Determinants of Vitellogenin B1 promoter Architecture: HNF3 and Estrogen Responsive Transcription within Chromatin

The liver-specific vitellogenin B1 promoter is efficiently activated by estrogen within a nucleosomal environment after microinjection into Xenopus laevis oocytes, consistent with the hypothesis that significant nucleosome remodeling over this promoter is not a prerequisite for the activation by the estrogen receptor (ER a ). This observation lead us to investigate determinants other than ER a of chromatin structure and transcriptional activation of the vitellogenin B1 promoter in this system and in vitro . We find that the liver-enriched transcription factor HNF3 has an important organizational role for chromatin structure as demonstrated by DNase I hypersensitive site mapping. Both HNF3 and the estrogen receptor activate transcription synergistically and are able to interact with chromatin reconstituted in vitro with three positioned nucleosomes. We propose that HNF3 is the cellular determinant which establishes a promoter environment favorable to a rapid transcriptional activation by the estrogen receptor.


INTRODUCTION
Vitellogenin is the precursor of the yolk proteins that represent a major source of nutrients for the developing embryo in oviparous vertebrates (1). Its liver-specific expression is controlled by estrogen in adult females (2) through its receptor which interacts as a homodimer with the estrogen responsive element (ERE) in the promoter of the vitellogenin genes (3,4). Functional studies of the Xenopus vitellogenin B1 promoter in cultured human breast cancer cell lines have revealed the presence of an estrogen responsive unit (ERU: region -334 to -302 relative to the start site of transcription) composed of two imperfect EREs (5). A third functional ERE was detected farther upstream in the region -555 to -543. The Xenopus vitellogenin B1 promoter (6) encompasses additional cis-acting elements for the liver-enriched factors HNF3 and C/EBP as well as for the ubiquitous protein CTF/NF1 (7). Tissue specificity may be achieved by a combination of these various transcription factors together with a specialized chromatin structure.
Chromatin is intimately related to the expression of eukaryotic genes in vitro and in vivo (8).
Nucleosome assembly over a promoter is implicated in establishing the repression of many inducible genes (9)(10)(11). Although nucleosomes were often reported to globally repress transcription, the precise position of a nucleosome on a promoter can also improve transcriptional activation (12)(13)(14)(15). Such a nucleosome, positioned in vitro on the promoter of the vitellogenin B1 gene in a region devoid of transcription factor binding sites (between -300 and -140 relative to the start site of transcription), potentiates estrogen receptor dependent transcriptional activation.
We have suggested that the wrapping of DNA around the histone octamer allows a better interaction between the distal estrogen receptor and proximal trans-activators (such as HNF3 and CTF/NF1) and/or the basal transcription machinery (14). Interestingly, HNF3 and NF1 are both able to modulate the position of a nucleosome (16)(17)(18) and HNF3 can replace linker histone on the serum albumin gene enhancer (19). Since HNF3 and NF1 bind to the vitellogenin B1 promoter, they may have important roles in the assembly of this regulatory DNA into organized chromatin structure in vivo.
In this study we have analyzed the interplay between the estrogen receptor and HNF3 on the vitellogenin B1 promoter within chromatin following microinjection into oocytes. We show that HNF3 induces DNase I hypersensitivity within the proximal regulatory elements concomitant to the activation of transcription. In addition, when expressed before chromatin assembly, HNF3 and the estrogen receptor activate transcription synergistically suggesting either that HNF3 establishes a chromatin structure favorable for subsequent association of the estrogen receptor binding or that they interact directly or through a common target. We have determined the positions of three nucleosomes over the vitellogenin B1 promoter in vitro. These nucleosomes do not interfere with ER and HNF3 recruitment and are not displaced.

Plasmid constructs
The pBS(-596/+8)VgB1 plasmid was obtained by subcloning the BamHI/RsaI fragment of upstream of the TATA box to generate the construct pB1(6ERE). Briefly, 2 ERU oligos designed from the wild type promoter (position -337 to -300 ), which contains two imperfect EREs, were flanked by AvaI sites (top strand: 5'TCG GGT CCA GTC ACT GTG ACC CAA CCC AAG   TTA TCG TGA CCT C3'; bottom strand : 5'CCC GAG AGG TCA TGA TAA CTT GGG TTG GGT CAC AGT GAC TGG A3'). The annealed oligonucleotides were multimerized and then introduced into the BglII site (position -41) of pBS(-596/+8)VgB1. This generated one construct with 1 ERU and one ERE. Since the wild type promoter contains already 3 EREs, the mutated promoter was named pB1(6ERE). The single stranded DNAs were prepared from phagemids induced with the VCS M13 helper phage (21).
The plasmids used for in vitro mRNA transcription were prepared as follows. The xERα coding sequence was amplified from pKCR2xERα (22) by PCR using oligonucleotides generating XbaI and PstI sites. The PCR fragment was then introduced into the same sites of the plasmid pSP64(polyA) (Promega) to generate the xERpSP64(polyA) construct. rHNF3αpMS(II) plasmid was obtained after introduction of the SacI/HindIII fragment (from pRB-HNF3α, gift from K.S. Zaret) into the same sites of pMS(II) which is a modified version of pSP64(polyA) containing additional unique restriction sites after the poly(A) segment (Melissa Stolow, unpublished).

In vitro mRNA synthesis and microinjection of Xenopus oocytes
Constructs for in vitro mRNA synthesis were linearized either with EcoRI (xERpSP64(polyA)) or with AvrII (rHNF3αpMS(II)). The mRNA in vitro synthesis was carried out with the SP6 mMessage mMachine Kit (Ambion) according to the recommended procedure. The isolation procedure of Xenopus laevis oocytes was performed as previously described (23). Stage VI oocytes were microinjected (13.8 nl)(Drummond nanoject) with the indicated amount of in vitro transcribed mRNA. After an overnight incubation in MBSH buffer (23) at 18°C, the oocytes nuclei (GV) were microinjected with the pBS(-596/+8)VgB1 as ssDNA (1ng). The oocytes were processed as described below generally after 6 hours incubation at 18°C.
Transcription assay and DNA recovery A group of 15 to 20 oocytes was collected per sample and homogenized in 10 µl per oocyte of 0.25M Tris-HCl at pH 8.0. The equivalent of 10 oocytes was used for RNA purification and transcription analysis and the remaining 5 oocytes equivalent was used for DNA recovery. Total RNA was prepared from the homogenate using the RNazol" method (Tel-Test Inc.). The RNazol" reagent (500µl) together with chloroform (50µl) was added to the homogenate (100µl) which was then incubated on ice for 15 minutes before centrifugation. The clear supernatant was then transferred to a fresh tube and precipitated with 1 volume of isopropanol. The washed RNA pellet (70% EtOH) was resuspended in 100µl of TE buffer and precipitated once again for at least one hour at 4°C with 30µl of 10M LiCl. The RNA pellet was then washed with EtOH 70% and dissolved in 20µl of DEPC-treated water. Half of the RNA sample (5 oocytes) was analyzed by primer extension.
Annealing occurred for 10 minutes at 65°C, 30 minutes at 55°C, 20 minutes at 37°C and 5 minutes at room temperature. The reverse transcription reaction was carried out at 42°C upon addition of 1µl 10 mM dNTPs and 100 Units of the Superscript"II RT (Gibco-BRL). The reaction was stopped after one hour by ethanol precipitation. The extension products were resolved on a 8% sequencing gel and visualized by autoradiography.
DNA was recovered from the remaining 5 oocytes equivalents of the homogenate (see above).
DNA recovery after injection is a control monitoring the microinjection efficiency. This DNA was also used for the supercoiling assay (see below). One volume of 1% SDS and 20mM EDTA pH 8.0 was added to the sample which was then treated for 2 to 3 hours with Proteinase K (200µg/ml) at 55°C. DNA was then extracted twice with phenol/chloroform and precipitated at room temperature with isopropanol. The washed DNA pellet (EtOH 70%) was then dissolved in 100µl of TE buffer, treated with RNase A (100µg/ml) for 1 hour at 37°C, extracted with phenol/chloroform and finally isopropanol precipitated. The recovered DNA was quantitatively analyzed by Southern Blot as described (21). Blots were then probed with random-primed labeled ssDNA from the vitellogenin B1 promoter.

DNase I hypersensitive site analysis
After injection, groups of 20 to 25 oocytes were collected and homogenized in 300 µl of digestion buffer (20 mM Tris-HCl pH 8.0, 50mM NaCl, 2mM MgCl 2 , 1mM EDTA, 1mM DTT and 5% glycerol). The homogenate was divided into three 100 µl fractions and digested for 1 minute (room temperature) with respectively 160 U, 120 U and 80 U of DNase I (Gibco-BRL). The reactions were stopped with 2 volumes of 1% SDS and 20mM EDTA and treated for 1 hour at 37°C with RNase A (100µg/ml). The samples were then incubated with Proteinase K (200µg/ml) for at least 3 hours. DNA was finally purified as described above (see Transcription assay and DNA recovery). The DNA pellets were resuspended in 50µl of 1x BamHI buffer and digested with 10 U of BamHI for 2 hours (Gibco-BRL). The samples were then once again extracted with phenol/chloroform and isopropanol precipitated. The DNA fragments were resolved on a 1.5 % agarose gel, blotted to a Zeta-probe® GT membrane (BioRad) and probed with a randomprimed labeled promoter fragment (-596/-445). As a control, free dsDNA (2ng per oocyte) was mixed with non injected oocytes and treated immediately with DNase I under the same conditions described above.

Micrococcal nuclease assay
The injected oocytes (20 per sample) were homogenized in 200 µl of MNase digestion buffer (10 mM Hepes pH 8.0, 50mM KCl, 2mM MgCl 2 , 3mM CaCl 2 , 1mM DTT, 0.1 % NP-40 and 8 % glycerol) and split into four tubes (50 µl aliquots). Increasing amounts of Micrococcal nuclease (Worthington) were added into the different aliquots (0.04, 0.2, 1 and 5 U) and digestion was allowed to proceed at room temperature for 20 minutes. The reaction was stopped with 1 volume of 1 % SDS and 20 mM EDTA pH 8.0. DNA was then purified according to the method described above (see Transcription assay and DNA recovery). The fragments following digestion were resolved on a 1.5 % agarose gel in 1x TBE and blotted into a Hybond-N+ membrane (Amersham) as indicated (21). The membrane was then probed with a random-primed labeled probe corresponding to the vitellogenin B1 promoter (region -596/+139).

Supercoiling assay for chromatin disruption
The supercoiling assay was performed according to a method described earlier (24). Generally, the equivalent of DNA recovered from one injected oocyte was loaded onto a 1.1 % agarose gel in 1xTPE buffer (40mM Tris, 30mM NaH 2 PO 4 , 10mM EDTA). Chloroquine (90µg/ml) was included in the gel and in the running buffer. Under these conditions the different topoisomers will be positively supercoiled, reflecting the differences in negative supercoiling depending on the original variations in nucleosomes density on the minichromosomes. Electrophoresis was performed overnight in the dark at 3.5V/cm. DNA was subsequently blotted onto a Hybond-N+ membrane (Amersham) by capillary transfer as described (21). The membrane was then probed with random-primed labeled ssDNA from the vitellogenin B1 promoter.

Nucleosome reconstitution
Nucleosomes were reconstituted on radiolabelled DNA fragments by salt/urea dialysis as described earlier (25). After reconstitution, the oligonucleosome cores were loaded on 10 to 30% sucrose gradients and centrifuged for 24 hours at 38000 rpm at 4°C in a Beckman SW41 rotor.
Chicken erythrocyte oligonucleosomes were prepared after removal of linker histone (26) and used as control for isolation of the native trinucleosome complexes.

Micrococcal nuclease mapping
Reconstituted chromatin (100ng of DNA) was digested with 0.01, 0.02 and 0.04 U of MNase (Worthington) for 5 minutes at 22°C in the presence of 0.5mM CaCl 2 . Digestions were started by concomitant addition of MNase and CaCl 2 and stopped by addition of 5mM EDTA, 0.25% (w/v) SDS and 1mg/ml proteinase K. After phenol/chloroform extraction and precipitation, the recovered DNA was 5' end-labeled with [γ↑ 32 P]ATP and T4 polynucleotide kinase. Endlabeled fragments were separated by electrophoresis in non denaturing 6% polyacrylamide gels.
DNA fragments of nucleosome core and chromatosome products were recovered and digested with restriction enzymes to determine microccoccal nuclease cleavage sites. The digestions were analysed on 8% denaturing gel.

Gel Mobility Retardation Assay
Recombinant HNF3α was purified by affinity chromatography from SF9 cells infected with an HNF3 recombinant baculovirus generated by the BaculoGold kit (Pharmingen). Purified hER was purchased at Panvera. Binding reactions were performed in a 15 µl final volume of 1X binding buffer (10mM Tris-HCl pH 8.0, 5mM MgCl 2 , 50mM NaCl, 150µg/ml BSA, 100µg/ml polydIdC) with 100 fmoles of labeled DNA and 300 fmoles of HNF3α or hER. Protein-DNA complexes were separated on a 0.25X TBE nondenaturing agarose gel.

DNaseI Footprinting
Binding reactions were performed as described above, except that polydIdC was omitted in the

Regulation of the vitellogenin B1 promoter by ERα and HNF3α
The estrogen receptor is only one of several transcription factors that regulate vitellogenin B1 gene expression. The basal promoter element (BPE, region -119/-32) encompasses multiple binding sites for liver-enriched factors like HNF3 and C/EBP and for the ubiquitous CTF/NF1 ( Figure 1A) (7). The contribution of these proteins has been already analyzed in vitro or by transient transfection (20,27). However these assays provide limited insights on how these transcription factors might function within chromatin. We have therefore decided to take advantage of the Xenopus oocyte microinjection system to examine this issue. Microinjection of DNA into the nucleus of Xenopus leads to chromatin assembly (23) which has been shown to influence the transcriptional regulation of the Xenopus HSP70 and TRβ promoters (11,28).
Oocytes also provide an efficient translational machinery which allows the study of the regulation of a given promoter by expressing specific transcription factors (29,30) by injecting mRNA encoding these components either before or after chromatin assembly.
The capacity of the intact ERα to activate transcription was tested after injection of a CAT reporter template fused to the vitellogenin B1 promoter (-596/+8), which contains three estrogen responsive elements (ERE) ( Figure 1A). Since chromatin assembly occurs with more rapid kinetics on microinjected ssDNA during second strand synthesis in comparison to the injection of dsDNA (23), we made use a of ssDNA template. The ssDNA was injected after the expression of ERα, meaning that the receptor was present during chromatin assembly. In contrast to an ealier report (31) we find that the exogenous ERα activates transcription in injected oocytes ( Figure 1B, compare lane 1 to lanes 3 and 4). In this particular experiment the addition of 17-β-estradiol (E2) did not stimulate transcription further ( Figure 1B, lane 4). This is not surprising since mature oocytes already contain enough endogenous estradiol to ensure receptor activity (32,33). Several lines of evidence from previous studies have suggested that NF1 somehow influences the activity of a chromatin template when the transcription factor is present during chromatin assembly. Depletion of an NF1-like activity reduces transcription by 5 fold from the vitellogenin B1 promoter during the assembly of chromatin in Xenopus oocyte extract (38). Moreover, the CTF/NF1 proline rich activation domain was shown to interact with the core histone H3 in yeast and mammalian cells in response to TGFβ (18). Interestingly, one NF1 DNA binding element is located downstream of the positioned nucleosome in the vitellogenin B1 promoter ( Figure 1A). NF1 cannot interact in vitro with nucleosomal DNA over the MMTV promoter regardless of its binding site orientation and position (39). It was therefore important to test NF1 ability to affect transcription when expressed either before or after chromatin assembly. In our microinjection assay however, expression of NF1 had no effect on transcription in the oocyte nucleus (data not shown).
In summary, these results indicate that HNF3 and ERα are both competent to activate transcription from a chromatin template when present during or after chromatin assembly ( Figure   1B and data not shown). The crystal structure of the DNA binding domain of HNF3 has striking similarities with the structured globular domain of the linker histone H5 (40,41). Interestingly HNF3 can direct the assembly of a nucleosome on the serum albumin enhancer where it is able to replace the linker histone H1 in vivo (16,17,19). Its presence during chromatin assembly process might therefore modify the nucleosomal organization on the vitellogenin B1 promoter. This suggests either that the vitellogenin B1 promoter chromatin organization is already preset for activation or that HNF3 may participate in nucleosomes remodeling, directly or indirectly. We next tested the role of HNF3 in the organization of chromatin structure directly. observed with the promoter containing 6 EREs ( Figure 3D, lanes 7 and 8). Surprisingly, the introduction of point mutations in every single EREs of the wild type promoter did not prevent the receptor dependent change of DNA topology ( Figure 3C, lanes 1 to 4). Thus, our MNase digestion and topology assays (Figure 3) suggest that a substantial alteration of the chromatin environment, dependent on the EREs, is not required for the activation of the vitellogenin B1 promoter by ERα.

In vitro nucleosome mapping over the vitellogenin B1 promoter
The lack of dramatic chromatin remodeling dependent on the estrogen receptor is consistent with the receptor binding in a linker region as predicted by the in vitro mapping of a nucleosome downstream of the ERU and functioning most effectively within a nucleosomal environment (14).
In order to gain more insight on the link existing between transcription and nucleosome positioning we decided to map nucleosomes in vitro over the entire promoter area. We also analysed the impact of linker histones on the position of nucleosomes.
We reconstituted nucleosome cores on a 724 bp template containing the vitellogenin B1 promoter (-596/+8) (Figure 4). Reconstituted chromatin was then fractionated on a sucrose gradient ( Figure 4A). Fractions containing trinucleosomes were subsequently pooled and as a control, an aliquot was refractionated on a second sucrose gradient with MNase treated erythrocyte chromatin ( Figures 4B and 4C). Next we made use of these reconstituted templates to map the position of nucleosomes ( The estrogen receptor and HNF3 associate with the vitellogenin B1 promoter within chromatin The nucleosome mapping described above is consistent with HNF3 and ER binding to internucleosomic regions. Therefore, these transcription factors should gain access to the promoter without difficulty. We have tested their DNA binding ability on the reconstituted vitellogenin B1 promoter (-418/+150) ( Figure 6). We show that purifed ER and HNF3 are indeed able to interact with their cognate DNA binding elements within reconstituted chromatin. This interaction is specific since it is competed away by specific oligonucleotides and not by random sequences (Figure 6, lanes 11 to 18). The specific complexes shifted by ER and HNF3 are different between free and reconstituted DNA indicating that nucleosomes are not displaced ( Figure 6, compares lanes 2 to 9 with lanes 11 to 18). We conclude that both proteins are competent for binding onto chromatin.
It is however necessary to perform DNase I footprinting experiments in order to determine whether ER and/or HNF3 interact with nucleosomes or in linker regions (Figure 7). The  (14). Alternatively, the increased DNase I accessibility over a long distance like observed on the vitellogenin B1 promoter, rather than on a discrete area, is consistent with the disruption of a higher order chromatin structure. Several lines of evidence suggest that HNF3 might influence the formation of higher order chromatin structures. First, the winged-helix domain of HNF3 is found in the linker histones H5 and H1 and they all share striking similarities (40,41,46). Second, HNF3 was shown to replace a linker histone on the edge of a positioned nucleosome on the serum albumin enhancer but was not able to compact internucleosomic DNA as does H1 or H5 (19). Although the Xenopus oocytes contain a linker histone variant (cleavage stage linker histone B4) (47), this protein binds to DNA with much reduced affinity compared to H1 (48) and it is likely that HNF3 would replace B4 in chromatin when a specific recognition site is present.
Since HNF3 is a liver-enriched transcription factor and that the vitellogenin B1 gene is expressed exclusively in hepatocytes, it is possible that HNF3 might preset the promoter for a rapid response to estrogen. A previous study indicated that a male liver extract contains all the specific factors involved in the derepression and induction on the vitellogenin B1 promoter, but sufficient ERα (49). More recently we have shown that the vitellogenin B1 proximal promoter is accessible to restriction enzyme in hepatocytes but not in erythrocytes where the gene is silent, indicating that chromatin structure is different in these cells (50). In the same study we could not detect any differences between estrogen induced females and uninduced males suggesting that a tissuespecific factor, not related to gender (possibly HNF3), is responsible for this defined chromatin structure. Furthermore, an in vivo fooprinting study has revealed that the (-81/-70) HNF3 binding site is occupied in male and female hepatocytes but not in erythrocytes (7). As importantly, it was shown that the amount of injected ER mRNA required to activate chromosomal vitellogenin genes in oocytes is reduced when receptor-free liver nuclear extracts are co-injected (33).
The vitellogenin B1 promoter is also regulated by the ubiquitous transcription factor CTF/NF1 which was reported to interact with the core histone H3 and to modulate the nucleosome position (18). However we observed neither an activation of transcription, nor an induction of DNase I hypersensitive site in the oocyte system (data not shown) even though NF1 was previously shown to activate transcription synergistically with ERα (27,51).
HNF3, nucleosome positioning and implications in the activation of the vitellogenin B1 gene.
HNF3 and ERα cooperatively activate transcription when present during chromatin assembly suggesting either that they somehow interact physically or that HNF3 helps the recruitment of ERα on a loosened chromatin environment. HNF3 was also implicated in directing the position of a nucleosome in vitro similar to that seen in vivo on the serum albumin enhancer (16,17).
In this study we show that HNF3 is indeed able to interact with chromatin in linker regions and on apparent nucleosomes without disturbing their positions. Therefore DNase I hypersensitivity ( Figure 2) does not necessary reflect displacement of nucleosomes from DNA (11,52). HNF3, like TR/RXR generates a robust DNase I hypersensitive site without significant change in DNA topology (data no shown) suggesting that nucleosome numbers remain constant. Preferential DNase I cleavage might reflect direct recruitment by HNF3 by protein-protein contacts or local disruption of chromatin folding. We propose that HNF3 has a dual function in the activation of the vitellogenin B1 promoter. In a first step, HNF3 replaces the linker histones leading to a decompaction of the chromatin fiber (19). In a subsequent step, HNF3 may prevent the sliding of nucleosomes over important regulatory elements. The priming of the promoter by HNF3 during liver differentiation (53) will elicit a rapid response to estrogen in females or hormone stimulated males. It is likely that the nucleosome (-85/+60) positioned over the TATA box will be significantly altered upon initiation of transcription but HNF3 is probably not involved directly in this process since it does not displace this nucleosome (Figure 7). One can speculate that additional cellular components would play in important role subsequent to the recruitment of ERα to the promoter such as the SWI/SNF chromatin remodeling complex (54) or an histone acetyltransferase activity.

The estrogen receptor and alterations in chromatin structure
Our experiments indicate that ERα in isolation induces a weak DNase I hypersensitive site when expressed during chromatin assembly (Figure 2). This is consistent with early in vivo experiments showing that Xenopus vitellogenin genes from estrogen-treated male liver cells display only a modest increase in DNase I sensitivity although transcription is efficient (55,56). A more precise study of the chicken vitellogenin (VTG II) gene, which is already marked in liver by internal and 3' end hypersensitive sites, demonstrates the appearance of strong additional hypersensitive sites in the 5' end region of the gene upon estrogen treatment (57,58). It seams that estrogen-related remodeling events are more important in chicken than in Xenopus, although we do not know whether this difference reflects more profound regulation specificities between both species. Importantly, the activation of the chromosomal Xenopus vitellogenin B2 gene in oocytes requires the expression of a huge excess of receptors in comparison to the level which is needed in a liver nucleus (33). However, co-injection of receptor-free liver nuclear extract significantly reduces the amount of receptor needed. This clearly suggests that the estrogen receptor does not easily activate transcription without other hepatic nuclear factors. This might also explain why estrogen induces strong DNase I hypersensitive sites in chicken liver but not in our oocytes experiments.
The estrogen receptor interacts with SWI2/SNF2 suggesting that the SWI/SNF complex participates in chromatin remodeling events (54). However, in the oocyte system, we could not detect a change of topology which depends directly on any of the described vitellogenin B1 promoter EREs (Figure 3) and the nucleosomal array is not dramatically altered by ERα. It is not clear whether the DNase I hypersensitive site stems only from the receptor binding to the promoter or from subsequent subtle chromatin alterations that are not detected in our crude assays. Activation of transcription by nuclear hormone receptors is believed to be mediated by a growing family of coactivators which includes SRC-1 (59), p300/CBP (60)(61)(62), ACTR (63) and P/CAF (64). All of the cofactors possess intrinsic histone acetyltransferase activity (63,(65)(66)(67).
Thus it is probable that part of ER dependent transcriptional activation of the vitellogenin B1 promoter will involve histone acetylation. The recent resolution of the nuclesome core particle has shown that one histone H4 tail makes intensive contacts with the face of an H2A/H2B dimer from an adjacent nucleosome (68). This is extremely relevent in the context of chromatin decompaction through acetylation.
Recently, two rotationally and translationally positioned nucleosomes were mapped in vivo on the human estrogen responsive pS2 promoter (69). Their positions were not affected by the estrogen receptor although the pS2 ERE is localized within the 5' edge of one nucleosome.
According to the in vitro assembly of tri-nucleosomes on the vitellogenin B1 promoter ( Figure 5) the ERU lies in a linker region or at the edge of a nucleosome (-359/-213). Therefore, the receptor should gain access to the promoter without difficulty as we have seen with the gel shift experiments ( Figure 6). Moreover, since the presence of this nucleosome in vitro potentiates activation of transcription by ERα (14), it seems unlikely that the receptor would favor a major chromatin perturbation in this region.
Unlike many promoters, the vitellogenin B1 gene does not require profound chromatin reorganisation at the nucleosomal level. We think however that local higher-order chromatin structure unfolding, probably by HNF3, is a key event in liver-specific activation of the vitellogenin B1 gene. Such a role for HNF3 in the regulation of a nuclear hormone regulated promoter is suggested for the first time.     ER and HNF3 interact with the reconstituted -418/+150 vitellogenin promoter fragment. Free (lanes 1 to 9) or reconstituted (lanes 10 to 18) DNA was incubated with either purified ER (lanes 2 to 5 and 11 to 14) or HNF3 (lanes 6 to 9 and 15 to 18). Binding was performed either in absence (ER, lanes 2 and 11; HNF3, lanes 6 and 15) or in presence of oligonucleotide competitors (ER, lanes 3 to 5 and 12 to 14; HNF3, lanes 7 to 9 and 16 to 18. ere (consensus