Filamentation by Escherichia coli subverts innate
defenses during urinary tract infection
Sheryl S. Justice*†, David A. Hunstad*‡, Patrick C. Seed*‡§, and Scott J. Hultgren*
Departments of *Molecular Microbiology and ‡Pediatrics, Washington University School of Medicine, St. Louis, MO 63110
Edited by M. J. Osborn, University of Connecticut Health Center, Farmington, CT, and approved November 8, 2006 (received for review July 26, 2006)
To establish disease, an infecting organism must overcome a vast
array of host defenses. During cystitis, uropathogenic Escherichia
coli (UPEC) subvert innate defenses by invading superficial um-
brella cells and rapidly increasing in numbers to form intracellular
bacterial communities (IBCs). In the late stages of the IBC pathway,
filamentous and bacillary UPEC detach from the biofilm-like IBC,
fluxing out of this safe haven to colonize the surrounding epithe-
lium and initiate subsequent generations of IBCs, and eventually
they establish a quiescent intracellular reservoir. Filamentous UPEC
are not observed during acute infection in mice lacking functional
Toll-like receptor 4 (TLR4), suggesting that the filamentous phe-
notype arises in response to host innate immunity. We investi-
gated SulA, a cell division inhibitor associated with the SOS
response, to gain insight into the role of filamentous UPEC in
pathogenesis. A transcriptional reporter from PsulA revealed spatial
and temporal differences in expression within IBCs, and it was
active in the majority of filamentous UPEC. Although UTI89 and
UTI89 ⌬sulA both formed first-generation IBCs equally well, UTI89
⌬sulA was sharply attenuated in formation of second-generation
IBCs and establishment of the quiescent intracellular reservoir. The
virulence of UTI89 ⌬sulA was restored in TLR4-deficient mice,
suggesting that filamentation facilitates the transition to addi-
tional rounds of IBC formation by subverting innate immune
responses. These findings demonstrate that transient SulA-medi-
ated inhibition of cell division is essential for UPEC virulence in the
murine model of cystitis.
bacterial filamentation 兩 SOS response 兩 pathogenesis
T he interaction between bacterium and host ultimately results
in symbiosis, disease, and/or clearance. The innate immune
response comprises a myriad of biochemical (e.g., oxidizing
agents), cytological (e.g., phagocytic cells), and biomechanical
(e.g., exfoliation of epithelial cells) tactics that prevent coloni-
zation and eradicate infectious microorganisms. Polymorphonu-
clear leukocytes (PMNs; neutrophils) are phagocytic cells that
represent a first line of defense, and they are recruited from the
bloodstream directly to the site of invading bacteria. The primary
role for PMNs is the neutralization of invading organisms by
using degradative enzymes, reactive oxygen species, and anti-
microbial peptides. Phagocytosis of bacteria results in the stim-
ulation of apoptosis within the PMN, modulating inflammation
and reducing collateral destruction of host tissues (1). Productive
pathogens use multiple mechanisms to evade and/or cope with
the stresses presented by these host defenses. Urinary tract
infections (UTIs) are an attractive model to study the molecular
details of this delicate balance between host and pathogen
because the bladder employs diverse tactics to combat infectious
organisms. Uropathogenic Escherichia coli (UPEC) have
adapted multiple mechanisms to evade these host defenses.
UTIs are among the most common infections in the United
States. Each year, at least 4 million women seek treatment for
UTIs, and ⬇$1.6 billion will be spent in the diagnosis and
treatment of UTIs (2). UPEC are the cause of 70–90% of all
community-acquired UTIs (2). Clinicians are frustrated by the
rise in antibiotic resistance among organisms that cause UTIs
and by frequent recurrences in healthy adult females after an
19884 –19889 兩 PNAS 兩 December 26, 2006 兩 vol. 103 兩 no. 52
initial UTI. Many women use antibiotics daily to reduce the risk
of recurrence only to suffer recrudescence when this prophylac-
tic regimen is discontinued. The use of in vitro and in vivo models
has allowed the description of events that delineate the acute and
chronic stages of UTI. Mannosylated uroplakin proteins coating
the surface of the superficial umbrella cells of the bladder serve
as the receptors for binding and invasion of UPEC by means of
the FimH adhesin at the tips of type 1 pili (3, 4). During murine
cystitis, intracellular bacterial communities (IBCs) appear ini-
tially as rod-shaped, loosely organized colonies with a doubling
time of 30 min (5). Between 6 and 8 h postinfection, the IBC
matures into a tightly packed, highly organized cluster of coccoid
bacteria with a doubling time greater than 45 min. Approxi-
mately 16 h postinfection, the bacteria on the outer surface of the
IBC regain a rod shape, become motile, detach from the IBC,
and eventually they escape from the superficial umbrella cell (5),
entering into the bladder lumen to attach to a naı̈ve epithelial cell
or to leave the host through micturition. During this escape
(fluxing), filamentous bacteria (up to 70 ␮m in length) are
observed on the luminal surface (5, 6). In the C3H mouse
background, this developmental cascade occurs in two cycles,
with the second generation displaying much slower kinetics than
the first (5). Exfoliation of the superficial umbrella cells occurs
at late stages of the acute infection (6) in an effort to eliminate
the bacterial burden. After superficial umbrella cell exfoliation,
UPEC establish a chronic, quiescent intracellular reservoir
(QIR) within the bladder epithelium (5, 7). Shortly after exfo-
liation, the urine is sterile on most days, but the bladder tissue
maintains a bacterial burden for weeks to months after acute
cystitis (7, 8). This QIR resists treatment with oral antibiotics; it
is apparently undetected by host immunity, and it can be the
source of a recurrent infection (7).
UPEC have developed specific strategies to bypass or over-
come host defenses. The bladder epithelium produces cytokines,
primarily through stimulation of host Toll-like receptor 4
(TLR4) by bacterial LPS, that recruit PMNs to assist in the
elimination of the bacteria. UPEC actively suppress the produc-
tion of cytokines produced by bladder epithelial cells (9). In
addition, the intracellular niche provides a safe haven for UPEC
evasion from infiltrating PMNs (5). Finally, our recent data
suggest that bacterial filamentation might confer resistance to
killing by PMNs (5).
Author contributions: S.S.J. and D.A.H. designed research; S.S.J. and D.A.H. performed
research; P.C.S. contributed new reagents/analytic tools; S.S.J., D.A.H., and S.J.H. analyzed
data; and S.S.J., D.A.H., and S.J.H. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS direct submission.
Abbreviations: IBC, intracellular bacterial community; MMC, mitomycin C; PMN, polymor-
phonuclear leukocyte (neutrophil); QIR, quiescent intracellular reservoir; UPEC, uropatho-
genic E. coli; UTI, urinary tract infection.
†Towhom correspondence should be addressed at the present address: Center for Micro-
bial Pathogenesis, Columbus Children’s Research Institute, 700 Children’s Drive, W532,
Columbus, OH 43205. E-mail: justices@ccri.net.
§Present address: Department of Pediatrics, Duke University, Durham, NC 27710.
© 2006 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0606329104
Fig. 1. Microscopic examination of IBC formation and UPEC filamentation. Female 7- to 9-week old C3H/HeN (a–e) or C3H/HeJ ( f) mouse strains were infected
with 107 cfu of UTI89/pCOMGFP (a and c) or UTI89 ⌬sulA/pCOMGFP (b, d, and f ) or UTI89 (e; red) plus UTI89 ⌬sulA/pCOMGFP (e; green) by transurethral
catheterization. Bladders were harvested at 6 h (a and b), 16 h (e), or 24 h (c, d, and f ) postinfection, and they were prepared as described in Materials and Methods
for visualization by confocal laser scanning fluorescent microscopy with three-dimensional reconstruction or as an optical slice (d). (Scale bars: 30 ␮m.)
Bacteria have developed multiple strategies to cope with
environmental stresses. DNA damage can result from certain
environmental stresses or insults (e.g., UV light, oxidative
conditions). The SOS response is one such strategy that is
designed to inhibit cell division while damaged DNA is repaired,
thereby preventing transmission of deleterious mutations to
daughter cells (for review, see ref. 10). On a molecular level,
mutations resulting in mismatched base pairs produce regions of
single-stranded DNA. RecA, the major bacterial recombinase,
binds single-stranded regions, and it is activated in the presence
of nucleotide triphosphates. Activated RecA stimulates the
autoproteolysis of the SOS transcriptional repressor, LexA.
LexA represses the transcription of at least 17 genes that are
involved in DNA repair, and it has recently been implicated in
the regulation of ⬎30 unlinked genes (11). The LexA regulon
includes a cell division inhibitor, SulA, to prevent transmission
of mutant DNA to new daughter cells. SulA specifically inhibits
FtsZ polymerization (12, 13), preventing FtsZ ring formation
(14–16) and thereby blocking septation. Once DNA repair is
complete, LexA repression of the SOS genes is restored. In
addition, Lon, the general cytoplasmic protease, degrades SulA,
restoring cell division capacity.
UPEC filamentation may serve as an additional mechanism to
evade host innate immune responses. The current study directly
tested a role for SulA-mediated cell division inhibition in the
pathogenic lifestyle of UPEC. Inactivation of sulA in UPEC sharply
attenuated virulence in the murine cystitis model, but only when the
TLR4-mediated response was intact, demonstrating that filamen-
tation is a critical component of the bacterial arsenal against the
innate immune response. Induction of the SOS response in vitro was
not sufficient to protect UPEC from PMN phagocytosis, suggesting
that factors in addition to filamentous morphology are required for
evasion of innate immunity.
Results
SulA Mediates Filamentation in Vivo. Bacterial filamentation has
been proposed as a mechanism of evasion of the innate immune
Justice et al.
MICROBIOLOGY
response in murine cystitis (5). In addition, filamentation may
play a role in human cystitis because filamentous UPEC are
observed in the urine of UTI patients (D. Rosen and S.J.H.,
unpublished data). To assess directly a role for filamentation in
cystitis, the bacterial system responsible for filamentation was
investigated. Because PMNs use oxidative molecules to eradicate
bacterial pathogens, the oxidative environment of the host
inflammatory response could lead to mutation of bacterial
DNA. DNA repair would be accomplished by induction of the
SOS response, ultimately leading to bacterial filamentation
caused by the cell division inhibitor, SulA (17). Therefore, a
kanamycin cassette disruption of sulA was transduced into our
prototypic cystitis UPEC isolate, UTI89. Cystitis was induced in
7- to 9-week-old female C3H/HeN mice by transurethral inoc-
ulation of 107 cfu of statically grown UTI89 or UTI89 ⌬sulA. The
ability of the sulA mutant to proceed through all stages of the
developmental pathway was evaluated by fluorescence micros-
copy or by recovery of cfu from the bladder. UTI89 and UTI89
⌬sulA were transformed with pCOMGFP (18) for ease of
visualization of intracellular bacteria by fluorescence micros-
copy. During the first 16 h postinfection, the size and morphol-
ogy of IBCs were indistinguishable between UTI89 and UTI89
⌬sulA (Fig. 1 a and b), suggesting that UTI89 ⌬sulA is competent
for first-generation IBC formation. Consistent with these mi-
croscopic results, the total numbers of bacteria recovered at 6
and 16 h from the bladders infected with UTI89 ⌬sulA were
indistinguishable from bladders infected with UTI89 (Fig. 2a).
However, at 16 h, when filamentous UTI89 is readily observed
in infected bladders (Fig. 1e), UTI89 ⌬sulA appeared only in
bacillary form, both intracellularly and on the luminal surface.
To characterize more fully the absence of filamentation in UTI89
⌬sulA, bladders coinfected with equivalent numbers of UTI89
and UTI89 ⌬sulA/pCOMGFP were studied by fluorescence
microscopy at 16 h postinfection. Wild-type UTI89 displayed red
fluorescence caused by the counterstaining of whole mounts
PNAS 兩 December 26, 2006 兩 vol. 103 兩 no. 52 兩 19885

Fig. 2. Virulence of UTI89 ⌬sulA. Female C3H/HeN (a) or C3H/HeJ (b) mice
were transurethrally inoculated with 107 UTI89 (Š), UTI89 ⌬sulA (‹), or UTI89
⌬lon (⽧). Bladders were harvested at 6 h, 16 h, 24 h, 48 h, or 2 weeks
postinfection and homogenized, and viable bacteria were enumerated as cfu
per bladder. Each graph is representative of three independent experiments.
The statistical significance was assayed by using the Mann–Whitney test (ns,
not significant; *, P ⫽ 0.03; **, P ⫽ 0.004; nd, not determined).
with a monomeric cyanine nucleic acid dye after paraformalde-
hyde fixation. Surface-associated UTI89 was observed in both
bacillary and filamentous forms; however, only bacillary UTI89
⌬sulA/pCOMGFP was observed on the surface (Fig. 1e). The
absence of filamentous UTI89 ⌬sulA during first-generation IBC
formation suggests that SulA is required for filamentation in
vivo.
Fluxing at the end of first-generation IBC formation leads to
invasion of naı̈ve superficial umbrella cells and generation of a
large number of second-generation IBCs (5). In marked contrast
to infections with UTI89 (Fig. 1c), UTI89 ⌬sulA was primarily on
the luminal surface at 24 h (Fig. 1d). Only four IBCs were
observed in 10 bladders infected with UTI89 ⌬sulA alone at 24 h
postinfection; typically, UTI89 forms 50 to several hundred IBCs
per bladder. In addition, at time points corresponding to the
second round of IBC formation (24 and 48 h), there was a
decrease of 5–7 orders of magnitude in cfu when bladders were
infected with UTI89 ⌬sulA compared with bladders infected
with UTI89 (Fig. 2a). Thirty percent of the bladders infected
with UTI89 ⌬sulA were sterile at 48 h postinfection. Sterile
bladders are rarely, if ever, observed at this time point after
infection with wild-type UTI89. These data suggest that SulA is
critical for UTI89 pathogenesis after first-generation IBC for-
mation. This time frame is the same as when filamentous bacteria
are typically observed, suggesting that the inability of UTI89
⌬sulA to filament dramatically attenuates virulence.
At 36–48 h postinfection, when filamentous bacteria are ob-
served in UTI89-infected bladders (5), filamentous UTI89 ⌬sulA
was not observed (data not shown), although the UTI89 ⌬sulA
19886 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0606329104
bacterial load was much lower than in wild-type infection at this
time point (Fig. 2a). After the second generation of IBC formation,
UPEC enter into a QIR (5, 7, 19). Bacteria were not recovered from
any of the bladders infected with UTI89 ⌬sulA at 2 weeks postin-
fection, whereas bladders infected with UTI89 all contained viable
cfu, reflecting the presence of a QIR (Fig. 2a) (103 to 105 cfu per
bladder) as observed in our earlier studies (7).
The inability of the sulA mutant to initiate effectively second-
generation IBC formation suggests that SulA-mediated bacterial
filamentation is important for the transition between genera-
tions of IBC formation. Moreover, the transition into a QIR is
also preceded by the emergence of a large number of filamentous
bacteria on the luminal surface (5). Thus, the pathogenesis of
murine cystitis hinges on SulA-mediated filamentation, a phe-
notype that might underlie reentry into naı̈ve superficial um-
brella cells or confer resistance to host immune responses,
including PMN engulfment and killing.
SulA Is Not Required in the Absence of Functional TLR4. PMN influx
into the bladder during UTI is greatly impaired in the C3H/HeJ
mouse strain (20) because of a mutation in the cytoplasmic tail
of TLR4, blocking transduction of LPS-initiated signals (21, 22).
Our previous studies indicated that filamentous UTI89 are not
observed during the acute infection in the C3H/HeJ mouse strain
(5). Therefore, we hypothesized that inactivation of sulA would
have no effect on the course of cystitis in the C3H/HeJ mouse.
UTI89 ⌬sulA persisted to the same extent as UTI89 after
inoculation into the bladders of C3H/HeJ mice (Fig. 2b) from 6 h
to 2 weeks. IBCs formed by UTI89 ⌬sulA in C3H/HeJ bladders
were similar in number and architecture at 24 h postinfection to
those observed in C3H/HeJ bladders infected with UTI89 (Fig.
1f ). Moreover, the ability of UTI89 ⌬sulA to establish a QIR was
restored in the Tlr4⫺/⫺ mouse strain (Fig. 2b), suggesting that the
requirement for SulA-mediated filamentation during IBC mat-
uration and establishment of cystitis relates to evasion of host
immunity.
UPEC Virulence Requires Recovery from Filamentation. Previous
studies indicated that cell division capacity was restored in
filamentous UPEC during UTI (5). To determine whether
restoration of cell division within the filamentous population is
essential for uropathogenesis, lon was inactivated in UTI89 to
prevent recovery from SulA-mediated filamentation. The cyto-
plasmic protease, Lon, degrades SulA to restore FtsZ polymer-
ization and septation after repair of damaged DNA (23). UTI89
⌬lon displayed a defect in cfu per bladder of 5 orders of
magnitude in C3H/HeN mice at 48 h (Fig. 2a), but virulence of
UTI89 ⌬lon was restored in the C3H/HeJ mice (Fig. 2b), similar
to that observed with UTI89 ⌬sulA infection.
When filamentous UTI89 on the bladder surface is visualized by
SEM, the majority of filaments exhibit a smooth surface (5). This
observation is consistent with previous studies reporting a smooth
surface for E. coli that have undergone filamentation mediated by
SulA (5, 6). However, occasionally, a partial septum is observed on
the surface of filamentous UPEC by SEM. Lutkenhaus and col-
leagues (24) have previously demonstrated that disassembly of
preformed septation complexes results in an aborted septum that
cannot be used in further septation events. Thus, the majority of
daughter cells arising after Lon-mediated restoration of cell division
would display the normal cell length that we observed. SulA-
mediated division inhibition has been described as a transient
process to prevent septation temporarily during the repair of DNA
damage (25). Our data suggest that the filamentation that occurs
during pathogenesis must also be a transient process and that
septation must be restored for continued progression through the
IBC cascade during acute infection.
Justice et al.
Fig. 3. Expression of sulA in vivo. Female C3H/HeN (a–d) or C3H/HeJ (e and f ) mice were transurethrally inoculated with 107 UTI89 attB::PsulA-gfp. Bladders were
harvested at 4 h (a), 6 h (b), or 16 (c–f ), and they were prepared as described in Materials and Methods for visualization by laser scanning fluorescent microscopy.
They were visualized with three-dimensional reconstruction (a–d) or as an optical slice (e and f ). (Scale bars: a, b, e, and f, 50 ␮m; c and d, 30 ␮m.)
sulA Is Expressed in Vivo. To demonstrate further that SulA is
involved in UPEC pathogenesis, the in vivo transcriptional
profile of sulA was evaluated. To observe directly expression of
sulA during UTI pathogenesis in vivo, the sulA promoter was
used to drive expression of the gene encoding GFP from a single
copy integrated at the attB site. In vitro, UTI89 attB::PsulA-gfp
exhibited strong fluorescence only on exposure to SOS-inducing
agents, e.g., mitomycin C (MMC; data not shown). Expression of
sulA was readily observed within a minority of the individual
bacteria in early IBCs in C3H/HeN mice infected with UTI89
attB::PsulA-gfp (Fig. 3 a and b). The number of bacteria within an
IBC that expressed sulA varied among IBCs depending on their
size and maturation state. Approximately 60% of the ⬎800 IBCs
observed within 10 murine bladders infected with UTI89
attB::PsulA-gfp exhibited differential sulA expression, both tem-
porally and spatially, by confocal microscopy. Fluorescence
emitted from GFP was also observed in the majority of filamen-
tous E. coli (Fig. 3 c and d) analyzed in 15 murine bladders
infected with UTI89 attB::PsulA-gfp, consistent with the hypoth-
esis that expression from PsulA leads to filamentation of UPEC
during cystitis. Infection of C3H/HeJ (Tlr4⫺/⫺) mice with UTI89
attB::PsulA-gfp did not lead to detectable expression of sulA
within ⬎600 IBCs analyzed (Fig. 3 e and f ). Thus, TLR4-
dependent innate responses lead to induction of sulA expression,
likely reflecting induction of the SOS response caused by host
production of DNA-damaging (e.g., oxidative) agents.
SulA Overproduction Does Not Protect from PMNs in Vitro. Our data
suggest that SulA is critical in uropathogenesis, but they do not
directly address the function of SulA in subverting innate
defenses in vivo. Previous studies have suggested that filamen-
tous UTI89 evades PMN killing (5). Current methods do not
permit isolation of sufficient numbers of filamentous bacteria
from the urine and/or bladders to assess directly the ability of in
vivo-produced filamentous UPEC to survive PMN phagocytosis
ex vivo. To circumvent this problem in vitro, UTI89 was trans-
formed with a plasmid (pGC165) that encodes sulA under
Justice et al.
MICROBIOLOGY
control of Plac. Filamentous and bacillary UTI89 were opsonized
with 50% normal human serum and challenged with purified
human PMNs. There was no apparent difference in the ability of
the PMNs to engulf and kill the filamentous and bacillary forms
of UTI89 (Fig. 4 a and b), suggesting that filamentation per se is
not sufficient for protection from PMNs. The observation that
size alone is not sufficient to protect the bacteria from engulf-
ment is not surprising, considering that PMNs engulf the much
larger hyphal form of Candida albicans (26). The ability of PMNs
to engulf the SulA-induced filaments in vitro suggests that there
are additional factors coregulated with the filamentous response,
needed to subvert PMNs. Such factors might represent other
components of the SOS regulon, or they may be produced
independent of the SOS response. Alternatively, the extended
length might provide a more secure attachment to the bladder
epithelium, with increased number of adhesins over a greater
surface area that would make removal by PMNs more difficult.
To test whether chemical induction of the SOS response
provides protection from PMN phagocytosis in vitro, UTI89 was
grown under static conditions with the addition of 300 ng/ml
MMC in early log phase at 37°C. PMNs were isolated from both
humans and mice as described in ref. 27, and UTI89 was grown
in the presence or absence of MMC and opsonized with normal
serum or mixed directly with human or mouse PMNs. As
visualized by microscopy, both human and mouse PMNs were
able to engulf both opsonized (data not shown) and nonopso-
nized (Fig. 4 c–f ) filamentous and bacillary UTI89 readily. From
these in vitro data, it appears that a putative bacterial component
for protection is not induced by brief exposure to a chemical
DNA-damaging agent. Additionally, it is unknown whether the
observed induction of sulA in vivo occurs as a component of the
SOS response. LexA-dependent induction mediated by DNA
damage is the only known mechanism that leads to sulA expres-
sion. Recently, microarray analysis of the SOS response in
laboratory (K-12) strains of E. coli have demonstrated that
additional genes are transcriptionally active on initiation of DNA
damage (11). Given the increased size of the UTI89 genome
PNAS 兩 December 26, 2006 兩 vol. 103 兩 no. 52 兩 19887
Fig. 4. PMN phagocytosis in vitro. Human and mouse PMNs were incubated with UTI89 under inducing conditions for 20 min, and they were visualized by using
phase microscopy (a, b, e, and f ) or fluorescent microscopy (c and d). (a) Opsonized, uninduced UTI89 with human PMNs. (b) Opsonized induced UTI89 Plac::sulA.
(c and d) MMC-induced UTI89 with mouse PMNs. (e) Uninduced UTI89 with human PMNs. ( f) MMC-induced UTI89 with human PMNs. (Scale bars: a–d, 30 ␮m;
e and f, 50 ␮m.)
relative to E. coli K-12 (28), it is likely that additional genes may
be present in the SOS regulon within pathogenic organisms.
However, the presence of other signals that induce sulA in vivo
cannot be eliminated. SOS induction for periods longer than 4 h
may result in the expression of additional molecules that may be
protective. The putative bacterial component for PMN protec-
tion may be produced in response to the same signal that results
in PsulA expression, or it may occur through an alternative signal
that occurs coincidently with the signal that induces PsulA in vivo.
Discussion
The data presented here provide strong evidence that filamen-
tation, mediated by SulA, is a vital component of the UPEC
pathogenic cascade within an immunocompetent host. The
transition of UPEC from the first to second generation of IBC
formation depends on SulA-mediated division inhibition and
Lon-mediated restoration of cell division competence. This
conclusion is based on the observation of a rapid decline in cfu
per bladder 24 h postinfection, which is the time point associated
with formation of second-generation IBCs (5). The restoration
of virulence of UTI89 ⌬sulA in C3H/HeJ mice argues that
SulA-mediated filamentation protects UPEC from components
of host innate defenses. However, overproduction of SulA or
chemical induction of the SOS response was insufficient to
confer protection from phagocytosis in vitro. Although sulA is
expressed in vivo during cystitis, the mechanism of its induction
is unknown. We propose that components of the innate immune
response (e.g., oxidative stresses) likely induce the SOS response
in vivo but that sulA induction in vivo might instead rely on signals
independent of the SOS pathway. Our data suggest that the
filamentous phenotype that is observed at the transition between
stages of murine cystitis is more than a morphologic one and that
filamentation may provide multiple means of protection for
UPEC. For example, previous reports have suggested that
induction of the SOS response and subsequent SulA-mediated
division inhibition protect against antibiotics that act on dividing
19888 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0606329104
cells (29). In addition, the potential roles of UPEC filamentation
in resistance to killing mediated by neutrophil extracellular traps
(30) and by antimicrobial peptides such as cathelicidin (31)
warrant further investigation.
In summary, we have demonstrated that in response to innate
immunity, UPEC employs filamentation as a means to evade this
immune response. Characterization of putative host signals and
further understanding of the regulation of sulA will offer insight
into the pathogenesis of diverse bacterial infections and will
foster the design of novel therapeutic agents.
Materials and Methods
Strains and Media. Bacterial strains were grown in Luria broth
statically at 37°C for 16 h to facilitate type 1 pilus production
(32). UTI89 is a clinical UPEC isolate obtained from a patient
with cystitis (8). The sulA::kan insertional mutation (33) was
transduced into UTI89 by using P1 transduction (34), and
transductants were selected by using 50 ␮g/ml kanamycin sulfate.
Transcriptional activity of sulA in vivo was assessed by introduc-
tion of the sulA promoter driving expression of the gene encod-
ing for green fluorescent protein (gfp) in single copy into UTI89.
Briefly, the sulA promoter region (⫺322 to ⫹51 relative to the
start ATG) was amplified by PCR from UTI89 genomic DNA by
using primers ACTGTATCGATCAACGGTCAGGCTGTA-
ACTACCA A A and ATGA ACCCGGGTGCTGTGGAT-
GAGAACGACGAATCA. The 373-bp product was restricted
with ClaI and SmaI, cloned into the same sites of pPSSH10,
sequenced for verification, and integrated into the ␭ attachment
site of E. coli MG1655 as described in ref. 35. The fusion was
moved into UTI89 by P1 phage transduction. Plasmid pGC165
contains sulA under the control of the Plac promoter (36). To
induce the SOS response in vitro, saturated cultures were diluted
1:1,000 and allowed to grow with aeration or statically at 37°C
until they reached early log phase (A600 ⫽ ⬇0.3). MMC (Sigma–
Aldrich, St. Louis, MO) was added to a final concentration of 300
ng/ml, and growth continued under the same conditions for
2–4 h.
Justice et al.
Murine Cystitis. Female C3H/HeN (Harlan, Indianapolis, IN) or
C3H/HeJ (Jackson Laboratories, Bar Harbor, ME) mice, 7–9 weeks
old, were infected with 107 cfu of E. coli transurethrally, in a volume
of 50 ␮l as described in ref. 6. At the times indicated, bladders were
aseptically harvested and homogenized for bacterial enumeration;
or they were bisected, splayed, and fixed in 3% paraformaldehyde
for 1 h for visualization by laser scanning confocal fluorescent
microscopy (Zeiss LSM510 Meta; Carl Zeiss, Inc., Thornwood,
NY). Fixed bladders were treated with 0.01% Triton X-100 in PBS
for 10 min, and then they were counterstained with a red fluores-
cent monomeric cyanine nucleic acid dye (ToPro3; Molecular
Probes, Eugene, OR) for an additional 10 min. Three-dimensional
image reconstruction was performed by using Volocity software
(Improvision, Lexington, MA).
PMN Isolation. Human blood was collected from peripheral veins
of healthy donors in accordance with a protocol approved by the
1. Kobayashi SD, Voyich JM, DeLeo FR (2003) Microbes Infect 5:1337–1344.
2. Foxman B (2003) Dis Mon 49:53–70.
3. Zhou G, Mo WJ, Sebbel P, Min G, Neubert TA, Glockshuber R, Wu XR, Sun
TT, Kong XP (2001) J Cell Sci 114:4095–4103.
4. Martinez JJ, Mulvey MA, Schilling JD, Pinkner JS, Hultgren SJ (2000) EMBO
J 19:2803–2812.
5. Justice SS, Hung C, Theriot JA, Fletcher DA, Anderson GG, Footer MJ,
Hultgren SJ (2004) Proc Natl Acad Sci USA 101:1333–1338.
6. Mulvey MA, Lopez-Boado YS, Wilson CL, Roth R, Parks WC, Heuser J,
Hultgren SJ (1998) Science 282:1494–1497.
7. Schilling JD, Lorenz RG, Hultgren SJ (2002) Infect Immun 70:7042–7049.
8. Mulvey MA, Schilling JD, Hultgren SJ (2001) Infect Immun 69:4572–4579.
9. Hunstad DA, Justice SS, Hung CS, Lauer SR, Hultgren SJ (2005) Infect Immun
73:3999–4006.
10. Janion C (2001) Acta Biochim Pol 48:599–610.
11. Quillardet P, Rouffaud MA, Bouige P (2003) Res Microbiol 154:559–572.
12. Mukherjee A, Cao C, Lutkenhaus J (1998) Proc Natl Acad Sci USA 95:2885–
2890.
13. Trusca D, Scott S, Thompson C, Bramhill D (1998) J Bacteriol 180:3946–3953.
14. Bi E, Lutkenhaus J (1993) J Bacteriol 175:1118–1125.
15. Justice SS, Garcia-Lara J, Rothfield LI (2000) Mol Microbiol 37:410–423.
16. Romberg L, Levin PA (2003) Annu Rev Microbiol 57:125–154.
17. Huisman O, D’Ari R, George J (1980) J Bacteriol 144:185–191.
18. Valdivia RH, Hromockyj AE, Monack D, Ramakrishnan L, Falkow S (1996)
Gene 173:47–52.
19. Mysorekar IU, Hultgren SJ (2006) Proc Natl Acad Sci USA 103:14170–14175.
20. McAdam KP, Ryan JL (1978) J Immunol 120:249–253.
Justice et al.
Institutional Review Board for human subjects research. Alter-
natively, female mice were treated with heparin (200 units s.c.)
for 10 min, then blood was harvested by cardiac puncture. Mouse
and human PMNs were isolated as described in ref. 27. Where
indicated, bacteria were opsonized with 50% normal human
serum in PBS for 30 min at 37°C before incubation with isolated
PMNs. Bacteria and PMNs were coincubated in RPMI medium
1640 with 10 mM unbuffered Hepes for up to 30 min at 37°C.
Samples were analyzed at specified intervals by using phase
microscopy or epifluorescent microscopy.
We thank L. Rothfield and M. J. Osborn for invaluable discussions. This
work was supported by National Institutes of Health Grants F32-
DK10168 (to S.S.J.), K08-DK067894 (to D.A.H.), K12-HD00850 (to
P.C.S.), and R01-DK51406 (to S.J.H.) and Office of Research on
Women’s Health Specialized Center of Research Grant P50-DK64540
with the Food and Drug Administration (to S.J.H.).
21. Vogel SN, Johnson D, Perera PY, Medvedev A, Lariviere L, Qureshi ST, Malo
D (1999) J Immunol 162:5666–5670.
22. Hoshino K, Takeuchi O, Kawai T, Sanjo H, Ogawa T, Takeda Y, Takeda K,
Akira S (1999) J Immunol 162:3749–3752.
23. Schoemaker JM, Gayda RC, Markovitz A (1984) J Bacteriol 158:551–561.
24. Addinall SG, Cao C, Lutkenhaus J (1997) J Bacteriol 179:4277–4284.
25. Maguin E, Lutkenhaus J, D’Ari R (1986) J Bacteriol 166:733–738.
26. Cockayne A, Odds FC (1984) J Gen Microbiol 130:465–471.
27. Boyum A (1968) Scand J Clin Lab Invest Suppl 97:77–89.
28. Chen SL, Hung CS, Xu J, Reigstad CS, Magrini V, Sabo A, Blasiar D, Bieri
T, Meyer RR, Ozersky P, et al. (2006) Proc Natl Acad Sci USA 103:5977–5982.
29. Miller C, Thomsen LE, Gaggero C, Mosseri R, Ingmer H, Cohen SN (2004)
Science 305:1629–1631.
30. Brinkmann V, Reichard U, Goosmann C, Fauler B, Uhlemann Y, Weiss DS,
Weinrauch Y, Zychlinsky A (2004) Science 303:1532–1535.
MICROBIOLOGY
31. Chromek M, Slamova Z, Bergman P, Kovacs L, Podracka L, Ehren I, Hokfelt
T, Gudmundsson GH, Gallo RL, Agerberth B, Brauner A (2006) Nat Med
12:636–641.
32. Langermann S, Palaszynski S, Barnhart M, Auguste G, Pinkner JS, Burlein J,
Barren P, Koenig S, Leath S, Jones CH, Hultgren SJ (1997) Science 276:607–
611.
33. D’Ari R, Huisman O (1983) J Bacteriol 156:243–250.
34. Silhavy TJ, Berman, ML, Enquist, LW (1984) (Cold Spring Harbor Lab Press,
Cold Spring Harbor, NY), pp 111–112.
35. Wright KJ, Seed PC, Hultgren SJ (2005) Infect Immun 73:7657–7668.
36. Huisman O, D’Ari R, Gottesman S (1984) Proc Natl Acad Sci USA 81:4490–
4494.
PNAS 兩 December 26, 2006 兩 vol. 103 兩 no. 52 兩 19889