Identification of previously unrecognized common elements in eukaryotic promoters. A ribosomal RNA gene initiator element for RNA polymerase I.

A new ribosomal RNA promoter element with a functional role similar to the RNA polymerase II initiator (Inr) was identified. This sequence, which we dub the ribosomal Inr (rInr) is unusually conserved, even in normally divergent RNA polymerase I promoters. It functions in the recruitment of the fundamental, TATA-binding protein (TBP)-containing transcription factor, TIF-IB. All upstream elements of the exceptionally strong Acanthamoeba castellanii ribosomal RNA core promoter, to within 6 base pairs of the transcription initiation site (tis), can be deleted without loss of specific transcription initiation. Thus, the A. castellanii promoter can function in a manner similar to RNA polymerase II TATA-less promoters. Sequence-specific photo-cross-linking localizes a 96-kDa subunit of TIF-IB and the second largest RNA polymerase I subunit (A133) to the rInr sequence. A185 also photo-cross-links when polymerase is stalled at +7.

A new ribosomal RNA promoter element with a functional role similar to the RNA polymerase II initiator (Inr) was identified. This sequence, which we dub the ribosomal Inr (rInr) is unusually conserved, even in normally divergent RNA polymerase I promoters. It functions in the recruitment of the fundamental, TATA-binding protein (TBP)-containing transcription factor, TIF-IB. All upstream elements of the exceptionally strong Acanthamoeba castellanii ribosomal RNA core promoter, to within 6 base pairs of the transcription initiation site (tis), can be deleted without loss of specific transcription initiation. Thus, the A. castellanii promoter can function in a manner similar to RNA polymerase II TATA-less promoters. Sequence-specific photocross-linking localizes a 96-kDa subunit of TIF-IB and the second largest RNA polymerase I subunit (A 133 ) to the rInr sequence. A 185 also photo-cross-links when polymerase is stalled at ؉7.
Because promoters for some small nuclear RNA genes switch between polymerase II and III as a result of simple sequence/ spacing alterations, promoters for eukaryotic RNA polymerases II and III are considered more similar to each other than to the RNA polymerase I promoter (reviewed in Ref. 1). Most polymerase II promoters contain a TATA box, or an initiator element (Inr), 1 or both (2). The Inr surrounds the transcription initiation site (tis) and the TATA box is upstream about 30 base pairs. The TATA box is the specific binding site for TATA-binding protein (TBP), the subunit common to the fundamental transcription factors (3) for all polymerases (4). On polymerase II genes with TATA boxes, TBP alone can nucleate the assembly of an initiation complex. In contrast, TATA-less promoters require additional factors or TFIID subunits, called TBP-associated factors (TAFs) (5). Drosophila TAF II 150 (dTAF II 150) specifically interacts with sequences including the Inr, tethering TBP to the promoter (6), and nucleating the assembly of the initiation complex. In humans, the functional dTAF II 150 homolog is CIF and is not tightly associated with TFIID (7). Promoters with both sequence elements are unusually strong, because they bind TFIID very efficiently.
The Acanthamoeba castellanii rRNA core promoter is also unusually strong, binding TIF-IB with a dissociation constant of approximately 30 pM. 2 We noted that this promoter, along with the promoters for other rRNA genes, contains a conserved sequence element which surrounds the tis (8). This is the only well-conserved sequence in eukaryotic rRNA promoters. In addition, Windle and Sollner-Webb (9) observed that in Xenopus laevis oocytes injected with huge quantities (approximately 1 ϫ 10 7 copies) of plasmid ribosomal DNA, the minimal rRNA promoter only encompassed this conserved sequence, although this seemed to be unique to this system and unusual assay condition. In Arabidopsis thaliana, mutations in the conserved sequence in the context of the full-length rRNA promoter decrease transcription in transient expression assays (10). However, the conserved element alone was incapable of directing transcription initiation. Similarly, point mutations of this sequence in A. castellanii alter transcriptional activity in an in vitro transcription system (11). The advent of the extremely sensitive phosphorimaging technology prompted us to reexamine whether this conserved sequence element could independently direct specific transcription initiation in the A. castellanii highly purified transcription system, and, if so, to begin an investigation of its mechanism.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-A. castellanii were grown in Neff's optimal growth medium in a gyratory shaker as described (12).
Transcription Factor and RNA Polymerase I-TIF-IB was purified through the second promoter-DNA affinity column chromatography step as described (13). RNA polymerase I was purified from whole cell extracts either using the standard ammonium sulfate method (14) or a modified KCl procedure (12). Both RNA polymerase I preparations gave identical results.
In Vitro Transcription Assays-AvaI-SalI DNA fragments from the plasmids pSBX60i, Ϫ12 and Ϫ6 deletions (15), were used as the templates for runoff transcription. The reaction conditions were as described in Ref. 16 with the following modifications: the KCl concentration was reduced to 100 mM, the DNA template was reduced to the amounts indicated on the figures, and the reactions were processed and analyzed as described in Ref. 13.
Primer Extension Assays-The runoff transcription reactions were scaled up 4-fold, and the transcripts were used as templates for primer extension reactions as described (17). The primer was the same as positions 891 (5Ј) to 908 of pBR322, which should yield a 66-nucleotide extension product.
Site-specific Photo-cross-linking-The photo-cross-linking reactions were carried out as described (18) with the following modifications. The indicated nucleotides were added to the reactions, and the reactions were incubated at 25°C for 10 min, followed by the addition of heparin to 20 g/ml. After an additional 10 min at 25°C, the reactions were exposed to UV light.

RESULTS AND DISCUSSION
The limits of the minimal A. castellanii rRNA promoter were reevaluated in vitro using highly sensitive phosphorimage technology to detect transcripts. Extensively purified TIF-IB and RNA polymerase I were used to test the 5Ј boundary of the * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
promoter. Previous studies (15,19) identified the region from Ϫ31 to ϩ8 as the minimal sequence necessary to obtain specific transcription initiation. Deletion beyond Ϫ31 resulted in a loss of transcription runoff RNA detectable by autoradiography. However, phosphorimage technology offers an approach which is capable of detecting isotopically labeled products at a significantly lower level. When the same deleted templates used previously were assayed using storage phosphor screens, transcription of genes deleted from their 5Ј end to Ϫ12 or even to Ϫ6 was detected (Fig. 1A, lanes 2, 3, 5, and 6). The wild type sequence in these templates ends at ϩ19, with the pBR322 sequence beyond. The efficiency of these deleted templates is significantly lower than the wild type template; for comparison, the amount of wild type template was decreased to 5% that used in the deleted template reactions (lane 1). Such weak transcripts would not have been detected using autoradiography and the correct exposure for the template containing the full promoter. These transcripts are correctly initiated. 1) They are dependent upon TIF-IB (lanes 4 and 7), which is required for RNA polymerase I to recognize the promoter (19,20). 2) Other restriction enzyme fragment templates produce correct runoff transcripts of predicted lengths (data not shown). 3) Both wild type and deletion mutant genes exhibit identical 5Ј ends in primer extension assays (Fig. 1B, lanes 2, 4, and 6). The primer used in these assays is complementary to the pBR322 vector sequence, so the assay is not detecting any contaminating rRNA which might be present in the protein preparations. The latter is extremely unlikely because these proteins have each been purified through multiple chromatographic columns. Multiple primer extension products have been observed before for in vitro rRNA transcripts in this and other systems and result either from artifacts of the method or heterogeneity of tis selection (20). They do not arise as artifacts because of secondary structure in a readthrough transcript originating upstream of tis. To test this notion, a T3 RNA polymerase transcript which originates upstream of the normal tis, thus having the same sequence as an RNA polymerase I readthrough product, was analyzed by primer extension. The T3 readthrough product does not produce any primer extension products in the region of the tis (lane 1), but does produce a strong full-length extension product (data not shown).
The minimal promoter detected in these assays is much smaller than in our previous study (15). Remarkably, both Ϫ12 and Ϫ6 are outside the region footprinted by the fundamental transcription initiation factor, TIF-IB, from A. castellanii (21,22). However, this region does exhibit enhanced bands in some footprinting experiments, and point mutations in this region affect transcription efficiency (11).
To map the TIF-IB and RNA polymerase I subunits along the template, we subjected template-committed complexes to sitespecific photo-cross-linking by incorporation of 5-[N-(p-azidobenzoyl)-3-aminoallyl]-deoxyuridine monophosphate into the template (18). 5-[N-(p-azidobenzoyl)-3-aminoallyl]-dUMP was incorporated into positions Ϫ1, Ϫ3, Ϫ5, and Ϫ7 in place of dTMP. In 9 separate experiments with several preparations of TIF-IB at various stages of purity, a specific 98-kDa protein photo-cross-linked weakly to the putative rInr sequence (Fig.  2A, lane 3). This protein was identified as TAF I 96 of TIF-IB based upon previously published criteria: TAF I 145 yields crosslinked bands of 153-154 kDa, TAF I 99 yields cross-linked bands of 109 -110 kDa, and TAF I 96 yields 96.8 -98-kDa, TAF I 91 yields 92.5-93-kDa, and TBP yields 39-kDa bands because of the short covalently linked DNA tag. (See also Ref. 18 for a complete discussion of assignments.) The experiment shown in Fig. 2A used TIF-IB in which no proteins other than TBP and the four TAF I s could be identified in silver-stained polyacrylamide gels. However, even in impure preparations of TIF-IB in which the TAF I s are not easily identified in the mass of contaminating proteins, no other proteins photo-cross-link to this position, indicating the specificity of this photo-cross-linking technique.
When RNA polymerase I is added to the committed complex, a protein of approximately 135 kDa photo-cross-links ( Fig. 2A,  lane 4). To verify that this is a component of RNA polymerase I, photo-cross-linking was performed across the peak of RNA polymerase I from a glycerol gradient using the same affinity probe (Fig. 2, B and C). The amount of photo-cross-linked 135-kDa product (Fig. 2C) correlates closely with RNA polymerase I activity determined in a specific runoff assay (Fig. 2B). The homogeneous RNA polymerase I used in this experiment only contains the two large subunits, 185 and 133 kDa, in this size range (14,23). On this basis, the photo-cross-linked subunit was identified as the second largest subunit, A 133 , of RNA polymerase I.
When RNA polymerase I is bound to the promoter, the efficiency of photo-cross-linking of TAF I 96 is significantly increased (lane 4). Thus, the binding of polymerase results in closer proximity of TAF I 96 to the template. Based on this and other data, 3 we propose that polymerase stabilizes TIF-IB binding and suggests an interaction between TAF I 96 and po-3 C. Radebaugh, unpublished data. FIG. 1. The rRNA promoter of A. castellanii can be deleted to ؊12 or ؊6 without complete loss of specific transcriptional activity. A, 3.8, 75, or 150 ng of DNA fragments containing the wild type or a 5Ј deletion to Ϫ12 or Ϫ6 (relative to the tis) were used in a runoff transcription assay. Specific transcripts of the expected 307 nucleotide length are labeled. B, primer extension analysis of the RNA products from the runoff assay. A T3 RNA polymerase transcript (lane 1) containing rRNA gene sequence from Ϫ55 to ϩ19 was used as a control for premature termination by reverse transcriptase. The transcripts from the reactions containing the wild type (W), Ϫ12, or Ϫ6 deletion templates with TIF- IB (lanes 2, 4, and 6) all gave products that correspond to the correct initiation site. The T3 control transcript (lane 1) and the reactions without TIF-IB (lanes 3, 5, and 7) did not contain any primer extension products in this region. lymerase I. Perhaps this interaction induces a conformational change in TIF-IB making it more accessible to the photo-crosslinking reagent. Following initiation and stalling of the polymerase at ϩ7 by addition of ATP and GTP, but omission of CTP, both the second largest (A 133 ) and the largest (A 185 ) subunits of polymerase I photo-cross-linked (lane 5). Under single-round transcription conditions, once polymerase has cleared the promoter following addition of all four ribonucleoside triphosphates, photo-cross-linking of the polymerase subunits is lost. This underscores the correct identification of the two polymerase subunits. Following promoter clearance, TAF I 96 photocross-linking to the probe reverts to the weak efficiency seen before polymerase was added (Fig. 2A, lane 6), consistent with the loss of a polymerase-induced conformational change in TIF-IB. TIF-IB is left behind to recruit successive RNA polymerase I molecules to the template, as previously shown (24).
The efficiency of cross-linking is very reproducible and extremely dependent upon the structure of the photo-cross-linking probe. We have tested initiation (Fig. 3, lanes 1-5) and stalled (Fig. 3, lanes 6 -10) complexes using a range of derivatives with arm lengths from 8.0 -20.9 Å. All of the derivatives cross-link the same subunits, but with different efficiencies. In the initiation complex, the 12.3-Å probe is most efficient at cross-linking TAF I 96, and the 12.8-Å probe cross-links A 133 best. There is a strikingly different dependence in the stalled complex, suggesting changes in probe conformation or subunit positional differences between the complexes. The 8.0-Å probe cross-links TAF I 96 and A 133 far more strongly than any of the other probes. In contrast, A 185 is cross-linked about equally by the 8.0-and 12.8-Å probes, but the 12.3-Å probe and probes longer than 12.8 Å are less efficient at photo-cross-linking this subunit. These probes also differ in the hydrophobicity of the linker arm, which may account for the apparent anomalous length dependence for the A 185 subunit. 4 We propose ribosomal RNA promoters contain a sequence which is functionally similar to the Inr element found in RNA polymerase II promoters, the rInr, and defines yet another class of Inr (2). Importantly, a similar small minimal promoter was found when large amounts of truncated (to Ϫ9) X. laevis rRNA genes were injected into oocytes (9), also demonstrating functionality of an Inr-like element in a non-polymerase II promoter. In further support of this notion, we note that all eukaryotic rRNA promoters have an element near the tis whose mutation alters promoter activity. We previously described a remarkably conserved sequence which is present in a large number of rRNA promoters (8): n(g/r)(g/r)Gt(T/A)aT-nTAgGG(a/g)gAn, where the underlined A is the tis. The conservation of this sequence in rRNA genes is significantly stronger than the consensus sequence for the polymerase II Inr (25). Point mutations of the A. castellanii rInr affect transcription (11). Similarly, point mutants of the homologous rInr sequence in A. thaliana affect transcriptional activity (10). However, even deletion of the A. castellanii rInr in the context of the full-length promoter does not affect subsequent RNA polymerase I binding (20). Thus, as is the case for Inr-less polymerase II promoters, TAF I 96 can be tethered to the tis by upstream promoter elements just as the Inr can tether TBP to TATA-less promoters (6). This reveals a functional similarity between promoters for RNA polymerases II and I which has previously gone unappreciated.
It has been argued that RNA polymerases II and III are more similar to each other because the sequences of their largest subunits are evolutionally more similar to each other than to polymerase I (26), and several snRNA gene promoters can be switched between polymerase II and III by rather simple deletions or promoter element spacing changes (reviewed in Refs. size standards marked on the stained gel with radioactive ink before exposure to storage phosphorscreen. The polypeptides which photo-cross-link to the DNA are identified at the right (18). B, runoff transcription assay of RNA polymerase I fractions from glycerol gradient rate zonal sedimentation (14). C, photo-cross-linking as in A, using fractions from the glycerol gradient shown in B.
FIG. 3. Photo-cross-linking near the tis is very distance-dependent. Nucleotide derivatives with various arm lengths were incorporated into the DNA as described in Fig. 2. Initiation (lanes 1-5) or stalled (lanes 6 -10) complexes were formed and photo-cross-linked. The distance between the dUMP 5-position and the arylazido group on the end of the arm of the derivative are shown at the top of each lane. The polypeptides which photo-cross-link are shown at the right. and 27). However, this ignores the fact that polymerases I and III share more subunits in common than polymerases II and III (in yeast, the AC40 and AC19, in addition to the five subunits common to all three enzymes (28)). The demonstration here that the structure of the promoters for polymerases I and II might be more similar than previously recognized suggests this supposition should be reexamined.