Superimposed Promoter Sequences of the Adenoviral E2 Early RNA Polymerase III and RNA Polymerase II Transcription Units*

The human adenovirus type 2 E2 early (E2E) transcriptional control region contains an efficient RNA polymerase III promoter, in addition to the well characterized promoter for RNA polymerase II. To determine whether this promoter includes intragenic sequences, we examined the effects of precise substitutions introduced between positions +2 and +62 on E2E transcription in an RNA polymerase III-specific, in vitro system. Two noncontiguous sequences within this region were necessary for efficient or accurate transcription by this enzyme. The sequence and properties of the functional element proximal to the sites of initiation identified it as an A box. Although a B box sequence could not be unambiguously located, substitutions between positions +42 and +62 that severely impaired transcription also inhibited binding of the human general initiation protein TFIIIC. Thus, this region of the RNA polymerase III E2E promoter contains a B box sequence. We also identified previously unrecognized intragenic sequences of the E2E RNA polymerase II promoter. In conjunction with our previous observations, these data establish that RNA polymerase II and RNA polymerase III promoter sequences are superimposed from approximately positions -30 to +20 of the complex E2E transcriptional control region. The alterations in transcription induced by certain mutations suggest that components of the RNA polymerase II and RNA polymerase III transcriptional machines compete for access to overlapping binding sites in the E2E template.

The human adenovirus type 2 E2 early (E2E) transcriptional control region contains an efficient RNA polymerase III promoter, in addition to the well characterized promoter for RNA polymerase II. To determine whether this promoter includes intragenic sequences, we examined the effects of precise substitutions introduced between positions ؉2 and ؉62 on E2E transcription in an RNA polymerase III-specific, in vitro system. Two noncontiguous sequences within this region were necessary for efficient or accurate transcription by this enzyme. The sequence and properties of the functional element proximal to the sites of initiation identified it as an A box. Although a B box sequence could not be unambiguously located, substitutions between positions ؉42 and ؉62 that severely impaired transcription also inhibited binding of the human general initiation protein TFIIIC. Thus, this region of the RNA polymerase III E2E promoter contains a B box sequence. We also identified previously unrecognized intragenic sequences of the E2E RNA polymerase II promoter. In conjunction with our previous observations, these data establish that RNA polymerase II and RNA polymerase III promoter sequences are superimposed from approximately positions -30 to ؉20 of the complex E2E transcriptional control region. The alterations in transcription induced by certain mutations suggest that components of the RNA polymerase II and RNA polymerase III transcriptional machines compete for access to overlapping binding sites in the E2E template.
Human subgroup C adenoviral proteins are synthesized in a strict temporal sequence during productive infection as a result of sequential activation of viral RNA polymerase II transcription units (see Refs. 1 and 2). The viral promoters recognized by components of the cellular RNA polymerase II machinery are analogous to those of cellular genes, indeed served as important models in early studies of RNA polymerase II transcription, but their activity is regulated by virus-specific mechanisms. The viral E1A proteins induce efficient transcription from early promoters by altering the activity or availability of a variety of cellular, sequence-specific transcriptional regulators or general initiation proteins (see Refs. 3,4). Expression of viral late genes requires viral DNA synthesis in the infected cell and sequence-specific, transcriptional activators, including the IVa 2 protein (5). Adenoviruses also depend on cellular RNA polymerase III (responsible for synthesis of tRNA, 5 S rRNA, and other small, host cell RNA species (see Refs. 6 -8)), to transcribe the viral VA RNA I and VA RNA II genes. The promoters of these genes comprise two intragenic sequences closely related to similarly located promoter sequences of tRNA genes and termed A and B boxes (see Refs. 6,9). These internal sequences are recognized by the RNA polymerase III-specific initiation protein TFIIIC (10 -15). This type of RNA polymerase III promoter architecture is designated type 2 (see Refs. 6,8,16). In infected cells, VA RNA I accumulates to a very high concentration, ϳ10 8 molecules per cell (17). This RNA is required for efficient synthesis of viral late proteins in productively infected cells (18), for it counters a cellular, anti-viral defense mechanism (see Refs. 19,20).
Although the adenoviral promoters recognized by the cellular RNA polymerase II and RNA polymerase III machines are generally typical of those of cellular genes, the E2 early (E2E) 1 transcriptional control region is unusual. The E2 RNA polymerase II transcription unit, which encodes the viral replication proteins, is expressed from a well characterized promoter (see Fig. 1) that contains a TATA-like sequence, two inverted binding sites for E2F, and an ATF recognition site (3,4). Each of these sequences is required for both basal transcription, and E1A protein-mediated stimulation of transcription in infected cells (21,22). However, RNA polymerase III also transcribes the 5Ј-end of this transcription unit both in vitro and in infected cells (23,24) to produce small RNA species of some 45 and 90 nucleotides (see Fig. 1). Efficient E2E transcription by RNA polymerase III requires sequences 5Ј to the site of initiation, notably the TATA-like sequence (23). 2 Thus, promoters directing transcription by these two cellular enzymes are at least partially superimposed. An analogous arrangement of RNA polymerase III and II transcription units had been observed previously at the P1 and P2 promoters of the human c-myc gene (25)(26)(27). These RNA polymerase III promoters support efficient transcription in vitro, but c-myc transcription by this enzyme is much less efficient than that by RNA polymerase II in Xenopus oocytes microinjected with the gene, and cannot be detected from the endogenous human genes of HeLa or HL60 cells (25,27,28). Thus, the adenoviral E2E transcriptional control region was the first documented to support transcription by both these cellular RNA polymerases in a normal biological context.
Also atypical is the low concentration attained by each small E2E RNA made by RNA polymerase III: 10 -20 copies per infected cell (24). This unusual property explains why E2E RNA polymerase III transcripts were not detected in the many previous studies of adenoviral gene expression. It also suggests that these RNAs are not likely to function in the same manner as the typical, abundant RNAs synthesized by RNA polymerase III. As a necessary prelude to genetic analyses of the function of the unorthodox adenoviral E2E RNA polymerase III transcription unit and its small RNA products, we are defining the sequences that comprise the RNA polymerase III promoter and their relationship to those of the overlapping (and essential) RNA polymerase II promoter. Here we report the results of mutational and biochemical analyses of intragenic sequences common to the E2E transcription units of RNA polymerases II and III.

EXPERIMENTAL PROCEDURES
Templates for in Vitro Transcription and Their Mutagenesis-The wild-type E2E template for in vitro transcription was the plasmid pEIIЈ, which contains the human adenovirus type 2 (Ad2) E2E sequence from position -249 to position ϩ200. This plasmid was constructed by first inserting a NarI fragment comprising bp 26,893 to 27,187 of the viral genome into the ClaI site of the polylinker of pSP73 (Promega). A DNA fragment corresponding to positions 27,105 to 27,341 in the Ad2 genome, generated by PCR, was then introduced via BsshII and XhoI sites within the E2E DNA fragment and the pSP73 polylinker, respectively. The pEIIЈ⌬t1 template, which was used as the parent in mutational analyses of intragenic promoter sequences (see "Results"), contained two T to A substitutions at positions ϩ48 and ϩ49 within the RNA polymerase III t1 termination site (see Fig. 1). Precise substitutions were introduced into the pEIIЈ⌬t1 template by the unique site elimination method (29). A control "maxigene" template, to serve as an internal control for in vitro transcription reactions, was constructed by ligating a 30-bp random DNA sequence flanked by PvuII sites into the PvuII site (ϩ62) of pEIIЈ. A second series of templates, comprising the E2E DNA sequence from position -249 to position ϩ62, or specifically mutated derivatives thereof, linked to the Photinus pyralis luciferase coding sequence was constructed for analysis of E2E transcription in whole cell extracts prepared from Ad2-infected cells. A DNA fragment encompassing the first 535 bp of the P. pyralis luciferase gene was amplified from pGL2 (Promega) using primers designed to introduce a PvuII and a BglII site at the 5Ј-and 3Ј-ends, respectively, of the PCR product. The PvuII/BglII-digested DNA was then ligated into similarly digested pEIIЈ⌬t1 DNA to place the E2E promoter region from -249 to ϩ62 immediately upstream of the luciferase sequence. The substitution mutations created in the pEIIЈ⌬t1 background were then introduced by swapping XhoI to PvuII fragments, or by the unique site elimination method. The nucleotide sequences of all pEIIЈ⌬t1 and ⌬t1/Luc templates were confirmed by the chain termination DNA sequencing method (30).
Plasmid DNA templates for in vitro transcription reactions were purified using the Qiagen Maxi-prep protocol. The concentrations of the different templates to be compared, as well as their quality (i.e. presence of largely supercoiled DNA), were confirmed by ethidium bromide staining of DNA, following electrophoresis in 1.4% agarose gels, cast, and run in 0.04 M Tris acetate, pH 8.0, containing 1 mM EDTA.
In Vitro Transcription by RNA Polymerase III-Transcription of E2E templates by RNA polymerase III was analyzed using nuclear extracts of HeLa cells prepared according to the procedure of Dignam et al. (31), in which at least 90% of the E2E transcripts are synthesized by RNA polymerase III (23). All transcription reactions contained 2 g/ml ␣-amanitin to inhibit RNA polymerase II, 0.1 pmol of template DNA, 0.05 pmol of the internal control template described in the previous section, 1.33 mg/ml nuclear extract protein, 25 mM Hepes-KOH (pH 7.9), 30 mM KCl, 6 mM MgCl 2 , 600 M each GTP, ATP, UTP, and CTP, 12% glycerol, and 2 mM dithiothreitol. Transcripts synthesized in 60 min at 30°C were purified and analyzed by primer extension, as described previously (32,33), using a saturating quantity (20 fmol) of a DNA primer complementary to positions ϩ68 to ϩ89 of the E2E transcription unit. The concentrations of the primer extension products of E2E and internal control transcripts were determined using a Molecular Dynamics PhosphorImager. Initial analyses of transcription from the pEIIЈ and pEIIЈ⌬t1 templates employed an RNase T1 protection assay with an E2E riboprobe specifically end-labeled at positions ϩ2, as described previously (24).
In Vitro Transcription by RNA Polymerase II-Transcription of wildtype and mutant E2E templates by RNA polymerase II was examined using whole cell extracts prepared as described previously (34) from HeLa cells infected with 15 plaque-forming units/cell Ad2 for 12 h, or from uninfected cells. Transcription reactions contained the quantities of template DNA listed in the figure legends, 0.1 pmol of the major late G-less cassette template, pML(C 2 AT) (35) as an RNA polymerase IIspecific internal control, 5.5 mg/ml Ad2-infected or uninfected cell extract protein, 67 mM KCl, 6.7 mM MgCl 2 , and other components as described above for RNA polymerase III transcription. The reactions also contained 40 M tagetitoxin, a concentration empirically determined to inhibit both E2E and VA RNA I transcription by RNA polymerase III (see "Results"). Transcripts synthesized in vitro were analyzed by reverse transcription from the primer described above or a primer complementary to the luciferase sequence corresponding to positions ϩ94 to ϩ113 of the pEIIЈ⌬t1/Luc transcription unit.
Binding of TFIIIC to RNA Polymerase III Promoter Sequences-Human TFIIIC was partially purified from nuclear extracts prepared from 3-4 ϫ 10 10 HeLa cells by sequential chromatography on phosphocellulose, DEAE-cellulose, and double-stranded DNA cellulose according to the procedure of Carey et al. (36). Binding of TFIIIC to VA RNA I or E2E RNA polymerase III promoters was examined using DNA fragments corresponding to positions ϩ6 to ϩ74 or positions -15 to ϩ100, respectively, of these transcription units. The VA RNA I DNA fragment was obtained by annealing complementary, purified oligonucleotides, while the E2E DNA fragment was generated by PCR from the pEIIЈ template with appropriate primers. Mutated E2E DNA fragments used as competitors in some experiments were made from mutated DNA templates with the same primers. All PCR products were purified by electrophoresis in 12% polyacrylamide gels cast, and run in 1ϫ TBE buffer (0.09 M Tris borate, pH 8.0, containing 2 mM EDTA). The concentration of DNA fragments were determined by ethidium bromide staining of the DNA following electrophoresis in 8% polyacrylamide gels with a DNA mass ladder (Life Technologies, Inc.) as standard. DNA fragments were end-labeled as described previously (33). Binding reactions contained the quantities of 32 P-labeled VA RNA I or E2E DNA fragments, human TFIIIC, and competitor DNA listed in the figure legends, 10 mM Hepes-KOH, pH 7.9, 50 mM KCl, 2.5 mM MgCl 2 , and 25 ng of poly(dI-dC)⅐(dI-dC) in a total volume of 20 l. Free and proteinbound DNA were separated by electrophoresis in 4% polyacrylamide (60:1, acrylamide:bis-acrylamide gels) cast in 0.5ϫ TBE buffer and run in 0.25ϫ TBE buffer at 150 V for 2 h at 4°C, following incubation of binding reactions at 30°C for 45-60 min. The quantities of specific TFIIIC⅐DNA formed in the presence of increasing concentrations of competitor complexes were measured using a Molecular Dynamics PhosphorImager. The molar excess concentrations of competitor required for 50% competition were then determined from plots of these values, averaged from several assays, versus competitor concentration, and used to calculate the affinity with which TFIIIC bound to the mutant DNAs relative to that for binding to wild-type E2E DNA set at 1.00.

Elimination of the RNA Polymerase III t1 Termination Site Does Not Alter the Efficiency of Transcription by This Enzyme-
The termination of E2E transcription by RNA polymerase III at two sites ( Fig. 1) complicates analysis of E2E transcription by the method of primer extension, which allows both transcription efficiency and the accuracy of initiation to be evaluated in a single assay. We therefore examined the effects of mutation of the t1 termination site (Fig. 1) on the products made by RNA polymerase III and the efficiency of transcription by this enzyme. Termination of transcription by vertebrate RNA polymerase III can be affected by a variety of signals (see Ref. 37 for a discussion). Nevertheless, a run of four (or more) Ts in the nontranscribed strand is commonly present. The last two of the four such T-A base pairs of the E2E t1 site (23) were therefore replaced by A-T base pairs, to create the pEIIЈ⌬t1 mutant template.
This template and its wild-type parent were transcribed in the RNA polymerase III-specific in vitro system described under "Experimental Procedures," and the E2E RNAs made from each template were compared using an RNase T1 protection assay with a complementary RNA probe specifically end-labeled at position ϩ2, where ϩ1 is defined as described in the legend to Fig. 1. As illustrated in Fig. 2A, RNA polymerase III transcription of the wild-type E2E template produced approximately equal quantities of the two predicted RNase T1 protec-tion products corresponding to RNA I and RNA II. However, no 55-nucleotide protection product corresponding to RNA I was observed when RNA polymerase III transcripts of the E2E⌬t1 template were examined in parallel ( Fig. 2A, compare lanes 1 and 2). Concomitantly, this mutation increased the concentration of E2E RNA II synthesized by RNA polymerase III in vitro ( Fig. 2A, lanes 1 and 2). Indeed, the concentration of the E2E RNA II transcribed from the E2E⌬t1 template was 106.5% (Ϯ1.5%) of the total concentration of the small E2E RNA species synthesized from the wild-type template. Because the ⌬t1 mutation prevented termination at the t1 site but did not inhibit RNA polymerase III transcription, the ⌬t1 template was used in all subsequent transcription assays as a surrogate wild-type.
Comparison of the Sites of Initiation of E2E Transcription by RNA Polymerases II and III-To map accurately the site(s) at which RNA polymerase III initiates E2E transcription, located previously in the vicinity of the major initiation site recognized by RNA polymerase II (23), primer extension was used to compare the 5Ј-ends of transcripts synthesized from the E2E⌬t1 template in reactions in which only one of the two enzymes was active. The fungal toxin ␣-amanitin (at a low concentration, 2 g/ml) and the bacterial phytotoxin tagetitoxin (40 M) were exploited to inhibit RNA polymerases II and III, respectively. The latter compound has been reported to be a specific inhibitor of RNA polymerase III transcription from the type 1 and type 3 promoters of the 5 S RNA and U6 genes, respectively (38). It inhibited completely RNA polymerase III transcription from both the Ad2 VA RNA I and E2E promoters by a concentration of 30 M, with no effect on RNA polymerase II transcription from the adenoviral major late promoter (data not shown). When RNA polymerase III was inhibited by the addition of tagetitoxin to the reaction, initiation was observed at a single site (Fig. 2B, lane 3) corresponding to the major cap site of E2E mRNAs made by RNA polymerase II in infected cells (39), by definition position ϩ1. In contrast, RNA polymerase III transcription of the E2E template, initiated predominantly from position ϩ2, although transcripts beginning at position Ϫ1 were also detected (Fig. 2B, lane 4). Thus, the two enzymes initiated transcription from distinct sites.
Identification of Intragenic Elements of the E2E RNA Polymerase III Promoter-Previous studies established that E2E sequences 3Ј to position ϩ62 are not necessary for efficient RNA polymerase III promoter activity in vitro (23), whereas deletion of sequences downstream of position ϩ2 or of positions ϩ22 to ϩ62 reduced transcription to undetectable levels (data not shown). We therefore examined the effects of precise substitutions within the region ϩ2 to ϩ62 on the efficiency and accuracy of transcription. Both the 10-and 5-bp substitutions shown in Fig. 3, which were named according to the positions of the mutated base pairs, were introduced into the ⌬t1 background, as described under "Experimental Procedures." The mutant and wild-type (⌬t1) templates were transcribed in vitro by RNA polymerase III, and the products were analyzed by primer extension. To allow accurate quantification of the effects of the mutations on the efficiency of transcription, an E2EЈ maxigene, which contains an insertion of a 30-bp random DNA sequence at position ϩ62, was included in all transcription reactions. Typical results of these analyses are shown in Fig. 4, A and B, and the effects of the mutations upon the efficiency of RNA polymerase III transcription are summarized in Fig. 4C. None of the substitutions introduced immediately downstream of the initiation sites, such as mut2-6, mut7-11, or mut7-16 (Fig. 4A, lanes 3 and 4; Fig. 4B, lanes [3][4][5], or between positions ϩ27 and ϩ42 (Fig. 4A, lanes 3 and 7; Fig. 4B, lanes 3 and 10) resulted in significant inhibition of RNA polymerase III transcription (Fig. 4C). In contrast, the quantities of E2E transcripts detected in reactions containing the mut22-31, mut42-51, and mut52-62 templates were severely reduced, to values of 0.22, 0.04, and 0.06, respectively, relative to the wild-type concentration set at 1.00 (Fig. 4A, lanes 6, 8, and 9; Fig. 4C). None of these changes in the transcribed sequence altered the stability of the RNA made in the in vitro transcription reactions (data not shown), indicating that these substitutions inhibited transcription. Various 5-bp substitutions made between positions ϩ42 and ϩ62 also severely inhibited transcription, as did the mut48 -57 substitution, which overlapped the 10-bp alterations initially introduced and analyzed (Fig.  4B, compare lanes 3, 11, 12, and 13; Fig. 4C). These properties suggest that the 21-bp segment occupying positions ϩ42 and ϩ62 comprises a single, functional promoter element. The data reported in Fig. 4, in conjunction with previous analysis of E2E templates carrying 3Ј-deletions (23), indicate that its 3Ј-boundary is close to position ϩ62, whereas its 5Ј-boundary can be placed between positions ϩ42 and ϩ47. Mutation of the sequence ϩ22 to ϩ26 also resulted in inhibition of transcription (Fig. 4B, lane 10; Fig. 4C), confirming the presence of a second intragenic promoter sequence. In contrast, templates carrying mutations between positions ϩ12 and ϩ21 directed somewhat more efficient transcription by RNA polymerase III, with the greatest increase (a factor of 2.5) observed when the sequence between positions ϩ14 and ϩ19 was altered (Fig. 4B, lane 6;  Fig. 4C).
These results indicated that the E2E RNA polymerase III promoter includes two blocks of internal sequence, occupying positions ϩ22 to ϩ26 and positions ϩ42 to ϩ62 of the transcription unit. This organization resembles that of type 2 RNA polymerase III promoters (see the introduction) as well as some that comprise both intragenic and essential upstream sequences, such as the promoter of the EBV EBER RNA 2 gene (40,41). We therefore searched the intragenic E2E RNA polymerase III promoter sequences for consensus A and B box sequences (6). As illustrated in Fig. 5, the functionally defined promoter sequence ϩ22 to ϩ26 is included within a sequence that resembles the A box consensus sequence of tRNA genes: both the inhibitory substitutions mut22-26 and mut22-31 (Fig. 4C) reduced the match to the A box consensus sequence from 7/12 (Fig. 5) to 4/12 (Fig. 3) positions. The alterations in initiation specificity to include the ϩ1 and ϩ3 sites induced by all substitutions within this intragenic E2E sequence (Fig. 4A,  lanes 4 and 5; Fig. 4B, lanes 5-8) provide strong support for its classification as an A box, for such change in initiation specificity is characteristic of mutations within the A box sequence of type 2 RNA polymerase III promoters (e.g. Refs. 42-44). Furthermore, as described in the next section, substitutions within the A box consensus sequence shown in Fig. 5 impaired binding of human TFIIIC to the E2E promoter. Although templates carrying the mut14 -19 and mut17-21 substitutions that alter the 5Ј portion of the A box consensus sequence were transcribed more efficiently than the wild-type E2E template (Fig. 4C), these mutations did not improve the match to the consensus. Rather, both these substitutions reduced the match, to 6/12 positions, and each eliminated one of the two invariant base pairs of the consensus sequence, both of which are represented in the wild-type E2E sequence (Figs. 3 and 5). An alternative explanation for these observations is discussed below. In contrast to the ready identification of an A box consensus sequence, it was not possible to locate a B box by inspection of the more downstream E2E sequence required for efficient RNA polymerase III transcription: this 21-bp segment contains two adjacent sequences that are weak matches to the B box consensus sequence (Fig. 5). Such ambiguity prompted us to evaluate more directly the question of whether the E2E RNA polymerase III promoter contains a B box sequence.
Human TFIIIC Components Recognize the Internal RNA Polymerase III Promoter Sequences-The A and B box sequences of type 2 RNA polymerase III promoters are recognized by components of TFIIIC (see the introduction). Thus, if the internal elements of the RNA polymerase III promoter described above included functional A and B sequences, TFIIIC should bind specifically to the E2E promoter. To test this prediction, the interactions of human TFIIIC with the internal sequences of the Ad2 VA RNA I and E2E RNA polymerase III promoters were compared using electrophoretic mobility shift assays (45,46). The human TFIIIC used in these experiments was partially purified by a previously described procedure (36), but was not subjected to B-box DNA-affinity chromatography (10). The latter procedure separates the B box binding component, TFIIIC2, from a second component, TFIIIC1, which is required for RNA polymerase III transcription and interacts strongly with termination signals and weakly with A box sequences (10,(13)(14)(15). Because a B box sequence had not been unambiguously identified in the E2E promoter, it seemed prudent to avoid separation of the promoter-binding proteins of TFIIIC.
As expected, human TFIIIC bound specifically to a DNA fragment spanning the A and B box sequences of the VA RNA I gene: formation of a pair of protein⅐DNA complexes was inhibited by homologous unlabeled DNA but not by an oligonucleotide unrelated in sequence (Fig. 6A, lanes 1-3). The mobility of these two complexes relative to free DNA and to one another suggests that they represent previously described TFIIIC2⅐DNA complexes containing transcriptionally active The positions of cDNA copied from RNA polymerase II transcripts initiating at position ϩ1 or from RNA polymerase III transcripts with 5Ј-ends at positions ϩ2 and Ϫ1 are indicated by the asterisk or arrows, respectively, at the left. These products were quantified using a PhosphorImager. The values obtained were corrected for variations in recovery of products by quantification of signals from a single-stranded, 32 P-labeled oligonucleotide was added to each reaction once transcription was complete and used to calculate the contributions of each RNA polymerase to the total transcription from the E2E transcriptional control region (shown at top).

FIG. 3. Sequences of mutated E2E
templates. The sequence of the surrogate wild-type E2E⌬t1 template from positions ϩ1 to ϩ68 is shown at the top, with the mutation in the t1 termination site shown in boldface and italics. The A box sequence is highlighted in gray. Positions that match the consensus A box sequence in Fig. 5 are shown in boldface. The sequences altered in the mutant templates listed at the right are shown below, with boldface indicating changes that maintain a match to the A box consensus. and inactive forms of this initiation protein (12,13). An E2E DNA fragment containing all sequences transcribed by RNA polymerase III also inhibited binding of TFIIIC to the internal promoter sequences of the VA RNA I gene (Fig. 6A, lanes 1 and  4). When this E2E DNA fragment was used as a probe, a set of specific protein⅐DNA complexes, two of which exhibited the same mobilities as the TFIIIC⅐VA I DNA complexes, was observed (Fig. 6A, lanes 5, 6, and 8). Their formation was inhibited by both homologous, unlabeled DNA and the DNA containing the VA RNA I promoter (Fig. 6A, lanes 5, 7, and 8). These results established that human TFIIIC specifically recognizes internal sequences of the E2E RNA polymerase III promoter.
We next examined the relationship between binding of TFIIIC to the internal E2E promoter sequence and the initiation of transcription by RNA polymerase III, by determining whether mutations that inhibited E2E transcription altered the binding of the protein. The abilities of wild-type DNA and mutant DNA fragments carrying specific substitutions to act as competitors of binding of TFIIIC to 32 P-labeled E2E DNA were compared as a function of competitor concentration. The mut27-31 substitution, which did not decrease the efficiency of E2E transcription by RNA polymerase III (Fig. 4, B and C), did not impair binding of TFIIIC to wild-type E2E DNA (Fig. 6B,  compare lanes 1-5 and 11-15). Mutations within the portion of the E2E sequence resembling an A box inhibited binding of TFIIIC, but by only a modest degree (Table I). In contrast, substitutions within the ϩ42 to ϩ62 sequence required for efficient transcription by this enzyme significantly impaired binding of TFIIIC (Fig. 6B, lanes 6 -10; Table I). The correlation between the effects of the substitution mutations on E2E transcription by RNA polymerase III (Fig. 4) and on binding of FIG. 5. Comparison of intragenic E2E RNA polymerase III promoter sequences to tRNA consensus A and B boxes. The sequence of the Ad2 E2E transcription units extending to position ϩ68 is shown at the top. Sequences required for efficient or accurate transcription by RNA polymerase III are underlined, whereas those necessary for maximally efficient transcription by RNA polymerase II are indicated by the line above the sequence. Consensus A and B box sequences (6,44) and typical distances between them are shown below, with invariant nucleotides in larger sized font, where R ϭ a purine, Y ϭ a pyrimidine, and N ϭ any nucleotide. The best matches of E2E sequences within the functionally defined promoter regions are indicated below, with conserved nucleotides in boldface.

FIG. 4. Identification of intragenic sequences required for E2E transcription by RNA polymerase III.
Transcription of the wild-type E2E template ⌬t1 or its mutated derivatives carrying 10-bp (A) or 5-bp (B) substitutions between positions ϩ2 and ϩ62 was carried out under the RNA polymerase III-specific conditions described under "Experimental Procedures." Transcription reactions contained 0.05 pmol of the internal control E2E maxigene template and no test template (lane 2 in A and B) or 0.1 pmol of wild-type E2E⌬t1 DNA (A and B, lane 3) or the mutant templates mut7-16, mut17-26, mut22-31, mut32-41, mut42-51, and mut 52-62 (A, lanes 4 -9, respectively) or mut2-6, mut7-11, mut12-16, mut14 -19, mut17-21, mut22-26, mut27-31, mut42-47, mut48 -57, and mut58 -62 (B, lanes 4 -13, respectively), which are designated m7-16 and so on. Transcripts were detected by primer extension, followed by electrophoresis of cDNA products under denaturing conditions and autoradiography (see "Experimental Procedures"). The positions of the cDNAs synthesized from the E2E (E2E) and internal control (IC) templates are indicated at the right, and the lengths of DNA markers (5Ј-end-labeled HaeIII fragments of pBR322 DNA run in lane 1 in each of the gels shown) are listed at the left. The specific E2E cDNAs were quantified directly using a Molecular Dynamics PhosphorImager, and values obtained for each mutant template were corrected using those of products of transcription of the internal control template. In C, these corrected values are expressed relative to that obtained from the wild-type E2E⌬t1 template, set at 1.0. The relative activity shown for each mutant template represents the mean of three independent experiments. TFIIIC components to E2E DNA (Table I) indicated that binding of TFIIIC to internal promoter sequences is required for RNA polymerase III transcription. Thus, the intragenic sequences required for E2E transcription by RNA polymerase III appear to be functionally equivalent to those of the well characterized type 2 promoters recognized by this enzyme.
Identification of Internal Sequences Required for RNA Polymerase II Transcription-As the contribution of transcribed sequences to the E2E RNA polymerase II promoter had never been examined, we also determined the effects of the intragenic mutations on transcription by this enzyme. Initial surveys of E2E transcription by RNA polymerase II and RNA polymerase III, distinguished by their sensitivities to 2 g/ml ␣-amanitin and 40 M tagetitoxin, respectively, indicated that RNA polymerase III transcription invariably predominated in both nuclear and whole HeLa cell extracts (data not shown). However, two conditions that increased the efficiency of E2E transcription by RNA polymerase II relative to that by RNA polymerase III were identified. Under in vitro conditions typically used for RNA polymerase II transcription, the proportion of the total E2E RNA synthesized by this polymerase was 2-to 3-fold greater in whole HeLa cell extracts prepared from cells harvested 8 -12 h after Ad2 infection than in uninfected cell extracts prepared in parallel (data not shown), reaching some 30% of the total (Fig. 2B). This difference could be the result of stimulation of RNA polymerase II transcription from the E2E promoter by the viral E1A proteins (see Ref. 4). We also observed that the fraction of E2E RNA made by RNA polymerase II in uninfected cell extracts increased to values like those supported by Ad2-infected cell extracts when the concentration of template DNA was reduced (data not shown). To avoid possible complications that might be introduced by effects of viral proteins, the latter condition was used to examine the effects of the E2E mutations described previously on RNA polymerase II transcription. A G-less cassette under the control of the viral ML promoter served as an internal control (see "Experimental Procedures"). Typical results of these experiments are shown in Fig. 7, A and B, and the quantification of several is summarized in Fig. 7C.
Substitution of sequences between positions ϩ32 and ϩ62 did not significantly alter the efficiency of RNA polymerase II transcription (Fig. 7B, compares lanes 1 and 5-8; Fig. 7C). However, mutation of the sequence adjacent to the initiation site (ϩ2 to ϩ11), as well as a more downstream segment (ϩ14 to ϩ21), resulted in inhibition of transcription by up to a factor of 8 (Fig. 7A, lanes 1-3 and 5-7; Fig. 7C). Thus, the E2E RNA polymerase II promoter extends for some 20 bp beyond the initiation site. Although substitution of positions ϩ12 to ϩ16 did not inhibit transcription (Figs. 7A, lanes 1 and 4, and 7C), direct inhibitory effects on RNA polymerase II transcription might be offset by relief of competition from RNA polymerase III transcription (see "Discussion"), for this substitution impinges on the A box of the RNA polymerase III promoter (Fig.  5). Substitution of positions ϩ27 to ϩ31 increased the efficiency of RNA transcription, by some 2.5-fold (Figs. 7B, lanes 1 and 4, and 7C). None of these intragenic sequence mutations altered the specificity of initiation of E2E transcription by RNA polymerase II (Fig. 7, A and B), indicating that this parameter is determined by RNA polymerase II promoter sequences lying upstream of the initiation site.

DISCUSSION
The promoters of many cellular and viral genes transcribed by RNA polymerase III include intragenic A and B boxes (see Refs. 6,8,16 for reviews). The adenoviral E2E RNA polymerase III promoter also contains two intragenic sequences required for efficient or accurate transcription by this enzyme (Fig. 4). The sequence of the initiation site-proximal promoter element (Fig. 5) and the alterations in initiation specificity observed when it was mutated (Fig. 4, A and B) allow it unambiguous identification as an A box. The specific binding of human TFIIIC to internal sequences of the E2E RNA polymerase III transcription unit (Fig. 6) confirms that this viral promoter contains a B box (10,(13)(14)(15). The observation that mutations within the sequence ϩ42 to ϩ62 strongly impaired both transcription (Fig. 4) and TFIIIC binding ( Fig. 6B; Table I) locate the E2E B box to within this 21-bp segment of the transcription unit. However, two nonoverlapping candidate sequences lie within this region (Fig. 5). Several arguments suggest that the E2E sequence located between positions ϩ51 and ϩ61 is the  binding site for TFIIIC2. This sequence includes three times as many of the invariant base pairs of the B box consensus sequences as the functionally defined sequence immediately 5Ј to it (Fig. 5). Furthermore, a B box at positions ϩ51 to ϩ61 would yield a separation of E2E A and B box sequences of 26 bp, closer to the range typically observed (see Refs. 6,8,16,47) and required for optimal type 2 promoter function (e.g. Refs. 42,48). Specific binding of the TFIIIC1 component of TFIIIC to the termination regions of the Ad2 VA RNA I gene and a Xenopus leavis tRNA 1 Met gene has been reported to stabilize binding of TFIIIC2 to the promoters to support maximally efficient transcription (14). The E2E t1 termination site for RNA polymerase III transcription (Fig. 1) is included within the 10-bp sequence lying 5Ј to the putative B box. Thus, TFIIIC1 may bind to a region that includes this sequence to stabilize otherwise low affinity binding of TFIIIC2 to the adjacent B box. Further experiments will be required to determine whether such a mechanism of TFIIIC binding accounts for the requirement for an unusually long, initiation site-distal, intragenic sequence.
The E2E RNA polymerase III promoter does not belong to one of the three types now generally recognized (see Refs. 6,8,16), but rather closely resembles that of the EBV EBER RNA 2 gene: both include intragenic A and B boxes and upstream TATA-like sequences (Fig. 5 (23, 40)) 2 and binding sites for sequence-specific activators of RNA polymerase II transcription that are necessary for maximal activity of the promoters (23,40). Despite such similarities, the EBV EBER RNA 2 gene is transcribed only by RNA polymerase III. One important determinant of such exclusivity is its TATA-like sequence, designated ETAB, located at positions Ϫ25 to Ϫ30 (41). The E2E TATA-like sequence, TTAAGAGT, is noncanonical, for it includes GC base pairs at positions analogous to those of the ETAB sequence. However, in contrast to this latter sequence, the E2E TATA sequence is also part of a functional RNA polymerase II promoter (see the introduction and Fig. 1) and behaves as a typical RNA polymerase II TATA sequence in assays for TBP binding and transcriptional function. 3 It will be of considerable interest to identify the features of the E2E TATA sequence that allow recognition by TBP-containing initiation proteins specific for both RNA polymerase II and RNA polymerase III.
Mutations introduced between positions ϩ14 and ϩ21, a sequence that includes most of the A box of the RNA polymerase III promoter (Fig. 5), significantly inhibited E2E transcription by RNA polymerase II (Fig. 7). Thus, a single intragenic E2E sequence functions as part of both the RNA polymerase II and the RNA polymerase III promoters (Fig. 5). This arrangement seems likely to account for the initially surprising observation that mutations introduced into the 5Ј-end of the A box altered initiation specificity but either did not reduce, or modestly increased, the efficiency of transcription by RNA polymerase III (Fig. 4). In the reactions specific for this enzyme, RNA polymerase II was inhibited by addition of an appropriate concentration of ␣-amanitin. This toxin binds to the largest subunit of RNA polymerase II and inhibits transcription after assembly of preinitiation complexes and formation of the first phosphodiester bond in nascent RNA (see Ref. 49). Consequently, when a mutation alters both the RNA polymerase III and the RNA polymerase II promoters, such as several that impinge on the A box, direct inhibition of RNA polymerase III transcription may be offset by inhibition of formation of RNA polymerase II initiation complexes. Thus, RNA polymerase III transcription would appear unchanged or stimulated, as indeed observed (Fig. 4). The properties of the E2E template with a substitution at the 3Ј-end of the A box (positions ϩ22 to ϩ26) provide additional support for the hypothesis that proteins mediating recognition of the RNA polymerase II and the RNA polymerase III promoters are in competition for superimposed binding sites: This mutation had no significant effect on the efficiency of transcription by RNA polymerase II (Fig. 7, B and  C), and inhibited E2E transcription by RNA polymerase III by some 90% (Fig. 4, B and C).
Substitutions between positions ϩ2 and ϩ11 that inhibited RNA polymerase II transcription did not stimulate RNA polymerase III transcription (Figs. 4 and 7). Conversely, when the B box-containing region of the RNA polymerase III promoter was crippled, no increase in transcription was observed in the RNA polymerase II-specific system, in which tagetitoxin inhibits elongation by RNA polymerase III with no apparent effect on preinitiation complex assembly (50). It therefore appears that binding of TFIIIC to the RNA polymerase III promoter neither interferes directly with recognition of the RNA polymerase II promoter, nor is the reaction rate-limiting for assembly 3 H. Chen and S. J. Flint, unpublished observation.

FIG. 7. Effects of intragenic E2E substitutions on RNA polymerase II transcription.
In vitro transcription reactions were as described under "Experimental Procedures" and contained 0.2 pmol of E2E⌬t1 luciferase (⌬t1) template DNA or the derivatives carrying the substitutions mut2-6, mut7-11, mut12-16, mut14 -19, mut17-21, mut17-26, mut22-26, mut22-31, mut27-31, mut32-41, mut42-51, mut52-62, and mut48 -62 (A, lanes 2-14, respectively), designated as in Fig. 4. The positions of the primer extension products of the E2E (E2E) and the pML(C 2 AT) internal control (ML) transcripts are indicated at the right. The results of quantification of the effects of these mutations on the efficiency of RNA polymerase II transcription, corrected using the ML internal control and normalized to the E2E⌬t1 template as described in the legend to Fig. 4, are shown in B. Each value is the mean of at least three independent determinations. of stable preinitiation complexes committed to transcription by RNA polymerase III under the conditions used in these experiments. A number of parameters, including absolute and relative concentrations of initiation proteins specific for each RNA polymerase, their affinities for binding sites in the template or in other proteins, and, in the cell, the relative concentrations of the RNA polymerase holoenzymes that are believed to mediate initiation of transcription (see Refs. [51][52][53], seem likely to govern which reactions will be in competition under different conditions. It will therefore be important to establish which sequences of the RNA polymerase II and RNA polymerase III promoters set the balance of E2E transcription by the two RNA polymerases in adenovirus-infected cells. Because E2E RNA I contains only the 5Ј-half of the sequence of the longer E2E RNA II (Fig. 1), it is unlikely that these RNA species conform to a common structure necessary for a common function. In fact, the run of four T-A base pairs of the E2E t1 termination site of Ad2 or Ad5 is not conserved even in the genomes of all human members of the Mastadenoviridae, and elimination of the t1 site does not impair replication of subgroup C adenoviruses. 4 These properties of the small E2E RNAs, and the low concentrations at which they are present in infected cells (24), suggest that production of functional RNA species may not be the primary purpose of transcription of E2E sequences by RNA polymerase III. Rather, such transcription might negatively regulate expression of the viral replication proteins encoded in the RNA polymerase II E2E transcription unit. The competition for access to superimposed sequences of the two E2E promoters discussed above is also consistent with such a regulatory role. Analogous function as regulators of RNA polymerase II transcription has been proposed (e.g. Refs. 26,54,55) or demonstrated (56) for specific RNA polymerase III transcription units present in cellular genomes. The identification of sequences that contribute to one or both of the adenoviral E2E promoters reported here will allow the role of the E2E RNA polymerase III promoter in the adenoviral genome to be tested, using genetic approaches, to determine whether the small E2E RNA species synthesized by RNA polymerase III perform a specific function during the adenoviral infectious cycle or are by-products of a less well understood regulatory mechanism.