Complete genome sequences and genomic characterization of five plasmids harbored by environmentally persistent Cronobacter sakazakii strains ST83 H322 and ST64 GK1025B obtained from powdered infant formula manufacturing facilities

Cronobacter sakazakii H322 and GK1025B were grown in 5 ml of Trypticase Soy Broth (TSB, BBL, Becton Dickinson, Franklin Lakes, NJ, USA) supplemented with 1% NaCl (TSBS), and incubated at 37 °C for 18 h with shaking conditions of 160 rpm (Thomas Scientific, Inc., Swedesboro, NJ, USA). Isolation of genomic DNA was performed using a 2 ml aliquot of each culture using the robotic QIACube workstation and the automated Qiagen DNeasy technology (Qiagen, Inc., Germantown, MD, USA) following the manufacturer’s recommendations as described by Jang et al. [17, 18].

The single-molecule real-time (SMRT) sequel sequencing technology [19] from PacBio (Pacific Biosciences, Menlo Park, CA, USA) was utilized to create high-quality long-read datasets of C. sakazakii strains H322 (SRR8305966) and GK1025B (SRR8305970). The initial processing of long-sequencing reads was carried out using the RS_HGAP_Assembly.2 protocol (default parameters) implemented in the Pacific Biosciences SMRT analysis portal (version 2.3.1). Quality filtering was performed automatically during assembly using the SMRT Portal P-filter module and using the Hierarchical Genome Assembly Process 3 (HGAP3) pipeline. For generating complete genomes, a hybrid assembly strategy with UniCycler assembly software [20] implemented on the Pathosystems Resource Integration Center (PATRIC) database web-server (https://​patricbrc.​org/​app/​Assembly2). Long-read short read archive (SRA) files from PacBio and corresponding WGS datasets of the strains obtained from sequencing runs performed on an Illumina MiSeq platform (Illumina, San Diego, CA, USA) [15, 16, 29] were combined following the instructions on the web-server. The Prokaryotic Genome Annotation Pipeline (PGAP) annotations [21] of these completed genomes, plasmid sequences, and their accession numbers were released under FDA GenomeTrakr Bioproject on NCBI (PRJNA258403), which is part of FDA’s foodborne pathogen research comprehensive Bioproject at NCBI (PRJNA186875). The RAST Seedviewer was used to help provide consistent and accurate genome annotations across the genomes and plasmid sequences [22].

The PROKSEE server (https://​beta.​proksee.​ca/​projects) was used to generate high-quality navigable maps of each circular plasmid as previously described [23]. Each Cronobacter plasmid’s sequence was submitted to CGE’s Plasmidfinder (https://​cge.​cbs.​dtu.​dk/​services/​PlasmidFinder/​) for in silico determination of incompatibility plasmids such as IncF, IncHI1, IncHI2, IncN, and IncI1 plasmids [24]. For prophage sequence identification, C. sakazakii strain FASTA data sets were uploaded to the PHASTER (PHAge Search Tool Enhanced Release) web server and pipeline (https://​phaster.​ca/​, last accessed 8.25.2021, [25, 26]). Mauve, Progressive Mauve, and Geneious suite 12.0 ([27]; https://​www.​geneious.​com/​) were used for alignment and visualization as needed. BLAST analysis for the presence of pH322_1 was performed on an in-house database of 683 genomes consisting of GenomeTrakr datasets along with publicly available genomes hosted at NCBI.

The characteristics of the completed genomes and closed plasmids harbored by C. sakazakii strains H322 and GK1025B are summarized in Table 1. Each genome consisted of a single circular chromosome of 4,350,614 bp and 4,362,605 bp in size, contained a GC content of 56.7% and 56.9%, and 4,146 and 3,693 coding DNA sequences (CDS), respectively. Two plasmids were identified as being harbored by C. sakazakii H322 and three plasmids were identified to be carried by C. sakazakii GK1025B. None of the five plasmids identified by genome sequencing were predicted on CGE's Plasmidfinder [24]. In addition to the closed plasmids generated from long-read sequencing, PHASTER analysis showed that C. sakazakii strain H322 hosted four intact prophage sequences (Additional file 2: Table S2) that were located on the chromosome which included a 47.4 Kbp Salmonella SEN34 (NCBI accession #: NC_028699), a 37.4 Kbp Enterobacteria mEp235 (NC_019708), a 43.5 Kbp Salmonella 118970_sal3 (NC_031940), and a 17.7 Kbp Enterobacteria P1 (NC_005856), prophage. An incomplete generalized transducing Salmonella bacteriophage ES18 prophage (NC_006949) prophage was also identified. Three intact Cronobacter prophage sequences were additionally identified by PHASTER analysis on the chromosome of GK1025B: Cronobacter ENT47670 (NC_019927), Cronobacter ESSI_2 (NC_047854), and Cronobacter phiES15 (NC_018454). The complete chromosomal sequences of the two C. sakazakii allowed for detailed annotation and identification of mobilome sequences that could be applied for comparative analysis with other strains of Cronobacter and related organisms.

A 100,741 bp pCS1-like closed plasmid, named pH322_1 was identified to be similar to the plasmid pseudomolecule initially predicted from H322 draft whole genome contig sequences by Chase et al. [15]. The sequence relatedness of pH322_1 to pCS1 harbored by C. sakazakii NCIMB 8272 (alias NCTC 8155) after PROKSEE analysis using the β-version of CGView Server [23] is shown in Fig. 1A. It had a GC content of 50.2% and harbored 137 CDS. Unique features contained on this plasmid included 12 mobile genetic elements comprising six copies of an Insertion Sequence 3 family transposase, three exonucleases (3–5′ exonuclease, SbcCD subunit D, and an unnamed exonuclease), a RecA recombinase, and a site-specific integrase (Additional file 1: Table S1). PGAP analysis also identified several phage-related genes on pH322_1 such as genes encoding for a phage exonuclease, and a phage tail fiber identified as side tail fiber protein, Stf (Additional file 1: Table S1). PHASTER analysis identified and confirmed the presence of an intact 96.9 Kb Salmonella SSU5 (NCBI accession number: NC_018843) prophage gene cluster [25] (Additional file 2: Table S2) located on pH322_1.

The second plasmid named pH322_2 and harbored by C. sakazakii H322 was 118,185 bp in size and contained a GC content of 56.8%. There were 118 CDS identified by PGAP annotation. Analysis using PROKSEE [23], showed the plasmid to be closely related to C. sakazakii virulence plasmid pSP291_1 harbored by ST4 C. sakazakii SP291 as described by Power et al. [30]. BLAST analysis showed that the virulence plasmid pH322_2 shares significant homology with the virulence plasmid backbones of pSP291_1 and pESA3 (data not shown). They share conserved features like the origin of replication gene, repA, two iron acquisition systems, an aerobactin-like siderophore (named Cronobactin, iucABCD/iutA), and an ABC ferric-iron transporter gene cluster (eitCBAD) as described by Franco et al. [31].

Complete sequences of three plasmids, pGK1025B_1, pGK1025B_2, and pGK1025B_3 was obtained by long-read sequencing and PGAP annotation (Additional file 1: Table S1 [21]). pGK1025B_1 was 101,769 bp in size and contained 141 CDS, of which 70 genes encoded for hypothetical proteins (Additional file 1: Table S1) and had a GC content of 51.1%. An intact 99.4 Kbp gene cluster encoding for a Salmonella SSU5 prophage (NC_018843) was identified using PHASTER and was like pH322_1 (Additional file 2: Table S2) described earlier. This is a slightly smaller sized prophage SSU5 gene cluster than what was reported by Kim et al. (103 Kbp; 2012) but is slightly larger than the prophage gene cluster present in pH322_1 (96.9 Kb). As was the case for pH322_1, the prophage gene cluster contained genes encoding for prophage structural proteins including terminase, capsid and tail proteins. Genes encoding for a lysin, an integrase, and a recombinase protein, and possessed a GC content of 51.1% were also noted.

pGK1025B_2 was identified as a slightly smaller version (120,182 bp) of the virulence plasmid, pESA3 (131,196 bp) that Franco et al. [31] described for C. sakazakii strain BAA-894. It contained 133 CDS, possessed a GC content of 56.6%, and harbored a homolog of repA, the plasmid’s origin of replication gene and a homolog of Cronobacter plasminogen activator, cpa (location: 6338 to 7276 bp). As described earlier, other noted genetic features among these virulence plasmids include: a siderophore aerobactin biosynthesis gene cluster (now named Cronobactin/siderophore receptor, iucABCD/iutA), a bicistronic toxin-antitoxin gene complex encoding for HigB/HipA, and a methyl-accepting chemotaxis protein I (serine chemoreceptor protein) gene.

pGK1025B_3 is a plasmid of 46,528 bp in size. It possesses a GC content of 51.1% and harbored 50 genes encoding for hypothetical proteins; however, it is a conjugative plasmid like pESA2 which is harbored by C. sakazakii BAA-894 [31, 33]. A ~ 16.4 Kbp type 4 secretion system gene cluster was found, and most notably it contains within this gene cluster a copy of a phospholipase D gene (plD, located between ~ 9308 to ~ 9841 bp, NCBI Locus Tag: AUM97_022060).

Initial sequence analyses described above suggested that pGK1025B_1 and pH322_1 belong to a unique category of plasmids called phage-plasmids (extrachromosomal DNA molecules that host intact prophage sequences and strictly behave like plasmids) that have been known since the 1960s [34, 35]. Sequence comparison of repA gene sequences from pH322_1 and pGK1025B_1 (locus_tags BTK77_021130 and AUM97_021275, respectively) suggested that these plasmids possessed a mutually exclusive origin of replication and different sub-groups of IncF1B category. It was reported that SSU5 phage bearing plasmids usually belong to IncF1B incompatibility group in other Enterobactereaceae members [35]. Some previously reported Cronobacter plasmids belong to this rare category of phage-plasmids along with the two plasmids identified in this study suggesting an expansion of genetic diversity among this emerging foodborne pathogen [15, 36]. We identified sequences containing significant homology to pGK1025B_1 and pH322_1 from among the known Cronobacter plasmids by BLAST analysis. The properties of these prophages are summarized in Additional file 2: Table S2 and suggest a prevalence of phage-plasmid like sequences of varied lengths in different plasmids containing homologous sequences to SSU5. Figure 2A shows the prophage gene cluster region from SSU5 in comparison with a few selected related C. sakazakii plasmids like pH322_1, pGK1025B-1, pCS1, pCsaC105731a, pCsa767a, and pCsaC757b. PHASTER analysis also identified and confirmed an intact Salmonella SSU5 prophage in the C. muytjensii plasmid pCmuyJZ38_1 [36] and a E. coli P1-like prophage on p505108-MDR and pGW1 (Additional file 2: Table S2). Analysis using the Mauve progressiveAlignment tool [28] revealed the variations in the size of the prophage region in these plasmids when compared with SSU5 as reported by Kim et al. [37] (Fig. 2B). To understand the distribution and prevalence of the phage-plasmids within Cronobacter species, we performed a BLAST analysis on 683 draft WGS genomes and closed plasmids representative of all seven species using the genes found on pH322_1. The results identified the phage-plasmid like sequences (from pH332_1) in three C. malonaticus strains (two ST129 and a single ST7), and 133 C. sakazakii strains which represented 18 different STs and these results are shown in Additional file 3: Table S3.

As noted above, the acquisition of plasmids containing prophages was a unique finding which was initially reported by Ikeda and Tomizawa for prophage P1 in Escherichia coli [34]. It was reported that rather than integrating its prophage genome into the host bacterium’s chromosome, its DNA was found to replicate as a circular plasmid in the lysogen. Prophage SSU5 is like E. coli prophages P1 and D6, which are also harbored on plasmids, are common in the Enterobacteriales, and were also among the first prophages found to be associated with plasmids [38]. Plasmids and prophages are key contributors to bacterial evolution and when found together as a single unit are often now referred to as phage–plasmids, which possess properties of both plasmids and prophages, for example, P1, N15 or SSU5. Biological characterization of these phage-plasmids is poorly understood. Pfeifer et al. screened over 2500 phages and 12,000 plasmids from across a diverse collection of bacterial phyla and identified 780 phage–plasmids grouped into eight distinct categories based on sequence features. This study further suggested a role for the phage-plasmids in genetically connecting phages and other mobile (and transducing) genetic elements [35]. Salmonella prophage SSU5 represents a different type of lysogenic phage with a circular phage-plasmid that is also very common in other members of the Enterobacteriales; however, they are only very rarely annotated as being phage related, much less as prophages [38]. Often a genome may contain an integration hot spot such as that found for Lactococcus lactis subsp. cremoris which contains 20% of its genome as IS elements [39]. This suggests that a genome can exist in an active evolutionary state, as it can readily accommodate new DNA and/or loose genome regions as well. Similarly, pH322_1 and pGK1025B_1 with an abundance of mobile genetic elements found in these phage-plasmid sequences (Additional file 1: Table S1) may represent such a genetic element. Furthermore, the fact that C. sakazakii strains H322 and GK1025B contain multiple plasmids may offer selective advantages to a bacterial host which may also reflect their adaptative abilities to persist within the nutrient-rich environment of a built environment, such as that of powdered infant formula manufacturing facilities [40].

The shared genome backbone (Additional file 1: Table S1) of pH322_2 and pGK1025B_2 with that of virulence plasmids, pESA3 [31] and pSP291_1 [30] is shown in Fig. 1B. Both plasmids harbored a Cronobacter plasminogen activator (cpa) encoding Protease VII or Omptin precursor (EC 3.4.23.49) (Additional file 1: Table S1) homologous to the Salmonella outer membrane protease, PgtE [1, 31, 32]. Both the plasmids contained a truncated ~ 13 kbp type six secretion system (T6SS) gene cluster which shares homology with a similar region harbored by pESA3, and pSP291_1. PROKSEE analysis showed that a similar truncated T6SS with a large deletion in the region for SP291_1 and pH322_2 compared to that of pESA3. In addition to the deletion within the T6SS gene cluster, pH322_2 also had a second large deletion of ~ 6 Kbp, which includes genes for a tyrosine-type recombinase/integrase, a hypothetical protein, and a phospholipase D. These results correlate with those reported by Franco et al. [31]; Tall et al. [32], Chase et al. [15], and Jang et al. [1] who had described the presence of a virulence plasmid like pH322_2, pGK1025B_2, pESA3 and pSP291_1 in a high percentage of C. sakazakii strains (629 of 652, 96%). Two functional T6SS clusters were reported by Wang et al., 2018 in the C. sakazakii strain ATCC12868 although the genome sequences are not available on NCBI for comparison [42]. In contrast, truncated T6SS segments on pESA3-like virulence plasmids reported by Franco et al. [31] and others had not been characterized in vivo or share sequence homology with the chromosomal clusters rendering their use just as a possible ‘signature sequence’ for this category of plasmids. A Cobalt ABC transporter gene cluster encoding for an ATP-binding protein (CbtL), permease protein (CbtK), and two copies of a substrate-binding protein gene (CbtJ) were found on pGK1025B_2, but not on pH322_2 Additional file 4: Figure S1.

Sequence alignment of newly described pGK1025B_3 compared with other conjugative class members produced on the PROKSEE server (Fig. 3) suggest that this plasmid represents a new conjugative plasmid that only has marginal sequence homology with pEAS2 from C. sakazakii BAA-894 [1, 33]. Interestingly, a known virulence gene coding a phospholipase D (PLD) was identified within a complete T4SS cluster harbored on pGK1025B_3 (Additional file 5: Figure S2). Results of a BLASTn analysis using the rep (CP078109.1) gene from pGK1025B_3 queried against Enterobacteriaceae and related endosymbionts (NCBI taxid:91347), showed a shared homology with many related rep genes. Alignment of these gene sequences, shown in Fig. 4 revealed that the rep gene of pGK1025B_3 clustered distinctly separated from a larger cluster of 91 rep genes of related plasmids of members of the Enterobacteriaceae. These results suggest that the rep gene of pGK1025B_3 may represent a novel Cronobacter origin of replication gene carried by a previously uncharacterized Cronobacter conjugative plasmid that harbors within its gene cluster a phospholipase D gene. Future surveillance studies to identify the prevalence of pGK1025B_3 like plasmids as well as functional genetic studies are needed. Phospholipase D (PLD) represents a heterogeneous group of lipolytic esterases, which are either secreted into the extracellular milieu, or directly injected into the host cell cytosol by a wide variety of Gram-positive and Gram-negative bacteria through Type 6 and Type 4 secretion systems [41]. It plays an important role in several host–pathogen physiological interactions involved in bacterial pathogenesis, including cell invasion, evasion of the host immune response through escape of or maturation avoidance within phagosomes, establishment of tissue colonization, and systemic spread. The contribution of the Cronobacter version of PLD in the pathogenicity of this organism needs to be further studied.