Expression of heregulin by mouse mammary tumor cells: role in activation of ErbB receptors.

The inappropriate activation of one or more members of the ErbB family of receptor tyrosine kinases [ErbB-1 (EGFR), ErbB-2, ErbB-3, ErbB-4] has been linked with oncogenesis. ErbB-2 is frequently coexpressed with ErbB-3 in breast cancer cells and in the presence of the ligand heregulin (HRG) the ErbB-2/ErbB-3 receptors form a signaling heterodimer that can affect cell proliferation and apoptosis. The major goal of the present study was to determine whether endogenous HRG causes autocrine/paracrine activation of ErbB-2/ErbB-3 and contributes to the proliferation of mammary epithelial tumor cells. Tyrosine-phosphorylated (activated) ErbB-2 and ErbB-3 receptors were detected in the majority of extracts from tumors that had formed spontaneously or as a result of oncogene expression. HRG-1 transcripts and protein were found in the epithelial cells of most of these mouse mammary tumors. Various mouse mammary cell lines also contained activated ErbB-2/ErbB-3 and HRG transcripts. A approximately 50 kDa C-terminal fragment of pro-HRG was detected, which indicates that the HRG-1 precursor is readily processed by these cells. It is likely that the secreted mature HRG activated the ErbB-2/3 receptors. Addition of an antiserum against HRG to the mammary epithelial tumor cell line TM-6 reduced ErbB-3 Tyr-phosphorylation. Treatment with HRG-1 siRNA oligonucleotides or infection with a retroviral construct to stably express HRG siRNA effectively reduced HRG protein levels, ErbB-2/ErbB-3 activation, and the rate of proliferation, which could be reversed by the addition of HRG. The cumulative findings from these experiments show that coexpression of the HRG ligand contributes to activation of ErbB-2/Erb-3 in mouse mammary tumor cells in an autocrine or paracrine fashion.

Genotoxic metabolites of estradiol in breast: potential mechanism of estradiol induced carcinogenesis.

Long term exposure to estradiol increases the risk of breast cancer in a variety of animal species, as well as in women. The mechanisms responsible for this effect have not been firmly established. The prevailing theory proposes that estrogens increase the rate of cell proliferation by stimulating estrogen receptor-mediated transcription and thereby the number of errors occurring during DNA replication. An alternative hypothesis proposes that estradiol can be metabolized to quinone derivatives which can react with DNA and then remove bases from DNA through a process called depurination. Error prone DNA repair then results in point mutations. We postulate that these two processes, increased cell proliferation and genotoxic metabolite formation, act in an additive or synergistic fashion to induce cancer. If correct, aromatase inhibitors would block both processes whereas anti-estrogens would only inhibit receptor-mediated effects. Accordingly, aromatase inhibitors would be more effective in preventing breast cancer than use of anti-estrogens. Our studies initially demonstrated that catechol estrogen (CE) quinone metabolites are formed in MCF-7 human breast cancer cells in culture. Measurement of estrogen metabolites and conjugates involved utilization of an HPLC separation coupled with an electrochemical detector. We then utilized an animal model that allows dissociation of estrogen receptor-mediated function from that of the effects of estradiol metabolites. Wnt-1 transgenic mice harboring a knock-out of ERalpha provides a means of examining the effect of estrogen deprivation in the absence of the ER in animals with a high incidence of breast tumors. ERbeta was shown to be absent in the breast tissue of these animals by RNase protection assay. In the breast tissue of these estrogen receptor alpha knock-out (ERKO)/Wnt-1 transgenic mice, we demonstrated formation of genotoxic estradiol metabolites. The ERKO/Wnt-1 breast extracts contained picomole amounts of the 4-catechol estrogens, but not their methoxy conjugates nor the 2-CE or their methoxy conjugates. The 4-CE conjugates with glutathione or its hydrolytic products (cysteine and N-acetylcysteine) were detected in picomole amounts in both tumors and hyperplastic mammary tissue, demonstrating the formation of CE-3,4-quinones. These results are consistent with the hypothesis that mammary tumor development is primarily initiated by metabolism of estrogens to 4-CE and, then, to CE-3,4-quinones, which may react with DNA to induce oncogenic mutations. The next set of experiments examined the incidence of tumors formed in Wnt-1 transgenic mice bearing wild type ERalpha (ER+/+), the heterozygous combination of genes (ER+/ER-) or ERalpha knock-out (ER-/-). To assess the effect of estrogens in the absence of ER, half of the animals were oophorectomized on day 15 and the other half were sham operated. Castration reduced the incidence of breast tumors in all animal groups and demonstrated the dependence of tumor formation upon estrogens. A trend toward reduction in tumor number (not statistically significant at this interim analysis) occurred in the absence of functional ER since the number of tumors was markedly reduced in ERKO animals which were castrated early in life. In aggregate, our results support the concept that metabolites of estradiol may act in concert with ER mediated mechanisms to induce breast cancer.

Catechol estrogen metabolites and conjugates in mammary tumors and hyperplastic tissue from estrogen receptor-alpha knock-out (ERKO)/Wnt-1 mice: implications for initiation of mammary tumors.

A novel model of breast cancer was established by crossing mice carrying the Wnt-1 transgene (100% of adult females develop spontaneous mammary tumors) with the ERKO mouse line, in which mammary tumors develop despite a lack of functional estrogen receptor-alpha. To begin investigating whether metabolite-mediated genotoxicity of estrogens may play an important role in the initiation of mammary tumors, the pattern of estrogen metabolites and conjugates was examined in ERKO/Wnt-1 mice. Extracts of hyperplastic mammary tissue and mammary tumors were analyzed by HPLC with identification and quantification of compounds by multichannel electrochemical detection. Picomole amounts of the 4-catechol estrogens (CE) were detected, but their methoxy conjugates, as well as the 2-CE and their methoxy conjugates, were not. 4-CE conjugates with glutathione or its hydrolytic products (cysteine and N-acetylcysteine) were detected in picomole amounts in both tumors and hyperplastic mammary tissue, demonstrating the formation of CE-3,4-quinones. These preliminary findings show that the estrogen metabolite profile in the mammary tissue is unbalanced, in that the normally minor 4-CE metabolites were detected in the mammary tissue and not the normally predominant 2-CE. These results are consistent with the hypothesis that the mammary tumor development is primarily initiated by metabolism of estrogens to 4-CE and, then, to CE-3,4-quinones, which may react with DNA to induce oncogenic mutations.

Induction of mammary gland development in estrogen receptor-alpha knockout mice.

Mammary glands from the estrogen receptor-a knockout (alphaERKO) mouse do not undergo ductal morphogenesis or alveolar development. Disrupted ERalpha signaling may result in reduced estrogen-responsive gene products in the mammary gland or reduced mammotropic hormones that contribute to the alphaERKO mammary phenotype. We report that circulating PRL is reduced in the female alphaERKO mouse. Implantation of an age-matched, heterozygous ERalpha pituitary isograft under the renal capsule of 25-day-old or 12-week-old alphaERKO mice increased circulating PRL and progesterone levels, and induced mammary gland development. Grafted alphaERKO mice also possessed hypertrophied corpora lutea demonstrating that PRL is luteotropic in the alphaERKO ovary. By contrast, ovariectomy at the time of pituitary grafting prevented mammary gland development in alphaERKO mice despite elevated PRL levels. Hormone replacement using pellet implants demonstrated that pharmacological doses of estradiol induced limited mammary ductal elongation, and estradiol in combination with progesterone stimulated lobuloalveolar development. PRL alone or in combination with progesterone or estradiol did not induce alphaERKO mammary growth. Estradiol and progesterone are required for the structural development of the alphaERKO mammary gland, and PRL contributes to this development by inducing ovarian progesterone levels. Therefore, the manifestation of the alphaERKO mammary phenotype appears due to the lack of direct estrogen action at the mammary gland and an indirect contributory role of estrogen signaling at the hypothalamic/pituitary axis.

A mouse mammary tumor virus-Wnt-1 transgene induces mammary gland hyperplasia and tumorigenesis in mice lacking estrogen receptor-alpha.

Estrogens have important functions in mammary gland development and carcinogenesis. To better define these roles, we have used two previously characterized lines of genetically altered mice: estrogen receptor-alpha (ER alpha) knockout (ERKO) mice, which lack the gene encoding ER alpha, and mouse mammary virus tumor (MMTV)-Wnt-1 transgenic mice (Wnt-1 TG), which develop mammary hyperplasia and neoplasia due to ectopic production of the Wnt-1 secretory glycoprotein. We have crossed these lines to ascertain the effects of ER alpha deficiency on mammary gland development and carcinogenesis in mice expressing the Wnt-1 transgene. Introduction of the Wnt-1 transgene into the ERKO background stimulates proliferation of alveolar-like epithelium, indicating that Wnt-1 protein can promote mitogenesis in the absence of an ER alpha-mediated response. The hyperplastic glandular tissue remains confined to the nipple region, implying that the requirement for ER alpha in ductal expansion is not overcome by ectopic Wnt-1. Tumors were detected in virgin ERKO females expressing the Wnt-1 transgene at an average age (48 weeks) that is twice that seen in virgin Wnt-1 TG mice (24 weeks) competent to produce ER alpha. Prepubertal ovariectomy of Wnt-1 TG mice also extended tumor latency to 42 weeks. However, pregnancy did not appear to accelerate the appearance of tumors in Wnt-1 TG mice, and tumor growth rates were not measurably affected by late ovariectomy. Small hyperplastic mammary glands were observed in Wnt-1 TG males, regardless of ER alpha gene status; the glands were similar in appearance to those found in ERKO/Wnt-1 TG females. Mammary tumors also occurred in Wnt-1 TG males; latency tended to be longer in the heterozygous ER alpha and ERKO males (86 to 100 weeks) than in wild-type ER alpha mice (ca. 75 weeks). We conclude that ectopic expression of the Wnt-1 proto-oncogene can induce mammary hyperplasia and tumorigenesis in the absence of ER alpha in female and male mice. The delayed time of tumor appearance may depend on the number of cells at risk of secondary events in the hyperplastic glands, on the carcinogenesis-promoting effects of ER alpha signaling, or on both.

Estrogen receptor residues required for stereospecific ligand recognition and activation.

The mouse estrogen receptor (mER) has been shown to exhibit stereospecific binding of certain stilbene estrogen agonists. The region of the mER involved in the stereochemical recognition of ligands was further defined using a stilbene isomer, Indenestrol B (IB). The IB compound has a chiral carbon bearing an ethyl substituent, and the wild type uterine mER has been shown to bind the enantiomers, IB-S and IB-R, with similar affinity. The wild type mER expressed in yeast exhibited a very minor preference for IB-S in transactivation (1.5-fold lower half-maximal dose than IB-R). The IB enantiomers could then be used to determine whether stereochemically distinct compounds with similar transcriptional activity utilize different amino acids in AF-2 for transactivation. Mutant mERs with glycine substitutions at Met521, His528, Met532, and Val537 were expressed in yeast and measured for IB-S- and IB-R-induced transactivation and ligand binding. The M532G mER showed a 124-fold and 50-fold reduction in transactivation induced by IB-S and IB-R, respectively, without a corresponding change in their ligand-binding affinities. Therefore, Met532 is required for transactivation induced by both IB enantiomers but does not discriminate based on stereospecificity. In contrast, the H528G mER displayed a gross change in stereospecific ligand recognition as illustrated by a 110-fold reduction in transactivation by IB-S and only a 8.8-fold decrease in activity by IB-R. As a result, H528G mER displayed a switch in ligand preference such that IB-R was now 8-fold more active than IB-S in transactivation. Therefore, His528 appears largely involved in transactivation specifically induced by IB-S but has a minimal influence in IB-S ligand binding. The remaining mutant mERs displayed wild type ligand binding and transactivation properties for the IB enantiomers illustrating no stereospecific recognition. These results imply that individual IB enantiomers bind to the mER with similar affinity but utilize at least one different amino acid within the AF-2 domain for signal transduction. The binding of stereochemically distinct ligands may alter the tertiary structure of the mER and cause repositioning of the AF-2 region that mediates transcription of specific genes and/or affect the binding of receptor-associated proteins, such as coactivators, which could influence transcription.

Estrogenic activity of a dieldrin/toxaphene mixture in the mouse uterus, MCF-7 human breast cancer cells, and yeast-based estrogen receptor assays: no apparent synergism.

The estrogenic activity of dieldrin, toxaphene, and an equimolar mixture of both compounds (dieldrin/toxaphene) was investigated in the 21-day-old B6C3F1 mouse uterus, MCF-7 human breast cancer cells, and in yeast-based reporter gene assays. Treatment of the animals with 17beta-estradiol (E2) (0.0053 kg/day x3) resulted in a 3.1-, 4.8-, and 7.8-fold increase in uterine wet weight, peroxidase activity, and progesterone receptor binding, respectively. In contrast, treatment with 2.5, 15 and 60 micromol/kg (x3) doses of toxaphene, dieldrin, or dieldrin/toxaphene (equimolar) did not significantly induce a dose-dependent increase in any of the E2-induced responses. The organochlorine pesticides alone and the binary mixture did not bind to the mouse uterine estrogen receptor (ER) in a competitive binding assay using [3H]E2 as the radioligand. In parallel studies, estrogenic activities were determined in MCF-7 cells by using a cell proliferation assay and by determining induction of chloramphenicol acetyl transferase (CAT) activity in MCF-7 cells transiently transfected with plasmids containing estrogen-responsive 5'-promoter regions from the rat creatine kinase B and human cathepsin D genes. E2 caused a 24-fold increase in CAT activity in MCF-7 cells transiently transfected with creatine kinase B and a 3.8-fold increase in cells transiently transfected with the human cathepsin D construct. Treatment of MCF-7 cells with dieldrin, toxaphene, or an equimolar mixture of dieldrin plus toxaphene (10(-8)-10(-5) M) did not significantly induce cell proliferation or CAT activity in the transient transfection experiment with both plasmids. The relative competitive binding of the organochlorine pesticides was determined by incubating MCF-7 cells with 10(-9) M [3H]E2 in the presence or absence of 2 x 10(-7) M unlabeled E2 (to determine nonspecific binding), toxaphene (10(-5) M), dieldrin (10(-5) M), and equimolar concentrations of the dieldrin plus toxaphene mixture (10(-5) M). The binding observed for [3H]E2 in the whole cell extracts was displaced by unlabeled E2, whereas the organochlorine pesticides and binary mixture exhibited minimal to nondetectable competitive binding activity. E2 caused a 5000-fold induction of beta-galactosidase (beta-gal) activity in yeast transformed with the human ER and a double estrogen responsive element upstream of the beta-gal reporter gene. Treatment with 10(-6)-10(-4) M chlordane, dieldrin, toxaphene, or an equimolar mixture of dieldrin/toxaphene did not induce activity, whereas 10(-4) M endosulfan caused a 2000-fold increase in beta-gal activity. Diethylstilbestrol caused a 20-fold increase in activity in yeast transformed with the mouse ER and a single estrogen responsive element upstream of the beta-gal reporter gene. Dieldrin, chlordane, toxaphene, and endosulfan induced a 1.5- to 4-fold increase in activity at a concentration of 2.5 x 10(-5) M. Synergistic transactivation was not observed for any equimolar binary mixture of the pesticides at concentrations of either 2.5 x 10(-5) M or 2.5 x 10(-4) M. The results of this study demonstrate that for several estrogen-responsive assays in the mouse uterus, MCF-7 human breast cancer cells, and yeast-based reporter gene assays, the activities of both dieldrin and toxaphene were minimal, and no synergistic interactions were observed with a binary mixture of the two compounds.

Mammary gland development and tumorigenesis in estrogen receptor knockout mice.

Estrogens are important for the development of the mammary gland and strongly associated with oncogenesis in this tissue. The biological effects of estrogens are mediated through the estrogen receptor (ER), a member of the nuclear receptor superfamily. The estrogen/ER signaling pathway plays a central role in mammary gland development, regulating the expression and activity of other growth factors and their receptors. The generation of the ER knockout (ERKO) mouse has made it possible to directly understand the contribution of ER in mammary development and has provided an unique opportunity to study estrogen action in carcinogenesis. A mammary oncogene (Wnt-1) was introduced into the ERKO background to determine if the absence of the ER would affect the development of tumors induced by oncogenic stimulation. The development, hyperplasia, and tumorigenesis in mammary glands from the ERKO/Wnt-1 mouse line are described. These studies provide the impetus to evaluate the effect of other oncogenes in mammary tumorigenesis in the absence of estrogen/ER signaling.

Mutational analysis of the estrogen receptor ligand-binding domain: influence of ligand structure and stereochemistry on transactivation.

The mouse estrogen receptor (mER) exhibits ligand stereochemical specificity for indenestrol A (IA), a stilbestrol estrogen. IA has a chiral C3 methyl group, and the mER preferentially binds the S-enantiomer (IA-S), resulting in elevated biological activity when compared with the IA-R enantiomer. To elucidate the mechanisms for this stereochemical recognition, we have constructed a series of mERs with individual amino acid substitutions at Met521, His528, Met532, and Val537. The abilities of yeast-expressed wild-type and mutant mERs to transactivate an estrogen-responsive reporter gene construct were measured in the presence of diethylstilbestrol (DES) and IA enantiomers. The concentration of IA-S required to induce half-maximal transactivation by wild-type mER was 10-fold lower than IA-R, which is attributed to the 15-fold greater binding affinity for IA-S. Wild-type mER displayed similar dose-response curves for IA-R and demethyl IA, which lacks a C3 methyl group, demonstrating that the presence and correct orientation of the C3 methyl group on the IA compound is required for high-affinity ligand binding and transcriptional activity. Each mutant exhibited a reduced preference for IA-S enantiomer with respect to transactivation, suggesting that this region of the mER functions in ligand stereochemical recognition and activation. A mutation at Met532 diminished DES- and IA-S-induced transactivation by 7.5-fold and 40-fold respectively, with minimal change on their binding affinity. These data suggest that Met532 is required for transactivation induced by the potent agonist, IA-S, and the M532G mutation effectively uncouples IA-S ligand binding from transactivation. Use of these stereochemically different ligands in combination with mutagenesis of the mER demonstrates that ligand structure could influence transactivation by specifically altering the conformation of the mER AF-2 region.

Sex hormone-binding globulin: gene organization and structure/function analyses.

Human plasma sex hormone-binding globulin (SHBG) and testicular androgen-binding protein (ABP) are homodimeric glycoproteins with a single steroid-binding site. They share the same primary structure and differ only with respect to the types of oligosaccharides attached to them. Both are products of a single gene (Shbg), the expression of which has been detected in several tissues including liver, testis, placenta, brain and endometrium. The transcript encoding SHBG in hepatocytes and ABP in Sertoli cells is identical. Several other transcripts result from differential exon utilization in sex steroid-responsive tissues, but their biological significance remains obscure. Wild-type and mutant forms of human SHBG have been produced in mammalian cells and Escherichia coli, and have provided insight into the structural and functional properties of SHBG and its related gene products.

Sex hormone-binding globulin/androgen-binding protein: steroid-binding and dimerization domains.

Plasma sex hormone-binding globulin (SHBG) and testicular androgen-binding protein (ABP) are homodimeric glycoproteins that share the same primary structure, and differ only with respect to the types of oligosaccharides associated with them. The biological significance of these differences is not understood, but enzymatically deglycosylated SHBG and a non-glycosylated SHBG mutant both bind steroids normally. Various affinity-labelling experiments, and studies of recombinant SHBG mutants have indicated that a region encompassing and including Met-139 in human SHBG represents an important component of its steroid-binding site. Analyses of chimeric proteins comprising various portions of human SHBG and rat ABP have also indicated that residues important for the much higher affinity of human SHBG for steroid ligands are probably located within the N-terminal portion of these molecules. Recent studies of SHBG mutants have confirmed this, and a deletion mutant containing only the first 205 N-terminal residues of human SHBG has been produced which dimerizes and binds steroids appropriately. The introduction of amino-acid substitutions between Lys-134 and Phe-148 of SHBG has also indicated that residues including and immediately N-terminal of Met-139 may influence steroid-binding specificity, while those immediately C-terminal of Met-139 represent at least a part of the dimerization domain. These studies have also demonstrated that dimerization is induced by the presence of steroid ligand in the binding site, and that divalent cations play an important role in this process. Together, these data have led us to conclude that SHBG is a modular protein, which comprises an N-terminal steroid-binding and dimerization domain, and a C-terminal domain containing a highly-conserved consensus sequence for glycosylation that may be required for other biological activities, such as cell-surface recognition.

Resolution of the steroid-binding and dimerization domains of human sex hormone-binding globulin by expression in Escherichia coli.

To determine the minimal sequence requirements for steroid binding and dimerization of human sex hormone-binding globulin (SHBG), the SHBG polypeptide and various SHBG deletion mutants were expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli. Fusion proteins containing the complete SHBG sequence, or the first 177 N-terminal residues of SHBG, bound steroids with high affinity and specificity. Further deletions from the C-terminus severely compromised steroid-binding activity, as did N-terminal deletions beyond residue 18 in the SHBG sequence. Thus, residues 18-177 in SHBG encompass a region required for its steroid-binding activity, and a disulfide bridge normally present between Cys-164 and Cys-188 in SHBG is not obviously essential for steroid binding. Most of the GST/SHBG fusion proteins undergo cleavage at 4 degrees C, releasing immunoreactive polypeptides that correspond approximately in size to their respective SHBG sequences. The 23-kDa immunoreactive cleavage product released from the fusion protein containing residues 1-205 in the SHBG sequence (SHBG 1-205) has a 50-fold greater steroid-binding capacity but a 7.5-fold lower affinity than its parent fusion protein. In addition, the 22-kDa immunoreactive polypeptide released from SHBG(1-194) binds steroid, and its dimerization is promoted by steroid ligands that bind SHBG with high affinity. These data suggest that the N-terminal region of SHBG dimerizes readily in the absence of GST and in doing so acquires steroid-binding sites.(ABSTRACT TRUNCATED AT 250 WORDS)

Serum distribution of two contraceptive progestins: 3-ketodesogestrel and gestodene.

A cross-over study of two oral contraceptive formulations, containing 30 micrograms ethinylestradiol in combination with 150 micrograms desogestrel (Marvelon) or 75 micrograms gestodene (Femovan), has been performed to compare the serum distribution and pharmacokinetics of gestodene and the active metabolite of desogestrel, namely 3-ketodesogestrel. Serum concentrations of both sex hormone-binding globulin (SHBG) and corticosteroid-binding globulin (CBG) were also measured and were increased more than 3-fold and 2-fold, respectively, on day 21 of the treatment cycle, with no statistically significant difference between treatment groups. In addition, 35 days after ingestion of either oral contraceptive had ceased, the serum SHBG and CBG concentrations were similar to the pretreatment values. During treatment cycles, increased serum SHBG levels were associated with a redistribution of 3-ketodesogestrel and gestodene such that the non-protein-bound (NPB) and albumin-bound fractions were reduced in concert with an increase in the relative proportions bound to SHBG. The proportion of gestodene bound to SHBG was consistently higher than that observed for 3-ketodesogestrel, and this undoubtedly reflects the higher affinity of SHBG for gestodene (Kd = 1.2 nM at 37 degrees C) when compared to 3-ketodesogestrel (Kd = 4.7 nM at 37 degrees C). It also probably accounts, in part, for the much higher total serum levels of gestodene (8.58 nmol/L) when compared to 3-ketodesogestrel (2.37 nmol/L) during the treatment cycles. Consequently, the absolute amounts of NPB, non-SHBG-bound, and SHBG-bound gestodene are significantly higher than those measured for 3-ketodesogestrel. It is concluded that ethinylestradiol-induced increases in serum SHBG levels during treatment with Marvelon or Femovan, influenced the distribution and total amount of 3-ketodesogestrel and gestodene in serum, respectively, and that this, combined with the higher affinity of SHBG for gestodene, results in a greater amount of bioavailable gestodene compared to 3-ketodesogestrel, despite the smaller dose of gestodene administered.

Steroid-binding and dimerization domains of human sex hormone-binding globulin partially overlap: steroids and Ca2+ stabilize dimer formation.

Human sex hormone-binding globulin (SHBG) is a homodimeric plasma glycoprotein with a single steroid-binding site for biologically active sex steroids, and a methionine at position 139 (M139) interacts with the photoaffinity ligand, delta 6-testosterone. We have introduced amino acid substitutions into this and other locations in the SHBG molecule and have examined their impact on steroid binding and dimerization. As a result, substitutions at residues 134-139 generate alterations steroid-binding specificity. In particular, substitutions at residues 134-138 were characterized by altered binding affinities for estradiol relative to 5 alpha-dihydrotestosterone (DHT), and one of them (R135L) also showed a 2-fold increase in affinity for C19 steroids with a 3 beta-hydroxy group. Unlike all of the other mutants we have examined, the M139W mutant has a 5-fold lower affinity for DHT, and its affinities for testosterone, 5 alpha-androstane-3 beta, 17 beta-diol, and estradiol also appear to be reduced to a similar extent. By contrast, M139W appears to bind androst-5-ene-3 beta, 17 beta-diol with only 2-fold less affinity than wild-type SHBG, while its affinity for 19-nortestosterone remains unaffected. Substitutions at other positions, including those immediately C-terminal to M139, had no effect on steroid-binding affinity and/or specificity. These data provide evidence that residues 134-139 influence the recognition of specific A/B ring conformations of steroid ligands and may constitute part of the steroid-binding domain. We have also found that substitutions at residues 138-148 impair dimerization and that this defect may be abrogated by occupancy of the steroid-binding site.(ABSTRACT TRUNCATED AT 250 WORDS)

Selective removal of glycosylation sites from sex hormone-binding globulin by site-directed mutagenesis.

Sex hormone-binding globulin (SHBG) is a homodimeric plasma glycoprotein with high affinity for sex steroid hormones. It contains two N-linked carbohydrate chains and one O-linked oligosaccharide per subunit, but their functional significance is not known. Site-directed mutagenesis of a human SHBG cDNA has enabled us to selectively disrupt the known glycosylation sites individually and in various combinations. The mutant cDNAs were expressed in Chinese hamster ovary (CHO) cells, and it was found that the presence of carbohydrates is not an absolute requirement for the secretion of SHBG from these cells, but the absence of both N-linked oligosaccharides reduced the amount of SHBG in the culture medium. In addition, the affinity and specificity of SHBG for steroid ligands was unaffected by the lack of one or more carbohydrate chains. Proportionally greater amounts (26-31%) of the mutants lacking a single N-linked carbohydrate chain failed to interact with Concanavalin-A. (Con-A) compared to normal SHBG produced by CHO cells (15%). Western analysis demonstrated that both consensus sites for N-glycosylation are used and that the typical heavy [mol wt (M(r)), approximately 51,000] and light (M(r), approximately 47,000) subunit size-heterogeneity was maintained regardless of the absence of an O-linked carbohydrate at residue 7. Furthermore, the SHBG mutants containing only one N-linked oligosaccharide comprise only a single subunit with a M(r) of approximately 47,000. This suggests that the heavy subunit contains two N-linked oligosaccharides, while only one of these sites is used on the light subunit. The M(r) of the various SHBG mutants were also examined by gel filtration, and this indicated that they are all produced as homodimers and that carbohydrates are not involved in subunit association.

Molecular analyses of a human sex hormone-binding globulin variant: evidence for an additional carbohydrate chain.

Genomic DNA was isolated from an individual who is homozygous for a sex hormone-binding globulin (SHBG) variant that resolves into three molecular weight forms of 56K, 52K, and 48K during electrophoresis under denaturing conditions (sodium dodecyl sulfate-polyacrylamide gel electrophoresis). This material was amplified using intron-specific oligonucleotide primers in a polymerase chain reaction to obtain the eight exons encoding SHBG. Sequence analysis of these exons revealed a point mutation encoding an amino acid substitution (Asp --> Asn) at residue 327 in the SHBG polypeptide, and the same mutation was identified in three siblings who also appear to be homozygous for this trait. This mutation introduces an additional consensus site for N-glycosylation at this position, and to confirm its utilization we introduced it into a human SHBG complementary DNA. The mutated complementary DNA was inserted into the pRc/CMV expression vector, and transfected into Chinese hamster ovary (CHO) cells. The product was secreted normally but proportionally less of it (54%) bound to concanavalin A when compared to normal SHBG produced by CHO cells (85%), or SHBG in the serum of either a normal individual or those who produce an electrophoretic variant (98%). Furthermore, when subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blotting, the SHBG variant produced by CHO cells consisted of a 60K subunit as well as the heavy (52K) and light (48K) subunits associated with normal SHBG produced by CHO cells or in serum. This additional subunit is larger than the variant in serum and probably reflects a greater degree of complexity in the carbohydrate structures added to recombinant SHBG during synthesis in CHO cells. Nevertheless, its steroid-binding affinity was equal to normal SHBG produced by CHO cells or SHBG in serum.

Structure/function analyses of human sex hormone-binding globulin by site-directed mutagenesis.

Human sex hormone-binding globulin (hSHBG) and rat androgen-binding protein (rABP) exhibit distinct affinities for sex-steroids. We therefore constructed and expressed a hSHBG/rABP hybrid cDNA encoding the N-terminal portion of hSHBG (205 residues) and the C-terminal portion of rABP (168 residues). The resulting chimera displayed similar steroid-binding characteristics as hSHBG and was recognised by a monoclonal antibody (S1B5) for hSHBG. We then created substitutions at Ser-133, His-136 and Met-139. The Asp-133 and Gln-136 mutants bound steroids in the same way as normal hSHBG while the steroid-binding affinity of Trp-139 was reduced. All three mutants cross-reacted similarly in a hSHBG radioimmunoassay, but Gln-136 was recognised poorly by the S1B5 antibody. These data imply that residues involved in steroid-binding are located within the N-terminal half of hSHBG and include Met-139, and that the S1B5 epitope is located in this region.

Expression and differential glycosylation of human sex hormone-binding globulin by mammalian cell lines.

The human sex hormone-binding globulin (SHBG) gene is responsible for the production of plasma SHBG by the liver and androgen-binding protein in the testis. Cell-specific glycosylation events during synthesis may account for minor differences in the biochemical properties of SHBG and androgen-binding protein, and we have, therefore, expressed a human SHBG cDNA in chinese hamster ovary (CHO) cells and a mouse hepatoma cell line (BW-1), and compared the products to SHBG in serum. The SHBG produced in this way is a homodimer of subunits that exhibit size microheterogeneity similar to SHBG in human serum, and its affinity for 5 alpha-dihydrotestosterone (Kd = 0.6 nM) and other steroids is essentially identical to that of natural SHBG. When medium from transfected CHO and BW-1 cells was subjected to Concanavalin-A (Con-A) chromatography, the relative amounts of SHBG retained by Con-A were 74% and 86%, respectively. In addition, when SHBG produced by CHO cells was separated into two fractions by Con-A chromatography and analyzed by polyacrylamide gel electrophoresis, SHBG that did not interact with Con-A migrated with a slightly larger apparent mol wt than that of SHBG that binds Con-A; this can be explained by the presence of triantennary, rather than biantennary, N-linked oligosaccharide chains. These data also demonstrate that the subunit microheterogeneity associated with plasma SHBG reflects differences in glycosylation during synthesis, which appear to be cell type specific.