The bacterium Francisella tularensis (Ft) is a potential weapon of bioterrorism when aerosolized. Macrophage infection is necessary for disease progression and efficient phagocytosis by human macrophages requires serum opsonization by complement. Microbial complement activation leads to surface deposition of a highly regulated protein complex resulting in opsonization or membrane lysis. The nature of complement component C3 deposition, i.e., C3b (opsonization and lysis) or C3bi (opsonization only) fragment deposition, is central to the outcome of activation. In this study, we examine the mechanisms of Ft resistance to complement-mediated lysis, C3 component deposition on the Ft surface, and complement activation. Upon incubation in fresh nonimmune human serum, Schu S4 (Ft subsp. tularensis), Fn (Ft subsp. novicida), and LVS (Ft subsp. holarctica live vaccine strain) were resistant to complement-mediated lysis, but LVSG and LVSR (LVS strains altered in surface carbohydrate structures) were susceptible. C3 deposition, however, occurred on all strains. Complement-susceptible strains had markedly increased C3 fragment deposition, including the persistent presence of C3b compared with C3bi, which indicates that C3b inactivation results in survival of complement-resistant strains. C1q, an essential component of the classical activation pathway, was necessary for lysis of complement-susceptible strains and optimal C3 deposition on all strains. Finally, use of Francisella LPS mutants confirmed O Ag as a major regulator of complement resistance. These data provide evidence that pathogenic Francisella activate complement, but are resistant to complement-mediated lysis in part due to limited C3 deposition, rapid conversion of surface-bound C3b to C3bi, and the presence of LPS O Ag.

Complement is a highly regulated and multifunctional system that is the major extracellular arm of innate immunity. Its activation results in three major potential outcomes for microbes: lysis upon assembly and insertion of the terminal membrane attack complex (MAC),4 opsonization, and the release of anaphylatoxins that enhance local inflammation (1, 2, 3). Three potential complement activation pathways exist. Surface Ag recognition by IgM or IgG initiates the classical pathway. Surface carbohydrate recognition by soluble lectins initiates the lectin-mediated pathway. Random H2O-catalyzed “tick-over” of C3 initiates the alternative pathway. Each pathway leads to assembly of C3 convertase, an enzymatically active complex formed by the cleavage fragments of upstream components. Both C4b2a (classical and lectin pathways C3 convertase) and C3bBb (alternative pathway C3 convertase) cleave C3a from native C3 to form C3b. A cryptic reactive thioester group within native C3 becomes available on C3b to form covalent linkages with either hydroxyl or amine moieties on microbial acceptor molecules (4). Bound C3b interacts with parent C3 convertases to form C5 convertases to initiate the terminal lytic pathway leading to insertion of the MAC. However, negative regulation of complement activity can result in further cleavage of C3b to smaller, inactive fragments (C3bi, C3dg, or C3d) that do not initiate the terminal lytic pathway (3). The C3b and C3bi fragments mediate opsonophagocytosis of bound organisms via their recognition by complement receptors (CR) on professional phagocytes (5). Macrophage CR-mediated entry for intracellular pathogens has long been regarded as a particularly important pathway to enhance survival following phagocytosis (6).

Francisella tularensis (Ft) is a Gram-negative facultative intracellular coccobacillus. Two subspecies of Ft exist that cause human disease. Type A, Ft subsp. tularensis, is endemic to North America. Type B, Ft subsp. holarctica, is endemic to Eurasia, Japan, and North America and is less virulent than type A (7, 8). Ft subsp. novicida (Fn) is avirulent in humans, except in rare cases (9), but causes a fulminant disease in mice. Like Fn, the live vaccine strain (LVS), which is derived from holarctica, is attenuated in humans but not in relevant animal models. LVS is undergoing Food and Drug Administration scrutiny as a vaccine since both the induced mechanism of immunological protection and the genetic basis of attenuation are undefined. Also, spontaneous changes occur with respect to colony morphology, resulting in phase variants (called gray variants) with reduced virulence and immunogenicity, which may affect vaccine utility. The genetic basis for phase variation, which also occurs for type A isolates, is also unknown (10). In fact, based on changes in LPS structure, multiple gray variant phenotypes have been described including variants that completely lack O Ag and variants that express antigenically distinct O Ag compared with LVS (11, 12).

Tularemia results from exposure to infected animal tissue, bites from arthropod vectors, or the direct ingestion or inhalation of Francisella (13). Pneumonic tularemia is the most serious form because untreated cases result in 30–60% mortality compared with 5–6% mortality associated with cutaneous disease (14, 15). Pneumonic tularemia develops upon inhalation of <10 CFU or via hematogenous spread of bacilli from peripheral sites to the lung (13, 16). Because of its highly infectious and lethal nature, Ft is designated as a category A select agent by the Centers for Disease Control and Prevention and is a potential weapon of bioterrorism. Francisella predominantly infects and replicates within macrophages and spreads systemically via the reticuloendothelial system (17).

Considering the extremely low infectious dose required to cause disease, it is paradoxical that in vitro studies examining macrophage infection by Ft require high experimental multiplicities of infection of at least 100:1 without opsonization. However, phagocytosis of type A Francisella, LVS, and Fn by human monocyte-derived macrophages increases dramatically in the presence of serum in a C3- and CR-dependent manner (18, 19). Complement-mediated opsonization has also been shown for the phagocytosis of LVS by human monocyte-derived dendritic cells (20). Given the importance of complement-mediated opsonization for uptake by phagocytes and a previous report that LVS does not fix C3 (21), we studied the nature of complement interactions with Francisella in nonimmune human serum. In this study, we examined several Francisella subspecies and variant strains to determine 1) whether complement activation leads to bacterial lysis, 2) the nature of C3 deposition and subsequent fragmentation, 3) complement activation pathways, and 4) resistance mechanisms to complement-mediated lysis.

Ft subsp. tularensis strain Schu S4, a Centers for Disease Control and Prevention clinical isolate, was provided by R. Lyons (University of New Mexico, Albuquerque, NM). Ft subsp. holarctica LVS (ATCC 29684) was provided by K. Elkins (Center for Biologics Research and Evaluation, U.S. Food and Drug Administration, Bethesda, MD). Fn (U112), LVSG, and LVSR were provided by F. Nano (University of Victoria, Victoria, British Columbia, Canada). LVSG is a spontaneous gray phase variant that rarely reverts to LVS when grown on chocolate II agar (11). LVSR was originally described as a capsule-negative strain (21) and was selected for its rough colony morphology after the mutagenesis of LVS by treatment with acridine orange. The LPS O Ag mutants, LVSΔwbtA and LVSΔwbtM, provided by D. Frank (Medical College of Wisconsin, Milwaukee, WI), were created by modified Himar1 transposon (HimarFT)-mediated mutagenesis of LVS (22). The complemented mutant strain (LVSΔwbtM:pFTNAT-wbtM) and LVSΔwbtM containing the empty pFTNAT complementation plasmid (LVSΔwbtM:pFTNAT) were also provided by Dr. D. Frank. Experiments using Schu S4 were conducted within biosafety level 3 (BSL3)-select agent-certified laboratories with adherence to federal- and institutional-select agent regulations. Bacteria were grown overnight (∼18 h) on chocolate II agar (BD Biosciences) at 37°C. For experiments using LVSΔwbtM and related strains, bacteria were grown overnight on Mueller-Hinton (MH) agar containing 2.1% MH broth (BD Biosciences), 0.5% NaCl, 1.6% agar, 1% protease peptone (BD Biosciences), 0.1% glucose, 2% Isovitalex (BD Biosciences), 0.025% ferric pyrophosphate, and 2.5% FBS (Invitrogen). Strains containing the pFTNAT plasmid were grown overnight on MH agar containing 50 μg ml−1 nourseothricin (Sigma-Aldrich). DH5α, a laboratory strain of Escherichia coli, was grown overnight on Luria-Bertani agar at 37°C before use.

Serum was isolated from healthy adult volunteers with no known exposure to Francisella according to a protocol approved by the Ohio State University Medical College Internal Review Board. The sera were processed to maintain optimal complement activity (23). Briefly, isolated nonheparinized whole blood was kept at room temperature for 1 h to allow for clot formation and then at 4°C to allow for clot contraction. The clot was removed by centrifugation at 500 × g for 15 min at 4°C. The serum fraction was collected, filter sterilized, aliquoted, and stored at −80°C. Heat inactivation (HI) was performed at 56°C for 30 min. C5-depleted (d), C8d, C1qd, and factor B (FB)-depleted (FBd) sera, and purified C1q and FB were purchased from Complement Technology and stored at −80°C. On the day of use, fresh sera were thawed at room temperature, then immediately chilled on ice until needed. A concentrated serine protease inhibitor mixture (containing AEBSF, aprotinin, elastatinal, and GGACK) was purchased from Calbiochem. Fifty percent hydroxylamine was purchased from Alfa Aesar. Other chemicals were purchased from Sigma-Aldrich.

Complement-mediated killing was conducted using fresh nonimmune or HI (negative control) sera. Bacteria were suspended in gelatin veronal buffer (GVB++; 0.1% gelatin, 5.5 mM barbital, 142 mM NaCl, 0.5 mM MgCl2, and 0.15 mM CaCl2; pH 7.3) at equalized concentrations by measuring the OD at 600 λ. For each assay, 2 × 106 bacteria were incubated with various serum concentrations for 1 h in microcentrifuge tubes (final volume of 200 μl in reaction buffer) at 37°C with slow agitation. For some experiments, 10 mM EDTA or 10 mM EGTA with 7 mM MgCl2 was added to fresh serum to differentiate potential complement activation pathways. In other experiments, C1qd, C1q-replete (r; achieved by adding purified C1q to C1qd at a final concentration of 200 μg/ml), FBd, and FBr (achieved by adding purified FB to FBd at a final concentration of 400 μg/ml) sera were used. Reactions were stopped by placing tubes on ice for 5 min. For some experiments, a serine protease inhibitor mixture diluted in ice-cold PBS was used to stop the reaction, and the results were the same as those using iced tubes. Ten-fold serial dilutions were plated to determine surviving CFU.

Fresh donor, HI donor, C5d, C1qd, and C1qr sera were used to evaluate complement component C3 deposition on Ft strains. After preblocking microcentrifuge reaction tubes in PBS containing 0.1% human serum albumin (ZLB Plasma) for 30 min at 37°C, 5 × 108 bacteria/reaction were incubated in 10% sera for various times at 37°C. Adding an ice-cold serine protease inhibitor mixture in blocking buffer terminated deposition and fragmentation. Unbound C3 was removed from bacterial pellets by washing in blocking buffer (three times) and PBS (one time) at room temperature (12,000 × g for 3 min). To ensure equal lane loading, aliquots were taken from the final PBS wash and plated to count CFU. After final resuspension in Laemmli’s sample buffer, samples were boiled for 2 min and separated by minigel 7.5% SDS-PAGE (Bio-Rad), followed by protein transfer to polyvinylidene difluoride membranes (Millipore). Membranes were blocked overnight at 4°C in advanced ECL blocking buffer at 2% (w/v), which was also used for Ab dilution (Amersham Biosciences). Goat antiserum to human C3 (diluted to 1/20,000; Quidel) served as the primary Ab for 1-h incubations at room temperature. HRP-conjugated rabbit anti-goat IgG (H + L) Ab (diluted to 1/20,000; Bio-Rad) was used as the secondary Ab for 1-h incubations at room temperature. Advanced ECL reagent was used for detection (Amersham Biosciences).

C8d serum was used to evaluate complement component C3, C5, and C7 deposition on Ft strains. After preblocking microcentrifuge reaction tubes as described for Western blotting experiments above, 3 × 108 bacteria/reaction were incubated in 10% fresh serum or serum containing 10 mM EDTA for 5 or 30 min at 37°C. Reactions were stopped and samples were washed as described above for Western blotting experiments. To ensure equal well loading, aliquots were taken from the final PBS wash and plated to count CFU. Bacteria (3 × 107) in suspension were added to medium-binding polystyrene wells in triplicate (Costar/Corning) and left to dry overnight. Wells were blocked overnight at 4°C with 3% OVA. After extensive washing with PBS, primary Abs were added for 1-h incubations at room temperature, which included goat antisera to human C3 (Quidel), human C5 and human C7 (CompTech) each diluted to 1/10,000 in blocking buffer. HRP-conjugated rabbit anti-goat IgG (H + L) Ab (diluted to 1/2,000; Bio-Rad) was used as the secondary Ab for 1-h incubations at room temperature. Substrate was added for 10 min at room temperature (Bio-Rad), and the reaction was stopped with 2% oxalic acid. Absorbance at 415 nm was measured on a 96-well plate reader (Molecular Devices). Values obtained from samples containing EDTA (≤0.2 in each case) were subtracted out in each case.

C3-bound bacterial pellets were obtained as described above for Western blotting. To examine the nature of the C3 fragments bound to each strain, hydroxylamine (NH2OH) was used to cleave thioester bonds formed between C3 and bacterial acceptor molecules as described (24), with modifications. Briefly, after completing C3 deposition reactions and removal of unbound C3, samples were solubilized by boiling in 1% SDS for 5 min. Control samples were prepared immediately for Western blotting and paired samples were first incubated in 2 M NH2OH in 20 mM Tris-buffered H2O (pH 10.5) for 1 h at 37°C. Western blotting was performed as described above except that 16.5-cm gels were used to create better separation between fragments and to allow for increased sample loading. Also, 5% milk was used to block and dilute Abs (primary at 1/2000 and secondary at 1/1000). Ab incubations were done for 1 h at room temperature. ECL reagents were used for detection (Amersham Biosciences). Band densitometry was analyzed using Image J software available through the National Institutes of Health web site (http://rsb.info.nih.gov/ij/).

Bacteria, cultured overnight on chocolate II agar plates, were suspended in PBS at a concentration of 3 × 1010 CFU/ml, as determined by OD to equalize the amount of bacteria, and subsequently washed twice in PBS by centrifugation at 12,000 × g for 4 min with resuspension. The final pellet was resuspended in 200 μl of Laemmli loading buffer, boiled for 10 min, and then incubated with proteinase K (Invitrogen) at a final concentration of 10 mg/ml for 2 h at 65°C. Samples were boiled again for 10 min and stored at −20°C until used. LPS was separated by 12% SDS-PAGE and silver stained as described (25). Briefly, gels were fixed overnight in solution containing 40% EtOH and 5% acetic acid. Gels were then incubated in 0.7% periodic acid in fixing solution for 7 min and subsequently washed with multiple exchanges of water. The staining solution (0.013% concentrated ammonium hydroxide, 0.02 N NaOH, and 0.67% silver nitrate (w/v) was applied with vigorous agitation for 10 min, then gels were washed three times (each 10 min) in water. Gels were developed in solution containing 0.275% monohydrous citric acid (w/v) and 0.0025% formaldehyde. Upon completion, development was stopped using 5% acetic acid.

For Western blotting, strains were grown overnight on chocolate II agar plates. Bacteria were suspended at a concentration of 1 × 1010 CFU/ml (as determined by OD to equalize the amount of bacteria), pelleted, boiled in Laemmli’s sample buffer for 10 min, then incubated with 10 mg/ml proteinase K at 65°C for 2 h. The samples were separated by 12% SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were blocked overnight with 5% dehydrated milk and immunoblotting was performed with anti-F. tularensis FB-11 Ab (diluted to 1/1000; Abcam) or with anti-Fn 5 mAb (diluted to 1/1000; ImmunoPrecise) for 4 h at room temperature using goat anti-mouse IgG (diluted to 1/4000; Bio-Rad) for 2 h at room temperature as the secondary Ab. Blots were developed with 5-bromo-4-chloro-3-indolyl phosphate/NBT (Sigma-Aldrich) as the substrate.

To determine associations of significance between or within groups where indicated, data were analyzed using unpaired (one-tailed) or one-sample Student’s t tests or one-way ANOVA followed by either Bonferroni’s or Dunnett’s multiple comparison posttests (using GraphPad Prism 4 software). CFU data were converted to log values to equalize variances. Differences between groups were considered statistically significant for p values <0.05.

Since complement-mediated opsonization is necessary for efficient phagocytosis of Francisella by macrophages and dendritic cells, we compared the ability of five strains to survive complement-mediated lysis in healthy nonimmune human serum (Table I). Strains included type A virulent Ft subsp. tularensis (Schu S4), Fn, Ft subsp. holarctica live vaccine strain (LVS), a gray phase variant of LVS (LVSG), and a putative capsule-negative strain derived from LVS (LVSR). We also included DH5α, a laboratory E. coli strain, as a positive control for complement activity in serum because of its marked sensitivity to complement-mediated lysis. Bacteria were incubated with 5 or 50% (fresh or HI) donor serum for 1 h, the reaction stopped, and the bacteria subsequently plated to count surviving CFU (Fig. 1). Because complement components C2 and FB are irreversibly denatured by mild heat, HI serum served as a negative control for complement-mediated lysis. In each experiment, CFU obtained from incubations in GVB++ alone (reaction buffer) or in HI serum were equivalent. Results are presented as the ratio of surviving CFU in fresh vs HI serum. Data show that virulent strains known to cause disease in either humans or animal models (Schu S4, Fn, and LVS) are resistant to lysis in both 5 and 50% serum. By comparison, LVSG and LVSR were both susceptible to complement-mediated lysis as survival decreased by up to 1,000- and 100,000-fold, respectively.

Table I.

Summary of F. tularensis strains used

F. tularensis SubspeciesVirulence (human)LPS AntigenicityaCorresponding Strain Used
tularensis +++ tularensis type Schu S4 
holarctica ++ tularensis type LVSb 
   LVSGc 
   LVSRd 
   LVSΔwbtAe 
   LVSΔwbtMf 
novicida +/− novicida type Fn (U112) 
F. tularensis SubspeciesVirulence (human)LPS AntigenicityaCorresponding Strain Used
tularensis +++ tularensis type Schu S4 
holarctica ++ tularensis type LVSb 
   LVSGc 
   LVSRd 
   LVSΔwbtAe 
   LVSΔwbtMf 
novicida +/− novicida type Fn (U112) 
a

Based on LPS structure and recognition by mAbs specific for either subsp. tularensis or subsp. novicida.

b

Live vaccine strain derived from subsp. holarctica that is avirulent in humans, but retains virulence in mice.

c

LVSG phase variant recognized by either anti-tularensis or anti-novicida mAb (11 ).

d

LVSR mutant (11 ).

e

Polar LVS O Ag mutant (22 ).

f

Nonpolar LVS O Ag mutant (22 ).

FIGURE 1.

Susceptibility to complement-mediated lysis differs among Francisella strains. In brief, 2 × 106 bacteria/reaction were incubated in either fresh nonimmune serum (separate donor for each experiment) or heat-inactivated (HI) serum (devoid of complement activity) at concentrations of 5 or 50% for 1 h at 37°C, then washed in ice cold PBS. DH5α, a laboratory strain of E. coli that is highly susceptible to lysis by complement, served as a positive control for serum complement activity. Data are presented as percent survival, i.e., CFU obtained from reactions in fresh serum normalized to CFU from reactions in HI serum. Means ± SEM are given (n = 3). ∗, p < 0.0001 for one-sample Student’s t test with theoretical mean of 100. ND, Not detected.

FIGURE 1.

Susceptibility to complement-mediated lysis differs among Francisella strains. In brief, 2 × 106 bacteria/reaction were incubated in either fresh nonimmune serum (separate donor for each experiment) or heat-inactivated (HI) serum (devoid of complement activity) at concentrations of 5 or 50% for 1 h at 37°C, then washed in ice cold PBS. DH5α, a laboratory strain of E. coli that is highly susceptible to lysis by complement, served as a positive control for serum complement activity. Data are presented as percent survival, i.e., CFU obtained from reactions in fresh serum normalized to CFU from reactions in HI serum. Means ± SEM are given (n = 3). ∗, p < 0.0001 for one-sample Student’s t test with theoretical mean of 100. ND, Not detected.

Close modal

Complement component C3 is the central component of the complement cascade. Since C3 fragment deposition is necessary for complement-mediated lysis and for opsonization, we examined deposition on each of the five Ft subspecies/variant strains described above by Western blotting. Bacteria were incubated in 10% fresh or HI donor serum for 30 min and then washed extensively to remove unbound C3 before sample preparation. Native C3 is a heterodimer composed of a 120-kDa α-chain and a 75-kDa β-chain. Upon activation, C3a is cleaved from the α-chain leaving C3bα′, a 110-kDa fragment. Upon fixation, C3bα′ covalently binds acceptors on the bacterial surface and, depending on the acceptor, will migrate as a band greater than 110 kDa. Also, both the binding of multiple acceptors of varying molecular mass and the further degradation of C3bα′ can result in a ladder-type banding pattern. The β-chain, however, is released by reduction of C3b cysteine bonds in sample buffer and is seen as a 75-kDa band.

C3 deposition was apparent on complement-resistant Ft strains (Schu S4, Fn, and LVS) indicative of complement activation (Fig. 2,A). C3 deposition on complement-susceptible strains (LVSG and LVSR) was performed using C5d serum (Fig. 2,B). C5 is a component of the terminal lytic pathway and its absence does not affect C3 deposition, but prevents bacterial lysis. To confirm that C3 deposition is unaffected, assays using resistant strains were performed with both C5d and fresh donor sera for comparison (Fig. 2 A). C3 associated with LVSG and LVSR was markedly increased compared with resistant strains. Also, we found that LVSG bound less C3 compared with LVSR.

FIGURE 2.

Complement component C3 deposition occurs in greater amounts on complement-susceptible strains of Francisella. In brief, 5 × 108 wild-type (A) or variant strain (B) bacteria/reaction were incubated in buffer alone or in 10% fresh, HI, or C5d serum for 30′ at 37°C, washed in ice-cold PBS containing protease inhibitors, boiled in Laemmli’s buffer, separated by 7.5% SDS-PAGE (5 × 107 bacteria/lane unless diluted), and examined for C3 by Western blotting. Aliquots from each sample were plated to count CFU just before lysis to equalize loading amounts. Goat antiserum to human C3 was used for detection. Control lanes contain 2 ng of purified native C3. The blot shown is representative of at least three independent experiments. C3α, 120-kDa chain; C3β, 75-kDa chain.

FIGURE 2.

Complement component C3 deposition occurs in greater amounts on complement-susceptible strains of Francisella. In brief, 5 × 108 wild-type (A) or variant strain (B) bacteria/reaction were incubated in buffer alone or in 10% fresh, HI, or C5d serum for 30′ at 37°C, washed in ice-cold PBS containing protease inhibitors, boiled in Laemmli’s buffer, separated by 7.5% SDS-PAGE (5 × 107 bacteria/lane unless diluted), and examined for C3 by Western blotting. Aliquots from each sample were plated to count CFU just before lysis to equalize loading amounts. Goat antiserum to human C3 was used for detection. Control lanes contain 2 ng of purified native C3. The blot shown is representative of at least three independent experiments. C3α, 120-kDa chain; C3β, 75-kDa chain.

Close modal

We performed ELISAs to better quantify total C3 deposition and to examine fixation of the downstream complement components C5 and C7. C8 is a component of the MAC and binds to the microbial surface subsequently to C7 deposition. Thus, we incubated each strain in 10% C8d serum for 5 or 30 min at 37°C to assess fixation of the upstream components. The above Western blotting results are supported by the finding that LVSR bound significantly greater amounts of total C3 (Fig. 3). LVSG bound an intermediate amount, but still more than each wild-type strain. At 5 min, compared with LVSR, wild-type strains did not bind C5 and LVSG bound significantly less than LVSR. At 30 min, C5 that bound to wild-type strains increased, but the amount bound to LVSG increased dramatically and was comparable to the amount fixed by LVSR. For C7 (a component of the MAC), a similar trend occurred in that there was little binding at 5 min except on LVSR. Then, at 30 min, binding to wild-type strains remained low and binding to LVSG increased to approach the level fixed by LVSR. Together, these findings provide strong evidence that the regulation of C3 deposition, which results in a marked reduction in the deposition of downstream components, is crucial for Francisella resistance to complement-mediated lysis.

FIGURE 3.

Quantitative analysis of complement components C3, C5, and C7 fixed by Francisella. For 5 and 30 min at 37°C, 3 × 108 bacteria/reaction were incubated in 10% fresh C8d serum or C8d serum containing 10 mM EDTA to block complement activity, washed in ice-cold PBS containing protease inhibitors, and resuspended in H2O. Aliquots from each sample were plated to count CFU. In brief, 3 × 107 bacteria/well were applied to 96-well plates and dried overnight. Goat antisera to human C3, C5, and C7 were used for detection. Absorbance at 415 nm was measured and values obtained using EDTA in serum were subtracted from matched values obtained using fresh serum. Means ± SEM are given (n = 3). For comparisons between LVS, LVSG, and LVSR, significant mean differences were determined by ANOVA followed by Bonferroni’s multiple comparison posttests. ∗, p < 0.05.

FIGURE 3.

Quantitative analysis of complement components C3, C5, and C7 fixed by Francisella. For 5 and 30 min at 37°C, 3 × 108 bacteria/reaction were incubated in 10% fresh C8d serum or C8d serum containing 10 mM EDTA to block complement activity, washed in ice-cold PBS containing protease inhibitors, and resuspended in H2O. Aliquots from each sample were plated to count CFU. In brief, 3 × 107 bacteria/well were applied to 96-well plates and dried overnight. Goat antisera to human C3, C5, and C7 were used for detection. Absorbance at 415 nm was measured and values obtained using EDTA in serum were subtracted from matched values obtained using fresh serum. Means ± SEM are given (n = 3). For comparisons between LVS, LVSG, and LVSR, significant mean differences were determined by ANOVA followed by Bonferroni’s multiple comparison posttests. ∗, p < 0.05.

Close modal

The nature of bound C3 fragmentation can determine the outcome of complement activation. C3b is necessary for activation of the terminal lytic pathway of complement, but smaller cleavage fragments (including C3bi, C3dg, and smaller fragments) do not initiate lysis. Of particular interest is C3bi because, like C3b, it mediates opsonophagocytosis. On Ft, inactivation of C3b to C3bi would account for both resistance to lysis and C3-mediated opsonization. We assessed the nature of C3 fragmentation by Western blot analysis (Fig. 4). Importantly, the study of fragment deposition by immunoblotting is confounded by the fact that C3α forms complexes via covalent bonds, thereby affecting band migration. At physiological pH, C3α acceptor thioester bonds form more readily than amide bonds (26). We used hydroxylamine treatment, which cleaves ester linkages, to release C3α fragments from acceptor and C5 convertase complexes. Treatment resulted in the disappearance of the majority of high molecular mass bands.

FIGURE 4.

The nature of bacteria-bound C3 fragments for different Francisella strains over time. In brief, 7.5 × 108 wild-type (A) or variant strain (B) bacteria/reaction were incubated in 10% C5d serum from 1 to 60 min at 37°C. After washing in ice-cold PBS containing protease inhibitors, bacterial pellets were lysed by boiling in 1% SDS. To allow for the quantitative assessment of fragmentation, hydroxylamine (NH2OH) was used to release C3bα′ (and smaller α-chain fragments) from covalent thioester bonds to bacterial acceptor molecules. Immunoblotting for C3 was performed as in Fig. 2, except that each lane contains 1 × 108 bacteria and a 16.5-cm gel was used for SDS-PAGE to allow for greater band separation. Control lanes contain 2 ng each of native C3 and C3bi. Each blot is representative of at least three independent experiments. C3α, 120-kDa native C3 α-chain; C3bα′, 110-kDa C3b α-chain; C3β, 75-kDa native C3 β-chain; and C3biα1′, 68-kDa C3bi α-chain fragment. C, Densitometry ratios of C3biα1′ to C3bα′ at early time points are compared for each strain. For variant strains, ratios ≤1 indicate a higher rate of C3b deposition compared with the rate of C3b to C3bi conversion. Mean ratios ± SEM for the densitometry of independent blots are given (n = 2 for Schu S4 and LVSG and n = 3 for LVS, Fn, and LVSR).

FIGURE 4.

The nature of bacteria-bound C3 fragments for different Francisella strains over time. In brief, 7.5 × 108 wild-type (A) or variant strain (B) bacteria/reaction were incubated in 10% C5d serum from 1 to 60 min at 37°C. After washing in ice-cold PBS containing protease inhibitors, bacterial pellets were lysed by boiling in 1% SDS. To allow for the quantitative assessment of fragmentation, hydroxylamine (NH2OH) was used to release C3bα′ (and smaller α-chain fragments) from covalent thioester bonds to bacterial acceptor molecules. Immunoblotting for C3 was performed as in Fig. 2, except that each lane contains 1 × 108 bacteria and a 16.5-cm gel was used for SDS-PAGE to allow for greater band separation. Control lanes contain 2 ng each of native C3 and C3bi. Each blot is representative of at least three independent experiments. C3α, 120-kDa native C3 α-chain; C3bα′, 110-kDa C3b α-chain; C3β, 75-kDa native C3 β-chain; and C3biα1′, 68-kDa C3bi α-chain fragment. C, Densitometry ratios of C3biα1′ to C3bα′ at early time points are compared for each strain. For variant strains, ratios ≤1 indicate a higher rate of C3b deposition compared with the rate of C3b to C3bi conversion. Mean ratios ± SEM for the densitometry of independent blots are given (n = 2 for Schu S4 and LVSG and n = 3 for LVS, Fn, and LVSR).

Close modal

Bacteria were incubated with 10% C5d serum from 1 to 60 min and we compared hydroxylamine-treated vs untreated samples. For resistant strains (Schu S4, LVS, and Fn), overall C3 fragment deposition occurred rapidly and increased over time (Fig. 4,A). In hydroxylamine-treated groups, the appearance of C3biα1′ (68-kDa C3biα′ fragment) occurred in excess of C3bα′ (110 kDa). For susceptible strains (LVSG and LVSR), the overall rate of C3 fragment deposition was increased compared with resistant strains (Fig. 4,B). Although C3bi deposition also occurred on LVSG and LVSR, the persistent and increasing appearance of C3bα′ was in stark contrast to its relative absence on resistant strains. Also, the greater amounts of C3bα′ on LVSR at early time points, compared with LVSG, correlate with its greater susceptibility to lysis. At early time points, ratios of C3biα1′ band densities to C3bα′ band densities illustrate more rapid conversion of C3b to C3bi on resistant strains (Fig. 4 C). Unfortunately, densitometry is less quantitative as the band intensities for C3 deposition become saturating for LVSG and LVSR. Ratios of <1 for LVSG and LVSR at early time points, before the saturation of band intensities, indicate that new C3 deposition occurs more rapidly than conversion of bound C3b to C3bi. For Francisella, these results indicate the importance of rapid conversion of C3b to C3bi for resistance to complement-mediated lysis. Also, we conclusively show that complement activation by Schu S4, LVS, and Fn leads to opsonization characterized by persistent C3bi deposition.

Once we confirmed that Francisella activates complement by demonstrating C3 deposition, we explored mechanisms of upstream activation. With the use of chelators in serum viability assays, we began to differentiate the relevant complement activation pathways targeting Ft. Initiation of the classical pathway is typically dependent upon surface recognition by IgM or IgG and subsequent recruitment of C1 (a complex formed by C1q, C1r, and C1s). The interaction between C1q and C1r is stabilized by Ca2+ and the serine protease activity associated with C1s requires Mg2+ (27). Similarly, lectin pathway-associated serine proteases require Ca2+ for activity. There is no Ca2+ requirement associated with the alternative pathway; however, Mg2+ is necessary for the interaction between FB and C3b. We used EDTA, which chelates both Ca2+ and Mg2+, to block all three activation pathways and EGTA, which chelates Ca2+, to specifically block the classical and lectin-mediated pathways (Fig. 5). The inclusion of EDTA and EGTA inhibited complement activity (restored viability) in serum concentrations as low as 1% where the classical and lectin, but not alternative, pathways are active. Bacterial lysis at concentrations higher than 10% was not completely inhibited by EGTA, indicative of some alternative pathway activity. Compared with the viability in concentrations of fresh serum above 10%, however, markedly less killing occurred in the presence of EGTA, suggesting the involvement of more than one pathway at these higher concentrations of serum. The implications of these data are that C3bBb (alternative pathway C3 convertase) formation occurs on both LVSG and LVSR, but that C4b2a (classical and lectin pathway C3 convertase) formation is predominant. Once C4b2a initiates C3b deposition, amplification of C3b deposition via C3bBb likely occurs.

FIGURE 5.

Complement activation by susceptible Francisella strains occurs via more than one pathway. LVSG and LVSR were incubated in increasing concentrations of fresh or HI serum or fresh serum containing 10 mM EDTA or 10 mM EGTA with 7 mM Mg2+ for 60 min at 37°C. EDTA inhibits all complement activation pathways by chelating both Ca2+ and Mg2+. EGTA (with Mg2+) inhibits the classical and lectin-mediated pathways by specific Ca2+ chelation. Viable bacteria were plated to count CFU. Results indicate that activation of the alternative pathway occurs at concentrations above 10%. Also, Ca2+-dependent (classical and/or lectin-mediated) pathways are activated in a concentration-dependent manner in serum concentrations as low as 1%. Bars represent means ± SEM of independent reactions in sera from separate donors (n = 3). For each serum test group, significant mean differences from controls (no serum) were determined by ANOVA. ∗, p < 0.01 for each Dunnett multiple comparison posttest. ND, Not detected.

FIGURE 5.

Complement activation by susceptible Francisella strains occurs via more than one pathway. LVSG and LVSR were incubated in increasing concentrations of fresh or HI serum or fresh serum containing 10 mM EDTA or 10 mM EGTA with 7 mM Mg2+ for 60 min at 37°C. EDTA inhibits all complement activation pathways by chelating both Ca2+ and Mg2+. EGTA (with Mg2+) inhibits the classical and lectin-mediated pathways by specific Ca2+ chelation. Viable bacteria were plated to count CFU. Results indicate that activation of the alternative pathway occurs at concentrations above 10%. Also, Ca2+-dependent (classical and/or lectin-mediated) pathways are activated in a concentration-dependent manner in serum concentrations as low as 1%. Bars represent means ± SEM of independent reactions in sera from separate donors (n = 3). For each serum test group, significant mean differences from controls (no serum) were determined by ANOVA. ∗, p < 0.01 for each Dunnett multiple comparison posttest. ND, Not detected.

Close modal

To further explore mechanisms of complement activation, we used C1qd serum (C1q-depleted serum deficient of classical pathway activity; Fig. 6,A) or FBd serum (serum deficient of alternative pathway activity; Fig. 6,B) to test survival of complement-susceptible strains. Lysis of LVSG did not occur in 5, 10, or 20% C1qd serum, but was restored in C1qr serum. Lysis of LVSR did not occur in 5% C1qd serum and occurred to only a small extent in 10 and 20% C1qd serum, but was completely restored in C1qr serum. In FBd and FBr sera, lysis occurred that was comparable to C1qr serum for both strains. Also, compared with FBd serum, lysis appeared to increase in 20% FBr serum (significant only for LVSR), which supports a minor role for alternative pathway activation. Taken together, results illustrated in Figs. 5 and 6 demonstrate two important findings. First, the classical pathway, but not the alternative pathway, is necessary for the optimal lysis of complement-susceptible strains at both low and high serum concentrations. Second, in the absence of an active classical pathway, either by Ca2+ chelation or by depletion of C1q, complement-susceptible strains can activate the alternative pathway to a small extent.

FIGURE 6.

Optimal lysis of variant Francisella strains is dependent upon C1q. LVSG and LVSR were incubated in 5–20% C1qd and C1qr (A) or FBd and FBr (B) sera for 1 h at 37°C. Viable bacteria were plated to count CFU. Minimal, but significant, lysis of LVSR occurs in 10 or 20% C1qd serum, which implicates a minor role for the alternative or lectin pathway. Repletion with C1q, however, restores optimal lysis of both strains. Conversely, repletion of FBd serum has no effect on lysis compared with FBd serum (except for LVSR at 20%). Bars are the means ± SEM of independent reactions (n = 3). For each serum test group, significant mean differences from controls (no serum) were determined by ANOVA. ∗, p < 0.01 for each Dunnett multiple comparison posttest. ∗∗, p < 0.05 for one-tailed Student’s t tests for significance between depleted and replete sera. ND, Not detected.

FIGURE 6.

Optimal lysis of variant Francisella strains is dependent upon C1q. LVSG and LVSR were incubated in 5–20% C1qd and C1qr (A) or FBd and FBr (B) sera for 1 h at 37°C. Viable bacteria were plated to count CFU. Minimal, but significant, lysis of LVSR occurs in 10 or 20% C1qd serum, which implicates a minor role for the alternative or lectin pathway. Repletion with C1q, however, restores optimal lysis of both strains. Conversely, repletion of FBd serum has no effect on lysis compared with FBd serum (except for LVSR at 20%). Bars are the means ± SEM of independent reactions (n = 3). For each serum test group, significant mean differences from controls (no serum) were determined by ANOVA. ∗, p < 0.01 for each Dunnett multiple comparison posttest. ∗∗, p < 0.05 for one-tailed Student’s t tests for significance between depleted and replete sera. ND, Not detected.

Close modal

After showing that C1q is required for the optimal lysis of LVSG and LVSR, we determined its role in mediating C3 deposition on both complement-resistant and -susceptible strains. As described above, C3 deposition was analyzed by Western blotting, but here we used 10% C5d, C1qd, and C1qr sera as sources of complement (Fig. 7). Since susceptible bacteria are lysed in C1q-sufficient serum within 1 h (Fig. 6), we tested survival at earlier time points and found that significant lysis did not occur in 15-min assays (data not shown). Thus, 15-min serum incubations were performed to evaluate C3 deposition on complement-susceptible strains in C1qr serum. Results demonstrate the requirement of C1q for optimal C3 deposition on each strain. Our cumulative data provide strong evidence that the classical pathway has a dominant role in mediating complement activation by Francisella.

FIGURE 7.

Deposition of C3 on both wild-type and variant strains is C1q dependent. In brief, 5 × 108 bacteria/reaction were incubated in 10% C5d, C1qd, or C1qr serum for 15 min at 37°C. Western blot analysis of C3 deposition was performed as in Fig. 2. Each blot is representative of at least three independent experiments.

FIGURE 7.

Deposition of C3 on both wild-type and variant strains is C1q dependent. In brief, 5 × 108 bacteria/reaction were incubated in 10% C5d, C1qd, or C1qr serum for 15 min at 37°C. Western blot analysis of C3 deposition was performed as in Fig. 2. Each blot is representative of at least three independent experiments.

Close modal

Thus far, our examinations of complement activation by LVS variant strains used LVSG and LVSR. Both have been shown to express structurally altered LPS O Ag compared with LVS (11). In this study, we used an LVS wbtA mutant (LVSΔwbtA), with deletion of a gene within the O Ag operon and devoid of LPS O Ag, to directly test the effect of LPS O Ag expression on complement activation (Table I and Fig. 8). Survival assays were performed as described above to examine effects on complement-mediated lysis. In 5 and 50% fresh donor serum, lysis of LVSΔwbtA did not differ significantly from lysis of LVSR (Fig. 8,A). Survival assays using fresh serum with the addition of chelators (data not shown) and using component deficient sera (Fig. 8,B) were repeated using LVSΔwbtA to assess important activation pathways. Results correlated with those previously described for LVSR in that a similar degree of lysis occurred for each serum group and that the classical pathway was necessary for optimal lysis. Next, Western blots of C3 deposition were performed to determine the effect of O Ag on the temporal nature and mechanism of C3 fixation. Like LVSR, and unlike LVS, LVSΔwbtA rapidly fixed total C3-derived fragments over time and specifically fixed a markedly increasing amount of C3bα′ (Fig. 8,C). Note that even at 1 min, when amounts of C3β are similar between the three strains, amounts of C3bα′ are greater on the susceptible strains. The use of C5d, C1qd, and C1qr sera as complement sources demonstrated that optimal C3 fixation by LVSΔwbtA requires functional classical pathway activity (Fig. 8 D).

FIGURE 8.

Complement susceptibility is determined by LPS O Ag expression by Francisella. LVSΔwbtA and LVSΔwbtM, LVS mutants devoid of O Ag, were tested for sensitivity to complement-mediated lysis and for C3 fixation. A, Survival assays in human nonimmune fresh donor and HI sera were done as described for Fig. 1. Mean ratios ± SEM are given (n = 3). ns, No significance by unpaired Student’s t test. B, LVSΔwbtA survival assays using C1qd, C1qr, FBd, or FBr serum and statistical analyses were done as described for Fig. 6. Means ± SEM are given (n = 3). ND, Not detected. C, Kinetic analyses of C3 fragments (released from bacterial acceptors using NH2OH) bound to LVS, LVSR, and LVSΔwbtA were done as described in Fig. 4. The blot is representative of two independent experiments. D, Immunoblotting was done as described in Fig. 4, showing the requirement of C1q for optimal C3 fixation by LVSΔwbtA. The blot is representative of three independent experiments. E, Whole cell bacterial lysates (containing 1 × 1010 CFU) were subjected to 12% SDS-PAGE and immunoblotting was performed using an anti-F. tularensis O Ag Ab. The blot shown is representative of two independent experiments. F, Survival assays in human nonimmune fresh donor and HI sera were done as described for Fig. 1 using LVS, LVSΔwbtM, LVSΔwbtM containing an empty complementation plasmid (LVSΔwbtM:pFTNAT), and the complemented mutant strain expressing O Ag (LVSΔwbtM:pFTNAT-wbtM). Mean ratios ± SEM are given (n = 3). G, Analysis of C3 fragments (released from bacterial acceptors using NH2OH) fixed by each strain described in F. C3 fixation occurred in 10% C5d serum at 37°C for 10 min. Western blots were performed as described in Fig. 4 and the blot shown is representative of two independent experiments.

FIGURE 8.

Complement susceptibility is determined by LPS O Ag expression by Francisella. LVSΔwbtA and LVSΔwbtM, LVS mutants devoid of O Ag, were tested for sensitivity to complement-mediated lysis and for C3 fixation. A, Survival assays in human nonimmune fresh donor and HI sera were done as described for Fig. 1. Mean ratios ± SEM are given (n = 3). ns, No significance by unpaired Student’s t test. B, LVSΔwbtA survival assays using C1qd, C1qr, FBd, or FBr serum and statistical analyses were done as described for Fig. 6. Means ± SEM are given (n = 3). ND, Not detected. C, Kinetic analyses of C3 fragments (released from bacterial acceptors using NH2OH) bound to LVS, LVSR, and LVSΔwbtA were done as described in Fig. 4. The blot is representative of two independent experiments. D, Immunoblotting was done as described in Fig. 4, showing the requirement of C1q for optimal C3 fixation by LVSΔwbtA. The blot is representative of three independent experiments. E, Whole cell bacterial lysates (containing 1 × 1010 CFU) were subjected to 12% SDS-PAGE and immunoblotting was performed using an anti-F. tularensis O Ag Ab. The blot shown is representative of two independent experiments. F, Survival assays in human nonimmune fresh donor and HI sera were done as described for Fig. 1 using LVS, LVSΔwbtM, LVSΔwbtM containing an empty complementation plasmid (LVSΔwbtM:pFTNAT), and the complemented mutant strain expressing O Ag (LVSΔwbtM:pFTNAT-wbtM). Mean ratios ± SEM are given (n = 3). G, Analysis of C3 fragments (released from bacterial acceptors using NH2OH) fixed by each strain described in F. C3 fixation occurred in 10% C5d serum at 37°C for 10 min. Western blots were performed as described in Fig. 4 and the blot shown is representative of two independent experiments.

Close modal

Based on the observed effects of wbtA deletion on complement activation by the mutant bacteria, we characterized O Ag expression by both wild-type and variant strains examined in our studies. Using whole cell Fn, LVS, LVSG, LVSR, and LVSΔwbtA lysates, we analyzed O Ag expression by Western blot (Fig. 8 E) and silver stain. Using an anti-F. tularensis-type O Ag Ab for immunoblotting, a typical laddering pattern is demonstrated for LVS with a grouping of full-length chains near the top of the membrane and clearer delineation of bands representing progressively shorter chains. For LVSG, an identical pattern is evident, but total O Ag expression is reduced compared with LVS. This implies that LVS and LVSG express O Ag of similar length, but that the amount of O Ag per bacterium (or the amount of O Ag expressing bacteria in a population) is reduced for LVSG. O Ag was not detected for LVSR or, as expected, LVSΔwbtA. These results were duplicated by silver stain (data not shown) in that O Ag banding was evident for LVS, was less evident for LVSG, and was absent for LVSR and LVSΔwbtA. Using an anti-F. novicida-type O Ag Ab, we detected full-length O Ag expression by Fn, but not by LVS or any of the LVS-derived variant strains (data not shown). Thus, wild-type strains, which resist complement-mediated lysis and rapidly mediate conversion of surface-bound C3b to C3bi, express abundant, full-length O Ag. LVSR and LVSΔwbtA, which are highly susceptible to complement-mediated lysis and rapidly fix high amounts of C3bα′, do not express O Ag. LVSG, which is moderately susceptible to lysis and fixes persistent C3bα′ (albeit less rapidly compared with LVSR and LVSΔwbtA), expresses a relatively intermediate amount of O Ag.

To confirm the regulatory role of O Ag in mediating resistance to complement-mediated lysis and in mediating conversion of C3b to C3bi, we used an additional mutant strain devoid of wbtM (LVSΔwbtM), which is a gene downstream of wbtA in the O Ag operon. In addition, we used the complemented strain containing the pFTNAT plasmid expressing functional wbtM as previously described (22). We found that compared with wild-type LVS and with the complemented mutant (LVSΔwbtM:pFTNAT-wbtM), LVSΔwbtM and LVSΔwbtM that contained an empty plasmid (LVSΔwbtM:pFTNAT) were susceptible to complement-mediated lysis in both 5 and 50% fresh donor serum at 1 h (Fig. 8,F). The degree of susceptibility strongly correlates with both LVSΔwbtA and LVSR. By ELISA, we also found that in 10% C8d serum, LVSΔwbtM fixed relatively high amounts of C3, C5, and C7 (comparable to LVSR in Fig. 3), but that LVSΔwbtM:pFTNAT-wbtM fixed less total C3 and minimal C5 and C7 (comparable to wild-type strains in Fig. 3; data not shown). Finally, we determined the nature of C3-derived fragments fixed in 10% C5d serum at 10 min. We found high amounts of C3bα′ on LVSΔwbtM and LVSΔwbtM:pFTNAT, but much higher amounts of C3biα1′ compared with C3bα′ fixed to LVS and to LVSΔwbtM:pFTNAT-wbtM (Fig. 8 G). We conclude that relative O Ag expression correlates strongly with the degree of complement activation by each strain.

The rapid onset and disease progression of pneumonic tularemia, despite low inoculation doses, implicates a failure of innate immune responses to control infection with Francisella. That Ft survives and replicates within phagocytes is particularly indicative of an ineffective microbicidal reaction to infection. However, bacilli must also survive exposure to extracellular mediators of innate immunity, including complement, as a prerequisite for cellular invasion. It has been established that Ft survives in whole blood both in vivo and ex vivo (28, 29). Importantly, pneumonic tularemia, the most severe form of tularemia, can develop secondary to cutaneous or mucosal infections. Clearly, Ft-complement interactions in serum or interstitial fluid would impact the outcome of secondary pneumonic disease. Complement is also abundant in bronchoalveolar fluid and, thus, likely affects primary pneumonic disease as well (30, 31, 32). Direct evidence of potent classical pathway activity in isolated human bronchoalveolar lavage fluid was demonstrated by C3 fixation by Mycobacterium tuberculosis and group B streptococcus (33, 34). The importance of complement in pulmonary immunity is exemplified in patients who are genetically deficient in either complement components or complement receptors, since they are at significantly increased risk for respiratory infection (35).

In addition to the direct mediation of pathogen lysis, complement modulates innate immune responses via the release of component fragments with anaphylactic activity and via opsonophagocytosis, which would affect cell-mediated responses. Matrix metalloproteinase 9-deficient mice have increased resistance to Schu S4 infection, likely due to its role in recruiting highly active neutrophils to the lung (36). Anaphylactic complement fragments may have a similar role, and since the rate of Francisella phase variation increases upon intracellular or in vivo growth, increased complement activation by gray variants might compound local inflammation. Unfortunately, the role of complement in mediating disease in animal models caused by type A strains or Fn (the most virulent strains for mice) has not been studied. Opsonization is influential beyond simply increasing the rate of particulate uptake. Depending on the identity of the opsonin (e.g., IgG vs C3bi) and associated surface receptors, downstream signaling events differ. Phagosomal trafficking, cytokine responses, and production of reactive oxygen intermediates are all influenced by the exact nature of receptor-ligand interactions (37).

Our data demonstrate complement activation by each Francisella strain tested. Several studies previously addressed susceptibility of Fn, LVS, and respective derivative strains to complement-mediated lysis (discussed below). We extended these studies to include Schu S4 and we provide an analysis of complement activation at the level of surface component deposition. Since it was shown that CR3 and CR4 have a major role in opsonophagocytosis of Ft by human monocyte-derived phagocytes (18, 19, 20) and the primary ligand for these receptors is C3bi, we hypothesized that Francisella would fix C3b and that conversion to C3bi would ensue. Conversion to C3bi and smaller fragments would also account for the ability of bacilli to survive extracellularly in vivo. We characterize the nature of Ft-bound C3-derived fragments and show directly that opsonization with C3bi occurs. Our finding that LVS fixes C3 contradicts an earlier study, which reported that LVS did not bind radioactively labeled C3 when both were added to human serum (21). Since we found that rapid C3b conversion occurs on the surface of LVS, it is possible that C3 cleavage adversely affected radioactive labeling in that study. We also compared the nature of C3 deposition on strains resistant to complement-mediated lysis to its nature on susceptible strains. We found that, on resistant strains, conversion of C3bα′ to C3biα′ occurred more rapidly than deposition of new C3bα′. We extended our studies to show that the classical pathway is required for optimal complement activation by each strain tested, a finding that has previously been limited to serum-susceptible variant strains. Finally, we found that O Ag expression is a major determinant of complement activation. The strength of this finding is increased by the fact that multiple LVS variant strains, generated by distinct methodologies, were used. These include the selection of a spontaneous colony morphology variant (LVSG), the creation of a strain by random mutagenesis of LVS followed by the selection of a rough colony morphology variant (LVSR), and the construction of targeted gene mutants (LVSΔwbtA and LVSΔwbtM).

A role for capsule has previously been implicated in the serum resistance of Francisella (38). We chose to study LVSR, a putative capsule negative (Cap) strain derived from LVS by Sandstrom et al. (21) and subsequently studied by Cowley et al. (11), to characterize the effect of capsule production on complement component deposition. For other bacteria, encapsulation has been shown to reduce C3 deposition and to protect against subsequent lysis (39, 40, 41). The specific nature of the Ft capsule has not been determined conclusively. Hood (42) reported that decapsulation of Ft occurs in hypertonic saline and that capsular material is biochemically distinct from the cell wall of decapsulated bacilli. Recently, a putative capsule locus in the LVS genome, containing orthologous genes to capB and capC of Bacillus anthrasis, was reported (43). In some studies, an electron-lucent material typical of a loose capsule can clearly be seen surrounding bacilli grown in defined medium (44, 45). Despite these reports, several laboratories, including ours, have been unable to conclusively identify capsular material by microscopy, possibly due to the use of rich medium for culturing (19, 46). In studies of serum resistance, Cap strains (including LVSR) were shown to activate complement (21, 47). Sorokin et al. (47) reported, however, that the Cap strain used in their study exhibited a truncated O Ag, which may have been the true determinant of complement susceptibility. To substantiate the designation of LVSR as a Cap mutant, electron micrographs comparing LVSR to LVS were presented by Sandstrom et al. (21). However, the represented LVS capsule does not resemble the images of Francisella capsule shown in the above-cited studies. Furthermore, O Ag expression by LVSR was not studied. We conclude, based on our studies, that mutations in LVSR affect O Ag expression and not capsule production. However, that O Ag is a constituent of capsular polysaccharide cannot be ruled out. Also, the possibility exists that encapsulation occurs only under specific growth conditions and provides an additional measure of protection against complement.

Because we did not detect encapsulation on serum-resistant wild-type strains, we explored alternative mechanisms of resistance, including the effect of LPS expression. Compared with other Gram-negative pathogens, Francisella LPS is unique (48). The Francisella LPS O Ag is composed of a repeating carbohydrate tetramer that is identical for all subspecies tested thus far except for Fn, which has two novel sugars in the tetramer (49, 50). It is unlikely, however, that this structural difference in O Ag affects complement activation since Schu S4 and LVS do not differentially fix C3 fragments compared with Fn (Fig. 2). Previously, O Ag mutant strains derived from both LVS and Fn have been found to be susceptible to lysis in serum, unlike the parent strains (46, 51, 52, 53). Also, LVS gray variants that lack O Ag are similarly susceptible to lysis in serum (12). We repeated these results using LVSΔwbtA (Fig. 8,A) and LVSΔwbtM (Fig. 8,F). Furthermore, we show similarities relating to serum resistance and C3 fixation between LVSR, LVSΔwbtA, and LVSΔwbtM (Fig. 8). Since the absence of O Ag expression is also similar for each strain (Fig. 8,E) and since complementation of the LVSΔwbtM mutant restores resistance to lysis and efficient C3b to C3bi conversion (Fig. 8, F and G), we conclude that loss of O Ag expression is the primary cause of serum susceptibility.

In addition to identifying a role for O Ag, we show that complement resistance is due, in part, to a reduction of C3 on wild-type strains. Our data indicate that this is the result of rapid conversion of fixed C3b to C3bi, which would limit C3 convertase-mediated amplification of C3b deposition (Fig. 4). C3b inactivation is a common resistance mechanism employed by Gram-negative bacteria that are human pathogens and C3bi fixation enables efficient uptake by host phagocytes (26, 54, 55, 56). We are currently evaluating two hypotheses to explain C3b cleavage based on known microbial mechanisms of complement evasion. The first involves direct bacterial protease expression and the second involves the recruitment of host-derived negative regulators of complement. Importantly, a sufficient characterization of the mechanisms of C3b inactivation would include the evaluation of both hypotheses, since neither is mutually exclusive and either might be affected by O Ag expression. O Ag expression is reduced or absent on each complement-susceptible strain used in the present study and each was derived by a distinct methodology. Thus, the existence of unidentified Ft serum resistance factors, unaffected by O Ag, is unlikely. We do not propose that O Ag acts as an acceptor for serum-derived negative regulators (involved in C3b cleavage) because C3b to C3bi conversion occurs on LVSR, LVSΔwbtA, and LVSΔwbtM (Fig. 8). That C3b conversion occurs on complement-susceptible strains, albeit at a slower rate than C3b deposition, signifies the relative importance of O Ag expression for survival. Other potential mechanisms of resistance to complement-mediated lysis cannot be ruled out such as steric hindrance to the formation of the MAC or the fixation of complement components by moieties distal to the outer membrane. However, we show that limited C5 and C7 (a component of the MAC) fixation occurs in the presence of O Ag (Fig. 3), which is consistent with its importance in regulating C3 deposition which then secondarily affects components downstream in the complement cascade.

The important role of the classical pathway was previously demonstrated by Sorokin et al. (47) who reported lysis of variant strains in serum. Consumption of complement activity, indicative of activation, in pooled human serum was mediated by whole bacteria, isolated outer membrane, and purified LPS. Significantly less complement consumption occurred in preabsorbed serum, which was depleted of specific bactericidal Abs by incubating pooled sera with acetone-dried bacteria at 4°C. Repletion of preabsorbed serum with anti-whole cell or anti-LPS Abs restored complement consumption. With respect to LVSR, Sandstrom et al. (21) reported that depletion of both IgM and C4 in human serum dramatically decreased lysis. In the present study, results using C1qd serum confirm the importance of the classical pathway not only for the lysis of variant strains, but also for complement activation by virulent strains (Figs. 6–8). Previously, we reported the identification of natural anti-Francisella Abs (using Fn) in human nonimmune serum, which might have a role in complement activation by the classical pathway (18).

In summary, we show that the classical pathway, and to a lesser extent the alternative pathway, is activated by Francisella wild-type and LPS variant strains. We present conclusive evidence that complement is activated, based on C3 fixation, by virulent Francisella strains including Schu S4, LVS, and Fn. Rapid conversion of C3b to C3bi on these strains contributes to their ability to resist complement-mediated lysis. LPS variant strains derived from LVS were susceptible to complement-mediated lysis, due in part to limited C3b to C3bi conversion that led to striking increases in C3bα′ fragment deposition and to increased binding of components of the terminal lytic pathway (including C5 and C7). Finally, we identify O Ag expression as a key determinant for the outcome of complement activation by Francisella because it regulates deposition of terminal lytic pathway components due to an increased rate of C3b to C3bi conversion.

We thank Dr. Michael Pangburn for helpful discussions and advice.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was sponsored by the National Institutes of Health/National Institute of Allergy and Infectious Diseases Regional Center of Excellence for Biodefense and Emerging Infectious Diseases Research Program. We acknowledge membership within and support from the Region V “Great Lakes” Regional Center of Excellence (National Institutes of Health Award 1-U54-AI-057153). C.D.C. is recipient of a National Institutes of Health Training Grant T32 AI055411.

4

Abbreviations used in this paper: MAC, membrane attack complex; d, depleted; r, replete(d); CR, complement receptor; FB, factor B; Fn, Francisella tularensis subsp. novicida; Ft, Francisella tularensis; HI, heat inactivated; LVS, Francisella tularensis subsp. holarctica live vaccine strain; LVSG, LVS gray phase variant; LVSR, LVS rough mutant; MH, Mueller-Hinton.

1
Muller-Eberhard, H. J..
1988
. Molecular organization and function of the complement system.
Annu. Rev. Biochem.
57
:
321
-347.
2
Walport, M. J..
2001
. Complement: first of two parts.
N. Engl. J. Med.
344
:
1058
-1066.
3
Gros, P., F. J. Milder, B. J. Janssen.
2008
. Complement driven by conformational changes.
Nat. Rev. Immunol.
8
:
48
-58.
4
Pangburn, M. K., H. J. Muller-Eberhard.
1980
. Relation of putative thioester bond in C3 to activation of the alternative pathway and the binding of C3b to biological targets of complement.
J. Exp. Med.
152
:
1102
-1114.
5
van Lookeren Campagne, M., C. Wiesmann, E. J. Brown.
2007
. Macrophage complement receptors and pathogen clearance.
Cell. Microbiol.
9
:
2095
-2102.
6
Aderem, A., D. M. Underhill.
1999
. Mechanisms of phagocytosis in macrophages.
Annu. Rev. Immunol.
17
:
593
-623.
7
Ellis, J., P. C. Oyston, M. Green, R. W. Titball.
2002
. Tularemia.
Clin. Microbiol. Rev.
15
:
631
-646.
8
Oyston, P. C., A. Sjostedt, R. W. Titball.
2004
. Tularaemia: bioterrorism defence renews interest in Francisella tularensis.
Nat. Rev. Microbiol.
2
:
967
-978.
9
Hollis, D. G., R. E. Weaver, A. G. Steigerwalt, J. D. Wenger, C. W. Moss, D. J. Brenner.
1989
. Francisella philomiragia comb. nov. (formerly Yersinia philomiragia) and Francisella tularensis biogroup novicida (formerly Francisella novicida) associated with human disease.
J. Clin. Microbiol.
27
:
1601
-1608.
10
Eigelsbach, H. T., W. Braun, R. D. Herring.
1951
. Studies on the variation of Bacterium tularense.
J. Bacteriol.
61
:
557
-569.
11
Cowley, S. C., S. V. Myltseva, F. E. Nano.
1996
. Phase variation in Francisella tularensis affecting intracellular growth, lipopolysaccharide antigenicity and nitric oxide production.
Mol. Microbiol.
20
:
867
-874.
12
Hartley, G., R. Taylor, J. Prior, S. Newstead, P. G. Hitchen, H. R. Morris, A. Dell, R. W. Titball.
2006
. Grey variants of the live vaccine strain of Francisella tularensis lack lipopolysaccharide O-antigen, show reduced ability to survive in macrophages and do not induce protective immunity in mice.
Vaccine
24
:
989
-996.
13
Dennis, D. T., T. V. Inglesby, D. A. Henderson, J. G. Bartlett, M. S. Ascher, E. Eitzen, A. D. Fine, A. M. Friedlander, J. Hauer, M. Layton, et al
2001
. Tularemia as a biological weapon: medical and public health management.
J. Am. Med. Assoc.
285
:
2763
-2773.
14
Heymann, D. L..
2000
. Tularemia. J. Chin, ed.
Control of Communicable Diseases Manual
532
-535. American Public Health Association, Washington, DC.
15
Cross, J. T., R. L. Penn.
2000
. Francisella tularensis (tularemia). G. L. Mandell, and J. Bennett, and R. Dolin, eds.
Principles and Practice of Infectious Diseases
2393
-2402. Churchill Livingstone, Philadelphia.
16
McCrumb, F. R..
1961
. Aerosol infection of man with Pasteurella tularensis.
Bacteriol. Rev.
25
:
262
-267.
17
Fortier, A. H., S. J. Green, T. Polsinelli, T. R. Jones, R. M. Crawford, D. A. Leiby, K. L. Elkins, M. S. Meltzer, C. A. Nacy.
1994
. Life and death of an intracellular pathogen: Francisella tularensis and the macrophage.
Immunol. Ser.
60
:
349
-361.
18
Balagopal, A., A. S. MacFarlane, N. Mohapatra, S. Soni, J. S. Gunn, L. S. Schlesinger.
2006
. Characterization of the receptor-ligand pathways important for entry and survival of Francisella tularensis in human macrophages.
Infect. Immun.
74
:
5114
-5125.
19
Clemens, D. L., B. Y. Lee, M. A. Horwitz.
2005
. Francisella tularensis enters macrophages via a novel process involving pseudopod loops.
Infect. Immun.
73
:
5892
-5902.
20
Ben Nasr, A., J. Haithcoat, J. E. Masterson, J. S. Gunn, T. Eaves-Pyles, G. R. Klimpel.
2006
. Critical role for serum opsonins and complement receptors CR3 (CD11b/CD18) and CR4 (CD11c/CD18) in phagocytosis of Francisella tularensis by human dendritic cells (DC): uptake of Francisella leads to activation of immature DC and intracellular survival of the bacteria.
J. Leukocyte Biol.
80
:
774
-786.
21
Sandstrom, G., S. Lofgren, A. Tarnvik.
1988
. A capsule-deficient mutant of Francisella tularensis LVS exhibits enhanced sensitivity to killing by serum but diminished sensitivity to killing by polymorphonuclear leukocytes.
Infect. Immun.
56
:
1194
-1202.
22
Maier, T. M., M. S. Casey, R. H. Becker, C. W. Dorsey, E. M. Glass, N. Maltsev, T. C. Zahrt, D. W. Frank.
2007
. Identification of Francisella tularensis Himar1-based transposon mutants defective for replication in macrophages.
Infect. Immun.
75
:
5376
-5389.
23
Horwitz, M. A., S. C. Silverstein.
1980
. Influence of the Escherichia coli capsule on complement fixation and on phagocytosis and killing by human phagocytes.
J. Clin. Invest.
65
:
82
-94.
24
Berger, M..
1990
. Third component of human complement: C3.
Methods Enzymol.
184
:
619
-628.
25
Tsai, C. M., C. E. Frasch.
1982
. A sensitive silver stain for detecting lipopolysaccharides in polyacrylamide gels.
Anal. Biochem.
119
:
115
-119.
26
Joiner, K. A..
1988
. Complement evasion by bacteria and parasites.
Annu. Rev. Microbiol.
42
:
201
-230.
27
James, K..
1982
. Complement: activation, consequences, and control.
Am. J. Med. Technol.
48
:
735
-742.
28
Forestal, C. A., M. Malik, S. V. Catlett, A. G. Savitt, J. L. Benach, T. J. Sellati, M. B. Furie.
2007
. Francisella tularensis has a significant extracellular phase in infected mice.
J. Infect. Dis.
196
:
134
-137.
29
White, J. D., J. R. Rooney, P. A. Prickett, E. B. Darrenbacher, C. W. Beard, W. R. Griffith.
1964
. Pathogenesis of experimental respiratory tularemia in monkeys.
J. Infect. Dis.
114
:
277
-283.
30
Bolger, M. S., D. S. Ross, H. Jiang, M. M. Frank, A. J. Ghio, D. A. Schwartz, J. R. Wright.
2007
. Complement levels and activity in the normal and LPS-injured lung.
Am. J. Physiol.
292
:
L748
-L759.
31
Cole, F. S., W. J. Matthews, Jr, T. H. Rossing, D. J. Gash, N. A. Lichtenberg, J. E. Pennington.
1983
. Complement biosynthesis by human bronchoalveolar macrophages.
Clin. Immunol. Immunopathol.
27
:
153
-159.
32
Strunk, R. C., D. M. Eidlen, R. J. Mason.
1988
. Pulmonary alveolar type II epithelial cells synthesize and secrete proteins of the classical and alternative complement pathways.
J. Clin. Invest.
81
:
1419
-1426.
33
Ferguson, J. S., J. J. Weis, J. L. Martin, L. S. Schlesinger.
2004
. Complement protein C3 binding to Mycobacterium tuberculosis is initiated by the classical pathway in human bronchoalveolar lavage fluid.
Infect. Immun.
72
:
2564
-2573.
34
Watford, W. T., A. J. Ghio, J. R. Wright.
2000
. Complement-mediated host defense in the lung.
Am. J. Physiol.
279
:
L790
-L798.
35
Figueroa, J. E., P. Densen.
1991
. Infectious diseases associated with complement deficiencies.
Clin. Microbiol. Rev.
4
:
359
-395.
36
Malik, M., C. S. Bakshi, K. McCabe, S. V. Catlett, A. Shah, R. Singh, P. L. Jackson, A. Gaggar, D. W. Metzger, J. A. Melendez, et al
2007
. Matrix metalloproteinase 9 activity enhances host susceptibility to pulmonary infection with type A and B strains of Francisella tularensis.
J. Immunol.
178
:
1013
-1020.
37
Underhill, D. M., A. Ozinsky.
2002
. Phagocytosis of microbes: complexity in action.
Annu. Rev. Immunol.
20
:
825
-852.
38
McLendon, M. K., M. A. Apicella, L. A. Allen.
2006
. Francisella tularensis: taxonomy, genetics, and immunopathogenesis of a potential agent of biowarfare.
Annu. Rev. Microbiol.
60
:
167
-185.
39
Reckseidler-Zenteno, S. L., R. DeVinney, D. E. Woods.
2005
. The capsular polysaccharide of Burkholderia pseudomallei contributes to survival in serum by reducing complement factor C3b deposition.
Infect. Immun.
73
:
1106
-1115.
40
Stevens, P., S. N. Huang, W. D. Welch, L. S. Young.
1978
. Restricted complement activation by Escherichia coli with the K-1 capsular serotype: a possible role in pathogenicity.
J. Immunol.
121
:
2174
-2180.
41
Cunnion, K. M., J. C. Lee, M. M. Frank.
2001
. Capsule production and growth phase influence binding of complement to Staphylococcus aureus.
Infect. Immun.
69
:
6796
-6803.
42
Hood, A. M..
1977
. Virulence factors of Francisella tularensis.
J. Hyg.
79
:
47
-60.
43
Su, J., J. Yang, D. Zhao, T. H. Kawula, J. A. Banas, J. R. Zhang.
2007
. Genome-wide identification of Francisella tularensis virulence determinants.
Infect. Immun.
75
:
3089
-3101.
44
Cherwonogrodzky, J. W., M. H. Knodel, M. R. Spence.
1994
. Increased encapsulation and virulence of Francisella tularensis live vaccine strain (LVS) by subculturing on synthetic medium.
Vaccine
12
:
773
-775.
45
Geisbert, T. W., P. B. Jahrling, J. W. Ezzell, Jr.
1993
. Use of immunoelectron microscopy to demonstrate Francisella tularensis.
J. Clin. Microbiol.
31
:
1936
-1939.
46
Raynaud, C., K. L. Meibom, M. A. Lety, I. Dubail, T. Candela, E. Frapy, A. Charbit.
2007
. Role of the wbt locus of Francisella tularensis in lipopolysaccharide O-antigen biogenesis and pathogenicity.
Infect. Immun.
75
:
536
-541.
47
Sorokin, V. M., N. V. Pavlovich, L. A. Prozorova.
1996
. Francisella tularensis resistance to bactericidal action of normal human serum.
FEMS Immunol. Med. Microbiol.
13
:
249
-252.
48
Gunn, J. S., R. K. Ernst.
2007
. The structure and function of Francisella lipopolysaccharide.
Ann. NY Acad. Sci.
1105
:
202
-218.
49
Prior, J. L., R. G. Prior, P. G. Hitchen, H. Diaper, K. F. Griffin, H. R. Morris, A. Dell, R. W. Titball.
2003
. Characterization of the O antigen gene cluster and structural analysis of the O antigen of Francisella tularensis subsp. tularensis.
J. Med. Microbiol.
52
:
845
-851.
50
Vinogradov, E., W. J. Conlan, J. S. Gunn, M. B. Perry.
2004
. Characterization of the lipopolysaccharide O-antigen of Francisella novicida (U112).
Carbohydr. Res.
339
:
649
-654.
51
Cowley, S. C., C. J. Gray, F. E. Nano.
2000
. Isolation and characterization of Francisella novicida mutants defective in lipopolysaccharide biosynthesis.
FEMS Microb. Lett.
182
:
63
-67.
52
Sebastian, S., S. T. Dillon, J. G. Lynch, L. T. Blalock, E. Balon, K. T. Lee, L. E. Comstock, J. W. Conlan, E. J. Rubin, A. O. Tzianabos, D. L. Kasper.
2007
. A defined O-antigen polysaccharide mutant of Francisella tularensis live vaccine strain has attenuated virulence while retaining its protective capacity.
Infect. Immun.
75
:
2591
-2602.
53
Li, J., C. Ryder, M. Mandal, F. Ahmed, P. Azadi, D. S. Snyder, R. D. Pechous, T. Zahrt, T. J. Inzana.
2007
. Attenuation and protective efficacy of an O-antigen-deficient mutant of Francisella tularensis LVS.
Microbiology
153
:
3141
-3153.
54
Ben Nasr, A., G. R. Klimpel.
2008
. Subversion of complement activation at the bacterial surface promotes serum resistance and opsonophagocytosis of Francisella tularensis.
J. Leukocyte Biol.
84
:
77
-85.
55
Kraiczy, P., R. Wurzner.
2006
. Complement escape of human pathogenic bacteria by acquisition of complement regulators.
Mol. Immunol.
43
:
31
-44.
56
Lambris, J. D., D. Ricklin, B. V. Geisbrecht.
2008
. Complement evasion by human pathogens.
Nat. Rev. Microbiol.
6
:
132
-142.