Purification and In Vitro Characterization of the Serratia marcescens ...

2 downloads 91730 Views 1MB Size Report
Mar 28, 2002 - novel class of prokaryotic zinc finger transcription factors. All ..... compiled by using Adobe Photoshop and Microsoft PowerPoint. VOL. 185 ...
JOURNAL OF BACTERIOLOGY, Mar. 2003, p. 1808–1816 0021-9193/03/$08.00⫹0 DOI: 10.1128/JB.185.6.1808–1816.2003 Copyright © 2003, American Society for Microbiology. All Rights Reserved.

Vol. 185, No. 6

Purification and In Vitro Characterization of the Serratia marcescens NucC Protein, a Zinc-Binding Transcription Factor Homologous to P2 Ogr Victor McAlister, Chao Zou,† Robert H. Winslow, and Gail E. Christie* Department of Microbiology and Immunology, Virginia Commonwealth University, Richmond, Virginia Received 28 March 2002/Accepted 16 December 2002

NucC is structurally and functionally homologous to a family of prokaryotic zinc finger transcription factors required for late gene expression in P2- and P4-related bacteriophages. Characterization of these proteins in vitro has been hampered by their relative insolubility and tendency to aggregate. We report here the successful purification of soluble, active, wild-type NucC protein. Purified NucC exhibits site-specific binding to a conserved DNA sequence that is located upstream of NucC-dependent Serratia marcescens promoters and the late promoters of P2-related phages. This sequence is sufficient for binding of NucC in vitro. NucC binding to the S. marcescens nuclease promoter PnucA and to the sequence upstream of the P2 late promoter PF is accompanied by DNA bending. NucC protects about 25 nucleotides of the PF upstream region from DNase I digestion, and RNA polymerase protects the promoter region only in the presence of NucC. Template DNA, RNA polymerase holoenzyme, and purified NucC are the only macromolecular components required for transcription from PF in vitro.

The NucC protein was originally identified as a positive regulator of extracellular nuclease and bacteriocin 28b production in Serratia marcescens (7, 12). NucC is a basic 75-aminoacid protein encoded by what appears to be a cryptic prophage and is a member of the bacteriophage P2 Ogr family of transcriptional activators. These proteins, which are required for late gene expression in P2- and P4-related phages, represent a novel class of prokaryotic zinc finger transcription factors. All of the members of this family are functionally interchangeable, at least to some extent, and have significant amino acid sequence similarity. Four invariant Cys residues, arranged in a Cys-X2-Cys-X22-Cys-X4-Cys motif, are essential for function and involved in the coordination of zinc (8, 15, 19, 21, 26). Binding sites for these proteins have been localized upstream of P2 and P4 late promoters and the S. marcescens nuclease promoter (1, 10, 13, 14, 33, 34); these sites include a conserved element of hyphenated dyad symmetry, TGT-N12-ACA. Genetic evidence indicates that members of this family of proteins activate transcription via a specific interaction with the C-terminal domain of the ␣ subunit of RNA polymerase (2, 31, 35). In vitro analysis of the mechanism of transcription activation by members of the P2 Ogr family has been hampered by the relative insolubility of these proteins. The binding sites in P2 and P4 late promoters were defined by using protein mixtures obtained by copurification with a MalE fusion protein (13, 14). The related 186B protein was purified as a Cd2⫹ derivative that retained partial activity (26). We have succeeded in purifying highly active native NucC protein and report here the analysis of the DNA binding specificity and in vitro activity of purified NucC.

MATERIALS AND METHODS Bacterial strains and plasmids. The strains and plasmids used in this study are listed in Table 1. Plasmid construction and DNA manipulations. Standard methods were used for DNA isolation and modification (27). Restriction and modification enzymes were obtained from commercial sources and used as recommended by the suppliers. For in vitro transcription or coupled transcription-translation, supercoiled plasmid DNA was purified by equilibrium centrifugation in CsCl. Plasmids for overexpression from the ␭ PL promoter were constructed by ligation of fragments containing the activator genes into pBB105, which had been digested with EcoRI and EcoRV. The ␾R73 ␦ gene was amplified from pBJ11 by PCR with primers SJK3 (5⬘-GACGAATTCGATGATGCGCTGCCCTT-3⬘) and SJK4 (5⬘ATCTCCTGATTAATGGTTG-3⬘; the underlined bases are the start and stop codons of the gene). The resulting 270-bp fragment was treated with T4 DNA polymerase, cleaved with EcoRI, and ligated with pBB105 to yield pSJK2. A 310-bp fragment containing the nucC gene was amplified from pSE380TacNucC by PCR with Vent DNA polymerase (New England Biolabs) and primers SM1 (5⬘-CGGAATTCTATGATGCACTGTCCACT-3⬘) and SM2 (5⬘-CACGTTGC ATTTGCGAG-3⬘). Following digestion with EcoRI, this fragment was ligated with pBB105 to yield pTG136. To fuse the P4 Psid promoter to cat, a fragment carrying Psid was excised from pSidZ by cleavage with EcoRI and BamHI. The 5⬘ overhang generated by EcoRI digestion was filled in by treatment with the Klenow fragment of DNA polymerase I prior to BamHI digestion. The resulting 703-bp fragment was ligated with the promoterless chloramphenicol acetyltransferase (CAT)-encoding vector pKK232-8, which had been cleaved with BamHI and SmaI to yield pTC12. Plasmid pFWT was constructed from overlapping synthetic oligonucleotides spanning the P2 F promoter that introduced a G at ⫺70, resulting in 5⬘ GGG, and a BamHI site at ⫹40. After annealing, this fragment was ligated with plasmid pMC1403, which had been cleaved with SmaI and BamHI; this fused the first five codons of P2 F in frame with lacZ. Derivatives of this promoter fusion carrying a C-to-A mutation at ⫺51 (pF51A) or a T-to-G mutation at ⫺64 (pF64) were constructed by amplifying this region with the flanking universal primers 1204 and 1212 (New England Biolabs) and a phosphorylated oligonucleotide containing the desired point mutation in the presence of thermostable DNA ligase as described by Michael (22). These fragments were ligated into pMC1403 in place of the wild-type promoter, and each mutation was verified by sequencing. The circular permutation vector pBendF51 was constructed by annealing the synthetic oligonucleotides 51UPB (5⬘-TCGAGGTTGTGCTGTCGATTAGACAA CCGGGA-3⬘) and 51DNB (5⬘-TCGATCCCGGTTGTCTAATCGACAGCAC AACC-3⬘) and ligating them with plasmid pBend2 (18), which had been cleaved with SalI.

* Corresponding author. Mailing address: P.O. Box 980678, Richmond, VA 23298-0678. Phone: (804) 828-9093. Fax: (804) 828-9946. E-mail: [email protected]. † Present address: Department of RARD, Aventis Pharmaceuticals, Bridgewater, NJ 08807. 1808

VOL. 185, 2003

PURIFICATION AND PROPERTIES OF NucC

TABLE 1. Strains and plasmids used in this study Bacterial strain or plasmid

E. coli C strains C-1a C-2322 C-2420 C-2448 Plasmids pBB105 pBend2 pBendF51 pBJ11 pFCAT100 pFWT pF51A pF64G pGC57 pGVB1 pKK232-8 pMC1403 pRS229 pSE380TacNucC pSidZ pSidZT pSJK2 pTC12 pTC13 pTG136 pTG198

Characteristic(s)

Reference or source

Prototrophic C-1a (P2 lg) ⌬(argF-lac)U169 C2420 rna::kan

28 9 13 13

Kanr ori/p15A ␭ PL cloning vector; ␭ cI(ts) Apr ori/ColE1 circular permutation vector pBend2 with 32-bp binding site fragment 1,164-bp ␾R73 NsiI fragment in pBluescriptSKIIN⫹ Apr ori/ColE1 P2 PF-cat Apr ori/ColE1 PF-lacZYA Apr ori/ColE1 PF51A-lacZYA Apr ori/ColE1 PF64G-lacZYA Kanr ori/p15A ␭ PL-P2 ogr; ␭ cI(ts) Kanr ori/p15A ␭ PL-P4 ␦; ␭ cI(ts) Apr ori/ColE1 promoterless cat vector Apr ori/ColE1 promoterless lacZYA PlacUV5-lacZ Apr Ptac-nucC lacIq Apr ori/ColE1 Psid-lacZYA Apr ori/ColE1 (rrnB T1)4 Psid-lacZYA Kanr ori/p15A ␭ PL-␾R73 ␦; ␭ cI(ts) Apr ori/ColE1 P4 Psid-cat Apr ori/ColE1 PF-trpA Kanr ori/p15A ␭ PL-nucC; ␭ cI(ts) Apr Kanr ori/ColE1 PnucA

2 18 This work 14 10 This This This 10 33 4 6 30 12 17 13 This This 20 This 34

work work work

work work work

Purification of NucC protein. Strain C2420 containing pTG136 was grown at 30°C in 850 ml of L broth (LB) plus 60 ␮g of kanamycin per ml and 50 ␮M ZnSO4 to an optical density at 600 nm of 0.7, and expression of NucC was induced by shifting the culture to 43°C for 2 to 3 h. Cells were harvested by centrifugation, washed with 1/10 volume of T5095 buffer (50 mM Tris-HCl, pH 9.5, 25°C), and frozen at ⫺70°C for at least 3 h. The frozen cell pellet was thawed in 20 ml of T5095 buffer at room temperature, lysed by incubation with 3 mg of lysozyme for 2 h on ice, and then pulse sonicated for 3 min on ice at a microtip setting of 3 (W225; Heat Systems Ultrasonics). Cell debris was removed by centrifugation for 20 min at 18,000 rpm (Sorvall SS-34 rotor), and ammonium sulfate was added to the supernatant of the cell lysate to a final concentration of 30% saturation. Crude NucC was collected by centrifugation for 30 min at 18,000 rpm, resuspended in 10 ml of buffer T5095, and applied to a 100-cm3 bed of Sephadex G-50 (Sigma) equilibrated with buffer T5095. The protein-containing flowthrough (36 ml) was applied to a 20-cm3 bed of Q-Sepharose Fast Flow (Pharmacia) equilibrated in buffer T5095. The column was washed with 5 bed volumes of buffer T5095, and NucC protein was eluted with buffer T5095 containing 125 mM NaCl. Purified NucC was dialyzed against 50 mM potassium HEPES (pH 7.5)–0.05 ␮M ZnSO4. NucC activity was detected by expression of ␤-galactosidase from the P4 Psid-lacZ fusion plasmid pSidZT in a coupled transcription-translation system. Protein purity was assessed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis with a Tricine buffer system; proteins were visualized by staining with Coomassie blue. Protein concentration was determined by using the Bio-Rad protein assay kit with bovine serum albumin (BSA) as the standard. Purified protein was stored in aliquots at ⫺80°C. In vitro transcription-translation assay. An S-30 extract was prepared from Lac⫺ Escherichia coli C strain C-2448 and used to monitor the activation of a Psid-lacZ fusion as described previously by Julien and Calendar (13, 14). Each 50-␮l reaction mixture contained 15 ␮l of diluted S-30 extract, 20 ␮l of S-30 reaction mixture adjusted to give a final Mg acetate concentration of 15 mM, 3.2 ␮g of plasmid pSidZT (or lacUV5 plasmid pRS229 as a control), and 1 to 5 ␮l of NucC-containing cell fractions or purified protein. Reaction mixtures were incubated for 1.5 h at 37°C, and then 550 ␮l of prewarmed Z buffer (23) containing 1 mg of o-nitrophenyl-␤-D-galactopyranoside per ml was added and incubation was continued at 28°C until a yellow color developed. Reactions were

1809

stopped by the addition of 500 ␮l of 1 M Na2CO3, and o-nitrophenol was measured by determining absorption at 420 nm. ␤-Galactosidase activity is expressed in nanomoles of o-nitrophenol per minute per milligram of protein. CAT assay. Cultures of C-1a transformed with pTC12 or pFCAT100 and a compatible plasmid containing an activator gene under ␭ PL control (pGC57, pGVB1, or pTG136) were grown at 30°C in LB containing 100 ␮g of ampicillin per ml and 60 ␮g of kanamycin per ml to an A600 of 0.5. Expression of the activator was induced by shifting the culture to 42°C. Cells were harvested 30 min after induction, and CAT activity was measured spectrophotometrically as described previously (10). Gel mobility shift assay. 32P-labeled DNA fragments of 154 bp containing the wild-type P2 F promoter and mutant derivatives were generated by incorporation of [␣-32P]dCTP during PCR amplification of promoter fragments cloned into pMC1403 by using flanking primers 1204 and 1212 (New England Biolabs). Fragments from the S. marcescens nuclease promoter were generated by end labeling of restriction fragments from pTG198 with [␥-32P]ATP and polynucleotide kinase. Labeled DNA was purified on agarose gels and eluted by using a GenElute agarose spin column (Sigma) or by phenol extraction. A small amount of labeled DNA was incubated with purified protein at 33°C for 20 min in a 10-␮l reaction mixture containing 40 mM Tris-HCl (pH 7.9), 1 mM MgCl2, 1 mM dithiothreitol (DTT), 100 ␮g of BSA per ml, and 0.1 mg of salmon sperm DNA per ml. Following the addition of 2.5 ␮l of loading dye (0.5% bromophenol blue, 0.5% xylene cyanol, and 50% glycerol in 2.5⫻ Tris-borate-EDTA [TBE]), protein-DNA complexes were resolved by electrophoresis at 4°C on 6% polyacrylamide gels (19:1 acrylamide/bisacrylamide ratio) in 0.5⫻ TBE. Measurement of DNA bending. Circular permutation plasmid pBendF51 was digested individually with MluI, BglII, NheI, SpeI, EcoRV, StuI, NruI, KpnI, PvuII, and BamHI. Purified NucC was incubated with the digested DNA as described in the previous section, and fragments were resolved by electrophoresis at 4°C on a 10% 75:1 acrylamide-bisacrylamide gel in 0.5⫻ TBE. Binding reactions were optimized to yield ⬃50% DNA binding for measurement of the relative mobilities of the bound DNA fragments. In vitro transcription. Single-round in vitro transcription was carried out essentially by the method of Kajitani and Ishihama (16) as modified by Zou et al. (38). A supercoiled plasmid containing the trp-a terminator and the P2 late promoter PF (pTC13) was used as the template. Template DNA (0.05 pmol) was preincubated with 30 pmol of NucC in 17 ␮l of transcription buffer (50 mM Tris-HCl [pH 7.5, 25°C], 3 mM MgCl2, 0.1 mM EDTA, 0.1 mM DTT, 0.5 mM spermidine, 15 mM NaCl, 100 mM K-Glu) containing 4 U of RNase inhibitor. One picomole of E. coli RNA polymerase (gift of Chuck Turnbough) was then added, and initiation complexes were allowed to form for 30 min at 37° prior to the addition of nucleoside triphosphates (each at 0.15 mM plus 1 ␮Ci of [␣-32P]UTP) and heparin (0.2 mg/ml). Transcription reactions were carried out for 5 min and stopped by placing the tubes on ice and adding 4 ␮l of sample buffer (80% formamide, 0.6% SDS, 48 mM EDTA, 20% glycerol, 0.06% bromophenol blue, 0.06% xylene cyanol). After incubation for 2 min at 95°C, the labeled transcripts were resolved by electrophoresis in 1⫻ TBE on a 6% polyacrylamide gel containing 6 M urea. DNase I footprinting. For DNase I footprinting, a 183-bp DNA fragment containing the wild-type P2 F promoter was amplified by PCR from plasmid pFWT by using Taq DNA polymerase (Invitrogen) and primers 1204M⫹n3 (5⬘-GCGTATCACGAGGACCTTTCGTCT-3⬘) and 1212⫹6n (5⬘-GGTTTTCC CAGTCACGACGTTGTA-3⬘). Primer 1212⫹6n was end labeled with [␥-32P] ATP and polynucleotide kinase so that the fragment was labeled on the template strand. Labeled fragments were gel purified as described above. Binding reactions for DNase footprinting of NucC in the absence of polymerase contained 20 fmol of template DNA in a 20-␮l volume of DNase footprinting buffer (50 mM potassium-HEPES [pH 7.5], 100 mM potassium glutamate, 10 mM MgSO4, 2 mM DTT, 0.05 ␮M ZnSO4, 100 ␮g of BSA per ml, 5% glycerol). The buffer for reaction mixtures containing RNA polymerase contained 150 mM potassium glutamate and 15 mM MgSO4; the other components were the same as those described above. Binding reaction mixtures were incubated for 2 h at 37°C prior to DNase I digestion. Various amounts of NucC and 85 nM RNA polymerase holoenzyme (Epicentre, Madison, Wis.) were added to the reaction mixtures where indicated. Partial DNA digestion was performed by adding 80 ␮l of DNase footprinting buffer containing 0.03 U of DNase I (Promega, Madison, Wis.) per ml and 0.1 mM CaCl2 and incubating the mixture for 20 s at 37°C. Digestion was stopped by addition of 120 ␮l of 60 mM EDTA–350 mM NaCl–1% SDS. Reaction mixtures were extracted with phenol-chloroform, ethanol precipitated, and resolved by electrophoresis on 6.5% polyacrylamide gels containing 7 M urea.

1810

MCALISTER ET AL.

J. BACTERIOL. TABLE 2. Activation of late promoters CAT sp acta

Promoter fused to cat

Activator

None None Ogr Ogr P4 Delta P4 Delta NucC NucC

P2 P4 P2 P4 P2 P4 P2 P4

⬍0.005 ⬍0.005 1.7 1.2 18 0.5 3.8 0.2

PF Psid PF Psid PF Psid PF Psid

a Micromoles of chloramphenicol acetylated per minute at 37°C per milligram of protein in cell extract. Each value represents the average of two determinations from each of at least two independent cultures.

FIG. 1. Lysis of a P2-lysogenic strain following induction of activator expression. E. coli strain C-2322 carrying either nucC plasmid pTG136 (E), ␾R73␦ plasmid pSJK2 (), or ogr plasmid pGC57 (■) was grown at 30°C in LB containing 60 ␮g of kanamycin per ml. Activator synthesis was derepressed by shifting the cultures to 42°C, and the optical density (OD) was monitored with a Klett-Summerson photoelectric colorimeter.

RESULTS AND DISCUSSION NucC is a satellite class activator of the P2/P4 family of transcription factors. The activators encoded by the P2-related phages and their satellite phages have been grouped into two functionally distinct classes that reflect their roles in vivo. Members of the helper class are poor activators of chromosomal (i.e., prophage) promoters, as judged by the failure of plasmid-encoded Ogr or Pag to cause lysis of a P2-lysogenic strain (14). In contrast, the Delta proteins encoded by satellite phages P4 and ␾R73 induce efficient expression of the P2 lysis genes from a prophage (11, 14). The satellite and helper classes of activators also show differences in promoter specificity, as assayed in vivo with reporter plasmids. The helper phage activators are generally less efficient on their own promoters and show better expression from the P4 late promoter Psid (1, 14), while the satellite phage activators work more efficiently at the helper phage late promoters (1, 10, 14). Since NucC activates transcription from chromosomal promoters in S. marcescens, we expected that it would fall into the satellite class of activators. Plasmid-encoded NucC does activate expression of the lysis genes from a P2 prophage, like the other activators in the satellite class. Within 40 min of induction of NucC expression, P2-lysogenic cells have begun to lyse (Fig. 1). NucC also shows promoter specificity characteristic of a satellite activator at P2 and P4 late promoters. Like P4 Delta, plasmid-encoded NucC is more effective than Ogr in activation of cat expression from the P2 late promoter PF in the reporter plasmid pFCAT100. Compared to Ogr, however, NucC is a relatively poor activator of cat expression from Psid (Table 2). By both of the previously defined criteria, therefore, NucC belongs in the satellite phage class of activators, along with the Delta proteins P4 and ␾R73. Overexpression and purification of active NucC protein.

Purification of NucC was carried out under nondenaturing conditions with a high-pH buffer to improve the solubility of the protein (pI of 8.1 predicted by the program Isoelectric in the GCG Wisconsin Package; Accelrys, Madison, Wis.). NucC protein was overexpressed in a Lac ⫺ E. coli C strain, C-2420, from plasmid pTG136, which encodes a ts ␭ repressor and carries the nucC gene under control of the ␭ PL promoter. Like P2 Ogr protein overexpressed from a similar plasmid (21), NucC accumulated in large amounts but was present primarily in the cell pellet at neutral pH. When the cells were lysed in a buffer at pH 9.5, NucC could readily be obtained in a soluble form after sonication. NucC was equally soluble in 50 mM Tris-HCl or 3-[cyclohexylamino]-1-propane sulfonic acid (CAPS) buffer at this pH. For purification, we employed 50 mM TrisHCl (pH 9.5; T5095), which has much less buffering capacity than CAPS, so that it would be easier to lower the pH for subsequent functional analysis. As shown in Table 3, more than 80% of the NucC activity could be recovered in a soluble form after sonication in T5095. This corresponds to only about 50% of the total NucC protein, however, as shown by comparison of lanes 1 and 2 in Fig. 2. This discrepancy could be due to the accumulation of inactive NucC protein during overexpression and/or incomplete solubilization of NucC protein in the crude lysate. Crude NucC (Fig. 2, lane 3) was obtained by precipitation with ammonium sulfate at 30% saturation. The precipitate was resuspended in 5 ml of buffer T5095 and desalted on a Sephadex G-50 column equilibrated with buffer T5095 before application to a Pharmacia Q-Sepharose Fast-Flow column also equilibrated with buffer T5095. NucC protein (Fig. 2, lane 4) was eluted from the Q-Sepharose column with 125 mM NaCl. NucC activity was monitored during purification by induction of expression of a P4 late promoter-lacZ fusion in a coupled in vitro transcription-translation assay system, and the yield of each purification step is presented in Table 3. The increase in the total units of activity obtained after the gel

TABLE 3. NucC purification Fraction

Amt of protein (mg)

Sp act (U/mg)

Fold purification

Crude lysate Cleared lysate 30% ammonium sulfate precipitate Sephadex G-50 filtrate Q-Sepharose 125 mM NaCl eluate

188 143 35 31 1.4

1.3 ⫻ 104 1.4 ⫻ 104 4.5 ⫻ 104 7.3 ⫻ 104 3.8 ⫻ 105

1 1.1 3.5 5.6 29

VOL. 185, 2003

FIG. 2. Purification of NucC. SDS–12% polyacrylamide gel electrophoresis of fractions from NucC purification, stained with Coomassie blue. Fractions are crude lysate (lane 1), cleared lysate (lane 2), 30% ammonium sulfate precipitate (lane 3), and 125 mM NaCl eluate (lane 4) from the Q-Sepharose column. This figure was compiled by using Adobe Photoshop.

filtration step is presumably due to the removal of either ammonium sulfate or another component in the crude lysate that has an inhibitory effect in the S-30 assay. The final yield of pure NucC from 850 ml of culture was 1.4 mg, and the specific activity of the purified protein was 3.8 ⫻ 105 U of NucC per mg. This is similar to the value previously reported for MalEDelta (1.4 ⫻ 105 U/mg; 13) and much higher than that reported for CdB (1.2 ⫻ 103 U/mg; 26). However, it is not possible to compare these values directly because they were assayed in different S-30 extracts. In addition, purified CdB was assayed on a different template and the effect of the Cd substitution on the activity of the protein is unknown. NucC exhibited the expected specificity for activation of transcription in vitro. As shown in Fig. 3, expression from a

FIG. 3. Specificity of NucC activation in a coupled in vitro transcription-translation assay. The DNA template is indicated by the promoter fused to lacZ. Reaction mixtures 1 to 4 contained P4 sid promoter plasmid pSidZT (0.2 pmol), and reaction mixtures 5 to 8 contained lacUV5 promoter plasmid pRS229 (0.1 pmol). Purified NucC (30 pmol) was added as indicated to reaction mixtures 2 to 5. Competitor DNA (1 pmol, reaction mixtures 3 and 7, or 10 pmol, reaction mixtures 4 and 8) was a 154-bp fragment generated by PCR amplification of P2 PF promoter variant 64G 51A.

PURIFICATION AND PROPERTIES OF NucC

1811

FIG. 4. Titration of DNA binding by NucC. The DNA template was a 154-bp PF fragment generated by PCR amplification from pFWT. Various amounts of purified NucC protein, as indicated, were incubated with approximately 1 ng of the labeled probe and resolved by electrophoresis in 0.5⫻ TBE on a 6% polyacrylamide gel (19:1) containing 5% glycerol at 4°C for 2.5 h at 100 V. This figure was compiled by using Adobe Photoshop and Microsoft PowerPoint.

Psid-lacZ fusion was dependent on NucC addition (lanes 1 and 2) while lacZ expression from the activator-independent lacUV5 promoter was unaffected by NucC (lanes 5 and 6). Expression from Psid was reduced by the addition of a DNA fragment carrying the upstream region from a variant of the P2 late promoter PF (lanes 3 and 4). This fragment competed specifically for NucC, since addition of the same amount of DNA had no effect on NucC-independent expression from the lacUV5 promoter (lanes 7 and 8). DNA binding activity of NucC. A gel mobility shift assay (Fig. 4) was used to estimate the DNA binding affinity of the purified NucC protein. The DNA fragment used as a probe carried the P2 late promoter PF. The protein concentration was in great excess over that of the target DNA, which permitted estimation of the apparent dissociation constant for the protein-DNA complex on the basis of the protein concentration required for binding of 50% of the DNA. The midpoint of the binding curve was determined by densitometry of the shifted fragments and resulted in an apparent Kd of 0.5 to 0.6 ␮M for NucC binding. We consider this a minimum estimate of DNA binding affinity because we do not know what fraction of NucC in our purified preparation is in an active form. The affinity of CdB reported by Pountney et al. (26) was almost 10-fold lower, at 4 ␮M, but this could be due to the replacement of Zn with Cd and/or the use of a different DNA target site, as well as differences in the way protein concentration was determined. The activator binding sites in P2 and P4 late promoters contain a conserved TGT-N12-ACA motif centered at about ⫺57 relative to the transcription start site. Analysis of mutants in vivo indicated that bases in this inverted repeat were important for activation of transcription (1, 10, 33). To investigate the role of this sequence motif in NucC binding, we analyzed three promoters for the formation of complexes with NucC by using a gel mobility shift assay (Fig. 5). We compared the wild-type P2 PF promoter with PF51A, a variant with a C-to-A change at ⫺51 to generate the consensus dyad motif, and with PF64G, a variant that alters the upstream half of the dyad to form a perfect inverted repeat with the nonconsensus wild-type PF sequence and therefore has nonconsensus sequences at both the upstream and downstream dyad elements. NucC binding to DNA carrying the 51A mutation was enhanced, and NucC binding to DNA carrying the 64G mutation was severely impaired, confirming the importance of the TGT-ACA inverted repeat in the interaction of NucC protein with DNA

1812

MCALISTER ET AL.

FIG. 5. Comparison of NucC binding to DNA fragments containing point mutations in the P2 PF activator binding site. Labeled DNA fragments of 154 bp were generated by PCR amplification from pFWT, pF51A, or pF64G as indicated. The triangles designate increasing NucC concentrations (0, 0.3, 0.6, 1.8, and 3.0 mM) in each set of five lanes. Complexes were resolved by electrophoresis in 0.5⫻ TBE on a 6% polyacrylamide gel. This figure was compiled by using Adobe Photoshop and Microsoft PowerPoint.

(Fig. 5). A fragment containing both the 64G and 51A mutations (i.e., a mutant upstream dyad and a consensus downstream dyad) bound with an affinity similar to that of wild-type PF (data not shown). The electrophoretic mobility shift experiments described above demonstrated NucC binding to a 154-bp fragment containing P2 PF. We next examined whether a fragment corresponding to just the region upstream of the core promoter is sufficient for NucC binding. Two 34-base complementary oligonucleotides, spanning the region from ⫺72 to ⫺43 of the F51A promoter, were end labeled with 32P and used as probes in a gel mobility shift assay. Each oligonucleotide was incubated with or without NucC protein, before or after annealing with unlabeled complementary oligonucleotide. As can be seen in Fig. 6A, the double-stranded fragment containing just this small upstream region was able to form a complex with NucC (lanes 1 and 3), indicating that this sequence contains all of the critical binding determinants. Furthermore, the boundary of the double-stranded region in the annealed pair of oligonucleotides extends just one nucleotide on either side beyond the region of P2 PF protected from cleavage by DNase I in the presence of the MalE-Delta or PSP3 Pag protein (14). This confirms that NucC has the same binding specificity as the other activators in this family. NucC binding requires that the DNA be double stranded, as evidenced by the lack of complex formation when the oligonucleotides were not annealed (Fig. 6A, lanes 5 and 7). Binding of NucC to fragments containing the region upstream of the S. marcescens nucA promoter was also examined. This promoter, which is one of the natural targets for NucC (12), also contains an upstream sequence element with the

J. BACTERIOL.

conserved TGT-N12-ACA motif. We have previously demonstrated a requirement for this region in NucC-dependent transcription and shown that P4 Delta protein binds to this sequence (34). Two different restriction fragments were used in the DNA binding assay (Fig. 7A). These fragments were similar in size but contained the NucC binding site at different positions within the fragment. The electrophoretic mobility of both fragments was retarded by NucC, confirming the formation of an NucC-DNA complex (Fig. 7B). Furthermore, the slower mobility of the bound fragment in which the NucC binding site is near the center indicated that binding of NucC to PnucA is accompanied by DNA bending. Measurement of DNA bending. To investigate further the apparent NucC-induced DNA bending, we introduced a small fragment carrying the upstream region of the P2 late promoter PF, from ⫺69 to ⫺42, into circular permutation vector pBend2 (18). We used the PF variant with a C-to-A transversion at ⫺51, which creates the consensus TGT-N12-ACA motif. Digestion of the resulting plasmid, pBendF51, with restriction enzymes that cleave in the flanking duplicated sequences results in a collection of fragments identical in size and sequence that carry the binding site at different locations. The electrophoretic mobilities of the different NucC-bound DNA frag-

FIG. 6. (A) Gel mobility shift assay with complementary 34-nucleotide oligonucleotides containing a NucC binding site. In lanes 1, 2, 5, and 6, oligonucleotide 34UP was 5⬘ end labeled; in lanes 3, 4, 7, and 8, oligonucleotide 34DN was 5⬘ end labeled. End-labeled fragments (1 ␮g) were incubated with 1 ␮g of NucC in a 15-␮l binding reaction mixture. In lanes 1 to 4, the two oligonucleotides were annealed before binding. Lanes 5 and 6 contain just the upper strand, while lanes 7 and 8 contain just the lower strand incubated under the same conditions. DS, double stranded; SS, single stranded. (B) Sequences of the two oligonucleotides used in this experiment. The sequence between ⫺72 and ⫺43 corresponds to that of the P2 PF promoter with the ⫺51A mutation to generate a consensus downstream half site. The nucleotides corresponding to the consensus TGT-ACA motif are in bold type. This figure was compiled by using Adobe Photoshop and Microsoft PowerPoint.

VOL. 185, 2003

PURIFICATION AND PROPERTIES OF NucC

1813

the relationship cos(␣/2) ⫽ ␮M/␮E (18). In this formula, ␮M and ␮E are the relative mobilities of DNA fragments with NucC in the middle and the end, respectively. The ␮M and ␮E values were determined from a graph of the relative mobilities of the DNA fragments as a function of the distances of the NucC binding site from the fragment ends (Fig. 8B). The apparent NucC-induced bend angle, calculated by extrapolating these data to the end of the fragment, was 85 to 89°. Sequence-specific binding of DNA by proteins often results in DNA bending, and promoter geometry has been demonstrated repeatedly to be a key element in transcriptional regulation (reviewed in reference 25). One of the consequences

FIG. 7. Binding of NucC to the S. marcescens nuclease promoter. (A) Map of PnucA indicating the restriction sites used to generate the two fragments used in this experiment. The position of the NucC binding site is indicated by the box, and an arrow indicates the transcription start site. Fragment a was generated by cleavage of pTG198 with ClaI and XbaI. Fragment b was generated by cleavage of pTG198 with HindIII and NdeI. The size of each fragment is indicated. (B) Gel mobility shift with end-labeled PnucA fragments. Each fragment was incubated with 0.4 ␮g of NucC, and the complexes were resolved by electrophoresis on 6% polyacrylamide. This figure was compiled by using Adobe Photoshop and Microsoft PowerPoint.

ments (Fig. 8A) were measured and graphed as a function of the distance in base pairs from the center position of the TGT-N12-ACA motif to the left end of each DNA fragment (Fig. 8B). An apparent NucC-induced DNA bend angle was calculated by using the following equation (32, 37): 1 ␮1 [1 ⫺ 2x1(1 ⫺ cos␣) ⫹ 2x12(1 ⫺ cos␣)] ⁄2 ⫽ 1 ␮2 [1 ⫺ 2x (1 ⫺ cos␣) ⫹ 2x 2(1 ⫺ cos␣)] ⁄2

2

2

In this equation, ␮1 and ␮2 denote the relative electrophoretic mobilities of NucC bound to DNA fragments 1 and 2, respectively, and x1 and x2 denote the fractional distances of the NucC binding site from the ends of those respective DNA fragments. ␣ is the calculated DNA bend angle. The electrophoretic mobilities of the 10 NucC-bound pBendF51 fragments were calculated pairwise (45 combinations) and averaged to obtain an apparent NucC-induced DNA bend angle of 89° with a standard deviation of 12°. A similar value for the bending angle was obtained by using

FIG. 8. DNA bending analysis. (A) NucC electrophoretic mobility shift assays using the 153-bp DNA binding site fragments released from pBendF51 by digestion with MluI (a), BglII (b), NheI (c), SpeI (d), EcoRV (e), PvuII (f), StuI (g), NruI (h), KpnI (i), and BamHI (j). (B) Plot of the relative mobility of NucC-DNA complexes as a function of the relative location of the NucC binding site within the 153-bp fragments. This figure was compiled by using Adobe Photoshop and Microsoft PowerPoint.

1814

MCALISTER ET AL.

FIG. 9. Transcription in vitro from the P2 late promoter PF in plasmid pTC13 in the absence (⫺) or presence (⫹) of purified NucC. The supercoiled template carries PF fused to an 82-bp synthetic fragment carrying the trp-a terminator. The position of the expected 150nucleotide transcript is indicated. Also shown is the constitutive 108nucleotide RNA I transcript synthesized from the ColE1-derived pUC9 vector. This figure was compiled by using Adobe Photoshop and Microsoft PowerPoint.

that accompany DNA bending is a phased alteration in the widths of the major and minor groves of the DNA helix. Recent evidence indicates that the interaction of the C-terminal domain of the ␣ subunit of RNA polymerase (␣CTD) with UP element DNA is increased by DNA bending accompanied by narrowing of the minor groove (36). Previous genetic evidence has implicated the ␣CTD in both promoter DNA recognition and direct interaction with P2 Ogr in P2 late gene transcription (35). The involvement of the ␣CTD in protein-protein and protein-promoter DNA contacts has also been demonstrated in the crystal structure of the E. coli catabolite activator protein (CAP)-␣CTD-promoter complex (3). The magnitude of CAPinduced DNA bending is similar to that which we observed for NucC, as measured by gel permutation analyses and calculated via the two methods described above (18, 37) and as determined in the crystal structures of CAP-promoter DNA complexes (3, 29). Whether a constriction in the minor groove of the promoter DNA recognized by the ␣CTD results from NucC-induced bending depends on the exact position(s) and direction of the bend, which cannot be determined from these measurements. In vitro transcription activation by purified NucC. A CsClpurified supercoiled plasmid carrying the P2 late promoter PF and the terminator from the trp operon attenuator was used as the template for in vitro transcription. NucC and the template were preincubated at 37°C for 20 min prior to the addition of RNA polymerase to optimize binding, and initiation complexes were allowed to form for 30 min prior to the addition of nucleoside triphosphates and heparin. A specific labeled transcript of the expected size was obtained only in the presence of NucC, while the constitutive RNA I transcript from the vector plasmid was synthesized in both reaction mixtures (Fig. 9). This is the first demonstration of activation of transcription in a purified in vitro system by a member of the P2 Ogr family of transcription factors. It confirms a direct requirement for this protein in activation of transcription from a P2 late promoter and proves that no additional factors are required.

J. BACTERIOL.

DNase I footprint analysis of NucC and RNA polymerase binding. A 187-bp DNA fragment corresponding to the entire P2 late promoter PF fragment cloned in pFWT was generated by PCR with primers complementary to the flanking vector sequence. This fragment was end labeled with 32P at the 5⬘ terminus of the template strand. NucC protects a region of about 25 nucleotides on the template strand from DNase I digestion (Fig. 10A). The 5⬘ end of the protected region extends to at least ⫺67; lack of DNase I cleavage on this template between ⫺68 and ⫺75 complicates precise assignment of the upstream boundary, but the junction between PF and vector sequences is at ⫺69. The 3⬘ boundary of the protected region is between ⫺44 and ⫺46; nucleotide ⫺46 is clearly protected, and ⫺43 is not, but lack of DNase I cleavage in the absence of protein at ⫺45 and ⫺44 prevents unambiguous assignment. This protected sequence corresponds to same region of PF protected by other members of the P2 Ogr family (13, 14). No apparent protection of the late promoter was obtained in the presence of RNA polymerase alone (Fig. 8B and unpublished results). This is consistent with a large body of in vivo data demonstrating a lack of detectable activity of this promoter in the absence of activator protein (1, 10), as well as with the in vitro data presented above. A specific interaction between RNA polymerase and P2 PF promoter DNA can be seen in the presence of NucC (Fig. 8B). The protected region obtained with the ternary complex encompasses the NucC bind-

FIG. 10. DNase I footprint analysis of NucC and RNA polymerase binding to P2 late promoter PF. The G and G ⫹ A lanes indicate Maxam-Gilbert sequencing reactions, and the bracketed regions indicate the approximate boundaries of the observed NucC and RNA polymerase binding sites. (A) DNase I footprint in the absence (⫺) and presence (⫹) of 1.3 ␮M NucC. (B) DNase I footprint in the presence of 85 nM RNA polymerase holoenzyme in the absence of NucC (⫺) and in the presence of increasing concentrations of NucC (8.7, 43, 87, and 430 nM). This figure was compiled by using Adobe Photoshop.

VOL. 185, 2003

ing site and extends in the 3⬘ direction to about ⫹20, corresponding to the expected footprint on the template strand for RNA polymerase in an open promoter complex (24). The amount of NucC required to obtain protection of the NucC binding site in the presence of RNA polymerase is about threefold less than that required to obtain equivalent protection in the absence of RNA polymerase (data not shown), suggesting that binding is cooperative. NucC activation of transcription from P2 late promoters is similar in a number of respects to activation by CAP at the lac promoter (reviewed in reference 5). In both cases, activation requires only RNA polymerase holoenzyme, promoter DNA, and the activator protein. NucC, like CAP, binds to a single DNA site upstream of the binding site for RNA polymerase (13, 14, 34; this study). Activation at P2 late promoters requires a determinant of the ␣CTD that overlaps the A287 determinant contacted by CAP (5, 35), consistent with a direct interaction between the activator and RNA polymerase equivalent to that made by CAP. A strict requirement for helical phasing (34) in NucC activation is also consistent with such a direct protein-protein interaction. Binding NucC and that by CAP bend DNA to apparently similar extents. We do not know whether the direction of the bend is similar, however, or whether bending plays a critical role in activation at P2 late promoters. The determinants for DNA binding by CAP and NucC are quite different, consistent with the fact that CAP is a helixturn-helix DNA binding protein while NucC more closely resembles eukaryotic zinc finger proteins. In addition, although both activators bind upstream of RNA polymerase, differences in the location of the binding site may affect interaction between the activator and RNA polymerase and/or between RNA polymerase and promoter DNA. The activator binding sites for the P2 and P4 late promoters are centered at about ⫺57, one-half of a helical turn closer to the binding site for RNA polymerase than is the binding site for CAP at the lac promoter. A surface of the ␣CTD that includes residues involved in UP element recognition (the 265 determinant) has been implicated in activation by Ogr (35), as well as CAP. The 265 determinant of the ␣CTD that interacts with CAP also interacts with a 6-bp DNA segment centered 19 bp from the center of the CAP binding site (3); this would be one helical turn upstream of the ⫺35 region in the lac promoter. However, a 6-bp PF sequence centered 19 bp from the center of the NucC binding site would overlap the ⫺35 determinant. A third region of the ␣CTD, the ⫺261 determinant, is proposed to mediate interactions with ␴70 at the lac promoter (3); this determinant is not required for activation at P2 late promoters (35). Interaction between NucC and promoter DNA appears to be required for the formation of an open promoter complex by RNA polymerase. The simplest interpretation of this result is that the mechanism of activation by NucC is solely that of recruitment of RNA polymerase to promoter DNA, as is believed to be the case for CAP at lac. However, on the basis of current evidence, we cannot rule out the possibility that NucC affects the transition from a closed to an open promoter complex. The availability of pure, active NucC will facilitate structure-function analysis of the mechanism of transcription activation, as well as detailed investigation of the determinants for

PURIFICATION AND PROPERTIES OF NucC

1815

DNA recognition by this novel family of prokaryotic zincbinding transcription factors. ACKNOWLEDGMENTS Chao Zou and Victor McAlister each made major and approximately equal contributions to this work. The purification procedure was developed by Chao Zou, who also performed initial characterization studies. Subsequent detailed characterization, bending analysis, and footprinting studies were carried out by Victor McAlister. This work was supported by grant R01GM52052 from the National Institutes of Health, by National Science Foundation grant MCB9982525 (to G.E.C.), and by National Institutes of Health MARC Fellowship F31 GM17594 (to V.M.). We thank Richard Calendar and Bryan Julien for strains and plasmids and for instruction in preparing the S-30 extract. Several past members of the laboratory constructed plasmids used during the course of this study, including Suzanne Kennedy and Te-Chung Lee. We also thank Tina Goodwin for plasmid construction and technical assistance. We gratefully acknowledge a gift of purified E. coli RNA polymerase from Chuck Turnbough and plasmid pBend2 from Sankar Adhya. REFERENCES 1. Anders, D. L. 1993. Ph.D. thesis. Virginia Commonwealth University, Richmond. 2. Ayers, D. J., M. G. Sunshine, E. W. Six, and G. E. Christie. 1994. Mutations affecting two adjacent amino acid residues in the alpha subunit of RNA polymerase block transcriptional activation by the P2 Ogr protein. J. Bacteriol. 176:7430–7438. 3. Benoff, B., H. Yang, C. Lawson, G. Parkinson, J. Liu, E. Blatter, Y. Ebright, H. Berman, and R. H. Ebright. 2002. Structural basis of transcription activation: the CAP-␣ CTD-DNA complex. Science 297:1562–1566. 4. Brosius, J. 1984. Plasmid vectors for the selection of promoters. Gene 27:151–160. 5. Busby, S., and R. H. Ebright. 1999. Transcription activation by catabolite activator protein (CAP). J. Mol. Biol. 293:199–213. 6. Casadaban, M. J., J. Chou, and S. N. Cohen. 1980. In vitro gene fusions that join an enzymatically active ␤-galactosidase segment to amino-terminal fragments of exogenous proteins: Escherichia coli plasmid vectors for the detection and cloning of translational initiation signals. J. Bacteriol. 143:971–980. 7. Ferrer, S., M. B. Viejo, J. F. Guasch, J. Enfedaque, and M. Regue. 1996. Genetic evidence for an activator required for induction of colicin-like bacteriocin 28b production in Serratia marcescens by DNA-damaging agents. J. Bacteriol. 178:951–960. 8. Gebhardt, K. 1994. Ph.D. thesis. University of Oslo, Oslo, Norway. 9. Ghisotti, D., S. Finkel, C. Halling, G. Deho, G. Sironi, and R. Calendar. 1990. The nonessential region of bacteriophage P4: DNA sequence, transcription, gene products, and functions. J. Virol. 64:24–36. 10. Grambow, N., N. K. Birkeland, D. L. Anders, and G. E. Christie. 1990. Deletion analysis of a bacteriophage P2 late promoter. Gene 95:9–15. 11. Halling, C. H. 1989. Ph.D. thesis. University of California, Berkeley. 12. Jin, S., Y. Chen, G. E. Christie, and M. J. Benedik. 1996. Regulation of the Serratia marcescens extracellular nuclease: positive control by a homolog of P2 Ogr encoded by a cryptic prophage. J. Mol. Biol. 256:264–278. 13. Julien, B., and R. Calendar. 1995. Purification and characterization of the bacteriophage P4 ␦ protein. J. Bacteriol. 177:3743–3751. 14. Julien, B., and R. Calendar. 1996. Bacteriophage PSP3 and ␾R73 activator proteins: analysis of promoter specificities. J. Bacteriol. 178:5668–5675. 15. Julien, B., D. Pountney, G. E. Christie, and R. Calendar. 1998. Mutational analysis of a satellite phage activator. Gene 223:129–134. 16. Kajitani, M., and A. Ishihama. 1983. Determination of the promoter strength in the mixed transcription system: promoters of lactose, tryptophan and ribosomal protein L10 operons from E. coli. Nucleic Acids Res. 11:3873– 3888. 17. Keener, J., E. C. Dale, S. Kustu, and R. Calendar. 1988. In vitro transcription from the late promoter of bacteriophage P4. J. Bacteriol. 170:3543–3546. 18. Kim, J., C. Zweib, C. Wu, and S. Adyha. 1989. Bending of DNA by generegulatory proteins: construction and use of a DNA bending vector. Gene 85:15–23. 19. King, R. A. 1993. Ph.D. thesis. Virginia Commonwealth University, Richmond. 20. Lee, T.-C. 1989. Ph.D. thesis. Virginia Commonwealth University, Richmond. 21. Lee, T.-C., and G. E. Christie. 1990. Purification and properties of the bacteriophage P2 ogr gene product: a prokaryotic zinc-binding transcriptional activator. J. Biol. Chem. 265:7472–7477. 22. Michael, S. F. 1994. Mutagenesis by incorporation of a phosphorylated oligo during PCR amplification. BioTechniques 16:410–412.

1816

MCALISTER ET AL.

23. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 24. Ozoline, O. N., and M. A. Tsyganov. 1995. Structure of open promoter complexes with Escherichia coli RNA polymerase as revealed by the DNase I footprinting technique: compilation analysis. Nucleic Acids Res. 23:4533– 4541. 25. Perez-Martin, J., and V. deLorenzo. 1997. Clues and consequences of DNA bending in transcription. Annu. Rev. Microbiol 51:593–628. 26. Pountney, D. L., R. P. Tiwari, and J. B. Egan. 1997. Metal-and DNA-binding properties and mutational analysis of the transcription activating factor, B, of coliphage 186: a prokaryotic C4 zinc finger protein. Protein Sci. 6:892–902. 27. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 28. Sasaki, I., and G. Bertani. 1965. Growth abnormalities in Hfr derivatives of Escherichia coli strain C. J. Gen Microbiol. 40:365–376. 29. Schultz, S. C., G. C. Shields, and T. A. Steitz. 1991. Crystal structure of a CAP-DNA complex: the DNA is bent by 90 degrees. Science 253:1001–1007. 30. Simons, R. W., B. C. Hoopes, W. R. McClure, and N. Kleckner. 1983. Three promoters near the termini of IS10: pIN, pOUT and pIII. Cell 34:673–682. 31. Sunshine, M., and B. Sauer. 1975. A bacterial mutation blocking P2 late gene expression. Proc. Natl. Acad. Sci. USA 72:2270–2774. 32. Thompson, J. F., and A. Landy. 1988. Empirical estimation of protein-

J. BACTERIOL.

33. 34. 35. 36.

37.

38.

induced DNA bending angles: applications to ␭ site-specific recombination complexes. Nucleic Acids Res. 20:9687–9705. Van Bokkelen, G. B., E. C. Dale, C. Halling, and R. Calendar. 1991. Mutational analysis of a bacteriophage P4 late promoter. J. Bacteriol. 173:37–45. Winslow, R. H., B. Julien, R. Calendar, and G. E. Christie. 1998. An upstream sequence element required for NucC-dependent expression of the Serratia marcescens extracellular nuclease. J. Bacteriol. 180:6064–6067. Wood, L. F., N. Y. Tszine, and G. E. Christie. 1997. Activation of P2 late transcription by P2 Ogr requires a discrete contact site on the C-terminus of the ␣ subunit of Escherichia coli RNA polymerase. J. Mol. Biol. 274:1–7. Yasuno, K., T. Yamazaki, Y. Tanaka, T. S. Kodama, A. Matsugami, M. Katahira, A. Ishihama, and Y. Kyogoku. 2001. Interaction of the C-terminal domain of the E. coli RNA polymerase alpha subunit with the UP element: recognizing the backbone structure in the minor groove surface. J. Mol. Biol. 306:213–225. Zhou, Y., X. Zhang, and R. H. Ebright. 1993. Identification of the activation region of catabolite gene activator protein (CAP): isolation and characterization of mutants of CAP specifically defective in transcription activation. Proc. Natl. Acad. Sci. USA 90:6081–6085. Zou, C., N. Fujita, and A. Ishihama. 1994. Asymmetric arrangement of two ␣ subunits within E. coli RNA polymerase: involvement of one ␣ subunit in contact with cAMP receptor protein. J. Mol. Biol. 236:1283–1288.