Gene, 119 (1992) 237-245 0 1992 Elsevier Science Publishers

GENE

B.V. All rights reserved.

237

0378-l 119/92/$05.00

06633

Functional analysis of the human transactivation assay in yeast t (Steroid

hormone;

scriptional

Saccharomyces

regulation;

cerevisiae; regulatable

estrogen

promoter;

receptor

inducible

growth;

using a phenotypic

heterologous

gene expression;

tran-

synergism)

Benoit Pierrat, David M. Heery, Yves Lemoine

and RCgine Losson

Laboratoire de G&Ptique Molkxdaire des Eucaryotes du CNRS. Unit6 184 de I’INSERM, Institut de Chimie Biologique, Fault& de Medecine. Strasbourg. France Received

by A. Hinnen:

6 April 1992; Accepted:

6 May 1992; Received

by publishers:

I June 1992

SUMMARY

We have constructed yeast strains in which the expression of the Saccharomyces cerevisiae URA3 gene is induced by the human estrogen receptor (hER). Promoter sequences required for both basal and activated transcription of URA3 were replaced with one or three estrogen-response elements (EREs) positioned upstream of the native TATA box. These constructs were each integrated at the TRPI locus of a yeast strain in which the natural URA3 gene had been deleted, and the integrants were transformed with low- or high-copy-number shuttle plasmids expressing wild-type or truncated derivatives of hER. Transformants were assayed for growth on uracil-deficient medium plus or minus estradiol (E2), for resistance to 5fluoroorotic acid (5FOA) and for activity of OMPdecase (orotidine-5’-monophosphate decarboxylase), the product of the URA3 gene. We show that the growth and 5-FOA-resistance (5-FOAR) phenotypes of these strains are strictly dependent upon the function of the receptor derivatives. Induction of URA3, measured by OMPdecase activity, was observed over a 20- to 2500-fold range depending on the receptor derivative, its expression level and the number of EREs in the responsive promoter. Both one- and three-ERE reporter strains expressing the full-length receptor are completely E,-dependent for growth, and display a 5-FOAR phenotype in the absence of the hormone. We demonstrate that the individual hER transactivation functions, TAF 1 and TAF2, are both functional in yeast, and that the hormone-dependent TAF2 is the more potent activator on our reporters. We show that hER displays strong homosynergism in yeast, and discuss the contributions of the two TAFs in hER synergism. The reporter system described here provides a phenotypic transactivation assay and constitutes a versatile genetic screening and selection system, as both phenotypic assays are amenable to positive selection strategies. In addition, the system may be adapted for the study of other inducible transcription factors.

INTRODUCTION

It has been shown that mammalian steroid hormone receptors and several other mammalian and viral transcrip-

Correspondence to: Dr. R. Losson, LGME du CNRS, 67085 Strasbourg Cedex, France. Tel. (33-88)37-12-55; Fax (33-88)37-01-48. ’ Dedicated to the memory

11 rue Humann,

tion factors retain at least some of their functions when expressed in yeast (Metzger et al., 1988; Schena and Yamamoto, 1988; Mak et al., 1988; McDonnell et al., 1989; Lambert et al., 1989; Ding et al., 1991; Purvis et al., 1991).

ing hER; kb, kilobase ribonucleotide;

aa, amino

acid(s);

bp, base

pair(s);

element;

5-FOA,

E,, estradiol;

ERE, estrogen-response

acid; HEGO,

wt hER; hER, human estrogen

receptor;

d, deletion; 5-fluoroorotic

hER, gene encod-

nilate

activation XGal,

function;

isomerase; sequence;

oligo, oligodeoxy-

orotidine-5’-monophosphatedecarboxylase;

ORF, open reading frame; PCR, polymerase phoglycerate kinase; R, resistance/resistant;

of Dr. J.-P. Lecocq.

, transactivation Abbreviations:

or 1000 bp; nt, nucleotide(s);

OMPdecase,

chain reaction; PGK, phosS., Saccharomyces; TAF-

TRPl, gene encoding

tsp, transcription

start

URA3, gene encoding

phosphoribosyl

point(s); OMPdecase;

5-bromo-4-chloro-3-indolyl-8_D-galactopyranoside.

anthra-

UAS, upstream wt, wild type;

238 For example, it has been demonstrated that the major known functions of the human estrogen receptor, namely DNA binding, hormone binding, nuclear localization and

the TATA box (Roy et al., 1990). Removal of the poly(dA:dT) element leads to at least a lOOO-fold decrease in expression of URA3, resulting in no detectable OMPdecase

transcriptional enhancement of a gene bearing an estrogen response element are operative when the receptor cDNA is expressed in yeast (Metzger et al., 1988). The molecular mechanism(s) by which this protein and other steroid hormone receptors modulate the expression of their target genes is currently of intense interest. Using a heterologous system such as yeast to dissect the transactivation functions of a mammalian transcription factor offers several important advantages including the utility of yeast genetic methods for mutation selection and complementation, and

activity and an inability to grow on media lacking uracil. However this deleted promoter is still inducible by PPRl, and induction restores the cell growth (Roy et al., 1990). We have taken these results into account to construct URA3 derived reporter genes, in which both of the cis-acting URA3 regulator elements, poly(dA:dT) and UASu, were deleted and replaced by one (IERE) or three (3ERE) estrogen response elements. The reporter constructs were stably integrated into the TRPI locus of a ura3-A strain. We show that the growth and the 5-FOA resistance of these reporter strains on selective media is strictly dependent upon the function of the human estrogen receptor, thus making it

the absence of interference from endogenous receptors or other proteins which may be encountered in mammalian systems and affect receptor function. In most cases described in the literature, transactivation by hormone receptors in yeast has been demonstrated by activation of 1ucZ reporter constructs preceded by a promoter containing the appropriate response elements. Thus, in yeast cells expressing a receptor, /3-galactosidase activity in cell free extracts can be increased many-fold by addition of the hormone in the culture medium. It is also possible to detect 1ucZ expression by the blue coloration of yeast colonies grown on media containing XGal. Although it has been suggested that this system may be used for genetic screening (Schena and Yamamoto, 1988), in our hands the poor reproducibility and heterogeneity of individual colony colour make the lucZ/XGal system unattractive for this purpose. To overcome this problem, we have developed the S. cerevisiae URA3 gene as a convenient reporter system. URA3 encodes an inducible enzyme, OMPdecase, involved in the biosynthesis of pyrimidines (Lacroute, 1968). The activity of this enzyme can be easily measured in cell free extracts and gives a sensitive and reliable indication of the level of URA3 expression. In addition to this quantitative measurement, two positive selectable phenotypes exist: URA3 transcription is required for growth on media lacking uracil, while conversely, lack of URA3 expression leads to resistance to 5-FOA (Boecke et al., 1987). 5-FOA is converted into a toxic metabolite by the URA3 gene product, and therefore can be used in the negative selection of URA3 expression. Basal expression of URA3 in a wt strain is normally sufficient to yield cells sensitive to 5-FOA (Boecke et al., 1987). This basal expression is dependent on a poly(dA:dT) tract located upstream from the TATA box between positions -216 to -157 with respect to the start codon (Roy et al., 1990). Under uracil starvation conditions, expression of URA3 is further induced fivefold by the PPRI gene product (Losson and Lacroute, 1981; 1983) which binds to a URA3 promoter upstream activation sequence (UAW) located between the poly(dA:dT) tract and

feasible to positively select mutations in either the receptor or the host genome which affect receptor activity. We demonstrate the utility of this system by comparing the abilities of wt and truncated derivatives of the estrogen receptor to activate the URA3 reporter gene in response to changes in their expression level and the number of ERE elements. Our results show that hER is able to homosynergize strongly in yeast. Its two independent transactivation functions, TAFl and TAF2, can both activate transcription in yeast, but their transactivation potentials and synergistic properties differ.

RESULTS

AND DISCUSSION

(a) Construction of an estradiol(E,)-dependent yeast strain The promoter of the 5’. cerevisiae FL100 URA3 gene was engineered to delete the two c&elements required for its expression: the poly(dA:dT) region which directs basal level of expression, and the upstream activating site (UASu) which is involved in PPRl activation of URA3. The HindIII-PstI fragment of the plasmid pER7 (Roy et al., 1990) which contains most of the wt URA3 promoter was excised and a series of overlapping oligos were then used to reconstruct a hybrid promoter containing three tandem estrogen response elements linked to the natural promoter at nt -139 (see Fig. 1A). Thus, the GA-binding factor sequences (Roy et al., 1990), the native TATA box, and the natural initiation sites of the URA3 promoter are conserved. We confirmed that this SERE-URA3 gene on the TRPl containing centromeric plasmid pER7 fails to support the growth on uracil deficient medium of the strain SL12 (uru3Al w-l-4) in which the coding sequence of the natural URA3 gene has been deleted (data not shown). The 3EREURA3 construct was then cloned into the EcoRV site of the TRPl gene, transcription of the two genes being divergent as shown in Fig. 1A. Next the disrupted TRPI gene was integrated into the genome of the yeast strain RL15 (rele-

239

URA3 UASu

TATA I, I,

v I

-100

PSd

HindIll -139

-216

W-I k

TM +802

ATG +t

Smal

URA3 TRPl (PL3)

0 0 0

WT

PL3

PL3

Fig. 1. Construction

of the Es-dependent

rsp of URA3. The coordinates

yeast strain PL3. (A) Structure

are determined

with respect

of the lJRA3 promoter

to the A of the ATG

start codon.

in wt (WT) and PL3 strains:

the bent arrows

The region of the wt URA3 promoter

between

indicate positions

-216 to -139 was replaced by three copies of the Xenopus luevis vitellogenin A2 estrogen-response element (ERE: Klein-Hitpass et al., 1986). A HindIH_,,,,-SmaI( + s2sj DNA fragment containing the URA3 coding sequence fused to the 3ERE-URA3 promoter was cloned into the EcoRV site of TRPZ in a reverse orientation (the heavy arrows indicate orientation of the ORFs). The TRPl gene disrupted by the ERE-URA3 gene was introduced back into the yeast strain RLl5 /eu2-Al rrpl::jlERE-URA3). start codon

(Mum-ura3-Al

The uru3-Al

to the AccI site located

leu2-Al deletions

have been previously

1981). Transformations as required.

introduced

27 nt downstream described

($80lacZAMlS)}

All DNA manipulations,

et al. (1989). (B): Growth phenotypes carrying the HIS3 selectable marker

1991) to produce

the integrant

PL3 (uru3-Al

from the PstI site at nt - 18 with respect

(S. Liang, and F. Lacroute,

personal

his3-A200

to the A of the ATG

communication).

The his3-A200

and Hieter (1989). Yeast were grown in YEPD or selective minimal medium (Sherman

(Becker and Guarente,

1991). Strain E. coli DH5 {F-,

was the usual E. coli recipient including

(Rotstein,

in the RL15 strain extends

from the stop codon

by Sikorski

were done by electroporation

gyrA, relA 1, A(lucZYA-argF)Ul69, (100 pg/ml)

his3-A200 leu2-Al) by gene replacement

deletion

ligation

and transformation

strain,

grown

endA 1, hsdR 17, r; m;,

in L-broth

medium,

of E. coli, were performed

supE44, t&l,

supplemented

essentially

and et al.,

recA 1,

with ampicillin

as described

by Maniatis

of the PL3 strain transformed with YEp90 or YEp90HEGO: The multicopy plasmid YEp90 is a 2 p-derived plasmid and a yeast PGK promoter and terminator cassette containing a unique EcoRI cloning site. It is derived from the

plasmid pTG848 (Loison et al., 1989) by exchanging the URA3 marker with the HIS3 marker, deleting the LEU2 gene by EcoRI digestion, filling in and religation, and changing the BglII site present downstream from the PGK promoter into an EcoRI site by site-directed mutagenesis (D. Metzger, R.L., J.-M. Bornert,

Y.L. and P. Chambon,

trol of the PGK promoter.

in press). The EcoRI

Plasmid YEp90HEGO

site of YEp90. The PL3 transformants at 100 nM ( + / - E,) and on 5-fluoorotic

contains

site in this PGK cassette

the complete

were grown exponentially acid containing

medium

allows the insertion

of cDNA

which are expressed

under the con-

coding sequence of the hER gene (Tora et al., 1989a) cloned into the unique EcoRI

on - histidine (5-FOA)

vant genotype ura3-Al his3-A200 leu2-Al) so that the resulting strain PL3 contains one copy of 3ERE-URA3 gene at the TRPl locus. The integration at TRPI was confirmed by PCR analysis (data not shown) and the structure of the PL3 genome at this locus is indicated in Fig. 1A. Note that

+ uracil medium,

and spot tested on minimal medium plus or minus E,

plus or minus E,. All the media contain

tryptophan

and leucine.

several attempts to integrate the reporter at the URA3 locus were unsuccessful for reasons which are not clear. There is circumstantial evidence that an essential gene lies 5’ to the URA3 gene (A. Roy, personal communication). It is possible that the poly(dA:dT) or other sequences in this

240 region are required

for normal

expression

of this upstream

trogens nafoxidine and 4-hydrox~amoxifen (Berry et al., 1990) can both induce the growth on selective medium of PL3 transformed with HEGO (data not shown). In contrast, we did not observe any weak agonist or antagonist activity with the ‘pure’ antiestrogen ICI 164 384 (Wakeling and Bowler, 1987). These results show that growth of PL3 is strongly dependent on the function of hER and that both selectable phenotypes, i.e., growth and 5-FOA resistance,

gene, making their deletion lethal to the cell. A yeast multicopy HIS’3 selectable vector (YEp90) expressing the human estrogen receptor HEGO under the control of the yeast PGK promoter, was introduced into PL3. The highly active PGK promoter is constitutively expressed during growth on glucose based media (Tuite et al., 1982), thus the level of receptor in the cells should be relatively high and presumable nonlimiting. Transformants were assayed for growth on selective medium lacking uracil or containing uracil plus SFOA, with or without E, (100 nM). Fig. IB clearly shows that the PL3 strain transformed with the control plasmid YEp90 is unable to grow

may be used for the selection of meaningful protein.

variants

of this

(b) Synergistic activation by estrogen receptor derivatives The number

of EREs

in hER-regulated

promoters

has

on medium lacking uracil in the presence or absence of E,, and is 5-FOA resistant. In contrast, while the PL3 strain

been shown to affect the efficiency of activation by the hER in mammalian cells (Martinez and Wahli, 1989; Ponglikit-

transformed with the plasmid YEp90HEGO also fails to grow on medium without uracil and estradiol, addition of hormone induces wt growth and sensitivity to 5-FOA. We found that as little as 5 nM of E, is sufficient to fully induce the growth of PL3 expressing intact hER and also that higher levels (over 1 FM) of the partially agonistic anties-

mongol et al., 1990). To study the effect of ERE number on activation in the system presented here, we constructed a second reporter strain (PLl) which is identical to PL3 except that it contains a single ERE instead of three EREs in the hybrid promoter. The wt human estrogen receptor HEGO (Tora et al., 1989a), or its truncated derivatives

HEGO

HE1 5

1 1

AA

r

-

I C ID m 595

ICll

282

I

I

F3

BlO Fig. 2. Western blot analysis of wt and truncated

mutants

hER. (A) Schematic

representation

of receptor derivatives:

HEGO

represents

the wt hER (Tora

et al. i989a), divided into six regions (A-F) from aa 1 to 595. The hatched box indicates the ho~one-binding domain. C is the DNA-binding domain. For further details and nomenclature, see Kumar et at. (1987). (Pane1 B) LcveIs of wt and mutant receptor proteins: Extracts of the PL3 strain transformed with ER-expression vectors (as indicated at the top) were laaded onto a 0.1 ?h SDS-lo% polyacrylamide get, electrophoresed, electrotransferred to nitrocellulose filters, and probed with the BlO and F3 monoclonal antibodies (recognizing epitopes in the N- and C-terminal regions of ER, respectively; S. Ah, Y. Lutz, J.P. Belloq and D. Metzger, derived from pRS313 (Sikorski and Hieter,

in preparation) as described in Towbin et al. (1979). The centromeric low-copy-number plasmid YCp90 was 1989) and contains the yeast PGK promoter and terminator isolated from YEp90 as a ClaI fragment.

241 HE1 5 and HEG19 containing the transactivation functions TAFI and TAF2, respectively (Tora et al., 1989b; Tasset

TABLE

et al., 1990), were expressed from the PGK promoter of the plasmid YEp90. Western analysis (immunoblot) confirmed that all three are expressed to approximately equivalent levels in yeast (Fig. 2). Activation of the UR43 reporters by these proteins was determined by spot test assays for growth on selective media (Fig. 3), and quantified by mea-

(HEGO)

I

OMPdecase

activity in PLl and its truncated

from the high-copy-number plasmid

and PL3 strains derivatives plasmid

assays,

both

reporter

the wt receptor

and HEG19)

expressed

YEp90 (A) or the low-copy-number

YCp90 (B)

OMPdacase activity Transforming plasmid PLl

suring the OMPdecase activity in cell free extracts prepared from transformants grown in the absence or presence of 100 nM E, (Table IA). In agreement with the growth

containing

(HE15

PL3

I - E2

+ Ez

- E2

+ E2

< 0.01

c 0.01

c 0.01

< 0.01

5 0.01

0.80

< 0.01

25.00

c 0.01

260

2.20

0.34

0 09

4.05

< 0.01

< 0.01

< 0.01

(A)

strains display an extremely low basal level of OMPdecase

YEp90 HE15

(A) PLl

L

PL3

I

I

I - E2

1 - E2

+ E2

+ E2

coo1 -x 0.01

HEG19

PI c

0.01

HEGO