BoneKEy-Osteovision | Commentary

Does sex make a difference? - sex-neutral nongenomic signaling by estrogens and androgens prevents bone cell death



DOI:10.1138/2001025

Estrogens (E) and androgens (A) are known to be bone-sparing (). Indeed, loss of either hormone as a result of menopause or andropause, or because of chemical or surgical gonadectomy, results in a rise in bone turnover and elevated bone resorption (). The net consequence of this activity is decreased bone density and increased risk of fraction (). While much of the action of E and A has been focused on the regulation of osteoclastogenesis, sex steroids also exert a variety of effects on osteoblasts (OB) (). The potentially anabolic actions of the two hormones are well defined, although their overall impact on the biology of the OB appears to be similar. These and other findings have led investigators to speculate that the biologically-active bone steroid in males is E rather that testosterone ().

E and A function via their individual receptors ER (estrogen receptor) and AR (androgen receptor) to regulate transcription through relatively well understood mechanisms (). The receptors are largely but not exclusively localized to the nucleus, are activated in a highly specific and selective manner by their respective ligands and become associated with specific regulatory elements located adjacent to certain gene promoters. These receptors can also modulate the activity of numerous heterodimeric regulatory factors including NF-κB, and AP-1 via a tethering mechanism that is dependent upon protein/protein interactions rather than receptor DNA binding (). Regardless of the mechanism, these actions precipitate a series of events that lead to tissue-specific changes in gene expression (). Steroid hormones also direct a variety of nongenomic events that involve actions at the level of the cellular membrane or in the cytoplasm (). The involvement of the nuclear receptors in these actions varies, however; some researchers report that the nuclear receptors are required while others find that they are not. Others suggest that novel forms of the receptors are mediators of these nongenomic actions. Finally, while most of these activities require ligand diffusion into the cell, others can be modulated through membrane-impermeable ligands (). The diversity of these actions suggests that steroids may operate to regulate cellular activity through multiple mechanisms.

The article by Kousteni and colleagues () describes work that reveals that E and A are capable of regulating the survival of bone forming OBs. Accordingly, E and A exert antiapoptotic effects on OBs. This effect is the opposite of that seen in bone resorbing osteoclasts (OC), and may form the basis for the imbalance that ensues following menopause between OBs and OCs that leads to osteoporosis. Mechanistically, the current studies show that E and A act in a rapid, nongenomic fashion to activate the ERK1/2 pathway that enhances OB survival in the presence of apoptotic agents such as etoposide. While none of these actions are necessarily surprising, what follows is truly unexpected. The authors discover that both17β-estradiol (E2) and 5α-DHT are equally effective in blocking apoptosis. Furthermore, the actions of each of these hormones can be abrogated by identical concentrations of both the ER antagonist ICI 182780 (ICI) and the AR antagonist flutamide (F). Finally, both ER and AR can mediate the effects of either ligand. These findings suggest a sex-neutral mechanism wherein either receptor can be activated by E or A with apparently identical cellular consequences. Using E as a paradigm, the authors show further that the antiapoptotic activity of ER is preserved when the receptor is targeted to the cytoplasm or membrane of the cell, but lost if it is targeted to the nucleus. Interestingly, while the antiapoptotic action of ER does not require DNA binding, it is dependent upon the ability of the receptor to bind estrogen and to regulate transcription via activation function 2 (AF-2). AF-2 is a region at the carboxy terminus of the ligand binding domain (LBD) of ER involved in comodulator interaction. Finally, the authors show that the antiapoptotic effects of E can be dissociated from its transcriptional effects using the novel ER ligand estren. These studies suggest a new mechanism to account for the similarity of action of both E and A on OB biology in both males and females.

These newly discovered activities of E and A in bone cells are highly provocative. As might be expected, however, the results raise new and pertinent questions of both a biological as well as technical nature. Until they are appropriately answered, it would seem unwise to adopt the findings of Kousteeni et al. () in their entirety.

Questions of a general biological nature are as follows. First, how significant is the role of E and A in OB survival in vivo? This research group and others have proposed a role for the sex steroids in OB or osteocyte survival and apoptosis, and have provided preliminary evidence in support of these biological effects (). Nevertheless, additional work will be necessary to establish firmly the nature of the signals that determine OB lifespan, to define the nature of the apoptotic processes, to assess whether sex steroids can oppose these natural signals thereby prolonging OB lifespan, and finally, to determine the features of this inhibitory process. Second, are the biological actions of E and A reported here representative of those that occur in vivo? Kousteni and colleagues () use a rather extensive series of cellular models that are both innovative and strategic to make their case. Unfortunately, the diversity and complexity of these cell types are likely to make it difficult to assess the relevance of these findings in vivo.

Numerous technical questions emerge directly from the authors’ experimental findings. As summarized earlier, the initial finding is as follows: 1) E2 and 5α-DHT both suppress apoptosis in OBs and 2) suppression is reversed nonspecifically by both ICI and F. These observations are surprising in view of the strong receptor selectivity characteristic of both ICI and F. Some lack of selectivity by ICI was noted through the progesterone receptor, but this relaxed selectivity did not extend to either AR or the glucocorticoid receptor (). A second finding is that the antiapoptotic effects of both E2 and 5α-DHT in HeLa cells can be mediated by either ER or AR. These results imply a lack of cognate receptor specificity for the activities of E2 and 5α-DHT and call into question the unique physical interaction and specificity that occurs between hormonal ligands and their respective receptors. It is satisfying that the activities of E and A are not mediated by the vitamin D receptor (VDR) or the retinoid X receptor (RXR) (except perhaps at the highest concentrations). Nevertheless, these negative controls introduce additional questions. Since 1,25(OH)2D3 (VD) can also suppress etoposide-induced apoptosis in the target cell line, how general is this antiapoptotic effect with respect to other hormones? Since VD also blocks apoptosis in the presence of RXR, which is the receptor for VD, VDR or RXR? Is RXR unable to discriminate between 9-cis retinoic acid and VD in the same way that ER and AR are unable to distinguish their two ligands? More importantly, why are endogenous VDR and RXR (known to be expressed in HeLa cells) unable to suppress etoposide-induced apoptosis. These findings seriously challenge our conceptual understanding of steroid hormone selectivity, specificity and action and, if correct, lead us inescapably to search for a new mechanism that can account for these interesting actions.

Kousteni et al. () next investigate this unusual mechanism using a variety of molecular reagents. In an elegant set of experiments, the authors show that the antiapoptotic action of E2 does not require nuclear localization of ER. Indeed, targeting the ER to the nucleus using an SV40 nuclear localization signal (NLS) prevents E2 action. Additional studies reveal that although the amino terminus and the DNA binding region of ER are extraneous, the LBD is required for the antiapoptotic effect. Interestingly, single amino acid changes in ER, one leading to a loss of hormone binding and the second that prevents comodulator interaction, compromise the receptor's antiapoptotic activity. The result of this latter mutation is particularly interesting since AF-2 represents a protein/protein interaction domain that is required for transcriptional regulation. Thus, AF-2 appears to function in some unknown way in the cytoplasm as well. The above studies support the idea that the antiapoptotic effects of E2 occur outside the nucleus and require a fully functional LBD, although they do not define a discrete mechanism.

The authors also take a novel approach to explore the idea that the antiapoptotic effects of estrogen are mediated by a mechanism unique from conventional nuclear actions. They introduce previously identified antagonist peptides into cells via expression vectors and assess their ability to block the hormone's effects on both transcription and apoptosis. Pioneered by McDonnell and associates (), these peptides represent novel helical motifs that interact directly with ERs and block transactivation. The authors find that at least one of the peptides is unable to reverse the antiapoptotic effect of E while blocking the classical E actions, whereas others block both the steroid's antiapoptotic activity as well as transcription in a receptor-selective manner. It is proposed that these results support the existence of a novel non-nuclear mechanism of action for E. Unfortunately, while this may be true, the results are also problematic. The antagonist peptides are expressed as fusion proteins comprised of the DNA binding domain and NLS of GAL4. As such, these peptides localize to the nucleus of target cells where they act (). Since the authors have shown in earlier experiments that such targeting of ER prevents E from exerting its antiapoptotic effects, it is unclear how these peptides can inhibit E action.

What are the signaling intermediates triggered by E2 that initiate the antiapoptotic response in osteoblasts? Through a series of rather elegant experiments using a variety of cell lines and macromolecules harboring well characterized mutations, the authors show that E2 acts to stimulate the src/MEK/ERK1/2 pathway. Others have shown similar activation of these pathways with estrogen () as well as androgens in cancer cells (). This effect is, nevertheless, highly transient and connected only indirectly to the antiapoptotic effect exhibited by E. It will therefore be extremely important to define the direct targets of ERK1/2 activation and to assess their ability to prolong the actual lifespan of the OB.

In a final set of experiments, the authors demonstrate that the transcription-inducing effects of E can be dissociated from its antiapopotic-inducing effects using synthetic ligands. Accordingly, while pyrazole regulates both activities of ER, estren modulates only its antiapoptotic activity. These findings parallel the discovery of ligands for other receptors that exhibit mechanistic selectivity (). Thus, they are extremely exciting from a potential therapeutic perspective. It must be stated, however, that the relative cellular activities of these two ligands in vitro are rather difficult to assess. First, little information or references are provided regarding the biological activity of these compounds, their metabolism or their affinity for ER. Second, the activities of these novel ligands need to be assessed in context with the activities of other bone-active estrogens such as E2, tamoxifen and raloxifene. Third, comparison of a synthetic transcriptional response (the C3 promoter) in HeLa cells with that of a bone fide biological response (antiapoptosis) in OBs seems problematic at best. Since a variety of endogenous transcriptional targets for E have been defined in OBs, why not explore estren regulation of these genes. At the very least, why choose the C3 promoter? Its relative activity following transfection is highly cell-specific () and its regulation by E mediated by a nonclassical DNA response element (). Thus, although the observations reported here with pyrazole and estren are extremely interesting, and predictive of a potentially new class of bone-active estrogens, further studies both in vitro and in vivo will be necessary to assess their absolute impact.

To summarize, Kousteni et al. () provide new and interesting data that demonstrate that sex steroids protect OBs through a novel mechanism that is nongenomic in nature. If true, selective targeting of this mechanism might lead to unique anabolic actions that could result in elevated bone formation. Additional experimentation is needed both in vitro and in vivo, however, before these newly found effects of steroids in bone can be fully embraced. This is due largely to the novelty of the findings and to the fact that many of its features are inconsistent with our current understanding of how steroids selectively interact with their receptors. It is suggested that the lack of specificity of E and A on either ER or AR results from a transient receptor-ligand interaction that results in a receptor conformation capable of unique short term biological actions (). This explanation is based upon the long held assumption that nongenomic actions are considered to be rapid whereas those that are genomic in nature are slower and more sustained. I would submit that both types of events actually occur in a rather similar time-frame. Accordingly, whether they are genomic or nongenomic, the initial events in steroid hormone signaling (nongenomic: MAPK activation, cAMP, Ca++) (8-10,17,18) (genomic: receptor activation, formation of functional dimers, association with DNA, initiation of transcription) () all occur within several minutes. Common to these rapid activities regardless of whether they are genomic or nongenomic are the downstream biological consequences; all occur in hours or days. The events characterized in the present work are no different. The cells are treated with E or A for 1 hour prior to administration of the apoptotic agent, and 6 hours later the number of apoptotic cells are counted. If the time-frame for the signaling actions seen here is indeed similar to those of classical genomic mechanisms, it is difficult to understand how a transient receptor conformation with a unique functional activity can account for the unusual effects seen in this interesting study.


Creative Commons License This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.