|Home | About | Journals | Submit | Contact Us | Français|
Francisella tularensis is capable of rampant intracellular growth and causes a potentially fatal disease in humans. Whereas many mutational studies have been performed with avirulent strains of Francisella, relatively little has been done with strains that cause human disease. We generated a near-saturating transposon library in the virulent strain Schu S4, which was subjected to high-throughput screening by transposon site hybridization through primary human macrophages, negatively selecting 202 genes. Of special note were genes in a locus of the Francisella chromosome, FTT1236, FTT1237, and FTT1238. Mutants with mutations in these genes demonstrated significant sensitivity to complement-mediated lysis compared with wild-type Schu S4 and exhibited marked defects in O-antigen and capsular polysaccharide biosynthesis. In the absence of complement, these mutants were phagocytosed more efficiently by macrophages than wild-type Schu S4 and were capable of phagosomal escape but exhibited reduced intracellular growth. Microscopic and quantitative analyses of macrophages infected with mutant bacteria revealed that these macrophages exhibited signs of cell death much earlier than those infected with Schu S4. These data suggest that FTT1236, FTT1237, and FTT1238 are important for polysaccharide biosynthesis and that the Francisella O antigen, capsule, or both are important for avoiding the early induction of macrophage death and the destruction of the replicative niche.
Much of the recent interest in Francisella tularensis, the etiological agent of tularemia, is due to concern about its potential use as an agent of bioterrorism coupled with an incomplete understanding of the molecular basis of its pathogenicity. F. tularensis is highly pathogenic by the pneumonic route, causing disease in humans with an inoculum as small as 10 organisms, and infection by this route carries a mortality rate of 30 to 60% if untreated (43, 67). Due to its extreme virulence and ease of aerosol dissemination, several nations have weaponized F. tularensis and the U.S. Centers for Disease Control and Prevention have classified this organism as a category A select agent (20). F. tularensis has a remarkably broad host range: it is capable of infecting over 250 known species from across the entire phylogenetic tree, including amoebae, insects, small mammals (such as rodents and lagomorphs), and primates (51). Tularemia is primarily a zoonosis, and humans are thought to be accidental hosts (23). The majority of human infections, the pneumonic infections reported on Martha's Vineyard in 2000 (21) being an notable exception, are cutaneous, lead to ulceroglandular disease, and ensue following exposure to infected animals or animal products (47). The ability of F. tularensis to infect such a wide range of eukaryotes suggests that this organism either co-opts cellular mechanisms common to all hosts, has the requisite virulence genes to adapt to many different intraorganismal environments, or both.
Despite the infectivity of Francisella for disparate hosts, relatively little is known about its virulence genetics. F. tularensis invades and replicates within many cell types: phagocytes, such as primary macrophages (human monocyte-derived macrophages [MDMs] and mouse bone marrow-derived macrophages) and macrophage-like cell lines (J774A.1 and THP-1), as well as in nonphagocytic cells such as bronchial airway epithelial cells, hepatocytes, human umbilical vein endothelial cells, and epithelium-derived tissue culture cell lines (i.e., HEp-2, A549, HBE, and HepG2) (17, 24, 31, 40, 61). Macrophages are known to bind and phagocytose F. tularensis using at least three receptors: complement receptor 3 (CR3), which binds to complement 3b (C3b) protein deposited upon the bacterium when exposed to fresh serum (14) (60); mannose receptor (MR) (60); and scavenger receptor A (53). Once internalized, F. tularensis organisms are able to alter intracellular trafficking of their phagosomes and prevent fusion with the lysosome, acquiring late endosomal markers such as lamp-1 transiently and altogether avoiding endosomes containing cathepsin D and S (15). Shortly thereafter, F. tularensis escapes from the phagosome and grows in the macrophage cytosol. The majority of genes known to be important for phagosomal escape and intracellular growth lie within the duplicated Francisella pathogenicity island (FPI), an ~30-kb region of the chromosome carrying the igl and pdp operons (46). Mutants with mutations in many of the genes in the FPI, such as iglABCD, vgrG, and iglI, as well as pdpA and pdpD, are defective for intracellular growth in both primary and tissue culture cells in vitro, and these genes may encode a novel type VI secretion system (4). Mutational analysis implicates genes such as mglA, sspA, fevR, fslA, pmrA, and migR in regulation of FPI genes (5-8, 10, 38). However, knowledge of other virulence genes residing outside the FPI that play important roles in intracellular growth is limited.
We undertook a high-throughput approach to identify genes important for the growth of virulent F. tularensis Schu S4 in primary human macrophages. F. tularensis has two major biovars: type B (F. tularensis subsp. holarctica), which is found throughout the Northern hemisphere, and type A (F. tularensis subsp. tularensis), which is found exclusively within North America and typically causes a more severe disease in humans than type B isolates. These biovars are further subdivided into clades that exhibit differences in virulence phenotypes (45). A significant research effort to date has focused on the virulence properties of the non-human pathogens Francisella novicida and the F. tularensis live vaccine strain (LVS), an attenuated F. tularensis subsp. holarctica variant generated in the Soviet Union in the 1950s. Although both of these strains are avirulent in healthy humans, they can grow in some human cells and cell lines in vitro and are virulent in the mouse model of infection, properties that make them attractive models for studying the pathogenesis of tularemia. Several random transposon mutant libraries generated in F. novicida and F. tularensis LVS (41, 50, 55, 66, 68, 70) have been screened using both in vitro tissue culture models and mice. In addition, Qin and Mann generated an ~700-member Tn5 (EZ-TN) mutant library in the type A strain F. tularensis Schu S4 and screened each clone through the hepatocyte-like cell line HepG2 to detect mutants defective in intracellular growth. Among the genes identified in this screen is FTT1236 (55).
Lipopolysaccharide (LPS) is a major component in the outer membranes of Gram-negative organisms, and O antigens (O Ags) are known virulence factors of many pathogenic bacteria that function in part to protect against damage by serum complement and antimicrobial peptides as well as mask bacterial surface antigens (16). LPS is composed of a lipid A moiety that secures it to the Gram-negative outer membrane on which a core polysaccharide and O Ag are assembled. Whereas the LPS of most bacterial species is recognized as a pathogen-associated molecular pattern (PAMP) through MD-2/TLR4 and is a potent activating signal for the innate immune system, the LPS of F. tularensis is nearly inert (27). Of note, the lipid A of F. tularensis has an atypical structure, does not bind to LPS-binding protein (LBP), and is therefore not recognized by TLR2 or TLR4 (3, 28, 52). In addition, the F. tularensis O Ag is structurally distinct from that of F. novicida (43). Although our understanding of the genetics of O-Ag biosynthesis in F. tularensis is incomplete, it is known that wbt operon mutants of F. tularensis LVS do not express an O Ag. These strains bind C3b more avidly and are more sensitive to bile salts and to complement-mediated lysis than wild-type strains (13, 39). These mutants are also defective for intracellular growth in J774A.1 cells and exhibit reduced virulence and dissemination in mice (41, 56, 62, 69). Of note, an LVS FTL0706 (designated FTT1238 in F. tularensis Schu S4) mutant lacks an O Ag and exhibits decreased replication within, and is cytotoxic to, J774A.1 cells (41). A corresponding Schu S4 FTT1238 mutant also lacks an O Ag, but its virulence properties have not yet been described.
In many pathogenic bacteria, capsular polysaccharides also contribute to serum resistance yet, unlike O Ags, diminish phagocytosis (16). Francisella has long been thought to have an extracellular structure that resembles that of a capsular polysaccharide. Observed by Hood in 1976 (30), the nature of this capsule has remained elusive. Acridine orange treatment was used to produce an undefined LVS mutant that appears to lack an extracellular polysaccharide structure (58). This strain was designated Cap− (alternately termed “rough” in more recent work) and is serum sensitive (13, 62). Furthermore, expression of the capsular polysaccharide was observed to be increased by repeated passage of LVS on Chamberlain's defined medium, which in turn increased virulence in mice (12). Recently, our group has identified a capsular polysaccharide using a new monoclonal antibody that recognizes this structure as distinct from LPS O Ag. This capsule has immunological properties distinct from those of purified LPS and is also a potential vaccine (2).
Advances in transposon delivery systems in our lab and others allowed us to generate a near-saturating random transposon mutant library in strain Schu S4 that is capable of high-throughput screening using the transposon site hybridization (TraSH) system previously used by Weiss et al. for F. novicida (70). Here we describe the utilization of this genetic system to identify and characterize new virulence genes important for F. tularensis Schu S4 entry and growth in human MDMs. Among the genes we identified is a locus required for LPS O-Ag and capsular polysaccharide biogenesis and for serum resistance and intracellular growth within MDMs. We further show that defects in intracellular growth are due to their induction of premature macrophage death, which deprives mutant organisms of their replicative niche.
All Francisella strains studied in this work are isogenic with F. tularensis subsp. tularensis Schu S4. Bacteria were routinely cultured on modified Mueller-Hinton plates (Acumedia, Lansing, MI) or in modified Mueller-Hinton broth supplemented with 150 mM NaCl (MMH-N). Where necessary to maintain plasmids, 25 μg/ml spectinomycin (Spec) or 25 μg/ml kanamycin (Kan) was added to the medium. F. tularensis wild-type Schu S4 and mutant strains expressing green fluorescent protein (GFP) were generated by cryotransforming them with pBB103-sGFP, a kind gift of Justin Schwartz (University of Iowa). Escherichia coli DH12S and DH5α λ pir were utilized in plasmid construction and were routinely cultured on Lennox agar and broth supplemented with 50 μg Spec or 50 μg Kan as appropriate. All work with Schu S4 was performed within the Carver College of Medicine Biosafety Level 3 (BSL-3) Core Facility, and all experimental protocols were reviewed for safety by the BSL-3 Oversight Committee of the University of Iowa Carver College of Medicine. Recombinant DNA work with Schu S4 was approved by the Institutional Biosafety Committee.
Heparinized venous blood was drawn from healthy adult volunteers using a protocol approved by the Institutional Review Board for Human Subjects at the University of Iowa, and all subjects provided informed consent. Peripheral blood mononuclear cells (PBMCs) were isolated as described previously (1, 60, 61). Briefly, PBMCs were isolated from venous blood by centrifugation on Ficoll-Hypaque (GE Healthcare, Piscataway, NJ), washed twice in HEPES-buffered RPMI 1640 with l-glutamine (RPMI) (Lonza, Walkersville, MD), seeded into Teflon jars at 2 × 106 cells/ml, and allowed to differentiate into monocyte-derived macrophages (MDMs) for 5 to 7 days in RPMI 1640 plus 20% autologous serum at 37°C with 5% CO2.
Plasmid pBDJ314 was generated through the addition of T7 promoters, pointing away from transposon sequences, to pBDJ303, the construction of which we previously described (8). Custom oligonucleotides (Integrated DNA Technologies, Coralville, IA) were designed (see the supplemental material for sequences) and the single-stranded oligonucleotides annealed to form a double-stranded fragment with overhanging ends compatible with restriction sites of pBB303 at either end of the mini-Tn5 transposable element (KpnI/EcoRI and SphI). Insertion of the promoters was confirmed by restriction digestion and sequencing. T7 promoter activity from either end of the mini-Tn5 element was confirmed by digesting pBDJ314 with EagI and performing in vitro transcription using T7 RNA polymerase, which yielded the expected 1.0-kb and 1.6-kb RNA products from the respective T7 promoters (data not shown). Plasmid pBDJ314 was next fused with plasmid pMKM219 as described previously (8) to generate a temperature-sensitive F. tularensis/E. coli shuttle vector and named pSL103. Plasmid pSL103 was cryotransformed into F. tularensis Schu S4 and selected on MMH-0.5% sheep's blood (MMH-B 0.5)-25 μg/ml Spec plates at 32°C, and one colony was inoculated overnight into MMH-Na broth with shaking at 32°C and grown overnight to an optical density at 600 nm (OD600) of ~0.15. This culture was diluted and plated on MMH-B 0.5-25 μg/ml Kan plates and incubated at 40°C to select for transposon insertions in the Francisella chromosome. Colonies were counted, swabbed into subpools of 100 to 200 independent mutants, swabbed into MMH-N-25 μg/ml Kan broth, and shaken overnight at 40°C to cure any remaining plasmid and to minimize the possibility of double transposon insertion mutants. These subpools were combined into pools of ~1,000 colonies, normalized by OD600, and frozen at −80°C in MMH-N broth plus 0.5 M sucrose and 10% glycerol.
MDMs were plated overnight into six-well dishes (Costar, Corning, NY) at a density of 3 × 105 MDMs/well. All 15 pools comprising our T7-capable mutant library were grown overnight in MMH-N broth to mid-log phase (OD600 of ~0.3 to 0.6), normalized by OD600, and combined. Genomic DNA (gDNA) was purified from this input pool using the DNeasy blood and tissue kit (Qiagen, Alameda, CA) according to the manufacturer's instructions. The library was pelleted by centrifugation, washed in Hanks balanced salt solution (HBSS) with calcium and magnesium (Gibco), opsonized by shaking incubation at 37°C in 50% normal human serum (NHS) for 30 min, washed in HBSS without cations (Gibco) to remove calcium and magnesium ions, resuspended in RPMI plus 2.5% NHS, and plated to determine the experimental multiplicity of infection (MOI). MDMs were asynchronously infected at an MOI of ~100 in 1 ml/well of RPMI plus 2.5% NHS for 1 h, washed with phosphate-buffered saline (PBS) (Gibco) to remove unbound bacteria, and incubated for 23 h in fresh RPMI 1640 plus 2.5% NHS. After 1 h, one well was routinely lysed in 1% saponin (Acros, Morris Plains, NJ) in PBS to determine the number of retained bacteria, which averaged 3.6 × 106 CFU/ml, representing approximately 20-fold coverage of the library/well (data not shown). MDMs were then lysed using 1% saponin in PBS, and a sample was taken from each well and plated to determine the extent of bacterial replication, which averaged 8.5 × 108 CFU/ml, approximately 200-fold growth from the 1-h time point (data not shown). Lysates of each well were then pelleted by centrifugation, washed in MMH-N broth to remove the saponin, and inoculated into MMH-N broth and grown for 12 h to mid-log phase. These cultures were then harvested for gDNA as described above, comprising the output pools.
Each DNA sample was divided in two and digested separately with BfaI and RsaI (NEB, Ipswich, MA). The digested DNA was used as the template for in vitro transcription with the AmpliScribe T7-Flash transcription kit (Epicentre, Madison, WI) following the manufacturer's protocol, except that 2 μg of digested DNA was used, and the reaction was allowed to proceed for 12 to 16 h. Purified RNA was used in a reverse transcription reaction using SuperScript II(−) (Invitrogen, Carlsbad, CA) and random hexamers as primers. cDNA was labeled with amino-allyl dUTP by using the Klenow enzyme (NEB, Ipswich, MA). The single-stranded DNA (ssDNA) containing amino-allyl dUTP from the mouse output or the library input pools was labeled with Cy5 or Cy3, respectively, before hybridization to our Francisella microarray as described previously (70). All raw data sets are freely available for download from the GEO database. Normalized data were downloaded from the Stanford Microarray Database according to the median log2 Cy5/Cy3 (logRAT2N). Filters for feature quality, including a Cy3 net median intensity of ≥150 and regression correlation of >0.6, were applied. To compare data from separate macrophage infection experiments, each experimental sample was zero transformed to the input/input control. Features (spots) missing values for ≥30% of the arrays were removed from the data set. The data sets were analyzed with the Significance Analysis of Microarrays (SAM) program by using the two-class analysis option to identify features that consistently deviated from the input and samples across all arrays with a false-discovery rate of 3.9%.
Site-directed mutation of genes indicated by negative selection to be putatively important for growth within human MDMs was achieved using a modified TargetTron (TA0100; Sigma-Aldrich, St. Louis, MO) mutagenesis system as previously described (57). Briefly, primers were designed through the TargetTron algorithm to modify the insertion site of the intron into two sites per targeted gene. Utilizing the TargetTron kit, a retargeted fragment was generated by PCR, cloned into the XhoI/BsrGI sites of the pKEK1140 temperature-sensitive Francisella-E. coli shuttle vector, and screened for LacZ activity. Clones were confirmed by BglII digest, cryotransformed into Schu S4, and plated at 30°C on MMH plus Kan. Transformants were restreaked on MMH plus Kan at 30°C, and many colonies from the second or third streak were streaked twice on nonselective media at 37°C, a restrictive temperature. Ten colonies were then patched to MMH with or without Kan to determine whether the plasmid had completely cured; genomic DNA from cured clones was then purified using a DNeasy blood and tissue kit (Qiagen) and subjected to PCR for confirmation of insertions. Generally, one in four colonies screened for one or both of the retargeted introns bore the desired insertion (data not shown).
To determine whether mutants with mutations in the genes chosen were negatively selected due to their inability to grow at a rate equivalent to that of wild-type Schu S4 in MMH-N broth, wild-type and mutant strains were grown to saturation with shaking at 37°C and diluted into fresh MMH-N broth to an OD600 of 0.1. Broth cultures were shaken at 200 rpm at 37°C, and the optical density was determined at intervals. The doubling time (T) was calculated according to the formula N = N0ekt, where T = (ln 2)/k, and growth indices were computed by dividing maximal mutant doubling times by maximal wild-type Schu S4 growth rates.
Bacteria were grown on MMH plates (with 25 μg spectinomycin for complemented mutants) for 2 days at 37°C with 5% CO2, resuspended and washed in HBSS with calcium and magnesium, and quantitated by OD600. Bacteria were then diluted to 1 × 107 CFU/ml in 50% normal human serum and incubated with shaking at 37°C for 90 min. Before and after incubation, bacteria were serially diluted in PBS, plated on MMH plates, and grown for 2 days at 37°C with 5% CO2. Colonies were enumerated to determine the concentration of viable cells both before and after treatment.
Isolation of Francisella capsule-like material and LPS was performed as described recently (2) with minor modifications. Bacteria grown as a lawn on MMH plates (with 25 μg Spec for complemented mutants) were collected by scraping and resuspended in a solution of 6 mM Tris (Research Products International Corp. [RPI], Mt. Prospect, IL), 10 mM EDTA (RPI), and 2% (wt/vol) sodium dodecyl sulfate (pH 6.8) (Amresco, Solon, OH) treated with proteinase K (New England Biolabs, Ipswitch, MA) and heat treated at 65°C to sterilize. Lysates were ethanol precipitated three times and treated with micrococcal nuclease (New England Biolabs) to digest chromosomal DNA. Samples were then phenol extracted, ethanol precipitated three more times, and separated into capsule and LPS fractions by centrifugation in the presence of Triton X-114 (Sigma-Aldrich). Fractions were lyophilized, and 5 μg of each was combined with NuPAGE (Invitrogen) sample reducing agent and sample buffer, boiled for 10 min, and electrophoresed on NuPAGE Novex 4 to 12% Bis-Tris gels in NuPAGE MES SDS running buffer. Samples were transferred to nitrocellulose and probed with primary antibodies 11B7 to detect capsule (2) and FB11 to detect LPS (QED Bioscience, San Diego, CA). Bands were visualized using goat anti-mouse IgG(H+L) conjugated to horseradish peroxidase (Jackson ImmunoResearch, West Grove, PA) and SuperSignal West Pico chemiluminescent substrate (Pierce Biotechnology, Rockford, IL).
To quantify the extent of replication, MDMs were plated in 24-well dishes (Costar) at a density of 1 × 105 cells/well and allowed to adhere overnight at 37°C. Monolayers were washed with PBS twice and infected in triplicate with unopsonized wild-type Schu S4, mutant, or complemented strains in RPMI plus 2.5% heat-inactivated (56°C, 30 min) NHS (hiNHS). After 1 h at 37°C, monolayers were washed twice with PBS to remove unassociated bacteria, and fresh RPMI plus 2.5% hiNHS was added. Wells were lysed in 1% saponin at intervals. Lysates were diluted in PBS, and viable bacteria were enumerated by plating on MMH as described above. For microscopic analysis, MDMs were seeded onto eight-well Nunc chamber slides (Thermo Fisher Scientific, Rochester, NY) at a density of 40,000 cells per chamber, allowed to adhere overnight, and washed and infected as described above.
For quantitation of MDM lysis by lactate dehydrogenase (LDH) release, macrophages were plated and infected as described above, except that 50 μl of culture supernatant was collected at intervals from triplicate wells and LDH in the supernatant was quantitated using the CytoTox-Fluor cytotoxicity assay (Promega, Madison, WI) according to the manufacturer's recommendations and read on an FLUOStar Optima microplate reader (BMG LabTech, Offenburg, Germany) using plate mode settings and averaging two readings per time point. Percent cytotoxicity was calculated by comparing an average of triplicate wells at each time point to an average of at least six uninfected positive-control wells lysed in 0.9% Triton X-100.
Macrophages infected with F. tularensis were processed for microscopic analysis as previously described (1) with minor modifications. Cells were fixed in 10% formalin, permeabilized by the addition of −20°C acetone and methanol (1:1), and blocked at 4°C for 5 days in PBS plus 0.5 mg/ml sodium azide and 10% bovine serum albumin (BSA), while lysed control slides were examined for sterility to be removed from the BSL-3 facility. Where possible (wild-type Schu S4 and mutants), GFP-expressing bacteria were utilized. The GFP-encoding plasmid could not be introduced into complemented mutants due to plasmid incompatibility between the complementation vectors and pBB103-sGFP. Consequently, complemented mutants were detected using anti-Francisella antiserum (Becton Dickinson and Co., Franklin Lakes, NJ) together with an Alexa Fluor 488-conjugated goat anti-rabbit IgG F(ab′)2 secondary antibody (Invitrogen). In all cases, the late endosome/lysosome-associated membrane protein-1 (lamp-1) was detected using a mouse anti-human lamp-1 monoclonal antibody (H4A3) from the Developmental Studies Hybridoma Bank at the University of Iowa (Iowa City) and an Alexa Fluor 568-conjugated goat anti-mouse IgG F(ab′)2 secondary antibody (Invitrogen). Images were obtained using a Zeiss LSM-510 confocal microscope (Carl Zeiss, Inc., Thornwood, NY).
In order to quantitate changes in nuclear morphology, uninfected macrophages and macrophages infected with wild-type Schu S4 or mutant bacteria were fixed at intervals after infection in 10% formalin. Cells were subsequently permeabilized with cold 70% ethanol and stored at −20°C. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) was then performed using the APO-BRDU kit (BD Pharmingen, San Diego, CA) according to the instructions of the manufacturer modified for use with cells adherent to chamber slides. Total nucleic acids were visualized by staining with 300 nM 4′,6′-diamidino-2-phenylindole (DAPI). Nuclei of infected cells were scored microscopically for pyknosis, as evidenced by nuclear rounding and condensation apparent by DAPI fluorescence, and for karyorrhexis, as evidenced by TUNEL positivity. In wells exposed to either Schu S4 or mutant bacteria, only the nuclei of cells containing bacteria were scored.
We used a comprehensive F. tularensis transposon insertion library, comprised of ~15,000 F. tularensis Schu S4 mutants, to screen for genes important for Francisella replication in primary human macrophages after infection in vitro. This library represents approximately 8-fold coverage of the 1,804 annotated open reading frames of the Schu S4 genome. Adequate representation of the genome in the library was confirmed by visualization of the entire library grown in vitro (data not shown). This library was used in toto for infections of human monocyte-derived macrophages (MDMs) using serum-opsonized bacteria at an MOI of ~100. At 24 h postinfection (hpi), the bacteria recovered from MDMs were pooled and compared with bacteria in the input sample taken at the time of infection using F. tularensis microarrays. Rigorous statistical analysis of the data was performed using Significance Analysis of Microarrays (SAM). We identified 202 genes as present in the input pool but absent from the MDMs after infection (see Table S1 in the supplemental material). These genes are referred to as negatively selected genes.
Many of the genes that we identified encode proteins involved in metabolism (such as pyrH, sucCD, and nadA) and protein synthesis, as is commonly observed in high-throughput screening of random Francisella mutants (35, 41, 50, 66, 68, 70). Notably, several genes residing within the wbt operon, which is required for O-Ag biosynthesis, were implicated in virulence. In contrast, mutants with mutations in genes of the Francisella pathogenicity island (FPI) did not appear to be negatively selected in our screen, most likely because they are duplicated in the genome. A subset of the genes implicated as important for intracellular survival and growth are represented in Fig. Fig.1.1. Perhaps of the most interest are the genes that were not identified in previous studies, several of which have no known function or do not have known orthologs in other microorganisms, and these may include virulence genes that participate in previously uncharacterized aspects of Schu S4 pathogenesis.
Of particular interest to us were several contiguous genes that were negatively selected in MDMs (FTT1235, FTT1236, and FTT1237 as well as neighboring gene FTT1238) that we hypothesized may constitute a locus required for intramacrophage growth (Fig. (Fig.2A).2A). We hypothesized that these genes were important for virulence, as they had been detected by our screen and others (41, 55, 70). We also predicted that these genes may be important for synthesis of extracellular polysaccharides. FTT1235 bears homology to LPS core mannosyl transferase lpcC from Brucella suis and appears, by protein family analysis (using the hmmsearch algorithm from HMMER 3.0 [www.hmmer.janelia.org] against all Pfam-A alignments as of 17 May 2010) to contain a conserved group 1 glycosyltransferase domain (amino acids 151 to 327) with an E value of 6e−41, whereas FTT1237 encodes a conserved group 8 glycosyltransferase domain (amino acids 5 to 264) at an E value of 1.9e−41. FTT1236 and FTT1238 exhibit no significant conserved domains (E value of <0.001). Although its function cannot be defined based upon the current analysis, FTT1238 is predicted to encode a membrane protein containing an estimated 10 transmembrane alpha-helices utilizing the TMpred algorithm (http://www.ch.embnet.org/software/TMPRED_form.html), and FTT1238 is known to be required for O-Ag synthesis in Schu S4 (41).
Based upon our TraSH data, the results of other studies, and our interest in polysaccharide synthesis, we selected FTT1235, FTT1236, and FTT1237 as targets to generate site-directed mutants using group II intron-mediated mutagenesis. We also selected FTT1238, based upon its proximity to the locus of interest, previous characterization of its LPS O-Ag phenotype, and negative selection in previous iterations of the TraSH screening through MDMs using slightly different experimental conditions (data not shown). We successfully generated intron insertions between nucleotides (nt) 202 and 203 of FTT1236, nt 409 and 410 of FTT1237, and nt 93 and 94 and nt 480 and 481 of FTT1238 (in all assays performed, the FTT1238 intron insertion in nt 480 and 481 was indistinguishable from the one between nt 93 and 94, and thus for all further experimentation, data are reported for the upstream mutant only). However, we were not able to generate a stable FTT1235 mutant, as sequential passage of bacteria transformed with the intron delivery plasmid at a restrictive temperature generated increasingly smaller colonies until viable cells could not subsequently be obtained, suggesting that a mutation in FTT1235 may be lethal under the conditions utilized (data not shown). To ensure that the genes we selected were not identified because of an inability to grow as rapidly as wild-type Schu S4 under the culture conditions utilized in our initial screening, we obtained growth curves for the FTT1236, FTT1237, and FTT1238 mutants as well as wild-type Schu S4 in broth culture and determined their maximal rates of replication. None of the mutants exhibited an in vitro growth rate significantly different from that of the wild type (data not shown).
Because our screen utilized bacteria that were opsonized with 50% normal human serum, we hypothesized that the mutants may have been negatively selected based upon sensitivity to serum complement. Consequently, we exposed wild-type Schu S4 and FTT1236, FTT1237, and FTT1238 mutants to 50% human serum for 90 min at 37°C and performed a viable cell count. All of the mutants were serum sensitive, consistent with disruption of outer membrane polysaccharide biosynthesis (Fig. (Fig.2B).2B). An FTT1236 mutant appears to be more sensitive to serum than either FTT1237 or FTT1238 mutants, though viability of all three strains was significantly diminished compared with that of wild-type Schu S4.
Due to the orientation of FTT1236 and FTT1237 and because merely 8 bp separate their coding sequences, we hypothesized that the genes composed an operon. In order to determine with certainty that the insertions into FTT1236, FTT1237, and FTT1238 were responsible for the mutants' sensitivity to serum and to functionally test this hypothesis, fragments carrying both FTT1236 and FTT1237, FTT1237 alone, and FTT1238 alone were cloned into plasmid pBB103, forming plasmids pSL129, pSL130, and pSL149, respectively. The FTT1237 gene was deleted from plasmid pSL129 by restriction digest with SalI and ligation, creating a plasmid with only a functional FTT1236 gene, pSL129Δ37. Mutants were cryotransformed with the appropriate complementation vectors and assayed for their sensitivity to killing by serum complement. Serum resistance was significantly increased by the addition of pSL129 to the FTT1236 mutant (here called FTT1236c), of pSL130 to the FTT1237 mutant (FTT1237c), and of pSL149 to the FTT1238 mutant (FTT1238c), indicating that mutations in the genes of interest accounted for serum sensitivity. Of note, complementation of the FTT1236 mutant could not be achieved by adding back either FTT1236 or FTT1237 alone, suggesting that transcription of both FTT1236 and FTT1237 is disrupted by a polar mutation in FTT1236. In contrast, the inability of pSL129Δ37 to complement a FTT1237 mutant confirmed the absence of functional FTT1237 in this plasmid (Fig. (Fig.2B2B).
To assess directly the roles of FTT1236, FTT1237, and FTT1238 in biosynthesis of LPS O Ag and capsule, these surface carbohydrates were isolated from wild-type, mutant, and complemented strains and analyzed by immunoblotting using monoclonal antibodies that recognize the Francisella O Ag (FB11) or capsular polysaccharide (11B7) (2). FTT1236, FTT1237, and FTT1238 mutants lacked O Ag, as determined by reactivity to antibody FB11, whereas in FTT1236c, FTT1237c, and FTT1238c mutants O-Ag production was restored as judged by this assay (Fig. (Fig.3A).3A). Addition of either the wild-type FTT1236 or FTT1237 gene alone in trans to the FTT1236 mutant or of the wild-type FTT1236 gene to the FTT1237 mutant did not restore O Ag, correlating with the inability of these plasmids to restore serum resistance. Although FTT1236 or FTT1238 mutants are capsule deficient or lack 11B7-recognized epitopes of capsule, an FTT1237 mutant retains 11B7-reactive material of significantly higher molecular weight than that of wild-type Schu S4 (Fig. (Fig.3B).3B). The FTT1237c mutant exhibited a wild-type molecular weight of the capsular material. The FTT1238c mutant demonstrated restored reactivity with 11B7, whereas the FTT1236c mutant exhibited only trace reactivity in comparison, possibly due to a stoichiometric effect of expression from a multicopy plasmid. Together, the serum sensitivity and the LPS and capsule immunoblotting data suggest that FTT1236 and FTT1237 compose an operon required, along with FTT1238, for the biosynthesis of O Ag and capsule in F. tularensis Schu S4.
Having determined that FTT1236, FTT1237, and FTT1238 mutants were serum sensitive, we used a complement-free infection model to assess bacterial fate in MDMs. To this end, MDMs were infected with Schu S4 or FTT1236, FTT1237, or FTT1238 mutants in medium supplemented with heat-inactivated normal human serum, samples were lysed at 8-h intervals for up to 24 hpi, and viable cell counts were performed. At 1 hpi, the mutants were reproducibly recovered at 3- to 5-fold-higher levels than wild-type bacteria. Thereafter, the FTT1236, FTT1237 and FTT1238 mutant strains grew at rates similar to that of wild-type Schu S4 over the first 16 h of infection, after which growth of mutant bacteria leveled off, whereas Schu S4 continued to grow within MDMs over the entire 48-h course of the experiment. Overall, mutant bacterial growth increased only 10-fold over 16 h, whereas Schu S4 growth increased 50-fold over the same time period and reached 300- to 500-fold increases over the 48 h (Fig. (Fig.44).
One possible explanation for the increased uptake and reduced growth exhibited by the FTT1236, FTT1237, and FTT1238 mutants was that their altered surface properties enhanced their adherence to the tissue culture wells or diminished the macrophages' ability to internalize them, as seen, for example, in bvrR/S mutants of Brucella abortus (64), and thus that a majority of the bacteria were growing extracellularly throughout the course of infection. To test this hypothesis, infected MDMs were treated with gentamicin at 4 hpi to kill extracellular bacteria. Addition of gentamicin did not significantly alter viable cell counts of bacteria in coculture with MDMs compared with those of untreated bacteria either immediately after treatment or over a 24-h time course. In contrast, bacteria exposed to gentamicin in the absence of MDMs were killed (data not shown). These data strongly suggested that the bacteria were internalized by macrophages and that the observed growth was intracellular.
To assess whether mutants exhibited defects in phagosome escape, we performed a microscopic analysis of MDMs infected with Schu S4 or mutant bacteria over a 32-h time course. At 1 hpi, <5% of wild-type or mutant bacteria colocalized with lamp-1, suggesting that both wild-type and mutant bacteria had successfully escaped the phagosome (Fig. (Fig.5).5). Further analysis of infection efficiency indicated that for Schu S4, 22% of MDMs were infected with 1.23 ± 0.57 bacteria per cell, whereas 94% and 95% of MDMs were infected with FTT1236 and FTT1238 mutants and contained an average of 6.02 ± 4.09 and 6.75 ± 4.72 bacteria each, respectively. These data are consistent with the differences in infection efficiency obtained by measurement of viable bacteria.
By 16 hpi, macrophages infected with mutant bacteria appeared to be significantly less healthy than those infected with wild-type bacteria despite the fact that the bacterial burden in Schu S4-infected cells was considerably higher (Fig. (Fig.66 to 8). Specifically, MDMs infected with FTT1236 (Fig. 8) or FTT1238 (Fig. (Fig.7)7) mutants exhibited increased rounding, blebbing, and nuclear condensation (Fig. (Fig.6).6). In contrast, the vast majority of MDMs infected with Schu S4 retained a normal flat, spread morphology despite their profound bacterial load. Complementation of mutants in trans with the corresponding gene restored near-wild-type growth patterns and macrophage morphology (Fig. (Fig.88 and and9).9). At 24 hpi, 92% of Schu S4-infected MDMs remained healthy, as indicated by their flat, spread morphology and large nuclei. In contrast, at 24 h after infection with FTT1236 or FTT1238 mutants, ~95% of MDMs appeared to be unhealthy or dead, as indicated by cell rounding, surface blebbing, and/or nuclear condensation and by decreased density of the monolayer. In a subset of MDMs infected with the mutants, these morphological changes were apparent as early as 16 hpi. The phenotype of the FTT1237 mutant appeared to be somewhat less severe than those of the other two, with only ~65% of infected MDMs appearing rounded at 24 hpi. Macrophages infected with complemented FTT1236c and FTT1237c strains appeared generally similar to Schu S4, with ~60% of infected cells displaying an adherent, spread flat phenotype at 24 hpi (Fig. (Fig.88 and and9).9). Complementation of the mutants in trans restored the ability of F. tularensis to grow intracellularly to high density without inducing cell death. Of note, bystander macrophages were not affected by growth of either Schu S4 or mutant bacteria within neighboring MDMs and remained healthy, flat, and spread on the substratum (Fig. (Fig.66 and and8).8). At 32 hpi, few cells infected with mutant bacteria remained, and cell debris was common. In contrast, ~75% of cells infected with Schu S4 still appeared to be relatively healthy despite extensive intracytosolic bacterial replication, whereas mutant bacteria were predominantly associated with cell debris and freely extracellular (Fig. (Fig.8).8). Extracellular bacteria were seldom observed in chambers infected with Schu S4, even at 32 hpi, as the majority of MDMs appeared to be intact (Fig. (Fig.6).6). Finally, neither wild-type nor mutant bacteria colocalized with lamp-1 at 16, 24, or 32 hpi, suggesting that autophagy was not induced under these conditions.
Taken together, these data indicate that FTT1236, FTT1237, and FTT1238 mutants are significantly attenuated for intramacrophage growth relative to Schu S4 despite enhanced infection in the absence of active complement. Moreover, we show that decreased intracellular growth is due, at least in part, to premature MDM death that deprives mutant organisms of their replicative niche.
In order to quantify the rate and extent of MDM death after infection with wild-type or mutant strains, we assayed MDM supernatants for the cytosolic enzyme lactate dehydrogenase (LDH) as an indicator of plasma membrane disruption. In keeping with the microscopy data, all of the mutant strains tested caused a significant release of LDH from infected macrophages, which was detectable as early as 8 hpi and increased over the course of the experiment (Fig. 10A). MDMs infected with FTT1236 mutant bacteria exhibited the most severe phenotype, showing the largest and most rapid release of LDH, whereas FTT1237 and FTT1238 mutants triggered slower release of LDH than did the FTT1236 mutant. Schu S4 was much less cytotoxic, with LDH release differing significantly from that from mock-infected MDMs only at the 24-h time point. Complementation of the interrupted genes and rescue of the corresponding surface polysaccharides dramatically decreased LDH release, suggesting that FTT1236, FTT1237, and FTT1238 are required to sustain MDM viability and therefore support intramacrophage growth.
We further assayed individual infected cells for nuclear condensation (pyknosis) and DNA fragmentation (karyorrhexis), which are indicators of cell death consistent with the induction of either apoptosis or pyroptosis. We now show that whereas nuclei of ~65% of MDMs infected with mutant bacteria appeared to be pyknotic and ~35% were positive for DNA fragmentation by TUNEL analysis at 16 hpi, MDMs infected with wild-type Schu S4 exhibited rates of pyknosis and karyorrhexis very similar to those of uninfected control macrophages (Fig. 10B). Combined, these data suggest that O Ag and/or capsule is required to prevent premature induction of macrophage death and subsequent destruction of the intracellular replicative niche of Francisella.
In this work, we present the first high-throughput screening of a near-saturating random mutant library in a virulent strain of Francisella tularensis through primary human macrophages by transposon site hybridization. Our screen uncovered a previously uncharacterized locus (FTT1235 to -8) of the Schu S4 chromosome that is required for O-Ag and capsule biosynthesis, among other new genes required for intramacrophage survival and growth. Three of the genes in this locus (FTT1236, FTT1237, and FTT1238) have previously been implicated in pathogenesis by other mutational studies of both virulent (55) and avirulent (41, 70) Francisella strains, but no characterization of the mechanism of intracellular attenuation was reported. Finally, of particular interest to our group, sequence analysis suggested that two of the four genes (FTT1235 and FTT1237) contain putative glycosyltransferase domains and might have a potential role in polysaccharide biosynthesis, as a Schu S4 FTT1238 mutant does not express O Ag (41).
In order to characterize the phenotypes of mutants with mutations in these genes, we constructed site-directed FTT1236, FTT1237, and FTT1238 Schu S4 mutants by utilizing group II intron-mediated mutagenesis adapted for Francisella (57). We determined that these mutants are serum sensitive and lack both O Ag and capsule as judged by immunoblotting. Complementation of these genes together and individually show that FTT1236 and FTT1237 compose an operon required for synthesis of O Ag and the Francisella O-Ag capsular polysaccharide recently described by our group (2). We also show that disruption of these genes enhances bacterial uptake by MDMs in the absence of fresh serum and that the mutants appear to be indistinguishable from Schu S4 in phagosome escape. However, these mutants exhibit a significant intramacrophage growth defect and are rapidly cytotoxic to MDMs compared with wild-type Schu S4. These phenotypes are reversed in the mutant strains by complementation in trans. Taken together, these data suggest that Schu S4 FTT1236, FTT1237, and FTT1238 mutants are attenuated in the MDM model due to their inability to express O Ag, capsule, or both, which in turn results in their inability to prevent rapid macrophage death.
Our application of transposon site hybridization to detect Schu S4 genes required in MDMs identified 202 genes as potentially involved in survival and growth within these macrophages, several of which had not previously been implicated in Francisella virulence. At the same time, this screen also confirms several genes identified in other high-throughput screens (41, 50, 55, 66, 68, 70) as important for virulence in several strains of Francisella and under other experimental conditions. Approximately 25% of the genes implicated in our study have been independently detected in other high-throughput screens. In keeping with this, a meta-analysis of high-throughput random mutant screens utilizing a variety of Francisella strains and model systems showed that 81% of the genes implicated in virulence were detected in only one study (of the nine examined) (44). As is commonly observed in high-throughput screening for Francisella mutants, genes likely involved in protein synthesis and various aspects of metabolism accounted for approximately 45% of the genes identified in our study. We expected these genes to be negatively selected, since they encode proteins that are required for Francisella growth and replication in its intracellular niche. It is also likely that some metabolic genes identified in our screen are necessary for growth within macrophages but not for growth in rich Mueller-Hinton broth. For example, it is likely that pyrH is unable to grow in MDMs due to deficits in pyrimidine biosynthesis, a pathway dispensable for growth in rich media (31, 60). Furthermore, several genes detected in our screen encode proteins essential for general bacterial survival, including genes for structural components of ribosomes such as rplD and rpsC. Although these genes have been identified in similar high-throughput screens (66, 70), it is likely that the transposon insertions employed either significantly alter the expression of these genes by affecting promoter or enhancer regions or result in truncations that encode partially functional products, thereby decreasing the bacterial growth rate in broth culture, macrophages, or both. The negative-selection technique used here also identified genes involved in cell envelope biosynthesis, especially those important for O-Ag synthesis in the wbt operon. This result is not surprising, as mutants with mutations in these genes are known to be serum sensitive and the bacteria used in our screen were opsonized. In contrast, our screen did not detect mutants with mutations in genes within the FPI, as the FPI is duplicated in the Francisella genome. Although Su et al. detected iglA, iglB, and iglC mutants in intranasally infected mice by signature-tagged mutagenesis (66), when examined more closely, an iglC transposon mutant resulted in a mere 6-fold defect in competition with the wild type and was the least significant virulence phenotype observed by this group. Further, a mutational analysis in LVS has shown that one copy of iglC is dispensable for intracellular pathogenesis (26), and these data were confirmed by us (61). These data suggest that F. tularensis mutants with mutations in a single copy of the FPI genes have, at most, a small defect in intracellular growth. The fact that these genes were not implicated in virulence in this screen suggests that the negative-selection protocol robustly and sensitively identifies genes in which mutations are likely to cause significant virulence defects.
To date, most studies of Francisella mutants with diminished or ablated O Ag have utilized F. tularensis LVS and F. novicida. Although images suggestive of capsule were observed as early as 1976 (30) and this structure was more abundant after serial passage on Chamberlain's defined medium (12), the polysaccharide has only recently been isolated and characterized (2). Early studies of an undefined rough mutant generated by acridine orange treatment of LVS showed increased complement deposition (58, 65) as well as serum sensitivity and decreased virulence in mice (58). Analysis of spontaneously arising “gray” variants of LVS, which display an altered (13, 19) or missing (29) O Ag, suggests that they too exhibit increased complement deposition, serum sensitivity, reduced intracellular growth in J774A.1 cells, and diminished virulence in mice (13, 29). To date, studies of O-Ag biosynthesis have been focused predominantly on the wbt locus. A wbtA mutant is among the first F. novicida transposon mutants unable to synthesize O Ag (18), and both a polar and nonpolar insertions within wbtA of LVS ablated O Ag, conferred sensitivity to complement-mediated lysis, impaired growth within J774A.1 cells, and reduced virulence and dissemination in the mouse model (13, 41, 56, 62). A spontaneous LVS wbtI gene mutant, as well as wbtC and wbtM insertion mutants, exhibit a similar lack of O Ag, complement sensitivity, and virulence defects in cells in vitro and in animal models (13, 39, 41). Of note, wbtDEF insertion mutations of Schu S4 and F. novicida ablate O-Ag biosynthesis in both strains but prevent intracellular growth only in Schu S4, which suggests strain-specific differences in the requirement for intact LPS for intracellular growth and may indicate divergent bacterial mechanisms of intracellular parasitism (69). Taken together, these data suggest that the wbt operon is important for biosynthesis of LPS, which is in turn important for intracellular growth and pathogenicity to mice. As expected, several genes within the wbt operon were identified in our negative-selection screen. However, prior to this study, only one gene outside the wbt locus, FTT1238, was known to be important for O-Ag biosynthesis in LVS and Schu S4. An LVS mutant with an insertion mutation in this gene exhibited reduced growth in J774A.1 cells and virulence in mice (41).
Consistent with the published data, we show that loss of O Ag and capsular polysaccharide renders Schu S4 FTT1236, FTT1237, and FTT1238 mutants sensitive to complement-mediated killing. It is noteworthy that although all three mutants are deficient, their surface polysaccharide immunoblot phenotypes are not identical. All three mutants lack an O-Ag ladder, similar to wbt operon mutants (18, 39, 41, 56, 62, 69). This phenotype is distinct from that of a wzy gene mutant, which exhibits only one repeating unit attached to lipid A (34). While both FTT1236 and FTT1238 mutants lack reactivity to anticapsule antibody 11B7, the FTT1238 mutant, as well as an FTT1237 mutant, retains a small amount of high-molecular-weight material reactive with anti-LPS antibody FB11. Because our recent data indicate that the structures of the sugar repeating unit of the capsule and O Ag are identical (2), it will be interesting to determine whether this high-molecular-weight material is capsular in origin or perhaps a biosynthetic intermediate. Of note, the FTT1237 mutant produces a capsular polysaccharide of increased molecular weight, which may indicate increased availability of O-Ag subunits in this mutant in the absence of an LPS O-Ag ladder. As the functions of the proteins encoded by these genes are unclear, further work will be required to elucidate their roles in O-Ag and capsule biosynthesis.
The increased uptake of FTT1236, FTT1237, and FTT1238 mutants is consistent with the antiphagocytic role of bacterial capsules. Although the mannose receptor (MR) plays a significant role in uptake of unopsonized wild-type F. tularensis (60), whether this or another receptor mediates entry of mutants in the absence of fresh serum is unknown. As the route of uptake can substantially impact downstream trafficking events, it is possible that a divergent receptor binding or mechanism of uptake may contribute to induction of the early cell death we observed. Furthermore, it is as yet unknown whether the induction of cell death results from signaling coincident with uptake or whether escape and/or intracytosolic replication is also required.
Most of the virulence gene mutants described to date, including those with mutations within the FPI (the best studied being iglC) or in genes regulating expression of FPI genes (such as mglA and fevR), retain their resistance to complement-mediated lysis and lose virulence due to their inability to escape the phagosome and replicate within the cytosol (6, 7, 38). A second class of mutants is capable of escape but incapable of intracytosolic growth, such as iglD mutants as well as those with mutations in genes involved in pyrimidine biosynthesis (31, 59, 61). We show here that FTT1236, FTT1237, and FTT1238 mutants are defective at a distinct stage in the Francisella life cycle. They do not exhibit an apparent defect in escape, as neither wild-type Schu S4 nor the mutants colocalized to any significant extent with lamp-1 at 1 hpi. Quantitation by viable cell count and microscopy shows that all strains are capable of intracytosolic replication within the first ~16 hpi. Thereafter, the wild type and mutants diverged such that Schu S4 continued to grow, filling the macrophage cytosol by 32 hpi, whereas the mutants did not. Thus, although the vast majority of macrophages infected with Schu S4 remained healthy despite an enormous bacterial load, the appearance of macrophages containing mutant bacteria was dramatically altered as indicated by cell rounding, surface blebbing, and nuclear condensation that progressed rapidly to destruction of the monolayer, which was confirmed by measurements of LDH release.
Whether MDMs infected with wild-type or mutant strains die by similar or distinct mechanisms remains to be determined, as nuclear condensation and surface blebbing can be associated with both apoptosis and pyroptosis. Consistent with this, Parmely and coworkers demonstrated that macrophages in the livers and spleens of mice infected with virulent F. tularensis strain KU49 show widespread caspase-3 activation that progresses to secondary necrosis and is accompanied by bacterial spread, whereas caspase-1 activation and localization of bacterial antigens to microgranulomas characterize mice infected with strain LVS or F. novicida U112 (48, 71). Macrophages infected with LVS in vitro can also undergo apoptosis (36, 37), whereas primed murine macrophages infected with F. novicida undergo rapid pyroptotic death that is dependent upon activated caspase-1 and AIM2-mediated signaling (32, 42).
Studies of mutants generated from LVS or F. novicida supports the hypothesis that preservation of macrophage viability is important for maintenance of this organism's replicative niche. Weiss and coworkers identified two mutants of F. novicida (with mutations in homologs of FTT0584 and FTT0748) that exhibited increased caspase-1- and ASC-mediated cytotoxicity in primed murine macrophages as well as decreased virulence in mice (70). In addition, an LVS tolC mutant is defective for growth within murine macrophages (25) and exhibits increased cytotoxicity in vitro (54) and diminished virulence in the mouse model (25, 54). The intracellular localization of the tolC mutant and the precise mechanism for this increased cell death are not yet understood but are thought to involve caspase-3 and not caspase-1 (54). The intramacrophage growth and cytotoxicity phenotypes of a Schu S4 tolC mutant remain unknown, but recent mouse survival experiments suggest that the contribution of the gene to virulence in this strain is modest (33). Notably, Maier and coworkers (41) selected the FTT1238 mutant for study due to its reduced toxicity to J774A.1 cells. Qin and Mann (55) also found that a Schu S4 FTT1236 mutant is incapable of growth in J774A.1 cells, but they did not determine whether these mutants induced host cell death. In contrast to the results of Maier et al., we show here that Schu S4 FTT1238 mutants exhibit increased cytotoxicity in human primary macrophages. Maier and coworkers did not characterize the intracellular growth of a Schu S4 FTT1238 mutant in J774A.1 cells, nor have we, and thus what accounts for the divergence in cytotoxicity remains to be determined. Possibilities include strain-specific factors and/or compensatory survival mechanisms in transformed J774A.1 cells that are missing within MDMs. Taken together, these data suggest that the mechanism of cell death induction may be dependent upon the Francisella strain and the host cell employed. Consequently, to gain an understanding of what cell death pathways are induced in human tularemia, it will be important to study mutants of human pathogenic strains of Francisella tularensis in primary human cells.
Although our results suggest that capsule and O Ag play a previously uncharacterized role in maintaining macrophage viability during infection with Schu S4, this notion is not without precedent, since rough mutants of other intracellular bacterial pathogens such as Coxiella and Brucella have similar phenotypes. In rough strains of both bacteria, mutants without O Ags are taken up more efficiently by macrophages (9, 49). Coxiella phase II mutants, which lack an O polysaccharide, activate human dendritic cells (DCs) (63). Brucella rough mutants induce cell death in macrophages that bears hallmarks of both apoptosis and oncosis (11, 22, 49). In both species, strains lacking O Ag are avirulent. In the case of Brucella, induction of early macrophage death is cell type specific; the extent to which mutants with mutations in FTT1236 to -8 undermine the viability of other host cells infected by this organism, such as neutrophils and epithelial and endothelial cells, is as yet unknown.
In summary, we demonstrate the high-throughput detection of 202 genes as being important for F. tularensis Schu S4 intracellular survival and growth in human primary macrophages by transposon site hybridization. Among these genes, we demonstrate that a previously uncharacterized locus, FTT1236 to -8, is required for both O-Ag and capsule biosynthesis. Mutants with mutations in this locus are phagocytosed more efficiently than wild-type bacteria, escape the phagosome, and initiate replication in the cytosol. However, infection is curtailed due to premature MDM death, and therefore these strains exhibit defects at a stage of the life cycle distinct from disruption of major regulatory factors or genes in the FPI. As previous studies have shown that Francisella-infected macrophages can die by caspase-1 or caspase-3 dependent mechanisms, it will be important in future studies to characterize the mechanism of death at the molecular level.
We thank the Carver College of Medicine Biosafety Level 3 Core Facility, Central Microscopy Research Facility, Matthew Faron, and Jed Rasmussen for their critical review of the manuscript, Blake Buchan, Justin Schwartz, Jason Barker, and Ramona McCaffrey for helpful insights, and Grace Lindemann for assistance with enumeration of macrophage bacterial uptake by microscopy.
S.R.L. was supported by a U.S. Department of Homeland Security Graduate Fellowship and a University of Iowa Presidential Fellowship. K.P. was supported by a predoctoral fellowship from the Agency for Science, Technology, and Research, Singapore. This study was supported by Project 14 of NIH grant U54AI057160 (to L.-A.H.A. and B.D.J.) via the Midwest Regional Center of Excellence (MRCE) for Biodefense and Emerging Infectious Diseases Research, grants AI063302 and AI065359 from the NIH-NIAID to D.M., and NIH grant P01 (AI044642) to M.A.A. and L.-A.H.A., which also funded some work in the lab of B.D.J.
Editor: J. B. Bliska
Published ahead of print on 15 November 2010.
†Supplemental material for this article may be found at http://iai.asm.org/.