Cronobacter spp. (formerly Enterobacter sakazakii) is an emerging opportunistic bacterial pathogen comprised of seven species: Cronobacter sakazakii, Cronobacter malonaticus, Cronobacter turicensis, Cronobacter muytjensii, Cronobacter condimenti, Cronobacter universalis, and Cronobacter dublinensis (Joseph et al., 2012b; Lu et al., 2014). This organism can cause necrotizing enterocolitis, bacteremia, and meningitis in neonates and infants, with a 40–80% mortality rate (Forsythe et al., 2014; Holy and Forsythe, 2014; Li et al., 2015). To date, only the three species C. sakazakii, C. malonaticus, and C. turicensis have been isolated from neonatal infections (Sonbol et al., 2013). Cronobacter spp. has been recovered from a wide variety of foods, with powdered infant formula (PIF) being of particular concern as it is the most significant source of Cronobacter strains resulting in neonatal infections (Iversen and Forsythe, 2004; Craven et al., 2010). Cronobacter strains cannot survive after the standard pasteurization procedures, therefore, the addition of non-heat treated materials and environmental contamination during filling and packaging are the plausible causes of Cronobacter contamination of PIF (Nazarowec-White and Farber, 1997). Besides, the strong resistant ability of the organism in dry PIF factory environments can increase the likelihood of post-pasteurization contamination (Riedel and Lehner, 2007).

Currently, antibiotic therapy is considered to be the common and preferred way to prevent the Cronobacter infection in humans (Depardieu et al., 2007). Many studies have confirmed that Cronobacter strains can be effectively eliminated by antibiotics, however, prolonged use of antibiotics is undesirable as it may result in the development of Cronobacter antibiotic resistance (Yoneyama and Katsumata, 2006; McMahon et al., 2007). Hochel et al. (2012) reported that all of 53 Cronobacter strains isolated from 399 retail food samples were resistant to erythromycin, and two of them were resistant to both erythromycin and tetracycline. Chon et al. (2012) found that 77.8% Cronobacter strains isolated from desiccated foods in Korea were not susceptible to cephalothin. The Cronobacter strains resistance to cefotaxime and streptomycin have also been isolated from various foods in Korea and commercial PIF in China (Lee et al., 2012). Caubilla-Barron and co-workers reported that the C. sakazakii strains isolated from two fatal neonatal infections expressed β–lactamase activity (Caubilla-Barron et al., 2007). In addition, the effective dose of antibiotics for treatment may change after long-term antibiotic use, therefore screening environmental and PIF isolates of Cronobacter for antibiotic resistance would be a useful comparison with clinical isolates.

The greater resistance of Cronobacter spp. to desiccation compared with other Enterobacteriaceae will enable their long term survival under a low water activity (a_w) condition such as in PIF (a_w: 0.2–0.5) which has a long shelf life of up to 1 year (Gurtler and Beuchat, 2007). In general, the viability of Cronobacter spp. in PIF decreases by 0.5–0.6 log cycles per month (Edelson-Mammel et al., 2005; Gurtler and Beuchat, 2007). Caubilla-Barron et al. (2007) found there was considerable reduction in the viable count of Cronobacter strains (about 3.34 log₁₀ cycles) in the first 6 months, whereas after the following 24 months, the average reduction in viability decreased by 1.88 log₁₀ cycles. It is important to note that some capsulated C. sakazakii strains were still recovered after 2.5 years, by contrast, both Salmonella enteritidis, and Escherichia coli were undetectable after 15 months (Barron and Forsythe, 2007). Some difference in desiccation resistance can be found among the different Cronobacter strains, for example, C. sakazakii NCTC11467^T has atypical growth characteristics compared with most of Cronobacter strains, and cannot be recovered from desiccated condition after 1 year (Iversen et al., 2004). Cronobacter spp. produce highly mucoid colonies on milk agar plates and the capsular material could be linked to virulence traits such as macrophage survival as well as desiccation resistance (Ogrodzki and Forsythe, 2015). Particular capsule profiles correlate with the meningitic C. sakazakii pathovar clonal complex (CC) 4 (Ogrodzki and Forsythe, 2015).

Cronobacter is a diverse genus and has been extensively studied using multilocus sequence typing for over 1,000 strains (Forsythe et al., 2014). This has revealed the high clonality within the genus as well as identified particular pathovars. C. sakazakii sequence type (ST) four identified using multi-locus sequence typing (MLST) was the dominant ST, and was associated with neonatal meningitis (Joseph and Forsythe, 2011). In addition, Cronobacter pathovars, associated with clonal complex are now recognized. Of particular relevance is the C. sakazakii CC4 pathovar which is associated with neonatal meningitis (Hariri et al., 2013; Forsythe et al., 2014). C. sakazakii ST8 and C. sakazakii ST1 were mainly isolated from clinical sources and PIF, respectively (Forsythe et al. 2014). C. sakazakii CC4 was also the genotype of ~25% of isolates recovered from the processing environment of PIF manufacturing plants (Sonbol et al., 2013). Why this CC predominates in PIF and neonatal infections is uncertain since no particular virulence genes have been detected (Joseph et al., 2012a). Therefore, a better understanding of the environmental fitness and antibiotic resistance of Cronobacter isolates with significant sequence type from PIF production areas is warranted.

In previous studies, our group isolated 66 Cronobacter sakazakii and four C. malonaticus strains from PIF and processing environments from 2009 to 2012, and identified their sequence type by MLST (Fei et al., 2015). In this study, we further characterized the strains according to their antibiograms and desiccation tolerance, and considered whether the differences corresponded with their sequence type.

C. sakazakii and C. malonaticus are the dominant species of Cronobacter spp. isolated from PIF and processing environment, and can infect infants and adults, respectively (Fei et al., 2015). Given the long term shelf-life of PIF (up to 1 year) and repeated isolation from PIF production plants (Craven et al., 2010; Sonbol et al., 2013), environmental fitness of Cronobacter is very important trait to understand as it may lead to increased persistence and neonatal exposure.

The phylogenetic relationship of concatenated sequences (3,036 bp) based on conventional MLST (7 loci), ribosomal-MLST (53 loci), and core genome MLST (1865 loci) reflects the whole genome phylogeny of the Cronobacter genus (Joseph et al., 2012b; Forsythe et al., 2014). The application of MLST has also led to the recognition of stable clonal complexes and sequence types associated with certain clinical presentations. Subsequently the diversity and clonal stability within the Cronobacter genus needs to be taken into consideration when interpreting laboratory studies. Therefore, Cronobacter strains with different sequence types were chosen for detailed study here.

In this study, 70 Cronobacter strains, previously isolated from PIF and processing environments, were used. These represented 19 sequence types of C. sakazakii and C. malonaticus, detailed information can be obtained from Cronobacter MLST databases (http://pubmlst.org/cronobacter/) (Fei et al., 2015). The main STs from PIF and processing environments, were the neonatal pathovars C. sakazakii ST4, ST1, and ST8, as well as ST64, ST12, ST21, ST258, and ST201.

The isolates were susceptible to most antibiotics, except for amoxicillin-clavulanate, and ampicillin. Previous reports revealed that the occurrence of susceptibility to cefotaxime, chloramphenicol, ciprofloxacin, gentamicin, and tetracycline in Cronobacter strains isolated from different sources (Chon et al., 2012; Xu et al., 2014). Chon et al. (2012) reported that only 5.6% strains from desiccated foods in Korea were resistance to ampicillin whereas in our study, all isolates were resistant to this antibiotic. Furthermore, we found that the MIC values of cefazolin varied across different STs, but were similar within each ST. This may reflect a connection between the ST and antimicrobial resistance.

The ability of eight Cronobacter isolates with different STs to resist the osmotic stress was evaluated in this study. Compared with C. malonaticus isolates, C. sakazakii isolates had a higher capacity to grow in the hyperosmotic media, in agreement with previous studies (Caubilla-Barron et al., 2007; Avelino Alvarez-Ordóñez et al., 2014). C. sakazakii CE13 (ST8) was the most resistant to osmotic stress in the eight strains, interestingly, C. sakazakii ATCC 29544^T reported to have atypical growth characteristics is also ST8 (Osaili and Forsythe, 2009). C. sakazakii CE1 (ST4) also showed a more osmotic stress resistance, which might be one of the major reasons why C. sakazakii CC4 is frequently isolated from infant food, ready-to-eat foods, potable water. This could result in higher infant exposure and therefore risk of infection (Forsythe et al., 2014).

Cronobacter spp. tend to persist more in PIF with a water activity (a_w) 0.25–0.30, compared with a_w 0.43–0.50, and the survival rate was not associated with whether the PIF was milk-based or soybean-based (Osaili and Forsythe, 2009). In general, this organism can persist in PIF for more than 1 year, therefore, monitoring of survival under desiccated conditions is warranted for the two main Cronobacter species, C. sakazakii and C. malonaticus. Breeuwer et al. (2003) reported that the reduction of Cronobacter strains (species undetermined) which had been dried in air and incubated for 46 days at 25°C ranged from 1.0 to 1.5 log cycles. Using similar desiccation conditions, the viable counts of Cronobacter strains maintained at 25°C for 45 days decreased by 0.86–1.76 log cycles in our study. Barron and Forsythe (2007) reported that the number of survivors of C. sakazakii, C. muytjensis, and C. turicensis under desiccated conditions in the first 30 days reduced by 0.58 log cycles on average, and decreased 3.34 log cycles during the first 6 months. While, under the similar desiccation conditions, the results in current study indicated that the reduction of Cronobacter strains declined by an average of 0.86 log cycles during the first month, and 2.25 log cycles during the first 6 months. The differences in reduction of Cronobacter strains between different reports indicated that a continuous assessment of resistance to dry stress was very necessary.

Interestingly, after kept in the desiccation conditions for 1 year, the total decrease of CE13 (ST8) was the least, followed by CE38 (ST12) and CE1 (ST4). This trend was similar to osmotic stress resistance. Mechanisms of resistance to these environmental stressor have previously been reviewed by Osaili and Forsythe (2009). The higher resistance to both osmotic and desiccation stresses by C. sakazakii compared with C. malonaticus may in part account for the predominance of C. sakazakii in PIF and processing environments and subsequent greater infant exposure.

In summary, this study contributes to an improved understanding of the environmental persistence of C. sakazakii and C. malonaticus and subsequent risk of infant exposure through contaminated PIF. In addition, the relatively low antibiotic resistance is reassuring for the current treatment of Cronobacter infections.

Conceived and designed the experiments: CM and PF. Performed the experiments: PF, YJ, CM, JF, and YZ. Generated and analyzed the data: PF and RL. Wrote the paper: PF, CM, and SF.