Synergism between Steroid Response and Promoter Elements during ...

3 downloads 1880 Views 3MB Size Report
Aug 22, 1990 - Cell-free Transcription* ... taining a nuclear factor 1 (NF-1)-binding site or a .... of the natural ovalbumin promoter in a cell-free system, we.
THEJOURNAL OF BIOLOGICAL CHEMISTRY The American Society for Biochemistry and Molecular Biology, Inc

Vol. 266, No. 9, Issue of March 25, pp. 5905-5910, 1991 Printed in U.S. A.

(c)1991 by

Synergism between Steroid Response and Promoter Elements during Cell-free Transcription* (Received for publication, August 22, 1990)

George F. Allan$, Nancy H. IngS, Sophia Y. TsaiS, Ganesan Srinivasans, NancyL. WeigelS, E.Brad Thompson#, Ming-Jer Tsai$, and Bert W. O’Malley$V From the $Department of Cell Biology, Baylor College of Medicine, Houston, Texas 77030 and the §Departmentof Human Biological Chemistry and Genetics, University of Texas Medical Branch, Galueston, Texas 77550

We haveanalyzedquantitativelytheinfluence of et al., 1988) and the progesterone receptor (Klein-Hitpass et distal promoter elements on steroid-responsive gene al., 1990) can enhance initiation complex formation on their expression in vitro. Functional synergism between en- respective promoters by recruiting and/or stabilizing their hancer and distal promoter elements was examined specificassembly on DNA. Stabilization by factorsbound using two model promoters, one containing a natural adjacenttothecapsite is likely to be a direct process, promoter (mouse mammary tumor virus long terminalinvolvingspecific protein-proteincontacts withindividual repeat) and one constructed artificially. Human glucomponents of the complex. More distant enhancer elements, cocorticoid receptor (GR) expressed in baculovirus in- ontheotherhand, may be unableto influence complex a mouse mammarytumor ducestranscriptionfrom assembly without the aid of additional factors bound at the virus long terminal repeat-containingDNA template. promoter. These interactionswould be reflected in functional Transcription is diminished by oligonucleotides containing a nuclearfactor 1 (NF-1)-binding site or a synergism between the two regulatory elements. Activation via the SV40 enhancer requires Spl-bindirig glucocorticoid/progesterone response element. Quanet al., 1983; Takahashi et titative analysis indicates that NF-1 andGR act syn- sites in the early promoter (Everett al., 1986), while activation of the heterologous P-globin gene ergistically during transcriptional activation. Incontrast, efficient activationby GR or purified chick pro- by this enhancer (Treisman and Maniatis, 1985) or multiple gesteronereceptor of a glucocorticoid/progesterone heat shock elements (Bienz and Pelham, 1986) is dependent response element-linked ovalbumin promoter does notupon distinct promoter sequences. Likewise, K enhancer-inrequire interaction with the chicken ovalbumin up- duced expressionfromthe heterologous thymidinekinase promoter is down-regulated by the removal of a linked imstream promoter (COUP) element in the distal promoter. Lack of synergism is not related to enhancer munoglobulin octamer sequence (Parslow et al., 1987). Steroid strength,sincethe glucocorticoid/progesterone re- hormone receptors also interact with downstream transcripsponse elements can be moved further from the pro- tion factors toinfluence cellular gene expression, since steroid moter or reduced to a single copy response element response elements (SREs)’ can cooperate with a variety of without increasing the dependence upon COUP. Strong heterologous binding sites, including those for nuclear factor synergism is restored following substitutionof an NF- 1 (NF-I), Spl, and octamer transcription factor, to stimulate 1 distal promoter element for the COUP element in the thymidine kinase promoter (Schule et al., 1988; Strahle et this construct. Our results suggest that synergismbe- al., 1988) and the @-globin promoter(Schatt et al., 1990). is Furthermore, glucocorticoid-dependent activation of the tween steroid response and distal promoter elements dependent upon the identity of the promoter element mouse mammary tumor virus (MMTV) long terminal repeat rather thanupon the inherent strength of the enhancer (LTR) is abolishedby removal of an NF-1-binding site in the element. distal promoter (Buetti and Kuhnel, 1986; Miksicek et al., 1987). Previous experimental analysisof the interactionsbetween gene regulatory factors hasrelied upon transient transfection The expression of a typical eukaryotic gene is controlledby multiple sequence elements within its 5”flankingregion. One analysis of recombinant DNA constructs inliving cells. Such way in which expression can be regulated by these elements studies are usually carried out by gross mutagenesis of regulatory sequences, whichmay remove overlapping binding sites is by controllingtherate of transcriptioninitiation.The level of a assembly of a preinitiation complex atthecapsiteis a or alterspatialrelationships.Furthermore,the transcription factor varies widely in different cell types and prerequisite to RNA synthesis (Lassar et al., 1983; Reinberg and Roeder, 1986; Buratowski et al., 1989). In theory, any of cannot be controlled. However, precise functional dissection the processes involved in its assembly is a potential point of and quantitation of these interactions can be carried out in control for transcriptional regulatory factors. For example, in cell-free assays using purified components. We have recently vitro transcription studies and template commitment assays established a progesteroneand glucocorticoid receptor-dependent in vitro transcription system using a minimal prohave shown that the major late transcription factor (Sawadogo and Roeder, 1985), activating transcription factor (Hai * This work was supported by National Institutesof Health grants, The costs of publication of this article were defrayed in part by the payment of pagecharges. Thisarticlemusttherefore be hereby marked “ndvertisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 7 To whom correspondence should be addressed.

’ Theabbreviations used areSREs,steroid response elements; COUP(-TF), chicken ovalbumin upstream promoter (transcription factor); GR,glucocorticoid receptor; GRE/PREs, glucocorticoid/progesterone response elements; MMTV-LTR, mouse mammary tumor virus long terminal repeat; NF-1, nuclear factor 1; PCR, polymerase chain reaction; PR, progesterone receptor; TAT, tyrosine aminotransferase; bp, base pair.

5905

5906

I n Vitro Synergism between Promoter and Steroid Response Elements

moter containing tandem glucocorticoid/progesterone response elements (GRE/PREs) adjacent to the TATA box of the chicken ovalbumin gene (Klein-Hitpass et al., 1990; Bagchi et al., 1990a; Tsai et al., 1990). In the present report, we describe an investigation of the requirements of steroid receptors for distal promoter-binding proteins during i n vitro activation of two complex promoters, the MMTV-LTR and the intact ovalbumin promoter. We find that synergistic cooperation between GRE/PREs and distalpromoter elements is dependent upon the identity of the transcription factor that binds to thedistal promoter.

A pllMTV-400 -70

-30 -20

I

L PPRSOV- 100 PRE PRE -50

-20

I

13.50 G-FREE

COUP

MATERIALS ANDMETHODS

Plasmid Constructs-pMMTV-400 was constructed by polymerase chain reaction (PCR)-generated synthesisof a fragment corresponding to the MMTV-LTR from -400 to -20. Plasmid pAHCAT (Nordeen et al., 1989) was used as template. Primers carried BglII overhangs (5’ end) or SstI overhangs (3’ end), andfragments were cloned into BglII- and SstI-digested pLovTATA (Klein-Hitpass et al., 1990). pPRE20V-100was generated by cloning of a PCR-generated fragment of the chicken ovalbumin promoter from -100 to -20 into BglII- and SstI-cut pPREzLovTATA. Linearized pSV.OG(-221) (Pastorcic et al., 1986) was used as PCR template. TheNF-1 recombinant (pPRE20VNF-100)was made by inserting overlapping oligonucleotides into BglII- and SstI-digested pPRE,LovTATA. Oligonucleotide sequences corresponded exactly to thatof the complete PCR fragment except that the region from -91 to -64 in the ovalbumin promoter was replaced by sequences from -84 to -61 in the MMTV-LTR. In Vitro Transcription Reactions and Receptor Preparations-All transcription assays were carried out as described in Bagchi et al. (1990a). For oligonucleotide competition analysis all reactions were adjusted to a final DNA concentration of 17 pg/ml with sonicated herring sperm DNA. Chick PR (form A) was purified as before (Klein-Hitpass et al., 1990). Human GRwas prepared as described by Srinivasan and Thompson (1990). Partially purified COUP-TF was prepared as before (Sagami et al., 1986). The resultant fraction contained 0.26 mg/ml of protein and was about 1%pure. RESULTS

NF-1 Is Required for Maximal Glucocorticoid Receptor-dependent Transcription from the M M T V Promoter-To examine enhancer-promoter interactions in a situation where the response elements are in their natural sequence context, the region of the MMTV-LTR from -400 to -20 was cloned upstream of the G-free cassette (pMMTV-400, Fig. 1A). Steroid responsiveness in vivo requires LTR sequences between -180 and -170 and sequences centered at -120 (Buetti and Kuhnel, 1986;Kuhnel et al., 1986),both of which bind purified GR and PR i n uitro (Scheidereit and Beato, 1984; von der Ahe et al., 1985). A third site centered at -70 is reported to be essential for steroid induction and binds the ubiquitous nuclear factor NF-1 (Nowock et al., 1985). pMMTV-400 was transcribed in the presence of a HeLa nuclear extract (Dignam et al., 1983) and humanGR prepared using a baculoviral expression system (Srinivasan and Thompson, 1990; Tsai et al., 1990). Twenty fmol of receptor induced a greater than 10-fold increase in specific transcription without a concomitant effect on activity of the adenovirus major late promoter used as an internal control (Fig. 2 A , lanes 1 and 2, and B ) . Extracts from cells infected with nonrecombinant baculovirus did not enhance specific transcription (Tsai et al., 1990). Transcription from the test template is virtually abolished by an excess of an oligonucleotide containing a GRE/PRE derived from the tyrosine aminotransferase (TAT) gene (Strahle et al., 1987) (lanes 3-5) but is unaffected by the presence of an oligonucleotide bearing a Xenopus vitellogenin A2 gene estrogen response element (Klein-Hitpass et al., 1989) (lanes 6-8). In the absence of added GR, the low basal activity of the promoter is eliminated



‘TATA’

L PPRE20VNF-100

L FIG. 1. Structure of the model promoter templates. A , the promoter regions shown were generated as described under “Materials and Methods” and inserted upstream of the G-free cassette. The TATA regions are represented by filled boxes and thedistal promoter elements by boxes with thin slashes (NF-1 in pMMTV-400) or thick slashes (COUPinpPRE20V-100). The endogenous hormone-response elements of MMTV (Buetti and Kuhnel, 1986; von der Ahe et al., 1985) and the tandem TAT GRE/PREs inserted upstream of the ovalbumin promoter are also shown. The site of specific transcription initiation is indicated by the arrow. B , pPRE,OVNF-100 was constructed as described in “Materials and Methods.”

by competition with an oligonucleotide containing the adenovirus 2 NF-1-binding site (Bradshaw et al., 1988a) (lanes 1 and 9).GR-dependent transcription in the presence of this oligonucleotide is severely curtailed (lanes 2 and l o ) , being reduced by more than &fold compared to activity inits absence (Fig. 2B). These data clearly show that the NF-1 transactivator is necessary for maximal induction of the MMTV-LTR by GR under these i n vitro assay conditions. Furthermore, quantitationof the specific transcripts produced from pMMTV-400 under each condition inFig. 2 allows calculation of the strengthof the resultant synergism between receptor and NF-1 transcription factor-binding sites. Synergism can be regarded to exist between the two binding sites if the activity of both actingtogether is greater the sum of their individual activities (Strahle et al., 1988; Tora et al., 1989; Lin et al., 1990). In Fig. 2B the difference between the activity in the absence of receptor (50X P R E oligo., +GR) plus the activity in the absence of transcription factor (50X NF-1 oligo., +GR) and the combined activity (NONE, +GR) is approximately 4-fold, suggesting that synergism between the GR and NF-1-binding sites contributes to 75% of MMTV promoter activation by GR. COUP TranscriptionFactor Is Dispensable for Activation of the Ovalbumin Promoter by Steroid Receptors-To establish a model template for examination of the role of distal promoter elements in receptor-dependent transcription, and also as a first step toward reproducing the functional complexity of the natural ovalbumin promoter in a cell-free system, we reintroduced ovalbumin gene flanking sequences up to position -100 (pPRE,OV-100, Fig. lA)to the minimal construct reported previously (Klein-Hitpass et al., 1990). The ovalbumin distal promoter contains an essential element centered at -80 (COUP) which binds a transcription factor, COUPTF, found in HeLa cells (Pastorcic et al., 1986; Sagami et al., 1986; Wang et al., 1987) and chick oviduct (Bagchi et al., 1987). Cloning of the factor has revealed that it is a member of the steroid receptor superfamily (Wang et al., 1989). The

Vitro In Synergism between Promoter and Steroid Response Elements A

r

Competitor:

--

GR:

”+

PRE

:: $

ERE

0’ I

-++

,“ $

5907

NF-I

0

1

-++

‘ ::

0‘

-++

-

-4 I

lAdML ”-

1

rrOOAdML

2

5 6 87

4

3

9

10 1 1 12 13 14 15 16 17 18

E 1

3 4

2

5

6 7 8

91011

100

TDeNmAp: l a l e

(ne)

n”

66

n’ w 3

luz - c o -

” 7 Competitor:

~

100 I

66

n“

luz

- n o - n

n. o

I

-GR

* GR

AdML

FIG. 2. Competition analysis of pMMTV-400. A, transcript ion reactions were performedin uitro as described in “Materials and Methods.”Reactionscontained 20 fmol of haculovirus-expressed human glucocorticoid receptor and -50- or 100-fold molar excesses of competitor oligonucleotide (oligo.)as indicated. Specific transcripts are shown for the test template (fillrd arrow) and adenovirus major late promoter internal control ( A d M L . open arroul). H, various low LKR exposures of autoradiograms from A were scanned using an lJltrascan XL laser densitometer. Specific transcripts from the test template were normalized against, those from the adenovirus major as shown. Lanes late promoter and relative activities were plotted containing a 100-fold molar excess of competitor were not included. From repeated experiments synergism between the regulatory elements was estimated at 9.8 ? 0.50-fold ( n = 3 ) (seetext). ERE, estrogen response elements.

1

2

3

4

5

6

7 8 9101112

c s +PR

100 ng

66

ng

33 ng

GRE/PREs in the ovalbumin promoter constructwere maintained 5’ to the promoter, placing their center approximately 50 base pairs (bp) upstream from the COUP element. pPRE,OV-100 was transcribed in a HeLa nuclear extract in the presence of either GR or PR (Fig. 3). T o ensure that all COUP-binding siteswere saturated, and consequently that any observed synergism was maximal, all reactions were supplemented with partially purified COUP-TF prepared as described under “Mat.erials andMethods”;theseconditions result in a slightly higher basal level of transcription. That the quantity added was indeed saturating was determined by I O 0 ng 66 ng PRE titration in an i n vitro transcription assay (data not shown) 0 COUP W N l OF TEMPLATE and by variation of the factor-to-template ratio (seebelow). Specific transcription is stimulated %fold by 2.0 pmol of PR FIG. 3. Competition analysis of pPRE20V-100. A, transcrip(lunes 1 and 10, Fig. 3A) and 5-fold by 30 fmol of GR (lunes tion reactions contained various amounts 01 test template, 2.6 p g of 2.0 pmol of nativechick 1 and 7, Fig. 3B). Receptor-enhanced transcription is elimi- a partiallypurifiedCOUP-TFfraction, nated in both cases by an excess of t.he GRE/PRE oligonucle- progesterone receptor, and a 200-fold molar excessof oligonucleotide (oligo.) as shown. H, as A, except for the ‘replacement of I’R hy 30 otide (lunes 2 and 11, Fig. 3A; lunes 2 and 8, Fig. 3R), but is fmol of GR. C, specific transcripts in lanrs 10-18 of A or lanrs 7-12 only marginally affected by a competitor bearing a mutant of H were quantitated hy scanning the gels in a Retascope 603 hlot GRE/PRE which is functionally inert (data not shown). Like-analyzer (Retagen). Following normalization against the adenovirus major late promoter internal control, data were plotted as before. wise, competition with an oligonucleotide containingthe COUP-TF-binding site (Sagami et al., 1986) virtually elimi- Synergism hetween the GRF,/I’KF,s and COUP was estimated at 1.5 nates transcription in the absence of receptor (lunes 1 and 3, +. 0.22-fold ( n = 6). Fig. 3, A and I?), while a mutant COUP oligonucleotide, to which the factor cannot bind in uitro, compet.es only weakly (Pastorcic et ul., 1986; Sagami et ul., 1986). PR- and GR-dependent transcriptionof pPRE,OV-100 re(data not shown). COUP-TF, therefore, mediates significant basal activity of the ovalbumin promoter in HeLa extracts, as mains relatively high in the absence of COUP-TF (lunes 10 has been previously shown using native promoter templates and 12, Fig. 3A; lunes 7 and 9, Fig. 3R), the reduction in HONE

5908

Vitro InSynergism between Promoter

activity being mainly due to elimination of the basal contribution of the distal promoter element. The data aredisplayed most clearly in Fig. 3C, where the level of specific transcription under each experimental condition has been quantitated and plotted. Activity with both receptor and transcription factor present (NONE) is approximately 1.5 times greater than the sum of the activities of PR or GR alone (+COUP oligo.) and COUP-TF alone (+PRE oligo.).Little or no synergism, therefore, exists between the enhanceranddistal promoter elements in the presence of either PR or GR. The strength of the interaction is unchanged when the ratio of protein to test template is increased by one-third (lanes 4-6 and 13-15, Fig. 3A; lanes 4-6 and 10-12, Fig. 3B; Fig. 3C) or %fold (lanes 7-9 and 16-18, Fig. 3A; Fig. 3C), indicating that all COUP sitesaresaturatedand consequently thatany observed synergism is not underestimated. Furthermore, the amount of synergism is the same even in the absence of supplemented COUP-TF (data not shown), indicating that the amount of transcription factor present in the HeLa extract is sufficient to have a maximal effect. Synergism with GREIPREs Is Dependent upon the Nature of the Dista1,PrornoterElement-The data presented thus far indicate that receptors bound to the tandem response elements at position -100 are capable of activating transcription independently of factors bound at COUP. In thisconstruction, therefore, the distal promoter element is dispensable for gene induction. This suggests that COUP-TF may be inherently unable to interact with GR and PR in thiscontext. Alternatively, since the tandem TAT GRE/PREs represent a relaet al., 1987; tively strong steroid-dependent enhancer (Strahle Klein-Hitpass et al., 1990),their strengthcombined with their proximity to the initiation site in pPRE20V-100 could override any need for downstream transcription factors. However, we did not observe any increased synergism with the COUP element even when the dual GRE/PREs were moved up to 600 bp further upstream from the initiation site, when they were moved400 bp downstream of the initiation site,or when they were replaced with a single GRE/PRE (data notshown). Therefore, weakened enhancer elements do not show an increased dependence upon COUP during activation by steroid receptors. We next decided to examine synergism between the response elements and a distal promoterwhere the COUP-TFbinding site has been replaced by that of the NF-1transcription factor, which has already been shown to act cooperatively within the MMTV promoter (Fig. 2). The resulting construct, pPRE20VNF-100 (Fig. lB), is identical to pPRE20V-100 except that the23-bp regionencompassing the COUP element in the ovalbumin promoter has been replaced by the 23-bp region encompassing the NF-1site in the MMTV-LTR. Constitutive expression of pPRE20VNF-100 is competed by excess NF-1-binding sites but not by COUP oligonucleotides, while the reverse pattern of competition is seen with pPRE20V-100 (data not shown). Thus, NF-1 present in the HeLa extract mediates basal activity of the recombinant promoter. Competition analysis of GR-dependent transcription (Fig. 4A) indicates that synergism between GRE/PREs and NF-1 promoter element is significantly greater than that observed with COUP (Fig. 4B). Synergism of similar strength is seen between the NF-1 promoter element anda single GRE/PRE (data not shown). These values are minimal estimates, since reactions were not supplemented with NF-1 and saturation of all binding sites cannotbe assured. However, it is clear that theextent to which receptors depend upon NF-1 for induction, within the context of the ovalbumin control sequences, is intermediate between that with COUP-TF and

and Steroid Response Elements with NF-1 bound to the MMTV-LTR. Therefore, because substitution of NF-1 for COUP enhances synergism with the GRE/PREs, COUP-TF must synergize poorly with GR and PR due to inherent properties of the factor itself rather than due to characteristics of the enhancer or of other modulating sequences in the promoter. DISCUSSION

In this report we have described a detailed analysis of the role of certain distal promoter elements in enhancer-dependent gene expression. The availability of a specific and sensitive receptor-dependent cell-free transcription system has allowed us to control the level of enhancer- and promoter-binding components in a way that made it possible to quantify functional synergism between the two. Such calculations are useful in determining the extent to which different regulatory elements combine to regulate gene expression under controlled conditions. Our analysis was based upon twomodel promoters: (i) the “natural” MMTV-LTR, containing a diffuse upstream hormone-responsive region and a prominent NF-1binding site in the distal promoter, and (ii) the “artificial” ovalbumin construct, containing discrete GRE/PREs from the TAT gene placed upstream of the first 100 bp of the native chicken ovalbumin promoter, including the COUP element at -80. Theseconstructs allowed examination of enhancer function in different sequence contexts and in combination with different distal promoter elements. Maximal induction of transcription from the MMTV promoter by GR is dependent upon the presence of NF-1 in the transcription reaction (Fig. 2). Removal of NF-1 by oligonucleotide competition reduces overall promoter activity drastically, while synergism between the control elements increases the sum of their individual activities 4-fold. These data derived from cell-free analyses are in general agreement with the published in vivo results of others, which showed that removal of the NF-1site led to theabolition of basal expression and a 10-fold reduction in the level of hormonal stimulation (Buetti and Kuhnel, 1986; Miksicek et al., 1987) and that expression of MMTV in transfected JEG-3 cells was dependent on cotransfection of NF-1 together with receptor coding sequences (Bruggemeier et al., 1990). Furthermore, we have confirmed the in uitro requirement for the NF-1-binding site using a linker-scanning MMTV mutant analyzed previously (Buetti and Kuhnel,1986).’ In uiuo, binding of NF-1andthe TATA factor to the MMTV promoter is only detectable following hormone treatment (Cordingley et al., 1987), indicating that bound steroid receptor facilitates the assembly of downstream transcription complexes including the distal promoter-binding factor. Furthermore, chromatin reconstitutionexperiments suggest that NF-1, unlike the receptor, is unable to bind to the nucleosomally organized promoter. Following receptor binding, it has been postulated that the chromatin structure in promoterproximal regions is perturbed sufficiently enough to allow access by exonuclease I11 and therefore, as a possible consequence, NF-1 (Pinaet al., 1990). Workman et al. (1988) have shown that no chromatin-like structures form on DNA at ratios of HeLa nuclear extract to template similar to those employed here. Thus, our in uitro results, in conjunction with the above, indicate that the MMTV distal promoter element and steroid-responsive enhancer could cooperate at two levels in uiuo: at the level of DNA binding, via alterations in chromatin structure, and at the level of activation, via proteinprotein interactions leading to the stabilization of preinitiation complexes. G. F. Allan and E. Buetti, unpublished data.

I n Vitro Synergism between Promoter and Steroid Response Elements

B.

A. -

Competitor: GR:

+-+ --++ -4

FIG. 4. Competitionanalysis

of ovalbumin/NF-1recombinant promoter. A , 66 ng of pPRE20VNF-100 were assayed with 30 fmol of GR and a 100-fold molar excess of NF-1 oligonucleotide. H , data were quantitated as in Fig. 3. Synergism was estimated a t 2.9 0.52-fold ( n = 4 ) .

5909

2

-8

6

-I>n >

t

2

"