Uropathogenic Escherichia coli Triggers Oxygen-Dependent Apoptosis in Human Neutrophils through the Cooperative Effect of Type 1 Fimbriae and Lipopolysaccharide

Abstract

Type 1 fimbriae are the most commonly expressed virulence factor on uropathogenic Escherichia coli. In addition to promoting avid bacterial adherence to the uroepithelium and enabling colonization, type 1 fimbriae recruit neutrophils to the urinary tract as an early inflammatory response. Using clinical isolates of type 1 fimbriated E. coli and an isogenic type 1 fimbria-negative mutant (CN1016) lacking the FimH adhesin, we investigated if these strains could modulate apoptosis in human neutrophils. We found that E. coli expressing type 1 fimbriae interacted with neutrophils in a mannose- and lipopolysaccharide (LPS)-dependent manner, leading to apoptosis which was triggered by the intracellular generation of reactive oxygen species. This induced neutrophil apoptosis was abolished by blocking FimH-mediated attachment, by inhibiting NADPH oxidase activation, or by neutralizing LPS. In contrast, CN1016, which did not adhere to or activate the respiratory burst of neutrophils, delayed the spontaneous apoptosis in an LPS-dependent manner. This delayed apoptosis could be mimicked by adding purified LPS and was also observed by using fimbriated bacteria in the presence of d-mannose. These results suggest that LPS is required for E. coli to exert both pro- and antiapoptotic effects on neutrophils and that the difference in LPS presentation (i.e., with or without fimbriae) determines the outcome. The present study showed that there is a fine-tuned balance between type 1 fimbria-induced and LPS-mediated delay of apoptosis in human neutrophils, in which altered fimbrial expression on uropathogenic E. coli determines the neutrophil survival and the subsequent inflammation during urinary tract infections.

Uropathogenic Escherichia coli strains are a selected subset of the fecal flora that express different virulence factors which enable them to colonize the urinary tract (27). In epidemiological studies, >90% of all E. coli isolates from urinary tract infection (UTI) patients express surface adhesive organelles called type 1 fimbriae (or mannose-sensitive fimbriae) (8, 25, 45). Further evidence supporting the role of type 1 fimbriae as a major urovirulence factor was obtained when workers screened the genes for the nine most common E. coli virulence factors and type 1 fimbriae were the only trait found in all 203 UTI isolates examined (11). In a mouse UTI model, type 1 fimbriae increase the virulence of E. coli for the urinary tract by promoting bacterial persistence and enhancing the inflammatory response to infection (7).

By recognizing d-mannose moieties, the type 1 fimbrial adhesin, FimH, mediates bacterial interactions with a variety of mannose-containing glycoproteins expressed on different host cells (15, 38, 54). In professional phagocytes, such as granulocytes and macrophages, the surface molecules implicated in this mannose-specific FimH recognition include CD11/CD18 (13, 14), CD66 (42), and CD48 (4). The responses to type 1 fimbrial binding include degranulation of mast cells, cytokine release from epithelial cells, and activation of the respiratory burst by granulocytes and macrophages (1, 13, 33, 39), all of which are important events in the inflammatory process. In a murine cystitis model, the FimH adhesin was the determinant of the adherence and exfoliation of bladder epithelial cells through an apoptosis-like mechanism involving caspase activation and DNA fragmentation (37). In addition, both in vitro and in vivo studies have demonstrated that type 1 fimbriated E. coli enhances neutrophil recruitment to the urinary tract in an adhesion-dependent manner specific for the FimH adhesin (16).

Human neutrophils, which are the first cells recruited to the site of infection, play a sentinel role in host defense against microbial invasion (3). In response to pathogens, neutrophils adhere to the endothelium and transmigrate into the infected tissue, where their activation induces production of reactive oxygen species (ROS) and release of granular enzymes that eliminate the intruding pathogen (3). A number of bacterial pathogens are, however, able to circumvent this bactericidal response by modulating the host cell response, such as by induction of apoptosis to eliminate key immune cells. Apoptotic neutrophils lose the ability to migrate, phagocytose, and generate a respiratory burst. Bacterium-induced host cell apoptosis may thereby serve as a pathogenic strategy to evade the host defense and thus modulate the pathogenesis of a variety of infectious diseases (53).

Neutrophils have a short life span, undergoing spontaneous apoptosis within 16 to 20 h in in vitro culture (44). The process of apoptosis down regulates the proinflammatory capacity of these cells and prepares them for removal from tissue by macrophages, thereby preventing them from releasing their toxic contents (43). While the life span of activated neutrophils is tightly regulated, the process of apoptosis can be accelerated or delayed by various inflammatory mediators, such as cytokines (23), bacterial products (2), or local conditions (for example, hypoxia [17]). Apoptosis which is promoted by neutrophil-derived ROS during phagocytosis (52) or inhibited by hypoxia and by the addition of antioxidants (17) suggests that oxygen-dependent mechanisms play a pivotal role in regulating the progression of inflammation.

Lipopolysaccharide (LPS) is a primary target of the innate immune response, and recognition of LPS involves serum proteins, such as LPS-binding protein (LBP) and soluble CD14. LPS complexes are transferred to the membrane-bound form of CD14 (glycosylphosphatidylinositol linked) (12, 46), and transmembrane LPS signaling is achieved by Toll-like receptor 4 (12, 20, 41, 46). An innate response against LPS involves the inhibition of neutrophil apoptosis, which is associated with prolongation of the functional life span of the neutrophil population (28).

In the present study, using an E. coli strain expressing type 1 fimbriae and an isogenic type 1 fimbria-negative mutant of this strain, we found that both type 1 fimbriae and LPS regulate neutrophil apoptosis.

MATERIALS AND METHODS

Reagents.

The chemicals used and their sources are as follows. Superoxide dismutase, catalase, and horseradish peroxidase were obtained from Boehringer (Mannheim, Germany); luminol (5-amino-2,3-dihydro-1,4-phtalazinedione), dimethyl sulfoxide, formyl-methionyl-leucyl-phenylalanine (fMLP), fluorescein 5-isothiocyanate (FITC), diphenylene iodonium (DPI), polymyxin B sulfate, phorbol 12-myristate 13-acetate (PMA), and purified LPS from Salmonella enterica serovar Typhimurium were obtained from Sigma Chemical Co. (St Louis, Mo.); an annexin V apoptosis detection kit and recombinant human tumor necrosis factor alpha (TNF-α) were obtained from R&D Systems (Abingdon, United Kingdom); Polymorph Prep and Lymphoprep were obtained from AXIS-SHIELD PoC AS (Oslo, Norway); tissue culture reagents were obtained from Invitrogen (Lidingö, Sweden); and a Limulus amebocyte lysate endotoxin assay kit was obtained from Charles River Endosafe (Charleston, S.C.). Krebs-Ringer phosphate buffer (KRG) containing 120 mM NaCl, 4.9 mM MgSO₄, 1.7 mM KH₂PO₄, 8.3 mM Na₂HPO₄, 1 mM CaCl₂, and 10 mM glucose (pH 7.3) was made at the Division of Clinical Microbiology (Linköping, Sweden).

Preparation and treatment of human neutrophils.

Human neutrophils were isolated from heparinized whole blood from healthy donors by gradient centrifugation as previously described (6). Briefly, neutrophils in the interphase of Polymorph Prep and Lymphoprep were collected and washed in phosphate-buffered saline (PBS), contaminating erythrocytes were removed by brief hypotonic lysis, and neutrophils were washed in KRG without Ca²⁺. Neutrophils (purity, about 98%) were resuspended in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum (FCS) and 2 mM l-glutamine (RPMI medium).

The NADPH oxidase activity of neutrophils was inhibited by preincubating the cells with 5 μM DPI for 15 min at 37°C and then was stimulated in the presence of the same inhibitor (9). DPI, at the concentration used, did not affect the phagocytosis of serum-opsonized bacteria (40).

Bacterial strains and growth conditions.

Wild-type E. coli strain 1177 (serotype O1:K1:H7) expressing type 1 fimbriae, isolated from children with their first episodes of acute pyelonephritis, the isogenic type 1 fimbria-negative mutant CN1016, and add-back mutant CN1018 were kindly provided by C. Svanborg (Department of Laboratory Medicine, Lund University Hospital, Lund, Sweden) (7). The strains were all hemolysin negative and expressed P fimbriae (7). The bacterial strains were grown statically overnight in Luria broth with appropriate selective antibiotics for expression of type 1 fimbriae as previously described (7). The bacterial concentration was estimated by spectrophotometry (HITACHI U-1100 spectrophotometer; Hitachi Ltd., Tokyo, Japan) and was verified by both microscopy and a CFU assay. Aliquots of a bacterial culture were washed by centrifugation in PBS (1,400 × g, 10 min, 4°C), resuspended in RPMI medium, and placed on ice until they were used.

For opsonization, aliquots of a bacterial culture were washed, resuspended in KRG supplemented with 20% human AB serum from healthy donors, and incubated for 30 min at 37°C. After washing, the bacteria were resuspended in RPMI medium. For heat inactivation, aliquots of a bacterial culture were resuspended in PBS and incubated at 56°C for 1 h and then resuspended in RPMI medium.

Yeast cell agglutination.

Prior to experiments, type 1 fimbrial expression and/or the integrity of the FimH adhesin was confirmed by mannose-sensitive yeast cell agglutination. Strains that agglutinated heat-inactivated and homogenized Saccharomyces cerevisiae in the absence of 1% d-mannose but not in the presence of 1% d-mannose were considered type 1 fimbriated (26). The type 1 fimbria-negative E. coli mutant CN1016 showed no agglutination.

Neutrophil-bacterium interaction.

Freshly isolated neutrophils in RPMI medium were plated in a 24-well tissue culture plate and prewarmed for 10 to 15 min at 37°C in a humidified CO₂ incubator (5% CO₂, 95% air), and subsequently they were stimulated with prewarmed bacteria or other stimuli at the concentrations indicated below. For the apoptosis assay, infection was stopped after 60 min by adding gentamicin (50 μg/ml), and incubation was continued for an additional 7 or 19 h at 37°C in 5% CO₂. At different times (see below), the cells were washed with PBS and used for further analysis.

To investigate the nature of the interaction between fimbriated or nonfimbriated E. coli and neutrophils, the bacteria were labeled with FITC, and the interaction was analyzed by both microscopy and flow cytometry (19). Bacterial aliquots were resuspended in an FITC solution (0.1 mg of FITC/ml in 0.1 M carbonate buffer, pH 9.6), incubated for 1 h at room temperature with rotation, washed, and then resuspended in RPMI medium. Where indicated below, the bacteria were subsequently opsonized with serum.

For microscopy analysis, neutrophils (1 × 10⁵ cells) were allowed to adhere to glass slides for 15 min at 37°C in a moist chamber. Nonadherent neutrophils were removed with warm KRG, and FITC-labeled bacteria were added at a ratio of 40 bacteria per neutrophil for 1 and 2 h at 37°C. The glass slides were then washed briefly in cold KRG and analyzed after addition of ethidium bromide (EtBr). A fluorescence microscope (Axioscope; Zeiss, Oberkochen, Germany) was used to count green intracellular and red extracellular bacteria. Experiments were performed in triplicate, and at least 200 cells per experiment were evaluated.

Flow cytometry analysis was used to assay the bacterial interaction with cells in suspension. This was done as follows. A total of 1 × 10⁶ cells were mixed with FITC-labeled bacteria at a ratio of 40 bacteria per neutrophil in RPMI medium and incubated at 37°C for 1 h. Reactions were terminated by adding cold KRG, and unbound bacteria were washed off by centrifugation with cold KRG (300 × g, 4°C, 7 min). The binding or phagocytosis of FITC-labeled bacteria (FL1) was measured by flow cytometry (FACS-Calibur; BD Biosciences) and was analyzed by using the CellQuest software. EtBr (50 μg/ml) was added to distinguish between intra- and extracellular bacteria. At least 10,000 cells were counted in each sample.

Assessment of neutrophil apoptosis.

As an indicator of early neutrophil apoptosis, we analyzed the exposure of phosphatidylserine on the surface of apoptotic cells by staining with FITC-conjugated annexin V according to the protocol of the manufacturer (R&D Systems). Neutrophils were washed once in binding buffer (150 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 10 mM HEPES [pH 7.4]), and specific binding of annexin V was achieved by incubating 10⁶ neutrophils in 60 μl of binding buffer containing a saturating concentration of FITC-annexin V for 15 min at 4°C in the dark. To discriminate between early apoptosis and necrosis, the cells were simultaneously stained with annexin V and propidium iodide (PI) before analysis. The binding of FITC-annexin V (FL1) and PI (FL2) to the cells was measured by fluorescence-activated cell sorting and was analyzed by using the CellQuest software program (40). Cells with increased FITC fluorescence (FITC positive and PI negative), corresponding to increased exposure of phosphatidylserine, were considered early apoptotic cells, while cells with increases in both FITC and PI fluorescence were considered late apoptotic or necrotic cells (50). At least 10,000 cells were counted in each sample, and a gate based on forward and side scatter was set to exclude cell debris. Trypan blue dye exclusion was performed in parallel to further distinguish viable or apoptotic cells from necrotic cells. Cells with an intact cell membrane did not take up the dye, whereas necrotic cells became blue.

Neutrophil respiratory burst activity.

The respiratory burst in neutrophils was measured by a luminol-amplified chemiluminescence (CL) assay with a six-channel LB9505 Bioluminat (Berthold Co., Wiblad, Germany) by using disposable 4-ml polypropene tubes. Neutrophils (1 × 10⁶ cells), luminol (20 μM), and horseradish peroxidase (4 U/ml) in KRG were prewarmed for 10 min at 37°C, and then the light emission was recorded continuously. After a baseline was established, prewarmed bacteria were added at different bacterium-to-cell ratios. To investigate the specific type 1 fimbria effect, bacteria were mixed with d-mannose (25 mM). To distinguish between extra- and intracellular generation of ROS, neutrophils were preincubated with superoxide dismutase (400 U/ml) and catalase (4,000 U/ml). These scavengers removed extracellular superoxide anion and hydrogen peroxide, respectively.

LPS detection.

To analyze if LPS contaminated the system and to estimate the bacterial LPS concentrations in supernatants of bacteria and neutrophils incubated together, a Limulus amebocyte lysate endotoxin assay kit was used. The samples were collected and centrifuged (5,000 × g, 10 min), and the LPS in the supernatants was determined as recommended by the manufacturer.

Statistical analysis.

All data are expressed below as means ± standard errors of the means. The significance of differences was analyzed by the Student t test (paired), unless indicated otherwise.

RESULTS

Adherence of type 1 fimbriated E. coli to human neutrophils.

To elucidate the nature of the interaction between neutrophils and type 1 fimbriated E. coli, we used both microscopy and flow cytometry to determine the binding and phagocytosis of FITC-labeled bacteria with either neutrophils adhering to glass slides or neutrophils kept in suspension. EtBr exclusion made it possible to distinguish between intracellular and extracellular bacteria. Microscopic analysis revealed that the type 1 fimbria-negative mutant (E. coli CN1016) did not bind to neutrophils, whereas both the wild type (E. coli 1177) and the add-back mutant (E. coli CN1018) bound to 70 to 90% of the neutrophils after 1 h of incubation (Fig. 1A). However, virtually all bacteria remained extracellular, and only 1 to 2% of the neutrophils ingested bacteria after incubation with E. coli 1177 and CN1018. As a control, about 80% of the neutrophils ingested opsonized CN1016 bacteria. Similar results were obtained when neutrophils were incubated with the bacteria for 2 h.

FIG. 1.

Type 1 fimbriated E. coli interacts with human neutrophils through adherence. E. coli strains 1177 (fimbriated wild type), CN1016 (FimH mutant), and CN1018 (add-back mutant) were FITC labeled and serum opsonized (ops-1016) as a positive phagocytic control. (A) Bacteria incubated for 1 h at 37°C with neutrophils on glass slides (bacterium-to-neutrophil ratio, 40:1) were analyzed by fluorescence microscopy. By using EtBr, the levels of adherence (red extracellular bacteria) and phagocytosis (green intracellular bacteria) were determined by determining the percentage of neutrophils interacting with one or more bacteria. The results are expressed as means ± standard errors of the means for three separate experiments in which 200 neutrophils were evaluated. (B) Bacteria and neutrophils (bacterium-to-neutrophil ratio, 40:1) incubated for 1 h at 37°C in suspension were analyzed with (dashed line) or without (solid line) EtBr by using flow cytometry. FL1-H detected an increase in green fluorescence emitted from neutrophils with adherent or ingested bacteria. The histograms are the representative fluorescence profiles for three separate experiments.

Flow cytometry analysis with cells in suspension verified the microscopic data obtained with adherent neutrophils. Type 1 fimbriated bacteria adhered to neutrophils but were not ingested since the increased green fluorescence was totally quenched with EtBr. Neutrophils incubated with the type 1 fimbria-negative mutant bound no bacteria (Fig. 1B).

Type 1 fimbriae stimulate apoptosis in human neutrophils.

To determine whether type 1 fimbriated E. coli regulates apoptosis in human neutrophils, cells were exposed to either a type 1 fimbriated strain or a type 1 fimbria-negative mutant for 7 h. Apoptosis was determined by detecting phosphatidylserine exposure with annexin V by using PI to counterstain late apoptotic or necrotic cells. We found that type 1 fimbriated E. coli had dual effects on apoptosis (i.e., antiapoptotic effects at low infection rates and proapoptotic effects with increasing infection rates), whereas the type 1 fimbria-negative mutant had antiapoptotic effects at all infection rates (Fig. 2A). When trypan blue exclusion was used, 98% ± 0.37% of the cells incubated with medium and 97% ± 0.58% of the cells incubated with CN1018 for 7 h retained an integrated cell membrane (i.e., resisted trypan blue staining) (means ± standard errors of the means; n = 5). The proapoptotic effects of CN1018 were strictly mediated by type 1 fimbriae because (i) blocking the type 1 fimbrial adhesin FimH with 25 mM d-mannose prior to stimulation not only abolished the proapoptotic effect of CN1018 but also decreased the spontaneous apoptosis of neutrophils (P < 0.01) (i.e., an antiapoptotic effect) and (ii) both live and heat-killed bacteria stimulated neutrophil apoptosis (i.e., structural components are determinants for the proapoptotic effects of CN1018) (Fig. 2B).

FIG. 2.

Type 1 fimbria-specific induction of human neutrophil apoptosis. Neutrophils infected with E. coli strains CN1016 (FimH mutant) and CN1018 (fimbriated add-back mutant) for 60 min at 37°C at different ratios of bacteria to neutrophils were cultured for 7 h in the presence of gentamicin. The graphs show apoptosis expressed as a percentage of the phosphatidylserine-positive cells determined by flow cytometric analysis of annexin V binding after 7 h of incubation. (A) Titration experiment indicating that there was dose-dependent increased apoptosis with the fimbriated E. coli. The data are means ± standard errors of the means for three separate experiments. (B) Heat-inactivated bacteria were washed with PBS prior to use. The results are expressed as means ± standard errors of the means for three separate experiments. An asterisk indicates that the value is significantly different from the control value (P < 0.01). Only 1 to 3% of the cells (treated with or without CN1018) were stained with trypan blue at the end of experiments, showing that the treatments did not cause cell necrosis (n = 3).

Intracellular production of ROS by type 1 fimbriated E. coli.

When luminol-amplified CL was used to measure ROS production, we found that the type 1 fimbria-expressing E. coli induced significant and sustained ROS production in neutrophils, with the peak reached after 40 to 50 min, whereas stimulation with the type 1 fimbria-negative mutant induced no significant increase in ROS production. The serum-opsonized mutant, however, gave rise to strong and more rapid ROS production than unopsonized wild-type bacteria (Fig. 3A). d-Mannose completely abrogated wild-type-induced production of ROS, indicating that the type 1 fimbriae are essential for triggering ROS production in neutrophils. However, d-mannose did not change ROS production induced by the serum-opsonized mutant (Fig. 3B), showing that there were different mechanisms for triggering ROS production in type 1 fimbria-expressing bacteria and the serum-opsonized mutant (i.e., the type 1 fimbrial adhesin FimH and phagocytosis, respectively). To further elucidate if ROS is generated intracellularly or released extracellularly, the neutrophils were preincubated with the extracellular scavengers superoxide dismutase and catalase prior to stimulation. These scavengers did not reduce the CL significantly, indicating that ROS was generated intracellularly. The CL values as total counts during 160 min of infection with and without scavengers were as follows: for CN1016, 13.6 × 10⁷ ± 2.3 × 10⁷ and 18.4 × 10⁷ ± 6.7× 10⁷, respectively; and for 1177, 248.2 × 10⁷ ± 9.6× 10⁷ and 291.7× 10⁷ ± 30.3× 10⁷, respectively (means ± standard errors of the means; n = 6).

FIG. 3.

Production of ROS in E. coli-stimulated neutrophils is dependent on type 1 fimbriae. E. coli strains 1177 (fimbriated wild type), CN1016 (FimH mutant), and CN1018 (add-back mutant) were used at an infection rate of 20:1 (ratio of bacteria to neutrophils), and the light emission was recorded as counts per minute with a luminol-amplified CL system. (A) Kinetic curves for E. coli-induced ROS production, showing that wild-type bacteria and the serum-opsonized mutant (ops-1016) had most of their effects within 60 min after infection (the arrow indicates the time of infection). The data are representative data for five independent experiments. (B) CL experiment with (striped bars) or without (open bars) 25 mM d-mannose, showing type 1 fimbria-specific ROS production and mannose-insensitive ROS production by the opsonized mutant, expressed as the total ROS production during 160 min (means ± standard errors of the means for five independent experiments). An asterisk indicates that the value is significantly different from the value for CN1016 and control cells (P < 0.01); a number sign indicates that the value is significantly different from the value for an untreated preparation (P < 0.01).

Type 1 fimbria-triggered apoptosis is ROS dependent.

ROS has been implicated as a common mediator of apoptosis in a variety of cells, including neutrophils (31). We used DPI, a flavoprotein inhibitor of NADPH oxidase (9), to investigate if ROS is involved in type 1 fimbria-induced apoptosis in neutrophils. DPI did not affect spontaneous apoptosis after 7 h (Fig. 4A), whereas it significantly reduced the type 1 fimbria-induced apoptosis (P < 0.01). Compared with the unopsonized mutant, the serum-opsonized mutant that triggered extensive ROS production caused a significant but modest increase in apoptosis that was sensitive to DPI (P < 0.01) (the values were 14.25 ± 4.15 and 3.92 ± 0.44 for the opsonized mutant and the opsonized mutant with DPI, respectively [means ± standard errors of the means]). DPI lowered the apoptotic rate induced by the opsonized mutant to a level similar to that for the unopsonized mutant, indicating that phagocytosis without ROS production is not enough for triggering apoptosis. When we plotted the ROS production induced by type 1 fimbriated strain CN1018 at different infection ratios (bacterium/neutrophil ratios of 1:1 to 40:1) versus the level of apoptosis, there was a correlation between ROS production and apoptosis induced by type 1 fimbriated E. coli (Fig. 4B).

FIG. 4.

Induction of neutrophil apoptosis by type 1 fimbriated E. coli is ROS dependent. Neutrophils that were pretreated with (dashed bars) or without (open bars) the NADPH oxidase inhibitor DPI (5 μM) for 15 min at 37°C were infected with E. coli strains CN1016 (FimH mutant) and CN1018 (fimbriated add-back mutant) at an infection rate of 20:1 (ratio of bacteria to neutrophils) and then cultured in the presence of gentamicin for 7 h (A). The results indicate levels of apoptosis, as determined by flow cytometric analysis of annexin V binding, and are expressed as means ± standard errors of the means for four separate experiments. One asterisk indicates that the value is significantly different from the control value (P < 0.01); two asterisks indicate that the value is significantly different from the value for CN1018 without an inhibitor (P < 0.01); one number sign indicates that the value is significantly different from the value for unopsonized CN1016; and two number signs indicate that the value is significantly different from the value for serum-opsonized CN1016 (ops-1016) without an inhibitor. (B) Percentage of apoptotic neutrophils after 7 h of incubation with CN1018 (see Fig. 2A) plotted versus the peak value of luminol-amplified CL in 60 min of infection induced by heat-killed CN1018 at different multiplicities of infection, ranging from 1 to 40 bacteria per neutrophil (the level of ROS production when the multiplicity of infection was 40 was defined as 100%). The CL results are means for four separate experiments; the Pearson correlation between parameters was P = 0.007.

Antiapoptotic effects of E. coli are mediated by LPS.

The results obtained with DPI and the low infection rate showed that type 1 fimbriated E. coli also has antiapoptotic properties. Since proinflammatory agents, such as bacterial LPS, can prolong neutrophil survival (28), we hypothesized that the type 1 fimbria-negative mutant exerted its antiapoptotic effect through LPS. Both the type 1 fimbria-negative mutant and soluble LPS had an antiapoptotic effect on neutrophils after 7 and 19 h of incubation (as shown for 19 h in Fig. 5A).

FIG. 5.

Nonfimbriated E. coli induces an LPS-mediated antiapoptotic effect on human neutrophils. Apoptosis was determined by flow cytometric analysis of annexin V binding after 19 h of incubation with CN1016 (FimH mutant) at a bacterium-to-neutrophil ratio of 1:1 or with 10 to 100 ng of purified LPS/ml, unless indicated otherwise. For panel B, bacterial LPS were neutralized (striped bars) or were not neutralized (open bars) with polymyxin B (final concentration, 1 μg/ml) 1 h before stimulation. In both panel A and panel B the results are expressed as means ± standard errors of the means for four separate experiments. An asterisk indicates that the value is significantly different from the control (cont.) value (P < 0.01), and a number sign indicates that the value is significantly different from the value for untreated CN1016 (P < 0.01). (C) Effects of serum depletion with RPMI medium (open bars) or RPMI medium without 10% FCS (striped bars), expressed as means ± standard errors of the means for three separate experiments. An asterisk indicates that the value is significantly different from the control value (P < 0.01), and a number sign indicates that the value is significantly different from the value for stimulation in the presence of FCS (P < 0.01).

To find out whether the mutant CN1016 bacteria could block apoptotic signals from other stimuli over a shorter time period, the neutrophils were incubated with 100 ng of TNF-α/ml (32) or the phorbol ester PMA (10⁻⁸ M) (31) for 2 h to induce apoptosis. The results showed that pretreatment for 30 min with mutant CN1016 bacteria suppressed TNF-α- and PMA-induced apoptosis by 42 and 27%, respectively (P ≤ 0.01 for both stimuli compared to no pretreatment with CN1016; n = 3), whereas the proapoptotic effect induced by the type 1 fimbria-bearing E. coli CN1018 strain was not affected by pretreatment with CN1016 bacteria (there was a 2% decrease in proapoptotic capacity; n = 3).

To verify that the antiapoptotic effect induced by the type 1 fimbria-negative mutant was mediated by LPS, the mutant was preincubated with polymyxin B. Polymyxin B is a cyclic decapeptide antibiotic that inhibits biological activities of LPS by high-affinity binding to the lipid A moiety (36). Spontaneous apoptosis was not affected by polymyxin B, giving a clear indication that the system was free of endotoxins. The type 1 fimbria-negative mutant with neutralized LPS lost its antiapoptotic capacity, suggesting that the antiapoptotic effect was exerted solely by bacterium-associated or released LPS (Fig. 5B). The LPS concentrations in supernatants of polymorphonuclear neutrophils incubated with the type 1 fimbria mutant CN1016 (at an infection rate of 1:1) or incubated with RPMI medium alone for 19 h were 218 ± 29.48 and 0.18 ± 0.015 ng/ml, respectively (means ± standard errors of the means; n = 3).

Previous studies have shown that LPS signaling requires soluble factors, such as LBP and CD14 (55). To investigate if this was the case, neutrophil apoptosis was also analyzed in a serum-free system, RPMI medium without FCS. Mutant CN1016 at both low and high infection rates (bacterium/neutrophil ratios of 1:1 and 20:1) induced the same antiapoptotic effect in the presence of serum, whereas only the high number of bacteria had an equivalent effect in the absence of serum. The antiapoptotic effect mediated by purified LPS was completely inhibited in serum-free medium (Fig. 5C).

To further investigate in what form LPS from CN1016 cells mediate the antiapoptotic effect in a serum-free system, an experiment with the high concentration (20:1, as shown in Fig. 5, but without neutrophils) was performed for 19 h. The bacteria and bacterium-derived constituents were isolated by differential centrifugation and tested with freshly isolated neutrophils. The bacterial cells obtained by centrifugation at 5,000 × g had no effect on neutrophil apoptosis, whereas the supernatant did have an effect (there was a 30% decrease in the spontaneous apoptosis). The supernatant from the 5,000-×-g centrifugation was further separated into a soluble fraction (not pelleted by centrifugation at 10,000 × g for 30 min at 4°C) and a particulate fraction containing membrane-LPS complexes (pellet), both of which mediated an antiapoptotic effect. We found that the soluble fraction had a strong antiapoptotic effect at a low dosage (28% decrease in the spontaneous apoptosis; P ≤ 0.01; n = 3), whereas the reconstituted particulate fraction had a stricter dose-dependent effect (27% decrease in the spontaneous apoptosis; P ≤ 0.01; n = 3). The soluble fraction was most likely to contain LPS and outer membrane-derived vesicles that gram-negative bacteria usually release (5), since centrifugation at 150,000 × g at 4°C for 3 h was used to pellet outer membrane vesicles in another study (51).

Proapoptotic effect of type 1 fimbriae requires LPS.

Since a small amount of E. coli releases considerable levels of LPS, the antiapoptotic effect seen with low infection rates and mannose treatment (Fig. 2) suggested that LPS could suppress the proapoptotic effect of the type 1 fimbriae. Polymyxin B-treated bacteria were used to study this balance between pro- and antiapoptotic signaling by the fimbriated bacteria and released LPS, respectively. Polymyxin B alone did not affect spontaneous apoptosis, but bacteria with neutralized LPS lost the capacity to trigger generation of ROS and induction of apoptosis in neutrophils (P < 0.01 for a comparison with untreated CN1018) (Fig. 6). It should be noted that polymyxin B-treated CN1018 still was able to adhere to neutrophils and agglutinate yeast cells in a mannose-sensitive manner to the same extent as heat-inactivated and untreated CN1018. Moreover, polymyxin B-treated neutrophils responded normally to fMLP-stimulated ROS production (Fig. 6B), and LPS alone was not sufficient to induce a respiratory burst even at microgram levels (data not shown). These results indicate that LPS served as a cofactor that was essential for type 1 fimbria-triggered ROS production and subsequent apoptosis in neutrophils.

FIG. 6.

Type 1 fimbria-triggered apoptosis and ROS production in human neutrophils require LPS. LPS on E. coli CN1018 (type 1 fimbriated add-back mutant) was neutralized for 1 h at room temperature with polymyxin B (PB) prior to stimulation. The final concentration of PB was 1 μg/ml, and the infection rate used was 20:1 (ratio of bacteria to neutrophils). (A) Apoptosis of neutrophils stimulated with E. coli CN1018 for 60 min at 37°C and additionally cultured for 7 h was determined by flow cytometric analysis of annexin V. The results are expressed as means ± standard errors of the means for three separate experiments. An asterisk indicates that the value is significantly different from the control value (P < 0.01), and a number sign indicates that the value is significantly different from the value for untreated CN1018 (P < 0.01). (B) ROS production from neutrophils stimulated with E. coli CN1018 or polymyxin B-treated CN1018 (PB-CN1018) was measured with a luminol-amplified CL system, and the results were recorded as light emission in counts per minute. The left arrow indicates the time of bacterial stimulation. The right arrow indicates the time of secondary stimulation (10⁻⁷ M fMLP) of neutrophils stimulated with polymyxin B-treated CN1018. The data are representative of the data from four independent experiments.

Priming effects of LPS.

LPS has a priming effect on fMLP-stimulated ROS production in neutrophils (55). We therefore analyzed whether LPS priming could affect the fimbria-mediated ROS production and subsequent apoptosis. Pretreatment of neutrophils with LPS (5 ng/ml) induced stronger ROS production upon stimulation with fMLP. However, it did not affect the magnitude and duration of ROS production in neutrophils stimulated with CN1018, although faster kinetics were observed in LPS-primed cells (Fig. 7). These results indicate that LPS priming cannot protect neutrophils from type 1 fimbria-induced apoptosis.

FIG. 7.

LPS priming induces faster ROS production in response to type 1 fimbriated E. coli. Neutrophils were pretreated with or without 5 ng of LPS/ml for 30 min at 37°C and stimulated with E. coli CN1018 (type 1 fimbriated add-back mutant) at an infection rate of 20:1 (ratio of bacteria to neutrophils) or fMLP (10⁻⁷ M), and the ROS production was measured with a luminol-amplified CL system and expressed as light emission in counts per minute. The arrow indicates the time of infection. The data are representative of the data from four independent experiments.

DISCUSSION

Type 1 fimbriated E. coli has a unique way of colonizing the host by adhering to the tissue. UTI isolates have a selective advantage by targeting receptors with high monomannose-binding specificity present in the urinary tract (47). Such binding provokes an immediate cytokine release, followed by polymorphonuclear neutrophil recruitment (16). The present study showed that binding of type 1 fimbriated E. coli induced apoptosis in human neutrophils without ingestion, whereas an isogenic type 1 fimbria-negative mutant did not induce apoptosis. The type 1 fimbrial adhesin FimH was the sole determinant for neutrophil adherence, whereas both the adhesin and LPS were required for apoptosis and ROS production. In contrast to spontaneous apoptosis, type 1 fimbria-induced apoptosis was strictly dependent on, and correlated to, intracellular generation of ROS. Moreover, in the absence of type 1 fimbriae E. coli mediated an antiapoptotic response.

It has been shown that E. coli can modulate apoptosis in human neutrophils (29, 35, 52). For example, verotoxin 2 produced by enterohemorrhagic E. coli (O157:H7), which is implicated in hemorrhagic colitis and hemolytic-uremic syndrome, modulates neutrophil apoptosis in an LPS-independent and heat-sensitive way (29). The present study showed that expression of type 1 fimbriae on uropathogenic E. coli, but not soluble factors or injected effector proteins or toxins released from the bacteria, is essential for the induction of neutrophil apoptosis. This conclusion is based on the following observations: (i) a type 1 fimbria mutant lacking the adhesive properties did not bind to or induce neutrophil apoptosis, whereas the add-back mutant had the proapoptotic phenotype of the wild-type bacteria; (ii) blocking the FimH adhesin with d-mannose completely abrogated the induced neutrophil apoptosis; and (iii) heat-inactivated bacteria had apoptotic effects similar to those observed with live bacteria. Because apoptotic cells are associated with the loss of bactericidal activities, these results suggest that the type 1 fimbria is a critical virulence factor that could help the bacteria evade host defenses. In this context, other workers have shown that the type 1 fimbria triggers the internalization of bacteria into bladder cells (34), subsequently causing apoptosis and rapid shedding of the infected bladder epithelial lining (37).

Type 1 fimbria-induced neutrophil apoptosis is an ROS-dependent process. A correlation between ROS and neutrophil apoptosis is in agreement with previous findings, which showed that Mycobacterium tuberculosis induced neutrophil apoptosis in an ROS-dependent manner (40), and with the observation that neutrophils from patients with chronic granulomatous disease having an inherited dysfunction in the ROS-producing enzyme NADPH oxidase showed decreases in both spontaneous and Fas-induced apoptosis (22). M. tuberculosis-induced NADPH oxidase activity triggered caspase 3 activity, as well as a shift in the Bax/Bcl-x_L ratio, ultimately leading to apoptosis. These results and other results support a role for mitochondria in cell death (57), where ROS-generating agents can peroxidate membrane lipids, leading to activation and release of mitochondrial apoptotic proteins, such as cytochrome c (56). Moreover, ingestion of opsonized bacteria triggers extensive ROS production without inducing apoptosis, whereas type 1 fimbria-triggered ROS production induces apoptosis. This suggests that the location of ROS production is vital for the induction of neutrophil apoptosis. Intraphagosomal ROS produced during phagocytosis could be less harmful than ROS produced outside such a well-defined compartment, as is the case for the adherent type 1 fimbriated bacterium. Previous experiments with macrophages showed that receptor-mediated phagocytosis protects cells from apoptosis by activating the serine/threonine kinase Akt (10). However, whether such a mechanism is also involved during receptor-mediated phagocytosis in neutrophils has not been determined yet.

The dominant intracellular generation of ROS induced by the 1177 strain differs from previous observations made in our laboratory (30) and by other workers (49). When type 1 fimbriated E. coli strains are compared, other underlying physicochemical surface properties (LPS, K antigen) affect the interaction (39). Therefore, different strains and FimH-coated microspheres may elicit different responses (49). Furthermore, mobilization of receptors may enhance the extracellular ROS production (21), suggesting that not only bacterial properties but also the activation state of the phagocyte determines the response.

In neutrophils, the type 1 fimbriae were required for adherence and induction of apoptosis. However, inactivating the biological activity of LPS by polymyxin B treatment abolished the proapoptotic effects of CN1018, indicating that FimH-mediated adhesion alone was not sufficient for a type 1 fimbria-mediated response. Polymyxin B-treated fimbriated bacteria still mediated adherence to neutrophils but did not induce any ROS production or apoptosis. Involvement of LPS in the type 1 fimbria-mediated interaction has been reported previously (18). Using a CD14-negative uroepithelial cell line, Hedlund et al. showed that type 1 fimbriae deliver a Toll-like receptor 4-dependent LPS signal to the mucosa, thereby activating a cytokine response (18). Taken together, these data, in contrast to data for isolated FimH (49), show that in bacteria both type 1 fimbriae and LPS are required for presentation and subsequent neutrophil activation, leading to ROS and apoptosis.

In addition, for E. coli to trigger neutrophil apoptosis (this study), it has been shown that purified LPS from E. coli and other gram-negative bacteria can inhibit or delay apoptosis through the activation of tyrosine kinases (48) and the mitogen extracellular signal-regulated kinase ERK (24). By using the type 1 fimbria-negative mutant lacking type 1 fimbriae, we can now mimic the antiapoptotic effect of purified LPS with both high and low concentrations of whole bacteria. We therefore propose that the modulation of neutrophil apoptosis is not always a matter of the bacterium-to-cell ratio, as previously proposed (35) or as shown for type 1 fimbriated E. coli. The dose-independent inhibition of neutrophil apoptosis was augmented by LBP, since serum depletion reduced the antiapoptotic effect of both bacterial and purified LPS. Furthermore, polymyxin B completely abolished the antiapoptotic effect of the type 1 fimbria-negative mutant. The observation that a high concentration of bacteria did not require serum to induce antiapoptosis, in contrast to purified LPS, showed that purified LPS and LPS released from E. coli differ in their dependence on cofactors. Furthermore, as determined with bacterium-derived fractions, our analysis suggests that LPS released from gram-negative bacteria in various forms, such as outer membrane vesicles or larger membrane fractions, has antiapoptotic capacity. Either these complexes released from bacteria could contain components that serve as cofactors or the complexes themselves could serve as vehicles presenting the LPS to the cell in a form that triggers a biological response. Since the type 1 fimbria-negative mutant inhibits apoptosis at very low infection rates and no cell interactions were seen, the data further suggest that E. coli exerts its antiapoptotic effect by releasing LPS. Furthermore, when the binding of fimbriated bacteria is inhibited with d-mannose, the type 1 fimbriae can no longer position the bacteria in a way in which the LPS can be properly presented, which suggests that the antiapoptotic effect seen in this case is also the effect of released LPS.

In conclusion, we show that type 1 fimbriae and LPS operate together to induce apoptosis in human neutrophils. The cooperative effects of these virulence attributes may function as a mechanism by which E. coli induces UTI. In contrast, LPS released from nonadherent bacteria has an antiapoptotic effect, suggesting that LPS can also serve as an important regulator of neutrophil survival in tissue.