Genotypic and phenotypic analysis of Enterobacter sakazakii strains from an outbreak resulting in fatalities in a neonatal intensive care unit in France.

Abstract

In 1994, an outbreak of Enterobacter sakazakii infections occurred in a neonatal intensive care unit in France from 5 May to 11 July. During the outbreak, 13 neonates were infected with E. sakazakii, resulting in 3 deaths. In addition, four symptomless neonates were colonized by E. sakazakii. The strains were subjected to 16S rRNA gene sequence analysis, genotyped using pulsed-field gel electrophoresis, and phenotyped for a range of enzyme activities. E. sakazakii was isolated from various anatomical sites, reconstituted formula, and an unopened can of powdered infant formula. A fourth neonate died from septic shock, attributed to E. sakazakii infection, during this period. However, 16S rRNA gene sequence analysis revealed that the organism was Enterobacter cloacae. There were three pulsotypes of E. sakazakii associated with infected neonates, and three neonates were infected by more than one genotype. One genotype matched isolates from unused prepared formula and unfinished formula. However, no pulsotypes matched the E. sakazakii strain recovered from an unopened can of powdered infant formula. One pulsotype was associated with the three fatal cases, and two of these isolates had extended-spectrum beta-lactamase activity. It is possible that E. sakazakii strains differ in their pathogenicities, as shown by the range of symptoms associated with each pulsotype.

Full text

Page 1

JOURNAL OF CLINICAL MICROBIOLOGY, Dec. 2007, p. 3979–3985
0095-1137/07/$08.00?0doi:10.1128/JCM.01075-07
Copyright © 2007, American Society for Microbiology. All Rights Reserved.
Vol. 45, No. 12
Genotypic and Phenotypic Analysis of Enterobacter sakazakii Strains
from an Outbreak Resulting in Fatalities in a Neonatal
Intensive Care Unit in France?
J. Caubilla-Barron,1E. Hurrell,1S. Townsend,1P. Cheetham,1C. Loc-Carrillo,1O. Fayet,2
M.-F. Pre `re,2and S. J. Forsythe1*
School of Biomedical and Natural Sciences, Nottingham Trent University, Clifton Lane, Nottingham, NG11 8NS, United Kingdom,1and
Laboratoire de Microbiologie et Ge ´netique Mole ´culaires, UMR 5100 CNRS et Universite ´ Paul Sabatier Toulouse III,
118 route de Narbonne, 31062 Toulouse cedex 9, France2
Received 25 May 2007/Returned for modification 11 July 2007/Accepted 25 September 2007
In 1994, an outbreak of Enterobacter sakazakii infections occurred in a neonatal intensive care unit in France
from 5 May to 11 July. During the outbreak, 13 neonates were infected with E. sakazakii, resulting in 3 deaths.
In addition, four symptomless neonates were colonized by E. sakazakii. The strains were subjected to 16S rRNA
gene sequence analysis, genotyped using pulsed-field gel electrophoresis, and phenotyped for a range of enzyme
activities. E. sakazakii was isolated from various anatomical sites, reconstituted formula, and an unopened can
of powdered infant formula. A fourth neonate died from septic shock, attributed to E. sakazakii infection,
during this period. However, 16S rRNA gene sequence analysis revealed that the organism was Enterobacter
cloacae. There were three pulsotypes of E. sakazakii associated with infected neonates, and three neonates were
infected by more than one genotype. One genotype matched isolates from unused prepared formula and
unfinished formula. However, no pulsotypes matched the E. sakazakii strain recovered from an unopened can
of powdered infant formula. One pulsotype was associated with the three fatal cases, and two of these isolates
had extended-spectrum ?-lactamase activity. It is possible that E. sakazakii strains differ in their pathogenic-
ities, as shown by the range of symptoms associated with each pulsotype.
Enterobacter sakazakii is an opportunistic pathogen associ-
ated with the ingestion of reconstituted infant formula and is a
rare cause of neonatal meningitis, necrotizing enterocolitis
(NEC), and sepsis (9, 10, 11, 23). Such cases often occur
among low-birth-weight preterm neonates, who are generally
more susceptible to gram-negative bacterial sepsis and endo-
toxemia associated with NEC (1, 26). The International Com-
mission on Microbiological Specifications for Foods (14) has
ranked E. sakazakii as a “severe hazard for restricted popula-
tions, life-threatening or substantial chronic sequelae or long
duration.” A number of reported E. sakazakii outbreaks have
been attributed to contaminated reconstituted infant formula
(4, 7, 13, 18, 31). Bowen and Braden (4) reviewed 46 cases of
invasive E. sakazakii infections and showed a link between
symptoms and birth weight but did not consider cases of NEC.
The virulence of E. sakazakii has been studied by Pagotto et
al. (23) and Mange et al. (21), who showed the presence of
enterotoxinsand adhesion
Townsend et al. demonstrated the translocation of E. sakazakii
and other intestinal bacteria across the rat intestinal wall in
response to the presence of lipopolysaccharide (28). They also
demonstrated that E. sakazakii causes chronic-patterned in-
flammation in the neonatal rat brain, invades capillary endo-
thelial brain cells, is taken up by macrophages, and induces
anti-inflammatory cytokine (interleukin-10) expression in vitro
to brain cells,respectively.
and in vivo at various levels according to strain (29). However,
these publications did not report the individual case details
associated with the isolates under study. Therefore, it is not
possible to directly consider correlations between in vitro and
in vivo studies.
This study analyzed 31 E. sakazakii strains isolated over a
3-month period in 1994 during a large E. sakazakii outbreak in
a neonatal intensive care unit (NICU) in France. This paper
primarily considers the genetic and phenotypic diversity of the
isolates and is not an epidemiological investigation. However,
where appropriate, neonatal details have been included. The
strains have been identified using 16S rRNA gene sequence
analysis, genotyped using pulsed-field gel electrophoresis
(PFGE), and phenotyped for a range of enzyme activities. In
addition, antibiograms and determination of extended-spec-
trum ?-lactamase (ESBL) production have been undertaken.
MATERIALS AND METHODS
Background. An outbreak of NEC and invasive infections occurred during the
period from 5 May to 15 July 1994 in a NICU. An internal hospital investigation,
which was not published, determined that the outbreak was due to E. sakazakii.
Details of the neonates from whom E. sakazakii was isolated are given in Table
1. All neonates were fed reconstituted powdered infant formula via enteral
perfusion. Formula was reconstituted every 24 h and stored in the refrigerator
until required. Every 4 to 6 h, syringes were filled from stored formula, and the
contents were fed to infants through enteric tubes at ambient temperature.
NEC case definition. Bell’s staging of NEC as modified by Walsh and Klieg-
man (32) was used by clinicians in the hospital. Infants with stage I disease (NEC
I) have suspected NEC, with systemic signs of temperature instability, apnea,
bradycardia, and lethargy. Stage II (definite NEC) is characterized by additional
symptoms of absence of bowel sounds, radiological signs of intestinal dilation,
ileus, and pneumatosis intestinalis. Stage III is advanced NEC, where the neo-
nate is critically ill with metabolic acidosis, neutropenia, signs of generalized
* Corresponding author. Mailing address: School of Biomedical and
Natural Sciences, Nottingham Trent University, Clifton Lane, Notting-
ham, NG11 8NS, United Kingdom. Phone: 115 8483529. Fax: 115
8486636. E-mail: Stephen.forsythe@ntu.ac.uk.
?Published ahead of print on 10 October 2007.
3979

Page 2

peritonitis, and marked tenderness and distension of the abdomen. Neonates
without NEC were also tested for E. sakazakii colonization.
Bacterial strains. In 1994, E. sakazakii strains were isolated from various
anatomical sites of 17 neonates (Table 2). These included sputum, feces, skin,
peritoneal fluid, and conjunctivae. E. sakazakii was also isolated from the re-
mains of prepared formula, unfinished formula collected during the outbreak
period, and an unopened can of powdered infant formula that was collected after
the end of the outbreak period. The putative E. sakazakii colonies were distin-
guished on lactose agar plates by their very mucoid aspect. The hospital identi-
fied the strains as E. sakazakii using API20E (BioMerieux), with gas production
as the confirmatory test. The isolates were kept in long-term storage at ?80°C
until the studies reported here. One strain (strain 766), originally identified as E.
TABLE 1. Clinical description of neonates
NeonateSex
Mode of
delivery
Gestation
age (wk)
Birth wt (g)
Start of
feeding (days)
Onset of
symptoms (days)
Symptom
A
B
C
D
E
F (died)
G
H (died)
I
J (died)
K
L
M
N
O
P
Q
R (died)
F
F
F
M
F
M
F
M
F
M
M
M
F
M
M
M
M
NR
NRa
Vaginal
Vaginal
Caesarean
Caesarean
Vaginal
NR
Vaginal
Caesarean
Caesarean
Caesarean
Caesarean
Vaginal
NR
Caesarean
NR
Vaginal
NR
NR
28
29
32
31
28
NR
31
31
32
30
33
29
NR
33
NR
41
NR
NR
1,200
1,380
1,800
1,470
1,000
NR
1,500
1,570
1,560
1,180
1,450
1,200
NR
2,090
NR
1,600
NR
NR
8
6
2
5
8
NR
1
2
2
6
2
8
NR
4
NR
2
NR
NR
13
18
17
27
28
NR
19
12
15
87
8
16
NR
20
NR
13
NR
No details
NEC II
Asymptomatic
NEC II
Asymptomatic
NEC III
No details
Meningitis
Septicemia
NEC II
NEC I
NEC II
NEC II
Digestive problems
Asymptomatic
Digestive problems
Asymptomatic
Sepsis
aNR, no record.
TABLE 2. PFGE profiles, biotyping, and plasmid and antibiotic profiles of patient, prepared formula, and powdered infant formula isolates
NeonateSymptoms
Isolation date
(day/mo/yr)
Isolation siteStrain no.
PFGE
profile
Antibiograma
A
B
No details
NEC II
23/03/1994
25/04/1994
29/04/1994
09/05/1994
08/06/1994
19/06/1994
07/04/1994
19/04/1994
11/05/1994
12/05/1994
24/05/1994
07/06/1994
09/06/1994
13/06/1994
13/06/1994
13/06/1994
14/06/1994
17/06/1994
26/06/1994
27/06/1994
27/06/1994
01/07/1994
15/06/1994
18/06/1994
20/06/1994
27/06/1994
27/06/1994
11/07/1994
11/07/1994
11/07/1994
04/06/1994
Trachea
Trachea
Trachea
Trachea
Stools
Stools
Peritoneal fluid
Sputum
Trachea
Trachea
Trachea
Trachea
Stools
Stools
Stools
Stools
Conjunctivae
Prepared formula
Skin
Stools
Stools
Stools
Stools
Stools
End of bottle
End of bottle
Prepared formula
Infant formula
Infant formula
Infant formula
Blood culture
699
703
704
708
696
690
701
691
767
709
705
695
706
697
702
692
694
712
707
730
711
698
700
693
713
714
715
716
717
718
766
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
4
4
4
NAb
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX, AMP, CTX, CXM (ESBL)
DOX
DOX
DOX, AMP, CTX, CXM (ESBL)
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX
DOX, AMP, CXM, AMC
C
D
E
F (died)
G
H (died)
C
B
J (died)
B
D
K
L
M
Asymptomatic
NEC II
Asymptomatic
NEC III
No details
Meningitis
Septicemia
NEC II
NEC II
NEC II
NEC II
NEC I
NEC II
NEC II
B
K
O
D
P
Q
NEC II
NEC I
Asymptomatic
NEC II
Digestive problems
Asymptomatic
R (died) Sepsis
aResistance profile for doxycycline (DOX), ampicillin (AMP), cefotaxime (CTX), cefuroxime (CXM), and amoxicillin-clavulanate (AMC).
bNA, not applicable.
3980 CAUBILLA-BARRON ET AL.J. CLIN. MICROBIOL.

Page 3

sakazakii, has been reidentified, by using 16S rRNA gene sequence analysis, as
Enterobacter cloacae.
Bacterial identification and 16S rRNA gene cluster group analysis. The re-
covered bacterial strains were identified using 16S rRNA gene sequence analysis
(Accugenix, Newark, DE) as described by Iversen et al. (15). The sequences were
compared with E. sakazakii sequences to determine the 16S rRNA gene cluster
group.
DNA isolation and PCR. Genomic DNA was prepared using the GenElute
bacterial genomic DNA kit (Sigma) and 1.5 ml of overnight culture grown in LB
broth according to the manufacturer’s instructions. By following methods pre-
scribed by Keyser et al. (19), primers Esak2 (5?-CCCGCATCTCTGCAGGAT
TCTC-3?) and Esak3 (5?-CTAATACCGCATAACGTCTACG-3?) were used to
amplify an 850-bp PCR product from a region of the E. sakazakii 16S rRNA
gene. Lehner et al. (20) used Esakf (5?-GCTYTGCTGACGAGTGGCGG-3?)
and Esakr (5?-ATCTCTGCAGGATTCTCTGG-3?) to amplify a 929-bp PCR
product, also from a region of the E. sakazakii 16S rRNA gene. The ompA gene
was amplified with primers ESSF (5?-GGATTTAACCGTGAACTTTTCC-3?)
and ESSR (5?-CGCCAGCGATGTTAGAAGA-3?), resulting in a 469-bp prod-
uct by using the PCR conditions described by Mohan Nair and Venkitana-
rayanan (22). The PCR protocols documented above were followed as described
in each publication using 2.5 U of GoTaq Flexi DNA polymerase, 5? Green
GoTaq Flexi buffer (Promega Corporation, Madison, WI), and a Genius ther-
mocycler (FGEN05TD; Techne Ltd., Cambridge, United Kingdom). E. sakazakii
strains NCTC 11467Tand ATCC 12868 were used as positive controls. PCR
products were visualized on 1% agarose gels stained with 0.5 ?g ml?1ethidium
bromide.
PFGE. PFGE of E. sakazakii was performed by following the Pulse Net USA
protocol for molecular subtyping of Escherichia coli O157:H7, nontyphoidal
Salmonella serotypes, and Shigella sonnei (6). The gel was run at switch times of
5 to 50 s for 20 h at 6 V in a CHEF-DR II system (Bio-Rad, Hercules, CA).
The PFGE patterns were analyzed by Bionumerics software, version 3.5 (Ap-
plied Maths, Sint-Martens-Latem, Belgium). The patterns were compared and
clustered by the unweighted-pair group method using arithmetic averages
(UPGMA) by using the Dice coefficient. The position tolerance was set to 1.5%,
and an optimization of 1.5% was applied during the comparison of PFGE
fingerprint patterns. PFGE patterns were interpreted according to the criteria of
Tenover et al. (27).
Antibiotic sensitivity testing. The susceptibilities of E. sakazakii to antimicro-
bial agents were determined by the disk diffusion method on Iso-Sensitest agar
(catalog no. CM0471; Oxoid Ltd.) according to the British Society for Antimi-
crobial Chemotherapy protocol (5). The antibiotics tested were amikacin, am-
picillin, cefotaxime, cefuroxime, cefpodoxime, ceftazidime, chloramphenicol,
ciprofloxacin, amoxicillin-clavulanate, doxycycline, gentamicin, imipenem,
piperacillin, and trimethoprim from Oxoid Ltd. UK (Basingstoke, United King-
dom). ESBL production was detected using the combination disc method as
described in HPA QSOP 51 (30) using ceftazidime-clavulanic acid, cefotaxime-
clavulanic acid, and cefpodoxime-clavulanic acid combination discs in compari-
son to individual-antibiotic ceftazidime, cefotaxime, and cefpodoxime discs ac-
cording to the manufacturer’s instructions (Mast Diagnostics, Bootle, United
Kingdom).
Biotyping. The biotype for each strain was determined according to the
Farmer et al. (8) biogrouping scheme as revised by Iversen et al. (17). Standard-
ized biochemical test strips (API20E, ID32, APIZYM) were employed according
to the manufacturer’s instructions (BioMe ´rieux UK). Additional tests of motility,
acid production from sugars, gas production from glucose, and malonate utili-
zation, the methyl red test, the Voges-Proskauer test, and the indole production
test were conducted as previously described. Bacterial isolates were subcultured
on tryptone soy agar (catalog no. 1.05458; Merck KGaA, Darmstadt, Germany)
prior to analysis.
Capsule production. Bacterial cultures were grown overnight at 37°C on milk
agar, which was composed of 3 g of agar (catalog no. LP0011; Oxoid Ltd.,
Basingstoke, United Kingdom) and 0.4 g of ammonium sulfate dissolved into 40
ml of distilled water. After autoclaving at 121°C for 15 min, the mixture was
combined with 200 ml of warm (55°C) liquid infant formula (Premium 1, milk-
based; Cow & Gate, Trowbridge, United Kingdom) and dispensed into petri
dishes. Each strain was evaluated for capsule production by visual comparison
with the colony morphology of E. sakazakii strains 1 (noncapsulated) and 2
(capsulated).
Protease activity. Skim milk powder (2%, wt/vol) was added to plate count
agar (tryptone glucose yeast agar; catalog no. CM325; Oxoid Ltd.) after auto-
claving to make SM-PCA. Plates were inoculated by a simple streak of the
bacteria and incubated at 37°C for 72 h. A positive result was indicated by zones
of clearing around bacterial growth.
Nucleotide sequence accession numbers. The GenBank accession numbers of
the E. sakazakii isolates sequenced in this study are AM778409 to AM778415.
RESULTS
Patients. During the 3-month outbreak period, 18 neonates
were infected or asymptomatically colonized, and there were
four deaths (Table 1). All infected neonates, except for one
(neonate D), were preterm. The average birth weight was 1,461
g, and birth weights ranged from 1,000 to 2,090 g. Nine neo-
nates had severe clinical symptoms: seven cases of NEC (one
[neonate F] with abdominal perforation), one case of septice-
mia (neonate I), and one case of meningitis (neonate H). An
autopsy of the latter neonate revealed cerebral lesions. All
neonates with NEC had birth weights of ?2,000 g. The neo-
nate with meningitis had a birth weight of 1,500 g. The onset of
illness was ?28 days for the majority (17/18) of neonates. The
exception was neonate K, who developed NEC I after 87 days
and had a low birth weight (1,180 g). Three neonates who died
(F, H, and J) had birth weights of 1,000 g, 1,500 g, and 1,560 g,
respectively. The first dates of illness for these neonates were
28, 19, and 15 days after birth. Four other neonates (C, E, O,
and Q) were colonized by E. sakazakii without any clinical
signs, and two neonates (N and P) had moderate digestive
problems.
Identification of bacterial isolates. Thirty-one strains were
recovered from the original outbreak collection. No isolates
were recovered for neonates I and N. All strains, except isolate
766, were confirmed as E. sakazakii using 16S rRNA gene
sequence analysis and were assigned to 16S rRNA gene cluster
group 1 (Table 2). The remaining strain (strain 766) was iden-
tified as E. cloacae. This strain was associated with the death of
neonate R through septic shock during the general E. sakazakii
outbreak and is considered in a separate section below.
PFGE typing of bacterial isolates. Three distinguishable E.
sakazakii pulsotypes were isolated from neonates and recon-
stituted formula. A fourth pulsotype was isolated from an
unopened can of infant formula (Fig. 1). The pulsotypes were
numbered in chronological order.
E. sakazakii pulsotype 1 was isolated from 23 March to 19
June and included strains from two cases of NEC and two
asymptomatic colonizations (Tables 1 and 2). No clinical
records were provided for neonate A, from whom the first
isolate of E. sakazakii was obtained. Six strains belonged to
pulsotype 1 and were isolated from either the trachea or stool
samples.
E. sakazakii pulsotype 2 was isolated from 7 April to 1 July
from neonatal peritoneal fluid, sputum, trachea, stools, con-
junctivae, and skin. This period overlapped with the period of
isolation of pulsotype 1. A total of 16 strains belonged to
pulsotype 2. These were obtained from one case of meningitis,
seven cases of NEC, two asymptomatic colonizations, and un-
used prepared formula. Three neonates (F, H, and J) colo-
nized by E. sakazakii pulsotype 2 died from NEC or meningitis.
Isolates 701, 767, and 695 from these neonates were obtained
from the peritoneal fluid and trachea. E. sakazakii pulsotype 2
strains 705, 706, and 707 were isolated from neonate B on
three occasions over 2 months (24 May, 9 June, and 26 June
1994). These strains were from the trachea, stool, and skin.
Their identical PFGE profiles formed a subcluster within the
VOL. 45, 2007ENTEROBACTER SAKAZAKII FATAL OUTBREAK STRAINS 3981

Page 4

pulsotype 2 cluster. Previously, E. sakazakii pulsotype 1 had
been isolated from the trachea of this neonate on 25 and 29
April 1994. E. sakazakii pulsotypes 1 and 2 were isolated 3 days
apart from neonate C on 9 and 12 May 1994. This neonate was
asymptomatically colonized, and the strains were isolated from
the trachea on both occasions. E. sakazakii pulsotype 1 was
isolated from neonate D on 8 June 1994, and pulsotype 2 was
isolated from the same neonate on 13 June and 1 July 1994. On
each occasion, the organism was isolated from the stool.
Pulsotype 3 was isolated over a 2-week period from 15 to 27
June. This period overlapped with both pulsotypes 1 and 2.
There were five strains in pulsotype 3. Two strains were from
the stools of neonates P and Q. Neonate P had a digestive
problem, and neonate Q was asymptomatic. Two further
strains were obtained 1 week apart (20 and 27 June 1994) from
leftover formula, and on the same day (27 June 1994) from
unused prepared formula.
An unopened can of powdered infant formula was sampled
for bacterial contamination on 11 July. This was after the
neonatal pulsotypes had been isolated. Three strains of E.
sakazakii (strains 716, 717, and 718) were isolated with PFGE
profiles that were more than three bands different from
pulsotypes 1 to 3. They were designated pulsotype 4.
Characterization of pulsotypes. Table 2 compares the phe-
notypic traits of the four pulsotypes. Pulsotypes 1 to 3 corre-
sponded to biotype 13, and pulsotype 4 corresponded to
biotype 5 (indole positive). APIZYM profiles did not differ
between pulsotypes. All strains had high levels (4 to 5 U) of
esterase-lipase and ?-galactosidase activities. Moderate (1 to 2
U) levels of ?-glucosidase activity were detected for all strains.
No proteolytic activity was detected for pulsotype 4 strains
growing on SM-PCA. In addition, pulsotype 4 strains did not
produce capsulated colonies on milk agar plates. Some strains
in pulsotypes 1 to 3 produced such profuse capsular material
that the colonies contacted the inverted petri dish lid.
All E. sakazakii strains were sensitive to ciprofloxacin, ami-
kacin, gentamicin, imipenem, piperacillin, and trimethoprim
and resistant to doxycycline. All but two strains (strains 695
and 767) of E. sakazakii were sensitive to ampicillin, cefo-
taxime, and cefuroxime. Strains 695 and 767 were also resistant
to cefpodoxime, ceftazidime, and chloramphenicol. Further
analysis using the combination disc method recommended by
FIG. 1. Dendrogram generated from PFGE profiles of E. sakazakii isolates (n ? 30) by Bionumerics software, version 3.5. Clustering was done
with UPGMA by using the Dice coefficient. Pulsotypes are identified on the left. The tolerance in the band was 1.5%, with an optimization of 1.5%.
3982CAUBILLA-BARRON ET AL.J. CLIN. MICROBIOL.

Page 5

the HPA (30) demonstrated that these strains possessed
ESBLs. These two strains, both pulsotype 2, were isolated in
two fatal cases (neonates H and J).
E. cloacae. Strain 766 was isolated as a blood culture from
neonate R, who died from septic shock. It was identified in
1994, using API20E, as E. sakazakii. However 16S rRNA
gene sequence analysis revealed that it was E. cloacae and
had been misidentified. Several methods for the identifica-
tion of E. sakazakii via PCR have been described and were
used in this study (19, 20, 22). PCR products of the indicative
size were obtained when genomic DNAs from E. sakazakii
strains NCTC 11467Tand ATCC 12868 were probed as previ-
ously described. However, no corresponding PCR product was
obtained for strain 766. This further confirmed that strain 766
was not E. sakazakii. Due to the lack of ?-glucosidase activity,
the organism did not produce characteristic blue-green E.
sakazakii colonies on the chromogenic E. sakazakii agar DFI
(catalog no. CM1055; Oxoid Ltd., Basingstoke, United King-
dom). It also failed to produce any yellow pigmentation on
tryptone soy agar after 48 h of incubation at 21°C. The
APIZYM profile differed from that of E. sakazakii by the lack
of detectable esterase and esterase-lipase activities. Antibio-
gram profiling showed that the strain was resistant to doxycy-
cline, ampicillin, cefuroxime, and amoxicillin-clavulanate.
DISCUSSION
This is the largest outbreak of E. sakazakii in a NICU, with
the most deaths, that has been reported to date. Unfortunately,
because of the long time between the outbreak and this study,
a number of clinical details were not available to us. Therefore,
this study is not an epidemiological outbreak investigation;
instead, it has focused on the diversity of the isolates. Never-
theless, neonatal information regarding gestation age, birth
weight, age of onset of symptoms, etc., are reported here (Ta-
ble 1) and will be of use for future studies on the risk factors for
E. sakazakii infections.
This study shows that, over the period from March to July
1994, neonates were infected by three pulsotypes of E. saka-
zakii for which no definitive sources were identified (Table 2).
The first pulsotype was isolated from March to June. It was
isolated from five neonates (A to E), including two cases of
NEC II, no fatal cases, and two asymptomatic neonates. The
two NEC II cases were cocolonized by pulsotype 2. Unfortu-
nately, no records were available for the first case, neonate A.
Pulsotype 2 was isolated from April to July. It was associated
with six NEC II or NEC III cases, one suspected NEC case,
one case of meningitis, and two asymptomatic neonates. There
were a total of three deaths. This pulsotype was also isolated
from prepared formula but not from an unopened can of
powdered infant formula. Pulsotype 3 was isolated over a
2-week period in June from two neonates: neonate O, with a
digestive problem, and neonate P, who was asymptomatic. It
was also isolated from unused formula and two leftover for-
mulas. Pulsotypes 1 to 3 were not isolated from unopened cans
of powdered infant formula analyzed in July, but only from
prepared formula on 17 and 27 June 1994 (Table 2). Pulsotype
3 was isolated from unfinished formula on two occasions (20
and 27 June 1994) and from prepared formula on 27 June
1994. No unopened cans of powdered infant formula were
tested at this time, and therefore the source of contamination
remains uncertain. E. sakazakii was isolated from an unopened
can of powdered infant formula (sample date, 11 July 1994).
This was 2 weeks after the last case. These strains had a unique
pulsotype. Previously, E. sakazakii has been isolated from in-
fant formula preparation equipment, the environment, and the
human throat (2, 12, 25). Therefore, given the ubiquity of E.
sakazakii, unattributable sources of E. sakazakii could have
been caregivers, cross-contamination from other neonates, the
environment, and batches of reconstituted infant formula. It is
not known if any neonates were fed the batch of powdered
infant formula from which E. sakazakii pulsotype 4 was iso-
lated (sample date, 11 July 1994), since there were no neonatal
cases in July with that pulsotype.
Multiple E. sakazakii pulsotypes were isolated from neo-
nates B, C, and D. Five isolates from the trachea, stool sam-
ples, and skin were obtained from neonate B over a 2-month
period (25 April to 26 June 1994) (Table 2). The three pulso-
type 2 strains (strains 705, 706, and 707) were from the trachea,
stools, and skin, respectively. Due to their identical PFGE
profiles, they formed a subcluster within the pulsotype 2 cluster
(Fig. 1). This indicates that the neonate was colonized by a
specific E. sakazakii strain from 24 May to 26 June. Neonate C
was asymptomatic. Pulsotype 1 and 2 strains 708 and 709 were
isolated 3 days apart (9 and 12 May 1994), both from the
trachea. Neonate D suffered from NEC II. Pulsotype 1 and 2
strains 696 to 698 were isolated from stool samples between 8
June and 1 July 1994. Therefore, multiple isolates from neo-
nates should be pulse typed; otherwise, clinical strains may not
be matched with source isolates. Prompt analysis of possible
sources is required, especially for substances such as powdered
infant formula, since many batches may be in use over a short
period of time.
The NICU feeding practices of reconstituting formula every
24 h and administering the formula over a 4- to 6-h period are
not currently recommended. After the outbreak, the following
internal recommendations were made: to chill the enteral feed-
ing syringe using cryogel and to change the syringe and syringe
end every 3 h. These are very similar to the 2-h maximum
period between preparation and administration recommended
by the FAO-WHO (9).
It should be noted that in 1994 E. sakazakii was not a well-
recognized neonatal pathogen and was not associated with
contaminated reconstituted infant formula. Therefore, the
hospital investigation was not initially focused on the infant
formula. The Belgian outbreak that associated E. sakazakii
with powdered infant formula, and in which two neonates died,
occurred in 1996, and the report of that outbreak was not
published until 2001 (31).
All strains were doxycycline resistant. Doxycycline is a tet-
racycline antibiotic that is not used for this patient group and
therefore has no clinical relevance. Cefotaxime and cefuroxime
are commonly used first-line treatments for neonates. There-
fore, the antibiograms of strains 767 and 695, from fatal cases,
are of particular interest. Both these strains had ESBL activity,
which may have been acquired by horizontal transfer from
other Enterobacteriaceae, since it was absent in the other pul-
sotype 2 strains. There are more than 30 variants of CTX-M
ESBLs, some of which have evolved from the chromosomal
?-lactamases of Kluyvera species (3). Cefpirome is stable to the
VOL. 45, 2007 ENTEROBACTER SAKAZAKII FATAL OUTBREAK STRAINS 3983

Page 6

inducible AmpC chromosomal ?-lactamases of Enterobacter
species, which give a false-positive ESBL result with other
cephalosporins that are used. The use of cephalosporins is not
recommended for the treatment of Enterobacter species infec-
tions, due to the selection of AmpC-derepressed mutants. E.
cloacae strain 766 was also from a fatal case and was resistant
to ampicillin, cefuroxime, and amoxicillin-clavulanate.
All E. sakazakii strains were in 16S rRNA gene cluster group
1, the largest of the four E. sakazakii genotypes (16). Pulsotype
2 isolates were from eight cases of NEC, one case of septice-
mia, and one case of meningitis. A total of three deaths were
associated with this pulsotype, and therefore it appears to be
more virulent than the other pulsotypes. In addition, two pulso-
type 2 strains (strains 696 and 767) from neonates H and J, who
died, had acquired ESBL activities. Pulsotype 4 strains, iso-
lated from an unopened can of powdered infant formula, dif-
fered from the other pulsotypes by the lack of protease activity
on SM-PCA and the absence of capsule production. Whether
these phenotypes are related to virulence determinants is un-
known, but the question is currently under investigation.
Strain 766 was initially identified as E. sakazakii by using
API20E and was isolated in a fatal case of septic shock (4 June
1994) during the NICU E. sakazakii outbreak (Table 1). How-
ever, 16S rRNA gene sequencing revealed that it did not fall
within any of the E. sakazakii cluster groups. Several PCR
protocols for the specific amplification of E. sakazakii have
been developed that amplify the 16S rRNA gene. More re-
cently, ompA-specific PCR primers were shown to distinguish
E. sakazakii from similar organisms (22). Our study used these
published protocols and PCR probes in order to obtain a
presumptive identification of strain 766 and to further under-
stand at what level the misidentification as E. sakazakii may
have occurred. While strains NCTC 11467Tand ATCC 12868
were correctly identified by these techniques, no PCR product
was obtained from strain 766.
Discrepancies in E. sakazakii identification between com-
mercial biochemical kits have been noted previously (15, 24).
Therefore, 16S rRNA gene sequence analysis was used as our
“gold standard” for identification. Commercially available spe-
cific E. sakazakii chromogenic agars are designed to detect
?-glucosidase activity. Therefore, E. cloacae strain 766 (?-glu-
cosidase negative) would not have been misidentified by rou-
tine use of these agars. Nor would the strain be regarded as a
presumptive E. sakazakii strain by use of PCR probes. This
further shows the need for rigorous testing and careful consid-
eration of rapid phenotypic tests.
This investigation highlights the need for accurate isolate
identification and prompt typing of neonatal and associated
isolates. The source of the three pulsotypes could not definitely
be identified as contaminated reconstituted infant formula.
Nevertheless, the feeding practices of preparation for 24-h
periods and prolonged (?2-h) administration could have en-
abled bacterial growth in the formula and increased the risk of
infection.
ACKNOWLEDGMENTS
We thank Stacey Brown for technical assistance, Julie Rainard for
translating documents, and Anna Bowen (CDC, Atlanta, GA) for
helpful discussions.
We also thank Nottingham Trent University, the Centre National de
la Recherche Scientifique, and the Universite ´ Paul Sabatier for finan-
cial support.
REFERENCES
1. Beck-Sague, C., P. Azimi, S. Fonseca, R. Baltimore, D. Powell, L. Bland, M.
Arduino, S. McAllister, R. Huberman, R. Sinkowitz, et al. 1994. Bloodstream
infections in neonatal intensive care unit patients: results of a multicenter
study. Pediatr. Infect. Dis. J. 13:1110–1116.
2. Block, C., O. Peleg, N. Minster, B. Bar-Oz, A. Simhon, I. Arad, and M.
Shapiro. 2002. Cluster of neonatal infections in Jerusalem due to unusual
biochemical variant of Enterobacter sakazakii. Eur. J. Clin. Microbiol. Infect.
Dis. 21:613–616.
3. Bonnet, R. 2004. Growing group of extended-spectrum beta-lactamases: the
CTX-M enzymes. Antimicrob. Agents Chemother. 48:1–14.
4. Bowen, A. B., and C. R. Braden. 2006. Invasive Enterobacter sakazakii disease
in infants. Emerg. Infect. Dis. 12:1185–1186.
5. British Society for Antimicrobial Chemotherapy. 2006. BSAC methods for
antimicrobial susceptibility testing, version 5. http://www.bsac.org.uk/_db
/_documents/version_5_.pdf.
6. CDC. June 2004. Pulse Net USA. One-day (24–48 h) standardized labo-
ratory protocol for molecular subtyping of Escherichia coli O157:H7,
non-typhoidal Salmonella serotypes, and Shigella sonnei by pulsed
field gel electrophoresis (PFGE). http://www.cdc.gov/pulsenet/protocols
/ecoli_salmonella_shigella_protocols.pdf.
7. Coignard, B., V. Vaillant, J.-P. Vincent, A. Lefle `che, P. Mariani-Kurkdjian,
C. Bernet, F. L’He ´riteau, H. Se ´ne ´chal, P. Grimont, E. Bingen, and J.-C.
Desenclos. 2006. Infections se ´ve `res a ` Enterobacter sakazakii chez des
nouveau-ne ´s ayant consomme ´ une pre ´paration en poudre pour nourrissons,
France, octobre-de ´cembre 2004. Bull. Epide ´miol. Hebdomadaire 2–3:10–13.
http://www.invs.sante.fr/beh/2006/02_03/beh_02_03_2006.pdf.
8. Farmer, J. J., III, M. A. Asbury, F. W. Hickman, D. J. Brenner, and The
Enterobacteriaceae Study Group. 1980. Enterobacter sakazakii: a new species
of “Enterobacteriaceae” isolated from clinical specimens. Int. J. Syst. Bacte-
riol. 30:569–584.
9. Food and Agriculture Organization–World Health Organization (FAO-WHO).
2004. Enterobacter sakazakii and other microorganisms in powdered infant
formula. Meeting report, MRA series 6. World Health Organization,
Geneva, Switzerland. http://www.who.int/foodsafety/publications/micro
/mra6/en/index.html.
10. Food and Agriculture Organization–World Health Organization (FAO-
WHO). 2006. Enterobacter sakazakii and Salmonella in powdered infant
formula. Meeting report, MRA series 10. World Health Organization,
Geneva, Switzerland. http://www.who.int/foodsafety/publications/micro
/mra10/en/index.html.
11. Forsythe, S. J. 2005. Enterobacter sakazakii and other bacteria in powdered
infant milk formula. Maternal Child Nutr. 1:44–50.
12. Gosney, M. A., M. V. Martin, A. E. Wright, and M. Gallagher. 2006. Entero-
bacter sakazkaii in the mouths of stroke patients and its association with
aspiration pneumonia. Eur. J. Intern. Med. 17:185–188.
13. Himelright, I., E. Harris, V. Lorch, and M. Anderson. 2002. Enterobacter
sakazakii infections associated with the use of powdered infant formula—
Tennessee, 2001. JAMA 287:2204–2205.
14. International Commission on Microbiological Specifications for Foods.
2002. Microorganisms in foods 7. Microbiological testing in food safety
management. Kluwer Academic/Plenum Publishers, New York, NY.
15. Iversen, C., P. Druggan, and S. Forsythe. 2004. A selective differential
medium for Enterobacter sakazakii, a preliminary study. Int. J. Food Micro-
biol. 96:133–139.
16. Iversen, C., M. Waddington, S. L. W. On, and S. Forsythe. 2004. Identifica-
tion and phylogeny of Enterobacter sakazakii relative to Enterobacter and
Citrobacter. J. Clin. Microbiol. 42:5368–5370.
17. Iversen, C., M. Waddington, J. J. Farmer III, and S. Forsythe. 2006. The
biochemical differentiation of Enterobacter sakazakii genotypes. BMC Mi-
crobiol. 6:94.
18. Jarvis, C. 2005. Fatal Enterobacter sakazakii infection associated with pow-
dered infant formula in a neonatal intensive care unit in New Zealand.
Am. J. Infect. Control 33:e19.
19. Keyser, M., R. C. Witthuhn, L. C. Ronquest, and T. J. Britz. 2003. Treatment
of winery effluent with upflow anaerobic sludge blanket (UASB)-granular
sludges enriched with Enterobacter sakazakii. Biotechnol. Lett. 25:1893–1898.
20. 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.
21. Mange, J. P., R. Stephan, N. Borel, P. Wild, K. S. Kim, A. Pospischil, and A.
Lehner. 2006. Adhesive properties of Enterobacter sakazakii to human epi-
thelial and brain microvascular endothelial cells. BMC Microbiol. 6:58.
22. Mohan Nair, M. K., and K. S. Venkitanarayanan. 2006. Cloning and se-
quencing of the ompA gene of Enterobacter sakazakii and development of an
ompA-targeted PCR for rapid detection of Enterobacter sakazakii in infant
formula. Appl. Environ. Microbiol. 72:2539–2546.
23. Pagotto, F. J., M. Nazarowec-White, S. Bidawid, and J. M. Farber. 2003.
3984 CAUBILLA-BARRON ET AL.J. CLIN. MICROBIOL.

Page 7

Enterobacter sakazakii: infectivity and enterotoxin production in vitro and in
vivo. J. Food Prot. 66:370–377.
24. Restaino, L., E. W. Frampton, W. C. Lionberg, and R. J. Becker. 2006. A
chromogenic plating medium for the isolation and identification of Enter-
obacter sakazakii from foods, food ingredients, and environmental sources. J.
Food Prot. 69:315–322.
25. Smeets, L. C., A. Voss, H. L. Muytjens, J. F. G. M. Meis, and W. J. G.
Melchers. 1998. Genetische karakterisatie van Enterobacter sakazakii—iso-
laten van Nederlandse patie ¨nten met neonatale meningitis. Ned. Tijdschr.
Med. Microbiol. 6:113–115.
26. Stoll, B. J., N. Hansen, A. A. Fanaroff, and J. A. Lemons. 2004. Enterobacter
sakazakii is a rare cause of neonatal septicemia or meningitis in VLBW
infants. J. Pediatr. 144:821–823.
27. Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Murray,
D. H. Persing, and B. Swaminathan. 1995. Interpreting chromosomal DNA
restriction patterns produced by pulsed-field gel electrophoresis: criteria for
bacterial strain typing. J. Clin. Microbiol. 33:2233–2239.
28. Townsend, S., J. Caubilla Barron, C. Loc-Carrillo, and S. Forsythe. 2007.
The presence of endotoxin in powdered infant formula milk and the influ-
ence of endotoxin and Enterobacter sakazakii on bacterial translocation in
the infant rat. Food Microbiol. 24:67–74.
29. Townsend, S. M., E. Hurrell, I. Gonzalez-Gomez, J. Lowe, J. G. Frye, S.
Forsythe, and J. L. Badger. 2007. Enterobacter sakazakii invades brain cap-
illary endothelial cells, persists in human macrophages influencing cytokine
secretion and induces severe brain pathology in the neonatal rat. Microbi-
ology 153:3538–3547.
30. United Kingdom Health Protection Agency. 2006. Laboratory detection and
reporting of bacteria with extended spectrum ?-lactamases. QSOP 51.
Health Protection Agency, London, United Kingdom.
31. van Acker, J., F. De Smet, G. Muyldermans, A. Bougatef, A. Naessens, and
S. Lauwers. 2001. Outbreak of necrotizing enterocolitis associated with En-
terobacter sakazakii in powdered milk formula. J. Clin. Microbiol. 39:293–
297.
32. Walsh, M. C., and R. M. Kliegman. 1986. Necrotizing enterocolitis: treat-
ment based on staging criteria. Pediatr. Clin. N. Am. 33:179–201.
VOL. 45, 2007 ENTEROBACTER SAKAZAKII FATAL OUTBREAK STRAINS 3985