|Home | About | Journals | Submit | Contact Us | Français|
Virulent phages have been used for many years to type Brucella isolates, but until recently knowledge about the genetic makeup of these phages remains limited. In this work the host specificity and genomic sequences of the original set (deposited in 1960) of VLA Brucella reference phages Tb, Fi, Wb, Bk2, R/C, and Iz were analyzed and compared with hitherto described brucellaphages. VLA phages turned out to be different from homonymous phages in other laboratories. The host range of the phages was defined by performing plaque assays with a wide selection of Brucella strains. Propagation of the phages on different strains did not alter host specificity. Sequencing of the phages TbV, FiV, WbV, and R/CV revealed nucleotide variations when compared to same-named phages previously described by other laboratories. The phages Bk2V and IzV were sequenced for the first time. While Bk2V exhibited the same deletions as WbV, IzV possesses the largest genome of all Brucella reference phages. The duplication of a 301 bp sequence in this phage and the large deletion in Bk2V, WbV, and R/CV may be a result of recombination caused by repetitive sequences located in this DNA region. To identify new phages as potential candidates for lysotyping, the host range and Single Nucleotide Polymorphisms (SNPs) of 22 non-reference Brucella phages were determined. The phages showed lysis patterns different from those of the reference phages and thus represent novel valuable candidates in the typing set.
Brucellae are highly infectious and facultative intracellular bacterial pathogens causing brucellosis, a frequent zoonosis with more than 500,000 human cases reported worldwide every year (de Figueiredo et al., 2015). Infections may lead to reproductive failure and abortion in animals and a feverish multiorgan disease in humans. Up to now, 12 species were allocated to the genus Brucella (Godfroid et al., 2013). Six of them, B. melitensis, B. abortus, B. suis, B. canis, B. ovis, and B. neotomae, recovered from goats, cattle, pigs, dogs, sheep, and desert rats, respectively, belong to the “classical” or “historical” Brucella species (http://www.bacterio.net/brucella.html). Later on, “novel” Brucella species were isolated from cetaceans (B. ceti) and pinnipeds (B. pinnipedialis; Foster et al., 2007), voles (B. microti; Scholz et al., 2008b), baboons (B. papionis; Whatmore et al., 2014), red foxes (B. vulpis; Scholz et al., 2016) and from a human breast implant infection (B. inopinata; Scholz et al., 2010). Recently, a novel Brucella spp. reservoir in amphibians (e.g., Big-eyed tree frog: Leptopelis verniculatus; African bullfrogs: Pyxicephalus edulis; White's tree frog: Litoria caerulea) was discovered (Eisenberg et al., 2012; Fischer et al., 2012; Whatmore et al., 2015; Soler-Lloréns et al., 2016). Genetically, all Brucella species are closely related exhibiting genome similarities of >90% at the nucleotide level (Al Dahouk et al., 2010). Furthermore, because of the close genetic relationship of several genetic loci (e.g., 16S rRNA, 98.7% and recA, 85.5%) and a biochemical profile similar to Ochrobactrum spp., particularly atypical Brucella species like B. microti and B. inopinata are often misidentified using commercially available biochemical test systems (Scholz et al., 2008a,c).
Identification and subtyping of brucellae is time-consuming and laborious. Suspicious colonies on agar plates are primarily identified by a slide agglutination test using polyvalent Brucella antiserum (anti-S serum; Alton et al., 1975). Alternatively, PCR detection systems targeting the bcsp31 gene and the intergenic sequence IS711 exist that are suited for the molecular detection of Brucella spp. (Baily et al., 1992; Hinic et al., 2008). Moreover, some multiplex PCR assays (e.g., AMOS, Bruceladder) were established for further species differentiation (López-Goñi et al., 2008; Mayer-Scholl et al., 2010). However, none of the available molecular typing systems cover all currently known species and biovars of the genus Brucella. In microbiological routine testing, the identification of species and biovars is based on specific properties of the bacteria (e.g., CO2 requirement, H2S production, urease activity, agglutination with monospecific A, R, and M sera, growth on media with thionin or basic fuchsin, metabolization of different substrates) and in particular susceptibility to lytic Brucella reference phages (Al Dahouk et al., 2010).
Phages which infect and lyse Brucella strains are known for over half a century (Parnas et al., 1958; Brinley-Morgan et al., 1960; Jablonski, 1962). After some basic characterization, a typing set comprising five reference phages [Tb (Tbilisi), Fi (Firenze), Wb (Weybridge), Bk (Berkeley), R/C] was developed (Corbel, 1984). Some years later the typing set was complemented by phage Iz (Izatnagar; Joint FAO/WHO Expert Committee on Brucellosis, 1986) and since then has been used in many diagnostic laboratories worldwide. The same holds true for a set of Brucella reference strains serving as controls for lysotyping. The original typing set has also been modified by adding other phages, e.g., S708, Bk2, F1, F25, and Np, some of which are mutants of the reference phages (Moreira-Jacob, 1968; Corbel et al., 1988; Rigby et al., 1989; Hammerl et al., 2014). All brucellaphages described so far have a podoviral morphology and are closely related, demonstrated by restriction analysis and southern hybridization (Segondy et al., 1988; Rigby et al., 1989). They are considered as a single taxonomic species (Corbel and Thomas, 1976; Ackermann et al., 1981). As a consequence, some phages possess an almost identical host specificity (Morgan, 1963; Calderone and Pickett, 1965). For a better identification and discrimination of isolates, Brucella lysotyping is mostly carried out by spot assays using both a routine test dilution (RTD), which is the highest dilution of a phage suspension producing confluent lysis of a propagator strain, and a 10,000 × RTD phage suspension. The main drawback of this procedure is that it cannot clearly distinguish between strains, in which the phages propagate, and those which are merely killed by the so-called lysis from without effect caused by a collapse of the cell wall in response to an overwhelming number of adsorbed phage particles (Corbel, 1984). Similar to growth inhibition, lysis from without is rather unspecific and more difficult to interpret than single plaques that unequivocally illustrate a phage infection (Jones et al., 1968). Furthermore, host range variants of Brucella reference phages have been isolated, particularly after changing the propagator strain (Corbel et al., 1988). Hence, it is not surprising that even studies, in which the same reference phages were tested, may show inconsistent results (Morris et al., 1973). For that reason it is important to know the biological and genetic properties of the typing phages exactly. First DNA sequences of phage Tb were reported by Zhu et al. (2009). The analysis of whole genome sequences of diagnostic brucellaphages confirmed their close relationship, even though phage Tb deposited in two different institutes revealed some sequence deviations, mainly point mutations (Flores et al., 2012; Farlow et al., 2014; Tevdoradze et al., 2015). Indeed, besides two InDels ~2.4 and 0.4 kb in size, the genomes of the investigated phages notably differ by single nucleotide polymorphisms (SNPs). Many of them were found in a gene probably encoding a tail collar protein, which has been suggested to be a determinant of host specificity. The importance of individual amino acids of the tail collar protein for the host range of the phages, however, has not yet been investigated.
In this work, the host specificity and genomes of six VLA Brucella reference phages (designated TbV, FiV, WbV, Bk2V, R/CV, and IzV herein) were analyzed in detail. The genome sequences of Bk2V and IzV will be presented for the first time. The phages were compared with homonymous phages originating from other laboratories. Bioinformatic analyses revealed DNA repeats within the phage genomes, which may be important for the acquisition, loss or duplication of DNA sequences. The host range determination of 22 non-reference phages disclosed some potential candidates useful for lysotyping.
Detailed information on all strains used in this study is given in Table S1. Cultivation of the bacteria was performed as previously described (Alton et al., 1975). Solid and overlay agar contained 1.8 and 0.7% (w/v) bacto-agar No. 1 (Oxoid, Wesel, Germany), respectively.
Relevant data on brucellaphages used in this study are given in Table S2. The reference phages TbV, Bk2V, WbV, FiV, R/CV, and IzV were obtained from the Weybridge bacteriophage collection in form of unopened lyophilized phage stocks produced by the OIE Brucellosis Reference Centre of the Veterinary Laboratories Agency (VLA, Addlestone, UK) in 1973. Lyophilized phages were suspended in 5 ml of SM-buffer (100 mM NaCl, 8 mM MgSO4 7H2O, 50 mM Tris-HCl, pH 7.5). The suspensions were used for plaque assays by the softagar overlay method as previously described (Sambrook and Russell, 2001). In contrast to phage R/CV, which was propagted on B. ovis strain 63/290, TbV, FiV, Bk2V, WbV, and IzV were propagated on B. abortus vaccine strain S19. Besides the standard reference phages which are globally used for Brucella-typing, other brucellaphages were included in the study. While lyophilized A422 and M51 stocks from 1960 were also provided by VLA, the origin of the phages F1, F1m, F1u, F25, F25u, F44, F45, F48, FO1, P, P2, 3, 6, 7, 10I, 12m, 24II, 45II, 212XV, and 371XXIX is unknown. Lyophilized stocks of these phages prepared at the BfR in 1973 were used for further investigation. However, there is no information available on how these phages were propagated before lyophilization. To ensure the purity of all brucellaphages, a three-fold successive single plaque separation was performed. After the third plaque separation purified phages were used for the preparation of high titer lysates (>109 PFU/ml). To accomplish this, 106 PFU were applied to 20 ml of an early logarithmic growing (McFarland 1.0–1.5) Brucella culture followed by incubation for 24–48 h on a rotational shaker (100 rpm) under microaerobic conditions. Thereafter, lysates were centrifuged for 10 min at 7,500 × g to remove cellular debris. Supernatants were subjected to sterile filtration (0.45 and 0.2 μm; Merck Millipore, Schwalbach, Germany) and DNaseI-/RNaseA-treatment (10 μg/ml wt/vol each; Roche, Mannheim, Germany). Phage particles were concentrated and purified by discontinuous CsCl-step gradients (CsCl, 1.3 to 1.7 g ml−1) as described previously (Sambrook and Russell, 2001). Phage bands recovered from CsCl-gradients were desalted using 100K Amicon Ultra centrifugal filter columns (Merck Millipore).
Host range analyses were carried out by spot assays on Brucella spp. reference and type strains (n = 26; Table S1) and field isolates as well as reference strains of Ochrobactrum spp. (n = 119), Yersinia enterocolitica O:9 (n = 7), Mesorhizobium sp. (n = 6), Sinorhizobium meliloti (n = 5), and Pseudomonas (n = 5; data not shown). Two hundred microliters of each strain were mixed with 5 ml of pre-warmed Brucella-broth soft agar (0.7%) and poured onto a lysogeny-broth (LB) agar plate. Ten microliter aliquots of 1:10 serial dilutions of each lysate were spotted onto the overlay agar. Agar plates were visually inspected after incubation for 24 and 48 h at 37°C. Phages that did not affect bacterial growth were classified as non-infectious (negative: −). Phages were classified as infectious (positive: +) when single plaques were identified in spotting zones of the tested lysates. In case of growth inhibition (GI) visible by an even but decreased bacterial growth within the spotting areas, the respective phage/strain combinations were further investigated by preparing phage lysates. For this purpose, 107 phages were applied to 20 ml of an early exponential growing Brucella culture (McFarland: 1.0–1.5). After incubation at 37°C for 48 h, lysates were purified as described above. Propagation of phages was determined by plaque assays (Sambrook and Russell, 2001). Four hundred microliters of a Brucella culture (McFarland: 5.0–7.0) were mixed with 100 μl aliquots of 1:10 serial dilutions of each lysate, incubated for 20 min at room temperature, added to 5 ml of pre-warmed Brucella-broth soft agar (0.7%) and poured onto LB agar plates. After incubation for 24 and 48 h at 37°C, agar plates were visually inspected for plaque activity. If phage replication occurred, the lysates contained more than 107 infectious particles. Strains which increased the phage titer were finally classified as susceptible (positive: +).
Phage DNA extraction from CsCl-purified particles was performed as previously described (Hammerl et al., 2016). Determination of phage genomic sequences was conducted using a Roche 454 genome sequencer FLX titanium system by GATC Biotech AG (Konstanz, Germany). Library generation and 454 FLX sequencing were carried out according to the procedure of the manufacturer (Roche/454 Life Sciences, Branford, Connecticut, USA). Sequence reads were assembled using the Roche/454 Newbler software at default settings (454 Life Sciences Corporation, Software release 2.3) resulting in one contig with an average sequence coverage of >100 per consensus base. Sequence analysis and alignments were carried out using Accelrys Gene v2.5 (Accelrys Inc., San Diego, CA, USA). Bioinformatic analysis and genome annotation were performed as described previously (Hammerl et al., 2014, 2016).
PCR was performed in an Eppendorf Mastercycler ep Gradient (Eppendorf, Hamburg, Germany) according to standard protocols. Single reactions were carried out with ~10 ng/μl phage DNA and 2.5 μl of each primer in a final volume of 50 μl using Qiagen DNA polymerase amplification components (Qiagen, Hilden, Germany). For PCR amplification the following parameters were used: initial template denaturation at 96°C for 120 s followed by 35 cycles including denaturation at 96°C for 15 s, annealing at 55°C for 5 min and elongation for 210 s at 72°C. A final elongation step at 72°C for 1 min was added. Purification of PCR products was performed by using the MSB spin PCRapace kit (Stratec, Birkenfeld, Germany). The nucleotide sequence of the PCR products was determined by Sanger sequencing (Eurofins Genomics, Ebersberg, Germany).
A comprehensive study with 10 Brucella species represented by 26 reference and type strains was performed to elucidate in which strains the group I to group VI Brucella reference phages TbV, FiV, WbV, Bk2V, R/CV, and IzV (Corbel, 1987) replicate and which strains are merely lysed by lysis from without at high MOIs. Lytic activity of the phages was quantified by plaque assays (titration) allowing an accurate determination of the hosts' susceptibility. If no plaque formation occurred and only growth inhibition was observed, we examined propagation of the phages using liquid cultures of the respective strains (see section Materials and Methods). The phages TbV (group I) and FiV (group II) showed almost identical lysis patterns (Table (Table1).1). They infected reference strains of the eight B. abortus biovars (bv1–7, 9), B. neotomae 5K33, B. microti CCM4915 and two strains of the recently identified species B. vulpis (Scholz et al., 2016). On the other hand, B. suis 1,330 (bv1) was only lysed by phage TbV, but not by FiV. Data that have been published on the susceptibility of B. suis 1,330 to phage Tbilisi are rather contradictory. While several articles reported that this strain was not lysed by the phage at RTD (Jones et al., 1968; Flores et al., 2012), other authors obtained plaques with both Tbilisi and Firenze (Rigby et al., 1989). Another deviation from published data concerns B. melitensis 16M (bv1), for which we and also others could not confirm infection by Tbilisi and Firenze, whereas Rigby et al. (1989) reported on plaque formation caused by these phages (Rigby et al., 1989). This raises the question about the reasons for these discrepancies? One possible explanation is that in our study and in the study of Rigby et al. (1989), phages were propagated on the B. abortus vaccine strain S19 (bv1), which makes host adaptation as a reason for different lysis patterns improbable. Furthermore, the slightly different methodologies applied to determine lysis patterns might not account for diverging results. It is more likely that the phages and/or indicator strains used in the two studies were actually not identical, perhaps due to mutations in genes important for phage infection. Of course, this presumption can only be confirmed by whole genome sequencing of the used strains and phages. For phage Tbilisi different variants (TbM and TbY) have already been described (Foster et al., 2007; Flores et al., 2012). In addition, it has been demonstrated that the propagation of brucellaphages on different indicator strains caused changes in the phage genome, which might alter host specificity (Tevdoradze et al., 2015). However, lysis patterns of the phages have not yet been determined and compared. We therefore address this issue in detail.
Unlike TbV and FiV, the phages WbV (group III), Bk2V (group IV), R/CV (group V), and IzV (group VI) revealed lysis patterns that correlated well with published data. The species B. ceti and B. inopinata were not infected by any phage. Five phages inhibited the growth of the B. ceti reference strain but replication of the phages did not occur. Though, a number of B. ceti strains isolated from cetaceans were lysed by Weybridge and Izatnagar in another study (Foster et al., 2007). In contrast, for B. inopinata which is currently represented only by its type strain BO1, no lytic phage has been found so far. Since this strain harbors an active prophage (Hammerl et al., 2016) that might affect the susceptibility to other phages, the importance of the prophage was analyzed in more detail (see next chapter). To avoid misinterpretation of phage typing results, plaque assays allowing a quantitative determination of the lytic activity are certainly better suited than spot assays using high titer lysates, since lysis from without effects are too variable to identify species unequivocally. We also studied infection of other species (Ochrobactrum spp., Mesorhizobium sp., Sinorhizobium meliloti, Yersinia enterocolitica O:9, Pseudomonas sp.) by the brucellaphages and detected growth inhibition of some strains (data not shown). As the zones of growth inhibition looked similar to halos obtained with Brucella, such results could be misleading.
In this set of experiments the question should be answered whether lytic specificity of the phages may be affected by a change of propagator strains. For that reason TbV, WbV, R/CV, and IzV were co-cultivated with B. abortus S19, B. melitensis 16M, B. suis 1,330 and B. ovis 63/290. After overnight incubation, phages were isolated and used for the next co-cultivation with the respective strain. This procedure was repeated 20 times corresponding to ~500 generations. Thereafter, host ranges of the phages were examined by plaque assays testing all reference and type strains (Table S1). Following this procedure, no change of lysis patterns was detected. Furthermore, no adaptation of the phages to new hosts was observed. Phage R/CV e.g., remained its specificity and infected exclusively rough strains. The data suggest that even though propagation of brucellaphages on different strains may cause genomic changes (Tevdoradze et al., 2015) this is not necessarily associated with an alteration of the host range. However, the number of phage particles released from individual strains can differ significantly (data not shown). Thus, lysates of the same phage exhibited different titers which may bias a result. To avoid diverging lysis patterns, we recommend to propagate diagnostic brucellaphages on the same indicator strain and to examine the phage genomes by sequencing if results are inconsistent.
Since many phages have been isolated from Brucella cultures, a lysogenic state termed pseudolysogeny has been suggested for these phages (Renoux and Suire, 1963). Lysogeny may influence the susceptibility of the bacteria to phages. We investigated B. abortus S19 colonies that had survived infection by phage TbV. The isolated colonies were passaged several times. While a release of phage particles after mitomycin C treatment (Hammerl et al., 2016) was not observed, TbV was identified in initial cultures by PCR. In addition, electron microscopy revealed TbV particles adsorbed to the cell wall (data not shown). Even after repeated cultivation, the phage was no longer detectable. The data indicate that strain B. abortus S19 may serve as carrier for TbV but that there is obviously no integration of the phage genome into the bacterial chromosomes. We also did not observe any immunity of the TbV-carrying bacteria against superinfection by the same or other Brucella reference phages. Similar results were published by other authors (Morris et al., 1973; Corbel and Morris, 1975). To elucidate whether TbV-induced cell lysis may be affected by endogenous prophages providing immunity to superinfection, a S19 derivative (S19lys) containing the temperate phage BiPBO1 (Hammerl et al., 2016) was studied. TbV lysed the lysogenic strain like the original strain without the prophage. Thus, BiPBO1 did not affect TbV propagation. However, lysates prepared with strain S19lys contained both TbV and BiPBO1 which could be easily identified by their different plaque morphologies. As the BiPBO1 prophage was induced by infection with TbV, lysogeny has to be taken into account when lysates of brucellaphages are prepared. Otherwise, incorrect results might be obtained when the phages are used for typing.
Previous studies on the Brucella reference phages Tbilisi, Weybridge, Bk2, and R/O (an instable variant of phage R/C) showed that their genomes cannot be distinguished by restriction analysis using the endonucleases BamHI, EcoRI, and PvuII, because identical fragment patterns were obtained (Segondy et al., 1988). In contrast, phage Nepean (Np) revealed some differences, e.g., an additional 1.0 kb fragment in the HindIII digest (Rigby et al., 1989). The hitherto sequenced brucellaphage genomes mainly differed by two major InDels and can thus be assigned to two groups (Flores et al., 2012; Farlow et al., 2014; Tevdoradze et al., 2015). The presence or absence of these sequences should be traceable by use of suitable restriction enzymes. We analyzed HindIII restriction patterns of the VLA reference phages in detail to ascertain whether the yet not sequenced phages Bk2V and IzV belong to one of the existing groups. As documented in Figure Figure1,1, phage TbV and FiV showed two additional restriction fragments, 5.0 kb and 2.8 kb in size, which were absent in WbV and R/CV. These fragments comprise DNA sequences that are missing in the latter phages. The fragments were also absent in the Bk2V restriction digest. This suggests that Bk2V may exhibit identical deletions as Weybridge and R/C. The two above mentioned restriction fragments of TbV and FiV were detected in IzV indicating that this phage resembles TbV and FiV in this respect. In summary, restriction analysis using HindIII is a fast and easy method to determine whether a new phage contains deletions similar to already known brucellaphages.
Host range analyses showed that the six VLA reference phages infected a distinct range of strains, even though some phages revealed almost identical lysis patterns. The only difference that was observed between TbV and FiV was B. suis strain 1,330 that was infected by TbV but not by FiV. Similarly, the host range of WbV, Bk2V and IzV only differed by one or two strains (Table (Table1).1). To allow a comparison between host specificity and genome variations, the phages were sequenced, two of them (Bk2V and IzV) for the first time. The remaining four phages (Tbilisi, Firenze, Weybridge, and R/C) have already been characterized in previous studies but none of the genomic sequences determined in this work was identical to those described by other authors. TbV showed eight SNPs compared to phage TbW deposited at the Félix d'Hérelle Reference Center for Bacterial Viruses, Université Laval, Canada (Farlow et al., 2014). Five SNPs are silent mutations, two others lying within ORF21 (neck protein) and ORF44 (hypothetical protein) caused an amino acid exchange and one (within ORF14) is an insertion resulting in a truncated protein of unknown function. All these SNPs plus 14 additional SNPs exist when TbV is compared with TbE isolated at Eliava, Tbilisi, Ukraine (Tevdoradze et al., 2015). Differences to TbM (Gamaleya Scientific Research Institute of Epidemiology and Microbiology, Moscow, Russia) are even more pronounced and have already been addressed by Farlow et al. (2014). Contrary to TbV, FiV exhibited only two SNPs compared to Firenze in Laval. One SNP is located in an intergenic region, the other SNP caused an amino acid exchange at the C-terminus of a hypothetical protein. WbV showed five single nucleotide deviations to its counterpart in Laval. Three of them are located in genes (ORF16 and ORF23) for structural proteins, one in ORF27 probably encoding a tail collar protein and one in ORF57 for a primase/DNA polymerase. All of them caused amino acid exchanges. The most pronounced discrepancies between two phages with identical designations were found in R/C. The genomic sequence of the VLA R/C phage strain is 45 bp shorter than that of R/C in Laval. Furhermore, nine SNPs and two InDels were identified. Most deviations (four amino acid exchanges and one deletion of two amino acids) were found in the tail collar protein. Two frame shift mutations leading to radically changed gene products are present in ORF11 and ORF14 encoding a hypothetical protein and a primase/DNA polymerase, respectively. Quite the opposite was observed for Bk2V. The host range of this phage differs significantly from that of Bk (Corbel, 1987; Farlow et al., 2014) but on the genome only one SNP located in the tail collar protein gene leading to an amino acid exchange was detected.
Phage IzV exhibits the largest genome (41,446 bp) of all hitherto described brucellaphages. It contains a 301 bp duplicated nucleotide sequence located between the ORFs 23 and 24, which code for tail fiber proteins. Apart from this deviation, the IzV genome composition is similar to those of TbV and FiV as it does not carry the two deletions present in other Brucella reference phages (Figure (Figure2).2). However, based on the SNP data the closest relative of IzV is not TbV or FiV, but phage Bk2V. Besides the two InDels there are only five SNPs in these two phages. All of the SNPs are similarly present in the phages TbV, FiV, WbV, and R/CV. The SNPs are located in genes for the large terminase subunit, neck protein, a hypothetical protein and the primase/DNA polymerase (Table S3). TbV and FiV revealed 17 additional SNPs which are spread all over the phage genomes (Figure (Figure2).2). Another interesting feature of IzV is that the gene for the primase/DNA polymerase contains an internal stop codon, caused by the deletion of a single nucleotide. Therefore, in IzV the largest gene of brucellaphages is splitted into two smaller ORFs. While the primase/DNA polymerase of TbV, for instance, comprises 780 amino acids, the ORFs 57 and 58 of IzV encode polypeptides of 496 and 284 amino acids. Because of the frame shift mutation, the eight C-terminal amino acids of the large IzV polypeptide diverge from the TbV protein. The small polypeptide exhibits no differences to the TbV sequence. Our analysis of the R/CV genomic sequence revealed that this phage also contains a stop codon within the primase/DNA polymerase gene. In this phage polypeptides of 236 and 555 amino acids are encoded. The data demonstrate that several variants of the primase/DNA polymerase exist in brucellaphages. Whether the two proteins of IzV and R/CV possess the same activity as their larger counterpart in TbV is unknown and has to be clarified by further experiments.
On the basis of host range and whole genome analyses, Farlow et al. (2014) divided the Laval reference phages into three groups. Group I is composed of Tbilisi and Firenze, group II includes Berkeley, R/C and Pr from Mexico and group III contains Weybridge and S708. Some of these phages were also investigated in this study. However, since the VLA and Laval reference phages are not identical, they cannot be easily compared. From the data obtained in our study, the VLA reference phages can be allocated to four groups. As in the classification mentioned above, TbV and FiV belong to group I. Group II consists of WbV and Bk2V, which exhibited an almost identical host range and which are genetically closely related. Phage R/CV represents group III. It possesses an unique host specificity and showed some deletions and frame shift mutations not occurring in the other reference phages. The fourth group is represented by IzV which does not fit to the other groups, neither by its host range, nor in terms of its genomic sequence. Due to the close overall DNA homologies of brucellaphages and because nucleotide variations can be observed after changing the host strain (Tevdoradze et al., 2015), the question arises, whether it really makes sense to group these phages. In addition, it should be considered that Brucella strains having the same designation do not need to be identical. This also can distort results, e.g., the determination of the host range of the phages.
Sequence determination of Brucella reference phages disclosed a 2.4 kb DNA fragment comprising two genes for tail fiber proteins present in TbV, FiV, and IzV, but absent in WbV, Bk2V, and R/CV (Figure (Figure3).3). Subsequent InDel analysis revealed some remarkable consistencies among various phages. Compared to the other reference phages, WbV, Bk2V, and R/CV show exactly the same deletion of 2,443 bp. In TbV, FiV, and IzV, the fragment is flanked by a 9 bp direct repeat termed RS-A and RS-B (5′-GACCAACCC-3′, Figure Figure3).3). A third copy of this sequence (RS-C) exists in reverse complement orientation ~700 bp apart from RS-A. By contrast, the WbV, Bk2V, and R/CV genomes contain only one copy of this sequence, adjacent to the deleted fragment. Notably, the 301 bp sequence that is duplicated in IzV also borders with one end on RS-A. At the other end of the duplicated sequence, a similar motif (5′-ACCAAACCC-3′) is located in reverse complement orientation (Figure (Figure3).3). This sequence does not exist in the other VLA Brucella reference phages. The duplication resulted in the generation of the new ORF 25 in IzV. These data suggest that the identified repeats may be important for the acquisition, loss or duplication of DNA sequences in brucellaphages. The additional 1.0 kb HindIII fragment identified in phage Nepean but not in other reference phages has also been suggested to be a repetition as it hybridized to Tbilisi DNA (Tevdoradze et al., 2015). It would be interesting to learn whether the repetition in this phage is similarly flanked by the repeats described above.
This study, and also those of other authors revealed several difficulties that may arise, when certain Brucella strains are typed using the existing reference phage typing set. The main reasons for this are the very similar host ranges of the phages and the fact that lysis from without effects, which are more difficult to interpret than plaques, have yet been included in the evaluation of lysis patterns. In addition, there are apparently host range variants of the reference phages, which further exacerbate the situation because it makes the comparison of data collected in different laboratories difficult. We therefore determined the host range of 22 non-reference phages deposited in the Brucella phage collection of the BfR (Table S2) to identify further candidates for typing. The analysis disclosed a very similar host specificity of the phages (Table (Table2).2). Though, the host range was clearly different from those of the reference phages. Like TbV and FiV, none of the phages lysed strains of B. melitensis and B. pinnipedialis. All phages infected at least two, most of them even three B. suis biovars, namely bv1, bv4, and bv5. In this regard, the phages resemble WbV, Bk2V, and IzV. Lysis of Brucella species was similar to the reference phages. The exact analysis of the phages' host range also revealed some individual differences. This particularly pertained to the B. suis bv5 strain 513 that was not lysed by five phages. Brucella microti CCM4915T was resistant only to one phage. Besides these specificities, all phages infected the same strains. It should, however, be emphasized that the phages produced a reduced number (up to 4 log units) of plaques on B. suis and B. microti strains. This has to be considered when the phages are used for typing. Two of the non-reference phages (A422 and M51) have been already studied by Morris et al. (1973). While we found plaques on B. suis with both phages, the other authors reported on lytic activity at RTD only for M51. It is conceivable that their A422 lysate did not contain enough active phage particles to cause lysis at RTD. Nevertheless, in spite of this discrepancy, the data of this study suggest that the reference phage typing set could be complemented by additional phages, which would make the discrimination of some Brucella species and strains more reliable.
The sequence analysis of the Brucella reference phages disclosed several SNPs (>50 SNPS, regions A–Q, Figure Figure2),2), most of them causing amino acid exchanges or deletions. Because no data were available about the genomes of the 22 non-reference phages, we examined all SNP positions by PCR using 23 different primer pairs. The study showed that all phages differed in at least one SNP and that none of the phages exhibited the two large deletions found in WbV, Bk2V, and R/CV. Taking into account all SNPs, two clusters of non-reference phages were assigned. One cluster (cluster B) consists of the phages A422 and M51 whose closest relatives are WbV, Bk2V, IzV, and R/CV. The remaining phages form a cluster (cluster C), which is related to cluster B (Figure (Figure4).4). As the neck and tail collar protein genes of brucellaphages have been suspected to be important for host specificity (Flores et al., 2012; Farlow et al., 2014; Tevdoradze et al., 2015), we focused on the occurrence of SNPs within these genes. In the six VLA reference phages, the neck and tail collar protein genes exhibited five and seven SNPs, respectively. The analysis of the non-reference phages revealed even higher numbers of SNPs, some of which are located at the same position as in the reference phages (Table S3). A comparison of SNPs within the neck and tail collar protein genes with the host ranges of the phages did not provide evidence for amino acids that may decide on the strains that are infected. However, among 18 non-reference phages that exhibited an identical host range, 10 amino acid exchanges were observed in the neck protein and 11 in the tail collar protein, indicating that these positions do not determine specificity. Hence, even though the neck and tail collar protein genes are hotspots for nucleotide variations, it remains open which sequences are the key factor defining the host range of the phages.
Lytic phages have been applied for decades to identify and discriminate Brucella species and biovars. Moreover, the set of reference phages used in different laboratories is basically the same. The phages were isolated many years ago and they were distributed to diagnostic laboratories worldwide. The same holds true for Brucella reference and type strains serving as controls. After sequencing of the first Brucella reference phages, it became obvious that there is not only one Tbilisi, Firenze, or Weybridge phage but some variants of the prototypes revealing distinct sequence alterations. Nucleotide variations particularly occur in genes probably involved in host specificity. Therefore, it is not surprising that the VLA reference phages investigated in this study partially exhibited a host range different from homonymous phages in other laboratories. However, it is important to point out that in our study, lytic activity was defined as the ability of the phages to form plaques, while rather unspecific lysis from without effects caused by extremely high numbers of phages were not evaluated. We think that this practice delivers more reliable data because single plaques are much easier to interpret than halos of lysis which may look very different. Results can also be biased when typing phages are propagated on different hosts, or when strains containing endogenous phages are infected. The B. abortus vaccine strain S19 is a well-suited host because it is susceptible to many Brucella phages and a Biosafety Level 2 (BSL2) organism. Thus, there are some issues that should be considered when brucellaphages are applied for typing. One main problem with lysotyping of Brucella strains is the similar host specificity of the reference phages. The situation could be improved by adding new phages to the typing set, which exhibit an individual host range. Our analysis of 22 non-reference brucellaphages revealed some new candidates that could be applied for routine diagnostics.
JH, HS, KN, and SH designed the study. JH, CG, CJ, and JR performed the experiments. JH, CG, CJ, HS, JR, SA, and SH analyzed the data. JH, CJ, and SH wrote the manuscript and prepared the tables and figures. All authors edited the manuscript.
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.
The study was financially supported by a grant of the German Federal Institute for Risk Assessment (1332-488). Parts of the study were conducted within the Ess-B.A.R. project (FKZ: 13N13982).
The Supplementary Material for this article can be found online at: http://journal.frontiersin.org/article/10.3389/fmicb.2017.00408/full#supplementary-material
Bacterial strains used in this study.
Bacteriophages used in this study.
Single Nucleotide Polymorphisms (SNPs) among brucellaphages.
ORF analysis of brucellaphage Iz.