Discussion
Cytosolic ER contains several sites that are sensitive to proteolytic cleavage. On limited digestion, ER fragments of 50 and 37 kDa and lower molecular weight were recorded in uterine cytosol preparations [42]. The presence of such isoforms has generally been attributed to proteolysis of the native ER [43,44]. Regarding ER prepared from breast cancer samples, we previously suggested [11] that proteolytic events that operate at the time of tissue processing might explain the observed extreme ER molecular heterogeneity. Data presented here clearly refute this hypothesis. A high proportion of these isoforms is derived from an intracellular proteolytic activity, as shown by electrophoresis of in situ [125I]TAZ-labelled ER and addition of protease inhibitors to cytosols. In this regard, the present results are in agreement with those reported recently by Navarro et al [45] who demonstrated that the inclusion of PMSF, aprotinin and leupeptin to human uterine cytosolic ER did not prevent the 8S to 4S transformation known to be associated with ER cleavage [46]. Hence, truncated ER isoforms are probably produced before contact with protease inhibitors.
On breast tumour slices, staining intensities (ISs) established with the ER1D5 monoclonal antibody used here have often been reported to be correlated with [3H]oestradiol binding capacities of corresponding cytosolic extracts [5,6,47]. Hence, a consensus seems to exist between immunohistochemical detection of the amino-terminal region of ER (ABC domains) and biochemical measurement of its carboxyl-terminal region (E domain). In agreement with this, we reported a significant correlation between immunohistochemical data and [3H]oestradiol-binding capacities measured by hydroxylapatite assay; ABC domains are indeed required for hydroxylapatite adsorption [12]. For both [3H]oestradiol- and [125I]TAZ labelled-ER isoforms, we found an excellent correlation between DCC values and hydroxylapatite adsorption data, suggesting the concomitant presence of free ABC and E domains in most cytosols (proteolytic fragments physically separated from each other at high ionic strength; high EI values). Hence, these domains would be held together within the cell nucleus because of the tendency of the receptor to be included into oligomeric structures. Corroborating these views of intracellular association between ER fragments, Fritsch et al [42] described large oligomeric complexes containing proteolysis-generated ER fragments in rat uterus.
When expressed individually as separate polypeptides in ER-negative breast cancer cells (MDA-MB-231), neither ABCD (AF-1 containing) nor EF (AF-2 containing) domains activated transcription from a hormone-inducible reporter gene (3ERE-pS2-CAT) [48]. Coexpression of the steroid receptor coactivator-1 protein with either ABCD or EF alone was also ineffective [49]. On the contrary, when expressed together, amino- and carboxyl-terminal ER regions interacted in an oestradiol-dependent manner to reconstitute ER transcriptional activity [48]. In this context, it should be emphasized that steroid receptor coactivator-1 has also been suggested to promote the association between separate amino- and carboxyl-terminal regions of ER, allowing full ER activation [49]. In view of these data, one may speculate that the secondary and tertiary structures of the receptor peptide required for its transcriptional ability would not be lost when it has been subjected to endogenous proteolytic activity. Hence, intracellular ER fragmentation should not necessarily imply lack of transcriptional activity.
Endogenous proteolysis is a mechanism of regulation of many cellular processes, such as cell cycle progression and transcriptional regulation [50,51]. Recent data concerning cell-specific regulation of oestrogen target gene expression in various rat tissues [52] and cell lines [53] demonstrated different levels of coactivators and corepressors. Hence, protein environment would be a factor of major importance in the ability of ER to transcribe genes. Although not presently demonstrated in breast cancer, the possibility of heterogeneous association between cleavage products of ER and other regulatory proteins could be proposed, generating a situation that may mimic the behaviour of chimaeric receptors. Thus, domain-swapping experiments in which the ER amino-terminal domain was switched with that of the glucocorticoid receptor yielded a receptor that upregulates transcription of glucocorticoid-responsive target genes when treated with oestradiol [54]. Investigation of homogenous association and/or heterogeneous association between cleaved ER regions and regulatory proteins would be informative with regard to the validity of this hypothesis.
In view of these considerations, it seems that analysis of the biological role and potential clinical relevance of endogenous ER proteolysis is of prime importance. The hydroxylapatite extraction assay described here, which is extremely simple, would provide a criteria for the detection of tumours characterized by a high amount of full-length ER (low EI). Such tumours may perhaps be considered an indication for adjuvant hormone therapy if we refer to earlier sucrose gradient sedimentation data concerning hormone dependency of tumours that express native and/or cleaved ER forms [17]; very high EI may perhaps be an index of poor prognosis. In practice, for clinical correlation studies devoted to analysis of these hypotheses, we propose that a hydroxylapatite extraction assay should be performed as a complement to the usual immunohistochemical test (or Abbott's ER enzyme immunoassay [4]), in view of the small size of most tumours that restrict the cytosol amount. If our speculations concerning the biological/clinical relevance of the detection of cleaved ER isoforms are verified, then the introduction of such a hydroxylapatite assay into routine practice would be helpful to orientate therapy.