ABSTRACT
Prophages switch from lysogenic to lytic mode in response to the host SOS response. The primary factor that governs this switch is a phage repressor, which is typically a host RecA-dependent autocleavable protein. Here, in an effort to reveal the mechanism underlying the phenotypic differences between the Salmonella temperate phages SPC32H and SPC32N, whose genome sequences differ by only two nucleotides, we identified a new class of Podoviridae phage lytic switch antirepressor that is structurally distinct from the previously reported Sipho- and Myoviridae phage antirepressors. The SPC32H repressor (Rep) is not cleaved by the SOS response but instead is inactivated by a small antirepressor (Ant), the expression of which is negatively controlled by host LexA. A single nucleotide mutation in the consensus sequence of the LexA-binding site, which overlaps with the ant promoter, results in constitutive Ant synthesis and consequently induces SPC32N to enter the lytic cycle. Numerous potential Ant homologues were identified in a variety of putative prophages and temperate Podoviridae phages, indicating that antirepressors may be widespread among temperate phages in the order Caudovirales to mediate a prudent prophage induction.
INTRODUCTION
Bacteriophages (phages), which are natural viral predators of bacteria, multiply by infecting specific host bacteria. Although there is an additional type of phage-host relationship called “steady-state infection,” which is exemplified by filamentous phages (1), phage genome replication generally occurs via two different developmental paths: the lytic cycle and the lysogenic cycle. In contrast to the lytic cycle, which results in immediate bursting of the host bacteria and the release of bacteriophage progeny, the lysogenic cycle involves the maintenance of the phage genome as a part of the host genome for several generations, typically by integrating into host chromosomes or, more rarely, by replicating as low-copy-number phage plasmids (2–4). The expression of genes necessary for progeny production and host cell lysis is tightly repressed by a phage regulatory system, but some physiological changes in the host induced by UV light irradiation or other DNA-damaging agents activate the lytic cycle by disabling the phage repressor. Phages fall into two categories: virulent phages that replicate strictly by the lytic cycle and temperate phages that can enter both the lytic cycle and the lysogenic cycle.
The lytic switch following lysogenic development has been well studied in the temperate phage lambda. In the lambda lysogenic phase, phage CI repressors form dimers and bind to specific operators to prevent expression of lambda early genes and subsequent late genes (5, 6). Upon host DNA damage, the activated host RecA protein induces CI proteolysis in a manner similar to the inactivation of the host SOS response regulator LexA (7–9). CI proteolysis leads to the expression of early and late genes, resulting in lytic development. This mechanism illustrates how lambda and other similar phages exploit the host cell SOS response to escape quickly from a potentially damaged host using the RecA-dependent cleavable repressor. Alternatively, some phages in the families Sipho- and Myoviridae utilize the LexA-regulated antirepressors instead of the cleavable repressor to associate their lytic switch to the host SOS response (10–12).
Here, in an effort to identify the factor(s) that causes a phenotypic difference between two very similar podoviral Salmonella phages, SPC32H and SPC32N, we found a novel Podoviridae phage lytic switch antirepressor. We observed that a single nucleotide change in the LexA-binding site, which overlaps with the promoter of the phage antirepressor gene, causes constitutive expression of the antirepressor Ant and consequent inhibition of phage repressor function in SPC32N. As a result, SPC32N could not establish lysogeny as clear plaque mutants. A LexA-dependent lytic switch involving an antirepressor, rather than repressor proteolysis, has been found previously in only sipho- and myoviral phages (10–12), and the podoviral SPC32H/N Ant protein had no significant homology to these known antirepressors. A database search identified many proteins with homology to Ant, suggesting the extensive use of antirepressor-mediated lytic induction among temperate phages in the order Caudovirales.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.The bacterial strains and plasmids used in this study are listed in Tables 1 and 2, respectively. All Salmonella mutants were derived from the prophage-cured Salmonella enterica serovar Typhimurium strain LT2 [referred to as LT2(c)] and its ΔLT2gtrABC1 (SR5003) derivative to exclude the effect of prophages and spontaneous phage resistance via O-antigen glucosylation, respectively (13, 14). Standard cloning procedures were used to construct the recombinant plasmids. Bacteria were grown aerobically at 37°C in LB medium supplemented with the following chemicals, as needed: ampicillin (Ap), 50 μg ml−1; kanamycin (Km), 50 μg ml−1; chloramphenicol (Cm), 25 μg ml−1; 5-bromo-4-chloro-3-indolyl-β-d-galactoside (X-Gal), 40 μg ml−1; l-arabinose, 0.2% (final concentration); isopropyl-β-d-thiogalactopyranoside (IPTG), 1,000 μM (final concentration); and mitomycin C (MMC), 1 μg ml−1 (final concentration). For the disc diffusion assay, 6-mm-diameter filter paper discs were soaked with 10 μl of arabinose, antibiotics, or MMC at the indicated concentrations, placed on the surface of the bacterium-inoculated solidified soft top agar (LB supplemented with 0.4% [wt/vol] agar and X-Gal, if necessary), and incubated at 37°C for 8 h.
Bacterial strains and bacteriophages used in this study
Plasmids used in this study
Bacteriophage.The bacteriophages used in this study are listed in Table 1. All phage mutants were derived from the temperate phage SPC32H, which was previously isolated from chicken fecal samples obtained from a traditional marketplace in South Korea (15). Routine phage spotting and double-agar overlay assays were conducted to determine the efficiency of plating (EOP) in specific bacteria (14, 15). For the morphological analysis, the phage stocks were negatively stained with 2% uranyl acetate (pH 4.0) as previously described (15) and were examined by transmission electron microscopy (LEO 912AB TEM; Carl Zeiss, Jena, Germany) at 120-kV accelerating voltage. The images were scanned with a Proscane 1,024 × 1,024-pixel charge-coupled device camera.
Bacteriophage genome sequencing and analysis.Phage nucleic acids that were extracted by the phenol-chloroform extraction method with protease K/SDS treatment (16) were pyrosequenced using the GS FLX Titanium system by Macrogen, Seoul, South Korea. The quality-filtered reads were assembled using the GS de novo assembler (v. 2.60), and the open reading frames (ORFs) that encode proteins of more than 35 amino acids in size were predicted using the software programs GeneMarkS (17), Glimmer 3.02 (18), and FgenesB (Softberry, Inc., Mount Kisco, NY, USA). The predicted ORFs were annotated based on the results of BLASTP (19), InterProScan (20), and NCBI Conserved Domain Database (21) analysis. tRNAscan-SE (22) and BPROM (Softberry, Inc.) were used to predict the tRNA sequences and the putative promoter/transcription factor-binding sites, respectively. Genomic comparison at the DNA level was visualized using the program Easyfig (23).
Construction of the Salmonella and phage mutants.The lambda red recombination method was used for in-frame gene deletion (24). To construct the noncleavable LexA protein, a point mutation in lexA {resulting in a G85D mutation in the amino acid sequence [lexA(G85D)]} was generated by lambda red recombination and double homologous recombination-based counterselection, as previously described (14), using the suicide vector pDS132 (25). The SPC32H lysogen [ΔLT2gtrABC1 (32H); SR5100] was isolated by sequential streaking of SPC32H-resistant clones from a lawn of phage-treated ΔLT2gtrABC1 and was verified by PCR amplification of the phage attachment (attR) site. The transcriptional recET::lacZ fusion was constructed using pCE70, as previously described (26, 27). Human influenza virus hemagglutinin (HA) epitope tagging of the specific gene(s) was also accomplished by lambda red recombination using oligonucleotides containing the HA tag sequence.
Phage mutants were induced from the SPC32H lysogen after the gene manipulations described above, with some modifications. Briefly, to generate SPC32H m1, a truncated tailspike gene (tsp::Kmr), which was constructed by lambda red recombination in the SPC32H lysogen, was replaced with the m1-containing tsp gene by double homologous recombination-based counterselection. The presence of the m1 mutation in the induced phage was confirmed by DNA sequencing. Similar methods were used to construct SPC32H m2 and SPC32H m12. The oligonucleotides used in this study are listed in Table 3.
Oligonucleotides used in this study
Bioluminescence reporter assay.The 197-bp fragment upstream of the ant gene in SPC32H (designated Pant_H) or SPC32N (designated Pant_N) was PCR amplified and cloned into pBBRlux (28), resulting in the transcriptional fusion of the operon luxCDABE to the putative ant gene promoter. S. Typhimurium strains harboring this reporter plasmid were cultured in 200 μl of fresh LB broth supplemented with appropriate antibiotics in a 96-well plate. The cellular bioluminescence of the culture and the absorbance at 600 nm (A600) were measured periodically using an Infinite 200 Pro plate reader (Tecan, Männedorf, Switzerland), and the results were expressed in arbitrary relative light units (RLU). To trigger the SOS responses, MMC was added to the culture after a 3-h incubation. The three independent assays with triple technical replications were performed.
Western blot analysis.At the mid-exponential phase, culture of the HA-tagged-gene(s)-containing Salmonella was treated by MMC, and portions of the culture were sampled at the indicated time points. Bacterial cells were harvested by centrifugation and were lysed with the B-Per reagent (Thermo Scientific, Illinois, USA). Soluble proteins (10 μg) from cell lysates were separated by 15% SDS-PAGE and electrotransferred to the polyvinylidene difluoride (PVDF) membrane. HA-tagged proteins and DnaK were detected with anti-HA and anti-DnaK antibodies, respectively. The chemiluminescence signals were developed using the West-Zol plus Western blot detection system (iNtRON Biotechnology, Gyeonggi-do, South Korea) after the goat anti-mouse IgG-horseradish peroxidase (HRP) (Santa Cruz Biotechnology, CA, USA) treatment, and then X-ray film was exposed to chemiluminescent light to detect the signals.
Bacterial two-hybrid assay.Protein-protein interaction was determined by the recovery of adenylate cyclase (CyaA) activity through heterodimerization of fusion proteins in the Escherichia coli BTH101 reporter strain (cyaA mutant) (29). The reporter strain harboring the fusion plasmid pair (e.g., pKT25-rep and pUT18c-ant) was streaked on LB agar supplemented with Km, Ap. and X-Gal or subjected to the β-galactosidase assay (30) to quantitatively measure the interaction.
Purification of proteins, rTEV protease treatment, and analytical size exclusion chromatography.Cultures of E. coli BL21(DE3) harboring pHIS-LexA, -Rep, or -Ant (optical density at 600 nm [OD600] = ∼0.15) were treated by 100 μM IPTG and incubated at 25°C for an additional 4 h. Cells were harvested by centrifugation and lysed in lysis buffer (20 mM Tris [pH 8.0], 500 mM NaCl, and 20 mM imidazole) by sonication on ice. Centrifuged (16,000 × g, 4°C, for 30 min) and filtered (0.22-μm filter; Millipore, Ireland) cell lysate was subjected to nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography (Qiagen, California, USA) according to the manufacturer's protocol with elution buffer (lysis buffer with 250 mM imidazole). The eluted protein was concentrated using a Vivaspin 20 instrument (3,000-molecular-weight cutoff [MWCO] polyethersulfone [PES]; Sartorius, Goettingen, Germany), and the buffer was changed (20 mM [Tris pH 8.0], 500 mM NaCl, and 50% glycerol) using a PD MidiTrap G-25 column (GE Healthcare, Buckinghamshire, United Kingdom). To remove the His6 tag from the purified proteins, recombinant tobacco etch virus (rTEV) protease (1:5 ratios in concentration) was treated for 6 h at 4°C in a cleavage buffer (10 mM Tris [pH 8.0], 150 mM NaCl, 0.5 mM EDTA, and 100 mM dithiothreitol [DTT]). For analytical size exclusion chromatography, a Superdex 200 10/300 GL column (GE Healthcare) was used. The column was equilibrated with a buffer consisting of 500 mM NaCl and 20 mM Tris [pH 8.0], and then purified proteins (500 μl of 0.8 μg μl−1) were loaded on to the column at a flow rate of 0.5 ml min−1.
Electrophoretic mobility shift assay (EMSA).The purified PCR fragments of the ant gene promoter region (APR) was γ-32P labeled using T4 polynucleotide kinase (TaKaRa, Japan). The labeled DNA (approximately 4 nM) was incubated with various concentrations of LexA for 30 min at 37°C in 20 μl of reaction mixture containing 1× binding buffer (10 mM HEPES [pH 8.0], 10 mM Tris [pH 8.0], 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, and 5% glycerol) and 1.1 μg of poly(dI-dC). For determination of Rep binding, various amounts of Rep were incubated for 15 min at 20°C with the 4 nM labeled DNA in the 20-μl reaction mixture. When appropriate, Rep was preincubated with various concentrations of Ant for 30 min at 20°C prior to incubation with labeled DNA. The samples were resolved by 6% native PAGE in 0.5× TBE buffer (45 mM Tris-borate [pH 8.3] and 1 mM EDTA). The gels were vacuum dried, and the radioactivity was analyzed using a BAS2500 system (Fujifilm, Tokyo, Japan).
RESULTS
Phenotypic and genomic characterization of the two related S. Typhimurium phages SPC32H and SPC32N.Previously, we isolated nine phages specific for S. Typhimurium from chicken fecal samples (15). Two of these phages, which originated from the same sample collection, exhibited distinct plaque morphologies on a lawn of S. Typhimurium: one phage (SPC32H) formed turbid plaques surrounded by a halo, but the other phage (SPC32N) formed clear plaques without a halo (Fig. 1A and B). Transmission electron microscopy (TEM) analysis revealed that both phages belonged to the family Podoviridae, since they had an isometric head (∼62.3 nm in diameter) and a short noncontractile tail (∼15.4 nm in length) with tail shaft and tail spikes (Fig. 1C and D). These two phages infected identical repertories of Salmonella strains using the O antigen (O-Ag) of Salmonella as the host receptor (data not shown).
Two similar S. Typhimurium-specific Podoviridae phages, SPC32H and SPC32N, produce morphologically distinct plaques. (A and B) Plaque morphology of SPC32H (A) or SPC32N (B). Dilutions (10 μl) of each phage stock were spotted onto a lawn of the S. Typhimurium LT2(c) ΔLT2gtrABC1 strain (SR5003). (C and D) TEM image of SPC32H (C) or SPC32N (D). Inset at the bottom left of each panel shows the enlarged virion morphology with a black scale bar (50 nm). The white arrow and arrowheads indicate the tail shaft and tail spikes, respectively.
Sequencing of SPC32H and SPC32N revealed that both phages contain 38,689 bp of double-stranded DNA with an identical G+C content of 50.16% and 51 predicted open reading frames (ORFs) with one Arg-tRNA. About half of the ORFs (24 ORFs) were annotated as hypothetical proteins, whereas the other annotated proteins were classified into the following modules: DNA packaging, virion structure morphogenesis, lysogenic conversion, host lysis, and DNA replication/recombination (Fig. 2A; see also Table S1 in the supplemental material). The predicted proteins included a phage integrase as well as a putative repressor, indicating that both phages might be temperate phages. BLASTP searches revealed that the SPC32H and SPC32N genomes closely resembled those of Salmonella phage ε15 and other ε15-like phages (31, 32). Indeed, whole-genome comparisons made at the DNA level revealed a significant degree of synteny between the genomes of SPC32H, ε15, and the ε15-like phage phiV10 (Fig. 2A). In particular, 34 out of 51 SPC32H gene products, including a small/large terminase, a head-to-tail joining protein, a putative major coat protein, a putative holin/endolysin, an integrase, a repressor, and a putative DNA replication protein, were highly similar (50 to ∼100% identity at the amino acid level) to those of ε15 (see Table S1). Genes for the putative SPC32H tail structure module (e.g., SPC32H_016, 017, and 018) had higher similarity to those of phiV10 than ε15 (Fig. 2A). Since these phages infect different hosts (i.e., S. Typhimurium for SPC32H, Salmonella enterica serovar Anatum for ε15, and E. coli O157:H7 for phiV10), differences were observed in the genes encoding tailspike proteins and the flanked lysogenic conversion module (which converts O-Ag to prevent superinfection). Taken together, these results suggest that SPC32H and SPC32N should be assigned to the class of ε15-like phages.
There are two single nucleotide differences between the genomes of ε15-like phage SPC32H and SPC32N. (A) DNA alignment of the genomes of phage ε15 (NC_004775.1), SPC32H, and phiV10 (NC_007804.2) using Easyfig. High sequence similarity between the genomes is indicated by the gray regions. SPC32H ORFs are indicated by numbered or annotated arrows. Phage functional modules are indicated under the arrows. ant, antirepressor; tsp, tailspike; oac, o-acetyltransferase; hol, holin; end, endolysin; int, integrase; rep, repressor. Note that the SPC32N genome is identical to that of SPC32H with the exception of two single nucleotide differences (see panel B). (B) Schematic representation of the location of the two single nucleotide differences, m1 and m2. The partial SPC32H genome sequence surrounding the two single nucleotide differences is shown. m1 (located within the tsp gene) and m2 (located in the intergenic region between SPC32H_020 and tsp) are indicated in bold, uppercase letters. The predicted −10 and −35 sites of the putative promoter for SPC32H_020 gene are boxed. The putative LexA-binding site (SOS box) and the putative repressor-binding site are underlined and doubly underlined, respectively. (C) Consensus sequence of the LexA-binding site from E. coli (8, 33, 34) and the putative LexA-binding sites from phage SPC32H and SPC32N. m2 is indicated with a gray background. Note that the LexA-binding site sequences for SPC32H and SPC32N shown here are reverse complements of the sequence shown in panel A.
A single nucleotide change is responsible for the phenotypic difference between the two phages.Interestingly, a comparison of the full genome sequences of SPC32H and SPC32N revealed only two nucleotide differences. One nucleotide difference, designated m1, is located within the tsp gene (SPC32H_021), which encodes a phage tailspike, and the other nucleotide difference, m2, is located in the intergenic region between a gene (SPC32H_020) encoding a hypothetical protein and the tsp gene (Fig. 2B). To verify whether m1, m2, or both single nucleotide differences were responsible for the differences between SPC32H and SPC32N, we mutated the SPC32H sequence to match that of SPC32N. As shown in Fig. 3A, we observed no significant changes in the turbidity of the lysis zone when the SPC32H m1 sequence was changed to that of SPC32N. However, the turbidity decreased dramatically when the SPC32H m2 sequence was replaced by that of SPC32N. We therefore investigated in detail how the single nucleotide difference at the m2 locus leads to this phenotypic difference.
Introducing the m2 sequence from SPC32N induces SPC32H to enter the lytic cycle. (A) High-titer phage stocks (>107 PFU ml−1; 10 μl) of SPC32H, SPC32N, and three mutant phages derived from SPC32H were spotted onto a lawn of S. Typhimurium LT2(c) ΔLT2gtrABC1. (B) SPC32H can lysogenize host Salmonella, whereas SPC32N cannot. Various template samples were PCR amplified with an attR-specific primer pair. M, DNA marker 1 Kb+ (Invitrogen); i, inner part of the lysis zone; e, edge of the lysis zone; gDNA, genomic DNA; ΔLT2gtrABC1(32H), SPC32H lysogen (SR5100). (C) DNA isolated from the lysis zones shown in panel A was PCR amplified using primers specific for the attR site to determine the lysogenization of each phage. Lanes 1 to 5 correspond to each lysis zone shown in panel A. Note that the introduction of m2 resulted in a disappearance of the lysogen-specific attR band (lanes 4 and 5).
Supplementation with the repressor induces the lysogenic development of the lytic cycle-biased phage SPC32N.Because lysogen formation is normally associated with plaque morphology, we investigated the ability of SPC32H and SPC32N to lysogenize. The SPC32H and ε15 integrases have 93% identity at the amino acid level, and both phages contain the highly conserved common core regions and arm-type binding sequences that are required for phage genome integration (31). This suggests that both phages may integrate their genome into the same attachment site, near the end of the Salmonella guaA gene. Therefore, to detect lysogenization by SPC32H or SPC32N, we PCR amplified the right end of the phage genome attachment site (attR site) using a primer pair that specifically anneals to the upstream region of the phage integrase gene (int) and within the guaA gene. The specific attR band was amplified from DNA isolated from the SPC32H lysis zone, whereas no band was detected using DNA isolated from the SPC32N lysis zone (Fig. 3B). The specific attR band was amplified from both colony and genomic DNA from the putative SPC32H lysogen [ΔLT2gtrABC1 (32H)] but not from DNA isolated from the parental Salmonella strain, SPC32H, or SPC32N (Fig. 3B). Furthermore, the SPC32H lysogen spontaneously produced phages that formed halo plaques during prolonged incubation (data not shown). These results clearly demonstrate that Salmonella can be lysogenized by SPC32H but not by SPC32N. The specific attR band was amplified from DNA isolated from the lysis zone of SPC32H m1 but not SPC32H m2 or m12 (Fig. 3C), confirming that m2 is the reason for phenotypic differences between SPC32H and SPC32N.
Because the phage repressor plays a critical role in the maintenance of the lysogenic state by repressing the expression of lytic genes and both phages have a putative repressor gene (rep), we assessed the deficiency of repression in SPC32N. When SPC32H and SPC32N were spotted onto lawns of a Salmonella strain overexpressing rep (ΔLT2gtrABC1 + prep), the EOP of both phages was significantly reduced (<10−5 for SPC32H and <10−2 for SPC32N), and both strains exhibited a more turbid lysis zone than the control (Fig. 4A). Moreover, the specific attR band was amplified from DNA isolated from the lysis zone of both phages (data not shown), suggesting that supplementation with the repressor can promote lysogenic development in SPC32N. These results suggest that SPC32N is defective in maintaining lysogeny, most likely due to an insufficient amount of the active repressor.
The novel antirepressor, encoded by SPC32H_020 (ant), induces the lytic development of SPC32H. (A) Supplementation with the putative repressor leads to the lysogenic development of the lytic cycle-biased phage SPC32N, while supplementation with the putative antirepressor results in the lytic development of SPC32H. Salmonella strains transformed with a control plasmid (pBAD24), a putative repressor-overexpressing plasmid (prep), or an SPC32H_020-overexpressing plasmid (pant) were infected with serially diluted (10-fold) stocks of SPC32H or SPC32N. l-Arabinose (0.2%, final concentration) was added to induce SPC32H_020 expression from pant. (B) The expression of the SPC32H_020 protein promotes the switch from lysogenic to lytic development. The SPC32H lysogen [ΔLT2gtrABC1 (32H); SR5100] and nonlysogen (ΔLT2gtrABC1; SR5003) strains were transformed with pant or a control plasmid (pBAD24), and the resulting strains were subjected to a disc diffusion assay with 10 μl of 15% l-arabinose. pant* indicates the plasmid encoding a frame-shifted ant gene. Arabinose-induced bacterial lysis was observed only in the SPC32H lysogen harboring pant.
A novel antirepressor encoded by SPC32H_020 governs the lytic switch.The m2 mutation is located 24 bp upstream of the start codon of the hypothetical protein SPC32H_020, suggesting that m2 may cause the observed phenotypic differences by affecting the expression of this protein. SPC32H_020 is a small, 86-amino-acid protein with no known conserved domain or motif. A BLASTP search identified 40 hypothetical proteins with more than 64% identity with SPC32H_020 but did not identify any protein with a known function. The proteins exhibiting high homology to SPC32H_020 were from members of the Enterobacteriaceae, such as E. coli, Salmonella spp., Klebsiella spp., Citrobacter spp., and Cronobacter spp., and from ε15-like phages, including ε15, TL-2011b, phiV10, SPN1S, and SPN9TCW (see Table S2 in the supplemental material). Interestingly, some larger proteins (>218 amino acids) with a relatively low identity (<42%) were annotated as putative antirepressors, suggesting the possibility of an antirepressor role for SPC32H_020.
To determine whether SPC32H_020 functions as an antirepressor, we measured the prophage induction efficiency from a Salmonella strain harboring a SPC32H mutant which lacks the gene SPC32H_020 [ΔLT2gtrABC1 (32H Δant) strain]. Compared with results for the WT phage lysogen, the spontaneous induction rate of the mutant phage lysogen was significantly lower (ca. 6 × 10−6-fold lower than that of the WT phage) (Table 4), indicating the critical role of SPC32H_020 in normal prophage induction. Furthermore, MMC treatment did not notably enhance induction of the mutant phage (1.11-fold increase) but did cause a 78.65-fold increase in induction of the WT phage (Table 4), suggesting a potential network between the SPC32H_020 gene and the host SOS response. Because the EOPs in Salmonella of the WT and mutant phage were similar (1.8 × 107 and 4.0 × 107 PFU ml−1, respectively), these results suggest that SPC32H_020 might act as an antirepressor. The function of the SPC32H_020 gene was further tested by a phage spotting assay using Salmonella harboring a plasmid overexpressing SPC32H_020 from a pBAD promoter (pant). As expected, both SPC32H and SPC32N generated clear lysis zones/plaques (Fig. 4A), and the lysogen-specific attR band was not PCR amplified from DNA isolated from the SPC32H lysis zone (data not shown). To verify the function of SPC32H_020 in lytic switching and prophage induction, an arabinose disc diffusion assay was conducted using Salmonella (ΔLT2gtrABC1) and the SPC32H Salmonella lysogen [ΔLT2gtrABC1 (32H)], both harboring pant. The SPC32H lysogen carrying pant underwent lysis in the presence of 15% arabinose, but no lysis was observed in the absence of pant (Fig. 4B). When the arabinose-inducible plasmid contained a frame-shifted ant gene (ant*), which was generated by inserting an additional adenine directly downstream from the SPC32H_020 start codon, the arabinose treatment did not induce lysis (Fig. 4B), suggesting that lytic switching is induced by the SPC32H_020 protein rather than the RNA. Taken together, our results strongly suggest that SPC32H_020 encodes a novel antirepressor protein that plays a significant role in the switch from the lysogenic cycle to the lytic cycle. We have annotated the SPC32H_020 gene as ant (antirepressor) and its gene product as Ant.
Comparison of prophage induction efficiencies
The m2 sequence in the SOS box causes constitutive expression of the antirepressor by SPC32N.The results described above indicate that the m2 mutation may allow the overexpression of ant in SPC32N. Using the BPROM software program, the −10 and −35 sites of the putative ant promoter and one LexA-binding site (SOS box), which overlaps the predicted −10 site, were predicted in the upstream region of the ant gene (Fig. 2B). Intriguingly, m2 is located in the consensus LexA-binding site sequence (8, 33, 34) (Fig. 2C). LexA is a transcriptional repressor that represses various SOS regulons, including LexA itself and the RecA protein, via binding to the SOS box. DNA damage induces the formation of activated RecA nucleoprotein filaments that promote autocleavage of LexA and consequent derepression of SOS regulons (8). Therefore, we hypothesized that the ant gene is an SOS regulon controlled by LexA and that m2 in the consensus LexA-binding site sequence might prevent the LexA-mediated repression of the ant gene in SPC32N.
To test this hypothesis, we first examined the promoter activity of the ant gene in both SPC32H and SPC32N via a bioluminescence reporter assay using luxCDABE. In contrast to the low number of RLU (relative light units) detected using the ant promoter from SPC32H (Pant_H), the promoter from SPC32N (Pant_N) exhibited approximately 2-log higher values (Fig. 5A and B). Treatment with MMC significantly increased the RLU produced from a clone harboring pPant_H::lux but not from a clone harboring pPant_N::lux (Fig. 5A and B), suggesting that Pant_H expression was activated by DNA damage, whereas Pant_N was expressed constitutively and independent of DNA damage. To elucidate whether these responses were associated with LexA, we constructed Salmonella mutants without the lexA gene or expressing a noncleavable form of LexA [lexA(G85D)] and measured the bioluminescence from the reporter plasmid pPant_H::lux. Both mutants were constructed in a ΔsulA background to suppress the lethality of the lexA deletion (35). This sulA deletion did not affect reporter gene expression (data not shown). In the absence of lexA, Pant_H activity was comparable to that observed in the lexA+ background in the presence of MMC, and the Pant_H activity was not affected by MMC treatment (Fig. 5B, ΔlexA). In addition, the lexA+ phenotype was partially rescued by in trans complementation of lexA (data not shown). In contrast, replacing LexA with the noncleavable form of LexA prevented promoter activation by MMC [Fig. 5B, lexA(G85D)], indicating that DNA damage activates the ant promoter through LexA proteolysis. The results were similar regardless of the presence of the SPC32H prophage [Fig. 5B, lexA+ versus lexA+(32H)], suggesting that no other factors, including the SPC32H repressor, are involved in ant gene regulation, despite the presence of a repressor-binding site immediately upstream of the ant promoter (Fig. 2B; also see below).
The ant promoter of SPC32H is activated by DNA damage via LexA proteolysis, whereas the SPC32N ant promoter is constitutively active, due to the inability of LexA to bind to the m2-containing consensus LexA-binding site. The RLU (relative light units) were calculated by dividing the measured bioluminescence by the A600 value. The mean and SD for three independent assays are shown on a log scale on the y axis (A and B). (A) Time course observation of ant promoter activity in the presence or absence of DNA damage. Salmonella strains harboring the bioluminescence reporter plasmid pPant_H::lux (luxCDABE fused to the putative ant promoter of SPC32H) or pPant_N::lux (luxCDABE fused to the putative ant promoter of SPC32N) were incubated at 37°C, and the bioluminescence, as well as the A600 of the culture, was measured every half-hour. The vertical arrows indicate MMC treatment (1 μg ml−1, final concentration; 3 h after incubation). (B) ant promoter activities of the various Salmonella strains at an A600 of ∼0.6, harboring the bioluminescence plasmid. MMC (1 μg ml−1, final concentration) was added after 3 h of incubation. lexA+, ΔLT2gtrABC1, SR5003; lexA+(32H), ΔLT2gtrABC1(32H), SR5100; ΔlexA, ΔLT2gtrABC1 ΔsulA ΔlexA, SR5158; lexA(G85D), ΔLT2gtrABC1 ΔsulA lexA(G85D), SR5176. ∗∗∗, P < 0.001. (C) LexA specifically binds to the putative ant gene promoter region of SPC32H but not to that of SPC32N, which contains m2. The γ-32P-labeled DNA fragment of the ant gene promoter region from SPC32H (APRH*) or from SPC32N (APRN*) was incubated with the indicated amounts of purified Salmonella LexA and was subjected to an electrophoretic mobility shift assay (EMSA). Corresponding unlabeled DNA fragments (APRH and APRN) were used for the competition analysis. The position of the unbound fragments (F) and fragments retarded by LexA binding (B) are indicated.
We next performed an electrophoretic mobility shift assay (EMSA) to show the binding of LexA to the SOS box within Pant_H or Pant_N. When the radiolabeled DNA fragment APRH* (ant gene promoter region from SPC32H) was incubated with an increasing amount of purified Salmonella LexA, a specific mobility shift was observed, and the APRH* fragment was released by the addition of the unlabeled competing cold probe APRH (Fig. 5C, lanes 1 to 8). In contrast, the unlabeled cold probe APRN (ant gene promoter region from SPC32N) was unable to compete with APRH* for LexA, and the APRN* fragment was not shifted in the presence of LexA (Fig. 5C, lanes 9 to14), confirming that LexA cannot repress ant expression in SPC32N due to an inability to bind to the SOS box containing m2.
Based on these results, we investigated the overall cascade of SPC32H induction using Salmonella strains lysogenized by a derivative of SPC32H containing lacZ transcriptionally fused to the putative recET genes. Because the phage recE and recT gene products, a 5′ → 3′ exonuclease and a single-strand DNA binding/annealing protein, respectively, promote homologous recombination to mediate the integration/excision of phage genome to/from the host chromosome (36–38), the expression of recET (and its orthologous genes) can be used as a reporter for prophage induction. As shown in Fig. 6, treatment with MMC but not other antibiotics activated the recET::lacZ fusion in the lexA+ background but not in the lexA(G85D) background, indicating that the DNA damage generated by MMC induces SPC32H induction dependent on LexA proteolysis. The expression of Ant clearly resulted in lysogen-specific lysis (Fig. 4B), supporting the idea that derepression of the ant gene via MMC-induced LexA proteolysis leads to phage induction. Taken together, these results demonstrate that the ant gene of SPC32H is negatively regulated by LexA and that m2 in the SOS box causes the dramatic phenotype differences between SPC32H and SPC32N by influencing ant expression.
DNA damage-induced LexA proteolysis followed by SPC32H ant expression induces the switch to lytic development. The lacZ gene, transcriptionally fused to the putative recET genes, was introduced into the SPC32H lysogens harboring an intact (lexA+) or noncleavable [lexA(G85D)] LexA, and the resulting strains were subjected to a disc diffusion assay with the following solutions: MMC, 0.5 mg ml−1 mitomycin C; Cm, 2.5 mg ml−1 chloramphenicol; Ap, 10 mg ml−1 ampicillin; D.W., distilled water. Note that the blue zone appears to surround the MMC disc in the lexA+ background only.
The antirepressor Ant interacts directly with the cognate repressor Rep.The putative repressor from SPC32H (designated Rep), encoded by SPC32H_041, is a 198-amino-acid protein that contains a helix-turn-helix motif. The RecA-mediated autocleavage site (Ala-Gly or Cys-Gly), a highly conserved site in cleavable repressors such as lambda CI (39), is not present in SPC32H Rep, strongly supporting the notion that SPC32H prophage induction involves the inhibition of Rep through means other than autocleavage assisted by RecA nucleofilaments. The immunodetection of HA epitope-tagged Rep demonstrated that the expression level of Rep remained virtually constant (i.e., was not cleaved) throughout a 1-h treatment with MMC (Fig. 7A). The lytic switch was activated by MMC in this experiment, as shown by the fact that HA-tagged Ant was expressed and accumulated after treatment with MMC (Fig. 7A, lower panel). The HA-tagged versions of the Rep and Ant proteins are fully functional (data not shown).
DNA damage induces Ant accumulation but not Rep cleavage, and the consequent binding of Ant to Rep inhibits the binding of Rep to specific operators. (A) Salmonella strains lysogenized by SPC32H expressing HA-tagged Rep (upper panel; ΔLT2gtrABC1 [32H rep-HA], SR5192) or both HA-tagged Rep and HA-tagged Ant (lower panel; ΔLT2gtrABC1 [32H rep-HA ant-HA], SR5197) were exposed to MMC for 1 or 2 h, respectively. The MMC-treated bacterial cultures were sampled at the indicated time points and subjected to the Western blotting to immunodetect the HA-tagged proteins. DnaK was used as an internal control. (B) Bacterial two-hybrid assays revealed the direct binding of Ant to Rep. The β-galactosidase activity of E. coli BTH101 reporter strains harboring the indicated plasmid pairs were measured. The activities are presented in Miller units. B, a backbone plasmid. (C) EMSA with purified Rep and Ant demonstrates the Ant-mediated inhibition of Rep binding to its operators. Mixtures of APRH* and the indicated amounts of Rep were incubated at 20°C for 15 min in 1× binding buffer supplemented with 1.1 μg of poly(dI-dC) and then electrophoresed on a 6% native acrylamide slab gel for EMSA. For competition analysis, unlabeled APRH fragments were added as cold probes to the mixture. When appropriate, Rep was preincubated with the indicated amounts of Ant at 20°C for 30 min and further incubated with APRH* as described above. The positions of the unbound fragments (F) and fragments retarded by Rep binding (B1 and B2) are indicated.
To explore the possible interaction between Rep and Ant, we performed a bacterial two-hybrid assay based on the restoration of β-galactosidase activity in E. coli cyaA mutant strain BTH101 (29). The reporter strain E. coli BTH101 expressing the combination of hybrid proteins (i.e., T25-Rep/T18-Ant or T25-Ant/T18-Rep) produced blue colonies on X-Gal plates (data not shown) and exhibited a significantly higher (approximately 10- to 50-fold) level of β-galactosidase activity than the negative control (i.e., E. coli BTH101 expressing the unfused T18 and T25 peptides) (Fig. 7B), indicating a heterodimerization of the hybrid proteins via interaction between Rep and Ant. Strong Rep-Rep and Ant-Ant interactions were also observed, implying the possibility of multimerization by each protein. Indeed, the results of analytical size exclusion chromatography demonstrated that Rep and Ant were able to dimerize and tetramerize, respectively (data not shown). Taken together, these results suggest that Rep and Ant can interact with themselves and each other.
EMSA using purified Rep and Ant revealed that Ant inhibits Rep target site binding. The high homology (97% identity) between SPC32H Rep and the ε15 repressor suggests that Rep may recognize the same DNA sequence (5′-ATTACCnnnnGGTAAT −3′) as the ε15 repressor. Radiolabeled APRH*, which includes the putative repressor-binding site as well as the SOS box, was also used in this assay. Two DNA-protein complex bands with different mobilities appeared when purified Rep was incubated with APRH* (Fig. 7C, lanes 1 to 4), suggesting that APR may have two Rep-binding sites with different affinities for Rep. A competition assay using a nonlabeled cold probe demonstrated the specificity of Rep-binding for APR (Fig. 7C. lanes 5 and 6). Notably, preincubation of Rep with purified Ant prevents the mobility shift of the APRH* fragment in an Ant concentration-dependent manner (Fig. 7C, lanes 7 to 9), suggesting that the specific interaction between Ant and its cognate repressor Rep interferes with Rep binding to its target DNA. Note that the protein concentrations indicated were calculated based on the assumption that the Rep and Ant stocks consisted entirely of active dimers and tetramers, respectively. The APRH* fragment clearly did not exhibit a mobility shift when incubated with Ant alone (Fig. 7C, lane 10), excluding the possibility that Ant inhibits Rep activity by competing for the Rep-binding site with Rep.
DISCUSSION
The goal of this study was to determine the cause for the phenotypic differences between two highly similar podoviral ε15-like phages, SPC32H and SPC32N. We detected two nucleotide differences between the two phage genomes, but only one, located within a noncoding region, was responsible for the phenotypic differences. This nucleotide polymorphism, m2, was located within a consensus LexA-binding site sequence that overlaps the −10 site of the promoter for SPC32H_020, which encodes a hypothetical protein (Fig. 2). This sequence difference prevents LexA from binding to its binding site, allowing the constitutive expression of a small hypothetical protein (Fig. 5), which we have identified as a novel antirepressor of the family Podoviridae. This antirepressor inhibits the binding of its cognate phage repressor to regulatory regions (Fig. 7C), resulting in a switch of the phage life cycle from lysogenic to lytic.
To date, at least two categories of lytic switch antirepression systems have been identified in temperate phages. The first system is represented by the Cro protein of several lambdoid phages, such as phage lambda, HK022, and HK97. In this system, the binding of the Cro protein to target operator sites prevents expression of the cI gene, which encodes the phage repressor CI (40). In contrast, the second system controls repressor activity at the protein level. For example, the antirepressor Tum from myoviral coliphage 186 binds directly to the phage repressor CI, preventing CI from binding to its operator sites (12). Notably, the latter system has been found in only a few temperate phages, including siphoviral coliphage N15 (11) and the siphoviral prophages Gifsy-1 and Gifsy-3 identified in S. Typhimurium strain 14028 (10). In the present study, we have elucidated the mechanism and regulation of an antirepressor, which belongs to the second category of lytic switch antirepression systems and is the first example of this type of system in the Podoviridae family of the order Caudovirales. A notable common feature of this second system type is the LexA-regulated initiation of antirepressor expression. Although the phage P22 also produces an antirepressor that inactivates the c2 repressor and prevents RecA-dependent c2 proteolysis (41, 42), it is unknown whether LexA regulates the expression of the antirepressor. However, the presence of a consensus LexA-binding sequence 38 bp upstream of the start codon of the antirepressor protein suggests that LexA may be involved in the regulation of the P22 antirepressor and consequently that the P22 antirepressor may be a member of the second category of antirepression systems.
Linking the host SOS response to the lytic switch is a fundamental strategy used by prophages to escape from damaged host cells. Compared to the RecA-dependent cleavable repressor system, such as the lambda CI (7, 40), the antirepression system appears to be more advantageous to the prophages. If host bacteria are able to repair DNA damage before prophage induction and survive (43), it would be more beneficial for the prophages to remain in the host cell. If lysis occurred, the induced phages would need to reestablish the prophage state in new host bacterial cells to stably maintain their genome as a part of a host genome. This superfluous step could easily be prevented by expressing the antirepressor in a LexA cleavage-dependent manner. Since antirepressor levels are reduced by the replenished LexA pool, lysogenic development could resume because inactivated, rather than degraded, repressors can be restored to function by dissociating from the antirepressor. Although we did not demonstrate the reversible binding of Ant and the recovery of Rep activity after Ant dissociation in the present study, Rep levels were stably maintained without degradation during MMC treatment (Fig. 6), suggesting that recycled Rep could be used in the resumed SPC32H lysogenic development. Indeed, the antirepressor Tum/repressor CI pair from coliphage 186 exhibits reversible Tum binding and the recovery of CI activity after dissociation from Tum (12). We are currently attempting to elucidate this issue by investigating the structure of the Rep-Ant complex as well as the individual proteins.
Considering the advantages of rapid resumption of the regulatory circuit, it is possible that this type of repressor/antirepressor system is widespread among the temperate phages. Remarkably, several homologues of SPC32H Ant (38 to 100% amino acid similarities) were identified in other Podoviridae phages and various bacteria in the family Enterobacteriaceae (see Table S2 in the supplemental material), most likely as a gene product of unknown function of prophages. Moreover, the phage antirepressors Tum (from Myoviridae coliphage 186), AntC (from Siphoviridae coliphage N15), GfoA (from Siphoviridae phage Gifsy-1), and Ant (from Podoviridae phage SPC32H) are distinct from each other at the amino acid sequence level (Fig. 8), suggesting that diverse repressor/antirepressor pairs are present in the order Caudovirales to allow for more prudent control of lytic/lysogenic switching. Therefore, as suggested by Mardanov and Ravin, the cleavable repressor system may not be the exclusive mechanism for lytic/lysogenic regulation in temperate phages (11). As recently illustrated by Lemire et al., antirepressors can mediate cross talk between prophages in polylysogenic strains (10). Thus, further studies regarding the trans activity of diverse phage antirepressors, including SPC32H Ant, would provide insight into the coordinated behavior of temperate phage subversion of their bacterial prey.
Amino acid alignment of the phage antirepressors. The amino acid sequences of Tum (from coliphage 186), AntC (from coliphage N15), GfoA (from Gifsy-1) and Ant (from SPC32H) were aligned using ClustalW2. There are no noticeable consensus residues, demonstrating the diversity of phage antirepressors in the order Caudovirales.
ACKNOWLEDGMENT
This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Ministry of Education, Science and Technology (no. 20090078983).
FOOTNOTES
- Received 6 August 2013.
- Accepted 19 August 2013.
- Accepted manuscript posted online 28 August 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JVI.02173-13.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
