664

Quinolone Antibiotics Induce Shiga Toxin–Encoding Bacteriophages,

Toxin Production, and Death in Mice
Xiaoping Zhang,a Aaron D. McDaniel,a Lucas E. Wolf,a Division of Geographic Medicine and Infectious Diseases, Tupper

Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019

Gerald T. Keusch, Matthew K. Waldor, Research Institute, New England Medical Center,
and David W. K. Acheson Boston, Massachusetts

Shiga toxin–producing Escherichia coli (STEC) cause significant disease; treatment is sup-
portive and antibiotic use is controversial. Ciprofloxacin but not fosfomycin causes Shiga
toxin–encoding bacteriophage induction and enhanced Shiga toxin (Stx) production from
E. coli O157:H7 in vitro. The potential clinical relevance of this was examined in mice colonized
with E. coli O157:H7 and given either ciprofloxacin or fosfomycin. Both antibiotics caused a
reduction in fecal STEC. However, animals treated with ciprofloxacin had a marked increase
in free fecal Stx, associated with death in two-thirds of the mice, whereas fosfomycin did not.
Experiments that used a kanamycin-marked Stx2 prophage demonstrated that ciprofloxacin,
but not fosfomycin, caused enhanced intraintestinal transfer of Stx2 prophage from one
E. coli to another. These observations suggest that treatment of human STEC infection with
bacteriophage-inducing antibiotics, such as fluoroquinolones, may have significant adverse
clinical consequences and that fluoroquinolone antibiotics may enhance the movement of
virulence factors in vivo.

Shiga toxin–producing Escherichia coli (STEC), such as
E. coli O157:H7, are emerging foodborne pathogens in the
United States and other parts of the world [1, 2]. Ingestion of
STEC can result in gastrointestinal symptoms, such as diarrhea
and hemorrhagic colitis, and may also progress to hemolytic-
uremic syndrome (HUS), a severe sequela of this infection [3].
STEC have become the most frequent cause of acute renal
failure in children in the United States in recent years and are
currently one of the most devastating foodborne pathogens in
many parts of the world. Little is known about the molecular
mechanisms leading to the development of the most serious
complications, such as HUS, but it is likely that different in-
dividuals respond to exposure to STEC in different ways. It is
not clear why some people develop a mild gastrointestinal ill-
ness, whereas others develop hemorrhagic colitis and still others
develop full-blown HUS. Many factors, including both host
and bacterial determinants, may be responsible for these dif-
Received 27 May 1999; revised 20 September 1999; electronically pub-
lished 8 February 2000.
Presented in part: IX International Congress of Bacteriology and Applied
Microbiology, Sydney, Australia, August 1999.
Financial support: NIH (AI-39067, AI-07329, and DK-34928). M.K.W.
is a Pew Scholar in the Biomedical Sciences and a Tupper Research Fellow.
a
Present affiliations: Department of Pharmaceutics, College of Pharmacy,
Rutgers, State University of New Jersey, Piscataway (X.Z.); University of
Southern California School of Medicine, Los Angeles (A.D.M.); Depart-
ment of Medicine, Northeast Health Systems, Beverly, Massachusetts
(L.E.W.).
Reprints or correspondence: Dr. David W. K. Acheson, Division of Geo-
graphic Medicine and Infectious Diseases, New England Medical Center,
750 Washington St., Boston, MA 02111 (Dacheson@Lifespan.org).
The Journal of Infectious Diseases 2000; 181:664–70
q 2000 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2000/18102-0034$02.00
ferences in response to STEC. It is clear that the production
of Shiga toxins by STEC is one of the most critical factors in
the development of HUS. It is also probable, but not proven
in humans, that the more toxin that is made, the more likely
a patient will develop significant medical problems. However,
the factors that control toxin production in the intestine and
the processes that are critical in the movement of toxin from
the intestinal lumen to their final target in the kidney, brain,
and other organs are as yet undetermined.
The Shiga toxin family is made up of 2 main groups, Shiga
toxin 1 (Stx1) and Shiga toxin 2 (Stx2), produced by E. coli
O157:H7 and many other STEC serotypes. Both are usually
encoded in the genomes of lysogenic lambdoid phages, thereby
allowing a mechanism for toxin gene dissemination via transfer
of bacteriophages [4, 5]. Recent studies suggest that the life
cycles of these Stx phages regulate toxin production and thereby
may influence pathogenesis [6]. In particular, Stx prophage in-
duction has been associated with increased Stx production in
vitro [7]. Stx prophages, like l prophages, are inducible with
agents that provoke the bacterial “SOS” response to DNA
damage and/or to the inhibition of DNA replication. Thus, the
induction of Stx prophages by the alkylating agent mitomycin
C is associated with significant increases in Stx production in
vitro [7]. In recA2 STEC strains, this SOS-triggering agent does
not induce Stx prophages or augment Stx production [7]. Be-
cause RecA is known to be required for the initiation of the
SOS response, but is not thought to directly regulate Stx pro-
duction, these findings indicate that phage induction is a re-
quired step in the augmentation of toxin production that occurs
following exposure of STEC to SOS-inducing agents.
The clinical relevance of prophage-mediated Stx production
JID 2000;181 (February) Quinolone Antibiotics Induce Shiga Toxins In Vivo
is suggested by the observation that patients treated with mi-
tomycin C have been reported to develop HUS [8]. Several
clinically useful antibiotics known to induce the SOS response,
including trimethoprim-sulfamethoxazole (commonly used to
treat diarrheal disease in children) and ciprofloxacin (commonly
used to treat diarrheal disease in adults), have also been shown
to enhance Stx production in vitro [9–11]. There have been no
large randomized controlled prospective studies to evaluate the
utility of antimicrobial treatment of STEC infection; however,
in 1 small retrospective study, treatment with sulfonamides,
which is known to induce bacteriophages, was associated with
an increased likelihood of developing HUS [12]. In contrast, a
retrospective analysis of the large 1996 O157:H7 outbreak in
Japan suggested that the use of the antibiotic fosfomycin, an
inhibitor of bacterial cell wall synthesis that is not known to
induce bacteriophages, was correlated with reduced likelihood
of developing HUS [13].
A variety of animal models have been developed for the study
of STEC [14], among which mouse models with orally delivered
STEC are widely used and relatively well characterized [15–18].
In the mouse models, animals given STEC orally frequently
develop neurologic and renal symptoms, usually culminating
in death. Coadministration of anti-Stx antibody prevents death
of the mice, thereby suggesting that Stx is the principal cause
of murine morbidity and mortality [18]. In mice, the principal
targets of Stx appear to be the central nervous system and
endothelial cells [16]. We recently demonstrated that E. coli Stx
phage lysogens produce infectious virions within the murine
gastrointestinal tract [19]. We hypothesized that antibiotic-
induced induction of Stx phage from intestinal O157:H7 in mice
may augment Stx production and lead to adverse consequences.
In this study, we used an STEC mouse model to investigate the
effects of 2 antibiotics on fecal STEC colony-forming units,
fecal toxin levels, intraintestinal phage production, and mouse
mortality. We sought to elucidate the potential risks associated
with the use of antibiotics that result in DNA damage and
phage induction in treating patients infected with STEC.
Materials and Methods
Chemicals and reagents. Ciprofloxacin (2 mg/mL ciprofloxacin
in 5% dextrose) was purchased from Bayer (West Haven, CT).
Kanamycin monosulfate, fosfomycin disodium salt, streptomycin
sulfate, and tetracycline hydrochloride were purchased from Sigma
(St. Louis), PBS from Gibco BRL (Grand Island, NY), Vent DNA
polymerase and restriction enzymes from New England Biolabs
(Beverly, MA), and pBluescript SK2 plasmid DNA from Stratagene
(La Jolla, CA).
Bacterial strains. E. coli O157:H7 strain 1:361 encodes Stx2
only and was isolated during a nationwide surveillance study [20].
1:361R, an StrR derivative of 1:361, was isolated by selecting for
a spontaneous streptomycin-resistant mutant. 1:361R was used for
all in vitro and in vivo studies described below. MC4100, argG:
Tn10, is a derivative of MC4100 (araD139, delta[lac]U169, strA,
665
thi) [21]; both were from our laboratory collection of bacterial
strains. MC4100 lysogens of Stx2 phage 933W [4] and 933W-Kn
(see below), MC4100 (933W) and MC4100 (933W-Kn), respec-
tively, were isolated as described elsewhere [7]. C600 (933W) is an
E. coli C600 lysogen of Stx2 phage 933W [4].
Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019
Construction of the 933W-Kn phage. 933W-Kn phage was con-
structed from the Stx2 phage 933W [4] by replacing most of the
Stx2 toxin operon promoter and the toxin A subunit gene (stx2A)
with the KnR gene from Tn903. A 524-bp Stx2 fragment was cloned
by use of the forward primer Stx2-1 (ACAG CTGCAGCG-
TTTCTGAACAG; PstI site underscored) and reverse primer Stx2-
2 (GCAG GGTACCACCGAACCCGTGT; KpnI site under-
scored). This fragment begins 50 bp upstream of the stop codon
of stx2A and ends 185 bp downstream of the stx2B gene stop
codon. This 524-bp fragment was cloned into a PstI/KpnI–cut de-
rivative of pBluescript SK2 that had previously had its 2 BspHI
sites converted to AvrII sites, yielding pSK/933W-1. Forward
primer Stx2-10 (GGCA GGATCCGAATGTAGTCAGTCAGA-
ACGGATG; BamHI site underscored) and reverse primer Stx2-7
(GCTG CTGCAG-TCATGACCAGACCCGGCGCAGA; PstI
and BspHI sites underscored) were used to clone a 364-bp region
from 933W phage. This 364-bp region spans from 65 bp upstream
of the ileX start codon [22] to a point 50 bp upstream of the stxA
start codon, including the Stx2 operon promoter [23]. The 364-bp
fragment was cloned into the BamHI/PstI–cut pSK/933W-1, yield-
ing pSK/933W-2. pSK/933W-2 was cut with BspHI and treated
with Klenow fragment, yielding a blunt-ended pSK/933W-2 back-
bone. A third pair of primers, forward primer DA101 (GCGC
AGTACTCTGTGTGGCCACGTTGTG; ScaI site underscored)
and reverse primer DA102 (GCGC AGTACTGTTCTAGAGG-
CCTGTGTG; ScaI site underscored), were used in a polymerase
chain reaction to clone the KnR gene (originally derived from
Tn903) with its associated promoter from plasmid p216 (obtained
from A. Camilli, Tufts University, Medford, MA). The KnR frag-
ment (1044 bp) was ligated into the blunted BspHI sites of the
pSK/933W-2 to yield pSK/933W-2-Kn. pSK/933W-2-Kn contained
the KnR gene flanked by the Stx2 operon promoter region (50) and
the stx2B region (30). This entire Stx2 promoter–KnR gene–stx2B
gene region insert was ligated into the partially digested suicide
vector pCVD442 [24] to yield pCVD442/933W-2-Kn. pCVD442/
933W-2-Kn was electroporated into C600 (933W). One of the C600
KnR isolates chosen randomly was induced with mitomycin C (500
ng/mL) in Luria-Bertani (LB) broth. Phage-containing supernatant
was then used to lysogenize C600. In C600 (933W-Kn), the re-
placement of stx2A by the KnR gene in the 933W-Kn prophage
was confirmed by polymerase chain reaction with the primers used
in the original cloning described above. The replacement of stx2A
by the KnR gene did not significantly alter the genome size of the
resulting phage (net 98-bp gain). As expected, E. coli cells lyso-
genized by this recombinant phage produced no Stx2 A subunit
by Western blot, and sonicated extracts of overnight cultures were
not cytotoxic to Vero cells [25] (data not shown). The ability of
933W-Kn phage to form plaques, to be induced from an E. coli
lysogen (e.g., C600 or MC4100), and to lysogenize a new E. coli
host (e.g., C600 or MC4100) remained intact and were no different
from the original 933W phage (data not shown). C600 cells lyso-
genized with 933W and 933W-Kn phages had identical growth
curves in LB medium (data not shown).
666 Zhang et al.
Quantitation of colony-forming units and plaque-forming units.
Samples from both in vitro and in vivo experiments were serially
diluted in PBS and plated onto LB agar plates containing strep-
tomycin at 500 mg/mL. Colony-forming units were counted after
overnight culture at 377C. Samples collected in the mouse lysogenic
conversion experiments were processed similarly and plated on ap-
propriate selective plates. Plaque-forming units were measured as
described elsewhere [19].
Stx2 concentration. Stx2 concentrations in bacterial cultures
were determined by EIA [26] by use of purified Stx2 as the standard.
Stx2 levels from murine fecal extracts were measured by means of
Premier EHEC (Meridian Diagnostics, Cincinnati). In the in vitro
experiments, 1.0-mL samples of various LB cultures of 1:361R were
spun for 5 min in a microcentrifuge. After centrifugation, each
supernatant was used for extracellular Stx2 determination, and
each pellet was resuspended in 1 mL of LB followed by sonication
(3 cycles of 30 pulses at 67% dutycycle and 4.5 output) on a Branson
Sonifier 450 (Branson Sonic Power, Danbury, CT). The sonicated
sample was then spun for 5 min, and the supernatants were used
for cell-associated Stx2 determination. In the in vivo experiments,
mouse fecal samples were suspended in PBS (see below) and spun
in a microcentrifuge for 5 min, and the supernatants were then
used for determination of the Stx2 concentration. Plating of these
fecal supernatants on LB-streptomycin plates showed no or neg-
ligible numbers of streptomycin-resistant cells, indicating that the
Stx2 levels measured represented primarily extracellular Stx2 in the
feces.
Effect of ciprofloxacin or fosfomycin on toxin expression in
vitro. Overnight cultures in LB medium were diluted in the same
medium to an optical density at 600 nm of ∼0.08 and then were
split into 3 25-mL cultures in flasks. No antibiotic was added to
the control culture, whereas ciprofloxacin and fosfomycin were
added to the second and third flasks, respectively. The colony-
forming units, plaque-forming units, and Stx2 concentrations at
time 0 were determined in the control flask. The 3 flasks were
incubated at 377C with shaking. At 3 and 6 h, samples were taken
from each of the 3 flasks, and optical density at 600 nm, colony-
forming units, plaque-forming units, and Stx2 concentration were
determined.
Mouse infection with E. coli 1:361R. The procedure used was
a modification of that described in our previous studies [19]. Sets
of 14 male CD-1 mice aged 5 weeks (Charles River Laboratories,
Wilmington, MA) were given drinking water containing strepto-
mycin (2 g/L) throughout each experiment. One day after strep-
tomycin was begun, the mice were starved for food for 24 h; then
each mouse was inoculated intragastrically with ∼ 3.0 3 10 9 1:361R
cells. Each inoculum comprised ∼0.15 mL of overnight culture that
had been washed once in PBS and resuspended in the same volume
of 20% sucrose. The day of inoculation was designated day 0. In
initial experiments, we found that on days 1 and 2 after 1:361R
inoculation there was a dramatic increase in the number of fecal
1:361R cells and fecal Stx2. One or 2 mice of each group of 14
died by day 4, after which no further deaths were observed. There-
fore, in subsequent antibiotic treatment experiments, 4 days after
inoculation with 1:361R, the surviving mice were divided into equal
groups. Mice received an intraperitoneal dose of either 30 mg of
ciprofloxacin or 15 mg of fosfomycin on days 4, 5, and 6. Control
mice received only PBS. The rationale behind drug dosage is as
JID 2000;181 (February)
follows. Previous in vitro experiments suggested that subinhibitory
concentrations of ciprofloxacin caused maximal increases in Stx
(data not shown). In initial experiments, we determined a dose of
ciprofloxacin that caused a significant drop in fecal 1:361R colony-
forming units (∼3 logs) but did not kill all of the organisms. The
Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019
dose of fosfomycin was adjusted to cause an equivalent drop in
fecal colony-forming units (∼3 logs). In all experiments, mice were
observed twice daily from day 0 to the end of each experiment on
day 18. On days 0–8 and on day 11, a fresh mouse stool sample
(2–3 pellets) was collected from each mouse and weighed. Each
stool sample was suspended in PBS at a ratio of 1 mL to 0.1 g of
stool. After stool debris were allowed to settle, the supernatant was
sampled for determination of colony-forming units. The remainder
was spun in a microcentrifuge for 5 min, and the second super-
natant was used for Stx2 determination as described above.
Effect of antibiotics on intraintestinal phage production. These
experiments were designed to measure intestinal 933W-Kn phage
production after ciprofloxacin and fosfomycin administration. A
mixture of overnight cultures of the donor strain, MC4100 (933W-
Kn) (∼ 2 3 10 9/mouse), and the recipient strain, MC4100, argG:
Tn10 (∼ 7 3 10 9/mouse), was administered together via a feeding
needle to CD-1 mice as described above. Unlike the 1:361R strain,
these E. coli K12 strains colonized the mouse gut poorly. Stool
colony-forming unit counts decreased rapidly, compared with those
of the 1:361R strain. Consequently, these experiments were com-
pressed to 8 days. Ciprofloxacin and fosfomycin were given on
days 2–4 at the same dosages as described above. From day 0 to
the end of the experiments on day 8, colony-forming units of the
donor, the recipient, and the in vivo–derived transductants were
determined by plating the stool samples onto 3 sets of LB agar
plates containing, respectively, streptomycin (500 mg/mL)
–kanamycin (50 mg/mL), streptomycin (500 mg/mL)–tetracycline
(7.5 mg/mL), and streptomycin (500 mg/mL)–kanamycin (50 mg/
mL)–tetracycline (7.5 mg/mL).
Results
Effect of ciprofloxacin or fosfomycin on Stx production in
vitro. Our first goal was to demonstrate the effect of cipro-
floxacin, an inhibitor of DNA gyrase, and fosfomycin, an in-
hibitor of peptidoglycan synthesis, on Stx2 production and bac-
teriophage induction in vitro from the clinical E. coli O157:H7
isolate, 1:361R. Three hours after the addition of ciprofloxacin
(30 ng/mL) to the 1:361R culture, we noted a 7-fold increase
in the number of plaque-forming units per milliliter of culture
supernatant and a 17-fold increase in Stx concentration com-
pared with that of the control culture (table 1). At the 6-h time
point, the ciprofloxacin-mediated induction of the Stx prophage
and Stx production were even more marked, with a 11000-fold
increase in the number of plaque-forming units per milliliter
and an ∼60-fold increase in the total Stx2 concentration in the
culture (table 1). The relative increase in both plaque-forming
units and toxin compared with control or fosfomycin treatment
were reproducible in subsequent experiments and at higher
doses of ciprofloxacin (40 ng/mL; data not shown).
This dramatic increase in Stx production occurred despite a
JID 2000;181 (February) Quinolone Antibiotics Induce Shiga Toxins In Vivo 667

Table 1. Ciprofloxacin induces Shiga toxin (Stx) prophage and augments Stx production
from Escherichia coli O157:H7 strain 1:361R.
Phage titer Stx concentration Cell no.
(pfu/mL 3103) (ng/mL) (cfu/mL 3106)
Time (h) Control Cpfx Fm Control Cpfx Fm Control Cpfx Fm

Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019

0 0.02 0.02 0.02 30 30 30 114.0 114.0 114.0
3 7.3 50.0 10.4 301 5325 231 1930.0 4.5 34.0
6 12.5 13,000 44.0 1406 81,800 522 4600.0 0.3 4200.0
NOTE. Overnight culture of 1:361R was diluted to optical density at 600 nm of 0.08 and split into
3 equal-volume cultures (time 0). At time 0, either ciprofloxacin (Cpfx, 25 ng/mL), fosfomycin (Fm, 800
ng/mL), or nothing (control) was added. Serial dilutions of chloroform-treated supernatants were plated
on E. coli strain C600. Polymerase chain reaction analysis was used to confirm that all 19 randomly picked
plaques were positive for stxA2 sequences. Stx2 concentration was determined in both culture supernatants
and sonicated pellets by EIA [22] with purified Stx2 as standard [23]. Total Stx2 is shown.

110,000-fold reduction in recoverable 1:361R colony-forming
units in the culture containing ciprofloxacin, probably a result
of cell killing by both the antibiotic and phage-mediated cell
lysis. Thus, ciprofloxacin resulted in marked Stx prophage in-
duction and increases in Stx production from Stx phage lyso-
gens, similar to our previously reported results with mitomycin
C [7]. Stx prophage induction and increase in Stx production
are not nonspecific responses to antimicrobial agents. Despite
the fact that the fosfomycin-treated cultures had recovered by
6 h, the drug resulted in a decline in 1:361R colony-forming
units at 3 h but, unlike ciprofloxacin, had no effect on the
induction of 1:361R Stx prophages or toxin production, com-
pared with controls (table 1).
Effect of ciprofloxacin or fosfomycin treatment in mice colo-
nized with E. coli O157:H7. To address the potential clinical
ramifications of antibiotic induction of Stx prophages and toxin
production, we then compared the effects of ciprofloxacin and
fosfomycin on intestinal toxin production and mortality in mice
colonized with 1:361R. To facilitate intestinal colonization with
this O157:H7 strain, CD-1 mice were given streptomycin [15]
for 2 days and then orally inoculated with 1:361R. In each set
of experiments, the mice were divided into equal groups on day
4. One group received an antibiotic (either ciprofloxacin or
fosfomycin) and the other group received PBS. In control mice,
the pattern of 1:361R intestinal colonization and toxin pro-
duction in individual mice was consistent (figures 1A and 2A).
Fecal Stx was detectable 1 day after inoculation and declined
thereafter as the 1:361R excretion dropped (figures 1A and 2A).
After treatment with either antibiotic, there was a more rapid
fall in fecal STEC colony-forming units than was observed in
the PBS control group (figure 1), with values differing by 3
orders of magnitude. Despite causing similar decreases in col-
ony-forming units, the antibiotics differed in their effect on fecal
Stx. Ciprofloxacin treatment resulted in an increase in fecal Stx
concentration in all 6 mice, whereas fosfomycin did not (figure
2). This increase in intraintestinal toxin production was accom-
panied by a dramatic clinical effect. Four of the 6 ciprofloxacin-
treated mice (those with the highest stool Stx concentrations)
died, whereas none of the fosfomycin-treated mice nor control
mice died. The mortality resulting from ciprofloxacin treatment
of 1:361R-colonized mice was reproduced in 2 subsequent ex-
periments. In the first of the repeat experiments, we observed
death in 4 of 7 ciprofloxacin-treated mice and in 0 of 6 PBS-
treated mice, and in the second experiment, 7 of 8 ciprofloxacin-
treated mice died. The mortality observed in the ciprofloxacin-
treated mice is not attributable to the antibiotic treatment alone,
because there were no deaths in a separate control group of 6
mice that were treated with streptomycin and then with cip-
rofloxacin as above but that were not colonized with 1:361R
(data not shown). There was no significant difference in the
amounts of fecal Stx2 detected in the fosfomycin-treated mice
compared with control (PBS-treated) mice (figure 2C vs. fig-
ure 2A).
Effect of ciprofloxacin on intraintestinal phage production in
mice. To explore whether treatment of mice with ciproflox-
acin leads to increased intraintestinal Stx phage production
from a lysogen as seen in vitro, we used an intraintestinal Stx
phage lysogenic conversion assay that we recently described
elsewhere [19], which involves intraintestinal transduction of a
marked E. coli recipient strain (a TcR derivative of E. coli K12
strain MC4100 [19]) to KnR by a Kn-marked Stx phage. This
phage, 933W-Kn, is harbored in an E. coli K12 strain, MC4100
(933W-Kn), the donor strain. After intragastric inoculation of
the donor and recipient strains to 5 mice, the number of donor
cells (SmR KnR), recipient cells (SmR TcR), and “transductant”
cells (those recipient cells lysogenically converted to kanamycin
resistance [SmR TcR KnR]) in fecal samples were determined
daily. Two days after the inoculation of the donor and recipient
strains, mice were injected intraperitoneally with either cipro-
floxacin, fosfomycin, or PBS at the same doses used in our
previous experiments. Because there is no biologically active
Stx-producing strain present in these experiments (replacement
of stxA by the KnR gene inactivates the toxin), no deaths were
observed in these experiments. As in the experiments with 1:
361R in mice, both ciprofloxacin and fosfomycin treatments
resulted in 100- to 1000-fold reductions in the number of donor
and recipient cells detected in fecal samples, compared with the
PBS-treated control group (both of these strains are sensitive
to the 2 antibiotics; table 2). However, significant numbers of
transductants were detected in all 5 of the mice treated with
668 Zhang et al. JID 2000;181 (February)

after induction of the Stx-encoding bacteriophages [7]. A num-

ber of reports have suggested that antibiotics are associated
with the development of HUS in patients infected with STEC
[12, 28]. Other in vitro studies have demonstrated that some
antibiotics, especially fluoroquinolones, will increase Stx pro-

Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019

duction [9, 10, 11, 29]. The purpose of this work was to take
these in vitro observations to a more critical level and to de-
termine whether fluoroquinolones induced toxin expression in
vivo and, if so, whether any potential clinical complications
might occur after therapy with fluoroquinolones in animals
infected with STEC.

Figure 1. Intestinal colonization with Escherichia coli O157:H7

strain 1:361R. After inoculation of 1:361R on day 0, colony-forming
units (cfu) per gram of stool were determined daily. Each line corre-
sponds to individual mouse. Treatments with either PBS (A), cipro-
floxacin (B), or fosfomycin (C) were given intraperitoneally on days
4–6 (shaded area). Because results for 2 control groups were virtually
identical, data from only 1 are shown (A).

ciprofloxacin and in none of the mice treated with fosfomycin

(table 2). One of the control mice also had a low level of trans-
ductants, suggesting that spontaneous Stx2 phage induction can
occur intraintestinally, similar to our previous results with Stx1
phage lysogens [19]. Spot checks of putative transductant cells
grown from fecal samples (SmR TcR KnR) with a polymerase
chain reaction assay for 933W-Kn phage DNA sequences con-
firmed that all of the colonies tested contained this phage.

Figure 2. Fecal Shiga toxin (Stx) concentrations after intragastric

Discussion
Little is known about the regulation of Stx expression in
vitro; nothing is known about its regulation in vivo. Stx1, like
Shiga toxin from Shigella dysenteriae, is iron-regulated in vitro
[27], but it is not clear whether this regulation is critical in vivo.
Production of both Stx1 and Stx2 has been shown to increase
inoculation of Escherichia coli O157:H7 strain 1:361R. Levels of Stx2
in stool samples (expressed as nanograms of Stx2 per gram of stool)
were determined daily after inoculation of 1:361R on day 0. Each line
corresponds to individual mouse. Treatments with either intraperito-
neal PBS (A), ciprofloxacin (B), or fosfomycin (C) were given intra-
peritoneally on days 4–6 (shaded area). †, mice that died within 24 h
of stool Stx determinations. Because results for 2 control groups were
virtually identical, data from only 1 are shown (A).
JID 2000;181 (February) Quinolone Antibiotics Induce Shiga Toxins In Vivo 669

Table 2. Ciprofloxacin enhances intraintestinal transduction of Kn-marked Shiga toxin (Stx) phage.
Ciprofloxacin treatment Fosfomycin treatment PBS treatment
Mouse Donor Recipient Transductant Donor Recipient Transductant Donor Recipient Transductant
1 4.0 3 105
8.0 3 105
1.1 3 104
9.2 3 105
2.8 3 104
UD 2.6 3 107
1.3 3 107
UD

Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019

2 1.6 3 105 5 3 104 1.7 3 104 1.0 3 103 1.0 3 104 UD 3.0 3 107 1.0 3 108 1.8 3 103
3 2.0 3 104 3 3 104 7.0 3 103 1.7 3 104 1.5 3 104 UD 2.0 3 107 2.0 3 107 UD
4 1.3 3 105 1.2 3 104 1.2 3 104 6.0 3 105 1.8 3 105 UD 1.2 3 108 6.0 3 106 UD
5 1.5 3 103 6.0 3 103 2.0 3 102 1.0 3 105 1.5 3 105 UD 4.9 3 107 1.2 3 108 UD
NOTE. On day 0, each mouse was orally inoculated with 2 3 109 cells of potential phage donor strain (MC4100 933W-Kn, SmR KnR)
and 5 3 109 cells of potential recipient strain (MC4100 argG:Tn10, SmR TcR); on days 2–4, mice were inoculated intraperitoneally with either
ciprofloxacin (30 mg/mouse), fosfomycin (15 mg/mouse), or PBS. Fecal samples were collected daily, weighed, resuspended in PBS, and
plated on media to allow determination of nos. of donor cells (SmR KnR), recipient cells (SmR TcR), and transductants (SmR KnR TcR). Data
are given as colony-forming units per gram of stool. Values shown are from day 3 after inoculation, time point representative of data from
entire week of study. Threshold level of detection is 102 cfu/g of stool; UD, undetectable at any time during experiment.

Ciprofloxacin and other fluoroquinolones are the antibiotics
most frequently prescribed for the treatment of diarrheal dis-
eases in adults. Such therapy is often empirical and prescribed
before results of stool cultures are available. We found that
ciprofloxacin treatment of mice colonized with E. coli O157:
H7 induced Stx prophages. The consequences of intestinal Stx
prophage induction were enhanced intraintestinal Stx produc-
tion, increased Stx phage transmission, and, most dramatically,
death of the ciprofloxacin-treated mice. Because fosfomycin and
ciprofloxacin treatments both reduced intestinal 1:361R excre-
tion similarly and because fosfomycin causes bacterial lysis (by
inhibiting cell wall synthesis), the absence of augmented intes-
tinal toxin levels and death in the mice treated with fosfomycin
indicates that the effects observed with ciprofloxacin cannot be
attributed just to antibiotic-related bacterial cell lysis and re-
lease of preformed toxin. Thus, expanding the scope of our in
vitro observations, ciprofloxacin, but not fosfomycin, enhanced
intraintestinal Stx2 production in mice colonized with E. coli
O157:H7. In this murine model, these ciprofloxacin-mediated
increases in intraintestinal toxin levels were sufficient to kill the
mice.
These findings suggest that quinolones, as well as other an-
tibiotics that can induce the bacterial SOS response with re-
sulting Stx prophage induction, should not be used as therapy
for STEC infection. Although the dose of ciprofloxacin that
we administered to mice colonized with E. coli O157:H7 was
only 20% of the dose (on a milligram per kilogram of body
weight basis) commonly given to patients with diarrhea, the
concentrations of ciprofloxacin in intestinal microenvironments
probably vary with the nature of intestinal contents, intestinal
mucus, bacterial flora, epithelial cell secretions, and gut motility.
When antibiotic therapy is initiated, concentrations required
for antibiotic-mediated Stx prophage and toxin induction may
be achieved in critical intestinal microenvironments well before
bactericidal concentrations can accumulate. Our in vitro ex-
periments demonstrated that even a 3-h exposure to ciproflox-
acin was enough to induce phage and elevate toxin expression.
Thus, we speculate that the use of higher doses of ciprofloxacin
will, in the initial phase, cause an elevation of toxin that may
be deleterious in patients, even if the organisms are subse-
quently killed by the antibiotic. These pharmacologic consid-
erations suggest that it may be very difficult to determine a safe
dose of fluoroquinolones for the treatment of STEC infection.
The cause of a diarrheal illness is usually not known when
patients first seek medical attention. Because STEC are asso-
ciated with ∼1% of stools submitted for analysis to clinical
microbiology laboratories in the United States [20], our data
suggest that empirical quinolone treatment of diarrheal disease
or, indeed, therapy with any other antibiotic known to cause
bacteriophage induction, should be reconsidered. If antimicro-
bial treatment of STEC infection is indicated, classes of anti-
biotics that do not induce Stx prophages (e.g., cell wall synthesis
inhibitors such as fosfomycin or protein synthesis inhibitors
such as macrolides) should be used. Similar caution should be
exercised in the choice of antimicrobial agents in the treatment
of other infections, in which prophage-encoded toxins are crit-
ical virulence factors [30, 31]. Clearly, many patients infected
with STEC go on to develop HUS without having received
antibiotic therapy, so there must be other important factors
relating to the control of toxin expression or its absorption
from the intestine. However, our observations do suggest that
caution should be exercised in relation to empirical treatment
of diarrheal disease in both children and adults.
Our data raise not only questions about the safety of the use
of fluoroquinolones in the treatment of diarrheal disease for
individual patients but also broader issues regarding the role
of antimicrobials in the dissemination of bacteriophage-en-
coded virulence factors in the environment. We found that Stx-
encoding phages are able to move from 1 strain of E. coli to
another in the intestines of mice and that antibiotic-mediated
prophage induction enhances this process. It is possible that
the use of phage-inducing antimicrobials in farm animals, es-
pecially at subtherapeutic levels, may inadvertently enhance the
dissemination of bacteriophage-encoded toxins. This could con-
tribute to the emergence of new pathogens and the spread of
Stx genes to other members of the Enterobacteriaceae family,
such as Citrobacter and Enterobacter species, both of which
670 Zhang et al.
have been associated with Shiga toxin–related HUS in humans
[32, 33].
16.
Acknowledgments
17.
We thank our colleagues A. Plaut, B. Davis, A. Camilli, and A. Kane
for their helpful comments.
18.
References
1. Acheson DWK, Keusch GT. Which Shiga toxin–producing E. coli are im- 19.
portant? ASM News 1996; 62:302–6.
2. Kaper JB, O’Brien AD. Escherichia coli O157:H7 and other Shiga
toxin–producing E. coli strains. Washington, DC: American Society for 20.
Microbiology Press, 1998.
3. Griffin PM. Escherichia coli O157:H7 and other enterohemorrhagic Escher-
ichia coli. In: Blaser MJ, Smith PD, Ravdin JI, Greenberg HB, Guerrant
RL, eds. Infections of the gastrointestinal tract. New York: Raven Press,
1995:739–61. 21.
4. O’Brien AD, Newland JW, Miller SF, Holmes RK, Williams-Smith H, Formal
SB. Shiga-like toxin converting phages from Escherichia coli strains that 22.
cause hemorrhagic colitis or infantile diarrhea. Science 1984; 226:694–6.
5. Huang A, Frieman J, Brunton JL. Characterization of a bacteriophage that
carries the genes for production of Shiga-like toxin 1 in Escherichia coli.
J Bacteriol 1987; 169:4308–12. 23.
6. Neely MN, Friedman DI. Functional and genetic analysis of regulatory
regions of coliphage H-19B: location of Shiga-like toxin and lysis genes
suggest a role for phage functions in toxin release. Mol Microbiol 1998;
24.
28:1255–67.
7. Muhldorfer I, Hacker J, Keusch GT, et al. Regulation of the Shiga-like toxin
II operon in Escherichia coli. Infect Immun 1996; 64:495–502.
25.
8. Acheson DWK, Donohue-Rolfe A. Cancer-associated hemolytic uremic syn-
drome: a possible role of mitomycin C in relation to Shiga-like toxins. J
Clin Oncol 1989; 7:1943.
26.
9. Karch H, Goroncy-Bermes H, Opferkuch W, Droll HP, O’Brien AD. Sub-
inhibitory concentrations of antibiotics modulate amount of Shiga-like
toxin produced by Escherichia coli. In: Adam D, Hahr H, Opterkuch W,
eds. The influence of antibiotics on the host-parasite relationship. Berlin:
27.
Springer Verlag, 1985:239–45.
10. Walterspiel JN, Ashkenazi S, Morrow AL, Cleary TG. Effect of subinhibitory
concentrations of antibiotics on extracellular Shiga-like toxin 1. Infection
28.
1992; 20:25–9.
11. Matsushiro A, Sato K, Miyamoto H, Yamamura T, Honda T. Induction of
prophages of enterohemorrhagic Escherichia coli O157:H7 with norflox-
29.
acin. J Bacteriol 1999; 181:2257–60.
12. Pavia AT, Nichols CR, Green DP, et al. Hemolytic-uremic syndrome during
an outbreak of Escherichia coli O157:H7 infections in institutions for
mentally retarded persons: clinical and epidemiological observations. J
Pediatr 1990; 116:544–51. 30.
13. Takeda T, Yoshino YY, Uchida H, Ikeda N, Tanimura M. Early use of
fosfomycin for Shiga toxin–producing Escherichia coli O157:H7 infection 31.
reduces the risk of hemolytic-uremic syndrome. In: Kaper JB, O’Brien
AD, eds. Escherichia coli O157:H7 and other Shiga toxin–producing E.
coli strains. Washington, DC: American Society for Microbiology Press, 32.
1998:385–7.
14. Moxley RA, Francis DH. Overview of animal models. In: Kaper JB, O’Brien
AD, eds. Escherichia coli O157:H7 and other Shiga toxin–producing E.
coli strains. Washington, DC: American Society for Microbiology Press, 33.
1998:249–60.
15. Wadolkowski EA, Burris JA, O’Brien AD. Mouse model for colonization
JID 2000;181 (February)
and disease caused by enterohemorrhagic Escherichia coli O157:H7. Infect
Immun 1990; 58:2438–45.
Karpman D, Connell H, Svensson M. The role of lipopolysaccharide and
Shiga-like toxin in a mouse model of Escherichia coli O157:H7 infection.
J Infect Dis 1997; 175:611–20.
Downloaded from https://academic.oup.com/jid/article-abstract/181/2/664/824154 by Macau Polytechnic Institute user on 18 September 2019
Kurioka T, Yunou Y, Kita E. Enhancement of susceptibility to Shiga
toxin–producing Escherichia coli O157:H7 by protein calorie malnutrition
in mice. Infect Immun 1998; 66:1726–34.
Wadolkowski EA, Sung LM, Burris JA, Samuel JE, O’Brien AD. Acute renal
tubular necrosis and death of mice orally infected with Escherichia coli
strains that produce Shiga-like toxin type II. Infect Immun 1990; 58:
3959–65.
Acheson DWK, Reidl J, Zhang X, Keusch GT, Mekalanos JJ, Waldor MK.
In vivo transduction with Shiga toxin 1–encoding phage. Infect Immun
1998; 66:4496–8.
Acheson DWK, Frankson K, Willis D, and the STEC Prevalence Study
Group. Multicenter prevalence study of Shiga toxin–producing Escheri-
chia coli [abstract C-205]. In: Proceedings of the 98th general meeting of
the American Society for Microbiology. Washington, DC: American So-
ciety for Microbiology, 1998.
Tesh VL, O’Brien AD. The pathogenic mechanisms of Shiga toxin and the
Shiga-like toxins. Mol Microbiol 1991; 5:1817–22.
Schmidt H, Scheef J, Janetzki-Mittmann C, Datz M, Karch H. An ileX tRNA
gene is located close to the Shiga toxin II operon in enterohemorrhagic
Escherichia coli O157 and non-O157 strains. FEMS Microbiol Lett
1997; 149:39–44.
Sung LM, Jackson MP, O’Brien AD, Holmes RK. Transcription of the Shiga-
like toxin type II and Shiga-like toxin type II variant operons of Escher-
ichia coli. J Bacteriol 1990; 172:6386–95.
Donnenberg MS, Kaper JB. Construction of an eae deletion mutant of en-
teropathogenic Escherichia coli by using a positive-selection suicide vector.
Infect Immun 1991; 59:4310–7.
Jacewicz M, Feldman HA, Donohue-Rolfe A, Balasubramanian KA, Keusch
GT. Pathogenesis of Shigella diarrhea. XIV. Analysis of Shiga toxin re-
ceptors on cloned Hela cells. J Infect Dis 1989; 159:881–9.
Donohue-Rolfe A, Acheson DWK, Kane AV, Keusch GT. Purification of
Shiga toxin and Shiga-like toxins I and II by receptor analog affinity
chromatography with immobilized P1 glycoprotein and production of
cross-reactive monoclonal antibodies. Infect Immun 1989; 57:3888–93.
Calderwood SB, Mekalanos JJ. Iron regulation of Shiga-like toxin expression
in Escherichia coli is mediated by the fur locus. J Bacteriol 1987; 169:
4759–64.
Carter AO, Borczyk AA, Carlson JAK, et al. A severe outbreak of Escherichia
coli O157:H7–associated hemorrhagic colitis in a nursing home. N Engl
J Med 1987; 317:1496–500.
Karch H, Strockbine NA, O’Brien AD. Growth of Escherichia coli in the
presence of trimethoprim-sulfamethoxazole facilitates detection of Shiga-
like toxin producing strains by colony blot assay. FEMS Microbiol Lett
1986; 35:141–5.
Waldor M. Bacteriophage biology and bacterial virulence. Trends Microbiol
1998; 6:295–7.
Kehl KS, Havens P, Behnke CE, Acheson DWK. Evaluation of the Premier
EHEC assay for detection of Shiga toxin–producing Escherichia coli. J
Clin Microbiol 1997; 35:2051–4.
Tschape H, Prager R, Streckel W, Fruth A, Tietze E, Bohme G. Verotoxi-
nogenic Citrobacter freundii associated with severe gastroenteritis and
cases of haemolytic uraemic syndrome in a nursery school: green butter
as the infection source. Epidemiol Infect 1995; 114:441–50.
Paton AW, Paton JC. Enterobacter cloacae producing a Shiga-like toxin
II–related cytotoxin associated with a case of hemolytic-uremic syndrome.
J Clin Microbiol 1996; 34:463–5.