The Cronobacter genus is an emergent group of bacterial pathogens in the Enterobacteriaceae family. The majority of infections (bacteraemia, and urinary tract infections) are in the adult population, however the most publicized cases are severe, and frequently fatal infections in neonates and infants [1, 2]. In these highly vulnerable populations, the organism is associated with necrotizing enterocolitis and a highly destructive form of meningitis in which the bacterium crosses the blood–brain barrier and causes abscess formation in the brain cavity [3, 4]. The genus is composed of seven species, and multilocus sequence typing has been used to describe the diversity of the genus [5, 6]. Evolutionary analysis suggests that the C. sakazakii species separated from the rest of the Cronobacter genus 15–23 million years ago (MYA) [6]. Recent whole genome studies have revealed that C. sakazakii is the only Cronobacter species that has the nanAKT gene cluster encoding for sialic acid utilization [7, 8]. Since sialic acid is found in breast milk, infant formula, mucin lining the intestinal tract and gangliosides in the brain [9], it is plausible that this metabolism may account for the predominance of C. sakazakii in neonatal and infant infections. However no laboratory studies investigating this trait have been undertaken to date.

Sialic acid can exist in nearly 50 different forms with the most studied being 2-keto-3-deoxy-5-acetamido-D-glycero-D-galacto-nonulosonic acid, often abbreviated as Neu5Ac. This sialic acid is generally found bound to sugars to form polysaccharides, and also can be bound to lipids or proteins to form sialo-glycoconjugates. With few exceptions, sialic acid conjugates are notably absent from many eukaryotic lineages, including most protostomes, plants, fungi, and protists. It is postulated that sialic acid synthesis evolved in animals and later emerged in bacterial pathogens and commensals either by convergent evolution or horizontal gene transfer. A number of microbial strategies have evolved which target host sialic acids for adherence, mimicry, and degradation [10, 11]. Some bacteria can produce sialidase (or neuraminidase), encoded by nanH, to cleave sialic acid from the glycoconjugate forms. This gene has low homology (<30%) across bacterial groups, and has not been described in many organisms [12, 13]. Although neonatal meningitic E. coli K1 is able to grow on sialic acid, it lacks the enzyme sialidase. However it is possibly able to obtain sufficient sialic acid from the activities of other sialidase-producing bacteria in the environment or from the host cells expressing the enzyme in conditions of inflammation [10, 13].

The uptake of sialic acid through the outer cell membrane of Gram negative bacteria is by an outer membrane porin, NanC. There are three different types of transporters for the inner membrane: NanT, a major facilitator superfamily (MFS) protein; TRAP, a tripartite ATP-independent periplasmic transport system; and an ATP-binding cassette (ABC) transporter. All members of the Enterobacteriaceae studied to date have shown the presence of the single-component NanT transport system [10, 11, 14]. Once transported into the cell, the Neu5Ac lyase (NanA) converts sialic acid (Neu5Ac) into N-acetylmannosamine (ManNAc) and phosphoenolpyruvate (PEP). NanK is an ATP-dependent kinase specific for ManNAc generating N-acetylmannosamine-6-phosphate (ManNAc-6-P). ManNAc-6-P epimerase (NanE) then converts the ManNAc-6-P into N-acetylglucosamine-6-phosphate (GlcNAc-6-P). GlcNAc-6-P deacetylase (NagA) and glucosamine-6-P deaminase (NagB) convert GlcNAc-6-P into fructose-6-phosphate, which is a substrate in the glycolytic pathway. NanR is the repressor that regulates the activity of these genes. The genes for the first three enzymes (nanA, nanK and nanE) are usually found clustered together forming the nan gene cluster [11]. However, there have been a few exceptions such as Citrobacter freundii and Edwardsiella tarda where the nanE gene is located in a separate region from the rest of the operon [14]. The genes encoding NagA and NagB are located adjacent to each other, but most often not necessarily in the vicinity of the nan gene cluster on the bacterial genomes [12–14]. The genes within the cluster show independent evolutionary histories. Several horizontal gene transfer events are noted in the phylogenetic trees for all three proteins. Most significantly, the NanA protein shows several possible horizontal gene transfer events between the Eukaryotes and Prokaryotes. Two examples are the clustering of Trichomonas vaginalis NanA protein sequences with Pasteurellaceae, and the Bacteroides, Yersinia and Vibrio branching closely with the Eukarya kingdom [11, 14].

The uptake of sialic acid into bacterial cells has been associated with a number of virulence factors. The bacterial glycolipid capsule is an example of molecular mimicry of the host as it aids the organism to overcome the immune responses of the host. Neonatal meningitic E. coli K1 uses sialic acid to decorate the cell surface, and Cronobacter does produce capsular material, especially when grown on milk [15]. Sun et al. [16] reported the O-PS gene for C. turicensis G3882 included N-acetylneuraminic acid synthetase and CMP-N-acetylneuraminic acid synthetase. However the accurate identification of this strain is uncertain as the 16S rRNA gene sequence (Accession no. HQ880409) does not match other strains of C. turicensis.

As previously reported for the genome sequenced strain C. sakazakii BAA-894, the gene cluster encoding for a putative sugar isomerase (YhcH) and the nanKTAR genes (encoding N-acetylneuraminate and N-acetylmannosamine degradation) are located at ESA_03609-13 [7, 8]. However, the remaining nan genes for sialic acid metabolism have not been described in detail, and these previous bioinformatic studies did not identify any candidate sialidase (NanH) genes [8].

C. sakazakii is associated with infections of low birth weight neonates, and this could be linked to a number of opportunities for the organism to utilize sialic acid for growth. Human milk contains sialic acid in the form of sialyloligosaccharides which are highest in colostrum and decreases by nearly 80% over the following 3 months after birth [9, 17]. Human milk from mothers of preterm infants contains 13–23% more sialic acid than milk from mothers of full-term infants at 3 of the 4 lactation stages. Similarly infant formulas contain sialic acid which may be bound to glycoproteins. Although the nutritional significance of sialic acid is unknown, it is plausible that it contributes to sialic acid accumulation in the brain. Breast milk also contains a variety of sialoglycans; secretory IgA, lactoferrin, and oligosaccharides. Human milk oligosaccharides are poorly digested and are substrates for intestinal bacterial metabolism, promoting bacterial growth in the intestinal tract. As a site of bacterial attachment, the intestinal microvilli of neonates have increased sialic acid and N-acetylglucosamine residues whereas adults have increased mannose, glucose, and fucose residues [17, 18]. Finally, the brain is the major site of sialic acid display in the form of gangliosides (sialylated glycolipids) and sialic acid may have a role in the structural and functional establishment of synaptic pathways [9]. It is possible that, like Neisseria meningitidis, Streptococcus pneumoniae and Haemophilus influenzae which cause bacterial meningitis in children <5 years, that C. sakazakii has a developmental dependence on access to the central nervous system.

Given the common clinical association of C. sakazakii infection of neonates with necrotising enterocolitis and brain damage, an improved understanding of sialic acid utilisation by the organism was warranted. This paper describes the plausible link between the recent evolution of sialic acid metabolism by C. sakazakii and its pathogenicity as well as re-investigates the possible presence of sialidase activity.

C. sakazakii is estimated to have branched from the rest of the Cronobacter members approximately 15–23 MYA [6], and is the Cronobacter species most frequently isolated from clinical infections of neonates. The absence of the core sialic acid utilization genes in the Cronobacter genus except for C. sakazakii, suggests their acquisition was by horizontal gene transfer instead of gene loss by the rest of the genus. The metabolism of sialic acid could have a role in the organisms’ virulence. The evolution of C. sakazakii post-dates key eras in the evolution of sialic acid biosynthetic pathway which occurred near the divergence of Coelomata (protostomes and deuterostomes ca.500 MYA), and milk producing mammals (Jurassic Era ca.200-145 MYA). Therefore the presence of sialic acid is a relatively recent evolution, and high levels are accessible to bacteria in milk, the intestines, and the brain.

The C. sakazakii genomes possess the genes required for the uptake and catabolism of sialic acid; Table 1. Although the nanAKTER genes are usually clustered together, in C. sakazakii nanE is located at a different region (ESA_00529) from the others (ESA_03610-12). This split has also been reported for Ed. tarda and Cit. freundii[14]. This possibly reflects a separate evolutionary lineage for the gene. All the C. sakazakii genomes encode for the NanC outer membrane porin and NanT inner membrane transporter protein. Hence, all C. sakazakii strains can potentially transport the exogenous sialic acid into the cytoplasm of their cells. Interestingly, all the Cronobacter species genomes also exhibited the presence of the genes (siaPQM) encoding for the TRAP transporter; Table 1. However laboratory growth studies confirmed that C. sakazakii was the only Cronobacter species which could grow on sialic acid as a carbon source and indicates that NanT is the active transporter in the process; Figure 5a. The acquisition of genes encoding for the utilization of exogenous sialic acid may have a major role in C. sakazakii colonisation of the human intestinal tract (via mucins) and the use of sialic acid in breast milk or infant formula as a nutrient source [9, 18].

The demonstration of C. sakazakii growth on the ganglioside GM1 (Figure 5b) indicated the possible possession of a sialidase enzyme, which despite previous sequence searches for nanH (the gene encoding for sialidase) had not been reported. Since the homology between sialidases is very low (<30%) [13], a more detailed search for the sialidase RIP and Asp-box motifs was undertaken. However no candidate genes were found with both RIP and Asp-box motifs. Nevertheless the growth on GM1 indicated a ganglioside degradative ability by C. sakazakii. GM1 is composed of galactose, N-acetyl-galactose, glucose as well as sialic acid units linked by β 1–3 and β 1–4 linkages, and these are joined to a steroid. Therefore it is proposed that GM1 is degraded to metabolisable sugar units by the activity of various β-galactosidases (ESA_01827, ESA_02977 & ESA_03417), β-acetyl-hexosaminidases (ESA_02237 and ESA_02655), esterases (ESA_00377 & ESA_00776) and lipases (ESA_02127 & ESA_02202), and therefore able to grow on this substrate.

The %GC content of the nanC gene was found to be 47%, considerably less than the 56% GC content of the overall Cronobacter genome. Slight aberrations were also seen in the %GC content values of the nanK and nanE genes, an observation previously noted in an evolutionary study of the nan clusters of other Enterobacteriaceae members such as Salmonella enterica, E. coli and Yersinia spp. [11]. These observations suggest a strong possibility of horizontal transfer events having influenced separate acquisitions for the nanAKT cluster in C. sakazakii, as well as the nanC and nanE clusters in the whole genus. Hence, it is possible that the nan clusters could have evolved in a mosaic pattern in this bacterial genus. A phylogenetic analysis was also conducted using the protein sequences encoded by the sialic acid utilization genes in C. sakazakii (Figures 1, 2, 3 and 4). This revealed that the nanATKR genes have evolved as a lineage in C. sakazakii, and independent of other closely related pathogens of the Enterobacteriaceae family. No close homology (50% cut-off) was found between the NanC protein sequences and those from most other members of the Enterobacteriaceae. This indicates the acquisition of the associated gene in C. sakazakii was from a more distantly related organism, currently unidentified.

It is also notable that the nanATK cluster in C. sakazakii is located adjacent to a stringent starvation gene homologue (sspA, ESA_03615) and therefore expression of this gene cluster could be responsive to environmental nutrient levels; Table 1. The presence of other related genes such as nagA and nagB, leading to the synthesis of fructose-6-phosphate, support the role of sialic acid as a carbon source in C. sakazakii. Whereas the absence of genes such as neuS, neuO and neuD in the genomes of the Cronobacter genus indicates that the organism does not decorate its cell surface with sialic acid, unlike E. coli K1 and N. meningitidis[10, 14].

Three rich mammalian sources of sialic acid for pathogenic or commensal bacteria are in the gastrointestinal tract, the brain and in human milk. These three locations of sialic acid correlate with the clinical presentations of C. sakazakii infections with respect to necrotizing enterocolitis and the intensive brain damage during neonatal meningitis. This is especially pertinent since most of the neonatal infections in C. sakazakii have been reported at a very early stage of growth with half in the first week and three quarters during the first month of age [1, 2, 4]. Apart from these mammalian sources of sialic acid, another feature with the organism is that sialic acid is often added to infant formula, a recognised source of neonatal Cronobacter infections [4, 6]. The unique utilization of sialic acid by C. sakazakii indicates a plausible adaptation of the organism to host. The source of the nan genes is not fully understood, but appears to be by multiple horizontal gene transfers, the sources of which are uncertain to date.