Bacteriophage ϕMAM1, a Viunalikevirus, Is a Broad-Host-Range, High-Efficiency Generalized Transducer That Infects Environmental and Clinical Isolates of the Enterobacterial Genera Serratia and Kluyvera

Abstract

Members of the enterobacterial genus Serratia are ecologically widespread, and some strains are opportunistic human pathogens. Bacteriophage ϕMAM1 was isolated on Serratia plymuthica A153, a biocontrol rhizosphere strain that produces the potently bioactive antifungal and anticancer haterumalide oocydin A. The ϕMAM1 phage is a generalized transducing phage that infects multiple environmental and clinical isolates of Serratia spp. and a rhizosphere strain of Kluyvera cryocrescens. Electron microscopy allowed classification of ϕMAM1 in the family Myoviridae. Bacteriophage ϕMAM1 is virulent, uses capsular polysaccharides as a receptor, and can transduce chromosomal markers at frequencies of up to 7 × 10⁻⁶ transductants per PFU. We also demonstrated transduction of the complete 77-kb oocydin A gene cluster and heterogeneric transduction of a plasmid carrying a type III toxin-antitoxin system. These results support the notion of the potential ecological importance of transducing phages in the acquisition of genes by horizontal gene transfer. Phylogenetic analyses grouped ϕMAM1 within the ViI-like bacteriophages, and genomic analyses revealed that the major differences between ϕMAM1 and other ViI-like phages arise in a region encoding the host recognition determinants. Our results predict that the wider genus of ViI-like phages could be efficient transducing phages, and this possibility has obvious implications for the ecology of horizontal gene transfer, bacterial functional genomics, and synthetic biology.

INTRODUCTION

Bacteriophages (phages) are ubiquitous obligate viral parasites of bacteria that reproduce in concert with their hosts in diverse natural environments (1, 2). Bacteriophage abundance correlates with bacterial population densities in a given niche, and the global phage population has been estimated to be over 10³¹ phages, reflecting over 10²⁵ infections per second (3, 4). This intimate interaction between phages and their hosts results in rapid coevolution, which has been observed under both natural environmental and laboratory conditions (5–7).

Since their discovery in the early 1900s, phages have been used as therapeutic agents in human infections (8, 9) but also as alternative biocontrol agents in agriculture, sewage treatment, and the food industry (10–13). Furthermore, phage display technology has proved useful in cancer studies and for the delivery of vaccines (12, 13). Additionally, transducing phages have been important tools in basic bacterial genetics, functional genomics, and synthetic biology applications (13–15).

Together with transformation and conjugation, transduction is one of the three main classes of horizontal gene transfer (HGT) in bacteria. In generalized transduction, phages accidentally package large fragments of host bacterial DNA instead of the viral genome. The generated transducing particles adsorb normally and inject DNA into a recipient bacterium. The injected bacterial DNA may integrate by homologous recombination into the genome of the recipient host, resulting in a stable bacterial transductant (14, 16). Transducing phages have been used extensively in genetic manipulations, including bacterial strain engineering, transposon mutagenesis, and plasmid transfer (14, 16).

Bacteria of the genus Serratia belong to the family Enterobacteriaceae. Serratia species are widely distributed, being commonly found in water, soil, plants, animals, and humans; some species are opportunistic human pathogens (17). Although Serratia plymuthica strains have been isolated from diverse environments, they are mostly associated with plants and are considered to be plant growth-promoting bacteria and excellent biocontrol agents for many fungal and oomycete plant diseases (18). Serratia plymuthica A153 was isolated from the rhizosphere of wheat (19) and produces several halogenated macrolides, haterumalides (20). Haterumalides—also known as oocydins—were among the first polyketides found to be synthesized by Serratia species (21), and they have been shown to exhibit anticancer, antifungal, antioomycete, and antihyperlipidemic properties (20–23).

Recently, we isolated and sequenced the new ViI-like Serratia phage ϕMAM1 (24). Here, we analyze the genome and report the morphological and biological characterization of ϕMAM1, a generalized transducing phage that is able to infect many environmental and clinical isolates from the Serratia and Kluyvera genera.

MATERIALS AND METHODS

Bacterial strains, plasmids, phages, culture media, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1 and in Table SA1 in the supplemental material. Serratia, Kluyvera, and their derivative strains were routinely grown at 30°C, unless otherwise indicated, in Luria broth (LB; 5 g yeast extract liter⁻¹, 10 g Bacto tryptone liter⁻¹, 5 g NaCl liter⁻¹), potato dextrose broth (24 g potato dextrose broth liter⁻¹), or minimal medium [0.1% (wt/vol) (NH₄)₂SO₄, 0.41 mM MgSO₄, 0.2% (wt/vol) glucose, 40 mM K₂HPO₄, 14.7 mM KH₂PO₄, pH 6.9 to 7.1]. Escherichia coli strains were grown at 37°C in LB. Escherichia coli DH5α was used for gene cloning. Media for the propagation of E. coli β2163 were supplemented with 300 μM 2,6-diaminopimelic acid (DAPA). When appropriate, antibiotics were used at the following final concentrations (in μg ml⁻¹): ampicillin, 100; kanamycin, 25 (E. coli strains) and 75 (Serratia strains); streptomycin, 50; and tetracycline, 10. Sucrose was added to a final concentration of 10% (wt/vol) when required to select derivatives that had undergone a second crossover event during marker exchange mutagenesis. Unless indicated otherwise, the growth temperature for S. plymuthica A153, 30°C, was used for the phage incubations. Phages were stored at 4°C in phage buffer (10 mM Tris-HCl, pH 7.4, 10 mM MgSO₄, 0.01%, wt/vol, gelatin) over a few drops of NaHCO₃-saturated chloroform.

TABLE 1.

ϕMAM1 host range

Strain	Relevant characteristics	Source or reference	EOPa	Plaque morphology	Avg plaque size (mm)d
Serratia plymuthica A153	Rhizosphere isolate, oocydin A positive	19	1	Clear	1.5
Serratia marcescens 1695	Clinical isolate, nonpigmented	42	—b	Turbid	—
S. marcescens 0006	Environmental isolate, nonpigmented	42	0.6c	Turbid	0.1
S. marcescens 1047	Clinical isolate, nonpigmented	42	0.5c	Turbid	0.3
S. marcescens 0026	Environmental isolate, nonpigmented	42	1.2	Clear	0.7
S. marcescens 3078	Clinical isolate, pigmented	42	1.2	Clear	1
S. marcescens 0038	Environmental isolate, pigmented	42	0.3	Clear	1
S. marcescens 3127	Clinical isolate, nonpigmented	42	0.3	Turbid	0.5
S. marcescens 0035	Environmental isolate, pigmented	42	0.1	Clear	0.5
S. marcescens 12	Clinical isolate, nonpigmented	42	0.3	Clear	0.1
S. marcescens 0040	Environmental isolate, pigmented	42	1.4	Clear	1.5
S. marcescens 2595	Clinical isolate, nonpigmented	42	—b	Turbid	—
S. marcescens 3127	Clinical isolate, pigmented	42	0.6	Clear	1
S. marcescens 3078V	Spontaneous nonpigmented variant of S. marcescens 3078	42	0.9	Turbid	0.8
S. marcescens ATCC 274	Clinical isolate, pigmented	66	0.5	Clear	0.5
S. marcescens S6	Clinical isolate, nonpigmented	76	0.6	Turbid	0.8
S. marcescens 365	Lab stock, pigmented	L. Debarbieux	6.1 × 10−3	Clear	0.1
S. marcescens 2170	Lab stock, pigmented	T. Watanabe	0.9	Clear	1
Kluyvera cryocrescens 2Kr27	Rhizosphere isolate	39	0.4	Clear	0.3

^aEOP was determined by dividing the phage titer of the ϕMAM1-sensitive strain by that of the wild-type reference strain, Serratia plymuthica A153.

^bNo plaques were observed. Adsorption assays confirmed that ϕMAM1 adsorbs to S. marcescens 1695 and S. marcescens 2595.

^cNo phage plaques were observed at 30°C. Phage plaques arose at 37°C in S. marcescens 0006 and S. marcescens 1047.

^dOn LB agar (0.35%, wt/vol, agar) lawns.

ϕMAM1 phage isolation and phage lysate preparation.

Treated sewage effluent was collected from the sewage treatment plant at Milton, Cambridge, United Kingdom (24). Briefly, a 10-ml sample of the effluent was filter sterilized. Then, 500 μl of the sterilized effluent was mixed with 200 μl of Serratia plymuthica A153 overnight culture and 4 ml of top LB agar (LBA; 0.35%, wt/vol, agar) and poured as an overlay onto LBA plates (1.5%, wt/vol, agar). The plates were incubated overnight at 30°C, and single phage plaques were picked with a sterile toothpick, placed into 0.2 ml phage buffer, and shaken with 20 μl of chloroform to kill any bacteria. The phages obtained were plaque purified three times. High-titer phage lysates were then obtained as described by Petty et al. (25). The phages were titrated by serial dilutions in phage buffer, and the phage titer (in PFU per milliliter) was determined.

ϕMAM1 genome sequencing.

Genomic DNA sequencing was performed using the 454 DNA pyrosequencing technology, as described previously (24). The preparation of the 454 library was done by nebulization using a GS FLX titanium rapid library preparation kit following the manufacturer's instructions (catalog no. 05608228001; Roche). The shearing was performed with nitrogen gas at a pressure of 30 lb/in² (2.1 bars) for 1 min. The assembly used 257,858 reads, or 102.6 MB, of raw data to give a 650× coverage of the genome.

Genome annotation and bioinformatics.

Annotation of the ϕMAM1 genome and identification of tRNAs were performed as described previously (24). Putative bacteriocins were identified using the web-based bacteriocin mining tool BAGEL2 (University of Groningen, Groningen, The Netherlands). Artemis software (Wellcome Trust Sanger Institute) was used to visualize and annotate the ϕMAM1 genome. Genome comparison analyses were performed by employing the Artemis comparison tool and EMBOSS Stretcher (European Bioinformatics Institute). The genomic organization and annotation of the seven previously reported ViI-like enterobacterial bacteriophages (26) were used to determine the location of ORF1 in ϕMAM1. The CGView comparison tool (27) was employed to visualize the ϕMAM1 genome and to generate Fig. 3. Multiple-sequence alignments of phage proteins were performed using the ClustalW2 program (European Bioinformatics Institute). To analyze and identify motifs in promoter regions, 100-bp sequences upstream of the start codons were extracted using the extractUpStreamDNA tool (http://lfz.corefacility.ca/extractUpStreamDNA/) and analyzed using the MEME suite of programs (28). Candidate late promoters in ϕMAM1 were identified by using the T4 late promoter conserved consensus sequence (TATAAATA). Rho-independent transcription terminators were identified by examining the secondary structure of the DNA using the ARNold program (http://rna.igmors.u-psud.fr/toolbox/arnold/index.php). The phylogenetic analyses were performed with MEGA software (v5.10) (29). The ϕMAM1 GenBank submission file was generated using the Sequin program (NCBI). The complete genome sequence of ϕMAM1 is available in GenBank under accession number JX878496 (24).

FIG 3.

Genome map of phage ϕMAM1. The physical locations by strand of all ORFs are shown on the outer ring. The color code representing the functional category of each gene is given where possible. The proposed structural genes (dark brown) are based on previous mass spectrometry analyses (11, 32). The region encoding the host recognition determinants, the tail spike proteins, is indicated with an asterisk. The three bluish and reddish internal rings represent the tBLASTx results against the ViI-like phages Vi01 (outer bluish-reddish ring), ϕSboM-AG3 (middle bluish-reddish ring), and LIMEstone1 (inner bluish-reddish ring). The GC content and the results of GC skew analysis are also shown. The genome map was generated using the CGView comparison tool (27).

Electron microscopy.

High-titer lysates for transmission electron microscopy were obtained as described by Petty et al. (25) using 0.35% (wt/vol) LB agarose instead of 0.35% (wt/vol) LB agar overlays. Twenty-five microliters of high-titer ϕMAM1 phage lysates (≥3 × 10¹⁰ PFU ml⁻¹) were absorbed onto 400-mesh copper grids with holey carbon support films (Agar Scientific, Stansted, United Kingdom). The copper grids were discharged in a Quorum/Emitech K100X system (Quorum, Ringmer, United Kingdom) prior to use. After 1 min, excess phage suspension was removed with filter paper and phage samples were negatively stained by placing the grids in 25-μl drops of 2% ammonium tungstate containing 0.1% trehalose (pH 7.0) or 2% aqueous uranyl acetate (pH 4.0) for 1 min. The grids were then blotted on filter paper to remove the excess solution and allowed to air dry for 10 min. Phages were examined by transmission electron microscopy in the Multi-Imaging Centre (Department of Physiology, Development and Neuroscience, University of Cambridge) using an FEI Tecnai G2 transmission electron microscope (FEI, OR, USA). The accelerating voltage was 120.0 kV, and images were captured with an AMT XR60B digital camera running Deben software.

Host range determination.

A collection of enterobacterial strains was assembled to test their sensitivity to ϕMAM1. Top agar overlays containing 4 ml of 0.35% (wt/vol) LBA and 200 μl of an overnight culture of the bacterial strain to be tested were poured onto a solid LBA plate and allowed to set. A 10-μl sample of a serial dilution of a high-titer phage lysate of ϕMAM1 was spotted onto each bacterial overlay, and the plates were incubated overnight at 30°C or 37°C. Ten microliters of phage buffer was also added as a negative dilution control. After incubation, each plate was scored for lysis. Efficiency of plaquing (EOP) was determined by dividing the phage titer of the ϕMAM1-sensitive strain by that of the wild-type reference strain, Serratia plymuthica A153. EOP assays were performed in triplicate.

Transduction assays.

Phage lysates were prepared on bacterial strains carrying the desired mutation or plasmid. The phages were initially tested for their ability to transduce the TnKRCPN1 transposon from oocJ::TnKRCPN1 and admK::TnKRCPN1 mutations into wild-type S. plymuthica A153. The transduction protocol was optimized on the basis of the incubation time, temperature, multiplicity of infection (MOI), and number of bacterial cells. An appropriate volume of a high-titer transducing lysate was added to a 10-ml overnight culture of the recipient strain (∼1 × 10¹⁰ total bacterial cells) to give the desired MOI. The mixture was incubated at 30°C for 30 min (1 h for S. marcescens 12), pelleted by centrifugation (4,000 × g for 10 min at 4°C), and washed twice with 10 ml of LB to remove any remaining lysate. The cells were resuspended in 5 ml of LB, 100-μl aliquots of the cell mixture were spread onto LBA plates containing the appropriate antibiotic, and the plates were incubated overnight. When the cell mixture was resuspended in small volumes (e.g., 300 μl of LB) and spread onto LBA plates, we observed a reduction in the transduction efficiency due to phage superinfection that caused the lysis of transductant colonies. The transductants obtained were streaked out three times prior to use to reduce or eliminate any bacteriophage carryover. Transductants were confirmed by screening the antibiotic-resistant colonies for coinheritance of a secondary phenotype of either antifungal and antibacterial properties, auxotrophy, or cellulose production (see the information in the supplemental material). During the transduction assays, controls for spontaneous resistance to antibiotics were determined by spreading 100 μl of the overnight culture onto LBA plates containing the relevant antibiotic. A 100-μl volume of the high-titer lysate was also spread on non-LBA plates to confirm lysate sterility. Transduction efficiency was defined as the number of transductants obtained per PFU.

One-step growth curve and burst size.

One-step growth curve experiments were performed as described previously (25). Briefly, an overnight bacterial culture was adjusted to an optical density at 600 nm (OD₆₀₀) of 0.02 in 25 ml LB in a 250-ml flask and grown in a water bath at 30°C with orbital shaking (225 rpm). At an OD₆₀₀ of 0.1, 10 ml of the bacterial culture was removed, placed into a sterile tube, and pelleted by centrifugation at 4,000 × g for 10 min at 4°C. Once the supernatant was removed, the pellet was resuspended in 10 ml of fresh LB and phage samples were added at an MOI of 0.001. The same amount of phages was also added to 10 ml of LB as a negative control. Phages were allowed to adsorb for 5 min at room temperature without shaking, and the supernatant containing any unadsorbed phages was removed by centrifugation at 4,000 × g for 10 min at 4°C. The pellets were resuspended in 10 ml LB, the suspension was added to 15 ml LB in a 250-ml conical flask, and the cultures were grown in a shaking water bath as described above. Every 10 min, samples were taken and immediately titrated to determine the number of PFU. The final growth curve represents the number of phages per initial infectious center.

Phage adsorption.

Phage adsorption experiments were carried out as described previously (25). Briefly, 10-ml overnight cultures of the bacterial host or nonhost bacterial control (Serratia marcescens Db11) were infected with ϕMAM1 at an MOI of 0.01, mixed briefly, and placed on a tube roller at 30°C. A bacterium-free negative control was created by adding the same amount of phage to 10 ml of LB. One-hundred-microliter samples were removed every 5 min and added to 900 μl of phage buffer and 30 μl of chloroform, and the components were mixed for 5 s and centrifuged at 13,000 × g for 1 min. The number of unadsorbed ϕMAM1 phage particles was determined by titrating the supernatant on 0.35% (wt/vol) LBA lawns, as described above. To measure phage adsorption to bacteria, the number of free ϕMAM1 particles per milliliter remaining in the supernatant was quantified and phage adsorption was expressed as a percentage of the number of PFU ml⁻¹ in the bacterium-free negative control.

Transposon mutagenesis and isolation of phage-resistant mutants.

Random transposon mutagenesis of S. plymuthica A153 using TnKRCPN1 was performed as follows. In a biparental conjugal mating, 500-μl volumes of overnight cultures of E. coli β2163(pKRCPN1) and S. plymuthica A153 were mixed, and cells were collected by centrifugation, resuspended in 30 μl of fresh LB, and spotted onto an LB agar plate supplemented with 300 μM DAPA. After overnight incubation at 30°C, the cells were scraped off the plate and resuspended in 1 ml of LB. ϕMAM1-resistant mutants were selected by mixing 100 μl of the resuspended mating patch with 200 μl phage lysate (≥1 × 10¹¹ PFU ml⁻¹) and 4 ml of 0.35% (wt/vol) LBA containing 75 μg ml⁻¹ kanamycin. The mixture was poured onto a solid LBA plate, allowed to set, and incubated overnight at 30°C. DAPA was not added to LBA medium to allow counterselection of the E. coli donor. The site of transposon TnKRCPN1 insertion in the ϕMAM1-resistant mutants was determined using random primed PCR following the method described previously (30).

Supplementary experimental procedures.

Information on the in vitro nucleic acid techniques, construction of the A153 OocKmSm strain, phenotypic analysis of transductants, and toxin-antitoxin abortive infection assays is given in the supplemental material.

RESULTS

Isolation of ϕMAM1.

During environmental screening for new bacteriophages infecting clinical and environmental isolates of Serratia, we isolated 17 new phages infecting the haterumalide-producing strain Serratia plymuthica A153 (24). Isolated, plaque-purified phage lysates of high titer were screened for their ability to transduce TnKRCPN1 chromosomal mutations into the A153 wild-type strain. Of the 17 new phages, ϕMAM1 was the only transducing phage and was selected for further characterization.

ϕMAM1 forms plaques ranging from 6.6 ± 0.9 to 1.2 ± 0.3 mm in diameter when the phage is titrated and plated in 0.2 to 0.8% agar overlays (see Fig. SA1 in the supplemental material). Within the range of temperatures tested, 25 to 37°C, no differences in plaque formation were observed (data not shown). ϕMAM1 viability was unaffected by the addition of chloroform during storage in phage buffer at 4°C, and phage lysates retained high viability during the period of this study, 2 years (>65% of the initial titer).

ϕMAM1 morphology.

Using transmission electron microscopy, the morphology of ϕMAM1 was determined. ϕMAM1 has an icosahedral and isometric head measuring 90 ± 2 nm from flat face to flat face. Its contractile tail measures 120 ± 2 nm long and 21 ± 2 nm in diameter when extended or 57 ± 2 nm long and 26 ± 2 nm in diameter when contracted (Fig. 1). The ϕMAM1 tail consists of an 11- ± 1-nm neck with a collar, a tail tube surrounded by a contractile sheath showing 24 transverse striations, a baseplate, and a complex system of tail spikes (Fig. 1). On the basis of its morphology and according to Ackermann's (31) classification, ϕMAM1 was grouped into the order Caudovirales within the Myoviridae family, which comprises 25% of the tailed bacteriophages and includes the Escherichia coli phage T4. Morphologically, ϕMAM1 is very similar to Salmonella phages Vi01 (32), ϕSH19 (33), and SFP10 (34), Shigella phage ϕSboM-AG3 (35), E. coli phages CBA120 (36) and PhaxI (37), and Dickeya phage LIMEstone1 (11). All these ViI-like enterobacterial bacteriophages have recently been described to be members of a newly proposed genus within the Myoviridae family, Viunalikevirus (26). The adsorption organelle in Viunalikevirus members shows two conformations consisting of prongs and umbrella-like filaments with rounded tips, both of which are associated with the baseplate (26). Both conformations were observed in ϕMAM1 (Fig. 1A and B), and each of the umbrella-like structures showed 4 short filaments attached to the baseplate by a stalk (Fig. 1B).

FIG 1.

Transmission electron micrographs of phage ϕMAM1. (A) Intact virion with folded prong-like structures. (B) Intact virion with unfolded prong-like structures. (Inset) Detail of an unfolded structure attached to the baseplate. (C) Intact virion with contracted tail, deformed head, and TSP structures. (D) ϕMAM1 phage particle with contracted tail, without the adsorption structures and with the central tube exposed. ϕMAM1 particles were stained with uranyl acetate (A) or ammonium tungstate (B to D) as described in Materials and Methods. Bars, 50 nm.

Biological characterization of ϕMAM1.

We evaluated the ability of ϕMAM1 to lyse host bacteria in liquid cultures by adding phages at different MOIs to S. plymuthica A153 cultures. ϕMAM1 infection first caused a reduction in the growth rate, followed by a cessation of bacterial growth 90 min after phage addition. Bacterial lysis caused a decrease in the turbidity of the culture 2 h after phage addition (Fig. 2A). However, a resumption of growth in A153 was observed after a prolonged incubation period (data not shown). This resumption was associated with the emergence of phage-resistant mutants, since A153 isolates from late infected cultures showed ϕMAM1 resistance in agar lawn assays.

FIG 2.

Biological characterization of ϕMAM1. (A) Serratia plymuthica A153 growth curves in the presence and absence of ϕMAM1. Phage was added at MOIs of 0.1 and 1 to A153 in the early exponential phase of growth, as described by Petty et al. (25). Arrow, time of addition of ϕMAM1. (B) Adsorption of ϕMAM1 to A153 and A153 phage-resistant (PR) mutants. The ϕMAM1-insensitive strain Serratia marcescens Db10 was used as a negative control for adsorption. Descriptions of the strains are provided in Table SA1 in the supplemental material, and the locations of the transposon insertion points are shown in Fig. SA5 in the supplemental material. (C) One-step growth curve of ϕMAM1. In all cases, error bars represent the standard deviations (n = 3). t, time.

Phage adsorption assays confirmed that ϕMAM1 adsorbs rapidly to S. plymuthica A153, with more than 95% of the phage particles being adsorbed within the first 5 min. The number of adsorbed phage particles reached an apparent maximum at 30 min postmixing (Fig. 2B), and this time was chosen as the incubation time in the A153 transduction assays (see below).

To determine the burst size of ϕMAM1, one-step growth curves were carried out. ϕMAM1 showed a latent period of ∼20 min and a rise period, where the phage particles were released, of about 40 min. The average burst size was more than 300 phage particles per initial infection center (Fig. 2C).

Host range of the virulent bacteriophage ϕMAM1.

The sensitivity to ϕMAM1 of a collection of 25 environmental and clinical isolates of Serratia marcescens (38) and 30 enterobacterial strains isolated from the rhizosphere of agronomic crops (39) was investigated by spotting phage lysates onto bacterial lawns. Bacterial clearing was observed on agar lawns of 17 S. marcescens strains and in the rhizobacterium Kluyvera cryocrescens 2Kr27 (Table 1; see also Fig. SA1 in the supplemental material). The plaques formed in the sensitive strains varied in size and were clear in most strains but turbid in S. marcescens 0006, 1047, 1695, 2595, 3127, 3078V, and S6. However, subsequent phage titrations showed no obvious plaque formation in S. marcescens 1695 and S. marcescens 2595. We showed that the observed lysis in the last two strains was not due to a bacteriocin present in the phage lysate (data not shown). Further adsorption assays confirmed that ϕMAM1 adsorbs to S. marcescens 1695 and S. marcescens 2595, but with low efficiencies (see Fig. SA2 in the supplemental material). For strains where ϕMAM1 was able to form plaques, the efficiency of plaquing (EOP) was similar, except on S. marcescens 365 (EOP, 6.1 × 10⁻³), suggesting that there was no host-controlled restriction system active against this phage in most strains (Table 1).

As ϕMAM1 formed turbid plaques in several strains, we used the protocol described by Petty et al. (25) to test for lysogeny in S. marcescens 0006, 1047, 3127, 3078V, and S6 after ϕMAM1 exposure. The results were universally negative, and it is reasonable to assume that this is either a virulent phage or a temperate phage with a low frequency of lysogeny.

Other enterobacterial strains belonging to the genera Dickeya, Escherichia, Pectobacterium, Citrobacter, Photorhabdus, Yersinia, Salmonella, Pantoea, and Enterobacter were also tested for susceptibility to ϕMAM1 infection. None of the strains tested showed signs of ϕMAM1 sensitivity.

ϕMAM1 is a generalized transducing phage.

As described above, ϕMAM1 was the only strain A153-infecting phage able to transduce mutations between A153 strains. No differences in the transduction efficiency were observed at different temperatures (e.g., 25, 30, and 37°C; not shown) or by using different number of cells (e.g., 10⁸, 10⁹, and 10¹⁰ total bacterial cells) (Table 2). The transduction efficiency was higher at MOIs of 0.01 and 0.1 (>10⁻⁶ transductants obtained per PFU). However, the transduction efficiency decreased at higher MOIs, probably due to killing of the transductants by ϕMAM1 superinfection (Table 2). In fact, during the transduction assays, the size of the transductant colonies after overnight incubation decreased as the MOI values increased, and some phage nibbling was observed at the edges of these colonies.

TABLE 2.

Frequency of transduction of ϕMAM1

No. of A153 cells	MOI	Transduction efficiencya
		Mean	SD
1 × 1010	10	<1.0 × 10−11
	1	3.4 × 10−7	4.1 × 10−8
	0.1	2.7 × 10−6	2.5 × 10−7
	0.01	3.3 × 10−6	6.7 × 10−7
1 × 109	10	<1.0 × 10−10
	1	3.0 × 10−7	1.8 × 10−8
	0.1	4.5 × 10−6	3.0 × 10−7
	0.01	5.8 × 10−6	9.7 × 10−7
1 × 108	10	<1.0 × 10−9
	1	2.8 × 10−7	5.9 × 10−8
	0.1	7.4 × 10−6	1.2 × 10−6
	0.01	6.8 × 10−6	0.9 × 10−6

^aThe efficiency of transduction was defined as the number of transductants carrying the oocJ::TnKRCPN1 marker obtained per PFU in the A153 recipient. The means and standard deviations of three independent experiments are shown.

Transduction efficiencies were determined in S. plymuthica A153 and S. marcescens 12 for several chromosomal markers, including mutations in the oocydin A biosynthetic genes oocJ and oocU, the antibiotic biosynthetic genes admH and admK, and the cellulose synthase regulator-encoding gene bcsB, among others. Transduction efficiencies of between 1.1 × 10⁻⁶ and 4.2 × 10⁻⁶ were observed for the different chromosomal markers in wild-type reference strain A153 (Table 3). The frequency ranged from 2.3 × 10⁻⁶ to 7.2 × 10⁻⁶ in S. marcescens 12 (Table 3).

TABLE 3.

ϕMAM1 transduction efficiency of different markers within Serratia plymuthica A153, Serratia marcescens 12, and Kluyvera cryocrescens 2Kr27

Donor strain	Relevant marker/plasmid	Recipient strain	Phenotype	Transduction efficiencya
				Mean	SD
Serratia plymuthica A153
VN1	admH::Km	S. plymuthica A153	Antibiotic negative	1.8 × 10−6	2.5 × 10−7
VN2	admK::Km	S. plymuthica A153	Antibiotic negative	1.1 × 10−6	1.8 × 10−7
MMnO13	oocJ::Km	S. plymuthica A153	Oocydin A negative	2.7 × 10−6	2.5 × 10−7
MMnO15	oocU::Km	S. plymuthica 153	Oocydin A negative	4.2 × 10−6	2.3 × 10−7
OocKmSm	Km, Sm	S. plymuthica A153	Oocydin A positive	2.1 × 10−7	3.1 × 10−8
BcsB	bcsB::Sm	S. plymuthica A153	Cellulose negative	1.3 × 10−6	1.1 × 10−7
A153	pECA1039-km3	S. plymuthica A153	ToxIN positive	2.5 × 10−7	7.4 × 10−8
Serratia marcescens 12
Sma12I	luxI::Km	S. marcescens 12	Lactone negative	5.3 × 10−6	1.3 × 10−7
Dho	narX::Km	S. marcescens 12	Pyrimidine auxotroph	6.6 × 10−6	2.7 × 10−7
Cps	cps::Km	S. marcescens 12	Uracil auxotroph	2.3 × 10−6	6.9 × 10−8
Sma12S	luxS::Km	S. marcescens 12	Autoinducer 2 negative	7.2 × 10−6	8.2 × 10−7
Intergeneric transduction
A153	pECA1039-km3	K. cryocrescens 2Kr27	ToxIN positive	2.7 × 10−7	2.6 × 10−8
A153	pECA1039-km3	S. marcescens 12	ToxIN positive	4.8 × 10−8	9.6 × 10−9

^aThe number of transductants showing coinheritance of antibiotic resistance and the listed phenotype was determined. Transduction efficiency is expressed as the number of transductants per PFU. The means and standard deviations of three independent experiments are shown. Transduction experiments were performed using 10¹⁰ cells with ϕMAM1 at an MOI of 0.1.

Transduction of a kanamycin derivative version of pECA1039 (40) from S. plymuthica A153 to S. marcescens 12 and Kluyvera cryocrescens 2Kr27 was also demonstrated. The plasmid pECA1039, isolated from the Gram-negative bacterium Pectobacterium atrosepticum (40), carries the first known type III toxin-antitoxin (TA) system, ToxIN, which is bifunctional as an abortive infection system active against many phages. However, ϕMAM1 was not aborted by the ToxIN system. Although the plasmid transduction efficiencies were relatively high, these were 1 order of magnitude lower than those obtained for the transduction of chromosomal markers (Table 3).

ϕMAM1 genome analysis.

ϕMAM1 has a double-stranded DNA of 157,834 bp (98.5 × 10⁶ Da) encoding 198 putative open reading frames (ORFs) and three tRNAs genes, tRNA-Pro (CCU), tRNA-Met (AUG), and tRNA-Cys (UGU) (Fig. 3; see Table SA2 in the supplemental material) (24). Three different start codons, ATG (94.0%), GTG (3.5%), and TTG (2.5%), were identified. The coding sequences of ϕMAM1 represent 91.6% of the genome, which has a GC content of 52.5%. On the basis of the BLASTP similarities and the presence of conserved domains, a predicted function was assigned to 39.8% of the identified ORFs. Within the 60.2% of the genes encoding hypothetical proteins, 27.3% of them were unique to ϕMAM1. Genome comparison analyses showed that the genome of ϕMAM1 is between 52.7 and 54.2% identical at the DNA level to the genomes of the seven sequenced ViI-like phages and that it shares between 59.0 and 64.6% of their putative ORFs (Fig. 3 and 4; see Tables SA2 and SA3 in the supplemental material).

FIG 4.

DNA homology between the genome of ϕMAM1 and the genomes of the seven sequenced ViI-like phages. The alignments represent the percent DNA homology between the genome of ϕMAM1 and the genomes of PhaxI (37%) (A), LIMEstone1 (11%) (B), CBA120 (36%) (C), ϕSH19 (33%) (D), SFP10 (34%) (E), Vi01 (32%) (F), and ϕSboM-AG3 (35%) (G). The regions encoding the host recognition determinants, the tail spike proteins, are highlighted with a box with a dashed border. Alignments were performed using the wgVISTA program (75).

The high homology observed between ϕMAM1 and ViI-like phages led us to investigate their phylogenetic relationships. Thus, the amino acid sequences of a structural protein (major capsid protein) and a nonstructural protein (DNA polymerase) were chosen to perform phylogenetic analyses. Previously, major capsid proteins and DNA polymerase were often used as phylogenetic markers (26, 34, 41). As expected from the phage morphology and the genomic comparison analyses, the phylogeny revealed that ϕMAM1 is closely related to ViI-like phages (see Fig. SA3 in the supplemental material).

Transcription, regulatory sequences, and translation.

The analysis of the promoter regions of the ϕMAM1 genes allowed us to identify the motif TTCAATAA(N₁₂)TATNAT in the promoters of 27 ϕMAM1 genes (see Fig. SA4A in the supplemental material). Most (81.4%) of the ϕMAM1 genes containing this upstream consensus sequence encode hypothetical proteins, making it difficult to predict in which stage of the lytic cycle they may be expressed. However, the motif shows similarity to the consensus sequence of the E. coli RpoD-dependent promoters [TTGACA(N_15-18)TATAAT] (42) and to the conserved motif identified in T4 early promoters (43), perhaps indicating that these genes are expressed in early stages of the phage lytic cycle.

In general, late transcription is responsible for the synthesis of phage structural components, besides other recombination and replication factors (43). Based on the consensus sequence of T4 late promoters (TATAAATA) (44), we identified 16 putative late promoters (see Fig. SA4B in the supplemental material). Seven of them, including P_{MAM_038} (upstream of a tail spike head-binding protein gene), P_{MAM_135} (baseplate hub subunit), P_{MAM_142} (Gp32 single-stranded DNA binding protein), and P_{MAM_143} (Gp19 baseplate tail tube) are expected to be late genes. Several proteins have been described to be involved in late transcription in T4-related phages, and ϕMAM1 encodes orthologs of all of them, including the transcription coactivator Gp33 (MAM_140), the sigma factor Gp55 (MAM_108), sliding clamp protein Gp45 (MAM_078), and the clamp-loader proteins Gp44 (MAM_079) and Gp62 (MAM_080) (43, 44). Thirty-six Rho-independent terminators were identified in the genome of ϕMAM1, and four of them were predicted to be on late transcripts (see Fig. SA4C in the supplemental material).

Structural components.

Mass spectrometry analysis identified 41 structural proteins in Vi01 (32) and 39 in LIMEstone1 (11), two ViI-like phages. However, the function of 33% (11) and 41% (32) of the identified structural proteins, respectively, is unknown. With the exception of 6 and 4 unknown structural proteins identified in Vi01 and LIMEstone1, respectively, orthologs for the remaining structural proteins were found in ϕMAM1 (see Table SA2 in the supplemental material). However, genome comparison analyses showed that major differences were observed in a 14.7-kb region encoding the host recognition elements, the tail spike proteins (TSPs) (Fig. 3 and 4; see Table SA2 in the supplemental material). Other ϕMAM1 structural proteins are described in the supplemental material.

Most of the sequenced ViI-like phages encode four tail spike proteins (26), and the presence of these TSPs has been confirmed by mass spectrometry analyses in Vi01 (32) and LIMEstone1 (11). The ϕMAM1 genome also encodes four putative TSPs, three of which (MAM_031, MAM_034, MAM_037) contain a pectate lyase domain (see Table SA2 in the supplemental material). The TSP MAM_031 is a 987-amino-acid protein, and BLASTP analyses showed that its first 262 amino acids are 45 to 49% identical (64% similar) to the N-terminal region of TSPs carried by all the seven sequenced ViI-like phages. However, the N-terminal regions of the other three putative TSPs (MAM_034, MAM_037, MAM_038) are less conserved and appear to be more specific to ϕMAM1. Thus, MAM_034 is 37 to 39% identical (53 to 55% similar) from amino acid 23 to amino acid 114 to TSPs present in SFP10, CBA120, and PhaxI, whereas amino acids 25 to 211 of the N-terminal region of MAM_037 are 34% identical (54% similar) to TSP-1 of ϕSH19. Finally, the first 97 residues of the 602-amino-acid TSP MAM_038 are 29% identical (46% similar) to the TSP Vi01_170c of the Salmonella phage Vi01 (see Table SA2 in the supplemental material). Interestingly, the BLASTP analyses also showed that the C-terminal regions of the four ϕMAM1 putative TSPs show similarity to several proteins of unknown function encoded by the genomes of Serratia plymuthica AS9 (45), Serratia plymuthica AS12 (46), and Serratia proteamaculans 568 (47), bacterial strains closely related to the ϕMAM1 host, S. plymuthica A153.

Capsular antigen is the receptor for ϕMAM1.

Group 1 capsular polysaccharides (CPSs) and lipopolysaccharide (LPS) O-antigen biosynthetic gene clusters of Enterobacteriaceae have important common features. In fact, research on the biosynthesis of LPS O antigens has been extrapolated to understand the assembly of CPSs (48). Thus, both CPSs and LPS O-antigen gene clusters share many genes, and they are located between the galF and gnd genes in strains belonging to the genera Escherichia, Salmonella, Shigella, and Enterobacter (49–52). However, it is the presence of the wza, wzb, and wzc genes that differentiates these CPS and LPS loci (48).

All the strain A153 phage resistance mutations mapped in a 17.6-kb region between galF and gnd (see Fig. SA5 in the supplemental material). The 17.6-kb region consists of 13 genes predicted to form an operon (see Fig. SA5 in the supplemental material), and consequently, the insertion of the TnKRCPN1 transposon may cause polar effects in the expression of the downstream genes. In this region, the wza, wzb, and wzc genes are involved in the synthesis, export, and regulation of CPSs (48, 53). The hexose-1-phosphate transferase-encoding gene (wbaP) was proposed to be involved in the initiation of capsular polysaccharide synthesis (48). The polysaccharide can be extended by the action of glycosyltransferases, four of which are encoded by genes in this region. The region also contains the gene wzx, encoding a flippase required for the translocation of the CPS biosynthetic units across the inner plasma membrane (53). Not surprisingly, the region also encodes a hypothetical protein that is 31% identical (49% similar) to the C-terminal region of the TSP MAM_034 protein.

During the characterization of the phage-resistant mutants, we noted that they exhibited a rough colony morphology characteristic of some CPS-defective mutants, and they showed the same antifungal activity as the parental A153 strain (see Fig. SA5 in the supplemental material). Adsorption of ϕMAM1 to all the phage-resistant mutants was abolished, confirming the role of CPSs as receptors (Fig. 2B).

Mobilization of the oocydin A gene cluster by transduction.

The abundance and diversity of secondary metabolite gene clusters in some bacterial strains and their irregular distribution raised the possibility of horizontal transfer of such gene clusters between bacterial strains (54). Recently, we identified a 77-kb gene cluster involved in the biosynthesis of the haterumalide oocydin A (ooc) (30). Given the high transduction efficiency and its large genome size, we decided to use ϕMAM1 as a genetic tool for mobilizing the ooc gene cluster. Thus, we constructed strain S. plymuthica A153 OocKmSm containing a kanamycin resistance cassette and a streptomycin resistance cassette up- and downstream of the ooc gene cluster, respectively. The OocKmSm strain was the donor strain used to transduce the ooc gene cluster into the non-oocydin A-producing strain OocEV, containing in-frame deletions in oocE and oocV. Among the transductants, 17.7% and 9.5% of the streptomycin- and kanamycin-resistant colonies, respectively, showed restoration of oocydin A production (Table 4; Fig. 5). The ooc gene cluster was transduced into OocEV at a frequency of 1.9 × 10⁻⁷, and the production of the haterumalide was fully restored in the kanamycin and streptomycin doubly resistant colonies (Table 4). We also tried to transduce the ooc gene cluster into the other ϕMAM1-sensitive strains. However, no transductants were obtained with these strains, presumably due to insufficient DNA homology between the donor (A153) and the recipient strains to enable homologous recombination.

TABLE 4.

Transduction efficiency of the 77-kbp oocydin A gene cluster by ϕMAM1^a

Antibiotic(s) used	Transduction efficiency		Restoration of oocydin A production in transductants (%)
	Mean	SD
Km	1.2 × 10−6	4.7 × 10−7	17.7 ± 4.8
Sm	2.1 × 10−6	4.2 × 10−7	9.5 ± 2.0
Km, Sm	1.9 × 10−7	1.7 × 10−8	100 ± 0

^aThe donor strain was OocKmSm, the relevant markers used in the selection plate were kanamycin (Km) and streptomycin (Sm), and the recipient strain was OocEV, which is a markerless, double-in-frame deletion mutant defective in the genes oocE and oocV (see Table SA1 in the supplemental material).

FIG 5.

Mobilization of the oocydin A gene cluster into the nonproducing strain Serratia plymuthica A153 OocEV through transduction mediated by phage ϕMAM1. Production of oocydin A in several A153 derivative strains is shown as the ability to inhibit the growth of the plant-pathogenic fungus Verticillium dahliae. The bioassays were performed as described previously (30). The assays were repeated at least five times, and representative results are shown. The bacterial strains used are described in Table SA1 in the supplemental material. wt, wild type.

DISCUSSION

Horizontal gene transfer (HGT) is a powerful source of genomic diversity and contributes actively to the rapid evolution of prokaryotes (55). In fact, it has been estimated that up to 80% of known bacterial and archaeal genomes may be affected by HGT (56). From an ecological perspective, the acquired genes may represent an adaptive advantage for the microorganisms under certain environmental conditions (57, 58). Transduction has been described in natural environments, such as in sewage, fresh water, and seawater and on plant leaves, reflecting its importance in microbial ecology and in the evolution of microbial populations (14, 56).

Several Serratia transducing bacteriophages have been identified previously (25, 59, 60). However, to our knowledge, ϕMAM1 is the first transducing phage described for Serratia plymuthica. Compared with other enterobacterial phages (25, 61, 62), the high transduction efficiency mediated by ϕMAM1, together with its large genome (it can transduce about 3% of the A153 host genome) and its wide host range, makes this new phage an extraordinary tool for bacterial genetics. Additionally, given the DNA homology between ϕMAM1 and the other sequenced ViI-like phages (Fig. 3 and 4; see also Table SA3 in the supplemental material), it remains possible that all these Viunalikevirus bacteriophages will be highly efficient transducers. Consistent with this notion, Salmonella phage Vi01 was shown to be a transducing phage 20 years ago (63).

We have recently shown that the oocydin A gene cluster is present in four plant-associated bacteria belonging to the genera Serratia and Dickeya (30). The fact that the genomic context of the ooc gene cluster in three of these bacteria is different, together with the identification of sequences similar to the sequence of a bacteriophage P4 integrase bordering the cluster in one of the strains, suggests that the gene cluster was previously acquired by HGT (30). High bacterial density has been associated with elevated transduction rates (64), perhaps indicating that the rhizosphere may be a niche conducive to higher transduction frequencies. In agreement with this idea, we have found that several Serratia strains isolated from the rhizosphere of the same crop are able to produce oocydin A (M. A. Matilla and G. P. C. Salmond, unpublished data). This observation indicates that the ooc gene cluster is more widespread than was perhaps expected, and it is possible that it has been transferred between rhizosphere strains by transduction.

ϕMAM1 contains three tRNA genes located together in a cluster. Although tRNA-Met is conserved within most of the sequenced ViI-like phages, thus far the tRNA-Pro and tRNA-Cys genes are unique to ϕMAM1. The function of phage tRNA is still an open debate, but it has been suggested that a positive correlation exists between the tRNA distribution and phage codon usage (65). Although this correlation is absent in ϕMAM1 (data not shown), phage tRNAs have been associated with overcoming differences in the GC content between the phages and their hosts (66). Interestingly, the GC content of the ϕMAM1 tRNA genes (56.2%) and that of the genome of its host, S. plymuthica A153 (55.9%) (24), are similar.

Many reports have reported promiscuity in enterobacterial phages (34, 35, 60, 67, 68), and ϕMAM1 is another noteworthy example. ϕMAM1 was found to infect 68% of the Serratia marcescens strains tested and a rhizobacterial strain from the Kluyvera genus. ϕMAM1-sensitive strains include 7 environmental and 10 clinical isolates, including the model bacterium S. marcescens 365 and the carbapenem-resistant strain S. marcescens S6. Additionally, the sequences of the C-terminal regions of the TSPs from ϕMAM1 show similarity to those of several hypothetical proteins encoded by the genomes of strains such as Serratia proteamaculans 568, a rhizobacterial strain with a genome highly similar to that of A153 (47). TSPs are involved in the recognition of a host receptor(s), and the analysis of their C-terminal regions has been predicted to be a rational approach for the identification of new phage hosts (26, 69, 70). Strain S. proteamaculans 568 shows interesting biocontrol properties, including plant growth promotion and production of exoenzymes (47); however, its genome does not contain genetic information for the synthesis of oocydin A. We were able to transduce the complete, functional ooc gene cluster back into the non-oocydin A producer strain OocEV at a high efficiency (Table 4; Fig. 5). Thus, ϕMAM1 might prove to be an effective tool for mobilizing the ooc gene cluster into non-oocydin producers, thereby engineering synthetic derivative strains with enhanced biocontrol properties by a process that perhaps mimics the process that is likely to be occurring naturally in the environment.

Mechanisms of bacterial defense against bacteriophages include the mutation of host receptors, restriction of the incoming phage DNA, abortive infection (i.e., through toxin-antitoxin [TA] systems), and CRISPR-Cas systems (71). TA systems are generally comprised of two genes encoding a toxin component and its cognate antitoxin (72). TA loci are widespread in bacteria, and they have been found in plasmids and bacterial chromosomes, strongly implicating HGT in the dissemination of these systems (72, 73). In 2009, we identified and characterized the first type III protein-RNA TA system, carried by the plasmid pECA1039 of the phytopathogen Pectobacterium atrosepticum (40). Using ϕMAM1, transduction of pECA1039-km3 between different species of Serratia and between two bacterial genera was demonstrated (Table 3). Given the importance of antiviral abortive infection systems in the population biology of bacteria that are constantly assaulted by viral parasites, these observations reinforce the notion that phages are key players in the dissemination of important genetic information and therefore in adaptive bacterial evolution in changing ecological niches. Interestingly, neither the P. atrosepticum TA system nor the type III TA system encoded by the genome of the insect-pathogenic bacterium Photorhabdus luminescens (72) was able to protect S. plymuthica A153 from ϕMAM1 infection (not shown), suggesting that this phage evolved the capacity to naturally circumvent the abortive infection system. However, another phage defensive mechanism(s) may be present in the ϕMAM1-sensitive strains S. marcescens 1695 and S. marcescens 2595, since no plaque formation was observed in these two strains.

The recognition of host receptors by ViI-like phages is dependent on tail spike proteins (TSPs) instead of T4-like tail fibers (26). On the basis of the crystal structure of the C-terminal TSP of the Salmonella phage Det7 (69), a phage morphologically similar to ϕMAM1, several authors have proposed a role for the TSP regions. Thus, the N-terminal region may bind to the phage baseplate, transferring the signal to the tail structure after phage adsorption, whereas the C-terminal region is responsible for the recognition of the host receptor (26, 36, 69). As observed in other ViI-like phages (26), the C-terminal region of the TSPs from ϕMAM1 contains domains with enzymatic activity. These domains with enzymatic activity are required for phage infection, and it has been proposed that they serve to separate the phage from cell debris after bacterial lysis (69). TSPs from the ViI-like Salmonella phages Vi01 (32) and Det7 (69) contain an acetyltransferase and an O-antigen binding domain, respectively, responsible for the recognition of the exopolysaccharide capsule and LPS as receptors. TSP-2 of the phage ϕSH19 contains a pectate lyase domain, and it was suggested that it may have a role in the cleavage of glycosidic bonds (33). In agreement with these observations, capsular antigen appears to be the receptor for ϕMAM1, and interestingly, three of its four TSPs contain a putative pectate lyase domain. Recently, capsular polysaccharides have also been proposed to be the receptor in the ViI-like Klebsiella phage o507-KN2-1 (74). Future synthetic biology approaches will involve engineering of highly efficient transducing phages with promiscuous host ranges through exploitation of structural knowledge that underpins the exquisite molecular specificity seen in these phage-host interactions.