Hormonal Regulation of Vitellogenin Genes: An Estrogen-Responsive Element in the Xenopus A2 Gene and a Multihormonal Regulatory Region in the Chicken II Gene

Emily P. Slater, Gerard Redeuilh, and Miguel Beato Institut fur Molekularbiologie und Tumorforschung Philipps Universitat 3550 Marburg, Germany Laboratoire des Hormones Institut National de la Sante et de la Recherche Medicale U33 Faculte de Medicine de Bicetre 94270 Bicetre, France

used to mediate estrogen induction of related genes in chickens and amphibians. (Molecular Endocrinology 5: 386-396, 1991)

Expression of the vitellogenin genes in avian and amphibian liver is regulated by estrogens. The DNA elements mediating estrogen induction of the various vitellogenin genes of chicken and Xenopus encompass one or more copies of a 13-mer palindromic sequence called the estrogen-responsive element (ERE). Here we show that upon incubation with the purified estrogen receptor (ER) from calf uterus the Xenopus vitellogenin A2 gene yields a DNase-l footprint over the ERE between -331 and -319. This element does not mediate the response to glucocorticoids or progestins in T47D cells. The three guanine residues in each half of the palindrome are protected against methylation by dimethylsulfate after incubation with ER, but not with glucocorticoid (GR) or progesterone (PR) receptors. In contrast, the chicken vitellogenin II gene exhibits multihormonal regulation by estrogens, progestins, and glucocorticoids in T47D and MCF7 cells. Regulation is mediated by the DNA region between -721 and - 5 9 1 that contains four binding sites for hormone receptors, as demonstrated by DNase-l footprints and methylation protection experiments. The two distal and most proximal binding sites are recognized by ER, GR, and PR, whereas the central binding site is only bound by ER and GR. At suboptimal concentrations, estrogens and progestins or glucocorticoids act synergistically. In experiments using a DNA fragment containing an ERE adjacent to a glucocorticoid-responsive element/progesterone-responsive element, ER and PR bind synergistically to their corresponding sites, perhaps explaining the functional synergism of both hormones. Thus, two very different regulatory elements are

INTRODUCTION

Regulation of the hepatic expression of egg yolk protein genes in avian and amphibian liver is a classical model system for the study of estrogen action. The DNA elements mediating induction by estrogens have been identified in several vitellogenin genes and have served to establish the consensus sequence for an estrogenresponsive element (ERE) (1 -5). The central core of the ERE is a 13-basepair (bp) palindrome, GGTCANNNTGACC, that shares a few essential nucleotides with the glucocorticoid-responsive element (GRE) (see Ref. 6 and references therein). As the GRE has been shown to mediate induction by progestins, androgens, and mineralocorticoids (7-10), the question arose of whether the EREs of the vitellogenin genes were also able to mediate induction of the adjacent promoter by other steroid hormones. Using natural or synthetic oligonucleotides, no evidence has been produced that the ERE palindrome can interact with steroid hormone receptors other than the estrogen receptor (ER). However, a few single mutations within the ERE are sufficient to convert it to a GRE (11), suggesting that there are only minor differences in the DNA recognition mechanisms of various hormone receptors. This idea is supported by the similarities between the amino acid sequence of the DNA-binding domains of various hormone receptors (Ref. 12 and references therein). The ERE of the chicken vitellogenin II gene at -600 is located adjacent to a potential GRE able to bind the glucocorticoid receptor (GR) in vitro (13). It has been

0888-8809/91 /0386-0396S03.00/0 Molecular Endocrinology Copyright© 1991 by The Endocrine Society

386

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 18:46 For personal use only. No other uses without permission. . All rights reserved.

Hormonal Regulation of Vitellogenin Genes

shown that a DNA fragment including the ERE and the GRE can mediate estrogen as well as glucocorticoid or progesterone induction of an adjacent heterologous promoter (14, 15). On the contrary, no evidence for multiple hormonal regulation of the Xenopus vitellogenin A2 gene has been reported, in agreement with the observed hormonal specificity of the hepatic vitellogenin gene expression in both birds and amphibia. Binding of the partially purified liver ER to the region containing the ERE of the chicken vitellogenin II gene has been reported (16). However, the limits of the DNase-l footprint were not precise enough to distinguish between binding to the ERE and that to the adjacent GRE. A recombinant ER has been shown to generate a footprint over the ERE region of the Xenopus vitellogenin A2 gene (17), but no attempts have been made to study the ability of these EREs to interact with other steroid hormone receptors. Here we show that the hormonal regulatory region of the chicken vitellogenin II gene is more complex than originally described and comprises four binding sites with different affinities for ER, progesterone receptor (PR), and GR, whereas the ERE of the Xenopus vitellogenin A2 gene contains a single ERE with low affinity for GR and PR. In addition, we show that the functional synergism observed between estrogens and progestins upon induction of the chicken vitellogenin II gene may at least partially be due to synergistic binding of the corresponding receptors to adjacent sites within the hormone-responsive element (HRE).

RESULTS Functional Studies with the Regulatory Region of Vitellogenin Genes To analyze the potential of the hormonal regulatory regions of chicken and frog vitellogenin genes to mediate regulation by steroid hormones in addition to estrogen, we used constructs containing these regions fused to the thymidine kinase (tk) promoter from herpes simplex virus and the chloramphenicol acetyl transferase (CAT) gene of E. coli. The constructs used for these studies are shown in Fig. 1A. The construct ChVittkCAT contains a fragment of the chick vitellogenin II gene from -721 to - 5 9 1 , whereas the construct XVittkCAT contains the region between -331 and -297 of the Xenopus vitellogenin A2 gene. Gene transfer experiments were performed in the human mammary carcinoma cell lines MCF-7 and T47D, both of which contain receptors for estrogen, glucocorticoids, and progestins. The results of a representative series of experiments are shown in Fig. 1, B-E. It is clear that, independent of the cell line, the ChVit-tkCAT construct responds to all three steroid hormones tested, whereas only estrogens were able to induce CAT activity after transfection of the XVit-tkCAT construct. These results were reproducibly observed in three different experiments. Analysis of the transcrip-

387

tion start point by RNase mapping (18) demonstrated that under all experimental conditions the correct transcripts were detected, and therefore, the hormonal effects were exerted at the relevant promoter (Fig. 2). Densitometric scanning of the bands corresponding to the transcripts originating from the tk promoter compared to those from the RSV promoter indicated that each hormone was able to induce transcription from the tk promoter. The data in Fig. 2 result from a 10-h induction, whereas in Fig. 1, B-E, the cells were treated for 48 h. The maximal response to the ChVit-tkCAT construct in MCF-7 cells was observed with dexamethasone (DEX; Fig. 1C), whereas in T47D cells all three hormones resulted in a similar response (Fig. 1B). When two hormones were added together to MCF-7 cells transfected with the chicken construct, a nearly additive effect was observed (Fig. 1C). In T47D cells, in which the response to the individual hormones was already very high, addition of two hormones also resulted in an additive effect (Fig. 1B). After transfection of these cells with the XVit-tkCAT construct, neither glucocorticoids nor progestins in addition to estrogen could influence the estrogen response. Thus, depending on the cell line and the source of the regulatory regions, different responses to the various hormones are found. Dose Dependence of the Biological Response Induction of XVit-tkCAT by estrogens has been shown to be dependent on concentrations of estrogen sufficient to saturate the ER (2). To verify that the observed effects of the other hormones in gene transfer experiments were mediated by the corresponding hormone receptors, we analyzed the dose dependence of the hormonal induction (Fig. 3). Half-saturation of the estrogen response of the ChVit-tkCAT construct in T47D cells was observed at 20 pM diethylstilbestrol (DES). With this construct, half-maximal induction by progestins was obtained with 0.1 nM R5020, suggesting a participation of the PR in this response. A half-maximal response to glucocorticoids was detected with 10 nM DEX, as expected from the affinity of the GR for this synthetic steroid. Thus, the individual hormones are active at concentrations compatible with their affinity to the relevant physiological receptors. At suboptimal concentrations of estrogen, the effect of added glucocorticoids or progestins was more than additive, and the corresponding dose-response curve was shifted to the left (data not shown). Therefore, lower concentrations of glucocorticoids or progesterone were needed for induction in the presence of estrogens, suggesting functional synergism between ER and either GR or PR. DNase-l Footprints and Methylation Protection Experiments with the Different Hormone Receptors To reach a better understanding of the biological findings summarized above, we studied the interaction of

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 18:46 For personal use only. No other uses without permission. . All rights reserved.

MOL ENDO-1991 388

Vol 5 No. 3

-721

-591

CAT

ChVH-tkCAT

CAT

XVIt-tkCAT

-331 -297

XVit-tkCAT

ChVit-tkCAT B

D

2000

2000

T47D

1000-

• control E3 DES • R5020 E3 DEX D DES • R5020 • DES • DEX II R5020 + DEX

MCF7

Fig. 1. Schematic Representation of the Chimeric Constructs A, HRE TK CAT constructs tested in transient transfection assays. ChVit-tkCAT contains sequences from the chicken viteliogenin II gene from -721 to - 5 9 1 . XVit-tkCAT contains Xenopus viteliogenin sequences from -331 to -297. Both regions were fused to the tk promoter of herpes simplex virus and the CAT gene of E. coli (7). B-E, Influence of steroid hormones on CAT activity after transient transfection of the constructs presented in A. Two human mammary carcinoma cell lines, T47D and MCF7, were transfected with each construct, grown for 48 h without hormone (control) or with 10~8 M DES, 10~8 M R5020, 10~7 M DEX, or combinations of two of the hormones, as indicated in the key on the right of the figure. The cells were harvested, and the cytosol extracts tested for CAT activity, which is expressed in picomoles or femtomoles per min/mg protein. The results with ChVit-tkCAT are presented in B and C, and those with XVit-tkCAT in D and E. B and D are the results obtained from T47D cells, and C and E those from MCF7 cells. Values are the average of three experiments. Error bars indicate the SD of the individual results.

the partially purified receptors for estrogens, glucocor-

strands (Fig. 4, A and B, lanes 4, 8, and 12). However,

ticoids, and progestins with the regulatory regions of

the pattern of protection against methylation by dime-

the chicken and Xenopus viteliogenin constructs.

thylsulfate at the N-7 positions of guanine residues was

DNase-l footprint and methylation protection experi-

different for the three hormone receptors. In the sense

ments with the Xenopus gene are depicted in Fig. 4.

strand, the ER protected the Gs at positions - 3 3 2 ,

Surprisingly, not only the ER, but also the GR and PR,

- 3 3 1 , -330, and - 3 2 2 , whereas neither the GR nor

generate DNase-l footprints over the region between

the PR produced a clear protection of these G residues

-331 and - 2 9 7 in both the sense and antisense

(Fig. 4A, compare lanes 6, 10, and 14). The GR pro-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 18:46 For personal use only. No other uses without permission. . All rights reserved.

Hormonal Regulation of Vitellogenin Genes

ChVit C

E

G

389

XVit P

C

E

210



150

— RSVCAT

1

3

4

5

tkCAT

6

Fig. 2. Analysis of CAT RNA after Transfection All three hormones, estrogen (E), glucocorticoid (G), and progestin (P), increase the concentration of tk CAT mRNA after transfection of T47D cells with plasmid ChVit-tkCAT. Lanes 1-4, RNA extracted from cells transfected with plasmid ChVit-tkCAT and incubated without hormones (1 ,C) or with DES (2,E), DEX (3,G), and R5020 (4,P) for 10 h. Lanes 5 and 6, RNA extracted from cells transfected with plasmid XVittkCAT and incubated without (5,C) or with DES (6,E) for 10 h.

-12

-io -a log concentration of hormone

Fig. 3. Dose-Response Curve of Transfected Plasmid ChVittkCAT Plotted are the CAT activities obtained after incubation with increasing concentrations of R5020 (•), DEX (•), and DES

The contact between the ER and the relevant guanine residues in the major groove of the double helix appears to be more intimate than that of the GR or PR. The results of similar experiments with the chicken vitellogenin II gene are shown in Fig. 5. Here a complex pattern of DNase-l protection was observed. With each of the hormone receptors tested there are several regions protected against DNase-l digestion. Within these regions there are G residues that are protected by the receptors from methylation, thus confirming the DNase-l footprinting data. The ER protects four regions: between -710 and -694, -670 and -655, -628 and -610, and -608 and -589 (Fig. 5, A and B). Within these regions the ER protected the Gs at positions -699, -665, - 6 6 1 , -646, -626, -625, -617, -603, and -594 on the sense strand and at positions -707, -705, -667, -623, -614, -600, and -590 on the antisense strand. The PR yields a similar footprint, although the extent of protection is not the same as with the ER. The region -628/-613 was previously described to be a binding site for the ER (16), and the region -608/-589 was found to bind the GR (13). Here we find that the PR binds to the GR region -608/-589 and, in addition, to the regions -710/-694 and -670/ -655 also bound by the ER, but the PR does not bind to the -628/-613 ER-binding region, as it is unable to protect the guanine residues in this region from methylation. Again, G residues within these other regions, at positions -699, -665, - 6 6 1 , -646, -603, and -594 on the sense strand and -707, -705, -667, -600, and -590 on the antisense strand, are protected by the PR against methylation (Fig. 5, A, C, and D). The GR protects these three regions and, in addition, the region between -628 and -613, similar to the ER. The results of methylation protection experiments are summarized in Fig. 5D. All three receptors contact G residues in the regions -710/-694, -670/-655, and -608/-589, whereas only ER and GR contact guanines in the region -628/-613. The guanine at position -615 is hypermethylated after incubation with GR (Fig. 5C) and protected after incubation with ER (Fig. 5B). A similar situation occurs at position -625, which is protected by ER and hypermethylated after incubation with GR (Fig. 5A).

(in-

Binding Synergism between ER and PR with an Oligonucleotide Containing ERE and GRE/ progesterone-responsive element (PRE)

hypermethylation at positions -331 and -330, and binding of the PR led to hypermethylation of position -330. A similar situation was found in the lower or antisense strand. Here the ER protected the Gs at -328, -320, and -319, whereas the GR and PR only weakly protected position -328 and led to hypermethylation of position -320 (Fig. 4B, compare lanes 6, 10, and 14). These results (summarized in Fig. 4C) suggest that although the three hormone receptors bind to the same region of the Xenopus vitellogenin A2 gene, they interact with the DNA sequence with different affinities.

The proximity of binding sites for the different receptors suggests the existence of functional synergism (14). To explore this possibility in more detail we constructed an oligonucleotide containing the combination of ERE and GRE found in chicken vitellogenin II (Fig. 6A). When this oligonucleotide was placed in front of the tkCAT reporter plasmid and transfected into T47D cells, it exhibited the expected behavior. Whereas an ERE oligo responded only to estrogens, the oligo containing ERE and GRE/PRE responded to estrogen, progestins, and glucocorticoids, but only when GR cDNA (19) was

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 18 November 2015. at 18:46 For personal use only. No other uses without permission. . All rights reserved.

MOL ENDO-1991 390

Vol 5 No. 3

B GR A

g

-

+

-

PR •

-

G DNose DMS

ER



DNose

-



-

PR

GR •

DMS DNose

-

ER



A

DMS

6

G DNose

DMS

DNase DMS DNose DMS

-33 1—

-297—

Hormonal regulation of vitellogenin genes: an estrogen-responsive element in the Xenopus A2 gene and a multihormonal regulatory region in the chicken II gene.

Expression of the vitellogenin genes in avian and amphibian liver is regulated by estrogens. The DNA elements mediating estrogen induction of the vari...
2MB Sizes 0 Downloads 0 Views