Occurrence and Characterization of Cronobacter spp. in Dehydrated Rice Powder from Chinese Supermarket

Cronobacter spp. are emerging food-borne pathogens and have been identified as causative agents of meningitis and necrotizing enterocolitis in infants. Dehydrated rice is popular with a wide range of people and it is frequently used as a substitute for infant milk powder to baby older than four months. The occurrence of Cronobacter spp. was investigated in 1,012 samples of dehydrated rice powder collected from 14 manufacturers in China during 2010 to 2012. The isolates were identified using fusA allele sequencing and subtyped using pulsed-field gel electrophoresis. Seventy-six samples (7.5%) contained Cronobacter spp. The prevalence among manufacturers ranged from 0-28.8%. The 76 isolates included 4 species [Cronobacter sakazakii (52 isolates) Cronobacter malonaticus (14 isolates), Cronobacter dublinensis (7 isolates), and Cronobacter muytjensii (3 isolates)]. Twenty-three unique fusA alleles and sixty-six PFGE-patterns were detected. All isolated strains were observed to be sensitive or to show intermediate susceptibility to eight tested antimicrobial agents. The study revealed serious contamination of dehydrated rice powder by Cronobacter spp., with prevalence varying among manufacturers in China. Identified Cronobacter species, fusA alleles, and subtypes were diverse.

Cronobacter spp. have been isolated from food and environmental, human, and clinical sources. A wide variety of foods including infant formula, cheese, dried foods, meats, and plant materials, as well as water can be contaminated with Cronobacter [14][15][16][17][18][19][20]. Dehydrated rice powder (DRP) is a good substitute for powdered infant formula (PIF) and is popular with a broad range of the population, but rice powder products have been found to be contaminated with Cronobacter [21][22]. Infant rice cereal reconstituted with water, milk, or infant formula supports copious growth of C. sakazakii, the species most commonly associated with infant disease [13,23]. However, investigation of Cronobacter presence in DRP is limited. To determine Cronobacter spp. contamination of DRP from manufacturers in China, we examined 1,012 samples from 14 manufacturers during 2010-2012 for the prevalence of Cronobacter spp. and the characteristics of the isolates including biochemical character, fusA alleles sequences, pulsed-field gel electrophoresis (PFGE) patterns, and antibiotic susceptibility.

Ethics statement
1012 DRP samples from 14 manufacturers and was collected from 6 supermarkets. The products that were found to be contaminated of Cronobacter in this study have been notified of the results to the manufacturers/supermarkets. On the request of the DRP manufacturers, the brand names and catalog/lot number of the DRP did not publishing but appear anonymously.

Sample collection
Sealed packages of DRP (n = 1,012) from 14 food manufacturers were purchased in 2010 to 2012 from different supermarkets in China (Table 1). These products contained additional small amount of ingredients such as zinc, Chinese yam powder, carrot powder and dried fruit for increased nutrition. The samples were immediately taken to the laboratory and examined for the presence of Cronobacter spp. under aseptic conditions on the day of arrival.

Isolation and identification of Cronobacter spp.
The procedure adopted for the isolation of Cronobacter spp. was based on   [24] and the original Enterobacter sakazakii technical specifications (ISO/TS 22964, 2006). Briefly, a 100 g DRP sample was homogenized with 900 mL buffered peptone water (BPW; OXOID, Hampshire, UK) and incubated at 37°C for 18 hr as a pre-enrichment step. Subsequently, 1 mL of the BPW suspension was transferred to 10 mL modified lauryl sulfate tryptose broth (OXOID, Hampshire, UK), and after further incubation at 42°C for 24 hr, the broth was streaked on Druggan Forsythe Iversen (DFI, Oxoid, Hampshire, UK) agar and incubated at 37°C for 24 hr. Greenish/blue colonies on DFI agar were considered presumptive Cronobacter spp. The suspected isolates were subjected to the biochemical galleries by VITEK 2 compact GN (bioMèrieix, France) for identification of Cronobacter, according to the manufacturer's instructions. Two reference strains Cronobacter muytjensii ATCC51329 and Cronobacter sakazakii ATCC 25944 purchased from American Type Culture Collection. Supplemental biochemical differentiation tests were performed according to Farmer (1980) and Iversen (2006) [25][26].

DNA extraction
Presumptive Cronobacter isolates were incubated overnight at 37°C in Luria Bertani broth. Genomic DNA was extracted using a Genomic DNA isolation kit (Qiagen, Germany) according to the manufacturer's instructions. Concentration and purity of DNA samples were estimated by Nanodrop2000 (Thermo Fisher Scientific, USA).

16S rRNA gene sequences amplification and sequencing
DNA sequencing for the 16S rRNA segment was performed as described by Iversen et al. [1]. PCR amplification using the primers P0 (5'-AGA GTT TGA TCC TGG CTC AG-3') and P6 (5'-GTACGG CTA CCT TGT TAC GA-3'). PCR reaction mixtures were conducted on a volume of 50 μL composed of 1×PCR buffer, 1× Q solution, 2 mM MgCl2, 0.25 mM dNTPs, 1 pM of each primer, 1 U Taq polymerase (Qiagen, UK), and~10ng chromosomal DNA with conditions: 3 min at 95°C; 30 cycles of 30 sec at 95°C, 30 sec at 54°C, and 2 min at 72°C; and a final extension of 5 min at 72°C. The PCR products were cut from 1% gel visualized under UV. After purification the amplified fragments were double stranded sequencing and triple repeats using the primers P0 and P6 by DNA analyzer (ABI 3730xl; USA). All 16S rRNA gene sequences of the 76 isolated strains were submitted to GenBank.

fusA gene sequences amplification and sequencing
The fusA gene PCR amplification was performed as described in Baldwin et al. [27]. After purification using a PCR purification Kit (Qiagen, Maryland, US) according to the manufacturer's instructions, the amplified fragments were double stranded sequencing and triple repeats using the sequencing primers by DNA analyzer (ABI 3730xl; USA). The fusA sequences were queried in MLST databases http://pubmlst.org/cronobacter/info/protocol.shtml to identify the species and fusA alleles sequences. A phylogenetic tree was constructed based on the sequences of the fusA alleles (438 bp) using the maximum likelihood algorithm in MEGA 6 (v. 6.06) [28].

PFGE of the isolates
Pulsed-field gel electrophoresis (PFGE) of Cronobacter isolates was performed as described by Brengi et al. [29]. Strains were grown overnight on blood agar plates and were harvested. XbaI was used as primary restriction enzyme. DNA fragments were separated by electrophoresis (CHEF Mapper, Bio-Rad Laboratories, Hercules, California, US) through a 1% (w/v) agarose gel (Seakem Gold, Rockland, Maine, US) in 0.5×TBE buffer using the program: initial switch time of 1.8 s and final switch time of 25 sec, 20 hr, 120°angle at 6 volts/cm. Gels were stained with GelRed (Biotium, CA, US) and visualized under U.V. light using a GelDoc XR+ system (Bio-Rad laboratories, Hercules, California, US). Dendrograms were constructed using BioNumerics v. 6.6 (Applied Maths, Sint-Martens-Latem, Belgium) and cluster analysis conducted using the DICE coefficient and unweighted pair group method with arithmetic means (UPGMA). Band position tolerance and optimization coefficient were 1.5%. A Salmonella serotype Braenderup strain (H9812) was chosen as the universal size standard.

Antimicrobial susceptibility testing
All isolates were tested using the broth microdilution method as described by the Clinical and Laboratory Standards (CLSI, M7-A7, 2006) [30] for susceptibility to 8 antimicrobial agents ( Table 2). Escherichia coli ATCC 25922 was used as reference stain. The minimal inhibitory concentrations (MIC) were read manually using a light box for interpretation in accordance with criteria provided by CLSI, 2010 [31].

Incidence of Cronobacter in DRP
Seventy-six of the 1012 tested DRP samples were found positive for Cronobacter by culture and 16S rRNA gene PCR amplification and sequencing. The contamination rate among manufacturers ranged from 0-28.8% (Table 1). Two manufacturers (A and H) showed >10% contamination rates, the highest 28.8%. Two manufacturers (J and L) contained no detectable Cronobacter, and the contamination rate in the remainder was 2.2-7.3% (Table 1).

PFGE
BioNumerics software analysis showed the 76 isolates to demonstrate 66 distinguishable Xba I PFGE patterns (Fig 2). A high degree of genetic diversity was revealed using PFGE after genomic DNA digestion; the discriminatory index was 0.9940, as calculated using Simpson's diversity index. Within the 66 PFGE-patterns, C. sakazakii, C. malonaticus, and C. dublinensis exhibited high diversity, while C. muytjensii showed a common pattern (Fig 2).

Discussion
Great attention has been focused on the contamination of PIF by Cronobacter spp. [32][33][34], while information concerning such contamination of DRP products is limited. In the present study, 1012 samples of DRP products from fourteen manufacturers were investigated for the presence of Cronobacter in 2010-2012. Cronobacter spp. contamination was revealed to be serious, occurring at an overall infection rate of 7.5%, with the highest level of contamination being 28.8%, much greater than seen in PIF (2-22%) [7]. Contamination rate varied widely among manufacturers. Some brands contained additional ingredients such as Chinese yam powder and dried fruit for increased nutrition. These ingredients added to rice powder without heat may increase the risk of contamination [35]. In our study, two manufacturers (A and H) that showed a high contamination rate were supplemented with Chinese yam powder, which may have been a contributing factor. Criteria regulating Cronobacter spp. have been established for infant formulas with an intended target age <6 months [36], and European Commission Regulation (No. 2073/2005; No. 1441/2007) requires a zero tolerance of Cronobacter presence in powdered formula intended for infants under 6 months of age. However, there is less attention and no established criteria for DRP Cronobacter contamination, although these foods are frequently used as a substitute for PIF to infant older than four months in some less affluent regions.
The previous study demonstrated that the 16S rRNA gene and the biotyping is not suitable for Cronobacter species identification [27]. Multilocus sequence typing (MLST) approach employing the alleles of 7 genes (atpD, fusA, glnS, gltB, gyrB, infB, and ppsA) showed can identify the Cronobacter species. Among the seven loci, fusA was observed to be the least number of shared alleles among species, and none of the profiles were shared between two or more species is adequate for speciation [2]. Based on fusA sequencing and analysis, we identified four species of Cronobacter from DRP. C. sakazakii and C. malonaticus to be the most common species (66/76), similar to results of other studies [37][38]. And the two species most commonly associated with human disease [13]. Products of one manufacturer contained four Cronobacter species. More than one species was detected in products of six manufacturers, and different species were isolated from the same batch. A single batch of DRP from manufacturer A contained both C. sakazakii and C. dublinensis (Fig 2), indicating that mixed contamination is possible and stresses the need to examine more than one colony from each batch of rice powder.
PFGE is well-established and widely used for the molecular subtyping of bacteria, including Cronobacter spp. [39][40][41]. Sixty-six PFGE patterns were observed among the 76 isolates. Species isolates in the same batch sometimes showed different genetic subtypes (Fig 2), indicating that contamination could have originated from different sources during the manufacturing process (Fig 2). Several manufacturers shared the same genetic subtypes (Fig 2), implying a common original source of contamination. We also observed the same PFGE-pattern in batches produced a year apart, implying ongoing contamination of the rice product (Fig 2). In manufacturer A, 38 isolated Cronobacter strains included four species, 14 allele fusA sequences (Fig 2), and 35 PFGE-patterns, revealing that Cronobacter contamination of a DRP product is complex and may originate from different sources, including harvest, transport, storage of the rice and the manufacturing process. The prevention of DRP Cronobacter contamination not only preventing the rice powder and ingredients contamination, but also get rid of the Cronobacter during manufacturing process.
In the biochemical test, three previously unreported biogroups were observed. Recently, new biogroups of Cronobacter have been identified [34]. It seems likely that further biogroups remain to be identified. In the antimicrobial susceptibility test, all natural Cronobacter isolates were susceptible or intermediately susceptible to chloramphenicol, trimethoprim/sulfamethoxazole, gentamicin, cefotaxime, nalidixic acid, ciprofloxacin, and tetracycline. Cronobacter spp. tend to be more sensitive to most antibiotics than other Enterobacteriaceae, although resistance to ampicillin and cephalothin resistance has developed [22,[42][43][44].

Conclusions
Cronobacter contamination of DRP products is a serious concern, with the contamination rate varying among manufacturers. The species, allele fusA sequences, and PFGE patterns of the identified Cronobacter isolates from DRP were diverse. The presence of these foodborne pathogens in rice powder products is a potential threat to human health, particularly for infants under 6 months of age and vulnerable adults. Greater attention should be paid to the contamination of DRP.