Microbiological survey and genomic analysis of Cronobacter sakazakii strains isolated from US households and retail foods

ABSTRACT Cronobacter species are opportunistic pathogens that are capable of causing morbidity and mortality, particularly in infants. Although the transmission dynamics involved in Cronobacter infections remain largely unknown, contaminated powdered infant formula (PIF) has been linked to 30% of Cronobacter sakazakii cases involving invasive illness in infants. As several lines of evidence have implicated the domestic environment in PIF contamination, we undertook a microbiological survey of homes (N = 263) across the US. Cronobacter spp. and C. sakazakii were isolated from 36.1% and 24.7% of US homes, respectively, with higher recovery rates observed for floor and kitchen surfaces. Multi-locus sequence typing indicated that the dominant strain was C. sakazakii ST4, the sequence type most commonly associated with neonatal meningitis. For comparison purposes, retail foods (N = 4,009) were also surveyed, with the highest contamination frequencies (10.1%–26.3%) seen for nut products, seeds, and grains/baked goods/flours. The sequence type profile of isolates recovered from homes mirrored that of isolates recovered from retail foods, with increased representation of ST1, ST4, ST13, ST17, and ST40. Analysis of 386 whole genomic sequences revealed significant diversity. Redundancies were only observed for isolates recovered from within the same domicile, and there were no identical matches with sequences archived at the NCBI pathogen database. Genes coding for putative virulence and antibiotic resistance factors did not segregate with clinically significant sequence types. Collectively, these findings support the possibility that contamination events occurring within the home should not be overlooked as a contributor to community-onset Cronobacter infections. IMPORTANCE Cronobacter sakazakii is an opportunistic pathogen that can cause significant morbidity and mortality in neonates. Its transmission dynamics are poorly understood, though powered infant formula (PIF) is thought to be the major transmission vehicle. How the PIF becomes contaminated remains unknown. Our survey shows that roughly 1/4 of US homes are contaminated with Cronobacter sakazakii, particularly in the kitchen setting. Our analyses suggest that the domestic environment may contribute to contamination of PIF and provides insights into mitigating the risk of transmission.

Infections occur in all age groups and can involve multiple anatomical sites, including the urinary tract, skin, respiratory tract, blood, gastrointestinal tract, central nervous system, conjunctiva, and bone (5)(6)(7)(8)(9)(10)(11).In adults, infections are generally non-life-threat ening and typically occur in the context of underlying health issues or advanced age (9,11).In contrast, infections in infants less than 2 months of age can manifest as necrotiz ing enterocolitis (NEC), sepsis, and meningitis, with fatality estimated at 40% (8,(10)(11)(12)(13)(14).Although the prevalence of Cronobacter infections in infants is relatively low (10,11), the potential for severe illness as well as lifelong sequelae (15) has earned Cronobacter spp. the status of most costly foodborne pathogen in the US, with an estimated per case healthcare cost of 1 million USD (16).
While the transmission dynamics of Cronobacter infections are largely unknown, a significant vehicle for transmission in infants is thought to be powdered infant formula (PIF) (10,(17)(18)(19)(20)(21)(22).As such, testing of available PIF containers is regularly undertaken in the course of source investigations involving infants.Detection of the organism in sealed, lot-matched containers denotes contamination originating during the manufacturing process and is referred to as intrinsic product contamination.Detection of the organism in previously opened PIF containers can signify extrinsic product contamination, though it does not eliminate the possibility of intrinsic contamination.Outside the US, cases of invasive Cronobacter illnesses in infants have been linked to both intrinsically contamina ted PIF and previously opened containers of PIF.Since 2004, the majority of non-US cases (85%) have occurred in the hospital setting, primarily as outbreak-associated, epidemiologically linked cases (10).Within the US, the pattern is strikingly different; since 2004, no cases have been linked to intrinsically contaminated PIF (10).Furthermore, the majority of US cases (78%) have primarily occurred in the community setting as sporadic, epidemiologically unlinked cases (10,20).
To better define the transmission dynamics, the US Center for Disease Control and Prevention (CDC) recently undertook a retrospective analysis of cases involving invasive Cronobacter illness in infants.From their analyses of 71 source investigations, contaminated opened PIF containers were identified in 30% (21 of 71) of cases, and contaminated surfaces and/or foods from within the home were identified in 44% (31 of 71) of cases (10).Transmission vehicles identified by these investigations included bottled PIF reconstitution water (10,13), kitchen sink surfaces (10), vacuum cleaner contents (23), breast pump components (10,24), pacifiers (10), bottle components (10), and food blenders (10).Additionally, 10% of source-investigated cases involved infants fed exclusively with expressed breastmilk.The clinical isolates in these instances matched environmental isolates recovered from breast milk samples, breast pump kit components, the kitchen sink, and the drying area next to the sink (10,(24)(25)(26).Although it is possible that some of these cases may involve intrinsic PIF contamination, these findings point at the possibility that PIF contamination in the US is occurring extrinsically, in either the hospital or home setting.
A number of surveillance studies have assessed the domestic environment as a possible source of infection.Within the US, Cronobacter spp.were recovered from 78.5% (51 of 65) of homes sampled in Georgia (27), while a separate study identified C. sakazakii in 26.9% (21 of 78) of homes in Tennessee (28).Sites within the domestic environment from which Cronobacter spp.have been reproducibly recovered include vacuum cleaner contents, floors, entryways, refrigerators, sinks, kitchen cleaning cloths, and kitchen sponges (27)(28)(29)(30)(31)(32)(33).Importantly, the conspicuous presence of Cronobacter spp., particularly C. sakazakii, in the domestic environment establishes clear opportuni ties for contamination of food and infant formula, potentiating foodborne illnesses.Therefore, to expand on these earlier, smaller studies, we undertook a large-scale survey of retail foods and households across the US.Isolates collected during the course of this survey underwent WGS analysis and sequence typing to assess the genetic diversity and potential clinical significance.The results of this study extend our understanding of Cronobacter transmission dynamics and support the use of mitigation strategies to reduce the risk of infection in the domestic environment.Importantly, the results of this study should be considered both narrowly and broadly, as caregiver awareness of the possible transmission risks arising from the domestic environment may play a crucial role in ensuring both the safety and security of infant formula.In this regard, in 2022, a major PIF manufacturer issued a voluntary recall based on the presumption of intrinsic contamination following four sporadic cases of community-onset invasive Cronobacter illness in neonates occurring over the period of September 2021 to February 2022 (34).However, subsequent product testing was unable to provide evidence for intrinsic contamination of PIF as the source of these infant illnesses (34).The recall led to nationwide shortages that limited formula availability for infants generally and for infants with special nutritional requirements over a period of several months (35).

Domestic environment samples
For the period of June 2022 to June 2023, households throughout the US were mailed sampling kits and overnight return packaging with re-freezable ice packs.The sam pling kits included MegaSampler Sponges with DE Neutralizing Broth (Weber Scien tific, Hamilton, NJ) and sterile sampling bags (Weber Scientific, Hamilton, NJ).Study participants were instructed to surface test bathroom areas (floors, sink, and counter tops), kitchen areas (floors, sink, countertops, and sponges), footwear, car mats, entryway floors, and refrigerator shelves using the sponges and to collect floor sweepings or vacuum cleaner contents into the sampling bag.The manufacturer's instructions for sampling were supplied with the kits.In some instances, study participants collected multiple samples for the same sample category or missed a sample category.In total, 2,810 environmental samples from 263 homes from 36 different states across the continental US were collected and analyzed.

Food samples
For comparison purposes, a total of 4,009 retail food items purchased across the US were analyzed throughout the period of March 2022 to August 2023.Each food sample tested was from a distinct lot.Food items included grains, milled flours, baked goods, confections, seeds and beans, raw beef, raw chicken, sprouts, nuts, nut butters, candies, snack foods, chocolate, pet food, trail mix, nutritional supplements, botanical products, noodles, pasta, fresh fruits, fresh vegetables, and spices/seasonings.Product descriptions are provided in Table 2.

Sample enrichment
Food samples (either 25 or 375 g) were enriched by the addition of M1 (GN) medium (either 75 or 1,125 mL, respectively) intended for gram-negative bacteria (Microbiologi que, Seattle, WA) as per the manufacturer's instructions and grown 26 hours t at 42°C.Swabs were placed in 100 mL of 2% peptone-buffered water and grown for 26 hours at 35°C.

PCR screening
Enrichments were screened for the presence of Cronobacter spp.using the Microbiolo gique Cronobacter and Salmonella Screening PCR Kit (Microbiologique, Seattle, WA) as per the kit instructions.Enrichment broths from presumptive positive PCR samples were then streaked onto RF medium and Harlequin medium plates to obtain isolated colonies.Confirmation was performed by colony PCR using the Cronobacter and Salmonella Confirmation PCR Kit (Microbiologique, Seattle, WA) as per the kit instructions.

Culture confirmation
All presumptive positive enrichment cultures were subjected to culture confirmation using the FDA Bacteriological Analytical Manual (BAM) method (36).Colonies from all culture-confirmed samples were subjected to whole-genome sequencing analysis.

Whole-genome sequencing and MLST sequence typing
Genomic DNA samples from 415 confirmed isolates were submitted for WGS.Genomic libraries were prepared using the Nextera XT DNA Sample Prep Kit (Illumina, CA), and genome sequencing was performed using the Illumina MiSeq desktop sequencer (Illumina, CA) loaded with a paired-end 2 × 250 cycle MiSeq reagent kit version 3. The raw sequencing reads were quality filtered using the default settings of the "fastp" version 0.20.0 software (37).After sequencing, FastQC v0.11.7 (38) was utilized to perform quality control checks on the raw sequence data.The filtered reads were then assembled using the SPAdes assembler (https://github.com/ablab/spades/tree/v3.15.4).The quality of the genome assemblies was evaluated using Quast version 4.6.3(39).Any data with metrics below 30 for R1 and R2, coverage less than 20×, an assembly sequenc ing length lower than 4 million base pairs or higher than 5 million base pairs, and more than 500 contigs were excluded from the analysis.The resulting de novo assem blies were then used for preliminary identification and cross comparisons using the MinHash algorithm implemented in an in-house mash database built using the RefSeq 210 database (40,41).From the whole-genome assemblies, seven-locus multi-locus sequence typing (MLST) was performed using the SRST2 version 0.2.0 software (https://genomemedicine.biomedcentral.com/articles/10.1186/s13073-014-0090-6).The assembled files were submitted to the TYGS server for genus-/species-level verification (42).This cross-referencing comparison confirmed the speciation of the Cronobacter species from the de novo assemblies.Sequence types were confirmed using MLST tool available at https://pubmlst.org/.The sequences were then compared against the NCBI Pathogen Detection Database (43).The SNP distances within the SNP cluster groups were compared using the CFSAN SNP pipeline (44).Sequences that satisfied the QC thresholds described above were deposited at the NCBI Pathogen Detection Database.The Biosample IDs for the assembled whole-genome sequences deposited at NCBI are listed in the Supplemental Material section.

Virulence analysis
Whole-genome sequences of 386 isolates were uploaded into the Bacterial and Viral Bioinformatics Resource Center (BV-BRC) (https://www.bv-brc.org)(45) and interrogated for factors associated with virulence, including secretion systems, adhesion factors, infection/invasion/host resistance factors, toxins, as well as antibiotic resistance enzymes and efflux pumps to ascertain the total number of genomes that contained each factor.In addition, factors associated with infection/invasion/host resistance were further analyzed for percent identity against reference sequences obtained from NCBI to compare genetic distinctions among the five most common sequence types identi fied in this study, namely ST1, ST4, ST13, ST17, and ST40.Sequences from NCBI included CsakCS931_RS20575 (Cpa), CsakCS931_RS08525 (EfeO), CsakCS931_RS09290 (Fha), CsakCS931_RS17840 (FkpA), CsakCS931_RS18715 (Hfq), CsakCS931_RS18895 (HlyIII), AFK66_RS20815 (IbeB), CsakCS931_RS21050 (IucA), CsakCS931_RS14245 (NanR), CsakCS931_RS15355 (SodA), and CsakCS931_RS03030 (WcaD).In all but one instance, the sequences used in the queries were specific for C. sakazakii.A novel allele was inferred on the basis of ≥0.002% difference in percent identity, and the numbering of alleles was arbitrarily assigned.

Phylogenetic analysis
Cronobacter MLST allele profiles were derived from https://pubmlst.org/bigsdb?db=pubmlst_cronobacter_seqdef&page=downloadAlleles&scheme_id=1&render=1.A blastn database of the Cronobacter MLST alleles was created on a local server using makeblastdb (https://www.ncbi.nlm.nih.gov/books/NBK569841/).MLST sequences were extracted from de novo assemblies using a local blastn database and a Python in-house script.The sequences were then concatenated to each sample, as performed by Rachlin et al. (46).Alignment was performed using MAFFT (47), followed by the creation of a phylogenetic tree using maximum likelihood estimation performed using Maga 11 software.The tree was visualized using iTOL version 6 software (https://itol.embl.de/).

Occurrence of Cronobacter spp. and Cronobacter sakazakii in US homes
To survey the domestic environment, samples were collected from a total of 263 homes located in 36 continental US states for the period of June 2022 to June 2023.On average, roughly 11 different samples were collected from each home.Samples were processed as described in the Materials and Methods section.PCR presumptive positive samples (N = 516, 382 of which were C. sakazakii and 134 of which were non-sakazakii Cronobacter spp.) underwent BAM method culture confirmation.Using this approach, 88.7% (339 of 38) of the C. sakazakii presumptive positive samples were confirmed positive.Corre spondingly, 67.2% (90 of 134) of the non-sakazakii Cronobacter spp.presumptive positive isolates were confirmed positive and subsequently speciated through WGS analysis.Of the 263 US homes tested, 36.1% were positive for Cronobacter spp., and 24.7% were positive for C. sakazakii.As for the other species, 6.5% of homes were positive for C. turicensis, 3.8% were positive for C. malonaticus, 3.4% were positive for C. dublinensis, 2.3% for C. muytjensii, and 0.8% were positive for C. universalis.Furthermore, 5.3% of homes were positive for two different species.For the 65 US homes that were positive for C. sakazakii, the following observations were made: 39 homes were positive for a single isolate, 6 homes were positive for multiple isolates (2-4) belonging to the same sequence type, 11 homes were positive for multiple sequence types (2-7), and 9 were positive for two or more different species including C. sakazakii (data not shown).In instances where multiple strains of the same ST were isolated from the same household, the number of SNPs (determined by WGS comparisons) ranged from 0 to 91.A total of 21 homes (8% of total) were positive for ST4, the sequence type associated with neonatal meningitis (13).Correspondingly, 28 homes were positive for one or more non-sakazakii species.

Genetic composition and diversity of Cronobacter spp. isolates
To ascertain a possible relationship between C. sakazakii isolates obtained from the domestic environmental samples and isolates recovered from retail foods, seven-locus multi-locus sequence typing was performed (Fig. 1) using the whole genomic sequences.For US homes (N = 248 C. sakazakii isolates total), 12 isolates (5%) did not match to previously assigned STs, and the remaining 236 isolates (95%) conformed to 51 different sequence types.The sequence type most commonly isolated from US homes was ST4, followed by ST40, ST17, ST13, ST1, and ST64.Correspondingly, for retail foods (N = 203 C. sakazakii isolates total), 12 isolates (6%) did not match to a previously defined ST, and the remaining 192 isolates (94%) matched to 45 different STs, with ST40 identified as the most common sequence type, followed by ST4, ST13, ST1, ST410, and ST1.
In comparison, the most commonly documented C. sakazakii ST at the MLST database (https://pubmlst.org/bigsdb?db=pubmlst_cronobacter_isolates,2,322 records) is ST4, followed by ST1, ST8, ST64, ST13, and ST40 (data not shown).Although not entirely overlapping, four of the top six sequence types are common to samples collected from homes and retail foods as well as the MLST database.Importantly, the most clinically significant C. sakazakii STs reported in the literature are ST1, ST4, ST8, and ST12 (48), where ST4 is the pathovar most commonly associated with meningitis in neonates (12,13).Aside from ST4, no other STs belonging to clonal complex 4 were identified in this survey.
Analysis of non-sakazakii species isolated from foods and domestic environmental samples (N = 97 isolates total) identified 50 isolates (52%) which were unique and did not match to known STs with the remaining 47 isolates (48%) matching to 24 established STs (Fig. 2).The most frequent non-sakazakii ST isolated from retail foods was ST167 C. dublinensis (three isolates) and, from homes, ST7 C. malonaticus (six isolates).Of note, ST7 is the most common sequence type isolated from adults with Cronobacter infections (49).a Culture-confirmed positive samples are expressed as the number of positive sites and parenthetically as the percentage of total samples in each category.
WGS comparisons made using 386 high-quality sequences obtained from isolates of both foods and homes revealed significant genetic diversity.With the exception of isolates collected from the same home, the remaining genomes (N = 269) were entirely unique and 118 clustered into 47 groups.None of the sequences were identical (SNPs <5) to sequences previously deposited at the NCBI pathogen browser.Several similarities were found with previously banked sequences.An isolate collected from a US home showed similarity (differing by 29 SNPs) to an isolate from an environmental sponge collected from a food manufacturing plant roughly 6 months apart in a separate state.This isolate was identified as ST12, a C. sakazakii pathovar associated with NEC in infants (50).A second isolate from a US home clustered with a 2008 food isolate, differing by 43 SNPs.Lastly, a third isolate obtained from a US home identified as ST4 clustered with a number of clinical isolates collected from 2003 to 2016 as well as environmental samples, including isolates from PIF (2011), breast milk (2016), and a breast pump (2011).In this instance, the genomic sequence differed by 50 or more SNPs.
Genetic diversity was further analyzed by assessing the population structure of 386 high-quality Cronobacter spp.genomic sequences based on assessment of MLST loci sequences (Fig. 3).While isolates clustered by ST, they did not robustly cluster by sample class (either retail foods or residences) nor by specific categories or foods or domestic  isolation sites.Of note, sizable recovery of C. sakazakii ST4 was observed for the pet food samples (1.4%, 6 isolated from 439 pet food samples), implying a novel transmis sion vehicle in the domestic environment.Tree nuts and peanuts also showed notable recovery of C. sakazakii ST4 (1.0%, 3 isolates out of 288 samples).Within the domestic setting, ST4 was most commonly recovered from kitchen surfaces (17 isolates), followed by floors (not including kitchen floors, 11 isolates).

Virulence features and antibiotic resistance markers of Cronobacter spp. isolates
Potential virulence and clinical significance were assessed in 386 sequenced isolates using the web-based bioinformatics tool BV-BRC (45).Included in the analysis were markers of secretion system components, host invasion/resistance factors, toxins, adhesion factors, invasion and host resistance factors, and antibiotic resistance enzymes/ efflux pumps.The occurrence of each marker in the 386 genomes is indicated in Table 3; markers that were not represented in the sequences were excluded in the tabulation.
Of the six different secretion systems, T1SS, T3SS, and T6SS were identified in most or all of the genomes.Evidence of T2SS was not identified in any of the genomes.T1SS was absent in some C. dublinensis isolates.T4SS, identified in only 27

DISCUSSION
Cronobacter spp.exhibit a wide-ranging ecological distribution, a property that can be attributed to their ability to withstand adverse conditions such as acid stress, temperature stress, prolonged desiccation, exposure to disinfectants, and osmotic stress (51)(52)(53)(54)(55)(56)(57).Recovery of Cronobacter spp.has been documented for a wide variety of environments and foods (55)(56)(57)(58)(59)(60)(61).With respect to foods, Cronobacter spp.have been most frequently recovered from plant-based foods, such as vegetables, legumes, fruits, sprouts, cereal products, flour-based confections, herbs, nuts, teas, and spices (29,30,(58)(59)(60)(61)(62)(63)(64)(65)(66)(67)(68)(69)(70)(71).This current survey of US retail foods similarly identified substantial contamina tion of plant-based products.The most frequently contaminated categories included grains/baked goods/flours (26.3% of samples were positive).However, recoveries from spices/seasonings, fruits, raw meat, and vegetables were negligible.This discrepancy may reflect distinctions in HACCP practices in different countries, such as the use of irradiation in the US or changes in quality control standards over time.Indeed, with the exception of one study examining foods from 44 countries (70), the remaining food studies have characterized Cronobacter burden in retail foods from outside North America, including from China, Korea, the Czech Republic, Slovakia, Jordan, India, Brazil, Ireland, and Switzerland.An interesting outcome of the retail food analysis was that the ST distribution profile of C. sakazakii isolates essentially mirrored that of isolates obtained from US homes, with biased frequencies noted for ST1, ST4, ST13, ST17, and ST40 in both sample categories.The occurrence of pathovars ST1 and ST4 is significant (48).The significance of the prevalence of ST4 is most notable due to its association with neonatal meningitis (13).Indeed, these distributions were similar to the ST distribution at the MLST database.These similarities may simply reflect the natural ecological distribution of different sequence types, though it is tempting to speculate that this apparent similarity may reflect a causal relationship between retail foods and introduction of organisms into the home setting.Similarly, the prevalence of Cronobacter spp.isolated from entry hallways and floors suggests that foot traffic may also serve as a vehicle for entry into the domestic environment as well.
Although not addressed by this study, Cronobacter spp.have also been isolated from powdered infant formulas, infant cereals, and powdered milk (54,61,72,73).Indeed, PIF is recognized around the world as a primary vehicle for transmission to infants  (10,(18)(19)(20)(21)(22)74).The connection with PIF can be attributed to several clustered cases and hospital-associated outbreaks that occurred in the Netherlands (1977-1981) (74), Tennessee, USA (1988) (18), France (1994) (22), Belgium (1998) (19), and the US across multiple states (2011) (13).However, a patient-linked organism was only recovered from unopened cans of PIF in the Belgium outbreak; in the remainder, recovery of patientlinked organism was only made from previously opened or previously prepared PIF (19).PIF, which is not a sterile product, is inherently prone to microbiological outgrowth if not handled according to the manufacturer's guidelines (21).Several outbreaks occurring in neonatal intensive care units have been attributed to inadequate hygienic preparation and temperature control of reconstituted PIF (20,22) Of note, US source investigations into cases of invasive Cronobacter infection in infants have been unable to link infections to unopened (intrinsically) contaminated PIF containers (10).Indeed, the only reported instance in the US of linkage with any type of powdered formula occurred in a Tennessee neonatal intensive care unit (NICU) in 2001.This case linked a clinical isolate to unopened containers of Portagen, a formula type intended for children and adults but not infants and therefore not subject to the same microbiological testing criteria and GMP requirements that were implemented by the FDA in 2014 (20).This case was the first and last instance of intrinsically contaminated powdered formula linked to infant infections in the US, though notably, it did not involve PIF.Accordingly, it remains highly plausible that the contamination events causing US infections are occurring in the hospital or community setting.Supporting the role for the environment in the transmission to infants is the fact that most NICUs in the US no longer use PIF due to a 2002 recommendation by the FDA to avoid feeding it in neonatology units.Yet since 2004, hospital-acquired infections have represented 23%-33% of invasive illness cases in infants [ (10), unpublished review of CDC cases 2020-2023].Furthermore, clustered, community-onset cases have never been epidemiologi cally linked to one another, implying a unique origin in each instance.
A number of studies have been undertaken to determine the occurrence of Cronobacter spp. in the domestic environment.Source investigation findings have linked 44% (31 of 71) of cases involving invasive illness in infants with environmental samples collected from homes.Furthermore, surveillance studies have documented Cronobacter spp.contamination in 31% (5 of 16 homes) of Dutch households (32), C. sakazakii contamination in 24.7% (21 of 85 homes) of Czech households (29), and C. sakazakii contamination in 26.9% (21 of 78 homes) of Tennessee households (28).A 2016 surveillance study of domestic households in Georgia showed that 78.5% (51 of 65) of homes tested positive for Cronobacter spp.( 27).This outcome is considerably higher than the outcomes obtained in the current study or abovementioned studies, and may reflect methodological differences.Within the domestic environment, Cronobacter spp.have been reproducibly isolated from vacuum cleaner contents, threshold floors, refrigerators, sinks, kitchen floors, cleaning cloths, and kitchen sponges (27)(28)(29)(30)(31)(32)(33), essentially mirroring sites of higher contamination identified by the current study.
Assessment of pathogenic potential indicated that many of the isolates shared genetic features that facilitate hematogenous spread as well as virulence and pathoge nicity by uropathogenic and enteropathogenic bacteria.These findings help explain the ability of Cronobacter to cause opportunistic infections of the blood, urinary, and gastro-intestinal systems.Similarly, the genomes contained a high proportion of antibiotic resistance genes that could potentially enable the organisms to evade an array of different antimicrobial and antiseptic classes.The identification of 14 antibiotic resistance marker genes, including those that rely on target alteration and efflux systems, is similar to the 19 that have been previously reported for 5 Cronobacter spp.isolates (75).A key distinction in this study was the occurrence of a marker for resistance to aminoglycosides in one of the isolates.While the identification of these factors at the genetic level does not guarantee their functionality, the widespread occurrence of many of these factors suggests that the Cronobacter spp.found in homes and on foods pose certain potential to cause disease.Furthermore, nested assessment of genetic variation in putative virulence factors revealed a pattern of inferred alleles, where alleles for SOD, HlyIII, EfeO, IucA, and Cpa appeared to segregate with pathovars ST1 and ST4, as opposed to ST13, ST17, and ST40.Of note, HlyIII is thought to enable persistence in blood through α-or beta-hemolytic activity.EfeO is associated with persistence and dissemi nation in the blood, as well as invasion of the CNS.Cpa enables persistence in blood through inactivation of α2-antiplasmin and complement components.In the absence of functional studies, this segregation of inferred VF alleles does not necessarily denote virulence causality.An alternate explanation for the apparent VF allelic segregation may simply involve clonality and, accordingly, warrants further detailed analyses.
This study contributes additional understanding obtained through detailed genetic analysis.In particular, the profound genetic diversity identified by this study helps explain the sporadic, epidemiologically unlinked nature of infections in the US.Furthermore, the dominance of pathovars ST1 and ST4 in retail foods and domestic samples provides insight into a potential exposure route to clinically significant strains in the home.
Additionally, these findings extend our understanding of Cronobacter transmission dynamics via the following inferences.First, introduction of Cronobacter into the home appears to relate to foot traffic, as supported by the contamination rates of entryway floors, footwear, and sweeping/vacuum cleaner dust reported in our study (Table 1).This risk could be mitigated by restricting use of outdoor footwear within the home and enhancing floor cleaning measures such as the use of approved disinfectants and vacuum cleaners fitted with HEPA filters.Second, the kitchen poses significant risk for food contamination, and the adoption of improved hygiene measures in the kitchen, including the use of approved disinfectants, may mitigate this risk.Third, the possibility that certain retail foods, including pet food, may be introducing Cronobacter into the home, implying the need to handle high-risk foods with more caution.Fourth, these data suggest that fecal-oral transmission is likely insignificant based on the paucity of isolates recovered from bathroom surfaces.Importantly, while these findings do not preclude the need to maintain stringent manufacturing practices and rigorous product testing, they do support the need to improve hygienic measures in the home setting.

FIG 1
FIG 1 Distribution of Cronobacter sakazakii MLST obtained from retail foods (blue) and US residences (orange).

FIG 2 FIG 3
FIG 2 Distribution of non-Cronobacter sakazakii MLST obtained from retail foods (blue) and US residences (orange).

TABLE 1
Recovery of Cronobacter spp.from different sites within US homes (N = 263)

TABLE 2
Recovery of Cronobacter spp.from US retail foods a Culture-confirmed positive samples are expressed as the number of positive sites and parenthetically as the percentage of total samples in each category.b Samples consisting mostly of sesame, chia, buckwheat, wheat berries, alfalfa, sunflowers, and various beans.c Samples consisting mostly of milk or dark chocolate candies, mints, chocolate-coated berries, and chocolate-or yogurt-coated pretzels.d Samples consisting mostly of ground beef (various fat content), chuck roast, sirloin, and heart.e Samples consisting mostly of ground chicken, chicken wings, and chicken-based sausage.

TABLE 3
Frequency of putative virulence factors and antibiotic resistance factors present in 386 Cronobacter spp.isolates

TABLE 3
Frequency of putative virulence factors and antibiotic resistance factors present in 386 Cronobacter spp.isolates (Continued) 1 a number of alleles segregated with pathovar sequence types.In this regard, all ST4 isolates shared unique alleles for SOD, HlyIII, EfeO, IucA, and Cpa.

TABLE 4
Nested analysis of putative virulence factors in Cronobacter sakazakii ST1, ST4, ST13, ST17, and ST40 isolates Inferred alleles indicated by boldface type segregate with pathovars ST1 and/or ST4.N = number of genomes for that sequence type.Numbers in parentheses refer to the number of occurrences of an inferred allele for a given sequence type.Alleles marked with an asterisk (*) indicate duplication of a gene within one or more of the genomes. a . The earliest microbiological survey of PIF published in 1988 indicated that 14% (20 of 141) of samples collected from 35 countries were contaminated with Cronobacter spp.(Enterobacter sakazakii) (17).A field survey conducted by the FDA in 2003 found Cronobacter spp. in 22.7% (5 of 22) finished product samples (unpublished findings).A more recent 2017 survey conducted in China, which assessed over 6,000 infant formula samples, reported a C. sakazakii contamination rate of <0.5% (73).