Isolation and identification of a pool of strains of Psychrobacter spp.
Bacterial strains were isolated from guano of little auks—the most abundant Arctic sea birds. A random selection of 88 isolates (Table S1, Supplementary Materials) (obtained from 6 independent guano samples) was examined for their colony and cell morphology, as well as basic physiological features (data not shown). Fragments of 16S rDNA were amplified by PCR from the strains and sequenced. Comparative analysis of the obtained sequences revealed that 17 of the strains (19.3 %) could be classified to the genus Psychrobacter. One of the randomly chosen strains (Psychrobacter sp. DAB_AL62B) was already described in our previous study (Lasek et al. 2012). The remaining strains were subjected to further analysis. To avoid the characterization of strains of clonal origin, isolates from different guano samples were examined.
Phylogenetic analysis was performed, based on the comparison of partial 16S rDNA sequences (1351 bp) of the strains and type strains representing the 34 Psychrobacter species described to date. The analyzed sequences of the strains DAB_AL32B, DAB_AL43B, DAB_AL60 and DAB_AL109bw were identical, and share 99.70 % similarity with DAB_AL25. The topology of the phylogenetic tree revealed that these five strains form a separate cluster linked to P. frigidicola DSM 12411 (99.11 % identity of the 16S rDNA sequences), isolated from ornithogenic soil in Antarctica (Bowman et al. 1996) (Fig. 1). The strains DAB_AL62B and DAB_AL12 were clustered in different groups. The former is most related to P. urativorans DSM 14009 (99.26 % identity) and P. cibarius JG-219 (99.18 % identity), isolated from ornithogenic soil and food products, respectively (Bowman et al. 1996; Jung et al. 2005), while the latter (most divergent among the “DAB” strains) is closest to P. cryohalolentis K5 (99.78 % identity), isolated from a cryopeg taken from the permafrost in the Kolyma lowland (Siberia, Russia) (Bakermans et al. 2006) (Fig. 1).
Characterization of the Psychrobacter spp. strains
A preliminary characterization of the isolated Psychrobacter spp. strains revealed that all of them fulfill the requirements of psychrophilicity (Morita 1975): they could grow at temperatures ranging between 4 and 25 °C, but not at ≥30 °C (optimum temperature 22 °C) (Figure S1, Supplementary Materials). These strains also grew in LB medium at pH values between 5 and 10, which is typical for neutrophilic bacteria (Slonczewski et al. 2009).
Possible resistance phenotypes, which are often determined by plasmids and transposons, were also examined. None of the Psychrobacter spp. strains were resistant to any of the tested antibiotics (ampicillin, chloramphenicol, kanamycin and tetracycline), but they showed resistance to several heavy metals: (1) low or moderate resistance to zinc, chromium(VI) and copper (MICs from 2 to 4 mM), and (2) moderate or high level resistance to arsenate—As(V) (MICs from 15 to 100 mM), except DAB_AL25 (MIC 2 mM). Two strains (DAB_AL12 and DAB_AL32B) also exhibited trigger level resistance to arsenite—As(III) (MICs of 1.5 and 2 mM, respectively).
Identification of a pool of Psychrobacter spp. plasmids
Plasmids are natural vectors that play a major role in the dissemination of accessory genetic information in HGT. The Psychrobacter strains analyzed in this study were found to carry nine circular plasmids (ranging in size from 2.9 to 14.9 kb) that are listed in Table 1. The highest number of plasmids was found in DAB_AL43B (4) and DAB_AL60 (2), while the remaining strains possess only single replicons or none at all.
The complete nucleotide sequences of these plasmids were determined. The relatively low average GC content of the obtained sequences (35.7–42.9 mol%) is typical for Psychrobacter spp. genomic DNA (e.g. Ayala-del-Río et al. 2010). The plasmids were found to contain from 2 to 13 open reading frames (ORFs), of sizes between 165 and 2133 bp (Table 1). Based on similarities to known genes, it was possible to predict functions for the polypeptides encoded by almost half of these ORFs. A summary of the ORFs, including their position, transcriptional orientation, the size of the encoded proteins, and their closest known homologs, is presented in Table S2 (Supplementary Materials).
Further analysis of the organization of the plasmids revealed the presence of several putative genetic modules responsible for (1) plasmid replication (REP), (2) stabilization (STA), and (3) mobilization for conjugal transfer (MOB). All plasmids also appear to carry different accessory genetic information (Fig. 2).
REP modules: structure and host range
Bioinformatic analysis of the plasmid genomes indicated the presence of two types of REP module: (1) repB-like (pP12P1, pP43BP1, pP43BP2, pP43BP3 and pP60P2) and (2) repA-like (pP32BP1, pP43BP4, pP60P1 and pP109bwP1). The sequences of these modules could be differentiated by their GC content: 35.7–39.2 mol% for repB-like and 41.3–42.9 mol% for repA-like plasmids (Table 1).
The characterized REP modules of pP12P1, pP43BP1, pP43BP2, pP43BP3 and pP60P2 have a structure that is typical for the replication systems of many theta-replicating plasmids (Chattoraj 2000). They contain a single ORF encoding a predicted protein with similarities to the initiator RepB protein, possessing nicking-closing (topoisomerase I like) activity, and a putative origin of replication (oriV), placed upstream of the repB gene (Fig. 3a). Comparative analysis revealed that closely related replication proteins are also encoded by plasmid 1 of P. cryohalolentis K5 (acc. no. NC_007968) and many plasmids of Acinetobacter spp. strains (e.g. pMAC and p11921 of Acinetobacter baumannii, acc. nos. NC_006877 and GU979000, respectively).
The putative oriVs of pP12P1, pP43BP1, pP43BP2, p43BP3 and pP60P2 contain four tandemly placed 22-bp long repeated sequences (IT1–IT4), i.e. putative iterons, which (as shown for other related plasmids of this type; Chattoraj 2000) may constitute binding sites for the Rep proteins. The IT repeats are located 81–132 bp upstream of the repB genes. They are identical in particular plasmids (with one exception being IT4 of pP43BP2, which differs from IT1 to IT3 by 5 mismatches); however, they show only limited reciprocal sequence similarity (Fig. 3a). Interestingly, the plasmids pP12P1, pP43BP1, pP43BP2 and pP60P2 contain a conserved palindromic sequence [5′-TAA(A/C)AGCTTTTA-3′] located 37–46 bp upstream of the IT1 repeats (Fig. 3a). In the case of pP43BP3, a similar sequence (5′-AAATTCATTT-3′) is situated between the IT2 and IT3 repeats (Fig. 3a). High conservation of the palindromic structure might suggest a role in replication initiation.
The REP modules of the second group of plasmids (pP32BP1, pP43BP4, pP60P1 and pP109bwP1) contain a single ORF encoding a putative protein in which three conserved regions can be distinguished: (1) the replicase domain, typical for plasmid DNA replication initiator proteins, (2) an alpha helical domain, found in the C-terminal regions of primases (PriCT-1), and (3) a HTH motif, that is most probably responsible for protein–DNA interactions (data not shown). The analyzed proteins share significant amino acid sequence homology with the RepA proteins of plasmid 1 of P. cryohalolentis K5 and pRWF101 of Psychrobacter sp. PRwf-1 (acc. nos. NC_007968 and NC_009516, respectively), and with the replication initiation protein of E. coli plasmid ColE2 (Yasueda et al. 1989). Careful inspection of the nucleotide sequences of the analyzed REP modules also revealed the presence of three DNA regions typical for the origins of replication of ColE2-type plasmids: (1) two direct repeats (L and R) 5′-CAGATAA-3′, (2) sites α and β, which determine the specificity of the interactions of Rep protein with the origin, and (3) a short sequence to which the Rep protein synthesizes a unique RNA primer that is crucial for the initiation of leading-strand DNA synthesis by DNA polymerase I (Fig. 3b) (Yagura et al. 2006).
The host range of REP modules representing both of the aforementioned groups of replicons was then examined. For this analysis we used two shuttle plasmids, pABW-12P1 and pABW-60P1, containing the REPs of pP60P1 and pP12P1, respectively, cloned (within XbaI restriction fragments) into the E. coli-specific, narrow host range (NHR) vector pABW1 (ColE1-type ori of pMB1). The ability of the shuttle plasmids to replicate was tested in (1) A. tumefaciens LBA 288R and P. versutus UW225 (Alphaproteobacteria), (2) Alcaligenes sp. LM16R (Betaproteobacteria), and (3) P. putida KT2442R (Gammaproteobacteria). The plasmids pABW-12P1 and pABW-60P1 were found to replicate exclusively in Psychrobacter spp., suggesting a relatively NHR.
Stable maintenance modules
Most of the analyzed plasmids appeared to lack stabilization systems, which are components of the vast majority of bacterial replicons. Only plasmid pP43BP3 carries a complete toxin–antitoxin (TA) system, possibly involved in the postsegregational elimination of plasmid-less cells from a bacterial population (Fig. 2). This putative TA module is composed of two short overlapping ORFs (14 bp overlap) encoding proteins with similarity to a number of RelB-like antitoxins (ORF5) and RelE-like toxins (ORF4) of relBE/parDE-type TA systems (Anantharaman and Aravind 2003). Two other plasmids, pP43BP2 and pP109bwP1, carry incomplete TA modules represented by single ORFs encoding RelB-like antitoxins.
Another plasmid, pP60P2, contains a putative type II restriction-modification (R-M) system. Similar to TA, such systems may increase the stability of plasmids by killing plasmid-less cells (e.g. Ichige and Kobayashi 2005; Dziewit et al. 2011b). The RM module of pP60P2 is composed of two divergently orientated ORFs: ORF5 and ORF6 (Fig. 2). BLAST searches revealed that the polypeptide encoded by ORF5 shares substantial homology with a large number of proteins annotated as m5C methyltransferases (MTases), with highest sequence similarity (60 %) to a putative MTase of Marivirga tractuosa DSM 4126 (acc. no. YP_004054233). The ORF6-encoded protein is similar to a putative restriction endonuclease (predicted recognition sequence 5′-CGCG-3′) encoded by Moraxella catarrhalis BC1 (acc. no. ZP_11632318).
ORF10 of pP60P2 (Fig. 2) encodes a predicted protein with a catalytic domain characteristic of serine recombinases, often recognized as resolvases in multimer resolution systems (MRS). MRS act to resolve plasmid oligomers and this activity increases the number of independent plasmid molecules available for distribution during cell division (Bahl et al. 2009). Homologs of ORF10 have been identified in other Psychrobacter plasmids, including pRWF101 of PRwf-1 and pP62BP1 of DAB_AL62B (Lasek et al. 2012).
MOB modules: structure and diversity
Many bacterial plasmids can be mobilized for conjugation by other self-transmissible elements (e.g. conjugative plasmids and integrative and conjugative elements—ICEs) encoding type 4 secretion systems. The mobilizable plasmids contain MOB modules encoding specific relaxosome components and an origin of transfer (oriT) (Francia et al. 2004; Garcillan-Barcia et al. 2009). The analyzed plasmids of Psychrobacter spp. were found to contain eight putative MOB modules, which may be classified, based on amino acid sequence similarities of their relaxases, into three distinct families (MOBP, MOBQ and MOBV).
Analysis of the MOBP family [MOBP5 (MOBHEN) clade] members, found in plasmids pP43BP1, pP43BP4, pP60P1 and pP109bwP1, revealed their structural divergence and permitted the identification of two subgroups. The single member of the first subgroup (MOB of pP43BP1) is composed of two overlapping ORFs (putative mobA and mobC), while the other three MOBP modules carry four convergently orientated ORFs (Fig. 4). The MobA relaxases of the two subgroups show partial amino acid sequence identity (≤36 %). The pP43BP1 MobA is most similar (44 % identity) to the relaxase/mobilization nuclease domain protein of plasmid pRWF101 of Psychrobacter sp. PRwf-1 (acc. no. NC_009516). In turn, the MobA proteins of pP43BP4, pP60P1 and pP109bwP1 are most similar (≥57 % identity) to the relaxase encoded by plasmid pKW1 of Pseudoalteromonas sp. 643A (Cieśliński et al. 2008). Interestingly, the three relaxases of this second subgroup are at least twice the length of other MobA proteins, and their C-terminal parts show no similarity to known protein sequences present in the GenBank (NCBI) databases. All of the analyzed MOBHEN modules contain predicted oriTs (located upstream of the mobC genes), which share sequence similarity with oriT of the mobilization system (MOBHEN clade member) of plasmid pSW200 of Erwinia stewartii SW2 (Fu et al. 1998) (Fig. 4).
The MOBQ family is the most abundant and diverse group of mobilization systems. Two such modules (encoding relaxases most similar to proteins classified within the MOBQ3 clade; Garcillan-Barcia et al. 2009) were identified within plasmids pP32BP1 and pP60P2. Both MOBs are composed of two non-overlapping, divergently orientated ORFs, encoding a putative relaxase (MobA) and a mobilization protein C (MobC), as well as a predicted oriT located within the mobA–mobC intergenic region (Fig. 4). The oriTs display 55 % sequence identity and they match the oriT consensus sequence of the MOBQ family (Francia et al. 2004) (Fig. 4). The relaxases of pP32BP1 and pP60P2 differ significantly in length and they exhibit only a moderate level of sequence similarity (40 %). Similarity searches revealed that the predicted proteins are most closely related (52 %) to the MobA protein encoded by a small (4658 bp) cryptic plasmid pMbo4.6 of Moraxella bovis ATCC 10900 (acc. no. NC_013500).
The MOB modules of plasmids pP43BP2 and pP43BP3 (Fig. 4) encode related relaxases (85 % identity), which exhibit ≥40 % amino acid sequence homology with the Mob protein of the mobilization system of a broad host range plasmid pBBR1 (2687 bp) of Bordetella bronchiseptica S87—the prototype of the MOBV2 clade (MOBV family) (Antoine and Locht 1992; Garcillan-Barcia et al. 2009). The predicted oriTs of these plasmids (both placed upstream of the mobA genes) are identical, and differ slightly from the oriT of pBBR1 (Fig. 4). They contain 8-bp long inverted repeats (IR) separated by a 3-bp spacer, which is a typical structure for oriTs of MOBs of the MOBV family (Szpirer et al. 2001; Guzman and Espinosa 1997).
Accessory genetic information
Besides the REP, MOB and STA modules, the analyzed plasmids were found to contain accessory genetic information (33 ORFs in total) of unknown function. Using the PRIAM program for automated enzyme detection we were able to assign putative functions to only four of the identified ORFs. ORF4 and ORF8 of pP60P2 and ORF6 of pP32BP1 (Fig. 2) encode putative Sel1 domain proteins, with weak homology (35–39 % amino acid sequence similarity; E value >1e−10) to beta-lactam hydrolases (EC 188.8.131.52)—a group of enzymes of varying specificity that hydrolyze penicillins or cephalosporins. ORF7 of plasmid pP60P2 encodes a putative protein with 99 % amino acid sequence similarity to subunit C of alkyl hydroperoxide reductase (EC 184.108.40.206) encoded by the ahpC gene of Psychrobacter arcticus 273-4. This enzyme reduces H2O2, organic peroxides and peroxynitrite (ONOO−), and therefore acts as an antioxidant and major scavenger of reactive oxygen species (ROS) generated in the cytoplasm of bacteria as a by-product of aerobic metabolism (Chen et al. 1998; Steele et al. 2010).