Page 1
Phylogeny
and Disease
Association of Shiga
Toxin–producing
Escherichia coli O91
Alexander Mellmann, Angelika Fruth,
Alexander W. Friedrich, Lothar H. Wieler,
Dag Harmsen, Dirk Werber, Barbara Middendorf,
Martina Bielaszewska, and Helge Karch
The diversity and relatedness of 100 Shiga toxin–pro-
ducing Escherichia coli O91 isolates from different patients
were examined by multilocus sequence typing. We identi-
fied 10 specific sequence types (ST) and 4 distinct clonal
groups. ST442 was significantly associated with hemolytic
uremic syndrome.
Shiga toxin–producing Escherichia coli (STEC) in-fections are public health concerns because of the se-
vere illnesses they cause, such as hemorrhagic colitis and
hemolytic uremic syndrome (HUS) (1). STEC constitute a
heterogeneous group of bacteria abundant in the reservoir
and in the environment (2). Transmission routes for human
STEC infection are numerous and include contact with ani-
mal excreta, person-to-person transmission, and inadvertent
ingestion of contaminated food and water. Many STEC se-
rotypes have been recovered from humans (3,4). Among
them, STEC O91 is the most common serogroup isolated
from adult patients in Germany (5,6). The strains within this
serogroup appear to be transmitted predominantly by food,
because 1) food vehicles have been identified as the only
risk factors for adults with sporadic STEC O91 infection in
Germany (6); 2) O91 is the second most frequently isolated
STEC serogroup in routine food samples (5); and 3) O91
is the only major STEC serogroup with no association be-
tween incidence of human infection and cattle density (7).
Whereas most human disease STEC serogroups pos-
sess, in addition to Shiga toxin, the eae gene encoding the
adhesin intimin (3,4,8), STEC O91 consistently lack this
virulence determinant (8,9). Despite frequent isolation of
STEC O91 from humans, the clonal relatedness of the se-
rotypes of this serogoup is poorly understood. Therefore,
we investigated 100 human STEC O91 isolates to deter-
mine the clonal structure of STEC O91 and its association
with disease.
The Study
A total of 100 STEC O91 isolates were obtained from
1997 through 2007 from patients with HUS (n = 4), bloody
diarrhea (n = 8), watery diarrhea without visible blood (n =
79), abdominal cramps without diarrhea (n = 1), or from as-
ymptomatic carriers (n = 8); samples were from Germany
(n = 96), Austria (n = 2; Austrian Reference Library, Inns-
bruck, Austria), Finland (n = 1; The National Public Health
Institute, Helsinki, Finland), and Canada (n = 1; Public
Health Agency of Canada, Guelph, Ontario, Canada). The
96 German O91 strains were recovered at the Institute of
Hygiene, University of Münster, Münster, and the Robert
Koch Institute, Wernigerode, Germany. The strains includ-
ed all human isolates of this serogroup that were recovered
during the study period in Germany and for which com-
plete clinical information was available. The strains corre-
spond to all O91 serotypes associated with human diseases
from sporadic cases in Germany in that interval. Thirty-five
strains have been described previously (4,8,10).
The age of patients from whom the STEC O91strains
originated ranged from 4 months to 89 years (median 28
years, interquartile range 12–38 years). The most severe
symptom was recorded for each patient. Diarrhea was
defined as ≥3 semisolid or liquid stools per day. Bloody
diarrhea was defined as diarrheal stools containing blood
visible to the naked eye. HUS was defined as a case of mi-
croangiopathic hemolytic anemia (hematocrit <30% with
peripheral evidence of intravascular hemolysis), thrombo-
cytopenia (platelet count <150,000/mm3), and renal insuf-
ficiency (serum creatinine concentration greater than the
upper limit of normal for age) (11). Asymptomatic carri-
ers were apparently healthy persons without diarrhea; their
stools were submitted as noted above.
Strains were isolated using Shiga toxin–encoding
genes as diagnostic targets (12) and then serotyped phe-
notypically (13). All strains were verified as O91 by using
PCR targeting wzy
O91
, a component of the rfb gene cluster
that synthesizes the O91 antigen (14). Multilocus sequence
typing (MLST) and phylogenetic analysis were performed
as described (4). All allelic sequences were deposited in the
E. coli MLST database (http://mlst.ucc.ie/mlst/dbs/Ecoli).
The minimum spanning tree was generated from all 100
O91 sequence types (STs) and compared with the HUS-
associated enterohemorrhagic E. coli (HUSEC) collection
(4) to display the distribution of the STs compared with all
known STs associated with HUS.
DISPATCHES
1474 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 15, No. 9, September 2009
Author affiliations: Institute of Hygiene and the National Consulting
Laboratory on Hemolytic Uremic Syndrome, Münster, Germany (A.
Mellman, A.W. Friedrich, D. Harmsen, B. Middendorf, M. Bielasze-
wska, H. Karch); Department of Periodontolgy, Münster (D. Harms-
en); Robert Koch Institute, Wernigerode, Germany (A. Fruth); Free
University Berlin, Berlin, Germany (L.H. Wieler); and Robert Koch
Institute, Berlin (D. Weber)
DOI: 10.3201/eid1509.090161
Page 2
Shiga Toxin–producing E. coli O91
We tabulated disease severity according to STs. To
study the relationship between ST442 and disease sever-
ity, patients were categorized into those with HUS, those
with bloody diarrhea, and, serving as reference group,
those with nonbloody diarrhea or those that asymptom-
atically excreted these organisms. Univariate associations
were computed by using exact logistic regression. p values
<0.05 were considered statistically significant. STATA re-
lease 10.0 (StataCorp LP, College Station, TX, USA) was
used for statistical analysis.
The 100 STEC O91 strains resulted in 10 different
STs. Of these, STs 33 and 442 were most common (63 and
20 isolates, respectively). Six additional STs (690, 1048,
1051, 1052, 1053, 1054) were single-locus variants of
ST33, indicating their close relationship. The 2 remaining
STs (641 and 1020) were not closely related to any other
ST of the serogroup O91 strains. Detailed analysis of the
7 housekeeping genes used for MLST demonstrated the
fumC gene sequence alone could differentiate 5 STs of the
O91 strains. The comparison of all O91 STs with all STEC
STs and serotypes associated with HUS (HUSEC collec-
tion) is displayed in the Figure.
H antigens associated with O91 were H8 (n = 1), H10
(n = 2), H14 (n = 52), H21 (n = 20), Hnt (H-antigen non-
typeable) (n = 10), and H– (H-antigen nonmotile) (n = 15).
In serotype O91:H8, O91:H10, and O91:H21 strains, the
STs were serotype-specific. However, ST33 and its single-
locus variants represented all strains of serotypes O91:H14,
O91: H–, and O91:Hnt.
A single sequence type, ST442, accounting for 20% of
all strains, was found among each of the four O91 isolates
from patients with HUS (Table 1), a highly significant as-
sociation (odds ratio [OR] 27.8, 95% confidence interval
[CI] 3.3–∞, p<0.01; Table 2). ST442 strains were also more
frequently isolated from patients with bloody diarrhea than
were strains belonging to other STs (3/20 [15.0%] vs. 5/80
[6.3%] respectively), but this difference was not statisti-
cally significant (Table 2). The overall association with
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 15, No. 9, September 2009 1475
Figure. Minimum spanning tree based on the multilocus sequence
typing allelic profiles portraying the clonal distribution of the 100
Escherichia coli O91:H8/H10/H14/H21/H–/Hnt isolates (highlighted
in gray) associated with different diseases in relation to the
hemolytic uremic syndrome–associated enterohemorrhagic E. coli
collection. Each dot represents a given sequence type, and the size
of each circle is proportional to the number of strains analyzed.
Connecting lines show the number of identical alleles between 2
STs (thick black line, 6 of 7 alleles identical; thick gray line, 5 alleles
identical; thick dashed line, 4 alleles identical; thin dashed lines of
increasing length, <3 alleles identical).
Table 1. Distribution of disease severity across 10 different sequence types of 100 STEC O91 strains isolated from humans*
Most severe symptom of patients infected with STEC O91
ST (Serotype) HUS (n = 4) BD (n = 8) D (n = 79) A (n = 9) Total no. strains
ST33 (O91:H14/H–/Hnt) 0 5 51 7 63
ST442 (O91:H21) 4 3 12 1† 20
ST641 (O91:H10) 0 0 2 0 2
ST690 (O91:H14) 0 0 8 1 9
ST1020 (O91:H8) 0 0 1 0 1
ST1048 (O91:H14) 0 0 1 0 1
ST1051 (O91:H14) 0 0 1 0 1
ST1052 (O91:H14) 0 0 1 0 1
ST1053 (O91:H14) 0 0 1 0 1
ST1054 (O91:Hnt) 0 0 1 0 1
*ST, sequence type; STEC, Shiga toxin–producing Escherichia coli; H, H-antigen nonmotile strains; Hnt, H-antigen nontypeable strains; HUS, hemolytic
uremic syndrome; BD, bloody diarrhea; D, diarrhea without visible blood; A, asymptomatic infection (1 patient infected with ST442 [STEC O91:H21] had
abdominal cramps without diarrhea).
†Abdominal cramps without diarrhea.
Page 3
severe disease, defined as either HUS or bloody diarrhea,
was strong (OR 7.8, 95% CI 1.83–36.6, p<0.01). Severe
illness was noted for 7 (35.0%) of 20 patients infected by
ST442 strains, but only for 5 (6.3%) of 80 patients infected
by STEC O91 of other STs (Table 2). Patients with bloody
diarrhea were younger (median age 12 years) than patients
who had mild or no symptoms (median age 20 years).
However, this difference was not observed for the 4 pa-
tients with HUS (median age 21 years); in this instance, 2
were adults, 1 was 39 months old, and 1 was unknown.
Conclusions
To gain insight into the clonal structure of STEC O91,
we determined the relatedness of 100 strains isolated from
patients and correlated the clonal lineage to the clinical
outcome of the infection. MLST analysis divided the O91
isolates into 10 different STs, whereas classical serotyping
identified only 4 complete serotypes (O- and H-antigen).
Moreover, MLST was able to type all 25 nonmotile (H–) or
nontypeable (Hnt) O91 strains. The analysis demonstrated
that the fumC gene from the 7 genes used for MLST was
the most heterogeneous and enabled strain differentiation
into 5 different STs, among these ST442. It might therefore
be a candidate for first-line single-locus sequence typing.
HUS or bloody diarrhea without HUS was significant-
ly associated with ST442, which was represented by sero-
type O91:H21 only. However, Pradel et al. also reported a
case of HUS associated with an O91:H10 isolate that could
be differentiated from O91:H21 by using ribotyping (15).
In our study, known virulence determinants such as cyto-
lethal distending toxin V or Shiga toxin 2d activatable by
elastase in O91:H21 strains (8,10) might contribute to the
higher virulence of O91:H21 (ST442). However, further
studies of the mechanisms behind the emergence of ST442
in Germany and additional analysis of global O91 isolates
are needed. With the MLST approach described, trends and
changes in STEC O91 epidemiology and human infections
can be carefully surveyed.
Acknowledgments
We thank M. Karmali, D. Orth, and A. Siitonen for kindly
providing O91 strains from their countries. We are also grateful
to P. Tarr for critical reading of the manuscript.
This study was supported by grants from the Federal Min-
istry of Education and Research Network Zoonoses (grant nos.
01KI07124, 01KI07120, 01KI07127, 1368-763) and from the
National Reference Laboratory Network of the Robert Koch
Institute.
Dr Mellmann is a consultant for hygiene and infectious dis-
eases at the University Hospital Münster. His professional inter-
ests include molecular identification and epidemiology of STEC
and other bacterial pathogens.
References
1. Karch H, Tarr PI, Bielaszewska M. Enterohaemorrhagic Escherichia
coli in human medicine. Int J Med Microbiol. 2005;295:405–18.
DOI: 10.1016/j.ijmm.2005.06.009
2. Caprioli A, Morabito S, Brugere H, Oswald E. Enterohaemorrhagic
Escherichia coli: emerging issues on virulence and modes of trans-
mission. Vet Res. 2005;36:289–311. DOI: 10.1051/vetres:2005002
3. Brooks JT, Sowers EG, Wells JG, Greene KD, Griffin PM, Hoekstra
RM, et al. Non-O157 Shiga toxin–producing Escherichia coli infec-
tions in the United States, 1983–2002. J Infect Dis. 2005;192:1422–9.
DOI: 10.1086/466536
4. Mellmann A, Bielaszewska M, Köck R, Friedrich AW, Fruth A,
Middendorf B, et al. Analysis of collection of hemolytic uremic
syndrome–associated enterohemorrhagic Escherichia coli. Emerg
Infect Dis. 2008;14:1287–90. DOI: 10.3201/eid1408.071082
5. Werber D, Beutin L, Pichner R, Stark K, Fruth A. Shiga toxin–pro-
ducing Escherichia coli serogroups in food and patients, Germany.
Emerg Infect Dis. 2008;14:1803–6. DOI: 10.3201/eid1411.080361
6. Werber D, Behnke SC, Fruth A, Merle R, Menzler S, Glaser S, et
al. Shiga toxin–producing Escherichia coli infection in Germany:
different risk factors for different age groups. Am J Epidemiol.
2007;165:425–34. DOI: 10.1093/aje/kwk023
7. Frank C, Kapfhammer S, Werber D, Stark K, Held L. Cattle density
and Shiga toxin–producing Escherichia coli infection in Germany:
increased risk for most but not all serogroups. Vector Borne Zoonot-
ic Dis. 2008;8:635–43. DOI: 10.1089/vbz.2007.0237
8. Bielaszewska M, Friedrich AW, Aldick T, Schurk-Bulgrin R, Karch
H. Shiga toxin activatable by intestinal mucus in Escherichia coli
isolated from humans: predictor for a severe clinical outcome. Clin
Infect Dis. 2006;43:1160–7. DOI: 10.1086/508195
9. Bettelheim KA. The non-O157 Shiga-toxigenic (verocytotoxigen-
ic) Escherichia coli; under-rated pathogens. Crit Rev Microbiol.
2007;33:67–87. DOI: 10.1080/10408410601172172
10. Bielaszewska M, Fell M, Greune L, Prager R, Fruth A, Tschäpe H,
et al. Characterization of cytolethal distending toxin genes and ex-
pression in Shiga toxin–producing Escherichia coli strains of non-
O157 serogroups. Infect Immun. 2004;72:1812–6. DOI: 10.1128/
IAI.72.3.1812-1816.2004
11. Tarr PI, Gordon CA, Chandler WL. Shiga toxin–producing Escheri-
chia coli and haemolytic uraemic syndrome. Lancet. 2005;365:1073–
86.
12. Mellmann A, Bielaszewska M, Zimmerhackl LB, Prager R, Harm-
sen D, Tschäpe H, et al. Enterohemorrhagic Escherichia coli in hu-
man infection: in vivo evolution of a bacterial pathogen. Clin Infect
Dis. 2005;41:785–92. DOI: 10.1086/432722
DISPATCHES
1476 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 15, No. 9, September 2009
Table 2. Univariate associations of STEC O91 of sequence type 442 with severe disease by use of exact logistic regression*
Severe disease ST442, no. (%) Non-ST442, no. (%) Odds ratio 95% Confidence interval p value
HUS 4 (20) 0 (0) 27.8† 3.29–f <0.01
BD 3 (15) 5 (6) 3.4 0.47–20.1 0.25
HUS or BD 7 (35) 5 (6) 7.8 1.83–36.6 <0.01
*Reference group consisted of persons who had nonbloody diarrhea or who were asymptomatic carriers. STEC, Shiga toxin-producing Escherichia coli;
ST, sequence type; HUS, hemolytic uremic syndrome; BD, bloody diarrhea.
†Mean unbiased estimate.
Page 4
Shiga Toxin–producing E. coli O91
13. Prager R, Strutz U, Fruth A, Tschäpe H. Subtyping of pathogenic Es-
cherichia coli strains using flagellar (H)-antigens: serotyping versus
fliC polymorphisms. Int J Med Microbiol. 2003;292:477–86. DOI:
10.1078/1438-4221-00226
14. Perelle S, Dilasser F, Grout J, Fach P. Identification of the O-antigen
biosynthesis genes of Escherichia coli O91 and development of a
O91 PCR serotyping test. J Appl Microbiol. 2002;93:758–64. DOI:
10.1046/j.1365-2672.2002.01743.x
15. Pradel N, Boukhors K, Bertin Y, Forestier C, Martin C, Livrelli V.
Heterogeneity of Shiga toxin-producing Escherichia coli strains
isolated from hemolytic-uremic syndrome patients, cattle, and food
samples in central France. Appl Environ Microbiol. 2001;67:2460–8.
DOI: 10.1128/AEM.67.6.2460-2468.2001
Address for correspondence: Alexander Mellmann, Institut für Hygiene,
Universität Münster, Robert Koch Str 41, 48149 Münster, Germany;
email: mellmann@uni-muenster.de
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 15, No. 9, September 2009 1477
End of preview.
