J Neuropathol Exp Neurol
Copyright Ó 2009 by the American Association of Neuropathologists, Inc.
Vol. 68, No. 6
June 2009
pp. 677Y690
ORIGINAL ARTICLE
OmpA Is the Critical Component for Escherichia coli
InvasionYInduced Astrocyte Activation
Hsueh-Hsia Wu, MS, Yi-Yuan Yang, PhD, Wen-Shyang Hsieh, MS, Chi-Hsin Lee, BS,
Sy-Jye C. Leu, PhD, and Mei-Ru Chen, PhD
From the Graduate Institute and Department of Microbiology, College of
Medicine, National Taiwan University (H-HW, M-RC); School of
Medical Laboratory Science and Biotechnology, College of Medicine,
Taipei Medical University (H-HW, Y-YY); Department of Laboratory
Medicine, Mackay Memorial Hospital (W-SH); and Graduate Institute
of Medical Sciences, College of Medicine, Taipei Medical University
(C-HL, S-JCL), Taipei, Taiwan.
Send correspondence and reprint requests to: Mei-Ru Chen, PhD, Graduate
Institute and Department of Microbiology, College of Medicine,
National Taiwan University, No 1, 1st section, Jen-Ai Road, Taipei
100, Taiwan; E-mail: [email protected]
Supported by Grant Nos. NSC92-2320-B-038-056 (to H-HW) and NSC952320-B-002-087-MY3 (to M-RC) from National Science Council and
Grant No. 9453 (to W-SH) from Mackay Memorial Hospital.
Online-only color figures are available at http://www.jneuropath.com.
ated with high mortality and morbidity. Most importantly,
approximately half of the survivors display adverse neurologic complications. The pathologic complications of bacterial meningitis include cerebritis, brain abscess, empyema,
and ventriculitis in the acute phase and the sequelae of cerebral atrophy and hydrocephalus (1). Neuronal injury associated with bacterial central nervous system (CNS) infection
involves multiple microbial and host factors. For an understanding of the pathogenesis of these complications, the
interactions between bacterial components and host cell
responses in the CNS need to be elucidated.
Escherichia coli strains with the K1 capsular polysaccharide are the most predominant Gram-negative bacteria associated with neonatal bacterial meningitis (2). The
meningitis-associated E. coli K1 strain can translocate from
the bloodstream to the CNS without disrupting the integrity
of the blood-brain barrier (BBB) (3). A high level of bacteremia and invasion through brain microvascular endothelial
cells (BMECs) seem to be determining factors that contribute to CNS infection (3). Several K1-associated components
participate in BMEC binding and invasion; these include
Fim H (4), K1 capsule (5), and outer membrane protein A
(OmpA) (6Y8).
Outer membrane protein A is one of the major outer
membrane proteins of E. coli and plays important roles in
maintaining the integrity of outer membrane and in bacterial
conjugation (9Y11). It is also the receptor for several bacteriophages (12Y15). Outer membrane protein A is encoded
by a 1038-bp open reading frame that consists of a 21-aminoacid leader peptide and the mature 325-amino-acid protein.
The N-terminal membraneYanchoring domain forms an antiparallel A-barrel, which has 8 transmembrane A-strands connected by 3 short periplasmic turns and 4 relatively large,
surface-exposed hydrophilic loops; the C-terminal domain interacts with the peptidoglycan layer in the periplasm to maintain outer membrane integrity (16). Outer membrane protein A
is highly conserved through the evolution of Gram-negative
bacteria and is important for the binding of E. coli to, and the
invasion of, BMECs (17).
Beyond the BBB, the most abundant cells in the CNS
are astrocytes that provide physical and nutritional support
for neurons. CNS injury results in astrocyte proliferation and
morphologic changes with enhanced expression of glial fibrillary acidic protein (GFAP), the major constituent of glial
intermediate filaments in mature astrocytes (18, 19). Astrocytes
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
677
Abstract
Escherichia coli is the major Gram-negative bacterial pathogen in
neonatal meningitis. Outer membrane protein A (OmpA) is a conserved major protein in the E. coli outer membrane and is involved
in several host-cell interactions. To characterize the role of OmpA in
the invasion of astrocytes by E. coli, we investigated OmpA-positive
and OmpA-negative E. coli strains. Outer membrane protein A+
E44, E105, and E109 strains adhered to and invaded C6 glioma cells
10- to 15-fold more efficiently than OmpA-negative strains. Actin
rearrangement, protein tyrosine kinase, and phosphoinositide
3Ykinase activation were required for OmpA-mediated invasion by
E. coli. In vitro infection of C6 cells and intracerebral injection into
mice of the E44 strain induced expression of the astrocyte differentiation marker glial fibrillary acidic protein and the inflammatory
mediators cyclooxygenase 2 and nitric oxide synthase 2. After intracerebral infection with E44, all C57BL/6 mice died within
36 hours, whereas 80% of mice injected with E44 premixed with
recombinant OmpA protein survived. Astrocyte activation and
neutrophil infiltration were reduced in brain tissue sections in the
mice given OmpA. Taken together, these data suggest that OmpAmediated invasion plays an important role in the early stage of
E. coliYinduced brain damage, and that it may have therapeutic use
in E. coli meningitis.
Key Words: Astrocyte, Cyclooxygenase 2, Escherichia coli, Glial
fibrillary acidic protein, Nitric oxide synthase 2, Outer membrane
protein A.
INTRODUCTION
Despite advances in anti-microbial treatments and intensive care, bacterial meningitis remains a disease associ-
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
are also involved in intracerebral immune regulation (20).
Activated astrocytes express cyclooxygenase 2 (COX-2) and
nitric oxide synthase 2 (NOS2) (21, 22), which may subsequently lead to damage of neurons and glial cells, activation
of BMECs, and overamplified inflammation responses
(23Y26). The initial response of astrocytes to E. coli infection
and the critical bacterial component(s) involved are not
presently known.
To provide insight into the pathogenic mechanisms
of inflammation of the CNS, we determined whether OmpA
protein mediates the invasion of astrocytes by E. coli. The
invasion-induced activation of astrocytes was demonstrated
by the expression of GFAP, COX-2, and NOS-2. The effects
of recombinant OmpA protein on intracerebral E. coli
challenge in C57BL/6 mice were also examined.
with individual bacterial strains at multiplicity of infection
(moi) of 0.1, 1, or 10 for 2 hours at 37-C, or at moi 10 for
0.5, 1, 1.5, 2, or 3 hours. Monolayers of cells were washed
with culture medium 3 times and lysed in 0.5% Triton X100. The released bacteria were enumerated by plating on
sheep blood agar plates (BAPs; Difco Laboratories). Assays were performed in duplicate and repeated at least 3
times. For intracellular bacteria assays, confluent monolayer cells (grown in 6-well plates) were incubated with
bacteria at moi of 0.1, 1, or 10 for 2 hours at 37-C or at
moi 10 for 0.5, 1, 1.5, 2, or 3 hours. The cells were washed
with culture medium 3 times and incubated with culture
medium containing 100 Kg/ml of gentamicin (Life Technologies, Gaithersburg, MD) for 2 hours to kill the extracellular bacteria. The monolayer cells were then washed 3
times and lysed in 0.5% Triton X-100. The released bacteria
were enumerated by plating on BAP. To examine the signaling pathways involved in the invasion process, C6 cells
were pretreated with individual inhibitor (Cytochalasin D,
genistein, or LY294002, all from Calbiochem, San Diego,
CA) at different concentrations for 1 hour before infection.
Cytochalasin D was dissolved in sterilized water, whereas
genistein and LY294002 were dissolved in dimethyl sulfoxide. Assays were performed in duplicate and repeated at
least 3 times.
MATERIALS AND METHODS
Cell Line
C6, a rat glioma cell line (ATCC CCL-107), was
purchased from Bioresource Collection and Research Center,
Hsinchu, Taiwan. Cells were cultured in Ham F10 medium
with 15% horse serum, 2.5% fetal bovine serum, and 50 U/ml
of penicillin-streptomycin (all from Life Technologies,
Gaithersburg, MD).
Chemicals, Bacteria, and Culture Medium
Chemicals were all purchased from Sigma (St.
Louis, MO) unless otherwise indicated. All Escherichia
coli strains were kindly provided by Dr. K.S. Kim (Division of Pediatric Infectious Diseases, School of Medicine,
Johns Hopkins University, Baltimore, MD); they are summarized in Table 1. E44 is K1 strain RS218 (O18:K1:H7)
isolated from the cerebrospinal fluid of a neonate with meningitis (17). E91 is an E44 mutant lacking the entire ompA
gene. E105 is an OmpA complemented mutant of E91 that
carries pRD87 (which contains the ompA gene on pUC9).
E111 is E91 with pUC9 vector control. E109 was obtained
by transformation of E91 with a pRD87 derivative that
lacks the C-terminal 53 amino acids of OmpA. MG1655
is a nonpathogenic strain that is noninvasive for the BBB.
Bacteria were grown in brain heart infusion broth with appropriate antibiotics (all from Difco Laboratories, Detroit,
MI). For infection experiments, overnight cultures were
expanded in brain heart infusion broth and incubation at
37-C for 2 to 3 hours to mid-log phase. Bacteria were
centrifuged and resuspended in cell culture medium without antibiotics.
Mouse Strain
C57BL/6 mice were obtained from the National Laboratory Animal Center of Taiwan and kept under pathogen-free
conditions. Animal procedures were performed in accordance
with an approved protocol (LAC-96-0031) under the institutional protocol of Taipei Medical University.
Adhesion and Invasion Assays
For cell-associated bacteria studies, confluent cultures of
C6 cells grown in 6-well plates (Corning, NY) were incubated
678
Intracellular Bacteria Survival
Confluent monolayer cells grown in 6-well plates were
infected with bacteria at moi 10 for 2 hours. Cells were
washed 3 times and then incubated with culture medium
containing 100 Kg/ml of gentamicin for 2 hours to kill the
extracellular bacteria. The cells were then incubated with
medium containing 20 Kg/ml of gentamicin for 2, 4, 24, or
48 hours, and recoverable colony-forming units (CFUs) were
determined on BAPs. Assays were performed in duplicate
and repeated at least 3 times.
Electron Microscopy
Confluent cultures of C6 cells grown in 2-well culture
slides were incubated with bacteria at moi 10 for 30 or
TABLE 1. Adhesion and Invasion of E. coli Strains into C6
Glioma Cells
Mean CFU (TSD)/Well
E. coli
Strain
E44
E91
E105
E109§
E111
MG1655
Relevant
Characteristic(s)*
K1 RS218 spontaneous
mutant
E44 OmpAj
E91 OmpA+
E91 truncated OmpA+
E91 OmpAj
K12 laboratory strain
Adhesion
Invasion
(6.4 T 0.1) 107
(4.5 T 0.3) 104
(5.2 T 0.4) 106†
(6.7 T 0.3) 107
(6.7 T 0.2) 107
(5.4 T 0.5) 106†
(9.0 T 1.5) 106†
(6.2 T 0.4) 103‡
(4.6 T 0.3) 104
(4.6 T 0.1) 104
(6.1 T 0.2) 103†
(5.7 T 0.5) 103†
SD, standard deviation.
*Specific characteristics of these strains are described in Reference 17.
†p G 0.01 compared with E44 by Student t test.
‡p G 0.05 compared with E44 by Student t test.
§C-Terminal 53 amino acids of OmpA are deleted in this strain.
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
OmpA in E. coli Invasion Into Astrocytes
90 minutes at 37-C, washed 3 times with prewarmed
phosphate-buffered saline (PBS), and fixed with 2.5%
glutaraldehyde. For transmission electron microscopy, samples
were fixed with 2% osmium tetroxide for 2 hours, washed 3
times with PBS, dehydrated in a gradient series of alcohol
concentrations, and embedded in Epon 812 (Serva, Heidelberg,
Germany). Ultrathin sections were stained with uranyl acetate
and lead citrate and examined with an EM 906 transmission
electron microscope (Zeiss).
were purified with Ni2+-charged sepharose according to
the manufacturer’s instructions (Amersham). His-OmpA and
His-enolase were eluted by an elution buffer (20 mmol/L of
sodium phosphate, 500 mmol/L of NaCl, and 500 mmol/L of
imidazole). The samples were massively dialyzed against
the elution buffer without imidazole then a PBS (pH 7.4) to
reduce the salt concentrations.
Electrophoresis and Immunoblotting
Electrophoresis was performed on 10% sodium dodecyl
sulfateYpolyacrylamide gel electrophoresis (SDS-PAGE).
Reagents for SDS-PAGE were from Bio-Rad (Hercules, CA).
After electrophoresis, proteins on the gel were electrotransferred onto a polyvinylidene fluoride (PVDF) membrane
(Amersham Pharmacia Biotech, Buckinghamshire, UK). After
transfer, the PVDF membranes were blocked with blocking
solution containing 5% skim milk in TBST (10 mmol/L Tris,
pH7.5, 100 mmol/L NaCl and 0.1% Tween 20) for 1 hour at
room temperature. Antibodies were purchased from Santa
Cruz Biotechnology (Santa Cruz, CA) unless otherwise
indicated. The PVDF membrane was incubated with a solution
containing mouse anti-GFAP antibody (1:100 diluted), rabbit
antiYCOX-2 antibody (1:200), rabbit antiYNOS-2 antibody
(1:200), or goat anti-actin antibody (1:200) in the blocking
buffer for 2 hours. After washing, the PVDF membrane was
incubated with peroxidase-linked secondary antibody (1:2000)
for 1 hour. Finally, the PVDF membrane was developed using
a chemiluminescence kit (Amersham).
Immunofluorescence Staining and Confocal
Analysis
Cryosections of mouse brains or slide-cultured C6
cells were fixed with 4% paraformaldehyde. Samples were
blocked with 1% bovine serum albumin and then incubated
with a solution containing mouse anti-GFAP antibody (1:50),
rabbit antiYCOX-2 antibody (1:50), rabbit antiYNOS-2 antibody (1:50), rat antiYLy-6G antibody (1:100; eBioscience,
San Diego, CA), or goat antiYE. coli antibody (1:100;
Abcam, Cambridge, UK) for 1 hour. After washing, samples
were incubated with fluorescein isothiocyanate (FITC)Y
labeled secondary antibody (1:200) or rhodamine-labeled
secondary antibody (1:50) for 1 hour. After washing, propidium iodide or Hoechst stain was used to stain DNA.
Slides were mounted in 50% glycerol-PBS and then examined with the TCS SP5 Confocal Spectral Microscope
Imaging System (Leica).
Expression and Purification of His-OmpA and
His-Enolase Protein
The DNA fragment containing full-length ompA and
enolase gene was amplified with polymerase chain reaction
using restriction enzyme sites containing primers, digested
with SacI and XhoI, and ligated into pET-21 expression
vector (Novagen, Darmstadt, Germany). The resultant plasmid was transformed into E. coli BL-21 (DE3) strain. Protein
expression was induced by 0.5 mmol/L of isopropyl-A-Dthiogalactopyranoside. His-OmpA and His-enolase proteins
Ó 2009 American Association of Neuropathologists, Inc.
Animal Experiments
C57BL/6 mice 8 to 12 weeks old were randomly distributed into groups. To assess the survival of C57BL/6 mice
after intracerebral bacterial administration, the protocol
developed by Tsao et al (23) was followed. Each group
of 5 mice was anesthetized with pentobarbital sodium salt
(50 mg/kg) by intraperitoneal injection, and then each brain
was infected with bacteria 5 105 in 20 Kl of PBS, or 5 Kg
of lipopolysaccharide (LPS; from E. coli O26:B6, Sigma) in
20 Kl of PBS by intracerebral injection. Phosphate-buffered
saline (20 Kl) was used as a negative control. The mice were
monitored for survival every 12 hours for 8 days. To investigate the role of recombinant OmpA protein in the survival
of C57BL/6 mice after intracerebral E44 administration, each
brain was infected with E44 5 105 in 20 Kl of PBS by
intracerebral injection in the absence or presence of OmpA
(8 Kg or 20 Kg) or 20 Kg enolase. Phosphate-buffered saline
(30 Kl) was used as a negative control. Survival of C57BL/6
mice was observed up to 8 days postadministration. For immunofluorescence staining and Western blotting, groups
of 3 C57BL/6 mice were anesthetized and then infected
with bacteria by intracerebral injection. Mice were killed
and brains were removed at 24 hours postchallenge. For
immunofluorescence staining, brains were embedded in OCT
compound, and 6-Km cryosections were cut. For Western
blotting, brains were homogenized with protein extraction
buffer (50 mmol/L of Tris HCl, pH 8.0, 150 mmol/L of NaCl,
1% NP-40, and the protease inhibitor cocktails [Bio-Rad],
including 0.2 mmol/L of PMSF, 20 Kg/ml of aprotinin, and
20 Kg/ml of leupeptin) and harvested for whole cell lysate.
To detect remnant bacteria in the brain, groups of 3 C57BL/6
mice were infected with bacteria by intracerebral injection.
Mice were killed at various times postchallenge. The brains
were aseptically removed and homogenized with 3% gelatin
in PBS. The samples were serially diluted, and CFU was
determined on BAP.
Statistics
The results were expressed as the
viation of 3 independent experiments.
differences between treatment groups
Student t test. Results were considered
calculated p value was less than 0.05.
mean T standard deThe significance of
was determined by
significant when the
RESULTS
OmpA of E. coli K1 Contributes to Adhesion to
and Invasion of C6 Cells
To provide an in vitro system for E. coli invasion of
astrocytes, we chose rat glioma C6 cells (27, 28). Pilot
679
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
experiments were performed with E44 (OmpA+), E91
(OmpA-), and a laboratory nonpathogenic control strain,
MG1655, to determine the optimal conditions for adhesion
and invasion. Monolayer cultured C6 cells were incubated
with bacteria at moi of 0.1, 1, or 10 for 2 hours. The cellassociated E. coli increased in a dose-dependent manner
(Fig. 1A). The effect of incubation time was then monitored
at moi 10. The results showed that attachment of the E. coli
strains increased steadily up to 3 hours (Fig. 1B). Next, the
invasion abilities of E44, E91, and MG1655 were monitored
by incubating bacteria at moi of 0.1, 1, or 10 with C6 cells for
2 hours (Fig. 1C) or at moi 10 for up to 3 hours (Fig. 1D).
Both the adhesion and invasion abilities of E44 are much
greater than those of E91 and MG1655. Because all strains
displayed dose-dependent and time-dependent invasion, moi
10- and 2-hour incubation were used subsequently as the
FIGURE 1. Adhesion and invasion of C6 cells by E. coli strains. (A) For dose-dependent adhesion assays, confluent cultures of C6 cells
were infected with bacteria at the indicated moi for 2 hours. Recoverable CFUs were determined on sheep BAPs. (B) For time-course
adhesion assays, confluent cultures of C6 cells were infected with bacteria at the indicated moi 10 for the indicated time periods, and
recoverable CFU were determined. (C) For dose-dependent invasion assays, confluent cultures of C6 cells were infected with bacteria
at the indicated moi for 2 hours. Cells were incubated with medium containing 100 Kg/ml of gentamicin for 2 hours to kill
extracellular bacteria. Recoverable CFUs were then determined. (D) For time-course invasion assays, confluent C6 cell cultures were
infected with bacteria at moi 10 for the indicated time periods. Cells were then incubated with medium containing 100 Kg/ml of
gentamicin for 2 hours to kill the extracellular bacteria and recoverable CFU determined. (E) To detect the survival of internalized
bacteria, confluent C6 cell cultures were infected with bacteria at moi 10 for 2 hours. After washing, the culture was incubated with
medium containing 100 Kg/ml of gentamicin for 2 hours to kill the extracellular bacteria and further incubated with medium
containing 20 Kg/ml of gentamicin for the indicated periods. Recoverable CFUs were then determined. Data represent the average
of 3 independent experiments; error bars indicate standard deviations.
680
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
OmpA in E. coli Invasion Into Astrocytes
standard assay conditions for this study. We then determined
whether OmpA influences the intracellular survival of E. coli.
The results showed that internalized bacteria decreased dramatically at 24 hours postincubation and became undetectable
at 48 hours postinfection (Fig. 1E). At the same time, the
growth of C6 cells was not affected by the infection with E.
coli, as indicated by viable cell number counts performed daily
for up to 7 days postinfection (data not shown).
To determine whether OmpA contributes to E. coli
invasion into C6 cells, a panel of E. coli strains generated in
E44 was used for adhesion and invasion assays. Among
them, E91 is OmpA-deficient; E105 and E109 are E91
complemented by a plasmid that carries a full-length or
truncated ompA, respectively; E111 contains the vector
control (Table 1) (17). We observed that OmpA+ strains
FIGURE 3. Signaling pathways involved in the invasion of E44
into C6 cells. Confluent cultures of C6 cells were pretreated for
1 hour before E44 infection with inhibitors and then infected
with E44 (moi 10) for 2 hours and assayed for invasion. The
inhibitors cytochalasin D (A), genistein (B), or LY294002 (C)
were added at the indicated concentrations. Cytochalasin D
was dissolved in sterilized water; genistein and LY294002 were
dissolved in dimethyl sulfoxide. Results are shown as relative
invasiveness with CFUs of untreated controls as 100%. Data
represent the average of 3 independent experiments; error
bars indicate standard deviations. **, p G 0.001; j, untreated
control; 0, solvent control.
FIGURE 2. Invasion of E44 into C6 cells under TEM. Confluent
monolayers of C6 cells were infected with bacteria at moi 10
for 30 (A) or 90 minutes (B); ultrathin sections were then
prepared for TEM (magnification: 8,000). TEM, transmission
electron microscopy.
Ó 2009 American Association of Neuropathologists, Inc.
E44, E105, and E109 adhered to and invaded into C6 cells
in numbers approximately 10- to 15-fold higher than the
OmpA- strains, E91, and E111. E91 behaved similarly to
MG1655 in adhesion and invasion assays (Table 1). Because the growth rates of various E. coli strains alone in
culture medium were similar (data not shown), the data
suggest that depletion of OmpA diminishes the adhesion
681
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
and invasion abilities of E. coli, rather than affecting its
growth per se.
The ability of bacteria to invade into C6 cells was
examined further by electron microscopy at 30 and 90
minutes post-E44 infection (Fig. 2A, B). The internalized
bacteria were surrounded by a membrane-like structure
and located within vacuoles. Intracellular E91 showed a
similar pattern, but with far less frequency (data not shown).
Invasion of E. coli was associated with the formation
of lamellipodia of C6 cells (Fig. 2B), suggesting that
FIGURE 4. Outer membrane protein AYmediated E. coli invasion induces GFAP expression in C6 glioma cells. Confluent cultures of
C6 cells were infected with E coli (moi, 10) for 2 hours, followed by culture medium containing 100 Kg/ml of gentamicin for
24 hours. Lipopolysaccharide (1 Kg/ml) treatment for 24 hours served as a positive control. (A, B) Western analysis of whole cell
lysate to detect GFAP and actin. (C) Immunofluorescence staining after infection of C6 cells grown on 22-mm glass coverslips
using anti-GFAP antibody. DNA was stained with PI and examined by confocal microscopy. E44-infected and LPS-treated cells
show enhanced GFAP immunoreactivity. (D) C6 cells were pretreated with 0.3 Kg/ml of CD, 100 Kmol/L of LY294002, or 100
Kmol/L of genistein for 1 hour before infection with E44. After 2 hours, the cells were washed and incubated with culture medium
containing 100 Kg/ml of gentamicin for 24 hours. Whole-cell lysate was collected, and Western analysis was performed to detect
GFAP and actin. C6 cells treated with inhibitor alone serve as a negative control. CD, cytochalasin D; PI, propidium iodine.
682
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
OmpA in E. coli Invasion Into Astrocytes
macropinocytosis seems to be involved in the internalization
process.
Actin Rearrangement, Protein Tyrosine Kinase,
and Phosphoinositide-3-Kinase Activation Are
Required for OmpA-Mediated E. coli Invasion
We next explored the cellular signaling pathways
involved in the E44 invasion process. It has been reported
FIGURE 6. Survival of C57BL/6 mice after infection with E. coli
strains. Groups of 5 C57BL/6 mice were anesthetized and
challenged with 5 105 bacteria or 5 Kg LPS by intracerebral
injection. Phosphate-buffered saline was used as a negative
control. The mice were monitored for survival up to 8 days.
in the case of E. coli K1, Listeria monocytogenes, group B
Streptococcus, and Moraxella catarrhalis that actin cytoskeleton rearrangement is a prerequisite of bacterial invasion
(29Y33). To determine whether actin rearrangement is
involved in E44 invasion, C6 cells were pretreated with the
microfilament-depolymerizing agent cytochalasin D at concentrations of 0.01, 0.03, 0.1, or 0.3 Kg/ml for 1 hour before
infection. Data obtained from invasion assays revealed that
E44 invasion was blocked by cytochalasin D in a dosedependent manner, and greater than 95% inhibition was
observed at 0.1 Kg/ml (Fig. 3A), suggesting that actin polymerization is required for OmpA-mediated E. coli invasion into C6 cells. Furthermore, invasion of E. coli K1 into
BMECs depends on tyrosine phosphorylation of focal adhesion kinase (34). To determine whether protein tyrosine
kinase activation is required for E44 invasion, C6 cells were
pretreated with the general protein tyrosine kinase inhibitor
genistein at increasing concentrations for 1 hour before infection. The results indicated that genistein blocked E44
FIGURE 5. Infection of OmpA+ E. coli induces astrocyte
activation and the expression of COX-2 and NOS-2 in vivo.
Groups of 3 C57BL/6 mice each were anesthetized and
challenged with 5 105 bacteria or 5 Kg LPS by intracerebral
injection. Phosphate-buffered saline was used as a negative
control; the mice were killed at 24 hours postinfection. (A, B)
Confocal images of immunofluorescence staining using mouse
anti-GFAP, rabbit antiYCOX-2, or rabbit antiYNOS-2 antibody,
followed by FITC-conjugated anti-mouse IgG or rhodamineconjugated anti-rabbit IgG antibody. DNA was stained with
Hoechst. (C, D) Western analysis of whole cell lysates detects
COX-2 or NOS-2 and actin. (E) Confocal images of consecutive
brain cryosections of E44-infected mouse brains at 24 hours
postinfection stained with mouse anti-GFAP or goat antiYE. coli
antibody, followed by FITC-conjugated anti-mouse IgG antibody or rhodamine-conjugated anti-goat IgG antibody. DNA
was stained with Hoechst. IgG, immunoglobulin G.
Ó 2009 American Association of Neuropathologists, Inc.
683
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
FIGURE 8. His-OmpA protects C57BL/6 mice from intracerebral infection of E44. Groups of 5 C57BL/6 mice were infected
with 5 105 E44 bacteria by intracerebral injection in the
absence or presence (8 or 20 Kg) of His-OmpA or 20 Kg Hisenolase. Phosphate-buffered saline served as a negative
control. Mice were monitored for survival at 12-hour intervals
for 8 days.
FIGURE 7. Effects of His-OmpA on E. coli invasion and GFAP
expression in C6 glioma cells. (A) Purified recombinant OmpA
and enolase proteins were displayed by 10% SDS-PAGE and
stained with Coomassie blue. The expected molecular weight
is 38 kDa for His-OmpA and 50 kDa for His-enolase. (B)
Immunoblotting of recombinant OmpA protein. Chicken antiOmpA IgY (1:5000) and HRP-conjugated donkey anti-chicken
IgY antibody (1:10000) were used. Immunoreactive bands
were visualized with DAB. (C) Confluent C6 cell monolayers
were infected with E44 (moi, 10) in the absence or presence
(4 or 40 Kg/ml) of His-OmpA or 40 Kg/ml His-enolase for 2
hours. The cells were then incubated with medium containing
100 Kg/ml of gentamicin for 2 hours to kill the extracellular
bacteria. After washing, recoverable CFUs were determined.
Results are expressed as relative invasiveness with CFU of E44
administration as 100%. Data represent the average of 3
independent experiments; error bars indicate standard deviations. **, p G 0.001. Confluent C6 cell cultures were infected
with E44 (moi, 10) in the presence (4 or 40 Kg/ml) or absence
of His-OmpA for 2 hours. Western analysis of whole cell lysates
was performed to detect GFAP and actin. (D) Confluent
cultures of C6 cells were infected with E44 (moi, 10) in the
absence or presence (40 Kg/ml) of His-OmpA or 40 Kg/ml Hisenolase for 2 hours. After washing, GFAP expression in whole
cell lysates was detected by immunoblotting. >OmpA, antiOmpA antibody; DAB, diaminobenzidine; HRP, horseradish
peroxidase; IgY, immunoglobulin Y.
684
invasion in a dose-dependent manner, and greater than 90%
inhibition was observed at a concentration of 30 Kmol/L
(Fig. 3B); this suggests that protein tyrosine kinase activation is also required for OmpA-mediated E. coli invasion.
As shown in Figure 2A and B, bacteria seemed to be
internalized through macropinocytosis that comprises a
phosphoinositide-3 (PI 3)YkinaseYdependent contractile
mechanism (35). To investigate the involvement of PI
3Ykinase during E44 invasion, C6 cells were pretreated with
the PI 3Ykinase inhibitor LY294002 at 3, 10, 30, or 100
Kmol/L for 1 hour before infection. E44 invasion was
blocked by LY294002 in a dose-dependent manner; greater
than 90% inhibition was achieved at a concentration of
30 Kmol/L (Fig. 3C), indicating that PI 3Ykinase is also
involved in OmpA-mediated E. coli invasion. Taken together, these results demonstrate that rearrangement of the
actin cytoskeleton and activation of protein tyrosine kinase
and PI 3Ykinase are all needed for OmpA-mediated E. coli
invasion of C6 cells.
OmpA-Mediated E. coli Invasion Activates C6
Glioma Cells
To determine whether C6 cells are activated by E. coli
infection, immunoblotting was used to detect GFAP levels.
The infection of OmpA+ strains E44, E105, and E109 but not
OmpA- strains E91, E111, and MG1655 induced GFAP
expression (Fig. 4A and B). By confocal microscopy, infection
of E44, but not E91 or MG1655, induced cytoplasmic
accumulation of GFAP (Fig. 4C). These results suggest that
E. coliYinduced GFAP expression is dependent on OmpA.
Next, invasion inhibitors were tested for their effects on E.
coliYinduced GFAP accumulation. C6 cells were pretreated
with 0.3 Kg/ml of cytochalasin D, 100 Kmol/L of LY294002,
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
OmpA in E. coli Invasion Into Astrocytes
or 100 Kmol/L of genistein for 1 hour before infection with
E44 and compared with mock-infected cells. Glial fibrillary
acidic protein expression induced by E44 was blocked sig-
nificantly in the presence of cytochalasin D, LY294002, or
genistein (Fig. 4D); these results suggest that OmpA-mediated
invasion is necessary for activation of C6 cells.
FIGURE 9. His-OmpA inhibits the activation of astrocytes and the expression of COX-2 and NOS-2 in C57BL/6 mice brains after
E44 infection. Groups of 3 C57BL/6 mice were anesthetized and infected with 5 105 of E44 by intracerebral injection in the
presence (8 or 20 Kg) or absence of His-OmpA. Phosphate-buffered saline served as a negative control. Mice were killed, and
brains were removed at 24 hours postinfection. (A, B) Immunofluorescence staining with mouse anti-GFAP, rabbit antiYCOX-2,
or rabbit antiYNOS-2 antibody, followed by FITC-conjugated anti-mouse IgG antibody or rhodamine-conjugated anti-rabbit IgG
antibody. Hoechst was used for DNA staining. (C, D) Brain whole cell lysates were analyzed for COX-2 or NOS-2 and actin by
immunoblotting.
Ó 2009 American Association of Neuropathologists, Inc.
685
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
Intracerebral Injection of OmpA+ E. coli Induces
the Activation of Astrocytes and the Expression
of COX-2 and NOS-2
periments, groups of mice were infected by intracerebral
injection with 5 103, 5 105, or 5 107 bacteria. The
100% lethal dose of E44 was 5 105, and the 100% lethal
dose of E91 and MG1655 was 5 107 (data not shown). The
infectious inoculums used were 5 105 in this study. Groups
of 3 C57BL/6 mice were anesthetized and then infected with
of E44, E91, MG1655, or 5 Kg of LPS by intracerebral
injection. At 24 hours postchallenge, the mice were killed,
and brains were removed and embedded in OCT for
cryosectioning. Because inflammatory molecules COX-2
We then determined whether OmpA is required for
activation of astrocytes by E. coli in vivo. Because the
invasive capability of E44 was 25- to 50-fold greater than
that of E91 in BMECs (17), we challenged mice by
intracerebral injection to avoid interference of the BBB
according to the protocol of Tsao et al (23). In pilot ex-
FIGURE 10. His-OmpA reduces neutrophil infiltration in C57BL/6 mouse brain after E44 infection. Consecutive brain cryosections
used in Figures 5 and 9 from mice that had been challenged with PBS (control), E44 (5 105), E44 and 20 Kg His-OmpA, E91,
MG1655, or 5 Kg LPS were stained with rat antiYLy-6G antibody and FITC-conjugated secondary antibody to demonstrate
neutrophil infiltration. Hoechst was used for DNA staining; the slides were observed with confocal microscopy. Bar = 25 Km.
686
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
and NOS-2 mediate a large amount of the neurologic damage
in CNS infections (23, 26), immunofluorescence staining
was performed and observed by confocal microscopy to
detect GFAP, COX-2, and NOS-2 expression in the brains
after infection. As shown in Figure 5A, E44 infection
induced accumulation of GFAP and expression of COX-2.
The overlay of both signals indicated that COX-2 expression
was colocalized with GFAP in activated astrocytes. In
parallel, infection of E44 also induced NOS-2 expression in
astrocytes expressing high levels of GFAP (Fig. 5B). Neither
the activation of astrocytes nor the expression of COX-2 or
NOS-2 was observed in E91- or MG1655-treated animals
(Fig. 5A, B). Furthermore, groups of 3 mice each were
infected with bacteria or 5 Kg of LPS by intracerebral
injection and analyzed for COX-2 and NOS-2 expression by
immunoblotting. Cyclooxygenase 2 and NOS-2 expression
were induced after infection with E44, but not E91 or
MG1655 (Fig. 5C, D). Furthermore, the consecutive brain
cryosections of E44-infected mouse brains at 24 hours
postinfection were costained with mouse anti-GFAP antibody
and goat antiYE. coli antibody. Colocalization of bacteria and
GFAP was observed in some areas with residual bacteria
(Fig. 5E). Taken together, these results suggest that OmpAmediated E. coli infection activates astrocytes and induces the
expression of COX-2 and NOS-2 in the mouse brain.
Intracerebral Injection of OmpA+ E. coli Strain
Causes the Death of C57BL/6 Mice
To identify the role of OmpA in the development of
E. coliYinduced CNS infection in vivo, groups of 5 mice
were infected with E. coli K1 strains (E44, E91, E105, E109,
and E111), MG1655, or 5 Kg of LPS by intracerebral injection and assessed for 8 days. Mice died within 48 hours of
intracerebral injection of the OmpA+ strains E44, E105, or
E109, whereas mice challenged with the OmpA- strains E91
or E111, MG1655, or with LPS lived for 8 days before they
were killed (Fig. 6). This suggests that OmpA-mediated
E. coli infection subsequently leads to the death of the mice.
Recombinant OmpA Protein Inhibits E. coli
Invasion and Activation of C6 Cells
Because purified OmpA protein inhibited the invasion
of BMECs by OmpA+ E. coli (8, 17), bacterially expressed
recombinant His-OmpA and the negative control His-enolase
were purified. Escherichia coli enolase is a glycolytic
enzyme that catalyzes 2-phosphoglycerate into phosphoenolpyruvate (36). The homogeneities of the purified His-OmpA
and His-enolase were revealed by Coomassie blue staining
and confirmed by immunoblotting (Fig. 7A, B). The purified
His-OmpA and His-enolase displayed molecular weights of
38 and 50 kDa on 10% SDS-PAGE, as predicted. To
examine the effect of His-OmpA on E44 invasion, C6 cell
monolayers were incubated with E44 in the absence or
presence of His-OmpA. There was approximately 50%
inhibition of invasion with 4 Kg/ml of His-OmpA and
approximately 83% inhibition with 40 Kg/ml of His-OmpA;
no inhibitory effect was observed in the presence of Hisenolase (Fig. 7C). By immunoblotting, the E44-induced
GFAP expression was suppressed significantly by His-OmpA
Ó 2009 American Association of Neuropathologists, Inc.
OmpA in E. coli Invasion Into Astrocytes
TABLE 2. Viable Bacteria in Brain Tissue After Intracerebral
Injection of E. coli
Mean CFU (TSD)/Brain in E. coli Strains Tested
Time After
Infection, hours
12
24
48
E44 + Recombinant
OmpA
E44
6
(2.3 T 0.4) 10
(2.4 T 0.1) 107
(6.0 T 0.3) 107
5
(2.0 T 0.3) 10
ND
ND
E91
(1.8 T 0.3) 105
ND
ND
Groups of mice (n = 3 per group) were given intracerebral injections of 5 105 E.
coli. The brains were collected at the indicated time points after injection, and the
bacteria remaining in the brains were quantified.
ND, not detectable.
in a dose-dependent manner (Fig. 7C). In contrast, Hisenolase did not inhibit E44-induced GFAP expression
(Fig. 7D). These data indicate that recombinant OmpA can
specifically prevent bacterial invasion and the subsequent
accumulation of GFAP in C6 cells.
Recombinant His-OmpA Protects C57BL/6 Mice
from E44-Induced Death
We next determined whether the presence of HisOmpA also protects mice from the death caused by intracerebral injection of E44. Groups of 5 C57BL/6 mice were
challenged by intracerebral injection with E44 alone or a
mixture of E44 and 8 or 20 Kg of His-OmpA or 20 Kg of
His-enolase, and survival of the mice was assessed for 8
days. The mice challenged with E44 alone or E44 + 20 Kg of
His-enolase died within 36 hours, whereas 40% of mice
survived after the injection of E44 + 8 Kg of His-OmpA, and
80% of mice with E44 + 20 Kg of His-OmpA to 8 days
postinfection (Fig. 8). Thus, recombinant His-OmpA can
protect C57BL/6 mice from intracerebral infection by E44.
Recombinant His-OmpA Inhibits the InfectionInduced Activation of Astrocytes and the
Expression of COX-2 and NOS-2 In Vivo
To determine whether recombinant His-OmpA affects
E44-induced activation of astrocytes and the expression of
COX-2 and NOS-2 in vivo, groups of 3 C57BL/6 mice were
infected with E44 alone or a mixture of E44 and 20 Kg of
His-OmpA. Mice were killed at 24 hours postchallenge, and
the brains were removed, embedded in OCT for cryosectioning, and the sections were examined by confocal microscopy.
The presence of His-OmpA reduced the E44-induced
expression of GFAP, COX-2, and NOS-2 in the mouse
brains (Fig. 9A, B). Further groups of mice were infected
with E44 alone or a mixture of E44 and 8 or 20 Kg of HisOmpA. At 24 hours postchallenge, the brains were removed,
homogenized with protein extraction buffer, and analyzed in
immunoblotting. The results confirmed the expression of
COX-2 and NOS-2 was inhibited specifically by His-OmpA
in a dose-dependent manner (Fig. 9C, D). Consecutive brain
sections were stained with antibody Ly-6G to observe
neutrophil infiltrates in the forebrain region. Consistent with
the survival experiment in Figure 6, intracerebral injection of
E44, but not E91, MG1655, or LPS, induced neutrophil
infiltration near the major blood vessels. In contrast, the
687
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
mixture of 20 Kg of His-OmpA with E44 reduced the
numbers of neutrophils in the forebrain region (Fig. 10).
Simultaneously, the remnant bacteria in brain infected with
E44 increased up to (2.4 T 0.1) 107 at 24 hours
postinfection and (6.0 T 0.3) 107 at 48 hours postinfection;
bacteria in brains infected with E91 or E44 with recombinant
OmpA were undetectable at 24 hours postinfection (Table 2).
These results demonstrate that mouse survival correlates with
bacterial clearance. Taken together, these data indicate that
His-OmpA can protect astrocytes from E44-induced activation, the expression of COX-2 and NOS-2, and neutrophil
infiltration in mouse brains, effects that may protect the mice
from severe neuronal damage and death.
OmpA experiments also demonstrate that OmpA protein, but
not the amount of Fim protein, contributes to the invasion of
E. coli into astrocytes.
Several host cell pathways are involved in the invasion of
E. coli K1 into BMECs, although they are not fully elucidated.
For example, rearrangement of the actin cytoskeleton was
shown to be a prerequisite for BMEC invasion (30), and
tyrosine phosphorylation of focal adhesion kinase and PI
3Ykinase activation are also required for E. coli invasion into
BMECs (34, 44). The intracellular signaling pathways
required for other microorganisms to invade into host cells
are diverse. For example, rearrangement of the actin cytoskeleton was also shown to be a prerequisite for BMEC invasion
by L. monocytogenes and group B Streptococcus (31, 32) and
for invasion of pulmonary epithelial cells by M. catarrhalis
(33). Invasion of BMEC by L. monocytogenes and of
epithelial cells by M. catarrhalis, however, does not require
phosphorylation of focal adhesion kinase (33, 34). Activation
of PI 3Ykinase also is involved in L. monocytogenes invasion
of BMEC and epithelial cells and M. catarrhalis invasion into
pulmonary epithelial cells (33, 34, 45), in contrast, invasion of
BMEC by group B Streptococcus was independent of PI
3Ykinase activation (3). Our present data indicate that
rearrangement of the actin cytoskeleton, protein tyrosine
kinase, and PI 3Ykinase activation are all required for E44
invasion of C6 cells. Thus, the invasion of bacteria into C6
cells may reflect the outcome of macropinocytosis, which is
usually induced by the interaction of microbial factors with
nonphagocytic cells, and followed by the activation of
tyrosine kinases and actin-mediated ruffling. This idea is
supported further by our electron micrographs showing that
adherence of E. coli to C6 cell was accompanied by the
formation of membrane ruffles and lamellipodia (Fig. 2B).
Escherichia coli was then internalized and located within
intracellular vacuoles, suggesting that a trigger-like uptake
mechanism is involved in this process. Nevertheless, some of
the macropinosome markers such as early endosomal autoantigen 1 or transferrin receptor for early endosome and Rab7
or Lamp 1 for late endosome and late endosome/lysosome
(46) should be investigated to clarify the mechanisms
involved in the E44 invasion of C6 cells in the future.
We observed that internalized bacteria decreased
dramatically at 24 hours postincubation in gentamicincontaining medium and became undetectable at 48 hours
postinfection (Fig. 1E). To examine whether exocytosis or
intracellular killing is mediated for bacteria clearance, we
performed an experiment that C6 cells were incubated in
antibiotic-free medium after E44 infection. At 24 hours
postincubation, (6.5 T 2.1) 105 CFU/ml were recovered
from the medium, suggesting that at least a portion of bacteria
may be exocytosed after invasion (data not shown); it was not
clear, however, whether intracellular killing occurred during
this process. In the studies of BMECs, vacuoles containing E.
coli K1 acquired markers for early endosome, late endosome,
and late endosome/lysosome, but not the lysomal enzyme
cathepsin D, suggesting E. coli K1 is able to direct the E. coli
containing vacuoles to escape from fusion with lysosome and
avoid the killing by lysosomal enzymes (7). Further study is
required to understand the fate of internalized E44 in C6 cells.
DISCUSSION
We identified OmpA as the primary bacterial component involved in adhesion, invasion, and activation of
astrocytes by E. coli. To our knowledge, this is the first report demonstrating that recombinant OmpA protein can
protect astrocytes from activation induced by invasion of E.
coli and, most importantly, protects the mice from death
caused by intracerebral infection of OmpA+ E. coli.
Outer membrane protein A was found to be the
major component of E. coli that mediates host-cell
interactions. In the noninvasive strain enterohemorrhagic
E. coli, OmpA functions as an adhesion molecule that binds
to HeLa, colonic epithelial and dendritic cells (37, 38). In
invasive strains, OmpA is important for the invasion of E.
coli K1 into macrophages and monocytes (39) and for
binding and invasion into BMECs (7, 17, 40). We found
that both the adhesion and invasion capacities of OmpA+ E.
coli K1 strains (E44, E105, E109) were approximately 10to 15-fold greater than those of OmpAj E. coli K1 strains
(E91, E111), that OmpA is the major determinant enabling
E. coli to adhere to and invade into astrocytes, and that its
binding is the most critical step for E. coli K1 invasion of
C6 cells.
Outer membrane protein A protein consists of 325
amino acids, and its N-terminal A-barrel domain is critical for
correct folding and function. Outer membrane protein A
proteins lacking the C-terminal residues 228 to 325 or
residues 194 to 325 are correctly incorporated into the outer
membrane and confer all known OmpA phenotypes (41, 42).
We also observed that E109, which expresses the truncated
OmpA lacking amino acids 301 to 325, has a similar ability
to invade into C6 cells to wild-type E44 and E105.
In addition to OmpA, type 1 fimbriae and other
bacterial components have been implicated in the adhesion
and invasion of E. coli to BMECs (3, 7). Expression of type 1
fimbriae is controlled by the inversion of a promotercontaining element upstream of the fim operon (43). In
OmpA-deleted mutants, this invertible element is predominantly in the OFF orientation, leading to decreased fimbrial
expression. It has been shown that the ompA deletion mutant
derived from the fim locked-ON background still exhibited
significantly lower adhesion and invasion abilities than the
fim locked-ON mutant, suggesting that OmpA is the primary
determinant for binding to BMECs (40). Our complementary
688
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
OmpA in E. coli Invasion Into Astrocytes
On the other hand, after bacterial infection, C6 cells were still
alive and multiplied as efficiently as uninfected control cells
up to 7 days in our study (data not shown). The anti-apoptotic
factor BclXL is induced after the invasion of E. coli K1 to
prevent apoptosis of macrophages (47).
In our pilot experiments, the OmpA- strain E91 and
MG1655 (OmpA+ , nonpathogenic and noninvasive strain)
had approximately 100-fold higher 100% lethal dose compared with the OmpA+ strain E44 (data not shown). Mice
died within 48 hours of intracerebral injection of the OmpA+
strains E44, whereas mice challenged with the OmpA- strains
E91 or MG1655 survived for 8 days (Fig. 6). The remaining
bacteria in the brains increased at 24 and 48 hours postinfection, whereas the bacteria in brains infected with E91
were undetectable at 24 hours postinfection (Table 2). These
results indicate that mice survival correlates with bacterial
clearance. We suggest intracerebral microglia cells are
responsible for the primary clearance of all the bacteria
challenged (48). Multiple factors may contribute additively to
the difference of clearance of OmpA+ or OmpA- bacteria. For
example, E44 was demonstrated to survive in both murine
and human macrophage cell lines and in monocytes and
macrophages of newborn rats (39). Therefore, E44 might
have a better chance of being sustained in microglia. Outer
membrane protein AYnegative, but not OmpAYposiive , E.
coli might simultaneously activate human neutrophils to
produce oxygen radicals (49). Additionally, the OmpA- strain
is more susceptible to membrane-acting bactericidal peptides
than the wild-type strain (49). Overall, OmpA+ bacteria have
all the advantages for intracerebral survival, which may lead
to the lethal effects on mouse.
Increased GFAP expression is linked to astrogliosis in
bacterial meningitis (50). Glial fibrillary acidic protein is
important for modulating astrocyte motility and shape by
providing structural stability to the astrocyte. We observed
that infection of OmpA+ E. coli K1 strains, but not OmpA- E.
coli K1 strains or MG1655, induced the expression of GFAP
by C6 cells, and that pretreatment with cytochalasin D,
LY294002, or genistein inhibited the invasion-induced
accumulation of GFAP, suggesting that the bacterial invasive
process is a prerequisite for GFAP expression. Thus, we
conclude that OmpA is a major determinant of the adherence,
invasion, and activation of astrocytes after E. coli infection.
Once bacteria infect the CNS, inflammatory molecules
such as COX-2, NOS-2, IL-8, and intracellular adhesion
molecule 1 are induced, leading to an increase in the
permeability of the BBB (23, 51Y53). The inflammatory effects then trigger transendothelial migration of neutrophils
that may result in further brain damage. Indeed, we demonstrate that intracerebral E44 infection induced COX-2 and
NOS-2 expression, and that COX-2 and NOS-2 colocalized
with GFAP in activated astrocytes. Simultaneously, intracerebral injection of E44, but not OmpA- bacteria, also
increased neutrophil infiltration in mouse forebrain (Fig. 10).
The inflammatory processes progress to include the destruction of the BBB, neutrophil infiltration, neuron damage, and
finally, the death of C57BL/6 mice within 48 hours.
We also demonstrated that OmpA-mediated binding
plays an important role in the early stage of E. coli K1 CNS
infection. Preincubation of BMEC with OmpA protein,
solubilized from the outer membrane of E44, inhibited the
invasion of OmpA+ E. coli (17). Similarly, the purified Nterminal amino acids 1 to 171 of OmpA also decreased the
association of E. coli K1 with BMEC in a dose-dependent
manner (8). Here, we used the purified recombinant fulllength OmpA protein to protect C6 cells from infection by
E44. Furthermore, recombinant OmpA protein protected
increased the survival of infected mice up to 80%. Simultaneously, the neutrophil infiltration in brain sections was
reduced in the presence of His-OmpA, suggesting that HisOmpA inhibited E44 adhesion to astrocytes, thus preventing
their activation and subsequent inflammatory processes.
Purified amino-terminal OmpA binds directly to
BMECs, whereas a derivative lacking all 4 extracellular
loops could not (8). In addition, short peptides corresponding
to loops 1 and 2 blocked OmpA+ E. coli invasion of BMECs
(17). It remains to be determined which parts of the external
loops are required for OmpA-mediated adhesion and invasion
of astrocytes. In bacterial meningitis, antibiotic therapy is the
first choice for management, but neurologic complications
often cannot be averted. Our results that recombinant OmpA
protein can prevent the invasion of astrocytes by E. coli and
the subsequent inflammation events suggest a new approach
for preventing or reducing the neurologic complications that
follow astrocyte activation during bacterial meningitis. The
possible protective effects and routes of administration of
OmpA protein for patients with bacteremia need to be
investigated further.
Ó 2009 American Association of Neuropathologists, Inc.
ACKNOWLEDGMENTS
The authors thank K.S. Kim, MD, Division of Pediatric
Infectious Diseases, School of Medicine, Johns Hopkins
University (Baltimore, MD), for the E. coli strains (E44,
E91, E105, E109, E111, and MG1655); Jean-San Chia, MD,
Graduate Institute and Department of Microbiology, College
of Medicine, National Taiwan University (Taipei, Taiwan),
for helpful discussion; and Tim J. Harrison of the Royal Free
and University College Medical School (University College
London) for critical reading and editing of the manuscript.
REFERENCES
1. Shah DK, Daley AJ, Hunt RW, et al. Cerebral white matter injury in the
newborn following Escherichia coli meningitis. Eur J Paediatr Neurol
2005;9:13Y17
2. Leib SL, Tauber MG. Pathogenesis of bacterial meningitis. Infect Dis
Clin North Am 1999;13:527Y48, vYvi. Review
3. Kim KS. Escherichia coli translocation at the blood-brain barrier. Infect
Immun 2001;69:5217Y22
4. Teng CH, Cai M, Shin S, et al. Escherichia coli K1 RS218 interacts with
human brain microvascular endothelial cells via type 1 fimbria bacteria
in the fimbriated state. Infect Immun 2005;73:2923Y31
5. Hoffman JA, Wass C, Stins MF, et al. The capsule supports survival but
not traversal of Escherichia coli K1 across the blood-brain barrier. Infect
Immun 1999;67:3566Y70
6. Khan NA, Shin S, Chung JW, et al. Outer membrane protein A and
cytotoxic necrotizing factorY1 use diverse signaling mechanisms for
Escherichia coli K1 invasion of human brain microvascular endothelial
cells. Microb Pathog 2003;35:35Y42
7. Kim KS. Pathogenesis of bacterial meningitis: From bacteraemia to
neuronal injury. Nat Rev Neurosci 2003;4:376Y85
689
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.
Wu et al
8. Shin S, Lu G, Cai M, et al. Escherichia coli outer membrane protein A
adheres to human brain microvascular endothelial cells. Biochem
Biophys Res Commun 2005;330:1199Y204
9. Koebnik R. Structural and functional roles of the surface-exposed loops
of the beta-barrel membrane protein OmpA from Escherichia coli. J
Bacteriol 1999;181:3688Y94
10. Klimke WA, Frost LS. Genetic analysis of the role of the transfer gene,
traN, of the F and R100-1 plasmids in mating pair stabilization during
conjugation. J Bacteriol 1998;180:4036Y43
11. Klimke WA, Rypien CD, Klinger B, et al. The mating pair stabilization
protein, TraN, of the F plasmid is an outer-membrane protein with two
regions that are important for its function in conjugation. Microbiology
2005;151:3527Y40
12. Schwarz H, Riede I, Sonntag I, et al. Degrees of relatedness of T-even
type E. coli phages using different or the same receptors and topology of
serologically cross-reacting sites. EMBO J 1983;2:375Y80
13. Morona R, Klose M, Henning U. Escherichia coli K-12 outer membrane
protein (OmpA) as a bacteriophage receptor: Analysis of mutant genes
expressing altered proteins. J Bacteriol 1984;159:570Y78
14. Riede I, Eschbach ML, Henning U. DNA sequence heterogeneity in the
genes of T-even type Escherichia coli phages encoding the receptor
recognizing protein of the long tail fibers. Mol Gen Genet 1984;195:
144Y52
15. Montag D, Riede I, Eschbach ML, et al. Receptor-recognizing proteins
of T-even type bacteriophages. Constant and hypervariable regions and
an unusual case of evolution. J Mol Biol 1987;196:165Y74
16. Pautsch A, Schulz GE. Structure of the outer membrane protein A
transmembrane domain. Nat Struct Biol 1998;5:1013Y17
17. Prasadarao NV, Wass CA, Weiser JN, et al. Outer membrane protein A
of Escherichia coli contributes to invasion of brain microvascular
endothelial cells. Infect Immun 1996;64:146Y53
18. Condorelli DF, Dell’Albani P, Kaczmarek L, et al. Glial fibrillary acidic
protein messenger RNA and glutamine synthetase activity after nervous
system injury. J Neurosci Res 1990;26:251Y57
19. Chiu FC, Norton WT, Fields KL. The cytoskeleton of primary astrocytes
in culture contains actin, glial fibrillary acidic protein, and the
fibroblast-type filament protein, vimentin. J Neurochem 1981;37:147Y55
20. Farina C, Aloisi F, Meinl E. Astrocytes are active players in cerebral
innate immunity. Trends Immunol 2007;28:138Y45
21. Lieberman AP, Pitha PM, Shin HS, et al. Production of tumor necrosis
factor and other cytokines by astrocytes stimulated with lipopolysaccharide or a neurotropic virus. Proc Natl Acad Sci U S A 1989;86:6348Y52
22. Murphy S, Simmons ML, Agullo L, et al. Synthesis of nitric oxide in
CNS glial cells. Trends Neurosci 1993;16:323Y28
23. Tsao N, Hsu HP, Lei HY. TNFalpha-induced cyclooxygenase 2 not only
increases the vasopermeability of blood-brain barrier but also enhances
the neutrophil survival in Escherichia coliYinduced brain inflammation.
Prostaglandins Other Lipid Mediat 1999;57:371Y82
24. Azuma I. Review: Inducer of cytokines in vivo: Overview of field and
romurtide experience. Int J Immunopharmacol 1992;14:487Y96
25. Dong Y, Benveniste EN. Immune function of astrocytes. Glia 2001;36:
180Y90
26. Salazar-Mather TP, Hokeness KL. Calling in the troops: Regulation of
inflammatory cell trafficking through innate cytokine/chemokine networks. Viral Immunol 2003;16:291Y306
27. Kronfeld I, Zsukerman A, Kazimirsky G, et al. Staurosporine induces
astrocytic phenotypes and differential expression of specific PKC
isoforms in C6 glial cells. J Neurochem 1995;65:1505Y14
28. Mangoura D, Sakellaridis N, Jones J, et al. Early and late passage C-6 glial
cell growth: Similarities with primary glial cells in culture. Neurochem
Res 1989;14:941Y47
29. Meier C, Oelschlaeger TA, Merkert H, et al. Ability of Escherichia coli
isolates that cause meningitis in newborns to invade epithelial and
endothelial cells. Infect Immun 1996;64:2391Y99
30. Prasadarao NV, Wass CA, Stins MF, et al. Outer membrane protein
AYpromoted actin condensation of brain microvascular endothelial cells is
required for Escherichia coli invasion. Infect Immun 1999;67:5775Y83
31. Greiffenberg L, Goebel W, Kim KS, et al. Interaction of Listeria
monocytogenes with human brain microvascular endothelial cells:
690
J Neuropathol Exp Neurol Volume 68, Number 6, June 2009
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
InlB-dependent invasion, long-term intracellular growth, and spread
from macrophages to endothelial cells. Infect Immun 1998;66:5260Y67
Nizet V, Kim KS, Stins M, et al. Invasion of brain microvascular endothelial cells by group B streptococci. Infect Immun 1997;65:5074Y81
Slevogt H, Seybold J, Tiwari KN, et al. Moraxella catarrhalis is
internalized in respiratory epithelial cells by a trigger-like mechanism
and initiates a TLR2- and partly NOD1-dependent inflammatory
immune response. Cell Microbiol 2007;9:694Y707
Reddy MA, Wass CA, Kim KS, et al. Involvement of focal adhesion
kinase in Escherichia coli invasion of human brain microvascular
endothelial cells. Infect Immun 2000;68:6423Y30
Lindmo K, Stenmark H. Regulation of membrane traffic by phosphoinositide 3Ykinases. J Cell Sci 2006;119:605Y14
Spring TG, Wold F. The purification and characterization of Escherichia
coli enolase. J Biol Chem 1971;246:6797Y802
Torres AG, Kaper JB. Multiple elements controlling adherence of enterohemorrhagic Escherichia coli O157:H7 to HeLa cells. Infect Immun
2003;71:4985Y95
Torres AG, Li Y, Tutt CB, et al. Outer membrane protein A of
Escherichia coli O157:H7 stimulates dendritic cell activation. Infect
Immun 2006;74:2676Y85
Sukumaran SK, Shimada H, Prasadarao NV. Entry and intracellular
replication of Escherichia coli K1 in macrophages require expression of
outer membrane protein A. Infect Immun 2003;71:5951Y61
Teng CH, Xie Y, Shin S, et al. Effects of ompA deletion on expression
of type 1 fimbriae in Escherichia coli K1 strain RS218 and on the
association of E. coli with human brain microvascular endothelial cells.
Infect Immun 2006;74:5609Y16
Henning U, Cole ST, Bremer E, et al. Gene fusions using the ompA
gene coding for a major outer-membrane protein of Escherichia coli
K12. Eur J Biochem 1983;136:233Y40
Bremer E, Cole ST, Hindennach I, et al. Export of a protein into the
outer membrane of Escherichia coli K12. Stable incorporation of the
OmpA protein requires less than 193 amino-terminal amino-acid
residues. Eur J Biochem 1982;122:223Y31
Smith SG, Dorman CJ. Functional analysis of the FimE integrase of
Escherichia coli K-12: Isolation of mutant derivatives with altered DNA
inversion preferences. Mol Microbiol 1999;34:965Y79
Reddy MA, Prasadarao NV, Wass CA, et al. Phosphatidylinositol 3Ykinase
activation and interaction with focal adhesion kinase in Escherichia coli
K1 invasion of human brain microvascular endothelial cells. J Biol Chem
2000;275:36769Y74
Ireton K, Payrastre B, Cossart P. The Listeria monocytogenes protein
InlB is an agonist of mammalian phosphoinositide 3Ykinase. J Biol
Chem 1999;274:17025Y32
Kerr MC, Teasdale RD. Defining macropinocytosis. Traffic 2009;10:
364Y371
Sukumaran SK, Selvaraj SK, Prasadarao NV. Inhibition of apoptosis by
Escherichia coli K1 is accompanied by increased expression of BclXL
and blockade of mitochondrial cytochrome c release in macrophages.
Infect Immun 2004;72:6012Y22
Kreutzberg GW. Microglia: A sensor for pathological events in the CNS.
Trends Neurosci 1996;19:312Y18
Fu H, Belaaouaj AA, Dahlgren C, et al. Outer membrane protein A
deficient Escherichia coli activates neutrophils to produce superoxide
and shows increased susceptibility to antibacterial peptides. Microbes
Infect 2003;5:781Y88
Eng LF, Ghirnikar RS, Lee YL. Glial fibrillary acidic protein:
GFAP-thirty-one years (1969Y2000). Neurochem Res 2000;25:1439Y51
Kim JM, Oh YK, Lee JH, et al. Induction of proinflammatory mediators
requires activation of the TRAF, NIK, IKK and NF-kappaB signal
transduction pathway in astrocytes infected with Escherichia coli. Clin
Exp Immunol 2005;140:450Y60
Galanakis E, Di Cello F, Paul-Satyaseela M, et al. Escherichia coli K1
induces IL-8 expression in human brain microvascular endothelial cells.
Eur Cytokine Netw 2006;17:260Y65
Selvaraj SK, Periandythevar P, Prasadarao NV. Outer membrane protein
A of Escherichia coli K1 selectively enhances the expression of
intercellular adhesion moleculeY1 in brain microvascular endothelial
cells. Microbes Infect 2007;9:547Y57
Ó 2009 American Association of Neuropathologists, Inc.
Copyright @ 2009 by the American Association of Neuropathologists, Inc. Unauthorized reproduction of this article is prohibited.