Mechanism of stimulation of ribosomal promoters by binding of the +1 and +2 nucleotides.

The rate of transcription of Escherichia coli ribosomal RNA promoters is central to adjusting the cellular growth rate to nutritional conditions. The +1 initiating nucleotide and ppGpp are regulatory effectors of these promoters. The data herein show that in vitro transcription is also regulated by the +2 nucleotide. Both the +1 and +2 nucleotides act by driving polymerase into an altered conformation rather than by increasing the lifetime of transcription complexes. The unique design of the ribosomal promoters may stabilize a distorted state of polymerase that is relieved by the binding of the two nucleotides required for transcription initiation.

The rate of transcription of Escherichia coli ribosomal RNA promoters is central to adjusting the cellular growth rate to nutritional conditions. The ؉1 initiating nucleotide and ppGpp are regulatory effectors of these promoters. The data herein show that in vitro transcription is also regulated by the ؉2 nucleotide. Both the ؉1 and ؉2 nucleotides act by driving polymerase into an altered conformation rather than by increasing the lifetime of transcription complexes. The unique design of the ribosomal promoters may stabilize a distorted state of polymerase that is relieved by the binding of the two nucleotides required for transcription initiation.
Transcription of ribosomal RNA is the limiting step in ribosome production (1). The ribosomal promoters sense the overall availability of nutrients and respond by producing amounts of rRNA that can support the maximal growth possible given the nutritional environment. Nucleotides have been shown to be critical effectors of this promoter response. The signal nucleotide ppGpp is produced in inverse concentration to growth rate, and this may help keep the rate of rRNA transcription at an appropriate level (2,3). The ribosomal promoters are also unusually sensitive to the concentration of the initiating nucleotide (4,5). During transitions to slow growth, the nucleotide concentration can be depressed, and this prevents unnecessary transcription of ribosomal components (3).
The mechanism by which nucleotides affect transcription from ribosomal promoters has received a great deal of attention. In the case of the initiating nucleotide, it was proposed that it works by stabilizing polymerase-promoter DNA complexes (4). This effect is thought to be specific to the ribosomal promoters because their complexes have short half-lives and are therefore uniquely unstable. ppGpp also reduces the complex half-life (6) and competes with the initiating nucleotide (5).
There are uncertainties associated with this mechanism of nucleotide control of ribosomal transcription. In order for the short lifetime of ribosomal complexes to reduce transcription, dissociation would need to occur before the first RNA bond is formed, but this is typically quite rapid (7)(8)(9)(10). How these unstable complexes are uniquely stabilized by the ϩ1 nucleotide is not known. In addition, recent experiments in other transcription systems have shown that the ϩ2 nucleotide can also have stimulatory effects. This occurs during initiation by the viral T7 RNA polymerase (11) and during elongation by the Escherichia coli RNA polymerase (12). Nucleotides that do not match the ϩ1 position on the template can stimulate the isomerization of E. coli RNA polymerase in complexes with fork-junction DNA. 1 The structure of such complexes is known (13) and gives no obvious clue as to how this stabilization by nucleotides could occur.
For these reasons we have re-evaluated the role of nucleotides as effectors of ribosomal transcription. The new data show that both the ϩ1 and ϩ2 nucleotides can act as effectors, unifying the properties of the bacterial and viral RNA polymerases. In these experiments, the nucleotides did not work by stabilizing transcription complexes but rather by causing a greater number of functional complexes to form. These effects can be rationalized in terms of the ribosomal promoter DNA sequence imposing a distorted conformation on RNA polymerase, which can be corrected by nucleotide binding.
In NTP stabilization experiments, 2 mM specified NTP was preincubated with the transcription complex for 5 min at 37°C. For heparin challenge experiments, 100 g/ml heparin was added to the transcription complex for 3 min at room temperature prior to the addition of the transcription nucleotide mix. In the half-life experiment, a scaled-up reaction with 2 nM holoenzyme was incubated with 1 nM pTH8-rrnB P1 in Buffer A for 5 min at 37°C. 20 nM BamHIII-cut pTH8-rrnB P1 competitor was then added. An aliquot was removed and added to a nucleotide mix as indicated above at various time intervals. Transcription was allowed to proceed for 10 min at 37°C. A control was run to obtain the background transcription with BamHIII-cut pTH8-rrnB P1, and this background was subtracted from the signal. In the control, 2 nM holoenzyme was preincubated with 20 nM BamHIII-cut pTH8-rrnB P1 for 5 min at 37°C, and then 1 nM uncut pTH8-rrnB P1 was added and transcription proceeded for 10 min at 37°C. The controls for reactions with the nucleotide used for the preincubation contained 2 mM of the indicated nucleotide in the incubation with holoenzyme-BamHIIIcut pTH8-rrnB P1. A formamide-urea dye mix was added to stop 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. reaction, and this was loaded onto a 6% urea gel. Bands were analyzed by a PhosphorImager (Amersham Biosciences).

High Concentration of Both the ϩ1 and ϩ2
Nucleotides Are Required at the rrnB P1 Promoter-A key property of the rrnB P1 promoter and its variants is that it transcribes poorly when the concentration of the ϩ1 initiating nucleoside triphosphate is low. Experiments indicated that nucleotide positions downstream from ϩ1 do not induce this property (4). We repeated these experiments under different conditions, namely substituting 37°C for lower temperature and using a template that contained the natural upstream sequence that enhances transcription (the "UP" element) (18,19). The experiment involved titrating the concentration of a single NTP, leaving the concentrations of the other 3 NTPs constant and fixed. The ribosomal promoter (rrnB P1) RNA sequence has an A at position ϩ1 and a C at position ϩ2 ( Fig. 1, top) Under these conditions, rrnB P1 transcription required a higher concentration of both the ϩ1 ATP and ϩ2 CTP. The data for the ϩ1 position replicated that obtained previously under different conditions (4). The ϩ2 effect had not been seen previously. The ATP and CTP curves are indistinguishable, indicating that the effect can be induced by either of the two nucleotides, which must bind polymerase and be condensed upon initiation. No similar effect was seen with GTP, which served as a control for a noninitiating nucleotide at this promoter. As the concentration of initiating nucleotides was increased beyond 200 M, transcription continued to increase (data not shown and see below).
Because the ϩ2 effect had not been seen previously, we repeated the type of experiment that previously supported the role of the ϩ1 nucleotide (4). The ϩ2 nucleotide was changed from C to G, and the experiment was repeated. The RNA sequence then had a ϩ2 G, and the ϩ1 A and all other nucleotides were left unchanged. The data (Fig. 2) show that the mutant promoter has a high concentration requirement for ATP and GTP, corresponding to its ϩ1 and ϩ2 positions. CTP, which behaved in this manner on the wild type promoter with a ϩ2 C, no longer did so when ϩ2 was changed to G. The magnitude of the effect was slightly reduced in the mutant, but the switch in specificity was clear. We concluded that both the ϩ1 and ϩ2 nucleotides have a higher concentration requirement at the rrnB P1 promoter.
The ϩ1 or ϩ2 NTP Can Activate Heparin-resistant Complexes-Previous experiments under lower ionic strength conditions have shown that, in the absence of initiating nucleotides, RNA polymerase binds the rrnB P1 promoter in a stable closed complex (20,21). Under similar conditions, preincubation with the ϩ1 (but not the ϩ2) nucleotide induce more complexes to survive dissociation in a heparin challenge protocol (4). In view of the additional effect of ϩ2 CTP in the above experiments, we reassessed the role of the ϩ1 and ϩ2 NTP under these experimental conditions. Fig. 3A shows the results of an experiment evaluating the effects of the ϩ1 or ϩ2 NTP in forming heparin-resistant complexes. In these experiments, each NTP was preincubated with the transcription complex, and then heparin was added for 3 min. At that point, transcription was initiated by adding the remaining NTPs. The data show that preincubation with either ϩ1 ATP (Fig. 3A, lane 2 versus lane 1) or ϩ2 CTP (lane 3 versus lane 1) increased transcription but that ϩ4 GTP had no effect (lane 4 versus lane 1). This is consistent with the specific need for high concentrations of the ϩ1 ATP and ϩ2 CTP in the titration experiments (Fig. 1). The effect of ϩ1 (reported previously in Ref. 4) is greater than that of ϩ2, which is clear but smaller.
Other experiments indicated that preincubation is not required to achieve the stimulatory effect of the nucleotide. Fig.  3B shows that, if either the ϩ1 or ϩ2 NTP is added after heparin, the results are unchanged from those described above. That is, there was still a specific stimulation (Fig. 3B,  is similar to panel A). In control experiments, the heparin was found to inactivate ϳ85% of the complexes if added directly to the polymerase prior to DNA (at the lac UV5 promoter, the effectiveness was greater than 95%, Fig. 3C). The results suggest that an initiating NTP can convert a poorly transcribing transcription complex to a better transcribing one, even in the presence of heparin.
It was reported previously that preincubation with the ϩ1 NTP increases transcription by stabilizing an intrinsically unstable transcription complex as assessed in a heparin challenge assay (4). The incomplete effect of heparin on inactivation at high ionic strength made this assay somewhat problematic. The prior heparin challenge assay used to determine complex lifetime was done using a lower KCl concentration (30 mM instead of 170 mM) (4), and we were able to reproduce the sensitivity to heparin under these low salt conditions (data not shown). We also found that the specificity of initiation was weak under these conditions, consistent with prior reports on general salt effects (22). To bypass these potentially complicating factors, we developed a modified assay to assess the lifetime of rrnB P1 promoter complexes at higher ionic strength, where NTP effects on transcription have been demonstrated (see above and Ref. 4).
High Concentrations of the ϩ1 and ϩ2 NTPs Do Not Significantly Stabilize Polymerase at the rrnB P1 Promoter-To measure the lifetime of RNA polymerase bound to the rrnB P1 promoter, we established a system wherein RNA polymerase does not efficiently reinitiate transcription from the template to which it is initially bound. In this system, RNA polymerase is prebound to the usual supercoiled template. A large quantity of competitor DNA was used, which was of the same template, except that it was cleaved prior to the usual transcription termination site so that its transcripts would be shorter and more distinguishable. Titration determined the amount of com-petitor that bound nearly all of the polymerase released after transcribing the original template. The small amount of reinitiated transcripts constituted a small background, which was corrected in the analysis. Under these conditions, polymerase initiated from the original template but could not rebind because of the large quantity of nearly identical promoter DNA, which acts as the competitor.
The response to the nucleotide was not changed under these conditions (data not shown). The effect of the nucleotide on inhibition by ppGpp was also tested using this competitor (Fig.  3D). ppGpp inhibited transcription 3-fold in the absence of nucleotide and only slightly less in the presence of the ϩ1 or ϩ2 NTP. The use of a DNA competitor did not appear to alter the regulatory properties of the promoter.
To measure the half-life of polymerase bound to the rrnB P1 promoter, complexes were assembled in the absence of NTPs, and competitor DNA was then added. At various subsequent times, aliquots were removed and NTPs were added to allow those complexes that had not dissociated to initiate transcripts. The amount of RNA at each time, corresponding to the number of functional complexes, was measured. The dissociation curve, which is plotted as the appropriate semilog plot, is shown in Fig. 4A. The half-life of the complexes under the experimental  1 and 3) and without heparin (lanes 2 and 4) for each promoter. Heparin was added to holoenzyme prior to the plasmid DNA in this case. D, same as in A but with HindIII-cut pTH8-rrnB P1 as the competitor instead of heparin, and 300 M ppGpp inhibitor was added with the remaining nucleotides to initiate transcription. conditions used here is 3-4 min, only slightly longer than that seen using the lower KCl conditions and heparin as a competitor (4).
When the experiment was repeated using transcription complexes, which had been preincubated with either the ϩ1 ATP (Fig. 4B) or the ϩ2 CTP (Fig. 4C), no significant change in the lifetime was seen. These lifetimes were much shorter than those obtained after the first bond formation (Ref. 6 and data not shown), ruling out the effects of NTP contamination in these experiments. In all three cases, the lifetime is 3-4 min (see legend to Fig. 4). It is important to note that these results were obtained under the same experimental conditions as described above, where preincubation with these nucleotides stimulated transcription. We infer that the cause of nucleotidedependent stimulation of transcription is not stabilization of polymerase at the promoter. Instead, the nucleotide must change some other property of the polymerase.

DISCUSSION
These experiments have addressed how the rrnB P1 ribosomal promoter is stimulated by the nucleotide, an effect thought to be central to regulation during growth phase transitions, where NTP concentrations change (3). The main new result was that this stimulation is associated with both the ϩ1 and ϩ2 initiating nucleotides. The effect was previously thought to be specific for ϩ1 NTP (4). Although the stimulation by ϩ2 was not anticipated, it helps unify our understanding of nucleotide effects; the ϩ2 NTP was previously shown to be an effector of elongation by E. coli RNA polymerase (12) and of initiation by T7 RNA polymerase (11). Under "Discussion," we place these results in context and suggest why ribosomal promoters are uniquely built to require higher concentrations of initiating nucleotides.
How Do the ϩ1 and ϩ2 NTPs Assist Ribosomal Transcription?-The data show that rrnB P1 transcription is dependent on high concentrations of the ϩ1 and ϩ2 NTPs, using two different protocols. First, low concentrations of an NTP corresponding to either the ϩ1 and ϩ2 RNA position led to low transcription (Fig. 1); low concentration of an NTP incorporated at a downstream position did not lower transcription (Fig. 1). This experiment differed from a prior one, which showed only a ϩ1 dependence (4) using a higher temperature (37°C) and using a promoter that contained the natural upstream sequence containing the UP element described previously under "Results." Second, when rrnB P1-polymerase complexes were preincubated with either the ϩ1 or ϩ2 NTPs, transcription was increased in a heparin challenge protocol. Taken together, the data show that transcription depends on the independent binding of the two nucleotides needed to form the first RNA bond.
The stimulation by the ϩ1 NTP has been suggested to be related to stabilizing the short-lived heparin-resistant complexes that form at ribosomal promoters (4). When we measured stability under conditions used here for transcription, the half-life was indeed short (3-4 min), but neither the ϩ1 nor the ϩ2 NTP lengthened it substantially. It is possible that the short half-lives are a consequence of unusual features of the transcription complex but are not the basis for stimulation by the ϩ1 and ϩ2 NTPs.
Instead, the ϩ1 and ϩ2 nucleotides appeared to increase the ability of bound RNA polymerase to transcribe. Fig. 3B shows that the ϩ1 or ϩ2 NTP can increase transcription even after heparin was added to the transcription complexes under these conditions (4). This indicates that the NTPs can convert inactive complexes to active ones. The conformation of polymerase at the rrnB P1 promoter appears to be distorted from that of a normal preinitiation complex, because NTPs have an abnor-mally high K m (Fig. 1 and reported previously for the ϩ1 NTP) (4), indicating that their binding sites are hindered or malformed. Thus, high concentrations of initiating NTPs are needed to fill these sites, but once bound, should induce restoration of the site configuration and activate the polymerase. The specific requirement for the ϩ1 or ϩ2 NTP is likely a consequence of the unique existence of these two sites in the preinitiation complex. rrnB P1 transcription needs high concentrations of these NTPs because its complexes, which contain these sites, may be distorted compared with the usual state at nonribosomal promoters.
A Model for the Unique Role of Nucleotides at Ribosomal Promoters-Even nonribosomal promoters have a slightly increased K m for binding the ϩ1 and ϩ2 NTPs compared with those used in elongation (9). Somehow this effect is exaggerated at ribosomal promoters to the extent that it allows the promoters to be regulated when NTP concentrations drop. Several features of the promoter are known to contribute to this effect (23). One prominent feature is the uniquely short spacer length of ribosomal promoters, which is central to control by both the initiating nucleotide (23) and growth rate regulation (24). The short spacer is interesting in several regards. It should distort the transcription complex, because a suboptimal-length spacer DNA cannot be accommodated without either the DNA changing its twist or the polymerase distorting to match the improperly presented Ϫ10 and Ϫ35 regions (25). Region 3 of senses the spacer length (26) and would likely be distorted at the ribosomal promoters. Two lines of evidence indicate that a region 3.2 distortion could be associated with creating a requirement for high concentrations of initiating NTPs. First, mutations and a truncation that includes region 3.2 lead to a decrease in abortive transcription (27,28), which can be compensated for by higher concentrations of initiating nucleotides (28). Second, region 3.2 is close to an NTP binding site, as demonstrated by its ability to cross-link to a bound ATP (29). Thus, the unique short spacer of the ribosomal promoters may cause region 3.2 to occlude the nearby NTP binding sites. This can be corrected by driving the binding with high concentrations of the initiating NTP.
Although the short spacer is a hallmark of ribosomal promoters and contributes to its regulation, another unusual feature is a long stretch of DNA between the downstream promoter element and the start site. The discriminator for ppGpp control is found here, and mutations (23) in this region result in a general loss of regulation. This long distance may increase the distortion of the complex. The manner of accommodation of the unusually large DNA segment could be affected by the sequence elements in the single-stranded DNA. Taken together, these features could be associated with distortions in the ϩ1 and ϩ2 NTP binding sites, which create the need for high concentrations to fill them. Once they are filled, the complex likely becomes undistorted, as a cross-linking study shows that NTP binding can alter the polymerase conformation to induce engagement of the ϩ1 position on the template strand. 1 These same features may also contribute to the down-regulation of ribosomal transcription by the nucleotide effector ppGpp. It was recently suggested that ppGpp competes with ATP for its binding site at rrnB P1 (5). ppGpp may be a particularly effective competitor at ribosomal promoters because the NTP binding sites are distorted and have a high K m for the ϩ1 and ϩ2 NTPs. Under certain conditions of nutritional deprivation, there may be synergistic down-regulation via nucleotides because the low ATP concentration leads to less transcription (3) and also allows more competition by the inhibitor ppGpp (5). Although the loss of NTPs may be accompanied by an increase in nucleoside diphosphate amounts, nucleo-side diphosphates have a significantly higher K m (not shown) and would be ineffective stimulators. When these nucleotide effects are coupled with repression caused by the reduction in DNA supercoiling (30), very substantial down-regulation of ribosomal transcription occurs.