Neutrophil extracellular traps (NETs) are an essential component of the antimicrobial repertoire and represent an effective means by which neutrophils capture, contain, and kill microorganisms. However, the uncontrolled or excessive liberation of NETs also damages surrounding cells and can contribute to disease pathophysiology. Alterations in the gut microbiota, as well as the presence of local and systemic markers of inflammation, are strongly associated with the manifestation of a spectrum of intestinal disorders, including chronic inflammatory bowel disease. Although probiotics exert beneficial effects on gut homeostasis, their direct effect on neutrophils, which are abundant in the setting of intestinal inflammation, remains unclear. In this study, we investigated the effects of nonpathogenic, enteropathogenic, and probiotic bacteria on the dynamics of NET formation. Using murine bone marrow–derived neutrophils and the neutrophil-differentiated human myeloid cell line d.HL-60, we demonstrate for the first time, to our knowledge, that probiotic Lactobacillus rhamnosus strain GG inhibits both PMA- and Staphylococcus aureus–induced formation of NETs. Moreover, probiotic L. rhamnosus strain GG had potent antioxidative activity: dampening reactive oxygen species production and phagocytic capacity of the neutrophils while protecting against cell cytotoxicity. Within the milieu of the gut, this represents a novel mechanism by which probiotics can locally dampen innate immune responses and confer desensitization toward luminal Ags.
Neutrophils are one of the first cells recruited to sites of microbial challenge and are elegantly equipped with an array of antimicrobial and proteolytic enzymes, reactive oxygen species (ROS), and anti-inflammatory molecules to combat infection (1, 2). The recent discovery of neutrophil extracellular traps (NETs) (3) has expanded the view of how neutrophils capture, contain, and kill microorganisms. Brinkmann et al. (3) first described that, following either microbial or pharmacologic activation, neutrophils release web-like extracellular structures (composed primarily of decondensed chromatin) that are highly decorated with antimicrobial and granule proteases. In a given niche, the formation of NETs, termed NETosis, is sufficient to trap bacteria (3, 4) and prevent further dissemination (5, 6). Moreover, its importance is underscored by evidence of persistent and recurrent infections in individuals with neutrophils that are not able to form NETs (7, 8). However, there is emerging evidence suggesting that uncontrolled or excessive NETosis, and the associated liberation of cell-free DNA and degradative proteases, damages surrounding cells and contributes to disease pathophysiology (9–11).
The intestinal tract serves as a unique interface between cells of the immune system and the Ag-rich, luminal environment. A delicate balancing act is maintained to ensure that robust host inflammatory responses are mounted against invasive pathogenic microorganisms and yet immune tolerance conferred to commensal inhabitants (12). Disturbance of the gut microbiota, termed “dysbiosis,” results in pathophysiological consequences (13).
Supplementation with beneficial microbes (also referred to as probiotics) can delay or prevent the onset of experimental colitis and, in humans, induce and maintain remission in individuals with ulcerative colitis (14, 15). In premature infants, the risk for developing necrotizing enterocolitis (aberrant inflammation of the intestine in response to initial bacterial colonization) is significantly reduced by supplementation with probiotics (16). Among their arsenal of properties, such as a strain-specific ability to induce colonization resistance, maintain intestinal epithelial integrity, increase protective goblet cell–derived mucus production, and stimulation of the host immune system, probiotics also can produce soluble antimicrobial and other modulatory factors that serve to counteract dysbiosis and protect against intestinal inflammation (17).
In the context of the intestinal tract, little is known about the ability of gut microorganisms to induce NETs. Given that neutrophils are abundant in the setting of colonic inflammation and the close proximity of these granulocytes to luminal and mucosa-associated microorganisms, this study sought to determine the effects of commensal, probiotic, and enteropathogenic bacteria on the formation of NETs. To our knowledge, we demonstrate for the first time that probiotic Lactobacillus rhamnosus strain GG (LGG) inhibits both pharmacologic and pathogen-induced NETosis. Moreover, neutrophils exposed to LGG have a reduced capacity to undergo oxidative burst and phagocytosis.
Materials and Methods
All animal work was approved by the Hospital for Sick Children’s Animal Research Ethics Board (approval number 22577) and adhered to the Canadian Council on Animal Care guidelines for humane animal use.
Probiotic LGG (ATCC 53103), and L. rhamnosus strain R0011 (R0011; Lallemand Health Solutions, Montreal, QC, Canada), and Escherichia coli strain Nissle 1917 (N.1917) (18) were grown for 16 h at 37°C in static nonaerated deMan–Rogosa–Sharpe or Luria–Bertani broth (Difco Laboratories, Detroit, MI), respectively. Aliquots of bacterial cultures subsequently were used to inoculate fresh broth and then cultured for an additional 3 h prior to use. Commensal E. coli strains HB101 (serotype O:rough) (19) and F18 (serotype O:rough:K1:H5) (20), enterohemorrhagic strains EDL933 (serotype O157:H7) (18) and CL56 (serotype O157:H7) (19), prototype translocating strain C25 (serotype O rough nonmotile) (21), and adherent invasive LF82 (serotype O83:H1 (22), a generous gift from Dr. A. Darfeuille-Michaud, UMR 1071 Inserm, Université d'Auvergne, Clermont-Ferrand, France), were grown in Penassay broth (Difco Laboratories), as above. Staphylococcus aureus (ATCC 25923) was grown in tryptic soy broth (Difco Laboratories). Following centrifugation (4000 × g, 5 min) and three washes in HBSS (Invitrogen), bacterial strains were normalized to a density ∼108 CFU/ml. In some experiments, bacteria were fixed in 4% formaldehyde, washed, and resuspended in HBSS. Loss of viability was confirmed by plating formalin-fixed bacteria onto agar plates (37°C for 48 h).
Bone marrow–derived and cultured neutrophils
Murine bone marrow–derived neutrophils (BMDNs) were harvested from male C57BL/6 mice, as described previously (23). Briefly, the tibia and femur were isolated and flushed with MEM alpha (Invitrogen), and the resultant cell suspension was centrifuged at 400 × g for 10 min. The cell pellet was resuspended in PBS (pH 7.4; without Ca2+/Mg2+; Life Technologies), layered on top of a Percoll density gradient (80/65%/55%; Sigma Aldrich), and centrifuged at 1000 × g for 30 min. BMDNs at the 80/65% interface were collected and washed in PBS, and contaminating RBCs were removed by hypotonic lysis. BMDNs were resuspended in HBSS (with Ca2+/Mg2+; Invitrogen) and normalized to 1 × 106 cells/ml.
The human promyelocytic cell line HL-60 (ATCC CCL-240) was maintained in IMDM supplemented with 20% heat-inactivated FBS (both from Invitrogen). To differentiate HL-60 cells into a mature neutrophil phenotype (d.HL-60), 2 × 105 cells/ml were incubated with 1% DMSO for 5 d and subsequently normalized to a concentration of 1 × 106 cells/ml in HBSS (with Ca2+/Mg2+; Invitrogen). For all experiments, 1 × 105 BMDNs or d.HL-60 cells were used.
Quantification and visualization of NETs
BMDNs and d.HL-60 cells were preincubated with either live or formalin-fixed LGG (1 h, 37°C, 5% CO2) and then activated to form NETs with the phorbol ester PMA (100 nM, 3 h), S. aureus (107 CFU, 3 h), or H2O2 (100 nM, 3 h). Extracellular DNA was stained with 5 μM SYTOX Green (Invitrogen), a fluorescent membrane-impermeable DNA dye, and fluorescence was quantified using a microplate reader equipped with filters to detect excitation/emission maxima: 485/520 nm (Victor X3; PerkinElmer). Wells containing both bacteria and neutrophils were normalized against the fluorescence emission from wells containing bacteria only. To visualize NETs, BMDNs and d.HL-60 cells were plated onto 12-mm poly-l-lysine–coated coverslips (BD) and activated, in the absence or presence of LGG, with PMA (as above). In some experiments, commensal, probiotic, and enterohemorrhagic E. coli strains (multiplicity of infection [MOI] 100, 37°C, 3 h) alone were used to assess NET activation ability. Cells were fixed with 4% formaldehyde (15–45 min, room temperature) and washed with PBS. For confocal fluorescence analysis, coverslips were first blocked with 3% BSA for 1 h and then stained with Abs against histone H3 (D1H2 XP rabbit mAb, 1:100 dilution; Cell Signaling), elastase (rabbit polyclonal Ab, 1:100; Abcam), and DNA (DAPI; Invitrogen). Tetramethylrhodamine isothiocyanate–conjugated goat anti-rabbit secondary Ab (1:400; Abcam) was used to visualize histone H3, whereas Alexa Fluor 488 goat anti-rabbit secondary Ab (1:1000; Invitrogen) was used to visualize elastase staining. Cells were mounted with Prolong Gold AntiFade reagent (Invitrogen), and confocal fluorescence imaging was performed on a Quorum WaveFX spinning disc confocal system mounted on an Olympus IX81 microscope. Images were acquired using 60×, 10×, and 40/0.75× objectives, Hamamatsu C9100-13 EM-CCD camera, and 405-, 491-, and 561-nm laser lines. Image acquisition was performed using PerkinElmer Volocity 6.2.1 software. For analysis, five random fields from each experimental treatment were captured with a 20× objective, and the total area occupied by DNA and histone H3–stained cells was quantified using ImageJ. Results are expressed as percentage cell area (per field of view) and were normalized per 100 cells.
ROS and superoxide production by BMDNs and d.HL-60 cells was quantified by luminol- and lucigenin-ECL, respectively. Briefly, 105 BMDNs and d.HL-60 cells were pretreated with live or formalin-fixed LGG (MOI 1, 10, and 100; 1 h, 37°C) or with bacterial strains N.1917, HB101, EDL 933, CL56, or LF82 (MOI 100) and then activated with PMA (100 nM, 1 h). To measure ROS levels, luminol (50 μM) and HRP (1.2 U/ml) were added to the reaction. In some experiments, BMDNs and d.HL-60 cells were activated with H2O2 (10 nM), and the resulting ROS levels were measured. To quantify PMA-induced superoxide production, lucigenin (5 μM) was added to the reaction mix. The resulting chemiluminescence was measured using a microplate reader (Victor X3; Perkin Elmer).
Lactate dehydrogenase (LDH) release from PMA- or S. aureus–treated BMDNs and d.HL-60 cells was quantified using the Cytoscan LDH Cytotoxicity Assay Kit (G-Biosciences). Briefly, BMDNs and d.HL-60 cells were treated with either live or formalin-fixed LGG (1 h, 37°C, 5% CO2) and then activated with PMA (100 nM) or S. aureus (107 CFU) for 1 h (37°C, 5% CO2). Oxidation and formation of formazan, which is produced in proportion to the level of LDH in cell supernatants, were measured at 490 nm in a microplate reader (Victor X3; PerkinElmer). LDH release was calculated as a percentage of maximum compared with lysed BMDNs or d.HL-60 cells.
Phagocytic capacity of d.HL-60 cells was quantified using the Vybrant Phagocytosis Assay Kit (Molecular Probes). Cells were preincubated with LGG, R0011, or E. coli strain HB101 (MOI 10, 100), for 1 h at 37°C, and allowed to adhere to the bottom of 0.01% poly-l-lysine–coated 96-well plates (PerkinElmer). Supernatants were aspirated and replaced with a suspension of fluorescein-labeled E. coli (K-12 strain) in HBSS (2 h at 37°C, 5% CO2). E. coli that were not engulfed by d.HL-60 cells during this time were aspirated, and the remaining extracellular fluorescence was quenched with trypan blue (0.25 mg/ml). Intracellular fluorescence was quantified using a microplate reader equipped with filters to detect excitation/emission maxima of 480/520 nm (Victor X3; PerkinElmer), and the results are presented as the percentage of phagocytosis compared with untreated d.HL-60 cells.
Data are presented as mean ± SEM. Comparisons among groups of data were made using one-way ANOVA followed by Dunnett post hoc analysis. An associated probability (p < 0.05) was considered significant.
LGG inhibits the formation of NETs
To investigate the efficacy of commensal, probiotic, and enterohemorrhagic bacteria to modulate the dynamics of NET formation, two neutrophil cell types were used: primary murine BMDNs and the human neutrophil–differentiated promyelocytic cell line d.HL-60. Two well-studied, gold standard activators of NETs, PMA (Fig. 1A) and the Gram-positive bacterium S. aureus (Fig. 1B, 1C), both induced the formation of NETs, as quantified by the detection of diphenylene iodinium– and DNase-sensitive extracellular DNA (SYTOX green). Although PMA induced the release of 10% of total DNA (above DNase baseline) from d.HL-60 cells and 20% of BMDNs, respectively, S. aureus promoted the release ∼20–60% of total DNA.
In the absence of stimuli, probiotic LGG lacked the ability to induce NETs. Interestingly, however, treatment of BMDNs and d.HL-60 cells with LGG (MOI 100) prior to PMA or S. aureus challenge potently abrogated the formation of NETs. The requirement for LGG to be live and viable was confirmed with results showing that formalin-fixed LGG had no effect on the capacity of BMDNs or d.HL-60 cells to release nuclear DNA. The absence of NETs was not due to local degradation by LGG cell wall–associated nucleases (Supplemental Fig. 1).
Immunofluorescence visualization of NETs (Fig. 2A–P), using the markers elastase and histone H3, colocalized with DNA, demonstrates that live, but not formalin-fixed, LGG inhibits the formation of these protease-rich DNA structures. In agreement with the measurement of externalized DNA using SYTOX green (above), quantification of microscopy images confirmed the inhibitory capacity of LGG on PMA-induced NETosis (Fig. 2Q).
Strain-specific ability of probiotic and enteropathogenic bacteria to activate NETs
The ability of commensal, probiotic, and enterohemorrhagic bacterial strains to induce NETosis also was assessed (Fig. 3). Although commensal E. coli strains HB101 and F18 had no effect on the capacity of BMDNs to form NETs (Fig. 3A, 3A1, 3B, 3B1), the probiotic N.1917 (Fig. 3C, 3C1) was an effective NET activator. R0011 (Fig. 3D, 3D1) did not induce NETs. Translocating C25 and enterohemorrhagic E. coli strains CL56 (but not EDL933) and adherent-invasive LF82 (Fig. 3E, 3E1, 3F, 3F1, 3G, 3G1, 3H, 3H1) also induced NET formation (quantification of microscopy images shown in Fig. 3I).
To assess their NET-inhibitory capacity, BMDNs were preincubated with the various bacterial strains and subsequently activated with PMA. Interestingly, of the bacterial strains that did not induce NETs, only LGG and R0011 had the capacity to further inhibit NET formation (Fig. 3J).
Protection against S. aureus–induced cell cytotoxicity
The induction of NETosis culminates in a time-dependent loss in cell viability, which is independent of apoptosis or necrosis (8). Although BMDNs and d.HL-60 cells challenged with PMA (60 min) induced minimal release of LDH (Fig. 4A), cells challenged with S. aureus (Fig. 4B) released significantly more LDH than did controls. This finding was most prominent in d.HL-60 cells. In line with the inhibitory effects of LGG on the formation of NETs, pretreatment of d.HL-60 cells with the probiotic protected against S. aureus–induced cell cytotoxicity, whereas formalin-fixed LGG did not rescue the cells. Of note, lactic acid production by probiotic LGG did not markedly change the pH (∼7) of the reaction mixture during the incubation period with BMDNs or d.HL-60 cells (data not shown).
LGG dampens neutrophil ROS production and phagocytosis
The ability of neutrophils to generate ROS is crucial for the formation of NETs (8). Cell activation with PMA mobilized the production of ROS (such as superoxide, H2O2, and peroxynitrite), a process that was inhibited in the presence of LGG in both BMDNs and d.HL-60 cells (Fig. 5A). Using a probe specific for superoxide, we show that live, but not formalin-fixed, LGG also inhibited PMA-induced superoxide production (Fig. 5B). In contrast, probiotic N.1917, commensal HB101, and pathogenic strains EDL933, CL56, and LF82 had no effects (Fig. 5C).
Given the role of ROS in canonical neutrophil functions, including phagocytosis, we next assessed the ability of d.HL-60 cells to phagocytose fluorescein-labeled E. coli (K-12 strain). Compared with controls without probiotic exposure, d.HL-60 cells pretreated with LGG (MOI 100) inhibited phagocytosis by ∼70% (Fig. 5D). Commensal E. coli strain HB101 and R0011 had no effect on the capacity of d.HL-60 cells to undergo phagocytosis.
LGG has potent antioxidative activity and protects against H2O2-induced NET activation
To determine whether exogenous administration of ROS could reverse the effects of LGG, BMDNs were pretreated with either the live or formalin-fixed probiotic. However, subsequent activation with H2O2 failed to induce NETs in cells that had been pretreated with live LGG (Fig. 6A). Moreover, detectable ROS levels were significantly reduced (Fig. 6B), indicating that probiotic LGG has potent antioxidative properties.
The intestinal tract serves as a reservoir for trillions of microorganisms, and the ability to discriminate and mount an appropriate host response to these microbes, whether commensal, opportunistic, or pathogenic, is essential for the maintenance of gut homeostasis. Neutrophils are abundant in the setting of intestinal inflammation, and this poses a challenging environment given the tremendous microbial burden. Despite evidence of gut pathogens, including Shigella flexneri (3) and S. aureus (8, 24), activating neutrophils to induce NETs, few studies have explored the implications of NETs on the gut microenvironment (25).
To our knowledge, we demonstrate for the first time that probiotic LGG inhibits both pharmacologic and microbial (PMA and S. aureus) induction of NETs. Moreover, LGG dampens ROS production and the phagocytic capacity of neutrophils, perhaps conferring a degree of hyporesponsiveness. Given the presence of NETs in inflamed intestine (3, 25), it is possible that some of the beneficial effects of LGG are attributable to its action on local neutrophils. Neutrophils produce high levels of an adenosine precursor, which, in the intestinal tract, stimulates passive water flux, providing the physiological basis of secretory diarrhea. Although NETs are localized to the intestine during experimental shigellosis (3), supplementation with probiotic lactobacilli protects against Shigella dysentariae 1–induced diarrhea in rats (26). In the clinical setting, LGG was shown to significantly reduce the duration of acute diarrheal illness in children (27).
Although NETs are essential for robust host responses to infection (7), their formation in a given microenvironment, such as the inflamed intestine, may have unintended consequences. Thus, the composition of NETs and their presence in various autoimmune and inflammatory disorders have fueled speculation regarding their roles in disease pathophysiology. Individuals with systemic lupus erythematosus and other vasculitides carry autoantibodies against neutrophil proteins (including elastase, myeloperoxidase, and lactoferrin), DNA, and histones. Indeed, such immune complexes activate NETs in vitro, and it was suggested that presentation of these Ags during NETosis likely precipitates inflammation in susceptible individuals (10). Approximately 60% of ulcerative colitis patients and 25% of Crohn’s disease patients carry autoantibodies against neutrophil proteins (28), although whether these activate NETs remains to be determined. DNase 1 expression, which is required for the degradation of NETs, is also reduced in inflammatory bowel disease (29). Interestingly, long pentraxin 3, a soluble innate immune receptor of which autoantibodies are present in patients with systemic lupus erythematosus, is localized to NETs and has been detected in crypt abscesses in colonic biopsies taken from patients with active ulcerative colitis (25).
The host response varies, depending on whether Gram-positive or Gram-negative bacteria are encountered (30). Furthermore, for a given bacterial species, strain-specific immunomodulation of host cells has been reported (30–33). The activation of NETs by E. coli was shown to compromise intestinal epithelial integrity (34). Given the dynamic shift in intestinal bacterial communities during health and disease, we also assessed the capacity of prototypical commensal (nonpathogenic), probiotic, and enterohemorrhagic bacteria to initiate NETosis. We show that, although commensal E. coli strains F18 and HB101 had no effect on NET formation, probiotic N.1917 was an effective NET activator. In adults, N.1917 has been used successfully for the management of Crohn’s disease and ulcerative colitis (35). However, proinflammatory effects, including potentiation of TLR4 and TLR5 activity (36), also were documented. The effects of various probiotic strains varied in their ability to inhibit NETosis, highlighting their often distinct underlying mechanisms of action (17). Proteomic analysis of LGG identified ≥50 proteins secreted into culture supernatants (37), including the soluble protein p40, which protects against cytokine-induced epithelial damage (38) and barrier dysfunction (39). Moreover, specific delivery of p40 to the mouse colon protects against colitis and epithelial cell apoptosis in an epidermal growth factor receptor–dependent manner (40). Mechanistically, we speculate that the NET-inhibitory effects of probiotic LGG may stem from an unidentified secreted factor.
R0011, like LGG, also inhibited phorbol ester–induced NETosis. However, the lack of effect on other neutrophil parameters, such as phagocytosis, suggests only a partial overlap in underlying mechanisms of action. Accordingly, although both L. rhamnosus strains secrete soluble proteins that promote intestinal epithelial homeostasis, R0011 lacks the functional pili that are present on LGG (41).
NETs can be induced by a variety of stimuli, including phorbol esters, fungi, parasites, microbial components (e.g., LPS), and ROS (42). Activation of TLR9 and platelet TLR4 induces potent NETosis (4, 6). The important role of the TLR-signaling pathway was confirmed by observations showing that neutrophils from MyD88−/− mice do not form NETs (6). Fuchs et al. (8) elegantly demonstrated time- and ROS-dependent externalization of chromatin and DNA in response to both PMA and S. aureus, two mechanistically distinct activators, which culminated in cell death independent of either apoptosis or necrosis. However, it is noteworthy that early induction of NETs by S. aureus occurs independently of ROS (24). In this study, the ability of LGG to inhibit both PKC-dependent PMA- and S. aureus–induced NETosis, while concomitantly dampening ROS production, suggests that a common downstream target of the NETosis pathway is disrupted. LGG protects against radiation-induced intestinal damage in a TLR-2/MyD88–dependent manner (43), the same pathway that is essential for mediating host responses to S. aureus (6). Interestingly, the ability of LGG to inhibit both exogenously administered ROS levels and H2O2-induced NET formation supports the antioxidative capacity of this probiotic. Indeed, some of the reported protective effects of LGG in the intestine are attributed to its modulation of oxidative parameters, including the induction of superoxide dismutase during Giardia intestinalis infection (44).
In the current study, we found that the effects of LGG on neutrophil function were also manifest by a reduced capacity of the cells to phagocytose E. coli. One of the mechanisms that pathogens use to evade neutrophils includes prevention of engulfment, through physical barriers, interference with opsonization, or inhibiting rearrangements of the F-actin cytoskeleton. It is also noteworthy that neutrophils treated with cytochalasin D, an inhibitor of actin polymerization and phagocytosis, are still able to form NETs (3).
The ability of LGG to inhibit S. aureus–induced NETs also translated to protection against cell cytotoxicity. S. aureus secretes pore-forming toxins that lyse neutrophils, including leukocidin, which was shown previously to induce the formation of NETs (24). Whether LGG secretes bacteriocins, which have antimicrobial activity against S. aureus, or directly disrupts the production of S. aureus–secreted toxins remains to be determined. Our results raise intriguing questions about the usefulness of probiotics, including LGG, in the context of infection. Indeed, Gan et al. (45) reported that Lactobacillus fermentum RC-14 and its secreted biosurfactant inhibit S. aureus infection and bacterial adherence to surgical implants.
In summary, to our knowledge, this is the first report identifying a probiotic bacterium, LGG, as an inhibitor of NETosis. We postulate that neutrophil hyporesponsiveness in a microenvironment rich in antigenic stimuli protects against the tissue-damaging properties of NETs. It is noteworthy that the inhibition of NETosis also may act as a double-edged sword, given the ability of NETs to ensnare pathogens and protect against infection. However, alongside observations that LGG was able to protect against S. aureus–induced cytotoxicity, this study highlights the dynamic interaction between beneficial bacteria and neutrophils and supports the usefulness of probiotics as gut-protective factors.
This work was supported by operating grants from the Canadian Institutes of Health Research (MOP-89894 and IOP-92890). P.M.S. is the recipient of a Canada Research Chair in Gastrointestinal Disease.
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived neutrophil
Lactobacillus rhamnosus strain GG
multiplicity of infection
Escherichia coli strain Nissle 1917
neutrophil extracellular trap
Lactobacillus rhamnosus strain R0011
reactive oxygen species.
The authors have no financial conflicts of interest.