The Role of the Alarmone (p)ppGpp in ςN Competition for Core RNA Polymerase*

Some promoters, including the DmpR-controlled ςN-dependent Po promoter, are effectively rendered silent in cells lacking the nutritional alarmone (p)ppGpp. Here we demonstrate that four mutations within the housekeeping ςD-factor can restore ςN-dependent Po transcription in the absence of (p)ppGpp. Using both in vitro and in vivotranscription competition assays, we show that all the four ςD mutant proteins are defective in their ability to compete with ςN for available core RNA polymerase and that the magnitude of the defect reflects the hierarchy of restoration of transcription from Po in (p)ppGpp-deficient cells. Consistently, underproduction of ςD or overproduction of the anti-ςD protein Rsd were also found to allow (p)ppGpp-independent transcription from the ςN-Po promoter. Together with data from the direct effects of (p)ppGpp on ςN-dependent Po transcription and ς-factor competition, the results support a model in which (p)ppGpp serves as a master global regulator of transcription by differentially modulating alternative ς-factor competition to adapt to changing cellular nutritional demands.

Escherichia coli holoenzyme RNA polymerase is composed of a core enzyme (E, 1 subunit composition ␣ 2␤␤ Ј) associated with one of seven sigma ()-factors that program the complex to engage and initiate transcription at different sets of promoters (1). Thus, the levels and binding properties of alternative -subunits together with factors that modulate their ability to associate with core RNA polymerase are critical for the relative composition of the multiple holoenzymes available for transcription of the distinct promoter classes within the prokaryotic genome. The seven different -factors of E. coli fall into two groups. The larger of these comprises six factors that share notable sequence and functional similarities to the major D ( 70 )-factor that is responsible for transcribing "housekeeping" genes (2). Recent structural studies have shown that the Dlike proteins comprise three globular domains that encompass previously identified conserved regions ( 2 (1.2 to 2.4), 3 (3.0 to 3.1), 4 (4.1 to 4.2)) that are tethered by flexible linkers (3)(4)(5)(6). In contrast, the alternative N ( 54 )-factor is in a class on its own, bearing little sequence homology to other -factors and determining recognition of the well conserved but unusual Ϫ24, Ϫ12 (TGGCAC-N 5 -TTGC) promoter sequences (7). In addition, there are significant differences in the action of the cognate holoenzymes at promoters. Unlike E D , which can undergo transition from the initial closed complex to the transcriptionally competent open complex without any other regulatory factor, the E N closed complex is kinetically and thermodynamically stable and E N cannot melt promoter DNA on its own. Transition to the open complex is dependent on interaction with, and nucleotide hydrolysis by, a member of the bacterial family of enhancer binding proteins (reviewed in Ref. 8).
Promoters of the N -dependent Ϫ24, Ϫ12 class direct transcription of genes involved in a variety of physiological processes responsive to nutrient limitation such as nitrogen assimilation and fixation, substrate-specific transport systems, and utilization of alternative carbon and energy sources. Appropriate environmental signals lead to activation of the cognate regulator by diverse mechanisms that result in a common active form of the regulator (9). For DmpR, which controls transcription of the N -Po promoter of an operon encoding the enzymes for metabolism of (methyl)phenols in Pseudomonas CF600, the activation mechanism is direct. Binding of aromatic phenolic pathway substrates to its N-terminal regulatory A-domain alleviates interdomain repression to give the active form of the protein (10,11), and DmpR mediated aromatic effectorand ATP-dependent transcription from Po can be fully reconstituted in vitro (12,13).
In addition to the specific DmpR-mediated control mechanism described above, transcription from the N -Po promoter is dependent on the unusual nucleotides guanosine tetraphosphate (ppGpp) and guanosine pentaphosphate (pppGpp), collectively referred to as (p)ppGpp. These molecules are heralds of metabolic stress and were originally identified as the mediators of the classical stringent response to down-regulate superfluous stable RNA synthesis upon amino acid starvation (reviewed in Refs. 14 and 15). Synthesis of these molecules by ribosome-associated RelA (p)ppGpp synthetase is activated by the arrival of an uncharged tRNA on the ribosome, whereas the dual function SpoT protein (which is primarily responsible for (p)ppGpp degradation) catalyzes (p)ppGpp synthesis in response to glucose starvation (16). Since their discovery, however, these signaling molecules have also been implicated in the up-regulation of transcription from many classes of promot-ers including those dependent on D (e.g. Refs. 16 and 17), the stationary phase -factor S (e.g. Refs. 18 -22), the heat shock -factor H (e.g. Refs. [22][23][24], as well as N (e.g. the Po and Pu promoters; Refs. [25][26][27]. Down-regulation of transcription from stringent promoters is believed to occur through the effects of binding of (p)ppGpp at the interface of the ␤and ␤Ј-subunit of RNA polymerase (28,29). Binding of (p)ppGpp destabilizes E D -promoter complexes, and, because open complexes formed by E D at rRNA P1 promoters are intrinsically very unstable, it has been proposed that these promoters are likely to be particularly sensitive to (p)ppGpp destabilization (30). Support for this mechanism comes from analysis of suppressor mutations within ␤ and ␤Ј (isolated by rescue of the polyauxotrophic growth defect of (p)ppGpp-deficient strains) that also destabilize ⌭ D -promoter open complexes in vitro (31)(32)(33). One possible consequence of down-regulation of transcription from the powerful stringent promoters is an increase in the pool of core RNA polymerase that is normally sequestered in producing these abundant transcripts. Thus, accumulation of (p)ppGpp and, by analogy, suppressor mutations has been proposed to increase the amount of E D available for (p)ppGpp-stimulated D -promoters that are difficult to saturate and have E D recruitment as the ratelimiting step (33)(34)(35).
The seven -factors of E. coli exhibit quite different affinities for core RNA polymerase (reviewed in Ref. 36), and (p)ppGpp has the potential to play multiple roles in their function. In addition to the potential to directly alter the promoter recognition and kinetics of transcriptional initiation at promoters, (p)ppGpp could also plausibly modulate the -association properties of the core RNA polymerase. Models involving the influence of (p)ppGpp on -factor competition for limiting available core have been put forward to explain poor induction of S -and H -dependent promoters in (p)ppGpp-deficient cells (20,21,37). Most recently evidence for a direct effect of (p)ppGpp on the competitive abilities and levels of S and H associated with core RNA polymerase has been presented (22). Primarily based on the ability of two mutations in D to restore transcription to a N -dependent promoter in the absence of (p)ppGpp, we have previously postulated that modulation of -factor competition may at least in part underlie the (p)ppGpp dependence of the Po promoter that requires the action of the structurally and functionally distinct N protein (25). Here, we dissect the potential of (p)ppGpp to directly affect N -dependent Po transcription and to modulate N competition by (i) identifying new suppressor mutations that restore transcription from Po in (p)ppGpp-deficient cells and (ii) performing in vivo and in vitro -factor competition in the presence and absence of (p)ppGpp. We present evidence that (i) the levels of alternative -factors have a large impact on the output from the Po promoter and (ii) the requirement for (p)ppGpp can be suppressed by four mutations in D that exhibit defects in competition against N , or by underproduction or sequestering of D . These results strongly support a model in which (p)ppGpp serves as a master regulator of transcription through -factor competition for limiting core RNA polymerase and brings the performance of the structurally and functionally distinct N protein within the global control of the stringent response.

EXPERIMENTAL PROCEDURES
Strains and Culture Media-Strains used are listed in Table I and were routinely cultured and assayed in Luria broth (LB; Ref. 38) supplemented with appropriate antibiotics for selection. Where indicated, assays were also performed using rich defined medium consisting of M9 minimal medium (38) supplemented with 22 amino acids (50 g/ml, Sigma kit), serine at 1 g/liter as the carbon source, all five nucleotides (20 g/ml), trace metals, and thiamine (0.05 mM). E. coli MG1655⌬lac and CF1693⌬lac strains carrying P trp -rpoD were generated by transferring (CAM⍀) P trp -rpoD from E. coli MC4100 (CAM⍀) P trp -rpoD into these strains by P1-mediated transduction using the tetracycline resistance aer-3075::Tn10 marker and testing for chloramphenicol resistance and/or indole-3-acrylic acid (IAA) dependence. Prior to analy- sis, all the rpo alleles were introduced in clean genetic backgrounds by P1-mediated transduction exploiting the thi-39::Tn10 marker for rpoBC alleles, aer-3075::Tn10 for rpoD alleles, and zhc-6::Tn10 for the rpoN allele. DNA sequencing was used to monitor co-transduction of the mutant alleles. Suppressor phenotypes were assessed using a Po-luxAB reporter and the ability of the strains to grow on M9 minimal media plates supplemented with 10 mM glucose and 100 g/ml thiamine.
Plasmids and DNA Manipulations-Plasmids and primers are listed in Tables II and III. Plasmids were constructed by standard recombinant techniques and the fidelity of all PCR-derived DNA confirmed by sequencing. Plasmids pVI466 and pVI684 are equivalent dmpR-Po-luxAB luciferase reporter plasmids carried on either an RSF1010-based vector pVI466 (copy numbers 16 -20) or a p15A-based vector pVI683 (copy numbers 18 -22) generated by replacing the EcoRI-to-NaeI kanamycin resistance gene and promoter region (bp 93-1440) of pPRO-Lar.A331 (Clontech) with an EcoRI-to-SmaI PCR-amplified spectinomycin/streptomycin gene generated from pUT-mini-Tn5-Sp/Sm using primers Sp1/Sp2. The NotI dmpR-Po-luxAB fragment of pVI466 was inserted into the unique NotI site of pVI683 to generate pVI684.
The dmpR-Po-dmpB plasmid pVI686 used as a reporter for P1linkage analysis was assembled from previously cloned regions of the dmp cluster of Pseudomonas CF600 and consists of a 4-kb HindIII-to-SmaI dmpR-Po-spanning fragment fused to a 2.3-kb SmaI-to-HpaI dmpB encoding fragment on the RSF1010-based cloning vector pVI398. The pVI687 dmpR-Po-tet selection plasmid has the same Po reporter fusion point as pVI686. Plasmid pVI687 was generated by first cloning a 2.4-kb NruI-to-SmaI dmpR-Po-spanning fragment into the EcoRV site of pBluescript SKϩ followed by insertion of a HindIII-to-KpnI fragment carrying the promoterless tetracycline resistance gene, generated from pBR322 using primers Tc1/Tc2.
The in vitro transcription template plasmid pVI695 is based on pTE103 and carries an EcoRI-to-BamHI Po promoter fragment generated using primers Po1 and Po2 (Table III). This Po promoter fragment spans from Ϫ408 to ϩ26 relative to the transcription start site and thus encompasses the DmpR binding sites (upstream activating sequences Ϫ170 to Ϫ127), the IHF binding site (-72 to Ϫ37), and the Ϫ24, Ϫ12 Po promoter (39).
The RpoN expression plasmid pVI688 carries the E. coli rpoN gene as an EcoRI-to-KpnI fragment spanning from 34 bp upstream of the ATG start to 7 bp downstream of the termination codon. The PCR-amplified fragment was generated using primers N1/N2 and cloned between the EcoRI-to-KpnI region of pEXT21. Plasmids pVI690 to pVI694, for expression of wild type and mutant D , were generated by assembling two PCR-amplified fragments between the NdeI and EcoRI sites downstream of the expression cassette of pJLA503. Primers D1/D2 were used to generate an NdeI-to-BamHI fragment with the NdeI site encompassing the ATG start codon and extending to the internal BamHI site of MG1655 rpoD. Primers D3/D4 were used to generate BamHI-to-EcoRI fragments from wild-type or mutant rpoD derivatives to reconstitute full-length rpoD from the internal BamHI to 12 bp downstream of the termination codon.

Selection of Second Site Suppressor
Mutants-Independent cultures of E. coli CF1693⌬lac harboring pVI687 (dmpR-Po-tet) were grown for 2 h with shaking at 30°C in 2 ml of LB supplemented with carbenicillin (50 g/ml) and 0.5 mM 2-methylphenol (the most potent aromatic effector of DmpR). Tetracycline was then added to 20 g/ml and the culture incubated under the same conditions for another 2 h prior to dilution and plating on equivalent solid media and growth overnight at 30°C. Equal numbers of large and small independent isolates were restreaked and screened for DmpR dependence of the tetracycline resistance phenotype by testing for Tc R in the presence and absence of 0.5 mM 2-methylphenol. Of the original 102 isolates, all exhibited DmpR effectordependent tetracycline resistance. However, 33 were discarded as being unstable, P1 phage-resistant, and/or having suppressor phenotypes too poor to map by P1 transduction as described below.
Localization and Identification of Mutations-Given the predominance of rpoBC mutants in a previous prototrophy rescue screen (40), isolates were first screened for alterations in rpoBC by using a nonisotopic RNA cleavage assay (NIRCA) kit (MutationScreener TM ) supplied by Ambion, Inc. and/or P1 transduction. For NIRCA analysis, each ϳ5-kb gene was amplified by colony PCR from each mutant strain using rpoB B1/B2 and rpoC C1/C2 primers and Expand High Fidelity Taq polymerase (Roche Molecular Biochemicals). Using these products as templates, five overlapping regions (of ϳ1000 bp) of each gene were amplified using primers that incorporated the T7 promoter region at the 5Ј end (BT7-1A/B to BT7-5A/B and CT7-1A/B to CT7-5A/B; Table  III). These products were then mixed with the corresponding section amplified from wild-type E. coli MG1655, transcribed to form RNA, and the hybridized products subjected to limited RNase cleavage. The sizes of RNase-cleaved bands resulting from mismatches were used to estimate the location of the mutation, which was subsequently confirmed by DNA sequencing. For the majority of isolates, assignment was made to the rpoBC region as soon as a mutation was detected during sequential analysis from the 5Ј region of rpoB through to the 3Ј region of rpoC. In the case of the three rpoB and three rpoC alleles further analyzed in this study, NIRCA analysis of the entire rpoBC region of each strain revealed the single genetic changes listed in Table I. However, even after introduction into a clean genetic background, because NIRCA analysis does not detect all mutations (see below), there is a small possibility that these derivatives may also contain additional closely linked but undetected mutations that contribute to their phenotype.
The NIRCA analysis assigned mutations in 53 isolates to rpoBC. Mutations in an additional 12 isolates were assigned to the rpoBC region by P1 linkage analysis. The assay system used to determine linkage frequencies utilized the thi-39::Tn10 marker and CF1693⌬lac harboring the reporter plasmid pVI686. Plasmid pVI686 mediates DmpR-Po-dependent expression of the dmpB-encoded catechol-2,3-dioxygenase, which converts catechol to the bright yellow product 2-hydroxymuconic semialdehyde. Co-transduction of thi-39::Tn10 and a suppressor mutation was scored among Tc R transductants grown in the presence of 0.5 mM 2-methylphenol by determination of the number of yellow versus white colonies after spraying with catechol (41). Ap R , E. coli rpoD expressed from P R P L This study pVI691 Ap R , rpoD-ƒDSA(536-538) expressed from P R P L This study pVI692 Ap R , rpoD-Y751H expressed from P R P L This study pVI693 Ap R , rpoD-P504L expressed from P R P L This study pVI694 Ap R , rpoD-S506F expressed from P R P L This study pVI695 Ap R , pTE103 carrying the Po promoter (Ϫ480 to ϩ26) and rna P1 of the vector This study The remaining four isolates with mutations unlinked to rpoBC were subjected to NIRCA analysis for mutations in both rpoD (using primers D5/D6 and DT7-1A to DT7-2B) and rpoN (using primers N3/N4 and NT7-1A to NT7-2B) (Table III). This analysis identified two suppressors, sup40 (with a mutation in rpoD) and sup35 (with mutations in both rpoD and rpoN). DNA sequence analysis of the entire rpoN and/or rpoD genes was used to confirm the genetic changes listed in Table I. In addition to changes that result in two substitutions within RpoN, the sup35 strain also possessed eight additional silent mutations in the wobble positions of codons within rpoN. The original sup-35 strain has also lost Km R associated with the ⌬relA251::Km of the host. Therefore, the detected mutant rpoD and rpoN alleles of this strain were only analyzed independently in authentic CF1693⌬lac.
Purified Proteins-Purification of E. coli N , IHF, ⌬A2*-His-DmpR, and DmpR-His has previously been described (12,39). Core RNA polymerase was purchased from Epicentre Technology. For purification of E. coli D and its mutant derivatives, expression plasmids pVI690 to pVI694 were introduced into the cognate CF1693⌬lac derivatives prior to purification as previously described (42).
Synthesis and Purification of ppGpp-Preparative-scale synthesis and purification of ppGpp was performed using two different methods that resulted in indistinguishable preparations that gave the same results in in vitro assays. Synthesis using native ribosome-associated RelA, prepared from LB-grown E. coli MG1655 pALS10 cells cultured in the presence of 1 mM isopropyl ␤-D-thiogalactopyranoside (IPTG) during the last 1 h of growth, was essentially as described (43) except that equimolar ratios of GTP and GDP were used as substrates. Synthesis using a purified His-tagged RelA protein was essentially as described in Ref. 27, with the minor modifications described in Ref. 22. Pure ppGpp was lyophilized and stored at Ϫ80°C until use. Purity of the preparations was monitored by thin layer chromatography on polyethyleneimine cellulose plates (Merck), using 1.5 M KH 2 PO 4 (pH 3.4) as chromatographic buffer. Concentrations of ppGpp were determined spectrophotometrically at A 260 using the molar extinction coefficient of 13,700.
In Vivo Luciferase Transcription Assays-Cultures for in vivo luciferase transcription assays were grown overnight in the test media, diluted and grown to early exponential phase, and then diluted once more prior to initiation of the experiment by the addition of 0.5 mM of the DmpR effector 2-methylphenol. For cultures of strains harboring the P trp -rpoD system, cultures were grown as described above in LB supplemented with 0.2 mM IAA, concentrated by centrifugation, and then used to inoculate the media indicated. Luciferase activity of LuxAB within whole cells was assayed with a 1:2000 dilution of decanal as described previously (44).
In Vitro Transcription-Assays (20 l) were performed at 37°C in transcriptional buffer containing 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM EDTA, and 0.275 mg/ml bovine serum albumin. Core RNA polymerase (10 nM) was allowed to associate with variable amounts of wild type or mutant D and/or N for 5 min in the presence of 4 mM ATP (required for DmpR activity) and ppGpp, where added. Open complex formation was initiated by the addition of 0.5 g of supercoiled pVI695, IHF (10 nM), DmpR-His (50 nM) or ⌬A2*-His-DmpR (100 nM), and the aromatic effector 2-methylphenol (0.5 mM). After 20 min of incubation to allow open complex formation, multiple-round transcription was initiated by adding a mixture of ATP, GTP, and CTP (final concentration, 0.4 mM each), UTP (final concentration, 0.06 mM), and [␣-32 P]UTP (5 Ci at Ͼ3000 Ci/ mmol, Amersham Biosciences). Re-initiation was prevented after 5 min by the addition of heparin (0.1 mg/ml), and 5 min later the reactions were terminated by adding 5 l of stop/load buffer (150 mM EDTA, 1.05 M NaCl, 14 M urea, 3% glycerol, 0.075% xylene cyanol, and 0.075% bromphenol blue). Samples were analyzed on a 7 M urea, 4.5% acrylamide sequencing gel and quantified using an Amersham Biosciences PhosphorImager.
Western Blot Analysis-Crude extracts of cytosolic proteins, SDSpolyacrylamide gel electrophoresis, transfer to nitrocellulose filters, and Western blot analysis with polyclonal rabbit anti-D and anti-E D (gift from M. Jishage) or mouse monoclonal anti-N sera (Neoclone) were as previously described (45). Antibody decorated bands were revealed using chemiluminescence reagents (Amersham Biosciences) as directed by the supplier. Differences in expression levels were assessed by comparison of different exposures of dilution series of crude extracts. Specificity of antisera was monitored using genetic control strains proficient and deficient in expression of the gene product and/or purified proteins.

Mutations in RNA Polymerase
Restore Transcription from Po in the Absence of (p)ppGpp-In wild type E. coli, transcriptional output from the Po promoter carried on a luciferase reporter plasmid (pVI466 dmpR-Po-luxAB) is induced by more than 3 orders of magnitude by the presence of aromatic effectors. This genetic system reproduces the exponential/stationary phase transition induction observed in the native system (Fig. 1A, and Ref. 44). The dependence of the Po promoter on (p)ppGpp leads to an ϳ10-fold decrease in maximal transcription from Po in (p)ppGpp-deficient strains (Fig. 1A, and Refs. 25, 26, and 44). We utilized this 10-fold effect on Po transcription to genetically select mutations that restored transcription to Po in the absence of (p)ppGpp (Fig. 1B).
To generate mutants we used a selection plasmid carrying the DmpR-controlled Po promoter driving transcription of a promoterless tetracycline resistance gene (pVI687, dmpR-Potet). This plasmid confers aromatic effector-dependent tetracycline resistance in the E. coli (p)ppGpp ϩ strain, but is incapable of conferring tetracycline resistant in a (p)ppGpp 0 counterpart. Spontaneous mutants were selected by asking for growth of (p)ppGpp 0 E. coli on LB containing tetracycline and an aro- matic effector of DmpR. P1 transduction and/or a non-isotopic RNA cleavage assay were used to determine the location of the mutations in the independent isolates as described under "Experimental Procedures." Of the 69 mutants that proved amenable to analysis, 65 harbored mutations in rpoBC (␤␤Ј), 1 had a mutation in rpoD ( D ), 1 had multiple mutations in rpoN ( N ) in addition to a mutation in rpoD, and 2 had mutations elsewhere in the genome. Thus, the majority of isolates harbor mutations in components of the RNA polymerase, and, of these, 97% had mutations in rpoBC and 3% had mutations in rpoD.
By screening a selection of the mutants using the dmpR-Po-luxAB luciferase reporter plasmid pV1466, we found that the degree of restoration of Po transcription in (p)ppGpp 0 CF1693⌬lac varied considerably from barely detectable to Ͼ4fold the level observed in the (p)ppGpp ϩ MG1655⌬lac counterpart in the case of one rpoB allele (data not shown). This suggests that the selection procedure is very sensitive and allows isolation of suppressors that differ by Ͼ40-fold in their ability to restore transcription from Po. The two rpoD alleles, the single rpoN allele, and six rpoBC alleles (Table I) were chosen for direct comparison with two previously isolated prototrophy-restoring rpoD alleles (P504L and S506F; Ref. 40) that also restore transcription to Po in (p)ppGpp 0 E. coli (25). The majority of the mutants shown in Fig. 1B restored Po N -dependent transcription to between 0.5-and 1.5-fold of the levels observed in (p)ppGpp ϩ MG1655⌬lac. However, the rpoD-35 (Y571H) and rpoN-35 (E150D/I165M) mutants are markedly poorer in their ability to restore transcription.
Mutant D Subunits Are Defective in Competition against N in Vitro-The data above clearly demonstrate that mutations in D can efficiently compensate for the effect of (p)ppGpp in vivo to restore transcription to the N -dependent Po promoter. Given the distinct classes of promoters recognized by the two -factors, it is unlikely that the D derivatives directly affect Po output. The most plausible interpretation of this data is that the mutant D subunits mediate their effects through deficient competition for limiting core RNA polymerase. To directly test this idea, we set up a multiple-round in vitro transcription (IVT) assay to simultaneously monitor the output of the N -dependent Po promoter and the D -dependent RNA1 promoter present on the template pVI695. This assay system, employing either a constitutively active form of DmpR deleted of its regulatory A-domain or aromatic effector-activated DmpR, results in clearly distinguishable N -and D -dependent transcripts ( Fig. 2A, compare lanes 1 and 2), which can be simultaneously monitored within the same reaction ( Fig. 2A, lane 4). Inclusion of 0.5 mM aromatic effector 2-methylphenol used in assays with DmpR-His did not alter specificity or the output from the IVT assay.
As a first step in the analysis, we added increasing amounts of the individual D subunits (Fig. 2B) or N (Fig. 2C) to a set concentration of core (10 nM) and supercoiled template (0.5 g). In addition, all assays contained effector-activated DmpR-His (50 nM) and ATP (4 mM) required for N -dependent transcription, and IHF (10 nM) required for optimal Po output (39). Consistent with previous data, comparison of the saturation curves for wild type D and N (closed symbols in Fig. 2, B and C) indicate that these two -factors have a similar high affinity for core (46 -48). For the D mutants, the plateau saturation values are all lower than observed for wild type D (Fig. 2B). Because formation of the holoenzyme is a prerequisite for promoter binding and transcriptional initiation by E D , transcription by E D can be considered as at least a second order co-operative binding event. Lower plateau values would thus be predicted for a defect in the initial binding step, which would lead to lower levels of the holoenzyme at equilibrium. However, it is possible or even probable that the mutations may also have some other defect in the complex series of events of transcriptional initiation that may also contribute to the lower net output.
To assess relative competitiveness of the different D -factors, we first performed a direct comparison of the ability of increasing concentrations of N to compete for 10 nM core polymerase in the presence of 20 nM D -wt or the most severely affected D -derivative ( D -ƒDSA-(536 -538)). The results in Fig. 3A illustrate the severe competitive defect of D -ƒDSA-(536 -538) and show that, at this fixed D concentration, greater than 5-fold more N is required to achieve a 50% reduction of RNA1 transcript levels with D -wt than with D -ƒDSA-(536 -538).
Based on the results above, we determined the relative competitiveness of all the mutant D subunits by measuring the ability of 100 nM N to compete for 10 nM core polymerase in the presence of 40 nM (Fig. 3B) or 20 nM (Fig. 3C) of the D mutants. The results show that the mutants differ substantially in their ability to compete, ranging from minimally effected ( D -Y571H) to severely affected ( D -ƒDSA-(536 -538)). Consistent with a critical role in -factor competition in vivo, the relative order of the defects of the these mutants follows the order of their ability to restore Po transcription in the absence of (p)-ppGpp (Fig. 1B) Po promoter output, we measured Po transcription in the presence of 200 nM to 800 M ppGpp in a multiple-round IVT assay. As shown in Fig. 4A, no major stimulatory effect of ppGpp was observed, and the highest concentration resulted in moderate inhibition. The minor stimulatory effect observed in the low micromolar ppGpp range was also obtained when the assay was performed with a constitutively active form of DmpR, ⌬A2-His-DmpR, deleted of its regulatory A-domain (data not shown). No further stimulatory effects were observed by changing assay parameters likely to modulate promoter kinetics, namely (i) reduction of DmpR concentrations from the saturating 50 nM concentration to as low as 2.5 nM, (ii) reducing temperature from 37°C to 30 or 20°C, or (iii) shortening open complex formation time from 20 min to 8 or 3 min (data not shown). Hence we conclude that ppGpp has no major stimulatory effects on DmpR-mediated Po transcriptional output un-der the in vitro conditions used.
To test whether ppGpp has any direct effect on N / D competition in vitro, we added ppGpp to a competition assay for limiting core (10 nM). In these experiments N was held at a constant concentration of 40 nM and challenged with increasing amounts of D from 0 to 100 nM. As shown in Fig. 4B, although D effectively competes with N , leading to a decrease in Ndependent Po transcription to ϳ20% at the highest concentration tested, addition of either 20 or 180 M ppGpp had no discernible effect. Using a similar experimental set-up, addition of 180 M ppGpp to an in vitro competition assay has recently been shown to result in an ϳ2-fold difference in inhibition of transcription from the H -dependent dnaK promoter caused by competing D (22). Thus, in contrast to the case of the Results are the average of duplicate determinations from two independent experiments. The same hierarchy of the mutations was also obtained using single-round IVT assays (data not shown). low affinity D -like H , ppGpp does not appear to have a detectable direct effect on competition between high affinity N and D in vitro.
The results outlined in this and preceding sections suggest that, although both N and D have similar high affinity for core when assessed in isolation (Fig. 2), clear differences between these two -factors become apparent under conditions of in vitro competition for limiting core. The data show that increasing concentrations of N are significantly poorer at outcompeting a fixed concentration of D than the converse. This is exemplified by the finding that a 2.5 M excess of D over N causes a ϳ4-fold decrease in N -dependent transcription (Fig.  4), whereas the equivalent molar excess of N over D causes only a ϳ1.5-fold decrease in D -dependent transcription (Fig.  3). These considerations prompted the series of experiments described below, designed to assess the effect of modulation of -factor levels on the transcriptional output from the Po promoter in vivo.
Effect of Modulating N and S Levels on in Vivo Transcription of Po-The levels of N in E. coli are constant throughout the different growth phases (36,49), and the cellular levels of both N and DmpR are independent of (p)ppGpp (Fig. 5A (inset), and Ref. 25). To test the effect of overexpression of N on Po transcription, the E. coli MG1655-derived rpoN gene was cloned under the control of the IPTG-inducible P tac promoter of a plasmid expression vector and introduced in the (p)ppGppdeficient and proficient strains. IPTG induction results in 10 -12-fold higher N levels in both strains as compared with vector control derivatives (Fig. 5A (inset) and data not shown). This high level of overexpression only partially restored transcription to the Po promoter in the (p)ppGpp-deficient strain, increasing transcription from 10 -14% to ϳ30% of the levels observed in the parental (p)ppGpp-proficient strain with native levels of N (Fig. 5A).
The maximal Po transcription level in the parental (p)ppGpp-proficient MG1655⌬lac strain is further enhanced 2-2.5-fold by overexpression of N (Fig. 5A). We also tested the effect of overexpression of N on transcription from Po in strains carrying the three rpoD alleles that all restore transcription in the absence of (p)ppGpp to above that observed in (p)ppGpp-proficient MG1655⌬lac (see Fig. 1B). The data shown in Fig. 5B demonstrate that, for all these three derivatives, overexpression of N results in a 2-2.5-fold enhancement of Po transcription in both (p)ppGpp-deficient and (p)ppGpp-proficient strains. However, the absolute output levels in the parental (p)ppGpp ϩ MG1655⌬lac derivatives are 25-45% lower than in the cognate (p)ppGpp 0 CF1693⌬lac derivatives. Because both the level and the competitive ability of S for available core are dramatically increased in (p)ppGpp-proficient strains (Ref. 22 and references therein), additional competition by S is likely to occur in the (p)ppGpp ϩ MG1655⌬lac derivatives. As shown in Fig. 5C, the presence or absence of S has the anticipated effects on the levels of Po transcription in vivo. As previously observed (44), lack of S in the (p)ppGpp-proficient RH90 strain results in a 2-fold increase in Po transcription, whereas overexpression of N in the absence of S results in a 3.5-fold net increase over the maximal levels of N -dependent Po transcription observed in the parental counterpart (Fig. 5C). Hence, we conclude that N is under significant competition with both S and D in (p)ppGpp ϩ cells.
Modulating D Levels Restores Output from Po in the Absence of (p)ppGpp-Both N and D levels in E. coli are constant throughout the different growth phases (49). The finding that D more readily out-competes N than the converse in vitro prompted us to assess the effect of decreased levels of D on transcription from Po in vivo. To achieve underproduction of D , we utilized a genetic system in which expression of D is under control of the P trp promoter, transcription from which can be regulated by varying levels of the IAA that counteracts the action of the Trp repressor (50). Culturing of both (p)pp-Gpp ϩ and (p)ppGpp 0 strains carrying the P trp -rpoD system in the presence of 0.2 mM IAA results in D levels comparable with those present in the wild-type (p)ppGpp ϩ counterpart (Fig. 6A, compare lanes 1-4). Reduced IAA (0.002 mM) causes a 2-3-fold underproduction of D in both P trp -rpoD strains, as compared with those cultured with 0.2 mM IAA (see Fig. 6A, lanes [5][6][7][8]. These culturing conditions did not alter the levels of the RNA polymerase ␣ subunit (Fig. 6A, lower panels) or ␤ ␤Ј subunits (data not shown).
When rpoD is under control of its native promoter, ϳ2-fold higher D levels are observed in the LB-grown (p)ppGpp 0 strain than in the (p)ppGpp ϩ strain (data not shown, and Ref. 40) and transcription from Po in (p)ppGpp-deficient strains is ϳ10% of that in (p)ppGpp-proficient strains (see Fig. 1). Culturing of P trp -rpoD strains in LB supplemented with 0.2 mM IAA to obtain D levels in the (p)ppGpp 0 CF1693⌬lac P trp -rpoD derivative equivalent to those in the (p)ppGpp ϩ counterpart results in a comparative increase in N -dependent Po transcription to ϳ75% of that in the presence of (p)ppGpp (Fig. 6B). Further 2-3-fold down-regulation of D levels by culturing with 0.002 mM IAA results in an additional increase in Po transcription in both strains (4 -5-fold), with transcription in the (p)ppGpp 0 strain now exceeding that in the presence of native D levels and (p)ppGpp by ϳ3-fold (Fig. 6B). Thus, consistent with a major role of -factor competition on Po transcription, downregulation of D levels can fully restore N -dependent Po transcription in the absence of (p)ppGpp. Down-regulation of D levels to below those in wild type cells also allows transcription from Po during exponential growth where (p)ppGpp levels are low and the Po promoter is normally silent (compare Figs. 6C and 1A). The finding that a 2-3-fold reduction in D causes a ϳ5-fold increase in N transcription (Fig. 6B), whereas a Ͼ10fold increase in N over native levels causes only a ϳ2.5-fold increase in Po output (Fig. 5), mirrors the in vitro finding that increased levels of N are poorer at out-competing D than the converse.
The E. coli Rsd protein has been proposed to act as an anti-D -factor during stationary phase by binding to free D and thereby reducing access to core RNA polymerase (51,52). Given the large effect of reduced levels of D on N -dependent Po transcription, we also determined the effect of increased Rsd levels on Po transcription in vivo. This was achieved by introducing this gene on a high copy number plasmid into the P trp -rpoD strains. As shown in Fig. 6B, Rsd overproduction alone increases Po transcription ϳ3-fold in both strains and is sufficient to restore Po transcription in the absence of (p)ppGpp to above that observed in the presence of (p)ppGpp and native levels of Rsd. Similar -fold increases in Po transcription by the presence of the Rsd plasmid were also observed in (p)ppGppproficient and deficient strains with rpoD in its native context (data not shown). Thus, like underproduction of D , overproduction of Rsd has a major impact on N -dependent Po transcription in vivo. The effects of these manipulations are more than additive, with simultaneous underproduction of D and overproduction of Rsd resulting in a 10 -12-fold increase over the maximum levels of Po transcription in cells with wild-type levels of these proteins (Fig. 6B). DISCUSSION Here we report on the function of (p)ppGpp as a master determinant of the outcome of N competition for limiting core RNA polymerase. This role for (p)ppGpp is based on data using transcription of the N -dependent Po promoter as a functional probe for E N activity in both in vivo and in vitro assays. Four D suppressor mutations that functionally mimic (p)ppGpp share the common property of being defective in their ability to compete with N for limiting core RNA polymerase in vitro (Figs. 2 and 3). These mutations were isolated by different genetic strategies, two (P504L and S506F) on the basis of their ability to restore prototrophy (40) and two (ƒDSA-(536 -538) and Y571H) on the basis of restoration of Po transcription, in (p)ppGpp-deficient E. coli (Fig. 1). Consistent with the common property of defects in -factor competition underlying the mechanism of suppression in both cases, the magnitude of the defect of these four D mutations follows the same hierarchy as their ability to restore Po transcription in (p)ppGpp 0 cells. This idea is further supported by the finding that underproduction of D and/or sequestering of D by Rsd in vivo can restore Po transcription in the absence of (p)ppGpp (Fig. 6). Moreover, underproduction of D allows E N -dependent Po transcription in (p)ppGpp ϩ cells during exponential growth where (p)ppGpp levels are low and the Po promoter is normally silent. Thus, we . Qualitative similar results were obtained when strains were cultured in rich defined minimal media (data not shown). C, immediate Po transcriptional response (open symbols) upon down-regulation of D levels in CF1963⌬lac P trp -rpoD (pVI684, pBR322) (circles) and its (p)ppGpp proficient counterpart (squares) MG1655⌬lac P trp -rpoD (pVI684, pBR322). Growth as measured by A 600 is indicated with closed symbols. conclude that (p)ppGpp per se is not an absolute requirement for N -dependent Po transcription. Rather, elevated synthesis of (p)ppGpp at the onset of stationary phase allows successful competition of N for the available core, resulting in elevated levels of E N sufficient to occupy and initiate transcription from the Po promoter. These findings do not preclude that sufficient E N exists during rapid exponential growth to allow transcription from certain high affinity, easy to saturate N promoters. Thus, akin to the model put forward to explain (p)ppGpp stimulation of specific D -promoters (33), the extent to which (p)ppGpp modulation of ⌭ N levels is manifested at different promoters will depend on their innate ability to recruit limiting ⌭ N .
The data outlined above clearly support a recently proposed model for (p)ppGpp as a master regulator of alternative -factor competition (22), and extend the model to include the structurally and functionally distinct N . Within this model (p)-ppGpp directly modulates interaction of alternative -factors with the increased pool of core RNA polymerase generated by (p)ppGpp-mediated down-regulation of stringent promoters. Facilitation of association of alternative -factor by (p)ppGpp has been postulated to explain both decreases in the levels of D -holoenzyme and increases in the levels of S -and H -holoenzymes in extracts from (p)ppGpp ϩ as compared with (p)p-pGpp 0 cells (22,40). Most directly, addition of ppGpp to an in vitro competition assay has been shown to have a direct stimulatory effect on the outcome of competition between the low affinity H -factor and D for core. It is as yet unclear whether (p)ppGpp has an inhibitory effect on D binding, a stimulatory effect on S and H binding to core, or both (22). We could not document any effect of ppGpp on N competition against D (Fig. 4) under assay conditions that gave the predicted 2-fold effect on H competition (data not shown). These results suggest that detection of direct effects of (p)ppGpp on -factor competition may be limited to low affinity D -like proteins, rendering systems dependent on S and H more sensitive to -factor competition. This idea is supported by the data from the D suppressor mutants (ƒDSA-(536 -538) and Y571H), where the effect of the poor suppressor D -Y571H that barely has detectable effects on N -competition ( Figs. 1 and 3), has markedly greater effects on in vivo and in vitro competition assays involving S and H (22).
The role of (p)ppGpp in -factor competition as described above does not exclude the possibility that transcription of some promoters are directly controlled by (p)ppGpp. Both positive and negative in vitro effects of adding (p)ppGpp to reconstituted transcription systems have been found; however, they are frequently notably lower than when assessed in vivo. More extensive positive stimulatory effects have been observed using coupled in vitro transcription-translation systems (53,54). We have only observed minor direct stimulatory effects of low micromolar concentrations of (p)ppGpp on E N transcription from the Po promoter in vitro (Fig. 4A). More substantial stimulation has been observed in a similar in vitro transcription assay employing the Po promoter and a truncated constitutively active form of XylR (27). We cannot as yet explain the differences in the stimulation levels observed in the two assays; however, it is possible that it is attributable to one of a number of differing properties between the two regulators (13). Nevertheless, the role of (p)ppGpp in determining the level of the pools of alternative holoenzyme RNA polymerases provides an in vivo mechanism to amplify even minor direct effects of (p)ppGpp at specific promoters.
The in vivo impact of -factor competition significantly influences the levels of N -dependent Po transcription, with overexpression of N , or lack of S , causing a ϳ2.5-fold increase, and underproduction and sequestering of D causing a Ͼ10-fold increase over the maximum level of Po transcription achieved in wild-type cells. Given the great impact of -factor competition on Po output, it may appear surprising that only one rpoN suppressor allele was isolated during the genetic selection. This mutant possesses two conservative substitutions (E150D and I165M) that both lie within a subportion of region II (amino acids 120 -215) of N that is intimately associated with core (reviewed in Ref. 8). These mutations only mediate a low level suppressor phenotype (Fig. 1); however, we were unable to overproduce and purify N-E150D/I165M using the D -dependent temperature-sensitive P R P L expression system that has been successfully employed for overexpression and purification of native N . As with two other overexpression systems that are ultimately dependent on D (lacI Q /P tac , and the phage T7 system P tac -T7 RNAP/P T7 ), induction conditions were found to result in rapid growth arrest. Although anecdotal, these findings suggest that N-E150D/I165M possesses an enhanced ability to compete with D , and that more pronounced increases in competitiveness might be lethal. In this respect it is interesting to note that, although both N and D exhibit similar affinities for core when assessed in isolation (Fig. 2), N is poorer at competing D than the converse in both in vivo and in vitro assays. The levels of these proteins within the cell are also disproportionate as compared with the number of genes requiring their activity. For an estimated ϳ30 potential N -dependent promoters, ϳ110 copies/cell of N are available, whereas only 600 -700 copies/cell of D are available for Ͼ1000 actively transcribed D -dependent promoters (36,55). The properties of constant comparatively high levels and high affinity of N , together with (p)ppGpp stimulation of its otherwise poor competitive ability against D , would provide a system for rapid alteration in occupancy and transcription of N -dependent promoters in response to nutritional changes in the environment without de novo synthesis of prerequisite . Thus, for the Po promoter controlling the enzymes for methylphenol metabolism, lack of (p)ppGpp and concomitant high -factor competition fulfill the same function as catabolite repression, namely causing silencing of energetically less favorable specialized catabolic functions until needed.
In the holoenzyme, lies spread out across the upstream face of the enzyme with each of the 2-4 domains and the linkers connecting them making extensive contacts with the core enzyme. The P504L and S506F substitutions lie just within the N-terminal end of the conserved 3.2-linker that joins the ␤1associated 3 and 4 that interacts with the ␤-flap. The 3amino acid insertion mutation ƒDSA-(536 -538) is located just within the first ␣-helix of the 4 -4.1 region, whereas the Y571H substitution is directly adjacent to the 4 -4.2 conserved region (6). Thus, consistent with their defects in competition against N , all these mutants could directly or indirectly decrease the overall affinity of the extensive D -core interaction.
The four mutant rpoD alleles employed in this study provided an important mechanistic tool to dissect the role of -factor competition without specifically affecting the kinetics of E N at N -dependent promoters. The least and most severely affected proteins (Y571H and ƒDSA-(536 -538), respectively) do not restore prototrophy, whereas the intermediately affected proteins (P504L and S506F) do (Table I). These results suggest that restoration of N -Po transcription may provide a broader activity window than prototrophy for isolating suppressor mutations within D . The genetic selection strategy, as anticipated, also identified many mutations in the ␤ (rpoB) and ␤Ј (rpoC) subunits of the transcriptional apparatus that restore transcription to Po in the absence of (p)ppGpp. These mutations include some (e.g. R454H; Fig. 1) that have previously been isolated on the basis of restoration of prototrophy (32). 2 RpoB-R454H and some of the other newly isolated rpoBC suppressors also restore the ability to grow on minimal media, whereas others do not (Table I and data not shown). Similarly, only 7 of 15 rpoBC alleles isolated on the basis of restoration of prototrophy also exhibited the phenotype of restoring activity to the Po promoter in (p)ppGpp-deficient strains (25). Thus, although some mutations in rpoBC mediate both phenotypes, others are specific to one phenotype. In addition to the potential to alter binding of D to core, mutations located within the ␤ and ␤Ј subunits have the potential to directly and differentially affect transcription kinetics at different classes of promoters. Our future analysis of these mutations is aimed at clarifying the degree to which such modulation and effects on -factor competition contribute to their different suppressor phenotypes.