Multiple mechanisms of NADPH oxidase inhibition by type A and type B Francisella tularensis

Francisella tularensis uses both pre- and post-assembly mechanisms to inhibit NADPH oxidase activity at its own phagosome and throughout infected human neutrophils.

Abstract

Ft is a facultative intracellular pathogen that infects many cell types, including neutrophils. In previous work, we demonstrated that the type B Ft strain LVS disrupts NADPH oxidase activity throughout human neutrophils, but how this is achieved is incompletely defined. Here, we used several type A and type B strains to demonstrate that Ft-mediated NADPH oxidase inhibition is more complex than appreciated previously. We confirm that phagosomes containing Ft opsonized with AS exclude flavocytochrome b₅₅₈ and extend previous results to show that soluble phox proteins were also affected, as indicated by diminished phosphorylation of p47^phox and other PKC substrates. However, a different mechanism accounts for the ability of Ft to inhibit neutrophil activation by formyl peptides, Staphylococcus aureus, OpZ, and phorbol esters. In this case, enzyme targeting and assembly were normal, and impaired superoxide production was characterized by sustained membrane accumulation of dysfunctional NADPH oxidase complexes. A similar post-assembly inhibition mechanism also diminished the ability of anti-Ft IS to confer neutrophil activation and bacterial killing, consistent with the limited role for antibodies in host defense during tularemia. Studies of mutants that we generated in the type A Ft strain Schu S4 demonstrate that the regulatory factor fevR is essential for NADPH oxidase inhibition, whereas iglI and iglJ, candidate secretion system effectors, and the acid phosphatase acpA are not. As Ft uses multiple mechanisms to block neutrophil NADPH oxidase activity, our data strongly suggest that this is a central aspect of virulence.

Introduction

Tularemia is a multisystem plague-like illness caused by the facultative intracellular bacterium Ft. The two major subspecies of Ft differ in geographic distribution and virulence. Highly virulent Ft subspecies tularensis (type A) is found in North America, whereas the less-virulent Ft subspecies holarctica (type B) is distributed throughout the Northern hemisphere [1]. Reservoirs of Ft relevant to human infection include rabbits, rodents, and ticks. Infection typically occurs by direct contact with infected animals, their carcasses, or arthropod vectors but can also result from inhalation of contaminated dust or ingestion of contaminated water. Inhalation of as few as 10 type A organisms is sufficient to cause a rapidly progressing and potentially fatal pneumonic infection, whereas infections with type B strains can be severe but rarely result in death [1]. An attenuated LVS of Ft holarctica was isolated years ago but is not licensed for use, in part, because its mechanism of attenuation is unknown [2]. Nevertheless, LVS retains many features of virulent Ft in vitro and for this reason, is studied widely [1, 2]. Ft virulence factors include an atypical LPS that exhibits little or no endotoxic activity but confers serum resistance together with other surface sugars and a duplicated region of the genome, called the FPI, which is essential for bacterial growth in macrophages and virulence in vivo [1].

PMN are key players in innate defense that use toxic ROS and cationic peptides to kill ingested microbes rapidly. Pivotal to oxidative host defense is the NADPH oxidase complex, a multi-component enzyme that catalyzes the conversion of molecular oxygen into superoxide anions [3]. In resting PMN, the enzyme is unassembled and inactive with subunits segregated in the membranes of specific granules and in the cytosol. During phagocytosis or when cells encounter soluble stimuli, the integral membrane subunits of the oxidase (gp91^phox/p22^phox heterodimers, also called flavocytochrome b₅₅₈) accumulate on forming phagosomes or at the cell surface, respectively, and phosphorylation of p47^phox triggers en bloc membrane translocation of the soluble subunits (p40^phox, p47^phox, and p67^phox) [3]. Rac2 translocates independently and is also essential for activity [3]. Persons with inherited defects in genes encoding any one of the NADPH oxidase components have chronic granulomatous disease and suffer from repeated, often life-threatening bacterial and fungal infections [4, 5], underscoring the role of this enzyme in innate immunity.

Results of several studies suggest that Ft-PMN interactions contribute to the pathogenesis of pneumonic tularemia. Neutrophils accumulate in alveoli, accounting for up to 50% of infected cells by Day 3, yet cannot control the infection [6–10]. Ft triggers PMN influx by up-regulating matrix metalloproteinase-9, which generates PMN chemoattractants via cleavage of the ECM [11]. When this enzyme is absent, neutrophil influx is nearly ablated, bacterial burden is diminished, and mice are able to survive an otherwise fatal type A Ft infection [11]. We and others [9, 12, 13] have shown that Ft is not killed efficiently by neutrophils in vivo or in vitro, but what accounts for this host defense defect is only beginning to be defined. When opsonized with normal serum, neither virulent type B Ft strains nor LVS activate PMN for production of ROS [12–15]. LVS phagosomes exclude flavocytochrome b₅₅₈, and within minutes, the ability of these cells to be activated by PMA or OpZ is impaired [13]. Blockade of the respiratory burst is followed by phagosome escape and bacterial persistence in the neutrophil cytosol [13], but how this is achieved is unknown, and similar studies of highly virulent type A Ft have not been performed.

We undertook this study to define further the mechanisms used by Ft to disrupt NADPH oxidase activity in human neutrophils. Our data indicate that Ft subspecies tularensis and holarctica disrupt neutrophil function. The results we obtained are noteworthy, as they demonstrate for the first time that Ft uses a multifaceted strategy to ensure blockade of the respiratory burst. During uptake of Ft opsonized with AS, NADPH oxidase assembly is disrupted at 2 points, as indicated by defects in flavocytochrome b₅₅₈ targeting and diminished phosphorylation of multiple PKC substrates, including p47^phox. In addition, an unusual post-assembly inhibition mechanism accounts for the ability of Ft to impair cell activation by all heterologous stimuli tested, despite their diverse mechanisms of action, and to diminish the ability of anti-Ft IS to stimulate ROS production and enhance bacterial killing. Studies of mutants that we generated in strain Schu S4 show that evasion of oxidative host defense requires the regulatory factor fevR but not the acid phosphatase acpA and also suggest that genes within the FevR regulon required for NADPH oxidase inhibition can be distinguished from those required for phagosome escape, such as iglI and iglJ. These data significantly advance our understanding of Ft-PMN interactions at the molecular level, reinforce the notion that blockade of neutrophil function is an important aspect of virulence, and confirm that observations made using the related organism Francisella novicida may not be applicable to type A and type B Ft [16].

MATERIALS AND METHODS

Materials

Tryptic soy broth and agar and cysteine heart agar were from Becton Dickinson (Sparks, MD, USA). Defibrinated sheep blood was from Remel (Lenexa, KS, USA), and Mueller Hinton agar was from Acumedia (Lansing, MI, USA). Endotoxin-free HBSS and PBS were from Mediatech, Inc. (Herndon, VA, USA). Endotoxin-free, Hepes-buffered RPMI 1640 (with and without phenol red) was from Lonza (Walkersville, MD, USA). Mouse anti-Ft LPS mAb T14 was from Novus Biologicals (Littleton, CO, USA). Rabbit anti-Ft antiserum was from BD Diagnostics (Sparks, MD, USA). IS, of known titer from 3 persons vaccinated with LVS, were obtained from Dr. Jeannine Petersen at the CDC (Ft. Collins, CO, USA). An antibody specific for serine-phosphorylated, active PKC substrates was from Cell Signaling Technologies (Danvers, MA, USA). Mouse mAb specific for gp91^phox (54.1) and p22^phox (44.1) [17, 18] were obtained from Dr. Algirdas Jesaitis (Montana State University, Bozeman, MT, USA). Rabbit antisera specific for p47^phox and p67^phox [19] were obtained from Dr. William Nauseef (University of Iowa, Iowa City, IA, USA). Rabbit anti-p40^phox mAb were from Epitomics (Burlingame, CA, USA). A mouse mAb specific for active Rac was from NewEast Biosciences (Malvern, PA, USA). Affinity-purified FITC- or rhodamine-conjugated donkey anti-rabbit and goat anti-mouse F(ab′)₂ secondary antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA, USA). [³²P]Orthophosphoric acid (285.6 Ci/mg) was from Perkin-Elmer (Waltham, MA, USA). Pierce SuperSignal West Pico ECL substrate kits were from Thermo Scientific (Rockford, IL, USA). Additional reagents were obtained from Sigma-Aldrich (St. Louis, MO, USA), unless indicated otherwise.

Neutrophil isolation

Heparinized venous blood was obtained from healthy adult volunteers in accordance with a protocol approved by the Institutional Review Board for Human Subjects at the University of Iowa. PMN were isolated using dextran sedimentation and density gradient separation on Ficoll-Hypaque, followed by hypotonic lysis of erythrocytes [17]. PMN (98% purity) were resuspended in HBSS without divalent cations, counted, and then diluted into appropriate media as indicated.

Cultivation of WT bacteria

Bacterial strains used in this study are shown in Table 1 and include virulent Ft subspecies tularensis (type A) strains Schu S4 [20, 21], TI0902 [22], MA00-2987 [23], and WY96-3418 [21, 23], as well as virulent Ft subspecies holarctica (type B) strains 1547-57 [24] and 1623-36 and the attenuated type B strain LVS [25]. Ft were grown routinely on CHAB for 48 h at 37°C in 5% CO₂. All manipulations of virulent type A and type B Ft were performed in a licensed BSL-3 facility with select agent approval and in accordance with all CDC and NIH regulatory and safety guidelines. S. aureus strain ALC1435 and H. pylori strain NCTC 11637 were cultured as described previously [26]. In all cases, harvested bacteria were washed 3 times with HBSS containing divalent cations prior to opsonization or use in experiments. Killed LVS was prepared by incubating washed bacteria in 10% formalin for 1 h. FK bacteria were washed 3 times with HBSS containing divalent cations prior to opsonization, and aliquots were plated on CHAB to confirm sterility.

TABLE 1.

Bacterial Strains Used in This Study

Strain	Description	Source	Ref.
LVS	Attenuated Ft holarctica, type B	ATCC 29684	[25]
1547–57	Virulent Ft holarctica, type B	Michael Apicella (U. Iowa)	[24]
1623–36	Virulent Ft holarctica, type B	Michael Apicella (U. Iowa)
Schu S4	Virulent Ft tularensis, type A	Michael Apicella (U. Iowa)	[20, 21]
acpA576	Schu S4 acpA mutant	This study
acpA576c	Schu S4 acpA mutant complemented with a WT copy of acpA in trans on pSL128	This study
fevR	Schu S4 fevR mutant	This study
fevRc	Schu S4 fevR mutant complemented with WT fevR in trans	This study
iglI153	Schu S4 iglI mutant	This study
iglI153c	Schu S4 iglI mutant complemented with a WT gene expressed in trans using pSL148	This study
iglJ189	Schu S4 iglJ mutant	This study
iglJ189c	Schu S4 iglJ mutant complemented with a WT gene expressed in trans using pSL148	This study
MA00-2987	Virulent Ft tularensis, type A	Michael Apicella (U. Iowa)	[23]
TI0902	Virulent Ft tularensis, type A	Michael Apicella (U. Iowa)	[22]
WY96-3418	Virulent Ft tularensis, type A	Michael Apicella (U. Iowa)	[21, 23]
ALC 1435	Staphylococcus aureus	Jon K. Femling (U. Iowa)	[26]
NCTC 11637	Helicobacter pylori (type strain)	ATCC 43504	[26]

Opsonization

Unless otherwise indicated, yeast zymosan particles, S. aureus, and all Ft strains were opsonized by incubation in 50% fresh human AS for 30 min at 37°C and then washed 3 times in HBSS prior to infection of PMN. Where noted, LVS was opsonized in a similar manner with the indicated concentrations of anti-LVS IS diluted in AS to ensure an adequate supply of active complement. H. pylori were not opsonized.

Respiratory burst

The rate and extent of ROS production by resting and stimulated PMN were measured using the luminol-ECL assay as described [13, 26–28]. Microtiter wells contained 1 × 10⁶ PMN loaded with 50 μM luminol in phenol red-free RPMI 1640 and 4% HSA. Stimuli were added to triplicate samples to achieve 200 nM PMA, 10 μM fMLF, OpZ at a MOI of 3:1, S. aureus at a MOI 5:1, or Ft at a MOI 50:1 (which results in 5–10 Ft/cell), and CL was measured every 30 s for 50–90 min at 37°C using NovoStar (for BSL-2 experiments) or LumiStar (for BSL-3 experiments) luminometers (both from BMG LabTech Inc., Durham, NC, USA). For BSL-2 experiments, samples were plated in Perkin-Elmer white OptiPlate 96 microtiter dishes, and measurements were taken from above, whereas studies in the BSL-3 laboratory used black, clear-bottom plates that were covered with lids, and CL was measured through the bottom of the wells. Because of these differences, data obtained from the 2 machines are similar but cannot be compared directly.

To assess the ability of Ft to disrupt PMN activation by other agonists, sequential stimulus assays were performed as we described [13]. In brief, PMN in microtiter wells were treated at Time 0 with buffer alone or infected with Ft to achieve 5–10 bacteria/cell, and luminol CL was measured at 30 s intervals for 10–15 min, after which, secondary stimuli were added (200 nM PMA, 10 μM fMLF, 50 μM diC8, 3:1 OpZ, 5:1 opsonized S. aureus, or 15:1 unopsonized H. pylori), and CL was measured for an additional 45–60 min at 37°C. Cells that were left in buffer throughout the experiment or cells that were stimulated with PMA, fMLF, diC8, OpZ, S. aureus, or H. pylori after an initial exposure to buffer served as negative and positive controls.

NBT staining was used as we described [13, 26, 28] to detect superoxide accumulation directly inside phagosomes. For sequential stimulus experiments, media containing 1 mg/ml NBT were added to PMN along with the second stimulus, and cells were fixed and processed for microscopy after 60 min at 37°C.

Neutrophil fractionation and cell-free NADPH oxidase assays

Diisopropyl fluorophosphate-treated PMN (5×10⁷/ml) in relaxation buffer (10 mM PIPES, 100 mM KCl, 3 mM NaCl, 3.4 mM MgCl₂), supplemented with 1 mM (Na)₂ATP and 1 mM PMSF, were disrupted by nitrogen cavitation (350 pounds/square inch, 20 min, 4°C) and collected drop-wise into EGTA (1 mM final concentration) [29, 30]. Nuclei and intact cells were removed by centrifugation, and the resulting postnuclear supernatants were fractionated on 2-layer, discontinuous Percoll density gradients, and β (specific and gelatinase granules), γ (plasma membrane and secretory vesicles), and cytosolic fractions were collected [29, 30]. Granules and membranes in the β and γ fractions were pelleted and washed to remove Percoll and then resuspended in relaxation buffer [30]. For cell-free NAPDH oxidase assays, 10⁷ cell equivalents of cytosol and β or γ fractions were mixed in the presence of 50 μM GTPγS, 55 μg/ml (Na)₂ATP, and 100 μM cytochrome c, with and without 50 μg/ml SOD [29]. After addition of 100 μM arachidonate, samples were incubated for 4–5 min at 37°C, and reactions were initiated by addition of 160 μM NADPH. Superoxide production was measured in a kinetic assay over a 10-min period as the amphiphile-dependent and SOD-inhibitable reduction of cytochrome c at A₅₀₀ using a SpectraMax Gemini microplate spectrophotometer (Molecular Devices Inc., Sunnyvale, CA, USA) [29]. Where noted, 5 × 10⁸ LVS were added to each reaction along with arachidonate.

Measurement of superoxide generated by xanthine oxidase

Activity of xanthine oxidase was measured in a microtiter assay using hypoxanthine as the substrate, and superoxide production was quantified as SOD-inhibitable reduction of cytochrome c as described [31]. Where noted, reactions also contained 5 × 10⁸ LVS.

Preparation of CM

Triplicate aliquots of PMN (5×10⁶ each) were diluted in 1 ml RPMI 1640 and incubated for 12 h at 37°C. This PMN CM was collected and sterile-filtered (0.2 μm) prior to use. Media conditioned by a combination of PMN and LVS were prepared in a similar manner, except that PMN were mixed with opsonized LVS at Time 0. To generate LVS-CM, 2.5 × 10⁹ bacteria were incubated for 12 h in RPMI 1640 prior to sterile filtration. Aliquots of each CM were plated on CHAB to confirm sterility. Effects of each CM on PMN oxidant production were tested using the luminol assay as described above. PMN treated with ″naïve″ RPMI 1640 or cells infected with opsonized LVS were used as controls.

Phagocytosis and intracellular killing

Resting PMN (5×10⁶/ml) in RPMI 1640 containing 10% AS were infected in triplicate with the indicated strains of opsonized Ft, as we described previously [13]. After 30 min at 37°C, PMN were washed 3 times (by 3 min, 1200 rpm centrifugations) to remove free bacteria and then resuspended in fresh, serum-free medium. After a total of 0.5–8 h at 37°C, infected PMN were lysed in 1% saponin, and live intracellular bacteria were quantified by plating diluted lysates on CHAB for enumeration of CFU [13, 28]. To quantify effects of LVS on phagocytic killing of S. aureus, resting PMN or cells that were preinfected with LVS for 15 min [13] were incubated with S. aureus (MOI 2.5:1) for 5 min at 37°C. Uningested bacteria were removed by washing (as described above), and after an additional 2–4 h at 37°C, cells were lysed in 0.5% saponin, and serial dilutions were plated for enumeration of CFU.

Immunofluorescence and confocal microscopy

Our established methods were used to assess the extent of NADPH oxidase assembly on Ft or OpZ phagosomes following synchronized phagocytosis [13, 26, 28, 32]. After fixation, permeabilization, and blocking, cells were stained with the antibodies listed above to detect NADPH oxidase components gp91^phox, p22^phox, p47^phox, p67^phox, p40^phox, or active Rac, alone or together with Ft. F(ab′)₂ secondary antibodies were conjugated to FITC or rhodamine. Samples were examined using an Axioplan2 fluorescence microscope or an LSM-510 confocal microscope (both from Carl Zeiss Inc., Thornwood, NY, USA). All studies were performed in triplicate, and at least 50 cells/condition and time-point were evaluated in each experiment.

In vivo phosphorylation of p47^phox and detection of active PKC substrates

In vivo phosphorylation of p47^phox in ³²Pi-labeled PMN was assessed as we described [17] with minor modifications. In brief, PMN in phosphate-free DMEM were labeled for 1 h at 25°C with 0.5 mCi/ml [³²P]orthophosphoric acid. After warming to 37°C, cells were left untreated or were stimulated for 10 min with 200 nM PMA, S. aureus (MOI 10:1), or live or killed LVS (MOI 100:1). For sequential stimulus experiments, labeled cells were treated with buffer alone or were infected with LVS for 15 min at 37°C prior to stimulation with PMA or OpZ (MOI 3:1) for 15 or 60 min. Each set of cells was lysed in ice-cold RIPA buffer containing protease and phosphatase inhibitors (0.09 TIU/ml aprotinin, 0.5 mg/ml leupeptin, 1 mM PMSF, 3 μM diisopropyl fluorophosphate, 50 mM NaF, 10 mM sodium pyrophosphate, and 0.2 mM sodium vanadate), and p47^phox was immunoprecipitated using anti-p47^phox antiserum and protein A sepharose beads. Immune complexes were resolved by 8% SDS-PAGE, and incorporation of ³²Pi into p47^phox was detected and quantified using a PhosphorImager (Typhoon 9410, Molecular Dynamics, Sunnyvale, CA, USA). In addition, immunoblots prepared from lysates of unlabeled cells were probed with mAb 2261 to detect active serine-phosphorylated PKC substrates [(R/K)X(S*)(Hyd)(R/K)]. Bands were detected using secondary antibodies conjugated to HRP and Pierce SuperSignal West Pico ECL reagents.

Generation of Schu S4 mutants and complemented strains

Mutants were generated in strain Schu S4 by Group II intron-mediated inactivation [33]. Specifically, acpA, iglI, and iglJ sequences were examined for insertion sites using the TargeTron algorithm (Sigma-Aldrich), and 2 sites within each gene (at nucleotides 42 and 576 of acpA, nucleotides 153 and 311 of iglI, and nucleotides 99 and 189 of iglJ) were selected. The Group II introns were retargeted by PCR amplifying the intron template in the TargetTron kit (Sigma-Aldrich, #TA0100) using primers generated by the algorithm, according to the manufacturer's instructions. In each case, the resulting fragments were cloned into the XhoI and BsrGI sites of pKEK1140, and plasmids were introduced into Schu S4 by cryotransformation [34]. Bacteria containing the retargeting plasmids were selected at 30°C on agar plates supplemented with 25 μg/ml kanamycin. After selection of putative mutants, single colonies were analyzed by PCR across the insertion site using primers CCAACCTTGCTCACAAAATCATGAG and GCAATCATCGTAGTATGGGTTAGG for acpA, CTCAAGATAAAGTAAATCTGATTAGTAATATGC and GCATTTGTGCAAAATTTGCATCC for iglI, and CCAGAATGACCCCGTAGAAAAAATTTC and CTTGATAAATACTCTGTAATTTAGTGAATGGTTTG for iglJ. Mutants containing the desired insertions were passaged twice to cure the plasmid, and single clones, designated acpA576, iglI153, and iglJ189, were studied further. To create complementation plasmids, the WT acpA or iglI-iglJ coding sequences were PCR-amplified from the Schu S4 chromosome, such that BamHI and SalI sites flanked each amplicon. Fragments were cloned into pBB103 with expression of the introduced genes driven by the Francisella groES promoter [33]. After verification by restriction digest, plasmids were introduced into acpA576, iglI153, or iglJ189 by cryotransformation to generate the trans-complemented strains acpA576c, iglI153c, and iglJ189c. A Schu S4 fevR mutant and complemented strain (fevRc) were also generated as we described previously for strain LVS [33].

For neutrophil activation assays, WT Schu S4 and all mutant strains were grown on CHAB, and all complemented strains were grown on CHAB containing 25 μg/ml spectinomycin. WT, mutant, and trans-complemented strains were tested in parallel for their ability to block neutrophil activation using the luminol CL assay as described above. Acid phosphatase activity associated with WT Schu S4, the acpA576 mutant, and acpA576c was assessed by overlaying bacteria that had been grown on Mueller Hinton agar for 48 h with a filter soaked in 0.2 M sodium acetate buffer (pH 6.0) + 4 μg/ml 5-bromo-4-chloro-3-indolyl phosphate in 20% dimethylformamide. Images were obtained 2 h after filter overlay.

Statistics

Statistical significance (P<0.05) was assessed using the f test, unpaired Student's t test, and/or one-way ANOVA where appropriate [13].

RESULTS

Virulent strains of Ft subspecies tularensis and holarctica do not activate human neutrophils

In marked contrast to S. aureus and Neisseria meningitidis, infection of human neutrophils with the attenuated Ft subspecies holarctica strain LVS does not trigger a respiratory burst, and this is a result, at least in part, of disruption of NADPH oxidase assembly on forming phagosomes [13, 17]. Here, we compared for the first time the ability of 4 virulent strains of Ft subspecies tularensis (Schu S4, MA00-2987, TI0902, and WY96-3418) and 2 virulent strains of Ft subspecies holarctica (1547-57 and 1623-36; Table 1) to activate human neutrophils using the luminol ECL assay. S. aureus and PMA were used as positive controls, and LVS was used as a negative control. The data in Fig. 1A demonstrate that in contrast to S. aureus but like LVS [13], neither Schu S4 nor strain 1547-57 triggered an oxidative burst in PMN (Fig. 1A). Similar data were obtained for all virulent Ft strains tested, and pooled data are shown in Fig. 1B, as total ROS normalized to neutrophils stimulated with 200 nM PMA. Confocal analyses revealed that Schu S4 and 1547-57 phagosomes also resembled LVS compartments [13] in their failure to acquire gp91^phox and p22^phox (Fig. 1C). In contrast, we routinely observe accumulation of NADPH oxidase components on at least 80% of S. aureus phagosomes [13]. Moreover, none of the Ft strains tested was efficiently killed by PMN, and at least 70% of ingested organisms remained viable, as judged by measurements of CFU obtained at 2 (Fig. 1D) or 4–8 hpi (not shown). These data demonstrate that disruption of the respiratory burst and evasion of intracellular killing are attributes shared by virulent Ft strains and LVS.

Figure 1. Neutrophils are not activated by virulent strains of Ft.

(A) Kinetics of ROS generation by uninfected PMN (UN) or cells stimulated with S. aureus (Sa), or Ft strains LVS, 1547-57, or Schu S4. Data indicate luminol CL in counts/s (cps) and are the average ± sem of triplicate samples from 1 experiment representative of 3. (B) Total ROS generated by uninfected PMN or cells infected for 90 min with LVS or virulent Ft strains 1547-57, 1623-36, MA00-2987, Schu S4, TI0902, or WY96-3418. Data are the average ± sem from 3 independent experiments performed in triplicate and are normalized to cells stimulated with 200 nM PMA. (C) Confocal images of PMN infected for 15 min with LVS, 1547-57, or Schu S4. Images show gp91/p22^phox in green and bacteria in red. Arrows indicate Ft phagosomes. (D) PMN were infected with LVS, 1547-57, or Schu S4, and viable intracellular bacteria were quantified at 30 min and 2 hpi by plating cell lysates for measurement of CFU. Data are the mean ± sem for triplicate samples from 1 experiment representative of 3.

Effects of Ft on phosphorylation of p47^phox and other PKC substrates

In previous work, we demonstrated that neither flavocytochrome b₅₅₈ nor soluble NADPH oxidase components p47^phox and p67^phox are present on LVS phagosomes [13]. Exclusion of gp91^phox /p22^phox heterodimers from Ft compartments (Fig. 1C) deprives soluble NADPH oxidase components of their membrane-docking site and is sufficient to prevent superoxide production [3, 19, 35] but does not preclude effects of Ft on other aspects of the enzyme assembly and activation process. Phosphorylation of multiple serine residues in p47^phox by PKC (and/or other kinases) is an essential, early signal required for NADPH oxidase assembly that triggers membrane translocation of p47/67/40^phox complexes [17, 36–38]. To assess the extent of p47^phox phosphorylation in vivo, we labeled neutrophils with [³²P]orthophosphate prior to stimulation with LVS or known NADPH oxidase agonists and quantified p47^phox phosphorylation after immunoprecipitation and SDS-PAGE. Our data confirm that p47^phox phosphorylation is low in resting neutrophils [17], yet increased 5- to 7-fold in cells stimulated for 10 min with S. aureus or PMA (Fig. 2A and B) [17, 39]. Similar to OpZ or N. meningitidis [17], p47^phox phosphorylation increased 3.2-fold in cells infected with FK LVS but only 2-fold in response to live LVS (Fig. 2A and B), whereas total p47^phox was unchanged (Fig. 2A and B). These data indicate that relative to other stimuli tested, p47^phox phosphorylation triggered by live Ft is diminished.

Figure 2. In vivo phosphorylation of p47 ^phox and other PKC substrates in LVS-infected PMN.

(A) PMN labeled with ³²P-orthophosphate were left untreated or were stimulated for 10 min at 37°C with 200 nM PMA, S. aureus (MOI 10:1), LVS (MOI 100:1), or FK-LVS (MOI 100:1). p47^phox was immunoprecipitated from cell lysates, and ³²Pi was detected by PhosphorImager analysis after SDS-PAGE. Total p47^phox was detected by immunoblotting and served as a loading control. (B) Quantitation of p47^phox phosphorylation. Data are the mean ± sem from 3 independent experiments (except for S. aureus; n=1) and are normalized to the uninfected PMN control. (C) PMN were left untreated or were stimulated for 10 min at 37°C with PMA, S. aureus, FK, or live LVS as in A. Immunoblots of normalized cell lysates (30 μg protein/lane) were probed to detect active, phosphorylated PKC substrates. (A and B) Data shown are from 1 experiment representative of 3.

To assess specificity for p47^phox, immunoblots of PMN lysates were probed with an antibody that detects a conserved motif found in active, serine-phosphorylated PKC substrates (Cell Signaling Technologies). Direct activation of PKC by PMA or infection with S. aureus triggered phosphorylation of many neutrophil proteins (Fig. 2C). Conversely, the overall signal obtained for cells infected with live Ft was low, suggesting impaired phosphorylation of multiple PKC substrates, including those with an apparent mass of 46–47 kDa, such as p47^phox and pleckstrin [40].

Virulent Ft strains and LVS inhibit neutrophil activation by diverse heterologous stimuli

Within 10 min of infection with LVS, the ability of neutrophils to produce ROS upon stimulation with PMA or OpZ is impaired significantly [13]. Whether this is also true for virulent Ft strains is unknown. Therefore, we used sequential stimulus luminol CL assays to assess oxidant production by control PMN or cells that were infected with type A or type B Ft 15 min prior to stimulation with PMA. Concordant with the data shown in Fig. 1, ROS were not generated by PMN maintained in buffer alone or by cells infected with the virulent type A Ft strain Schu S4, the virulent type B strain 1547-57, or the attenuated type B strain LVS (Fig. 3A and B). As expected [13], addition of PMA to cells pretreated with buffer triggered a robust respiratory burst (Fig. 3A, red symbols). However, this response was diminished profoundly by prior infection with Schu S4, 1547-57, or LVS (Fig. 3A, blue, purple, and gold symbols). Data from 3 independent experiments show that total PMA-stimulated ROS production was inhibited by 77.1 ± 1.8%, 88.2 ± 1.2%, or 88.6 ± 1.0%, respectively (n=3; P<0.002 vs. buffer-pretreated control). Similar data were obtained using the other virulent Ft strains in our collection (not shown) or when OpZ (Fig. 3B) or diC8 (65% inhibition, not illustrated) was used as the second stimulus.

Figure 3. Virulent Ft strains and LVS impair neutrophil activation by diverse heterologous stimuli.

All panels show sequential stimulus luminol CL assays. The first stimulus was added at Time 0, and the second stimulus was added after 15 min (A, B, and D) or 10 min (C) at 37°C. Graphs show luminol CL in counts/s and in each case, are the mean ± sem of triplicate samples from a representative experiment. (A) PMN were left in buffer, B-, or infected with Ft strains LVS, 1547-57, or Schu S4 prior to stimulation with PMA. B-B, Neutrophils treated with buffer only. (B) Same as A, except OpZ was used as the second stimulus. ZYM, Zymosan. (C) Effect of LVS preinfection on cell activation by 10 μM fMLF. (D) Effect of LVS preinfection on PMN activation by S. aureus. (Inset) Intracellular killing of S. aureus by control (–) or LVS-infected (+) PMN measured after 2 h at 37°C. Data are the mean ± sem (n=3).

OpZ activates signaling pathways coupled to CD11b/CD18 and dectin-1 [41], whereas PMA and diC8 mimic diacylglycerol and activate PKC directly [42]. To determine whether Ft impairs neutrophil activation by stimuli that act by other mechanisms, we analyzed ROS production by PMN that were treated with the formyl peptide fMLF, opsonized S. aureus, or unopsonized H. pylori in the presence or absence of LVS. fMLF triggers NADPH oxidase activation at the plasma membrane via G protein-coupled receptor signaling [43]. Like OpZ, S. aureus binds CD11b/CD18 and triggers NAPDH oxidase activation on forming phagosomes [44], whereas H. pylori enters neutrophils via lectinophagocytosis [45] and triggers a strong respiratory burst, while also diverting NADPH oxidase complexes away from bacterial phagosomes [26]. We now show that prior infection with LVS was sufficient to curtail ROS production triggered by fMLF (Fig. 3C), S. aureus (Fig. 3D), or H. pylori (not illustrated). As the central role of oxidants in control of S. aureus is well documented [46], we hypothesized that intracellular killing of this pathogen might be compromised. Indeed, we found that LVS reduced intracellular killing of S. aureus by 60.4 ± 7.4% (n=4; P=0.001; Fig. 3D, inset). Taken together, our data demonstrate that type A and type B Ft impair neutrophil activation by all NADPH oxidase agonists tested, despite their diverse mechanisms of action.

Ft profoundly impairs superoxide accumulation inside OpZ phagosomes

How Ft impairs PMN activation by other stimuli is unknown. To examine inhibition at the level of single cells, we used NBT staining to detect superoxide inside OpZ phagosomes. As shown in Fig. 4, >85% of OpZ phagosomes in control neutrophils contained dark or moderately dark formazan deposits, indicative of robust, local NADPH oxidase activity, whereas the vast majority of OpZ phagosomes in LVS-infected cells was unstained (formazan-negative) or exhibited only trace NBT staining. As diminished phagocytosis could inhibit cell activation indirectly, we quantified OpZ uptake and found that control and LVS-infected PMN contained an average of 2 ± 1 zymosan particles (n=3). These data demonstrate definitively that Ft inhibits superoxide anion accumulation inside OpZ phagosomes without perturbing other aspects of neutrophil function, such as phagocytosis.

Figure 4. Francisella inhibits superoxide accumulation inside OpZ phagosomes.

(A) NBT staining shows the extent of superoxide accumulation (blue formazan deposits) inside OpZ phagosomes of control (con) and LVS-preinfected neutrophils. Arrows indicate OpZ. (B) Similar to A, except PMN were stained to detect LVS (red) as well as superoxide (gray-black formazan). Arrows indicate OpZ. (C) Percentage of OpZ phagosomes in control or LVS-preinfected PMN that contain robust, medium, low, or undetectable formazan deposits. Data are the mean ± sem (n=3). **P < 0.005 vs. OpZ control in each category.

Ft prolongs NADPH oxidase assembly on OpZ phagosomes

Although mechanisms that control respiratory burst termination are not well defined, waning superoxide production coincides with enzyme disassembly and translocation of p47^phox, p67^phox, and p40^phox from the membrane back to the cytosol [17, 47, 48]. Typically, the percentage of OpZ phagosomes exhibiting enrichment for p47^phox and p67^phox declines sharply by 60–90 min after particle uptake [17, 26]. We confirmed these data here and extended our studies to include p40^phox with similar results (Fig. 5A–D). Rac2 is also required for NADPH oxidase function, and the data in Fig. 5E and F indicate that the active, endogenous form of this GTPase also accumulated on 15-min OpZ phagosomes. Although fewer OpZ compartments were enriched for active Rac when assayed 60 min after uptake, the decline was less pronounced than for soluble phox proteins and was not statistically significant (P=0.055).

Figure 5. LVS prolongs NADPH oxidase assembly on OpZ phagosomes.

Accumulation of p40^phox (A and B), p47^phox(C), p67^phox (D), and active Rac (E and F) on 15- and 60-min OpZ phagosomes in control and LVS-preinfected (LVS pre.) PMN was detected using confocal microscopy. (A and E) Representative images show p40^phox and active Rac in green and LVS in red. OpZ was detected using DIC. Arrows and arrowheads indicate positive and negative OpZ phagosomes, respectively. (B–D and F) Graphs show the percentage of 15- and 60-min OpZ phagosomes enriched for each marker at each time-point and are the mean ± sem from 3 to 5 independent experiments. *P < 0.05.

As Ft diminished total ROS production but did not prevent initiation of the response in the first seconds to minutes after addition of fMLF, S. aureus or other heterologous stimuli {(Fig. 3) and ref. [13]}, we hypothesized that Ft might enhance disassembly of NADPH oxidase complexes as a means to curtail an ongoing respiratory burst. However, a direct comparison of 15- and 60-min OpZ phagosomes in control and LVS-infected PMN revealed the opposite outcome (Fig. 5A–F). Thus, whereas oxidase components accumulated to a similar extent on 15-min OpZ phagosomes in the presence or absence of LVS, disappearance of p47/67/40^phox and active Rac from 60-min OpZ phagosomes was inhibited by Ft, and local enzyme enrichment was sustained relative to the buffer-pretreated controls. We therefore conclude that the ability of Ft to disrupt an ongoing respiratory burst is not coupled to enzyme disassembly but is characterized instead by sustained membrane accumulation of dysfunctional oxidase complexes.

LVS does not impair PMA- or OpZ-stimulated p47^phox phosphorylation in vivo

In addition to the central role of p47^phox phosphorylation in NADPH oxidase assembly and initiation of the respiratory burst, continuous phosphorylation of this protein is required to maintain superoxide production [47, 48]. To determine whether Ft affected p47^phox phosphorylation as a means to truncate an ongoing respiratory burst, ³²Pi-labeled neutrophils were pretreated with buffer or LVS and then stimulated with PMA or OpZ. Fifteen and 60 min later, in vivo phosphorylation of p47^phox was quantified. Immunoblots prepared from unlabeled cell lysates were also probed to detect active PKC substrates. LVS had no discernable effect on PMA-triggered phosphorylation of p47^phox (Fig. 6A) or other PKC substrates (Fig. 6B), despite profound inhibition of oxidant production (Fig. 3A). Similar data were obtained using OpZ as the second stimulus (not shown). A comparison of the data in Figs. 4–6 with Figs. 1 and 2 demonstrates definitively that Ft can use pre- and post-assembly mechanisms to inhibit oxidant production at its own phagosome and throughout infected cells and that inhibition can be achieved despite phosphorylation of p47^phox.

Figure 6. LVS does not impair PMA-triggered phosphorylation of p47^phox or other PKC substrates.

(A) In vivo phosphorylation of p47^phox was assessed in ³²Pi-labeled PMN that were maintained in buffer alone (BB), pretreated with buffer and then stimulated with LVS (BL), pretreated with buffer and then stimulated with PMA (BP), or infected with LVS and then stimulated with PMA (LP). Lysates were prepared 15 and 60 min after addition of the second stimulus, and ³²P-p47^phox was detected by PhosphorImager analysis after immunoprecipitation and SDS-PAGE. Data shown are from 1 experiment representative of 3. (B) Active PKC substrates in PMN lysates were detected by immunoblotting. Cells were stimulated as in panel A, and were processed 15 min after addition of the second stimulus. Each lane contains 30 μg protein. Data shown are representative of 3 independent determinations.

CM contribute to NADPH oxidase inhibition

Next, we considered the possibility that PMN function may be disrupted by host or bacterial factors in CM. To test this, we prepared media that were conditioned by LVS, PMN, or LVS-infected PMN (see Materials and Methods).

None of the CM activated neutrophils directly (Fig. 7A). On the other hand, media condition by neutrophils alone reduced total PMA-stimulated ROS production by 10–15%, whereas LVS CM or media conditioned by a combination of LVS and PMN reduced ROS production by 20–25% (Fig. 7A). As the inhibitory effect of each CM was reproducible but modest compared with whole LVS (70–85% inhibition; Fig. 7A), our data suggest that factors in CM contribute to but do not completely account for the ability of Ft to inhibit the respiratory burst.

Figure 7. Clues to the mechanism of NADPH oxidase inhibition.

(A) Effects of CM on cell activation by PMA. Neutrophils were treated at Time 0 with RPMI (Naïve media), media conditioned by LVS (LVS CM), neutrophils (PMN CM), or a combination of LVS and PMN (Dual CM). PMN infected with LVS at Time 0 were used as controls, and in all cases, ROS were detected using luminol. Open symbols indicate cells exposed to media or bacteria only. Closed symbols indicate neutrophils stimulated with PMA after 10 min. (B) LVS inhibits NADPH oxidase (NOX) activity in the cell-free assay yet does not affect xanthine oxidase (XO). In each case, superoxide production was measured in a kinetic assay as a SOD-inhibitable reduction of cytochrome c at A₅₅₀. Data indicate maximum rates normalized to the no-LVS control and are the mean ± sem of 3 independent determinations. *P = 0.013 versus control.

Ft inhibits NADPH oxidase activity in a cell-free assay but does not scavenge superoxide or prevent its generation by xanthine oxidase

Next, we tested the ability of Ft to diminish superoxide production in the cell-free NADPH oxidase assay using purified cytosol and membranes obtained from resting neutrophils that were disrupted by nitrogen cavitation and fractionated on Percoll gradients [29, 30] (see Materials and Methods). Reaction mixtures contained 10⁷ cell equivalents of cytosol and membranes, and arachidonate was used as the activating amphiphile. Superoxide production was measured in a kinetic assay as a SOD-inhibitable reduction of cytochrome c at 550 nm. In this assay, LVS inhibited the maximum rate of superoxide production by 71 ± 16% (P=0.013; n=3; Fig. 7B). Specificity is indicated by the fact that LVS had no significant effect on superoxide production by xanthine oxidase (5±2% inhibition; n=3; Fig. 7B).

Ft acid phosphatase is dispensable for NADPH oxidase inhibition by strain Schu S4

Ft virulence factors essential for NADPH oxidase inhibition are not well defined. One candidate is the periplasmic acid phosphatase AcpA, which in purified form, diminishes activation of porcine neutrophils by PMA and fMLF [49]. To test the role of AcpA in NADPH oxidase inhibition by a virulent, human pathogenic strain of Ft, we used Group II intron retargeting [33] to disrupt acpA in Schu S4, generating mutant clone acpA576 (Fig. 8A and B). The apcA576 strain grew normally on agar plates yet lacked acid phosphatase activity, as judged by filter paper overlay (Fig. 8C). Complementation with a WT acpA gene expressed in trans restored enzyme activity, validating the mutant phenotype (Fig. 8C, strain acpA576c). Despite the paucity of acid phosphatase activity, infection of PMN with strain acpA576 did not trigger a respiratory burst, and by this assay, mutant organisms were indistinguishable from WT Schu S4 and acpA576c (Fig. 8D). Strain acpA576 was also indistinguishable from Schu S4 and the complemented strain in its ability to inhibit neutrophil activation by PMA (Fig. 8E) or other agonists (not shown).

Figure 8. Generation and characterization of a Schu S4 acpA mutant.

(A) The diagram shows the site of Group II intron-mediated insertion into the acpA gene of the Schu S4 chromosome. (B) Confirmation by PCR of acpA disruption in mutant clone acpA576. Lane M, Markers; lane 1, Schu S4; lane 2, acpA576. (C) Acid phosphatase activity of Schu S4 (sector 1), acpA576 (sector 2), and the complemented strain acpA576c (sector 3). Left image, Bottom view of the agar plate showing bacterial growth; right image, filter paper overlay showing bacterial acid phosphatase activity. (D) ROS generated by uninfected PMN or cells infected for 90 min at 37°C with Schu S4, acpA576, or acpA576c were detected using luminol. Data indicate total ROS normalized to cells stimulated with 200 nM PMA and are the average ± sem (n=3). (E) Sequential stimulus luminol assay. PMN were pretreated with buffer or infected with Schu S4, acpA576, or acpA576c for 15 min at 37°C and then stimulated with PMA. Graph shows total PMA-stimulated ROS normalized to the buffer-pretreated control. Data are the mean ± sem (n=3). *P < 0.05 versus other samples.

Role for fevR, but not iglI or iglJ, in NADPH oxidase inhibition by strain Schu S4

Recent data of Barker et al.[50] suggest that the FPI encodes a secretion system, and IglI is a candidate effector required for virulence that is secreted into the medium as well as the host cell cytosol. FevR controls expression of many Francisella genes, including those in the FPI, and we demonstrated recently a role for FevR in NADPH oxidase inhibition by LVS [33]. Nevertheless, the genes within the FevR regulon required for this aspect of virulence are unknown. As a secreted virulence factor could account for the ability of Ft to inhibit ROS production at the Ft phagosome and throughout infected cells, we used Group II intron retargeting to disrupt iglI and the adjacent gene iglJ in Schu S4, generating mutant strains iglI153 and iglJ189 (Table 1 and Fig. 9A and B). We also generated a fevR mutant in Schu S4, as we described previously for LVS [33], and then compared the ability of WT and mutant organisms to activate PMN. Our data demonstrate that fevR was required for NADPH oxidase inhibition by type A Ft during initial infection and upon exposure to other stimuli (Fig. 9C and D). In this regard, it is of interest that the complemented strain, fevRc, diminished PMA-stimulated ROS to a greater extent than WT Schu S4 (Fig. 9D). On the other hand, IglI and IglJ appeared dispensable, as the iglI153 and iglJ189 mutant strains (Fig. 9C and D) and their trans-complemented counterparts (not shown) were indistinguishable from the Schu S4 parent. Control experiments confirmed inactivation of iglI, iglJ, and fevR and rescue by trans complementation, as all 3 mutant strains were defective for phagosome escape in human macrophages, whereas WT and complemented strains were not (unpublished data). These data indicate that distinct components of the FevR regulon are required for blockade of the respiratory burst and phagosome escape.

Figure 9. Characterization of Schu S4 iglI, iglJ, and fevR mutants.

(A) Diagram shows the site of Group II intron-mediated insertion into the iglI and iglJ genes in the Schu S4 chromosome. (B) Confirmation by PCR of iglI and iglJ disruption in mutant clones iglI153 and iglJ189. Lane M, Markers; lanes 1 and 3, Schu S4; lane 2, iglI153; lane 4, iglJ189. (C) ROS, generated by uninfected PMN or cells infected for 90 min at 37°C with Schu S4, fevR, fevRc, iglI153, or iglJ189, were detected using luminol. Data are the average ± sem (n=3) and are normalized to the uninfected PMN control. *P ≤ 0.025 versus other samples. (D) Sequential-stimulus luminol assay. PMN were pretreated with buffer or infected with Schu S4, fevR, fevRc, iglI153, or iglJ189 for 15 min and then stimulated with PMA. Graph shows total PMA-stimulated ROS normalized to the buffer-pretreated control. Data are the mean ± sem (n=3). *P < 0.05 versus buffer; **P < 0.025 versus all other samples.

LVS attenuates the ability of IS to confer PMN activation and bacterial killing

Under normal circumstances, ligation of FcRs by IgG-coated organisms enhances host defense and microbe clearance. However, studies of human volunteers in the 1960s found that with regard to Ft, antibodies are a reliable marker of infection or vaccination, but their concentration does not correlate with the efficacy of the immune response [51]. Recent studies of mice suggest that compared with T cells, antibodies contribute little anti-Ft defense, particularly during infection with type A strains [2, 52–54]. At the same time, the results of a 1980 study suggest that IS-opsonized LVS can activate PMN [15], but interpretation of these data is complicated by the fact that Lofgren et al. used a high concentration of IS (75%) to opsonize LVS, which likely caused formation of large bacterial aggregates. As it is now appreciated that changes in particle size can affect outcome, independent of the receptor engaged [55], the effects of IS on Ft-PMN interactions are not well defined.

To revisit this issue, we obtained from Dr. Jeannine Petersen (CDC) IS of known titer from 3 persons vaccinated previously with LVS and tested their effects over a range of concentrations. Data obtained for all 3 sera were similar with results for LVS opsonized with IS #1 (agglutinating titer 1:1000), shown in Fig. 10. To recapitulate conditions used previously [15], we first opsonized LVS with a 1:8 dilution of IS #1 and confirmed that these organisms triggered a robust respiratory burst (Fig. 10A and B). As expected, these samples also contained large LVS aggregates, many of which bound to PMN but were too big to be internalized. Under these conditions, formazan deposits were present, not only inside PMN but also covered the cell surface (Fig. 10C, right panel). In marked contrast, additional studies revealed that ROS production declined sharply as the concentration of IS was reduced, such that neutrophils infected with LVS that were opsonized with a subagglutinating concentration of IS #1 (1:1064) triggered no more ROS than cells infected with AS-opsonized organisms (Fig. 10A and B). These results suggested that Ft may be able to undermine the ability of IS to confer PMN activation, and in this regard, our studies of bacteria opsonized with intermediate concentrations of IS were perhaps the most informative. For example, LVS opsonized with IS (1:32) triggered a moderate respiratory burst (Fig. 10A and B), and confocal analysis showed that 88 ± 7% of phagosomes containing these organisms accumulated NADPH oxidase components, yet only 9 ± 2% contained superoxide, as judged by NBT staining (Fig. 10C and D). Moreover, although IS (1:32) increased infection efficiency compared with AS-opsonized LVS, the ability of PMN to kill these organisms was not enhanced (Fig. 10E). Thus, we show for the first time that Ft can overcome the ability of low-moderate antibody concentrations to control infection via ROS production.

Figure 10. Post-assembly NADPH oxidase inhibition mechanisms also curtail PMN activation by IS-opsonized LVS.

(A) Rate and extent of ROS production by PMN that were left untreated, stimulated with OpZ, infected with AS-opsonized LVS, or infected with LVS opsonized with the indicated dilution of IS. (B) Pooled data from 3 independent experiments show peak luminol CL (mean±sem; n=3) for untreated PMN or cells stimulated in parallel with OpZ, AS-opsonized LVS, or LVS that were opsonized with IS diluted 1:8, 1:16, 1:32, or 1:1024. (C) NBT staining of PMN infected for 60 min with AS-, IS (1:32)-, or IS (1:8)-opsonized LVS. Bacteria are shown in red, and the black arrow and arrowheads indicate NBT-positive and NBT-negative phagosomes, respectively. White arrowheads indicate uningested LVS aggregates. (D) PMN infected with AS- or IS-opsonized LVS for 15 min were stained to show gp91/p22^phox or p47/67^phox in red and LVS in green. IS was used at 1:32 dilution. Arrows and arrowheads indicate positive and negative phagosomes, respectively. (E) PMN were infected with LVS opsonized with AS or IS (1:32), and viable intracellular bacteria were quantified at 30 min and 2 hpi by plating cell lysates for measurement of CFU. Data are the mean ± sem for triplicate samples from a representative experiment.

DISCUSSION

Although it is clear that Ft evades killing by phagocytes, including human neutrophils, the underlying mechanisms of inhibition are only beginning to be defined. We demonstrated previously that LVS opsonized with AS disrupts NADPH oxidase assembly as it enters PMN and that the ability of infected cells to generate ROS upon stimulation with PMA or OpZ is impaired [13]. Nevertheless, how this complex pattern of inhibition is achieved is unknown, and whether human pathogenic strains of Ft differ in their ability to alter neutrophil function is unclear. This study compares the ability of virulent type A and type B Ft strains to disrupt neutrophil oxidative defense mechanisms, and our findings are significant for 3 reasons. First, we show that all Ft strains tested impair neutrophil activation. During uptake of Ft opsonized with AS, enzyme assembly is disrupted, as indicated by diminished phosphorylation of p47^phox and phagosome exclusion of flavocytochrome b_558. Second, we show that Ft can also act distal to assembly to inhibit the function of NADPH oxidase complexes at the membrane. This latter mechanism not only impairs the responses of infected cells to heterologous stimuli but also curtails ROS production during phagocytosis of Ft opsonized with IS and as such, suggests that this pathogen may diminish the efficacy of the adaptive immune response. Finally, using mutants generated in Schu S4, we show that pre- and post-assembly NADPH oxidase inhibition mechanisms require the Ft virulence regulator fevR but not iglI, iglJ, or the acid phosphatase acpA. These data demonstrate that distinct components of the FevR regulon are required for phagosome escape and NADPH oxidase inhibition and also indicate that the mechanisms of Ft-mediated NADPH oxidase inhibition are distinct from those used by F. novicida.

The two subspecies of Ft that account for nearly all cases of human tularemia differ in geographic distribution. Ft subspecies holarctica is found throughout the Northern hemisphere, whereas Ft subspecies tularensis is restricted to North America. Recently, type A strains of Ft were divided into 2 clades; A1 strains are located in central and eastern United States, and A2 strains are more abundant than A1 strains in the west [21, 56]. Moreover, for reasons that are as-yet unclear, the case-fatality rate for humans infected with A1 strains is much higher (24%) than type B (7%) or A2 (0%) isolates [57]. With regard to the organisms used for this study, Schu was isolated originally from a human tularemia patient in Ohio in 1941 [58]. The highly virulent subclone Schu S4 was isolated in 1951 and has since been studied widely [20]. Schu S4 and the Massachusetts strain (MA00-2987) belong to clade A1 [23], the Wyoming strain (WY96-3418) belongs to clade A2 [21, 23], and isolates from 2 patients in Iowa are type B (1547-57 and 1623-36) [24]. TI0902 is a type A strain of an unknown subtype that was isolated from a cat [22]. Determining the extent to which Ft strains differ in their ability to disrupt the host immune response is an area of active investigation, and the results of this study demonstrate that all strains tested can inhibit PMN production of ROS. We therefore conclude that NADPH oxidase inhibition is a conserved virulence determinant that is shared by A1, A2, and B strains and is retained in the attenuated vaccine strain, LVS.

Pathogens that survive inside phagocytes must withstand, avoid, or inhibit toxic ROS. To this end, several organisms prevent NADPH oxidase activation by disrupting an aspect of the enzyme assembly process. For example, compartments containing Leishmania donovani acquire gp91^phox and Rac but not p47^phox or p67^phox, and this has been attributed to defects in p47^phox phosphorylation [59–61]. Coxiella burnetti also prevents membrane translocation of p47/67^phox in neutrophils by an unknown mechanism [62], whereas Salmonella enterica serovar Typhimurium uses proteins encoded in the second pathogenicity island to prevent phagosome acquisition of flavocytochrome b₅₅₈ in macrophages [63]. Anaplasma phagocytophilum is an obligate intracellular pathogen that infects neutrophils and their precursors and exhibits a more complex inhibition strategy. In mature PMN, this pathogen disrupts NADPH oxidase assembly via effects on targeting of flavocytochrome b₅₅₈ and Rac, and in HL-60 promyelocytes, it diminishes expression of genes that encode NADPH oxidase subunits [64].

The results of this study demonstrate for the first time that Ft also uses a complex and multifaceted strategy to inhibit NADPH oxidase activity throughout PMN. Forming phagosomes containing type A and type B Ft are not only devoid of flavocytochrome b₅₅₈ (as occurs in macrophages infected with Salmonella), but also, phosphorylation of p47^phox and other PKC substrates is diminished relative to other agonists tested (similar to the effects of Leishmania amastigotes). Thus, live Ft disrupts multiple aspects of the enzyme assembly and activation process via effects on membrane and soluble NADPH oxidase components, and defects in flavocytochrome b₅₅₈ targeting likely synergize with diminished p47^phox phosphorylation to ensure blockade of the respiratory burst during infection with AS-opsonized organisms (Figs. 1 and 2). In future studies, it will be of interest to determine whether all phosphorylation sites in p47^phox are affected equally or whether there is a selective defect in generation of the most highly acidic forms of the protein, as occurs in persons with flavocytochrome-deficient, chronic granulomatous disease [65], and to ascertain whether Ft inhibits or is merely a poor agonist of PKC.

Perhaps the most striking finding of this study is our demonstration that Ft uses a distinct post-assembly inhibition mechanism to impair cell activation by heterologous stimuli. In this case, there is no defect in targeting of flavocytochrome b₅₅₈ or active Rac, phosphorylation of p47^phox, or membrane translocation of p47/67/40^phox complexes, yet the overall magnitude of the respiratory burst is diminished profoundly (Figs. 3–6). A defect in ROS production distal to enzyme assembly is consistent with the ability of Ft to impair NADPH oxidase activation by all stimuli tested, despite their diverse mechanisms of action. Ft is therefore distinct from Neisseria gonorrhoeae, which can curtail PMN activation by fMLF but not PMA [66], and its effects also differ from the stimulus-selective defects in superoxide attributed to disruption of p40^phox [5] or defects in the ClC3 anion channel that inhibit phagocytosis as well as oxidant production [30]. Under these conditions, diminished ROS production cannot be attributed to accelerated or enhanced enzyme disassembly and is characterized instead by sustained membrane accumulation of dysfunctional NADPH oxidase complexes. In keeping with this, phosphorylation of p47^phox and other PKC substrates appears normal when assayed 15 or 60 min after addition of PMA or OpZ, results that distinguish the effects of Ft from PKC inhibitors that curtail an ongoing respiratory burst by preventing the sustained phosphorylation of p47^phox required for maintenance of enzyme assembly and activity [47, 48]. The mechanism of inhibition that we describe here is also consistent with the ability of Ft to inhibit NADPH oxidase activity in the cell-free reconstitution assay, as in this system, superoxide production is independent of the upstream signaling and phosphorylation events that occur in intact cells, and arachidonate is used to induce the conformational changes in p47^phox required for NADPH oxidase assembly and activation [67, 68]. To our knowledge, a post-assembly mechanism of NADPH oxidase inhibition has not been described for other pathogens and as such, may be unique to Ft. As we show that a similar post-assembly NADPH oxidase inhibition mechanism curtails the ability of IS to confer PMN activation and enhance intracellular killing (Fig. 10), our data also provide a potential explanation for the limited role of antibodies in host defense against Ft [2, 51–54] and strongly suggest that this bacterium undermines antibody efficacy as part of its virulence strategy. Consistent with this, Schu S4 may also stimulate antibody degradation [69].

The virulence factors required for evasion of oxidative host defense are poorly defined. We have shown that LVS carA, carB, and pyrB mutants activate PMN and are killed rapidly by intraphagosomal ROS [28]. However, these genes are part of the pyrimidine biosynthetic pathway, and their contribution to virulence is indirect. Many bacterial pathogens use type III, type IV, or type VI secretion systems to introduce effector proteins directly into host cells as a means to disrupt phagocytosis, phagosome maturation, or ROS production [63, 70, 71]. A subset of genes in the FPI is similar to type VI secretion system components [71, 72], and the first candidate effector of this secretion system, IglI, was identified recently [50]. IglI is also encoded in the FPI, has been detected in the cytosol of infected macrophages and in the surrounding media, and is essential for phagosome escape [50]. Whether IglI or another secreted effector blocks ROS production during tularemia is unknown, but this concept is attractive, as it is compatible with the rapid, global inhibition of NADPH oxidase activity that ensues following Ft uptake as well as the effects of CM and the ability of Ft to inhibit NADPH oxidase activity in the cell-free assay. FevR is a regulatory factor that controls the expression of ∼100 Ft genes, including but not limited to those in the FPI [73], and a recent study by our group suggested for the first time that FevR is required, not only for phagosome egress but also for blockade of the PMN respiratory burst [33]. Nevertheless, the genes within the FevR regulon that affect NADPH oxidase activity directly are unknown. To begin to address this, we generated fevR, iglI, and iglJ mutants in the highly virulent type A Ft strain Schu S4. A direct comparison of mutant and WT organisms revealed that all 3 genes were essential for phagosome escape and bacterial growth in human macrophages (our unpublished data), but unlike fevR, neither iglI nor iglJ was required for inhibition of the respiratory burst during infection with Ft alone or upon exposure of infected cells to other stimuli (Fig. 9). These data are significant, not only because they confirm the role of FevR in Schu S4 virulence [16] but also because they provide the first evidence to suggest that distinct components of the FevR regulon are required for phagosome escape and evasion of oxidative host defense.

Another gene implicated in NADPH oxidase inhibition is acpA, which encodes a periplasmic acid phosphatase. The purified enzyme impairs activation of porcine neutrophils by fMLF and PMA and exhibits broad phosphomonoesterase activity, as indicated by its ability to cleave ATP, NADP⁺, and various phosphopeptides in vitro [49]. Despite this, several lines of evidence indicate that neither AcpA nor other acid phosphatases is required for NADPH oxidase inhibition by human pathogenic strains of Ft. First and foremost, the acpA gene is not intact in the Ft holarctica genome [16, 74]. Consequently, the enzyme is truncated at its N terminus, lacks the signal sequence required for export to the periplasm, and may be unstable, as AcpA protein and activity are nearly undetectable or absent in bacterial lysates, as judged by analyses of clinical Ft holarctica isolates as well as LVS [16, 74–76]. Despite this, the first studies to demonstrate inhibition of human PMN activation by Ft used type B strains [14, 15] and were confirmed by us using LVS and two clinical type B isolates (Fig. 1). Second, we generated for this study an acpA mutant in Schu S4 to test directly the role of this enzyme in NADPH oxidase inhibition by organisms of the tularensis biovar. Total acid phosphatase activity in the apcA576 mutant strain was reduced by >90%, confirming published data [16], but this had no effect on the ability of these organisms to inhibit NADPH oxidase activity during bacterial uptake or upon exposure of infected neutrophils to heterologous stimuli (Fig. 8). Third, acpA is not a component of the FevR regulon [77], and we have shown that fevR is essential for NADPH oxidase inhibition by LVS [33] as well as Schu S4 (Fig. 9), whereas acpA is not. Altogether, these data demonstrate that acpA is dispensable for evasion of oxidative host defense by type A and type B Ft.

Among other organisms in the genus Francisella is F. novicida. This bacterium is not a human pathogen but is often used as a Ft surrogate, as it causes a tularemia-like disease in mice and does not require BSL-3 containment. While this manuscript was being revised, Mohapatra et al. [78] reported that F. novicida also inhibits the oxidative burst. In this case, inhibition appears to occur by a mechanism that differs in several respects from what we report here and in previous work for Ft. Specifically, the data indicate that F. novicida relies predominantly, or perhaps exclusively, on diminished phosphorylation of p47^phox to curtail ROS production [78] and as such, may not be able to prevent phagosome accumulation of flavocytochrome b₅₅₈ or to act at a post-assembly stage to inhibit the activity of enzyme complexes at the membrane. Furthermore, diminished phosphorylation of p47^phox in infected cells was attributed to the concerted action of 4 acid phosphatases, as a quadruple deletion mutant lacking acpA, acpB, acpC, and hapA cannot prevent oxidase assembly or activation at the F. novicida phagosome and is killed rapidly by PMN [78]. These data are of interest, as there are 6 acid phosphatase genes in the F. novicida genome, acpA–D, and hapA-B [16]. Of these, only acpB is conserved in all type B Ft strains, and the others are disrupted or are nonfunctional pseudogenes [16]. In contrast, type A Ft express acpB and acpC as well as acpA [16]. The activity of AcpB is extremely low, and recent data indicate that deletion of acpA alone or together with acpB and acpC in Schu S4 has no effect on pathogenicity, as judged by bacterial growth in macrophages or virulence in a mouse intranasal infection model, whereas fevR is essential [16]. Similarly, we show here that fevR is essential for NADPH oxidase inhibition, but acpA is not, and as such, our data further support the notion that acid phosphatases are dispensable for virulence of human pathogenic strains of Ft and are also in keeping with the idea that high acid phosphatase activity correlates with diminished virulence for organisms in the Francisella genus [16, 76].

In summary, the results of this study demonstrate for the first time that Ft uses a multifaceted strategy to inhibit production of toxic ROS by human neutrophils. We show that Ft can block enzyme assembly or act at a later stage to disrupt the activity of NADPH oxidase complexes at the membrane. Our results account for the ability of Ft to impair cell activation by all heterologous stimuli tested and provide a potential explanation for the limited efficacy of specific antibodies in host defense during tularemia. Using targeted mutagenesis, we show that distinct components of the fevR regulon are required for phagosome egress and evasion of oxidative host defense and also identify important differences between human pathogenic Ft and F. novicida. Although much remains to be determined, the results of this study significantly advance our understanding of the pathogenesis of tularemia and strongly suggest that disruption of oxidative defense mechanisms is a central aspect of virulence.

AUTHORSHIP

R.L.M. designed and performed experiments and cowrote the manuscript. J.T.S., S.R.L., and B.W.B. designed and performed experiments and edited the manuscript. J.G.M. designed experiments and edited the manuscript. B.D.J. edited the manuscript. L-A.H.A. designed experiments and cowrote the manuscript.