Proc. Natl. Acad. Sci. USA Vol. 87, pp. 9168-9172, December 1990 Biochemistry

Cloned yeast and mammalian transcription factor TFIID gene products support basal but not activated metallothionein gene transcription RAVI KAMBADUR, VALERIA CULOTTA, AND DEAN HAMER Laboratory of Biochemistry, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892

Communicated by Philip Leder, September 6, 1990

We have studied the role of TFIID in the transcription and regulation of the CUPI gene of S. cerevisae. The CUPI gene encodes a small Cu-binding metallothionein that protects cells against Cu toxicity (23). At low, physiological concentrations of Cu, the CUPI gene is transcribed at a low basal level, whereas at high potentially toxic concentrations of Cu, transcription is strongly induced. This response is mediated by the ACEl protein (ref. 24; referred to as CUP2 in refs. 25 and 26). ACEl consists of two domains: an amino-terminal Cu-dependent DNA binding domain and a carboxyl-terminal acidic activation domain (27). At low Cu concentrations, ACEl is constitutively synthesized but cannot specifically bind DNA because it is unfolded. In the presence of excess Cu, the amino-terminal domain undergoes a conformational switch that allows it to specifically bind to multiple sites in the upstream activation region (UAS) (27). Once bound to the promoter, ACEl activates transcription by stimulating the formation of a committed transcription complex (28). Although the activation of CUPI gene transcription by ACEl has been studied both genetically and biochemically (23-31), no information is available on the role of general transcription factors in this process. Here we show by genetic analysis, in vitro transcription reactions, and DNase protection experiments that the transcription of the CUPI gene requires TFIID. However, the cloned yeast and human TFIID gene products alone are incapable of activation by ACEl, suggesting that they lack a component(s) or modification(s) required for transcriptional regulation.

Transcription factor 1ID (TFIID), the ABSTRACT "TATA binding factor," is thought to play a key role in the regulation of eukaryotic transcriptional initiation. We have studied the role of TFIID in the transcription of the yeast metallothionein gene, which is regulated by the copperdependent activator protein ACEL. Both basal and induced transcription of the metallothionein gene require TFIID and a functional TATA binding site. Crude human and mouse TFIuD fractions, prepared from mammalian cells, respond to stimulation by ACEL. In contrast, human and yeast TFIID proteins expressed from the cloned genes do not respond to ACE1, except in the presence of wheat germ or yeast total cell extracts. These results indicate that the cloned TFIHD gene products lack a component(s) or modification(s) that is required for regulated as compared to basal transcription.

The regulated transcription of eukaryotic coding genes requires both gene-specific factors that recognize specific DNA sequence motifs and general initiation factors that interact with common core promoter elements. The best characterized of the general factors is transcription factor IID (TFIID; also known as BTF1 or DB), which binds to the "TATA box" sequences typically found between positions -60 to -120 for yeast genes and at about position -30 for higher eukaryotic genes (1-5). The consensus binding sequence for TFIID is TATAAA/T (2, 6, 7), but a variety of other sequences are also recognized with various affinities (8). The binding of TFIID to the TATA sequence is an early step in the formation of an active transcription complex (9, 10). Until recently, it was difficult to study the precise role of TFIID in transcriptional regulation because of its low abundance and difficulty of purification. The discovery that the yeast Saccharomyces cerevisae contains an abundant activity that can functionally substitute for mammalian TFIID in the transcription of minimal promoter templates (11-13) allowed the purification and cloning of a gene encoding a 27-kDa yeast polypeptide that possesses transcriptional complementation and specific TATA-binding activities (14-18). This polypeptide is encoded by the SPT15 locus, which is essential for normal yeast cell growth and is known to be involved in transcription by virtue of the suppression of Ty element promoter insertions by certain alleles (15). The S. cerevisae gene has been used to clone homologous sequences from fission yeast, Neurospora, Drosophila, and humans (19-22). The human gene encodes a 37.5-kDa polypeptide that possesses TATA-binding activity and complements a TFIID-depleted human extract for transcription from the adenovirus E1B promoter. The yeast and human proteins are 80% homologous in the carboxyl-terminal 181 amino acids but highly divergent in the amino-terminal region.

MATERIALS AND METHODS Mutagenesis, Plasmid Constructions, and in Vivo Assays. Site-directed mutagenesis was performed by the method of Vandeyar et al. (32). For in vivo assays, the mutagenized promoters Were subcloned into the galK indicator plasmid YSK57 and transformed into yeast strain BR10 for galactokinase assays (27). For in vitro analyses, the mutagenized promoters were subcloned into p8CAT (28). The construction of ATATAA was similar to that of 5'A34CAT (33) except that the inserted oligonucleotide contained a Pst I site in place of the TATA sequence at nucleotides -24 to -29 and that the vector was UASc: mTATA:CAT (28). TATAA* and TAGGG* were constructed by inserting the appropriate oligonucleotides into the Pst I site of the ATATAA vector. In these latter constructs, the TATAA or TAGGG sequences are 7 base pairs (bp) upstream of the TATAA sequence in the wild-type promoter. Preparation of Cloned TFIID and ACEl Proteins and Footprinting Reactions. Yeast TFIID was prepared either by SP6 polymerase transcription of pSH227 followed by translation in a reticulocyte lysate (16) or from Escherichia coli carrying the yeast TFIID gene expression plasmid pASY2D (18). The

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: TFIID, TFIIA, etc., transcription factor lID, transcription factor IIA, etc., respectively; UAS, upstream activation region.

9168

Proc. Natl. Acad. Sci. USA 87 (1990)

Biochemistry: Kambadur et al. bacterial TFIID was purified by the method of Buratowski et al. (12) through the Mono S stage. Human TFIID was prepared by bacteriophage T3 RNA polymerase transcription of pKB104 followed by translation in a reticulocyte lysate (19). ACE1 protein was prepared either by translation in a wheat germ extract (27, 28) or by overproduction in bacteria using a phage T7 vector (D.H. and S. Hu, unpublished data). The bacterial Cu-ACE1 was purified by heparin-Sepharose, CM-Sepharose, and Sephadex G-75 chromatography. DNase I footprinting reactions (50 ,ul) Contained 20 mM Tris HCI (pH 8.0), 10% (vol/vol) glycerol, 0.01% Nonidet P-40, 1 mM 2-mercaptoethanol, 5 mM MgSO4, 2.5 mM CaC12, 1% polyvinyl alcohol, 20 pg of poly(dG-dC), 5 fmol (10,000 cpm) of end-labeled DNA probe, and an amount of bacterially produced yeast TFIID predetermined to give an optimal footprint over the TATA box. The reaction mixtures were processed as described (8). In Vitro Transcription and Preparation of TFYI Fractions. In vitro transcription reactions using a mouse nuclear extract were performed and assayed by primer extension as described (28). TFIID-depleted extracts were prepared by passing the nuclear extract, adjusted to 475 mM KCI, over a phosphocellulose column and collecting the flow-through fraction. A crude mouse TFIID fraction was obtained by elution of the phosphocellulose column with 800 mM KCl. The factor was further purified and concentrated by heparinSepharose chromatography (34). A crude human TFIID fraction was similarly prepared from a HeLa cell transcription extract. Micrococcal nuclease-treated wheat germ extract was purchased from Amersham. Yeast whole-cell extracts were prepared by the method of Hahn and Guarente (35).

RESULTS CUP) Gene Transcription Requires a Functional TATA Sequence. To begin our analysis of the role of TFIID in yeast metallothionein gene transcription, we determined whether or not CUPI gene expression requires a TATA sequence. The CUPI sequence is as follows:

9169

Table 1. Expression of TATA sequence mutants in vivo GalK activity + Cu - Cu Sequence Mutant 100 ± 8 (n = 3) 11 ± 2 Wild type 124 ± 9 (n = 3) -95/TAGGGA 12 ± 2 -93/G 7 ± 3 (n = 6) 1 ± 0.4 -77/TAGGAA -75/GG 123 ± 15 (n = 5) 7 ± 3 -45/TAGGAA -43/GG 95 ± 9 (n = 2) 10 ± 1 -33/TAGAT -31/GG 108 ± 19 (n = 4) 8 ± 2 -77/TATAAG -72/G Names and the sequences of the mutant derivatives are shown as well as the relative galK expression levels for the mutants. Data are normalized to a value of 100 for the wild-type promoter in the presence of Cu. Values are the mean ± SD for n transformants.

initiation site at position +1. Fig. lA shows that the -77/ TAGGAA mutant abolished this transcript whereas mutations in the -43 and -31 sequences had no effect. Mutation of the -93 sequence also had no effect (data not shown). Interestingly, the -77/TATAAG mutation had a 2- to 4-fold increased level of transcription, both in this experiment and in assays using TFIID-depleted extract supplemented with either yeast or mouse TFIID (see below). These results show that the CUPI gene requires one specific TATA sequence, located at a typical distance from the initiation site for yeast, for efficient basal and Cu-induced transcription in vivo and in vitro. The requirement for a TATA sequence was further studied using hybrid promoters containing a functional yeast upstream activation region, UASc, linked to TATA and initiation sequences derived from the mouse metallothionein I gene (28). Fig. 1B shows that a construct carrying the wild-type mouse sequences gave rise to a transcript with the same initiation site as for the mouse metallothionein I gene in vivo (33). Replacement of the TATA sequence by a Pst I site, o

A

co

Cb M

Cu:

C)

;

/

C,

C:'

c)

/

t'

-+-+-+-+-+

-75

-93

TATGGATTGTCAGAATCATATAAAAGAGAAGCAAA -43

-31

TAACTCCTTGTCTTGTATCAATTGCATT&TAATAT The CUPI gene promoter contains four potential TATA sequences between the UAS and the initiation site (bold type). The initiation site in yeast is at position +1 and the UASc region is between positions -145 and -104. Each site was altered to TAGGXX, mobilized into a CUPJ:galK plasmid containing the intact CUPI promoter, and assayed for galK activity in cells grown in the presence or absence of Cu (27). Mutation of the sequence -77/TATAAA to the nonconsensus sequence -77/TAGGAA greatly diminished transcription from the CUPI promoter. The levels of both Cu-induced and basal expression were reduced by >10-fold (Table 1). In contrast, mutation of this sequence to TATAAG, which has been shown to be functional in combination with certain UAS elements (36), had no significant effect on expression in vivo. Mutations at the TATA-like sequences at positions -93, -43, and -31 also did not significantly affect expression. The importance of the -77/TATAAA sequence for CUP! gene transcription was also tested in vitro using a nuclear extract from mouse. We have shown (28) that this heterologous extract can specifically transcribe the CUPI gene in the presence of the ACE1 regulatory factor and Cu ions, that the template is transcribed only once, and that the major transcript initiates at position -47 relative to the Yeast in vivo

181

tt.P *Aw

....:: .:.......

B V

v f.

Cu:

0

T

- + -

104-..

+

p - +

"'T, 00 /,T AT - +

-

+ M

4-111

4-104

FIG. 1. Analysis of TATA mutants in vitro. (A) CUP1:CAT construct (lanes wt) and its mutant derivatives, as indicated, were transcribed in a complete mouse extract supplemented with in vitro-synthesized apoACE1 in the presence (+) or absence (-) of 35 ,uM Cu-acetonitrile as described (28). The arrow marks the position of the 181-nucleotide product of CUP1:CAT transcription. The sequences of the mutants are shown in Table 1. (B) UASc:mTATA: CAT construct (lanes wt) and its mutant derivatives were transcribed in the presence (+) or absence (-) of Cu-acetonitrile as above. The arrows mark the 104- and 111-nucleotide transcription products of UASc:mTATA:CAT and UASc:TATAA*:CAT, respectively. M is

Msp l-digested pBR322 fragments with sizes of 160, 147, 123, 110, and 90 bp.

9170

Biochemistry: Kambadur et al.

Proc. Natl. Acad. Sci. USA 87 (1990)

generating construct ATATAA, completely abolished this transcript. In construct TATAA*, in which the TATAA sequence is restored in a position 7 bp upstream of its normal site and with altered flanking sequences, transcription was restored to 10% of the wild-type level and the initiation site was shifted -7 bp upstream. In contrast, a construct with the mutant sequence TAGGG inserted at this position did not restore transcription. These results show that the yeast TATA sequence can be functionally replaced by a mouse TATA sequence in vitro and that the position of this sequence sets the initiation site in a mammalian extract. Binding of TFIID to the TATA Sequence. To determine if the functional TATA sequences act as binding sites for TFIID, we performed DNase I footprinting experiments using yeast TFIID produced in bacteria. Fig. 2A shows that yeast TFIID strongly protected the CUP] promoter over a 16-bp region (positions -79 to -64) encompassing the functidnal TATA sequence -77/TATAAA. Protection was also observed over -33/TATAAT and other A+T-rich sequences in some experiments, but only at higher protein concentrations than were required to protect the -77/TATAAA sequence. Altering the -77/TATAAA sequence to the nonfunctional sequence TAGGAT completely eliminated the footprint. Similar experiments using the UASc-mouse metallothionein I hybrid promoters are shown in Fig. 2B. TFIID, at an appropriate concentration, strongly protected the mouse promoter for =20 bp (positions - 36 to -17) centered over the -29/TATAAA sequence. Deletion of the TATAA sequence eliminated the footprint whereas replacement with TATAA* resulted in a somewhat weaker footprint appropriately shifted upstream. The nonfunctional TAGGG* mutant gave no footprint. These results show that TFIID binds to the

A

B

Cu:. apoACEl:

+

l.

+

+

.

.-

CuACEl:

LI

MMM

TFIID:

Y Nr H H M M (r it)(r) (r) (m) inm

TFIID:

)m) (m) (m)

*MO

2 3 4

1

5

CuACEl: TFIID:

+ H H HHHM M (r) (r) (h) (h) (m) (m) +

P*

2 3 4 5

1

C

A.

B

A

functional TATA sequences and that there is complete correlation between the TATA sequence requirements for transcription in vivo and in vitro and for TFIID binding activity. TFIID Requirement for in Vitro Transcription and Regulation by ACEL. To further study the dependence of CUPI gene transcription on the TFIID factor, in vitro transcription reactions were performed using a TFIID-depleted extract and various sources of TFIID. The mouse transcription extract was passed over a phosphocellulose column to separate a 0.475 M KCI fraction containing RNA polymerase II and factors TFIIA, TFIIB, and TFIIE/F from a 0.8 M KCI fraction containing the TFIID factor and other polypeptides

D Con WGX 4 CuACE1: TFIID: Y Y Y Y (b) (b) (b) t

* -

6 7

8

YX

Con,

y )b y y b~ (b) (bi i5lb)

M

w

*-±

± (3 (3

M

-@

,;;

At

Act

TEIID

H1 ar_HiIO

-W

VIG. 2. Binding of TFMID to the TATA box. (A) Binding to the CUP] promoter. Reaction mixtures contained (+) or lacked (-) bacterially produced yeast TFIID as indicated. The probes were prepared by cutting CUP1:CAT or the -75/GG mutant with HindIII at position +52, end-labeling with [32P]ATP and T4 polynucleotide kinase, and recutting with Spe I at position -227. Lane M contains Msp I-digested pBR322 end-labeled fragments as markers. The position of the -77/TATAAA sequence is indicated. (B) Binding to hybrid 'promoters. The probes were prepared by cutting UASc: mTATA:CAT or its mutant derivatives with HindlII at position + 18, end-labeling with [32P]ATP and T4 polynucleotide kinase, and recutting with EcoRI upstream of UASc. The positions of the -29/ TATAA sequence and the -35/TATAA* sequences are shown.

1 2 3

4

5

6

l

2

3 4

5

6

11

8

FIG. 3. In vitro transcription using various sources of TFI1D. (A) TFIID is required for in vitro transcription. UASc:mTATA was transcribed in a TFIID-depleted mouse extract containing (+) or lacking (-) apoACE1 translated in a wheat germ translation mixture (28), 35 ttM Cu-acetonitrile, and 5 ,ul of mammalian TFIID obtained by phosphocellulose chromatography of a mouse nuclear extract. CON, transcription in the complete nuclear extract in the presence of apoACE1 and Cu. (B) Transcription with cloned human and yeast TFIID gene products compared to a mouse TFIID fraction. Transcription of UASc:mTATA in the TFIID-depleted extracts was conducted in the presence (+) or absence (-) of 1 ,ul of Cu-ACE1 purified from an overproducing bacterial strain. Lanes 1-6 contain reaction mixtures supplemented with 10 Al of reticulocyte lysate translation mixture programmed with yeast [Y(r)] or human [H(r)] or no (-) TFIID mRNA to give a final concentration of TFIID protein of 5 fmol/Al, as estimated by SDS/gel electrophoresis and scintillation counting. Lanes 7 and 8 contain reaction mixtures supplemented with 1 .lI of mouse TFIID [M(m)] obtained by phosphocellulose and heparin-Sepharose chromatography of a mouse L cell nuclear extract. (C) Transcription with cloned human TFIID versus a human TFIID fraction. Transcription reaction mixtures, as in Fig. 2B, contained (+) or lacked (-) Cu-ACE1 and were supplemented as indicated with TFIID translated in a reticulocyte lysate [H(r)], human TFIID from HeLa cells [H(h)], or mouse TFIID from L cells [M(m)]. Lanes 1 and 2 contained 5 .lI of a reticulocyte lysate programmed with human TFIID mRNA, lanes 3 and 4 contained 5 IL1 of human HeLa cell TFIID fraction, and lanes 5 and 6 contained 1 A.1 of mouse L cell TFIID fraction. (D) Reconstitution of regulated transcription with the cloned TFIID gene product. The transcription reaction mixtures containing 0.5 ul of bacterially produced yeast TFIID and containing (+) or lacking (-) bacterially produced Cu-ACE1 were further supplemented as indicated with 2 ul of a whole cell extract from either wheat germ (WGX) or yeast (YX) cells or with extract buffer (Con).

Biochemistry: Kambadur et al. (34). Fig. 3A shows an experiment in which the TFIIDdepleted 0.475 M KCI fraction was supplemented with various combinations of the crude TFIID fraction, apoACE1 protein (translated in a wheat germ extract), and Cu. Only in the presence of all three components was efficient transcription observed, and the ratio of transcription with and without ACEl protein was >10-fold. Similar results were obtained using bacterially produced highly purified Cu-ACE1 and the mouse TFIID fraction further purified by heparin-Sepharose chromatography (Fig. 3B, lanes 7 and 8) or a human TFIID fraction from HeLa cells (Fig. 3C, lanes 3 and 4). Again, efficient transcription was observed only in reaction mixtures containing both TFIID and Cu-ACE1, and the stimulation of transcription obtained with ACEl was >10-fold. Quite different results were obtained when cloned human or yeast TFIID gene products were used in place of the TFIID fractions from mammalian cells. Fig. 3B shows that the human and yeast factors, translated in a reticulocyte lysate, gave rise to a basal level of transcription that was readily detected compared to reaction mixtures supplemented with a control reticulocyte lysate (lanes 3-6 versus lanes 1 and 2). The basal level of transcription supported by the cloned TFIID proteins was 5-fold higher than that obtained with the concentration of crude

mouse

TFIID fraction used in this

experiment (lane 7); however, since the amount of TFIID polypeptide present in the mouse fraction was not determined and since TFIID was limiting in these reactions (see below), the relative efficiencies for basal transcription cannot be quantitatively compared. The addition of Cu-ACE1 to reaction mixtures containing the cloned TFIID proteins did not affect transcription, even though it strongly stimulated transcription in parallel reactions containing the mouse TFIID fraction (lanes 7 and 8). This was not due to an ACEl inhibitor in the reticulocyte lysate, since reactions containing the mouse TFIID fraction showed a good response to ACEl in the presence of an equivalent amount of lysate (data not shown). The lack of response of the cloned TFIID proteins to ACEl could in principle be due to a species difference since the cloned proteins were obtained from human or yeast whereas the crude fraction was obtained from mouse. To test this possibility, we compared cloned human TFIID to a human TFIID fraction from HeLa cells. Fig. 3C shows that reaction mixtures containing the human TFIID fraction responded well to ACEl whereas reaction mixtures containing human TFIID synthesized in vitro gave a high basal level of transcription and failed to respond to ACEL. These results led us to consider the possibility that the cloned TFIID gene products lack a component(s) or modification(s) that is necessary for activation by Cu-ACE1. To restore regulation, we supplemented reaction mixtures containing bacterially produced yeast TFIID with whole cell wheat germ or yeast extracts (Fig. 3D). In both cases, the extracts strongly reduced the level of basal transcription and partially restored regulation by Cu-ACE1. The levels of Cu-ACE1 stimulation in such reactions were typically 4- to 7-fold, compared to 10- to 50-fold in reaction mixtures containing the crude human or mouse TFIID fractions. The activity in the wheat germ extract appeared to be labile as it was destroyed by a 1-hr incubation at room temperature under translation conditions (data not shown). A trivial explanation for the above results could be that the cloned TFIID polypeptides were present in excess, thereby eliminating the need for ACEL. Table 2 shows that this was not the case. Even in reaction mixtures containing low concentrations of bacterially produced yeast TFIID or of in vitro-synthesized human TFIID, such that basal transcription levels were close to those observed with the mouse and human TFIID fractions, the addition of ACEl stimulated transcription no more than 1.8- to 2.5-fold. This shows that the lack of regulation in reactions programmed with the

Proc. Natl. Acad. Sci. USA 87 (1990)

9171

Table 2. In vitro transcription with various concentrations of cloned TFIID gene products ACE+ + TFIID ratio Exp A Mouse (m) (1.8 Al) 2 100 50 Yeast (b) (0.04 Al) 0.8 1.4 1.8 Yeast (b) (0.5 Al) 61 79 1.3 B Human (h) (5 Al) 4 100 25 Human (r) (1 ,ul) 6 15 2.5 Human (r) (10 AI) 38 27 0.7 Transcription reactions, as in Fig. 3, contained or lacked CuACE1 purified from bacteria and the indicated amount of mouse TFIID fraction from L cells (mouse), yeast TFIID from bacteria (yeast), human TFIID fraction from HeLa cells [human (h)], or human TFIID translated in a reticulocyte lysate [(human (r)]. Transcription levels were quantitated by laser densitometry of the autoradiogram and normalized to a value of 100 for mouse TFIID in the presence of ACEL. m, from mouse; b, from bacteria.

cloned TFIID polypeptides is not simply due to the presence of excess transcription factor.

DISCUSSION We have shown that the transcription of the yeast metallothionein gene requires the general transcription factor TFIID and its TATA binding site. Evidence for the critical role of TFIID is 3-fold. (i) Mutations in the -77/TATAAA sequence of the CUPI promoter or in the -29/TATAAA sequence of a yeast-mouse hybrid promoter greatly reduce transcription. (ii) In vitro transcription of the yeast and hybrid metallothionein genes in a TFIID-depleted cell extract requires the addition of TFIID, which can be supplied either from mammals or yeast. (iii) Yeast TFIID binds to the functional TATA sequences of the yeast and hybrid promoters but not to nonfunctional mutant derivatives. In vitro transcription reaction mixtures containing crude human or mouse TFIID fractions showed a strong response to ACEl whereas reaction mixtures containing cloned human and yeast TFIID proteins did not. In addition, the cloned proteins gave higher basal levels of transcription than the crude fractions under the reaction conditions typically used. The lack of responsiveness of the cloned TFIID was not due solely to species differences since this effect was observed using cloned and crude TFIID both derived from human. It was also not due to saturation of the reactions with TFIID, since equivalent results were obtained over a range of factor concentrations. Rather, we propose that the cloned gene products lack a component(s) or modification(s) required for regulated as compared to basal transcription. Three speculations concerning the nature of this "missing link" are shown in Fig. 4 and discussed below. As shown in Fig. 4A, there could be a "bridge protein(s)" or "coactivator(s)" that is present in the mammalian-cell TFIID fractions but not in the cloned gene products. The function of the "bridge protein" would be to mediate interactions between ACEl and TFIID, whereas a "coactivator" might also contact additional transcription factors. This class of model would explain why transcription reaction mixtures containing the cloned TFIID proteins do not respond to stimulation by ACEL. As shown in Fig. 4B, there could be an "anti-TFIID protein(s)" whose negative effect on TFIID is alleviated by ACEL. Such a protein could, for example, block the activation function of TFIID in the absence of ACE1, similar to the ability of GAL80 to block GAL4 in the absence of galactose (37). Alternatively, the protein could specifically or nonspecifically bind to the TATA sequence, thereby restricting access of TFIID to the template unless ACEl is present. This

9172

Proc. Natl. Acad Sci. USA 87 (1990)

Biochemistry: Kambadur et al.

transcription extract. We thank C. Klee, Carl Wu, members of the Hamer lab, and the reviewers for comments on the manuscript.

A

Bridge or Coactivator

AC'E F

I

E I 1

UASc

TATA

1. Breathnach, R. & Chambon, P. (1981) Annu. Rev. Biochem. 50, 349-383. 2. Chen, W. & Struhl, K. (1985) EMBO J. 4, 3273-3280. 3. Hahn, S., Hoar, E. & Guarente, L. (1985) Proc. Natl. Acad. Sci. USA 82, 8562-8566.

4. Nagawa, F. & Fink, G. (1985) Proc. Natl. Acad. Sci. USA 82,

B

Anut'

i/Anti

VTFIIDe lTEIIDj

AEft ACEI1i UASc c _

ACE

UASc

8557-8561. 5. McNeil, J. B. & Smith, M. (1986) J. Mol. Biol. 187, 363-378. 6. Myers, R., Tilly, K. & Maniatis, T. (1986) Science 232, 613-618. 7. Chen, W. & Struhl, K. (1988) Proc. Natl. Acad. Sci. USA 85, 2691-2695. 8. Hahn, S., Buratowski, S., Sharp, P. & Guarente, L. (1989) Proc. Natl. Acad. Sci. USA 86, 5718-5722. 9. Buratowski, S., Hahn, S., Guarente, L. & Sharp, P. (1989) Cell 56, 549-561.

10. Workman, J., Roeder, R. & Kingston, R. (1990) EMBO J. 9, 1299-1308. 11. Cavallini, B., Huet, J., Plassat, J., Sentenac, A., Elgy, J. & Chambon, P. (1988) Nature (London) 334, 77-80. 12. Buratowski, S., Hahn, S., Sharp, P. & Guarente, L. (1988) Nature (London) 334, 37-42.

13. Horikoshi, M., Wang, C., Fujii, H., Cromlish, J., Wei, P. & Roeder, R. (1989) Proc. NatI. Acad. Sci. USA 86, 4843-4847.

FIG. 4. (A-C) Three speculative models (described in text) for transcription regulation by ACE1 and TFIID.

model would explain why the cloned TFIID proteins give higher basal transcription levels than the mammalian cell fractions and why yeast and wheat germ whole cell extracts repress basal transcription and partially restore ACEl regulation. As shown in Fig. 4C, TFIID might undergo a modification(s) that is necessary for regulation by ACEL. For example, TFIID might undergo a critical phosphorylation or glycosylation that does not occur in bacteria or in the reticulocyte lysate. Alternatively, TFIID might undergo a modification, such as multimerization, which allows it to bind to promoter sites other than the TATA sequence; in this case, the function of ACEl might be to displace or transport TFIID to the functional TATA sequence. Our results clearly show that TFIID can support basal transcription independently of its ability to respond to ACEL. The availability of pure TFIID and ACEl proteins may facilitate the search for additional TFIID cofactors, subunits, or modifying activities that mediate interaction with this type of acidic activator protein. After this paper was submitted, it was reported that transcriptional activation by the Spl, CTF, and USF factors also requires "coactivator" or "adaptor" proteins and that different cofactors may be involved for different classes of activator protein (38-41). It was also shown that a GAL4-VP16 hybrid protein can repress activated transcription from an unrelated UAS sequence, presumably by titrating a positively acting TFIID cofactor as proposed in Fig. 4A (42, 43). Our results do not rule out the possibility that TFIID can interact with acidic activator proteins directly (44), since transcription might require another factor to activate or different activators might have different factors. Rather, they suggest that different types of interactions between TFIID and activator proteins contribute to the diversity of eukaryotic transcriptional regulation. R.K. and V.C. contributed equally to this paper. We thank C. Kao, M. Schmidt, and A. Berk for providing the human TFIID clone prior to publication and for the yeast TF1ID clone, S. Hahn for the yeast TF1ID in vitro transcription plasmid, and M. Falzon for HeLa cell

14. Cavallini, B., Faus, I., Matthes, H., Chipoulet, J., Winsor, B., Elgy, J. & Chambon, P. (1989) Proc. Natl. Acad. Sci. USA 86,9803-9807. 15. Eisenmann, D. M., Dollard, C. & Winston, F. (1989) Cell 58, 1183-1191. 16. Hahn, S., Buratowski, S., Sharp, P. & Guarente, L. (1989) Cell 58, 1173-1181. 17. Horikoshi, M., Wang, C., Fujii, H., Cromlish, J., Weil, P. & Roeder, R. (1989) Nature (London) 341, 299-303. 18. Schmidt, M., Kao, C., Pei, R. & Berk, A. (1989) Proc. Natl. Acad. Sci. USA 86, 7785-7789. 19. Kao, C., Lieberman, P., Schmidt, M., Zhou, Q., Pei, R. & Berk, A. (1990) Science 248, 1646-1650. 20. Lewin, B. (1990) Cell 61, 1161-1164. 21. Fikes, J., Becker, D., Winston, F. & Guarente, L. (1990) Nature (London) 346, 291-294. 22. Hoffmann, A., Horikoshi, M., Wang, C., Schroeder, S., Weil, P. & Roeder, R. (1990) Genes Dev. 4, 1141-1148. 23. Hamer, D., Thiele, D. & Lemont, J. (1985) Science 228, 685-690. 24. Thiele, D. (1988) Mol. Cell. Biol. 8, 2745-2752. 25. Welch, J., Fogel, S., Buchman, C. & Karin, M. (1989) EMBO J. 8, 255-260.

26. Buchman, C., Skroch, P., Welch, J., Fogel, S. & Karin, M. (1989) Mol. Cell. Biol. 9, 4091-4095.

27. Furst, P., Hu, S., Hackett, R. & Hamer, D. (1988) Cell 55,705-717. 28. Culotta, V. C., Hsu, T., Hu, S., Furst, P. & Hamer, D. (1989) Proc. Natl. Acad. Sci. USA 86, 8377-8381. 29. Furst, P. & Hamer, D. (1989) Proc. Natl. Acad. Sci. USA 86, 5267-5271.

30. Huibregtse, J. M., Engelke, D. & Thiele, D. (1989) Proc. NatI. Acad. Sci. USA 85, 65-69. 31. Szczypka, M. & Thiele, D. (1989) Mol. Cell. Biol. 9, 421-429. 32. Vandeyar, M. A., Werner, M., Hutton, C. & Batt, C. (1988) Gene 65, 129-123. 33. Culotta, V. C. & Hamer, D. (1989) Mol. Cell. Biol. 9, 1376-1380. 34. Dignam, J., Martin, P., Shastry, B. & Roeder, R. (1983) Methods Enzymol. 101, 582-598. 35. Hahn, S. & Guarente, L. (1988) Science 240, 317-319. 36. Harbury, P. A. & Struhl, K. (1989) Mol. Cell. Biol. 9, 5298-5304. 37. Ptashne, M. (1988) Nature (London) 335, 683-689. 38. Peterson, M., Tanese, N., Pugh, B. & Tjian, R. (1990) Science 248, 1625-1630. 39. Hoey, T., Dynlacht, B., Peterson, M., Pugh, B. & Tjian, R. (1990) Cell 61, 1179-1186. 40. Pugh, B. & Tjian, R. (1990) Cell 61, 1187-1197.

41. Hoffmann, A., Sinn, E., Yamamoto, T., Wang, J., Roy, A., Horikoshi, M. & Roeder, R. G. (1990) Nature (London) 346, 387390.

42. Berger, S., Cress, W., Cress, A., Trizenberg, S. & Guarente, L. (1990) Cell 61, 1199-1208. 43. Kelleher, R., Flanagan, P. & Kornberg, R. (1990) Cell 61, 12091215. 44. Stringer, K., Ingles, C. & Greenblatt, J. (1990) Nature (London) 345, 783-786.