Identification of Inhibitors of a Bacterial Sigma Factor Using a New High-Throughput Screening Assay

Abstract

Gram-negative bacteria are formidable pathogens because their cell envelope presents an adaptable barrier to environmental and host-mediated challenges. The stress response pathway controlled by the alternative sigma factor σ^E is critical for maintenance of the cell envelope. Because σ^E is required for the virulence or viability of several Gram-negative pathogens, it might be a useful target for antibiotic development. To determine if small molecules can inhibit the σ^E pathway, and to permit high-throughput screening for antibiotic lead compounds, a σ^E activity assay that is compatible with high-throughput screening was developed and validated. The screen employs a biological assay with positive readout. An Escherichia coli strain was engineered to express yellow fluorescent protein (YFP) under negative regulation by the σ^E pathway, such that inhibitors of the pathway increase the production of YFP. To validate the screen, the reporter strain was used to identify σ^E pathway inhibitors from a library of cyclic peptides. Biochemical characterization of one of the inhibitory cyclic peptides showed that it binds σ^E, inhibits RNA polymerase holoenzyme formation, and inhibits σ^E-dependent transcription in vitro. These results demonstrate that alternative sigma factors can be inhibited by small molecules and enable high-throughput screening for inhibitors of the σ^E pathway.

INTRODUCTION

Gram-negative bacteria are remarkably successful pathogens, and the increasing prevalence of antibiotic resistance in these species presents a significant threat to human health (1, 2). A major factor contributing to the success of these bacteria in virulence and in evading antibiotic action is their ability to maintain the integrity of the outer compartment of the cell, the cell envelope, which consists of the outer membrane, the periplasmic space, the peptidoglycan layer, and the cytoplasmic membrane. The σ^E pathway is one of the key regulatory systems used by Gram-negative bacteria to adapt to challenges to the cell envelope encountered in the environment, including those presented by the host during infection (3, 4).

Sigma factors, such as σ^E, are the subunits of RNA polymerase (RNAP) responsible for promoter recognition and transcription initiation (5, 6). The majority of transcription in the cell is directed by the housekeeping sigma factor, σ⁷⁰. However, most bacteria also possess a series of alternative sigma factors that are activated by particular stresses and redirect RNAP to promoters for genes required to respond to the stress in question (6, 7). σ^E is a member of the largest and most widespread group of alternative sigma factors, referred to as the group 4 or ECF (extracytoplasmic function) sigma factors (8, 9). ECF sigma factors have been implicated in stress survival, virulence, and antibiotic resistance in many pathogens (3, 4, 10, 11).

rpoE, the gene encoding σ^E, is essential for viability of Escherichia coli K-12 and Yersinia enterocolitica (12–16). rpoE is also likely required for viability in adherent-invasive E. coli (associated with Crohn's disease), Haemophilus ducreyi, and Bordetella pertussis, because deletion mutants could not be obtained (17–19). Likewise, deletion mutants were obtained only under certain conditions and suppressors arose frequently in Vibrio cholerae (20). In bacterial pathogens that do not require σ^E for viability, mutants lacking σ^E are often attenuated for virulence. These bacteria include Salmonella enterica serovar Typhimurium, E. coli UTI89, Pseudomonas aeruginosa, and Klebsiella pneumoniae (21–24). In addition, V. cholerae ΔrpoE strains were still highly attenuated for virulence despite the appearance of suppressor mutations that allowed growth in culture. Given these phenotypes, the σ^E pathway presents a potential target for new antibacterials.

In Escherichia coli and related bacteria, the major role of the σ^E pathway in cell envelope homeostasis is to control the integrity and composition of the outer membrane by two major mechanisms. First, σ^E transcribes several small RNAs (sRNAs) that act in conjunction with the Hfq protein to silence the gene expression of outer membrane porins and a major cellular lipoprotein (25, 26). Second, σ^E transcribes genes encoding proteins required for folding and delivery of porins to the outer membrane as well as genes required for the export of lipopolysaccharide (LPS) to the outer membrane (27). In this manner, σ^E ensures proper porin production, controls the amount and identity of the porins produced, and ensures proper LPS export to the outer membrane (27, 28).

The regulatory pathway controlling σ^E in E. coli has been studied extensively, and genes encoding the major players in the pathway are found in the genomes of other bacteria that have homologues of σ^E (8). σ^E activity is strongly inhibited by the anti-sigma factor RseA, an inner membrane protein (29, 30). RseA binds σ^E with high affinity and prevents σ^E from binding core RNAP (31). Stresses that disrupt the proper delivery of LPS and outer membrane porins to the outer membrane trigger proteolysis of RseA, freeing σ^E to interact with RNA polymerase and initiate the transcription of genes required to combat the stress (32–34). A low basal level of degradation of RseA provides sufficient free σ^E to maintain the viability of strains of E. coli that require σ^E activity (32, 35, 36).

The bacterial cell envelope is a proven target for antibiotic action. Targeting of the σ^E pathway presents a new approach to simultaneously disrupt several components of this compartment. Drugs that block the σ^E pathway would prevent the ability of the bacterium to ensure envelope integrity and to modulate the cell envelope during infection, resulting in cell death for pathogens in which σ^E is essential for viability or reducing the virulence of pathogens in which σ^E is important for causing disease. Currently, no inhibitors that target any step in the σ^E pathway are available.

To determine if the σ^E pathway can be inhibited by small molecules, an assay compatible with high-throughput screening (HTS) was developed. The assay was used to identify inhibitors from libraries of cyclic peptides generated in E. coli by using SICLOPPS (split-intein circular ligation of proteins and peptides), a genetic system based on spontaneous protein splicing by inteins. SICLOPPS has been used to isolate inhibitors of several bacterial proteins, including the ClpXP protease, Hfq, and the Dam methyltransferase (37–39). One of the inhibitory cyclic peptides inhibited σ^E-dependent transcription by decreasing the affinity of σ^E and core RNAP, demonstrating that this assay is effective and that inhibitors of σ^E can be obtained.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains used in this study are described in Table 1. Mutant alleles ydcQ, lacI^q, and rseA were mobilized in the appropriate strains by using P1 phage transduction, and the antibiotic resistance markers were removed by using FLP recombinase (40, 41). E. coli strains were grown in LB at 30°C with aeration unless otherwise noted. Where appropriate, 100 μg/ml ampicillin, 30 μg/ml kanamycin, 30 μg/ml chloramphenicol, and 0.0002% arabinose were added.

TABLE 1.

Strains and plasmids

Strain or plasmid	Description	Reference
Strains
BW27786	DE(araBAD) DE(rhaBAD) DE(araFGH) ϕ(ΔaraEp PCP13-araE ΔrhaD-rhaB)	63
SEA5036	BL21(DE3) slyD::kan pLysS pPER76	44
SEA6805	BW27786(prpoErybB)(pompC′-yfp)	This study
SEA6809	BW27786(pTrc99a)(pompC′-yfp)	This study
SEA6833	BW27786 ΔydcQ ϕλrpoHP3::lacZ	This study
SAE006	BW27786 lacIq	38
SAE008	SAE006(pRybB)(pompC′-yfp)	38
SAE009	SAE006(pJV300)(pompC′-yfp)	38
SAE018	SEA6805(pSI18)	This study
SAE019	SEA6805(pSI19)	This study
SAE020	SEA6805(pSI24)	This study
SAE021	SEA6805(pSI24Am7Q)	This study
SAE028	SAE008(pSI24)	This study
SAE051	SEA6833(pSRE)(pLC245)	This study
SAE052	SEA6833(pSRE)(pTrc99a)	This study
SAE057	SAE051(pSI24)	This study
SAE173	SEA6809 ΔrseA	This study
SAE174	SAE020 ΔrseA	This study
SAE197	SEA6805(pSI24Am7K)	This study
SAE200	SEA6805(pRI20)	This study
SAE248	SAE197 ΔrseA	This study
Plasmids
pTrc99a	Plasmid containing the Ptrc promoter; Ampr	64
pSB4K5	BioBrick vector; Kanr	65
pEGFP-N2	Plasmid carrying egfp; Kanr	66
pRybB	Same as pFM1-1 (pBRpLac carrying rybB; Ampr)	67
pJV300	sRNA control vector; Ampr	68
prpoErybB	pTrc99a carrying rpoE and rybB; Ampr	This study
pLC245	pTrc99a carrying rpoE; Ampr	27
pompC'-yfp	pSB4K5 with ompC′-yfp; Kanr	38
pSRE	pSB4K5 carrying egfp; Kanr	This study
pARCBD	SICLOPPS system plasmid; Cmr	50
pSI5	pARCBD encoding SGWWDAV*; Cmr	This study
pSI18	pARCBD encoding SGWSER*T; Cmr	This study
pSI19	pARCBD encoding SGWAD*CK; Cmr	This study
pSI24	pARCBD encoding SGWLGS*P; Cmr	This study
pSI24Am7K	pARCBD encoding SGWLGSKP; Cmr	This study
pSI24Am7Q	pARCBD encoding SGWLGSQP; Cmr	This study
pPER76	pET15b carrying rpoE; Ampr	69
prpoEN80C/C165A	pET15b encoding σE N80C/C165A; Ampr	This study

Plasmid constructions.

Plasmids used in this study are described in Table 1, and oligonucleotide sequences are listed in Table 2. To make prpoErybB, the rrnBT1T2 transcription terminator was amplified from pTrc99a by using primers rrnbT1Ba and rrnBT2X. The rybB gene and its promoter were amplified from E. coli genomic DNA by using primers rybBX and rybBSI. Both PCR products were digested with XbaI and ligated by using T4 DNA ligase, and the resulting rrnBT1T2-rybB product was amplified by PCR using primers rrnbT1Ba and rybBSI. The amplified DNA was digested with BamHI and SalI and ligated into pLC245 cut with the same enzymes.

TABLE 2.

Oligonucleotides

Oligonucleotide	Sequence
rrnbT1Ba	TATTAGGATCCTCAGAAGTGAAACGCCGTAGCG
rrnBT2X	TATTATCTAGATCAGGGTTATTGTCTCATGAGC
rybBX	ATAATTCTAGAAAACTGAAGTTGCCCTGAAAATG
rybBSI	ATAATGTTCGACTAAGCCGCTATCGCGCGAGGAG
rybBE	ATAATGAATTCAAACTGAAGTTGCCCTGAAAATG
rybBB	TATTATCATGACTAACCTCCTGACATCAAAGAAAAGCAGTGGCAC
egfpB	TATTATCATGAGCAAGGGCGAGGAGCTGT
egfpP	ATAATCTGCAGTTACTTGTACAGCTCGTCCATG
rpoE_N80C_for	GTATCGGATTGCTGTATGTACAGCGAAAAATTA CC
rpoE_N80C_rev	GGTAATTTTTCGCTGTACATACAGCAATCCGATAC
rpoE_C165A_for	CACCGTACCTACCGGAGCATCCATGATAG GGC
rpoE_C165A_rev	GCCGCTATCATGGATGCTCCGGTAGGTACGGTG
rybB_EcoRI_for	CATGGTATGGCCAGGATTAGG
rybB_XhoI_rev	GAGGGTTGCAGGGTAGTAG
E24K_top	GGGCGATCGCCCACAATTCCGGCTGGTTGGGCTCGAAGCCGTGC
E24K_bottom	TTAAGCACGGCTTCGAGCCCAACCAGCCGGAATTGTGGGCGATCGCCCCAT

For the construction of pSRE, the rybB promoter was amplified from E. coli genomic DNA by using primers rybBE and rybBB. The egfp gene was amplified from pEGFP-N2 by using primers egfpB and egfpP. Both PCR products were digested with BspHI and ligated, and the resulting rybB-egfp product was amplified by PCR using primers rybBE and egfpP, digested with EcoRI and PstI, and ligated into pSB4K5 cut with the same enzymes.

Site-directed mutagenesis to generate the σ^E_N80C/C165A variant was performed by using the QuikChange mutagenesis kit (Stratagene) with primer pair rpoE_N80C_for and rpoE_N80C_rev and primer pair rpoE_C165A_for and rpoE_C165A_rev, according to the manufacturer's instructions.

Chemical synthesis of SI24.

The linear peptide was purchased from the Huck Institute of the Life Sciences Macromolecular Core Facility (Pennsylvania State University, Hershey, PA). Head-to-tail chemical cyclization was initiated by dissolving 100 mg linear peptide, 70 mg 1-ethyl-3-(3′-dimethylaminopropyl)carbodiimide, and 66 mg 1-hydroxy-7-azabenzotriazole in 75 ml dimethylformamide under argon. The reaction mixture was incubated at room temperature for 12 h and evaporated under reduced pressure to near dryness. The resulting residue was triturated in diethyl ether to obtain a crude cyclized product. Purification was performed on a preparative high-performance liquid chromatography system (Waters Corporation) by using a C₁₈ reverse-phase column (Varian Dynamax) with a combination of acetonitrile and water, both containing 0.1% trifluoroacetic acid as the mobile phase. Excess solvent was removed by lyophilization, and the resulting solid was analyzed by electrospray ionization mass spectrometry (positive mode) to confirm its mass.

Screen for inhibitors of the σ^E pathway.

The SGWX₅ SICLOPPS library was constructed as previously described (42). E. coli strain SEA6805 was transformed with the SICLOPPS plasmid library, and the resulting colonies were scraped, grown overnight, diluted 100-fold, and grown to an optical density at 600 nm (OD₆₀₀) of 0.2. Expression of rpoE and rybB was induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to 1 mM for 1 h, ompC′-yfp expression was induced by the addition of anhydrous tetracycline (AHT) to 100 ng/ml for 3 h, and cells were sorted by fluorescence-activated cell sorting (FACS) using a Beckman Coulter Elite cell sorter equipped with Autoclone for the detection of yellow fluorescent protein (YFP) fluorescence intensity. Cells were selected based on YFP fluorescence intensity, and the brightest 0.01% of cells were selected and deposited onto agar plates for clonal growth. Cells from each colony were grown as described above and imaged by epifluorescence microscopy, and the fluorescence intensity was measured by using Simple PCI software (Compix, Inc.). Plasmid DNA was prepared from selected clones and sequenced. Peptide sequences were determined by conceptual translation of the DNA sequences.

Protein sequence analysis.

E. coli strain SAE020 was grown to an OD₆₀₀ of 0.6, and expression of SI24 was induced by the addition of arabinose to 2% for 5 h. Cells were harvested by centrifugation, and the cell pellet was resuspended in 50 ml chitin buffer (20 mM Tris-HCl [pH 7], 500 mM NaCl, 1 mM EDTA, 0.1% Tween 20, 1 mM phenylmethylsulfonyl fluoride [PMSF]). Cells were lysed by sonication, and debris was removed by centrifugation. The clarified lysate was passed over a column equilibrated in chitin buffer and loaded with 1 ml chitin resin (New England BioLabs). The column was washed with 20 volumes of chitin buffer, and 200 μl 3× SDS-PAGE buffer (188 mM Tris-HCl [pH 6.8], 3% SDS, 30% glycerol, 0.01% bromophenol blue, 15% β-mercaptoethanol) was added to the resin. The mixture was boiled, the protein was resolved on a 12% SDS-polyacrylamide gel, and bands corresponding to SI24 were excised. The sequence was determined at the Proteomics and Mass Spectrometry Core Facility at Pennsylvania State University, University Park, PA. A Thermo LTQ Orbitrap Velos mass spectrometer with a Dionex Ultimate 3000 Nano-LC system was used to analyze chymotryptic peptides. The data were processed by using Proteome Discoverer 1.3 (Thermo Scientific).

Fluorescent reporter and cell staining assays.

E. coli strains were grown with or without the addition of arabinose and examined under inducing and noninducing conditions. For induced cultures, IPTG was added at an OD₆₀₀ of 0.2 to a final concentration of 1 mM, the cultures were grown for 1 h, AHT was added to 100 ng/ml, and cultures were grown for an additional 3 h. Cells were examined by using epifluorescence microscopy. For propidium iodide staining, cultures were grown and induced as described above, and propidium iodide was added to 3 μg/ml. Cells were incubated with shaking for 30 min, harvested by centrifugation, resuspended in 50 μl LB, and examined by using epifluorescence microscopy. All experiments were done in triplicate.

For assays in microtiter plates, cultures of an E. coli prpoErybB pompC′-yfp strain and an E. coli pTrc99a pompC′-yfp strain were grown in flasks, and IPTG was added at an OD₆₀₀ of 0.2 to a final concentration of 1 mM. Following the addition of IPTG, 50-μl aliquots were transferred into wells of 384-well trays. Trays were incubated at 37°C for 45 min with shaking at 300 rpm, AHT was added, and trays were incubated at 37°C for an additional 3 h. Fluorescence was measured with excitation at 500 nm and emission at 540 nm, and the fluorescence intensity was normalized to the OD₆₀₀ of the well. Z′ values were calculated for each day, as described previously (43), and Z′ values for 3 different days were averaged. The E. coli pTrc99a pompC′-yfp strain was used as the positive control.

Western blotting.

E. coli strains were grown as described above for protein sequence analysis, and equivalent OD₆₀₀ units were harvested. Cells were lysed in SDS-PAGE buffer, separated on 15% SDS-polyacrylamide gels, and transferred onto a Hybond-P polyvinylidene difluoride (PVDF) membrane (GE Healthcare). SICLOPPS proteins were detected by using rabbit polyclonal antibodies raised against the chitin-binding domain (CBD) (S6654S; New England BioLabs).

Protein purification and dye labeling.

N-terminally His-tagged wild-type σ^E and σ^E_N80C/C165A proteins were purified as previously described and stored in storage buffer (20 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM dithiothreitol [DTT], 15% glycerol) (44). Proteins were quantified by measuring the absorbance at 280 nm (ε = 14,770 M⁻¹ cm⁻¹). DTT levels were maintained at 1 mM throughout storage. Core RNAP and σ⁷⁰ were gifts from Katsuhiko Murakami (Pennsylvania State University) and purified as described previously (45).

Immediately before labeling of N-terminally His-tagged wild-type σ^E and σ^E_N80C/C165A, the storage buffer was removed by chromatography with a Bio-Gel P4 desalting column (Bio-Rad) equilibrated with a solution containing 50 mM Tris (pH 8.0), 500 mM NaCl, 0.5 mM EDTA, and 10% glycerol. Proteins were then incubated on ice with 100 μM BODIPY-FL maleimide (Invitrogen/Molecular Probes) diluted from a 2.5 mM BODIPY-FL stock in acetone. Unreacted dye was removed by using a Bio-Gel P4 desalting column equilibrated in a solution containing 50 mM Tris (pH 8.0), 500 mM NaCl, 0.5 mM EDTA, and 10% glycerol. The degree of dye labeling was determined by measuring the absorbance of the labeled protein at 505 nm (ε₅₀₅ = 79,000 M⁻¹ cm⁻¹; ε₂₈₀ = 1,300 M⁻¹ cm⁻¹) and was typically >50%.

Fluorescence anisotropy.

Anisotropy assays were performed with Greiner black 384-well microplates by using a microplate reader (Infinite M1000; Tecan) at 30°C, with excitation at 470 nm and emission at 514 nm. The G factor was determined to be 1.13 by using 1 nM fluorescein in 0.01 M NaOH. Binding reaction mixtures (20 μl) containing 100 nM BODIPY-σ^E_N80C/C165A and 0 to 1 mM SI24 or 5 nM BODIPY-σ^E_N80C/C165A and 0 to 1 μM RNAP in transcription buffer (50 mM Tris-HCl [pH 8.0], 10 mM MgCl₂, 50 mM NaCl, 5% glycerol, 10 mM 2-mercaptoethanol, 0.01% NP-40) were incubated at 30°C for 30 min, and anisotropy measurements were performed. The fractional occupancies were calculated as

$$\theta =(P{-P}_0)/(P_{\max \hspace{0.17em}}{-P}_0)$$ (1)

where P₀ and P are polarization values before and after the addition of ligands, respectively, and P_max is the polarization value at saturation. Binding constants were obtained by fitting of the results to the equilibrium binding equation for a bimolecular association, where E, S, and ES are the concentrations of BODIPY-σ^E_N80C/C165A, the ligand (RNAP or SI24), and the bound complex, respectively, and K_D is the equilibrium dissociation constant (46):

$$\left[ES\right]=\frac{{\left[E\right]}_0+{\left[S\right]}_0+K_D-{\left\{{\left({\left[E\right]}_0+{\left[S\right]}_0+K_D\right)}^2-4\cdot {\left[E\right]}_0\cdot {\left[S\right]}_0\right\}}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}{2}$$ (2)

Competition assays were used to determine the affinity of unlabeled σ^E and σ⁷⁰ for core RNAP and to measure the K_I for inhibition of holoenzyme formation by SI24. In these assays, reaction mixtures (30 μl) contained 5 nM BODIPY-σ^E_N80C/C165A, 75 nM core RNAP, and either 0 to 4 mM SI24 or 0 to 1 μM unlabeled sigma factor. BODIPY-σ^E_N80C/C165A was incubated with competitors for 20 min at 30°C, core RNAP was added, the mixture was incubated for 30 min, and fluorescence anisotropy was then measured. The 50% inhibitory concentrations (IC₅₀s) were calculated by using the Hill equation (47), and equilibrium dissociation constants (K_I) were calculated from the IC₅₀ as follows, where K_D refers to the affinity of BODIPY-σ^E_N80C/C165A and RNAP (48):

$$K_I={\text{IC}}_{50}/(1+[\text{RNAP}]/K_D)$$ (3)

In vitro transcription.

For multiple-round transcription reactions, σ^E holoenzyme was formed by incubating 0.25 μM core RNAP with 0.25 μM σ^E for 30 min at 30°C in transcription buffer. Transcription reactions were initiated by adding the σ^E holoenzyme to the transcription template along with nucleotides and [α-³²P]UTP. The transcription reaction mixtures (10-μl final volume) contained 25 nM E. coli RNAP core enzyme, 25 nM σ^E, 10 nM transcription template containing positions −70 to +100 of the rybB gene, 200 μM ATP, 200 μM CTP, 200 μM GTP, and 20 μM [α-³²P]UTP in transcription buffer. Transcription was allowed to proceed for 10 min at 30°C and terminated by the addition of Tris-borate-EDTA (TBE) buffer with 8 M urea, 10 mM EDTA, 0.04% bromophenol blue, and 0.04% xylene cyanol. Samples were heated for 2 min at 95°C, cooled for 5 min on ice, and applied onto a 6% polyacrylamide gel (19:1 acrylamide-bisacrylamide) with 7 M urea. Bands were visualized on a Typhoon 9410 instrument (GE Healthcare) and quantified by using ImageQuant (Molecular Dynamics). To assess the effects of SI24, the cyclic peptide was incubated with σ^E for 30 min at 30°C before the addition of core RNAP or was added after the addition of core RNAP. The 215-bp transcription template DNA fragment was generated by PCR of the rybB promoter from genomic DNA using primers RybB_EcoRI_for and RybB_XhoI_rev.

For single-round transcription experiments, 0.25 μM σ^E was preincubated with 0.25 μM core RNAP for 30 min at 30°C to form the holoenzyme. Template DNA containing the rybB promoter was added to the σ^E holoenzyme, and reaction mixes were incubated for 30 min at 30°C to allow for open-complex formation. A single round of transcription was initiated by the addition of nucleoside triphosphates (NTPs) and heparin. The transcription reaction mixtures (20-μl final volume) contained 25 nM core RNAP, 25 nM σ^E, 10 nM transcription template, 200 μM ATP, 200 μM CTP, 200 μM GTP, 20 μM [α-³²P]UTP, and 25 μg/ml heparin in transcription buffer. Following a 10-min incubation at 30°C, reactions were terminated, and transcripts were separated by polyacrylamide gel electrophoresis and visualized as described above for the multiple-round transcription assays. The effects of SI24 were measured by incubating the cyclic peptide with σ^E for 30 min at 30°C before holoenzyme formation or after holoenzyme formation but before the addition of the DNA template to form open complexes. In all cases, IC₅₀s were calculated by using the Hill equation (47), and K_I values were determined by using equation 3 (48). The effect of SI24 on transcription by σ⁷⁰-RNAP was measured by using the T7A1 promoter as the transcription template (49) (gift from Katsuhiko Murakami) in single-round assays, as described above for σ^E.

RESULTS

Screen for inhibitors of the σ^E pathway.

To identify inhibitors of the σ^E pathway, a cell-based assay with positive readout for inhibition was developed. In this assay, σ^E transcribes the RybB sRNA, which is a member of its regulon. A gene fusion between the region of ompC that is targeted by RybB and yfp allows σ^E pathway activity to be monitored by YFP fluorescence (Fig. 1A). In the absence of an inhibitor, σ^E transcribes rybB, and RybB-Hfq represses the production of OmpC′-YFP, resulting in dark cells. Conversely, if an inhibitor of the σ^E pathway is present, OmpC′-YFP is produced, and the cells are fluorescent. To ensure that sufficient quantities of each component are present to give an accurate readout of pathway activity, the rpoE and rybB genes are expressed from a plasmid (prpoErypB). On this plasmid, rpoE expression is controlled by an IPTG-inducible promoter, and rybB expression is controlled by its native σ^E-dependent promoter. ompC′-yfp is expressed from a tetracycline-regulated promoter on a separate plasmid (pompC′-yfp).

FIG 1.

Screen for cyclic peptide inhibitors of the σ^E pathway. (A) Schematic representation of the σ^E pathway reporter. (Left) When the pathway is functional, σ^E directs transcription of RybB, which binds an Hfq hexamer. Hfq-RybB represses the translation of ompC′-yfp mRNA and targets it for degradation, preventing the production of OmpC′-YFP. (Right) Inhibition of a step in the pathway, such as σ^E-dependent transcription or RybB-Hfq activity, derepresses translation, resulting in OmpC′-YFP production and fluorescent cells. (B) pSI24 increases fluorescence in the E. coli prpoErybB pompC′-yfp strain. With no inhibitor, RybB represses ompC′-yfp expression, and E. coli prpoErybB pompC′-yfp cells have low YFP fluorescence intensity (prpoErybB). When SI24 is produced in the E. coli prpoErybB pompC′-yfp strain (prpoErybB+pSI24), the fluorescence intensity is similar to that observed for the positive-control strain (No prpoErybB).

The maximum fluorescence that can be produced from the reporter under screening conditions was determined by constructing an isogenic control strain that contains pompC′-yfp and the pTrc99a vector lacking rpoE and rybB. This strain mimics strong inhibition of the σ^E pathway. OmpC′-YFP fluorescence intensity in the control strain was 4-fold higher than that in the strain with prpoErybB (Fig. 1B). The difference in fluorescence intensity between the two strains indicates that the assay has the sensitivity to detect inhibitors that reduce the activity of the σ^E pathway.

Inhibitors of the σ^E pathway were identified from the SGWX₅ library of cyclic peptides produced by using SICLOPPS (50). In this library, five codons in the cyclic peptide gene (corresponding to the X amino acids) are randomized at the DNA level, and the SGW sequence is constant to ensure efficient cyclization. Cyclic peptide production was induced in the E. coli prpoErybB pompC′-yfp strain, and FACS was used to isolate cells with the highest fluorescence intensity from a population of 2 × 10⁶ cells. Cultures of the selected clones were grown, and OmpC′-YFP production was measured in the presence and absence of cyclic peptide expression by using epifluorescence microscopy. To ensure that the increased fluorescence was caused by the SICLOPPS plasmid and not mutations in the reporter strain, plasmids encoding putative inhibitors were purified and retransformed into the E. coli strain carrying prpoErybB and pompC′-yfp, and the fluorescence assay was repeated. Four plasmids caused >2-fold-higher fluorescence than that of the negative control (E. coli prpoErybB pompC′-yfp strain with no cyclic peptide). The strongest inhibition was observed for cells carrying pSI24. The fluorescence intensities of cells with pSI24 showed a bimodal distribution. Ten percent of cells had a fluorescence intensity within 2 standard deviations of the average for the negative control and were scored “dark.” The remaining 90% of cells with pSI24 had a fluorescence intensity 300% ± 13% higher than that of the negative control. Based on these data, pSI24 was chosen for further characterization (Fig. 1B).

Lysine is inserted at the TAG codon of pSI24.

Sequencing of the four plasmids with >2-fold-increased fluorescence, including SI24, showed that each one contained a TAG codon in the randomized region of the sequence (Table 3). The presence of a stop codon suggested that either the inhibition of the σ^E pathway was the result of an uncircularized, truncated SICLOPPS protein or the read-through of the stop codon resulted in a complete SICLOPPS protein that could produce an inhibitory cyclic peptide. The SICLOPPS plasmids encode a chitin-binding domain (CBD) at the C terminus of the full-length SICLOPPS protein. The CBD is removed during the first step of cyclization, generating the I_N-CBD fragment (42, 50). Because the full-length SICLOPPS protein and I_N-CBD have different sizes (22 kDa and 19 kDa, respectively), Western blotting to detect the CBD can be used to measure the expression of the full-length protein and determine the ability of the protein to initiate cyclization. Western blotting of extracts from a strain with pSI24 revealed that the full-length protein and the I_N-CBD were produced albeit at low levels (Fig. 2A). This result suggested that read-through did occur at the stop codon. The 22-kDa protein was purified by using chitin affinity chromatography and identified by liquid chromatography-tandem mass spectrometry (LC-MS/MS) (Fig. 2). The mass spectrometry results confirmed that the full-length SICLOPPS protein was present and revealed that a lysine residue was inserted at the TAG codon in the randomized-sequence region. These data indicated that a cyclic peptide of the sequence SGWLGSKP was produced at low levels in cells with pSI24.

TABLE 3.

Sequences of cyclic peptide inhibitors

Plasmid	Sequence of inhibitora	% fluorescent cells
pSI24	SGWLGS*P	90
pSI5	SGWWDAV*	80
pSI18	SGWSER*T	82
pSI19	SGWAD*CK	85

^aAmino acid sequence of the randomized region determined by conceptual translation of the plasmid DNA sequence. An asterisk indicates a stop codon.

FIG 2.

SI24 contains lysine at position 7. (A) Schematic diagram of the SICLOPPS reaction. In the first step, the I_N-CBD is spliced out to give a lariat intermediate, and in the second step, the cyclic peptide is released. (B) Representative Western blot using an anti-CBD antibody showing the relative amounts of full-length protein and I_N-CBD produced from E. coli prpoErybB pompC′-yfp cells with no SICLOPPS plasmid, with expression induced from pSI24, and with expression induced from an unrelated SICLOPPS plasmid. (C) Purified full-length SICLOPPS protein from pSI24 resolved by SDS-PAGE. The lysate from cells expressing an unrelated SICLOPPS protein is shown for comparison. (D) MS/MS spectrum from one chymotryptic peptide of the SI24 protein from panel C. The masses of the parent ion (495.9) and fragmentation ions are indicated in black, with b⁺¹ and y⁺¹ assignments shown in gray.

Expression of cyclic SGWLGSKP is toxic.

To determine if cyclic SGWLGSKP was responsible for σ^E pathway inhibition in cells containing pSI24, the stop codon in the randomized region of pSI24 was changed to a lysine codon to generate pSI24Am7K. When cyclic peptide expression was induced from pSI24Am7K in the E. coli strain carrying prpoErybB and pompC′-yfp, the lag phase following dilution of the culture grown overnight was extended, and even after growth resumed, the cultures grew more slowly than isogenic strains with pSI24 or no SICLOPPS plasmid (Fig. 3A). Epifluorescence microscopy of samples taken after 7 to 8 h of culture growth showed that cultures with pSI24Am7K contained a small percentage of fluorescent cells of the typical size of E. coli cells, but the majority of cells formed long filaments (Fig. 3B). To assess whether the expression of cyclic peptide from pSI24Am7K affected viability, cells were stained by propidium iodide, a membrane-impermeable dye that is excluded from viable cells. Propidium iodide was retained in most cells with pSI24Am7K but not in cells expressing an unrelated cyclic peptide, RI20, that does not cause filamentation or alter growth (Fig. 3C). These data suggest that expression from pSI24Am7K is toxic and that the toxicity is not due to the SICLOPPS reaction or to general properties of cyclic peptides. The toxicity produced by expression from pSI24Am7K is consistent with the cyclic peptide inhibiting the σ^E pathway, because σ^E activity is required for the viability of E. coli K-12 strains.

FIG 3.

Expression from pSI24Am7K is toxic. (A) Representative growth curves of the E. coli prpoErybB pompC′-yfp pSI24 strain with (open circles) and without (filled circles) induction and the E. coli prpoErybB pompC′-yfp pSI24Am7K strain with (open triangles) and without (filled triangles) induction. (B) Fluorescence micrographs of the E. coli prpoErybB pompC′-yfp and E. coli pompC′-yfp strains, as described in the legend of Fig. 1B. Expression from pSI24Am7K or pSI24Am7Q in the E. coli prpoErybB pompC′-yfp strain results in fluorescence and filamentous growth. (C) Propidium iodide staining of E. coli prpoErybB pompC′-yfp cells with pSI24Am7K or a plasmid encoding an unrelated cyclic peptide, pRI20, with and without induction. Red indicates the permeability of cells to propidium iodide. Samples were taken after 6 to 7 h of growth in panels B and C.

The toxic effect of expression from pSI24Am7K suggested that pSI24 passed the screen because the amber codon allowed the low-level expression of a highly active cyclic peptide. To determine if the lysine inserted at the amber codon is required for activity, a variant of pSI24 with a glutamine codon at position 7 (pSI24Am7Q) was tested. Expression from pSI24Am7Q in the E. coli prpoErybB pompC′-yfp strain resulted in fluorescent cells and caused slow growth and filamentation, which was similar to the phenotype observed for cells with pSI24Am7K (Fig. 3C). These results indicate that lysine at position 7 is not required for SI24 activity and suggest that pSI24 with an amber codon at position 7 was identified in the screen because cyclic SGWLGSKP inhibits the σ^E pathway but is toxic when expressed at typical levels.

SI24 inhibits σ^E-dependent transcription.

Because repression of the prpoErybB pompC′-yfp reporter requires both σ^E-dependent transcription of rybB and Hfq-RybB-mediated repression of ompC′-yfp, SI24 could inhibit either step of the σ^E pathway. To test if SI24 inhibits RybB activity, plasmid prpoErybB was replaced with prybB, which carries the rybB gene under the control a σ⁷⁰-dependent IPTG-inducible promoter. In the E. coli prybB pompC′-yfp strain, rybB expression and ompC′-yfp repression are not dependent on σ^E. When SI24 was expressed in the E. coli prybB pompC′-yfp strain, there was not an increase in fluorescence: 75% of cells showed no increase in fluorescence intensity, and the remaining 25% were ≤1.2-fold brighter (Fig. 4A). These data suggested that SI24 does not inhibit Hfq-RybB activity.

FIG 4.

SI24 inhibits σ^E-dependent transcription. (A) SI24 does not inhibit Hfq-RybB activity. With no inhibitor, RybB represses ompC′-yfp expression, and E. coli prybB pompC′-yfp cells have low YFP fluorescence intensity (prybB). When SI24 is produced (prybB+pSI24), the fluorescence intensity is similar to that in a strain with no inhibitor. Cells are fluorescent when RybB is not present (No prybB). (B) SI24 inhibits production of GFP from a σ^E-dependent promoter in the E. coli prpoE pSRE strain. With no inhibitor, rpoE expression causes GFP production and fluorescent cells (prpoE). When SI24 is produced (prpoE+pSI24), fluorescence intensity is decreased. Cells are not fluorescent in a strain with no RpoE expression (No prpoE).

To determine if SI24 inhibits σ^E-dependent transcription, egfp was placed under the control of the σ^E-dependent P_rybB promoter to make the pSRE reporter. Because basal levels of σ^E activity are low in unstressed E. coli cells, a plasmid with rpoE under the control of an IPTG-inducible promoter was transformed into the E. coli pSRE strain to increase the level of green fluorescent protein (GFP) production from P_rybB. Overproduction of σ^E in this strain resulted in a 4-fold increase in the average GFP fluorescence intensity. When SI24 expression and overproduction of σ^E were induced in the same cells, the average GFP fluorescence intensity decreased by 64% (Fig. 4B), indicating that SI24 inhibits σ^E-dependent transcription.

SI24 inhibits transcription in vitro.

The in vivo data suggest that SI24 inhibits σ^E-dependent transcription, either directly or by altering some aspect of cellular physiology. To determine if SI24 acts directly on σ^E-dependent transcription, cyclic SGWLGSKPM was chemically synthesized and added to multiple-round transcription reactions in vitro. In this assay, RNAP goes through several rounds of holoenzyme formation, transcription initiation, elongation, and termination, recapitulating the fundamental steps of transcription that occur in the cell. When SI24 was incubated with σ^E before RNAP was added to form the holoenzyme, SI24 inhibited transcription with an IC₅₀ of 270 μM and a K_I of 37 ± 17 μM (Fig. 5A). When SI24 was added to σ^E after holoenzyme formation, a small concentration-dependent effect on transcription was observed, with an IC₅₀ of 1.25 mM (Fig. 5B). These data indicate that SI24 is a transcription inhibitor and suggest that SI24 may act at a step before holoenzyme formation.

FIG 5.

SI24 inhibits transcription in vitro. (A) Multiple-round and single-round transcription assays with SI24 incubated with σ^E before the addition of core RNAP. (B) Multiple-round and single-round transcription assays with SI24 after σ^E holoenzyme formation. (C) Single-round transcription assays with SI24 incubated with σ⁷⁰ before the addition of core RNAP. Representative gels and plots showing averages of data from 2 independent experiments, with whiskers indicating standard deviations, are shown for multiple-round (filled circles) and single-round (open circles) assays. Solid lines show fits to the Hill equation.

To confirm that SI24 inhibits σ^E-dependent transcription, similar experiments were performed by using a single-round transcription assay. In these experiments, heparin was included in the reaction mixture to prevent rebinding of polymerase to DNA, thereby limiting transcription to a single round. When SI24 was added to σ^E prior to the addition of core RNAP, transcription was inhibited in a concentration-dependent manner, with a K_I of 62 ± 6 μM (Fig. 5A). However, when σ^E was incubated with core RNAP to form the holoenzyme before SI24 was added, little inhibition was observed (Fig. 5B). These data are consistent with the results from the multiple-round assays. The observations that SI24 is more potent when added before holoenzyme formation in multiple-round transcription assays and that it is active only when added before holoenzyme formation in the single-round transcription assay suggest that SI24 interferes with the formation of an active σ^E-RNAP holoenzyme. The multiple-round transcription assay involves the reassociation of σ^E and RNAP to form a holoenzyme and initiate a new round of transcription following termination, providing an opportunity for SI24 to act, whereas this recycling does not occur in the single-round assay.

SI24 binds σ^E in vitro and inhibits holoenzyme formation.

Because SI24 inhibited transcription only when it was preincubated with σ^E, the simplest explanation for the mechanism of action of SI24 is that it binds σ^E or core RNAP and disrupts holoenzyme formation. To test if SI24 binds σ^E directly, cyclic SI24 was used in fluorescence anisotropy experiments with BODIPY-labeled σ^E_N80C/C165A. In this σ^E variant, the naturally occurring cysteine was replaced with alanine, and a cysteine residue was introduced at position 80 in place of asparagine, so the protein could be labeled at a surface-exposed region that does not interfere with binding to core RNAP (51). Incubation of BODIPY-σ^E_N80C/C165A with increasing concentrations of SI24K altered the anisotropy of BODIPY-σ^E_N80C/C165A in a concentration-dependent manner. SI24 bound BODIPY-σ^E_N80C/C165A with a K_D of 32 ± 8 μM (Fig. 6A). The K_D was similar when the anisotropy experiment was repeated by using wild-type σ^E labeled with BODIPY at cysteine 165, indicating that the position of the label did not influence peptide binding (not shown).

FIG 6.

SI24 binds σ^E and inhibits interactions with core RNAP. (A) Fluorescence anisotropy experiments using SI24 and BODIPY-σ^E_N80C/C165A were used to determine the dissociation constant for binding (K_D = 32 ± 8 μM). (B) Fluorescence anisotropy experiments using core RNAP and BODIPY-σ^E_N80C/C165A were used to determine the dissociation constant for binding (K_D = 8.7 ± 5 nM). Solid lines in panels A and B show fits to equation 2. (C) Competition binding experiments were used to determine the affinity of core RNAP for σ^E and σ⁷⁰. Preformed complexes with 5 nM BODIPY-σ^E_N80C/C165A and 75 nM RNAP were challenged with the unlabeled σ^E or σ⁷⁰ competitor. (D) Competition binding experiments show that SI24 inhibits holoenzyme formation. SI24 was added to binding reaction mixtures containing 5 nM BODIPY-σ^E_N80C/C165A and 75 nM core RNAP. Solid lines in panels C and D show fits to the Hill equation.

Given that SI24 binds σ^E, we next examined whether SI24 binding could interfere with σ^E holoenzyme formation. Fluorescence anisotropy was used first to determine the binding affinity of σ^E and core RNAP and then to determine whether SI24 alters the binding affinity. Core RNAP bound to BODIPY-σ^E_N80C/C165A with a K_D of 8.7 ± 5 nM (Fig. 6B). Because the σ^E variant contains two mutations and a fluorescent label, which might affect binding to core RNAP, the binding affinity of wild-type σ^E and core RNAP was determined by using a competition assay in which increasing amounts of unlabeled wild-type σ^E were added to the core RNAP BODIPY-σ^E_N80C/C165A complex. The K_I for wild-type σ^E and core RNAP was 4 ± 1 nM, which is close to the K_D measured with the labeled σ^E variant. To ensure that the anisotropy assay provided accurate measurements of binding affinity, the binding constant for core RNAP binding with σ⁷⁰, the housekeeping sigma factor, was measured by using the same competition assay with σ⁷⁰ as the competitor. σ⁷⁰ bound core RNAP with a K_I of 3 ± 2 nM (Fig. 6C), in agreement with previously reported K_D values (52).

Having established the assay, the ability of SI24 to disrupt the interaction between core RNAP and BODIPY-σ^E_N80C/C165A was tested. When BODIPY-σ^E_N80C/C165A was preincubated with SI24, SI24 decreased the affinity of BODIPY-σ^E_N80C/C165A for core RNAP in a concentration-dependent manner (Fig. 6D). The K_I for binding of SI24 and BODIPY-σ^E_N80C/C165A calculated from the inhibition data was 53 μM, similar to the K_D value obtained by using the direct binding assay. The K_I values for σ^E binding with core RNAP calculated from the multiround transcription and single-round transcription experiments were similar to those measured by the anisotropy experiment, indicating that binding of SI24 with σ^E could account for the observed inhibition.

E. coli sigma factors share some conserved regions, including those required to bind core RNAP (7). To determine if SI24 is specific for σ^E or can inhibit the binding of other sigma factors with core RNAP, the effect of SI24 on σ⁷⁰-dependent transcription was assayed. When SI24 was incubated with σ⁷⁰ before holoenzyme formation in a single-round transcription experiment, some inhibition was observed. However, higher concentrations of SI24 were needed to inhibit σ⁷⁰-dependent transcription than σ^E-dependent transcription; the K_I for SI24 inhibition of σ⁷⁰-dependent transcription was >100 μM (Fig. 5C). Therefore, although SI24 can inhibit σ⁷⁰-dependent transcription in vitro, it preferentially inhibits σ^E.

In a wild-type E. coli strain, most σ^E in the cell is bound to the anti-sigma factor RseA, so the concentration of free σ^E is low. If SI24 acts by inhibiting σ^E-dependent transcription in vivo, increasing the free concentration of σ^E should counteract the effects of SI24. To test this prediction, the active concentration of σ^E in the cell was increased by deleting the gene encoding RseA. σ^E activity is 25- to 30-fold higher in a ΔrseA strain than in a wild-type strain (S. E. Ades, unpublished observations; 29). In the ΔrseA strain, cyclic peptide expression from pSI24 did not affect fluorescence from the E. coli prpoErybB pompC′-yfp strain, and cyclic peptide expression from pSI24Am7K did not alter the growth rate or morphology of the cells (Fig. 7). Western blotting demonstrated that the expression levels of the full-length SI24 SICLOPPS protein and the first cyclization product from pSI24 and pSI24Am7K were the same in the wild-type and ΔrseA strains. These results suggest that additional σ^E can overcome inhibition by SI24 and are consistent with SI24 acting by preventing σ^E binding with core RNAP.

FIG 7.

Deletion of rseA suppresses the phenotypes caused by SI24. (A) SI24 does not increase fluorescence when expressed in the E. coli prpoErybB pompC′-yfp ΔrseA strain. Results of assays for SI24 expression in the E. coli prpoErybB pompC′-yfp strain (wild type [wt]), performed as described in the legend of Fig. 1B, are shown for comparison. (B) Representative growth curves of the E. coli prpoErybB pompC′-yfp ΔrseA pSI24Am7K strain with (open triangles) or without (filled triangles) induction. Growth curves for the E. coli prpoErybB pompC′-yfp pSI24Am7K strain with (open circles) and without (filled circles) induction, performed as described in the legend of Fig. 3A, are shown for comparison.

SI24 is not an effective inhibitor when added to E. coli cultures.

For SI24 to act as an antibiotic, it must be effective when added exogenously to cultures of bacterial cells. However, when chemically synthesized SI24 was added to cultures of the E. coli prpoErybB pompC′-yfp strain, the peptide had no effect on growth or OmpC′-YFP fluorescence. These data indicate that although SI24 is an effective inhibitor of σ^E in vitro and when it is expressed inside the cell, either SI24 cannot cross the cell envelope of E. coli or it does not accumulate to high enough concentrations in the cytoplasm to block σ^E activity.

Validation of the σ^E pathway inhibitor screen under HTS conditions.

Screening for high YPF fluorescence in the E. coli prpoErybB pompC′-yfp reporter strain successfully identified inhibitors of the σ^E pathway by using a FACS-based screen with a library of compounds expressed in vivo. To determine if the reporter could be used to screen compound libraries in a high-throughput format, YFP fluorescence from E. coli prpoErybB pompC′-yfp and E. coli pompC′-yfp strains was measured in 384-well microtiter plates. The average fluorescence in the E. coli pTrc99a pompC′-yfp strain, the positive-control strain mimicking full inhibition of σ^E activity, was 5-fold higher than that in the E. coli prpoErybB pompC′-yfp strain, with a Z′ value of 0.6 ± 0.05. These data indicate that the assay is appropriate for high-throughput screening.

DISCUSSION

The results described here characterize a cell-based assay that can efficiently identify inhibitors of the σ^E pathway. The SI24 cyclic peptide binds σ^E with a K_D of ∼32 μM, inhibits binding of σ^E to RNAP, and inhibits transcription from σ^E-dependent promoters in vitro and in vivo (Fig. 5 and 6). The affinity of σ^E for RNAP is 1,000-fold higher than that for SI24, so high concentrations of SI24 (0.1 to 1 mM) are required to inhibit σ^E-dependent transcription in vitro under equilibrium binding conditions (Fig. 5A). However, reporter assays demonstrated that SI24 inhibits σ^E-dependent transcription in vivo even when it is expressed from pSI24 and requires read-through of a stop codon for production (Fig. 1B).

What is responsible for the increased effectiveness of SI24 in vivo? Transcription of σ^E-dependent promoters in E. coli requires the binding of free σ^E to free core RNAP, and the reaction is likely to be controlled by kinetics and not thermodynamics. The binding affinity of σ^E and RseA is 0.2 nM (data not shown), and there is ∼2.5-fold more RseA than σ^E in E. coli, so there is little free σ^E at equilibrium (53). However, RseA is an unstable protein in E. coli. The half-life of RseA is ∼8 min (54), so turnover of RseA will produce free σ^E that can interact with core RNAP or SI24. The concentration of free core RNAP available to bind to σ^E is limited by its engagement in transcription elongation and by binding with σ⁷⁰. The affinity of core RNAP for σ⁷⁰ is 3 ± 2 nM (Fig. 6C), and the concentration of σ⁷⁰ in the cell is higher than the concentration of core RNAP (and significantly higher than the concentration of free σ^E) (55). However, sigma factors are released from core RNAP after transcription initiation (56). After transcription termination, core RNAP can bind σ^E or another sigma factor, and it is likely that kinetic competition for sigma factor binding controls the amount of σ^E-dependent transcription. In this model, SI24 could inhibit σ^E-dependent transcription in vivo by decreasing the rate of association of σ^E and core RNAP. Consistent with this kinetic regulation model, SI24 did not inhibit transcription in vitro when it was added after holoenzyme formation in the single-round transcription assay (Fig. 5B).

SI24 can inhibit transcription by σ⁷⁰ in addition to σ^E but with lower affinity. Therefore, it is possible that the toxicity associated with expression from pSI24Am7K is due in part to the inhibition of σ⁷⁰. However, there are several indications that SI24 acts by inhibiting σ^E. First, increasing the amount of free σ^E in the cell by deleting rseA eliminates the effects of pSI24Am7K, consistent with SI24 acting through σ^E. Second, there is much more σ⁷⁰ than σ^E in E. coli, and concentrations of cyclic peptides are unlikely to reach a level at which they can fully inhibit σ⁷⁰. Third, clear selectivity for σ^E was seen when SI24 was expressed from plasmid pSI24. Transcription of ompC′-yfp depends on σ⁷⁰, but the expression of ompC′-yfp increased in the presence of SI24 (Fig. 1B), consistent with SI24 inhibiting σ^E more efficiently than σ⁷⁰. Overall, the work presented here demonstrates that inhibitors of σ^E-dependent transcription can be obtained by using the prpoErybB pompC′-yfp assay.

SI24 is not a lead compound for drug development because it does not enter E. coli cells when added exogenously. However, elucidation of the binding interface between SI24 and σ^E would reveal interactions that could be targeted for structure-based drug design. Moreover, the assay used to identify SI24 is appropriate for high-throughput formats and could be used to identify inhibitors from libraries of molecules with good pharmaceutical properties. Because repression of ompC′-yfp requires both σ^E activity and Hfq-RybB activity, the assay could identify inhibitors of either pathway.

In this screen, plasmids from all the selected cells contained a stop codon. In pSI24, this stop codon is read with low efficiency by tRNA^Lys, which is consistent with previously reported data about the UAG nonsense codon being misread by tRNA^Lys_UUU (where UUU is the anticodon) (57). It is notable that the read-through allowed the identification of a potent inhibitor that was toxic when a sense codon was used. The growth and morphological defects produced by pSI24Am7K were severe enough that cells with the sense codon would have been unlikely to survive the screening process. Plasmids with amber codons have also been selected in other screens (37). In addition to tRNA^Lys, amber codons can be read at low efficiency by tRNA^Gln and tRNA^Tyr. It is not yet known what peptides were produced by these other nonsense plasmids, but stop codon read-through might provide a means to identify peptides and plasmids produced by SICLOPPS or other mechanisms that are extremely potent.

Bacterial RNA polymerase has been explored as a target for drug development for many years. The majority of polymerase inhibitors bind in or near the active site and interfere with RNA synthesis (15, 16). Recent HTS and rational design approaches have identified compounds that interfere with sigma-core interactions (58–60). However, these compounds were designed to inhibit transcription by the housekeeping sigma factor. Inhibitors of alternative sigma factors involved in infection and virulence, such as σ^E, present a different approach to antibiotic development, because such compounds will interfere with a pathogen's ability to cause disease. One other inhibitor of an alternative sigma factor, σ^B from Listeria monocytogenes, has been reported (61). However, this inhibitor does not bind to σ^B and instead appears to act by interfering with the signaling pathway leading to the activation of σ^B (62). SI24 is the first compound reported to bind to an alternative sigma factor and interfere with transcription by that sigma factor. This assay is currently being used in small-molecule screening with the goal of identifying more selective inhibitors of σ^E. In addition, the assay can be readily adapted to find inhibitors of other alternative sigma factors by changing the sigma factor gene carried on the plasmid and replacing the promoter for rybB with the appropriate promoter sequence.

This is the third published study in which a SICLOPPS library was used in conjunction with FACS to validate antibiotic targets and screens (37, 38). We suggest that this is a useful route to validate assays for high-throughput screening. The SICLOPPS system provides a complex library that is very inexpensive to produce. Likewise, FACS allows rapid isolation of positive clones at low cost. All assays that have been developed for SICLOPPS/FACS have adapted easily to a microtiter plate format for HTS. In addition to validating the screening assay, the SICLOPPS library yields cyclic peptides that can be used as controls for HTS.