Characterization of Cronobacter recovered from dried milk and related products

Background Cronobacter is a recently proposed genus consisting of six genomospecies that encompass the organisms previously identified as Enterobacter sakazakii. Cronobacter are opportunistic pathogens and are known to cause serious infections in infants, particularly neonates. High case fatality rates have been associated with infections and acute sequelae can occur in survivors with severe ramifications on neurological development. Infant formula has been identified as one route of transmission for infection in infants. However, the primary reservoirs for subsequent contamination of foods with Cronobacter remain undefined due to the ubiquitous nature of these organisms. More recently, infections in adults have been reported, especially amongst the elderly and patients who are immunocompromised. To help prevent the transmission of infection, it is important to identify the main food sources for Cronobacter. The aim of this study was to identify and characterize Cronobacter isolated from dried-milk and related products available in an Egyptian food market. Results In total sixteen Cronobacter strains were isolated from 152 dairy-based products. These were identified and characterized using pheno- and genotyping experiments. Real-time PCR confirmed the detection of Cronobacter. Following antibiotic susceptibility tests, 3 strains showed resistance to trimethoprim and/or neomycin. Phenotype profiles were generated based on key biochemical distinguishing tests. Pulsed-field gel electrophoresis (PFGE) identified 8 PFGE types amongst the collection of strains. Repetitive sequence based PCR (rep-PCR) analysis identified 3 rep-PCR types amongst the collection of strains. Sequencing of the recN gene was used to differentiate among the recently described species of Cronobacter. Conclusion This study identified the presence of Cronobacter in dried milk and related products sourced from the Nile-Delta region of Egypt. Although the majority of the strains were susceptible to the antibiotics tested, resistance was observed in three isolates, highlighting the risks associated with Cronobacter contamination in foods. Phenotype and genotype analysis should be applied to further characterize Cronobacter spp. and prevent its transmission into food products.


Background
Recent taxonomic work by Iversen et al. [1,2] has led to an alternative classification of the organism, Enterobacter sakazakii, and the proposal of a newly defined genus, Cronobacter. Cronobacter spp. are considered emerging opportunistic pathogens and are associated with outbreaks of infections amongst infants, in particular neonates [3][4][5]. Symptoms include bacteremia, necrotizing enterocolitis and meningitis, with case fatality rates as high as 80% being reported. The prognosis for survivors is also poor, with neurological development being severely affected in many cases [6]. More recently the association of Cronobacter with infections in adults has been investigated. Gosney et al. [7] described the isolation of Cronobacter from seven adult stroke patients. See et al. [8] reported a case of bacteremia in a 75 year old woman who presented with a splenic abscess. In total, thirteen cases of Cronobacter infections in adults have been documented from 1985 to present.
The primary origins of Cronobacter spp. remain unknown. Due to its ubiquitous nature, Cronobacter can be isolated from a wide variety of foods including milk, cheese, dried foods, meats, water, vegetables, rice, bread, tea, herbs and spices [9][10][11][12][13][14]. Surveillance studies have detected Cronobacter in infant formula production, food processing, households and clinical environments. Powdered infant formula (PIF) has been epidemiologically linked to cases of infection in infants, thus research has specifically focused on the monitoring of PIF products for the presence of Cronobacter. However, less is known regarding the prevalence of Cronobacter in other dairy foods. Recently, El-Sharoud et al. [15] examined dairy products from an Egyptian market for the occurrence of the organism. Cronobacter was isolated from skimmed milk and a related imitation soft cheese.
Identifying foods that may contain Cronobacter is important to discover the possible routes for transmission of infection. With the indication that Cronobacter spp. infect both infants and vulnerable adults it is important that a wider variety of foods now be evaluated. The aim of this study was firstly, to determine if Cronobacter could be found present in dried milk and related products and secondly, to characterize any isolates recovered.

Bacterial Cultures
A summary of the isolates characterized in this study can be seen in Table 1.

Isolation & Identification
In total 152 dairy-based products obtained within the Nile-Delta region of Egypt were tested for the presence of Cronobacter. Additionally, strain CFS-FSMP 1511 from the environment of a milk powder factory was obtained from Nestlé Research Centre, Lausanne, Switzerland. Samples included full-fat milk powder (n = 15), skimmed milk powder (n = 37), dried whey (n = 5), dried ice-cream (n = 5), dried artificial cream (n = 5), Sahlab (n = 10), PIF (n = 35), stored-cheese (n = 10), and fresh Domiatti cheese (n = 10), Kariesh cheese (n = 10) and Ras cheese (n = 10) ( Table 2). Collected samples represented different, commercially available brands of each product type. Domiati cheese is a traditional Egyptian, highly salted, enzymecoagulated, soft cheese. Similarly, Ras cheese, also typically Egyptian is a hard cheese that is prepared from raw cow's milk or a mixture of cow and buffalo's milk. Kariesh cheese is a traditional Egyptian soft cheese that is produced by acid coagulation of skim milk by culturing with lactic acid bacteria. Sahlab is a dried blend consisting of dried skim milk, starch and nuts that is reconstituted in boiling water and served as a hot drink. Isolation was performed according to a modification of the International Organization for Standards Technical Specification on the detection of E. sakazakii (ISO/TS 22964). In brief, samples were diluted 1:10 (w/v) in buffered peptone water (BPW)  [16]. In short, a primer set and probe targeting the dnaG gene located internally to the macromolecular synthesis (MMS) operon was applied [17].
The Cronobacter genus currently consists of six genomospecies [18]. To this end, isolates confirmed as Cronobacter were speciated using biochemical differentiation tests as described by Iversen et al. [19] and recN gene sequence analysis (Kuhnert P., Korczak B.M., Stephan R., Joosten H., Iversen C: Phylogeny and whole genome DNA-DNA similarity of Enterobacter and related taxa by multilocus sequence analysis (MLSA)).

Molecular Subtyping
Pulsed-field gel electrophoresis (PFGE) was applied as described previously [21]. Analysis was carried out using BioNumerics software V3.0 (Applied Maths, Sint-Martens-Latem, Belgium). A dendrogram was generated using the DICE coefficient and unweighted pair group method with arithmetic mean (UPGMA). A band tolerance and optimization coefficient of 1.5% was applied.

Es-R_seq 5'-TGGGTAACGCACATCACCTGAGT-3'
A maximum parsimony phylogenetic tree (Figure 1) was generated using BioNumerics software (Applied Maths, recN gene sequencing clustering analysis of Cronobacter species (Colours relate to the phenotypes in Table 3) Figure 1 recN gene sequencing clustering analysis of Cronobacter species (Colours relate to the phenotypes in Table 3).
Belgium). Gaps were not considered an extra state. The Jukes-Cantor correction was used to compensate for divergence being a logarithmic function of time due to the increased probability of a second substitution at a nucleotide site slowing the increase in the count of differences as divergence time increases [23]. Felsenstein bootstraps (1,000 simulations) were applied to assess the level of confidence for each clade of the observed trees based on the proportion of bootstrap trees showing the same clade [24]. The topology of the maximum parsimony tree was optimized using simulated annealing. [This is a heuristic approach that occasionally accepts a worse tree during the course of the search allowing it to escape local optima. This method is more economical than the more usual heuristic searches (stepwise addition and hill-climbing), which can require many random re-starts, especially with large data matrices].

Isolation & Identification
A total of sixteen Cronobacter strains were isolated from various food products ( Table 1). Some of the non-Cronobacter strains isolated included Citrobacter freundii, Enterobacter cloacae, Proteus vulgaris and putative Vibrio cholerae.
Presumptive positive isolates produced blue-green colonies on DFI agar and were identified as Cronobacter (E. sakazakii) using ID 32E test strips. Real-time PCR analysis confirmed the detection of Cronobacter isolates. Biochemical tests were performed in order to distinguish the phenotypes of the Cronobacter isolates and contribute to the speciation of the collection of strains. The results of these tests are shown in Table 3.

Antibiotic Susceptibility
The antibiotic susceptibility testing indicated that all isolates were susceptible to streptomycin, compound sulphonamides, ampicillin, gentamicin, spectinomycin and furazolidone. Isolates CFS-FSMP 1500 and CFS-FSMP 1512 were found to be resistant to neomycin and isolate CFS-FSMP 1510 was reported as resistant to trimethoprim and neomycin. All other isolates were found susceptible to these two antimicrobial agents (Table 4).

PFGE Analysis
Macrorestriction of Cronobacter genomic DNA with XbaI yielded 10 to 17 DNA fragments ranging in size from 48.5 to 1,000 kbp. A dendrogram was compiled which illustrates the fingerprint pattern similarities between the vari- ous Cronobacter isolates ( Figure 2). In total, 8 pulse-types (denoted 1 through 8) were identified that showed indistinguishable similarity.

rep-PCR Analysis
The rep-PCR typing yielded between 18 and 25 DNA fragments that ranged in size from 150 to 3,500 bp. A dendrogram representing the genetic relatedness amongst the isolates was composed (Figure 3). Amongst the collection, 3 rep-PCR cluster groups (denoted A, B and C) were identified that exhibited identical similarity. A maximum parsimony tree was produced that displays the genetic relationship amongst the collection of strains (Figure 1). For comparison purposes a representative for each of the other Cronobacter spp., C. dublinensis (EU569474), C. genomospecies 1 (EU569479), C. muytjensii (EU569492), C. turicensis (EU569523) and two novel Enterobacter species, E. helveticus (EU569447) and E. pulveris (EU569451), which represent the closest related species of Cronobacter, were also included in the analysis.

Discussion
The focus of this study was to test a collection of dried milk and related products available in Egypt for the presence of Cronobacter. While PIF has been identified as one vehicle of transmission for infection in infants, less is known regarding other dried dairy products. More recent reports have also identified Cronobacter infections in immunocompromised adults, further highlighting the need to identify these organisms' primary origin for contamination. The food products tested included milk powders, PIF, dried whey, dried ice-cream, Sahlab and cheese and all were obtained from the Nile-Delta region of Egypt. In total, a collection of sixteen Cronobacter isolates were recovered from the foods tests and these were characterized using both pheno-and genotyping methods. The results of the biochemical assays identified the presence of 5 phenotype profiles amongst the collection of isolates (Table 3).
PFGE and rep-PCR analysis was performed for molecular characterization of the isolates. PFGE typing identified 8 pulse-type cluster groups exhibiting ≥ 95% similarity. Analysis using rep-PCR typing identified 3 cluster groups that showed ≥ 95% similarity. Interestingly, rep-PCR clustered all the C. malonaticus isolates into a single cluster, denoted as rep-PCR type A, while the C. sakazakii isolates formed two distinct clusters, rep-PCR types B and C. Isolates CFS-FSMP 1507 and 1509 produced unique phenotype profiles when compared with the other strains in the collection. PFGE analysis also grouped the latter two isolates into distinct clusters, pulse-types 6 and 5 respectively. Further work is needed to determine whether or not these strains represent unique subtypes of C. sakazakii. Sequencing of the recN gene was applied to further characterize the isolates and confirm the species identification. This method was chosen as it has shown a higher discriminatory power with regard to the speciation of Cronobacter isolates when compared to 16S rRNA sequencing (Kuhnert P., Korczak B.M., Stephan R., Joosten H., Iversen C: Phylogeny and whole genome DNA-DNA similarity of Enterobacter and related taxa by multilocus sequence analysis (MLSA)). The method identified two Cronobacter species recovered in this study, C. sakazakii and C. malonaticus. Figure 1 outlines the genetic relationship between the Cronobacter species type strains, the isolates from this study and the genetically closest Enterobacter species. As expected, the isolates recovered from the foods studied, clustered with the type strains of C. sakazakii and C. malonaticus.
Antimicrobial susceptibility testing indicated that all isolates were susceptible to ampicillin, compound sulphonamides, furazolidone, gentamicin, spectinomycin and streptomycin. These findings are in agreement with the data obtained by Stock and Wiedemann [25]. In their study they identified Cronobacter as being more susceptible to β-lactam antibiotics, including ampicillin, when compared with the Enterobacter species, E. amnigenus, E. cancerogenus and E. gergoviae. Interestingly, the Cronobacter isolates screened in their study were naturally susceptible to neomycin. The isolates CFS-FSMP 1500, 1510 and 1512 were resistant to this antibiotic. Neomycin is an aminoglycoside antibiotic, the mode of action of which is to bind to the 30S ribosomal subunit of bacteria. A possible reason behind this observed resistance could be an alteration to the binding site protein of the 30S subunit.
Such an occurrence has previously led to streptomycin resistance, another aminoglycoside compound. In the Stock and Wiedemann study [25] all Cronobacter and Enterobacter strains tested were susceptible to antifolate compounds. However, in our study isolate CFS-FSMP 1510 was resistant trimethoprim. Trimethoprim is an antifolate compound and acts by inhibiting dihydrofolate reductase enzymes in susceptible bacteria. Resistance in Gram-negative bacteria has previously been reported and it is believed that the mechanism of resistance lies within the expression of plasmid and/or transposon mediated dihydrofolate reductase genes.

Conclusion
This study identified and characterized Cronobacter isolates recovered from dried milk and related food products. Although the majority of the strains were susceptible to the panel of antibiotics tested, resistance patterns PFGE analysis showing the clustering of Cronobacter isolates recovered from dairy products  observed in three isolates may indicate increasing risks to public health associated with the presence of Cronobacter in foods. Phenotypic and genotypic analysis should be applied to further monitor and characterize the presence of Cronobacter in food production environments and prevent its transmission thereby improving food safety and quality.
Authors' contributions WME isolated the cultures and contributed to the outline of the study. SOB performed PFGE analysis of the isolates and contributed to the drafting of the manuscript. CN performed the biochemical profiling of the collection of strains and participated in drafting the manuscript. CI carried out recN gene sequence analysis and alignments and helped draft the manuscript. SF conceived of the study, and participated in its design and helped to draft the manuscript. BH coordinated the study and carried out realtime PCR detection, rep-PCR molecular subtyping of the isolates and drafted the manuscript. All authors read and approved the final manuscript.
rep-PCR analysis illustrating the relatedness of Cronobacter isolates recovered from dairy products Figure 3 rep-PCR analysis illustrating the relatedness of Cronobacter isolates recovered from dairy products.