Binding of the Transcription Effector ppGpp to Escherichia coli RNA Polymerase Is Allosteric, Modular, and Occurs Near the N Terminus of the b * -Subunit*

Among the prokaryotae, the nucleotide ppGpp is a second messenger of physiological stress and starva-tion. The target of ppGpp is RNA polymerase, where it putatively binds and alters the enzyme’s activity. Previous data had implicated the b -subunit of Escherichia coli RNA polymerase as containing a single ppGpp binding site. In this study, a photocross-linkable derivative of ppGpp, 6-thioguanosine-3 * ,5 * -(bis)pyrophosphate (6-thio-ppGpp), was used to localize the ppGpp binding site. In in vitro transcription assays, 6-thio-ppGpp inhibited transcription from the argT promoter identically to bona fide ppGpp. The thio group of 6-thio-ppGpp is directly photoactivatable and is thus a zero-length cross-linker. Cross-linking of RNA polymerase was directed primarily to the b * -subunit and could be competed effi-ciently by native ppGpp but not by GTP or GDP. Cyanogen bromide digestion analysis of the cross-linked b * subunit was consistent with an extreme N-terminal cross-link. To assess allosteric consequences of ppGpp binding to RNA polymerase, high level trypsin resistance in the presence and absence of ppGpp was monitored. Trypsin digestion of RNA polymerase bound to ppGpp

Within the bacterial domain of the kingdom prokaryotae exists a general and ubiquitous response to nutritional and environmental stress, the stringent response (1). This general stress response is mediated by high level accumulation of the transcription effector guanosine-3Ј,5Ј-(bis)pyrophosphate (ppGpp). A major effect of elevated ppGpp levels is an immediate and severe reduction of stable rRNA and tRNA gene transcription (2). The cessation of stable RNA syntheses halts the major energy consuming activities of the cell, transcription and translation. This period of metabolic inactivity allows the cell to utilize its remaining energy reserves to adapt to stressful growth conditions through induction of specific "stress genes" (3). Once adaptation is near completion, ppGpp levels decrease and growth resumes. Failure to reduce ppGpp levels results in a severe reduction of cell viability (4).
Several lines of evidence suggest that ppGpp exerts its effects by directly binding to RNA polymerase (RNAP). 1 Certain rifampicin-resistant mutants of the ␤-subunit of RNAP display increased intracellular sensitivity to ppGpp (5,6). Spontaneously occurring mutants that confer survival under artificial and prolonged exposure to toxic levels of ppGpp were mapped to the rpoB gene, encoding the ␤-subunit of RNAP (7). Mutant strains devoid of ppGpp, ppGpp 0 strains, are incapable of surviving nutritional deprivation; however, specific mutants of the 70 -, ␤-, or ␤Ј-subunits of RNAP restores normal survivability to ppGpp 0 strains (8). Fluorescence quenching studies of RNAP in the presence of increasing concentrations of a fluorescently labeled ppGpp analogue (1-aminonapthalene-5-sulfonate-ppGpp) is consistent with binding of ppGpp to a single binding site on RNAP (9). Cross-linking analyses by Chatterji et al. (10) using a radioactive photocross-linkable derivative of ppGpp, 8-azidoguanine-3Ј,5Ј-(bis)pyrophosphate (8-azido-ppGpp), demonstrated predominant cross-linking of ppGpp to the ␤-subunit. In the same study (10), it was also observed that both N-and C-terminal partial trypsin digestion fragments of the ␤-subunit were cross-linked by 8-azido-ppGpp, suggesting a modular ppGpp binding site analogous to that of the nucleotide binding site at the catalytic center of RNAP (11). Despite extensive studies on RNAP-ppGpp interactions, a precise localization of the ppGpp binding site on RNAP is lacking.
An allosteric mechanism of ppGpp action on RNAP is generally invoked as mediating its transcriptional effects, although this has not been extensively studied. In this context, allostery refers to the inducement of functionally relevant conformational changes of RNAP as a result of ppGpp binding at a location other than the catalytic site. Consistent with this notion, ribonucleoside triphosphates do not compete with ppGpp for RNAP binding (12). In addition, circular dichroism studies have revealed a small but significant change in total ␣-helical content of RNAP following addition of ppGpp (13). To date, this single previous study (13) represents the only physical evidence of induced conformational change of RNAP by ppGpp. Clearly, further studies characterizing the nature of ppGpp allostery are warranted.
The aim of the present investigation is 2-fold: 1) to substantiate and refine the location of the ppGpp binding site on E. coli RNAP; and 2), to begin to elucidate the nature of ppGppinduced conformational changes of RNAP. Toward the first of these goals, we synthesized a new photocross-linkable deriva-tive of ppGpp, 6-thioguanosine-3Ј,5Ј-(bis)pyrophosphate (6thio-ppGpp). The 6-thio-ppGpp derivative is a zero-length photocross-linking reagent, because the thiol group is directly photoactivatable and has a van der Waals radius similar to the oxygen for which it is substituted (14). Photoaffinity labeling of RNAP with 6-thio-ppGpp indicates that the N terminus of the largest subunit of RNAP, ␤Ј, is the predominant location of cross-linking. To begin to investigate the conformational consequences of ppGpp-RNAP interactions, we assessed alterations in trypsin sensitivity of RNAP in the presence and absence of ppGpp. Three major ppGpp-dependent trypsin-resistant fragments of RNAP were observed. Identification of these trypsin-resistant fragments and their superimposition onto the three-dimensional structure of RNAP indicates an overlap with and close proximity to the 6-thio-ppGpp crosslinking site. These results have ramifications for both definitive identification of the ppGpp binding site on RNAP as well as for the allosteric consequences of ppGpp-RNAP interaction.

EXPERIMENTAL PROCEDURES
Enzymes and Chemicals--Escherichia coli RNA polymerase was purified from E. coli K12 strain MG1655 using the method of Burgess and Jendrisak (15) with the modifications of Lowe et al. (16). Final 70 holoenzyme separation from core enzyme was accomplished by heparin-Sepharose chromatography (17). Crude RelA-ribosome preparations were prepared according to Cashel (18). Chemicals and solvents were reagent grade, used without further purification, and purchased from Sigma-Aldrich. All radioactive nucleotides were obtained from ICN Pharmaceuticals Inc.
Synthesis of ppGpp and Derivatives-Synthesis conditions for ppGpp were as described (18). For synthesis of radioactive ppGpp, 0.5 mCi of [␥-32 P]ATP (4500 Ci/mmol) were diluted with 1 mM cold ATP and 0.5 mM GDP or 6-thioguanosine-5Ј-triphosphate (6-thio-GTP) in a 0.1 ml reaction volume mixed with 60 A 260 units of crude RelA-ribosomes at room temperature for 12 h. Final purification of ppGpp was on a QAE-Sephadex A-25 column as described (19). Nucleotides were typically eluted with a 0 -0.5 M linear gradient of triethylammonium bicarbonate, pH 7.5. Elution products were visualized by autoradiography after separation by polyethyleneimine-cellulose thin layer chromatography (20). It was noted that 6-thioguanosine derivatives chromatographed with lower mobility on polyethyleneimine-cellulose plates than the corresponding guanosine containing compounds. The ppGpp-containing fractions were pooled and concentrated through removal of the triethylammonium bicarbonate solvent by lyophilization in a vacuum centrifuge at room temperature. The product was resuspended in 50 l of water followed by precipitation with 2% NaI in acetone, the pellet was washed three times with 1 ml of acetone and a final wash with 1 ml of ether. The final product was resuspended in 50 l of water. 6-Thio-ppGpp exhibited an ultraviolet light spectrum typical for ppGpp with a characteristic maximum at 253 nm and a shoulder peak between 265 and 280 nm at pH 7.0 (19).
NMR Analyses of ppGpp and 6-Thio-ppGpp-The proton and phosphorus NMR spectra of purified ppGpp and 6-thio-ppGpp were obtained at 500 MHz on a Varian INOVA 500 (Chemistry Instrumentation Center, State University of New York at Buffalo) in 10% D 2 O pH 4.0, spectra was collected for 27 h. 1 H and 31 P chemical shifts were compared and found to be nearly identical between ppGpp and 6-thio-ppGpp. Assignments of proton shifts were based on previous published spectra for 6-thio-GTP (14).
In Vitro Transcription-RNA synthesis was carried out with a mixture of argT and lacUV5 (10 nM each) promoter-bearing linear templates in 50 mM Tris-HCl, pH 8.0, 100 mM KCl, 10 mM MgCl 2 , 10 mM 2-mercaptoethanol, 250 M GTP, ATP, and CTP, and 20 M UTP with 15 Ci of [␣-32 P]UTP (800 Ci/mmol) in 20-l reaction volumes at 37°C for 10 min. Increasing amounts of GDP, ppGpp, or 6-thio-ppGpp were incubated as indicated with 100 nM RNA polymerase for 15 min at room temperature and then warmed to 37°C for 2 min prior to initiating the reaction by adding an equal volume of prewarmed template/substrate mix. Reactions were terminated and processed according to Hsu (21). RNA was fractionated on a 7 M urea-6% polyacrylamide gel (19:1 acrylamide:bisacrylamide) in Tris-boric acid-EDTA buffer.
Cross-linking-All cross-linking experiments were carried out in 25 mM HEPES, pH 7.9, 5 mM MgCl 2 , 100 mM KCl, 25 g/ml bovine serum albumin, 5% glycerol. E. coli RNA polymerase (1 M) was incubated with radioactively labeled 6-thio-ppGpp in a 20-l reaction volume. Samples were mixed, incubated for 15 min at room temperature, and then transferred to ice and irradiated with 302 nm ultraviolet light (MacroVue UV-20 transilluminator; surface intensity, 9000 W/cm 2 ) for 20 min at a distance of 4 cm in an open Eppendorf tube covered with a polystyrene filter to remove stray radiation of Ͻ290 nm, thus avoiding protein-protein cross-linking. Photoaffinity labeling reactions were terminated by the addition of 5ϫ SDS sample buffer (0.625 M Tris-HCl, pH 6.8, 5% sodium dodecyl sulfate, 5% 2-mercaptoethanol, 0.2% bromphenol blue, 25% glycerol) to a 1ϫ concentration. Samples were heated to 65°C for 5 min and then resolved by electrophoresis through a SDS-10% polyacrylamide gel (30:0.8 acrylamide:bisacrylamide) to separate all RNA polymerase subunits and through a SDS-4% polyacrylamide gel to resolve ␤and ␤Ј-subunits using Tris-glycine buffer.
Mapping of Cross-links-Radioactive 6-thio-ppGpp cross-linked ␤Јsubunit was extracted from a denaturing SDS-4% polyacrylamide gel following electrophoresis. The region of the polyacrylamide gel containing ␤Ј-protein was excised. Protein was eluted by diffusion out of polyacrylamide by overnight shaking of macerated gel slices in 0.03% SDS at 37°C. Eluted protein was concentrated by freeze-drying and then resuspended in water to a final concentration of 1-2% SDS (22). Partial cyanogen bromide (CNBr) degradation of isolated ␤Ј-protein was performed as described previously (23). Cleavage products were fractionated by electrophoresis on a SDS-7-16% gradient polyacrylamide gels in a Tris-glycine buffer system.
Trypsin Digestion Studies-RNA polymerase (0.25 M) was incubated for 15 min at room temperature with 200 M of either GDP or ppGpp in 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 10 mM magnesium acetate, 5% glycerol, 0.1 mM EDTA. After a 2-min warming at 37°C, samples were mixed with increasing amounts of tosyl-L-phenylalanine chloromethylketone-treated trypsin (0, 0.1, 0.3, 0.6, 1.2, and 2.5 M) in a 20-l reaction volume and incubated at 37°C for an additional 5 min. Trypsin digestions were stopped by the addition of phenylmethylsulfonyl fluoride to a final concentration 10 mM, immediately followed by the addition of 5ϫ SDS-sample buffer to a 1ϫ concentration and incubation at 105°C for 2 min. Total trypsin fragments were resolved in a SDS-12% polyacylamide gel and visualized by colloidal Coomassie G-250 staining (24). Specific trypsin fragments were identified by Western blot analyses using monoclonal antibodies against N-terminal, middle, and C-terminal epitopes of ␤Ј (7RC78, 7RC74, and NT73, respectively) and ␤ (7RB145, 7RB135, and NT63, respectively) (Ref. 25  Imaging and Quantification-Radioactive samples were visualized by autoradiography and, when necessary, imaged on a Bio-Rad Molecular Imager and quantified using the Molecular Analyst f software (Bio-Rad Laboratories Inc.).

Matrix-assisted Laser Desorption/Ionization Time-of-flight Mass Spectrometry (MALDI-TOF MS)-Protein fragments were excised from
Coomassie-stained gels and analyzed at Borealis Biosciences Inc., Toronto Canada by MALDI-TOF MS.
The reactivity of 6-thio-GTP was comparable with GTP in the RelA-mediated reaction (data not shown). The conversion of pppGpp to ppGpp is mediated by the enzyme guanosine pentaphosphatase, which is present in crude RelA-ribosome preparations (18). Highly purified 6-thio-ppGpp was separated and ppGpp Binding to the ␤Ј-Subunit of E. coli RNA Polymerase obtained as described by Cashel (18). The structure of purified 6-thio-ppGpp was verified by nuclear magnetic resonance (NMR). Both phosphorus ( 31 P) and proton ( 1 H) spectra of bona fide ppGpp and 6-thio-ppGpp were comparable and indicative of very similar structures (Fig. 1). Unlike the proton NMR spectra reported for 6-thio-GTP (14), 6-thio-ppGpp shows a strong chemical shift specific for a mercapto group proton (Fig. 1). Thus, in solution at pH 4.0, most of 6-thio-ppGpp bears a mercapto group at position 6 of the guanine moiety rather than a thio group as depicted in Fig. 1. The predominance of the mercapto over the thio form of 6-thio-ppGpp may simply be because of the acidic pH under which NMR was performed. This is unclear, however, because the absolute pK a of the thio group of 6-thio-ppGpp is unknown, and sufficient material was not obtained to be able to determine this experimentally. Because the photoreactive conformation of the 6-thioguanosine moiety is the thio form, cross-linking to RNA polymerase is only possible if the bound form of 6-thio-ppGpp is in the thio form and not the mercapto form, as appears to be the case (Fig. 3).
Transcription Activity of 6-Thio-ppGpp-To test the activity of 6-thio-ppGpp compared with ppGpp, their ability to inhibit transcription of the tRNA promoter, P argT , was assayed. The activity of the argT promoter is negatively affected by increased ppGpp levels in vivo (26). Here, we demonstrate for the first time a specific inhibition of P argT in vitro by ppGpp under standard ionic and transcription conditions. At low ionic conditions, Ͻ50 mM KCl, the inhibitory effects of ppGpp were diminished (data not shown). A 4 -6-fold inhibition of RNA synthesis from P argT was observed at the highest concentration of ppGpp tested (Fig. 2). As an internal control, transcription from a P lacUV5 template was monitored in an equimolar mixed template reaction. The transcription from P lacUV5 was also inhibited between 20 -35%, as previously noted (27), but ppGpp preferentially inhibited transcription from P argT . The inhibition of transcription of P argT by 6-thio-ppGpp paralleled that of ppGpp.
Cross-linking of 6-Thio-ppGpp to RNA Polymerase-Increasing amounts of radioactive 6-thio-ppGpp were mixed with 1 M RNA polymerase and subjected to photoactivating ultraviolet light irradiation. Electrophoretic analysis of radioactively labeled RNAP on denaturing polyacrylamide gels revealed the presence of cross-linker on all RNAP subunits. At limiting concentrations of cross-linker, however, labeling appeared predominant for ␤and/or ␤Ј-subunits (Fig. 3A). Further crosslinking experiments were all performed at an equimolar concentration of RNAP and 6-thio-ppGpp (1 M ea). In competition experiments, cross-linking of ␤Јand ␤-subunits was dramatically reduced in the presence of a 200-fold excess of cold genuine ppGpp but not GTP (Fig. 3B). Electrophoresis of photoaffinity-labeled RNAP on low percentage denaturing 4% polyacrylamide gels to resolve ␤Јand ␤-subunits showed that the majority (ϳ90%) of label was associated with the ␤Ј subunit (Fig. 3C). Competition with cold ppGpp had little effect on residual ␤-subunit cross-linking, but it essentially eliminated cross-linking of the ␤Ј-subunits (Fig. 3D). No competition with 6-thio-ppGpp binding was observed in experiments with either GTP or GDP (Fig. 3D). In addition, we observed no effect on cross-linking in the presence of nonspecific or promoter containing DNA, RNA, or rifampicin (data not shown).
Mapping of 6-Thio-ppGpp Cross-linking on the ␤Ј-Subunit of RNAP-The labeled ␤Ј-subunit shown in Fig. 3C was excised and extracted from the 4% polyacrylamide gel and subjected to partial CNBr digestion. Following digestion, the fragments were resolved on a gradient polyacylamide gel and visualized by phosphorimaging. The actual partial CNBr digest is shown in Fig. 4 in comparison with idealized computer-generated profiles of a single-hit partial CNBr digest of ␤Ј. Idealized digestions were generated assuming either extreme C-or Nterminal labeling of ␤Ј, respectively. The profile of the experimental CNBr digestion fragments aligns well with the idealized CNBr N-terminally labeled fragment ladder, particularly the three very specific bands of 102, 130, and 152 amino acid residues (Fig. 4). The pattern and intensity of bands is comparable with what has been obtained previously for N-terminally labeled ␤Ј (28). These results are consistent with 6-thio-ppGpp cross-linking between amino acid residues 29 and 102 of the ␤Ј-subunit of RNAP, because no label was found in association ppGpp-dependent Trypsin-resistant Fragments of RNAP-To begin to explore the allosteric consequences of ppGpp binding to RNAP, resistance to trypsin proteolysis was used to probe for ppGpp stabilized domains of RNAP. Limiting proteolysis has been used as a means of defining subtle alterations in the conformation of RNAP upon promoter binding (29) as well as for the definition of major RNAP structural domains (30). In contrast, we employed in this study an excess of trypsin to assay for protection of small highly structured domains of RNAP formed upon ppGpp binding. A range of trypsin concentration between 0 and 2.5 M was used to digest a solution of 0.25 M RNAP preincubated with 250 M GDP or ppGpp. Trypsin fragments were resolved on denaturing polyacylamide gels. Following Coomassie staining, three prominent ppGppdependent trypsin-resistant fragments were revealed with an apparent molecular mass of ϳ100, 40, and 28 kDa, respectively (Fig. 5A, indicated by asterisks). The same fragments were also obtained in the presence of GDP; however, they had a much lower resistance to trypsin in comparison with ppGpp (Fig. 5A). The binding of ppGpp does not, therefore, appear to induce formation of unique conformers of RNAP. Instead, it seems to stabilize the existing domain structure and/or sterically hinder trypsin accessibility. In control experiments, bovine serum al-bumin trypsin sensitivity was identical in the presence or absence of either GDP or ppGpp (data not shown). Attempts to cleave RNAP cross-linked with radioactive 6-thio-ppGpp did not give a clear or consistent pattern of digestion; we conjecture that this failure was the result of technical problems, e.g. an interference with trypsin activity by the presence of nonspecific UV-induced cross-links. Perhaps in future experiments, the exact identification of the conformational change of RNAP induced following 6-thio-ppGpp cross-linking may be achieved using methods other than trypsin digestion to probe for changes in protein conformation.
The identity of the tryptic fragments stabilized in the presence of ppGpp was determined by Western blot analyses. The trypsinized samples shown in Fig. 5A were processed for Western analyses and probed with a battery of monoclonal antibodies (mAb) against various epitopes of the ␤and ␤Ј-subunits (see "Experimental Procedures"). The ␤Ј-subunit-specific mAb, 7RC78, recognizes an epitope located between amino acid residues 115 and 236. Western blots of trypsin fragments with 7RC78 reveals that the 100-and 28-kDa ppGpp-protected fragments contain the N terminus of the ␤Ј-subunit (Fig. 5B). The ␤-subunit-specific mAb, NT63, recognizes an epitope between amino acid residues 922 and 1099 and is the most C-terminal epitope recognized by available RNAP mAbs (Burgess and colleagues (15,16,25,29,37)). Western blot analysis with NT63 is consistent with the 40kDa fragment containing the C terminus of ␤ (Fig. 5C). Given the fact, however, that the epitope of NT63 is a considerable distance from the actual C terminus of the ␤-subunit, it was difficult to make a definitive C-terminal assignment to this fragment. For this reason, MALDI-TOF MS analysis was performed on the 40-kDa trypsin stable fragment. MALDI-TOF MS analysis (see "Experimental Procedures") indicated that the 40-kDa ␤ fragment spans amino acid residues 958 -1328 Ϯ 10. Thus, the C-terminal assignment of the 40-kDa ␤-subunit band was confirmed. Assignment of the approximate end points of the 28-kDa fragment by MALDI-TOF MS was not successful because of contamination by identical size fragments of the ␣-subunit. Fortunately, the close proximity of the ␤Ј-specific mAb, 7RC78, allows confident N-terminal assignment to the trypsin stable 28-kDa fragment.  ppGpp Binding to the ␤Ј-Subunit of E. coli RNA Polymerase DISCUSSION The global stress regulator, ppGpp, a ligand and effector of RNAP, has been studied since its discovery by Cashel and Gallant (31) over 30 years ago, yet little is known about how it binds to and modifies RNAP. Here, we report the synthesis of a novel photocross-linkable derivative of ppGpp, 6-thioguanosine-3Ј,5Ј-(bis)pyrophosphate. This cross-linkable ppGpp derivative has allowed us to obtain the first indication that ppGpp binds near the N terminus of the large subunit of RNAP, ␤Ј. This compound has the distinct advantage over preceding cross-linkable ppGpp derivatives of carrying a zero-length cross-linking group. The chemical structure of 6-thio-ppGpp in solution, confirmed by NMR spectroscopy, reveals a predominance of the mercapto group (analogous to an enol group in ppGpp) rather than the thio group (analogous to a keto group in ppGpp) at the sixth position of the guanine moiety at pH 4.0 ( Fig. 1). At this same pH level, a mercapto group was not noted or reported for the precursor, 6-thio-GTP (14). Thus, the presence of the 3Ј-pyrophosphate on ppGpp may change the local chemistry of the thio group on the guanine ring. No enol proton was detectable in bona fide ppGpp, indicating that this property is unique for the thiolated ppGpp. It is not clear which is the preferred form of 6-thio-ppGpp at pH 7.9, at which crosslinking and transcription analyses were performed. Equilibrium competition experiments and relative affinity measurements of 6-thio-ppGpp compared with ppGpp could not be performed because of the nonequilibrium conditions necessary for efficient 6-thio-ppGpp cross-linking (see "Experimental Procedures"). If 6-thio-ppGpp at pH 7.9 is predominantly in the mercapto conformation, however, it does not appear to affect dramatically its efficacy in inhibiting transcription (Fig. 2). Fundamental chemistry dictates that the thio and not the mercapto conformation of the 6-thioguanosine moiety is subject to photoactivation. With this in mind, the observed cross-linking of RNAP by 6-thio-ppGpp (Fig. 3) is consistent with a preferred binding to RNAP of the thio form of 6-thio-ppGpp. It is likely, therefore, that in natural ppGpp the keto group at position 6 of the guanine moiety is important for the binding of ppGpp to RNAP.
Predominant cross-linking of radioactively labeled 6-thio-ppGpp occurs within the first 102 amino acid residues of the N terminus of the ␤Ј-subunit of RNAP (Figs. 3 and 4). This location overlaps with the conserved region A of ␤Ј (␤Ј A ). In the recently solved crystal structure model of Thermus aquaticus RNAP, the extreme N terminus of the ␤Ј-subunit, including ␤Ј A , is disordered and lacks electron density (32). The crystal data together with our finding that ppGpp binds to and induces trypsin resistance to the N-terminal portion of ␤Ј (Fig. 5B) leads us to propose that ppGpp binding induces a higher order structure of this region.
The N terminus of ␤Ј is implicated as playing a crucial role in many aspects of the transcription process. Region ␤Ј A contains a zinc finger, which is essential for stable DNA association of the elongating transcription complex (33) and has been crosslinked to the double-stranded DNA at the lagging end of the transcription bubble (34). Additionally, within the transcription complex the extreme N terminus of the ␤Ј-subunit crosslinks along the entire length of the "extruded" nascent RNA (34,35). This close proximity of the extruded RNA to the N terminus of ␤Ј is "replaced" in a paused transcription complex by an alternate proximity of the RNA to the so-called "flap" structure of the ␤-subunit (36). This repositioning or "switching" of the path of RNA within the complex is thought to be part of the mechanism of transcriptional pausing (36). Finally, a primary determinant of 70 binding is found between amino acid residues 260 and 309 of the ␤Ј-subunit (37). These multifaceted functions associated with the N terminus of ␤Ј will be considered below in light of the putative binding of ppGpp near this region and the known effects of ppGpp.
The binding of ppGpp to RNAP has been proposed to destabilize specifically the open complex of rRNA promoters (27,38). Thus, ppGpp binding and restructuring of the region of RNAP near the N terminus could disrupt the function of the ␤Ј A zinc finger and/or 70 interaction and could lead to collapse of the open complex. Another well documented effect of ppGpp is that it decreases transcription elongation rates, primarily by increasing the pause times at naturally occurring pause sites (12, 39 -41). Because the path of nascent RNA, extruded across the N terminus of ␤Ј in elongating RNAP, is altered to putatively bring about transcriptional pausing (36), ppGpp binding might directly influence this process. Thus, the binding of ppGpp to FIG. 5. Trypsin-resistant fragments of RNAP in the presence of ppGpp. Increasing concentrations of trypsin, as indicated, were incubated with 250 nM RNAP for 5 min and fractionated on denaturing 10% polyacylamide gels. A, Coomassie-stained gel of total protein. Known proteins and the position of migration of relative molecular weight markers are indicated. The major trypsin-resistant fragments are indicated by asterisks. B, Western blot of the same total protein samples as shown in A probed with monoclonal antibody 7RC78, which recognizes an epitope between amino acid residues 115 and 236 of the ␤Ј-subunit. C, Western blot of the same total protein samples as shown in A probed with monoclonal antibody NT63, which recognizes an epitope between amino acid residues 922 and 1090 of the ␤ -subunit. this particular region of RNAP may be linked mechanistically to many of the observed effects of ppGpp-RNAP interactions.
Previous data (see the Introduction) have indicated that ppGpp binding is localized to the ␤-subunit of RNAP. However, in this study we have determined that the N terminus of the ␤Ј-subunit is in close proximity to bound ppGpp. Although our results seemingly conflict with these previous observations, upon closer examination they actually corroborate earlier data concerning ppGpp-RNAP binding. We have synthesized and used for the first time a zero-length cross-linking ppGpp derivative, 6-thio-ppGpp. Preceding studies (10) used an azido group cross-linker, which replaces the proton at position 8 of the guanine moiety of ppGpp with an azido group (-N 3 ). The thio group at position 6 of the guanine moiety of 6-thio-ppGpp is 8 -10 Å distant from the azido group of 8-azido-ppGpp. Given the considerable separation of these two cross-linking groups, the thio group of 6-thio-ppGpp predictably would be in a different location and thus provide unique cross-links compared with 8-azido-ppGpp. Our trypsin resistance studies (Fig. 5) are congruent with the formation of a highly structured N-terminal portion of the ␤Ј-subunit and C-terminal portion of the ␤-subunit of RNAP as a consequence of ppGpp binding. These results are consistent with the fact that the N and C termini of ␤Ј and ␤, respectively, are spatially close in the three-dimensional model of prokaryotic RNAP and constitute an intertwined interface of the two subunits (32). These observations lead us to propose that the binding of ppGpp is allosteric and that the site of binding is modular and located close to the intersubunit interface comprising the N-and C-terminal portions of the ␤Јand ␤-subunits, respectively.