The TATA Motif Specifies the Differential Activation of Minimal Promoters by Varicella Zoster Virus Immediate-early Regulatory Protein IE62*

The immediate-early IE62 protein of varicella zoster virus is an acidic transcriptional activator capable of up-regulating many viral and cellular promoters with varying efficiencies. We demonstrate that, in the context of a minimal promoter, a TATA element is both sufficient and essential for IE62-mediated transcriptional activation. Differential levels of activation by IE62 in this context were conferred by a panel of naturally occurring sequence variations within the TATA motif itself. TATA motif-specific, differential induction was not obtained when the IE62 acidic activation domain was targeted as a GAL4 fusion protein to the same panel. The prototype acidic transactivator, VP16 of herpes simplex virus, failed to discriminate between these different TATA motifs when they were placed into an appropriate responsive promoter context. Nonetheless, a chimeric IE62 polypeptide substituted with the VP16 activation domain retained the ability to differentially modulate minimal promoters with various TATA motifs. Taken together with its binding to TATA box-binding protein (TBP) and transcription factor IIBin vitro, we suggest that IE62 has the unusual ability to achieve differential levels of transcriptional activation through different TATA motifs, which may be accomplished either directly or indirectly by recognizing conformational variations in DNA-bound TBP, TBP-transcription factor IIA/B, or TBP-TATA-associated factor complexes.

The immediate-early IE62 protein of varicella zoster virus is an acidic transcriptional activator capable of up-regulating many viral and cellular promoters with varying efficiencies. We demonstrate that, in the context of a minimal promoter, a TATA element is both sufficient and essential for IE62-mediated transcriptional activation. Differential levels of activation by IE62 in this context were conferred by a panel of naturally occurring sequence variations within the TATA motif itself. TATA motif-specific, differential induction was not obtained when the IE62 acidic activation domain was targeted as a GAL4 fusion protein to the same panel. The prototype acidic transactivator, VP16 of herpes simplex virus, failed to discriminate between these different TATA motifs when they were placed into an appropriate responsive promoter context. Nonetheless, a chimeric IE62 polypeptide substituted with the VP16 activation domain retained the ability to differentially modulate minimal promoters with various TATA motifs. Taken together with its binding to TATA box-binding protein (TBP) and transcription factor IIB in vitro, we suggest that IE62 has the unusual ability to achieve differential levels of transcriptional activation through different TATA motifs, which may be accomplished either directly or indirectly by recognizing conformational variations in DNA-bound TBP, TBP-transcription factor IIA/B, or TBP-TATA-associated factor complexes.
Varicella zoster virus (VZV) 1 causes chickenpox as a primary infection, and shingles as a reactivated infection. The lytic cycle of this virus begins with the expression of its major immediateearly protein, IE62, a DNA-binding trans-regulatory protein containing 1310 amino acid residues with a predicted molecular mass of 140 kDa (1). IE62 is a potent transactivator that stimulates transcription not only of target VZV genes, but also a variety of heterologous viral and cellular genes (2,3). Although there is no definitive genetic proof as yet that IE62 is essential for VZV replication, its ability to regulate the expression of VZV genes of all three putative kinetic classes (immediate-early, early, and late) as well as its sequence homology and functional similarity to the essential ICP4 transactivator of herpes simplex virus (HSV) argue that it serves a critical role in the VZV replicative cycle. Both VZV IE62 and HSV-1 ICP4 bind to specific ATCGTC DNA sequences at the cap site of their own promoters, and these interactions are thought to be involved primarily in negative transcriptional autoregulation. However, the mechanism by which these proteins up-regulate the transcription of responsive viral promoters has not yet been well defined.
Accumulating evidence indicates that transcriptional activators, in general, enhance transcription by influencing or stabilizing functional interactions among the general transcription factors including TATA box-binding protein (TBP), TFIIA, and TFIIB, that are essential for the formation of a functional pre-initiation complex (reviewed in Refs. 4 -12). In addition, the in vitro function of transcriptional activators requires a number of co-factors that are not essential for basal transcription in minimal reconstituted systems. Many of these co-factors are physically associated with the TBP polypeptide (TATAassociated factors or TAFs) and constitute the holoTFIID fraction along with TBP. Certain transactivators have been shown to interact specifically with TBP and one or more TAFs, and these interactions appear to function as molecular adapters that bridge between the transactivator and the general transcription initiation machinery and serve to recruit TBP and enhance a rate-limiting step in transcription (reviewed in Refs. 5, 8, and 10).
An intriguing aspect of IE62 function is its capacity to activate in trans a diverse array of promoters that lack any apparent conserved motifs in their promoter and upstream regions (2,3,13). In addition, the magnitude of responsiveness to IE62 varies widely from promoter to promoter. Curiously, in both HSV and VZV, most of the immediate-early and early class promoters tend to have classical close-to-consensus TATA motifs, whereas late class promoters typically have highly divergent, non-consensus (but still AT-rich) TATA-like motifs. Both types of promoters respond to IE62 and ICP4, although in HSV other accessory viral factors clearly contribute to both maximal and appropriate temporal responses of late promoters.
Previous deletion analysis of the VZV IE62 promoter revealed that a minimal reporter construct possessing the TATA box with only 15 base pairs (bp) of adjacent upstream sequence retains responsiveness to IE62-mediated induction but not to its HSV homolog ICP4 (13). Because this upstream sequence lacks consensus recognition motifs for binding to IE62 or any other known transcription factors, we have focused our atten-tion on the TATA element itself as a potential key factor in mediating the transcriptional effects of IE62. In this report, we present evidence to suggest that a novel TATA motif-dependent mechanism is both essential and sufficient for IE62-mediated transcriptional activation.

EXPERIMENTAL PROCEDURES
Target Reporter Gene Plasmids Containing Panels of Different TATA Motifs-Plasmids p62CAT⌬258 and p62CAT⌬45 have been described previously (13). In pIFNTATA, a 60-bp segment (TTGTCCCTATTTA-AGAGAGATGTACACAGCAGGCTCTCAGAGAACCTGTAGGAGAA-ACT) extending from Ϫ38 to ϩ22 of the murine ␣4 interferon gene was placed immediately upstream from the coding region of the chloramphenicol acetyltransferase (CAT) reporter gene (14). The TATA element and the putative initiator element (Inr), respectively, of the interferon gene are shown in bold print. In pIFN⌬TATA and pIFNTATA⌬CAP, either 5 or 3 bp of the TATA or the Inr, respectively, were deleted by site-specific mutagenesis. In pIFNTATAMCAP the putative Inr element was changed from TCTCAG to TCTAAG. To create a TATA only synthetic promoter (pTATAx5CAT), a 44-bp oligonucleotide consisting of five tandem copies of the ␤-casein TATA sequence TATATATA was inserted at the XbaI site upstream of the CAT coding region in the plasmid pCAT-B (Basic, Promega Corp.). A parallel control plasmid (p⌬TATAx5CAT) contained a 44-bp oligonucleotide consisting of five tandem copies of the mutant sequence TATCGATC. In pE1bTATA, a single copy 12-bp (TAATATAGGAGA) synthetic oligonucleotide bearing the adenovirus E1b gene TATA motif was inserted immediately upstream of the CAT gene (15).
For analysis of the responsiveness of diverse TATA elements to IE62, a panel of target constructs was created. To do so, the TATA element in the pIFNTATA target plasmid was modified by site-specific mutagenesis to resemble the TATA motifs of several viral and cellular genes. For example, p(T)TTTTAA bears the TTTTAA motif of the VZV IE62 gene; p(T)ATTAAA bears the ATTAAA motif of the VZV thymidine kinase (TK) gene; p(T)ATTTAAATT bears the ATTTAAATT motif of the VZV glycoprotein C gene; pTATAAAA bears the TATAAAA motif of the adenovirus MLP and of the HSP90 genes; pTATATATA(A) bears the TATATATA motif of the cellular ␤-casein gene (16); and pTATATAA bears the TATATAA motif of the human cytomegalovirus (HCMV) major immediate-early (MIE) promoter. The distance between the TATA box and Inr element (25 bp) remained unchanged for TATTAAA, TTTTTAA, TATAAAA, and TATATAA mutants. However, this distance was reduced by 2 bp for TATATATAA mutant and by 3 bp for TATT-TAAATT mutant.
Effector Genes and CAT Assays-The pCMV62 plasmid expresses the VZV IE62 protein (3) from the HCMV MIE promoter. The pRL45 plasmid, which expresses both the immediate-early proteins IE1 and IE2 of HCMV, has been described previously (18). Plasmid p62GAL (9 -120) expresses the chimeric protein GAL4-IE62 containing the activation domain of IE62 (aa 9 -120) fused to the GAL4 DNA binding domain (aa 1-147) and has also been described previously (19). The VP16 plasmid (pGH62) expresses the intact HSV-1 VP16 effector protein from its cognate promoter (17). The plasmid pGH114 has been described earlier (13) and expresses ICP4 from the simian cytomegalovirus immediate early promoter. The pPRV180 plasmid expresses the IE180 of pseudorabies virus and was a gift from P. Sheldrick. The E1A expression plasmid (20) was a gift from G. Nabel. The plasmid pCMV62/ 16AD is a derivative of pCMV62 and contains the coding region of HSV VP16 activation domain (aa 411-455) obtained as a BglII-EcoRI fragment from pVP16 (CLONTECH), cloned in-frame 5Ј to the IE62 coding segment extending from codon 96 to 1310. Effector and target reporter gene plasmid DNAs were electroporated into A3.01 cells (a human T lymphocytic cell line). The CAT activity was determined as described previously (3), quantitated with a PhosphorImager scanner (Molecular Dynamics), and expressed as percentage of conversion of [ 14 C]chloramphenicol to its acetylated derivatives.
RNase Protection Assay-In RNase protection assays, 40 g of total cellular RNA extracted from transfected A3.01 T lymphocytic cells were hybridized with an antisense RNA probe corresponding to CAT gene for 12 h at 56°C in 40 mM PIPES (pH 6.7), 350 mM NaCl, 1 mM EDTA, and 80% formamide. As internal controls, antisense RNA probes corresponding to the CD4 cell surface marker and two housekeeping genes, L32 ribosomal gene and glyceraldehyde-3-phosphate dehydrogenase, were also included. Samples were then treated with RNase A (40 g/ml) and RNase T1 (2 g/ml) for 45 min at 30°C. The reactions were terminated by adding 50 g of proteinase K and 10 l of 10% SDS and incubated at 37°C for 15 min. After phenol-chloroform extraction and ethanol precipitation, the samples were resolved on 6% denaturing polyacrylamide gel and subjected to autoradiography as well as phosphorimage analysis.
In Vitro Protein Matrix Affinity Binding Assays-Three overlapping segments (aa 9 -410, 234 -734, and 734 -1310) spanning nearly the entire length of the 1310-aa IE62 protein, were cloned into an in vitro transcription/translation vector (21) to generate portions of the IE62 polypeptide in a rabbit reticulocyte lysate system. Agarose beads coupled to purified intact 33-kDa human TFIIB synthesized in Escherichia coli were purchased from Santa Cruz Biotechnology (catalog SC4001AC). TBP-N (N-terminal) and TBP-C (C-terminal) GST-fusion proteins were expressed in E. coli and purified by incubating the bacterial supernatants with glutathione-Sepharose beads as described previously (22). For analysis of purity of bead-bound fusion proteins, the beads were boiled in sample buffer and subjected to SDS-PAGE, followed by visualization of proteins with Coomassie Blue stain. Equal amounts (25 g) of bead-bound proteins were incubated with in vitro translated, 35 S-labeled polypeptides representing segments of the IE62 protein. The binding buffer contained 140 mM NaCl, 50 mM Tri-HCl, 0.5% Triton X-100, 2 mM dithiothreitol, and 100 g/ml ethidium bromide to eliminate nonspecific protein DNA-interactions (23). After extensive washing to remove the unbound material, the beads were resuspended in protein sample buffer, boiled for 5 min, and resolved by SDS-PAGE.
Electrophoretic Mobility Shift Assay (EMSA)-Electrophoretic mobility shift assays with purified recombinant human TATA-binding protein TBP synthesized in E. coli (Promega TFIID, catalog no. E3081) were performed with synthetic, double-stranded oligonucleotides 34 bp in length. The entire sequence of the top strand of each TATA motif oligonucleotide evaluated by EMSA is shown in Table I. The binding reactions were performed in a buffer containing 10% glycerol, 20 mM Tris-HCl (pH 7.0) 80 mM KCl, 10 mM MgCl 2 , and 2 mM dithiothreitol. As nonspecific carrier DNA, 1 g of poly(dG-dC):poly(dG-dC) was included in the binding reaction. One hundred femtomoles of each 32 P-labeled, double-stranded oligonucleotide pair were incubated with indicated amounts of purified bacterially expressed human TBP (Promega) at room temperature for 20 min, loaded onto a nondenaturing 5% polyacrylamide gel, and then electrophoresed for 4 h at 150 V. The authenticity of the commercial TBP preparation was confirmed using an antibody specific for human TBP (Upstate Biotechnology Inc.) to supershift the complexes formed with TATA box containing oligonucleotide. In competition experiments, 100 fmol of the 32 P-labeled, doublestranded HCMV MIE TATATAA oligonucleotide probe was first incubated with 5 ng of TBP for 10 min and then the competitor oligonucleotide was added and incubated for another 10 min. The electrophoresis buffer was 0.5ϫ TBE with 2.5 mM MgCl 2 and 0.05% Nonidet P-40.

RESULTS
The TATA Element Is Essential for IE62-mediated Transcriptional Activation-The immediate-early protein encoded by the open reading frame 62 constitutes the major transcriptional regulatory protein of VZV. However, the mechanistic aspects of its transcriptional activity remain poorly understood, although its HSV counterpart ICP4 has been extensively studied. Our previous studies indicated important differences between these two related regulatory proteins of alphaherpesviruses. For example, unlike ICP4 which represses its cognate promoter activity, IE62 augments the transcription from its own promoter as well as that of the ICP4 promoter. In addition, despite ICP4's repressive effects on its own promoter, it enhanced the transcription from the IE62 promoter. However, the upstream cis-elements required for the induction of IE62 promoter by IE62 and ICP4 differed (13). Confirming our previous observations, as shown in Fig. 1 (panel A), the IE62 promoter with 258-bp upstream sequences with reference to its transcription start site (p62CAT⌬258) responded positively when cotransfected with either IE62 or ICP4 (120-or 25-fold, respectively). In contrast, an IE62 promoter version having only 15-bp upstream sequences together with its TATA box (p62CAT⌬45) responded only to IE62 (36-fold).
Scrutiny of the IE62 promoter region in p62CAT⌬45 revealed no recognition motifs for binding to IE62 or any other known transcription factors other than the TATA box. Considering that IE62 activates a wide array of viral and cellular promoters and the fact that the core promoter element-TATA box is a common motif present in all these responsive promoters, we focused on the TATA element as a potential mediator of the IE62 transcriptional activity.
To explore the contribution of the TATA element to IE62mediated induction, a CAT reporter gene containing a 60-bp minimal promoter derived from the murine ␣4 IFN promoter elements (pIFNTATA) (14) was electroporated into the human T-lymphocytic cell line A3.01 with or without a VZV IE62 effector plasmid. T lymphocytes have been implicated as important vehicles for the growth and spread of VZV in vivo (see Ref. 3 and references therein), and specific T cell tropism for VZV has been demonstrated (24). Our previous studies with VZV promoter targets were also carried out in these cells (3,13,19).
The minimal pIFNTATA promoter target was induced by the cotransfected IE62 effector plasmid ( Fig. 1, panel B), which expressed the intact IE62 protein under the control of the HCMV MIE promoter. Neither a control plasmid containing the HCMV MIE promoter region alone (data not shown), nor plasmids expressing the IE62 homologs ICP4 of HSV and IE180 of pseudorabies virus, the ICP0 protein of HSV, or the adenovirus E1A protein, resulted in any obvious increase in CAT activity when cotransfected with the target plasmids. The HCMV IE1/IE2, however, was able to activate pIFNTATA significantly ( Fig. 1, panel B).
Two synthetic minimal core promoters consisting, first, of a 12-bp TAATATA element from the adenovirus E1b promoter in pE1bTATA CAT (15) and, second, one containing three tandem copies of either a wild-type TATATATA motif only or a ⌬TATA mutant motif inserted 5Ј to the CAT reporter gene in pCAT-B vector (Promega Corp.) were also tested. Co-transfection with VZV IE62 gave 4-fold induction of pE1bTATA CAT and 26-fold induction of p(TATA)3 CAT-B, but gave no effect on either the pCAT-B or p(⌬TATA)3 CAT targets (data not shown).
All three minimal promoters used above lacked any recognized upstream elements other than the TATA boxes, but pIFNTATA does contain a putative pyrimidine-rich Inr element (CTCTCAG), of a type that can substitute for TATA in some circumstances. Therefore, to assess the relative contributions of the TATA element and the Inr element in IE62-mediated transactivation, we performed transient expression assays with a set of minimal pIFNTATA derivatives in which both of these elements were individually mutated ( Fig. 1, panels B and C). Deletion of 5 bp of the ␣-IFN TATA element itself (pIFN⌬TATA) led to complete ablation of IE62 responsiveness ( Fig. 1, panel B), while deletion or mutation of the Inr element FIG. 1. A, the cis elements in VZV IE62 promoter necessary for positive modulation by VZV IE62 and HSV ICP4 are separable. The two reporter constructs p62CAT⌬258 and p62CAT⌬45 contain promoter sequences of VZV open reading frame 62 gene extending up to 258 and 45 bp upstream of its transcription start site respectively. B, the presence of a TATA element is both sufficient and essential for the activation of minimal promoters by VZV IE62. The pIFNTATA target plasmid contains a 60-bp minimal promoter derived from the mouse ␣4 interferon promoter elements between Ϫ38 and ϩ22 driving the expression of a CAT reporter gene. The pIFN⌬TATA plasmid is a derivative of pIFNTATA with a 5-bp deletion in the TATA element. C, IE62-mediated activation of pIFNTATA target promoters containing mutant and deleted Inr elements. The pIFNTATA⌬CAP and pIFNTATAMCAP plasmids are derivatives of pIFNTATA with either a 3-bp deletion or a single point mutation in the initiator element (CAP) as described under "Experimental Procedures." The effector proteins VZV IE62 and HSV ICP4 were expressed from pCMV62 and pGH114, respectively (3), and are driven by the CMV immediate-early promoters. The IE1 and IE2 effector proteins of HCMV were expressed from plasmid pRL45 (18). Other effector plasmids used are described under "Experimental Procedures." Equal amounts of the effector and target plasmid DNAs (10 g) were electroporated into the A3.01 human T-lymphocytic cell line, and the CAT activity was measured as described under "Experimental Procedures." (pIFNTATA⌬CAP and pIFNTATAMCAP) did not significantly affect the IE62 responsiveness of the minimal promoter (Fig. 1,  panel C). Although pIFN⌬TATA failed to respond to IE62, it still retained the ability to respond to HCMV IE1/IE2. Effector plasmid pRL45 expresses both IE1 and IE2 splice variants, and it is likely that the pIFN⌬TATA induction seen with pRL45 is actually due to the IE1 polypeptide since IE1 polypeptide but not IE2 has been shown to enhance transcription from TATA less promoters (25). The above experiments imply that a TATA element is required for responsiveness of a minimal promoter to IE62, but not to certain other viral activators.
Naturally Occurring Sequence Variations within the TATA Motif Confer Differential Induction Levels-IE62 is a potent activator of diverse eukaryotic promoters, although the extent of responsiveness to it varies widely from promoter to promoter. For example, the VZV TK promoter was induced 125fold by IE62, whereas the VZV gpC promoter was refractory to it (2, 3). These observations, together with the above data regarding the importance of the TATA element to IE62 responsiveness, led us to consider that the different TATA motifs themselves, rather than different upstream elements, may be most important in modulating the level of IE62 responsiveness of a promoter. To test this hypothesis, the TATA element within the mouse ␣4 IFN promoter (TATTTAA) in the pIFN-TATA plasmid was substituted for by the TATA elements derived from several other well defined cellular and viral promoters (Table I and Fig. 2).
Most of the modifications of the mouse ␣4 IFN promoter TATA element did not result in obvious changes in the basal activity of the promoter itself, within the limits of sensitivity of the CAT assays as shown in Fig. 2 (ranged between 0.3% and 0.5% acetylation of chloramphenicol substrate) or when determined by a sensitive RNase protection assay for in vivo expressed CAT mRNA (see below). Nonetheless, the panel of minimal promoter constructs bearing substituted TATA elements displayed dramatic differences in the magnitude of their responsiveness to activation by VZV IE62. For example, conversion of the IFN TATTTAA motif to resemble the TTTAAATT sequence from the VZV glycoprotein C gene almost abolished IE62 responsiveness (17-fold decrease) and substitution with TTTTAA from the VZV IE62 promoter reduced induction 4-fold, whereas substitution with the TATAAAA consensus sequence of the Ad2 MLP and HSP90 promoters conferred a 258-fold activation of CAT expression in the presence of IE62. In fact, where direct comparisons could be made (e.g. TK, gpC, and IE62) the relative degree of responsiveness of each reporter construct to IE62 (54-, 3-, and 13-fold, respectively) resembled the previously reported (3,13) order of responsiveness to IE62 of the original promoter regions from which the different TATA elements were derived. These experiments suggested that the TATA motif itself greatly influences overall IE62 responsiveness.
To validate the differential activation of various TATA modified promoter constructs by IE62, it is crucial to assess whether the substitution of different TATA motifs influenced the basal promoter activity appreciably. Therefore, the basal activity of each minimal promoter construct examined in Fig. 2, was directly measured by a sensitive RNase protection assay following transfection of TATA modified promoter CAT plasmid DNA into A3.01 T cells. As evident from the data in Fig. 3, all minimal promoter versions were transcriptionally active, including the pIFN⌬TATA (which still contains the core promoter Inr element to initiate transcription), resulting in appropriately sized protected fragment with the CAT probe (visible after 6-day exposure). Probes for housekeeping genes L32 ribosomal gene and glyceraldehyde-3-phosphate dehydrogenase gene as well as the abundantly expressed CD4 gene were included as controls for RNA integrity and quantitation standards. Upon quantitation of CAT transcripts from each promoter, the differences in intensity were less than 2-fold. For example, the intensity of the protected fragment generated from the consensus TATATAA motif differed from that of the TTTTTAA and ⌬TATA motifs by 20% and 36%, respectively. Although this was somewhat surprising, a review of published literature revealed that, of seven motifs examined in Fig. 3, five motifs (TATATAA, TATATATAA, TATAAAA, TATTTAA, and TATTAAA) have previously been shown to confer high transcriptional activity (26,27), and our results are thus concordant with the published data. However, the remaining two motifs, TTTTTAA and TATTTAAATT, from VZV IE62 and gC promoters, respectively, to our knowledge have not been eval-   GGACTTTGTCCCTACAAAGAGAGATGTACAG uated previously. Nonetheless, the abundant expression of both IE62 and gC polypeptides despite their highly divergent TATA motifs during VZV infection very likely indicate their transcriptional competence. Taken together, these data suggest that within the limits of sensitivity of each assay employed the modifications of TATA elements assessed in this study do not appreciably alter the basal activity of the minimal promoter and the differential activation seen in the presence of IE62 with various TATA motifs is most likely to be a function of the VZV IE62 itself. TATA Motif-dependent, Differential Induction Is Not Conferred by the Activation Domain of IE62-We next sought to determine whether TATA motif-dependent, differential induction is conferred by the activation domain of IE62 itself or by some other segment of the IE62 protein. This is of special relevance in the light of recent identification of a coactivator protein (PC4) that mediates acidic activator transcription and contains SEAC motifs in common with VZV IE62 and many other viral transactivators (28,29). The strong acidic activation domain of IE62 is localized to the negatively charged N-terminal region (aa 9 -120) of the 1310-amino acid protein (19,30). Previously, we have shown that the IE62 activation domain (aa 9 -120) when expressed as a fusion protein with the yeast GAL4 DNA-binding domain (GAL4-IE62) efficiently activates promoters with upstream GAL4 binding motifs but not promoters without such sites (19). In addition, the deletion of this single activation domain of IE62 results in ablation of transcriptional activity, although the deletant molecule is still capable of competitively interfering with the transcriptional activity of the native IE62 (19). To assess whether the activation domain itself is responsible for both transcriptional activation as well as sensing the differences in the TATA motif leading to differential responsiveness, five tandem GAL4-binding sites were placed upstream of the TATA box in each of the panel of TATA substituted promoters described earlier in Fig. 2. The insertion of GAL4 binding sites neither affected the basal activities of the promoters nor their ability to respond differentially to native IE62 when cotransfected (data not shown). When this panel was cotransfected with the GAL4-IE62 chimeric plasmid, all promoters responded vigorously (Fig. 4). However, the levels of activation varied by less than 2-fold upon substitution of the consensus TATA motif with a variety of other non-consensus TATA motifs, which in the context of minimal promoters lacking the GAL4-binding sites led to nearly 100-fold TATA-dependent differences in activation by the full-length IE62 protein. Although these findings suggest that the activation domain is not sufficient to impart differential induction under these circumstances, one could argue that the tethering of activation domain via a GAL4 bridge may not represent natural conformation or recruitment of IE62. However, if a region other than the activation domain of IE62 is responsible for the TATA motif-dependent, differential activation of promoters, then, theoretically, if the modular activation domain of IE62 is replaced with a heterologous activation domain, the chimeric protein might still be able to display TATA motif-dependent differential effects mimicking a more natural setting if such a molecule is transcriptionally functional. In selecting an appropriate heterologous activation domain, the preponderance of acidic residues, as well as the comparable sizes and transcriptional strengths of the modular domains (19), favored VP16, the prototype of acidic transcriptional activators (31). However, it was crucial that the substituted activation domain be free of any inherent TATA motif-dependent effects. Although HSV VP16 remains one of the most extensively studied viral transcriptional activators, the influence of the TATA motif, if any, for VP16-mediated transcriptional activity has not been described. We therefore first examined whether the TATA motif influences VP16-mediated transcriptional activity.
The promoter for the HSV immediate-early gene encoding the ICP0 protein is efficiently activated by VP16. VP16 responsiveness is mediated primarily by an interaction with the cellular Oct-1 protein and subsequent binding to multiple upstream Oct-1/TAATGARAT DNA elements in the ICP0 promoter (32,33). As shown in Fig. 5 (panel A), VP16-mediated activation of the ICP0 promoter was changed minimally (range: 11-15-fold) by substitution of its TATAAG element with several of the TATA motifs examined earlier in Fig. 2. Nonetheless, it still remains possible that the apparent insignificance of the TATA element for the VP16-induced activation of ICP0 promoter resulted from compensatory effects of other transcription factors such as Sp-1 that binds to the upstream elements in this complex promoter. To address this issue, we devised a strategy to direct VP16 to the minimal promoter panel used in Fig. 2. A 23-bp synthetic Oct-1/TAATGARAT element was inserted immediately upstream of the TATA box. As shown in Fig. 5 (panel B), when this panel was cotransfected with a VP16 expression plasmid, once again there was no differential TATA motif-dependent induction of the target plasmids. Equally important was the fact that in the context of a minimal promoter, or for that matter in the context of a natural target promoter (ICP0 promoter), VP16 was capable of inducing transcription in the absence of a TATA box.
Having confirmed that VP16 is devoid of any inherent TATA motif-dependent effects, a chimeric IE62 with VP16 activation domain substituting the cognate activation domain of IE62 was constructed (IE62/16AD). When IE62/16AD was co-expressed with the minimal promoter panel examined in Fig. 2, with modified TATA elements, the induction of the minimal promoters were differentially modulated as was seen with the native FIG. 3. Measurement of basal promoter activity by RNase protection assay. Twenty micrograms of each plasmid DNA were transfected into A3.01 cells, and after 36 h cellular RNA was extracted. Forty micrograms of cellular RNA were hybridized with a 209-nt, 32 P-labeled, antisense riboprobe corresponding to CAT gene. The expected size of the protected fragment is 180 nt. As internal controls labeled antisense riboprobes for CD4 (191 nt in length and when protected yields a 162-nt fragment), L32 ribosomal (141 nt in length and when protected yields a 113-nt fragment) and glyceraldehyde-3-phosphate dehydrogenase gene (124 nt in length and when protected yield a 96-nt fragment) were also included in the RNase protection assay. The protected RNA fragments were resolved on 6% sequencing gel and subjected to autoradiography and phosphorimage analysis. IE62 (compare Fig. 2 versus Fig. 6). These results taken together with the lack of TATA specificity obtained with VP16 (Fig. 5), and also by the HCMV IE1/IE2 proteins (see Fig. 1,  panel B), supports the notion that TATA motif-dependent, differential induction is an inherent property of IE62 and not a general property of viral transactivators and a region other than the activation domain of IE62 dictates this unique phenomenon.
A Region of IE62 That Interacts with TBP and TFIIB in Vitro Maps Outside of the Activation Domain-Given the clear involvement of TATA element in IE62-mediated transcriptional activation, we assessed whether IE62 physically interacts with the essential components of the basal transcriptional machinery that assemble on the TATA element. The results of proteinprotein affinity binding studies, shown in Fig. 7, indicate that it does interact with both TBP and TFIIB. When various in vitro translated, 35 S-labeled, VZV IE62 polypeptide segments were incubated with purified, bacterially expressed TBP or TFIIB immobilized on beads, the IE62 protein bound to TFIIB (lanes 2) and to the evolutionarily conserved C-terminal segment (lanes 4) of TBP (GST/TBP-C) but not to its N-terminal segment (lanes 3) (GST/TBP-N). The region of IE62 that interacted most strongly with the basal transcription factors mapped between aa 234 and 734, well downstream from the only defined activation domain, between aa 9 and 120. A weaker interaction also occurred with the aa 734 -1310 region, suggesting that IE62 possesses multiple domains that contact basal transcription factors. This situation contrasts with the results for several other well defined viral transactivators such as HSV VP16, Ad2 E1A, and EBV Zta, which bind strongly to the C-terminal basic repeat domain of TBP in vitro directly through their activation domains (34 -37). The HCMV IE2, however, resembles VZV IE62 in that it binds to TBP strongly through a region that is independent of its acidic activator domain (22,38).
Nonetheless, the activation domain is critical for transcriptional activity of the full-length, 1310-amino acid IE62 protein, since the deletion of 85 amino acids extending from codon 10 to 95 (IE62⌬AD) resulted in complete ablation of the transcriptional activity (Fig. 8). Moreover, consistent with the notion that high affinity binding to both TBP and TFIIB are integral to transcriptional activity of IE62, an 80-amino acid deletion (aa 416 -506) within this region (IE62⌬416 -506) resulted in loss of transcriptional activity. The expression and intracellular distribution profiles of both mutants IE62⌬AD and IE6262⌬416 -506 were identical to that of full-length, wild-type IE62 (data not shown), and both mutants competitively inhibited the transcriptional activity of full-length IE62 as shown in Fig. 8. When 10 g of IE62 plasmid was cotransfected with either 5 or 15 g of IE62⌬AD, the induction of the reporter gene was reduced by 70% or 90%, respectively. Similarly, cotransfection of IE62 plasmid with 5 or 15 g of IE62⌬416 -506 resulted in 60% or 85% reduction in wild-type IE62-mediated inducibility, suggesting that the transcriptional-deficient mutants are still capable of efficiently sequestering cellular components involved in IE62-mediated transcription. Alternate possibility that the competitive inhibition may operate at the DNA binding step rather than at the level of transcription is unlikely since the target construct pIFNTATA lacks any IE62 binding ATCGTC sites.

TATA Motifs That Confer Widely Differing Responsiveness to IE62 Display Comparable Levels of Binding to TBP in Vitro-
TFIID is a high molecular weight, multiprotein complex consisting of TBP, which specifically recognizes and binds to the TATA boxes of class II gene promoters (39,40), and a number of associated factors (TAFs). While both consensus (TATAAA or TATATAA) and non-consensus TATA elements bind TFIID (41), it has been suggested that the strength of the interaction between TFIID and the TATA element correlates with the basal promoter activity (42). Certain transcriptional activators such as Zta and ICP4 of EBV and HSV, respectively, induce transcription by enhancing or stabilizing TBP binding to the TATA box. Promoters with TATA elements that bind TBP suboptimally are usually responsive to these activators but not the promoters with TATA elements that avidly bind TBP (35,43). It should be noted, however, that the promoters that responded to IE62 most vigorously, in contrast, possessed apparently consensus or near consensus TATA motifs, for example TATATATAA, TATATAA, or TATAAAA. Although only limited information exists in literature as to the binding affinities of TBP to various TATA motifs (41,43), two of the motifs i.e., TATAAAA and TATATATAA, that conferred high activity in the presence of IE62 have previously been shown to display apparent dissociation constants (K d ) of 2 ϫ 10 Ϫ9 M, a value that is among the highest reported for any TATA motif (41). Since no information exists in the literature with regard to TBP binding efficiency to other TATA motifs evaluated in our study, we assessed the ability of each of the motifs to bind purified recombinant human TBP in an in vitro electrophoretic mobility shift assay to examine any possible link between TBP binding to a particular TATA motif and the level of its responsiveness to IE62. The synthetic oligonucleotides representing different TATA motifs used as probes in the EMSA are listed in Table I. An additional probe with TACAAA motif of adenovirus EIIa promoter, which was originally used by Huang et al. (42) to demonstrate minimal TBP binding by DNase foot printing assay, was included in our TBP EMSA to validate our assay conditions. First, the amount of TBP added to the binding reaction was titrated (20, 8, and 4 ng of TBP) against a fixed amount of each probe (100 fmol) and the fraction of probe that FIG. 4. The N-terminal acidic activation domain of IE62 is not sufficient to mediate TATA motif-dependent, differential modulation. A 100-bp oligonucleotide cassette containing five tandem GAL4 binding sites (68) was placed upstream of the TATA box in the same panel of target reporter genes used earlier in Fig. 2. The chimeric protein GAL4-IE62 expressed from p62GAL(9 -120) contains the activation domain of IE62 (aa 9 -120) fused to the DNA binding domain (aa 1-147) of yeast GAL4 protein. Because of the sensitivity of the targets to the GAL4 chimeric effector, the amounts of the target and effector DNAs used in co-transfections were reduced to 100 ng each. complexed with the added TBP was quantitated. As shown in Fig. 9 (panel A), at high concentrations of TBP (20 ng), all TATA motifs used in the present study displayed significant TBP-binding profiles, with a difference less than 3-fold between the highest (TATATAA) and the lowest (TATTTAA). A similar trend was evident at low TBP (4 ng) concentration as well. The apparent reduction in binding to TATTAAA probe in Fig. 9 at low TBP concentrations, however, was not reproducible. In contrast, the probe with a deletion in the TATA motif (⌬TATA) failed to elicit any TBP binding at all. Having failed to detect any gross differences in direct binding of TBP to our panel of TATA motifs, we sought to confirm this observation by a competition binding assay in which TBP binding to the consensus TATATAA motif was competed with a 100-fold excess of unlabeled competitor sequence. Again, as seen in Fig. 9 (panel B), no significant differences were noted. Thus, no definitive correlation between the binding of TBP to specific motifs and the extent of IE62-mediated activation could be discerned. For example, the TTTTTAA motif from the IE62 promoter and the TATTTAAATT motif from the glycoprotein C promoter bound TBP strongly, yet the ␣IFN-CAT reporter genes containing these elements responded weakly to IE62 (13-and 3-fold, respectively). The fact that the most vigorous IE62-mediated activation was achieved with TATA motifs that display the highest affinities toward TBP, taken together with the observation that the motifs that were relatively refractory to IE62 induction still bound TBP efficiently, therefore, argue in the case of IE62, for a mechanism that is fundamentally different from what has been proposed for ICP4 and Zta transcriptional activators.

DISCUSSION
The VZV IE62 transcriptional regulatory protein possesses an unusual, TATA motif-dependent mechanism by which it mediates activation of simple, minimal eukaryotic promoter targets. Large variations in the levels of activation of various promoters tested were evident according to which particular TATA sequence was present in them. The one other viral transactivator known to activate minimal promoters via a TATA-dependent mechanism is the adenovirus E1A protein. E1A, however, exhibits an absolute requirement for one specific motif, namely TATAA, for transactivation to occur (44,45). Similarly, the muscle-specific enhancer of the myoglobin gene has been shown to function in the presence of its cognate TATAAAA motif, but not with the TATTTAT element from the SV40 promoter (46). The activity of cellular transcription factor ATF has also been shown to be sensitive to TATA motif changes (47).
The results presented here argue that VZV IE62 acts primarily by a mechanism that does not involve or require upstream targeting sites or factors and that it is unusually sensitive to the nature of the specific TATA motifs present. Such a mechanism differs significantly from the strong and relatively specific requirements for upstream targeting by other well studied viral and eukaryotic transactivators. For example, HSV VP16, Ad2 E1A, and EBV Zta interact in vitro with TBP and TFIIB (and TAFs) through their activation domains, whereas VZV IE62 interacts with TBP and TFIIB through a region(s) that is distinct from its activation domain. VP16 and E1A principally target upstream sites through indirect protein: protein interactions with Oct-1, and Rb:E2F or ATF-2, respectively, whereas Zta binds directly to upstream ZRE motifs and acts as a bridge between these sites and TFIID, resulting in TFIIA-dependent stabilization of TBP binding to weak TATA motifs such as GATAAA (35,48). The HCMV IE1/IE2 transactivators are also clearly different from IE62 in that they target promoters lacking TATA motifs. Both HCMV IE2 and HSV ICP4 also interact with TBP and TFIIB in vitro through regions that are distinct from their activation domains (22,49). Although ICP4 represses upstream activator-mediated transcription including that of Sp-I, presumably through formation of a tripartite complex consisting of TBP, TFIIB, and ICP4 itself, the activation induced by ICP4 correlates with its interaction with TFIID perhaps via TAF250 requiring the C terminus of ICP4 (49 -51). IE62 also differs from ICP4 in possessing a powerful N-terminal activator domain that functions effectively in a GAL4 fusion protein in comparison to a weak, promoter-specific modular activation domain recently reported in ICP4 (19,52). In addition, our previous work (13) showing FIG. 5. Lack of TATA motif specificity for VP16-mediated activation. A, in the ICP0-CAT gene, the VP16-responsive promoter elements extending from Ϫ800 to ϩ120 of the ICP0 HSV immediate-early gene drive the expression of a CAT reporter cassette. The plasmids pICP0CAT⌬TATA, pICP0CATMTTTTTAA, pICP0CATMTATATAA, and pICP0CATMTATAAAA are derivatives of ICP0-CAT in which the native TATAA element was changed by site-directed mutagenesis to introduce a variety of different TATA motifs matching those used in Fig. 2. The VP16 plasmid used expresses intact HSV VP16 protein from its own promoter. Nevertheless, because of the high sensitivity of ICP0-CAT to VP16-mediated activation, the target and the effector DNAs used in the co-transfections were reduced to 2 g each. B, a 23-bp synthetic Oct-1/TAATGARAT cassette was inserted immediately upstream of the TATA box in the panel of minimal promoter plasmids with different TATA motifs shown in Fig. 2, and the resultant set of new plasmids were designated with VP prefix. This panel was then cotransfected with a plasmid expressing VP16. In cotransfections 10 g of effector and 10 g of target DNA were used. The -fold inductions are the mean value of three independent experiments for each target. that the IE62 promoter positively responds to both ICP4 and IE62 indicates that the promoter elements required for activation by VZV IE62 and HSV ICP4 are different, a finding that was confirmed in the present study by the absence of activation of pIFNTATA by ICP4 (Fig. 1). While the TATA box is dispensable for ICP4-mediated activation of promoters, it does appear to influence the magnitude of activation such that it activates promoters with TATA boxes with low affinity for TBP more efficiently than those with high affinity for TBP (43). The related pseudorabies virus IE180 protein also has been shown to activate suboptimally utilized promoters and contains an N-terminal activator domain (53), but VZV IE62 differs from it in that the TATA motif alone is a sufficient cis-acting element to mediate transactivation by IE62, but not by IE180 (54).
Simon et al. (44) showed that, despite similar binding to TFIID and in vitro functionality for the Ad MLP TATAAAA motif and the SV40 TATTTAT motif, only the former responds to E1A transactivation. In this regard, VZV IE62 resembles E1A, except that a much broader pattern of TATA motifs are recognized (i.e. weak or non-responsive for TTTTTAA and TATTTAAATT; moderately responsive for TATTTAA and TAT-TAAA; highly responsive for TATAAAA, TATATAA, and TATATATAA).
The yeast GCN4 and GAL4 proteins (both acidic transactivators) also differ from IE62 in their ability to interact functionally with certain TATA motifs. In fact, this finding was originally used to invoke models for either two different TATA proteins or two types of acidic activator domains with different selectivities (55). All evidence at present however, points to there being only one universal TBP species.
X-ray crystallographic and biochemical analyses indicate that TBP binds directionally to DNA as a monomer, but it exists in unbound form in solution as a stable homodimer (56 -59). Unlike most DNA-binding transcription factors, TBP binds DNA in the minor groove in a two-step process (60) with relatively low discrimination of specific base pairs. In so doing TBP bends and conformationally distorts the helix, while undergoing conformational changes itself (56,60,61). Dissociation of unbound dimers and of DNA-bound monomers of TBP are relatively slow processes compared with the rate of association of monomeric TBP with DNA (62). Stabilization of TBP binding to either TATA elements by TFIIA (and TFIIB) ( 19) or an IE62 mutant carrying a deletion in the TBP/TFIIB binding region (IE62⌬416 -506) were cotransfected with the minimal promoter pIFNTATA alone or in combination with the full-length IE62 to determine the importance of these regions for transcriptional activity of IE62 as well as whether the mutant protein interferes with the transcriptional activity of native IE62.
promoters, are thought to be key rate-limiting steps in the formation and processing of initiation complexes. Indeed, binding between TBP and TFIIA has been found recently to be an essential step in transactivation by yeast acidic activator protein domains (9). TFIIA binding also appears to alter the conformation of bound TBP.
Our EMSA results (Fig. 9) indicate that all of the TATA like motifs studied here (both consensus and non-consensus alike) exhibit similar affinity for TBP. These results closely resemble those obtained by Hahn et al. (41) for purified yeast holo-TFIID on larger DNA fragments by DNase I footprinting of a number of consensus and non-consensus TATA motifs. It seems, therefore, unlikely that either TBP or TBP-TAF complexes directly discriminate between different types of TATA motifs examined in this study. The highly uniform levels of transactivation of both our GAL4-responsive and VP16-responsive promoter panels containing different TATA motifs also suggest that neither basal levels (as supported by CAT activity and RNase protection data) nor upstream-element-mediated activation are influenced by particular TATA motifs in the panel tested. However, it could also be argued that the TATA motifs themselves become irrelevant when multiple upstream response elements are present so as to offset TATA box requirement for transcription by compensatory effects of other transcription factors binding to these upstream elements.
Considering the inability of the TBP protein to discriminate significantly among the different TATA motif oligonucleotides used in our in vitro EMSA binding studies, it appears unlikely that TBP interactions with the TATA motif alone can directly account for the specificity of IE62 transactivation. Similarly, neither the GAL4 fusion proteins containing the IE62 N-terminal acidic activation domain, nor Oct-1/TAATGARAT-targeted intact VP16, could discriminate functionally between the different TATA motifs. It also appears unlikely that TATA specificity is attained directly through the properties of any particular subclass of general factors or acidic activator domains.
Instead, these properties reside within the core non-activator domain segment of IE62 itself as indicated by our domain-swap experiment (Fig. 6). Although IE62 is also a sequence-specific DNA-binding protein, it preferentially recognizes consensus motifs containing ATCGTC elements, and it neither bound to the TATATAA oligonucleotide probes by EMSA in our work nor protected TATA motif regions in previous foot printing analyses (65). The homologous HSV ICP4 protein has also been shown previously by EMSA and DNA footprinting assays to form a cooperative tripartite DNA-bound complex with TFIIB and TBP (or holo-TFIID), but only on a target DNA probe containing both a TATA motif and an ICP4 binding site (49). Therefore, we are left to conclude that a direct interaction between IE62 with TBP, most likely as a complex with TFIIA or TFIIB or with a specific TAF or co-activator (66), can recognize subtle conformational or stability differences (at either the DNA or protein level) that are imparted by particular base pair arrangements within the TATA motifs, and that these altered interactions manifest themselves as different levels of transactivation.
Emerging evidence from in vivo studies in yeast with yTAF II 145 implicate distinctions among different core promoters (67). Our work reinforces this notion that there is a level of discrimination between different types of TATA motifs (particularly for transactivation) that is not currently understood. The VZV IE62 protein is particularly intriguing in this regard, because the effect occurs in the absence of upstream targeting motifs, and certain non-consensus TATA motifs, particularly TATTAAA, are sufficient to mediate activation. In addition, the ability of VZV IE62 to bind to TBP in vitro through an additional internal non-activator domain suggests that it combines the properties expected of both a transactivator and an adapter or co-activating factor.
Acknowledgments-I thank Thomas Waldmann for support and encouragement and John Hay, Stephen Straus, Gary Hayward, Joseph FIG. 9. Both TATA motifs that are highly responsive to IE62 as well as TATA motifs that are refractory to IE62 show similar TBP-binding profiles by EMSA with purified recombinant human TBP. A, 32 P-labeled, double-stranded oligonucleotide probes representing various TATA motifs evaluated in this study were used to measure any quantitative differences in TBP binding. TACAAA motif of adenovirus EIIa promoter which has been previously reported to be deficient in TBP binding (42) was included as a control. The binding reactions contained a fixed amount of probe (100 fmol) and 1 g of poly(dG-dC) as nonspecific DNA. For each probe, three different amounts of TBP (20, 8, and 4 ng) were tested along with a control reaction without any added TBP (Ϫ). The amount of probe that complexed with TBP was determined by PhosphorImager scanning and is given as a percentage below the corresponding amount of TBP in the reaction. B, binding of TBP to consensus TATATAA motif was competed with 100fold excess of unlabeled competing DNA sequences. In the competition experiments, 100 fmol of 32 P-labeled TATATAA probe was used with 5 ng of TBP.