Comparative Genomic Characterization of the Highly Persistent and Potentially Virulent Cronobacter sakazakii ST83, CC65 Strain H322 and Other ST83 Strains

Abstract

Cronobacter (C.) sakazakii is an opportunistic pathogen and has been associated with serious infections with high mortality rates predominantly in pre-term, low-birth weight and/or immune compromised neonates and infants. Infections have been epidemiologically linked to consumption of intrinsically and extrinsically contaminated lots of reconstituted powdered infant formula (PIF), thus contamination of such products is a challenging task for the PIF producing industry. We present the draft genome of C. sakazakii H322, a highly persistent sequence type (ST) 83, clonal complex (CC) 65, serotype O:7 strain obtained from a batch of non-released contaminated PIF product. The presence of this strain in the production environment was traced back more than 4 years. Whole genome sequencing (WGS) of this strain together with four more ST83 strains (PIF production environment-associated) confirmed a high degree of sequence homology among four of the five strains. Phylogenetic analysis using microarray (MA) and WGS data showed that the ST83 strains were highly phylogenetically related and MA showed that between 5 and 38 genes differed from one another in these strains. All strains possessed the pESA3-like virulence plasmid and one strain possessed a pESA2-like plasmid. In addition, a pCS1-like plasmid was also found. In order to assess the potential in vivo pathogenicity of the ST83 strains, each strain was subjected to infection studies using the recently developed zebrafish embryo model. Our results showed a high (90-100%) zebrafish mortality rate for all of these strains, suggesting a high risk for infections and illness in neonates potentially exposed to PIF contaminated with ST83 C. sakazakii strains. In summary, virulent ST83, CC65, serotype CsakO:7 strains, though rarely found intrinsically in PIF, can persist within a PIF manufacturing facility for years and potentially pose significant quality assurance challenges to the PIF manufacturing industry.

Keywords: Cronobacter; environmental microarray; persister; whole genome sequencing.

FIGURE 1

Phylogenetic Neighbor Net (SplitsTree4) analysis of 114 C. sakazakii and phylogenetically related strains, which were generated from the gene-difference matrix (Supplementary Tables S1, S2) and then overlaid with sequence types (ST). The microarray experimental protocol as described by Jackson et al. (2011) and as modified by Tall et al. (2015) were used for the interrogation of the strains for the analysis. The phylogenetic tree illustrates that the Cronobacter microarray could clearly separate the strains according to species and respective ST. The ST83 strains (^∗) clustered together along with a single ST83 strain which was obtained from an United States diary powder manufacturing facility. Bar marker represents 0.01 gene differences.

FIGURE 2

Comparative microarray probe set summarization of two Swiss ST83 strains H2399 and H1191 compared to strain H322 using scatter plots showing RMA summarized probe set intensities from each strain’s data points which are color-coded based on their intensities in H322. The y-axis show the RMA summarized probe set intensities from H322 (blue) and the x-axis shows the probe set intensities from H2399 and H1191 (red), respectively.

FIGURE 3

Phylogenetic analysis of C. sakazakii strains of different ST: Allelic variations in conserved genes spanning the whole genome were used for this analysis. Alleles present in homologs of 1000 C. sakazakii strain BAA-894 chromosomal loci (reference gene set for this analysis) in 44 genomes were detected by BLAST-based workflow (Tall et al., 2015). SNPs in 37,711 positions in these genomes were concatenated and the cladogram was built using Neighbor-Joining algorithm available on MEGA7 phylogenetic suite (Kumar et al., 2016). The bootstrap consensus (values located on the nodes) in the tree are inferred from 500 replicates. 44 genomes separated into eight STs and multiple clusters reflecting microarray-based clusters. Five ST83 isolates group together with A31 showing some differences in the cluster confirming other results.

FIGURE 4

Comparative genomic analysis to identify pCS1-like sequences: Mauve global alignment of the pCS1-like sequences in H322: (track 4) with C. sakazakii NCTC 8155 pCS1_NCTC 8155 (track 1), C. sakazakii pNCIM8272 (track2) and C. sakazakii 8399 pCsak8399 (track 3). H322 contigs mapping to pCS1 plasmid (CP012254) were used to create an artificial molecule to represent pH322 (track 4; pH322_pseudomolecule). This artificial scaffold was compared with pCS1 and pCS1-like plasmid contigs from strains NCIMB8272 (track2) and Csak8399 (track3). In addition to pESA3-like plasmid, isolates such as strain H322 reported in this work contain a large plasmid highly comparable to pCS1 plasmid.

FIGURE 5

Results of the zebrafish embryo infection experiments with five C. sakazakii ST83 strains, the two clinical strains ATCC29544^T and NM 1242 as well as E. coli Xl1 blue.