The complete genome sequence of Cronobacter sakazakii ATCC 29544^T, a food-borne pathogen, isolated from a child’s throat

Cronobacter sakazakii ATCC 29544^T was routinely cultivated using Luria-Bertani (LB) medium at 37 °C with shaking at 220 rpm. Bacterial cells were harvested in the mid-exponential growth phase using centrifugation at 16,000×g for 1 min and its genomic DNA was isolated using G-spin™ Genomic DNA Extraction Kit for Bacteria (iNtRON Biotechnology, Seongnam, South Korea). The concentration and purity of extracted DNA were determined by NanoVue (GE healthcare, Little Chalfont, United Kingdom).

The complete genome of C. sakazakii ATCC 29544^T was sequenced at Macrogen, Seoul, South Korea, using a hybrid of PacBio RS II (Pacific Biosciences, Menlo Park, CA, USA) and Illumina HiSeq 2500 (San Diego, CA, USA). The sequence reads from PacBio RS II and Illumina HiSeq 2500 platforms were assembled using HGAP (version 2.0) and ALLPATHS-LG (version r47449), respectively. The final genome coverages were on average 1321 X Illumina and 73 X PacBio, respectively.

The ORFs were predicted using Glimmer3 [11] and GeneMark.hmm [12]. The gene prediction results were confirmed by manual curation. The genes of rRNA and tRNA were predicted using RNAmmer 1.2 [13] and tRNAscan-SE [14], respectively. The genome annotation was conducted using NCBI BLASTP [15] and a predicted protein analysis using InterProScan 5 [16] for the prediction of protein functions.

Phage-associated gene clusters in the genome sequences of C. sakazakii ATCC 29544 were searched using PHAST server [17].

Cronobacter sakazakii ATCC 29544^T was obtained from American Type Culture Collection (ATCC) and its morphological observation using a transmission electron microscopy (TEM) showed a traditional shape of C. sakazakii as a short rod (Additional file 1: Figure S1). In addition, this strain was confirmed to C. sakazakii using 16S rRNA gene sequencing. For genome sequencing, the raw read sequences were selected and assembled when their quality scores were more than 40 as cutoffs. The complete genome sequence after genome assembly was also used for confirmation using ANI analysis with previously reported complete genome sequences of C. sakazakii.

The complete genome of C. sakazakii ATCC 29544^T is composed of a circular chromosome and three plasmids (Fig. 1). The chromosome is 4,511,265 bp in DNA length with a GC content of 56.71%, 4380 ORFs, 22 rRNA genes, and 83 tRNA genes. The plasmids, designated pCSK29544_p1, pCSK29544_p2 and pCSK29544_p3, are 93,905, 4938 and 53,457 bp in DNA length with GC contents of 57.02, 54.88 and 50.07%, respectively. The plasmids have predicted ORFs of 72, 6, and 57.

Cronobacter sakazakii infects human neonates and infants via mostly contaminated reconstituted infant formula milk, causing serious human diseases, including bacteremia, necrotizing enterocolitis, and even meningitis [6]. The high survival rate of C. sakazakii, even under extremely dried conditions, as in milk formula powder, has not been investigated at the molecular level. To understand this extraordinary property of C. sakazakii, molecular studies have been recently performed. Interestingly, the gene clusters associated with biosynthesis of capsular polysaccharides (CSK29544_00281-00284) and cellulose (CSK29544_01124-01127) were detected. Recently, the biofilm formation and cellulose (as a component of biofilm) production of C. sakazakii were experimentally confirmed [18, 19], suggesting that these gene clusters may be involved in the biofilm formation. This biofilm formation of C. sakazakii may contribute to the survival in the infant formula conditions [19]. However, exopolysaccharide (EPS) production was not observed in C. sakazakii ATCC 29544^T [20], suggesting that capsular polysaccharide gene cluster may be associated with biofilm formation, not with EPS production. In particular, C. sakazakii is the only Cronobacter species that has the nanKTAR gene cluster to utilize sialic acid [21]. Interestingly, C. sakazakii ATCC 29544^T has this gene cluster (CSK29544_00587-590), indicating its ability to use it. This unique ability may be involved in its adaptation to the milk conditions because sialic acid is one of the components of milk [22].

Upon human infection, C. sakazakii invades brain microvascular endothelial cells (BMECs), barriers to protect the brain from infection of meningitic pathogens, including Escherichia coli K1 and Neisseria meningitides, which cause meningitis in neonates and infants [23, 24]. Outer membrane protein A (OmpA) of C. sakazakii is a critical determinant, contributing in vitro invasion of human BMECs by enhancing cell adhesion [24]. In addition, a few genes (ibeA, ibeB, and yijP) in meningitic E. coli were also suggested to be associated with the invasion of human BMECs [25‐27]. Interestingly, two genes, ompA (CSK29544_03699) for BMEC adhesion and the ibeB-homologous cusC (pCSK29544_3p0028) for the penetration of BMECs, were detected in the genome of C. sakazakii ATCC 29544^T−, suggesting this strain may invade human BMECs. However, ibeA and yijP were not detected in the genome. It is noteworthy that this ibeB-homologous cusC is located in the gene cluster of the complete copper- and silver-resistance cation efflux system as encoded by cusABCF (CSK29544_3p0026–CSK29544_3p0031) in the plasmid pCSK29544_3p, suggesting this BMEC invasion ability may be transferrable to other C. sakazakii or that the strain ATCC 29544^T may have been acquired from other C. sakazakii for human BMEC invasion. The presence of a conjugational transfer protein encoded by traX (CSK29544_3p0007) supports this hypothesis.

Cronobacter sakazakii ATCC 29544^T harbors three plasmids, pCSK29544_1p (1p), pCSK29544_2p (2p), and pCSK29544_3p (3p). Interestingly, two large plasmids (1p and 3p) encode virulence-associated genes, related to iron uptake and BMEC invasion. Therefore, these two plasmids may contribute to host dominance and pathogenesis, respectively.

Iron is an essential element for dominant survival and colonization via bacterial competition for iron uptake because it plays an important role in the electron transport system for energy production [28]. To accomplish this dominant survival and colonization against other bacteria, C. sakazakii has an iron acquisition system, including siderophore production (iucABCD/iutA) and an ABC-type transport system (eitCBAD) in the plasmid pESA3 [29]. The strain ATCC 29544^T also has this privileged iron acquisition system, including siderophore biosynthesis (iucABCD/iutA; CSK29544_1p0024–CSK29544_1p0028) and an ABC-type iron transport system (eitCBAD; CSK29544_1p0056–CSK29544_1p0059). However, this iron acquisition system is present in the plasmid 1p, similar to pESA3 of C. sakazakii BAA-894 [29]. Interestingly, C. sakazakii BAA-894 has multiple copies of pESA3 plasmids in a host, which is highly homologous to the plasmid 1p, suggesting the plasmid 1p may be multi copied, too [30]. These results indicate that the host strain may take advantage of better iron uptake ability in the given environments, probably due to the presence of multiple copies of this iron acquisition system encoded in the plasmid 1p.

Recent studies have revealed that milk formula contains an at least six-times-higher arsenic concentration than that of breast milk, suggesting bacterial survival in milk formula may require arsenic resistance [31]. In addition to the previously mentioned BMEC invasion ability of C. sakazakii ATCC 29544^T via the ibeB-homologous cusC gene in the cation efflux system, the plasmid 3p has an arsenic resistance system as encoded by arsRDABC, suggesting the strain ATCC 29544^T may have arsenic resistance activity for survival in formula milk powder and that this ability may have been acquired via plasmid conjugational transfer.

To date, five complete genome sequences of C. sakazakii and their annotation data are available in the GenBank database. To compare these genomes, whole genomic DNA sequence-based average nucleotide index (ANI) analysis and Roary matrix-based protein sequence analysis were performed. ANI analysis showed that C. sakazakii NCTC 8155 and SP291 are the most closely related, forming a taxonomical group, and the strains ATCC 29544^T, ATCC BAA894, and ES15 are also similar to this group (Fig. 2a). Subsequent Roary matrix-based protein sequence analysis of these five genomes supports this relationship among them (Fig. 2b). However, the strain ATCC 29544^T is more closely related to the group in the Roary matrix-based tree.

Moreover, pan genome analysis of these five genomes was conducted and their strain-specific regions were detected (Fig. 1). Interestingly, three functional gene clusters and seven prophage regions are ATCC 29544^T-specific. In addition to the iron acquisition system consisting of siderophore production and ABC-type transport system in the plasmid 1p, ATCC 29544^T-specific iron uptake-associated gene cluster containing fecABDCEI for dominant uptake of ferric dicitrate is present in the chromosome, probably related to survival in the iron-limited condition [32]. The second arsenic resistance gene cluster consisting of arsRCB for arsenate sensing, reduction, and efflux pumping was also detected, which is similar to the gene cluster in the plasmid 3p, probably for dominant survival in the human gut containing low concentration of arsenic after human infection [33] (Additional file 1: Fig. S2A). Comparative analysis of their protein sequences showed that they are >94% identical to each other, probably for synergistic arsenic resistance activity. Unlike other two gene clusters shared by the chromosome and plasmid, two copies of lactose operons are present in the chromosomes of ATCC 29544^T, NCTC 8155, and SP291, probably related to the Roary matrix-based tree (Fig. 2b). While one of these operons (CSK29544_04271–04273) are shared by all C. sakazakii genomes, the second lactose operon (CSK29544_00439–00442) is ATCC 29544^T-specific. This second lactose operon is similar to other lactose operon in Klebsiella and Enterobacter (data not shown), suggesting that this operon may be originated from other genus bacterium. The presence of multiple mobile elements containing transposases and relatively low protein sequence identity support this hypothesis (Additional file 1: Fig. S2B). Furthermore, pan genome analysis showed strain-specific region (Fig. 1). Most of genes in this region belong to prophages and they have very low homology among them, indicating that these prophages are also strain-specific. Interestingly, the strain ATCC 29544^T contains seven strain-specific prophage regions with five intact prophages and two incomplete prophages (Table 1). However, most of genes in the prophages encode hypothetical proteins.