All strains used in this study were obtained from the strain collections of the Bundeswehr Institute of Microbiology (Munich) and Federal Institute for Risk Assessment (Berlin). Information on the relevant strains is summarized in Table 1. If not otherwise indicated bacteria were cultivated aerobically in lysogeny broth (LB) (Carl Roth, Karlsruhe, Germany) at 28°C for 24 h under shaking conditions (180–200 rpm). Solid and overlay agar contained 1.8% (w/v) and 0.6% (w/v) bacto-agar No. 1 (Oxoid, Wesel, Germany), respectively.

To optimize the induction of Ochrobactrum prophages, various stress factors (mitomycin C, heat, UV) were investigated. At an adsorption (A_588nm) of 0.2–0.3, 10 ml bacterial cultures were treated with different mitomycin C concentrations (0.5–10 μg/ml), heat (50, 60, or 70°C for 30 and 60 s) or UV (for 30 and 60 s). Thereafter, cultures were incubated under standard conditions for 16–18 h. UV treatment was performed by applying aliquots of the culture into petri dishes (d = 90 mm), which were placed in a distance of 10 cm to an UV lamp (corresponding 45 J m⁻²). UV radiation treatment was performed for 30 or 60 s. To obtain cell-free lysates, samples were centrifuged for 15 min at 6,000 × g and filtrated through 0.22 μm sterile filters (GE Healthcare, Munich, Germany). Lytic activity was quantified by spotting 10 μl aliquots of a 1:10 dilution series onto lawns of susceptible Ochrobactrum strains. All strains were investigated in triplicate under the respective induction conditions. For further characterization, plaques of POA1180 and POI1126 were purified by a three-fold single plaque purification procedure.

Phages POA1180 and POI1126 were recovered by mitomycin C (0.5 μg/ml) treatment of strains O. anthropi O1180 and O. intermedium O1126, respectively. High titer lysates were prepared from 200 ml cultures. Cell-free lysates were obtained by centrifugation at 6,000 × g for 30 min and filtration of the supernatants through 0.22 μm sterile pore-size filters (GE Healthcare, Munich, Germany). To remove bacterial DNA and RNA, phage lysates were supplemented with 10 mM MgCl₂, 10 μg ml⁻¹ DNaseI and RNase A (Roche, Mannheim, Germany) followed by an incubation at 37°C for 2 h. Phage particles were concentrated by ultracentrifugation (Beckman) for 4 h at 200,000 × g. Concentrated phages were resuspended in SM-buffer and purified by discontinuous gradient (CsCl, 1.35–1.65 g ml⁻¹) centrifugation at 141,000 × g for 18 h (Sambrook and Russell, 2001). After centrifugation phage bands were removed and desalted by 100K Amicon Ultra centrifugal filter columns (Merck Millipore, Schwalbach, Germany).

The host range of the phages (e.g., POA1180, POI1126) was determined by spot assays. This was performed by applying 100–200 μl of each Ochrobactrum spp. strain to 6 ml LB soft agar (0.6%) and pouring of the overlay agar onto a LB agar plate. Aliquots of 1:10 serial dilutions of phage lysates were dropped onto the lawn of the solidified overlay agar. After 24 h spotting areas were visually inspected for plaque formation. Lytic activity was examined on strains of Ochrobactrum spp. (n = 119), Brucella spp. (26 reference and type strains), Mesorhizobium (n = 6), Sinorhizobium (n = 5), Pseudomonas (n = 5) and Yersinia enterocolitica O:9 (n = 4).

CsCl-purified phages were applied to pioloform-carbon-coated, 400-mesh copper grids (Plano GmbH, Germany), for 10 min, fixed with 2.5% aqueous glutaraldehyde solution for 1 min, stained with 2% aqueous uranyl acetate solution for 1 min and examined using a JEM-1010 (JOEL, Tokyo, Japan) transmission electron microscope at 80 kV accelerated voltage.

Phage DNA was extracted from CsCl-purified particles by proteinase K/SDS treatment and ethanol precipitation (Sambrook and Russell, 2001). Thereafter, the phage DNA was resuspended in 0.5 × TE-buffer (pH 8.0) for further analyses. Sequencing libraries were prepared using the Nextera XT DNA Sample Preparation Kit according to the recommendations of the manufacturer. Paired-end sequencing was performed on the Illumina MiSeq benchtop using the MiSeq Reagent v3 600-cycle Kit (2 × 300 cycles) (Illumina, CA, USA). Raw reads were assembled de novo using tools of the Pathosystems Resource Integration Center resulting in a single contig with an average sequence coverage of more than 120 and 100 per consensus base for POA1180 and POI1126, respectively. DNA hybridization was conducted using the Roche digoxygenin hybridization system on positively charged nylon membranes according to the manufacture's procedure (Roche, Heidelberg, Germany).

To identify putative prophage sequences in the available Ochrobactrum spp. genomes of GenBank (NCBI), the Phage Search Tool-PHAST was used (Zhou et al., 2011). Sequence analysis and alignments were carried out using Accelrys Gene v2.5 (Accelrys Inc., San Diego, CA, USA). Identification of genetic elements like ORFS, transcription terminators, and tRNAs on the phage genome was conducted as previously described (Hammerl et al., 2011, 2016). Similarity and identity values were determined at the NCBI homepage using different BLAST algorithms (Johnson et al., 2008). Annotation of the phage genomes was performed by using Sequin (https://www.ncbi.nlm.nih.gov/Sequin/).

The complete nucleotide sequences of the O. anthropi POA1180 and O. intermedium POI1126 phage genomes were submitted to GenBank under the accession numbers KX669658 and KY417925, respectively.

To get a first overview on the presence of prophage DNA in Ochrobactrum spp., 19 strains representing several species whose genome sequences have been published, were analyzed by automated in silico analyses using the PHAge Search Tool (PHAST) (Zhou et al., 2011). We identified prophage sequences (6.9–91.8 kb in size) in all analyzed Ochrobactrum species. Between two and ten putative phage genomes were detected in each strain (Table 2). Highly prevalent was a 17.3–91.8 kb DNA region that showed significant similarity to the Sinorhizobium meliloti phage 16-3, which possesses a genome of 60 kb (Deak et al., 2010). The second most frequent prophage was represented by a DNA region of 25.3–56.8 kb revealing close relationship to phage AmM-1 isolated from the Rhizobiales deep-sea bacterium Aurantimonas sp. whose genome has a size of 47.8 kb (Yoshida et al., 2015). Most of the remaining prophages in Ochrobactrum spp. are similar to other Rhizobium phages but homologies to phages of Rhodobacter, Pseudomonas, various Enterobacteriaceae and even Bacillus were also detected (Table 2). According to the results obtained by in silico analyses, some of the prophages may be complete and intact. The GC content of the prophages ranges between 49.7 and 65% whereas the average GC content of O. anthropi is 56%, suggesting that some Ochrobactrum strains may have acquired temperate phages from different hosts through horizontal gene transfer (Table S2).

In this study, 125 Ochrobactrum strains of various species deposited in the strain collection of the BfR were analyzed in terms of inducible prophages. We first compared the efficacy of various stress factors (mitomycin C, UV, high temperature) on the release of phage particles. For this purpose, the strains O. anthropi O1180 (Figure 1A) and O. intermedium O1126 (data not shown) were treated with different concentrations (0.5, 2.5, and 10.0 μg/ml) of mitomycin C, heat (50, 60, 70°C for 30 and 60 s) and UV (45 J m⁻² for 30 and 60 s). Thereafter, the strains were further incubated at 28 and 37°C. At both temperatures bacterial growth was inhibited in response to the tested stress factors, particularly mitomycin C, even though prophage induction at 37°C occurred earlier in most preparations. As shown in Figure 1B even the untreated control of both strains released 10⁴ to 10⁵ phages per milliliter (Table 3). Under stress conditions titers up to 2 × 10⁸ pfu/ml were determined. In general, all tested stress factors were suitable to induce prophages, albeit phage titers achieved in the individual approaches differed significantly from each other. O. anthropi strain O1180 revealed much stronger temperature dependence than O. intermedium O1126. Some phage titers obtained at 28°C were two to three orders of magnitude higher than those of the corresponding lysates at 37°C. Moreover, whereas UV treatment was the most efficient method to induce prophages in O. anthropi, the highest phage titers in O. intermedium were achieved with a mitomycin C concentration of 0.5 μg/ml. Thus, Ochrobactrum strains diverge significantly regarding their sensitivity to prophage inductors. Based on the results of this study, subsequent induction experiments with a broad range (n = 125) of Ochrobactrum spp. strains were performed with 0.5 μg/ml mitomycin C at 28°C.

The 125 prepared lysates were tested for their lytic activity by spotting aliquots on all 125 Ochrobactrum spp. strains. As an additional indicator for released particles, DNA was isolated from the lysates. Plaques were observed with ~60% of the lysates. Up to 20 strains were susceptible to each lysate. In most cases several Ochrobactrum species were lysed. To get a deeper insight into the released phages, 19 lysates originating from O. anthropi, O. intermedium, Ochrobactrum tritici, O. dulfi, and Ochrobactrum sp., from which significant amounts of encapsidated phage DNA could be isolated (data not shown) were selected for further analyses. Electron microscopic examination revealed phage particles in all lysates, even though four of them did not show lytic activity. According to their morphology, the identified phages belong to different families (Table 4). Most prominent was the family Myoviridae represented by phages with an isodiametric or elongated head and a contractile tail. Some lysates contained particles with a long, non-contractile and very short tail, which are members of the Siphoviridae and Podoviridae, respectively. However, it is important to mention that a number of lysates exhibited various morphotypes indicating that the respective strains released several phages simultaneously.

Previous experiments demonstrated that most of the investigated Ochrobactrum strains harbored one or more prophages inducible by mitomycin C, heat and UV. To obtain data on individual phages present in the 19 selected lysates, high titer lysates were prepared by mitomycin C treatment of large culture volumes of the respective Ochrobactrum strains (see Section Materials and Methods). Upon purification by CsCl density-gradient centrifugation, 19 phages were isolated and characterized in terms of their host range and genetic relationship. Similar to the data obtained with non-purified lysates, the majority of phages infected several Ochrobactrum species, particularly O. anthropi and O. intermedium, but also strains of O. tritici, O. oryzae, and O. pseudintermedium (Table 5). By contrast, classical and atypical Brucella strains were not lysed. Plaques were even and mostly small and rather turbid, as shown for the myovirus POA1180 and the podovirus POI1126 (Figure 2). On the other hand, phage P1218 isolated from O. intermedium revealed a very narrow host range and exclusively lysed three strains belonging to this species. Finally, four phages (P1142, P1166, P1184 and P1196) did not infect any Ochrobactrum spp. strain. In some cases growth inhibition of indicator strains was observed indicating that phage replication did not occur in those strains.

To determine the genetic relationships of the Ochrobactrum phages amongst themselves and to the temperate Brucella inopinata phage BiPBO1, restriction patterns of the phage genomes were compared, followed by Southern hybridization (Figure 3). EcoRV restriction analyses revealed numerous fragments and demonstrated that the phages possess double-stranded DNA. Most restriction patterns were unique but the O. dulfi phages P1205 and P1206 exhibited identical profiles. In addition, two O. intermedium phages (P1132 and P1135) and the phages P1126 and P1187, isolated from O. intermedium and O. tritici, respectively, showed very similar patterns. However, the latter two phages differ in their host specificity indicating at least some minor genetic variations. According to the obtained restriction fragments (Figure 3A), genome sizes between 32 and 63 kb were calculated (data not shown). To determine DNA homologies between the phages, Southern hybridizations were performed using three phage DNAs as probes; myovirus POA1180 isolated from O. anthropi (Figure 3B), podovirus POI1126 (Figure 3C) isolated from O. intermedium and B. inopinata phage BiPBO1 (Figure 3D), which already revealed significant homologies to some Ochrobactrum prophages (Hammerl et al., 2016). Phage POA1180 did not hybridize to most of the other phages. Weak homologies were detected to POI1126 and to the two O. dulfi phages whose restriction patterns were identical. Phage POI1126 hybridized strongly to the O. anthropi phage P1129 and to the O. tritici phage P1187. Moreover, the homologous restriction fragments were similar in size corroborating the close relationship of these phages. The temperate B. inopinata phage BiPBO1 showed relationship to the O. anthropi phages P1129 and P1157 (Figure 3). Based on these results it can be suggested that temperate Ochrobactrum phages are genetically highly diverse. To further characterize this diversity, the genomic sequences of myovirus POA1180 and podovirus POI1126 were determined.

The genome of phage POA1180 has a size of 41,655 bp with a GC-content of 56.6%. Fifty-eight putative gene products were assigned of which 55 are located on the same DNA strand. For 32 gene products a functional prediction could be made (Table S1). It is notable that almost no similarities were detected to proteins of other phages. Instead, very similar proteins are encoded by other Ochrobactrum strains and strains of the genera Brucella, Agrobacterium, Bartonella, Burkholderia, Rhizobium, and Stappia. The highest similarities exist to a prophage recently identified in Brucella strain 10RB9215 isolated from an exotic frog (Scholz et al., 2016). This indicates that closely related prophages are widespread in these bacteria but that reports on the respective phages are scarce. Based on homologies to other proteins, a gene map of phage POA1180 was constructed (Figure 4). The left half of the POA1180 genome mainly contains genes for DNA packaging and virion assembly. We identified putative genes for the small (ORF01) and large subunit of the terminase (ORF02), capsid protein, and several tail proteins (e.g., tail tube protein, tape measure protein, and tail fiber protein). The right half of the POA1180 genome contains genes for various proteins. First and foremost, two genes (ORF31 and ORF32) were identified which may confer resistance to chromium. Their products are very similar to proteins of Rhizobium, Mesorhizobium, and Sinorhizobium. While ORF31 encodes a chromate transport protein, the ORF32 product is closely related to chromate resistance proteins. Chromium resistance has already been reported for the O. tritici strain 5bvl1 but in this strain, the resistance genes are located on a transposon (Morais et al., 2011). Transposase genes (ORF42 and ORF43) are also present on the POA1180 genome, but whether the chromate resistance genes of this phage can be mobilized, is not known. The POA1180 ORF46 product is similar to sulfate permeases belonging to the CysZ family and may mediate the uptake of sulfate subsequently utilized for the synthesis of sulfur-containing compounds in the cell. Another gene that might be important for Ochrobactrum metabolism is ORF53 probably encoding a NAD(P) transhydrogenase. In addition, the right half of the POA1180 phage genome contains genes for an ATPase, transcriptional regulators, a partitioning protein, an exonuclease and the endolysin. However, for most genes located in this region, functional predictions could not be made.

Phage POI1126 possesses a genome of 60,065 bp with a GC-content of 56.2%. Eighty putative genes were assigned, which are equally located on both DNA strands (Figure 4). The function of 31 gene products could be predicted on the basis of similarities to known proteins. The closest related phages of POI1126 are the S. meliloti phage PCB5, Erwinia pyrifoliae phage PEp14 and the Burkholderia cenocepacia phages DC1, Bcep22, and BcepIL02. While the lifestyle and morphology of PCB5 have not been documented yet, the other phages are podoviruses like POI1126 (Gill et al., 2011; Lynch et al., 2012). The three Burkholderia phages were originally isolated from soil. It has still to be clearified whether these phages are virulent or temperate since they were unable to form stable lysogens. Although, a recombinase gene and an attP site have been detected on the phage genomes the phages may be unable to successfully integrate into the host chromosome. In contrast, phage POI1126 was isolated from an O. intermedium strain by induction with mitomycin C suggesting a temperate lifestyle. The analysis of the POI1126 genome disclosed that the phage is almost identical to an unnamed plasmid of the O. intermedium strain LMG 3301 (ACQA01000004). We only found one nucleotide variation (deletion) in ORF30. The frame shift mutation led to an exchange of seven C-terminal amino acids in the predicted DNA-methylase protein. The strong homologies to the plasmid of strain LMG 3301 inspired us to analyse the plasmid content of O. intermedium strain O1126. This strain indeed contains a plasmid that showed an identical EcoRV restriction pattern as phage POI1126 (Figure 5). From these data, it can be assumed that both O. intermedium strains, O1126 and LM3301, harbor a temperate phage whose prophage replicates as plasmid. The sequence of the LMG 3301 plasmid deposited at NCBI is framed by a terminal direct repeat of 389 bp. We analyzed the corresponding DNA regions of POI1126 and the plasmid prophage by PCR but did not detect this repetitive sequence. As a consequence the first and last ORF of the LMG 3301 sequence encoding putative DNA methylases are merged in strain O1126. The resulting methylase gene (ORF30) is by far the largest gene of POI1126 (14.526 bp) and preceded by a suitable ribosome binding site. The genome analysis of the phage revealed two ORFs (38 and 39) probably forming an operon that may be important for plasmid maintenance. Their products are similar to partitioning proteins. Other plasmid-associated genes could not be determined. On the other hand, a putative integrase gene (ORF65) was identified whose function is yet not known. The gene map of phage POI1126 shows that the left half of the genome mainly contains genes for structural proteins and virion assembly, while only few ORFs in the right half could be functionally assigned. It is possible that this part of the genome harbors other genes, which are required for plasmid replication.

The phages POI1126 and POA1180 possess several genes encoding potential methyltransferases (Table S1). To find out whether the genomes of the phages are modified, their DNAs were digested with several restriction endonucleases. None of the obtained restriction patterns of POA1180 coincided with patterns determined by in silico analysis. Figure 6A presents EcoRI and EcoRV profiles of the POA1180 DNA. Using the NEB cutter software, restriction patterns of CpG methylated phage DNA were predicted but also these patterns did not agree with patterns obtained by digestion (Figure 6A). However, a superimposition of the computer-generated EcoRI profiles (but not of the EcoRV profiles) corresponded well with the restriction pattern observable in agarose gels. This suggests that the POA1180 phage DNA contains some modifications, which diverge from modifications known in E. coli. On the other hand, the EcoRI and EcoRV restriction patterns of phage POI1126 were in good agreement with patterns predicted for unmethylated DNA (Figure 6B). Thus, the DNA of this phage is either not modified or contains modifications which cannot be traced by the restriction endonucleases used.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.