1191
Journal o f Food Protection, Vol. 78, No. 6, 2015, Pages 1191-1196 doi: 10.4315/0362-028X.JFP-14-563 Copyright © , International Association for Food Protection
Research Note
Reevaluation of a Suspected Cronobacter sakazakii Outbreak in Mexico EMILY E. JACKSON, 1 JULIO PARRA FLORES,2 EDUARDO FERNANDEZ-ESCARTIN,3 a n d STEPHEN J. FORSYTHE1* 1School o f Science and Technology, Nottingham Trent University, Clifton Lane, Nottingham, NG11 8NS, UK; 2Departamento de Nutrition y Salud Publica, Universidad del Bio-Bio, Chilian, Nuble, 3780000, Chilian, Chile; and 3Departamento de Investigation y Posgrado en Alimentos, Facultad de Quimica, Universidad Autonoma de Queretaro, Queretaro, 76010. Mexico MS 14-563: Received 26 November 2014/Accepted 30 January 2015
A BSTR A C T In 2010, two infants became ill at a hospital in Mexico. Subsequently, a range of clinical, environmental, and powdered and rehydrated infant formula isolates were identified by using a combination of phenotyping and PCR probes. The strains were clustered according to pulsed-field gel electrophoresis. The causative agent was reported as Cronobacter sakazakii, with powdered infant formula (PIF) identified as the likely source of the infections. This new study further characterized the isolates from this outbreak by using multilocus sequence typing and whole genome sequencing of selected strains. Though four PIF isolates and one hospital environmental isolate were identified as C. sakazakii sequence type 297 by multilocus sequence typing, they were isolated 6 months prior to the outbreak. Genotypic analyses of patient isolates identified them as Enterobacter hormaechei and Enterobacter spp. The pulsed-field gel electrophoresis profile of the Enterobacter spp. isolates matched those of isolates from previously unopened tins of PIF. E. hormaechei was only isolated from the two infants and not PIF. The reevaluation of this outbreak highlights the need for accurate detection and identification assays, particularly during outbreak investigations in which incorrect identifications may mislead the investigation and attribution of the source. Though the species responsible for the symptoms could not be determined, this outbreak demonstrated the possible transmission of Enterobacter spp. from PIF to infants. These are possibly the first reported cases of Enterobacter spp. infection of infants from bacterialcontaminated PIF.
In 2010, two infants in a hospital in Mexico who were fed reconstituted powdered infant formula (R-PIF) devel oped bloody diarrhea (3). Both infants were treated with antibiotics and recovered from their illness (3). The reported analysis of isolates from PIF, R-PIF, and patient fecal samples led to the conclusion that the illnesses were caused by Cronobacter sakazakii (3). PIF was suspected as the source of the bacteria, especially as the water used to prepare the formula was not sufficiently heated to ensure destruction of microorganisms (3). The original outbreak investigation utilized cultural isolation, phenotyping, and PCR probe-based identification of suspect isolates (3, 14, 16, 19). However, difficulties using these methods for identification of Cronobacter isolates have been reported (2, 9-11, 20). Additionally, cultural identification methods have misidentified Entero bacter species as Cronobacter species (9, 16, 20). The current study was undertaken to further characterize the isolates associated with this outbreak. Isolates were identified using multilocus sequence typing (MLST) and full genome sequencing. These results suggest that this particular outbreak was caused by one or more species of
* Author for correspondence. Tel and Fax: + 4 4 (0)115 848 3529; E-mail: [email protected].
the genus Enterobacter, not C. sakazakii, as originally reported. M A TER IA LS A N D M E TH O D S Bacterial strains and cultural confirmation. This study used bacterial strains from the previous report and additional strains not included in the original publication of the outbreak (8731, 8733, and 8756) (3). These latter strains were included in the current study, as they were isolated from PIF and the hospital environment at approximately the same time as strains 8710 and 8718, which were analyzed in the previous report. All PIF strains were isolated from unopened tins (400 g). Strains were cultured on tryptic soy agar (TSA; Oxoid Thermo Scientific, Basingstoke, UK) and incubated at 37°C for 24 h. An isolated colony from each plate was streaked to DrugganIversen-Forsythe agar (DFI; Oxoid Thermo Scientific) and incubated at 37°C for 24 h. Both agars were examined for typical colony morphologies. DNA extraction. DNA was extracted from strains 8701, 8706, 8710, 8718, 8731, 8733, and 8756 for genotypic characterization. Strains 8706 and 8701 were selected as patient isolates, as each strain represents one of the two main pulsed-field gel electrophoresis groups identified in the original outbreak investigation and strains from both groups were isolated from both infants (3). In the original outbreak investigation, strains were inconsistently assigned to groups based either on banding
1192
JACKSON ET AL.
J. Food Prot., Vol. 78, No. 6
TABLE 1. Bacterial strains and sourcesa Strain no.
Source
Date of isolation
fusA allele
PIF unopened PIF unopened PIF unopened PIF unopened Bottle-washing area
21 July 2009 21 July 2009 28 July 2009 28 July 2009 1 Aug. 2009
1 1 1 1 1
Washing zone R-PIF PIF unopened PIF unopened R-PIF R-PIF R-PIF R-PIF R-PEF Feces (infant 1) Feces (infant 2)
10 19 24 24 24 24 24 24 24 24 24
Aug. 2009 Aug. 2009 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010
NDrf ND ND ND ND ND ND ND ND 81 ND
ND ND ND ND ND ND ND ND ND Enterobacter spp. ND
Feces Feces Feces Feces Feces
24 24 24 24 24
Jan. Jan. Jan. Jan. Jan.
ND ND 75 ND ND
ND ND E. hormaechei ND ND
Identity
PTl* 8710 8731° 8718 8733' 8756c
C. C. C. C. C.
sakazakii sakazakii sakazakii sakazakii sakazakii
ST297 ST297 ST297 ST297 ST297
PT2 8740' 8715 8705 8720 8708 8716 8724 8721 8707' 8706 8709 PT3 8700 8719 8701 8702 8703
(infant (infant (infant (infant (infant
1) 1) 2) 2) 2)
2010 2010 2010 2010 2010
a Isolates shown in bold were considered to be part of the outbreak. PIF, powdered infant formula; R-PIF, reconstituted powdered infant formula; ST, sequence type. h PT, pulsotype. In the original investigation, strains were assigned to groups A to D, based on date of isolation. Strains were reassigned to PTs based on the original pulsed-field gel electrophoresis analysis to better illustrate their relatedness. 1 Pulsotype unpublished data. Isolates not included in original outbreak investigation (3). d ND, not determined.
profile or date of isolation. Strains were reassigned to groups that correspond to pulsotypes (PT) based on profile similarity only; see Table 1. Pulsed-field gel electrophoresis groups A and B were renamed here as PT2 and PT3, respectively. It should be noted that PT2 and PT3 strains were isolated from both patients. The original analysis showed that strains 8710 and 8718 clustered together. These have been termed PT1 in this article. Strains 8710 and 8718 were selected as PIF isolates, and strain 8756 was further analyzed, as it was isolated from the hospital environment (bottle-washing sink). Strains 8731 and 8733 were further analyzed because they displayed colony morphologies typical of Cronobacter spp. on TSA and DFI. DNA was extracted from these strains using the GenElute Bacterial Genomic Miniprep Kit (Sigma, Gillingham, UK), according to the manufacturer’s instructions. Extracted DNA was stored at —20°C and was thawed on ice before use. Genome sequencing. The extracted genomic DNA of strains 8701, 8706, and 8756 was sequenced by Exeter Sequencing Services at Exeter University (UK). Assembled genomes were uploaded to the Cronobacter MLST database at http://pubmlst.org/ cronobacter (ID: 855-7). Full genomes of additional species and strains were accessed through the Cronobacter database at http:// pubmlst.org/cronobacter. The sequences of the a-glucosidase encoding genes, palQ and palZ were extracted from the partial genome sequence of “Enterobacter sakazakii” (accession: AM075208) (15). Using the
BLAST function of the Cronobacter PubMLST database, the genomes of strains 8701, 8706, and 8756 were examined for the presence of these genes. Genes were considered present if >90% of the nucleotide sequence was detected. The genomes were also searched for virulence associated genes using either the BLAST function or the Genome Comparator tool of the Cronobacter PubMLST database. Genes encoding iron uptake systems (entABCDEFS, fepA l, fepA2, fepBCDEG, flwABCD), fimbriae (fim Al, fimA2, fim D l, fimD2, fimD3, fimD4 and fimG), and hemolysins (hemABDEFGHKLNX) were exam ined. The accession numbers for fhuABCDE are ESA_03187-90 and ESA_02242. The remaining genes are available via the Genome Comparator tool of the Cronobacter PubMLST database. Multilocus sequence typing. DNA from strains 8710, 8718, 8731, and 8733 was amplified and sequenced for the seven-loci MLST scheme (atpD, fusA, glnS, gltB, gyrB, injB, and ppsA; concatenated length 3,096 bp), as described by Baldwin et al. (/). The MLST profiles for strains 8701, 8706, and 8756 were obtained by extracting the allele sequences from the full genome sequences by using the database at http://pubmlst.org/cronobacter. Allele numbers and sequence types (STs) were assigned, as described by Baldwin et al. (1). Species identifications were based upon the sequence of the fusA gene included in this scheme (13). Ribosomal MLST (rMLST) analysis was performed on strains 8701, 8706, and 8756, using the Genome Comparator tool available at http://pubmlst.org/cronobacter. This tool extracts the
REEVALUATION OF C. SAKAZAKII OUTBREAK
J. Food Prot., Vol. 78, No. 6
selected gene sequences from the available full genomes. A total of 53 genes were included in this analysis, with a total concatenated length of 22,511 bp. Concatenated sequences were aligned using MAFFT at the time of extraction. Construction of phylogenetic trees. Phylogenetic trees were constructed for the six-loci MLST and rMLST sequences by using the neighbor-joining method within MEGA 5. Only six loci were used in construction of the MLST tree, as strain 8706 contained a truncated glnS gene. The total concatenated length of the sequences was 2,673 bp. The strength of the phylogenetic analysis was assessed using the bootstrap method with 1,000 replicates. Average nucleotide identity. To further identify the species involved in this outbreak, the genomes from strains 8701, 8706, and 8756 were compared with the genomes of a variety of Enterobacter spp. Those species clustering closest to strains 8701 and 8706 were used in the average nucleotide identity (ANI) analysis, performed as described by Goris et al. (7), at http://enveomics.ce.gatech.edu/ani. An ANI value of &95% indicates that two strains belong to the same species (7). RESULTS
Typical Cronobacter colony morphologies (yellow on TSA; blue-green on DFI) were observed for only strains 8710, 8718, 8731, 8733, and 8756, which were isolated in July and August 2009 from unopened tins of PIF and in the hospital environment. The remaining isolates showed cream-colored, glossy colonies on both media, indicating a lack of pigment production and oc-glucosidase activity. To confirm the lack of a-glucosidase genes, the genomes of strain 8701, 8706, and 8756 were examined for two a-glucosidase-encoding genes: palQ and palZ (accession: AM075208) (15). Strain 8756 matched the full-length sequence of palQ (1,677 bp), while strains 8701 and 8706 matched only 315 and 318 bp, respectively. Similarly, the full length of palZ (1,656 bp) was detected in strain 8756, but only 166 and 369 bp were matched in strains 8701 and 8706, respectively. Examination of the genomes of strain 8701 and 8706 revealed the presence of some putative virulence factors. Although fim D l was present in 8706 and partially present in 8701, all other fimbriae genes were absent in both strains. Conversely, all hemolysin genes were present in the both strains. Additionally, the iron uptake system fhuABCDE was found to be present in both strains, though a partial sequence was detected forESA_03190 in 8706. Sequences matching the other two hemolysin gene clusters were also detected, though at least one gene from each cluster was missing in both strains. The fusA gene sequence was extracted from the full genome sequence of strains 8701, 8706, and 8756. Additionally, the fusA gene from strains 8710, 8718, 8731, and 8733 was sequenced, as described by Baldwin et al. (1). Allele numbers were assigned and uploaded to the database at http://pubmlst.org/cronobacter/ (12). Strains 8710, 8718, 8731, 8733, and 8756 were all found to match fusA allele 1 (Table 1), identifying them as C. sakazakii (13). These five strains were found to belong to ST297 when all seven loci were sequenced. The patient strains 8701 and 8706 had fusA alleles 75 and 81, respectively
1193
(Table 1). FusA allele 75 presumptively identifies strain 8701 as Enterobacter ludwigii, while allele 81 identifies strain 8706 as Enterobacter spp. As the Cronobacter PubMLST database is dedicated to Cronobacter species, it does not contain representatives of all Enterobacter species. As the patient strain 8706 contained a truncated glnS gene, the six remaining MLST loci were used for phylogenetic analysis, shown in Figure 1. As expected from the fusA sequences and MLST, the PIF and environmental strains 8710, 8718, 8731, 8733, and 8756 cluster with C. sakazakii. The clinical strains 8701 (from infant 2) and 8706 (from infant 1) cluster with members of the Enterobacter genus, supporting the claim that these isolates are not C. sakazakii. Strain 8701 clustered closest to Enterobacter hormaechei ATCC 49162T, with very strong bootstrap support. Strain 8706 clustered closest to E. ludwigii Ec-WSUl, though with a low bootstrap value of 67%. The 53 loci of the rMLST were also used to construct a phylogenetic tree (Fig. 2). As with the six loci MLST, strain 8756 clustered with C. sakazakii, supporting its identifica tion based on the fusA sequence. Strains 8701 and 8706 again cluster with members of the Enterobacter genus. Strain 8701 again clustered closest to E. hormaechei ATCC 49162t . Strain 8706 also clustered near E. hormaechei but on its own branch. The locations of both strains are supported by high bootstrap values. The results of the two-way ANI analysis are shown in Table 2. This analysis showed the closest match of E. hormaechei ATCC 49162T for patient strain 8701 (95.59%) and Enterobacter asburiae LI for patient strain 8706 (91.28%). The PIF strain 8756 showed a very high similarity (98.22%) to C. sakazakii ATCC 29544T. D ISC U SSIO N
The original description of the clinical symptoms of bloody diarrhea did not match the normal description of symptoms associated with C. sakazakii infection and warranted further analysis (3,6). Furthermore, initial analysis revealed that the clinical isolates did not produce typical Cronobacter colony morphologies on TSA or DFI, further supporting the possibility that they were unlikely to be C. sakazakii-, see Table 1. Consequently, genotypic analysis of selected isolates was undertaken in an attempt to identify the species responsible for the illnesses reported in Mexico. As expected based on the colony morphologies, strains 8701 and 8706 were found to lack the a-glucosidase genes, palQ and palZ (15). This further supports the proposal that these strains are not Cronobacter spp. Conversely, strain 8756 was found to contain both palQ and palZ. as expected, based on the typical Cronobacter colony morphology displayed on DFI. Further examination of the genomes for putative virulence factors revealed a possible hemolysin and at least one iron uptake system in strains 8701 and 8706. The presence of these potential virulence factors supports the claim that one or both of these strains caused the illnesses. MLST confirms that the PIF isolates from July and August 2009 were C. sakazakii, genotyped as ST297. The
1194
JACKSON ET AL.
J. Food Prot., Vol. 78, No. 6
FIGURE 1. Neighbor-joining phylogenet ic tree based on the six concatenated atpD, fusA, gltB gyrB, infB, and ppsA sequences (concatenated length: 2,673 bp: littp:ll pitbmlst.org/cronobacteij, showing the po sition o f strains 8701, 8706, 8710. 8718. 8731, 8733, and 8756 within the genera Siccibacter, Franconibacter, Cronobacter, Citrobacter, and Enterobacter. Bootstrap values based on 1,000 replications are shown at branch nodes. Bar, one substitu tion per 100 nucleotide positions.
40 p 98
- C ronobacter con d im en ti LM G 26250T - C ronobacter m u ytjen sii ATCC51329T
L
------ C ronobacter dublinensis LM G 23823' ----- C ronobacter tu ric em is LM G 23827T
100
Cronobacter universalis NCTC9529T
----- C ronobacter m a lo n a ticu s LM G 238261 — — C ronobacter sa ka za kii ATCC2 9 5 4 4 T 100
100
8756 8710 100
8718 8733 8732
------------------ F ra n conibacterhelveticus LMG2 3 73 2T ---------------- F ranconibacter p u lveris G -6 0 1 /0 5 T ------------- Siccibacter colletis LM G28204T ---- Siccibacter turicensis LM G 23730T ---------- C itrobacter y o u n g a e ATCC29220T
100
94 100
---------------- C itrobacter freu n d ii ATCC8090T ---------- C itrobacter rodentium ICC168 -------------- C itrobacter am a lo n a ticu s A SM 73105v ---------- E ntero b a cter m assiliensis |C 163 ------------------- E ntero b a cter aerogenes KCTC21901 --------------------------- Yokenella regensburgei ATCC43003 -------------------------------- E n tero b a cter asburiae LI ------------------ Enterobacter cancerogenus ATCC35316 ------------------------ Enterobacter cloacae ATCC13047T 100 |------------------ Enterobacter horm aechei ATCC49162T
99 72
34 90 100 79
I--------------------8701
------------------------------ Enterobacter ludw igii ATCC49162T ------------------------------- 8 7 0 6
I-----
001
H
PubMLST Cronobacter database, which contains over 1,000 strains, does not contain any clinical strains with this sequence type, and it is unrelated to C. sakazakii STs that have been linked to neonatal illnesses (13). Converse ly, fusA sequencing identified patient strains 8701 and FIGURE 2. Neighbor-joining phylogenet ic tree based on 53 concatenated ribosomal proteins genes (rMLST) sequences (concat enated length: 22,511 bp: http./lpubmlst. orglcronobacter), showing the position of strains 8701, 8706, and 8756 within the genera Siccibacter, Franconibacter, Crono bacter, Citrobacter, and Enterobacter. Boot strap values (>70%) based on 1,000 replications are shown at branch nodes. Bar, five substitutions per 1,000 nucleotide positions.
8706 as Enterobacter species. Given the Cronobacter PubMLST does not contain representative fusA profiles for all Enterobacter species, further phylogenetic analysis based on 6 or 53 loci was undertaken to identify these isolates.
Cronobacter sakazakii ATCC295441 8756
------------- Cronobacter malonaticus LMG238261 — Cronobacter universalis NCTC9529T
----- Cronobacter turicensis LMG238271 ------------------------------------ Cronobacter condimenti LMG262501 — Cronobacter dublinensis LMG238237
—
Cronobacter muytjensii ATCC51329T
-
Franconibacter pulveris G-601/05T
Franconibacter helveticus LMG23732T mo |- --------------------------- Siccibacter colletis LMG282041
------------------------------- Siccibacter turicensis LMG237301 loo |- ---------- Citrobacter freundii ATCC8090T '- ------ -Citrobacteryoungae ATCC29220T Citrobacter rodentium ICC168
------ Citrobacter amalonaticus ASM73105v -------- Yokenella regensburgei ATCC43003 ---------------- Enterobacter aerogenes KCTC2190T ------------------- Enterobacter asburiae LI ----------------- Enterobacter cancerogenus ATCC35316 ------ Enterobacter ludwigii ATCC491621 looi— Enterobacter hormaechei ATCC491627 I— 8701 -8 7 0 6 — Enterobacter cloacae ATCC13047T
- Enterobacter massiliensis JC163
i------0.005
H
1195
REEVALUATION OF C. SA KA ZAK II OUTBREAK
J. Food Prot., Vol. 78, No. 6
TABLE 2. Average nucleotide identity (ANI) values for strains 8756, 8706, and 8701 compared with Cronobacter sakazakii and a variety of Enterobacter speciesa Species
C. sakazakii E. asburiae E. hormaechei Enterobacter cancerogenus E. cloacae E. ludwigii Enterobacter aerogenes
Strain 8756 (PT1) 8706 (PT2) 8701 (PT3) ATCC 295441 LI ATCC 491 62t ATCC 35316 ATCC I 3047t ECWSU1 KCTC2190t
8756 (PT 1; %)
82.92 83.01 9 8 .2 2
83.20 83.10 82.99 82.80 82.78 83.07
8706 (FT2; %)
88.93 84.00 9 1 .2 8
88.88 87.84 89.18 88.76 83.65
8701 (PT3; %)
83.67 89.15 9 5 .5 9
88.09 88.52 87.85 83.64
a The highest ANI value for each strain is shown in bold. PT, pulsotype.
The phylogenetic analyses showed that strain 8701 (PT3), from infant 2, clustered with E. hormaechei, supported by bootstrap values of 100% on both trees, (Figs. 1 and 2). The ANI analysis showed a 95.59% match with E. hormaechei ATCC 49162T. An ANI value of >95% has been suggested for species demarcation (7). The high ANI value and strongly supported phylogenetic analyses lead to the conclusion that strain 8701, isolated from a fecal sample from infant 2, is E. hormaechei. As reported previously, the isolates from PT3 were resistant to several antibiotics, including cefotaxime and chloramphenicol (3). High antimicrobial resistance has been reported for E. hormaechei strains, which had been isolated from neonates and misidentified as C. sakazakii (19). Because the PT3 isolates were isolated from both patients, but not from PIF or R-PIF, we are uncertain of the source. The Enterobacter spp. isolates belonging to PT2 were isolated from PIF, R-PIF, and both patients (3). The exact species could not be detennined, as fusA allele 81 has not been linked to a particular Enterobacter species. The phylogenetic analysis based on six loci showed strain 8706 clustered with E. ludwigii, with low bootstrap support (Fig. 1). The rMLST analysis, based on 53 loci, showed strain 8706 clustered near E. hormaechei, but on its own branch, supported by high bootstrapping values (Fig. 2). The closest ANI match to strain 8706 was E. asburiae LI, with a value of 91.28% (Table 2). This value is not high enough to positively identify strain 8706 as E. asburiae. The heteroge neity of the Enterobacter cloacae complex did not allow for identification based on 16S rRNA sequences, and the species of this isolate remains unknown (8). Though the isolate could not be identified to the species level, these analyses indicate that it is not C. sakazakii, as originally reported. This reevaluation shows the importance of accurate detection and identification methods. The phenotypic and PCR-based methods used in the original outbreak investi gation can be unreliable for identification of Cronobacter spp. (2, 9—11, 19). The PCR probe methods utilized by Flores et al. (3) for species identification targeted the 16S rRNA and rpoB genes, both of which have exhibited problems in the past (2, 10, 11). The 16S rRNA PCR method detects only “E. sakazakii,” and the method targeting rpoB has been shown to produce false-positive
results with strains belonging to the newly defined genus Franconibacter (10, 14, 18, 19). Though the true cause of the infant illnesses cannot be determined, it is suspected that they were caused by E. hormaechei, as this species has been implicated in previous neonatal intensive care unit outbreaks (20, 21). Addition ally, Enterobacter spp., including E. hormaechei, have been previously isolated from PIF and have been misidentified as “E. sakazakii,” the fonner name for Cronobacter spp. (9, 17, 20). Accurate identification is essential for tracing out breaks, and standards must be written that will ensure protection from the hazards present in food products. As the Enterobacter spp. isolates belonging to PT2 were obtained from the unopened tins of PIF and patient samples, PIF is suspected as a possible source of these isolates. It is important to note that during the outbreak investigation in Mexico, it was found that the water used to rehydrate the PIF was only heated to approximately 40°C, which is not hot enough to kill microorganisms present (3). This oversight may have exposed the infants to the pathogens, bringing preparation practices and the microbiological criteria specified in PIF standards into question. Currently, Cronobacter species and Salmonella species are the only organisms with “ clear evidence of causality (category A)’’ in illnesses linked to PIF (5). Category B organisms (“ causality plausible, but not yet demonstrated” ) include Citrobacter koseri, Citrobacter freundii, Enterobacter cloacae, Escherichia vulneris, Escherichia coli, Hafnia alvei, Klebsiella pneumoniae, Klebsiella oxytoca, Pantoea agglomerans, Serratia spp., and Acinetobacter spp. (5, 6). The designation of some organisms may need to be revised to reflect outbreaks that have now been linked to species other than Cronobacter and Salmonella. The outbreak in Mexico demonstrates that species currently permitted in PIF can be transmitted to infants when appropriate measures, such as heating reconstitution water to at least 70°C, are not taken (4-6). Opportunistic pathogens in the PIF may, then, go on to cause illnesses in a susceptible host. For example, problems with infection control measures may have contributed to an outbreak of E. hormaechei in Pennsylvania (21). It was found that health care workers were not changing their gloves between
1196
JACKSON ET AL.
patients and that hand washing facilities were sometimes not properly stocked, which may have- contributed to cross contamination between the infants. Cross-contamination may also have occurred in the hospital in Mexico. C. sakazakii ST297 was recovered from PIF and the hospital’s bottle-washing sink, suggesting cross contamination from the PIF. The possible cross-contamina tion, coupled with the use of reconstitution water heated to approximately 40°C, suggests the potential for an outbreak, though no illnesses were reported at the time. In conclusion, suspected Cronobacter infant isolates from an outbreak in Mexico were identified as two species of Enterobacter. Though the species responsible for the symptoms could not be determined, this outbreak demon strated possible transmission of Enterobacter spp. from PIF to infants. Additionally, attention must be paid to infection control measures in hospitals to prevent exposure to and spread of pathogens. ACKNOWLEDGMENTS This work was partially supported by DIUBB through project 143720 and a Santander Research Mobility Grant. This publication made use of the Cronobacter Multilocus Sequence Typing Web site (http://pubmlst.org/ cronobacter/), developed by Keith Jolley and sited at the University of Oxford. The development of this site has been funded by the Wellcome Trust.
J. Food Prot., Vol. 78, No. 6
8.
9.
10.
11.
12.
13.
14.
15.
REFERENCES 1.
Baldwin, A., M. Loughlin, J. Caubilla-Barron, E. Kucerova, G. Manning, C. Dowson, and S. Forsythe. 2009. Multilocus sequence typing of Cronobacter sakazakii and Cronobacter malonaticus reveals stable clonal structures with clinical significance which do not correlate with biotypes. BMC Microbiol. 9:223. doi: 10.1186/ 1471-2180-9-223. 2. Cetinkaya, E., S. Joseph, K. Ayhan, and S. J. Forsythe. 2013. Comparison of methods for the microbiological identification and profiling of Cronobacter species from ingredients used in the preparation of infant formula. Mol. Cell. Probes 27:60-64. 3. Mores, J. P., S. A. Medrano, J. S. Sanchez, and E. Femandez-Escartin. 2011. Two cases of hemorrhagic diarrhea caused by Cronobacter sakazakii in hospitalized nursing infants associated with the consump tion of powdered infant formula. J. Food Prot. 74:2177-2181. 4. Food and Agricultural Organization of the United Nations, World Health Organization. 2004. Enterobacter sakazakii and other organisms in powdered infant formula. Meeting report. Microbiolog ical risk assessment series. Meeting report. Available at: http://www. who.int/foodsafety/publications/mra6-enterobacter-sakazakii/en/. Ac cessed 19 February 2015. 5. Food and Agricultural Organization of the United Nations, World Health Organization. 2006. Enterobacter sakazakii and Salmonella in powdered infant formula. Meeting report. Microbiological risk assessment series. Available at: http://www.who.int/foodsafety/ publications/mralO/en/. Accessed 19 February 2015. 6. Food and Agricultural Organization of the United Nations, World Health Organization. 2008. Enterobacter sakazakii (Cronobacter spp.) in powdered formulae. Meeting report. Microbiological risk assessment series. Available at: http://www.who.int/foodsafety/ publications/mra_followup/en/. Accessed 19 February 2015. 7. Goris, J., K. T. Konstantinidis, J. A. Klappenbach, T. Coenye, P. Vandamme, and J. M. Tiedje. 2007. DNA-DNA hybridization values
16.
17.
18.
19.
20.
21.
and their relationship to whole-genome sequence similarities. Int. J. Syst. Evol. Microbiol. 57:81-91. Hoffmann. H., and A. Roggenkamp. 2003. Population genetics of the nomenspecies Enterobacter cloacae. Appl. Environ. Microbiol. 69: 5306-5318. Ivy, R. A., J. M. Farber, F. Pagotto, and M. Wiedmann. 2013. International Life Science Institute North America Cronobacter (formerly Enterobacter sakazakii) isolate set. J. Food Prot. 76:40-51. Jackson, E. E., H. Sonbol, N. Masood, and S. J. Forsythe. 2014. Genotypic and phenotypic characteristics of Cronobacter species, with particular attention to the newly reclassified species Cronobacter helveticus, Cronobacter pulveris, and Cronobacter zurichensis. Food Microbiol. 44:226-235. Jaradat, Z. W., Q. O. Ababneh, I. M. Saadoun, N. A. Samara, and A. M. Rashdan. 2009. Isolation of Cronobacter spp. (formerly Enterobacter sakazakii) from infant food, herbs and environmental samples and the subsequent identification and confirmation of the isolates using biochemical, chromogenic assays, PCR and 16S rRNA sequencing. BMC Microbiol. 9:225. doi:10.1186/1471-2180-9-225. Jolley K A, and M. C. Maiden. BIGSdb: scalable analysis of bacterial genome variation at the population level. BMC Bioinformatics 2010:11:595. Joseph, S., H. Sonbol, S. Hariri, P. Desai, M. McClelland, and S. J. Forsythe. 2012. Diversity of the Cronobacter genus as revealed by multilocus sequence typing../. Clin. Microbiol. 50:3031-3039. Lehner, A., C. Fricker-Feer, and R. Stephan. 2012. Identification of the recently described Cronobacter condimenti by an rpoB-genebased PCR system. J. Med. Microbiol. 61:1034-1035. Lehner, A., K. Riedel, T. Rattei, A. Ruepp, D. Frishman, P. Breeuwer, B. Diep, L. Eberl, and R. Stephan. 2006. Molecular characterization of the a-glucosidase activity in Enterobacter sakazakii reveals the presence of a putative gene cluster for palatinose metabolism. Syst. Appl. Microbiol. 29:609-625. Lehner, A., T. Tasara, and R. Stephan. 2004. 16S rRNA gene based analysis of Enterobacter sakazakii strains from different sources and development of a PCR assay for identification. BMC Microbiol. 4:43. Muytjens, H. L., H. Roelofs-Willemse, and G. H. Jaspar. 1988. Quality of powdered substitutes for breast milk with regard to members of the family Enterobacteriaceae. J. Clin. Microbiol. 26: 743-746. Stephan, R., C. J. Grim, G. R. Gopinath, M. K. Mammel, V. Sathyamoorthy, L. H. Trach, H. R. Chase, S. Fanning, and B. D. Tall. 2014. Re-examination of the taxonomic status of Enterobacter helveticus, Enterobacter pulveris and Enterobacter turicensis as members of the genus Cronobacter and their reclassification in the genera Franconibacter gen. nov. and Siccibacter gen. nov. as Franconibacter helveticus comb, nov., Franconibacter pulveris comb. nov. and Siccibacter turicensis comb, nov., respectively. Int. J. Syst. Evol. Microbiol. 64:3402-3410. Stoop, B.. A. Lehner, C. Iversen, S. Fanning, and R. Stephan. 2009. Development and evaluation of rpoB based PCR systems to differentiate the six proposed species within the genus Cronobacter. Int. J. Food Microbiol. 136:165-168. Townsend, S. M., E. Hurrell, J. Caubilla-Barron, C. Loc-Carrillo, and S. J. Forsythe. 2008. Characterization of an extended-spectrum betalactamase Enterobacter hormaechei nosocomial outbreak, and other Enterobacter hormaechei misidentified as Cronobacter (Enterobac ter) sakazakii. Microbiology 154:3659-3667. Wenger, P. N., J. I. Tokars, P. Brennan, C. Samel, L. Bland. M. Miller, L. Carson, M. Arduino, P. Edelstein, S. Aguero, C. Riddle, C. O ’Hara, and W. Jarvis. 1997. An outbreak of Enterobacter hormaechei infection and colonization in an intensive care nursery. Clin. Infect. Dis. 24:1243-1244.
Copyright of Journal of Food Protection is the property of Allen Press Publishing Services Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
Journal o f Food Protection, Vol. 78, No. 6, 2015, Pages 1191-1196 doi: 10.4315/0362-028X.JFP-14-563 Copyright © , International Association for Food Protection
Research Note
Reevaluation of a Suspected Cronobacter sakazakii Outbreak in Mexico EMILY E. JACKSON, 1 JULIO PARRA FLORES,2 EDUARDO FERNANDEZ-ESCARTIN,3 a n d STEPHEN J. FORSYTHE1* 1School o f Science and Technology, Nottingham Trent University, Clifton Lane, Nottingham, NG11 8NS, UK; 2Departamento de Nutrition y Salud Publica, Universidad del Bio-Bio, Chilian, Nuble, 3780000, Chilian, Chile; and 3Departamento de Investigation y Posgrado en Alimentos, Facultad de Quimica, Universidad Autonoma de Queretaro, Queretaro, 76010. Mexico MS 14-563: Received 26 November 2014/Accepted 30 January 2015
A BSTR A C T In 2010, two infants became ill at a hospital in Mexico. Subsequently, a range of clinical, environmental, and powdered and rehydrated infant formula isolates were identified by using a combination of phenotyping and PCR probes. The strains were clustered according to pulsed-field gel electrophoresis. The causative agent was reported as Cronobacter sakazakii, with powdered infant formula (PIF) identified as the likely source of the infections. This new study further characterized the isolates from this outbreak by using multilocus sequence typing and whole genome sequencing of selected strains. Though four PIF isolates and one hospital environmental isolate were identified as C. sakazakii sequence type 297 by multilocus sequence typing, they were isolated 6 months prior to the outbreak. Genotypic analyses of patient isolates identified them as Enterobacter hormaechei and Enterobacter spp. The pulsed-field gel electrophoresis profile of the Enterobacter spp. isolates matched those of isolates from previously unopened tins of PIF. E. hormaechei was only isolated from the two infants and not PIF. The reevaluation of this outbreak highlights the need for accurate detection and identification assays, particularly during outbreak investigations in which incorrect identifications may mislead the investigation and attribution of the source. Though the species responsible for the symptoms could not be determined, this outbreak demonstrated the possible transmission of Enterobacter spp. from PIF to infants. These are possibly the first reported cases of Enterobacter spp. infection of infants from bacterialcontaminated PIF.
In 2010, two infants in a hospital in Mexico who were fed reconstituted powdered infant formula (R-PIF) devel oped bloody diarrhea (3). Both infants were treated with antibiotics and recovered from their illness (3). The reported analysis of isolates from PIF, R-PIF, and patient fecal samples led to the conclusion that the illnesses were caused by Cronobacter sakazakii (3). PIF was suspected as the source of the bacteria, especially as the water used to prepare the formula was not sufficiently heated to ensure destruction of microorganisms (3). The original outbreak investigation utilized cultural isolation, phenotyping, and PCR probe-based identification of suspect isolates (3, 14, 16, 19). However, difficulties using these methods for identification of Cronobacter isolates have been reported (2, 9-11, 20). Additionally, cultural identification methods have misidentified Entero bacter species as Cronobacter species (9, 16, 20). The current study was undertaken to further characterize the isolates associated with this outbreak. Isolates were identified using multilocus sequence typing (MLST) and full genome sequencing. These results suggest that this particular outbreak was caused by one or more species of
* Author for correspondence. Tel and Fax: + 4 4 (0)115 848 3529; E-mail: [email protected].
the genus Enterobacter, not C. sakazakii, as originally reported. M A TER IA LS A N D M E TH O D S Bacterial strains and cultural confirmation. This study used bacterial strains from the previous report and additional strains not included in the original publication of the outbreak (8731, 8733, and 8756) (3). These latter strains were included in the current study, as they were isolated from PIF and the hospital environment at approximately the same time as strains 8710 and 8718, which were analyzed in the previous report. All PIF strains were isolated from unopened tins (400 g). Strains were cultured on tryptic soy agar (TSA; Oxoid Thermo Scientific, Basingstoke, UK) and incubated at 37°C for 24 h. An isolated colony from each plate was streaked to DrugganIversen-Forsythe agar (DFI; Oxoid Thermo Scientific) and incubated at 37°C for 24 h. Both agars were examined for typical colony morphologies. DNA extraction. DNA was extracted from strains 8701, 8706, 8710, 8718, 8731, 8733, and 8756 for genotypic characterization. Strains 8706 and 8701 were selected as patient isolates, as each strain represents one of the two main pulsed-field gel electrophoresis groups identified in the original outbreak investigation and strains from both groups were isolated from both infants (3). In the original outbreak investigation, strains were inconsistently assigned to groups based either on banding
1192
JACKSON ET AL.
J. Food Prot., Vol. 78, No. 6
TABLE 1. Bacterial strains and sourcesa Strain no.
Source
Date of isolation
fusA allele
PIF unopened PIF unopened PIF unopened PIF unopened Bottle-washing area
21 July 2009 21 July 2009 28 July 2009 28 July 2009 1 Aug. 2009
1 1 1 1 1
Washing zone R-PIF PIF unopened PIF unopened R-PIF R-PIF R-PIF R-PIF R-PEF Feces (infant 1) Feces (infant 2)
10 19 24 24 24 24 24 24 24 24 24
Aug. 2009 Aug. 2009 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010 Jan. 2010
NDrf ND ND ND ND ND ND ND ND 81 ND
ND ND ND ND ND ND ND ND ND Enterobacter spp. ND
Feces Feces Feces Feces Feces
24 24 24 24 24
Jan. Jan. Jan. Jan. Jan.
ND ND 75 ND ND
ND ND E. hormaechei ND ND
Identity
PTl* 8710 8731° 8718 8733' 8756c
C. C. C. C. C.
sakazakii sakazakii sakazakii sakazakii sakazakii
ST297 ST297 ST297 ST297 ST297
PT2 8740' 8715 8705 8720 8708 8716 8724 8721 8707' 8706 8709 PT3 8700 8719 8701 8702 8703
(infant (infant (infant (infant (infant
1) 1) 2) 2) 2)
2010 2010 2010 2010 2010
a Isolates shown in bold were considered to be part of the outbreak. PIF, powdered infant formula; R-PIF, reconstituted powdered infant formula; ST, sequence type. h PT, pulsotype. In the original investigation, strains were assigned to groups A to D, based on date of isolation. Strains were reassigned to PTs based on the original pulsed-field gel electrophoresis analysis to better illustrate their relatedness. 1 Pulsotype unpublished data. Isolates not included in original outbreak investigation (3). d ND, not determined.
profile or date of isolation. Strains were reassigned to groups that correspond to pulsotypes (PT) based on profile similarity only; see Table 1. Pulsed-field gel electrophoresis groups A and B were renamed here as PT2 and PT3, respectively. It should be noted that PT2 and PT3 strains were isolated from both patients. The original analysis showed that strains 8710 and 8718 clustered together. These have been termed PT1 in this article. Strains 8710 and 8718 were selected as PIF isolates, and strain 8756 was further analyzed, as it was isolated from the hospital environment (bottle-washing sink). Strains 8731 and 8733 were further analyzed because they displayed colony morphologies typical of Cronobacter spp. on TSA and DFI. DNA was extracted from these strains using the GenElute Bacterial Genomic Miniprep Kit (Sigma, Gillingham, UK), according to the manufacturer’s instructions. Extracted DNA was stored at —20°C and was thawed on ice before use. Genome sequencing. The extracted genomic DNA of strains 8701, 8706, and 8756 was sequenced by Exeter Sequencing Services at Exeter University (UK). Assembled genomes were uploaded to the Cronobacter MLST database at http://pubmlst.org/ cronobacter (ID: 855-7). Full genomes of additional species and strains were accessed through the Cronobacter database at http:// pubmlst.org/cronobacter. The sequences of the a-glucosidase encoding genes, palQ and palZ were extracted from the partial genome sequence of “Enterobacter sakazakii” (accession: AM075208) (15). Using the
BLAST function of the Cronobacter PubMLST database, the genomes of strains 8701, 8706, and 8756 were examined for the presence of these genes. Genes were considered present if >90% of the nucleotide sequence was detected. The genomes were also searched for virulence associated genes using either the BLAST function or the Genome Comparator tool of the Cronobacter PubMLST database. Genes encoding iron uptake systems (entABCDEFS, fepA l, fepA2, fepBCDEG, flwABCD), fimbriae (fim Al, fimA2, fim D l, fimD2, fimD3, fimD4 and fimG), and hemolysins (hemABDEFGHKLNX) were exam ined. The accession numbers for fhuABCDE are ESA_03187-90 and ESA_02242. The remaining genes are available via the Genome Comparator tool of the Cronobacter PubMLST database. Multilocus sequence typing. DNA from strains 8710, 8718, 8731, and 8733 was amplified and sequenced for the seven-loci MLST scheme (atpD, fusA, glnS, gltB, gyrB, injB, and ppsA; concatenated length 3,096 bp), as described by Baldwin et al. (/). The MLST profiles for strains 8701, 8706, and 8756 were obtained by extracting the allele sequences from the full genome sequences by using the database at http://pubmlst.org/cronobacter. Allele numbers and sequence types (STs) were assigned, as described by Baldwin et al. (1). Species identifications were based upon the sequence of the fusA gene included in this scheme (13). Ribosomal MLST (rMLST) analysis was performed on strains 8701, 8706, and 8756, using the Genome Comparator tool available at http://pubmlst.org/cronobacter. This tool extracts the
REEVALUATION OF C. SAKAZAKII OUTBREAK
J. Food Prot., Vol. 78, No. 6
selected gene sequences from the available full genomes. A total of 53 genes were included in this analysis, with a total concatenated length of 22,511 bp. Concatenated sequences were aligned using MAFFT at the time of extraction. Construction of phylogenetic trees. Phylogenetic trees were constructed for the six-loci MLST and rMLST sequences by using the neighbor-joining method within MEGA 5. Only six loci were used in construction of the MLST tree, as strain 8706 contained a truncated glnS gene. The total concatenated length of the sequences was 2,673 bp. The strength of the phylogenetic analysis was assessed using the bootstrap method with 1,000 replicates. Average nucleotide identity. To further identify the species involved in this outbreak, the genomes from strains 8701, 8706, and 8756 were compared with the genomes of a variety of Enterobacter spp. Those species clustering closest to strains 8701 and 8706 were used in the average nucleotide identity (ANI) analysis, performed as described by Goris et al. (7), at http://enveomics.ce.gatech.edu/ani. An ANI value of &95% indicates that two strains belong to the same species (7). RESULTS
Typical Cronobacter colony morphologies (yellow on TSA; blue-green on DFI) were observed for only strains 8710, 8718, 8731, 8733, and 8756, which were isolated in July and August 2009 from unopened tins of PIF and in the hospital environment. The remaining isolates showed cream-colored, glossy colonies on both media, indicating a lack of pigment production and oc-glucosidase activity. To confirm the lack of a-glucosidase genes, the genomes of strain 8701, 8706, and 8756 were examined for two a-glucosidase-encoding genes: palQ and palZ (accession: AM075208) (15). Strain 8756 matched the full-length sequence of palQ (1,677 bp), while strains 8701 and 8706 matched only 315 and 318 bp, respectively. Similarly, the full length of palZ (1,656 bp) was detected in strain 8756, but only 166 and 369 bp were matched in strains 8701 and 8706, respectively. Examination of the genomes of strain 8701 and 8706 revealed the presence of some putative virulence factors. Although fim D l was present in 8706 and partially present in 8701, all other fimbriae genes were absent in both strains. Conversely, all hemolysin genes were present in the both strains. Additionally, the iron uptake system fhuABCDE was found to be present in both strains, though a partial sequence was detected forESA_03190 in 8706. Sequences matching the other two hemolysin gene clusters were also detected, though at least one gene from each cluster was missing in both strains. The fusA gene sequence was extracted from the full genome sequence of strains 8701, 8706, and 8756. Additionally, the fusA gene from strains 8710, 8718, 8731, and 8733 was sequenced, as described by Baldwin et al. (1). Allele numbers were assigned and uploaded to the database at http://pubmlst.org/cronobacter/ (12). Strains 8710, 8718, 8731, 8733, and 8756 were all found to match fusA allele 1 (Table 1), identifying them as C. sakazakii (13). These five strains were found to belong to ST297 when all seven loci were sequenced. The patient strains 8701 and 8706 had fusA alleles 75 and 81, respectively
1193
(Table 1). FusA allele 75 presumptively identifies strain 8701 as Enterobacter ludwigii, while allele 81 identifies strain 8706 as Enterobacter spp. As the Cronobacter PubMLST database is dedicated to Cronobacter species, it does not contain representatives of all Enterobacter species. As the patient strain 8706 contained a truncated glnS gene, the six remaining MLST loci were used for phylogenetic analysis, shown in Figure 1. As expected from the fusA sequences and MLST, the PIF and environmental strains 8710, 8718, 8731, 8733, and 8756 cluster with C. sakazakii. The clinical strains 8701 (from infant 2) and 8706 (from infant 1) cluster with members of the Enterobacter genus, supporting the claim that these isolates are not C. sakazakii. Strain 8701 clustered closest to Enterobacter hormaechei ATCC 49162T, with very strong bootstrap support. Strain 8706 clustered closest to E. ludwigii Ec-WSUl, though with a low bootstrap value of 67%. The 53 loci of the rMLST were also used to construct a phylogenetic tree (Fig. 2). As with the six loci MLST, strain 8756 clustered with C. sakazakii, supporting its identifica tion based on the fusA sequence. Strains 8701 and 8706 again cluster with members of the Enterobacter genus. Strain 8701 again clustered closest to E. hormaechei ATCC 49162t . Strain 8706 also clustered near E. hormaechei but on its own branch. The locations of both strains are supported by high bootstrap values. The results of the two-way ANI analysis are shown in Table 2. This analysis showed the closest match of E. hormaechei ATCC 49162T for patient strain 8701 (95.59%) and Enterobacter asburiae LI for patient strain 8706 (91.28%). The PIF strain 8756 showed a very high similarity (98.22%) to C. sakazakii ATCC 29544T. D ISC U SSIO N
The original description of the clinical symptoms of bloody diarrhea did not match the normal description of symptoms associated with C. sakazakii infection and warranted further analysis (3,6). Furthermore, initial analysis revealed that the clinical isolates did not produce typical Cronobacter colony morphologies on TSA or DFI, further supporting the possibility that they were unlikely to be C. sakazakii-, see Table 1. Consequently, genotypic analysis of selected isolates was undertaken in an attempt to identify the species responsible for the illnesses reported in Mexico. As expected based on the colony morphologies, strains 8701 and 8706 were found to lack the a-glucosidase genes, palQ and palZ (15). This further supports the proposal that these strains are not Cronobacter spp. Conversely, strain 8756 was found to contain both palQ and palZ. as expected, based on the typical Cronobacter colony morphology displayed on DFI. Further examination of the genomes for putative virulence factors revealed a possible hemolysin and at least one iron uptake system in strains 8701 and 8706. The presence of these potential virulence factors supports the claim that one or both of these strains caused the illnesses. MLST confirms that the PIF isolates from July and August 2009 were C. sakazakii, genotyped as ST297. The
1194
JACKSON ET AL.
J. Food Prot., Vol. 78, No. 6
FIGURE 1. Neighbor-joining phylogenet ic tree based on the six concatenated atpD, fusA, gltB gyrB, infB, and ppsA sequences (concatenated length: 2,673 bp: littp:ll pitbmlst.org/cronobacteij, showing the po sition o f strains 8701, 8706, 8710. 8718. 8731, 8733, and 8756 within the genera Siccibacter, Franconibacter, Cronobacter, Citrobacter, and Enterobacter. Bootstrap values based on 1,000 replications are shown at branch nodes. Bar, one substitu tion per 100 nucleotide positions.
40 p 98
- C ronobacter con d im en ti LM G 26250T - C ronobacter m u ytjen sii ATCC51329T
L
------ C ronobacter dublinensis LM G 23823' ----- C ronobacter tu ric em is LM G 23827T
100
Cronobacter universalis NCTC9529T
----- C ronobacter m a lo n a ticu s LM G 238261 — — C ronobacter sa ka za kii ATCC2 9 5 4 4 T 100
100
8756 8710 100
8718 8733 8732
------------------ F ra n conibacterhelveticus LMG2 3 73 2T ---------------- F ranconibacter p u lveris G -6 0 1 /0 5 T ------------- Siccibacter colletis LM G28204T ---- Siccibacter turicensis LM G 23730T ---------- C itrobacter y o u n g a e ATCC29220T
100
94 100
---------------- C itrobacter freu n d ii ATCC8090T ---------- C itrobacter rodentium ICC168 -------------- C itrobacter am a lo n a ticu s A SM 73105v ---------- E ntero b a cter m assiliensis |C 163 ------------------- E ntero b a cter aerogenes KCTC21901 --------------------------- Yokenella regensburgei ATCC43003 -------------------------------- E n tero b a cter asburiae LI ------------------ Enterobacter cancerogenus ATCC35316 ------------------------ Enterobacter cloacae ATCC13047T 100 |------------------ Enterobacter horm aechei ATCC49162T
99 72
34 90 100 79
I--------------------8701
------------------------------ Enterobacter ludw igii ATCC49162T ------------------------------- 8 7 0 6
I-----
001
H
PubMLST Cronobacter database, which contains over 1,000 strains, does not contain any clinical strains with this sequence type, and it is unrelated to C. sakazakii STs that have been linked to neonatal illnesses (13). Converse ly, fusA sequencing identified patient strains 8701 and FIGURE 2. Neighbor-joining phylogenet ic tree based on 53 concatenated ribosomal proteins genes (rMLST) sequences (concat enated length: 22,511 bp: http./lpubmlst. orglcronobacter), showing the position of strains 8701, 8706, and 8756 within the genera Siccibacter, Franconibacter, Crono bacter, Citrobacter, and Enterobacter. Boot strap values (>70%) based on 1,000 replications are shown at branch nodes. Bar, five substitutions per 1,000 nucleotide positions.
8706 as Enterobacter species. Given the Cronobacter PubMLST does not contain representative fusA profiles for all Enterobacter species, further phylogenetic analysis based on 6 or 53 loci was undertaken to identify these isolates.
Cronobacter sakazakii ATCC295441 8756
------------- Cronobacter malonaticus LMG238261 — Cronobacter universalis NCTC9529T
----- Cronobacter turicensis LMG238271 ------------------------------------ Cronobacter condimenti LMG262501 — Cronobacter dublinensis LMG238237
—
Cronobacter muytjensii ATCC51329T
-
Franconibacter pulveris G-601/05T
Franconibacter helveticus LMG23732T mo |- --------------------------- Siccibacter colletis LMG282041
------------------------------- Siccibacter turicensis LMG237301 loo |- ---------- Citrobacter freundii ATCC8090T '- ------ -Citrobacteryoungae ATCC29220T Citrobacter rodentium ICC168
------ Citrobacter amalonaticus ASM73105v -------- Yokenella regensburgei ATCC43003 ---------------- Enterobacter aerogenes KCTC2190T ------------------- Enterobacter asburiae LI ----------------- Enterobacter cancerogenus ATCC35316 ------ Enterobacter ludwigii ATCC491621 looi— Enterobacter hormaechei ATCC491627 I— 8701 -8 7 0 6 — Enterobacter cloacae ATCC13047T
- Enterobacter massiliensis JC163
i------0.005
H
1195
REEVALUATION OF C. SA KA ZAK II OUTBREAK
J. Food Prot., Vol. 78, No. 6
TABLE 2. Average nucleotide identity (ANI) values for strains 8756, 8706, and 8701 compared with Cronobacter sakazakii and a variety of Enterobacter speciesa Species
C. sakazakii E. asburiae E. hormaechei Enterobacter cancerogenus E. cloacae E. ludwigii Enterobacter aerogenes
Strain 8756 (PT1) 8706 (PT2) 8701 (PT3) ATCC 295441 LI ATCC 491 62t ATCC 35316 ATCC I 3047t ECWSU1 KCTC2190t
8756 (PT 1; %)
82.92 83.01 9 8 .2 2
83.20 83.10 82.99 82.80 82.78 83.07
8706 (FT2; %)
88.93 84.00 9 1 .2 8
88.88 87.84 89.18 88.76 83.65
8701 (PT3; %)
83.67 89.15 9 5 .5 9
88.09 88.52 87.85 83.64
a The highest ANI value for each strain is shown in bold. PT, pulsotype.
The phylogenetic analyses showed that strain 8701 (PT3), from infant 2, clustered with E. hormaechei, supported by bootstrap values of 100% on both trees, (Figs. 1 and 2). The ANI analysis showed a 95.59% match with E. hormaechei ATCC 49162T. An ANI value of >95% has been suggested for species demarcation (7). The high ANI value and strongly supported phylogenetic analyses lead to the conclusion that strain 8701, isolated from a fecal sample from infant 2, is E. hormaechei. As reported previously, the isolates from PT3 were resistant to several antibiotics, including cefotaxime and chloramphenicol (3). High antimicrobial resistance has been reported for E. hormaechei strains, which had been isolated from neonates and misidentified as C. sakazakii (19). Because the PT3 isolates were isolated from both patients, but not from PIF or R-PIF, we are uncertain of the source. The Enterobacter spp. isolates belonging to PT2 were isolated from PIF, R-PIF, and both patients (3). The exact species could not be detennined, as fusA allele 81 has not been linked to a particular Enterobacter species. The phylogenetic analysis based on six loci showed strain 8706 clustered with E. ludwigii, with low bootstrap support (Fig. 1). The rMLST analysis, based on 53 loci, showed strain 8706 clustered near E. hormaechei, but on its own branch, supported by high bootstrapping values (Fig. 2). The closest ANI match to strain 8706 was E. asburiae LI, with a value of 91.28% (Table 2). This value is not high enough to positively identify strain 8706 as E. asburiae. The heteroge neity of the Enterobacter cloacae complex did not allow for identification based on 16S rRNA sequences, and the species of this isolate remains unknown (8). Though the isolate could not be identified to the species level, these analyses indicate that it is not C. sakazakii, as originally reported. This reevaluation shows the importance of accurate detection and identification methods. The phenotypic and PCR-based methods used in the original outbreak investi gation can be unreliable for identification of Cronobacter spp. (2, 9—11, 19). The PCR probe methods utilized by Flores et al. (3) for species identification targeted the 16S rRNA and rpoB genes, both of which have exhibited problems in the past (2, 10, 11). The 16S rRNA PCR method detects only “E. sakazakii,” and the method targeting rpoB has been shown to produce false-positive
results with strains belonging to the newly defined genus Franconibacter (10, 14, 18, 19). Though the true cause of the infant illnesses cannot be determined, it is suspected that they were caused by E. hormaechei, as this species has been implicated in previous neonatal intensive care unit outbreaks (20, 21). Addition ally, Enterobacter spp., including E. hormaechei, have been previously isolated from PIF and have been misidentified as “E. sakazakii,” the fonner name for Cronobacter spp. (9, 17, 20). Accurate identification is essential for tracing out breaks, and standards must be written that will ensure protection from the hazards present in food products. As the Enterobacter spp. isolates belonging to PT2 were obtained from the unopened tins of PIF and patient samples, PIF is suspected as a possible source of these isolates. It is important to note that during the outbreak investigation in Mexico, it was found that the water used to rehydrate the PIF was only heated to approximately 40°C, which is not hot enough to kill microorganisms present (3). This oversight may have exposed the infants to the pathogens, bringing preparation practices and the microbiological criteria specified in PIF standards into question. Currently, Cronobacter species and Salmonella species are the only organisms with “ clear evidence of causality (category A)’’ in illnesses linked to PIF (5). Category B organisms (“ causality plausible, but not yet demonstrated” ) include Citrobacter koseri, Citrobacter freundii, Enterobacter cloacae, Escherichia vulneris, Escherichia coli, Hafnia alvei, Klebsiella pneumoniae, Klebsiella oxytoca, Pantoea agglomerans, Serratia spp., and Acinetobacter spp. (5, 6). The designation of some organisms may need to be revised to reflect outbreaks that have now been linked to species other than Cronobacter and Salmonella. The outbreak in Mexico demonstrates that species currently permitted in PIF can be transmitted to infants when appropriate measures, such as heating reconstitution water to at least 70°C, are not taken (4-6). Opportunistic pathogens in the PIF may, then, go on to cause illnesses in a susceptible host. For example, problems with infection control measures may have contributed to an outbreak of E. hormaechei in Pennsylvania (21). It was found that health care workers were not changing their gloves between
1196
JACKSON ET AL.
patients and that hand washing facilities were sometimes not properly stocked, which may have- contributed to cross contamination between the infants. Cross-contamination may also have occurred in the hospital in Mexico. C. sakazakii ST297 was recovered from PIF and the hospital’s bottle-washing sink, suggesting cross contamination from the PIF. The possible cross-contamina tion, coupled with the use of reconstitution water heated to approximately 40°C, suggests the potential for an outbreak, though no illnesses were reported at the time. In conclusion, suspected Cronobacter infant isolates from an outbreak in Mexico were identified as two species of Enterobacter. Though the species responsible for the symptoms could not be determined, this outbreak demon strated possible transmission of Enterobacter spp. from PIF to infants. Additionally, attention must be paid to infection control measures in hospitals to prevent exposure to and spread of pathogens. ACKNOWLEDGMENTS This work was partially supported by DIUBB through project 143720 and a Santander Research Mobility Grant. This publication made use of the Cronobacter Multilocus Sequence Typing Web site (http://pubmlst.org/ cronobacter/), developed by Keith Jolley and sited at the University of Oxford. The development of this site has been funded by the Wellcome Trust.
J. Food Prot., Vol. 78, No. 6
8.
9.
10.
11.
12.
13.
14.
15.
REFERENCES 1.
Baldwin, A., M. Loughlin, J. Caubilla-Barron, E. Kucerova, G. Manning, C. Dowson, and S. Forsythe. 2009. Multilocus sequence typing of Cronobacter sakazakii and Cronobacter malonaticus reveals stable clonal structures with clinical significance which do not correlate with biotypes. BMC Microbiol. 9:223. doi: 10.1186/ 1471-2180-9-223. 2. Cetinkaya, E., S. Joseph, K. Ayhan, and S. J. Forsythe. 2013. Comparison of methods for the microbiological identification and profiling of Cronobacter species from ingredients used in the preparation of infant formula. Mol. Cell. Probes 27:60-64. 3. Mores, J. P., S. A. Medrano, J. S. Sanchez, and E. Femandez-Escartin. 2011. Two cases of hemorrhagic diarrhea caused by Cronobacter sakazakii in hospitalized nursing infants associated with the consump tion of powdered infant formula. J. Food Prot. 74:2177-2181. 4. Food and Agricultural Organization of the United Nations, World Health Organization. 2004. Enterobacter sakazakii and other organisms in powdered infant formula. Meeting report. Microbiolog ical risk assessment series. Meeting report. Available at: http://www. who.int/foodsafety/publications/mra6-enterobacter-sakazakii/en/. Ac cessed 19 February 2015. 5. Food and Agricultural Organization of the United Nations, World Health Organization. 2006. Enterobacter sakazakii and Salmonella in powdered infant formula. Meeting report. Microbiological risk assessment series. Available at: http://www.who.int/foodsafety/ publications/mralO/en/. Accessed 19 February 2015. 6. Food and Agricultural Organization of the United Nations, World Health Organization. 2008. Enterobacter sakazakii (Cronobacter spp.) in powdered formulae. Meeting report. Microbiological risk assessment series. Available at: http://www.who.int/foodsafety/ publications/mra_followup/en/. Accessed 19 February 2015. 7. Goris, J., K. T. Konstantinidis, J. A. Klappenbach, T. Coenye, P. Vandamme, and J. M. Tiedje. 2007. DNA-DNA hybridization values
16.
17.
18.
19.
20.
21.
and their relationship to whole-genome sequence similarities. Int. J. Syst. Evol. Microbiol. 57:81-91. Hoffmann. H., and A. Roggenkamp. 2003. Population genetics of the nomenspecies Enterobacter cloacae. Appl. Environ. Microbiol. 69: 5306-5318. Ivy, R. A., J. M. Farber, F. Pagotto, and M. Wiedmann. 2013. International Life Science Institute North America Cronobacter (formerly Enterobacter sakazakii) isolate set. J. Food Prot. 76:40-51. Jackson, E. E., H. Sonbol, N. Masood, and S. J. Forsythe. 2014. Genotypic and phenotypic characteristics of Cronobacter species, with particular attention to the newly reclassified species Cronobacter helveticus, Cronobacter pulveris, and Cronobacter zurichensis. Food Microbiol. 44:226-235. Jaradat, Z. W., Q. O. Ababneh, I. M. Saadoun, N. A. Samara, and A. M. Rashdan. 2009. Isolation of Cronobacter spp. (formerly Enterobacter sakazakii) from infant food, herbs and environmental samples and the subsequent identification and confirmation of the isolates using biochemical, chromogenic assays, PCR and 16S rRNA sequencing. BMC Microbiol. 9:225. doi:10.1186/1471-2180-9-225. Jolley K A, and M. C. Maiden. BIGSdb: scalable analysis of bacterial genome variation at the population level. BMC Bioinformatics 2010:11:595. Joseph, S., H. Sonbol, S. Hariri, P. Desai, M. McClelland, and S. J. Forsythe. 2012. Diversity of the Cronobacter genus as revealed by multilocus sequence typing../. Clin. Microbiol. 50:3031-3039. Lehner, A., C. Fricker-Feer, and R. Stephan. 2012. Identification of the recently described Cronobacter condimenti by an rpoB-genebased PCR system. J. Med. Microbiol. 61:1034-1035. Lehner, A., K. Riedel, T. Rattei, A. Ruepp, D. Frishman, P. Breeuwer, B. Diep, L. Eberl, and R. Stephan. 2006. Molecular characterization of the a-glucosidase activity in Enterobacter sakazakii reveals the presence of a putative gene cluster for palatinose metabolism. Syst. Appl. Microbiol. 29:609-625. Lehner, A., T. Tasara, and R. Stephan. 2004. 16S rRNA gene based analysis of Enterobacter sakazakii strains from different sources and development of a PCR assay for identification. BMC Microbiol. 4:43. Muytjens, H. L., H. Roelofs-Willemse, and G. H. Jaspar. 1988. Quality of powdered substitutes for breast milk with regard to members of the family Enterobacteriaceae. J. Clin. Microbiol. 26: 743-746. Stephan, R., C. J. Grim, G. R. Gopinath, M. K. Mammel, V. Sathyamoorthy, L. H. Trach, H. R. Chase, S. Fanning, and B. D. Tall. 2014. Re-examination of the taxonomic status of Enterobacter helveticus, Enterobacter pulveris and Enterobacter turicensis as members of the genus Cronobacter and their reclassification in the genera Franconibacter gen. nov. and Siccibacter gen. nov. as Franconibacter helveticus comb, nov., Franconibacter pulveris comb. nov. and Siccibacter turicensis comb, nov., respectively. Int. J. Syst. Evol. Microbiol. 64:3402-3410. Stoop, B.. A. Lehner, C. Iversen, S. Fanning, and R. Stephan. 2009. Development and evaluation of rpoB based PCR systems to differentiate the six proposed species within the genus Cronobacter. Int. J. Food Microbiol. 136:165-168. Townsend, S. M., E. Hurrell, J. Caubilla-Barron, C. Loc-Carrillo, and S. J. Forsythe. 2008. Characterization of an extended-spectrum betalactamase Enterobacter hormaechei nosocomial outbreak, and other Enterobacter hormaechei misidentified as Cronobacter (Enterobac ter) sakazakii. Microbiology 154:3659-3667. Wenger, P. N., J. I. Tokars, P. Brennan, C. Samel, L. Bland. M. Miller, L. Carson, M. Arduino, P. Edelstein, S. Aguero, C. Riddle, C. O ’Hara, and W. Jarvis. 1997. An outbreak of Enterobacter hormaechei infection and colonization in an intensive care nursery. Clin. Infect. Dis. 24:1243-1244.
Copyright of Journal of Food Protection is the property of Allen Press Publishing Services Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
