Cell, Vol. 65, 493-505,

May 3, 1991, Copyright

0 1991 by Cell Press

Direct Interaction between Spl and the BPV Enhancer E2 Protein Mediates Synergistic Activation of Transcription Rong Li, Jonathan D. Knight, Stephen P. Jackson,* Robert Tjian, l and Michael R. Botchan Department of Molecular and Cellular Biology Division of Biochemistry and Molecular Biology *Howard Hughes Medical Institute University of California Berkeley, California 94720

Summary The physical interaction of heterologous site-specific DNA-binding proteins is an important theme in eukaryotic transcriptional regulation. In this paper, we show that the cellular transcription factor Spl and the BPV-1 (bovine papillomavirus type 1) enhancer protein E2 activate transcription synergistically from two papilloma viral promoters and a series of synthetic promoter constructs in transient transfection experiments. Furthermore, Spl can target E2 to a promoter region even in the absence of a specific E2 DNA-binding motif. Biochemical experiments establish that Spl enhances E2 binding to its sites and that the two proteins form a specific complex. Spl sequesters distally bound E2 to the promoter region by formation of stable DNA loops, visualized by electron microscopy. These experiments substantiate the notion that enhancer binding proteins are targeted to promoter regions by direct interaction with proteins that bind proximal to the transcriptional start site. Introduction The regulation of metazoan transcription is governed by the combined action of various sequence-specific DNAbinding proteins (Dynan and Tjian, 1985; Ptashne, 1988; Mitchell and Tjian, 1989). The binding sites for some regulatory proteins, such as Spl (Dynan and Tjian, 1983), ATF (Lee et al., 1987), and NFl (Jones et al., 1987), are generally located proximal to the start site, and the factors themselves are expressed in a wide variety of cell types. By contrast, other site-specific DNA-binding proteins, which regulate spatial and temporal patterns of gene expression, bind regulatory elements either at remote positions (enhancers) or intermingled with the sites for the ubiquitous promoter elements (Miiller and Schaffner, 1990). A striking characteristic of many eukaryotic DNAbinding activators is their ability to activate gene expression synergistically. Multimerization of the DNA-binding sites for many (but not all) transcription factors, either in homologous or heterologous combinations, greatly enhances the strength of a promoter. Characteristically, the combination of multiple elements results in significantly more potent activation than would the sum of the activation potentials of the individual parts. The basic mechanisms that underlie this synergism are of fundamental importance, but remain unknown. Moreover, this issue is inti-

mately related to an important regulatory problem. Since the general transcription machinery appears to be conserved in all cell types, the rules that govern synergistic interactions between the site-specific DNA-binding proteins will probably limit the extent to which permutations of a finite number of such factors can be used to generate a large diversity of tissue-specific promoters. It appears that certain War/s-activators can work synergistically in combination with many other trans-activators. For example, derivatives of the yeast activator GAL4 can stimulate transcription in synergy with various mammalian activators such as ATF, USF, and the glucocorticoid receptor (Lin et al., 1988; Kakidani and Ptashne, 1988; Carey et al., 1990; Lin et al., 1990). Given the promiscuity of these interactions, it is reasonable to postulate that a third factor mediates the synergistic interactions between these DNA-binding factors, even though postulating a third factor with universal recognition capabilities is no more parsimonious than imagining a common interacting domain(s) on each DNA-binding protein. Another view of synergy posits that certain combinations of factors work together in particular aspects of the transcription reaction. In fact, results from work on U2 gene transcription (Tanaka et al., 1988) “composite” glucocorticoid-responsive elements (Diamond et al., 1990) and the GT-1 enhancer motif (Fromental et al., 1988) provide good examples of how certain trans-activators act synergistically only in specific combinations. It is relevant to point out that the protein targets for some of these DNA-binding factors are not extremely promiscuous, but show species or factor specificities (Pugh and Tjian, 1990). Additional study is required to resolve this dichotomy of ideas on the nature of interactions between enhancer binding proteins and promoterproximal binding factors. In this paper, we describe the synergistic interactions between the bovine papillomavirus type 1 (BPV-1) protein E2 and the cellular transcription factor Spl . The E2 protein is an essential viral enhancer factor that binds as a dimer to a specific DNA sequence and is capable of activating transcription both in vivo and in vitro (Spalholz et al., 1985, 1987; Androphy et al., 1987; Moskaluk and Bastia, 1987; Dostatni et al., 1988; McBride et al., 1988; Li et al., 1989; Dostatni and Yaniv, personal communication). It has been established that E2 binding elements display synergistic enhancer activity when positioned in tandem or when widely separated (Spalholz et al., 1988; Thierry et al., 1990). This correlates with its in vitro binding cooperativity and the ability of the protein to form DNA loops upon binding to two well-separated sites (Knight et al., 1991). The amino-terminal part of the 410 amino acid protein is predicted to be an acidic amphipathic a helix and is important for transcriptional activation (Giri and Yaniv, 1988; McBride et al., 1989). The DNA-binding and dimerization domain has been localized to the carboxy-terminal 100 amino acids (McBride et al., 1988, 1989). To explore the functional synergism between E2 and Spl, we have utilized both BPV promoters and a series of

Cell 494

1

Figure

2

3

45

6

7

8

9

10

11

1. Map of the BPV-1 Genome

Solid boxes represent the 17 E2 binding sites found in the genome. The open bars lettered El to E8 represent region. Five promoters (PI-P5) in this region are also shown. The upstream regulatory region (URR), which (EZRE), is enlarged at the bottom.

synthetic gene constructs in transient transfection assays. Various fusion proteins have been used in cotransfection assays to verify the specific interaction between the two proteins. To address the question of whether the synergism observed in vivo involves direct protein-protein interaction, we have applied multiple biochemical approaches, such as cross-linking and coimmunoprecipitation, DNAase I footprint analysis, and electron microscopy. Our results show that direct interaction between these two heterologous DNA-binding proteins is likely to explain, in part, their synergistic activity. Results Spl Binding Sites Are Important for E2 Stimulation of BPV Promoters In the BPV viral DNA, E2 stimulates several widely separated promoters through EP-responsive enhancer elements (E2RE) in the upstream regulatory region (URR) (see Figure 1; Spalholz et al., 1987; Haugen et al., 1987; Hermonat et al., 1988; Prakash et al., 1988; R. L. and M. B., unpublished data). One of these E2-regulated promoters, P4 (or P2440), which is located 2.4 kb away from the enhancer, expresses the product of E2 itself (Spalholz et al., 1985; Hermonat and Howley, 1987; Vaillancourt et al., 1990) as well as the major BPV transforming protein E5 (Prakash et al., 1988). Another viral promoter, P5 (or P3089), internal to the E2 open reading frame (ORF), is responsible for the expression of the C-terminal half of the E2 polypeptide (E2-C) (Stenlund et al., 1985; Baker and Howley, 1987). E2-C antagonizes E2 activation, as well as repressing the basal level of transcription (Lambert et al., 1987; Choe et al., 1989). While studying the BPV-1 P5 promoter, we observed that the PSCAT construct (shown in Figure 2A) can be activated by E2, even in the absence of the E2 binding sites located in the URR, although the E2 stimulation is greatly enhanced by the URR sequences placed in either orientation downstream of the chloramphenycol transacetylase (CAT) gene (data not shown). The construct PSCAT

the eight ORFs in the BPV-1 early bears the E2-responsive elements

allowed us to ask what role the weak E2 binding sites BS16 and BS17 might play in modulating this response. Surprisingly, point mutations that abolish both E2 binding sites (Figure 2A, 1617CAT) decrease, but do not eliminate, P5 responsiveness to E2 activation (Figure 28, top panel). In contrast, a chimeric protein that contains the VP1 6 activation domain fused to the E2 DNA-binding domain stimulates P5 transcription in a strictly DNA-site-dependent manner (Figure 2B, bottom panel). This behavior of E2 is reminiscent of previous reports showing that E2 can stimulate transcription from a group of viral and cellular promoters that lack obvious E2 binding motifs (Haugen et al., 1988; Heike et al., 1989). We reasoned that E2 may be targeted by another factor that is bound to all of these promoters. Interestingly, Spl binding sites have been identified in most of these promoters. Indeed, for E2 siteindependent stimulation of certain lymphokine genes, a GC box element is contained within the sequences that have been shown to mediate E2 responsiveness (Heike et al., 1989). Furthermore, we demonstrated that Spl binds to a DNA sequence between BS16 and BS17 in the P5 promoter (Figure 2C, lanes 3,4, and 5.). Although no consensus Spl binding sites (Dynan and Tjian, 1985; Kadonaga and Tjian, 1986) were found in this region, the two related sequences GGGTGG and AGGTGG were present. To establish that these two sequences did specify Spl recognition, two small deletions (indicated by the brackets in Figure 2A) were introduced into the P5 promoter region, thus creating the construct P5SMCAT. These deletions abolished Spl binding, as shown by the DNAase I footprint assay (Figure 2C, lanes 7, 8 and 9). It is difficult to establish unequivocally that an in vitro Spl binding site actually functions as such in vivo. To explore this point further, we transfected the PSCAT construct into the Drosophila Schneider cell line, which lacks endogenous Spl but is fully responsive to exogenous Spl . As shown in Figure 2D, the P5 promoter is readily stimulated by Spl, confirming that P5 is indeed an Splresponsive promoter. As expected, the CAT plasmid bearing the deletion mutant of Spl binding sites, namely,

Functional 495

Spl(ng)

Interaction

between

E2 and Spl

0 50 100 0 50 100 III PSCAT PSSMCAT Schneider cells

Figure 2. Spl Stimulates Transcription Is Important for E2 Stimulation

EZWd

0 0.1 0.5 0 0.1 0.5

III PSCAT cv-1

PSSMCAT cells

of the BPV P5 Promoter

and

(A) Diagram of the PBCAT, 1617CAT, and P5SMCATconstructs. Numbers 16 and 17 represent the two E2 binding sites in the P5 promoter region. The DNA sequence shown here is the region protected by Spl in the DNAase I footprint assays. The bracketed regions underneath are the nucleotides deleted in PBSMCAT plasmid. The point mutations in E2 binding sites 16 and 17 are described in Experimental Procedures. These mutations abolished the E2 binding to both sites (data not shown), The open and filled bars represent the wild-type and mutant binding sites, respectively. (B) Quantitation of the cotransfection results showing the E2 and VP16-E2 stimulation of transcription from P5CAT and 1617CAT. The x-axis denotes the micrograms of E2 or VP16-E2 expression vector used in each transfection. The y-axis denotes the percentage of acetylation of chloramphenicol. The amount of reporter plasmids used was kept at 5 ug in each transfection. Both the E2 and VP16-E2 expression vectors are driven by the cytomegalovirus promoter. The E2 gene used in CAT assays in this paper contains a mutation that changes the initiator ATG (Met) of E2-C to ATC (lie) (Stenlund and Botchan, 1990). (C) DNAasel footprint assay showing Spl binding in the P5 region. Lanes l-5, DNA fragment bearing wild-type P5 region from Ncol (nucleotide 2878) to Fspl (nucleotide 3023); lanes 6-10, the same restriction fragment from the mutant construct PBSMCAT. In both cases, the fragments were labeled with [y-3ZP]ATP at the Ncol site. Lanes 1 and 10, AG sequencing ladder; lanes 2 and 6, no protein control; lanes 3 and 7,80 ng of Spl was added; lanes 4 and 8,27 ng of Spl ; lanes 5 and 9, 9 ng of Spl. (D) Cotransfection assay showing Spl stimulation of P5 transcription. The indicated reporter plasmids (5 pg) were transfected either with or without Spl expression vector into Drosophila Schneider cells. The percentage of acetylation of chloramphenicol is indicated at the top of each lane for this experiment and in all subsequent CAT assays. Spl was expressed from the promoter for the Drosophila actin gene promoter. (E) Cotransfection assay performed in CV-1 cells illustrating the importance of the Spl sites for E2 stimulation. In each experiment, 5 ng of reporter plasmids was used.

P5SMCAT, did not respond effectively to Spl . To determine if the Spl sites in P5 are important for E2 activation, we introduced the Spl site mutants, along with E2, into mammalian CV-1 cells. Figure 2E demonstrates that the wild-type construct PSCAT is stimulated 40-fold by E2, while the mutant PSSMCAT is stimulated only 4-fold. Repeated experiments over a range of E2 concentrations showthat the P5SMCATconstruct isseverely limited in its responsiveness to E2, although some low level activity is observed. This point is discussed further in the section of Results concerning the physical interaction of E2 and Spl . These experiments indicate that the presence of the Spl binding sites is important for the specific activation of the P5 promoter mediated by E2. The finding of a novel Spl recognition sequence of functional significance in the P5 promoter region prompted us to scrutinize the BPV-1 genome for similar motifs. A computer search revealed another GT box at the BPV P4 promoter region (see Figure 3A). DNAase I footprint that Spl indeed binds to this sea48says demonstrated quence (Figure 3B, lanes 2 and 3). To assess the functional significance of this sequence on the basal promoter activity of P4, a single point mutation was generated at one of the most conserved nucleotides in Spl binding motifs (the G to C transition indicated in Figure 3A). The mutation does not change any codon assignment in the overlapping El ORF. As predicted, the mutation abolishes Spl binding (Figure 38, lanes 5 and 6). Intact BPV plasmid DNA bearing this mutation was tested for its transforming ability in mouse Cl27 cells. In three independent experiments, 1 u.g of the wild-type BPV-1 DNA (pMLBPV-100) gave rise to 66,214, and 100 transformed cell colenies, respectively, whereas only 3,5, and 0 morphologically transformed cell colonies could be detected with the mutant DNA (pMLBPV-1OOSM). Given the fact that the P4 promoter provides the major avenues for the expression of the E2 activator and the E5 oncogene, we surmised that this point mutation must have severely disrupted the promoter function. To measure basal expression levels from the P4 promoter, we used the BPV-E5 CATvector described by Prakash et al. (1988). In this vector, the CATgene replaces the E5 ORF in the BPV-1 genome. The CAT gene is therefore expressed from P4 through aspliced mRNA, as illustrated in Figure 3A. This mRNA reflects the most abundant BPV-1 mRNA (Stenlund et al., 1985; Yang et al., 1987) and as shown by Prakash et al. (1988), deletions of the P4 promoter essentially eliminate CAT activity in the construct. Furthermore, it has been established that this promoter is activated by E2 and requires URR sequences for this stimulation (Hermonat et al., 1988; Prakash et al., 1988). The wild-type and the Spl binding mutant plasmids (E5SMCAT) were transfected in side by side assays into CV-1 cells. The basal-level activity measured at 20 pg of the input plasmid is almost completely abolished with the mutant (Figure 3C). The BPVE5CAT construct is essentially an intact BPV-1 genome, and it is therefore likely that the construct itself produces E2, accounting for the nonlinear increase in the CAT signal when the reporter plasmid goes from 5 to 20 pg. To show that this particular

Cell 496

A URR

Spl

P4

CAT

Figure 3. Spl Binding Sites at BPV P4 Region Are Crucial for the Promoter Function

(A) Schematic diagram of the structure of BPVEBCAT, the P4 reporter plasmid. This plasmid contains the whole BPV early region with the CAT gene fused in and replacing the E5 CRF. AGGAGGGGTGGAG Depicted here is the URR sequence, which bears the E2-responsive elements, the P4 proC moter, and the Spl site in that region. D, donor, A, acceptor. The dotted lines represent the splicing pattern that leads to the CAT tranC scripts. The DNA sequence of the bottom (“lo) 12 1.1 13 9.2 9.7 0.6 0.6 1.6 3.1 1.4 strand (5’to 3’) shows the Spl binding site and 123 45 67 the nucleotide mutated in E5SMCAT plasmid. (B) DNAase I footprint assay showing the Spl binding to the P4 promoter region. Lanes 1, 2, and 3, fragments of wild-type BPV sequence from EcoRl (nucleotide 2113) to Sphl (nucleotide 2617). Lanes 4, 5, and 6, same fragment 0 0 0.2 0.5 1.0 0 0 0.2 0.5 1.0 but with the point mutation introduced into the Spl site (EBSMCAT). In both cases, the frag20 5 5 5 5 20 5 5 5 5 ments were labeled with [Y-~~P]ATP at the II EcoRl site. Because of the distant location of E5SMCAT BPVE5CAT the Spl site from the labeled end, AG sequence ladders do not resolve well around the region of interest. The same labeled fragment was digested with BstEll, which fortuitously cut D (“lo) 1.2 12 6.3 0.8 9.1 25 at the Spl site; the BstEll-EcoRI fragment was then used as a marker (lane 7). Lanes 1 and 4, no protein control; lanes 2 and 5,60 ng of Spl ; lanes 3 and 6, 27 ng of Spl. (C)An Spl site is required for P4 transcription. The indicated reporter plasmids were introduced into CV-1 cells either with or without an E2 expression vector. To compare the basal VP16/E2(ug) 0 0.2 0.5 0 0.2 0.5 level of transcription between the wild-type and the mutant P4 reporter plasmids, 20 ug of DNA uu of each plasmid was transfected into the cells. BPVEBCAT E5SMCAT To test for E2 stimulation, the amount of reporter plasmid used was kept at 5 ug in each reaction. Cells transfected with only carrier DNA gave rise to a background of 0.55% acetylation. (D) The fusion protein VP16-E2 can stimulate P4 transcription independent of the Spl site. Both the wild-type and the mutant P4 CAT reporter plasmids (5 pg) were transfected into CV-1 cells along with the VP16-E2 plasmids at the indicated amounts.

BPV-E5CAT

GT motif flanking the P4 promoter responds directly to Spl , we synthesized a 22 base oligonucleotide containing the wild-type or the mutant Spl site (from nucleotide 2387 to nucleotide 2409 of the BPV-1 genome) and cloned it in front of the minimal promoter of the HSV (herpes simplex virus) TK gene (from -37 to +51). The construct bearing the wild-type oligonucleotide, but not the mutant one, can be readily stimulated by exogenously added Spl in Drosophila Schneider cells (data not shown). These studies are in agreement with the recent published work of Spalholz et al. (1991); that work has shown that the zinc finger domain of the Spl protein binds to the GT motif at the P4 promotor and that Spl is involved in its regulation. Some discussion is required to analyze the data addressing the importance of this particular Spl site for E2 stimulation of P4. The Spl binding site shown in Figure 3 overlaps the BPV-1 E2 binding site 15, which is the weakest one in the viral genome (Li et al., 1989). In fact, the G-C transition depicted in Figure 3 should increase the binding of E2 to this site (Spalholz et al., 1991), as the mutant site now contains the sequence ACCAGGAGCGGT, where the match is closer to the consensus E2 binding

motifs (ACCGNJGGT). Interestingly, as shown in Figure 3C, this mutant is not stimulated by E2. Thus, it would tentatively appear that the Spl binding is more important than E2 binding. This point is corroborated by another mutation, which destroys E2 binding at this site but leaves Spl binding intact. This particular mutant converts the wild-type sequence BAGGAGGWGG to &CJIGGAGGmGG (the nucleotides in bold face depict the Spl site, while the underlined nucleotides show the E2 recognition sequence). In this case, basal activity is unaffected, and the E2 activation is at wild-type level (data not shown). In agreement with these results, Howley and colleagues have recently concluded that the E2 binding sites in the URR region are critical for E2 stimulation of P4, but BS15 is not (Spalholz et al., 1991). The results shown in Figure 3D provide another important control for these experiments; a VP18-E2 fusion protein is capable of activating this mutant promoter (E%MCAT), indeed, at higher levels than it does the wild-type promoter. This shows that the mutation which disrupts the Spl binding site does not inadvertently create an E2 repressor binding site and that unlike E2, VP1 6-E2 can stimulate P4 transcription inde-

Functional 497

Interaction

between

4xSpl

(%)

Ww)

Sp1E2

e

SplMEP

w

tkE2

4-w

tkME2

^I-/

1.7

75

96

0

20

100

E2 and Spl

TATA?

0.66 2.9 7.2

0

CAT

5xE2

0.7

4.0

3.4 0.72

0

20

100

20 100

0

1.4

1.0

20

100

uuuu SplE2

Figure 4. Synergistic Proximal GC Boxes

Spl ME2 Activation

between

tkE2 Distal E2 Binding

tkME2 Sites and

(Top) Diagrams of various CAT constructs. “4 x Spl” and “5 x E2” rep resent the repeats of these binding sites at the indicated positions, respectively. The shaded boxes are the mutant E2 binding sites. The minimal promoter region of the HSV TK gene contains only the TATA element and the transcription start site (from sequence -37 to f51). (Bottom) These reporter constructs (5 pg) were transfected into CV-1 cells either with or without the E2 expression vector, and CAT assays were performed. In the case of Spl E2, the signal is well beyond the linear response of the assay. From multiple assays within the linear range, we estimate that the E2 induction is approximately lOO-fold.

pendent of the Spl site. We therefore conclude that in the case of P4, as in the case of P5, an Spl site is critical for both basal-level activity and EPspecific activation. Activation through Distal E2 Binding Sites Is Greatly Facilitated by the Presence of Proximal GC Boxes The data presented above are consistent with the suggestion that Spl binding sites serve some role in facilitating E2 activation of various BPV-1 promoters. However, since the P4 and P5 promoters are likely to have other binding sites for cellular transcription factors, the results do not necessarily imply a direct interaction between Spl and E2. For example, Spl may interact with some other cellular factor that binds to the P4 or P5 promoter and that interacts with E2. This putative other factor may, in turn, require Spl to function, so that deletion of Spl sites would result in lowered basal and E2 responsiveness. This possibility is not unreasonable, given the finding that the VP16-E2 fusion gene can activate both P4 and P5, even when the Spl sites are mutant. This shows that other DNA-binding sites must contribute to the basal activity of the P4 promoter. To investigate the linkage between E2 and Spl activities directly, we utilized a series of simple reporter plasmids (Figure 4) that contain the TATA box from sequence -37 to +51 of the HSV TK gene. In one case, we inserted four GC boxes immediately upstream of the TATA element and placed five repeats of a strong E2 binding

site downstream of the CAT gene (Spl E2). This construct resembles the natural arrangement of the URR relative to the Spl sites proximal to the P4 promoter. A series of control plasmids was also constructed. Plasmid tkE2 contains the wild-type E2 sites downstream of the CAT gene, but does not have the proximal Spl sites. In two other constructs (Spl ME2 and tkMEP), mutant E2 sites that do not bind E2 (characterized previously in Hirochika et al., 1988) were placed in identical positions to the wild type. As can be seen in Figure 4, the proximal Spl sites allowed for an induction by E2 of at least 50-fold. In contrast, the constructs that contained the TATA element, but no Spl sites, were barely activated by E2. Interestingly, Spl ME2 reproducibly showed a weak response to E2 (roughly lo-fold). We do not believe that this is due to weak E2 binding to these mutant E2 sites, because we found that a basic TK TATA promoter with Spl sites, but without the mutant sites, was activated by E2 to an identical extent (data not shown). Rather, we feel it is more likely, in view of the data presented below, that E2 is localized to the promoter by association with Spl protein already bound to the GC boxes. E2 Interacts with a GAL4-Spl Chimera to Superactivate Transcription from a Promoter Containing GAL4 Binding Sites Although the data presented above show that Spl binding sites mediate the activation by E2, this could be due either to Spl or perhaps to one of the other cellular factors that have been shown to bind the GC motifs. To determine whether Spl indeed mediates the E2 response, we utilized a GAL-Spl vector that produces a chimeric protein of the GAL4 DNA-binding domain and Spl (G. Gill, University of California at Berkeley) as a way to modulate endogenous activator levels in CV-1 cells. As a control to ascertain the Spl specificity, we also utilized the GAL4 activator itself, a factor well known to activate transcription in a variety of eukaryotic cells (Kakidani and Ptashne, 1988; Webster et al., 1988; Sadowski et al., 1988; Gill et al., 1990). The transactivator and reporter plasmids used in these experiments are shown in Figure 5A. Gill (unpublished data) has shown that the GAL-Spl fusion protein can activate transcription in a GAL4 binding site-dependent manner. While optimal GAL-Spl activation is achieved in CV-1 cells with microgram quantities of the activator gene (Gill, unpublished data), we used smaller quantities to optimize for superactivation by E2. E2 on its own does not stimulate the GALCresponsive promoter (data not shown), nor does it substantially affect the activity of GAL4 (Figure 5B). However, as can clearly be seen in Figure 58, E2 markedly superactivates the GAL-Spl fusion protein over a range of GAL-Spl levels. Figure 5C quantitates and compares the fold stimulation of E2 upon the wild-type GAL4 and GAL-Spl fusion protein. This experiment shows that tethering Spl to a GALCresponsive promoter allows the transcription to be E2 responsive and thus substantiates the notion that Spl itself is able to mediate the E2 activation. Spl and E2 Physically Interact A series of biochemical tests were used to investigate the

Cell 498

A GAL4

SV40

DNAB

5xGAL4 w-7

TATA

I)

GAL-Spl

GSBCAT

+

E2

++

+++

-

-

+

++

+

+

+++

+

CAT

GAL-Sol + E2’-u-f++

++ +++

+

++ +++

C 30

20

10

0 0

Figure 5. The GAL-Spl Responsive Promoter

100 200 ng of GAL4 or GAL-Spl DNA

Fusion

Protein

Can Target

300

E2 to the GALC

(A) Diagrams of the GAL4, GAL-Spl, and the reporter constructs. Both GAL4 and GAL-Spl were expressed by an SV40 early promoter. “5 x GAL4” in the reporter plasmid represents the five repeats of GAL4 binding sites. (B) The reporter plasmid (5 ug) was transected with the indicated expression vectors into CV-1 cells. In this experiment, both GAL4 and the fusion GAL4-Spl were kept at low concentrations (+, 10 ng; ++, 50 ng; +++, 200 ng) to achieve an optimal E2 superactivation. Wherever indicated, 8 pg of E2 expression vector was used. (C) The magnitude of E2 superactivation is plotted at various levels of the GAL4 and GAL-Spl DNA. Superimposed on the E2 stimulation of the GAL4-responsive promoter is its effect on the expression of GAL4 and GAL-Spl, since both genes were driven by the SV40 early promoter, which contains multiple Spl binding sites. This may account for the slight increase (