| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Departments of
*
Medicine and
Pediatrics and Molecular Microbiology, Washington University School of Medicine, St. Louis, MO 63110
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
,
and MHC class I were not significantly affected, suggesting
preferential activation of a specific subset of epithelial genes
directed toward defense against bacteria. Induction of ICAM-1 required
direct bacterial interaction with the epithelial cell surface and was
not reproduced by purified H. influenzae
lipooligosaccharide. Consistent with a functional role for this
response, induction of ICAM-1 by H. influenzae mediated
increased neutrophil adherence to the epithelial cell surface.
Furthermore, in an in vivo murine model of airway infection with
H. influenzae, increased epithelial cell ICAM-1
expression coincided with increased chemokine levels and neutrophil
recruitment in the airway. These results indicate that ICAM-1
expression on human respiratory epithelial cells is induced by
epithelial cell interaction with H. influenzae and
suggest that an ICAM-1-dependent mechanism can mediate neutrophil
adherence to these cells independent of inflammatory mediator release
by other cell types. Direct induction of specific epithelial cell genes
(such as ICAM-1 and IL-8) by bacterial infection may allow for rapid
and efficient innate defense in the airway. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
One mechanism for epithelial cells to participate in airway defense is through coordination of leukocyte influx and activation by expression of adhesive surface proteins and secretion of chemotactic molecules. Increased cell-surface expression of the adhesive glycoprotein ICAM-1 in response to inflammatory stimuli is one important example of epithelial cell contribution to leukocyte recruitment (4, 5). The interaction of ICAM-1 with CD18/ß2 integrin-containing counterreceptors on leukocytes is a crucial mechanism for leukocyte adhesion and activation by airway epithelial cells (4, 6, 7). A second example of epithelial cell capacity to coordinate leukocyte influx in the airway is by secretion of the CXC chemokine IL-8, which targets neutrophils to sites of attack through its chemoattractant and activating properties (8). Although pathways for airway epithelial cell expression of cell adhesion molecules and secretion of mediator molecules in response to inflammatory cytokines and viruses have been described (4, 9, 10, 11), the role of epithelial cells in the airway response to bacterial infection is less well developed. However, pilin-mediated adherence of Pseudomonas aeruginosa has been shown to stimulate airway epithelial cell production of IL-8 (12, 13), providing precedent for the possibility that epithelial cells in the respiratory system initiate leukocyte recruitment following direct detection of bacterial stimuli. Based on the fact that the human lung is exposed to many different bacterial species, it seems logical that multiple detection mechanisms may have evolved to allow epithelial cells to respond effectively to a wide variety of potentially pathogenic microbes.
Haemophilus influenzae is a pleomorphic Gram-negative bacillus that frequently colonizes human respiratory mucosa and often produces localized respiratory tract disease (14, 15). Encapsulated isolates of H. influenzae are designated serotypes a–f on the basis of expression of one of six antigenically distinct polysaccharide capsules (3). Isolates of H. influenzae that fail to agglutinate with typing antisera against known capsular structures are designated nontypable. Although serotype b strains account for most cases of bacteremic and systemic illness due to H. influenzae, infection with this serotype is becoming less common due to recent widespread use of vaccines directed against the serotype b capsule (15). We have focused on nontypable strains because these organisms are responsible for the majority of disease caused by H. influenzae that is localized to the respiratory tract, including otitis media, sinusitis, bronchitis, and pneumonia (15, 16). Patients with chronic bronchitis, bronchiectasis, and cystic fibrosis are particularly predisposed to infection with nontypable H. influenzae (14, 17). Direct contact between H. influenzae and airway epithelial cells has been demonstrated during airway infection, and attachment to epithelial cells is mediated by specific bacterial surface molecules (3, 18). Although epithelial cell receptors for these bacterial adhesive proteins have been more difficult to identify, the use of surface molecules on epithelial cells to detect and respond to infection by H. influenzae seems likely. Accordingly, we hypothesized that epithelial cells in the lung might have the capacity to detect infection with H. influenzae directly and then respond through activation of specific airway defense genes such as ICAM-1 and IL-8.
In the present study, we examined the airway epithelial response to H. influenzae using both an in vitro epithelial cell-bacteria interaction system and an in vivo murine infection model. Our results indicate that H. influenzae interaction with airway epithelial cells rapidly induces IL-8 production, ICAM-1 expression, and ICAM-1-dependent neutrophil adherence. Lipooligosaccharide (LOS)3 released from H. influenzae is not sufficient for direct induction of epithelial cell ICAM-1; instead, other molecules that appear to be present on the bacterial cell surface are required for this epithelial cell response. Epithelial cell ICAM-1 expression coincides with increased chemokine levels and neutrophil recruitment in a murine model of airway infection with H. influenzae, suggesting the physiologic relevance of this response to airway defense against bacteria. Direct induction of specific epithelial genes (such as ICAM-1 and IL-8) may allow for rapid targeting and activation of neutrophils at sites of H. influenzae infection, resulting in efficient innate defense in the airway.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following bacteria were used: nontypable H.
influenzae strain 12; a strain 12 mutant deficient in expression
of the high m.w. (HMW) 1 and 2 adhesins (19); nontypable
H. influenzae strain N187; a strain N187 mutant deficient in
expression of the HMW1, HMW2, and Haemophilus adherence and
penetration protein (Hap) adhesins (20); nontypable
H. influenzae strain 11; a strain 11 mutant deficient in
expression of the H. influenzae adhesin (Hia)
(21); H. influenzae serotype b strain Eagan; an
Eagan mutant lacking capsule expression (22); P.
aeruginosa strains PAK and PA01 (gifts from S. Lory, University of
Washington) (13, 23); Escherichia coli stains
HB101 and DH5 (Life Technologies, Gaithersburg, MD); clinical
isolates of Staphylococcus aureus (CDC1, 97-033191; CDC2,
97-033192; Centers for Disease Control and Prevention, Atlanta, GA); a
clinical isolate of Streptococcus pneumoniae (St. Louis
Children’s Hospital); and Bordetella pertussis strains
Tohama I (virulent phase) and Tohama III (isogenic avirulent phase)
(gifts from W. Goldman, Washington University) (24).
H. influenzae was grown on chocolate blood agar supplemented
with 1% Isovitalex (25) or in brain-heart infusion broth
supplemented with haemin and NAD (22). P.
aeruginosa and E. coli were grown on Luria-Bertani (LB)
agar or in LB broth. S. aureus was grown on trypticase soy
agar with 5% sheep blood (Becton Dickinson, Cockeysville, MD) or in
Nutrient Broth No. 2 (Oxoid, London, U.K.). S. pneumoniae
was grown on trypticase soy agar with 5% sheep blood or in Todd-Hewitt
broth (Difco, Detroit, MI). B. pertussis was grown on agar
or in broth containing modified Stainer-Scholte medium with supplements
as described previously (24). To prepare aerated,
log-phase bacteria, colonies from a fresh plate were inoculated into
3–20 ml of broth, and the culture was incubated with rotary shaking at
37°C until a turbidimetric estimate (OD600 =
0.6–0.7) of 
1–3 x 109 CFU/ml was
achieved (1.5–2 h). Bacteria were quantitated by plating serial
dilutions of broth cultures, culture media, epithelial cells, or
homogenized lung or agar particles on solid agar media appropriate for
each organism. LOS (a gift from M. Apicella, University of Iowa) was
purified from nontypable H. influenzae strain 2019 by the
phenol/chloroform/petroleum ether method (26).
Diphosphoryl lipid A (a gift from N. Qureshi, University of Wisconsin)
was purified from Rhodobacter sphaeroides as described
previously (27).
Epithelial cell culture and infection
The A549 respiratory epithelial cell line (American Type Culture
Collection CCL-185; Manassas, VA) was cultured in RPMI 1640 medium
supplemented with 10% heat-inactivated (56°C for 1 h) FBS, 2 mM
L-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin,
and 0.25 µg/ml amphotericin B. The BEAS-2B normal airway epithelial
cell line (a gift from J. Lechner, National Cancer Institute)
(28) was cultured in LHC-8e medium (29) in
flasks coated with collagen/albumin as described previously (4, 9, 10, 11, 30). HUVEC (Clonetics, San Diego, CA) were maintained in
supplemented endothelial basal medium in flasks coated with gelatin up
to passage 10 as described previously (4, 6, 7, 9). The
cystic fibrosis airway epithelial cell line IB3–1 (genotype
F508/W1282X) and the matched rescued cell line C38 (gifts from P.
Zeitlin, Johns Hopkins University) (31, 32) were cultured
in LHC-8 medium (29) supplemented with 5% FBS, 100 U/ml
penicillin, 100 µg/ml streptomycin, 0.2 mg/ml imipenem (Merck, West
Point, PA), 80 µg/ml tobramycin (Eli Lilly, Indianapolis, IN), and
2.5 µg/ml amphotericin B in flasks coated with collagen/albumin.
Cells were cultured at 37°C in 5% CO2 and
incubated with bacteria in the same media without antibiotics except as
indicated. We have found that an inoculum of 
200 CFU/epithelial cell
(107 CFU/ml of media) results in maximal ICAM-1
induction, correlating with a minimal requirement for bacterial growth
to reach stationary phase levels
(108–109 CFU/ml of
media).
Enzyme-linked immunoassays
Protein levels on the surface of epithelial and endothelial
cells at confluence on 24- or 96-well tissue culture plates were
determined using an ELISA as described previously (4) with
minor modifications. In some experiments, epithelial cells were
separated from bacteria using a Transwell culture system with a
0.1-µm pore size polycarbonate membrane (Corning-Costar, Acton, MA),
culturing the epithelial cells in the lower chamber and bacteria in the
upper chamber. After incubation with bacteria, epithelial cell
monolayers were washed in buffer consisting of PBS with 0.5% BSA, 1 mM
CaCl2, and 1 mM MgCl2 and
then fixed for 15 min at 25°C in 1% paraformaldehyde in PBS. The
cell monolayer was thoroughly washed, and nonspecific Ags were blocked
by exposure to a blocking solution (3% BSA in washing buffer) for 90
min at 37°C. The cells were then exposed to 1 µg/ml of murine mAb
84H10 against human ICAM-1 (Immunotech, Westbrook, ME), murine mAb
B9.12.1 against human HLA-A,B,C/MHC class I (Immunotech), or control
murine mAb for 1 h at 37°C. Concentrations of mAb that gave a
maximal response on Ab-titration curves were used. Primary mAb binding
was detected by incubation with 0.04 µg/ml
F(ab')2 fragment-specific, goat anti-mouse
IgG conjugated to HRP (Pierce, Rockford, IL) in blocking solution
followed by 3,3',5,5'-tetramethylbenzidine plus
H2O2 substrate (Pierce).
The color reaction was stopped after 15 min by addition of an equal
volume of 2 M H2SO4, and
absorbance was determined at 450 nm using a spectrophotometric
microplate reader (Bio-Rad, Hercules, CA). Values for control mAb were
subtracted from values for ICAM-1 and MHC class I to control for
background Ab binding. Cytokines (IL-1ß, IL-8, and TNF-) released
into epithelial cell culture media were assayed using quantitative
sandwich enzyme-linked immunoassay kits (R&D Systems, Minneapolis, MN).
According to the manufacturer, the sensitivity of these assay systems
for IL-1ß is <1 pg/ml, for IL-8 is <10 pg/ml, and for TNF- is
<0.18 pg/ml.
RNA blot analysis
Total cellular RNA was isolated from epithelial cell monolayers
after bacterial infection using guanidinium isothiocyanate lysis
followed by either centrifugation through cesium chloride or a column
isolation kit (Qiagen, Santa Clarita, CA) and was subjected to RNA blot
analysis as described previously (4, 11). Probes were
labeled with [-32P]dCTP (>3000 Ci/mmol) by
the random-primer technique (Multiprime; Amersham International, Little
Chalfont, U.K.) and included: 1) a 1.4-kb XhoI fragment from
pCD1.8 containing human ICAM-1 cDNA in pCDM8 (a gift from M. Dustin,
Washington University and D. Staunton, Harvard University)
(33); 2) a 0.55-kb XbaI-HindIII
fragment from pHcGAP containing human GAPDH cDNA in pBR322 (American
Type Culture Collection).
Bacterial adherence assay
Bacterial adherence to epithelial cells at confluence on 24-well tissue culture plates was quantified using an assay described previously (19). In brief, bacteria were inoculated onto epithelial cell monolayers in antibiotic-free media, and plates were centrifuged at 165 x g for 5 min to facilitate bacteria-epithelial cell contact. After incubation at 37°C in 5% CO2 for 30 min, monolayers were rinsed five times with PBS to remove nonadherent organisms and then released with 0.05% trypsin plus 0.5% EDTA in PBS. Well contents were agitated, and serial dilutions were plated on chocolate blood agar for bacterial quantitation.
Neutrophil adherence assay
Neutrophil adherence to epithelial cells at confluence on
96-well tissue culture plates was quantified using an assay modified
from ones described previously (4, 6, 7). Neutrophils in
whole blood from healthy donors were isolated by a one-step gradient
through Polymorphoprep (Nycomed Pharma AS, Oslo, Norway) followed by
two cycles of hypotonic lysis to remove contaminating erythrocytes.
Cell suspensions (>95% neutrophils) were labeled with 1 µg/ml of
calcein AM fluorescent marker (Molecular Probes, Eugene, OR) in
calcium- and magnesium-free (CMF) HBSS containing 0.1% BSA for 30 min
at 25°C. The labeled neutrophils were collected by centrifugation,
counted, incubated in CMF-HBSS plus BSA containing 100 U/ml TNF- (a
gift from Genentech, South San Franscisco, CA) for 15 min at 37°C to
activate adhesive proteins, and resuspended in CMF-HBSS plus BSA.
Epithelial cell monolayers were washed thoroughly, and 3 x
105 neutrophils were added to each well in 100
µl of HBSS plus BSA containing calcium and magnesium. In some tissue
culture wells, neutrophils were added that had been pretreated with 30
µg/ml of blocking murine mAb R15.7 against human
ß2 integrin/CD18 (a gift from R. Rothlein,
Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT) (34)
or control murine mAb for 30 min at 37°C. In other wells, the cell
monolayer was pretreated with 50 µg/ml of blocking F(ab) of murine
mAb R6.5 against human ICAM-1 (a gift from R. Rothlein)
(34) or control murine mAb F(ab) for 30 min at 37°C. Ab
concentrations that caused maximal inhibition of neutrophil adherence
were used (34, 35), and Abs were not removed during the
subsequent adherence period. Neutrophil adherence was allowed to occur
for 15 min at 37°C. Fluorescence of all neutrophils added to the well
was quantitated using a fluorescence microplate reader (Molecular
Devices, Sunnyvale, CA) set at 
ex = 485 nm and
em = 538 nm. Nonadherent neutrophils were then
removed by filling each well with HBSS plus BSA and centrifuging
inverted plates for 2 min at 200 x g. This step was
repeated, 100 µl HBSS plus BSA was added, and the fluorescence of
neutrophils adherent to the monolayer in each well was determined using
the microplate reader. Background fluorescence was determined from
wells in which neutrophils were not added, and this background value
was subtracted from values obtained from experimental wells.
Measurements of fluorescence were converted to number of adherent
neutrophils/well using the formula: (adherent fluorescence/total
fluorescence) x (3 x 105
neutrophils). Preliminary experiments established a high correlation
(r = 0.99) between measurements of neutrophils obtained
by determining fluorescence vs the values obtained for manual cell
counting (4, 6, 7).
Murine lung infection model
Nontypable H. influenzae strain 12 was incorporated
into agar particles using a technique modified from ones described
previously (36). Bacteria were harvested from a 20-ml
aerated, log-phase culture by centrifugation (10,000 x
g, 4°C, 10 min) and were resuspended in 2x infection
buffer (1x = 100 mM NaCl, 10 mM KCl, 10 mM sodium phosphate, pH 7.4,
10 mM glucose, 2% casamino acids) at a concentration of 2 x
109 CFU/ml. The bacterial suspension was warmed
to 45°C, and 3 ml was mixed with 3 ml of molten 4% Noble Agar
(Difco) at 45°C. The bacteria-agar mixture was aspirated into a
syringe, and 3 ml was immediately injected through a 23-g needle into
27 ml of rapidly stirring 1x infection buffer at 4°C. This produced
a suspension containing bacteria at 108 CFU/ml
incorporated into amorphous agar particles ranging in size from 30 to
200 µm. Four-week-old C57BL/6 mice (Taconic, Germantown, NY) housed
under pathogen-free conditions were anesthetized with 87 mg/kg of
ketamine HCl and 13 mg/kg of xylazine HCl injected i.m. Mice were then
placed in an intubation apparatus (37) and orotracheally
intubated with a 22-gauge i.v. catheter. A 30-µl volume of PBS,
control agar particles, or bacteria-agar particle suspension at 37°C
(followed by 100 µl of air) was injected into the airway, and the
mice were allowed to recover.
Lung infection quantitation and immunohistochemistry
After a specified duration of infection, mice were anesthetized and then euthanized by cervical dislocation, the lungs were exposed, and the pulmonary vascular system was flushed via the right ventricle with sterile saline. For murine chemokine quantitation, the diaphagmatic lobe of the right lung was ligated at its root, removed, and stored at -80°C. After harvest at all time points, lung tissue was homogenized in 0.5 ml of 50 mM sodium phosphate, pH 6.0 containing 0.5% hexadecyltrimethylammonium bromide using a glass tissue grinder (Kontes Glass, Vineland, NJ) followed by sonication using a sonic dismembrator (Fisher Scientific, Pittsburgh, PA) set at 30%. Samples were cleared by centifugation at 15,000 x g for 15 min at 4°C, and murine CXC chemokines (KC and macrophage inflammatory protein-2 (MIP-2)) were assayed in supernatants using quantitative sandwich enzyme-linked immunoassay kits (R&D Systems). The sensitivity of these assay systems for KC is <2.0 pg/ml and for MIP-2 is <1.5 pg/ml. For bacterial quantitation, the left lung was ligated at its root and removed under sterile conditions, minced with a razor blade in a petri dish with 0.5 ml of PBS, and homogenized using a glass tissue grinder. Serial dilutions were then inoculated onto solid agar media for culture. For tissue staining, the trachea was cannulated with a 22-gauge i.v. catheter, and the right lung was inflated under 25 cm H2O pressure and fixed with 10% buffered formalin for 18 h. Paraffin-embedded 6-µm sections were deparaffinized in a D-limonene-based clearing solution (Stephens Scientific, Riverdale, NJ) and rehydrated in graded ethanol solutions. Serial sections stained with hematoxylin and eosin were used to assess inflammatory cell influx near agar particles. For ICAM-1 immunostaining of lung sections, endogenous peroxidase activity was blocked by treatment with 5% hydrogen peroxide for 5 min, and nonspecific Ags were blocked by exposure to a blocking solution (3% nonimmune goat serum in PBS) for 30 min at 25°C. Slides were then incubated with 2 µg/ml of hamster mAb 3E2 against murine ICAM-1 (PharMingen, San Diego, CA) or isotype-matched control hamster mAb for 18 h at 4°C. Primary Ab binding was detected by incubation with 7.5 µg/ml biotinylated goat anti-hamster IgG Ab (Vector Laboratories, Burlingame, CA) in blocking solution followed by streptavidin-conjugated HRP and 3,3'-diaminobenzidine plus H2O2 substrate (Vector Laboratories). Tissue sections were counterstained with hematoxylin, dehydrated in graded ethanol, and mounted for microscopy (Model D-7082; Carl Zeiss, Thornwood, NY). Images were acquired using a computerized digital camera system (SPOT-2; Diagnostic Instruments, Sterling Heights, MI) interfaced with Photoshop 5.0 software (Adobe Systems, San Jose, CA).
Statistical analysis
Neutrophil and bacterial adherence assays and enzyme-linked immunoassay were performed two to four times to insure reproducible results, and values from representative experiments were analyzed for statistical significance using a one-way ANOVA for a factorial experimental design. The multicomparison significance level for the one-way ANOVA was 0.05. If significance was achieved by one-way analysis, post-ANOVA comparison of means was performed using Scheffe F tests (38).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
ICAM-1 protein levels on the surface of cultured human respiratory
epithelial cell monolayers were examined using an enzyme-linked
immunoassay based on specific anti-ICAM-1 mAb (4).
These experiments revealed low constitutive ICAM-1 expression on A549
respiratory epithelial cells and a marked increase in ICAM-1 levels
after incubation for 24 h with nontypable H. influenzae
strain 12 (Fig. 1
A).
Incubation of BEAS-2B human bronchial epithelial cells with H.
influenzae strain 12 also resulted in a significant increase in
ICAM-1 expression. The relative magnitude of induction was less marked
with BEAS-2B cells, reflecting relatively high constitutive levels of
ICAM-1 on these cells that do not recapitulate low levels of expression
observed on tracheal and bronchial epithelial cells in vivo (4, 39). We also tested IB3–1 and C38 airway epithelial cells and
found that these cell lines responded to H. influenzae by
induction of cell-surface ICAM-1 at levels similar to those observed
with BEAS-2B cells (results not shown). Examination of the relationship
between inoculum and ICAM-1 induction revealed that lower inocula
induced less ICAM-1 at early time points. The effect of inoculum size
was less pronounced after 12 h, likely reflecting attainment of
stationary phase and bacterial concentrations of
108–109 CFU/ml by this
point, regardless of the initial inoculum (A.G.F. and D.C.L.,
unpublished observation).
|
B). Taken together, these results suggest that H.
influenzae has the ability to directly activate epithelial cell
ICAM-1 expression through control of mRNA level, similar to regulation
of this gene by certain cytokines (4). P. aeruginosa also induces ICAM-1 expression on respiratory epithelial cells
To determine whether induction of ICAM-1 is a uniform response of
respiratory epithelial cells to bacterial infection, we examined the
epithelial response to several other respiratory pathogens. In
performing these experiments, we incubated bacteria with A549 cell
monolayers and then verified that each bacterial strain achieved a
density of at least 108 CFU/ml (results not
shown). Following incubation with epithelial cells for 24 h,
H. influenzae type b strain Eagan induced cell-surface
ICAM-1 expression similar in magnitude to that seen with nontypable
H. influenzae strain 12 (Fig. 2A). In contrast, isolates of
S. pneumoniae and B. pertussis failed to
stimulate ICAM-1 expression. Experiments with P. aeruginosa
were complicated by significant cytotoxic effects on the monolayer
after 12 h of infection, possibly due to release of exotoxins or
proteases. To circumvent this problem, we performed RNA blot analysis
after infection of the monolayer with bacteria for 8 h (the time
point of peak ICAM-1 mRNA expression after H. influenzae
infection). Under these conditions, infection with P.
aeruginosa induced slightly higher levels of epithelial cell
ICAM-1 mRNA than did H. influenzae, while strains of
E. coli and S. aureus had minimal effect (Fig. 2B). Although we cannot exclude the possibility that the
culture system used affected the phenotype or growth of these bacterial
species differently or that the isolates tested lack appropriate
virulence factors required for this effect, our results suggest that
selected bacterial species important in human lung infections have the
capacity to induce respiratory epithelial cell ICAM-1 expression.
|
Recent studies examining the interaction between P.
aeruginosa and respiratory epithelial cells suggest that
epithelial cells may participate in neutrophil recruitment to sites of
infection through release of cytokines such as IL-8 (12, 13). To examine the effect of H. influenzae infection
on other epithelial cell inflammatory mediators, we performed
enzyme-linked immunoassays for epithelial cell expression of MHC class
I and release of IL-1ß, IL-8, and TNF-. Using A549 cells, we
observed minimal effects on MHC class I surface expression and IL-1ß
and TNF- secretion (results not shown). In contrast, similar to
observations with H292 lung epithelial cells (40),
infection with H. influenzae resulted in high levels of IL-8
release (Fig. 3A). The time
course for induction of IL-8 was similar to the relatively rapid
expression of ICAM-1 in response to H. influenzae (Fig. 3B), suggesting preferential activation of a specific
epithelial cell gene subset directed at early defense against bacteria.
Because of the coordinated effects of IL-8 and ICAM-1 on neutrophils,
this epithelial cell gene subset may be particularly important in
mediating neutrophil recruitment and activation in the airway in
response to bacteria.
|
To further define the nature of the interaction between H.
influenzae and epithelial cells that results in increased ICAM-1
expression, we performed a set of experiments in which incubation
conditions were adjusted to affect specific characteristics of the
interaction (e.g., bacterial viability, direct bacteria-epithelial cell
contact). In these experiments, sufficient inocula of nontypable
H. influenzae strain 12 (108 CFU/ml)
were added to epithelial cells to insure that treatment effects were
not simply due to inhibition of bacterial growth. We found that
treatment of H. influenzae with a bacteriostatic
concentration of chloramphenicol sufficient to selectively inhibit
bacterial protein synthesis (20, 40, 41) and killing of
H. influenzae using a bactericidal dose of gentamicin had no
significant effect on bacterial capacity for induction of epithelial
cell ICAM-1 (Fig. 4A),
suggesting that induction of ICAM-1 is mediated by a constitutive
bacterial factor. However, sterile filtrate from a viable culture of
H. influenzae in an amount equivalent to the volume added
for a stationary-phase level of infection had minimal effect on
epithelial cell ICAM-1 levels, indicating that the relevant bacterial
factor(s) is not released from the organism or does not pass through a
0.2-µm pore-size polyethersulfone filter. To support this conclusion,
direct interaction of H. influenzae with epithelial cells
was compared with interaction between bacteria and epithelial cell
monolayer separated (in media during the incubation period) by a
0.1-µm pore-size polycarbonate membrane. Despite the fact that a
similar density of bacteria was achieved when H. influenzae
was cultured above the filter compared with incubation in direct
contact with epithelial cells (>109 CFU/ml),
separation of bacteria from epithelial cells by a semipermeable
membrane almost completely eliminated ICAM-1 and IL-8 induction (Fig. 4B).
|
108 CFU/ml (Fig. 4C). Bacteria that
were treated with gentamicin and then washed induced levels of
epithelial cell ICAM-1 that were similar to inocula that were untreated
or gentamicin treated but not washed. H. influenzae that
were treated with paraformaldehyde and then washed induced lower levels
of ICAM-1, and this may reflect alteration of bacterial surface Ags
induced by this treatment or a mild loss of bacteria that we observed
with washing (results not shown). These findings indicate that H.
influenzae induction of ICAM-1 is mediated by a factor not
released from the bacterial surface and suggest that
bacteria-epithelial cell contact is required for ICAM-1 induction. The
release of a soluble factor from the bacteria that has physical
characteristics that preclude significant permeability through
separation membranes used in our experiments (e.g., tendency to adhere
to the membranes or form large aggregates) seems less likely.
Interestingly, treatment of epithelial cells with dexamethasone
resulted in partial inhibition of ICAM-1 induction by H.
influenzae (Fig. 4A), suggesting that the epithelial
signaling pathway activated by the bacteria is affected by
corticosteroids. H. influenzae LOS does not induce ICAM-1 expression on respiratory epithelial cells
LPS is a component of the outer membrane of Gram-negative
bacteria, but is also released during growth and death of these
bacteria (42). Unlike enteric Gram-negative bacteria,
Haemophilus, Bordetella, Moraxella,
and Neisseria species synthesize a form of LPS containing an
oligosaccharide linked to lipid A without repeating subunit O-Ag
polysaccharide chains (referred to as lipooligosaccharide or LOS)
(42, 43). H. influenzae LOS is highly
hydrophobic, and its biochemical characteristics may limit permeability
through the membranes we tested. Therefore, to directly assess the role
of H. influenzae LOS in epithelial cell activation, we
tested a highly purified form of LOS from nontypable H.
influenzae (26) and used diphosphoryl lipid A
purified from Rhodobacter sphaeroides (RsDPLA), which
functions as a competitive LPS antagonist (27, 44).
Biologic activity of these preparations was verified using the
LPS-responsive HUVEC model, which revealed that H.
influenzae LOS causes induction of ICAM-1 on HUVEC that is
inhibitable with 10-fold higher concentrations of RsDPLA (Fig. 5A). However, induction of
ICAM-1 on A549 cells by both viable and gentamicin-treated H.
influenzae was not inhibited by RsDPLA (Fig. 5B).
Because RsDPLA does not have the capacity to block high levels of LPS
due to its partial agonist activity at high concentrations (Fig. 5A), we tested purified H. influenzae LOS on A549
cells at concentrations (1–100 µg/ml) previously shown to be
released by the bacteria (42). We found no induction of
ICAM-1 on A549 cells by this LOS preparation despite high levels of
induction on HUVEC (Fig. 5C). Although these results do not
rule out the possibility that high levels of LOS are necessary for
H. influenzae induction of epithelial cell ICAM-1
expression, they argue that LOS is not sufficient and another component
of H. influenzae is required for this effect.
|
A number of proteins that promote adherence to epithelial cells have been identified on the surface of nontypable and type b strains of H. influenzae, and expression of these adhesins varies depending on the isolate (3). Approximately 75% of nontypable strains express high-molecular-weight adhesins referred to as HMW1 and HMW2 (45). Most of the remaining nontypable strains and virtually all type b strains express an adhesin called Hia or a homologous protein designated Hsf (21, 46). Among nontypable strains, an Hia homologue is universally absent from strains that express HMW1/HMW2-like proteins (21, 47). Another H. influenzae adhesin is the Hap serine protease (48), which is present in strains expressing HMW1 and HMW2 and possibly in other strains as well. In addition, selected isolates express hemaglutinating pili (49, 50).
To examine the potential role of each of these known H.
influenzae adhesins in inducing epithelial cell ICAM-1 expression,
we measured ICAM-1 levels on A549 cells incubated with a series of
wild-type H. influenzae strains or isogenic
adhesin-deficient mutants. In these experiments, the three wild-type
nontypable strains (strains 12, N187, and 11), which lack the genes
involved in pilus biogenesis, induced similar levels of ICAM-1
expression (Fig. 6A).
Comparison of strain 12 to an isogenic mutant lacking HMW1 and HMW2
revealed a significant difference in adherence to A549 cells but no
difference in ability to induce ICAM-1 expression (Fig. 6, A
and B). Similarly, comparison of strain N187 and an isogenic
mutant lacking HMW proteins and Hap demonstrated reduced adherence by
the mutant strain but equivalent capacity to induce ICAM-1 expression
on A549 cells. Strain 11 and a strain 11 mutant deficient in expression
of Hia exhibited comparable levels of adherence to A549 cells
(suggesting absence of the Hia receptor on these cells) and induced
similar levels of ICAM-1. We also examined the effect of the type b
capsule on H. influenzae adherence to A549 cells and
induction of ICAM-1 expression. Of note, this bacterial capsule is
known to interfere with the function of some adherence mechanisms but
not others (22). We found that encapsulated strain Eagan
demonstrated minimal stable adherence to A549 cells (Fig. 6B). Nevertheless, this strain was still effective at
inducing epithelial cell ICAM-1 expression (Fig. 6A).
Considered together, these results indicate that H.
influenzae induction of epithelial cell ICAM-1 is independent of
the HMW1, HMW2, Hia, and Hap adhesins, which promote high-affinity
bacterial adherence. In addition, induction is independent of
piliation, which appears to be associated with lower-affinity adherence
(51). It is possible that ICAM-1 expression is mediated by
other bacterial surface molecules involved in low-affinity, unstable
interactions with epithelial cells.
|
To assess the functional activity of ICAM-1 induced on respiratory
epithelial cells by H. influenzae, neutrophil adherence to
A549 cells was determined before and after incubation with H.
influenzae. Infection of A549 cells for 24 h with H.
influenzae strain 12 resulted in enhanced epithelial-neutrophil
adherence using an in vitro assay system (Fig. 7). Specific blocking mAb were added to
the assay to determine the contribution of epithelial cell ICAM-1 and
its leukocyte counterreceptors, members of the
ß2 integrin (CD18) family. We found that
increased epithelial-neutrophil adherence after epithelial cell
infection with H. influenzae was almost completely
inhibitable by treatment with anti-ICAM-1 or anti-CD18 mAbs.
The more efficient blockade that we observed with anti-CD18
compared with anti-ICAM-1 mAb is consistent with other reports
(4, 5) and may reflect interactions between other
epithelial cell surface molecules and leukocyte CD18 receptors. The
expression of functional ICAM-1 in response to H. influenzae
infection confirms that this mechanism could contribute to neutrophil
recruitment or activation in the airway.
|
To examine the in vivo epithelial cell response to H.
influenzae, we modified a murine model of bacterial airway
infection that has been used to examine the pulmonary response to
respiratory pathogens, including H. influenzae (36, 52). In this model, small airway entrapment of agar particles
impairs bacterial clearance from the lung, resulting in infection and
leukocyte recruitment in the airway. Because a murine structural
homologue for human IL-8 has not been identified (53), we
determined the levels of KC and MIP-2, which are functional CXC
chemokine homologues for IL-8 due to their ability to induce neutrophil
chemotaxis (54, 55, 56). Similar to other models of pulmonary
infection (54, 55, 56), high levels of these chemokines were
detected in the lungs of mice infected with H. influenzae
for 24 h (Fig. 8). KC and MIP-2
levels were lower after 72 h of infection compared with the 24-h
time point. The expression of these chemokines may account for
selective neutrophil recruitment early in the course of bacterial
infection in the lung.
|
).
Immunohistochemical analysis of lungs from these mice revealed few
leukocytes and little ICAM-1 in the airway epithelium at 24 h
after injection of either PBS or agar particles without bacteria.
High-level constitutive expression of ICAM-1 on type I alveolar
epithelial cells served as a positive control for ICAM-1 staining in
each sample (39). There was also minimal leukocyte influx
and ICAM-1 expression in the airways of mice infected for 4 h by
particles containing H. influenzae. However, we observed
greatly increased numbers of leukocytes (particularly neutrophils)
located near agar particles and increased airway epithelial cell ICAM-1
staining in samples from mice infected for 24 h. Thus, airway
epithelial cell ICAM-1 expression after infection with H.
influenzae was time dependent in a pattern that correlated with
our in vitro studies. Increased leukocyte numbers persisted in the
airways of mice after 72 h of infection, although airway ICAM-1
staining was decreased at this time point compared with the 24-h time
point. Leukocytes localized predominantly in the airway lumen during
infection, consistent with the concept that epithelial cell expression
of ICAM-1 is responsible for a component of leukocyte recruitment
and/or activation in the airway.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Using four respiratory epithelial cell lines, we observed that
infection with H. influenzae directly stimulated ICAM-1
expression, suggesting that this mechanism could function in concert
with cytokine-dependent expression and may be generalized to airway
epithelial cell behavior in vivo. Given that unstimulated type II
alveolar, tracheal, and bronchial epithelial cells generally express
low ICAM-1 levels in vivo (4, 39), we performed many of
our experiments using the A549 epithelial cell line, which also
displays low levels of constitutive ICAM-1, allowing us to maximize
detection of increased expression in response to bacteria.
Interestingly, direct induction of ICAM-1 expression and cytokine
secretion in response to microbial infection has also been reported for
nonrespiratory epithelial cells, although the mechanism and cytokine
profile of response is not always the same. These examples include: 1)
increased IL-8, GM-CSF, TNF-, and ICAM-1 by human mucosal epithelial
cells after infection with Neisseria gonorrhoeae (58, 59); 2) increased IL-6, IL-8, and ICAM-1 by human uroepithelial
cells after infection with uropathogenic E. coli
(60); and 3) increased IL-8, TNF-, GM-CSF, monocyte
chemotactic protein-1, and ICAM-1 by human intestinal epithelial cells
after invasion by several strains of Gram-positive and Gram-negative
bacteria (61, 62). Thus, our observation of respiratory
epithelial cell secretion of IL-8 and expression of ICAM-1 in response
to H. influenzae infection fits well with the concept that
host epithelial cells interacting with potentially pathogenic organisms
have evolved mechanisms to activate defense responses when infection
occurs. Bacterial induction of host cell mediators is generally rapid,
suggesting that these mechanisms allow for early mobilization of
inflammatory cells to defend the host at sites of infection.
Mammalian cell responses to the presence of bacteria are often initiated through detection of specific bacterial products, such as components of LPS, pili, or peptidoglycan (13, 24, 63, 64). Determinants for activation of epithelial cell defense genes in response to bacteria depend on physical characteristics of the bacterial stimuli, including whether factors are constitutively produced, expressed on the cell surface, and/or released by the bacteria. Our findings suggest that increased epithelial ICAM-1 expression after H. influenzae infection requires direct epithelial cell contact (possibly through a low-affinity interaction) with a constitutive molecule on the bacterial cell surface. However, for several reasons, we directed initial experiments toward determining if released LOS played a role in epithelial cell ICAM-1 induction. First, LPS induces a variety of responses in mammalian cells and may regulate cytokine gene expression in airway epithelial cells (26, 63, 64). Second, we could not initially exclude the possibilities that a bacterial extracellular factor that induces epithelial cell ICAM-1 may have physical characteristics (presumably other than molecular size) precluding significant permeability through 0.2- and 0.1-µm membranes or that bacteria-associated LOS may activate epithelial cells. Lastly, partially purified LOS from nontypable H. influenzae has been reported to increase cytokine release and ICAM-1 expression on isolated airway epithelial cells from smokers with lung cancer (65). However, the H. influenzae LOS used for this previous study was a partially purified preparation used at high concentration, and we postulate that impurities in that preparation might include other non-LOS factor(s) that induce ICAM-1 expression on epithelial cells.
Our work provides multiple lines of evidence that LOS is not responsible for induction of epithelial cell ICAM-1 by H. influenzae. We were unable to induce this effect by high levels of purified LOS from H. influenzae. As a corollary, we were unable to inhibit induction of epithelial cell ICAM-1 expression using a specific competitive inhibitor of LPS. In addition, we found that incubation with other LPS-releasing Gram-negative bacteria (B. pertussis and E. coli) at similar levels of infection did not induce epithelial cell ICAM-1. Furthermore, the absence of serum and thus the serum-associated LPS cofactors LPS-binding protein and soluble CD14 (64, 66) in culture media had no effect on H. influenzae induction of epithelial cell ICAM-1 (A.G.F. and D.C.L., unpublished observation). We and others have previously been unable to induce ICAM-1 expression or IL-8 release in human respiratory epithelial cells by several different preparations of LPS from other organisms, including E. coli (Refs. 13 and 67 and A.G.F. and D.C.L., unpublished observation). Although there are clearly structural differences between H. influenzae LOS and other forms of LPS, the biologic activity of LPS resides in the lipid A moiety, and the presence and structural characteristics of lipid A are highly conserved in diverse Gram-negative bacteria, including E. coli and H. influenzae (63, 64, 68). The relative lack of sensitivity for airway epithelial cell ICAM-1 expression in response to LPS likely reflects the fact that epithelial cells are continuously exposed to many bacterial products in the environment, and therefore other factors are used by epithelial cells to differentiate bacterial presence from infection. Interestingly, epithelial expression of the adenoviral oncoprotein E1A changes the epithelial phenotype into one that is LPS responsive, suggesting that factors in the environment may alter LPS signaling pathways in epithelial cells (69, 70). Although our results indicate that epithelial cells in the airway are not responsive to LOS alone, they do not rule out the possibility that LOS may be necessary or contribute to ICAM-1 induction by another factor. This system could be analogous to that for induction of epithelial cell IL-6 in which it appears that other as yet unidentified factors work in synergy with H. influenzae LOS (40, 65, 71).
Bacterial attachment to epithelial cells is fundamental to the process of colonization and to the pathogenesis of disease and is mediated by bacterial factors called adhesins, which interact with specific host cell-surface structures (3). Adhesins on the surface of H. influenzae include polymeric structures such as hemagglutinating pili and monomeric or oligomeric molecules, including HMW1, HMW2, Hia, and Hap. Nontypable strain 12 does not express pili (3, 50), yet this strain was capable of activating epithelial cell ICAM-1 expression, suggesting that other H. influenzae factors mediate this response. Isogenic mutants lacking HMW1, HMW2, Hia, and/or Hap retained the capacity to induce epithelial cell ICAM-1, suggesting that these adhesins are not essential for this effect. Similarly, an encapsulated type b strain was capable of inducing ICAM-1 despite the fact that encapsulation interferes with the function of nonpilus adhesins and virtually eliminated adherence to A549 cells. Our work suggests that the factors that induce epithelial cell ICAM-1 are not the high-affinity adhesins that are presumably used by H. influenzae for initial bacterial adherence and colonization. It is interesting to speculate that progression from colonization to disease requires interaction between low-affinity bacterial factors and the epithelial cell surface, thereby activating the inflammatory response.
The possible physiologic significance of our findings was indicated by
demonstrating increased ICAM-1-dependent neutrophil adherence to
epithelial cells in vitro after infection with H.
influenzae. This finding was supported further by a murine model
of pulmonary infection in which airway infection resulted in neutrophil
recruitment and increased CXC chemokine levels that coincided with
increased epithelial cell ICAM-1 expression. In this model, agar
particles were used to impair clearance of bacteria because we have
found that mice without impaired airway function generally clear
H. influenzae within 24 h (L.P. and D.C.L., unpublished
observation). This requirement for decreased airway clearance may be
similar in humans in which preexisting lung diseases associated with
airway dysfunction (e.g., cystic fibrosis, bronchiectasis, bronchitis)
predispose patients to infection with H. influenzae. The in
vivo model system used for our studies does not differentiate the role
of direct interaction between bacteria and epithelial cells from the
better understood cytokine-dependent mechanisms for inflammatory gene
regulation. IFN-
and TNF- appear to be the predominant cytokine
stimulants of airway epithelial cell ICAM-1 expression (4, 5, 67), and epithelial cell induction of ICAM-1 by LPS in vivo may
be TNF- dependent (72). However, rapid induction of
ICAM-1 and other mediators by direct bacterial interaction with
epithelial cells, independent of inflammatory mediator release by other
cell types, may allow for early targeting of neutrophils to sites of
H. influenzae infection resulting in efficient innate
defense in the airway. Taken together, our findings suggest a model for
leukocyte migration and activation in the airway in which expression of
cell adhesion molecules and chemokines by epithelial cells in response
to bacterial infection may be mediated both indirectly through
communication with other cell types by cytokines and/or directly by
bacteria in the airway.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Dwight C. Look, Washington University School of Medicine, Campus Box 8052, 660 South Euclid Avenue, St. Louis, MO 63110. E-mail address:
3 Abbreviations used in this paper: LOS, lipooligosaccharide; HMW, high m.w.; CMF, calcium- and magnesium-free; Hap, Haemophilus adherence and penetration protein; Hia, H. influenzae adhesin; LB, Luria-Bertani; LHC, Laboratory of Human Carcinogenesis; MIP-2, macrophage inflammatory protein-2; RsDPLA, Rhodobacter sphaeroides diphosphoryl lipid A.
Received for publication July 20, 1999. Accepted for publication February 4, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
in human airway epithelial cells. Am. J. Physiol. 263:L79.
1+ T cell adherence and tissue retention. J. Immunol. 155:1117.[Abstract]
response element binding proteins with the intercellular adhesion molecule-1 (ICAM-1) gene promoter controls the pattern of expression on epithelial cells. J. Biol. Chem. 269:8952.
factor (RpoN) of RNA polymerase. Proc. Natl. Acad. Sci. USA 86:1954.
