance. Although care must be taken not to overinterpret a nonsignificant result, this effect could be explained by low concentrations of testosterone remaining within the medium despite our use of charcoal stripped serum.
We have previously reported that in ROS 17/2.8 cells, which are female, the proliferative response to estrogen is mediated through IGF-I and the proliferative response to strain is mediated through IGF-II.(3) It appears that primary cultures of male rat osteoblast-like cells act similarly to female ROS 17/2.8 cells and mediate the proliferative effects of strain through IGF-II, and those of estrogen and testosterone after aromatization mediate effects of strain through IGF-I.
The principal relevance of our current data is the understanding of the mechanisms by which bone cells achieve and maintain the structural suitability of the skeleton. We assume that bone architecture is adjusted to be able to withstand functional loading because resident bone cells respond to loading-induced strain in their surrounding matrix and appropriately influence and coordinate bone formation and resorption. Our data relate only to the early proliferative responses to mechanical strain of cells of the osteoblast lineage. Further studies are required to ascertain whether they also apply to other aspects of bone cells' adaptive responses to strain or to other consequences of mechanical loading.
Our previous finding that in rat female osteoblast-like cells strain appears to exert its proliferative effects through the ER(1) suggested that the diminished effectiveness of strain-related control of bone architecture in women after the menopause could be related to down-regulation of ER number in the absence of estrogen. Such down-regulation has been reported to occur in the resident cells of trabecular bone in ovariectomized rats.(43)
The inference from our current experiments is that the osteoregulatory effects of mechanical strain in male cells are mediated by the ER as in females. This is consistent with the report that the bones of a man with a genetic mutation and no ER were thin and fragile.(44) In the absence of the ER-based machinery by which resident bone cells process strain-related information, this individual's skeleton would be deprived of the mechanically derived stimulus by which normal robustness of the skeleton is achieved and maintained. Essentially, this situation is similar to that which we hypothesize to occur in the postmenopausal skeleton if deprived of the effects of mechanical stimulation by diminished ER number. The anatomical differences between the two situations can be explained by the resident bone cells of the ER-deficient man being deprived of mechanical stimulation during growth, whereas those of postmenopausal women would be putatively (and less completely) deprived after maturity.
In conclusion, the data presented here suggest that primary cultures of LOB derived osteoblast-like cells from male rats proliferate in response to mechanical strain by similar use of the ER as those from females. In these cells from both sexes the proliferative responses to strain and estrogen do not compete but can be blocked by both the SERMs ICI 182,780 and tamoxifen. This suggests that strain and estrogen stimulate proliferation via different domains of the ER.
The androgen receptor is present in male rat primary LOB-derived osteoblasts and is responsive to 5α-dihydrotestosterone. However, it does not appear to be involved in these cells' proliferative responses to strain or estrogen. The proliferative response to testosterone appears to follow its aromatization to estrogen.
As in females, the proliferative response of male rat primary LOB-derived osteoblasts to sex hormone is mediated by IGF-I and that to strain by IGF-II.
We are grateful to Dr. Alan Wakeling of Zeneca, Ltd. for ICI 182,780 and to Dr. R.O. Neri (Shering-plough, NJ, U.S.A.) for hydroxyflutamide. This work was supported by the Wellcome Trust. J.S.P. is a Wellcome Trust Career Development Fellow.