|Home | About | Journals | Submit | Contact Us | Français|
Enterococcus faecalis is a member of the intestinal and oral microbiota that may affect the etiology of colorectal and oral cancers. The mechanisms by which E. faecalis may contribute to the initiation and progression of these cancers remain uncertain. Epidermal growth factor receptor (EGFR) signaling is postulated to play a crucial role in oral carcinogenesis. A link between E. faecalis and EGFR signaling in oral cancer has not been elucidated. The present study aimed to evaluate the association between E. faecalis and oral cancer and to determine the underlying mechanisms that link E. faecalis to EGFR signaling. We report the high frequency of E. faecalis infection in oral tumors and the clinical association with EGFR activation. Using human oral cancer cells, we support the clinical findings and demonstrate that E. faecalis can induce EGFR activation and cell proliferation. E. faecalis activates EGFR through production of H2O2, a signaling molecule that activates several signaling pathways. Inhibitors of H2O2 (catalase) and EGFR (gefitinib) significantly blocked E. faecalis-induced EGFR activation and cell proliferation. Therefore, E. faecalis infection of oral tumor tissues suggests a possible association between E. faecalis infection and oral carcinogenesis. Interaction of E. faecalis with host cells and production of H2O2 increase EGFR activation, thereby contributing to cell proliferation.
Oral and oropharyngeal squamous cell carcinoma (OSCC) is one of the 10 most common cancers worldwide (28, 32). The global incidence of OSCC is increasing, with over 300,000 people diagnosed with the disease each year (13, 55, 74). The prognosis for these patients is poor, with an overall 5-year survival rate of 60%. The major known genetic, environmental, and social (e.g., alcohol and tobacco use) risk factors for OSCC (1, 68) are not associated with all cases and cannot explain the increase in incidence (24, 28, 33, 45). Other than these clinical risk factors, viruses (e.g., human papillomavirus [HPV] types 16 and 18) and bacteria that reside throughout the oral cavity are speculated to play a role in initiation and progression of oral cancer (1, 28, 48). A better understanding of how bacteria may be involved in oral carcinogenesis is critical in order to develop effective biomarkers and therapies that can improve early diagnosis and patient outcomes.
Enterococcus faecalis is a member of the human commensal microbiota. As a facultative anaerobe, it is well adapted to survive within numerous complex niches of the human host, including the gastrointestinal tract (56) and the oral cavity (62), where it is associated with asymptomatic, persistent endodontic infections (69). Despite the fact that E. faecalis is most commonly considered a member of the normal flora, the microorganism has emerged as a human pathogen of significant concern (56). In our ongoing study of the microbiota in oral precancerous and cancerous lesions, we determined that the abundance of E. faecalis is significantly increased in cancerous lesions (data not shown). In the human gastrointestinal tract, the level of E. faecalis was shown to be increased significantly in human colon cancers (2). In interleukin-10 (IL-10) knockout mice (3), E. faecalis is associated with colitis and progression to colorectal cancer. Infection of colonic epithelial cells with E. faecalis induces chromosomal instability, most likely by substantial production of oxidants (72), which are among the virulence factors considered to be of key importance in the carcinogenic process. One such oxidant, hydrogen peroxide (H2O2), a small, uncharged molecule capable of freely diffusing across the cell membrane, can activate several signaling pathways, including the epidermal growth factor receptor (EGFR) pathway (9, 57).
The EGFR signaling pathway is a chief regulator of cell proliferation in various cell types, including epithelial, endothelial, and fibroblastic cells (15, 58). Abnormal transactivation of the EGFR has been described in the development and prognosis of malignancies (36, 37, 53). The EGFR and its ligand, transforming growth factor alpha (TGF-α), are commonly overexpressed in head and neck squamous cell carcinomas (HNSCC) (36), and expression increases as oral premalignant lesions progress to invasive OSCC (13, 64). Growing evidence suggests that the underlying cause of abnormal EGFR activation in cancers may involve bacterial infection. For example, Helicobacter pylori (5) and Mycoplasma hyorhinis (22) showed enhanced EGFR activation in human gastric cancer cells. As a result, our objectives in the present study were to determine (i) whether there are associations between the frequencies of E. faecalis infection and oral cancers and (ii) whether H2O2 production by E. faecalis could lead to the activation of EGFR (activated EGFR is phosphorylated EGFR [pEGFR]) and to cell proliferation in oral cancer cells.
In this report, we demonstrate the frequent infection of E. faecalis in oral tumors and a clinical link to EGFR activation. These clinical findings were validated in our in vitro model, which demonstrated EGFR activation by E. faecalis in oral cancer cells. This induction appears to be cell line independent, with E. faecalis able to enhance cell proliferation in various cell types, including epithelial (oral cancer cells), endothelial (human umbilical vein endothelial cells [HUVEC]), and fibroblastic (human gingival fibroblast [hGF]) cells. Hydrogen peroxide produced by E. faecalis or synergistically through interaction of E. faecalis with host cells activates EGFR in a dose-dependent manner. Catalase, an inhibitor of H2O2, significantly blocked E. faecalis-induced EGFR activation and cell proliferation, in accordance with the results for the EGFR inhibitor gefitinib. Relative to the case with E. faecalis, little effect was observed in response to treatments with additional Gram-positive oral bacteria (Streptococcus anginosus and Streptococcus mitis) associated with HNSCC or with bacterial ligands (lipoteichoic acid [LTA] from E. faecalis and lipopolysaccharide from Escherichia coli). We postulate that the elevated frequency of E. faecalis infection may promote EGFR activation through H2O2, resulting in abnormal cell proliferation in oral cancer.
Oral specimens were obtained from 10 male and 10 female oral cancer tumor patients with an average age of 59 years. The control group consisted of 10 male and 10 female cancer-free patients with an average age of 58 years. Ethical approval for the study was granted by the institutional review boards of The State University of New York at Buffalo and The Roswell Park Cancer Institute (RPCI), Buffalo, NY. Written consent was received from all participants involved in the study. Tumor tissues from keratinized oral mucosa were removed from advanced oral lesions (high-grade lesions and OSCC) under aseptic conditions at the Department of Dentistry and Maxillofacial Prosthetics at RPCI. Histological grading of tissues was completed in the RPCI Department of Pathology and Laboratory Medicine. For the control group, normal tissues from keratinized oral mucosa without any sign of inflammation from cancer-free patients treated at the School of Dental Medicine, The State University of New York at Buffalo, were collected from tissue discarded during implant surgery, esthetic crown lengthening, or soft tissue grafting. All specimens were immediately placed into irradiated sterile saline solution, transferred to the laboratory, and incubated in an antibiotic cocktail (twice, with penicillin and streptomycin) and subsequently in povidone-iodine (Betadine) solution for 5 min for surface bacterial decontamination (27).
Total genomic DNA was isolated using a DNeasy blood and tissue kit (Qiagen, Valencia, CA) and mechanical lysis. The tissues mentioned above were incubated in tissue lysis buffer (ATL; Qiagen) and proteinase K, followed by mechanical lysis with a FastPrep FP120 machine (MP Biomedical, Solon, OH), treatment with RNase A, and purification on Qiagen columns. One tube containing 10 μl of sterile water was extracted in parallel to monitor possible contamination.
The PCR used to determine the occurrence of E. faecalis was performed in a 25-μl reaction mixture containing 10 μl DNA template, 2.5 μl of 10× PCR buffer, 2 mM MgCl2, 0.1 μl Taq DNA polymerase, a 0.2 mM concentration of each deoxynucleoside triphosphate (dNTP), and 1 μM (each) specific primers. DNA extracted from E. faecalis strain OG1RF and sterile, ultrapure water were included instead of the sample as positive and negative controls, respectively. 16S rRNA gene-directed species-specific primers were as follows: forward (5′-3′), GTTTATGCCGCATGGCATAAGAG; and reverse (5′-3′), CCGTCAGGGGACGTTCAG. The DNA amplification conditions for PCR with species-specific primers were 5 min of initial denaturation at 95°C followed by 36 consecutive cycles of 95°C for 30 s, 60°C for 1 min, and 72°C for 1 min (14). DNA samples that did not yield PCR products with the E. faecalis-specific primers were reamplified in an autonested PCR using the same primer pair. The amplification parameters were modified by reducing the number of cycles to 30 cycles and using 2 μl of template (67). The positive amplicons were validated by sequencing. The chi-square test with Yates' correction was applied using Prism 5 (GraphPad Software, Inc., La Jolla, CA) to analyze statistically significant differences between normal (tissues from cancer-free patients) and tumor (advanced lesions from oral precancer and cancer patients) tissues with regard to the number of E. faecalis-positive samples. The significance level was set at P values of <0.05.
Formalin-fixed, paraffin-embedded oral tissues (10 each of the normal and tumor tissues mentioned above) were cut into 4-μm sections. The sections were deparaffinized by submersion in xylene for 5 min. This was repeated two times and followed by sequential immersion in 100%, 95%, 70%, and 50% ethanol (5 min each). Slides were washed with sterile phosphate-buffered saline (PBS, pH 7.2) for 1 min and incubated in 50 mM Tris-EDTA (TE) buffer (pH 7.4) with lysozyme (10 mg/ml) at 37°C for 20 min, followed by rinsing with PBS (3 times). Sections were then treated with 50 mM TE buffer containing proteinase K (7 μg/ml) at 37°C for 20 min, followed by rinsing with PBS (3 times). Sections were pretreated with a hybridization buffer (0.9 M NaCl, 20 mM Tris-Cl [pH 7.5], and 0.1% [wt/vol] SDS) at 48°C for 20 min (26). Prewarmed hybridization buffer containing 25% formamide and 0.1 μM oligonucleotide probe (Alexa Fluor 594-labeled Enfl84 probe [5′-CCTCTTTCCAATTGAGTGCA-3′] for E. faecalis  and Alexa Fluor 488-labeled EUB338 probe [5′-GCTGCCTCCCGTAGGAGT-3′] for universal bacteria) was added to the sections (26). Following overnight incubation in a dark humid chamber at 48°C (26), the slides were washed with posthybridization solution (0.9 M NaCl, 20 mM Tris-Cl [pH 7.5]) for 20 min. Sections were counterstained with 0.025% (wt/vol) concanavalin A (ConA)-Alexa Fluor 647 conjugate (Molecular Probes) for 20 min, followed by 0.2 μg/ml DAPI (4′,6-diamidino-2-phenylindole; Molecular Probes) for 2 min. Each slide was washed with water, air dried, and subsequently mounted with ProLong Gold antifade reagent (Invitrogen). Slides were imaged on an Olympus FV1000 laser scanning confocal microscope and digitally photographed using a Fluoview 1000 system. Five fields at a magnification of ×60 were randomly chosen and scored for colocalization of the Enfl84 and EUB338 probes (41). Results are expressed as means and standard errors of the means (SEM). Statistical significance was determined by Student's t test.
Six representative specimens of normal and tumor tissues were chosen randomly. After deparaffinization and enzymatic treatment as described above, sections were blocked in PBS containing 5% goat serum at room temperature for 1 h. Sections were then incubated for 1 h in primary antibody followed by 1 h in secondary antibody. The primary antibody was against the activated form of EGFR (pEGFR) (a mouse monoclonal antibody specific for tyrosine-phosphorylated EGFR [clone 74; BD Transduction Laboratories]). The secondary antibody was an Alexa Fluor 488-conjugated goat anti-mouse IgG (Molecular Probes) (38). The sections were subsequently prehybridized, hybridized with the Enfl84 probe, posthybridized, counterstained, and mounted as described above. Slides were imaged on an Olympus FV1000 laser scanning confocal microscope and digitally photographed using a Fluoview 1000 system.
A253 and CAL27 human oral cancer cells from ATCC and primary human gingival fibroblasts (hGFs) (generously provided by Rosemary Dziak, The State University of New York at Buffalo) were maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen) and 1% penicillin-streptomycin. HUVEC were purchased from ATCC and maintained in Ham's F-12 medium (F12K; Invitrogen) with 0.1 mg/ml heparin sodium salt from porcine intestinal mucosa (Sigma), 0.05 mg/ml endothelial cell growth supplement (ECGS; Sigma), 10% FBS, and 1% penicillin-streptomycin. Prior to treatments, cells grown to 80% confluence were serum starved overnight in DMEM containing 1% antibiotic. For the EGF and hydrogen peroxide treatments, cells were incubated in the same medium supplemented with 100 ng/ml EGF (R&D Systems) (38, 51) or 1 μM hydrogen peroxide (Sigma) (57). For signaling inhibition, cells were pretreated with 1 μM gefitinib (Cayman Chemical Company) (7) or 1,200 units/ml catalase (59, 73) from Micrococcus sp. (Sigma) for 1 h. Lipoteichoic acid from E. faecalis (100 ng/ml; Sigma) (6) and lipopolysaccharide from E. coli (100 ng/ml; Sigma) (4) were used as additional controls.
E. faecalis VF, V5, and V8 are human isolates from oral squamous cell carcinoma tissues collected in the current study. E. faecalis CDC551 and OG1RF are human isolates from elsewhere (generously provided by Jose A. Lemos, University of Rochester). Streptococcus salivarius strain KB005, Streptococcus anginosus strains ATCC 33397 and NCTC10708, and Streptococcus mitis strains SK141 and NCTC10712 are oral streptococci obtained from Frank A. Scannapieco (The State University of New York at Buffalo). These strains were routinely grown in brain heart infusion broth (BHI; Difco) supplemented with 10% FBS at 37°C.
Total EGFR and phosphorylated EGFR (pEGFR) in cultured cells were measured using a RayBio cell-based EGFR (activated) enzyme-linked immunosorbent assay (ELISA) kit. Briefly, A253 or CAL27 cells were grown in 96-well microtiter plates at a density of 30,000 cells/well and maintained in serum-free medium for 24 h prior to addition of bacteria. Overnight bacterial cultures were washed in PBS (3 times) and then added to cells in serum-free medium at a multiplicity of infection (MOI) of 10, 100, or 1,000, with the MOI defined as the number of bacterial CFU per cultured cell. Bacteria were cocultured with oral cancer cells for 45 min. Aliquots of serum-free medium in the absence and presence of EGF (100 ng/ml) for 45 min served as negative and positive controls, respectively. Bacteria in the absence of cancer cells were included to confirm that bacteria alone were not producing an activating signal during the experimental procedure. After 45 min, cells were washed 3 times with PBS and then fixed. Levels of total EGFR and pEGFR were examined according to the manufacturer's instructions. The optical density of each well was measured at 450 nm in a microplate spectrophotometer (BioTek EL808) (66). The functional activity of EGFR in oral cancer cells was calculated based on the level of pEGFR normalized with total EGFR expression.
A253 or CAL27 cells seeded at 1 × 105 cells/ml in glass-bottom dishes (MatTek) were left untreated (control), treated with EGF (100 ng/ml; positive control), or infected with E. faecalis strain VF at an MOI of 1,000:1 for 45 min. For inhibition of signaling, oral cancer cells were pretreated for 1 h with gefitinib or catalase as described in “Cell culture” before being treated with EGF or E. faecalis strain VF. Cells were then washed with PBS and fixed with 4% formaldehyde diluted in PBS for 15 min at room temperature (59). Cells were permeabilized for 30 min at room temperature with PBS containing 0.2% Triton X-100 and then blocked in PBS containing 0.2% Triton X-100 and 5% goat serum for 30 min. Cells were subsequently incubated for 1 h with primary antibody followed by 1 h with secondary antibody as described above (38). Cells were counterstained with 0.2 μg/ml DAPI (Molecular Probes) for 2 min. Each dish was washed with water and subsequently mounted with ProLong Gold antifade reagent (Invitrogen). Cells were imaged on an Olympus FV1000 laser scanning confocal microscope and digitally photographed using a Fluoview 1000 system.
A253 and CAL27 cells, HUVEC, and hGFs were plated at 1 × 105 cells/ml in serum-free medium in 96-well plates for 24 h before the assay, followed by infection with E. faecalis VF at an MOI of 1,000:1, treatment with 100 ng/ml LTA from E. faecalis, or no treatment (negative control) for 24 h. Bacteria in the absence of cancer cells were included to confirm that bacteria at an MOI of 1,000:1 did not produce an additional signal during the experimental procedure. To assess cell proliferation, we used a bromodeoxyuridine (BrdU) cell proliferation assay (Exalpha Biologicals) according to the manufacturer's instructions. After stimulation, cells were incubated in culture medium containing BrdU (2 μl/ml) for 2 h before termination of the experiment. The cells with incorporated BrdU were fixed and permeabilized, and their DNA was denatured by treatment with fixing solution. An anti-BrdU monoclonal detector antibody was incubated with the cells for 1 h, followed by horseradish peroxidase-conjugated goat anti-mouse antibody for 30 min at room temperature. After treatment with the chromogenic substrate tetramethylbenzidine (TMB) and addition of stop solution, the amount of BrdU was quantified spectrophotometrically with a microplate spectrophotometer at a wavelength of 450 nm, with a reference wavelength of 570 nm (23).
A253 cells were seeded at 1 × 105 cells/ml and grown on glass-bottom dishes (MatTek) for 2 days. After incubation in serum-free medium for 2 days, cells were treated with E. faecalis VF at an MOI of 1,000:1, 100 ng/ml EGF (R&D Systems), or 1 μM hydrogen peroxide (Sigma) or left untreated as a negative control for 24 h. For inhibition of signaling, oral cancer cells were pretreated for 1 h with gefitinib or catalase as described in “Cell culture” before being treated continually with VF, EGF, or hydrogen peroxide until termination of the experiment. Cells were then washed with PBS and fixed with 4% formaldehyde diluted in PBS for 30 min at room temperature. Cells were treated with Peroxidazed-1 (Biocare Medical) for 15 min and then incubated for 30 min at room temperature with PBS containing 0.2% Triton X-100 and 5% FBS. Cells were subsequently incubated for 1 h with primary antibody followed by 1 h with secondary antibody. The primary antibody used was a rabbit monoclonal antibody specific for Ki-67 (clone SP6; Biocare Medical). The secondary antibody was a horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (MACH 2; Biocare Medical). The Ki-67-positive staining was developed by diaminobenzidine (DAB; Biocare Medical), followed by hematoxylin as a counterstain (76).
Quantification of hydrogen peroxide production was measured using an Amplex Red hydrogen peroxide-peroxidase assay kit (Invitrogen) according to the methods of Chen et al. (12) and the manufacturer, with minor modifications. Briefly, a reaction mixture containing 50 mM Amplex Red reagent and 0.1 U/ml horseradish peroxidase in 0.05 M sodium phosphate buffer, pH 7.4, was added to a 96-well microtiter plate and incubated at 37°C for 10 min. Bacterial cultures (E. faecalis VF, V8, V5, CDC551, and OG1RF; S. salivarius KB005; S. anginosus ATCC 33397 and NCTC10708; and S. mitis SK141 and NCTC10712) were grown overnight in BHI with 10% FBS, centrifuged, and subsequently washed in PBS (3 times). Bacteria were resuspended, diluted to an optical density at 600 nm (OD600) of 0.1 in DMEM containing 25 mM glucose, and incubated at 37°C for 30 min. Aliquots of diluted bacteria were added to the prewarmed reaction mixture. A series of hydrogen peroxide concentration standards together with DMEM as the blank control were also included in the plate. The absorbance was read at 560 nm every 5 min for 60 min, using a fluorescence plate reader (SpectraMax M5; Molecular Devices). Background fluorescence determined for a blank control was subtracted from each value. Hydrogen peroxide production was calculated from the standard curve.
The subjects enrolled in this study were classified either as possessing oral tumors (high-grade lesions and OSCC) or as cancer-free controls (oral tissues from cancer-free patients). E. faecalis frequencies in the two patient groups are shown in Fig. 1A. E. faecalis DNA was detected by PCR in 20/20 (100%) oral tumors. The frequency of E. faecalis in control samples was significantly lower (5/20 samples [25%]) (P < 0.0001). Positive PCR results were confirmed by direct sequencing of the 16S rRNA gene, and the results were identical to published E. faecalis 16S rRNA gene sequences.
Clinical features (age, gender, and smoking status) were compared between patients who were E. faecalis positive and E. faecalis negative. The results are shown in Table S1 in the supplemental material, with no significant differences observed (P > 0.05).
Given the increased frequency of E. faecalis DNA in oral tumors, we next asked whether E. faecalis could be detected in oral tumor histological sections, and if so, where it was found. To address this question, we performed 16S rRNA gene FISH, using a pan-16S rRNA gene probe (EUB338) (26) to detect a broad range of bacteria and also using an E. faecalis-specific 16S rRNA gene probe (Enfl84) (71), on oral tumor biopsy specimens (10 specimens) and normal oral tissue (10 specimens). We found that E. faecalis was generally observed within the prickle cell layer of the tumor tissue, where the cell shape is oval and represents the body of the epithelium, and was seen less frequently at the superficial layer of the mucosa or tissue border, where the cell is flat and contains small oval nuclei. Some examples of bacteria (EUB338; green) and E. faecalis (Enfl84; red) (Fig. 1B, upper panels) are indicated in the photomicrographs. On the other hand, fewer bacteria, including E. faecalis, were detected in the control group. Those bacteria that were detected in the control group were observed solely on the tissue surface, which contains thin, flat, and nonnucleated cells (Fig. 1B, lower panels). Quantitation of the FISH results (Fig. 1C) showed that E. faecalis was enriched in oral tumors relative to its level in normal tissue (P = 0.0008).
One of the major pathogenic mechanisms of oral carcinogenesis involves the EGFR signaling pathway. Several reports have shown that EGFR is overexpressed in 80 to 100% of HNSCC (36, 60), which can lead to invasive OSCC (64). Furthermore, our study found that E. faecalis was strongly associated with oral tumors. Therefore, we hypothesized that the presence of E. faecalis within oral tumors may involve EGFR activation. To first address this hypothesis, we performed FISH with oligonucleotide probes to detect E. faecalis, together with an immunofluorescence assay to detect activated EGFR. Generally, when EGFR interacts with its ligand, the ligand-receptor complexes are internalized, with most pEGFR localized within small vesicular structures distributed throughout the cytoplasm and a small amount of pEGFR localized in the nucleus (38, 51). In this study, similar immunostaining patterns for pEGFR were obtained in oral tumors but not in the healthy control tissues (data not shown). Our results showing substantially less EGFR activation in healthy control tissues are consistent with a previous study where EGFR expression was significantly lower in normal controls, contributing to less EGFR activation (64). We also noted that E. faecalis was localized to the phosphorylated EGFR area, which appeared in the prickle cell layer of oral tumor tissues (Fig. 2A, panels iv and v). However, we did not observe punctate vesicles containing pEGFR when E. faecalis was located in the superficial part of the oral tumor (Fig. 2A, panel vi), as described for Fig. 1B regarding the oral mucosa histology. These findings suggested that E. faecalis could be associated clinically with pEGFR in oral tumor specimens. Therefore, we investigated whether this association also occurs in vitro. We isolated and propagated E. faecalis from oral cancer tissues and confirmed the strain identities by 16S rRNA gene sequencing. Among the cancer isolates, E. faecalis VF appeared to show the highest capacity to promote EGFR activation (data not shown). As a result, we selected this strain to further determine the role and underlying mechanism of E. faecalis in EGFR activation and cell proliferation. We infected the oral cancer cell line A253 with increasing doses of E. faecalis VF (MOIs of 10 to 1,000) to analyze EGFR activation by ELISA (Fig. 2B) and immunofluorescence assay (Fig. 2C). As shown in Fig. 2B, EGFR activation was significantly increased after E. faecalis infection, with approximately 1.3-fold induction compared with the uninfected controls. Similar levels were also observed in our EGF-treated positive control.
To confirm that E. faecalis infection results in increased EGFR activation, we next performed immunofluorescence analysis of infected cells by using an antibody that recognizes pEGFR. A253 cells were either left untreated, infected with E. faecalis VF, or treated with EGF for 45 min. As shown in Fig. 2C, positive pEGFR staining in EGF-treated cells resulted in punctate vesicular compartments within the cell. The pEGFR staining pattern and intensity of cells infected with E. faecalis VF were similar to those of EGF-treated cells. These results support the clinical association between E. faecalis infection and pEGFR activation in oral tumors. Conversely, EGFR activation was absent in control cells. Consistent with these findings, Fig. S1A and B in the supplemental material show that E. faecalis was also able to induce EGFR activation with another oral cancer cell line, CAL27, suggesting that the ability of E. faecalis to enhance EGFR activation is independent of the oral cell line used.
A downstream effect of EGFR signaling is enhanced cell proliferation. During oral carcinogenesis, growth signaling can become dysregulated through increased EGFR activation (13), thereby contributing to cell proliferation in various cell types, including epithelial, endothelial, and fibroblastic cells (15, 58, 60). We initially sought to determine whether E. faecalis could also enhance cell proliferation in these cells. The oral cancer epithelial cell lines CAL27 and A253 were cultured with either E. faecalis VF (MOI of 1,000) or LTA from E. faecalis (100 ng/ml) (6). LTA was used to further examine whether the external cell structure of E. faecalis was important in this signaling mechanism. After serum starvation for 24 h, E. faecalis VF significantly increased BrdU incorporation, a sign of cell proliferation, by 1.5-fold and 1.4-fold in CAL27 (P < 0.001) and A253 (P < 0.01) cells, respectively, while lipoteichoic acid from E. faecalis resulted in only 1.1-fold and 1.07-fold induction, respectively (P > 0.05) (Fig. 3A and andB).B). In addition, there was a 40% to 50% increase in BrdU incorporation in cells infected with E. faecalis VF compared with those treated with lipoteichoic acid for both CAL27 (P < 0.01) and A253 (P < 0.05) cells. The pattern observed in the BrdU incorporation assay corresponded to the level of EGFR activation in cells treated with either E. faecalis or lipoteichoic acid in comparison with controls (see Fig. S2A in the supplemental material), suggesting that there could be factors other than cell structure, e.g., lipoteichoic acid, which can enhance EGFR activation and cell proliferation.
The tumor microenvironment is composed of various cell types, including fibroblasts and endothelial cells (35). To assess whether E. faecalis could also increase cell proliferation in these cell types, we treated hGF cells and HUVEC under the same conditions as those used for the CAL27 and A253 cell lines. After infection of hGF cells with E. faecalis, we found that the average level of BrdU incorporation was elevated 2.4-fold and 2-fold relative to those with the control (P < 0.001) and lipoteichoic acid (P < 0.001), respectively. Elevated BrdU incorporation (a 1.2-fold induction) was also observed in lipoteichoic acid-treated cells compared to control cells (P < 0.001) (Fig. 3C). Moreover, a similar, statistically significant increase occurred when HUVEC were treated with E. faecalis VF and lipoteichoic acid (Fig. 3D). These findings suggest that E. faecalis strongly promotes proliferation in various tumor microenvironment cell types.
After demonstrating that E. faecalis could promote EGFR activation and cell proliferation, we next examined whether E. faecalis-activated EGFR contributes to increases in cell proliferation. To block EGFR activation, A253 cells were preincubated with gefitinib (1 μM) (7), an EGFR kinase inhibitor, for 1 h and then treated with either E. faecalis VF or EGF for 45 min. EGFR activation was analyzed by ELISA and immunofluorescence assay. We observed a significant decrease in EGFR activation when gefitinib was incubated prior to the addition of either E. faecalis VF (P < 0.05) or EGF (P < 0.01) (Fig. 4A). Confocal microscopy analysis also showed that after cells were exposed to EGF or E. faecalis VF, the prior gefitinib treatment prevented EGFR activation, to levels similar to control levels (Fig. 4B). We additionally tested the effects of E. faecalis and EGF on CAL27 cells after preincubation with gefitinib under the same conditions. Again, we found a marked reduction in EGFR activation as a result of gefitinib treatment in both the ELISA and the immunofluorescence assay (see Fig. S2B and C in the supplemental material). Having shown that gefitinib reduced EGFR activation, we sought to determine if this block affected cell proliferation after E. faecalis infection. Cell proliferation of A253 cells incubated under similar experimental conditions was quantitated by a BrdU incorporation assay. As shown in Fig. 4C, EGF (P < 0.01) and E. faecalis VF (P < 0.01) significantly enhanced BrdU incorporation (1.2-fold) relative to the control level. Preincubation of A253 cells with gefitinib alone showed a nonsignificant decrease in BrdU incorporation compared to the untreated cell control (P > 0.05), suggesting that this concentration of gefitinib was not toxic to the cell. A significant decrease occurred, however, when gefitinib was incubated prior to addition of both EGF (P < 0.001) and E. faecalis (P < 0.001). This attenuation was significantly less than that for both nontreatment and treatment with gefitinib alone (P < 0.001).
The Ki-67 protein is expressed robustly throughout all phases of the cell cycle, except for in G0 resting cells, making Ki-67 an exceptional marker for detecting actively proliferating cells (8, 53). Ki-67 was detected in cells treated under the same experimental conditions as those for the BrdU assay described above, using immunocytochemical staining to confirm cell cycle acceleration and proliferation. As shown in Fig. 4D, the number of Ki-67-positive cells (with DAB chromogen at the nucleus) increased considerably after EGF or E. faecalis VF was added compared to the numbers for controls (nontreatment and treatment with gefitinib). In contrast, gefitinib-pretreated cells exposed to EGF or E. faecalis VF showed a marked decrease in the ability to promote cell proliferation, indicating that the inhibitory effect on the stimulus-treated cells after pretreatment with gefitinib was not due to cytotoxicity from the inhibitor. These findings suggest that E. faecalis VF-enhanced cell proliferation is mediated by EGFR signaling pathway activation.
Several studies indicate that exposure to oxidative stress, such as the presence of H2O2, preferentially facilitates EGFR activation (38, 58, 59), enhances cell proliferation (43, 57, 70), and thereby promotes tumorigenesis (29). Because E. faecalis can generate a considerable amount of oxidants, such as superoxide and H2O2 (30), our next series of experiments was designed to determine whether H2O2 production by E. faecalis VF could induce cell proliferation via EGFR activation. We first evaluated the effect of H2O2 on EGFR activation. Initially, A253 cells were left untreated or were pretreated with either gefitinib (1 μM) or catalase from Micrococcus lysodeikticus (1,200 units/ml) to hydrolyze peroxides (59, 73) for 1 h. No changes in EGFR activation (P > 0.05) were observed with these treatments (Fig. 5A). Addition of 1 μM H2O2 (57) or EGF increased the level of EGFR activation 1.3-fold (P < 0.05) or 1.5-fold (P < 0.01), respectively. Preincubation with gefitinib significantly decreased the level of EGFR activation after cells were exposed to H2O2, to levels similar to the activation levels observed with the untreated control and inhibitors (gefitinib and catalase) alone. A greater decrease in EGFR activation occurred with catalase treatment, indicating that catalase plays a primary role in inhibiting H2O2-induced EGFR activation.
To determine the level of H2O2 generated by E. faecalis VF, bacteria were incubated with DMEM for 30 min, with or without the addition of catalase. As the MOI increased from 10 to 1,000, the level of H2O2 production by E. faecalis VF significantly increased compared to the control level (without bacteria) (P < 0.001). The level of H2O2 production was significantly reduced when catalase was added (P < 0.001) (Fig. 5B). The dose-dependent increase in H2O2 production by E. faecalis VF correlated with the level of pEGFR we observed after A253 cells were exposed to E. faecalis VF at high MOIs (Fig. 2B). Conversely, heat-killed E. faecalis VF failed to enhance H2O2 production and the pEGFR level (data not shown). Previous studies reported that host cells produced more hydrogen peroxide after exposure to E. faecalis (11). Therefore, in our experiments, it was possible that the hydrogen peroxide produced by both E. faecalis and the cancer cells synergistically enhanced pEGFR. To determine hydrogen peroxide production by A253 cells in response to E. faecalis, we measured the amount of hydrogen peroxide produced after the cells were exposed to E. faecalis. As shown in Fig. 5C, infection of A253 cells with E. faecalis VF significantly enhanced hydrogen peroxide production. The addition of catalase dramatically attenuated the level of hydrogen peroxide. Moreover, the amount of hydrogen peroxide in A253 cells infected with E. faecalis VF was significantly higher than that in E. faecalis alone. As a control, we tested heat-killed E. faecalis VF that was not able to produce reactive oxygen species (ROS) and found that A253 cells incubated with heat-killed bacteria failed to enhance hydrogen peroxide production, similar to case with the control (A253 cells) or heat-killed bacteria alone. Taken together, these data suggest that production of H2O2 by E. faecalis VF and cancer cells could synergistically promote EGFR activation. Another plausible explanation for these results is that interaction of E. faecalis VF with A253 cells stimulates this bacterium to produce more ROS.
We next assessed whether catalase can block E. faecalis VF-induced EGFR activation in A253 cells. Preincubation with catalase significantly decreased the level of pEGFR activation in E. faecalis VF-exposed cells, to the basal level (for DMEM with or without catalase) (P < 0.01). A consistent pattern of pEGFR activation was observed when catalase or gefitinib was used as a treatment before the addition of H2O2 or EGF (Fig. 5D). To confirm these findings, immunofluorescence staining showed that when cells were preexposed to catalase, E. faecalis VF-induced pEGFR activation was abolished, similar to the case in cells preexposed to gefitinib. Similar reductions were also observed in EGF-treated cells (positive control) after preexposure to gefitinib (Fig. 5E). To extend these findings, we performed the experiments again, but using CAL27 cells. The profiles of EGFR activation by ELISA and confocal microscopy analysis were similar to those for A253 cells and decreased significantly with E. faecalis VF-infected cells preexposed to catalase (see Fig. S3A and B in the supplemental material). This suggests that H2O2 from E. faecalis VF and cancer cells promotes EGFR activation in both cell lines.
Our previous data showed that lipoteichoic acid from E. faecalis partially enhanced cell proliferation (Fig. 3A and andB)B) compared to that with whole E. faecalis VF, indicating that cellular factors other than surface structures (e.g., lipoteichoic acid) were able to promote cell proliferation. Therefore, we set out to determine whether this cell proliferation was due primarily to H2O2 production from E. faecalis VF and cancer cells. A253 cells were incubated for 1 h with H2O2 or EGF or for 45 min with E. faecalis after pretreatment with either catalase or gefitinib. Cell proliferation was analyzed by the BrdU incorporation assay. Aliquots of serum-free medium in the absence and presence of inhibitors (catalase and gefitinib) served as controls. As shown in Fig. 5F, all three stimuli (H2O2, EGF, and E. faecalis VF) were able to significantly enhance BrdU incorporation compared with controls. In contrast, catalase-treated cells in the presence of stimuli showed a significant decrease in BrdU incorporation, with approximately 15% less incorporation than the basal level (P < 0.01). A similar response, approximately 30% less than the basal level (P < 0.001), was found after treatment with gefitinib. However, cells treated with catalase or gefitinib alone showed no significant difference in BrdU incorporation relative to untreated cells (P > 0.05). These data suggested that the inhibitory effect in the stimulus-treated cells after pretreatment with catalase and gefitinib was not due to cytotoxicity from inhibitors. To confirm these data, A253 cells were treated under the same conditions, stained for Ki-67 by immunocytochemistry, and visualized by microscopy (Fig. 5G). We found that Ki-67-positive cells appeared more frequently in cells exposed to E. faecalis VF or H2O2 (positive control) than in untreated cells. Catalase, on the other hand, decreased Ki-67 expression after exposure to stimuli. A similar decrease was also observed in H2O2-treated cells cocultured with gefitinib. Taken together, these results indicate that H2O2 production by E. faecalis alone or by both E. faecalis and cancer cells via a synergistic interaction enhances EGFR activation, contributing to cell proliferation in oral cancer cells. These data provide evidence that antioxidants such as catalase are potential candidates for protection against E. faecalis-induced EGFR activation and cell proliferation.
Numerous reports have observed an association of oral streptococci, including S. anginosus and S. mitis (49, 50, 61), with HNSCC. However, the role of these oral streptococci in oral carcinogenesis remains unclear. In the present study, we found that H2O2 production by E. faecalis and cancer cells augments the pathogenicity of these microorganisms by enhancing cell proliferation through EGFR activation, potentially giving rise to neoplastic transformations in oral cancer. To determine whether this trait could be found in other bacteria associated with HNSCC, H2O2 production by E. faecalis cancer isolates (VF, V5, and V8) was first quantified in vitro and compared to H2O2 production of other clinical E. faecalis isolates (CDC551 and OG1RF), S. anginosus, and S. mitis. As shown in Fig. 6A, we found that the level of H2O2 production was substantially elevated in all E. faecalis strains compared to all other bacteria tested. The average level of H2O2 production for all strains of oral streptococci was significantly lower than that generated by E. faecalis (P < 0.001). Lipoteichoic acid from E. faecalis or lipopolysaccharide from E. coli alone had no effect on H2O2 production compared with a no-bacterium control. These results indicate that among Gram-positive bacteria associated with oral cancer, E. faecalis is notable in generating H2O2.
Our study found that E. faecalis VF could induce cancer cells to generate increased levels of endogenous hydrogen peroxide, either by production of exogenous hydrogen peroxide or by an unknown mechanism. To determine whether this phenotype could be found in other bacteria associated with HNSCC, the oral streptococcus strains used for Fig. 6A were incubated with serum-starved cells for 45 min. As shown in Fig. 6B, the amount of hydrogen peroxide produced from cancer cells and any E. faecalis strain was significantly higher than that produced from cancer cells incubated with the oral streptococcal strains (~15-fold increase; P < 0.001), lipoteichoic acid (~28-fold increase; P < 0.001), lipopolysaccharide (~28-fold increase; P < 0.001), or no bacteria (~28-fold increase; P < 0.001). These data support the view that hydrogen peroxide produced by E. faecalis may be one factor essential for increased ROS production by cancer cells.
To determine whether the H2O2 production from these bacteria would correlate with an increase in EGFR activation, we examined the level of EGFR activation by comparing H2O2-producing E. faecalis strains to Gram-positive bacteria associated with HNSCC that produced less H2O2. Lipoteichoic acid, lipopolysaccharide, and a no-bacterium control were also included. We found that the level of pEGFR activation was markedly increased after CAL27 cells were exposed to any of the E. faecalis strains. A slight increase was also observed when lipoteichoic acid from E. faecalis was added, but this difference was not significant (P > 0.05). The average level of pEGFR activation from all E. faecalis strains was significantly greater than that for cells cultured with oral streptococci (P < 0.001), lipoteichoic acid (P < 0.01), lipopolysaccharide (P < 0.01), or no bacteria (P < 0.01) (Fig. 6C). These findings strongly suggest that substantial H2O2 production by E. faecalis and E. faecalis-infected cancer cells is a trait frequently found in E. faecalis infection but not in oral streptococcal infection associated with HNSCC. This trait could ultimately lead to increased pEGFR activation in oral cancer cells by E. faecalis.
EGFR is one of the most commonly overexpressed receptor tyrosine kinases in head and neck malignancies (17, 36). The overexpression of EGFR has been correlated with a poor prognosis for OSCC patients (13, 53). In addition, EGFR inhibitors have been shown to have an antiproliferative effect in preclinical models of OSCC (25, 65). Therefore, determining the underlying causes of EGFR activation in oral cancer is of substantial clinical importance. The implication of bacterial infection by H. pylori (37) and M. hyorhinis (22) as the underlying cause of EGFR overexpression and activation in gastric cancers supports our interest in the potential role of bacteria in OSCC etiology.
In this study, we began to explore the relationship between E. faecalis and OSCC. E. faecalis DNA was identified in 100% (20/20 samples) of the oral tumors examined, a frequency rate that is statistically significant compared to that for the healthy control group, with 25% (5/20 samples) of control samples containing E. faecalis (Fig. 1A). Our results for normal tissue with no clinical signs of inflammation are similar to those of other studies that found the frequency of E. faecalis in healthy oral environments to be around 20% (62). Confounding factors such as age, gender, and smoking status of patients, which could be interpreted as biases, did not interfere with our results.
Histological analysis (Fig. 1C) revealed that E. faecalis is found more frequently in oral tumors than in control tissue. These results are similar to those of previous reports in which the level of E. faecalis in colorectal cancer patients was higher than that in healthy volunteers (2). We found E. faecalis located in the prickle cell layer, indicating middle or deep parts of the tumors, rather than on the superficial layer or surface as we observed in the normal controls (Fig. 1B). In accordance with this discovery, E. faecalis-infected cancer isolates were able to invade oral cancer cells and remain viable for at least 72 h under serum- and antibiotic-free conditions (data not shown). Similar observations of bacteria located deep inside tumors but not on the surface or ulcerated portions of the cancer tissue have been reported for oral streptococci associated with head and neck cancer (50, 63) and for Fusobacterium associated with colorectal carcinoma (41). These findings suggest that E. faecalis is well adapted to survive within the evolving tumor microenvironment.
We have shown by confocal microscopy analysis that E. faecalis and pEGFR-positive cells are in close association within oral cancer tissues (Fig. 2A). Conversely, this localization is rarely found in healthy control tissue, corresponding to the lower frequency of E. faecalis (data not shown). This concept is in line with the work of Wong et al. (75), who reported that H. pylori infection was associated with EGFR overexpression in patients with chronic active gastritis but not in healthy controls. Therefore, exploration of a causal link between E. faecalis and pEGFR activation in oral cancer is justified.
Other investigators have previously reported that infection with E. faecalis causes chromosomal instability in colonic epithelial cells (72), likely a result of oxidants that are abundantly produced by E. faecalis (30). In this study, we investigated the role of E. faecalis in EGFR activation in the oral cancer cell lines A253 and CAL27. We found that the level of EGFR activation was increased in both cell lines (Fig. 2B and andC;C; see Fig. S1A and B in the supplemental material), indicating that EGFR activation by E. faecalis is cell line independent. Our focus on E. faecalis infection did not exclude other members of the bacterial flora that may enhance pEGFR activation. Accordingly, we also used lipopolysaccharide from E. coli, which represented an external structure from Gram-negative bacteria, and lipoteichoic acid from E. faecalis, which represented an external structure from Gram-positive bacteria. Similar to previous studies which reported that E. coli lipopolysaccharide can stimulate pEGFR in certain cell types, our data indicate that the role of lipopolysaccharide in EGFR activation may be cell line dependent (4). Likewise, the low level of EGFR activation observed with lipoteichoic acid from E. faecalis (Fig. 6C) suggests that there may be E. faecalis factors other than external cell structures which enhance EGFR activation.
In this study, we focused on the role of H2O2 production by E. faecalis in EGFR activation. Among reactive oxygen species, including superoxide, H2O2, and the hydroxyl radical, H2O2 has the chemical stability necessary to reach significant steady-state concentrations in preclinical models (9, 57). Moreover, H2O2 is small and uncharged, allowing it to diffuse across cell membranes, which makes it an ideal candidate for a signaling molecule (52, 57). For these reasons, we sought to evaluate the possible role of H2O2 production by E. faecalis in EGFR activation and cell proliferation. Our results showed that H2O2 production by E. faecalis increased directly with the MOI (Fig. 5B). Moreover, infection of A253 cells with E. faecalis VF significantly enhanced hydrogen peroxide production, but this was not enhanced in heat-killed bacterium-treated A253 cells or in E. faecalis alone (Fig. 5C). A similar trend was seen with the level of pEGFR (Fig. 2B), suggesting that H2O2 from E. faecalis and cancer cells could synergistically enhance pEGFR. We also found that H2O2 added exogenously or produced by E. faecalis increased activation of EGFR in a manner similar to that for EGF (positive control) (Fig. 5A and andD).D). These results are supported by several studies showing that H2O2 alone can enhance EGFR activation (38, 42, 54, 59). The precise mechanism by which H2O2 activates EGFR is not known, but H2O2 may promote pEGFR activity by releasing EGFR dimerization constraints under physiologic conditions. Such constraints include intracellular inhibitors or structural elements attached to intact plasma membranes (19) as a result of the activation of kinases (21). In an alternative mechanism, H2O2 activation of EGFR is due to the inhibition of phosphatases, which results in separation of tyrosine phosphatases from their respective receptor kinases (16, 21).
Several studies have associated the oral microbiota with head and neck malignancies, although no direct causal association between specific bacteria or mechanisms have been demonstrated (10, 28, 48). Therefore, in addition to E. faecalis, we included three oral streptococci previously associated with OSCC, i.e., S. anginosus, S. salivarius, and S. mitis (50), in our analysis of H2O2 production. Production of H2O2 by E. faecalis or E. faecalis-infected cancer cells was significantly higher than that by any of the three oral streptococcal species or oral streptococcus-infected cancer cells (Fig. 6A and andB),B), in agreement with previous analyses of E. faecalis and H2O2 (11, 30, 31). The high-level in vitro production of H2O2 by E. faecalis and E. faecalis-infected cancer cells was reflected by a greater stimulation of EGFR than that seen with oral streptococci (Fig. 6C). These data suggest that synergistically H2O2-induced EGFR activation by E. faecalis and cancer cells is clinically important and deserves particular attention.
In our analysis of the association between E. faecalis-mediated EGFR activation and cell proliferation, we showed that cell proliferation in oral cell lines (A253 and CAL27), hGFs (primary human gingival fibroblasts), and HUVEC was enhanced in response to E. faecalis infection (Fig. 3) and in A253 cells treated with exogenous H2O2 or EGF (Fig. 4C and and5F).5F). As we expected, the induced activation of EGFR by these stimuli (E. faecalis, H2O2, or EGF) was decreased when cells were preincubated with gefitinib (an EGFR blocker) or catalase (an enzyme capable of decomposing H2O2 to water and oxygen) (Fig. 5C and andD;D; see Fig. S3A and B in the supplemental material), two inhibitors of H2O2-induced EGFR activation in cancer cells (44, 47). These results indicate that E. faecalis-induced cell proliferation depends on H2O2 production and extend the recent finding that H2O2 promotes cell proliferation through EGFR activation (54). Our data suggest that activation of EGFR by E. faecalis may result in an uncontrolled proliferative signal to various cell types inside the tumor microenvironment, contributing to tumor progression.
Approximately 1.2 million cancers per year worldwide can be attributed to microbial infections (48). One of the microorganisms involved, H. pylori, is capable of enhancing cell proliferation through the EGFR signaling pathway (34, 37). The tumorigenic mechanisms employed by H. pylori and E. faecalis appear to share common features. For example, previous studies have reported secretion of IL-8 from cells after exposure to H. pylori (5, 34). This cytokine can induce EGFR activation by way of metalloproteinase-dependent EGFR ligand shedding, resulting in the activation of numerous cell signaling pathways, such as the mitogenic extracellular signal-regulated kinase 1/2 (ERK1/2) signaling cascade (34, 37, 46). Likewise, our study found that E. faecalis can increase cell proliferation and the expression of both IL-8 and EGFR ligands such as TGF-α, while treatment with catalase, gefitinib, GM6001 (matrix metalloproteinase [MMP] inhibitor), or PD98059 (ERK1/2 inhibitor) results in an inhibitive effect (data not shown). These results suggest that H2O2 production by E. faecalis and cancer cells synergistically increases EGFR activation, thereby potentially promoting an autocrine loop contributing to cell proliferation. Our findings are in agreement with previous studies that support a role for H2O2 in the EGFR signaling pathway (18, 58) and demonstrate that H2O2 treatment initiates stimulation of the EGFR and ectodomain shedding of the EGFR and EGF-like ligands via MMP activation. The EGFR activation and upregulation of these ligands then elicit signaling events through EGFR/MEK/ERK phosphorylation, thereby contributing to cell proliferation (18, 42, 47, 54, 57, 58).
Many host and environmental factors can contribute to cancer development through biological shifts in tumoricidal and tumorigenic factors within the tumor microenvironment (39, 40). Antioxidant enzymes such as catalase are deficient in the tissues and erythrocytes of OSCC patients, and the level of oxidative stress is high. The oxidant-antioxidant imbalance results in tissue damage and promotion of various pathological processes, including cancer (20). Other than genetic and environmental factors, bacterial virulence characteristics can generate a substantial amount of oxidants, such as H2O2, that may play an additional role in disease outcome.
In summary, we have shown that the frequency of E. faecalis in oral tumors is significant and is clinically related to EGFR activation. Our in vitro studies support clinical findings relative to the potential role of E. faecalis-mediated EGFR activation in oral cancer. In this pathway, H2O2 production by E. faecalis and cancer cells promotes EGFR activation through direct stimulation and/or metalloproteinase-dependent EGF-like signals (e.g., TGF-α). Activated EGFR may be a significant contributor to MEK/ERK activation, thereby contributing to cell proliferation. Attenuation of E. faecalis-induced EGFR activation and cell proliferation by gefitinib and catalase (Fig. 7), as well as blockage of TGF-α activation by gefitinib, catalase, and MMP and ERK1/2 inhibitors, suggests the potential clinical applications of catalase and EGFR inhibitors in the treatment of E. faecalis-induced oral carcinogenesis.
We are grateful to the study subjects. We acknowledge essential contributions from the staffs of the Department of Dentistry and Maxillofacial Prosthetics at the Roswell Park Cancer Institute and the Department of Periodontics & Endodontics at The State University of New York at Buffalo for tumor and control specimens and patient data collection. We thank Linda Callahan and Maria Jepson for confocal microscope analysis; R. Dziak, J. A. Lemos, and F. A. Scannapieco for providing bacterial strains and primary hGF cells; Johanna Schwingel for reviewing the manuscript; and Benjamin Greene for DNA extraction.
This work was partially supported by an Alliance grant from the Roswell Park Cancer Institute.
Published ahead of print 30 July 2012
Supplemental material for this article may be found at http://iai.asm.org/.