Human adrenodoxin: cloning of three cDNAs and cycloheximide enhancement in JEG-3 cells.

Adrenodoxin is an iron-sulfur protein serving as an electron transport intermediate for two mitochondrial steroidogenic cytochromes P450. We have cloned and sequenced three human adrenal adrenodoxin cDNAs. The longest 5'-untranslated region was 131 bases long, and the coding sequences, identical in all three clones, predict a preprotein of 180 amino acids. The 3'-untranslated regions were 235, 596, and 776 bases long due to the presence of alternate polyadenylation sites. RNA transfer blots showed multiple size species of adrenodoxin mRNA consistent with finding multiple polyadenylation sites. Similar sized cross-hybridizing RNA species are found abundantly in the adrenal and testis and to a lesser degree in RNA from human fetal brain, spleen, placenta, kidney, liver, and intestine, as well as in cultured fibroblasts, suggesting the same or a very similar iron-sulfur protein is found in mitochondria of nonsteroidogenic tissues. JEG-3 cells, a transformed progesterone-producing line of trophoblastic origin, accumulate mRNAs for cytochrome P450scc (the mitochondrial cholesterol side-chain cleavage enzyme), adrenodoxin, and the fos oncogene when stimulated with 8-bromo-cyclic AMP. Addition of actinomycin D to such cultures blocked cAMP-induced accumulation of mRNAs for cytochrome P450scc and adrenodoxin. Addition of cycloheximide or puromycin to such cultures substantially reduced basal levels and markedly attenuated the cAMP-induced accumulation of cytochrome P450scc mRNA, but augmented the accumulation of adrenodoxin and fos mRNAs in additive and multiplicative fashions, respectively. These data indicate that the cAMP-induced synthesis of the steroidogenic machinery is not wholly dependent on cycloheximide-sensitive protein mediators.

Adrenodoxin is an iron-sulfur protein serving as an electron transport intermediate for two mitochondrial steroidogenic cytochromes P450. We have cloned and sequenced three human adrenal adrenodoxin cDNAs. The longest 5'-untranslated region was 131 bases long, and the coding sequences, identical in all three clones, predict a preprotein of 180 amino acids. The 3'-untranslated regions were 235, 696, and 776 bases long due to the presence of alternate polyadenylation sites. RNA transfer blots showed multiple size species of adrenodoxin mRNA consistent with finding multiple polyadenylation sites. Similar sized cross-hybridizing RNA species are found abundantly in the adrenal and testis and to a lesser degree in RNA from human fetal brain, spleen, placenta, kidney, liver, and intestine, as well as in cultured fibroblasts, suggesting the same or a very similar iron-sulfur protein is found in mitochondria of nonsteroidogenic tissues. JEG-3 cells, a transformed progesterone-producing line of trophoblastic origin, accumulate mRNAs for cytochrome P45Oscc (the mitochondrial cholesterol side-chain cleavage enzyme), adrenodoxin, and the fos oncogene when stimulated with 8-bromo-cyclic AMP. Addition of actinomycin D to such cultures blocked CAMP-induced accumulation of mRNAs for cytochrome P45Oscc and adrenodoxin. Addition of cycloheximide or puromycin to such cultures substantially reduced basal levels and markedly attenuated the CAMP-induced accumulation of cytochrome P45Oscc mRNA, but augmented the accumulation of adrenodoxin and fos mRNAs in additive and multiplicative fashions, respectively. These data indicate that the CAMP-induced synthesis of the steroidogenic machinery is not wholly dependent on cycloheximide-sensitive protein mediators.
The first and rate-limiting step in the synthesis of steroid hormones from cholesterol is mediated by a single mitochondrial species of cytochrome P450, termed P45Oscc, which *This work was supported by Grant 6-396 from the March of Dimes and Grant DK 39773 from the National Institutes of Health (to W. L. M.), by a grant from the Mellon Foundation and Grant HD 06274 from the National Institutes of Health (to J. F. S.), and by a scholarship from the ARCS Foundation (to J. P.-L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper hos been submitted to the GenBankTM/EMBL Data Bank with accession number(s) J03548.
W To whom all correspondence should be addressed. catalyzes the 20-hydroxylation, 22-hydroxylation, and sidechain cleavage of cholesterol, all on a single active site, to yield pregnenolone and isocaproic acid (for review, see Ref. 1). This unique P45Oscc is encoded by a single gene lying on human chromosome 15 (2,3). Each of the three enzymatic activities of P45Oscc requires the transfer of a pair of electrons from NADPH via a flavoprotein, termed adrenodoxin reductase, and an iron-sulfur protein, termed adrenodoxin. These same two electron transport intermediates also serve the other mitochondrial steroidogenic enzyme, P450cll. Bovine adrenodoxin is 114 amino acids long (4), arising by cleavage from a larger 186-amino acid precursor having amino and carboxyl extensions of 58 and 14 amino acids, respectively (5). However, little is known about human adrenodoxin.
We report the cloning of three human adrenodoxin cDNAs.
The sequences of these cDNAs indicate they arose from a single gene but that the transcripts are polyadenylated at three alternative sites. The sizes of the predicted corresponding mRNAs correlate well with the sizes seen on RNA transfer blots. Like adrenodoxin mRNA in cultured bovine adrenocortical cells, human adrenodoxin mRNA accumulation is induced by cyclic AMP. However, unlike the bovine system, cycloheximide augments, rather than eliminates, this CAMPinduced mRNA accumulation.

MATERIALS AND METHODS
The construction of our human adrenal cDNA library in X g t l O has been described (2). The library was screened as described (6)  (v/v) heat-inactivated fetal bovine serum on plastic dishes (Nunc, Denmark) in 5% COZ, 95% air with changes of medium every 48 h. Treatments with 1.5 m M 8bromo-CAMP (Sigma) and/or 20 pg/ml cycloheximide (Sigma) were done as described under "Results." Cellular RNA was extracted in 4 M guanidine thiocyanate, electrophoresed, blotted to nylon membranes, and probed, as described (9). Fetal tissues were obtained under approved protocols from deliveries by elective therapeutic cervical dilatation and evacuation. Gestational ages were estimated by fetal foot length, regardless of gestational history (10). Fetal and adult tissue RNAs were prepared, blotted, and probed as described (11). Probes used were phAdx-6 (described below), hSCC-71 (3), and pfos-1 (12). Several clones were identified by probing with the 5'-half of the bovine cDNA lying on a 520-base pair PstI/XbaI fragment. The three clones giving the strongest hybridization signal, designated hAdx-2, -6, and -7, were plaque-purified, and their inserts were subcloned in pUC-19. The cloned cDNAs in hAdx-2, -6, and -7 were approximately 1200, 900, and 1250 base pairs, respectively, all longer than the full-length 800base pair bovine cDNA (5). Although these clones had different restriction endonuclease mapping patterns, all possessed an XbaI site, as did the bovine cDNA, therefore, we sequenced all three clones. The relationship of the three clones and their sequencing strategies are shown in Fig. 1. All three clones encode full-length mature adrenodoxin, although hAdx-2 lacks the region encoding most of the prepeptide. The varying lengths of the three clones are due to the use of three alternate polyadenylation signals found 20-26 bases upstream from poly(A) stretches in each clone. Only one of these polyadenylation signals, that used in clone hAdx-7, has the AATAAA consensus sequence (13); the apparent polyadenylation signals used in hAdx-2 is ATAAA, while that used in hAdx-6 is ATTAAA (Fig. 2). A fourth potential polyadenylation signal, ATTAAA, found 66 bases downstream from the translational stop codon, is used in bovine adrenodoxin mRNA (5). However, no clones using this site were found, and RNA transfer blots cannot determine the presence or absence of human adrenodoxin mRNA molecules terminating at this site.
Tissue Distribution and Size of Adrenodoxin mRNA-Human adrenal, testis, and placenta all synthesize steroid hormones and contain P45Oscc mRNA (11,14); hence, these tissues are also expected to be rich sources of adrenodoxin mRNA. An RNA gel transfer blot (Fig. 3) shows that adrenodoxin mRNA is most abundant in the testis and adrenal but that appreciable quantities are found in the placenta. Furthermore, some adrenodoxin mRNA is also detected in each of the other six tissues examined. While some steroidogenic activity has been ascribed to these tissues, especially in the fetus (15), no P45Oscc mRNA was detected in similar RNA gel transfer experiments (11). Thus, it is unlikely that the encoded adrenodoxin in these nonsteroidogenic tissues participates in electron transfer to P45Oscc. Our hAdx-6 cDNA may be cross-hybridizing with mRNA for a different but structurally related iron-sulfur protein. Alternatively, it is possible that the same adrenodoxin iron-sulfur protein mediates electron transport to other nonsteroidogenic mitochondrial cytochrome P450 enzymes in these other tissues. Since the RNA transfer blot was probed and washed under highly stringent conditions, it is most likely that all these tissues contain the same adrenodoxin mRNA.
All tissues contained the same pattern of adrenodoxin mRNAs ranging from 1.0 to 1.7 kilobases. These three species of RNA are poorly distinguished in the blot in Fig. 3, since large amounts of RNA were loaded, and some degradation of RNA often occurs in fetal tissues before they are frozen. However, bands of these three sizes are seen in RNA from JEG-3 cells (Figs. 4 and 5) and also from primary cultures of human granulosa cells (9), and from primary cultures of human fetal adrenal or testicular cells (16). An additional band of about 3.8 kilobases is also seen in most blots. The nature of this RNA is unknown. It may represent an unprocessed nuclear precursor or another species of adrenodoxin mRNA having a very long 3"untranslated region. However, examination of Southern blots and multiple genomic clones indicates there is only one functional adrenodoxin gene;' hence, this RNA does not arise from another related gene.
Regulation of Expression of Adrenodoxin-The principal hormonal stimulators of steroid hormone synthesis are ACTH in the adrenal and the gonadotropins, luteinizing hormone and follicle-stimulating hormone, in the gonad. All work through cell-surface receptors to stimulate intracellular cyclic AMP, which in turn stimulates accumulation of mRNA for P45Oscc (15, 17), mediated principally by increased gene transcription (18). Cyclic AMP also stimulates steroidogenesis and accumulation of P45Oscc mRNA in the placenta (3,15,19). We examined the hormonal regulation of adrenodoxin and P45Oscc mRNAs in the transformed cytotrophoblast tumor cell line JEG-3.
JEG-3 cells accumulate mRNA for both P45Oscc and adrenodoxin for up to 48 h while in the presence of 1.5 mM 8-Br-CAMP. By contrast, mRNA encoded by the fos oncogene accumulates very rapidly, reaching its maximal value by 30 min and then diminishing steadily thereafter (Fig. 4), a time course similar to that seen in activated fibroblasts (20); however, densitometric scanning of the data in Fig. 4 shows that 6 h after stimulation the amount of fos mRNA is 5-fold greater than control. Incubation of JEG-3 cells with 20 pg/ml cycloheximide for 30 min before adding 8-Br-CAMP had varying effects on the mRNA for P45Oscc, adrenodoxin, and fos meas- ured 6 h later (Fig. 5). Cycloheximide quickly reduced basal P45Oscc mRNA to undetected amounts and reduced the CAMP-stimulated P45Oscc mRNA to much less than baseline amounts. By contrast, both cycloheximide and 8-Br-cAMP had stimulatory effects on adrenodoxin mRNA; when administered together these effects were additive as shown by laser densitometric quantitation of the data in Fig. 5. Both cycloheximide and 8-Br-CAMP also stimulated accumulation of fos mRNA, though to a greater extent than their stimulation of adrenodoxin mRNA. When administered together, laser densitometry shows that the stimulatory effects of cycloheximide and 8-Br-CAMP on fos are multiplicative. Thus, cycloheximide and 8-Br-CAMP appear to exert at least three different classes of effects on JEG-3 cell mRNA. These effects of cycloheximide on P45Oscc and adrenodoxin mRNAs could be duplicated by incubating the cells with 200 PM puromycin for 30 min before adding the 8-Br-CAMP. By contrast, adding 2 pg/ml actinomycin D 30 min before stimulating with 8-Br-cAMP blocked the increases in P45Oscc and adrenodoxin (not shown). Thus, two drugs inhibiting protein synthesis by different mechanisms stimulate accumulation of adrenodoxin mRNA, while inhibiting the accumulation of P45Oscc mRNA.

DISCUSSION
The sequence of mature human adrenodoxin is highly homologous to the bovine protein (105 of 114 amino acids, A. P45Oscc 92.1%), while the processed amino-and carboxyl-terminal extensions of the preprotein are only 50% homologous (29 of 58 residues) and 42.8% homologous (6 of 14 residues), respectively. Thus, aside from the need to preserve residues dictating transport to mitochondria, there appears to be little evolutionary pressure to maintain the sequence of amino-terminal peptide. Some investigators have described molecular evolution in terms of the so-called "unit evolutionary period", defined as the length of time in millions of years required for a 1% amino acid sequence difference to arise in two related peptides (21). Assuming the ancestors to cattle and human beings diverged 85 million years ago, the unit evolutionary period for mature adrenodoxin would be 9.3, while that for the amino-and carboxyl-terminal extensions would be 0.6. The existence of proteins with clearly defined domains of conserved and nonconserved regions severely limits the usefulness of the unit evolutionary period concept, as discussed earlier (22).
Steroid hormone synthesis is largely confined to the adrenals, gonads, and placenta, although some steroidogenic activity has been described in virtually all tissues in the human fetus (23). However, much of this "ectopic" steroidogenesis appears to be due to enzymes other than those functioning in the adrenals and gonads, as the mRNAs for P45Oscc, P450c17, and P450c21 could not be detected in transfer blots of RNA from human fetal ovary, kidney, muscle, liver, intestine, or spleen (11). Therefore, detecting adrenodoxin mRNA in these tissues was unexpected. As these tissues lacked detectable P45Oscc mRNA, it is likely that the adrenodoxin in these tissues is functioning as an electron transport intermediate for other mitochondrial cytochromes P450. Thus, the term "adrenodoxin" may be inappropriately parochial for this widely distributed protein.
Cycloheximide-mediated accumulation of fos mRNA has been described by several groups (24-26). However, the stimulatory effect of cycloheximide on adrenodoxin mRNA in JEG-3 cells stands in sharp contrast to earlier studies of adrenodoxin in primary cultures of bovine adrenocortical cells. In the bovine adrenal system, the mRNAs for P45Oscc and adrenodoxin (as well as for the microsomal steroidogenic enzymes P450c17 and P450c21) always respond in parallel: ACTH and cAMP stimulate these mRNAs, and cycloheximide ablates that stimulation (5, 18). These observations suggested the presence of a rapidly turning over cycloheximide-sensitive "steroid hormone-inducing protein" (18). While a 30-amino acid "steroidogenesis activator polypeptide" has recently been isolated (27,28), the relationship of steroidogenesis activator polypeptide to the hypothetical steroid hormone-inducing protein, if any, is unknown. However, the parallel stimulation of P45Oscc and adrenodoxin mRNAs seen in bovine adrenal cells is clearly not a generalized phenomenon among steroidogenic tissues. Normal human cytotrophoblasts as well as transformed trophoblastic cells respond to cAMP with accumulation of adrenodoxin mRNA.3 Similarly, cycloheximide does not inhibit the accumulation of P45Oscc and adrenodoxin mRNAs in primary cultures of human ovarian granulosa cells stimulated with 8-Br-CAMP (9). The responses of the JEG-3 cells employed in our present study differ. Unlike the human granulosa cells but like the bovine adrenocortical cells, cycloheximide inhibits the accumulation of P45Oscc mRNA in response to CAMP. Unlike the bovine adrenal cells, cycloheximide stimulates accumulation of adrenodoxin mRNA and promotes a further additive increase in adrenodoxin mRNA in cells stimulated with 8-Br-CAMP. Thus, the differences among the bovine adrenocortical system and human systems we have studied appear to involve multiple factors including species, cell type, cell transformation, and hormonal pretreatment. One possible mechanism for cycloheximide's stimulatory effect on adrenodoxin mRNA might involve a rapidly turning over cycloheximide-sensitive nuclease specific for AU-rich regions. AU-rich sequences increase mRNA turnover when incorporated into the 3'-untranslated regions of otherwise stable mRNAs in transformed cells, and cycloheximide treatment of such cells increases accumulation of these modified mRNAs (29). The 3"untranslated region of adrenodoxin mRNA is 72% AU. In cultured human granulosa cells, cycloheximide similarly stimulates accumulation of mRNA for the low density lipoprotein receptor, which also has an AU-rich 3"untranslated region (30). While these data do not rule out the existence of rapidly turning over cycloheximide-sensitive steroid hormone-inducing proteins, they indicate that cellular strategies for the control of steroidogenesis are varied and complex and are unlikely to fit into a single common overall scheme.