Cronobacter species (formerly known as Enterobacter sakazakii) in powdered infant formula: a review of our current understanding of the biology of this bacterium

Authors

  • Q.Q. Yan,

    1.  UCD Centre for Food Safety, WHO Collaborating Centre for Research, Reference & Training on Cronobacter, School of Public Health, Physiotherapy & Population Science, University College Dublin, Dublin, Ireland
    Search for more papers by this author
  • O. Condell,

    1.  UCD Centre for Food Safety, WHO Collaborating Centre for Research, Reference & Training on Cronobacter, School of Public Health, Physiotherapy & Population Science, University College Dublin, Dublin, Ireland
    Search for more papers by this author
  • K. Power,

    1.  UCD Centre for Food Safety, WHO Collaborating Centre for Research, Reference & Training on Cronobacter, School of Public Health, Physiotherapy & Population Science, University College Dublin, Dublin, Ireland
    Search for more papers by this author
  • F. Butler,

    1.  UCD School of BioSystems Engineering, University College Dublin, Dublin, Ireland
    Search for more papers by this author
  • B.D. Tall,

    1.  US Food and Drug Administration (FDA), Division of Virulence Assessment, OARSA, Centre for Food Safety and Applied Nutrition, MOD 1 Facility, Virulence Mechanisms Branch (HFS-025), 8301 Muirkirk rd, Laurel, MD 20708, USA
    Search for more papers by this author
  • S. Fanning

    1.  UCD Centre for Food Safety, WHO Collaborating Centre for Research, Reference & Training on Cronobacter, School of Public Health, Physiotherapy & Population Science, University College Dublin, Dublin, Ireland
    Search for more papers by this author

Séamus Fanning, UCD Centre for Food Safety, School of Public Health, Physiotherapy & Population Science, UCD Veterinary Sciences Centre, University College Dublin, Belfield, Dublin 4, Ireland. E-mail: sfanning@ucd.ie

Summary

Cronobacter species (formerly known as Enterobacter sakazakii) are opportunistic pathogens that can cause necrotizing enterocolitis, bacteraemia and meningitis, predominantly in neonates. Infection in these vulnerable infants has been linked to the consumption of contaminated powdered infant formula (PIF). Considerable research has been undertaken on this organism in the past number of years which has enhanced our understanding of this neonatal pathogen leading to improvements in its control within the PIF production environment. The taxonomy of the organism resulted in the recognition of a new genus, Cronobacter, which consists of seven species. This paper presents an up-to-date review of our current knowledge of Cronobacter species. Taxonomy, genome sequencing, current detection protocols and epidemiology are all discussed. In addition, consideration is given to the control of this organism in the manufacturing environment, as a first step towards reducing the occurrence of this pathogen in PIF.

Introduction

Cronobacter species (formerly known as Enterobacter sakazakii) are Gram-negative rod-shaped, motile pathogenic bacteria of the family Enterobacteriaceae. These organisms are regarded as opportunistic pathogens linked with life-threatening infections predominantly in neonates (infants <4 weeks of age) (Bar-Oz et al. 2001; Gurtler et al. 2005, Anonymous 2006a,b; Mullane et al. 2007a). Clinical syndromes of Cronobacter infection include necrotizing enterocolitis (NEC), bacteraemia and meningitis, with case fatality rates ranging between 40 and 80% being reported (Bowen and Braden 2006; Friedemann 2009). Infections in older infants and among immuno-compromised adults, mainly the elderly, have also been noted (Bowen and Braden 2006; Gosney et al. 2006; See et al. 2007). The bacterium has been isolated from a range of food sources including dairy-based foods, dried meats, water, rice and others (Baumgartner et al. 2009; Chap et al. 2009; Healy et al. 2010). Surveillance studies detected Cronobacter in a variety of different environments including households, livestock facilities, food manufacturing operations, in particular PIF production facilities (Bar-Oz et al. 2001; Kandhai et al. 2004; Mullane et al. 2007b; Kilonzo-Nthenge et al. 2008). Contaminated powdered infant formula (PIF) has been epidemiologically linked with many of the infections reported (Bowen and Braden 2006). Controlling the microbiological load in infant food products and understanding the optimal growth conditions and epidemiology would contribute positively towards a reduction in the health risk to vulnerable individuals.

Classification of Cronobacter

Cronobacter species were originally referred to as yellow-pigmented Enterobacter cloacae, later being reclassified as a new species, E. sakazakii in 1980 (Farmer et al. 1980). Using partial 16S ribosomal DNA (rDNA) and hsp60 sequencing, Iversen et al. (2004a) divided 126 Cronobacter isolates into four clusters, suggesting that the genus may require re-classification. Later and following further extensive polyphasic analysis, Iversen et al. (2007a, 2008) proposed the reclassification of these bacteria into a new genus called Cronobacter. Originally, six species (C. sakazakii, C. malonaticus, C. turicensis, C. muytjensii, C. dublinensis and C. genomospecies 1) were defined and comprised of the 16 biogroups described in Table 1. A new species (C. condimenti) was identified recently by Joseph et al. (2011), and in addition, C. universalis now replaces the original C. genomospecies 1.

Table 1.   Distribution of biogroups among the genus Cronobacter
Cronobacter speciesBiogroups
Cronobacter sakazakii sp. nov.1, 2–4, 7, 8, 11, & 13
Cronobacter malonaticus sp. nov.5, 9, &14
Cronobacter turicensis sp. nov.16
Cronobacter muytjensii sp. nov.15
Cronobacter condimenti sp. nov.1
Cronobacter universalis sp. nov.Separate genomospecies
Cronobacter dublinensis sp. nov.
 Cronobacter dublinensis subsp.  dublinensis subsp. nov.12
 Cronobacter dublinensis subsp.  lausannensis subsp. nov.10
 Cronobacter dublinensis subsp.  lactaridi subsp. nov.6

This re-classification by Iversen et al. (2007a, 2008) was subsequently supported by both optical mapping and genome sequencing data which confirmed the revision of the taxonomy (Kotewicz and Tall 2009; Kucerova et al. 2010). Although C. sakazakii and C. malonaticus were found to be closely related and difficult to distinguish by 16S rDNA sequence analysis, a seven loci (atpD, fusA, glnS, gltB, gyrB, infB, ppsA) multilocus sequence typing (MLST) scheme was developed to discriminate between these two species (Baldwin et al. 2009). Furthermore, recent findings reported by Joseph and Forsythe (2011) identified a highly stable sequence type (denoted as ST4) within Cronobacter sakazakii and which was responsible for a large proportion of severe neonatal infections, especially neonatal meningitis.

Two of the species genomes were subsequently published (as discussed below), and currently, a collaborative effort is underway to complete the genome sequences of a further 15 isolates of six of the seven Cronobacter species adding substantially to our knowledge of the core genome of this group of bacteria and highlighting particular species-specific features of interest.

Another method of classification used is O-antigen typing and studies describing the nature of the O-antigen associated with Cronobacter species have been reported (Mullane et al. 2008a; Jarvis et al. 2011). The O-antigen is a component of the lipopolysaccharide (LPS) structure located on the outer surface of Gram-negative bacteria and is responsible for serological diversity. Mullane et al. (2008a) initially developed a molecular serotyping method, based on long-range amplification of the rfb-encoding locus (in Gram-negative enteric bacteria located between galF-gnd) followed by MboII digestion (Fig. 1). Using this approach, a PCR-RFLP profile was generated which can be compared across several isolates. Based on this approach, the first two O-serotypes were characterized and denoted as O:1 and O:2. More recently, another five additional O-antigens were serologically identified by Sun et al. (2011) and these correlated with the previously reported PCR-RFLP profiles. Jarvis et al. (2011) extended the original molecular-characterization scheme to include other Cronobacter species and defined new molecular O-serotype gene clusters. Two of these O-serotype gene clusters were shared among C. sakazakii and C. muytjensii, as well as C. malonaticus and C. turicensis strains (Tall, unpublished data). The structural composition of several O-serotypes has now been described (MacLean et al. 2009; Czerwicka et al. 2010; Maclean et al. 2011; Arbatsky et al. 2010, 2011; Shashkov et al. 2011).

Figure 1.

 Restriction fragment length profiles of amplified rfb-encoding loci of Cronobacter following MboII digestion. A selection of some of the serotypes previously classified. Lane 1, ATCC®BAA 894, Cronobacter sakazakii O1; Lane 2, E830, C. sakazakii O2; Lane 3, E615, Cronobacter malonaticus O1; Lane 4, E618, C. malonaticus O2; Lane 5, E464, Cronobacter dublinensis O6; Lane 6, E797, Cronobacter genomospecies 1 O9; Lane M1, 100 bp DNA ladder (New England Biolabs, Hertfordshire, England) and Lane M2, 1 kb DNA ladder (New England Biolabs).

Genomes of the genus Cronobacter

The first sequenced Cronobacter genome, C. sakazakii ATCC®BAA-894 was published by Kucerova et al. (2010). It revealed a single chromosome of 4·4 Mb (57% GC) along with two plasmids, denoted as pESA2 and pESA3 (31-kb, 51% GC and 131-kb, 56% GC, respectively). The source of this isolate was reported to be contaminated PIF used in a neonatal intensive care unit and which gave rise to an outbreak in 2001 in Tennessee, USA. Using comparative genomic hybridization (CGH) techniques, representative isolates including C. malonaticus, C. turicensis, C. muytjensii and C. dublinensis were further investigated, and those genes considered to be part of the core species genome along with other markers unique to C. sakazakii were identified (Kucerova et al. 2010).

Cronobacter sakazakii ATCC®BAA-894 contained approximately 4392 genes as part of its core genome. However, using the CGH approach, 4382 unique, annotated genes from both the chromosome and plasmids were noted, and only 54·9% of genes were common to all C. sakazakii with 43·3% being common across all Cronobacter species. Interestingly, 21 genes were found to be unique in five of the C. sakazakii tested, and these encoded proteins were involved in pilus assembly, a phosphotransferase system (PTS), an acid transporter, N-acetylneuraminate lyase and a toxin/antitoxin system.

The genome of C. turicensis z3032 was published in early 2011 (Stephan et al. 2011) in an attempt to further determine virulence factors and mechanisms of pathogenicity in this bacterium. Following the deaths of two newborn infants in 2005, the latter isolate was cultured from the blood of one child with meningitis (Mange et al. 2006). In this sequenced isolate, the genome was 4·4 Mb (57% GC) in size and contained three plasmids of sizes approx 138-kb pCTU1 (56%), 22-kb pCTU2 (49%) and 54-kb pCTU3 (50% GC). Two hundred and twenty-three genes were annotated as virulence- and disease-related encoding open reading frames (ORFs); however, 9·27% (413 of 4455) encoded proteins were of unknown function. Because these latter ORFs lacked similarity to sequences already in the current databases, these could potentially have important pathogenic functions.

Franco et al. (2011a) reported the in silico analysis of two plasmids, one from C. sakazakii ATCC®BAA-894, pESA3 and the other from Cronobacter turicensis z3032, pCTU1, each of which contained a RepFIB replicon. Both plasmids possessed two iron acquisition systems (eitCBAD and iucABCD/iutA) essential for survival and successful pathogenesis. Ninety-seven per cent of 229 strains, representing seven of the eight Cronobacter species, possessed a RepFIB plasmid. The presence of a Cronobacter plasminogen activator-encoding gene (cpa) [encoded on pESA3], a type 6 secretion system (T6SS) [also encoded on pESA3] and a filamentous haemagglutinin/adhesin (FHA) gene locus (located on pCTU1) suggested the existence of unique virulence determinants in these species. The cpa-encoding gene encodes an outer membrane protease implicated in serum resistance, a feature that would facilitate Cronobacter species in crossing the blood–brain barrier and causing meningitis (Franco et al. 2011b). The T6SS acts to translocate putative effector proteins aiding in bacterial pathogenesis, while FHA pCTU1 contains fhaB, fhaC genes (which encode proteins with similarly identity to a transported and transporter protein as part of a two-partner secretion system) and five associate putative adhesins. Further studies are now in progress to extend our understanding of the functional roles of these plasmid-encoded loci.

Virulence characteristics of Cronobacter

Information from epidemiological studies along with in vitro mammalian tissue culture assays has shown that Cronobacter isolates demonstrate a variable virulence phenotype (Caubilla-Barron et al. 2007; Townsend et al. 2007, 2008). Only isolates of C. sakazakii, C. malonaticus and C. turicensis have been linked with neonatal infections (Healy et al. 2010; Kucerova et al. 2010).

Currently, few clues as to the mechanisms involved are known, though, data from genome sequencing efforts highlighted several potential markers that may be helpful candidates for future studies (Kucerova et al. 2010; Stephan et al. 2011).

The first putative Cronobacter virulence factors to be described were enterotoxin-like compounds produced by four of 18 isolates studied (Pagotto et al. 2003). Using conventional tissue culture-based assays, Cronobacter is known to attach to intestinal cell lines in vitro and survive within macrophages for periods of time (Townsend et al. 2008). Franco et al. (2011b) recently demonstrated resistance to complement-mediated killing of C. sakazakii, and this was associated with the presence of cpa which is contained on the pESA3 plasmid.

The outer membrane protein A, encoded by the ompA gene, is probably the best characterized virulence marker. This was originally reported by Nair and Venkitanarayanan (2006) and shown to be required for binding of the bacterium to human brain microvascular endothelial cells (BMEC) (Nair et al. 2009). More recently, Kim and Loessner (2008) reported that the disruption of tight junctions significantly enhanced adherence of C. sakazakii to Caco2 cells in culture and that the same marker was required for basolateral invasion (Kim et al. 2010a,b). The ompA-encoding gene is thought to be present in all Cronobacter strains tested, and this marker has also been linked with invasive Escherichia coli, which causes neonatal meningitis (Prasadarao et al. 1996; Kim 2000).

At the epithelial cell surface, C. sakazakii infection results in damage of these cells, following the recruitment of greater numbers of dentritic cells, compared with macrophages and neutrophils (Emami et al. 2011). Using a NEC mouse model, these effects were shown to be mediated through OmpA and involved inducible NO synthase (iNOS).

From an analysis of the annotated genes in C. sakazakii ATCC®BAA-894, Kucerova et al. (2010) highlighted several markers including, ibeA, ibeB, yijP and ompA, which were previously identified in other organisms associated with invasion of BMEC (Prasadarao et al. 1996; Huang et al. 1999, 2001; Wang et al. 1999). Interestingly, ibeB (a gene synonymous with cusC), which belongs to a cluster of genes encoding a copper and silver resistance cation efflux system, facilitates the invasion of BMEC cells (Franke et al. 2003); although this gene was found in the reference strain C. sakazakii ATCC®BAA-894, the genes ibeA and yijP produced no matches. When assessed, it was found that the complete cation efflux operon (cusA, cusB and cusC) and its regulatory gene cusR were present in isolates associated with neonatal infections (including C. sakazakii, C. turicensis and C. malonaticus) and absent in the other tested strains (Kucerova et al. 2010).

Detection protocols

Conventional bacteriological culture

The first detection method developed for Cronobacter species was described by Muytjens et al. (1988). Based on this protocol, the US Food and Drug Administration (US FDA) recommended a method to isolate and enumerate E. sakazakii from powdered infant formula in 2002. In 2006, the International Organization for Standardization (ISO 2009) and the International Dairy Federation developed a technical standard protocol for the detection of Cronobacter species from milk-based powdered formula known as ISO/TS 22964 (Anonymous 2006a,b) (described in Table 2). More recently, the US–FDA method was revised to combine both a PCR assay and two newly developed chromogenic agars for detection (Chen et al. 2009; Chen 2011).

Table 2.   Detection protocols for Cronobacter in PIF
ProcedureFDA (Original)ISO/TS 22964FDA (revised)
Pre-enrichmentMake 1 : 10 (w/v) of sample in distilled water, incubated overnight at 36°CMake 1 : 10 (w/v) of sample in BPW, incubated at 37°C for 18 ± 2 hMake 1 : 10 (w/v) of sample in BPW, incubated at 36°C for 6 h
Selective enrichmentTransfer 10 ml pre-enrichment to 90 ml EE broth, incubated overnight at 36°CTransfer 100 μl pre-enrichment to 10 ml mLST/vancomycin medium, incubated at 44°C for 24 ± 2 h 
Selection/isolationMake an isolation streak and spread plate from each EE broth onto VRBG agar, incubated overnight at 36°CStreak from the cultured mLST/ vancomycin medium one loopful on the chromogenic agar in Petri dishes, incubated at 44°C for 24 ± 2 hCentrifuge 40 ml samples, 3000 g, 10 min and resuspend pellet in 200 μl PBS; Spread 100 μl onto chromogenic media, incubated overnight at 36°C
ConfirmationPick five presumptive positive colonies and streak onto TSA, incubated overnight at 25°CSelect five typical colonies and streak on TSA agar, incubated at 25°C for 48 ± 4 hPick two typical colonies from each chromogenic media confirmed with real-time PCR, API 20E, Rapid ID 32 E
IdentificationYellow colonies are confirmed with the API 20E test kitSelect one yellow colony from each TSA plate for biochemical characterization 
Detection time (days)563

Pre-enrichment of the PIF samples to be tested is a requirement in these three protocols, and the time duration varies from a maximum overnight period (ranging from 18 to 24 h) to a minimum time period of 6 h, followed by selective enrichment and subsequent isolation using selective agars/media. Typical colonies are confirmed using a selective agar and/or a suitable real-time PCR assay, with the final identification based on either biochemical/molecular characterization. In the revised US–FDA protocol, there is one enrichment step which is then followed by a molecular method used for quick confirmation. This approach eliminates two days from the detection procedure compared with the original protocol.

Selective media for Cronobacter, including Leuschner–Bew agar (Leuscher and Bew 2004), Druggan-Fosythe-Iversen agar (Iversen et al. 2004b), Oh-Kang agar (Oh and Kang 2004), ESPM agar (Restaino et al. 2006) and HiCrome™Cronobacter spp. agar (Sigma-Aldrich, Switzerland), have been developed, based on the α-glucosidase enzyme marker (Iversen et al. 2004b) identified by Muytjens et al. (1984) and β-cellobiosidase activity (Restaino et al. 2006) which is present in all Cronobacter strains. Moreover, violet red bile agar (VRBA), MacConkey agar and desoxycholate agar, which are selective for Gram-negative bacteria, are also available for the isolation of Cronobacter (Druggan and Iversen 2009; Forsythe 2009) from foods.

However, despite the availability of selective agar media, some were shown to insufficiently support the growth of all Cronobacter strains (Iversen and Forsythe 2007) and other related species, such as Enterobacter helveticus, Enterobacter pulveris and Enterobacter turicensis, which are often found in the same ecological niches. Therefore, improvements in the design of the selective media for the isolation and identification of Cronobacter were required.

O’Brien et al. (2009) described the design of a one-step pre-enrichment and enrichment protocol using a chromogenic medium. In this detection strategy, the specific broth developed [denoted as Cronobacter enrichment broth (CEB)] facilitated a shortened two-day culture method for the detection of Cronobacter species in PIF. Mullane et al. (2006) utilized a cationic-magnetic-bead capture technique to improve the sensitivity of detection for Cronobacter from PIF.

The accuracy and reliability of commercially available identification kits for Cronobacter have been questioned, with reports of false-negative and false-positive identification (Restaino et al. 2006; Iversen and Forsythe 2007). However, Gen III is currently the only commercially available identification kit with the original six species included (Healy 2010).

Immuno-based detection protocols

Immuno-based assays are convenient for detection methods that can be applied to detect specific bacteria. Assays using monoclonal antibodies are widely used in research as rapid detection tools. These approaches can improve the sensitivity and specificity for detection. Commercial kits based on enzyme-linked immunosorbent assay (ELISA) technology have been developed, and the VITEK® immunodiagnostic assay system (denoted as VIDAS®, bioMerieux Vitek Inc., Hazelwood, MO, USA) has been used as a rapid detection platform for Salmonella, E. coli O157:H7, Listeria species, Campylobacter jejuni and Staphylococcus species enterotoxins. The VIDAS®Salmonella method has been validated and certified by the Association of Official Analytical Chemists (AOAC) as an approved method of analysis in foods. This protocol was also approved by other regulatory organisations including Health Canada, the European Microbiological Method Assessment Scheme (EMMAS) and German Normalization Institute (DIN). Research on a VIDAS®-based Cronobacter method is currently in progress; early results suggest that the method is fast, reliable and sensitive for the detection of Cronobacter in a range of test matrices (Q.Q. Yan, unpublished data).

Molecular-based detection protocols

Molecular detection techniques have always been regarded as useful tools to extend our understanding of the epidemiology of an organism. Usually, these assays are designed to target unique genes present in the pathogen of interest. Many of the more recent assay formats are based on real-time PCR and several have been designed for the specific detection of Cronobacter (Malorny and Wagner 2005; Seo and Brackett 2005; Drudy et al. 2006; Nair and Venkitanarayanan 2006; Kothary et al. 2007; Zhou et al. 2008). Targets used (Table 3) include the 16S rRNA gene, 16S -23S rDNA intergenic region, the dnaG gene, the ompA gene, the 1,6 α-glucosidase-encoding gene (gluA) and a zinc-containing methalloprotease (zpx). More recently, a PCR-based method developed by Stoop et al. (2009) was extended by A. Lehner, C. Fricker-Feer and R. Stephan (unpublished data) to include the new species described. The latter protocol facilitated the detection of all seven known species within the Cronobacter genus using a mismatch-PCR-based approach. The applications of these molecular-based protocols can support the traditional culture-based approaches (Fig. 2). Interestingly, using the latter protocol, C. malonaticus and C. sakazakii could not be differentiated using this approach, and therefore, a second PCR was required to accurately identify each of these species. In 2011, Yan et al. (2011) published a PCR- and array-based biomarker verification study for the detection and identification of Cronobacter spp. Here, these authors propose to elucidate virulence markers which may be helpful as biomarkers for differentiating Cronobacter spp. and Salmonella spp. from other food-borne pathogens. While these putative markers were identified, further validation experiments are currently being conducted.

Table 3.   Gene targets useful for the detection of Cronobacter and related species
 Gene targetsReference
Genus loci
 ribosomal  DNA (rDNA)16S rRNAIversen et al. (2004a)
23S rRNADerzelle et al. (2007)
tRNAGluHassan et al. (2007)
FISHIversen et al. (2007a)
Almeida et al. (2009)
 1,6 α-glucosidasegluAIversen et al. (2007b)
 MMS operondnaGSeo and Brackett (2005)Drudy et al. (2006)
 Zinc-containing  metalloproteasezpxJaradat et al. (2009)
 Outer membrane  protein AompANair and Venkitanarayanan (2006)
Species loci
 rfb (O-antigen)wehC [CsakO:1] & wehI [CsakO:2]Mullane et al. (2008a)
wzx [CsakO:3; CturO:1;Cmuy O:1; and Cmal O:1 and O:2 ]Jarvis et al. 2011)
 β-subunit of RNA  polymeraserpoBStoop et al. (2009)
Other gene targets
 RNaseP infB (initiation factor)  
Figure 2.

 A 1·0% agarose gel showing the amplification of rpoB amplicons used to identify six of the genus species (Stoop et al. 2009). Lane 1, Cronobacter sakazakii SP291; Lane 2, Cronobacter malonaticus E766; Lane 3, C. malonaticus E766 (following a confirmatory second PCR); Lane 4, Cronobacter muytjensii ATCC®51329; Lane 5, Cronobacter dublinensis CFS237; Lane 6, Cronobacter turicensis E626; Lane 7, Cronobacter genomospecies 1 E797 and Lane M, 100 bp DNA ladder.

Molecular subtyping has long been regarded as a useful approach that can be applied to elucidate the nature of those bacteria colonizing a particular ecological niche. Mullane et al. (2007b) applied pulsed-field gel electrophoresis (PFGE) to characterize and track Cronobacter species in a PIF processing facility (Fig. 3). The study provided a basis for the development of efficient intervention measures contributing to the reduction of Cronobacter in the PIF manufacturing environment. A similar approach using the second generation subtyping method, multiple-locus variable-number tandem-repeat analysis (MLVA), was subsequently applied to subtype a collection of geno- and phenotypically diverse Cronobacter isolates (Mullane et al. 2008b). However, a standardized PFGE protocol is close to completion and has already been validated by PulseNet, a network of national and regional laboratory networks dedicated to tracking food-borne infections.

Figure 3.

 Pulsed-field gel electrophoresis (PFGE) profiles used to characterize and track Cronobacter species in a PIF processing facility. Lane 1, factory sample 1; Lane 2, factory sample 2; Lane 3, factory sample 3; Lane 4, factory sample 4; Lane 5, factory sample 5; Lane 6, factory sample 6; Lane 7, factory sample 7; Lane 8, factory sample 8; Lane 9, factory sample 9 and Lane M, Salmonella Braenderup, molecular marker, genomic DNA digested with XbaI. The arrow heads at the foot of the image show that in lanes 3 and 7, these isolates have the same PFGE profile and would be considered to be indistinguishable. Similarly, isolates contained in lanes 4 and 5 have another indistinguishable profile cluster.

A significant reduction in both time and cost associated with genome sequencing has made molecular detection methods increasingly accessable. Multilocus sequence analysis (MLSA) based on recN, rpoA and thdF genes was used in describing the genomic similarity of Cronobacter genes by Kuhnert et al. (2009). El-Sharoud et al. (2009) applied recN gene sequence analysis to isolate Cronobacter strains recovered from dried milk and related products. A similar MLST typing scheme using the following seven housekeeping genes: atpD, fusA, glnS, gltB, gyrB, infB, ppsA (3036 nt concatenated length) has been developed by Baldwin et al. (2009), and a database containing defined sequence types covering all Cronobacter spp. is currently being maintained at Oxford University. The database and MLST analytics can be accessed at http://www.pubMLST.org/cronobacter. However, the scheme has not been applied for use in any ongoing epidemiologic investigation.

Whole-genome sequencing efforts can be expected to facilitate the correct identification of the bacterial species being studied; it can also provide detailed information regarding the unique geno- and phenotypic features. Moreover, these approaches can be used for comparative purposes to rapidly and simultaneously investigate the presence/absence of all annotated genes or coding sequences or/and polymorphisms that may contribute to a specific morphology or physiology. Microarray-based comparative genomic indexing analysis of Cronobacter genus was originally performed by Healy et al. (2009), and this study identified species-specific genes that could be evaluated as candidate markers for inclusion in a molecular-based detection protocol. More recently, next generation sequencing has been used for the comprehensive analysis of eight Cronobacter genomes to better understand the pathogenicity and evolution of the genus alongside the characterization of virulence genes (Ji 2011). While this approach might reveal species-specific genomic information, in essence, this is a method that will provide much more detail necessary for subtyping an organism.

Epidemiology of Cronobacter species

Cronobacter infections are rare and are often underreported, especially in developing and less-developed countries (Estuningsih and Sani 2008; Friedemann 2009). Thus, the epidemiology of Cronobacter species is incomplete and poorly described. Bowen and Braden (2006) first attempted to describe the epidemiology related to these infections. These authors analysed the clinical case notes of 46 invasive infant E. sakazakii infections between 1961 and 2005. These included 12 infants presenting with bacteraemia, 33 with meningitis, and 1 with a urinary tract infection. Infants presenting with bacteraemia had higher birth weights (2454 g), and a gestational age of 37 weeks, with infections occurring at a younger age (6 days), compared with those infants presenting with meningitis. Meningitis was reported to have a high mortality rate (42%) and many of the survivors (more than 74%) suffer chronic neurological and developmental complications (Reij and Zwietering 2008). More recently, Friedemann (2009) analysed 120–150 neonatal Cronobacter-confirmed infections based on data published between 2000 and 2008. The overall lethality of the 67 invasive infections noted was 26·9%. Lethality of Cronobacter meningitis, bacteraemia and necrotizing enterocolitis (NEC) was calculated to be 41·9% (P < 0·0001), <10% and 19·0% (P < 0·05), respectively. Interestingly, this study identified two key risk factors, a longer gestational age at birth and parentage not from Europe, as significant factors for a higher reporting probability of neonatal Cronobacter meningitis based on a logistic regression models.

Many infections in newly born babies are transmitted from mother to child through the mother’s birth canal, which has been suspected as a source of Cronobacter infections (Townsend and Forsythe 2008; Kandhai 2010). In most cases, both the route of exposure and the incubation period are generally unclear. Two early described outbreaks demonstrated a clear relationship between Cronobacter isolates from infected patients and the isolates cultured from unopened cans of PIF of the same batch consumed by these same patients (Clark et al. 1990; Block et al. 2002). Although powdered infant formula is regarded as an important source of this pathogen, environmental or extrinsic sources of contamination should not be excluded (Noriega et al. 1990). It has been reported that plant material may be the natural habitat for Cronobacter species (Schmid et al. 2009). Moreover, reports on Cronobacter species infections in immune-compromised adults (Gosney et al. 2006; See et al. 2007) may indicate other potential sources of contamination, such as the home environment or transient carriage states present in adult care takers, among others (Kandhai 2010; unpublished data and personal communication with Anna Bowen, CDC). It was estimated that the annual incidence rate among the low birth weight infants (i.e. weight <2500 g; children <12 months of age) was 8·7 per 100 000 infants in the United States of America (FAO/WHO 2006). Similarly, another study estimated an incidence rate of 9·4 per 100 000 among very low birth weight infants (i.e. weight < 1500 g) (Stoll et al. 2004).

Additionally, the prevalence of Cronobacter species infections in adults is increased in the elderly who have experienced strokes that have affected their abilities to swallow (dysphagia) and may therefore require reconstituted powdered protein supplements as part of their daily diet (Gosney et al. 2006; FAO/WHO 2008). This is a problem that is likely to become more common because of the ageing of the world’s population, and as trends for consumption of synthetic, dehydrated formulas for such patients increase.

Control in manufacturing environment

Describing the epidemiology of Cronobacter in a PIF production environment can be regarded as a useful first step in an attempt to reduce the bacterial load and control dissemination. Mullane et al. (2008c) investigated Cronobacter species in a powdered milk protein manufacturing facility and highlighted the importance attached to the correct installation and maintenance of air filters to reduce the dissemination of Cronobacter and other biological hazards in the food production setting. Furthermore, these strategies facilitate the distinction between transient colonizing bacteria and those that can persist for long periods of time (Osaili and Forsythe 2009). These latter organisms could be regarded as having adapted to the production environment. Data from on-going surveillance studies identified a Cronobacter sakazakii isolate that demonstrated a remarkable phenotype. This isolate adapted and acquired a tolerance to temperatures of 60°C for periods of time. The bacterium also remained in a viable state when in a desiccated state for several weeks (Cooney, unpublished data). This finding clearly supports the need to continuously monitor Cronobacter species in the production environment and also to identify those isolates that persist. An understanding the molecular mechanisms associated with such characteristics may be helpful as a means of eliminating them.

Walsh et al. (2011) provided further insights into the variability among strains in terms of their survival characteristics. Environmental strains appeared to survive better in dry ingredients, when tested in inulin and lecithin over a 338-day period. Also, clinical strains appeared to be more thermotolerant compared with their environmental counterparts. This resistance may be linked to a phenotype involving the production of extracellular polysaccharide (EPS). Walsh et al. (2011) proposed, based on these data, that the ability to produce EPS reduces thermotolerance. Based on these observations, clinical strains may have patho-adaptation resulting in a less desiccation-resistant phenotype.

Gajdosova et al. (2011) characterized an 18-kbp region from a collection of Cronobacter isolates and compared this locus to members belonging to other genera, such as Enterobacter, Citrobacter and Escherichia, where its presence was positively correlated with increased thermotolerance. The contribution of the 22 open reading frames annotated within this region to the thermotolerance phenotype remains unclear and requires further detailed experimental investigation.

Farmer et al. (1980) observed that many of the isolates that were studied originally had two different colony types referred to as types A and B. Type A colonies were described as ‘either dry or mucoid, crenated (notched or scalloped edges), and rubbery when touched with a loop’. Type B colonies were described as possessing ‘a typical smooth colony appearance, easily removed with a loop’. Based on similar descriptions of colonies of Salmonella (Anriany et al. 2006), we currently related these Cronobacter colony descriptions to those reported by Zogaj et al. (2003) as colonies expressing the rugose phenotype. Various studies using other enteric organisms such as Salmonella (Anriany et al. 2006), Vibrio cholerae (Ali et al. 2002) and Grimontia hollisae (formerly Vibrio hollisae) (Curtis et al. 2007) have shown that strains expressing the rugose phenotype impart: (i) a resistance to desiccation and antimicrobial agents such as hypochlorite; (ii) an increased ability to form biofilms; and (iii) the reversible rugose to smooth colony phase variation that was originally described by Farmer et al. (1980), respectively. It has been reported that Cronobacter species possesses a bcsABZC operon (Grimm et al. 2008) which is required for cellulose expression. Grimm et al. (2008) showed that a strain of E. sakazakii, ES5, which was used in development of a bacterial artificial chromosome (BAC) library, possesses the genetic machinery for cellulose biosynthesis. These studies also suggested that the overexpression of an exopolysaccharide composed of cellulose may play a role in rugosity, but the involvement and expression of other bacterial exopolysaccharides in rugosity should not be discounted. Understanding how these phenotypes evolve at a molecular level may facilitate the development of strategies to eliminate the persistent population of Cronobacter. Several such strategies have been suggested recently. These include the use of biocides and natural antibacterial compounds, such as essential oils and polyphenols, all of which are effective against Cronobacter (Brul 2004; Manach et al. 2004; Kim et al. 2010a,b).

Biocides to control Cronobacter

In attempts to ensure food safety and improve hygiene measures, the use of biocides and chemical-based disinfection protocols to control the microbial ecology of the production environment is in widespread use throughout the modern food industry. Several studies evaluated the ability of Cronobacter to survive treatment with common biocides used for this purpose. A study by Condell et al. (2012) evaluated the efficacy of eight commercially available biocide formulations against a collection of 90 Cronobacter species cultured from various origins: food, water, clinical and environmental. This study determined that each biocide formulation was completely effective in killing all Cronobacter strains tested, when cultured in a planktonic state, at the working concentration recommended by the manufacturer (Fig. 4). Mean minimum inhibitory concentration (MIC) values determined for these biocides ranged from 0·2 to 50% of the recommended working concentration. However, when the biocide formulations were re-analysed for their efficacy against surface-dried bacterial cells and Cronobacter contained in a biofilm, they all displayed a reduced killing effect. Five of the biocides were ineffective at killing Cronobacter contained in a biofilm at twice the reported working concentration and three were unable to eradicate surface-dried Cronobacter after an hour of contact time. These results further emphasized a critical need to understand the role of stress responses in Cronobacter species.

Figure 4.

 Minimum inhibitory concentration of planktonic Cronobacter measured against eight food industry biocide formulations.

In a similar approach, Kim et al. (2007) reported that different disinfectants varied in their lethality to Cronobacter and showed reduced activity against Cronobacter in a biofilm, or when dried on stainless steel.

Both of these studies indicated that biofilm or surface-dried-associated phenotypes may contribute to the persistence of Cronobacter in the production environment. Cleaning regimes should consider this possibility, incorporating control strategies to prevent the development of biofilm and surface-dried communities in the food processing environment.

Other studies evaluated the use of natural biocides as food additives as an alternative natural means to control Cronobacter. Al-Holy et al. (2010) studied the effect of using natural biocides as food additives for the control of Cronobacter. In their study, lactic acid (LA), copper sulfate and monolaurin were used to inactivate Cronobacter species. Data showed that the use of a synergistic interactive combination of LA and copper sulfate could be beneficial to control Cronobacter in PIF industry.

Although biocides play an important role, their widespread use is not without its risks. The current scientific literature continues to report on the links between the over-use of biocides in the clinical and domestic environments and the subsequent selection of bacterial isolates displaying an increased tolerance to these agents concomitant with the emergence of cross-resistance to clinically important antibiotics. Condell et al. (unpublished data) evaluated the ability of Cronobacter to develop a tolerance to commercial food-grade biocide formulations and to active biocidal compounds contained in these products. Sublethal exposure of Cronobacter failed to increase their tolerance to any of the formulations tested or the actives contained therein. Nonetheless, the potential for selection and dissemination of pathogens, such as Cronobacter in the food chain, which have become biocide tolerant, is an area that should be considered and monitored in the designing and implementation of cleaning regimes.

Essential oils

Essential oils, aromatic-based liquids derived from plants, have been shown to inhibit some food-borne pathogen (Brul 2004). These oils may demonstrate potential for use in the food industry, particularly as consumer demands for pathogen-free food along with the minimal use of artificial preservatives continues to increase (Brul 2004). The capacity for the use of essential oils in the control of Cronobacter has been evaluated, with work mainly focusing on ‘trans’-cinnamaldehyde (TC), a component of bark extract from the cinnamon plant (Amalaradjou and Venkitanarayanan 2011a).

Studies examined the inhibition of Cronobacter on abiotic surfaces (Amalaradjou and Venkitanarayanan 2011b) as well as evaluating its use in reducing the tolerance of Cronobacter to environmental stresses (Amalaradjou and Venkitanarayanan 2011c). These studies determined that TC was an effective agent in the inhibition and inactivation of Cronobacter biofilms, in the reduction of Cronobacter tolerance to desiccation, acid and osmotic stresses and in enhancing the killing effect of heat treatments. Indeed, this compound was shown to down-regulate genes involved in biofilm formation (Amalaradjou and Venkitanarayanan 2011b) in addition to many important stress regulators within the genus, such as rpoS, phoP/phoQ and ompR (Amalaradjou and Venkitanarayanan 2011c). Further work by Amalaradjou and Venkitanarayanan examined the effect of TC on Cronobacter by analysing alterations in the proteome following exposure to TC. This work determined that TC disrupts Cronobacter metabolism, down-regulating proteins involved in amino acid metabolism as well as the F0F1 ATPase, disrupting the production of ATP. TC was also found to inhibit proteins involved in active transport across the membrane, flagellar biosynthesis, many genes associated with bacterial survival and defence such as catalase, superoxide dismutase and metaloprotease, as well as OmpA, a known Cronobacter virulence factor involved in adherence and invasion. The ability of TC to disrupt proteins associated with motility and survival in the host indicated that it may have a compromising effect on Cronobacter virulence (Amalaradjou and Venkitanarayanan 2011a). These data suggest that TC may be a promising agent useful in the control of Cronobacter. However, further work is required to identify and verify an appropriate application protocol for essential oils in the control of food-borne pathogens.

Polyphenols

Polyphenols are compounds that are abundant in nature. They are produced by plants and animals and often found in food and soil. Plants produce polyphenols as a defence mechanism to protect against infections, thus many plant polyphenols elaborate an antibacterial activity (Manach et al. 2004). Polyphenols have been evaluated as a potential food additive for the control of Cronobacter. A study by Kim et al. (2010a,b) concluded that red muscadine juice, a rich source of polyphenols, displayed strong antimicrobial action against Cronobacter, with tannic acid showing the greatest effect. This study suggested that the red muscadine juice had the potential for use in baby food as an inhibitor of Cronobacter (Kim et al. 2010a,b).

Prebiotics

Prebiotics have been emerging over recent years as a beneficial food ingredient. These are known to stimulate the growth of beneficial bacteria in the intestine and have been recently reported to inhibit bacterial adherence to host cell surfaces in vitro (Gibson et al. 2005). Prebiotics, in particular polydextrose (PDX) and galactooligosaccharides (GOS), have been evaluated for use in the prevention of infection by Cronobacter. A study by Quintero et al. (2011) determined that GOS and a PDX/GOS combination had an inhibitory effect on the adherence of Cronobacter to intestinal-derived cells in tissue culture experiments. This work indicated that these prebiotics are inhibitory during the initial step in the establishment of a Cronobacter infection and therefore show some potential in the prevention of Cronobacter-related illness (Quintero et al. 2011). Furthermore, prebiotics have the added advantage of being food grade agents, and further work in this area could be important in the development of prebiotics as a natural, noninvasive and safe method for the control of Cronobacter infection.

Risk analysis

The implementation of microbiological criteria is one of the control measures that should be employed to reduce the risk of Cronobacter infection associated with PIF. There are several types of sampling plans that can be employed for the microbiological testing of PIF. The most common approach is attribute testing, which can be used to determine the presence/absence of Cronobacter.

A workshop on Cronobacter was convened in 2004, jointly by the United Nations (UN)-Food and Agriculture Organization (FAO) and World Health Organization (WHO) in response to a request for scientific advice from the Codex Alimentarius Committee on Food Hygiene (EC 2005), to provide information or guidance to PIF manufacturing companies and parents in regard to Cronobacter infections (FAO/WHO 2004). For any single lot of PIF, the level of contamination and the within-lot variability will determine the likelihood that a sample will be positive for Cronobacter and thus accepted or rejected. Three parameters can be used to characterize PIF production: the mean log concentration (MLC) (CFU g−1) of Cronobacter across all PIF lots, the between-lot standard deviation (σb) across all PIF lots and the within-lot standard deviation (σw) for individual PIF lots. Therefore, related process controls should be developed according to these parameters.

In recent years, Hazard Analysis and Critical Control Point (HACCP) systems designed for prevision and management of contamination has been widely accepted and applied. HACCP programmes are mandatory in the food industry, but infant formula establishments are not required at this stage to have quality system standards such as ISO 9000 in place. Nevertheless, all infant formula industry-related manufactures are required to have in place effective good manufacturing practice (GMP) and related quality control procedures, which help these food factories to monitor the product line, manage the production quality and further improve products and processes.

Future considerations

Future attention to improve the control of Cronobacter should focus on five aspects as follows (FAO/WHO 2004):

  • 1 For manufacturing and factories:
    • i Implementing an effective environmental monitoring programme, such as GMP and HACCP, to control the microbiological hazards from the raw materials, during the entire processing chain, until the final products, so as to minimize the entry of Cronobacter into the PIF environment and avoid the growth/persistence of this pathogen in PIF products.
    • ii Improving PIF product labels and communicating with consumers to create awareness of the correct method to be used for reconstituting PIF products.
    • iii Collaborating with researchers and governments to provide assistance on solving Cronobacter-related issues.
  • 2 For governments and intergovernmental bodies:
    • i Setting a standard regulation and/or legislation directive for Cronobacter to guide food manufacturing towards improved control in the quality of their PIF products and further reduce the risk of Cronobacter infection.
    • ii Educating healthcare professionals to provide high-quality training to parents and professional caregivers to ensure PIF is prepared, handled and stored properly.
  • 3 For hospitals:
    • i Using commercial sterile liquid formula or formula, which has undergone an effective point of use decontamination procedure and which is to be given to high-risk infants.
    • ii Educating parents in relation to the proper way of raising children being fed with PIF.
    • iii Assisting developing countries in establishing effective measures to minimum risk on Cronobacter infection.
  • 4 For researchers and public health officials:
    • i Developing a better understanding of the ecology, virulence and other characteristics of Cronobacter as a means of developing effective ways to reduce contamination in reconstituted PIF.
    • ii Investigating and reporting of sources and vehicles and establishing laboratory-based domestic and international networks such as an integrated food safety system (IFSS), a mandate put forth by of the United States Food Safety and Modernization Act of 2011 (US-FSMA) which will house the Pathogen-Annotated Tracking Resource Network system (PATRN) (Tall 2010) as a prototype of an IFSS.
    • iii Developing effective and rapid Cronobacter detection protocols for the PIF industry.
  • 4 For consumers:
    • i Being aware of the healthcare information-related infants.
    • ii Obtaining scientifically grounded assistance from professionals, such as caregivers, doctors, PIF researchers.

Ancillary