INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Reiterative transcription (also referred to as pseudotemplated transcription, transcriptional slippage, and RNA polymerase stuttering) is a reaction catalyzed by a number of bacterial, phage, viral, and eukaryotic RNA polymerases (15, 16, 19). In this reaction, nucleotides are repetitively added to the 3' end of a nascent transcript due to slippage between the transcript and the DNA or RNA template. Typically, slippage occurs between a homopolymeric sequence in the transcript and at least three complementary bases in the template (37). In most cases, the mechanism involves one or more rounds of a one-base upstream shift of the transcript so that the same nucleotide in the template specifies multiple residues in the transcript (10, 13). Recent studies indicate that reiterative transcription plays an important role in the expression and regulation of a number of bacterial and viral genes by a variety of mechanisms (13, 18, 20, 29).

One of these genes is the upp gene of Escherichia coli. The upp gene encodes the pyrimidine salvage enzyme uracil phosphoribosyltransferase, which catalyzes the formation of UMP from uracil and phosphoribosylpyrophosphate (4). The upp gene appears to be the first gene of an operon that also contains the uraA gene, encoding uracil permease, and perhaps a third, uncharacterized gene designated b2496 (3, 7). A sequence resembling a strong intrinsic transcriptional terminator is located between the upp and uraA genes (3). The function of this terminator-like sequence in the expression and regulation of downstream genes is unknown.

Expression of the upp gene, and presumably cotranscribed genes, is negatively regulated over a sixfold range by pyrimidine availability (4, 30, 35). This regulation occurs mainly by UTP-sensitive selection of alternative transcriptional start sites, which produces transcripts that differ in the ability to be productively elongated (35). The upp initially transcribed region contains the sequence GATTTTTTTTG (nontemplate strand) (Fig. 1). Transcription is initiated primarily at the first two bases in this sequence, designated G6 and A7 (numbering from the promoter 10 region). High intracellular levels of UTP, due to ample pyrimidine availability or synthesis, favor initiation at position A7. However, the resulting transcripts are subject to reiterative transcription (i.e., repetitive UMP addition) within the 8-bp T · A tract in the initially transcribed region. These transcripts, with the general sequence AUUUU_n (where n equals 1 to >50), are not extended to include downstream sequences and are eventually aborted. In contrast, low intracellular levels of UTP, caused by pyrimidine limitation, strongly favor initiation at position G6. This start site switch appears to be caused by inhibition of initiation at position A7, which relies on a high concentration of UTP to form the critical first internucleotide bond of the transcript (23, 26). Transcripts initiated at position G6 can, at least in part, avoid reiterative transcription and be elongated normally. This effect is apparently due to the formation of a relatively stable hybrid between the 5' end of the G6 transcript and the DNA template.

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Sequences of the initially transcribed regions of the upp, codBA, pyrBI, and carAB P1 promoters. The nontemplate strand sequence is shown. The 10 region of each promoter is underlined, and the major transcriptional start sites are indicated by asterisks. Start sites are identified in the text according to their positions downstream from the 10 region; e.g., the major upp start sites are designated G6 and A7.

Several other E. coli operons encoding pyrimidine metabolic enzymes are regulated by mechanisms that employ reiterative transcription. The codBA operon, which encodes the pyrimidine salvage enzymes cytosine permease and cytosine deaminase, is regulated over a 30-fold range by a mechanism that is entirely analogous to that described for the upp gene (29). Interestingly, the codBA initially transcribed region, GATTTTTTG, contains only a 6-bp T · A tract, and the G and A start sites (i.e., G7 and A8) are 1 base further downstream from the promoter 10 region than in the upp promoter (Fig. 1). The latter difference contributes significantly to the higher range of codBA regulation. (S. M. Dylla and C. L. Turnbough, Jr., unpublished data).

The pyrBI operon, which encodes the two subunits of the pyrimidine biosynthetic enzyme aspartate transcarbamylase, is regulated over a sevenfold range by a UTP-sensitive reiterative transcription mechanism that differs fundamentally from the upp and codBA mechanisms (20). The pyrBI initially transcribed region, AATTTGCC, contains a 3-bp T · A tract and a single transcriptional start site (Fig. 1). The key regulatory event occurs after the synthesis of the first five bases of the transcript, AAUUU, at which point the transcript can reversibly slip on the DNA template and has the potential to engage in reiterative transcription. The extent of reiterative transcription versus strictly templated transcription is controlled by the intracellular level of UTP. High UTP levels favor reiterative transcription, which produces transcripts with the sequence AAUUUU_n (where n equals 1 to >30). These transcripts are always aborted. In contrast, low UTP levels favor the addition of a G residue as the sixth base in the transcript. This addition precludes reiterative transcription, and the AAUUUG transcript may be elongated to a full-length pyrBI transcript. The carAB operon, which encodes the two subunits of the pyrimidine biosynthetic enzyme carbamoyl phosphate synthetase, is regulated over a threefold range by a mechanism equivalent to that described for the pyrBI operon (Fig. 1) (12). The function of the four control mechanisms described above is to adjust pyrimidine salvage and biosynthetic enzyme levels to the cellular need for pyrimidine nucleotides.

A comparison of the reiterative transcription control mechanisms for the upp, codBA, pyrBI, and carAB operons and a rudimentary understanding of the requirements for reiterative transcription raise an obvious question. Is the very long T · A tract in the upp initially transcribed region required for regulation? In this study, we investigated this question. We constructed upp promoter mutations that systematically shorten the T · A tract in the initially transcribed region and examined the effects of these mutations on upp expression and regulation. The results indicate, unexpectedly, that a very long T · A tract (i.e., 7 or 8 bp) is, in fact, required for normal regulation. Additional examination of upp transcription provided an explanation for this requirement.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. E. coli K-12 strain CLT42 [F car-94 (argF-lac)U169 rpsL150 thiA1 relA1 deoC1 ptsF25 flbB5301 rbsR] (32) was used as the parent in the construction of seven lambda lysogens used in this study. These strains were constructed by inserting into the CLT42 chromosomal lambda attachment site a single copy of a recombinant lambda bacteriophage that carries either the wild type or a mutant version of a upp::lacZ gene fusion. The wild-type fusion contains the wild-type upp promoter region, while the six mutant fusions contain 1- to 6-bp deletions in the T · A tract of the upp initially transcribed region (i.e., T tract in Fig. 1). The construction of two of the lysogens, which carry either the wild-type or 6-bp deletion promoter region in the gene fusion, was described previously (35). Essentially the same procedure was used here to construct five additional lysogens in which the upp::lacZ gene fusions contain 1- to 5-bp deletions in the T · A tract of the upp initially transcribed region. A brief summary of the steps involved in the construction of all seven lysogens is provided below.

Construction of gene fusions. Construction of upp::lacZ gene fusions employed the multicopy plasmid pMLB1034, which contains the lacZ gene without a promoter, a ribosomal binding site, and the first eight codons for -galactosidase (34). This plasmid contains an EcoRI/SmaI/BamHI cloning site immediately preceding the lacZ gene. Gene fusions were made by first digesting plasmid pMLB1034 with EcoRI and BamHI and then ligating the linear plasmid to an EcoRI-BamHI restriction fragment containing the wild-type upp promoter region (from 100 to +127, numbering from the first transcriptional initiation site) or an equivalent fragment containing a mutant promoter region. The downstream sequence in these promoter region fragments extends through upp codon 30. The resulting plasmids were transformed into strain CLT240 (CLT42 pcnB80 zad::Tn10). The pcnB80 mutation of this strain reduces the plasmid copy number, which is essential for maintaining (at least some) mutant fusion plasmids free of secondary mutations that reduce upp::lacZ expression. All fusion constructions were confirmed by DNA sequence analysis.

Transfer of gene fusions from plasmids to the E. coli chromosome. Wild-type and mutant upp::lacZ gene fusions carried on derivatives of plasmid pMLB1034 (in strain CLT240) were individually transferred to the chromosome of strain CLT42 by using phage lambda RZ5 (31). The presence of a single prophage at the lambda attachment site on the chromosome was determined by PCR analysis (35).

Restriction digests, ligations, transformations, PCR, site-directed mutagenesis, and DNA preparations. Conditions for restriction digests, ligations, and transformations were as previously described (32). PCR amplification of DNA was performed with Pfu DNA polymerase (Stratagene), using the reaction mixture recommended by the supplier. PCR conditions were: 95°C for 5 min; then 95°C for 1 min, 60°C for 1 min, and 72°C for 2 min for 30 cycles; and finally 72°C for 5 min. Site-directed mutations were introduced into the upp promoter region by using a PCR-based procedure similar to that described by Barettino et al. (6). The resulting mutations were verified by DNA sequence analysis. DNA was prepared essentially as previously described (35).

Media and culture methods. Cells used for enzyme assays and RNA isolations were grown in NC medium (1) supplemented with 10 mM NH₄Cl, 0.4% (wt/vol) glucose, 0.015 mM thiamine hydrochloride, 1 mM arginine, and either 1 mM uracil or 0.25 mM UMP. Cultures were incubated at 37°C with shaking and grown to an optical density at 650 nm of 0.5 (mid-log phase). Culture densities were measured with a Gilford model 260 spectrophotometer, and doubling times (which vary slightly with culture density when UMP is the pyrimidine source) were determined between optical densities of 0.1 and 0.2

Enzyme assays. Cell extracts were prepared by sonic oscillation (31). -Galactosidase activity (24) and protein concentration (22) were determined as previously described.

Isolation of cellular RNA and primer extension mapping. Cellular RNA was isolated quantitatively as described by Wilson et al. (36). Primer extension mapping of the 5' ends of upp::lacZ transcripts was performed as described by Liu and Turnbough (21), except that 50 µg of RNA from uracil-grown cells and 42 µg of RNA from UMP-grown cells were used for analysis. The different amounts of RNA, which were isolated from the same mass of cells, reflect the different levels of stable RNA in cells growing at different rates. The primer used in these experiments was 5'TTTTCCCAGTCACGACGTTG, which was labeled with ³²P at the 5' end (5). This primer hybridizes to the lacZ sequence just downstream from the fusion junction in the upp::lacZ transcript. In these experiments, the primer was in large molar excess. In addition to the standard procedure of identifying transcript-specific primer extension products by alignment with bands in a sequencing ladder, identities were confirmed by spiking sequencing reactions with primer extension products prior to analysis by gel electrophoresis (data not shown).

In vitro transcription. Purified RNA polymerase holoenzyme containing ⁷⁰ was prepared as previously described (8, 9, 11). DNA templates for in vitro transcription were gel-purified PvuII restriction fragments (derived from plasmid DNA) containing either the wild-type or mutant upp promoter regions. These fragments were identical to the EcoRI-BamHI restriction fragments used for construction of upp::lacZ gene fusions, except for several base pairs at each end. Transcription reaction mixtures (10 µl) contained 10 nM DNA template; 100 nM RNA polymerase; 20 mM Tris-acetate (pH 7.9); 10 mM magnesium acetate; 100 mM potassium glutamate; 0.2 mM Na₂EDTA; 0.1 mM dithiothreitol; 200 µM each ATP, CTP, and GTP; and either 50 or 1,000 µM UTP. The reaction mixture included either [-³²P]ATP or [-³²P]GTP (purchased from NEN) at a specific activity of 0.625 Ci/mmol. Reactions were initiated by addition of RNA polymerase, and the reaction mixtures were incubated at 37°C for 15 min. Heparin (1 µl of a 1-mg/ml solution) was then added to the mixture, and incubation was continued for an additional 10 min to permit the completion of elongating transcripts. Reactions were terminated by adding 10 µl of stop solution (7 M urea, 2 mM Na₂EDTA, 0.025% [wt/vol] each bromophenol blue and xylene cyanol) and placing the samples on ice. The samples were heated at 100°C for 3 min, and an equal volume of each sample was removed and run on a 25% polyacrylamide (29:1 acrylamide-bisacrylamide ratio)-50 mM Tris-borate (pH 8.3)-1 mM Na₂EDTA sequencing gel containing 7 M urea (28). Transcripts were visualized by autoradiography and quantitated by scanning gels with a Molecular Dynamics PhosphorImager. Transcripts were identified as previously described (35).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of shortening the T · A tract in the upp initially transcribed region on upp::lacZ expression and regulation. To examine the regulatory role of the long T · A tract in the upp initially transcribed region, we first constructed seven isogenic E. coli strains each containing either a wild-type or a mutant upp::lacZ gene fusion. Each gene fusion was carried on a lambda bacteriophage and inserted in single copy into the chromosomal lambda attachment site of strain CLT42 (car-94 lacZYA). The wild-type fusion contains the wild-type upp promoter region, and the six mutant fusions contain 1- to 6-bp deletions in the 8-bp T · A tract of the upp initially transcribed region. The promoters for the upp::lacZ gene fusions (and the fusions themselves) are hereafter referred to as T_n promoters (and fusions), where n equals the number of T · A base pairs present in the initially transcribed region.

View this table: [in this window] [in a new window]   TABLE 1.   Effects of (T · A)n mutations on upp::lacZ expression and regulationa

Analysis of upp::lacZ transcripts initiated at wild-type and mutant promoters. To further elucidate the effects of the deletions in the T · A tract, we used quantitative primer extension mapping to measure the levels and determine the start sites of upp::lacZ transcripts synthesized in wild-type and mutant fusion strains grown on uracil or UMP. Cellular RNA was isolated from cultures that were essentially identical to those described in Table 1. The primer used in these experiments hybridizes to the lacZ sequence contained in the fusions and therefore detects only upp::lacZ transcripts in the lacZYA strains. It should also be noted that nonproductive transcripts initiated at positions G6 and A7 (e.g., GAU_n and AU_n) are not detected in this assay. The results are shown in Fig. 2. In the case of the wild-type (T₈) and T₇ fusions, only transcripts initiated at the G6 start site were detected in uracil- and UMP-grown cells (Fig. 2A). The relative levels of these transcripts closely paralleled the cellular -galactosidase activities described above (Fig. 2B and Table 1), as expected for transcriptionally controlled fusions. In the case of the fusions containing shorter T · A tracts, transcripts initiated at both upp start sites (i.e., G6 and A7) were detected, indicating avoidance of reiterative transcription by A7 transcripts. At least with the T₆, T₅, and T₄ fusions, G6 transcript levels were also increased (Fig. 2A), indicating that these transcripts more efficiently avoided reiterative transcription. As with the wild-type and T₇ fusions, the relative levels of total upp::lacZ transcripts specified by the T₆, T₅, T₄, T₃, and T₂ fusions closely reflected cellular -galactosidase activities.

View larger version (28K): [in this window] [in a new window]   FIG. 2.   Levels of upp::lacZ transcripts initiated at the wild-type (T8) and mutant (Tn) promoters. Cellular RNA was quantitatively isolated from cells grown on either uracil (R) or UMP (M). Transcript levels were measured by primer extension mapping as described in Materials and Methods. (A) Autoradiograph of the 10% polyacrylamide sequencing gel that was used to separate the primer extension products. The autoradiograph was cut between the T7 and T6 lanes, and one part was placed beneath the other to facilitate labeling. The dideoxy sequencing ladder of the wild-type upp promoter region, which was used to identify transcripts, was generated with the same primer that was used for primer extension. (Note that the DNA sequence is for the template strand and is complementary to the transcript sequences.) The bands corresponding to G6 and (where detected) A7 transcripts are indicated. The positions of these bands in the gel are different for each promoter, reflecting the number of U residues specified by the T · A tract. Bands marked with an asterisk represent major transcripts produced by in vivo degradation of G6 and A7 transcripts. (B) Total productive (i.e., upp::lacZ) transcript levels in uracil- and UMP-grown cells were quantitated using a PhosphorImager and plotted in arbitrary units. Total transcript levels included G6 and A7 transcripts, degradation products, and escaped stuttering products in the case of the T3 fusion. (C) The levels of productive A7-initiated transcripts in uracil-grown cells were plotted as percentages of total productive transcript levels, calculated using the formula (A7/G6 + A7) × 100. In the case of the T3 fusion, escaped stuttering products were counted as A7 transcripts.

Analysis of short transcripts initiated at wild-type and mutant upp promoters in vitro. To directly analyze the effects of the deletions in the T · A tract of the upp initially transcribed region on reiterative transcription, we examined the short transcripts produced by transcription of wild-type and mutant upp promoter regions in vitro. These short transcripts include both the aborted products of reiterative transcription and the products of simple abortive initiation involving strictly templated transcription. Transcripts up to approximately 12 bases in length can be produced by simple abortive initiation at the upp promoter. We analyzed G6 and A7 transcripts separately by using either [-³²P]GTP or [-³²P]ATP, respectively, to label the 5' ends of transcripts. Transcription reaction mixtures contained either 50 or 1,000 µM UTP (200 µM each other nucleoside triphosphate) for the synthesis of G6 and A7 transcripts, respectively, to maximize initiation at the start site of interest. These UTP concentrations (i.e., 50 and 1,000 µM) roughly mimic those found in cells grown under conditions of pyrimidine limitation or excess, respectively (2, 25). Transcripts produced in vitro were separated on a 25% polyacrylamide sequencing gel and visualized by autoradiography.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   Analysis of short G6 transcripts initiated at wild-type and mutant upp promoters. DNA templates containing either the wild-type (T8) or a Tn mutant promoter were transcribed in vitro in reaction mixtures containing 200 µM each ATP, CTP, and [-32P]GTP and 50 µM UTP. The autoradiograph is of the 25% polyacrylamide gel used to separate the transcripts. The numbers on the left indicate transcript length (in nucleotides); brackets are used to indicate equal-length transcripts with different sequences, which in most cases cause these transcripts to migrate differently in the gel. In the case of the wild-type promoter (i.e., lane T8 only), several other transcripts are marked. Transcripts with the sequences GAU9 and GAU8G are labeled. Two longer GAUn transcripts (i.e., GAU10 and GAU11) are marked with asterisks. The longest unambiguously assigned product of simple abortive initiation, GAU8GU, is marked with a diamond.

View larger version (54K): [in this window] [in a new window]   FIG. 4.   Analysis of short A7 transcripts initiated at wild-type and mutant upp promoters. DNA templates containing either the wild-type (T8) or a Tn mutant promoter were transcribed in vitro in reaction mixtures containing 200 µM each [-32P]ATP, CTP, and GTP and 1,000 µM UTP. The autoradiograph is of the 25% polyacrylamide gel used to separate the transcripts. The lengths (in nucleotides) of transcripts with the sequence AUn (where n is 3) are shown on the left. The lengths of transcripts produced by simple abortive initiation are shown on the right. Note that the identity of the minor 5-mer transcript initiated at the T4 promoter is unknown. Also note that the AU2GU transcript, which is produced by simple abortive initiation at the T2 promoter, comigrates with the AU3G transcript initiated at the T3 promoter. Unlabeled transcripts initiated at the T2 promoter are discussed in the text.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although extensive reiterative transcription can occur at promoters that contain a T · A tract in the initially transcribed region as short as 3 bp (12, 17, 20), a much longer T · A tract is clearly required for normal pyrimidine-mediated regulation of upp expression in E. coli. The 8-bp T · A tract in the wild-type upp promoter can be shortened by a single base pair with minimal effect, but longer deletions significantly and progressively reduce the range of regulation. Regulation involving reiterative transcription appears to be completely eliminated when the T · A tract contains three or fewer base pairs.

The progressive loss of regulation caused by the deletion mutations is due primarily to regular increases in the level of upp expression in cells grown under conditions of pyrimidine excess (Table 1). These increases are due, in large part, to the avoidance of reiterative transcription by a significant fraction of A7 transcripts (Fig. 2A), which are the predominant products of transcriptional initiation at the upp promoter in uracil-grown cells (35). The fraction of A7 transcripts that avoid reiterative transcription increases as the length of the T · A tract decreases, as clearly indicated by the quantitative primer extension mapping of cellular transcripts (Fig. 2C).

A similar but less dramatic pattern of avoidance of reiterative transcription by A7 transcripts is observed in vitro (Fig. 4). Transcription of wild-type and mutant upp promoters indicates that avoidance of reiterative transcription by A7 transcripts, resulting in strictly templated transcription, can occur only when the T · A tract is shortened to six or fewer base pairs. The level of strictly templated transcription, as indicated by the products of simple abortive initiation, increases as the length of the T · A tract is reduced. However, a high level of reiterative transcription still occurs in vitro at all promoters except the T₂ promoter, at which reiterative transcription is abolished. The fact that reiterative transcription is so robust at promoters like T₃ and T₄, while regulation is severely restricted at these promoters, is somewhat surprising. One possible explanation is that only a very low percentage of initiation events need to be redirected from the reiterative to the strictly templated transcription pathway to produce the maximum level of full-length transcripts. Another explanation, at least for the T₃ promoter, is that some transcripts can enter the reiterative transcription pathway and then switch to the strictly templated mode to produce full-length transcripts. Clear evidence for such a switch at the T₃ promoter is provided by the results of quantitative primer extension mapping of cellular transcripts (Fig. 2A). Why such a switch is restricted to the T₃ promoter remains to be determined.

Avoidance of reiterative transcription by A7 transcripts initiated at mutant upp promoters may explain the loss of most, if not all, regulation of upp expression. However, avoidance of reiterative transcription at the mutant promoters is not restricted to A7 transcripts. The data from quantitative primer extension mapping of cellular transcripts indicate that approximately one-third of the G6 transcripts initiated at the wild-type upp promoter are subject to reiterative transcription (Fig. 2B). This reiterative transcription appears to be affected by shortening of the T · A tract similarly to that observed with the A7 transcripts. For example, reducing the length of the T · A tract to six or fewer base pairs causes an increase in the level of cellular G6 transcripts, which in this case can be detected in cells grown under conditions of either pyrimidine excess or limitation (Fig. 2A). However, the pattern of these increases is different than that observed with A7 transcripts. There is not a steady increase in the level of G6 transcripts with progressive shortening of the T · A tract. Instead, the highest levels of G6 transcripts are observed with the T₆ and T₅ promoters, with progressively lower levels of G6 transcripts with promoters T₄, T₃, and T₂. This pattern may indicate that shortening of the T · A tract to 6 bp is sufficient to achieve maximum avoidance of reiterative transcription by G6 transcripts. Consistent with this idea is the observation that in vitro, reiterative transcription involving G6 transcripts was sharply reduced by shortening of the T · A tract to six or fewer base pairs (Fig. 3). The relative decreases observed in vivo in G6 transcript levels with promoters T₄, T₃, and T₂ may reflect secondary effects on promoter strength or transcript stability. Consistent with these proposals, in vitro production of all short G6 transcripts at promoters T₄, T₃, and T₂ was relatively low (Fig. 3), perhaps indicating reduced promoter activity, and in vivo degradation of G6 and A7 upp::lacZ transcripts was greatly enhanced in the case of the T₂ promoter (Fig. 2A).

Overall, the results of this study indicate that a long T · A tract in the initially transcribed region of the wild-type upp promoter is necessary to achieve a particular balance between reiterative and strictly templated transcription involving G6 and A7 transcripts. This balance establishes a maximum or physiologically appropriate level of upp regulation. A long run of seven or eight T · A base pairs is required to ensure that essentially all A7 transcripts will engage in nonproductive reiterative transcription. A run of eight T · A base pairs is still short enough to allow the majority of G6 transcripts to avoid reiterative transcription and produce translatable mRNA. Presumably, the avoidance of reiterative transcription by G6 transcripts can occur by virtue of an rG · dC base pair formed by the 5' G residue of the transcript and its complementary C residue in the DNA template. The formation of this strong base pair inhibits upstream slippage of the nascent transcript that is a prerequisite for reiterative transcription.

The ability of G6 transcripts to avoid reiterative transcription is likely to be related to another feature of the transcriptional initiation complex, and that is the length of the RNA-DNA hybrid formed between the nascent transcript and the DNA template. The results presented in this paper are consistent with the formation of an 8-bp RNA-DNA hybrid during initiation of G6 transcripts. Such a hybrid would explain the elimination of essentially all reiterative transcription involving G6 transcripts at mutant promoters containing six or fewer base pairs in the T · A tract. The nascent transcripts synthesized at these promoters would be part of a relatively stable hybrid, anchored by a 5' rG · dC base pair, during the addition of all of the U residues specified by the T · A tract. The fact that about one-third of the G6 transcripts initiated at the T₇ and wild-type (T₈) promoters engage in reiterative transcription indicates that the RNA-DNA hybrid, at least a permanent hybrid, does not extend beyond 8 bp. The factors that allow most of the latter transcripts to avoid reiterative transcription (assuming that the 8-bp hybrid is correct) remain to be determined. Thus, the suggested length of the RNA-DNA hybrid at the upp promoter is the same as, or similar to, the 8- to 9-bp hybrid detected during transcriptional elongation (27, 33). It will be of interest to determine in future studies if the RNA-DNA hybrid length is the same at all promoters. If not, then the capacity to engage in reiterative transcription during initiation could be affected by this difference.