Genome-Wide Screening Identifies Six Genes That Are Associated with Susceptibility to Escherichia coli Microcin PDI

Abstract

The microcin PDI inhibits a diverse group of pathogenic Escherichia coli strains. Coculture of a single-gene knockout library (BW25113; n = 3,985 mutants) against a microcin PDI-producing strain (E. coli 25) identified six mutants that were not susceptible (ΔatpA, ΔatpF, ΔdsbA, ΔdsbB, ΔompF, and ΔompR). Complementation of these genes restored susceptibility in all cases, and the loss of susceptibility was confirmed through independent gene knockouts in E. coli O157:H7 Sakai. Heterologous expression of E. coli ompF conferred susceptibility to Salmonella enterica and Yersinia enterocolitica strains that are normally unaffected by microcin PDI. The expression of chimeric OmpF and site-directed mutagenesis revealed that the K₄₇G₄₈N₄₉ region within the first extracellular loop of E. coli OmpF is a putative binding site for microcin PDI. OmpR is a transcriptional regulator for ompF, and consequently loss of susceptibility by the ΔompR strain most likely is related to this function. Deletion of AtpA and AtpF, as well as AtpE and AtpH (missed in the original library screen), resulted in the loss of susceptibility to microcin PDI and the loss of ATP synthase function. Coculture of a susceptible strain in the presence of an ATP synthase inhibitor resulted in a loss of susceptibility, confirming that a functional ATP synthase complex is required for microcin PDI activity. In trans expression of ompF in the ΔdsbA and ΔdsbB strains did not restore a susceptible phenotype, indicating that these proteins are probably involved with the formation of disulfide bonds for OmpF or microcin PDI.

INTRODUCTION

Escherichia coli strain 25 (E. coli 25; cattle origin) has an in vitro and in vivo competitive advantage against other E. coli strains that is linked to the production of the microcin PDI (MccPDI) (1–3). The inhibitory phenotype was first observed in vitro (4) and later called “proximity-dependent inhibition” (PDI), because inhibition occurred only when competing cells were in close proximity to sensitive cells (1). MccPDI appears to be most closely related to class IIa microcins, and the cluster of genes that encode MccPDI and associated immunity, activation, and export are located on a conjugative plasmid (2).

E. coli produces various antimicrobial bacteriocins that are classified as colicins or microcins. Microcins are distinguished by their lower molecular mass (<10 kDa) and require active transport across the membrane of producing cells. Microcins typically have a narrow spectrum of activity that is mediated through specific receptors expressed on the surface of susceptible bacteria. To date, 16 microcins have been described, including MccPDI. The receptors for seven of these microcins have been identified and include the outer membrane proteins Cir, FepA, Fiu, FhuA, and OmpF, all of which normally function in iron and other nutrient uptake (5–7).

While the gene encoding MccPDI has been identified and gene knockout and complementation studies have confirmed its inhibitory activity (2), little is known about how this microcin functions. The purpose of the current study was to identify any receptor and associated proteins required for MccPDI to recognize and inhibit susceptible E. coli. The E. coli Keio collection, a single-gene deletion library, was screened for mutants able to grow in the presence of the MccPDI-producing strain E. coli 25. The Keio collection is comprised of individual gene knockouts for all nonessential genes that are expressed by the PDI-susceptible E. coli strain BW25113. Screening of the full library followed by validation experiments demonstrated that strains lacking atpA, atpF, dsbA, dsbB, ompF, or ompR were no longer sensitive to PDI, indicating that these genes are required for MccPDI to inhibit susceptible cells. Subsequent heterologous expression of E. coli ompF in Salmonella enterica and Yersinia enterocolitica conferred MccPDI susceptibility to these species. The mechanism by which MccPDI inhibits susceptible E. coli is not known, but here we demonstrate that MccPDI inhibition of susceptible cells requires a functional ATP synthase complex.

MATERIALS AND METHODS

Bacterial strains, media, and culture conditions.

Bacterial strains used in this study (Table 1) were cultured in LB-Lennox medium (Difco, Franklin Lakes, NJ) or in M9 minimal medium (6 g liter⁻¹ Na₂HPO₄, 3 g liter⁻¹ KH₂PO₄, 0.5 g liter⁻¹ NaCl, 1 g liter⁻¹ NH₄Cl, 2 mg liter⁻¹ thiamine, 1 mM MgSO₄, 0.1 mM CaCl₂, and 0.2% glucose) at 37°C with shaking (200 rpm) unless otherwise indicated. Electroporation experiments were conducted using super optimal broth (SOB; 20 g liter⁻¹ Bacto tryptone, 5 g liter⁻¹ Bacto yeast extract, 0.5 g liter⁻¹ NaCl, 2.5 mM KCl, and 10 mM MgCl₂) and super optimal broth with catabolite repression (SOC; SOB with 20 mM glucose). The E. coli Keio knockout collection (termed the Keio collection), a set of single-gene deletion mutants for all nonessential genes in E. coli K-12 strain BW25113, was purchased from Thermo Fisher Scientific Inc. To make it possible to isolate the wild-type BW25113 strain from a mixed culture, the strain was sequentially passaged five times in LB broth with increasing concentrations of antibiotic until it was capable of growing in the presence of 30 μg ml⁻¹ nalidixic acid (1). Keio collection knockout strains originally were constructed by inserting a kanamycin resistance cassette into the gene of interest. For the current project, these strains were selected in culture and on agar plates with the addition of kanamycin. Depending on the experiment, antibiotics were added to media at the following concentrations: tetracycline, 50 μg ml⁻¹; chloramphenicol, 34 μg ml⁻¹; kanamycin, 50 μg ml⁻¹; nalidixic acid, 30 μg ml⁻¹; and ampicillin, 100 μg ml⁻¹.

TABLE 1.

Bacterial strains used in this study

Strain name	Designation	Relevant genotype/phenotypea	Reference or source
E. coli 25	E25	Wild type; SSuTr PDI+	1
E25ΔtraM	E25ΔtraM	SSuTr Cmr PDI+, traM deletion prevents conjugation	3
E25ΔmcpM	E25ΔmcpM	SSuTr Cmr PDI−, mcpM deletion eliminates PDI inhibition phenotype	3
Keio collection
BW25113	Kwt	Nalr, Keio collection wild-type K-12 strain	46
BW25113 (pMMB207 only)	Kwt(vector ctrl)	Nalr Cmr, BW25113 with empty pMMB207 vector	This study
51H12 ΔatpA (p207::atpA)	KΔatpA(atpA)	Kanr Cmr, atpA deletion complemented with atpA	This study
53B1 ΔatpF (p207::atpF)	KΔatpF(atpF)	Kanr Cmr, atpF deletion complemented with atpF	This study
65F11 ΔdsbA (p207::dsbA)	KΔdsbA(dsbA)	Kanr Cmr, dsbA deletion complemented with dsbA	This study
67G8 ΔdsbB (p207::dsbB)	KΔdsbB(dsbB)	Kanr Cmr, dsbB deletion complemented with dsbB	This study
7G4 ΔompF (p207::ompF)	KΔompF(ompF)	Kanr Cmr, ompF deletion complemented with ompF	This study
5F1 ΔompR (p207::ompR)	KΔompR(ompR)	Kanr Cmr, ompR deletion complemented with ompR	This study
7G4 ΔompF (p207::dsbA)	KΔompF(dsbA)	Kanr Cmr, ompF deletion expressing recombinant dsbA	This study
7G4 ΔompF (p207 only)	KΔompF(vector ctrl)	Kanr Cmr, ompF deletion with empty pMMB207 vector	This study
51H12 ΔatpA (p207::ompF)	KΔatpA(ompF)	Kanr Cmr, atpA deletion expressing recombinant ompF	This study
53B1 ΔatpF (p207::ompF)	KΔatpF(ompF)	Kanr Cmr, atpF deletion expressing recombinant ompF	This study
65F11 ΔdsbA (p207::ompF)	KΔdsbA(ompF)	Kanr Cmr, dsbA deletion expressing recombinant ompF	This study
67G8 ΔdsbB (p207::ompF)	KΔdsbB(ompF)	Kanr Cmr, dsbB deletion expressing recombinant ompF	This study
5F1 ΔompR (p207::ompF)	KΔompR(ompF)	Kanr Cmr, ompR deletion expressing recombinant ompF	This study
O157:H7 Sakai	O157:H7	Wild type; Nalr	47
ΔatpA	O157:H7ΔatpA	Nalr Kanr, atpA deletion	This study
ΔatpF	O157:H7ΔatpF	Nalr Kanr, atpF deletion	This study
ΔdsbA	O157:H7ΔdsbA	Nalr Kanr, dsbA deletion	This study
ΔdsbB	O157:H7ΔdsbB	Nalr Kanr, dsbB deletion	This study
ΔompF	O157:H7ΔompF	Nalr Kanr, ompF deletion	This study
ΔompR	O157:H7ΔompR	Nalr Kanr, ompR deletion	This study
ompF heterologous expression, chimeric expression, and in cis His tagged
7G4 ΔompF (p207::Se.ompF)	KΔompF(Se.ompF)	Kanr Cmr, ompF deletion complemented with Salmonella ompF	This study
7G4 ΔompF (p207::ompFEc1–72+Se68–363)	KΔompF(ompFEc+Se)	Kanr Cmr, ompF deletion complemented with N-terminal sequence of E. coli ompF (bases 1–72) and C-terminal sequence (bases 68–363) of Salmonella ompF	This study
7G4 ΔompF (p207::ompFSe1–68+Ec73–362)	KΔompF(ompFSe+Ec)	Kanr Cmr, ompF deletion complemented with N-terminal sequence of Salmonella ompF (bases 1–68) and C-terminal sequence (bases 73–362) of E. coli ompF	This study
7G4 ΔompF (p207::Ec.ompFΔSYGG)	KΔompF(Ec.ΔSYGG)	Kanr Cmr, ompF deletion complemented with E. coli ompF sequence containing an SYGG deletion	This study
7G4 ΔompF (p207::Ec.ompFΔKGN)	KΔompF(Ec.ΔKGN)	Kanr Cmr, ompF deletion complemented with E. coli ompF sequence containing a KGN deletion	This study
S. enteritidis 22577(p207::Ec.ompF)	Se22577(Ec.ompF)	Nalr Cmr, Se22577 expressing recombinant E. coli ompF	This study
Y. enterocolitica JB580v(p207::Ec.ompF)	JB580v(Ec.ompF)	Nalr Cmr, JB580v expressing recombinant E. coli ompF	This study
BW25113::ompF-His	Kwt::ompF-His	Nalr Kanr, in cis His-tag insertion at C-terminus of ompF	This study
S. enteritidis 22577::ompF-His	Se22577::ompF-His	Nalr Kanr, S. enteritidis strain 22577 with in cis His tag insertion at C terminus of ompF	This study
Other
BW25113ΔtonBΔtolA	KΔtonBΔtolA	Kanr Cmr, tonB and tolA deletion strain, BW25113 background	This study
BW25113ΔtonBΔtolQ	KΔtonBΔtolQ	Kanr Cmr, tonB and tolQ deletion strain, BW25113 background	This study
BW25113ΔexbDΔtolQ	KΔexbDΔtolQ	Kanr Cmr, exbD and tolQ deletion strain, BW25113 background	This study
BW25113ΔexbDΔtolR	KΔexbDΔtolR	Kanr Cmr, exbD and tolR deletion strain, BW25113 background	This study
S. enteritidis 22577	Se22577	Wild type, Nalr, Salmonella enterica serovar Enteritidis 22577	This study
S. enteritidis 21679	Se21679	Wild type, Nalr, Salmonella enterica serovar Enteritidis 21679	This study
Y. enterocolitica JB580v	JB580v	Nalr, Yersinia enterocolitica strain JB580v	48

^aSSuT^r, resistant to streptomycin, sulfonamide, and tetracycline antibiotics.

Screening for loss of MccPDI susceptibility.

The Keio collection was employed to identify genes associated with susceptibility to MccPDI. Each mutant was grown overnight at 37°C without shaking in a 96-well plate containing 150 μl fresh LB (with kanamycin) per well. A 10-ml culture of E. coli 25 also was started at this time in LB (with tetracycline) and incubated at 37°C with shaking (200 rpm). The following day each mutant was individually placed into coculture with E. coli 25 in a sterile, U-bottom 96-well plate with 200 μl M9 medium per well. A 96-pin replicator (Boekel Scientific, Feasterville, PA) was used to transfer overnight cultures (∼1 μl) of each strain for these competition experiments. The replicator was sterilized 3× between each use by submerging pins into 70% ethanol and flaming. The E. coli 25 culture was poured into a sterile plastic trough and transferred in the same manner. Competition cultures were incubated overnight at 37°C, with shaking at 100 rpm. Approximately 24 h later cocultures (∼1 μl) were transferred onto LB agar containing kanamycin or tetracycline to select for the Keio strains or E. coli 25, respectively. The plates were incubated at 37°C for at least 6 h. Growth on tetracycline verified the presence of E. coli 25 in the culture. No growth on kanamycin indicated that the Keio knockout strain being tested was still susceptible to PDI. Growth of a Keio mutant strain on the kanamycin plate indicated the identification of gene knockout strains that potentially were no longer susceptible to PDI.

Direct competition assays.

Bacterial strains were grown individually overnight in LB medium with the appropriate antibiotics. Equal volumes of each competing strain were inoculated into fresh M9 medium at a 1:200 dilution, and then the mixed cultures were incubated at 37°C for 24 h. When necessary, isopropyl-β-d-thiogalactopyranoside (IPTG; 100 μM unless specified otherwise) and chloramphenicol were added into M9 medium during the competition. Monocultures of each competing strain also were prepared as controls after inoculation into fresh M9 medium at the same dilution. The ATP synthase inhibitor resveratrol (Sigma-Alrich) was added at a final concentration of 25 μg ml⁻¹ to determine if a functional ATP synthase complex is required for susceptibility to MccPDI (8). To quantify the CFU of each strain following competition, each culture was serially diluted in a 96-well plate containing sterile PBS and dropped (5 μl) onto LB agar supplemented with the appropriate antibiotics to select for each competing strain. Agar plates (3 per enumerated dilution) were incubated overnight at 37°C, and colonies were tallied to estimate total CFU.

Complementation of six Keio mutant strains.

Genomic DNA from strain BW25113 was extracted by using a DNeasy blood and tissue kit by following the manufacturer's instructions (Qiagen, Valencia, CA). Primers (Table 2) were used to amplify the open reading frames of six genes from BW25113 genomic DNA with the addition of sequence to synthesize Flag tags at the C terminus of each recombinant protein. Each amplified fragment was digested with corresponding restriction enzymes and cloned into the vector pMMB207 (9), resulting in six recombinant plasmids: p207::atpA, p207::atpF, p207::dsbA, p207::dsbB, p207::ompF, and p207::ompR. The pMMB207 vector contains a chloramphenicol resistance gene. These plasmids were individually electroporated into E. coli S17-1 λpir, after which these constructs were introduced by conjugation into the corresponding six different mutant strains from the Keio collection (strains 51H12, 53B1, 65F11, 67G8, 7G4, and 5F1) to generate the KΔatpA(atpA), KΔatpF(atpF), KΔdsbA(dsbA), KΔdsbB(dsbB), KΔompF(ompF), and KΔompR(ompR) complemented strains (Table 1). KΔompF(dsbA) and KΔompF(vector ctrl) strains also were constructed as controls to confirm that complementation was specific for a targeted trait. For conjugation, donor S17-1 λpir strains and recipient strains were inoculated separately into LB broth with appropriate antibiotics and incubated overnight at 37°C or 30°C. Overnight cultures then were mixed (500 μl of each) in a 1.5-ml tube, and the cells were pelleted by centrifugation (1,500 × g, 5 min, room temperature). The pellet was washed once with 1 ml of LB and then resuspended in 50 μl of LB. The conjugation mixture was spotted onto an LB agar plate and incubated overnight at 37°C or 30°C. Conjugants were isolated by directly streaking onto selective plates. All resulting complemented strains were competed against the E. coli 25ΔtraM strain in M9 medium with IPTG and chloramphenicol as described above. For competition experiments with the KΔompF(ompF) strain, the concentration of IPTG was reduced to 20 μM. The E. coli 25ΔtraM strain was used for these experiments because it produces the PDI phenotype against susceptible strains but is nonconjugative due to the insertion of a chloramphenicol gene in traM (3).

TABLE 2.

PCR primers used in this study

Primer name	Sequencea (5′–3′)	Purpose
atpA_p207_KpnI	CGGGGTACCATGCAACTGAATTCCACCGA	Used to construct recombinant plasmid for atpA complementation
atpA_p207_HindIII	CCCAAGCTTTTACTTGTCGTCATCGTCTTTGTAGTCCCAGGATTGGGTTGCTTTGA
atpF_p207_EcoRI	CCGGAATTCATGGTGAATCTTAACGCAACAA	Used to construct recombinant plasmid for atpF complementation
atpF_p207_SalI	ACGCGTCGACTTACTTGTCGTCATCGTCTTTGTAGTCCAGTTCAGCGACAAGTTTATC
dsbA_p207_EcoRI	CCGGAATTCATGAAAAAGATTTGGCTGG	Used to construct recombinant plasmid for dsbA complementation
dsbA_p207_SalI	ACGCGTCGACTTACTTGTCGTCATCGTCTTTGTAGTCTTTTTTCTCGGACAGATATTTC
dsbB_p207_EcoRI	CCGGAATTCATGTTGCGATTTTTGAACCA	Used to construct recombinant plasmid for dsbB complementation
dsbB_p207_SalI	ACGCGTCGACTTACTTGTCGTCATCGTCTTTGTAGTCGCGACCGAACAGATCACGTT
ompF_p207_EcoRI	CCGGAATTCATGATGAAGCGCAATATT	Used to construct recombinant plasmid for ompF complementation
ompF_p207_SalI	ACGCGTCGACTTACTTGTCGTCATCGTCTTTGTAGTCGAACTGGTAAACGATACC
ompR_p207_EcoRI	CCGGAATTCATGCAAGAGAACTACAAGA	Used to construct recombinant plasmid for ompR complementation
ompR_p207_SalI	ACGCGTCGACCTACTTGTCGTCATCGTCTTTGTAGTCTGCTTTAGAGCCGTCCGGTA
ΔatpA_H1_P1	CTTGCAGACGTCTTGCAGTCTTAAGGGGACTGGAGC	Used to generate atpA knockout strain
ΔatpA_H2_P2	GCCTTGCGGCCTGCCCTAAGGCAAGCCGCCAGACGT
atpA_up	TGCTGCGATGGAAAAACGTC
atpA_down	TTCTGGACGCTTGCGATCTT
ΔatpF_H1_P1	TTTATTTAAAGAGCAATATCAGAACGTTAACTAAATAGAGGCATTGTGCT	Used to generate atpF knockout strain
ΔatpF_H2_P2	GGCGTAGGGGCGAGCTACCGTAATAAATTCAGACATCAGCCCCTCCCTCC
atpF_up	ATCGCTGTAGGTCTGGGTCT
atpF_down	ATGTCCTGCCAGCGTTCTAC
ΔdsbA_H1_P1	GTGAATATTCACGGGCTTTATGTAATTTACATTGAA	Used to generate dsbA knockout strain
ΔdsbA_H2_P2	AATTAACACCTATGTATTAATCGGAGAGAGTAGATC
dsbA_up	AGCGGCAGGATGCATTATCA
dsbA_down	GGGAAGATTACTGGCTGCGA
ΔdsbB_H1_P1	AATAAATATAGCGGCAGGAAAAAAGCGCTCCCGCAGGAGCGCCGAATGGA	Used to generate dsbB knockout strain
ΔdsbB_H2_P2	AATGAATTGGTTTAAACTGCGCACTCTATGCATATTGCAGGGAAATGATT
dsbB_up	CAATGGCAGATGAAGCGAGC
dsbB_down	TGCAAATGGGCTGGATAGCA
ΔompF_H1_P1	GTTGTCAGAATCGATCTGGTTGATGATGTAGTCAAC	Used to generate ompF knockout strain
ΔompF_H2_P2	GTGATCGTCCCTGCTCTGTTAGTAGCAGGTACTGCA
ompF_up	CGCTATCAGGGTAACGGGAG
ompF_down	AGCACTTTCACGGTAGCGAA
ΔompR_H1_P1	GGGCAAATGAACTTCGTGGCGAGAAGCGCAATCGCC	Used to generate ompR knockout strain
ΔompR_H2_P2	CTTACAAATTGTTGCGAACCTTTGGGAGTACAAACA
ompR_up	TGTTGCGAACCTTTGGGAGT
ompR_down	AGCAAGGTGACGATGAGCAA
Ec.ompF_ H1_ His_P1	GACGACACCGTTGCTGTGGGTATCGTTTACCAGTTCCATCATCACCATCACCATTAA	In cis addition of His tag (C terminal) to ompF in E. coli strain BW25113
Ec.ompF_ H2 _P2	AAGTCCTGTTTTTTCGGCATTTAACAAAGAGGTGTG
Se.ompF_ H1_ His_P1	GACGATCAGGCGGCTGTCGGTATTACTTACCAGTTCCATCATCACCATCACCATTGA	In cis addition of His tag (C terminal) to ompF in S. enteritidis strain 22577
Se.ompF_ H2_P2	AGTCCTGTTTTTGAGGCATAAAACAAAGGGGTCTG
Ec1–72_EcoRI	CCGGAATTCATGATGAAGCGCAATATT	Used to construct KΔompF(ompFEc+Se)
Ec1-72_R	CGGTGTTAATTTGAGTTTCCCCTTTAAAACCA
Se69-363_F	GGAAACTCAAATTAACACCGATCTGACCGGT
Se69-363_SalI	ACGCGTCGACTCAATGGTGATGGTGATGATGGAACTGGTAAGTAATACCGA
Se1-68_EcoRI	CCGGAATTCATGATGAAGCGCAAAATCCT	Used to construct KΔompF(ompFSe+Ec)
Se1-68_R	CGGAATTGATCTGCGTTTCCCCTTTAAA
Ec73-362_ F	GGAAACGCAGATCAATTCCGATCTGACCG
Ec73-362_SalI	ACGCGTCGACTTAATGGTGATGGTGATGATGGAACTGGTAAACGATACC
ΔSYGG-F	AATGGCGACATGACCTATG	Used to construct KΔompF(Ec.ΔSYGG)
ΔSYGG-R	GTTTTCACCGTTACCCTTG
ΔKGN-F	GGTGAAAACAGTTACGGTG	Used to construct KΔompF(Ec.ΔKGN)
ΔKGN-R	GGAAAAATAATGCAGACCAAC
ΔtolA_H1_P1	CTTGCTTGAAAGAGAGCGGGTAACAGGCGAACAGTTTTTGGAAACCGAGA	Used to knock out tolA for the ΔtonB knockout strain (Keio collection no. 67A9)
ΔtolA_H2_P2	CCGTGGCAACCGGTGCCTGATGTTGACCGTCCGAACAGTCAACATCGCGA
tolA_up	TAATGACGCAGCCTATCTAA
tolA_down	ATGCCTGCTTCATCATATCT
ΔtolQ_H1_P1	AAAATGAAGCCTCGTGCGCTTCCCAAGTCTATTGTCGCGGAGTTTAAGCA	Used to knock out tolQ for the ΔtonB knockout strain (Keio collection no. 67A9)
ΔtolQ_H2_P2	ATTTCGGACTTGAGATCGCGACGACCTCGTCCACGCGCTCTGGCCATGGC
tolQ_up	GTCAACGCCGAGAATACT
tolQ_down	ACCAATACCAGACACTTCAA
ΔtolR_H1_P1	TTCTGCACCGCCAGGCGTTTACCGTTAGCGAGAGCAACAAGGGGTAAGCC	Used to knock out tolR for the ΔexbD knockout strain (Keio collection no. 57A12)
ΔtolR_H2_P2	AACTGTTCGCCTGTTACCCGCTCTCTTTCAAGCAAGGGAAACGCAGATGT
tolR_up	TGGAACTGAATTACGACAAC
tolR_down	CAACTACCGCACCTGAAT
P1	TGTGTAGGCTGGAGCTGCTTCG	Used to amplify the kanamycin or chloramphenicol resistance genes
P2	CATATGAATATCCTCCTTA
C1	TTATACGCAAGGCGACAAGG	Used to verify the resistance gene cassette integration at the desired location combined with gene specific primers
C2	GATCTTCCGTCACAGGTAGG
K1	CAGTCATAGCCGAATAGCCT
K2	CGGTGCCCTGAATGAACTGC

^aFor gene deletion mutants, the kanamycin or chloramphenicol primers P1 and P2 were added to the 3′ end of homologous extension sequences (H1 and H2) of the E. coli-specific gene. Restriction enzyme sites are underlined. PCRs were performed in a volume of 50 μl containing 45 μl of Platinum PCR supermix (Invitrogen, Carlsbad, CA), 1 U Taq DNA polymerase, 22 mM Tris-HCl (pH 8.4), 55 mM KCl, 1.65 mM MgCl₂, 220 μM each deoxynucleoside triphosphate, and stabilizers. PCR was carried out under the following conditions: 2 min at 94°C for 1 cycle; 30 s at 94°C, 30 s at 55°C, and 2 min at 68°C for 30 cycles; and 10 min at 68°C for 1 cycle.

PCR-mediated gene deletion.

Gene-specific knockouts were generated using the homologous recombination system described by Datsenko and Wanner (10). Briefly, the gene of interest was replaced with a PCR-generated kanamycin or chloramphenicol resistance gene that originated from pKD4 or pKD3, respectively. Each primer incorporated a 36- to 50-nucleotide (nt) segment complementary to the region flanking the gene of interest as described in Table 2. PCR products were column purified using a QIAquick PCR purification kit (Qiagen), digested overnight at 37°C with DpnI (New England BioLabs, Ipswich, MA), purified again, and suspended in 30 μl 10 mM Tris, pH 8.0. The target strains carrying the λ Red plasmid pKD46 (Amp^r) were grown in SOB medium with 1 mM l-arabinose at 30°C to an optical density at 600 nm (OD₆₀₀) of ∼0.6. The cells then were made electrocompetent by washing twice with ice-cold water and once with 10% glycerol and concentrating the cells 100-fold in 10% glycerol. Electrocompetent cells (50 μl) were pulsed with ∼150 ng of purified PCR product using a Gene Pulser Xcell (Bio-Rad, Hercules, CA). SOC medium was immediately added to the cells that were then incubated for 2 h at 30°C. Cells were plated on LB with kanamycin or chloramphenicol and incubated overnight at 30°C to select for transformants. PCR was used to verify the resistance gene cassette integrated at the desired location using primers within the antibiotic resistance gene combined with genomic primers adjacent to the sequence of interest (Table 2).

Heterologous ompF expression.

MccPDI does not function against Salmonella or Yersinia (see Results). To determine if susceptibility can be conferred to these bacteria, the recombinant expression plasmid for E. coli ompF (p207::ompF) was introduced by conjugation from E. coli S17 λpir into S. enterica serovar Enteritidis 22577 and Yersinia enterocolitica JB580v to generate Se22577(Ec.ompF) and JB580v(Ec.ompF) strains, respectively (Table 1). Both strains then were competed against the E. coli 25ΔtraM or E. coli 25ΔmcpM strain in M9 medium with IPTG and chloramphenicol as described above. For these experiments, the E. coli 25ΔmcpM strain, which does not synthesize a functional MccPDI (3), was used as a negative control. Heterologous expression of the recombinant OmpF was confirmed by Western blotting.

Western blotting.

Strains were harvested from broth culture by centrifugation (13,800 × g, 2 min, 4°C), and the pellet was resuspended in Laemmli sample buffer and boiled for 5 min. The samples were electrophoresed by using Mini-Protean precast gels and subsequently transferred to a low-fluorescence polyvinylidene difluoride (LF-PVDF) membrane using the Trans-Blot Turbo transfer system (Bio-Rad). Membranes were blocked with 5% skim milk in Tris-buffered saline and probed with a His tag monoclonal antibody (1:2,500 dilution; Novagen), Flag-tagged monoclonal antibody (1:1,000; Thermo Scientific), or anti-DnaK monoclonal antibody (1:1,000; Enzo Life Sciences, Farmingdale, NY) for 1 h at room temperature. Blots then were incubated with anti-mouse DyLight 488 conjugate (1:5,000; Thermo Scientific) for 1 h at room temperature, followed by visualization using a ChemiDoc MP imaging system (Bio-Rad).

In cis expression of ompF fused with a C-terminal His tag.

To generate an in-frame, 6-histidine tag on the C terminus of the endogenous ompF gene in E. coli strain BW25113 and Salmonella serovar Enteritidis strain 22557, the stop codon of ompF was replaced with a 6× histidine-stop sequence and a kanamycin resistance gene by using methods described by Datsenko and Wanner (10). Briefly, 36-nt sequences flanking the stop codon of ompF were chosen as homologous extension arms (H1 and H2) (Table 2). The His tag sequence was added to the C terminus of arm H1. These primers then were used to amplify the kanamycin resistance gene cassette from the pKD4 plasmid, and the remaining protocol was followed as described above.

OmpF chimera from E. coli and S. enterica serovar Enteritidis.

Splice overlap extension PCR (11) was employed to construct two chimeric ompF sequences (ompF_{Ec1–72+Se68–363} [i.e., amino acids 1 to 72 of E. coli ompF and amino acids 68 to 363 of serovar Enteritidis ompF] and ompF_{Se1–68+Ec73–362}) based on templates from E. coli (strain BW25113) and S. enterica serovar Enteritidis (strain 22577). For the ompF_{Ec1–72+Se68–363} construct, primers Ec1-72_EcoRI, which incorporated an EcoRI site, and Ec1-72_R were paired, while primers Se69-363_F and Se69-363_SalI, which incorporated a SalI site, were paired to amplify the corresponding regions (Table 2). Both fragments contained a 10-bp complementary sequence and were used as the templates for the next PCR procedure, which used primers Ec1-72_EcoRI and Se69-363_SalI. The resulting fragment, containing amino acids 1 to 72 of E. coli ompF and amino acids 68 to 363 of serovar Enteritidis ompF, was inserted into the pMMB207 vector, generating recombinant plasmid p207::ompF_{Ec1–72+Se68–363}, to express the chimeric sequence of E. coli ompF and serovar Enteritidis ompF (ompF_Ec+Se). A similar strategy was used to construct and express the ompF_{Se1–68+Ec73–362} chimeric sequence (ompF_Se+Ec) (Table 1).

Site-directed mutagenesis.

Based on the chimeric OmpF data and amino acid variation between E. coli, S. enterica, and Y. enterocolitica, we identified the amino acid positions from OmpF that were most likely to play a functional role in binding to MccPDI. The corresponding DNA sequences in E. coli strain BW25113 were deleted by using mutagenesis primers (Table 2) and a Q5 site-directed mutagenesis kit (New England BioLabs). The manufacturer's protocol was followed to generate both the KΔompF(Ec.ΔSYGG) and KΔompF(Ec.ΔKGN) strains (Table 1).

Statistical analysis.

One-way analysis of variance (ANOVA) was used to compare experimental results with pairwise comparisons made by using a Dunnett's one-way multiple-comparison test (SigmaPlot, version 12.5; Systat Software, Inc., San Jose, CA).

RESULTS

Six genes in E. coli are requisite for, or are correlated with susceptibility to, MccPDI.

Mutants from the Keio collection (n = 3,985) were screened using a high-throughput 96-well plate method. Following two rounds of screening, six mutants (51H12, 53B1, 65F11, 67G8, 7G4, and 5F1) (Table 1) were consistently identified as resistant to inhibition by E. coli 25. These mutants then were placed into direct competition experiments (M9 medium) with E. coli 25, and all six strains were able to grow to a population density of 5.6 to 7.6 log CFU after 24 h, whereas the isogenic control (BW25113) only grew to a density of ∼3.5-log CFU (Table 3). These knockout strains grew to a slightly higher density (6.5 to 8.3-log CFU) when competed against the E. coli 25ΔmcpM MccPDI knockout strain, which is defective in MccPDI synthesis and also harbors a constitutively expressed chloramphenicol resistance gene (Table 3).

TABLE 3.

Gene knockout strains from the Keio collection that are no longer susceptible to killing by E. coli 25

Strain IDa	Gene knockout	Strain or gene functionb	Log10 CFU/ml ± SEM recovered after coculture withc:
			E. coli 25	E. coli 25ΔmcpM
BW25113	None	Wild-type E. coli K-12 from Keio collection; positive control	3.48 ± 0.30	8.20 ± 0.04
51H12	atpA	ATP synthase, F1 complex, α subunit; also called papA	5.61 ± 0.20	6.57 ± 0.26
53B1	atpF	ATP synthase, Fo complex, β subunit; also called papF	6.81 ± 0.39	7.33 ± 0.17
65F11	dsbA	Periplasmic protein disulfide isomerase; involved with disulfide bond formation	7.37 ± 0.06	8.22 ± 0.11
67G8	dsbB	Disulfide oxidoreductase; involved with disulfide bond formation membrane protein; oxidizes periplasmic DsbA	7.60 ± 0.38	8.33 ± 0.17
7G4	ompF	Outer membrane protein, porin	6.98 ± 0.14	7.97 ± 0.08
5F1	ompR	Transcriptional regulatory protein OmpR; response regulator for osmoregulation	6.33 ± 0.33	8.27 ± 0.07

^aStrain identifier (ID) or Keio collection clone ID.

^bPutative gene function as defined in the Ecogene 3.0 database using the structural and functional annotation of the E. coli K-12 genome (http://ecogene.org).

^cAverage CFU (±SEM) recovered after coculture with E. coli 25 (n = 3 independent replicates); recovery of >5 log₁₀ CFU/ml was considered resistant to MccPDI.

To rule out the contribution of polar effects from insertion of a kanamycin resistance gene during construction of the Keio collection knockout strains, complementation experiments demonstrated that susceptibility to MccPDI could be restored in all cases, while the KΔompF(dsbA) and KΔompF(vector ctrl) control strains maintained resistance to MccPDI (Fig. 1). To validate these findings, independent deletion mutants for the six genes were generated in an E. coli O157:H7 Sakai background. Competition assays with E. coli 25 confirmed loss of sensitivity to MccPDI (see Fig. S1 in the supplemental material). Collectively, these data indicate that, at a minimum, these six genes each need to be expressed for E. coli to be susceptible to MccPDI.

FIG 1.

Complementation restores susceptibility to MccPDI for the six Keio collection strains. The ompF mutant strain (KΔompF) with recombinant dsbA and empty pMMB207 vector were used as controls, indicating that complementation effects were specific. Results are expressed as the difference between mean log CFU (± standard errors of the means [SEM]; n = 3 independent experiments) during coculture and that during monoculture. *, significant ANOVA followed by a Dunnett's one-way multiple-comparison test versus control group (P < 0.001).

OmpF is necessary for susceptibility to MccPDI.

Because OmpF is a confirmed receptor for other microcins (12) and is the only known outer membrane protein from our list of six candidates, we hypothesized that heterologous expression of E. coli ompF in other bacterial species would make them susceptible to MccPDI. S. enterica and Y. enterocolitica strains synthesize OmpF homologues (see Fig. S2 in the supplemental material), but these species of bacteria are not susceptible to MccPDI (see Fig. S3). In trans expression of E. coli ompF (Ec.ompF) in S. enterica strain 22577 and Y. enterocolitica strain JB580v (Fig. 2A) conferred susceptibility to MccPDI (Fig. 2B).

FIG 2.

Heterologous expression of E. coli ompF converts resistant strains into MccPDI-susceptible strains. (A) Western blot detection of E. coli OmpF synthesis in Salmonella enteritidis (strain 22577) and Yersinia enterocolitica (strain JB580v). Lane 1, wild-type strains; lane 2, Salmonella 22577(Ec.ompF) and Yersinia JB580v(Ec.ompF) without IPTG induction; lane 3, Salmonella 22577(Ec.ompF) and Yersinia JB580v(Ec.ompF) after induction with IPTG. (B) Competition with E. coli 25ΔtraM (microcin-producing) and E. coli 25ΔmcpM (no microcin production, negative control) after heterologous expression of E. coli ompF in Salmonella and Yersinia. Results are expressed as the difference between mean log CFU (± SEM; n = 3 independent experiments) during coculture and monoculture. *, significant ANOVA (P < 0.001) for comparisons between paired competitions (E. coli 25ΔtraM strain versus E. coli 25ΔmcpM strain).

These findings are consistent with OmpF being necessary for MccPDI to inhibit susceptible E. coli. Nevertheless, similar results might arise if susceptibility to MccPDI is a function of the abundance of OmpF in the outer membrane (E. coli or otherwise). Under this scenario, the expression of recombinant OmpF would exceed the threshold abundance needed for a bacterium to become susceptible to MccPDI. To test this alternative, the abundance of in cis His-tagged OmpF was qualitatively assessed for both the E. coli and S. enterica serovar Enteritidis strains. Western blotting showed that for the culture conditions employed in this study, the total protein level of native OmpF in S. enterica is substantially greater than that of E. coli (Se222577::ompF-His and Kwt::ompF-His, respectively) (Fig. 3A). Furthermore, when recombinant S. enterica ompF was expressed in the E. coli KΔompF(Se.ompF) strain, it did not recover the susceptible phenotype when competed against the E25ΔtraM strain (Fig. 3B). These results are consistent with a specific MccPDI-E. coli OmpF interaction rather than a quantitative effect dependent on OmpF abundance.

FIG 3.

N terminus of E. coli ompF drives specificity of MccPDI. (A) Western blot detection of OmpF synthesis from E. coli strain BW25113 (Bw) and S. enteritidis strain 22577 (Se). For both strains the endogenous ompF gene was fused with a C-terminal His tag to enable detection using an anti-His antibody. Endogenous DnaK served as a relative loading control. (B) Competition with E. coli 25ΔtraM (microcin producing) or E. coli 25ΔmcpM (no microcin production) with the KΔompF strain after complementation with E. coli ompF (Ec.ompF) or Salmonella ompF (Se.ompF) or one of two chimeric ompF constructs (ompF_Ec+Se and ompF_Se+Ec). Results are expressed as the difference between mean log CFU during coculture and monoculture (n = 3 independent replicates; error bars indicate SEM). *, significant ANOVA (P < 0.001) for comparisons between paired competitions (E. coli 25ΔtraM strain versus E. coli 25ΔmcpM strain).

The N terminus of E. coli OmpF drives the specificity of MccPDI.

The extracellular regions (loops) of OmpF differ between E. coli and MccPDI-resistant species (S. enterica, Y. enterocolitica, and Klebsiella pneumoniae; see Fig. S2 in the supplemental material), suggesting that these regions serve as a specific ligand for MccPDI (13). To narrow the possibilities, two chimeric ompF sequences were constructed (ompF_{Ec1–72+Se68–363} and ompF_{Se1–68+Ec73–362}) (Table 1). After complementation into the Keio collection ΔompF strain (7G4), competition against an MccPDI-producing strain (the E25ΔtraM strain) demonstrated that the OmpF_{Ec1–72+Se68–363} (ompF_Ec+Se) chimera restored susceptibility to MccPDI, while the second chimera, OmpF_{Se1–68+Ec73–362} (ompF_Se+Ec), did not (Fig. 3B). Consequently, the N-terminal sequence (amino acids 1 to 72) of the E. coli OmpF probably is required for interaction with MccPDI.

Based on these findings, the interspecific diversity of OmpF amino acid composition (see Fig. S2 in the supplemental material), and the crystal structure of OmpF (PDB accession number 2ZFG), two potential binding regions were identified (K₄₇G₄₈N₄₉ and S₅₃Y₅₄G₅₅G₅₆), both of which are located in the first extracellular loop of E. coli OmpF. These regions subsequently were deleted as part of a recombinant protein, and the modified sequences were expressed in a ΔompF background [KΔompF(Ec.ΔKGN) and KΔompF(Ec.ΔSYGG), respectively]. Competition between these strains and an MccPDI-producing strain (the E25ΔtraM strain) indicated that deletion of K₄₇G₄₈N₄₉ blocked MccPDI susceptibility, whereas deletion of S₅₃Y₅₄G₅₅G₅₆ had no impact on MccPDI susceptibility (Fig. 4). Together, these results indicate that K₄₇G₄₈N₄₉ within the first extracellular loop of E. coli OmpF is a binding site for MccPDI. Alternatively, it is possible that some of our OmpF chimeras and mutants, which lacked an MccPDI-susceptible phenotype, were not displayed in the membrane properly due to either improper folding or trafficking.

FIG 4.

KGN motif located on the first extracellular loop of E. coli OmpF is the putative binding site for MccPDI. KΔompF(Ec.ΔSYGG) and KΔompF(Ec.ΔKGN) are ompF mutant strains expressing E. coli ompF sequence containing deletions for the KGN or SYGG peptides, respectively (see the text). Competition experiments included either a microcin-producing strain (E25ΔtraM) or nonproducing strain (E25ΔmcpM). Results are expressed as the difference between mean log CFU during coculture and monoculture. Experiments were replicated independently (n = 2). Error bars indicate SEM. *, significant ANOVA (P < 0.001) for comparisons between paired competitions (E. coli 25ΔtraM strain versus E. coli 25ΔmcpM strain).

MccPDI activity requires a functional ATP synthase complex.

Preliminary screening of the Keio collection demonstrated that ΔatpA and ΔatpF strains are no longer susceptible to MccPDI (Table 2). Both AtpA and AtpF are components of E. coli ATP synthase, which contributes to the production of ATP (14). Direct competition experiments between wild-type E25 and other knockout strains for the ATP synthase complex (KΔatpB [strain 54D1], KΔatpC [strain 45E6], KΔatpD [strain 45F6], KΔatpG [strain 89C11], and KΔaptI [strain 71G2]) did not confer resistance to MccPDI (see Fig. S4A in the supplemental material). In contrast, both KΔatpE (strain 54C1) and KΔatpH (strain 54A1) strains were resistant to MccPDI (see Fig. S4A).

Strains of E. coli with nonfunctional ATP synthase can grow on agar plates that are supplemented with glucose but not on plates that are supplemented with succinate or other C4-dicarboxylates (15). Accordingly, we cultured the Keio collection ATP synthase mutants on M9 agar plates supplemented with succinate or glucose to identify which mutations no longer have a functional ATP synthase complex. Only the mutant strains that were resistant to MccPDI (KΔatpA, KΔatpE, KΔatpF, and KΔatpH strains) were unable to grow on the succinate-supplemented plates, although they could grow on the glucose-supplemented plates (see Fig. S4B in the supplemental material). Consequently, MccPDI activity correlates with the presence of a functional ATP synthase complex. To determine if a functional ATP synthase is required for susceptibility, we conducted competition experiments in the presence of a previously described ATP synthase inhibitor, the polyphenol resveratrol (8). Inhibition of ATP synthase activity resulted in reduced susceptibility to MccPDI (see Fig. S4C). To ensure that this experimental design did not impact microcin production in the MccPDI-producing strain, both atpA and atpF were deleted from E. coli 25, resulting in strains that lack a functional ATP synthase complex. In both cases, the inhibitory phenotype from MccPDI was not affected. Consequently, MccPDI inhibition requires a functional ATP synthase within the susceptible strain. Interestingly, we found that in trans expression of E. coli ompF in two of the four atp gene knockouts (ΔatpA and ΔatpH mutants) restored susceptibility to MccPDI (Fig. 5). Similar in trans expression of ompF in the remaining two atp gene deletion strains (ΔatpF and ΔatpE mutants), however, did not restore susceptibility to MccPDI.

FIG 5.

In trans expression of E. coli ompF restores susceptibility to MccPDI in KΔatpA, KΔatpH, and KΔompR strains. Results are shown for competition with E25ΔtraM (microcin producing) or E25ΔmcpM (no microcin production) after complementation with ompF in different Keio collection strains (KΔatpA, KΔatpF, KΔdsbA, KΔdsbB, KΔompR, KΔatpH, and KΔatpH deletion strains). Results are expressed as the difference between mean log CFU counts during coculture and monoculture. Experiments were replicated independently (n = 3). Error bars indicate SEM. *, significant ANOVA (P < 0.001) for comparisons between paired competitions (E. coli 25ΔtraM strain versus E. coli 25ΔmcpM strain).

Potential roles of DsbA, DsbB, and OmpR in OmpF synthesis and maturation.

The ΔdsbA and ΔdsbB strains from the Keio collection were no longer susceptible to MccPDI (Table 2). Deletion of dsbA is known to limit the abundance of OmpF in the E. coli cell membrane (16); consequently, finding dsbA and dsbB linkage to MccPDI susceptibility was not unexpected. Interestingly, in trans expression of recombinant E. coli ompF in these two strains did not recover the susceptible phenotype (Fig. 5). OmpR is known to regulate the expression of ompF and ompC via a two-component regulatory system in E. coli (17). This is consistent with our findings whereby the gene knockout for ompR lost susceptibility to MccPDI (Table 2) while complementation of ompR restored susceptibility (Fig. 1), as did in trans expression of E. coli ompF in the KΔompR strain (Fig. 5).

The Ton and Tol systems are not involved in MccPDI activity.

Many characterized microcins and colicins employ the Ton or Tol system in conjunction with siderophore receptors (FepA, FhuA, and CirA) to enter susceptible cells, whereas others use only a porin protein (OmpF, OmpA, and OmpW) to gain entry into susceptible cells (12, 18). Preliminary screening of the Keio collection did not identify any of the corresponding Ton and Tol knockout strains as being resistant to MccPDI (data not shown). This does not eliminate the possibility that redundant activity from these systems would mask effects from single-gene knockouts. Consequently, a series of double knockouts were constructed in the wild-type Keio collection strain (ΔtonB ΔtolA, ΔtonB ΔtolQ, ΔexbD ΔtolQ and ΔexbD ΔtolR); however, competition against microcin-producing E. coli 25 showed no reduction in inhibition (see Fig. S5 in the supplemental material).

DISCUSSION

Microcins require outer membrane receptors on the target bacteria before they can recognize and inhibit these cells. The receptor-microcin interaction is highly specific, which explains the narrow spectrum of microcin activity (19). To identify proteins required for MccPDI activity, we screened a single-gene knockout library of the E. coli K-12 genome and identified six knockouts with reduced susceptibility to MccPDI (atpA, atpF, dsbA, dsbB, ompF, and ompR). OmpF is probably the only surface-exposed protein in this subset. Data from different experiments involving recombinant protein constructs, including heterologous expression in two different bacterial species, confirmed that OmpF serves as the receptor for MccPDI binding with a putative binding site being located in the first exterior loop of OmpF. OmpR is a response regulator involved in expression of ompF and ompC in E. coli (20). Consequently, it was not unexpected that the ompR knockout strain was resistant to MccPDI, and that subsequent in trans expression of ompF restored susceptibility for the ΔompR strain.

Some colicins employ both OmpF and BtuB to translocate into susceptible cells (21), but the ΔbtuB strain (55F12) from the Keio collection was fully susceptible to MccPDI (data not shown). Because others have demonstrated that the 55F12 strain harbors an intact copy of btuB in addition to the deleted locus (22), we generated an independent ΔbtuB strain, but it also remained susceptible to MccPDI (data not shown). It should be noted, however, that there was no positive control for this experiment, and we cannot be certain that the parental background that we used (E. coli 186) does not also harbor a second copy of this gene. Nevertheless, this finding is consistent with the absence of any reports linking microcins to BtuB. Because the 96-well plate assay used in this study was principally a screening tool of undefined sensitivity, we also confirmed gene knockout strains for all known receptors or translocators used by other microcins and colicins (ΔfhuA, ΔfepA, Δcir, Δfiu, ΔompW, ΔompA, and Δtsx strains) (5–7, 21, 23–25). All were fully susceptible to MccPDI in direct competition assays against the wild-type E. coli 25 strain (data not shown). Therefore, it appears that OmpF is the sole receptor for MccPDI, although we cannot rule out the possibility that more than one additional protein could serve as a redundant coreceptor that would not be detected by screening a library of single-gene knockout strains. If this was the case, these proteins would have to be highly conserved, because heterologous expression of E. coli ompF in Salmonella and Yersinia is sufficient to make these species susceptible to MccPDI.

All colicins and microcins described to date are thought to penetrate the outer membrane barrier to reach their site of action (periplasm or cytosol) in the target cells (26, 27). Molecules with a mass of approximately 600 Da are able to diffuse through the OmpF porin (28), but the predicted mass of MccPDI is approximately 12,000 Da (immature protein) and 10,000 Da after maturation (2). This mass difference suggests MccPDI cannot freely diffuse through the OmpF pore as a method for gaining entry into susceptible cells and may require a partner system for active transport. Microcins B17 and C are known to interact with OmpF (12), but there is no evidence that they require either the Ton or Tol system. Instead, these microcins rely on inner membrane systems, SbmA (29) and Yej (30), respectively, for active transport across the cell membrane. We individually tested these and other potential microcin transport-related genes (27) from the Keio collection (ΔsbmA, ΔyejA, ΔyejV, ΔyejE, ΔyejF, ΔmanY, and ΔmanZ) but found no loss of susceptibility to MccPDI (data not shown).

Given these results, it is possible that OmpF is the sole receptor for MccPDI. Under this scenario, and assuming that MccPDI crosses the susceptible cell membrane, MccPDI is probably secreted from the producing cell as a linear peptide. This mode of secretion has been predicted for colicin A and colicin N (31, 32). Colicin N is thought to bind the periphery of OmpF on the cell surface (33) and cross the outer membrane in an unfolded state, after which it forms a channel in the lipopolysaccharide (LPS) adjacent to OmpF (34). Future experiments will investigate the specific mechanism of MccPDI entry into susceptible E. coli.

Our screen of the Keio collection also identified ΔdsbA and ΔdsbB mutants as being required for MccPDI susceptibility. DsbA and DsbB are thiol-redox enzymes that catalyze disulfide bond formation between cysteine residues for proteins that are transported into the periplasm (35–38). In addition, Pugsley (16) reported that mutations in dsbA reduced the transcription of ompF, but when recombinant ompF was expressed in these knockout strains (this study), susceptibility to MccPDI was not restored. If deletion of dsbA or dsbB only impacted transcription of ompF, then complementation of ompF using an inducible promoter should have restored the MccPDI-susceptible phenotype. The lack of restoration indicates that these proteins play a role in posttranslational modification of OmpF, either directly or indirectly (36), through trafficking or through folding, although OmpF lacks any cysteine residues.

It also is possible that these proteins play a role in posttranslational modification of the microcin protein, assuming that it enters the periplasm (consistent with entry as a linear peptide). Cysteine is the least abundant amino acid found in proteins (39) and is also one of the most reactive (40, 41). For these reasons, if a protein contains an even number of cysteines and is predicted to function outside the cytoplasm, it is likely that the cysteines are paired to form disulfide bonds (42). MccPDI has four cysteines in the C terminus, consistent with the prediction that disulfide bond formation occurs. Consequently, DsbA and DsbB may serve two functions in this system, an indirect impact on posttranslational modification of OmpF and a direct interaction with MccPDI.

E. coli ATP synthase is an important enzyme that provides energy for the cell by catalyzing the synthesis of ATP during oxidative phosphorylation (14, 43). A cluster of genes (atpIBEFHAGDC) is responsible for encoding the subunits of this large enzyme complex (44). Initial screening of the Keio collection knockout library demonstrated that the ΔatpA and ΔatpF strains were not susceptible to MccPDI activity; additional direct competition experiments indicated that the ΔatpE and ΔatpH strains also were not susceptible to MccPDI. Microcin H47, a class IIb microcin, targets the AtpB, AtpE, and AtpF proteins that form the membrane-bound F_o proton channel of ATP synthase (a, c, and b subunits) (25). In the case of MccPDI, deletion of the four genes that encode proteins for the c, b, δ, and α subunits of ATP synthase (atpE, atpF, atpH, and atpA) confers resistance to MccPDI activity while also causing the loss of ATP synthase activity. Deletion of the remaining five ATP synthase component genes did not affect susceptibility to MccPDI or ATP synthase activity. Using an ATP synthase inhibitor, we showed that a functional ATP synthase is necessary for MccPDI inhibition, and we confirmed that the loss of ATP synthase activity within the microcin-producing strain had no impact on the inhibition phenotype. Consequently, whereas microcin H47 exerts its activity by inhibiting the ATP synthase (45), MccPDI activity requires the presence of a functional ATP synthase in the susceptible cell.

In summary, a preliminary screen of the Keio collection identified six gene knockouts that were no longer susceptible to MccPDI activity. We show that E. coli OmpF is necessary to confer susceptibility to MccPDI through complementation experiments, independent gene knockout experiments (E. coli O157:H7), and heterologous expression of the recombinant OmpF in Salmonella and Yersinia. We speculate that DsbA and DsbB are required for proper folding and/or trafficking of OmpF for folding MccPDI upon entry into the susceptible cell. OmpR is a known transcriptional regulator of ompF. Finally, a functional ATP synthase is required for a strain to be susceptible to MccPDI activity.