Neutrophils are the primary immune cells that respond to inflammation and combat microbial transgression. To thrive, the bacteria residing in their mammalian host have to withstand the antibactericidal responses of neutrophils. We report that enterobactin (Ent), a catecholate siderophore expressed by Escherichia coli, inhibited PMA-induced generation of reactive oxygen species (ROS) and neutrophil extracellular traps (NETs) in mouse and human neutrophils. Ent also impaired the degranulation of primary granules and inhibited phagocytosis and bactericidal activity of neutrophils, without affecting their migration and chemotaxis. Molecular analysis revealed that Ent can chelate intracellular labile iron that is required for neutrophil oxidative responses. Other siderophores (pyoverdine, ferrichrome, deferoxamine) likewise inhibited ROS and NETs in neutrophils, thus indicating that the chelation of iron may largely explain their inhibitory effects. To counter iron theft by Ent, neutrophils rely on the siderophore-binding protein lipocalin 2 (Lcn2) in a “tug-of-war” for iron. The inhibition of neutrophil ROS and NETs by Ent was augmented in Lcn2-deficient neutrophils compared with wild-type neutrophils but was rescued by the exogenous addition of recombinant Lcn2. Taken together, our findings illustrate the novel concept that microbial siderophore’s iron-scavenging property may serve as an antiradical defense system that neutralizes the immune functions of neutrophils.
This article is featured in In This Issue, p.4181
As the vanguard of the innate immune system, neutrophils are specialized to combat microbial transgression via the generation of reactive oxygen species (ROS) and the phagocytosis or degranulation of antimicrobial mediators, including myeloperoxidase (MPO) and lipocalin 2 (Lcn2). Moreover, neutrophils can release DNA to form weblike neutrophil extracellular traps (NETs) to entrap and promote the killing of invading pathogens (1). The immune pressure exerted by neutrophils curtails the growth of many bacteria, yet the Gram-negative Enterobacteriaceae could bloom and thrive in the inflamed gut despite the intense inflammatory milieu. Archetypal members of the Enterobacteriaceae family, such as Escherichia coli and Salmonella enterica spp., are capable of exploiting host-derived nitrate, reactive nitrogen species, and reactive oxygen species (ROS) to support their anaerobic respiration and growth during inflammation (2–6). However, the mechanisms that allow these bacteria to neutralize the antimicrobial activity of neutrophils are not well understood.
Iron is an essential nutrient for almost all aerobic organisms, yet its participation in redox reactions could aggravate oxidative stress. Accordingly, the host induces hypoferremia or “anemia of inflammation” as an acute-phase response to sequester the labile iron pool (LIP; alias catalytic/reactive iron) (7). This scarcity of iron evokes E. coli and Salmonella to increase their production of enterobactin (Ent), a prototypical catecholate-type siderophore (Greek: “iron carrier”), to acquire iron from the host. The host responds by eliciting the siderophore-binding protein Lcn2 (alias siderocalin or neutrophil gelatinase-associated lipocalin) to sequester Ent (8, 9). Such a “tug-of-war” between Lcn2 and Ent for iron represents one of the most well-characterized host–pathogen interactions that are highly conserved in humans and other mammals (8).
In this study, we report that Ent can inhibit neutrophil ROS generation, NET formation, degranulation, and phagocytosis without affecting neutrophil polarization and chemotaxis. Our molecular analysis revealed that Ent exerts its immunoregulatory effects by chelating the intracellular iron and LIP in neutrophils. The capacity of Ent to impact neutrophil function was notably augmented in Lcn2-deficient neutrophils compared with wild-type (WT) neutrophils but rescued by the addition of exogenous Lcn2. Taken together, our findings illustrate the novel concept that Ent may serve as the microbial anti-radical, which confers a survival advantage to bacteria against neutrophils.
Materials and Methods
Iron-free Ent (E. coli), pyoverdine (Pseudomonas fluorescens), ferrichrome (Ustilago sphaerogena), deferoxamine mesylate salt, 2,3 dihydroxybenzoic acid (2,3-DHBA), PMA, LPS (E. coli 0128: B12), Ca+2 ionophore (A23187), DMSO, ferric chloride, PIPES, Histopaque-1077 and -1119, RPMI 1640, Luria-Bertani (LB) broth, dextran, BSA, paraformaldehyde, saponin, fMLF, and kanamycin were procured from Sigma (St. Louis, MO). Leukotriene B4 (LTB4) was from Cayman Chemical (Ann Arbor, MI). Carrier-free mouse rLcn2 (free from endotoxin, siderophore, and iron) was obtained from Cell Signaling Technology (Danvers, MA). Chrome azurol S (CAS) was purchased from Acros Organics (Geel, Belgium).
Lcn2-knockout (KO) mice generated by Dr. S. Akira (Japan) were obtained via Dr. K. Smith (University of Washington). MPO-deficient (MpoKO) and NADPH oxidase 2 (NOX2)-deficient (Nox2KO) mice were procured from the Jackson Laboratory (Bar Harbor, ME). Mpo/Nox2-double KO (DKO) and peptidyl arginine deiminase 4 (PAD4)-deficient (Pad4KO) mice (10) were generated in The Pennsylvania State University animal facility. All of the mice are on the C57BL/6 (BL6) background and were crossed with BL6 WT mice for >10 generations.
This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health and was approved by the Institutional Animal Care and Use Committee at The Pennsylvania State University (#43376-1).
Human neutrophil (HL-60) cell line
Human neutrophils (HL-60 cells) were cultured in complete RPMI 1640 containing 1.25% DMSO to induce differentiation into neutrophil-like cells, as described previously (11).
Isolation of bone marrow–derived neutrophils and peripheral neutrophils
Bone marrow-derived neutrophils (BMDNs) were isolated from mice (6–8 wk old; sex matched) using the Histopaque gradient method (12). This method yielded >95% pure and >99% viable Ly6G+ neutrophils, as confirmed by flow cytometry. Blood samples from anonymous human donors were obtained from the National Institutes of Health Blood Bank research program. Heparinized whole blood from healthy donors was obtained by venipuncture. Human peripheral neutrophil (PMNs) were isolated from whole blood using dextran sedimentation (3% dextran in 0.9% NaCl), followed by differential centrifugation over Histopaque 1077, as previously described (13, 14). Residual erythrocytes were lysed using hypotonic shock in 0.2% NaCl and neutralization in 1.6% NaCl.
Intracellular ROS assay
Freshly isolated BMDNs/PMNs (2 × 105 cells per well) were preincubated with Ent (0, 2.5, 12.5, or 25 μM) or the indicated siderophores (2,3-DHBA, deferoxamine [DFO], ferrichrome and pyoverdine; 25 μM) for 30 min before treatment with PMA (50 nM), LPS (1.0 μg/ml), or calcium ionophore (5.0 μM) for 3 h at 37°C and 5% CO2. Cells were washed with PBS, stained with 5 μM CellROX Deep Red Reagent (Molecular Probes) for 30 min at 37°C in the dark, and washed twice with PBS. Fluorescence was measured by flow cytometry (Accuri c6; BD Biosciences) and analyzed using FlowJo software (Becton Dickinson). Intracellular ROS was expressed as the fold change of mean fluorescence intensity (MFI) normalized to the controls.
SYTOX Green (5.0 μM) was added to the BMDNs/PMNs at the initial stage of PMA stimulation described above. The plates were read in a SpectraMax Gemini fluorescence microplate reader (Molecular Devices, Sunnyvale, CA) with a filter setting of 485 nm (excitation)/527 nm (emission). The data were presented as fold change of NET formation normalized to the controls (15). To visualize NETs, BMDNs (5 × 105 per well) were seeded in eight-well glass chambered slides (Lab-Tek II) in incomplete RPMI 1640 media, treated with PMA and Ent as described, and incubated for 3 h at 37°C and 5% CO2. Cells were fixed with 3.7% formaldehyde for 10 min and stained with anti-mouse citrullinated histone H3 (H3-Cit) mAb overnight at 4°C (Abcam) and Fluoroshield with DAPI (Sigma) for 5 min. Fluorescent microscopic images were acquired at 40× magnification in different fields using an Olympus FV10i-LIV confocal microscope.
Nonpathogenic WT E. coli (EcK12) and its isogenic mutant of 3-dehydroquinate synthase (∆aroB) were acquired from The Coli Genetic Stock Center (Yale University). An isogenic mutant of ferrienterobactin permease (∆fepA; EcK12), ∆fepA/∆aroB-double mutant (EcK12), and GFP-expressing E. coli (BL21) were gifts from Dr. K. Postle (The Pennsylvania State University). E. coli mutant of ferric uptake regulator (∆fur; EcK12) was a gift from Dr. J. Imlay (University of Illinois).
Stimulation of BMDNs with bacteria
E. coli and its isogenic Ent mutants (∆aroB, ∆fepA, and ∆fur) were grown overnight in LB media (containing 50 μg/ml kanamycin) at 37°C with shaking (200 rpm). Equal numbers of bacterial CFU were adjusted based on their OD at 600 nm, pelleted and resuspended in incomplete RPMI 1640 media, and heat killed at 90°C for 5 min. Heat-killed bacteria were added to PMA-treated BMDNs at a multiplicity of infection (MOI) of 100:1 and incubated for 3 h at 37°C and 5% CO2. Intracellular ROS and NETs were analyzed as described above.
Bacteria survival assay
BMDNs and PMNs (2 × 106 cells per well in 1 ml of incomplete RPMI 1640 media without antibiotic) were plated in six-well plates. Ent (25 μM) was added to the respective wells and incubated for 1 h at 37°C and 5% CO2. Next, cells were infected with E. coli at an MOI of 50:1 and incubated for 3 h at 37°C and 5% CO2. Postinfection, culture supernatants were collected, serially diluted, and plated on LB agar plates. The neutrophil fractions were pelleted, washed, resuspended in 1 ml of 50% PBS containing Triton X-100 (0.1%) to release phagocytosed bacteria, and plated on LB agar plates. Plates were incubated overnight at 37°C for CFU counting.
GFP-expressing E. coli (16) were opsonized by incubating equal volumes of heat-inactivated 10% mouse serum and bacterial suspension in PBS for 1 h at 37°C. BMDNs were treated with Ent (25 μM) at 37°C for 30 min. After three washes in PBS, bacteria were added to the BMDNs at a 50:1 bacteria/neutrophils ratio and incubated at 37°C for 2 h. Cells were fixed with ice-cold 1% formaldehyde in PBS and washed twice with PBS to remove extracellular bacteria. The percentage of phagocytosed bacteria was determined using a BD LSRFortessa flow cytometer and analyzed using FlowJo software.
CAS agar plates and CAS liquid reagent were prepared as described previously (17). Equal CFU of isogenic E. coli Ent mutants were spotted on a CAS agar plate, incubated overnight at 37°C, and monitored for the formation of an orange halo. CAS remains blue in color when complexed with iron, but it turns orange when iron is chelated by other iron chelators. To estimate Ent released by bacteria, the culture supernatants (diluted 4× in 100 μl) were added to the CAS liquid reagent (100 μl) and incubated at room temperature for 20 min, and the change in absorbance was read at 630 nm. Results were estimated using a standard curve generated from iron-free Ent.
MPO and Lcn2 protein levels in the BMDN culture supernatant and cell lysates (prepared in RIPA buffer containing 1× protease inhibitor mixture) were analyzed using DuoSet ELISA kits (R&D Systems, Minneapolis, MN), according to the manufacturer’s instructions.
Lactate dehydrogenase assay
Lactate dehydrogenase (LDH) levels in the BMDN culture supernatant were measured using a kit from Randox (Crumlin, U.K.), according to the manufacturer’s instructions.
BMDN cell lysates were subjected to Western blot using a standard procedure and probed with Abs to H3-Cit (a marker for NETs), total H3, and PAD4 (Abcam).
BMDN viability and apoptosis were measured using the FITC Annexin V Apoptosis Detection Kit (BD Biosciences), according to the manufacturer’s instruction. Results were analyzed using an LSR Fortessa flow cytometer and FlowJo software and presented as the percentage of early (Annexin V+) and late apoptotic (Annexin V+ and propidium iodide+) cells.
Neutrophil polarization analysis
Eight-well chambered coverglass slides (Lab-Tek; Thermo Fisher Scientific) coated with 1% BSA were used. PMNs (2.5 × 105 cells per well) were incubated with vehicle (DMSO) or treated with Ent (100 μM) for 2 h at 37°C. Post-Ent treatment, neutrophils were stimulated for 5 min at 37°C with fMLF (10 nM) or LTB4 (100 nM). Nonadherent cells were washed away, and the remaining adherent cells were fixed with 4% paraformaldehyde for 10 min. Cells were later permeabilized using 0.05% saponin containing PBS, blocked, and stained for F-actin and nuclei using Rhodamine Phalloidin and DAPI (Molecular Probes; Thermo Fisher Scientific). Fluorescence microscopy of samples was performed using a 40× objective lens on a Zeiss observer microscope (Zeiss Axiovert S100 microscope). Multiple fields were acquired for each condition, and a minimum of 100 cells was analyzed for each experiment. PMNs displaying enriched F-actin at the protrusive front were considered polarized, and the percentage of such polarized cells in each condition was quantified visually.
Chemotaxis analysis using under-agarose assay
Six-well plates (In Vitro Scientific) coated with 1% BSA were used. The assay was performed as previously described (13). Briefly, 0.5% SeaKem ME Agarose (Lonza) in 50% each of PBS and RPMI 1640 without phenol red was allowed to solidify in wells. Three 1-mm diameter wells were carved out 2 mm apart. Vehicle (DMSO)- or Ent-treated (100 μM for 2 h at 37°C) cells were stained with CellTracker Red CMTPX Dye (Life Technologies; Thermo Fisher Scientific), according to the manufacturer’s protocol. Cells were washed and resuspended in RPMI 1640 without phenol red, and 10 μl (5 × 104 cells) of vehicle- or Ent-treated cells was added to the outermost wells and allowed to migrate toward the center well containing 10 μl of fMLF (100 nM) or LTB4 (250 nM). Fluorescence microscopy of samples was performed using a 1× objective lens (Zeiss Axiovert S100 microscope), and images of the same field of view were acquired every 45 s for 60 min. Image sequences from under-agarose videos were used to track ≥20 cells per condition in each experiment. Each image sequence was loaded on ImageJ, and cells were followed using the Manual Tracking plugin. The output files containing the x and y coordinates were then used to measure the accumulated distance and directionality using the Chemotaxis plugin available for ImageJ.
Intracellular labile iron assay
BMDNs were incubated for 15 min at 37°C and 5% CO2 with 0.5 μM Calcein-AM (Sigma). The cells were washed twice and treated with Ent or DFO (25 μM) for 2 h. Following washing with PBS, cells were analyzed using an LSRFortessa flow cytometer, and MFI was calculated using FlowJo software. The levels of intracellular labile iron were calculated by subtracting the difference in the MFI (ΔF) before and after treatment with iron chelators (18). Percentage chelation was calculated using the formula (ΔF/control MFI) × 100.
Bleomycin-detectable labile iron assay
The assay was performed as outlined by Burkitt et al. (19), with modifications described previously (20). In principle, ascorbic acid converts Fe+3 into Fe+2, which, in combination with bleomycin, causes DNA breaks. Ent or DFO was incubated with Fe+3 (0.005–200 μM) for 10 min prior to the addition of the assay mixture (20) and incubated at 37°C for 2 h. Bleomycin/iron-induced damage to DNA corresponds to loss of the ethidium bromide–enhanced fluorescent signal (excitation 510 nm; emission 590 nm) and is presented as the percentage of labile iron.
All experiments were performed in triplicate, and data presented are representative of three independent experiments. All values in the results are expressed as mean ± SEM. Statistical significance between two groups was analyzed using an unpaired, two-tailed t test. Data from more than two groups were compared using one-way ANOVA, followed by the Dunnett post hoc test (when comparing the mean of each column with the mean of the control column) or the Tukey multiple-comparison test (when comparing the mean of each column with the mean of every other column). The p values <0.05 were considered statistically significant and are denoted as *p < 0.05, **p < 0.01, and ***p < 0.001. All statistical analyses were performed with GraphPad Prism 7.0 software (GraphPad, La Jolla, CA).
Ent inhibits neutrophil ROS generation
Neutrophils are one of the professional immune cells that mediate bacterial killing during infection. Our recent finding that E. coli–derived Ent potently inhibits the activity of purified MPO from human neutrophils (21) raises the possibility that Ent may also modulate other neutrophil functions. To advance toward this hypothesis, we isolated BMDNs from BL6 mice and treated them with Ent. Although we observed a substantial level of intracellular ROS in BMDNs, treatment with Ent inhibited ROS generation in a dose- and time-dependent manner (Fig. 1A, 1B). We repeated the experiments using BMDNs stimulated with PMA, which activates NADPH oxidase and upregulates their ROS generation. Ent likewise mitigated PMA-induced ROS generation in BMDNs in a dose- and time-dependent manner (Fig. 1C–E). Moreover, Ent was effective in inhibiting the generation of ROS in BMDNs activated by an array of potent neutrophil stimulators, including LPS and Ca+2 ionophore (Fig. 1F). Similar results were observed when experiments were repeated using a human neutrophil-like cell line (nHL-60) and human PMNs (Fig. 1G–I).
To assess the effect of Ent-mediated inhibition on ROS generation, we next measured the levels of ROS in Ent-treated WT BMDNs in comparison with BMDNs deficient in NOX2 and MPO as negative controls. As anticipated, the genetic loss of NOX2 (22, 23) abrogated the capacity of BMDNs to upregulate ROS in response to PMA stimulation (Fig. 1J). Presumably as the result of having intact NOX2, MPO-deficient BMDNs could still generate ROS to some extent in response to PMA, but this was abrogated upon treatment with Ent (Fig. 1J). Intriguingly, the loss of detectable ROS in Ent-treated WT and MPO-deficient BMDNs was comparable to that in MPO/NOX2 double-deficient BMDNs, which are ROS deficient (Fig. 1J); such an outcome further affirms the efficacy of Ent in inhibiting the generation of ROS in neutrophils.
To rule out whether Ent-mediated inhibition of neutrophil ROS is due to increased cell death, we assessed the markers of early and late apoptosis in Ent-treated BMDNs. However, the percentage of apoptotic cells was comparable between Ent- and vehicle-treated BMDNs (Fig. 1K), indicating that Ent-mediated effects are not due to altered cell viability. Next, we assayed the cell-free supernatant for LDH, whose release indicates cytosolic leakage and cell death (24). The reduced LDH release from Ent + PMA–treated BMDNs affirms the efficacy of Ent in inhibiting neutrophil activation in comparison with PMA-treated BMDNs (Fig. 1L).
Ent inhibits NET formation and degranulation
Because ROS are indispensable for the production of NETs (25), we envisioned that Ent-mediated inhibition of ROS generation may impede the formation of NETs as well. Indeed, we observed a 1.7–2-fold reduction in the release of NETs from Ent + PMA–treated BMDNs compared with BMDNs treated with PMA alone (Fig. 2A, 2B). The inhibition of NET generation by Ent correlates with the decrease in the levels of NET markers (H3-Cit and PAD4) (26) in Ent + PMA–treated BMDNs compared with PMA-treated controls (Fig. 2C). Quantification of extracellular DNA with SYTOX Green indicates that WT BMDNs could generate NETs, even at the basal level, compared with NET-deficient BMDNs isolated from Pad4KO mice (10, 27) as negative control (Fig. 2D). However, the NETs generated at the basal level in WT BMDNs was prevented by treatment with Ent (Fig. 2E, 2F). The formation of NETs increased substantially upon stimulation with PMA, yet Ent was effective in inhibiting PMA-induced NETs in a dose- and time-dependent manner (Fig. 2G, 2H). Similar inhibition of NET formation by Ent was observed when experiments were repeated with LPS- or Ca+2-ionophore treated BMDNs (Fig. 2I), PMA-treated human PMNs (Fig. 2J, 2K), and the nHL-60 cell line (Fig. 2L).
To investigate whether Ent could inhibit neutrophil degranulation, we assayed the cell-free supernatants from BMDNs for granular markers, such as MPO and Lcn2, which correlate with the release of primary and secondary granules, respectively (28, 29). The addition of Ent to control and PMA-treated BMDNs blocked the release of their primary granules, as reflected by the reduced levels of MPO in the supernatant (Fig. 2M). However, Ent did not impair the capacity of BMDNs to degranulate their secondary granules or Lcn2 in response to PMA (Fig. 2N), thus suggesting the specificity of Ent in inhibiting the release of primary, but not secondary, granules.
We next evaluated the effect of Ent on the acquisition of neutrophil polarity and chemotaxis. Human PMNs were stimulated with subsaturating concentrations of two independent and potent chemoattractants, fMLF: and LTB4. We found that pretreatment with Ent, even at 100 μM, had little or no impact on the ability of PMNs to polarize in response to fMLF or LTB4 (Supplemental Fig. 1A, 1B). Furthermore, Ent-treated PMNs exhibited a chemotactic response to fMLF and LTB4 that was comparable to vehicle-treated PMNs. Indeed, the total distance migrated and directionality of migration were unchanged by Ent treatment (Supplemental Fig. 1C–E). These findings establish that Ent has no significant impact on the ability of PMNs to polarize and chemotaxis in response to fMLF and LTB4.
Ent protects bacteria against the antibactericidal activity of neutrophils
To address whether Ent-producing bacteria can inhibit neutrophil ROS and NETs, we took advantage of E. coli isogenic mutants that are deficient in Ent (ΔaroB) or overexpress Ent (ΔfepA and Δfur). We first determined that ΔfepA and Δfur mutants produced more Ent than WT E.coli based on the extent of halo formation on the CAS agar plate. By using the more sensitive CAS liquid assay, we noted that Δfur secreted a greater amount of Ent into the culture supernatant, followed by ΔfepA, WT, and ΔaroB strains (Fig. 3B). The discrepancy in the levels of siderophores detected on CAS agar (ΔfepA > Δfur) and CAS liquid (Δfur > ΔfepA) is likely due to technical differences (e.g., growth conditions) in the assays; nevertheless, it is sufficient to establish that ΔfepA and Δfur produced more Ent than WT and ΔaroB strains. Although the bacteria themselves induce NET formation, we observed that their level of Ent production was inversely correlated with their capacity to inhibit basal ROS production from BMDNs (Fig. 3C, 3D). When PMA-stimulated BMDNs were treated with heat-killed Ent mutant bacteria, ROS and NETs responses were inhibited in the order of Δfur > ΔfepA > WT > ΔaroB, which correlates with the levels of Ent production in these strains (Fig. 3E–G). The bacterial culture supernatants, which contain Ent, were also effective in inhibiting PMA-induced ROS and NETs from BMDNs in the order of Δfur > ΔfepA > WT > ΔaroB (Fig. 3H, 3I). Our observations that WT E. coli and isogenic mutants were not as effective as their culture supernatants in inhibiting the BMDNs’ PMA-induced ROS could be explained by the possibility that Ent accumulated more in the culture supernatant. However, the unanticipated reduction in PMA-induced ROS by ΔaroB supernatant suggests the possibility that other unknown bacterial metabolites may also play a role in alleviating the oxidative responses.
The inhibition of neutrophil ROS and NET responses by Ent may confer a survival advantage to Ent-producing bacteria. Consistent with this notion, we observed that BMDNs treated with exogenous Ent were impaired in mediating the killing of WT E. coli; viable bacteria recovered from Ent-treated BMDNs were ∼7.3-fold higher in the assay supernatant and ∼4-fold higher in the lysed neutrophil fraction compared with bacteria recovered from vehicle-treated BMDNs (Fig. 4A). Similar results were observed in human PMNs (Fig. 4B). To assess whether endogenously produced Ent also offers similar protection, we next examined the survivability of ΔfepA and ΔfepA/ΔaroB E. coli mutants against BMDNs. The Ent-overexpressing ΔfepA mutant was substantially protected from the bactericidal activity of BMDNs compared with the Ent-deficient ΔfepA/ΔaroB mutant, which had poor survival (Fig. 4C). In addition to promoting the survivability of extracellular and phagocytosed bacteria, Ent inhibited the capacity of BMDNs to phagocytose bacteria by ∼1.3-fold compared with vehicle-treated control (Fig. 4D, 4E).
Ent-mediated inhibition of neutrophil ROS and NETs is counterregulated by Lcn2
Given that Ent did not impair the release of Lcn2 from neutrophils, we envisioned that the secreted Lcn2 may be able to neutralize the effects of Ent. Although Ent inhibited PMA-induced ROS by 1.8-fold in WT BMDNs relative to PMA-only control, the extent of inhibition increased to 3.6-fold in Lcn2-deficient BMDNs (Fig. 5A). Furthermore, Ent inhibited PMA-induced NETs by 2.4-fold in WT BMDNs and 4.3-fold in Lcn2-deficient BMDNs compared with their corresponding PMA-only treated BMDNs (Fig. 5B). These observations suggest that Ent is nearly twice as effective at inhibiting ROS and NETs in Lcn2-deficient neutrophils than in WT neutrophils. Because Lcn2 is known to bind Ent in a stoichiometry ratio of 1:1, we investigated whether adding rLcn2 to Ent in such a ratio would protect Lcn2-deficient neutrophils from being inhibited. Treatment with exogenous rLcn2 indeed rescued Lcn2-deficient neutrophils from Ent and restored their capacity to generate basal and PMA-induced ROS and NETs (Fig. 5C–F).
Ent inhibits neutrophil functions by chelating cytosolic labile iron
The intricate interaction between Lcn2 and Ent reported in this article is highly evocative of their tug-of-war for iron, which compelled us to investigate whether alterations in cellular iron may provide the mechanistic explanation for how ROS and NETs are regulated by Lcn2 and Ent. Using the cytochemical calcein-AM method (30–32), we demonstrate that Lcn2-deficient BMDNs contain ∼2-fold more cytosolic LIP compared with WT BMDNs (Fig. 6A); such a result could potentially explain our observations that Lcn2-deficient BMDNs generated more PMA-induced ROS than did WT BMDNs (Fig. 5A).
The LIP assay used in this study requires the addition of DFO to chelate iron from calcein, thus allowing the detection of displaced calcein as a measurement of cellular LIP (30–32). Based on this principle, we envision that the calcein-AM assay can be modified to mechanistically examine the effects of Ent on BMDNs (i.e., by using Ent as the iron chelator instead of DFO). The use of Ent in the modified assay yielded higher MFI readings compared with DFO (Fig. 6B); the higher sensitivity of Ent may be due to its superior affinity and kinetics in chelating iron compared with DFO, which is known to be a slow iron chelator (33). Although the level of cytosolic LIP in Lcn2-deficient BMDNs was still detected as being 2-fold higher than in WT BMDNs in this modified assay, our use of Ent provided several key mechanistic insights. First, the readout itself indicates that Ent can permeate the neutrophil membrane and remain intact in the cytosolic compartment. Second, the readout also implicates that Ent retains its iron-chelating property and potently chelates cytosolic LIP in the neutrophils. The majority of neutrophil oxidative responses require LIP (34); thus, its reduced bioavailability due to chelation by Ent could be a potential mechanism by which Ent hijacks neutrophil functions.
To ascertain whether Ent could quench the reactivity of LIP, we modified the bleomycin-based LIP-detection assay that was originally developed to quantify LIP in biological samples (19). In our modified assay system, we instead prepared a series of samples with known concentrations of labile iron and tested whether Ent could quench their detection. Intriguingly, 50 μM of Ent completely prevented the detection of labile iron ranging from 0.005 to 50 μM but not when the concentration of labile iron exceeded 50 μM (Fig. 6C). Such findings coincide with the well-accepted notion that Ent binds iron in a stoichiometry ratio of 1:1 (35); it also affirms that Ent can effectively quench the reactivity of bound iron but not the excess unbound iron.
To address whether chelation of labile iron could be the underlying mechanism of Ent-mediated inhibition of neutrophil functions, we next tested other siderophores for their ability to inhibit PMA-induced ROS and NETs in BMDNs. A weak iron chelator, such as 2,3-DHBA (a monomeric Ent), modestly mitigated PMA-induced neutrophil ROS and NETs (Fig. 6D, 6E). Other iron chelators, such as ferrichrome (fungal siderophore from U. sphaerogena) and the bacterial siderophores pyoverdine (from P. fluorescens) and DFO (from Streptomyces pilosus), mitigated PMA-induced ROS and NETs to a greater extent (Fig. 6D, 6E). However, our data suggest that Ent mediated a superior inhibition in the order Ent > pyoverdine > DFO > ferrichrome (Fig. 6D, 6E), which is consistent with Ent having an unmatched affinity (Kd of 10−49 M) for iron (36).
The chelation of LIP by Ent may impede neutrophil ROS responses that are dependent on labile iron. To test whether Lcn2 could prevent Ent from chelating cytosolic LIP, we next performed the modified calcein-AM assay using Ent and rLcn2. The addition of Ent modestly increased the MFI reading in WT BMDNs (indicating that Ent was chelating iron from the intracellular calcein–iron complex), but the increase was more pronounced in Lcn2-deficient BMDNs (Fig. 6F). The addition of rLcn2 remarkably suppresses the MFI reading of the elevated labile iron in Lcn2-deficient BMDNs to levels that are comparable to those in WT BMDNs (Fig. 6F).
The iron-acquisition system of E. coli encodes the enzymes for the synthesis of Ent, the archetype siderophore for Gram-negative bacteria. However, the production of Ent is metabolically expensive, given the number of enzymes that are needed to generate Ent from its aromatic amino acid precursors (37). Accordingly, it was presumed that Ent may exert a plethora of key physiological functions that could account for its high metabolic cost (38), aside from acquiring iron to support bacterial growth. Such a notion was demonstrated in studies in which Ent was shown to be fundamental for E. coli to establish colonization in the mammalian gut (39), develop mature biofilms (40, 41), alleviate oxidative stress (42, 43), and neutralize the antimicrobial activity of MPO (21). The multifaceted properties of Ent are further exemplified by our findings that it could inhibit an array of neutrophil functions, including ROS generation, NET formation, degranulation, phagocytosis, and bacterial killing activity. In addition to conferring a survival advantage to E. coli, Ent-mediated inhibition of mucosal ROS generation may be of advantage to other gut commensals in achieving stable colonization (44). Such a notion is exemplified by the intricate Ent-dependent cross-talk among the microbial community (41, 45, 46) and the fact that various non–Ent-producing bacteria have, nonetheless, evolved to express the receptor for Ent (47).
Various biological processes depend upon the reactivity of labile iron, a redox engine, to catalyze chemical reactions that are essential to life. The oxidative responses of neutrophils are particularly iron dependent (48), in addition to the fact that many of their proinflammatory mediators (i.e., MPO, NADPH oxidase, and iNOS) are heme proteins. Investigating how Ent hijacks neutrophil activity revealed that the hydrophobic Ent could permeate cellular membranes and chelate the cytosolic labile iron (free/catalytic iron) of neutrophils. Our observations that other siderophores (i.e., pyoverdine, ferrichrome, DFO) likewise inhibited neutrophil ROS and NETs further suggest that the chelation of iron may explain their inhibitory effects. The inhibition of NET formation by siderophores is rather unprecedented and may allude to the possible requirement of labile iron during the formation of NETs. Analogous to the ability of Streptococcus spp. to escape NETs by expressing DNases (49, 50), the production of siderophore by E. coli and other bacteria may be a key mechanism that allows them to evade NET-mediated killing. We considered that the loss of NET responses may be secondary to the inhibition of ROS production or MPO (21) by Ent, in light of the contention that the NET response requires the activity of NADPH oxidase and MPO (22, 23, 25, 51). Nevertheless, further studies are needed to investigate whether Ent could directly modulate NET formation.
The inhibitory effects of Ent on neutrophils can be counterregulated by Lcn2 that sequesters bacterial siderophores with a high affinity (Kd = 0.41 nM) (35). Lcn2 is expressed by a variety of immune and nonimmune cells, although neutrophils are the richest source of Lcn2 (52). Although Lcn2 is regarded as the primary defense against the iron theft by Ent (35), our group reported previously that Lcn2 also protects the activity of MPO from being inhibited by Ent (21). In the current study, we further highlight the importance of Lcn2 in shielding the neutrophils and their functions from being compromised by Ent. However, the tug-of-war between Lcn2 and Ent may exert selective immune pressure on E. coli to evolve or acquire the ability to produce Lcn2-resistant or stealth siderophores (e.g., pyoverdine) (53). In such a regard, the inability of Lcn2 to neutralize the stealth siderophores may potentially prevent neutrophil functions from effectively mediating the clearance of pathogens (54).
The bacterial siderophore from S. pilosus, DFO, has been widely used clinically to treat patients with iron overload and hereditary hemochromatosis. However, the use of DFO can potentially promote bacterial and fungal infection (55–57); such side effects are postulated to be due to the delivery of iron-bound DFO to the pathogens (58). Our findings suggest that the siderophore-induced loss of neutrophil activity might also promote a state of “functional neutropenia,” thus exacerbating susceptibility to infection. The development of iron chelators without the side effects of inhibiting neutrophil functions could be further explored in future studies on optimizing siderophore-based therapy in treating iron-related disorders.
There may be circumstances in which neutrophil inactivation would be desirable, as in the case of inflammatory disorders whereby dysregulated neutrophils could inflict collateral damage to bystander tissues. Such a possibility is suggested by the clinical use of the probiotic E. coli Nissle 1917 to treat gut inflammation (59); its efficacy in preventing the colonization of enteropathogens is purported to be due to its unique capacity to express four types of siderophores: Ent, aerobactin, yersiniabactin, and salmochelin (60). Although E. coli Nissle 1917 was reported to exhibit anti-inflammatory properties, its siderophore production may explain, in part, its ability to immunoregulate host inflammatory responses. The ability of siderophore to quench the redox potential of labile iron may also safeguard host and bacteria from the iron-induced generation of free radicals via Fenton reaction. The emerging concept that bacterial siderophores are not merely iron chelators, but are also immunoregulatory metabolites, could certainly be investigated in future studies to determine their therapeutic potential in treating inflammatory diseases.
We thank Dr. A. Catharine Ross and Dr. Na Xiong for their critical input. We acknowledge the Huck Institutes of the Life Sciences for seed funding through the Flow Cytometry and Microscope Facilities. We thank and dedicate this study to our long-standing collaborator, Dr. Niels Borregaard, for helpful discussions and support of this study.
This work was supported by National Institutes of Health Grant R01 DK097865 (to M.V.-K.). V.S. is supported by a Crohn's and Colitis Foundation of America Research Fellowship Award.
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived neutrophil
chrome azurol S
2,3 dihydroxybenzoic acid
citrullinated histone H3
labile iron pool
mean fluorescence intensity
multiplicity of infection
neutrophil extracellular trap
NADPH oxidase 2
peptidyl arginine deaminase 4
reactive oxygen species
The authors have no financial conflicts of interest.