Skip to main content
Nucleic Acids Research logoLink to Nucleic Acids Research
. 2000 Jul 15;28(14):2634–2642. doi: 10.1093/nar/28.14.2634

The analysis of chimeric human/rainbow trout estrogen receptors reveals amino acid residues outside of P- and D-boxes important for the transactivation function

Fabrice G Petit 1, Yves Valotaire 1, Farzad Pakdel 1,a
PMCID: PMC102667  PMID: 10908317

Abstract

The amino acid sequence of rainbow trout estrogen receptor (rtER) is highly conserved in the C domain but presents few similarities in the A/B and E domains with human estrogen receptor α (hER) [NR3A1]. A previous study has shown that rtER and hER have differential functional activities in yeast Saccharomyces cerevisiae. To determine the domain(s) responsible for these differences, chimeric human/rainbow trout estrogen receptors were constructed. The A/B, C/D or E/F regions of rtER were replaced by corresponding regions of hER and expressed in yeast cells. Ligand-binding and transcription activation abilities of these hybrid receptors were compared with those of wild-type rtER or hER. Surprisingly, our data revealed that the human C/D domains play an important role in the magnitude of transactivation of ER. Two other chimeric ERs carrying either a C or D domain of hER showed that the C domain was responsible for this effect whereas the D domain did not affect hybrid receptor activities. Moreover, a chimeric hER carrying the C domain of rtER showed maximal transcriptional activity similar to that observed with rtER. Gel shift assays showed that, whereas rtER and hER present a similar binding affinity to an estrogen response element (ERE) element, the rtER C domain is responsible for a weaker DNA binding stability compared to those of hER. In addition, the human C domain allows approxi­mately 2 times faster associ­ation of ER to an ERE. Utilization of reporter genes containing one or three EREs confirms that rtER requires protein–protein interactions for its stabilization on DNA and that the C domain is involved in this stabilization. Moreover, AF-1 may be implicated in this synergistic effect of EREs. Interestingly, although E domains of these two receptors are much less conserved, replacement of this domain in rtER by its human counterpart resulted in higher estradiol sensitivity but no increase in the magnitude of transactivation. Data from the chimeric receptors, rtER(hC) and hER(rtC), demonstrated that rtER AF-1 and AF-2 activation domains activated transcription in the presence of estradiol similar to both AF-1 and AF-2 hER. This implies that these domains, which show poor sequence homology, may interact with similar basal transcription factors.

INTRODUCTION

Estrogen receptors (ERs) belong to the superfamily of steroid/thyroid hormone receptors and are ligand-inducible trans­criptional enhancer factors that regulate transcription upon binding to the hormone-responsive element of the promoters (15).

The presence of ER in the rainbow trout has been described in the male liver cytosol by measuring estradiol (E2) binding parameters (6). Rainbow trout ER cDNA has been cloned from a cDNA library constructed from a female liver in vitello­genesis (7,8). ER contains six functionally distinct domains denoted A–F (9). Between human (10) and rainbow trout ER (hER and rtER, respectively) [NR3A1], the homologies in amino acid sequences are extremely variable according to different domains. The most highly conserved region is the C domain (92% homology) which is characterized by the presence of two zinc fingers responsible for the DNA binding (11). Discrimination of the estrogen responsive element (ERE) involves three amino acid residues located in the first zinc finger and called the P-box (1214). The C domain is also responsible for a constitutive dimerization (15), which involves the formation of the D-box by five amino acid residues located in the second zinc finger (14,16). The N-terminal region (or A/B domain) is the less conserved domain between hER and rtER (20% homology) and contains a transcriptional activation function AF-1 that functions in a hormone-independent manner (4,17,18) and is promoter and cell type-specific (1820). The E domain presents 60% of similar residues between hER and rtER and contains the hormone binding domain (HBD) (21), a hormone-dependent transcriptional activation function AF-2 (18,20) and a hormone-dependent dimerization function (22,23). It has been reported that the full receptor activity may result from the synergistic effect of both AF-1 and AF-2 transactivation domains in most promoter contexts (17,24). Nevertheless, receptor activity can be modulated by antiestrogens such as 4-hydroxytamoxifen (OHT) or ICI 164,384 (ICI). OHT presents a partial agonistic activity depending on the AF-1 transactivation function, the promoter context and the cell type (18,19), whereas ICI is considered as a pure antiestrogen (23,2527).

As described above, rtER and hER share some important structural differences. Therefore it was interesting to see whether these structural differences were also responsible for functional differences. We previously developed a yeast system expressing stable rtER (28). A comparative study with hER showed that these receptors exhibited differential functional activities when tested on different reporter genes in the presence of hormone or antihormones. When tested on a two EREs reporter gene, rtER presented three main functional differences with hER: (i) in the absence of E2, only rtER exhibited a basal transcriptional activity; (ii) the minimal E2 concentration efficient for the transactivation was 10-fold higher for rtER than that required for hER; and (iii) the maximal transcription activity of rtER in presence of 10 nM E2 represented only 26% of the maximal activity of hER (28). Therefore, to define a specific region (or regions) within the receptor responsible for the differential activities, several chimeric human/trout ERs were generated by domain swapping. Our data demonstrated that although the DNA-binding domains of rtER and hER exhibited very high similarity, substitution of this domain in rtER with those of hER enhanced the trans­activation ability of rtER. These results highlight that few differences in the amino acid residues of this domain between hER and rtER may strongly contribute to the DNA–receptor interaction. On the other hand, switching rtER HBD with the same domain from hER increased E2-binding affinity ~3-fold.

MATERIALS AND METHODS

Plasmid constructions

All constructions of human/trout estrogen receptor chimeras were performed using the polymerase chain reaction (PCR) as previously described (29). A portion of hER can be joined to a portion of rtER using ‘inside’ primers (38 bp long) which share the junction region. The two primary PCRs were performed separately to produce, respectively, the portions of hER and rtER that we wanted to join. After mixing the purified primary PCR products, a 3′ extension of the heteroduplex templates was performed to generate a fragment that is the sum of the two overlapping products. Annealing at 52°C was accomplished for the first 10 cycles for the formation of the recombinant heteroduplex molecules, followed by an annealing at 60°C and addition of ‘outside’ primers for the 30 subsequent cycles.

The human/trout ER chimeras were constructed using the PCR procedure described above. The hER A/B region encoding amino acids 1–179 was joined to the rtER C/D/E/F region encoding amino acids 143–575 to generate the rtER(hAB) chimera. The hER C/D region encoding amino acids 180–301 was joined to rtER A/B and E/F regions encoding amino acids 1–142 and 267–575, respectively, to generate the rtER(hCD) chimera. The hER E/F region encoding amino acids 302–595 was joined to the rtER A/B/C/D region encoding amino acids 1–266 to generate the rtER(hEF) chimera. The hER C region encoding amino acids 180–262 was joined to rtER A/B and D/E/F regions encoding amino acids 1–142 and 225–575, respectively, to generate the rtER(hC) chimera. The hER D region encoding amino acids 263–301 was joined to rtER A/B/C and E/F regions encoding amino acids 1–224 and 267–575 to generate the rtER(hD) chimera. The rtER C region encoding amino acids 143–224 was joined to hER A/B and D/E/F regions encoding amino acids 1–179 and 263–595, respectively, to generate the hER(rtC) chimera.

The rtER-M1 mutated receptor was generated by using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The yeast expression plasmid YEprtER was used as template. Then the mutation was confirmed by sequencing. The hER-M1 mutated receptor was constructed according to the method previously described (30). The single-stranded template was prepared from phage M13 containing hER cDNA. After mutagenesis and verification by sequencing, the hER-M1 insert was subcloned into the SalI site of the expression vector pCMV5. The 1100 bp NotI/BglII fragment of hER inserted into the yeast expression plasmid YEpucG (31) was removed and replaced by the 1100 bp NotI/BglII fragment of pCMV5-hER-M1. The presence of the mutation was confirmed by sequencing. Oligonucleotides used in muta­genesis were: 5′-GGCTTGCGTAAGGACCGCCGCGGTGG­GCGG­GTTCTCAGG-3′ and 5′-CCTGAGAACCCGCCCACCGCGGCGGTCCTTACGCAAGCC-3′ for the rtER-M1, and 5′-GTT­TCAACATTCTCCCTCC*TCGGTCTTTTCGTATCCC-3′ for hER-M1 (*, suppression of TCT).

The oligonucleotides 5′-TCGAGCCATGGTCACAGTGA­C­CGGTCACAGTGACCGGTCACAGTGACCC-3′ and 3′-CGG­TACCAGTGTCACTGGCCAGTGTCACTGGCCAGTGTC­ACTGGGAGCT-5′ containing three ERE consensus binding sites (19) were cloned into the XhoI site of the CYC1 promoter of pLGΔ-178 (32), resulting in pLGΔ-178/3EREc. All constructs were verified by sequencing.

The other plasmids used in this study were the expression vectors YEphER (31), a gift from B. S. Katzenellenbogen, YEprtER (28), and the reporter genes, pLRΔ21-U1ERE and pLRΔ21-U3ERE (19), gifts from P. Chambon.

Yeast strains

The yeast strains used in this study were BJ2168 (a leu2 trp1 ura3-52 prb1-1122 pep4-3 prc1-407 gal2) (Yeast Genetic Stock Center, Berkeley, CA), BJ-ECZ (a leu2 trp1 ura3-52 prb1-1122 pep4-3 prc1-407 gal2::URA3–2ERE–CYC1 –lacZ) (31), a gift from B. S. Katzenellenbogen, and FL100 (ura3-373–251–328, trp1-4, ppr1-Δ1) (33), a gift from P. Chambon. Yeast cells were transformed using a lithium acetate method and selected by growth on complete minimal medium [0.13% dropout powder lacking uracil and tryptophan, 0.67% yeast nitrogen base, 0.5% (NH4)2SO4 and 1% dextrose] (28).

Whole and nuclear yeast extracts

Yeast cells were grown in selective medium (see above) at 30°C with vigorous shaking (300 r.p.m.) to an absorbance, at 600 nm, of 1.5–2.0 for whole extracts or 5.0 for nuclear extracts. The cells were harvested, cell walls were removed using lyticase (Sigma, St Quentin Fallavier, France), and the spheroplasts were lysed according to McDonnell et al. (34) to obtain a whole cell extract. Nuclear extracts were performed from spheroplasts as previously described (35).

Estradiol-binding assays

Estradiol-binding assays were performed as previously described (28). Aliquots of 45 µg of protein from whole cell extract (1 mg/ml) were incubated at 4°C for 16 h with [3H]estradiol (Amersham Corp., Buckinghamshire, UK) at increasing concentrations from 0.25 to 20 nM. Non-specific binding was determined in the presence of a 150-fold excess of unlabeled ligand. Free and bound ligand were separated by the addition of an equal volume of dextran-coated charcoal (0.5% charcoal, 0.05% dextran) in TEG (10 mM Tris–HCl pH 7.4, 1.5 mM EDTA, 10% glycerol). Samples were incubated for 8 min at 4°C with periodic resuspension. The charcoal was pelleted by a 10 min centrifugation at 10 000 g, and an aliquot of each supernatant containing bound ligand was withdrawn for liquid scintillation counting. Affinity of rtER for estradiol was determined by the Scatchard method (36).

Western blot analysis

Western blots were performed as previously described (28). Nuclear yeast extracts (10 µg) were fractionated on 10% SDS–polyacrylamide gel and transferred to Immobilon-P type polyvinylidene difluoride (PVDF, Millipore, Bedford, MA). Blots were first incubated with the purified anti-rtER antibodies or with the monoclonal anti-hER antibody (H222) for at least 12 h, then 1 h with the secondary conjugated antibody goat anti-rabbit or anti-mouse IgG-alkaline phosphatase (Tropix Massachusetts, Bedford, MA). After several washes, the blots were revealed by CSPD chemiluminescent detection system from Tropix kit and visualized by autoradiography for 15–30 s.

In vitro expression of ERs

rtER, rtER(hC), hER and hER(rtC) were released from YEpucG using BamHI and subcloned into pBluescript-KSII (Stratagene). Rabbit reticulocyte lysate and T7 RNA polymerase (Promega, Madison, WI) were used to in vitro translate these different receptors in the presence of TRAN35S LABEL™ (ICN, Orsay, France). An aliquot of this reaction mixture was analysed by fractionation on 10% SDS–poly­acrylamide gel. The gel was dried and auto­radiographed.

DNA binding assays

An aliquot of 4.0 µl of in vitro translation products was incubated in a buffer containing 10 mM Tris–HCl pH 7.5, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.1 mg/ml BSA, 1 mM PMSF, 80 ng poly dI-dC and 5% glycerol for 5 min at 4°C. For competition assays, samples were incubated with a 1–50-fold molar excess of unlabeled oligomer. Then, samples were incubated with 32P-labeled EREc (30 000 c.p.m.) for 15 min at room temperature. The probe was end-labeled with the Klenow fragment of Escherichia coli DNA polymerase I (Boehringer Mannheim, Meylan, France). After incubation, the reactions were loaded onto a 5% polyacrylamide gel and electrophoresed at 4°C for 3 h at 200 V in 0.5× TBE buffer. After electrophoresis, the gels were dried and exposed to a Kodak Biomax film. The binding was analyzed using an Instantimager (Packard, Rungis, France). We performed the on- and off-rate assays as previously described (37). For the on-rate experiment, the reaction was incubated with poly dI-dC for 15 min, then 32P-labeled EREc (60 000 c.p.m./lane) was added and the reaction mixture was further incubated for 0, 5, 10, 15, 20 and 25 min. The samples were then loaded on a 5% polyacrylamide gel. For the off-rate assay, the reaction was incubated with poly dI-dC for 15 min, then the probe was added for 25 min. A 3-fold molar excess of cold EREc was added and reaction samples were loaded on a 5% polyacrylamide gel after 0, 5, 10, 20 and 30 min incubation. ER–EREc complexes were quantitated as described above.

The EREc oligomer used was the same one present in the reporter plasmid pLRΔ21-U1ERE (19), 5′-TCGAGCCATGG­TCACAGTGACCC-3′.

β-galactosidase assays

β-galactosidase assays were performed as previously described (28) in the presence of increasing 17β-estradiol concentrations for the transactivation assays. The antiestrogens OHT and ICI 164,384, kindly provided by Dr A. Wakeling (ICI Pharmaceuticals, Macclesfield, UK), were used at 10 µM. The β-galactosidase activity was measured at 420 nm using the NpGal (o-nitrophenyl β-d-galactopyranoside) substrate and expressed in Miller units (38).

RESULTS

Construction of various chimeric human/trout ERs

Recently, a new isoform of rtER has been cloned. This form of rtER, named rtERL, contains 45 additional amino acids at the N-terminal region of the previous form of rtER, now called rtERS (39). Interestingly, rtERL showed almost no hormone-independent activity, suggesting that these additional amino acids (A domain) repress the hormone-independent activity of ER (39) (R.Métivier, F.Petit, Y.Valotaire and F.Pakdel, in preparation). Since we have previously studied the short isoform rtERS, we used the same isoform, referred as rtER, in the present study.

We have previously shown that hER and rtER have different transactivation properties in terms of potency and efficacy (28). To further investigate specific domain(s) responsible for such a phenomenon, chimeric human/trout ERs were constructed by domain swapping (Fig. 1A). In these chimeric receptors, the human A/B, C/D, C, D or E/F domains have been put in place of the trout homologous domains. In another chimeric receptor, the human C domain has been replaced by the corresponding trout domain. Western blotting showed that the apparent molecular mass for each chimeric ER was in agreement with the molecular mass deduced from corres­ponding DNA insert (Fig. 1B). However, from these experiments, we observed that for the chimeras rtER(hCD), rtER(hC) and rtER(hD), a doublet is observed, suggesting the existence of a specific degradation or another translation product in yeast (Fig. 1B). To verify the E2-binding capacity, Scatchard analyses were performed at 4°C with whole cell extracts from ER-transformed yeast in the presence of [3H]E2 (Table 1). As expected, rtER(hAB) and rtER(hCD) showed dissociation constants (Kd) of ∼0.82 and 1.18 nM, respectively, which are close to that for rtER (Kd = 1.35 nM). On the other hand, rtER(hEF), like hER, showed a much higher E2-binding affinity (Kd = 0.56 nM for rtER(hEF) and 0.23 nM for hER).

Figure 1.

Figure 1

Transactivation analysis of wild-type rtER, hER and chimeric ERs in yeast. (A) Schematic representation of rtER, hER and the chimeric ERs rtER(hAB), rtER(hCD), rtER(hEF), rtER(hC), rtER(hD) and hER(rtC). The regions containing the DBD and HBD are shown in full segment and hatched from right to left, respectively. Percentages indicate amino acid identity between hER and rtER. (B) Aliquots of 10 µg of nuclear extracts of untransformed (-ER) or indicated receptor transformed yeast were resolved on a 10% polyacrylamide gel and electrotransfered to a nylon membrane. Western blots were performed using a specific anti-rtER antibody-Ab3 (left) or an anti-hER antibody-H222 (right). The position of the molecular weight marker (66 KDa) is indicated at the left side. For the transactivation analyses, yeast strain BJ-ECZ containing the reporter gene 2ERE-CYC1-lacZ integrated in the genome and expressing either no receptor (-ER), rtER, hER, rtER(hAB), rtER(hCD), rtER(hEF) (C), rtER(hC), rtER(hD) or hER(rtC) (D) were grown in selective medium containing an increasing concentration of E2. The β-galactosidase activity was measured in yeast extracts. Values represent the means ± SEM from at least four separate experiments.

Table 1. Saturation analyses of the [3H]E2-binding of rtER, hER and chimeric ER receptors expressed in yeast.

Receptor Dissociation constant (Kd)
rtER 1.35 ± 0.48 nM
hER 0.23 ± 0.05 nM
rtER(hAB) 0.82 ± 0.20 nM
rtER(hCD) 1.18 ± 0.11 nM
rtER(hEF) 0.56 ± 0.20 nM

Aliquots of yeast extracts (45 µg protein) containing ER were incubated with increasing concentrations of [3H]-E2, in the presence and absence of 150-fold excess of radioinert E2, at 4°C for 16 h. Specific ligand binding was determined by subtracting non-specific from total binding. Affinity of the receptor for E2 was estimated by Scatchard analysis of the data from saturation curves. The Kd values indicated were determined from at least four independent experiments.

Importance of the DNA-binding domain for the magnitude of the ER transcriptional activity

rtER and hER present high homology in the primary sequence of the DNA-binding domain (92%) whereas important differences exist in the A/B domain (20% homology) and in the ligand-binding domain (60% homology) (7; Fig. 1A). Since the N-terminal and ligand-binding domain encompass the AF-1 and AF-2 transactivation functions of the receptor, respectively, we have suggested the implication of these domains in the differential functional activities of rtER and hER (28). The ability of wild-type and different chimeric ERs to activate gene transcription was studied. Recombinant yeast strains BJ-ECZ containing the reporter gene 2ERE-CYC1-lacZ integrated in the genome and stably expressing each receptor were grown in the presence of increasing concentrations of E2 under conditions described in Materials and Methods, and lacZ expression was assayed. Figure 1C and D shows the β-galactosidase activity dose–response curves obtained for different receptors. Note that in the absence of ER, β-galactosidase activity expressed from this reporter gene was <0.05 U (Fig. 1C). rtER, rtER(hAB) and rtER(hCD) were able to stimulate the β-galacto­sidase activity from ∼1 nM E2 whereas hER and rtER(hEF) required a 10-fold lower concentration (≈0.1 nM). This observation correlates well with the measures of E2-binding presented in Table 1 showing that the hEF domain is responsible for the high E2-binding affinity. Surprisingly, at high concentration of E2 (>10–8 M), rtER(hCD) showed a maximal trans­criptional activity similar to that obtained with hER (70–80 U β-galactosidase activity). However, transcriptional activity of rtER(hAB) as well as rtER(hEF) was roughly similar to that for rtER (15–20 U β-galactosidase activity) (Fig. 1C). To further examine the domain responsible for the high transcriptional activity of rtER(hCD), two chimeric trout ERs possessing the human C or D domains were constructed (Fig. 1A). In a transactivation assay, rtER(hD) presented a similar dose–response curve compared to rtER, whereas rtER(hC) was able to highly induce the β-galactosidase activity as hER or rtER(hCD) (Fig. 1C and D). To confirm the important role of DNA-binding domain in the transactivation function, we replaced the hER DNA-binding domain with those of rtER to generate the chimeric receptor hER(rtC) (Fig. 1A). As shown in Figure 1D, the maximal hER(rtC) transcriptional activity was similar to those obtained with rtER. Together, these results indicate that although the main differences in the primary sequences of rtER and hER are located in the A/B and E domains, differences in transcriptional activity in the presence of estradiol are not related to these domains but rather to the DNA-binding domain. The weaker magnitude of E2 stimulation of trans­cription mediated by rtER compared to hER may be a consequence of lower DNA-binding stability.

The rtER and hER DNA-binding domains present a similar ERE binding affinity but a different DNA-binding stability

To examine whether the rtER C domain is responsible for lower ERE binding activity when compared to the hER counterpart domain, we performed gel shift experiments. rtER, rtER(hC), hER and hER(rtC) were in vitro translated in the presence of 35S-labeled methionine/cysteine and run on a SDS–PAGE. Note that the number of methionines/cysteines present in both rtER and rtER(hC) and both hER and hER(rtC) is similar (38 versus 40, respectively). Therefore, we observed that each receptor was efficiently in vitro translated at a similar level (Fig. 2A). As previously described (40), a smallest ER 50 kDa form was produced in vitro for hER and hER(rtC). These two receptors possess the same internal translation initiation, apparently absent in rtER mRNA (Fig. 2A). Then, we examined the relative binding affinity of rtER, rtER(hC), hER and hER(rtC) for a consensus ERE using gel shift assays (Fig. 2B). The relative binding of ER to an ERE was quantified by an Instantimager and the measures show that both rtER and hER(rtC) present weaker relative binding to an EREc than do both hER and rtER(hC) (Fig. 2C). On the other hand, the relative binding affinities were similar for these four receptors (Fig. 2D). These different data suggest that the rtER and hER C domains present a similar relative ERE binding affinity. At equal amount of receptor, the intensity of the retarded complexes was 3 times lower for the rtER C domain containing complexes than those of the hER C domain (Fig. 2C). This observation could be a consequence of a different stability on an ERE between the rtER and hER C domains. To elucidate this possibility, we investigated how the rtER or hER C domains influence the kinetics of ER–EREc interactions. For this purpose, we compared rtER and rtER(hC), which differ by the C domain. The results obtained in the association kinetic study (on-rate) showed that rtER forms a maximal interaction after 25 min (Fig. 3A). The half-life value for the rate of rtER–ERE complex formation was ∼14 min. The presence of the hER C domain results in rapid maximal formation of an rtER(hC)–ERE complex (15 min, Fig. 3A). Moreover, 50% of the maximal retarded complex was reached 8 min after incubation of rtER(hC) with [32P]EREc. The kinetics of complex dissociation (off-rate) were also analyzed (Fig. 3B) and the results showed that the half-life of rtER is ∼30 min. rtER(hC), which binds an EREc more strongly than rtER (Fig. 2B), formed a very stable complex since the half-life value for loss of 50% is not reached 30 min after incubation with 3-fold excess of cold EREc (Fig. 3B). These results suggest that the hER C domain is responsible for a greater stability on an EREc than its counterpart in rtER.

Figure 2.

Figure 2

Examination of the relative DNA-binding affinity of wild-type and chimeric ERs to a consensus ERE. (A) rtER, rtER(hC), hER and hER(rtC) were in vitro translated in reticulocyte lysate in the presence of 35S-labeled methionine/cysteine, run on a 10% SDS–PAGE and autoradiographed. (B) Binding reactions containing 4 µl of in vitro translated ER were incubated with increasing amounts of consensus ERE before the addition of 0.08 pmol of 32P-labeled EREc. The molar excess of unlabeled ERE added was 0-fold (lanes 1, 8, 15 and 22), 1-fold (lanes 2, 9, 16 and 23), 2-fold (lanes 3, 10, 17 and 24), 5-fold (lanes 4, 11, 18 and 25), 10-fold (lanes 5, 12, 19 and 26), 20-fold (lanes 6, 13, 20 and 27) and 50-fold (lanes 7, 14, 21 and 28). Analysis of protein binding by gel retardation was accomplished as described in Materials and Methods. The arrows 1 and 2–3 show the specific ER–ERE complexes for rtER and rtER(hC), and hER and hER(rtC), respectively. The figure shows the result of a representative gel retardation out of three. Gels were then scanned with an Instantimager (Packard). The relative ER amount bound to the ERE without any competition (lanes 1, 8, 15 and 22) was determined (C) and the relative binding (%) of the different receptors were plotted against the molar excess of competitor EREc (D).

Figure 3.

Figure 3

Kinetics effects of the C domain on ER–EREc binding. (A) Association kinetics (on-rate) of rtER and rtER(hC) on EREc. Equal amounts of in vitro translated rtER or rtER(hC) were incubated with poly dI-dC for 15 min, then 32P-labeled EREc was added and the reaction mixture was further incubated for 0, 5, 10, 15, 20 and 25 min. The samples were then loaded on a 5% polyacrylamide gel. (B) Dissociation kinetics (off-rate) of rtER and rtER(hC) on EREc. The reactions were incubated with poly dI-dC for 15 min, then the probe was added for 25 min. A 3-fold molar excess of cold EREc was added and reaction samples were loaded on a 5% polyacrylamide gel after 0, 5, 10, 20 and 30 min incubation. ER–EREc complexes were quantitated using an Instantimager (Packard). Results are represented as the means ± SEM from three or four separate experiments. A representative experiment is showed on top of the graph for each receptor.

rtER requires protein–protein interactions to highly induce a reporter gene

To further investigate the function of the DNA-binding domain in the transcriptional activity of ERs, chimeric ERs were tested on two new reporter genes containing one or three EREs. FL100 yeast strain was transformed with the different reporter plasmids pLRΔ21-U1ERE or pLRΔ21-U3ERE (19) and with expression vectors containing the different chimeric ERs. Note that for maximal estrogenic induction, we used 1 µM E2 with FL100 versus 10 nM E2 with BJ2168 as previously reported (28). In the absence of ER, induction of β-galactosidase was very low for the two reporter genes (<0.11 U) (Fig. 4). Upon addition of E2, only hER, rtER(hCD) and rtER(hC) were able to induce β-galactosidase with the reporter gene containing one ERE (Fig. 4A). The induction of β-galactosidase by rtER, rtER(hAB), rtER(hEF), rtER(hD) or hER(rtC) was very weak (maximal induction ≈1 U) in comparison with those obtained for hER (≈15 U), rtER(hCD) (≈9 U) and rtER(hC) (≈5 U) (Fig. 4A). These data confirm the implication of DBD in the stabilization of the receptor on DNA. Note also that the relative lower transactivation activity of rtER(hCD) or rtER(hC) with the reporter gene pLRΔ21-U1ERE compared to that obtained with hER may be a consequence of a lower expression of this receptor within the cells.

Figure 4.

Figure 4

Transcriptional activation of reporter genes containing one or three EREs, by wild-type and chimeric ERs. Yeast strain FL100 containing either reporter plasmid pLRΔ21-U1ERE (A), or pLRΔ21-U3ERE (B), expressing no receptor (-ER), rtER, hER, rtER(hAB), rtER(hCD), rtER(hEF), rtER(hC), rtER(hD) or hER(rtC) were grown in the absence or presence of 10–8 or 10–6 M E2. The β-galactosidase activity was measured in yeast extracts. Values represent the means ± SEM from at least four separate experiments.

When we used the reporter plasmid pLRΔ21-U3ERE containing three EREs (Fig. 4B), the level of β-galactosidase induction was much higher (maximal E2-induction, 50–360 U) than with the reporter gene containing one ERE (maximal E2-induction, 1–15 U) confirming the existence of a synergistic effect of EREs to achieve high estrogen inducibility (19,28,41,42). Nevertheless, the synergistic effects were not the same for all receptors (Fig. 4B). Thus, it was ∼230-, 8-, 40-, 30-, 320-, 50-, 200- and 84-fold with rtER, hER, rtER(hAB), rtER(hCD), rtER(hEF), rtER(hC), rtER(hD) and hER(rtC) respectively (Fig. 4). These results suggest the necessity of protein–protein interactions to stabilize the ER dimers on DNA and therefore to highly induce the reporter gene (28). The presence of the human A/B domain leads to a lower β-galacto­sidase induction (50 versus 165 U, Fig. 4B) reflecting a decrease of the synergistic effect of rtER (40- versus 230-fold, Fig. 4), and this suggests the implication of the A/B domain in the synergistic effect.

Effect of an amino acid mutation on the transcriptional activity of rtER and hER

The DNA-binding domain presents high homology in the amino acid sequence between rtER and hER (Fig. 5A). However, seven amino acids differ between hER and rtER, and an additional arginine is found in hER (position 260, Fig. 5A). Interestingly, this arginine was absent in all fish ERs and ERβs already known and cloned (data not shown). To investigate whether this basic residue may be involved in the interaction with DNA, we have decided to either add it to rtER (denoted rtER-M1) or remove it from hER (denoted hER-M1) (Fig. 5). The ability of wild-type and mutated ERs to activate trans­cription was studied by using the yeast strain BJ2168 containing the reporter plasmid pLRΔ21-U1ERE (Fig. 5B), which showed a marked difference in induction between rtER and hER. Yeast stably expressing each ER was grown in the presence or absence of 10–9 or 10–8 M estradiol. At 10–9 M estradiol, a weak increase of rtER or rtER-M1 transcriptional activity was observed whereas the induction by hER or hER-M1 represents 50% of the maximal induction (Fig. 5B). Interestingly, at 10–8 M estradiol, rtER-M1 showed a maximal transcriptional activity 2.7-fold higher than those obtained with the wild-type receptor. However hER-M1 exhibited a maximal transcriptional activity similar to that obtained with the wild-type hER (Fig. 5B). Similar results were also obtained using the reporter gene 2ERE-CYC1-lacZ integrated into the genome (data not shown). These data indicated that, although addition of an arginine at position 223 in rtER (corresponding to the arginine 260 in hER) may contribute to a higher magnitude of the rtER transcriptional activity, this residue is probably not the only one responsible for the different DNA-binding and/or transcriptional activity found between these receptors.

Figure 5.

Figure 5

Transactivation analysis of wild-type rtER, hER and mutated ERs in yeast. (A) Schematic representation of the two zinc fingers present in the rtER DBD. The closed circle amino acids represent those of hER different from those of rtER. Conserved cysteine residues that function to coordinate zinc ions are in bold. Helix 1 contains P-box amino acid residues that provided deoxynucleotide contacts and fit into the major groove of the DNA helix. The three P-box residues, indicated by closed circles, are involved in binding site discrimination. The five amino acid residues of the D-box are involved in the dimerization between the DBD of homodimeric ERs. The arrowhead shows the arginine we added or removed in rtER or hER, respectively, to create rtER-M1 and hER-M1, respectively. (B) Yeast strains BJ2168 containing the reporter plasmid pLRΔ21-U1ERE and expressing rtER, rtER-M1, hER or hER-M1 were grown in selective medium in the presence or absence of 10–9 or 10–8 M E2. The β-galactosidase activity was measured in yeast extracts. Values represent the means ± SEM from at least seven separate experiments.

DISCUSSION

Using a simple eucaryotic organism such as yeast Saccharomyces cerevisiae, which offers a basal transcriptional machinery without tissue- and/or species-specific transcription factors, we have previously shown that stably expressed human or trout ERs exhibited different transactivation properties when tested on different reporter genes in the presence of hormone or anti­hormones (28). Since weak similarity exits in the primary structure between these two receptors for some domains (7), it was not possible to define critical amino acids responsible for these differential activities by doing site-directed mutagenesis. We have therefore undertaken the characterization of specific region(s) involved in these functional differences by generating several chimeric human/trout ERs by domain swapping.

Data showed that all chimeric ERs activate gene transcription in an E2 dose–response manner. However, dose–response curves obtained for rtER (28), or chimeric rtER(hCD), rtER(hC), rtER(hD) and rtER(hAB) receptors were shifted to the right compared to hER, rtER(hEF) or hER(rtC). These results together with direct determination of E2-binding affinity (Table 1) confirm that lower E2 sensitivity of rtER compared to hER may be attributed to the divergent amino acids in the HBD between these two receptors. Indeed, the four amino acids (G521, H524, L525 and M528), previously described to be important for the binding of E2 and the transcriptional activity of the receptor in presence of E2 (43), are located in a highly conserved region among all ERs. Nevertheless, even in the presence of higher E2 concentration, wild-type rtER or chimeric rtER(hAB), rtER(hD), rtER(hEF), as well as hER(rtC) receptors were much less potent transcriptional activators than hER. Interestingly, rtER(hCD) or rtER(hC) were able to induce β-galactosidase activity to a maximal level similar to that obtained with hER. Therefore, the magnitude of β-galactosidase induction by rtER or hER can be attributed to the activity of the C domain. Moreover, this result was confirmed by using the reverse chimera hER(rtC) which showed a maximal activity in presence of E2 similar to those of rtER. Gel mobility shift assays demonstrated that, while both receptors presented a similar DNA-binding affinity, the C domain of rtER is responsible for low association and fast dissociation rates compared to its counterpart in hER. It appears therefore that hER–DNA complex remains much more stable compared to rtER–DNA complex. Unlike other domains of ER, the C domain is the most conserved region between rtER and hER (92% homology) (7). This region contains two zinc fingers that are responsible for DNA binding (11), which includes discrimination of the ERE by the P-box residues (1214) and hormone-independent dimerization by the D-box residues (14,16). As these different sub-domains are totally conserved between rtER and hER (Fig. 5A), the different transcriptional activity of both receptors should involved amino acid(s) located outside of D- and P-boxes. Indeed, in the C domain, seven amino acid residues differ between both ERs (Fig. 5A) and an additional Arg (position 260) is found for hER. A study on nerve growth factor inducible-B (NGFI-B) defined a group of six amino acids (noted A-box), located in the D domain, as critical for interaction with the base pairs upstream of the recognition motif and involved in high-affinity DNA binding (44). Moreover, a region (noted T-box) located between the zinc finger modules and the A-box of RAR, RXR and TR appeared to be critical for high DNA-binding affinity (4446). The new estrogen receptor subtype (called ERβ) (4749) also shows some differences in the DNA-binding domain amino acid sequence in comparison with the hER. It has been shown that mouse ERβ did not bind as strongly to a vitellogenin A2-ERE probe as mouse ERα (49) and hERβ binds to DNA with a weaker affinity than that of hERα (50). However, this obser­vation was not reproduced by another laboratory (51). Neverthe­less, the differences in the DBD amino acid sequence between both types of ER might suggest a role for these residues in the binding stability of ERβ to DNA. Moreover, in mammalian cells, it has been shown that ERβ is a much less E2-dependent transactivation factor than ERα (48,49,52). In a preliminary study of site-directed mutagenesis to characterize amino acid residues implicated in the transcriptional activity of rtER or hER, we focused on the absence of a basic residue (Arg) in all cloned fish ERs and mammalian ERβ and its presence in the other ERs. The first data showed that the addition of this arginine in rtER enhances its activity, which may reflect an increase of receptor–DNA complex stability. In contrast, the deletion of the arginine 260 in hER did not change the trans­criptional activity of the receptor. These results suggest that this residue might be involved in the interaction of ER with DNA but it is obviously not the only amino acid conferring a more stable interaction with DNA. The involvement of other residues alone or in combination should be considered. The characterization of these amino acids is currently under investigation.

Alignment of rtER with hER indicated that the receptors share 60% amino acid similarity in E domains and low sequence homology in their N-terminal (A/B) domains (7,8), suggesting that each receptor may possess distinct trans­activation functions. In fact, the results obtained with rtER(hCD), rtER(hC) or hER(rtC) suggest that the synergistic activity of trout AF-1 and AF-2 is similar to hER. Therefore, the amino acid residues implicated in this synergistic effect are probably the same. For the AF-2 activation domain, the core sequence is highly conserved among the nuclear receptors including rtER (53).

In a previous study (28), we showed that ICI was unable to activate hER in yeast. However, an agonist activity of ICI has been reported with rtER. Mahfoudi et al. (54) showed that the AF-1 region was required, in combination with AF-2, for a transcriptionally active configuration of the mutants, which is dependent on ICI or tamoxifen activation. In mammalian and yeast cells, the partial agonist activity of OHT can be ascribed to the activity of AF-1, whereas AF-2 activity is inhibited by OHT (18,19,55). Since hER and rtER, occupied with ICI or OHT, differently activated gene transcription in yeast (28), we used chimeric human/trout ERs to further study the different action modes of these antiestrogens. As expected, only the presence of the human or trout A/B domain influences the transcriptional activity of wild-type or chimeric receptors in the presence of OHT, indicating that AF-1 is required for OHT agonism (data not shown). Recently, using ERα/ERβ chimeric receptors (52), it has been shown that the antiestrogen agonism of OHT or 2-phenylbenzofuran was dependent on the A/B domain of ERα. As expected, using the chimeric receptors, we observed that the agonistic effect of ICI requires the cooperativity of both the rtER AF-1 and AF-2 regions to obtain a similar activity to the rtER (data not shown). This cooperativity between AF-1 and AF-2 would be absent for hER bound to ICI. The hER conformational changes induced by ICI might be different from those of rtER. These conformational changes would prevent the interactions of AF activator domains with the transcriptional machinery for hER (56).

The molecular mechanisms by which ERs regulate gene expression are well studied but not completely understood. Utilization of human/trout ER chimeras highlights an important implication of the DNA-binding domain in the transactivation function. Therefore, different transactivation properties between hER and rtER are opportunities for understanding the mechanisms of action of ER domains.

Acknowledgments

ACKNOWLEDGEMENTS

We are grateful to Dr B. S. Katzenellenbogen (University of Illinois, Urbana-Champaign) for providing the yeast expression vector, YEpucG, and the yeast reporter plasmid, pLGΔ-178, and to Prof. P. Chambon and Dr D. Metzger (Laboratoire de Génétique Moléculaire des Eucaryotes du Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale-U.184, Strasbourg) for providing the reporter plasmids, pLRΔ21-U1ERE and pLRΔ21-U3ERE, and the yeast strain, FL100. This work was supported by the Centre National de la Recherche Scientifique, the Institut National de la Recherche Agronomique, and fellowships from the French Ministère de l’Enseignement Supérieure et de la Recherche and the Association pour la Recherche contre le Cancer to F. Petit. We also thank Dr P. Le Goff for reviewing the manuscript.

REFERENCES

  • 1.Green S. and Chambon,P. (1986) Nature, 324, 615–617. [DOI] [PubMed] [Google Scholar]
  • 2.Beato M. (1989) Cell, 56, 335–344. [DOI] [PubMed] [Google Scholar]
  • 3.Evans R.M. (1988) Science, 240, 889–895. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Green S. and Chambon,P. (1988) Trends Genet., 4, 309–314. [DOI] [PubMed] [Google Scholar]
  • 5.Yamamoto K.R. (1985) Annu. Rev. Genet., 19, 209–252. [DOI] [PubMed] [Google Scholar]
  • 6.Maître J.-L., Mercier,L., Dolo,L. and Valotaire,Y. (1985) Biochimie, 67, 215–225. [DOI] [PubMed] [Google Scholar]
  • 7.Pakdel F., Le Guellec,C., Vaillant,C., Le Roux,M.-G. and Valotaire,Y. (1989) Mol. Endocrinol., 3, 44–51. [DOI] [PubMed] [Google Scholar]
  • 8.Pakdel F., Le Gac,F., Le Goff,P. and Valotaire,Y. (1990) Mol. Cell. Endocrinol., 71, 195–204. [DOI] [PubMed] [Google Scholar]
  • 9.Krust A., Green,S., Argos,P., Kumar,V., Walter,P., Bornert,J.-M. and Chambon,P. (1986) EMBO J., 5, 891–897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Green S., Walter,P., Kumar,V., Krust,A., Bornert,J.-M., Argos,P. and Chambon,P. (1986) Nature, 320, 134–139. [DOI] [PubMed] [Google Scholar]
  • 11.Green S., Kumar,V., Theulaz,I., Wahli,W. and Chambon,P. (1988) EMBO J., 7, 3037–3044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Danielsen M., Hinck,L. and Ringold,G.M. (1989) Cell, 57, 1131–1138. [DOI] [PubMed] [Google Scholar]
  • 13.Mader S., Kumar,V., de Verneuil,H. and Chambon,P. (1989) Nature, 338, 271–274. [DOI] [PubMed] [Google Scholar]
  • 14.Umesono K. and Evans,R.M. (1989) Cell, 57, 1139–1146. [DOI] [PubMed] [Google Scholar]
  • 15.Kumar V. and Chambon,P. (1988) Cell, 55, 145–156. [DOI] [PubMed] [Google Scholar]
  • 16.Dahlman-Wright K., Wright,A., Gustafsson,J.-Å. and Carlstedt-Duke,J. (1991) J. Biol. Chem., 266, 3107–3112. [PubMed] [Google Scholar]
  • 17.Kumar V., Green,S., Stack,G., Berry,M., Jin,J.-R. and Chambon,P. (1987) Cell, 51, 941–951. [DOI] [PubMed] [Google Scholar]
  • 18.Berry M., Metzger,D. and Chambon,P. (1990) EMBO J., 9, 2811–2818. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Metzger D., Losson,R., Bornert,J.-M., Lemoine,Y. and Chambon,P. (1992) Nucleic Acids Res., 20, 2813–2817. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Tora L., White,J., Brou,C., Tasset,D., Webster,N., Scheer,E. and Chambon,P. (1989) Cell, 59, 477–487. [DOI] [PubMed] [Google Scholar]
  • 21.Kumar V., Green,S., Staub,A. and Chambon,P. (1986) EMBO J., 5, 2231–2236. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Fawell S.E., Lees,J.A., White,R. and Parker,M.G. (1990) Cell, 60, 953–962. [DOI] [PubMed] [Google Scholar]
  • 23.Fawell S.E., White,R., Hoare,S., Sydenham,M., Page,M. and Parker,M.G. (1990) Proc. Natl Acad. Sci. USA, 87, 6883–6887. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Tzukerman M.T., Esty,A., Santiso-Mere,D., Danielian,P., Parker,M.G., Stein,R.B., Pike,J.W. and McDonnell,D.P. (1994) Mol. Endocrinol., 8, 21–30. [DOI] [PubMed] [Google Scholar]
  • 25.Wakeling A.E. and Bowler,J. (1987) J. Endocrinol., 112, R7–R10. [DOI] [PubMed] [Google Scholar]
  • 26.Wakeling A.E. and Bowler,J. (1988) J. Steroid Biochem., 31, 645–653. [DOI] [PubMed] [Google Scholar]
  • 27.Wilson A.P.M., Weatherill,P.J., Nicholson,R.I., Davies,P. and Wakeling,A.E. (1990) J. Steroid Biochem., 35, 421–428. [DOI] [PubMed] [Google Scholar]
  • 28.Petit F., Valotaire,Y. and Pakdel,F. (1995) Eur. J. Biochem., 233, 584–592. [DOI] [PubMed] [Google Scholar]
  • 29.Higuchi R. (1990) In Innis,M.A., Gelfand,D.H., Sninsky,J.J. and White,T.J. (eds), PCR Protocols: a Guide to Methods and Applications. Academic Press, Inc., San Diego, pp. 177–183.
  • 30.Pakdel F. and Katzenellenbogen,B.S. (1992) J. Biol. Chem., 267, 3429–3437. [PubMed] [Google Scholar]
  • 31.Wrenn C.K. and Katzenellenbogen,B.S. (1993) J. Biol. Chem., 268, 24089–24098. [PubMed] [Google Scholar]
  • 32.Guarente L. and Mason,T. (1983) Cell, 32, 1279–1286. [DOI] [PubMed] [Google Scholar]
  • 33.Roy A., Exinger,F. and Losson,R. (1990) Mol. Cell. Biol., 10, 5257–5270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.McDonnell D.P., Nawaz,Z., Densmore,C., Weigel,N.L., Pham,T.A., Clark,J.H. and O’Malley,B.W. (1991) J. Steroid Biochem. Mol. Biol., 39, 291–297. [DOI] [PubMed] [Google Scholar]
  • 35.Petit F.G., Métivier,R., Valotaire,Y. and Pakdel,F. (1999) Eur. J. Biochem., 259, 385–395. [DOI] [PubMed] [Google Scholar]
  • 36.Scatchard G. (1949) Ann. N. Y. Acad. Sci., 51, 660–672. [Google Scholar]
  • 37.Wang W., Dong,L., Saville,B. and Safe,S. (1999) Mol. Endocrinol., 13, 1373–1387. [DOI] [PubMed] [Google Scholar]
  • 38.Miller J.H. (ed.) (1972) Experiments in Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York, NY.
  • 39.Pakdel F., Métivier,R., Flouriot,G. and Valotaire,Y. (2000) Endocrinology, 141, 571–580. [DOI] [PubMed] [Google Scholar]
  • 40.Barraille P., Chinestra,P., Bayard,F. and Faye,J.-C. (1999) Biochem. Biophys. Res. Commun., 257, 84–88. [DOI] [PubMed] [Google Scholar]
  • 41.Klein-Hitpab L., Kaling,M. and Ryffel,G.U. (1988) J. Mol. Biol., 201, 537–544. [DOI] [PubMed] [Google Scholar]
  • 42.Martinez E. and Wahli,W. (1989) EMBO J., 8, 3781–3791. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Ekena K., Weis,K.E., Katzenellenbogen,J.A. and Katzenellenbogen,B.S. (1997) J. Biol. Chem., 272, 5069–5075. [DOI] [PubMed] [Google Scholar]
  • 44.Wilson T.E., Paulsen,R.E., Padgett,K.A. and Millbrandt,J. (1992) Science, 256, 107–110. [DOI] [PubMed] [Google Scholar]
  • 45.Lee M.S., Kliewer,S.A., Provencal,J., Wright,P.E. and Evans,R.M. (1993) Science, 260, 1117–1121. [DOI] [PubMed] [Google Scholar]
  • 46.Predki P.F., Zamble,D., Sarkar,B. and Giguère,V. (1994) Mol. Endocrinol., 8, 31–39. [DOI] [PubMed] [Google Scholar]
  • 47.Kuiper G.G.J.M., Enmark,E., Peltohuikko,M., Nilsson,S. and Gustafsson,J.A. (1996) Proc. Natl Acad. Sci. USA, 93, 5925–5930. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Mosselman S., Polman,J. and Dijkema,R. (1996) FEBS Lett., 392, 49–53. [DOI] [PubMed] [Google Scholar]
  • 49.Tremblay G.B., Tremblay,A., Copeland,N.G., Gilbert,D.J., Jenkins,N.A., Labrie,F. and Giguere,V. (1997) Mol. Endocrinol., 11, 353–365. [DOI] [PubMed] [Google Scholar]
  • 50.Cowley S.M., Hoare,S., Mosselman,S. and Parker,M.G. (1997) J. Biol. Chem., 272, 19858–19862. [DOI] [PubMed] [Google Scholar]
  • 51.Pettersson K., Grandien,K., Kuiper,G.G.J.M. and Gustafsson,J.-Å. (1997) Mol. Endocrinol., 11, 1486–1496. [DOI] [PubMed] [Google Scholar]
  • 52.McInerney E.M., Weis,K.E., Sun,J., Mosselman,S. and Katzenellenbogen,B.S. (1998) Endocrinology, 139, 4513–4522. [DOI] [PubMed] [Google Scholar]
  • 53.Danielian P.S., White,R., Lees,J.A. and Parker,M.G. (1992) EMBO J., 11, 1025–1033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Mahfoudi A., Roulet,E., Dauvois,S., Parker,M.G. and Wahli,W. (1995) Proc. Natl Acad. Sci. USA, 92, 4206–4210. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Metzger D., Ali,S., Bornert,J.-M. and Chambon,P. (1995) J. Biol. Chem., 270, 9535–9542. [DOI] [PubMed] [Google Scholar]
  • 56.McDonnell D.P., Clemm,D.L., Hermann,T., Goldman,M.E. and Pike,J.W. (1995) Mol. Endocrinol., 9, 659–669. [DOI] [PubMed] [Google Scholar]

Articles from Nucleic Acids Research are provided here courtesy of Oxford University Press

RESOURCES