Sep 27, 1993 - 'Present address: Eisai London Research Laboratories, BernardKatz. Building, University College London, Gower Street, London WC1E.
The EMBO Journal vol. 13 no. 1 pp. 147 - 1 57, 1994
Cooperativity in vivo between the E2 transactivator and the TATA box binding protein depends on core promoter structure
Jonathan Ham', Gertrud Steger and Moshe Yaniv2 UnitI des Virus Oncogenes, VA 1644 du CNRS, D6partement des Biotechnologies, Institut Pasteur, 25 rue du Dr Roux, 75724 Paris, France 'Present address: Eisai London Research Laboratories, Bernard Katz Building, University College London, Gower Street, London WC1E 6BT, UK 2Corresponding author Communicated by M.Yaniv
The E2 transactivator protein of bovine papillomavirus 1 (BPV-1) can strongly stimulate complex promoters such as that of the herpes simplex virus thymidine kinase gene but does not efficiently activate minimal promoters that only contain E2 binding sites and a TATA box. Here we show that overexpression of the human, but not yeast, TATA box binding protein (TBP) in transfection experinents overcomes this block and enables E2 to activate a minimal TATA box-conaining promoter. This suggests that recruitment of the TFHID complex to such promoters is normally a rate limiting step for transcriptional activation by E2 in vivo. In contrast, minimal promoters that contain an initiator element in addition to a TATA box are efficiently activated by E2 on its own and this activation is only moderately enhanced by TBP overexpression. In such E2-responsive promoters the TATA box or initiator can be functionally replaced by SP1 binding sites. Both the initiator binding protein, TFII-I, and SP1 have been found to interact physically with components of the TFIID complex. Since either TBP overexpression or the presence of an initiator or SP1 binding sites can increase activation by E2, it seems likely that the principal role of the E2 activation domain is to affect a step in the formation of the transcription initiation complex that occurs after TFTLID has bound to the promoter. Sequential action of transcription factors, such as TFH-I, SP1 and E2, may be one type of mechanism underlying the widely observed phenomenon of transcriptional synergy. Key words: BPV- 1 E2/initiator/TATA box/TFIID/ transactivator
Introduction The regulatory regions of RNA polymerase 11-transcribed genes contain two tpes of functional element: core promoter elements such as the TATA box (Breathnach and Chambon, 1981) and initiator (Smale and Baltimore, 1989), which specify the site at which the RNA polymerase II transcription initiation complex is assembled, and upstream promoter elements and enhancer sequences which bind activator proteins that increase the rate of transcription initiation [see Mitchell and Tjian (1989) for review]. RNA polymerase II Oxford University Press
is guided to the initiation region by a number of accessory proteins referred to as general or basal transcription factors (GTFs; reviewed by Sawadogo and Sentenac, 1990; Roeder, 1991; Zawel and Reinberg, 1992; Drapkin et al., 1993). The results of in vitro experiments with partially or wholly purified GTFs suggest that the transcription initiation complex is assembled in an ordered sequence (Van Dyke et al., 1988; Buratowski et al., 1988; Maldonado et al., 1990). In the case of promoters that contain a TATA box the first factor to bind is TFIID, which recognizes the TATA sequence. TFIID is in fact a large multisubunit complex consisting of the TATA box binding protein (TBP), an evolutionarily conserved polypeptide which specifically binds to the TATA element (reviewed by Greenblatt, 1991; Rigby, 1993), and a number of TBP associated factors (TAFs), which have been shown to be necessary for the functioning of many activators, i.e. function as coactivators (Dynlacht et al., 1991; Tanese et al., 1991; Zhou et al., 1992). The binding of TFIID is stimulated by another GTF, TFIIA, which on its own does not bind to DNA [see Roeder (1991) and Zawel and Reinberg (1992) for references]. Once TFIID has bound to the promoter, TFIIB is recruited and then the TFIID-A-B complex is recognized by RNA polymerase II in association with TFIIF. TFIIE, H and J then bind to complete the formation of the pre-initiation complex (Zawel and Reinberg, 1992). A number of core promoters have been described which contain initiator elements in addition to, or as an alternative to the TATA box [see Roeder (1991) and Zawel and Reinberg (1992)]. Like the TATA box, an initiator element (INR) can specify the site at which the transcription initiation complex is formed and where the mRNA starts, usually within the initiator sequence (Smale and Baltimore, 1989). A protein, TFII-I, has recently been purified from mammalian cells which specifically recognizes the initiator elements (consensus sequence YAYTCYYY) of the adenovirus major late, TdT and HIV-1 promoters and which strongly stimulates the binding of the TBP (or TFIID) to DNA (Roy et al., 1991, 1993). Furthermore, TFII-I can substitute for TFIIA in an in vitro transcription system reconstituted with purified components (Roy et al., 1991) and different pathways of preinitiation complex formation can occur on TATA+ INR-, TATA- INR+ and TATA+ INR+ core promoters (Roy et al., 1993). It has been suggested that one consequence of this may be that the different types of core promoter might respond differently to a given activator protein (Roy et al., 1993). How do the activator proteins that bind to upstream promoter elements and enhancers increase the rate of transcription initiation by RNA polymerase II? One model for the mechanism of transcriptional activation proposes that activators directly contact one or more of the GTFs and thereby increase the number of transcription initiation complexes formed or their rate of formation. In support of this model, a number of activator proteins have now been 147
J.Ham, G.Steger and M.Yaniv
shown to interact directly with one or more components of the transcription initiation complex (reviewed by Carey, 1991; Struhl, 1991; Ham et al., 1992; Drapkin et al., 1993). TFIID has attracted considerable attention as a potential target since it is the first factor to bind in the case of promoters with a TATA box and its binding appears to be a prerequisite for the subsequent assembly of the other GTFs and RNA polymerase II into a stable preinitiation complex. A number of activators-herpes simplex virus (HSV) VP16, adenovirus ElA, Epstein-Barr virus Zta and the cellular p53 protein-have all been found to interact directly with the TBP in vitro (see Ham et al., 1992; Seto et al., 1992), whereas SPi interacts indirectly with the TBP by contacting TAF 110 (Hoey et al., 1993), which in turn is anchored to the TBP by means of an interaction with TAF 250 (Weinzierl et al., 1993). Certain activators may interact with more than one target. For example, as well as binding to the TBP in vitro, VP16 has been found to bind to TFIIB (Lin and Green, 1991; Lin et al., 1991; Roberts et al., 1993), and to require a coactivator activity (White et al., 1991). Furthermore, kinetic experiments have suggested that the principal effect of VP16 on transcription occurs after binding of TFIID (White et al., 1992). We are studying the mechanism by which the E2 protein of bovine papillomavirus 1 (BPV-1) activates transcription. E2 is a sequence-specific DNA binding protein which binds as a dimer to the palindromic sequence ACCGNNNNCGGT, which occurs in multiple copies in papillomavirus genomes (reviewed by Ham et al., 199 lb; McBride et al., 1991; Steger et al., 1993). E2 has the properties of an enhancer factor, that is it can activate the transcription of promoters in BPV-1 that are at some distance (kilobases away) from its specific DNA binding sites (e.g. P2443, Spalholz et al., 1991). Furthermore, E2 can activate heterologous promoters such as that of the HSV thymidine kinase (TK) gene when its binding sites are cloned either upstream or downstream of the transcription initiation site (Thierry et al., 1990). Interestingly, E2 is unable to activate efficiently a minimal promoter consisting of E2 sites and the TK TATA box and initiation site alone (Ham et al., 1991a; Li et al., 1991). However, if binding sites for upstream promoter factors such as SP1 or USF are cloned next to the TK TATA box, the ability of E2 to activate transcription is restored (Ham et al., 1991a; Li et al., 1991). The target of the E2 transcriptional activation domain is unknown at present. However, in the course of studying the mechanism by which the BPV-1 E2 represses transcription of the HPV18 P105 promoter in vitro, we obtained results that suggested that the TBP might be one target of the E2 activation domain (Dostatni et al., 1991). Normally the HPV18 P105 promoter is repressed by the binding of the BPV-1 E2 protein to an E2 site 3 bp upstream of the P105 TATA box (Thierry and Yaniv, 1987; Thierry and Howley, 1991). In vitro, the binding of E2 to this site appears to destabilize the complex between the TBP and the TATA box (Dostatni et al., 1991). However, if the distance between the E2 site and TATA sequence is increased to 8, 13 or 18 bp, E2 and the TBP bind simultaneously and cooperatively (Dostatni et al., 1991; G.Steger, J.Ham and M.Yaniv, in preparation). This suggests that E2 can actually increase the stability of the TBP-TATA box complex provided that the binding sites for E2 and the TBP are not too close together. Following the recent cloning of the TBP and other GTFs
it has become possible to study the consequences of overexpressing particular GTFs and mutant derivatives of them in transient transfection experiments (Berkenstam et al., 1992; Colgan and Manley, 1992; Colgan et al., 1993). Here we have used TBP overexpression as a way of studying the mechanism by which E2 activates transcription in vivo.
Results The E2 protein of BPV- 1 and the human TATA box factor cooperatively activate transcription in vivo It has previously been observed, in transient transfection experiments, that the E2 transactivator protein of BPV-1 does not efficiently activate a minimal promoter consisting of E2 DNA binding sites and the TATA box and transcription initiation site of the HSV TK promoter linked to the bacterial chloramphenicol acetyl transferase (CAT) gene (Ham et al., 1991a; Li et al., 1991). However, in another series of experiments we have shown that in vitro, purified E2 protein and the TBP bind cooperatively to oligonucleotides containing a single E2 site 8 bp away from the HPV18 P105 TATA box (Dostatni et al., 1991). We therefore wondered whether the P105 TATA box (TATAAAA) might be functionally different to that of the TK promoter (TATTAAG) and whether insertion of E2 sites very close to the P105 TATA sequence (8 bp away) might allow E2 to activate a minimal promoter in transient transfection experiments. To address this question we set up a transfection system in which we could alter the intracellular concentration of either E2 or the human TBP. We used a mouse embryonic carcinoma cell line, RAC65, in which the endogenous level of TBP can be increased at least several-fold by transfection of an expression vector for this protein (data not shown; Berkenstam et al., 1992). Transient transfection of a TBP expression vector has also been shown to increase TBP levels in other mammalian cell lines (e.g. mouse P19 cells, Berkenstam et al., 1992; human C33 cells, J.Ham, unpub-
lished observations) and in Drosophila Schneider cells (Colgan and Manley, 1992) suggesting that many cell lines may resemble RAC65 cells in this respect. To test the effect of overexpressing the TBP on transcriptional activation by E2, various CAT reporter plasmids together with a reference 3-galactosidase ($-gal) expression vector were transfected into RAC65 cells either alone or with expression vectors for E2 (pC59), human TBP (pSG5.hTBP) or both proteins together. 24 hours after transfection, extracts were prepared for CAT and ,-gal assays, as described in Materials and methods. CAT activities were normalized for (-gal activity and can therefore be compared between the different figures. Typical results are shown in Figure 1A. In this experiment several different constructions were tested: the first (pmET105) contains a mutated E2 binding site that is unable to bind E2 in vitro (see Dostatni et al., 1991), 8 bp away from the P105 TATA box. As expected, this construction was not activated by E2, but cotransfection of the human TBP expression vector increased the basal level of the promoter 5-fold, suggesting that the endogenous level of TBP in RAC65 cells is not optimal for the basal activity of the P105 TATA box. Expression of E2 at the same time as the TBP slightly reduced the induction by TBP. In contrast, a construction with four functional E2 binding sites 8 bp upstream of the -
Interaction of BPV-1 E2 protein and TBP in vivo
Normalized CAT activity
E2 TBP E2+TBP
.t**ll|ls.I .. ..
E2+TBP E2 TBP E2+TBP
E2 T E2
E2 TBP E2+TBP
E2 TBP E2+TBP
E2 TBP E2+TBP
E2 TBP E2+TBP
4 E2 sites
E2 + TBP
2 E2 sites
4 E2 sites
Distance between E2 sites and the P105 TATA box (bp) Fig. 1. E2 and human TBP cooperate to synergistically activate the transcription of a minimal promoter that only contains E2 binding sites and a TATA box. (A) RAC65 cells were transiently transfected with various CAT reporter plasmids together with expression vectors for E2 or the TBP, as indicated. The structure of the reporter plasmids is shown to the left of the bar graph of relative CAT activity. mE2 represents a mutated E2 site (ACCGAAAACCCT). The mutated P105 TATA box (CCTAAAA) is shown as an X. SVeCAT contains the enhancer and early promoter of SV40 upstream of the bacterial CAT gene. The normalized CAT activities shown are the mean of the results obtained in three independent transfection experiments. Normalized CAT activity was calculated as described in Materials and methods. (B) Effect of distance between E2 sites and the TATA box on transcriptional activation by E2 in the presence or absence of pSG5.hTBP. Constructs containing blocks of two or four E2 sites cloned at various distances upstream of the P105 TATA box were transfected into RAC65 cells together with expression vectors for E2 and human TBP as indicated. Distance (in bp) between the E2 sites and P105 TATA box was plotted against fold activation by E2. Fold activation was calculated for each construct by dividing the normalized CAT activity obtained in the presence of E2 or E2 + TBP by the normalized CAT activity obtained in the absence of co-transfected expression vectors. In HPV18 the normal spacing between the P105 TATA box and upstream E2 sites is 3 bp.
J.Ham, G.Steger and M.Yaniv
P105 TATA box (p4ET105) could be weakly activated by
E2 alone (3-fold). Co-transfection of the human TBP increased the basal activity of this promoter 3-fold but simultaneous expression of E2 and TBP led to a very strong synergistic activation of the basal level (150-fold activation). On the other hand, a target promoter with four E2 sites and a mutated TATA box (p4EmT; the bases mutated are shown in Figure 4A) was not activated by E2, even when pSG5.hTBP was co-transfected, indicating that the functional cooperativity between E2 and the TBP requires both E2 sites and an intact TATA box. The degree of synergy between E2 and the TBP depends on the number of E2 sites in the reporter construct. As shown in Figure IA, a single E2 site does not allow significant activation by E2 in mammalian cells, even when TBP is overexpressed (pET105; Figure 1A). In contrast, constructs with two, three or four E2 sites can be activated by E2 in the presence of pSG5.hTBP, the level of activation by E2 increasing more than additively as the number of E2 binding sites is increased. The molecular basis of this synergy between E2 dimers is unknown [see Ham et al. (199lb) for a discussion]. Finally, the activity of the SV40 enhancer and promoter linked to CAT (SVeCAT in Figure 1A) was not significantly altered by expression of either E2 or TBP, an important control since the E2 expression vector, pC59, contains this promoter. Therefore the synergy between E2 and TBP does not result from an increase in the concentration of E2 due to activation of the SV40 promoter in pC59 by overexpression of TBP. From these results it is clear that a target promoter consisting of four E2 sites 8 bp upstream of the P105 TATA box and initiation site is only weakly activated by E2 alone in RAC65 cells, as was previously shown for the TK core promoter in C33 cells, a human cervical cancer cell line (Ham et al., 1991a). However, expression of human TBP at the same time as E2 increases the level of CAT activity in the presence of E2. TBP overexpression has a similar effect on E2 activation of the P105 core promoter in human C33 cells (.Ham, unpublished observations). Thus, the inability of E2 on its own to activate a promoter containing only E2 sites, a TATA box and initiation site appears to be a generalized phenomenon, since it has been observed with two different TATA sequences in two different cell lines. To investigate whether the cooperativity between E2 and the TBP was affected by the distance between their respective binding sites, we tested spacings of 3, 8, 13, 18 or 23 bp between groups of either two or four E2 sites and the P105 TATA box (Figure IB; distances of >23 bp were not tested). With a spacing of 3 bp, the normal spacing in the HPV18 P105 promoter, E2 does not significantly activate transcription in the presence of the TBP. This is consistent with the observation that this spacing does not allow E2 and the TBP to bind cooperatively to DNA (Dostatni et al., 1991). Then, in the case of groups of either two or four E2 sites, the degree of activation by E2 in the presence of the TBP increases steadily as the distance between the E2 sites and TATA box is increased (8-23 bp spacings). 'Fold activation' by E2 may increase with distance either because at close spacings E2 dimers interfere with the recruitment of the multisubunit TFIID complex or other GTFs to the promoter, or because when E2 sites are further away DNAbound E2 dimers may activate transcription more efficiently because they make better contacts with the transcription initiation complex than at closer spacings. Interestingly, there is no helix-tum phase dependence in the synergism between 150
E2 and TBP suggesting that perhaps it does not require direct protein -protein contact between the two proteins. Alternatively, phasing effects may not be apparent because E2 dimers are flexible enough to be able to contact the initiation complex even when they are bound to the opposite side of the DNA helix. Does cooperativity between E2 and human TBP require the E2 transcriptional activation domain? Figure 2A shows that three deletion mutations previously shown to reduce transactivation of a complex promoter by E2 (McBride et al., 1989) also affect E2 activation of the P105 minimal promoter in the presence of the TBP, indicating that the previously defined E2 transactivation domain is necessary for cooperativity with the TBP. The TBP contains two regions, an evolutionarily conserved carboxy-terminal core domain, which is necessary and sufficient for TATA box recognition and pre-initiation complex formation in vitro, and an amino-terminal domain, which varies in length and amino acid sequence between different species (see, for example, Hoffmann et al., 1990). The function of the latter is not yet clear. To determine whether both of these regions are necessary for cooperativity with E2 in vivo, we tested a truncated form of the TBP which only contains the carboxyterminal domain (hcore). When transfected into RAC65 cells together with E2 and p4ET105, the hcore increased activation by E2 almost as effectively as the full-length human TBP (Figure 2B), indicating that the amino-terminus is dispensable for cooperativity with E2. Since E2 can efficiently activate transcription in yeast (Saccharomyces cerevisiae) when its DNA binding sites are cloned upstream of the CYCJ promoter (Lambert et al., 1989) and since the carboxy-terminal core domains of the human and yeast TBPs show 81 % identity at the amino acid level, we investigated whether the yeast TBP could also increase transcriptional activation of p4ET105 by E2. Surprisingly, the yeast TBP was unable to cooperate with E2 in RAC65 cells. The same expression vector (pSG5.yTBP) has previously been shown to express functional yeast TBP in RAC65 cells (Berkenstam et al., 1992). To check that the inability of the yeast TBP to cooperate with E2 was not due to an inhibitory effect of the amino-terminus of the yeast protein, we tested a hybrid, yNhC, in which the amino-terminus of the yeast TBP had been grafted on to the hcore (Berkenstam et al., 1992). This protein cooperated efficiently with E2 (Figure 2B) suggesting that it is the difference (19%) between the core regions of the yeast and human proteins which is responsible for their differing behaviour in RAC65 cells. The adenovirus major late promoter can be activated by E2 on its own Since different pathways of preinitiation complex formation can occur in vitro on TATA+ INR-, TATA- INR+ and TATA+ INR+ core promoters it has been suggested that the different types of core promoter might respond differently to a given activator protein (Roy et al., 1993). To test this idea we decided to compare p4EMLC, a reporter plasmid containing four E2 sites 24 bp upstream of the adenovirus major late core promoter (nucleotides -44 to + 10), which has a consensus TATA box and initiator element (see Figure 4A) with p4ET105, which contains four E2 sites upstream of the P105 TATA box. The typical result of a transient transfection experiment in RAC65 cells with these constructs is shown in Figure 3A. When p4EMLC was transfected together with the E2 expression vector pC59
Interaction of BPV-1 E2 protein and TBP in vivo
ccrserved ccre drma. i
BPV-1 E2 TRAN SACTIVATION
... . .. . ....... . I
Al-1 5 1-52 A92-1 v
Fo;o activation -1
ITB P hcore
5_-TBP .21-1 .9 1-2 - TBP
+ yNhC E2 -j 1. .-
Ncrrral'zed CAT activity Itvn-
2 C2 2
Normalized CAT activity
Fig. 2. Regions of E2 and the TBP required for cooperativity in vivo. (A) Cooperativity between E2 and the TBP requires the E2 transcriptional activation domain. A diagram of the BPV-1 E2 protein is shown at the top with the previously defined transactivation, hinge and DNA binding domains marked. Numbers indicate positions in the amino acid sequence. Principal in vivo phosphorylation sites are marked. The regions deleted in the three mutants (A1 -15, Al -52 and A92- 161) are shown beneath the transactivation domain. p4ET105 was transfected into RAC65 cells with various combinations of expression vectors for TBP and E2 or the three E2 deletion mutants as indicated on the left side of the bar graph that shows the resulting CAT activity. (B) The yeast TBP does not increase transcriptional activation by E2 even though the conserved core domain of the human protein does. The structures of the human and yeast TBPs and mutant derivatives (hcore and yNhC) are shown. The variable amino-terminal regions of the human and yeast proteins are 156 and 61 amino acid residues long, respectively. The carboxy-terminal core domain, which is highly conserved between different species, is 180 residues long and contains a direct repeat (shown by arrows) and a repeat of basic residues (+ + + + +). This region of the protein folds into a saddle-shaped structure and is responsible for DNA binding and interactions with TAFs and GTFs [see Nikolov et al. (1992) and Zhou et al. (1993)]. p4ET105 was transfected into RAC65 cells together with the expression vectors shown to the left of the bar graph of normalized CAT activity. Fold activation, shown on the right, was calculated by dividing the nonnalized CAT activity for each extract by the value obtained for p4ET105 alone (-).
alone, E2 increased the basal level of the major late core was very weakly activated by E2 alone (2-fold) in this experiment. Furthermore, expression of the TBP at the same time as E2 only led to a modest increase in the level of activation of p4EMLC by E2 (77-fold versus 71-fold without TBP) whereas activation of p4ET105 was greatly increased, as shown before. This effect is also illustrated in the dose -response experiment shown in Figure 3B, where the amount of pSG5.hTBP transfected was varied and the activity of the two promoters in the presence of E2 was determined. Finally, as previously observed in C33 cells (Ham et al., 1991a), activation of the major late core promoter by E2 required the presence of E2 binding sites (data not shown). Thus in contrast to a core promoter containing only the HPV18 P105 TATA box (e.g. that in p4ET105), the major late core promoter in p4EMLC can be efficiently activated by E2 and overexpression of the TBP leads to no further increase in activation, suggesting that the normal level of TBP in RAC65 cells is not a limiting factor for activation of this promoter by E2. What might be responsible for this difference? There are two main differences between the promoter regions in p4ET105 and p4EMLC (see Figure 4A). Firstly, the major late core promoter contains promoter 71-fold. In contrast, p4ET105
initiator element in addition to a TATA box (Roy et al., 1991), whereas there are no good matches to the INR consensus in the P105 promoter. Secondly, although the sequences of the two TATA boxes are identical, their contexts are very different. Notably, the major late TATA box is embedded in a very GC-rich region. The distances between the E2 sites and the two TATA sequences are also different-8 bp in p4ET105 versus 24 bp in p4EMLC-but this can be discounted as an important factor since the results of the spacing experiment indicate that as the distance between four E2 sites and the P105 TATA box is increased, activation by E2 alone does not increase significantly and, regardless of the spacing, is always greatly increased by coexpressing the TBP (Figure 1B). To test whether it was the presence of an initiator element that enabled E2 to activate efficiently the major late core promoter, we decided to clone the major late INR sequence downstream of the P105 TATA an
box. Two constructions were made in which four E2 sites were positioned either 8 bp (in p4ET 1051) or 23 bp (in p4E23T1051) upstream of the P105 TATA box and major late INR element. When transfected into RAC65 cells, both constructs were activated by E2 on its own (Figure 4B; compare p4ET1O5I with p4ET105, and p4E23TlO5I with p4E23TlO5). This indicates that the addition of the major late INR to the P105 TATA box creates a minimal promoter 1 51
J.Ham, G.Steger and M.Yaniv
TRFres o rc
;t E P
*.- / ;
1-.'. -- L. -,.-
Fig. 3. E2 can efficiendy activate the adenovirus major late core promoter on its own. (A) p4ET105 and p4EMLC were transfected into RAC65 together with the expression vectors indicated beneath the CAT chromatogram. CAT assays were performed as described in Figure 1A. Normalized CAT activities are shown at the bottom together with the structures of the CAT reporter plasmids. Sequences derived from the P105 or major late promoters are shown by thin lines and plain text or bold lines and italics respectively. (B) Dose-response curve showing how varying the amount of pSG5.hTBP transfected into RAC65 cells affects transcriptional activation of p4ET105 (A) or p4EMLC (*) by E2. cells
that can be efficiently activated by E2 in cells containing the normal endogenous level of TBP. A construction containing four E2 sites, a mutated P105 TATA box and the major late initiator (p4E I) was not activated by E2 even when the TBP was coexpressed, indicating that a TATAINR+ core promoter does not respond to E2. Because an initiator element can independently specify the site of transcription initiation (Smale and Baltimore, 1989; Carcamo et al., 1991) it was important to determine whether or not the addition of the adenovirus major late initiator to the P105 TATA box had altered the pattern of RNA initiation on the P105 core promoter. We therefore transfected p4E23T105 and p4E23T105 I into RAC65 cells together with pC59 and pSG5.hTBP. 24 h after transfection total cytoplasmic RNA was prepared and the pattern of RNA initiation was determined by primer extension analysis. In the case of p4E23T105 four start sites were observed: at approximately -70, -29, -2 and +1 (Figure 5). The major start site is at -2 and this together with the + 1 start corresponds to the cluster of initiation sites around position 105, originally mapped by Thierry et al. (1987) using the RNase protection technique (these are shown in Figure 4A). The additional, weaker start sites at -70 and -29 might be a cell type specific difference or might result from the fact that we are studying the isolated P105 core promoter rather than the full-length promoter with its associated upstream elements. Importantly, the addition of the adenovirus major late initiator in p4E23T105 I does not alter the pattern of RNA start sites, indicating that the initiator's effect on transcriptional activation by E2 is due to an increase in the rate of initiation at the normal P105 start sites rather than a switch in promoter usage, i.e. from the P105 TATA box to the initiator. It also suggests that it is the TATA box that plays the dominant role in fixing the pattern of RNA initiation in this situation. Although the hybrid promoters in p4ET105 I and
p4E23T105 I can be efficiently activated by E2 on its own (20-fold and 29-fold respectively), this activation can still be increased a further 3- to 4-fold by expressing the TBP at the same time as E2. This contrasts with the behaviour of the major late promoter in p4EMLC and suggests that the INR element is not the only factor responsible for the difference between the major late and P105 minimal promoters. To test this idea we decided to introduce point mutations into the INR element in the context of the major late core promoter. Three bases in the initiator element (see Figure 4A) were simultaneously mutated (p4EmIt). The same mutations have previously been shown to abolish binding of TFII-I to the INR in vitro (Roy et al., 1991). In another construction (p4Emli) the major late TATA box was mutated (see Figure 4A). Mutation of the INR element reduces the ability of E2 to activate the major late core promoter on its own and increases the effect of TBP expression on E2 activation (Figure 6) but does not alter the pattern of RNA initiation on the major late promoter (data not shown; Du et al., 1993). However, this TATA+ INRconstruct is still activated more efficiently by E2 on its own than those, e.g. p4ET105, that contain the P105 TATA box. One possible explanation for this difference is that maybe the context of the major late TATA box (GC-rich sequences) facilitates activation by E2-perhaps the conformation of the TFIID -promoter complex is different from that formed on the P105 TATA sequence. In fact, when bound to the major late TATA box, TFIID gives rise to a footprint that extends much further downstream of the TATA sequence than when it is bound to the adenovirus Elb or E4 TATA boxes (Chiang et al., 1993). When the major late initiator element is mutated so as to destroy TFII-I binding, this extended footprint is still seen, suggesting that it depends on the context of the major late TATA box rather than on the presence of the initiator (Chiang et al., 1993). Furthermore, the major late TATA box appears to be intrinsically stronger in in vitro
Interaction of BPV-1 E2 protein and TBP in vivo
HPV1 8 P105
TATA.AAA GATGTGAGAAACACACCACAATACCAAGGCGCGCTTTGAAGATCT 0~~~~~~ S so
INR co-nsersus YAYTCYYY -~1
AdPenio v i'ru s .iirLa:e
Normalized CAT activity
E2 ±t TBP 0
p4 ET 1r. 5
-11-. 7 I''-.1-11-i",
.1 ...X.I..:. ",
E2 TBP E2+TBP Fig. E2.
4. Addition of the adenovirus
(A) Structure of
and Roeder found in the
major late initiator element to the P105 TATA box increases the ability P105 and adenovirus major late core promoters. Nucleotide sequences are
of the from
P105 core promoter to respond to Thierry et al. (1987) and Sawadogo
(1985) respectively. TATA and initiator elements are boxed. The consensus sequence derived from the sequences of the initiator elements Point mutations in major late, TdT and HIV-1 promoters by Roy etr al. (1991) is shown. This is the initiator sequence bound by
previously mapped for thie P105 promoter by RNase protection are represented by black dots (Thierry etr al., 1987). The nucleotide at position +2 (marked with a X) in the P105 core promoter was changed from a T to an A to destroy an ATG codon. (B) Reporter plasmids in which the adenovirus major late initiator had been added to the P105 core promoter were transfected into RAC65 cells together with the expression vectors indicated. The structure of the different plasmids is shown on the left. P105 sequences are representedl by thin lines and plain text, major late sequences by bold lines and italics. The distance between E2 sites and the TATA box is indicated. The values for normalized CAT activity are the mean of the results obtained in three independent trasfection experiments. The column to the right of the bar graph of normalized CAT activity shows the 'fold effect' of TBP coexpression on activation by E2. Addition of the initiator element to the P105 COre promoter reduces the effect of TBP coexpression by 10-fold. the TATA boxes and initiator, described in the text,
indicated. The RNA
transcription experiments than the ElIb or E4 TATA boxes. Perhaps the P105 TATA sequence resembles the latter two TATA boxes. Alternatively, it is possible that the major late core promoter or upstream polylinker or vector sequences contain as yet uncharaterized binding sites for transcription factors which facilitate activation by E2 in a similar manner to the INR element. For example, we note that there are potential SPI binding sites around the major late TATA sequence (see Figure 4A). In fact, purified SPI, protects two regions of the major late promoter in DNase I footprinting experiments in vitro (G.Steger, unpublished observations). The observation that mutation of the TATA box in the major
core promoter greatly reduces activation by E2 but does entirely abolish it (see p4Emli in Figure 6) while the homologous P105-derived promoter was inactive Figure 4B) would be consistent with the idea that another transcription factor binding site may be present.
activate TA TA-less promoters that
sites for SPi
presented so far demonstrate that on its own the protein can efficiently activate target promoters that contain the P105 TATA box and major late initiator element. Those that contain only the P105 TATA box can The results BPV-1 E2
J.Ham, G.Steger and M.Yaniv
n Mil F' \iR
Fig. 5. Addition of the major late initiator element to the P105 core promoter does not alter the pattern of mRNA initiation. RAC65 cells were transfected with p4E23T105I or p4E23T105 together with pC59 and pSG5.hTBP. 24 h after transfection total cytoplasmic RNA was isolated as described in Materials and methods. 40 Ag of RNA was used for each primer extension reaction. The 17mer primer hybridizes to the 5' end of the CAT gene [see Ham et al. (1991a)] and was also used to sequence the promoter of p4E23TI05I. These sequence reactions (CTAG) were run with the primer extension products on an 8% denaturing polyacrylamide gel. The extension products corresponding to the -29, -2 and + 1 start sites are indicated by arrows. The positions of the TATA box and initiator in the p4E23TlO5I sequence are shown.
also be strongly activated by E2 provided that the TBP is overexpressed. In contrast, promoters that only contain an initiator element either do not respond to E2 (in the case of p4EI) or respond very weakly (in the case of p4Emli) even when the TBP is overexpressed, suggesting that E2 does not efficiently activate TATA-less promoters. However, we thought this unlikely since the BPV-1 genome contains promoters, such as P3080, which appear to lack a TATA box but which are nevertheless activated by E2. It is not yet known whether P3080 contains an initiator or analogous element at the RNA start site but it has been shown to contain SPI binding sites, which are essential for its activation by E2 (Li et al., 1991). We therefore wondered whether the addition of SP1 binding sites to an initiator element would create a promoter that could be activated by E2. Two copies of the distal GC box from the HSV TK promoter were cloned between E2 sites and the major late initiator element-see p4ESmli in Figure 6. When transfected into RAC65 cells
this construction was activated 29-fold by E2 alone, demonstrating that E2 can activate a TATA-less promoter provided that it contains more than just an initiator element. Interestingly, when p4ESmli was transfected with the expression vectors for both E2 and the TBP, the overall level of CAT activity was lower than that obtained with E2 alone. A similar effect was observed with a construction containing two SP1 binding sites and the major late TATA box (p4ESmlt Figure 6; note that a different scale has been used for this construct) suggesting that it is related to the SP1 sites rather than to whether or not a promoter contains a TATA element. We have also tested a plasmid that contains two USF binding sites cloned between two E2 sites and the major late core promoter and found that, in this situation, coexpression of the TBP does not affect transcriptional activation by E2 (data not shown). In contrast, a similar TATA+ INR- promoter that contains SP1 sites behaved differently: coexpression of the TATA binding protein reduced the level of activation by E2 (data not shown).
Our experiments have defined three types of core promoter that differ in the way they respond to the E2 transactivator. The first class consists of core promoters that only contain an initiator element, which are not activated by E2; the second type are those that just contain a TATA box, which are weakly activated by E2 on its own but can be strongly activated when the TBP is overexpressed; finally, there are core promoters that consist of either (i) a TATA box + an initiator element, (ii) SP1 sites + a TATA box or (iii) SP1 sites + an initiator, which are efflciently activated by E2 alone. These results suggest that a core promoter must contain at least two elements to be able to respond to E2 and that such elements, the TATA box, initiator or SPi sites, are interchangeable to a large extent. As previously shown (Ham et al., 1991a) promoters that contain E2 sites, SP1 sites and a TATA box are very strong whereas even when TBP is overexpressed those that contain E2 sites and a TATA box or SPI sites and an initiator are still relatively weak. The present experiments also demonstrate that in vivo transfection studies are still a valuable tool for the study of the mechanisms regulating transcription initiation. What might be the molecular basis for the in vivo cooperativity between E2 and the TBP that we have observed with minimal promoters that only contain E2 binding sites and a TATA box? One possibility is that E2 recruits the TBP to the promoter by means of a protein-protein interaction, as suggested by in vitro binding experiments with purified E2 and TBP (Dostatni et al., 1991), and that perhaps there is not normally enough TBP in transfected cells for all of the promoters that have bound E2. However, if this were the case overexpression of the TBP might be expected to reduce activation by E2 after a certain intracellular concentration of TBP was reached due to a 'squelching' effect, which was not observed in the series of experiments described here. Furthermore, TBP is not limiting for E2-activated promoters that contain two or more core promoter elements, as discussed above. An alternative hypothesis is that perhaps transcriptional activation by E2 is blocked by negatively acting factors, such as Drl, Dr2, NC1 and NC2, that directly associate with the TBP [see Drapkin et al. (1993) for review] and that overexpression
Interaction of BPV-1 E2 protein and TBP in vivo
Normalized CAT activity 0
ML 24 bp
I E2 E2
E2 E2 ML
I E2 2 E
E2 TBP E2+TBP
E2 I E2
I EI2 ML
E2 TBP E2+TBP
...................... ........................ .,- " ""
Fig. 6. Effect of mutating the TATA box or initiator in the context of the major late core promoter. Major late promoter sequences are represented by bold lines and italics. The mutated TATA box or mutated initiator (see Figure 5A for the bases mutated) are shown by Xs. The sequence of the SPI sites is the same as that of the distal GC box from the HSV TK promoter. The direction in which the GC boxes point indicates their orientation: an arrow pointing to the right represents 5'-CCCCGCCC-3'. Note that the scale is different for p4ESmlt because the normalized CAT activities for this construct are so high. Normalized CAT activities are the mean of the results obtained in three independent transfection experiments.
of the TBP titrates away such factors. Yet another possibility, which we favour, is that TBP overexpression might increase activation by E2 because E2 predominantly acts after the TBP has bound to the promoter. Two different mechanisms can be envisaged. Firstly, if E2 interacts directly with the TBP it might do so only after it has bound to the TATA box; for example, a conformational change in the TBP that would allow interaction with E2 might result from interaction of TBP with DNA. Consistent with such an idea, recent in vitro experiments (G.Steger, J.Ham and M.Yaniv, in preparation) show that E2 does not affect the rate at which the TBP binds to the P105 TATA box but instead acts to reduce the rate at which the TBP dissociates from the DNA. However, we have also found that the E2 transcriptional repressor, which lacks most of the activation domain and which is unable to activate transcription in vivo [see Ham et al. (1991a), McBride et al. (1991) and Steger et al. (1993) for references], can also stabilize TBP binding. This indicates that a direct interaction between E2 and the DNA-bound TBP can only be one part of the mechanism by which E2 activates transcription and is on its own not sufficient to explain activation in vivo. Taken together with our in vivo results, the findings of these in vitro experiments suggest an alternative hypothesis: that the principal role of the E2 activation domain is to act at a step in preinitiation complex formation that occurs after TEIID has bound to the promoter. In contrast to the human TBP or its conserved carboxyterminal core domain, the yeast TBP does not cooperate with E2 in mammalian cells, even though E2 can efficiently activate transcription in yeast. The yeast TBP can replace the human protein in basal transcription in a HeLa in vitro transcription system (Buratowski et al., 1988; Horikoshi
et al., 1989) and the yeast and human TBPs are interchangeable for the response to acidic transcriptional activators in vitro (Kelleher et al., 1992). However, the yeast TBP does not allow SP1, which has a glutamine-rich activation domain, to stimulate transcription in a mammalian in vitro system (Pugh and Tjian, 1990) and the SP1 activation domain does not function in yeast (cited in Pugh and Tjian, 1992). The TBP without its associated TAFs will support basal transcription in vitro, but in the case of all the activators tested so far TAFs are necessary for stimulation over the basal level (Dynlacht et al., 1991; Tanese et al., 1991; Zhou et al., 1992). Although it is not yet known whether this requirement is also true for E2, it is possible that the yeast TBP does not cooperate with E2 in RAC65 cells because it is unable to associate with certain mammalian TAFs necessary for transcriptional activation by E2 in vivo. We have found that addition of an initiator element or SPI binding sites to a TATA box substitutes for overexpression of the TBP to a large degree, i.e. a minimal promoter containing E2 sites, a TATA box and initiator or SP1 sites is efficiently activated by E2 on its own. This phenomenon can be explained by our hypothesis that the major role of the E2 transactivation domain in vivo is to stimulate a step in the formation of the pre-initiation complex that occurs after TYIID has bound to the promoter. The initiator binding protein, TFII-I, binds cooperatively to DNA with the TBP or TFIID (Roeder, 1991; Roy et al., 1993). SP1 also interacts directly with the TFIfl) complex by binding to TAF 110 (Hoey et al., 1993). Therefore, in the case of promoters that contain an initiator or SP1 binding sites in addition to a TATA box, TFII-I or SP1 would recruit TFIID to the promoter and the number of promoters occupied by TFIID
J.Ham, G.Steger and M.Yaniv at normal intracellular TBP concentrations would be higher than in the case of promoters that only contain a TATA box. This would lead to a high level of transcriptional activation by E2 but synergy between E2 and the TBP would be low because the presence of the initiator or SPI binding sites compensates for TBP overexpression. Surprisingly, we have observed that promoters containing only E2 sites and the adenovirus major late initiator are not efficiently activated by E2 even when the TBP is overexpressed. Like those that contain a TATA box, TATAINR+ promoters have been shown to require TFIID in in vitro transcription experiments (Smale et al., 1990; Pugh and Tjian, 1991) and TFII-I can apparently recruit TFIID to such promoters (Roy et al., 1993). Addition of SPI sites to a TATA- INR+ promoter enables it to respond to E2, suggesting that the presence of SP1 allows a pathway of preinitiation complex formation that favours activation by E2. Perhaps stable association of TFIID with a TATA-less promoter in vivo requires the presence of SP1 binding sites in addition to the initiator element. Some functional differences between promoters with and without TATA boxes have recently emerged (Colgan and Manley, 1992; Mack et al., 1993). Colgan and Manley (1992) found that overexpression of the TBP in Drosophila Schneider cells increased the activity of a promoter containing SPI sites and a TATA box but reduced the activity of a promoter containing SP1 sites and an initiator element. We have observed that when E2 sites are cloned upstream of a core promoter containing SPI sites and an initiator, overexpression of the TBP reduces the overall activity of the promoter in the presence of E2. However, we have also seen this effect with promoters containing a TATA box in addition to E2 sites and SPI sites suggesting that in our system this phenomenon is related to SPI rather than to whether a promoter is TATA+ or TATA-. This is confirmed by the fact that the ability of E2 to activate a similar construction in which the SPI sites are replaced with USF sites is not affected by overexpression of the TBP. Perhaps a TAF essential for the functioning of SP1, such as TAF 110 (Hoey et al., 1993), becomes limiting when the TBP is overexpressed in RAC65 cells, resulting in the formation of incomplete TFHD complexes that cannot interact with SP1, i.e. overexpression of the TBP would squelch transactivation by SPI. Since activation by E2 is augmented by SPI this would lead to a reduction in the level of CAT activity in the presence of E2. The fact that TBP overexpression does not affect the ability of E2 to activate promoters that contain USF sites suggests that SP1 and USF might activate transcription by different mechanisms. For example, activation by USF might require a TAF that does not become limiting when the TBP is overexpressed. In conclusion, the results presented here are consistent with a model in which the E2 transactivator acts after TFIID has bound to the promoter. On the other hand, proximal promoter factors such as TFII-I and SP1 have been shown to interact directly with TFIID. Such factors may increase transcriptional activation by E2 because they recruit TFIID to the promoter. This type of mechanism-the sequential action of transcription factors that affect different steps in the assembly of the transcription initiation complex-may be one way of explaining how different transcription factors cooperate to activate transcription synergistically in vivo. It also suggests that the ability of different promoters to respond
to an E2-responsive enhancer will depend on their affmity for TFIID, which in turn will be determined by core promoter structure.
Materials and methods Plasmid constructions C4Treporterplasmids. Plasmids containing E2 binding sites and the HPV18 P105 TATA box were constructed as follows. The plasmid TKM (Thierry et al., 1990) contains the HSV TK promoter from -109 to +55 upstream of the CAT open reading frame and SV40 transcription termination signals. The TK promoter was excised by digestion with XbaI and BgM and replaced with two double-stranded oligonucleotides ligated head to tail. One, E8T, with an XbaI site at the 5' end, contained a single E2 site 8 bp away from the P105 TATA box [upper strand: 5'-CTAGACCGAAAACGGTGATCCGTATATAAAAGATGTG; lower strand: 5'-TTCTCACATCTTTTATA TACGGATCACCGTTTTCGGT (the E2 site and TATA box are underlined)]. The other oligonucleotide contained the P105 mRNA start site and had a Bgll site at its 3' end [upper strand: 5'-AGAAACACACCACAATACCAAGGCGCGCTTTGAA; lower strand: 5'-GATCTTCAAAGCGCGCCTTGGTTATTGTGGTGTGT (nucleotide 105 is underlined)]. Additional E2 sites were cloned into the unique XbaI site at the 5' end of E8T using the following double-stranded oligonucleotide: upper strand, 5'-CTAGACCGAAAACGGTG; lower strand, 5'-CTAGCACCGTTTTCGGT. To alter the spacing between the E2 site and P105 TATA box, four pairs of oligonucleotides were synthesized. These were identical in sequence to E8T except for the region between the E2 site and the TATA box. The spacer sequences (upper strand 5' to 3') were: GTA (E3T, 3 bp spacing), GATCGTACTCGTA (E13T, 13 bp spacing), GATCGTACTAGCTACGTA (E18T, 18 bp spacing) and GATCGTACTAGCTACTGATCGTA (E23T, 23 bp spacing). To mutate the E2 binding site or P105 TATA box, oligonucleotides based on E8T containing the following mutations were synthesized: ACCGAAAACCCT (for a mutated E2 site) or CCTAAAA (mutated TATA box). These mutations abolish in vitro binding of E2 or TBP respectively [see Dostatni et al. (1991)]. To replace the sequences that normally surround the P105 mRNA start site with the adenovirus major late promoter INR element, the following double-stranded oligonucleotide was co-ligated with E8T or E23T: upper strand, 5'-AGAAACACACCACGTCCTC4CTCTC7TCCTTGAA; lower strand, 5'-GATCTTCAAGGAAGAGAGTGAGGACGTGGTGTGT (adenovirus major late sequences are underlined and the initiator consensus is shown in italics). Constructs containing E2 binding sites linked to the adenovirus major late core promoter and variants thereof were derived from p4EMLC (previously called pC18, see Figure 2B in Ham et al., 1991a). p4EMLC contains a polylinker (5'-Hindu, SphI, PstI, Sall, XbaI, BamHI) upstream of the adenovirus major late core promoter (-44 to + 10) linked to CAT. A block of four E2 binding sites is cloned at the XbaI site and lies 24 bp upstream of the major late TATA element. To mutate the major late TATA box or INR element, p4EMLC was digested with BamHI and XhoI (the latter cuts at the very 5' end of the CAT gene), to remove the wild-type core promoter sequences, which were then replaced with double-stranded oligonucleotides containing the appropriate mutations: TCTACAA to mutate the TATA box or CACATACT to mutate the INR element (the mutated bases are underlined). To clone SPI binding sites at the BamHI site between the E2 sites and the TATA box, an oligonucleotide containing the distal GC box from the TK promoter was used. This has BamrHI-compatible ends. Upper strand: 5'-GATCTAAACCCCGCCCAGCG; lower strand: 5'GATCCGCTGGGCGGGGTTTA. Standard procedures were used to construct all plasmids (Sambrook et al., 1989). Promoter structures were verified by direct sequencing of plasmid DNA using a USB Sequenase version 2.0 kit. Oligonucleotides were synthesized on a Milligen/Biosearch Cyclone Plus DNA synthesizer.
Expression vectors The E2 expression vector pC59 and the series of E2 deletion mutants are described in McBride et al. (1989). These proteins are all expressed under the control of the SV40 early promoter. pSG5.hTBP, the expression vector for the full-length human TBP, is described in Berkenstam et al. (1992) as phTFIID and consists of a full-length human TBP cDNA cloned in the SV40-based expression vector pSG5 (Green et al., 1988). The hcore, yeast TBP and yeast NhC were also expressed in pSG5 and are described in
Interaction of BPV-1 E2 protein and TBP in vivo
Cell culture and transient transfections RAC65 cells are derived from the mouse P19 embryonic carcinoma cell line (Jones-Villeneuve et al., 1983) and were grown in 1:1 DMEM:F-12 supplemented with 7.5% fetal calf serum in gelatin-coated dishes and passaged every 2 days. 24 h before transfection cells were seeded at a density of 1.25 x 105 per 6 cm diameter dish. Cells were re-fed with 2 ml of medium - 4 h before transfection and calcium phosphate-DNA precipitates were prepared by the standard procedure (Wigler et al., 1977). 0.5 ml of precipitate was added to each dish and this contained 2 Ag of CAT reporter plasmid, 1 jig of RSV (3-gal, 0.5 ug of E2 expression vector (pC59) or the same vector without insert, 0.5 jig of pSG5.hTBP or pSG5 and 2 /g of pGEM3 as carrier. After overnight incubation the precipitates were washed away and fresh medium was added for an additional 24 h. Cells were harvested and extracts prepared and assayed for CAT and ,3-gal activity as described by Herbomel et al. (1984). Normalized CAT activity was calculated by dividing the percentage of [14C]chloramphenicol converted to the monoacetylated form by the ,8-gal activity obtained for the same extract. The values shown represent the average of the results from three independent transfection experiments. Fold activation by E2 tends to vary somewhat between experiments because basal levels vary. Isolation of total cytoplasmic RNA from transfected cells and primer extension 9 cm diameter dishes were seeded with 3.75 x 105 RAC65 cells and transfected with 6 yg CAT reporter plasmid, 3 isg RSV (-gal, 1.5 jig pC59, 1.5 Ag pSG5.hTBP and 6 Ag pGEM3 as carrier. 24 h after transfection cytoplasmic RNA was isolated and primer extension analysis carried out essentially as described in Ham et al. (1991a). 40 ug of cytoplasmic RNA was used for each primer extension reaction together with 6 x 104 c.p.m. of the 17mer CAT primer, labelled using T4 polynucleotide kinase.
Acknowledgements We would like to thank Francoise Thierry and Karen Philpott for helpful discussions and critical reading of the manuscript. We are also indebted to Maria del Mar Vivanco Ruiz, Anders Berkenstam and Henk Stunnenberg for kindly providing us with RAC65 cells and to Alison McBride for sending us the E2 deletion mutants. This work was supported by grants from ARC, LNFCC, INSERM and FRMF. J.H. received fellowships from SERC/NATO and ARC. G.S. was an EMBO long-term fellow.
References Berkenstam,A., Vivanco Ruiz,M.d.M., Barettino,D., Horikoshi,M. and Stunnenberg,H.G. (1992) Cell, 69, 401-412. Breathnach,R. and Chambon,P. (1981) Annu. Rev. Biochem., 50, 349-383. Buratowski,S., Hahn,S., Sharp,P.A. and Guarente,L. (1988) Nature, 334, 37-42. Carcamo.,J., Buckbinder,L. and Reinberg,D. (1991) Proc. NatlAcad. Sci. USA, 88, 8052-8056. Carey,M. (1991) Curr. Opin. Cell Biol., 3, 452-460. Chiang,C.-M., Ge,H., Wang,Z., Hoffman,A. and Roeder,R.G. (1993) EMBO J., 12, 2794-2762. Colgan,J. and Manley,J.L. (1992) Genes Dev., 6, 304-315. Colgan,J., Wampler,S. and Manley,J.L. (1993) Nature, 362, 549-553. Dostatni,N., Lambert,P.F., Sousa,R., Ham,J., Howley,P.M. and Yaniv,M. (1991) Genes Dev., 5, 1657-1671. Drapkin,R., Merino,A. and Reinberg,D. (1993) Curr. Opin. Cell Biol., 5, 469-476. Du,H., Roy,A.L. and Roeder,R.G. (1993) EMBO J., 12, 501-511. Dynlacht,B.D., Hoey,T. and Tjian,R. (1991) Cell, 66, 563-576. Green,S., Issemann,I. and Sheer,E. (1988) Nucleic Acids Res., 16, 369. Greenblatt,J. (1991) Cell, 66, 1067-1070. Ham,J., Dostatni,N., Arnos,F. and Yaniv,M. (1991a) EMBO J., 10, 2931 -2940. Ham,J., Dostatni,N., Gauthier,J.-M. and Yaniv,M. (1991b) Trends Biochem. Sci., 16, 440-444. Ham,J., Steger,G. and Yaniv,M. (1992) FEBS Lett., 307, 81-86. Herbomel,P., Bourachot,B. and Yaniv,M. (1984) Cell, 39, 653-662. Hoey,T., Weinzierl,R.O.J., Gill,G., Chen,J.-L., Dynlacht,B.D. and Tjian,R. (1993) Cell, 72, 247-260. Hoffmann,A., Sinn,E., Yamamoto,T., Wang,J., Roy,A., Horikoshi,M. and Roeder,R.G. (1990) Nature, 346, 387-390. Horikoshi,M., Wang,C.K., Fujii,H., Cromlish,J.A., Weil,P.A. and Roeder,R.G. (1989) Proc. Natl Acad. Sci. USA, 86, 4843-4847.
Jones-Villeneuve,E.M.V., Rudnicki,M.A., Harris,J.F. and McBumey,M.W. (1983) Mol. Cell. Biol., 3, 2271-2279. Kelleher,R.J., Flanagan,P.M., Chasman,D.I., Ponticelli,A.S., Struhl,K. and Komberg,R.D. (1992) Genes Dev., 6, 296-303. Lambert,P.F., Dostatni,N., McBride,A.A., Yaniv,M., Howley,P.M. and Arcangioli,B. (1989) Genes Dev., 3, 38-48. Li,R., Knight,J.D., Jackson,S.P., Tjian,R. and Botchan,M.R. (1991) Cell, 65, 493-505. Lin,Y.-S. and Green,M.R. (1991) Cell, 64, 971-981. Lin,Y.-S., Ha,I., Maldonado,E., Reinberg,D. and Green,M.R. (1991) Nature, 353, 569-571. Mack,D.H., Vartikar,J., Pipas,J.M. and Laimins,L.A. (1993) Nature, 363, 281 -283. McBride,A.A., Byrne,J.C. and Howley,P.M. (1989) Proc. NatlAcad. Sci. USA, 86, 510-514. McBride,A.A., Romanczuk,H. and Howley,P.M. (1991) J. Biol. Chem., 266, 18411-18414. Maldonado,E., Ha,I., Cortes,P., Weis,L. and Reinberg,D. (1990) Mol. Cell. Biol., 10, 6335-6347. Mitchell,P.J. and Tjian,R. (1989) Science, 245, 371-378. Nikolov,D.B., Hu,S.-H., Lin,J., Gasch,A., Hoffmann,A., Horikoshi,M., Chua,N.-H, Roeder,R.G. and Burley,S.K. (1992) Nature, 360, 40-46. Pugh,B.F. and Tjian,R. (1990) Cell, 61, 1187-1197. Pugh,B.F. and Tjian,R. (1991) Genes Dev., 5, 1935-1945. Pugh,B.F. and Tjian,R. (1992) J. Biol. Chem., 267, 679-682. Rigby,P.W.J. (1993) Cell, 72, 7-10. Roberts,S.G.E., Ha,I., Maldonado,E., Reinberg,D. and Green,M.R. (1993) Nature, 363, 741-744. Roeder,R.G. (1991) Trends Biochem. Sci., 16, 402-408. Roy,A.L., Meisterernst,M., Pognonec,P. and Roeder,R.G. (1991) Nature, 354, 245-248. Roy,A.L., Malik,S., Meistererust,M. and Roeder,R.G. (1993) Nature, 365, 355-359. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning, A Laboratory Manual. 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sawadogo,M. and Roeder,R.G. (1985) Cell, 43, 165-175. Sawadogo,M. and Sentenac,A. (1990) Annu. Rev. Biochem., 59, 711-754. Seto,E., Usheva,A., Zambetti,G.P., Momand,J., Horikoshi,N., Weimmann,R., Levine,A.J. and Shenk,T. (1992) Proc. NatlAcad. Sci. USA, 89, 12028-12032. Smale,S.T. and Baltimore,D. (1989) Cell, 57, 103-113. Smale,S.T., Schmidt,M.C., Berk,A.J. and Baltimore,D. (1990) Proc. Natl Acad. Sci. USA, 87, 4509-4513. Spalholz,B.A., Vande Pol,S.B. and Howley,P.M. (1991) J. Virol., 65, 743 -753. Steger,G., Ham,J., Thierry,F. and Yaniv,M. (1993) In Doerfler,W. and Bohm,P. (eds), Virus Strategies. Verlag Chimie, Weinheim, pp. 263-278. Struhl,K. (1991) Curr. Biol., 1, 188-191. Tanese,N., Pugh,B.F. and Tjian,R. (1991) Genes Dev., 5, 2212-2224. Thierry,F. and Howley,P.M. (1991) New Biol., 3, 90-100. Thierry,F. and Yaniv,M. (1987) EMBO J., 6, 3391-3397. Thierry,F., Heard,J.M., Dartnann,K. and Yaniv,M. (1987) J. Virol., 61, 134-142. Thierry,F., Dostatni,N., Arnos,F. and Yaniv,M. (1990) Mol. Cell. Biol., 10, 4431-4437. Van Dyke,M.W., Roeder,R.G. and Sawadogo,M. (1988) Science, 241, 1335-1338. Weinzierl,R.O.J., Dynlacht,B.D. and Tjian,R. (1993) Nature, 362, 511-517. White,J., Brou,C., Wu,J., Burton,N., Egly,J.M. and Chambon,P. (1991) Proc. Natl Acad. Sci. USA, 88, 7674-7678. White,J., Brou,C., Wu,J., Lutz,Y., Moncollin,V. and Chambon,P. (1992) EMBO J., 11, 2229-2240. Wigler,M., Silverstein,S., Lee,L.S., Pellicer,A., Cheng,Y.C. and Axel,R. (1977) Cell, 11, 223-232. Zawel,L. and Reinberg,D. (1992) Curr. Opin. Cell Biol., 4, 488-495. Zhou,Q., Lieberman,P.M., Boyer,T.G. and Berk,A.J. (1992) Genes Dev., 6, 1964-1974. Zhou,Q., Boyer,T.G. and Berk,A.J. (1993) Genes Dev., 7, 180-187.
Received on August 4, 1993; revised on September 27, 1993