Enterohemorrhagic Escherichia coli O157:H7 Disrupts Stat1-Mediated Gamma Interferon Signal Transduction in Epithelial Cells

Abstract

Enterohemorrhagic Escherichia coli (EHEC) O157:H7 is a clinically important bacterial enteropathogen that manipulates a variety of host cell signal transduction cascades to establish infection. However, the effect of EHEC O157:H7 on Jak/Stat signaling is unknown. To define the effect of EHEC infection on epithelial gamma interferon (IFN-γ)-Stat1 signaling, human T84 and HEp-2 epithelial cells were infected with EHEC O157:H7 and then stimulated with recombinant human IFN-γ. Cells were also infected with different EHEC strains, heat-killed EHEC, enteropathogenic E. coli (EPEC) O127:H6, and the commensal strain E. coli HB101. Nuclear and whole-cell protein extracts were prepared and were assayed by an electrophoretic mobility shift assay (EMSA) and by Western blotting, respectively. Cells were also processed for immunofluorescence to detect the subcellular localization of Stat1. The EMSA revealed inducible, but not constitutive, Stat1 activation upon IFN-γ treatment of both cell lines. The EMSA also showed that 6 h of EHEC O157:H7 infection, but not 30 min of EHEC O157:H7 infection, prevented subsequent Stat1 DNA binding induced by IFN-γ, whereas infection with EPEC did not. Immunoblotting showed that infection with EHEC, but not infection with EPEC, eliminated IFN-γ-induced Stat1 tyrosine phosphorylation in both dose- and time-dependent fashions and disrupted inducible protein expression of the Stat1-dependent gene interferon regulatory factor 1. Immunofluorescence revealed that EHEC infection did not prevent nuclear accumulation of Stat1 after IFN-γ treatment. Also, Stat1 tyrosine phosphorylation was suppressed by different EHEC isolates, including intimin-, type III secretion- and plasmid-deficient strains, but not by HB101 and heat-killed EHEC. These findings indicate the novel disruption of host cell signaling caused by EHEC infection but not by EPEC infection.

Infection of humans with the bacterial enteropathogen enterohemorrhagic Escherichia coli (EHEC) serotype O157:H7 causes outbreaks of hemorrhagic colitis and hemolytic uremic syndrome; the latter is related to production of Shiga-like toxins (5) and is an important cause of acute kidney failure (5, 54). The related pathogen enteropathogenic E. coli (EPEC) is an important cause of prolonged watery diarrhea in infants in underdeveloped countries (5, 11, 51). Once ingested, these bacteria establish infection by manipulating host epithelial cell signal transduction cascades (5, 11, 51). For example, activation of phospholipase Cγ, phosphoinositide 3-kinase, and 5-lipoxygenase (20) and mobilization of the second messengers inositol trisphosphate and calcium (6) are important for forming characteristic attaching and effacing lesions and rearrangements of the host cell cytoskeleton. However, these bacteria disrupt intestinal epithelial barrier function by manipulation of differential signal transduction cascades; EHEC does this via protein kinase C and the myosin light-chain kinase (38), and EPEC does it via a myosin light-chain kinase-dependent, protein kinase C-independent mechanism (55). Moreover, while EPEC infection causes tyrosine phosphorylation of the translocated intimin receptor (4), EHEC infection does not (17, 23), an observation that emphasizes that these infections are established by different mechanisms.

Multiple cytokines transduce intracellular effects on target cells via the Janus kinase-signal transducer and activator of transcription (Jak/Stat) signal transduction cascade (16). Gamma interferon (IFN-γ) is a cytokine that is critical to host defense (41), and it is produced in murine models of EPEC infection (13) following infection of mice with the attaching-effacing pathogen Citrobacter rodentium (13, 14) and in humans infected with EPEC (47). Indeed, IFN-γ knockout mice infected with C. rodentium carry a higher bacterial load and exhibit greater susceptibility to mucosal and gut-derived systemic infections than wild-type littermates (42). Following stimulation of cell surface IFN-γ receptor chains 1 and 2, the associated Jak1 and Jak2 phosphorylate specific tyrosine residues on the cytoplasmic tails of the receptor that serve as docking sites for the normally latent cytoplasmic Stat1 protein. Subsequent tyrosine phosphorylation of Stat1 by Jak2 allows Stat1 to dimerize, translocate to the nucleus, and modulate transcription by binding specific DNA sequences in promoters of responsive genes (39).

Microbes can interfere with cytokine signaling (53). This disruption occurs as a result of cytokine degradation, decreased cytokine release, cytokine receptor downregulation, and disruption of cytokine second messengers (53). Recent studies have begun to reveal the molecular mechanisms involved in these host responses to infection. For example, lipopolysaccharide activates suppressor of cytokine signaling 3 (SOCS3) in macrophages (45), a member of a family of inducible proteins that negatively regulates Jak/Stat signaling (27). Also, infection of macrophages with Ehrlichia chafeensis activates protein kinase A (29), and infection with Listeria monocytogenes activates p38 mitogen-activated protein kinase-dependent expression of SOCS3 to disrupt IFN-γ signaling (46).

The aim of the present study was to determine if EHEC O157:H7 infection disrupts IFN-γ signal transduction in human epithelial cells by defining the effects of infection on Stat1 DNA binding, tyrosine phosphorylation, and subcellular localization and inducibility of a Stat1-dependent gene. Our results revealed a signal transduction pathway that is important to host defenses which is compromised at the level of Stat1 activation by EHEC infection but not by EPEC infection.

MATERIALS AND METHODS

Eukaryotic cell culture.

The transformed human laryngeal epithelial cell line HEp-2 (= ATCC CCL-23) was cultured in minimum essential medium supplemented with 15% fetal bovine serum (FBS), 2.5% penicillin-streptomycin, 1.8% sodium bicarbonate, and 1.2% Fungizone (all obtained from Life Technologies, Grand Island, N.Y.) (18). The transformed human colonic epithelial cell line T84 (= ATCC CCL-248; American Type Culture Collection, Manassas, Va.) was cultured in a 1:1 mixture of Dulbecco's modified Eagle medium and Ham's F-12 medium supplemented with 10% (vol/vol) FBS, 2% penicillin-streptomycin, 2% sodium bicarbonate, and 0.6% l-glutamine (2). Cells were grown at 37°C in a 5% CO₂ atmosphere. HEp-2 and T84 cells were seeded (2 × 10⁶ to 3 × 10⁶ cells) onto 6-cm petri dishes (Falcon) and grown to confluence for whole-cell or nuclear protein extraction. Prior to bacterial infection and protein extraction, cells were incubated in medium containing 1% FBS and no antibiotics for 20 h at 37°C. For Stat1 immunofluorescence experiments, 1 × 10⁵ HEp-2 and T84 cells were grown in Labtek four-well chamber slides (Miles Scientific, Naperville, Ill.).

Growth of bacteria and conditions of infection.

The E. coli strains used in this study (Table 1) were kept in 5% sheep blood agar plates at 4°C and cultured in static, nonaerated Penassay broth (Difco Laboratories, Detroit, Mich.) overnight at 37°C prior to infection of tissue culture cells (18), unless otherwise noted. Cultures of EHEC O157:H7 strain CL-8 KO1 on agar or in broth were supplemented with 150 μg of carbenicillin per ml and 50 μg of nalidixic acid per ml (31). For some experiments (see Fig. 7B), strains CL-8 and CL-8KO1 were first grown overnight in Penassay broth; then they were diluted 1:50 in 10 ml of cell culture medium and grown for an additional 3 h, and 5 ml of the resulting growth was used to infect T84 cells for 4 h. Cultures of strain CVD451 on agar and in Penassay broth were supplemented with 50 μg of kanamycin per ml (19). For infection of epithelial cells, bacteria were pelleted from broth cultures by centrifugation at 3,000 rpm (Microfuge 18; Beckman Coulter, Palo Alto, Calif.) for 5 min and washed in cell culture medium. The bacteria were resuspended in 0.025 ml (total volume) of tissue culture medium and then added to the host cells. The same volume of medium alone served as a vehicle control. In some experiments, EHEC O157:H7 strain CL-56 was heat killed by boiling for 15 min, and the absence of growth was confirmed by plating samples onto sheep blood agar plates.

TABLE 1.

E. coli strains employed in this study

Strain	Serotype	Description	Sourcea	Reference(s)
CL-56	O157:H7	Wild-type EHEC	HC, HUS	38, 40
CL-8	O157:H7	Wild-type EHEC	HC, HUS	31
CL-8 KO1	O157:H7	eaeA-negative mutant of parental strain CL-8	Lab	31
86-24	O157:H7	Wild-type EHEC	HC, HUS	19
CVD451	O157:H7	sepB-negative mutant (type III secretion deficient) of parental strain 86-24 (a kind gift from J. Kaper)	Lab	19
7785-5	O157:H7	Wild-type EHEC	HC, HUS	50
2-45	O157:H7	Plasmid pO157-cured derivative of parental strain 7785-5	Lab	50
85-289	O157:H7	Wild-type EHEC	HC, HUS	50
85-170	O157:H7	Spontaneous Stx-1- and Stx-2-negative mutant of parental strain 85-289	Lab	50
CL-15	O113:H21	Wild-type eaeA negative	HC, HUS	7
HB101	O:rough	Nonpathogenic lab strain	Commensal	38
E2348/69	O127:H6	Wild-type EPEC	Infant diarrhea	18

^aHC, hemorrhagic colitis; HUS, hemolytic-uremic syndrome.

FIG. 7.

Suppression of IFN-γ-induced Stat1 tyrosine phosphorylation is dependent on a heat-sensitive factor and is intimin independent. (A) Whole-cell protein extracts from T84 cells analyzed by immunoblotting showed that Stat1 tyrosine phosphorylation (upper panel) was inducible by IFN-γ treatment (50 ng/ml, 30 min) (lane 2). Live EHEC O157:H7 strain CL-56 prevented IFN-γ stimulation of Stat1 tyrosine phosphorylation, while heat-killed CL-56 (Hk CL-56) did not (lane 6). EHEC O113:H21 strain CL-15 (wild type, intimin deficient) also prevented subsequent activation of Stat1 tyrosine phosphorylation by IFN-γ, while infection with the commensal strain E. coli HB101 did not. Bacteria alone had no effect (n = 3). The lower panel shows the levels of nonphosphorylated Stat1. (B) EHEC O157:H7 strain CL-8 and its eaeA insertional mutant CL-8KO1 both disrupted IFN-γ-activated Stat1 tyrosine phosphorylation (n = 2 or 3).

Eukaryotic cell treatment with Stx-1 and Stx-2.

Shiga-like toxin 1 (Stx-1) and Stx-2 were kindly provided by M. A. Karmali (Health Canada, Guelph, Ontario, Canada). Confluent HEp-2 cells were incubated in medium containing 1% FBS and no antibiotics for 20 h at 37°C and then treated with Stx-1 or Stx-2 at a dose of 50 ng/ml for 6 h before IFN-γ stimulation (50 ng/ml, 30 min). Subsequently, whole-cell protein lysates were collected, stored at −70°C, and analyzed by immunoblotting.

Immunoblotting.

For whole-cell protein extraction, cell monolayers were washed three times with ice-cold phosphate-buffered saline (PBS), scraped with a rubber policeman into 1 ml of PBS, and pelleted by centrifugation at 12,000 rpm for 10 s. Subsequently, the cell pellet was resuspended in 0.15 ml of RIPA buffer (1% NP-40, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate in PBS) supplemented with 50 mM NaF, 150 mM NaCl, 1 mM Na₃VO₄, 20 μg of phenylmethylsulfonyl fluoride per ml, 15 μg of aprotinin per ml, 2 μg of pepstatin A per ml, and 2 μg of leupeptin per ml (all chemicals were obtained from Sigma Aldrich) by vortexing, sheared by passage through a 25-gauge needle three times, and left at 4°C for 30 min. The lysate was then cleared by centrifugation at 12,000 rpm (Sorval SS34 rotor; Mondell Scientific, Guelph, Ontario, Canada) for 10 min at 4°C, and the supernatant was stored at −70°C as the whole-cell protein extract (3).

An equal volume of whole-cell protein extract was then added to 2× loading buffer, boiled for 5 min, and electrophoresed through a 5% polyacrylamide stacking gel and an 8% polyacrylamide-sodium dodecyl sulfate separating gel at 111 V for 1.5 h at room temperature. Proteins were then electrophoretically transferred onto a nitrocellulose membrane (BioTrace NT; Pall Corporation, Ann Arbor, Mich.) at 4°C and 100 V for 1.25 h and then blocked with Tris-buffered saline containing 0.05% Tween 20 (TBST) supplemented with 5% low-fat milk for 30 min at room temperature. The membranes were probed overnight with shaking at 4°C with either anti-phosphotyrosine Stat1 antibody (1:1,000; Upstate Biotechnology, Lake Placid, N.Y.), anti-Stat1 p84/p91 antibody (1:5,000; E-23; Santa Cruz Biotechnology, Santa Cruz, Calif.), anti-interferon regulatory factor 1 (1:500; C-20; Santa Cruz Biotechnology), or anti-actin (1:1,000; I-19; Santa Cruz Biotechnology) in TBST-5% low-fat milk. The membranes were then washed in TBST, rinsed in distilled H₂O, probed with horseradish peroxidase-conjugated donkey anti-rabbit antibody (1:1,000 to 1:5,000; Santa Cruz Biotechnology) for 1.5 h at room temperature, washed again in TBST, and rinsed again in distilled H₂O. Bands were then visualized by chemiluminescence (Western blotting Luminol reagent; Santa Cruz Biotechnology) by using Kodak Biomax MR film.

Electrophoretic mobility shift assay (EMSA).

Nuclear protein extracts were collected by the method of Andrews and Faller (1); 15 μg of aprotinin per ml, 2 μg of pepstatin A per ml, 2 μg of leupeptin per ml, and 20 μg of phenylmethylsulfonyl fluoride per ml (all chemicals were obtained from Sigma Aldrich) were added to the extraction buffers (2, 33), and the preparations were stored at −70°C. The protein concentration of each sample was determined by the Bio-Rad assay.

EMSAs were performed as previously described (2, 33). Briefly, 15 μg of nuclear protein extract was mixed with binding buffer (250 mM Tris-Cl [pH 7.5], 40 mM NaCl, 10 mM EDTA [pH 8], 2.5 mM dithiothreitol, 10 mM spermidine, 5% autoclaved distilled H₂O, 25% glycerol) and incubated with ∼3 × 10⁵ cpm of [α-³²P]dCTP (NEN Life Science Products, Boston, Mass.)-end-labeled double-stranded oligonucleotides containing a Stat1 binding sequence (52) for 20 min at room temperature. Indicator dye (0.25% [wt/vol] bromphenol blue, 5% [wt/vol] glycerol) was added to each sample, and the samples were electrophoresed through a 5% polyacrylamide gel (ratio of acrylamide to bisacrylamide, 40:1) containing 1.25% (vol/vol) glycerol, 0.7% (wt/vol) ammonium persulfate, 0.05% (vol/vol) N,N,N′,N′-tetramethylethylenediamine, and 1.25% (vol/vol) Tris borate-EDTA (10× TBE buffer [89 mM Tris borate, 2 mM EDTA; pH 8]) at 120 V for 2 to 3 h at room temperature in 0.25× TBE buffer. The gels were then dried and visualized by autoradiography.

The specificity controls included nonradioactively labeled hSIE probe as a cold competitor (25-fold molar excess), the polyclonal p84/p91 Stat1 antibody (1 μg; Santa Cruz Biotechnology) for EMSA supershift, and both an irrelevant DNA probe (Stat6; mutant cold competitor) and an isotype-matched irrelevant antibody (Stat6).

Immunofluorescence.

HEp-2 and T84 cells were washed three times with ice-cold PBS, fixed in 4% paraformaldehyde for 30 min, permeabilized for 4 min with 0.1% Triton X-100, and blocked for 30 min in 2% bovine serum albumin (BSA)-0.1% Triton X-100 (25). Subsequently, the cells were probed for Stat1 with anti-p84/p91 Stat1 (1:100; Santa Cruz Biotechnology) in 2% BSA-0.1% Triton X-100 overnight at 4°C, washed in PBS, incubated with rhodamine-red goat anti-rabbit secondary antibody (1:100; Jackson Labs) in BSA-Triton X-100 for 1 h at room temperature, and washed in PBS. Vectashield (Vector Labs, Burlingame, Calif.) mounting medium for fluorescence was added, and slides were sealed with coverslips and examined under immunoflourescence conditions (Leitz Dialux 22; Leica Canada, Willowdale, Ontario, Canada).

RESULTS

EHEC O157:H7 infection prevents IFN-γ-induced Stat1 DNA binding in T84 and HEp-2 cells.

Bacterial infections interfere with IFN-γ-induced Stat1 DNA binding in macrophages (46). To determine whether EHEC infection interferes with Stat signaling in epithelial cells, two epithelial cell lines were employed in the present study. HEp-2 laryngeal epithelial cells were used because they are a widely employed model system for EHEC O157:H7 adhesion and modulation of signaling in host cells (34, 43) and studies of apoptosis (3, 21). In complementary studies T84 cells were used as a model polarized intestinal epithelial monolayer (37, 38). As shown in Fig. 1, there was no constitutive Stat1 DNA binding, whereas IFN-γ induced Stat1 DNA binding in HEp-2 and T84 cell lines. Supershift and elimination of the inducible band in an EMSA by incubation with a Stat1 antibody, but not by incubation with a Stat6 antibody, confirmed that the band was Stat1. Preincubation and successful competition with the nonradiolabeled Stat1 DNA oligonucleotide probe as a cold competitor, but not with a Stat6 probe as a mutant cold competitor, before incubation with the radioactive Stat1 probe showed the specificity of the radioactive probe for Stat1 (Fig. 1).

FIG. 1.

IFN-γ activates Stat1 DNA binding in T84 and HEp-2 cells. EMSA analysis of nuclear protein extracts from T84 (A) and HEp-2 (B) cells showed no constitutive Stat1 DNA binding, but there was a dose-dependent increase in Stat-1 DNA binding in response to IFN-γ stimulation (arrow). Supershift (tailless arrow) and partial elimination with a Stat1 antibody (Ab) (lane S1), but not with a Stat6 antibody (lane S6), confirmed the identity of the band as Stat1. The specificity of the radiolabeled double-stranded DNA oligonucleotide probe was confirmed by competition of the Stat1 band with a nonradiolabeled Stat1 probe (cold competitor) (lane cc) but not with a Stat6 probe (mutant cold competitor) (lane mcc). The EMSA results are representative of the results of three experiments. NS, nonspecific; FP, free probe.

After the IFN-γ-induced band was confirmed to be Stat1, HEp-2 and T84 cells were shown to respond similarly to IFN-γ stimulation with dose-dependent activation of Stat1 DNA binding (Fig. 1). EMSA further revealed that 6 h of EHEC O157:H7 infection (strain CL-56; multiplicity of infection [MOI], 100:1) (Fig. 2B), but not 30 min (Fig. 2A) of EHEC O157:H7 infection, eliminated the ability of IFN-γ to activate Stat1 DNA binding in T84 cells. Comparable results were obtained with 6 h of EHEC infection and IFN-γ stimulation of HEp-2 cells (Fig. 2C). Conversely, EPEC O127:H6 strain E2348/69 (MOI, 100:1; 30 min or 6 h of infection) was unable to disrupt IFN-γ-stimulated Stat1 DNA binding in either cell line (Fig. 2). Bacteria alone did not elicit Stat1 DNA binding.

FIG. 2.

EHEC infection, but not EPEC infection, prevents IFN-γ-stimulated Stat1 DNA binding in T84 and HEp-2 cells. Nuclear protein extracts analyzed by EMSA showed that neither T84 nor HEp-2 cells exhibited constitutive Stat1 DNA binding (arrow) (lane 1), while IFN-γ induced Stat1 DNA binding (lane 2). T84 cells infected with EHEC O157:H7 strain CL-56 for 6 h (B), but not T84 cells infected with EHEC O157:H7 strain CL-56 for 30 min (A), showed diminished IFN-γ (50 ng/ml, 30 min)-stimulated Stat1 DNA binding activity compared to cells stimulated with IFN-γ alone (compare lanes 2 and 4 in panel B). EPEC O127:H6 strain E2348/69 infection did not eliminate Stat1 DNA binding (n = 3). (C) In HEp-2 cell nuclear protein extracts, EHEC infection for 6 h (MOI, 100:1), but not EPEC infection for 6 h (MOI, 100:1), eliminated subsequent IFN-γ (50 ng/ml, 30 min)-stimulated Stat1 DNA binding activity (n = 3). NS, nonspecific.

EHEC O157:H7 infection, but not EPEC infection, prevents Stat1 tyrosine phosphorylation.

To determine the effects of EHEC infection and EPEC infection on IFN-γ-induced tyrosine phosphorylation of Stat1, which is necessary for dimerization and DNA binding (39), an immunoblot analysis of whole-cell protein extracts was performed. As shown in Fig. 3, IFN-γ (50 ng/ml, 30 min) elicited tyrosine phosphorylation of Stat1 in both T84 (Fig. 3A) and HEp-2 (Fig. 3B) cells, while bacterial infection alone had no effect. EHEC O157:H7 strain CL-56 infection eliminated the ability of IFN-γ to cause Stat1 tyrosine phosphorylation in both a dose-dependent manner (Fig. 3) and a time-dependent manner (Fig. 4). Maximum effects were observed with an EHEC O157:H7 MOI of 100:1 at 6 h postinfection in both cell lines. In contrast, the related pathogen EPEC strain E2348/69 was unable to prevent Stat1 tyrosine phosphorylation in either HEp-2 cells or T84 cells after 6 h of infection (Fig. 3 and 4) or after 8 h of infection (data not shown) in T84 cells.

FIG. 3.

Infection of T84 and HEp-2 cells by EHEC, but not infection of T84 and HEp-2 cells by EPEC, leads to dose-dependent inhibition of IFN-γ-stimulated Stat1 tyrosine phosphorylation. (A) Whole-cell protein extracts analyzed by immunoblotting showed that untreated T84 cells had no constitutive Stat1 tyrosine phosphorylation (lane 1), whereas IFN-γ treatment (50 ng/ml, 30 min) activated phosphorylation of Stat1 (in the upper panel the upper band of a doublet is the Stat alpha isoform, and the lower band is the Stat1 beta isoform) (n = 3) (lane 2). Infection with EHEC O157:H7 strain CL-56 at an MOI of 100 (6 h) completely prevented subsequent Stat1 activation by IFN-γ. An EHEC MOI of 10 partially eliminated Stat1 activation, while EPEC (MOI, 100 and 10) was unable to prevent Stat1 activation (n = 3). Bacteria alone had no effect. The lower panel shows that nonphosphorylated Stat1 protein levels among samples were similar. (B) Untreated HEp-2 cells had little constitutive Stat1 tyrosine phosphorylation (lane 1), whereas IFN-γ treatment (50 ng/ml, 30 min) activated Stat1 (lane 2). Six hours of infection with EHEC (MOI, 1) or EPEC (MOI, 100) was unable to prevent IFN-γ-induced Stat1 tyrosine phosphorylation. However, EHEC infection for 6 h at MOIs of 100 (lane 3) and 10 (lane 4) did prevent Stat1 activation (n = 2). Bacterial infection alone did not elicit Stat1 tyrosine phosphorylation (data not shown). NS, nonspecific.

FIG. 4.

Infection of T84 and HEp-2 cells by EHEC O157:H7 leads to time-dependent inhibition of IFN-γ-stimulated Stat1 tyrosine phosphorylation. (A) Whole-cell protein extracts from T84 cells analyzed by immunoblotting showed that Stat1 tyrosine phosphorylation (upper panel) was inducible upon IFN-γ treatment (50 ng/ml, 30 min) compared with nonstimulated cells. EHEC infection for 20 min followed by IFN-γ stimulation (lane 4), compared to the time-matched control containing medium plus IFN-γ (lane 3), was unable to eliminate the ability of IFN-γ to cause Stat1 tyrosine phosphorylation. However, infection for 2 and 5 h did prevent Stat1 activation (n = 3). EHEC infection alone did not elicit Stat1 tyrosine phosphorylation (data not shown). The lower panel shows the protein loading levels of nonphosphorylated Stat1. (B) Untreated HEp-2 cells showed little constitutive Stat1 tyrosine phosphorylation (upper panel, lane 1), whereas IFN-γ treatment (50 ng/ml, 30 min) activated Stat1 (lane 2). Infection with EHEC (MOI, 100) for 30 min or 1 or 3 h did not prevent subsequent IFN-γ-induced Stat1 tyrosine phosphorylation, whereas infection for 6 h did prevent such phosphorylation (n = 2).

EHEC O157:H7 infection does not prevent IFN-γ-induced nuclear accumulation of Stat1.

Immunofluorescence analysis of HEp-2 cells was employed to determine the subcellular localization of Stat1 after EHEC infection, because tyrosine phosphorylation of Stat1 generally leads to its nuclear localization, followed by binding to DNA (39). In uninfected HEp-2 cells Stat1 was distributed throughout the cytoplasm (Fig. 5A), but it translocated to the nucleus upon IFN-γ stimulation (50 ng/ml, 30 min) (Fig. 5B). EHEC O157:H7 infection alone caused a partial perinuclear redistribution of Stat1 (Fig. 5C). HEp-2 cells infected with EHEC for 6 h and then stimulated by IFN-γ showed nuclear accumulation of Stat1 (Fig. 5D). Comparable findings were observed with T84 cells, in which infection did not prevent IFN-γ-induced nuclear localization of Stat1 (Fig. 5E to H).

FIG. 5.

Immunofluorescence of HEp-2 and T84 cells for Stat1. (A) HEp-2 cells. Stat1 is distributed throughout the unstimulated HEp-2 cells, and it localizes predominantly to the nucleus upon IFN-γ treatment (50 ng/ml, 30 min) (B). (C) EHEC infection for 6 h (MOI, 100:1) leads to partial perinuclear redistribution of Stat1, but it does not completely prevent nuclear localization induced by IFN-γ treatment after such infection (n = 2) (D). (E) T84. Stat1 is localized to the cytoplasm of unstimulated cells, and it translocates to the nucleus upon IFN-γ treatment (50 ng/ml, 30 min) (F). EHEC infection for 6 h (MOI, 100:1) affects neither the constitutive distribution of Stat1 (G) nor the IFN-γ-induced nuclear localization of Stat1 (H).

EHEC infection prevents IFN-γ-induced activation of Stat1-dependent IRF-1.

Interferon regulatory factor 1 (IRF-1) is a transcription factor activated by IFN-γ in a Stat1-dependent manner (28, 48). Whole-cell protein lysates from T84 cells analyzed by immunoblotting showed that constitutive levels of IRF-1 protein expression were enhanced by IFN-γ stimulation (50 ng/ml, 4 h). However, following infection with EHEC O157:H7 strain CL-56, but not following infection with EPEC O127:H6 strain E2348/69, IFN-γ induced less IRF-1 protein expression (Fig. 6).

FIG. 6.

EHEC infection prevents IFN-γ-induced activation of Stat1-dependent IRF-1. (Upper panel) IRF-1 protein is constitutively expressed in whole-cell lysates from T84 cells analyzed by immunoblotting (lane 1). IFN-γ stimulation (50 ng/ml, 4 h) upregulates IRF-1 expression (lane 2), which is blocked by infection with EHEC strain CL-56 (lane 4) but not by infection with EPEC strain E2348/69 (lane 6) (6 h; MOI, 100:1) (n = 3). (Lower panel) Actin levels were assayed to monitor protein loading levels for samples.

Bacterial factors involved in EHEC suppression of IFN-γ-Stat1 signaling.

T84 cells were infected with EHEC O157:H7 strain CL-56 (live or heat killed), EHEC O113:H21 strain CL-15, and the commensal strain E. coli HB101 (Table 1). Figure 7A shows that infection with EHEC strain CL-56 (6 h; MOI, 100:1), but not infection with heat-killed CL-56 or HB101, prevented IFN-γ from stimulating Stat1 tyrosine phosphorylation in T84 cells, indicating that a lipopolysaccharide-independent, heat-labile factor of the pathogen is responsible for the observed effect on Stat1 signaling.

Since participation of a heat-labile EHEC factor in this signaling process was implied, the roles of various EHEC O157:H7-derived proteins in suppression of IFN-γ-induced Stat1 tyrosine phosphorylation were determined. EHEC and EPEC express different isoforms of intimin, a 94-kDa outer membrane protein which is involved in the intimate attachment of bacteria and in the modulation of host cell signal transduction responses (5, 51). Strain CL-15 is an EHEC strain (serotype O113:H21) that lacks intimin, and strain CL-8KO1 is an insertional mutant with an insertion in the eaeA (intimin) gene of wild-type parental strain CL-8 (serotype O157:H7). Infection with CL-15 (Fig. 7A) and infection with CL-8KO1 (Fig. 7B) both prevented IFN-γ-induced Stat1 tyrosine phosphorylation in T84 cells, indicating that a bacterial factor other than intimin is responsible for suppressing IFN-γ signal transduction.

EHEC strains harbor plasmid pO157, which is not present in EPEC strains (5). Moreover, EHEC and EPEC inject proteins, which have functional differences, into host cells via a type III secretion system (5). Thus, a plasmid-cured EHEC O157:H7 strain and a sepB type III-deficient EHEC O157:H7 mutant and their parental counterparts were tested for the ability to disrupt Stat1 activation. Figure 8A shows that both wild-type strain 7785-5 and its plasmid-cured derivative, strain 2-45, suppressed IFN-γ-induced Stat1 tyrosine phosphorylation equally. Figure 8B shows similar results for parental EHEC O157:H7 strain 86-24 and its sepB mutant when IFN-γ-induced Stat1 tyrosine phosphorylation was examined.

FIG. 8.

EHEC O157:H7 suppression of IFN-γ-induced Stat1 tyrosine phosphorylation is pO157, type III secretion, and Stx-1 and -2 independent. (A) Whole-cell protein extracts from T84 cells analyzed by immunoblotting showed that IFN-γ-inducible (50 ng/ml, 30 min) Stat1 tyrosine phosphorylation is suppressed equally by the pO157-positive EHEC O157:H7 strain 7785-5 and its plasmid-cured counterpart (strain 2-45; 6 h; MOI, 100:1) (n = 2). (B) Whole-cell protein extracts from T84 cells analyzed by immunoblotting showed that IFN-γ-inducible Stat1 tyrosine phosphorylation is suppressed equally by the parental EHEC O157:H7 strain 86-24 and its sepB mutant (e.g., type III secretion-deficient) strain CVD451 (6 h; MOI, 100:1) (n = 2). (C) Whole-cell protein extracts from HEp-2 cells analyzed by immunoblotting showed that IFN-γ-inducible (50 ng/ml, 30 min) Stat1 tyrosine phosphorylation (upper panel) is suppressed by both the Stx-1- and -2-positive EHEC O157:H7 strain 85-289 and its derivative, Stx-1- and -2-negative EHEC O157:H7 strain 85-170 (6 h; MOI, 100:1). Treatment with purified Stx-1 or Stx-2 (50 ng/ml, 6 h) followed by IFN-γ stimulation did not prevent Stat1 tyrosine phosphorylation. Neither bacteria nor toxins alone had an effect on Stat1 tyrosine phosphorylation. The lower panels show the levels of nonphosphorylated Stat1 (n = 2 or 3).

EHEC strains, but not EPEC strains, elaborate Shiga-like toxins that modulate host cell signal transduction (3, 5, 21). HEp-2 cells are known to express the Stx-1 and Stx-2 receptor globotriaosylceramide and to respond to Shiga-like toxin treatment with functional changes, such as apoptosis (3, 21, 37). Thus, HEp-2 cells were treated with purified Stx-1 and Stx-2 (50 ng/ml, 6 h) and then stimulated with IFN-γ. Neither Stx-1 nor Stx-2 prevented IFN-γ stimulation of Stat1 tyrosine phosphorylation in HEp-2 cells (Fig. 8C). Wild-type EHEC O157:H7 strain 85-289, which produces Stx-1 and Stx-2, and its spontaneous Stx-1- and Stx-2-negative mutant, strain 85-170, similarly suppressed IFN-γ-induced Stat1 tyrosine phosphorylation (Fig. 8C), indicating that the inhibitory effects are independent of Shiga-like toxin production by EHEC O157:H7. Neither purified toxin nor bacteria alone had an effect on tyrosine phosphorylation of Stat1. Together, these results indicate that EHEC suppression of IFN-γ-Stat1 signaling is common among different strains and that it is independent of intimin, type III protein secretion, pO157, and Shiga-like toxins.

DISCUSSION

This is the first study to show that EHEC O157:H7 suppresses IFN-γ-Stat1 signaling in epithelial cells. In this study we employed complementary techniques, including EMSA, immunoblotting, and immunofluorescence, to define the effects of EHEC infection on IFN-γ-induced Stat1 DNA binding, tyrosine phosphorylation, nuclear translocation, and gene expression. EHEC, but not heat-killed EHEC, EPEC, or the commensal strain E. coli HB101, downregulated epithelial IFN-γ-Stat1 signal transduction. Furthermore, EHEC suppression of IFN-γ-Stat1 signaling was found in multiple EHEC strains independent of lipopolysaccharide but dependent on a heat-labile bacterial factor other than intimin, a type III secreted protein, a plasmid-encoded protein, or Stx-1 and Stx-2.

Recent evidence has demonstrated that microbes disrupt cytokine signal transduction (53). Indeed, EHEC and EPEC inhibit IFN-γ secretion from lymphocytes (26), and IFN-γ knockout mice are less competent to clear infections by the attaching-effacing pathogen C. rodentium (42), suggesting that IFN-γ release following such infections is important yet compromised. IFN-γ has biological effects on intestinal epithelial cells (12, 32, 33) and activates Stat1 in intestinal epithelium (33). Thus, we determined whether EHEC modulated IFN-γ-Stat1 signal transduction in epithelial cells, one of the first lines of defense against enteric pathogens. Previous work by McKay et al. (33) showed that IFN-γ activates Stat1 in T84 cells. Here we confirmed these observations and extended them to another epithelial cell line, HEp-2. We also showed that after infection with EHEC, but not after infection with EPEC, IFN-γ was less able to stimulate Stat1 DNA binding in nuclear protein extracts from the two different epithelial cell lines. This is similar to reports that Listeria monocytogenes (46) and adenovirus (22) prevent IFN-γ-stimulated Stat1 DNA binding in infected macrophages and airway epithelia, respectively.

Tyrosine phosphorylation of Stat1 allows for dimerization and the ability to bind DNA (39). EHEC, but not EPEC, prevented the tyrosine phosphorylation of Stat1 in two epithelial cell lines in a dose- and time-dependent manner. These results are in agreement with the diminished Stat1 DNA binding observed by EMSA and are consistent with reports that infection with adenovirus (22), E. chaffeensis (29), or L. monoctyogenes (46) or incubation with lipopolysaccharide (45) and Porphyromonas gingivalis membrane vesicles (44) also disrupts IFN-γ-Stat1 signaling. The functional relevance to suppression of IFN-γ-induced Stat1 DNA binding and tyrosine phosphorylation by EHEC infection, but not by EPEC infection, was confirmed by the inhibition of the IFN-γ-Stat1-dependent induction of IRF-1 protein expression. This finding indicates that IFN-γ-induced Stat1-dependent gene expression is suppressed in epithelial cells, and we speculate that this may be important in a host response to infection (12, 30).

The cause of the signaling divergence observed between EHEC and EPEC remains to be determined. Heat-killed EHEC O157:H7 strain CL-56 did not downregulate IFN-γ-Stat1 signaling, unlike heat-killed L. monocytogenes (46), suggesting that a lipopolysaccharide-independent, heat-labile factor, such as a bacterial outer membrane protein, could be involved. Since EHEC and EPEC express intimins (products of the eaeA gene) with different homologies on the bacterial cell surface and since intimin is involved in modulation of host cell signal transduction responses (5, 51), the abilities of two different intimin-deficient EHEC strains to modulate Stat1 signaling were assessed. Both EHEC O157:H7 eaeA insertional mutant strain CL-8KO1 and the wild-type eaeA-negative EHEC O113:H21 strain CL-15 disrupted IFN-γ-induced Stat1 tyrosine phosphorylation, indicating that the signaling is independent of intimin. Furthermore, since the functions of EHEC- and EPEC-secreted proteins can differ (4, 5), we tested an EHEC O157:H7 sepB mutant strain deficient in type III secretion and showed that it disrupted IFN-γ-induced Stat1 tyrosine phosphorylation like the parental strain from which it was derived. This finding is consistent with the observation that EHEC O113:H21 strain CL-15 disrupted IFN-γ-induced Stat1 tyrosine phosphorylation even though it does not possess the locus of enterocyte effacement -encoded intimin gene eaeA and thus likely lacks the locus of enterocyte effacement-encoded type III secretion system (7).

EHEC O157:H7 strains carry a plasmid, designated pO157, which is absent from EPEC (5). However, the ability of plasmid-cured EHEC O157:H7 strain 2-45 to disrupt Stat1 activation was identical to that of the parental strain, strain 7785-5. Furthermore, EHEC, but not EPEC, elaborates Stx-1 and Stx-2 (5). As HEp-2 cells are used as a model system for EHEC O157:H7 infection (3, 21, 34, 43) and they express the Stx-1 and Stx-2 receptor globotriaosylceramide (3, 21, 37), we tested the abilities of EHEC deficient in toxin production and purified toxins to block IFN-γ-stimulated Stat1 tyrosine phosphorylation in this cell line. As shown in Fig. 8C, Shiga-like toxin-positive and -negative EHEC had similar inhibitory effects on Stat1 tyrosine phosphorylation, and purified Stx-1 or Stx-2 did not inhibit IFN-γ-induced Stat1 tyrosine phosphorylation. Thus, disruption of IFN-γ-Stat1 signaling is attributable to a heat-sensitive EHEC factor other than intimin, a type III secreted protein, a plasmid-encoded factor, or Shiga-like toxins. Determining the specific bacterial factor of EHEC strains that inhibits epithelial Stat1 activation requires further investigation.

For Stat1 to become tyrosine phosphorylated, an IFN-γ receptor 1-IFN-γ receptor 2 complex with functional associated Jak1 and Jak2 proteins must be available on the eukaryotic cell (39). The finding that Stat1 tyrosine phosphorylation is diminished by EHEC infection also could be explained if expression of the IFN-γ receptor is altered. Indeed, Mycobacterium avium infection of murine macrophages reduces the expression of IFN-γ receptor 1 and IFN-γ receptor 2 to disrupt such signaling (15). Disruption of Jak protein expression or activity could also explain our findings, since P. gingivalis membrane vesicles can cause degradation of Jak1 and Jak2 and disrupt IFN-γ signaling (44).

It is recognized that EHEC and EPEC can differ in terms of signal transduction elicited in infected host cells (5, 8, 17, 23). For example, EHEC disruption of intestinal epithelial barrier function is protein kinase C dependent (38), whereas EPEC disrupts the gut monolayer in a protein kinase C-independent fashion (55). When forming characteristic attaching and effacing lesions, the EPEC-derived translocated intimin receptor (Tir) is tyrosine phosphorylated (4), while the homologous protein in EHEC (EspE) is not phosphorylated due to the lack of such a tyrosine residue (4, 17). Furthermore, EPEC recruits the signal transduction adapter proteins Grb2 and CrkII to sites of intimate attachment, whereas EHEC does not (9). Here, we demonstrated another level of divergence, since EHEC, but not EPEC, prevented IFN-γ-stimulated tyrosine phosphorylation and DNA binding of Stat1 in epithelial cells. It is noteworthy that EPEC does initiate the tyrosine dephosphorylation of several host proteins in infected HeLa epithelial cells (24). Together with our results, this finding indicates that there are differences in the downregulation of host cell tyrosine phosphorylation by EPEC and EHEC.

Tyrosine phosphorylation of Stat proteins is generally followed by dimerization of the proteins and translocation to the nucleus (39). However, even though EHEC infection prevents IFN-γ-stimulated DNA binding, tyrosine phosphorylation of Stat1, and expression of the Stat1-dependent gene product IRF-1, unexpectedly, it did not completely prevent the nuclear localization of Stat1. Previous studies have shown that nuclear localization of Stat1 can occur in the absence of tyrosine phosphorylation to maintain constitutive gene expression (35, 36) and that phosphorylated Stat1 dimers translocate to the nucleus only upon cytokine stimulation, to modulate gene expression (36). Moreover, Stat1 is dephosphorylated by a nuclear phosphatase in fibroblasts (10), which was recently identified as the nuclear form of the T-cell phosphatase TC45 in HeLa cells (49). These observations suggest that upon IFN-γ stimulation, Stat1 reaches the nucleus, where it is either dephosphorylated or inactivated in EHEC O157:H7-infected, IFN-γ-stimulated epithelial cells.

In summary, in this study we identified the novel ability of EHEC, but not of EPEC, to disrupt IFN-γ-Stat1 signal transduction in infected epithelial cells. These findings have implications regarding the mechanism of bacterial evasion of the host immune system and suggest the possibility that the host immune response could be downregulated to support bacterial survival (26, 53). Moreover, these findings highlight the fact that EHEC and EPEC elicit different responses in infected host epithelia. Since antibiotics may be harmful to humans if they are given during the course of EHEC infections (54), this is an important consideration in the development of novel therapeutic strategies to combat these enteric infections.