Screening of Antibiotic and Virulence Genes from Whole Genome Sequenced Cronobacter sakazakii Isolated from Food and Milk-Producing Environments

The objective of this study was to use whole-genome sequencing (WGS) to screen for genes encoding for antibiotic resistance, fitness and virulence in Cronobacter sakazakii strains that had been isolated from food and powdered-milk-producing environments. Virulence (VGs) and antibiotic-resistance genes (ARGs) were detected with the Comprehensive Antibiotic Resistance Database (CARD) platform, ResFinder and PlasmidFinder tools. Susceptibility testing was performed using disk diffusion. Fifteen presumptive strains of Cronobacter spp. were identified by MALDI-TOF MS and ribosomal-MLST. Nine C. sakazakii strains were found in the meningitic pathovar ST4: two were ST83 and one was ST1. The C. sakazakii ST4 strains were further distinguished using core genome MLST based on 3678 loci. Almost all (93%) strains were resistant to cephalotin and 33% were resistant to ampicillin. In addition, 20 ARGs, mainly involved in regulatory and efflux antibiotics, were detected. Ninety-nine VGs were detected that encoded for OmpA, siderophores and genes involved in metabolism and stress. The IncFIB (pCTU3) plasmid was detected, and the prevalent mobile genetic elements (MGEs) were ISEsa1, ISEc52 and ISEhe3. The C. sakazakii isolates analyzed in this study harbored ARGs and VGs, which could have contributed to their persistence in powdered-milk-producing environments, and increase the risk of infection in susceptible population groups.


Introduction
Cronobacter is a genus of enteropathogenic bacteria consisting of seven species: Cronobacter sakazakii, C. malonaticus, C. universalis, C. turicensis, C. muytjensii, C. dublinensis and C. condimenti [1][2][3]. C. sakazakii and C. malonaticus are the species with the greatest clinical significance as they have been detected in both individual cases as well as in outbreaks [4].
C. sakazakii mainly affects immunocompromised groups ranging from premature newborns infants to older adults [5][6][7]. In infants, the serious symptoms include life-threatening meningitis, septicemia and necrotizing enterocolitis (EN), while in adults, urinary tract infections are more common. The mortality rate associated with the infection of this pathogen is between 15 and 80%, whereas, for cases with neonatal meningitis and septicemia, rates of 15  Most C. sakazakii were of the neonatal meningitic pathovar sequence type ST4 (60%, 9/15) [5]. The remainder were C. sakazakii ST83 (13.3%, 2/15); and single strains of ST1, ST93, ST822, and ST260. The prevalent C. sakazakii serotype was O:2 and only the ST1 isolate was serotype O:1. There was no correlation between source of isolation, or serotype, or STs for the C. sakazakii ST4 strains ( Table 2).
Analysis of the 14 C. sakazakii isolates using cgMLST comprising 3678 core genes supported the conventional 7-loci MLST clustering of strains, but also showed that the ST4 strains could be further distinguished ( Figure 1).

Figure 1.
Minimum spanning tree (MST) of fourteen C. sakazakii strains. The isolates are represented as colored circles according to their sequence type (ST) as defined using the 7-loci MLST scheme (STs). Black numbers on the connection lines indicate the number of allelic differences between isolates from the cgMLST scheme comprising 3678 target genes for C. sakazakii. Isolates falling under the cluster threshold of 10 alleles are marked in grey as clusters.

Antibiotic Genes Detected in C. sakazakii Genomes
Twenty genes associated with antibiotic resistance were identified in the 14 C. sakazakii genomes. Some of these genes were classified as global regulators, such as HNS and CRP, which were associated with resistance-nodulation-cell division (RND) antibiotic efflux pump AMR families and rsmA a global translational regulatory protein RsmA. Other genes were related to antibiotic efflux, reduced antibiotic permeability, target modification or antibiotic inactivation (Supplementary Table S1). C. sakazakii strains carried the same genes ( Figure 2) that were involved in general bacterial porin with reduced permeability to betalactams, small multidrug resistance (SMR) antibiotic efflux pump, ATP-binding cassette (ABC) antibiotic efflux pump, resistance-nodulation-cell division (RND), glycopeptide resistance gene cluster, Van ligase, antibiotic-resistant GlpT, and Penicillin-binding protein mutations conferring resistance to beta-lactam antibiotics related to phenotypic profiles (Supplementary Table S1). Three genes (EF-tu, CSA-1 and qacJ) were identified in seven, four and two genomes, respectively ( Figure 2). The EF-tu gene with a SNP R234F was detected in 46.7% (7/15) of the genomes. This gene belongs to the AMR gene family elfamycin and their mechanism of resistance is antibiotic target alteration. The gene CSA-1 was found in 26.7% (4/15) of the genomes. This gene belongs to AMR gene family CSA beta-lactamases, conferring resistance to cephalosporin by antibiotic inactivation. Lastly, the qacJ gene was identified in 13.3% (2/15) of the genomes of the associated gene family small multidrug-resistance (SMR) antibiotic efflux pump disinfecting agents and antiseptics (Supplementary Table  S1).
Although most of the genes were identified in all the strains, three genes were absent in the analyzed genomes ( Figure 2). The gene kpnF was not detected in 60% (9/15) of the genomes. This gene belongs to a major facilitator superfamily (MFS) antibiotic efflux pump. The gene CSA-2 was not found in 40% (6/15) of analyzed genomes. This gene encodes for a CSA beta-lactamase inactivating cephalosporins. The gene qacG was not identified in 13.3% (2/15) of genomes. This gene encodes for a small multidrug-resistance (SMR) antibiotic efflux pump associated with resistance to disinfecting agents and anti- Three genes (EF-tu, CSA-1 and qacJ) were identified in seven, four and two genomes, respectively ( Figure 2). The EF-tu gene with a SNP R234F was detected in 46.7% (7/15) of the genomes. This gene belongs to the AMR gene family elfamycin and their mechanism of resistance is antibiotic target alteration. The gene CSA-1 was found in 26.7% (4/15) of the genomes. This gene belongs to AMR gene family CSA beta-lactamases, conferring resistance to cephalosporin by antibiotic inactivation. Lastly, the qacJ gene was identified in 13.3% (2/15) of the genomes of the associated gene family small multidrug-resistance (SMR) antibiotic efflux pump disinfecting agents and antiseptics (Supplementary Table S1).
Although most of the genes were identified in all the strains, three genes were absent in the analyzed genomes ( Figure 2). The gene kpnF was not detected in 60% (9/15) of the genomes. This gene belongs to a major facilitator superfamily (MFS) antibiotic efflux pump. The gene CSA-2 was not found in 40% (6/15) of analyzed genomes. This gene encodes for a CSA beta-lactamase inactivating cephalosporins. The gene qacG was not identified in 13.3% (2/15) of genomes. This gene encodes for a small multidrug-resistance (SMR) antibiotic efflux pump associated with resistance to disinfecting agents and antiseptics; it uses the antibiotic efflux as a mechanism of resistance (Supplementary Table S1).
The genes kpnF, CSA-1, and qacJ were not identified in the C. dublinensis genome (510180-19); however, these genes were detected in most C. sakazakii genomes.

Discussion
C. sakazakii ST1, ST4 and, to a lesser extent, ST83, are the pathovars that have been most frequently found in PIF marketed in different countries, in PIF production plants and in invasive clinical cases such as fatal meningitis and septicemia [5,[25][26][27][28][29]. The other strains, ST93, ST822 and ST260, have not been associated with cases of disease and are of less clinical importance. In this study, we identified nine C. sakazakii isolates of ST4 (CC4) serotype Csak O:2. These were mostly from environmental sources (5/9) ( Table 2), whereas C. sakazakii 1 ST1 (CC1) serotype Csak O:1 was isolated from food. A previous study carried out on Cronobacter in the Americas showed that most of the isolates came from clinical, environmental and infant formula samples from North America (57.4%, n = 465) and Brazil (42.6%, n = 465). In addition, the study reported a total of 75 sequence types, with the most frequent being C. sakazakii ST4 (CC4) and ST1 (CC1) [30].
Analysis using a 3678 loci cgMLST scheme revealed a group of eight indistinguisable C. sakazakii ST4 isolates from food and the production environment. A previously reported multicenter study assessing the incidence of C. sakazakii throughout Europe found that 76.6% (59/77) of the human clinical strains of Cronobacter spp. corresponded to C. sakazakii. Twelve C. sakazakii isolates were ST4, and eight were ST1 [31]. There was only one allele difference between isolates associated with epidemiological outbreaks in Austria in 2016 and 2009, respectively.
C. dublinensis has primarily been reported in environmental samples and food ingredients. This is primarily considered an environmental commensal and has rarely been reported in clinical samples [4,32].
In the present study, the C. sakazakii strains were resistant to cephalothin, ampicillin tetracycline, ceftazidime and amoxicillin-clavulanic acid. In addition, two strains showed an MDR profile (ampicillin, cephalothin and tetracycline); one was ST 260 (510176- 19), and the other was ST4 (510177-19). These two strains were isolated from different foods. Compared to other studies, the resistance observed in this study is low. For example, in Teheran city (Iran), of the 25 C. sakazakii strains recovered from PIF, 96% were MDR; these were mainly resistant to amoxicillin-clavulanic acid, amoxicillin, ampicillin, cefoxitin, cefepime, erythromycin, and ceftriaxone, and totally susceptible to trimethoprim/sulfamethoxazole and levofloxacin antibiotics [33]. In China, C. sakazakii from PIF and processing environments showed isolates resistant to amoxicillin-clavulanic acid, ampicillin, and cefazolin [34].
MDR isolates have been reported from domestic kitchens in middle Tennessee, (USA); the highest resistance was to penicillin (76.1%), tetracycline (66.6%), ciprofloxacin (57.1%), and nalidixic acid (47.6%) [35]. C. sakazakii antibiotic resistance is diverse and depends on the source and geographic location of the strain [33][34][35]. The genomic analysis has enabled the detection of genes involved in antibiotic resistance. The genes are related to cell permeability to beta-lactams and antibiotic efflux pumps. In other studies, blaCTX genes are commonly detected in strains resistant to cephalosporins, whereas, in this study, these genes were not detected [36,37]. Another detected gene was EF-Tu, which is included in the AMR gene family elfamycin and their mechanism of resistance is antibiotic target alteration. This gene has evolved to be a multifunctional protein in a wide variety of pathogenic bacteria [38]. Functions related to EF-Tu vary among microbial species; however, there is a common role of adherence and immune regulation [38]. The AmpC beta-lactamases designated CSA-1 and CSA-2 (from C. sakazakii) were found in the 14 C. sakazakii strains and have been reported previously [39].
The genomic analysis of the Cronobacter strains revealed a number of virulence-related genes. Several putative genes of the Type VI Secretion System cluster were found. The Type VI Secretion System (T6SS) is a protein secretion machinery that transports protein toxins into eukaryotic cells [40,41] and bacteriolytic effectors to target cells [42]. Some genes involved in the iron uptake category (fur, iroN and PSEEN_RS11615), responsible for the production of metal binding proteins, were identified in genomic sequences, and are associated with bacterial virulence [43]. The flagella operon was found in all genomes. Flagella has been involved in adherence and proinflammatory response in different pathogens [16]. The most frequently occuring gene, the outer membrane protein A (OmpA), was identified in the genomes. OmpA plays a critical role in host cell invasion [44].
In our research, all 14 C. sakazakii strains encoded the cpa gene. The cpa gene is possibly involved in serum resistance, as well as in the systemic spread of C. sakazakii. The isolates harbored the nanAKT cassette coding for the use of exogenous sialic acid. Only the C. sakazakii species and a few C. turicensis strains can catabolize this [45]. Sialic acid is found in gangliosides of the brain, and occurs naturally in breast milk [46]. Powdered infant formula is supplemented with this monosaccharide due to its association with brain development [4]. Sialic acid also regulates the expression of enzymes, such as sialidase and adhesins, or inhibits the transcription factors of the fimB gene involved in the adhesion and invasion of epithelial cells [47].
Another important observation was the presence of rpoS, ibpA, ibpB and clpk genes in the C. sakazakii genomes. These genes are involved in protection against environmental stress [48]. For example, the RpoS regulon, which acts as a transcriptional factor in response to general stress, develops cross-protection against other environmental disturbances, such as the response to oxidative stress and the response to heat stress. In E. coli, for example, it represents 10% of genes [49]. On the other hand, the genes encoding for heat shock proteins Hsp15, Hsp20 and HspQ, were found in thermotolerant C. sakazakii [50,51]. Therefore, when these environmental C. sakazakii are subjected to desiccation or drying processes, as in the preparation of PIF or other foods, they may become more resistant to these stresses, enabling them to persist either in the environments of these production plants or in the final products.
Sixty percent of the C. sakazakii strains in our study carried the IncFIB(pCTU3) plasmid, which has been previously reported in environmental sources and food [52,53], as shown in Table 3. The majority of strains carrying the plasmid were C. sakazakii ST4. This plasmid is associated with encoding for efflux pumps associated with heavy metals, which could allow the microorganism to adapt to the changing environmental conditions of osmotic stress [54].
In this study, C. sakazakii ST4 and ST1 were isolated from environmental strains ( Table 2), and harbored genes that encode for antibiotic resistance proteins and virulence factors (Figures 2 and 3). Consequently, there is the potential risk that these isolates could persist in the environment and become a source of food contamination, therefore presenting a risk to the health of consumers.

Sampling
A total of 15 presumptive strains of Cronobacter spp. were isolated from 2000 samples from food (n = 997), milk-producing environments (n = 855), and feces from live poultry (n = 148) on the territory of the Czech Republic in 2015. Sampling was performed by instructed veterinarians. Solid, loose and slurry materials were collected in 60-200 mL sterile plastic containers (Dispolab Ltd., Brno, Czech Republic) or in original packaging. Swabs were taken using Transbak system containing Amies broth with active carbon (Dispolab Ltd., Brno, Czech Republic). Environmental swabs, feces and meat samples were stored and transported to the laboratory at +4 • C. Other materials were transported at +21 • C. These strains were analyzed at the bacteriology department, Palacký University Olomouc, Czech Republic (Supplementary Table S3).

Isolation and Identification Methods of Cronobacter spp.
Cronobacter spp. strains were isolated according to the method described by Iversen et al. [55]. Food and environment samples were pre-enriched in buffered peptone water (BPW), followed by Enterobacteriaceae enrichment broth (BD Difco, Sparks, MD, USA), then on Brilliance CM 1035 chromogenic agar (Oxoid Thermo-Fisher, UK) and purified on trypticase soy agar (BD Difco, Sparks, MD, USA). Prior to sequencing, the strains were presumptively identified using Matrix Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS) (Bruker, Billerica, MA, USA) and with the MALDI Biotyper Compass IVD 4.1.60 software (Bruker, Billerica, MA, USA) described by Lepuschitz et al. [56]. The identification of the Cronobacter spp. strains was confirmed with the Ribosomal Multilocus Sequence Typing (rMLST) software available at https: //pubmlst.org/species-id (accessed on 2 February 2023) [57].

Whole-Genome Sequencing (WGS)
Cronobacter spp. isolates were cultured in Columbia blood agar plates (bioMérieux, Marcy-l'Étoile, France) at 37 • C for 24 h. For WGS, DNA was isolated from bacterial cultures with the MagAttract HMW DNA Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The quantification of input DNA was performed with a Qubit 2.0 fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) and the double-stranded DNA (dsDNA) BR assay kit (Thermo Fisher Scientific). Nextera XT chemistry (Illumina Inc., San Diego, CA, USA) was used to prepare sequencing libraries for a 2 × 300 bp paired-end sequencing run on an Illumina MiSeq sequencer. Samples were sequenced to achieve a minimum of 80-fold coverage using standard Illumina protocols. The resulting FASTQ files were quality-trimmed and de novo assembled with the SPAdes v3.11.1 software. Contigs were filtered for a minimum of 5-fold coverage and 200 bp minimum length with the Ridom SeqSphere+ software v8.3 (Ridom, Münster, Germany) [31]. Raw reads were quality-controlled using FastQC v0.11.9. Trimmomatic v0.36 [58] was used to remove adapter sequences and to trim the last 10 bp of each sequence and sequences with quality scores < 20. Reads were assembled using SPAdes v3.11.1 [59]. Contigs were filtered for a minimum coverage of 5× and a minimum length of 200 bp using SeqSphere+ software v8.3.0 (Ridom GmbH, Würzburg, Germany) [58] (Supplementary Table S3).

Sequence Type (ST) and Core Genome Multilocus Sequence Typing (cgMLST) of Cronobacter sakazakii
A total of 3678 targets were used for core genome multilocus sequence typing (cgMLST) scheme of Cronobacter spp., with strain ATCC BAA-894 as a reference using a target gene loci task template of the Ridom SeqSphere+ software v8.3.0 (Ridom, Münster, Germany) [31,60]. According to this cgMLST scheme, isolates were visualized with a minimum spanning tree (MST) to establish their genotypic relationships and define those isolates with maximum differences of 10 alleles as clusters [31]. In addition, the sequences of the seven housekeeping genes of the conventional MLST for C. sakazakii (atpD, fusA, glnS, gltB, gyrB, infB and ppsA) were extracted and cross-checked against the Cronobacter MLST database https://pubmlst.org/organisms/cronobacter-spp/ (accessed on 2 February 2023) [61].

Determination of Serotypes
The profiles of gnd and galF genes specific to the Cronobacter spp. serotype O region were determined by WGS sequence analysis with the BIGSdb tool available in the PubMLST database (http://pubmlst.org/cronobacter/) (accessed on 2 February 2023) [18].

Detection of Antibiotic Resistance and Virulence Genes
The existence of virulence genes was confirmed by ResFinder tool from the Center of Genomic Epidemiology (CGE) (http://www.genomicepidemiology.org) (accessed on 5 March 2023) [63]. Thresholds for the target scanning procedure were set with a required identity of ≥90% to the reference sequence and an aligned reference sequence ≥99%. The Comprehensive Antibiotic Resistance Database (CARD) with the "perfect" and "strict" default settings for sequence analysis [64], the Task Template AMRFinderPlus 3.2.3 available in Ridom SeqSphere+ v7.8.0 software using the EXACT method at 100%, and BLAST alignment for protein identification available in the AMRFinderPlus database were used for antimicrobial resistance genes. For the search for genes and proteins associated with virulence and fitness, the Virulence Factor Database (VFDB), available at http://www. mgc.ac.cn/VFs/main.htm (accessed on 5 March 2023), was downloaded and compared with each of the 15 genomes using the blast tool ncbi-blast-2.13.0+-x64-arm-linux.tar.gz, available at https://ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LAST/ (accessed on 5 March 2023) on command line.

Conclusions
The C. sakazakii isolates analyzed in this study harbored ARGs and VGs, which contributed to their persistence in powdered-milk-producing environments, and increased the risk of infection in susceptible population groups due to the presence of multi-resistant strains to antibiotics, reducing the possibility of treating of the infection [67,68].