ABSTRACT
IglE is a small, hypothetical protein encoded by the duplicated Francisella pathogenicity island (FPI). Inactivation of both copies of iglE rendered Francisella tularensis subsp. tularensis Schu S4 avirulent and incapable of intracellular replication, owing to an inability to escape the phagosome. This defect was fully reversed following single-copy expression of iglE in trans from attTn7 under the control of the Francisella rpsL promoter, thereby establishing that the loss of iglE, and not polar effects on downstream vgrG gene expression, was responsible for the defect. IglE is exported to the Francisella outer membrane as an ∼13.9-kDa lipoprotein, determined on the basis of a combination of selective Triton X-114 solubilization, radiolabeling with [3H]palmitic acid, and sucrose density gradient membrane partitioning studies. Lastly, a genetic screen using the iglE-null live vaccine strain resulted in the identification of key regions in the carboxyl terminus of IglE that are required for intracellular replication of Francisella tularensis in J774A.1 macrophages. Thus, IglE is essential for Francisella tularensis virulence. Our data support a model that likely includes protein-protein interactions at or near the bacterial cell surface that are unknown at present.
INTRODUCTION
Francisella tularensis is a small, nonmotile, Gram-negative, intracellular bacterium and the causal agent of the zoonotic disease tularemia (1). The majority of human tularemia infections are caused by the two subspecies that differ in their geographic distribution and their potential virulence for humans. F. tularensis subsp. tularensis (type A) is found only in North America and is responsible for a severe, potentially lethal disease following exposure to as few as 10 CFU. F. tularensis subsp. holarctica (type B) is present throughout the Northern Hemisphere but is associated with milder disease that is rarely fatal (2, 3). In nature, tularemia arises following exposure to infected animals, especially rodents and lagomorphs, or through the bites of blood-feeding arthropods (1, 3). However, because of its low infection dose and high associated mortality, especially following aerosol exposure, F. tularensis has been designated by the Centers for Disease Control and Prevention as a tier 1 biothreat agent, with high potential for illegitimate use. Several decades ago, an attenuated live vaccine strain (LVS) was developed from F. tularensis subsp. holarctica but remains unapproved as a vaccine owing to questions regarding the genetic nature of its attenuation and other safety concerns (1). Both LVS and a related organism, F. tularensis subsp. novicida, lack significant virulence for immunocompetent humans but can cause a tularemia-like illness in mice. As such, these two strains are often used as biosafety level 2 experimental surrogates for more virulent strains of F. tularensis. Although such studies have led to many important findings regarding the biology of Francisella, there are several instances where the outcome of infections with these lower-virulence strains was not predictive of the outcome of infection with a virulent type A Francisella strain, such as Schu S4 (4–6).
The pathogenesis of F. tularensis is poorly understood at the molecular level, but the ability to invade and replicate in macrophages appears to be key for productive infection (7). Francisella enters these cells by triggering the formation of asymmetric, pseudopodal loops that promote initial uptake into spacious vacuoles that progressively acquire some early and late endosomal markers (i.e., EEA-1, Rab5, Rab7, CD63, LAMP-1, and LAMP-2) but that exclude others (i.e., cathepsin D) (8–11). Shortly after entering the macrophage, the phagosomal membrane surrounding the bacterium is degraded and the bacteria are released into the host cytosol, where extensive intracytosolic replication occurs (7). This leads to eventual apoptotic (12, 13) and pyroptotic (14) lysis of the host cell and the subsequent release of the intracellular bacteria within. In some instances, especially late in the infection cycle of murine bone marrow-derived macrophages (BMMs) but not human blood monocyte-derived macrophages, some cytosolic F. tularensis reenters the endocytic pathway in a process that involves the formation of autophagosomes (15, 16). Whether this serves to protect the bacterium from actions of the host inflammasome pathway or, rather, is a host-mediated process aimed at destroying the pathogen remains a subject of debate. However, recent studies suggest that the formation of these LAMP-1-positive autophagosomes, also known as Francisella-containing vacuoles (FCVs), are not bactericidal per se but instead serve as a clearance mechanism for replication-deficient or damaged cytosolic Francisella via a ubiquitin-LC3-SQSTM1-LC3 pathway (17). Francisella also actively interferes with host intracellular signaling pathways and innate immune responses (9, 18–20) and promotes active immune suppression during early pulmonary residence (21). The factors employed by virulent forms of this bacterium to orchestrate these numerous changes are, as yet, poorly defined.
Genetic studies have implicated the Francisella pathogenicity island (FPI), an ∼30-kb gene cluster consisting of 16 to 19 open reading frames, as being critical for phagosomal escape and intracellular survival (reviewed in reference 22). The FPI gene cluster, which is present in all sequenced Francisella genomes evaluated to date, encodes several genes (i.e., pdpB, vgrG, dotU, iglI, and iglJ) that are thought to encode a type VI secretion system (T6SS) (23–25) on the basis of limited sequence homology to the same genes in other bacteria (26, 27). Interestingly, the FPI is present at a single copy in low-virulence F. tularensis subsp. novicida but is duplicated in type A and type B strains. The presence of two identical copies in the most virulent forms of F. tularensis has limited all but a few genetic studies (8, 28–32) to lower-virulence F. tularensis subsp. novicida (which is more easily manipulated via standard genetic techniques owing to the presence of a single copy of the FPI). Nevertheless, there is a growing body of evidence that most of the genes in the FPI are required for virulence and/or intramacrophage growth in at least one subspecies (reviewed in reference 22). On the basis of this evidence, it has been proposed that Francisella may encode specialized multicomponent machinery to facilitate secretion of effector molecules either to the bacterial cell surface or directly into host cells. Evidence to support this directly, however, is lacking at present.
iglE encodes a small, hypothetical protein that was previously identified as 1 of 16 to 19 genes in the FPI required for intracellular replication of low-virulence F. tularensis subsp. novicida (33–36). However, these authors did not address potential polar effects on downstream vgrG gene transcription, and attempts to complement the iglE mutation with a multicopy extrachromosomal plasmid were unsuccessful (36); thus, the phenotype of the resultant mutants could not be unequivocally attributed to the loss of iglE alone. IglE was also identified as one of eight secreted proteins on the basis of a comprehensive reporter assay of FPI-encoded proteins coupled to TEM (a β-lactamase) (37). IglE has a putative signal sequence and lipobox signature motif, suggesting that its gene may encode a bacterial lipoprotein, but it lacks significant homology to other known proteins. The crystal structure of a recombinant modified form of IglE has been solved (38), but there are few additional data to substantiate how this protein might contribute to intracellular pathogenesis. Here we report that iglE is required for phagosomal escape and the intracellular survival of virulent type A Schu S4 in bone marrow-derived macrophages. We also show that IglE is essential for murine pathogenesis and that these effects are not due to polar effects on downstream vgrG transcription, as expression of iglE in trans from distal attTn7 fully restored virulence to the iglE-null strains. Finally, a genetic screen developed with LVS was employed to identify key regions of IglE that are required for intracellular replication in J774A.1 macrophages. These genetic analyses, along with additional biochemical studies, demonstrate that IglE is an outer membrane-localized lipoprotein and identify critical residues near its carboxyl terminus that are required for intracellular replication of LVS in J774A.1 macrophages. Our data further suggest that the same key regions are essential for the ability of F. tularensis subsp. tularensis Schu S4 to cause lethal disease in mice.
MATERIALS AND METHODS
Bacterial strains and culture conditions.Strains and plasmids used in this study are listed in Table 1. For routine cultivation, F. tularensis was grown on modified Mueller-Hinton broth (Mueller-Hinton supplemented with 1.23 mM CaCl2, 1.03 mM MgCl, 0.1% [wt/vol] glucose, 0.025% [wt/vol] ferric pyrophosphate, and 2% [wt/vol] IsoVitaleX [BD Biosciences] [sMHB]) or on modified Mueller-Hinton agar (Mueller-Hinton supplemented prior to autoclave sterilization with 1% [wt/vol] tryptone powder, 0.5% [wt/vol] NaCl, and 1.6% (wt/vol) agar and, once cooled to 55°C, further supplemented with 2.5% [vol/vol] heat-inactivated donor calf serum, 0.1% [wt/vol] glucose, 0.025% [wt/vol] ferric pyrophosphate, and 2% [wt/vol] IsoVitaleX [BD Biosciences] [sMHA]). Brain heart infusion (BHI) broth was prepared as previously described (39). For initial recovery of transconjugants, a modified chocolate agar (CA+; Mueller-Hinton agar supplemented prior to autoclave sterilization with 1% [wt/vol] tryptone powder, 0.5% [wt/vol] NaCl, and 1.6% [wt/vol] agar and, once cooled to 55°C, further supplemented with hemoglobin supplement [BD Biosciences], 0.1% [wt/vol] glucose, and 2% [wt/vol] IsoVitaleX BD Biosciences]) was employed. Escherichia coli DH5α or XL-1 Blue was used as the host for routine plasmid manipulation. E. coli S17.1 was used as a host for bacterial conjugation. Where needed, Francisella growth media were supplemented with 200 mg/liter hygromycin (Hyg), 10 mg/liter kanamycin (Kan), 100 mg/liter polymyxin B, 25 mg/liter ampicillin (Amp), or 16 mg/liter vancomycin. E. coli was grown using Luria-Bertani broth or agar further supplemented with 200 mg/liter hygromycin, 30 mg/liter kanamycin, or 100 mg/liter ampicillin. Where needed, sucrose was added to a concentration of 8% (wt/vol) prior to autoclave sterilization.
Bacterial strains and plasmids used in this study
Cloning, expression, and purification of recombinant IglE protein.A portion of the iglE open reading frame(s) [ORF(s); FTT1701/FTT1346], without the predicted N-terminal signal sequence, was amplified from F. tularensis subsp. tularensis Schu S4 genomic DNA using high-affinity Platinum Taq polymerase (Invitrogen) with the oligonucleotide primers GP149 (GGggatccAGTGATGGTTTGTATATCA) and GP150 (CGctcgagTTAATCTTTTTCTATGCTA). The resultant PCR product was directionally cloned into the pProEX HTb vector (Invitrogen) using engineered BamHI and XhoI sites, respectively (lowercase letters in the primer sequences), which created an amino-terminal 6× polyhistidine (H6×) fusion. Recombinant protein was induced for 3 h from clone TP254 containing a sequence-verified insert by the addition of isopropyl-β-d-thiogalactopyranoside to 0.5 mM. Recombinant IglE bearing a polyhistidine amino-terminal fusion tag (H6×-IglE) was then purified by affinity chromatography using equilibrated Ni-nitrilotriacetic acid-agarose (Qiagen, Valencia, CA), as described previously (40). Protein purity and yield were assessed by SDS-PAGE and with a DC protein assay kit (Bio-Rad).
Animal care and use.All procedures involving animals were approved by the University of Texas (UT) Southwestern Medical Center Institutional Animal Care and Use Committee and the Biological and Chemical Safety Advisory Committee. Animals were housed in microisolator cages at the UT Southwestern Animal Resource Center and fed irradiated food and water ad libitum.
Antibody generation and immunoblotting.Polyclonal antisera against H6×-IglE were generated by injecting ∼0.02 mg of recombinant H6×-IglE protein intraperitoneally into 6-week-old female Sprague-Dawley rats (Harlan, Indianapolis, IN), after first emulsifying with an equal volume of complete Freund's adjuvant (Sigma) essentially as described previously (40). Immunoblots were performed using calibrated whole-cell lysates (WSLs) prepared in 1× SDS lysis buffer with 2-mercapetoethanol to an equivalent of 0.2 optical density (OD) units, measured at 600 nm (OD600) per 0.1 ml of SDS lysis buffer and boiling for 10 min. Procedures were otherwise as described by Huntley and associates (40).
Generation of markerless deletion mutants.Splicing-overlap extension (SOE) PCR (41) was used to generate an iglE deletion-insertion cassette in which the majority of the coding region of iglE (encompassing nucleotides 12 to 378 of the 378-bp ORF) was replaced with the FLP recombination target (FRT)-flanked Pfn-kanamycin resistance cassette (FRT-Pfn-kan-FRT) from pLG66a (42) (see the supplemental material for details on strain and plasmid construction). The resultant ΔiglE::FRT-Pfn-kan-FRT amplicon was cut with ApaI and ligated to pTP163 (see the supplemental material), which is a modified version of a previously described sacB-containing suicide plasmid, pMP590 (43). The resultant construct (pTP364) was then transferred without selection from E. coli S17.1 into Francisella by conjugation using filter paper mating. Transconjugants were initially recovered on CA+ with kanamycin and polymyxin B, passaged once on sMHA plus kanamycin to allow secondary recombination, and then passaged on sMHA supplemented with kanamycin and 8% (wt/vol) sucrose to select for clones that had undergone deletion of the wild-type gene and intervening sequences, including the sacB sucrose sensitivity marker. The FRT-flanked kanamycin resistance cassette was then removed essentially as described previously (42) using a modified version of the unstable plasmid pMP829 (44) bearing the sacB counterselection marker from pMP590 (43) and expressing the Flp recombinase obtained from pFFlp-hyg (42) (see the supplemental material). The resultant FLP helper plasmid, pTP512, was introduced into Francisella by electroporation. Hygromycin-resistant (Hygr) transformants were passaged once on drug-free sMHA and patched to screen for sensitivity to kanamycin (i.e., excision of the FRT-flanked kanamycin resistance gene). The resultant kanamycin-sensitive clones were passaged once more through a drug-free intermediate (usually overnight growth in sMHB) and then recovered onto sMHA supplemented with 8% sucrose to select for clones that had lost the unstable FLP helper plasmid pTP512. Excision of the FRT-Pfn-kan-FRT sequence leaves behind a short FRT scar and allows the FRT-Pfn-kan-FRT marker to be recycled for a second round of gene inactivation, which was necessary in order to inactivate both copies of iglE (FTT1701/FTT1346) in two successive rounds of selection.
Genetic complementation via Tn7.For complementation studies, IglE functions were provided in trans from attTn7 using a mini-Tn7 delivery system described previously for use in Francisella (45). Briefly, a PrpsL-iglE amplicon was generated by SOE PCR using Schu S4 genomic DNA as a template. PstI and BamHI sites, engineered into the flanking primers (see the supplemental material), allowed directional cloning into plasmid pUC18T-mini-Tn7T (46). A kanamycin resistance cassette (FRT-flanked PgroE-aphA from pTP086) was ligated into the BamHI site of mini-Tn7 to facilitate selection of recombinant clones. Tn7 helper functions were provided from pTP181, a derivative of the E. coli-Francisella shuttle vector pMP829 (44) expressing the tnsABCD genes from pTNS2 (46). LVS-based and Schu S4-based ΔiglE1 ΔiglE2::FRT clones (TP509 and S4-046, respectively) were electrotransformed with pTP181 and recovered on sMHA containing hygromycin. Following purification of an isolate from a single colony, hygromycin-resistant clones were electrotransformed with pTP565 and recombinant clones were recovered on sMHA containing kanamycin. Kanamycin-resistant clones were passaged once on sMHA lacking hygromycin to promote helper plasmid loss, which was confirmed by patching for antibiotic sensitivity. Integration of Tn7 containing PrpsL-IglE at attTn7 near the 3′ glmS region was verified by diagnostic PCR using the primer pair GP022 (TTTACGATACCGCTTCAGCT) and GP023 (AAGGCTGATATCGCAATAGT), whose binding sites flank attTn7 (see reference 45).
Real-time PCR.RNA was extracted from LVS or TP509 (ΔiglE1 ΔiglE2::FRT) grown at 37°C with slow aeration (115 rpm) in BHI broth (pH 6.8) to an OD600 of ∼0.3 to 0.7. RNA was extracted using the TRIzol reagent (Invitrogen) and then further purified and DNase I digested using a RiboPure kit (Ambion). For quantitative PCRs, 0.1 to 0.2 μg of isolated RNA samples from two independent experiments was analyzed using TaqMan master mix (Applied Biosystems, Foster City, CA), reverse transcriptase (RT), and target-specific probe (2.5 pmol/μl) and primers (2 pmol/μl each). Amplification and fluorescence detection were conducted in an ABI Prism 7500fast sequence detector (PerkinElmer, Applied Biosystems) with a program of 40 cycles, with each cycle consisting of 95°C for 15 s and 60°C for 1 min. Primers and probes were designed using Primer Express software (PerkinElmer, Applied Biosystems). A control that contained all the reagents listed above but that lacked reverse transcriptase was also included to ensure the absence of DNA. All reactions were performed in triplicate. Analysis of relative gene expression employed the 2−ΔΔCT (where CT is the threshold cycle) method (47), employing gyrA as an internal calibrator. Analysis included an adjustment for primer binding efficiency on the basis of standard curves calculated using a genomic DNA control template. Details of the primer and probe sets used for real-time PCR can be found in the supplemental material.
Macrophage culture and infection.Bone marrow cells were isolated from the femurs of 6- to 10-week-old female C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) and differentiated into macrophages for 5 days at 37°C in 7% CO2 in 1 g/liter glucose in Dulbecco's modified Eagle medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen), 10% L929 cell-conditioned medium, and 2 mM l-glutamine in non-tissue-culture-treated petri dishes. After 5 days, loosely adherent BMMs were washed with phosphate-buffered saline (PBS), harvested by incubation in chilled cation-free PBS supplemented with 1 g/liter d-glucose on ice for 10 min, resuspended in complete medium, and replated in 24-well cell-culture-treated plates at a density of 1 × 105 macrophages/well. BMMs were further incubated at 37°C under a 7% CO2 atmosphere for 48 h, and the BMM cultures were replenished with complete medium at 24 h before infection. Immediately prior to infection, a few colonies from a freshly streaked sMHA plate were suspended in sMHB, and the OD600 was measured to estimate bacterial numbers. Bacterial suspensions were then diluted in complete medium, and 0.5 ml was added to chilled BMMs at a multiplicity of infection (MOI) of 50. Bacteria were centrifuged onto macrophages at 400 × g for 10 min at 4°C, and infected BMMs incubated for 20 min at 37°C under a 7% CO2 atmosphere, including an initial, rapid warm-up in a 37°C water bath to synchronize bacterial uptake. Infected BMMs were then washed 5 times with DMEM to remove extracellular bacteria and incubated for 40 min in complete medium and then for an additional 60 min in complete medium containing 100 mg/liter gentamicin to kill extracellular bacteria. Thereafter, infected BMMs were incubated in gentamicin-free medium until processing. The number of viable intracellular bacteria per well was determined in triplicate for each time point. Infected BMMs were washed 3 times with sterile PBS and then lysed with 1 ml of sterile 1% saponin for 3 min at room temperature, followed by repeated pipetting to complete lysis. Serial dilutions of the lysates were rapidly plated onto sMHA and incubated for 3 days at 37°C under 7% CO2 before enumeration of the CFU.
Immunofluorescence microscopy.BMMs grown on 12-mm glass coverslips in 24-well plates were infected, washed 3 times with PBS, fixed with 3% paraformaldehyde, pH 7.4, at 37°C for 20 min, washed 3 times with PBS, and then incubated for 10 min in 50 mM NH4Cl in PBS in order to quench free aldehyde groups. Samples were blocked and permeabilized in blocking buffer (10% horse serum, 0.1% saponin in PBS) for 30 min at room temperature. Cells were labeled by incubating coverslips inverted onto drops of primary antibodies diluted in blocking buffer for 45 min at room temperature. Primary antibodies used were mouse anti-F. tularensis lipopolysaccharide (US Biological, Swampscott, MA) and rat anti-mouse LAMP-1 (clone 1D4B, developed by J. T. August and obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA). Bound antibodies were detected by incubation with 1:500 dilutions in blocking buffer of Alexa Fluor 488–donkey anti-mouse and Alexa Fluor 568–donkey anti-rat antibodies for 45 min at room temperature. Cells were washed twice with 0.1% saponin in PBS, once in PBS, and once in H2O and then mounted in Mowiol 4-88 mounting medium (Calbiochem, Gibbstown, NJ). Samples were observed on a Carl Zeiss LSM 710 confocal laser scanning microscope for image acquisition. Confocal images of 1,024 by 1,024 pixels were acquired and assembled using Adobe Photoshop CS3 software.
Infection of C3H/HeN mice.Seven- to 8-week old female C3H/HeN mice were used for infection. All animals were housed in animal biosafety level 3 facilities. Mice were anesthetized with ketamine plus xylazine and infected intranasally, dropwise, with 0.02 ml (0.01 ml per nostril) of either the Schu S4 parent strain (CDC 1001), the ΔiglE1 ΔiglE2::FRT mutant (S4-046), or the complemented ΔiglE1 ΔiglE2::FRT-Tn7-iglE clone (S4-050), the ΔiglE1 ΔiglE2::FRT-Tn7-iglE (Cys22-Gly) clone (S4-074), or ΔiglE1 ΔiglE2::FRT-Tn7-iglE Δ(Asn96-Asp125) clone (S4-072). Actual infection doses were determined by plating in triplicate onto sMHA, and mice were monitored daily for signs of morbidity and mortality. For experiments requiring tissue harvest, lungs, spleens, and the left lateral lobe of the liver were aseptically harvested from mice and placed in Whirl-Pak bags (Nasco, Fort Atkinson, WI). Three to 5 ml of PBS was added, and tissues were homogenized for 1 min in a stomacher (Seward, Worthing, West Sussex, United Kingdom). Organ homogenates were serially diluted 10-fold in PBS, and 0.02 ml of each dilution was plated onto sMHA containing 100 mg/liter polymyxin B, 25 mg/liter ampicillin, and 16 mg/liter vancomycin. Parallel studies confirmed that the presence of these antibiotics does not interfere with F. tularensis growth. After 72 h of incubation at 37°C in an atmosphere of 5% CO2, the number of CFU was determined for each dilution, and the average number of CFU/organ was calculated.
TX-114 extraction of Francisella proteins.Triton X-114 (TX-114) phase-partitioning studies were preformed on F. tularensis subsp. holarctica LVS (RG004) cultivated in 4 ml of sMHB with moderate aeration until the OD600 reached ∼0.2. The cells were pelleted by centrifugation and washed in 10 ml 1× PBS. Precondensed TX-114 (in 150 mM NaCl, 10 mM Tris, pH 7.5) was added to a final concentration of 2.5% (wt/vol), and the sample was agitated overnight at 4°C and then centrifuged at 6,000 × g at 4°C for 10 min in a prechilled 4°C rotor to remove wholly insoluble material. The supernatant was collected and placed at 30°C for 30 min to allow phase separation. The mixture was centrifuged at 2,500 × g at 25°C for 20 min, and the upper TX-114 detergent-insoluble (IN) phase and lower detergent-soluble (DT) phase were carefully collected. The IN phase was further purified by adjusting the TX-114 concentration to 2.5% (wt/vol) and repeating the phase partitioning twice, as described above. Combined DT phases were further purified by adding an equal volume of aqueous 0.06% (wt/vol) TX-114 and repeating the phase-partitioning process. Proteins in the IN and DT fractions were harvested by precipitation with 2 volumes of 100% ethanol. Samples were rinsed with 70% ethanol, allowed to air dry, and suspended in 10 mM Tris. Equal amounts of protein were prepared for SDS-PAGE, and protein localization was monitored by immunoblotting with monospecific rat polyclonal antisera for FopA (40) or IglE, FTT0507, or FTT0825c.
Radiolabeling with [3H]palmitic acid.Mid-log-phase cultures (OD600, ∼0.14 to 0.2) of TP509 (LVS ΔiglE1 ΔiglE2::FRT) or TP569 LVS (ΔiglE1 ΔiglE2::FRT-Tn7-iglE) grown in Chamberlain's defined medium (CDM) at 37°C with aeration were back-diluted into 5 ml prewarmed CDM containing 0.25 mCi (0.05 ml) of [9,10-3H(N)]palmitic acid (PerkinElmer) to yield an OD600 of ∼0.03. Cultures were incubated overnight at 37°C with aeration to facilitate incorporation of the radiolabeled precursor. The final OD600 of each culture was 0.54 and 0.57, respectively. Cell pellets were recovered by centrifugation at 6,000 × g for 20 min at 10°C and frozen at −20°C until used. Frozen cell pellets were lysed in 0.1 ml B-PER II reagent (Thermo Fisher Scientific) further supplemented with EDTA-free complete protease inhibitor cocktail (Roche). Insoluble material was removed by centrifugation, and IglE protein was recovered from the WCLs by immunoprecipitation using rat polyclonal antisera to IglE (GS99) after first coupling to Dynabeads (Invitrogen) according to the manufacturer's instructions. Naive rat serum coupled to Dynabeads was used in control reactions to show the specificity of the IglE immunoprecipitation reaction. Proteins were prepared for SDS-PAGE and either subjected to immunoblotting with IglE-specific rat polyclonal antisera or treated with En3Hance gel autoradiography enhancement reagent (PerkinElmer), as detailed by the manufacturer. The dried gel was then exposed to CL-X Posure film (Thermo Scientific) at −80°C for 11 days.
Subcellular localization of Francisella proteins.A previously described osmotic lysis and sucrose density gradient centrifugation procedure (40, 48) was used to monitor the subcellular localization of IglE or an IglE Δ(Asn96-Asp125) variant. Sequential gradient fractions were collected dropwise, and the density (g/ml) of each was determined on the basis of the refractive indices. Samples representing each density fraction were prepared for SDS-PAGE, and IglE localization was monitored by immunoblotting with rat anti-IglE antisera. Rat polyclonal antisera against the F. tularensis Pal or SecY protein (40) were used as outer membrane (OM) and inner membrane (IM) localization controls, respectively.
IglE mutagenesis and screening through J774 macrophages.Mutant alleles of iglE were constructed by PCR-mediated site-directed mutagenesis of parent plasmid pTP418 (see the supplemental material). An inverse PCR approach was used to generate specific internal in-frame iglE deletion mutations [i.e., iglE Δ(Tyr2-Iso23), iglE Δ(Asn96-Asp125), iglE Δ(Ser110-Asp125), iglE Δ(Ser120-Asp125)] (see Fig. 8). This results in the addition of a single nontemplated proline residue at the site of each deletion. Hence, a control plasmid (pTP795) bearing no internal deletion but only the nontemplated proline residue was also included. The iglE-H6× construct (pTP744) and the iglE Δ(Ser120-Asp125*) construct (pTP750) were generated using standard PCR approaches. The latter bears the same Ser120-Asp125 deletion mutation as in pTP789, but without the addition of the nontemplated proline, and serves as an additional control. The iglE (Cys22-Gly) point mutation was generated using an SOE PCR approach with specific mutagenic primers in which the nucleotides corresponding to the codon encoding cysteine 22 (UGU) were altered to GGU, encoding glycine 22. The resultant PCR amplicons were cut with the restriction enzymes indicated in the supplemental material and ligated to similarly restricted pTP418 or pUC18T-mini-Tn7T (46). In each case, the FRT-flanked PgroE-aphA cassette from pTP086 was cloned into a distal BamHI site to facilitate selection for Tn7 insertion into the chromosome. Following sequence confirmation, each allele was integrated into the TP563 chromosome using the Tn7 delivery system described above. The mutant alleles were again confirmed by sequence analysis, and protein production was monitored by immunoblotting using rat polyclonal IglE antisera.
Rescue of IglE function was assayed in the immortalized murine J774A.1 macrophage cell line (TIB-67; American Type Culture Collection) using a modified gentamicin protection assay. Briefly, J774A.1 cells were cultured in DMEM containing the GlutaMAX supplement and 4.5 g/liter d-glucose. DMEM was further supplemented with 10% (vol/vol) heat-inactivated fetal bovine serum (FBS; Sigma). J774A.1 macrophages were removed from the culture dishes and seeded to tissue culture 12-well plates at 2.5 × 105 cells/well 1 day prior to infection. J774A.1 cells were infected at an MOI of 20 following opsonization of freshly prepared F. tularensis subsp. holarctica LVS cultures or derivatives with DMEM plus 10% complement-preserved mouse serum (Valley Biomedical, Winchester, VA). For infections, medium was aspirated from J774A.1 cells and 0.3 ml of the adjusted bacterial suspension was added to each well (∼1 × 107 CFU/well). The actual infection dose was confirmed by serial dilution and plating on sMHA. To help promote synchronized infection, the culture plates were centrifuged at 500 × g for 10 min at 20°C. Phagocytosis was allowed to proceed for 1 h at 37°C in an atmosphere of 5% CO2. Macrophages were washed once in PBS at 37°C and then incubated for 1 h at 37°C with DMEM supplemented with 10% FBS and 50 mg/liter gentamicin to kill extracellular F. tularensis. Wells were washed once in PBS at 37°C and incubated in DMEM supplemented with 10% FBS and 5 mg/liter gentamicin. At 2 or 22 h postinfection (p.i.), the macrophages were lysed in 0.25% (wt/vol) saponin in 1× PBS with repeated pipetting to complete lysis. Serial dilutions of the lysates in PBS were then rapidly plated onto sMHA, and the plates were incubated for 3 days at 37°C in an atmosphere of 5% CO2 prior to enumeration of the CFU. The number of viable intracellular CFU per well was determined in duplicate. The attenuation index was calculated as the fold change in the number of CFU between 2 and 22 h p.i. for each test strain over the number of CFU of the complemented parent strain (iglE+) and represents the average of at least two independent determinations ± standard deviations.
IglE protein stability assays.The stability of IglE or genetically modified IglE variants expressed ectopically in the LVS ΔiglE1 ΔiglE2::FRT background was monitored in cultures treated at mid-log phase (OD600, ∼0.1 to 0.19) with 1 mg/liter rifampin and 4 mg/liter chloramphenicol. Samples were recovered at various times after drug addition and chilled on ice, and then the cell pellets were recovered by centrifugation (16,000 × g, 4 min, 4°C). Samples were standardized by adjusting the cell pellets to 0.2 OD unit (OD600) per 0.1 ml SDS-PAGE lysis buffer. Cell lysates were separated by SDS-PAGE, and IglE or FopA was detected following immunoblotting with monospecific rat polyclonal antiserum for IglE or FopA, respectively.
RESULTS
Two-step deletion of duplicate copies of iglE from the Schu S4 and LVS chromosomes.We modified plasmid pMP590 (43), to facilitate efficient two-step gene inactivation of the duplicated iglE locus, using homologous recombination with an FRT-flanked kanamycin resistance cassette and SacB-assisted allelic replacement (Fig. 1). After recovery and passage to ensure purity, clones that had undergone the desired deletion-replacement of one of two copies of iglE (henceforth, iglE1 clones) were selected on sMHA with kanamycin and sucrose. The FRT-flanked kanamycin resistance cassette was then excised, resulting in a markerless gene deletion, by introducing an unstable hygromycin-resistant shuttle vector expressing the FLP recombinase and SacB (pTP512). A second round of homologous recombination was then employed to inactivate the second duplicated copy of iglE (henceforth, iglE2 clones). Helper plasmid loss and gene deletion replacement were monitored at each step by patching for antibiotic resistance or sensitivity and by diagnostic PCR employing primer pairs that flank the targeted regions (data not shown).
Schematic depiction of iglE and flanking genes in F. tularensis subsp. holarctica LVS, F. tularensis subsp. tularensis Schu S4, and associated derivatives following allelic replacement by a two-step, positive selection strategy. Only the relevant portion of the FPI is shown.
Growth of Schu S4 ΔiglE1 ΔiglE2::FRT is attenuated in murine BMMs.Previous studies employing iglE mutants derived from low-virulence F. tularensis subsp. novicida have suggested that iglE is 1 of 16 to 19 genes in the FPI required for intracellular replication (33–36). However, the role of this duplicated genetic locus in virulent type A F. tularensis subsp. tularensis Schu S4 had not been previously tested. We therefore infected BMMs with Schu S4 ΔiglE1 ΔiglE2::FRT (S4-046) and measured intracellular growth using a gentamicin protection assay. In contrast to the virulent Schu S4 parent strain, intracellular replication of the Schu S4 ΔiglE1 ΔiglE2::FRT strain (an iglE-null mutant) was strongly impaired in BMMs, and this strain was steadily cleared from these cells over a 24-h time course (Fig. 2A). Because other FPI mutants have previously been shown to exhibit defects in phagosomal escape, we next asked whether the Schu S4 ΔiglE1 ΔiglE2::FRT mutant was impaired in this regard using a previously described phagosomal integrity assay and confocal immunofluorescence microscopy of bacterial colocalization with LAMP-1-positive membranes (as measures of vacuolar versus cytosolic location) (49). Unlike the Schu S4 wild-type parent strain, which escaped from its original phagosome by 1 h p.i. (∼86% of bacteria were cytosolic; Fig. 2B and C) and replicated extensively in the cytosol by 10 h p.i. (Fig. 2C, top), the ΔiglE1 ΔiglE2::FRT mutant remained enclosed within a vacuole (∼62% of bacteria) at 1 and 10 h p.i. (Fig. 2B and C, middle), consistent with the intracellular killing observed in the gentamicin protection assay (Fig. 2A). Hence, deletion of iglE (FTT1701/FTT1346) abolishes the ability of Schu S4 to escape from the phagosome and survive and replicate intracellularly.
Intracellular replication and phagosomal escape of Schu S4 are dependent on IglE. (A) Schu S4 (black), the iglE-null strain (Schu S4 ΔiglE1 ΔiglE2::FRT; red), and the iglE-null mutant complemented in trans from attTn7 (Schu S4 ΔiglE1 ΔiglE2::FRT-Tn7-iglE; blue) were used to infect BMMs seeded in 24-well plates at an MOI of 50. The intracellular numbers of CFU were determined at various times p.i. Data are presented as the means ± SDs from a representative experiment performed at least twice. (B) Intracellular trafficking of Schu S4 and derivatives in BMM. At various times p.i., infected macrophages were subjected to a phagosomal integrity assay to enumerate the percentage of cytosolic bacteria. Data are the means ± SDs of three independent experiments. Symbols are as defined for panel A. (C) Representative confocal micrographs of BMMs infected for 1 h or 10 h with Schu S4 and derivatives. Samples were processed for immunofluorescence labeling of bacteria (green) and LAMP-1-positive vacuoles (red). Single-channel images of the boxed areas are shown in the magnified insets. White arrows, bacteria of interest.
Complementation of the unmarked ΔiglE1 ΔiglE2::FRT mutant in F. tularensis subsp. tularensis Schu S4 and LVS.As previous studies using low-virulence F. tularensis subsp. novicida failed to successfully complement the intracellular growth defect arising from loss of iglE using a multicopy plasmid-based transcomplementation approach, possibly owing to gene dosage effects (36), it was unclear whether the intracellular growth defect arising from the loss of iglE was due to possible polar effects on downstream FPI genes. We therefore employed a modified Tn7 delivery system (45, 46) to insert a copy of iglE under the control of the Francisella rpsL promoter (PrpsL) in attTn7 near the glmS gene. Loss of IglE expression was confirmed by immunoblotting of the ΔiglE1 ΔiglE2::FRT mutants, whereas restoration of IglE expression to elevated levels was observed for the Tn7-transcomplemented clones (TP569 and S4-050; Fig. 3A). We next tested our Schu S4-based ΔiglE1 ΔiglE2::FRT-Tn7-iglE complemented clone for growth in BMMs. Intracellular replication was fully restored to wild-type levels in the complemented Schu S4 ΔiglE1 ΔiglE2::FRT strain constitutively expressing iglE from attTn7 (Fig. 2A), thus indicating that loss of iglE1 and iglE2 was responsible for this defect. Complementation of the ΔiglE1 ΔiglE2::FRT mutant also restored phagosomal escape and cytosolic replication (Fig. 2B and C, bottom). As genetic complementation of the intramacrophage growth defect of the Schu S4 ΔiglE1 ΔiglE2::FRT mutant was accomplished in trans, these data clearly indicate that the loss of iglE and not polar effects on downstream FPI genes was indeed responsible for the intracellular growth defect. This observation is important, as two independent real-time quantitative PCR studies indicated a slight but highly reproducible decrease in vgrG (FTT1702) transcript amounts by ∼30% in strain TP509 bearing FRT-marked null mutations in both iglE1 and iglE2 (relative quantity of vgrG, 0.67 and 0.73, respectively, compared to that in the LVS parent strain) (data not shown). As no similar reductions were observed for the unlinked control gene, FTT1714c (relative quantity of FTT1714c, 0.98 and 1.04, respectively), we interpret this result to mean that the loss of iglE or the retention of the FRT scar in the deletion mutant has in some way impacted transcription levels or the transcript stability of vgrG, which is likely cotranscribed with iglE (Fig. 1).
Immunoblot analysis of IglE or IglE variants in F. tularensis. (A) Calibrated whole-cell lysates of LVS, Schu S4, the iglE-null strains (ΔiglE), or the iglE-null strains expressing wild-type iglE (iglE+), iglE Δ(Asn96-Asp125), or iglE (Cys22-Gly) alleles from attTn7 were separated by SDS-PAGE and analyzed by immunoblotting with rat polyclonal IglE antisera. FopA levels were monitored by immunoblotting with rat polyclonal FopA antiserum as a loading control. (B) Differences in the mobility of IglE (wild-type) and IglE (Cys22-Gly) observed in SDS-polyacrylamide gels followed by immunoblotting with rat polyclonal IglE antisera.
The virulence of Schu S4 ΔiglE1 ΔiglE2::FRT is attenuated in C3H/HeN mice.Intramacrophage growth is considered to be a hallmark of Francisella pathogenicity in vivo (7). To determine whether the F. tularensis subsp. tularensis Schu S4 ΔiglE1 ΔiglE2::FRT mutant was defective for virulence in vivo, C3H/HeN mice were infected i.n. with 447 CFU Schu S4 or 356 CFU Schu S4 ΔiglE1 ΔiglE2::FRT (>10 times the i.n. lethal dose for Schu S4 [strain CDC 1001] in our hands) and monitored for survival and/or bacterial burdens in the lung, liver, and spleen at various times postinfection. Use of an elevated i.n. infection dose served to ensure that any attenuation observed for the iglE deletion mutant was not marginal relative to that observed for the wild-type virulent parent. Whereas increasing concentrations of bacteria were recovered from the lungs, spleens, and livers of Schu S4-infected mice at days 3 and 5 postinfection, the Schu S4 ΔiglE1 ΔiglE2::FRT mutant was not recovered from any of these tissues at any time point (limit of detection, 150 CFU/spleen or lung and 446 CFU/liver), indicating that the ΔiglE1 ΔiglE2::FRT mutant exhibits a significant (P < 0.001, day 3 and 5, lung and spleen; P < 0.01, day 3; P < 0.001 day 5, liver; Bonferroni posttests) defect for in vivo survival and/or replication (see Fig. 5). Consistent with this observation, all of the Schu S4-infected mice either were severely moribund or had succumbed to disease by day 5 (0 of 8 survived; median time until death, 5 days), whereas none of the ΔiglE1 ΔiglE2::FRT mutant-infected mice died or showed any overt signs of illness during the infection (P < 0.0001, log-rank [Mantel-Cox] test) (Fig. 4). Indeed, we observed no evidence for illness in C3H/HeN mice infected i.n. with an even higher i.n. challenge dose (∼35,600 CFU) of the Schu S4 ΔiglE1 ΔiglE2::FRT mutant, indicating that the 50% lethal dose for this strain is greater than 3.6 × 104 CFU (Fig. 4). As with our infection of BMMs, the virulence defect in the ΔiglE1 ΔiglE2::FRT mutant was restored in the complemented strain constitutively expressing iglE in trans from attTn7 (no significant difference between Schu S4 and the ΔiglE1 ΔiglE2::FRT-Tn7-iglE complemented clone; P < 0.0005 for the ΔiglE1 ΔiglE2::FRT mutant versus the ΔiglE1 ΔiglE2::FRT-Tn7-iglE complemented clone, log-rank [Mantel-Cox] test) (Fig. 4). All animals infected via the i.n. route with ∼183 CFU of the ΔiglE1 ΔiglE2::FRT-Tn7-iglE mutant succumbed to infection, with only a slight delay in the median time until death (7 days for animals infected with the ΔiglE1 ΔiglE2::FRT-Tn7-iglE complemented clone) relative to that for animals infected with the Schu S4 parent (5 days for animals infected with wild-type Schu S4) being observed (Fig. 4). Organ CFU burdens in the lungs, livers, and spleens of animals infected with the complemented clone were also slightly reduced relative to those in the tissues of animals infected with the fully virulent parent strain Schu S4 on day 3 (P < 0.05, lung; not significant, liver; P < 0.01, spleen; Bonferroni posttests) but reached equivalent levels by day 5. Importantly, these values were significantly elevated over those for animals infected with the ΔiglE1 ΔiglE2::FRT mutant (P < 0.001, days 3 and 5, lung; P < 0.01, day 3; P < 0.001, day 5, spleen; P < 0.05, day 3; P < 0.001, day 5, liver; Bonferroni posttests), which could not be recovered from any tissues at any time point tested (Fig. 5). The slight delay in colonization and lethality in mice infected with the ΔiglE1 ΔiglE2::FRT-Tn7-iglE complemented clone most likely reflects elevated constitutive expression of iglE from PrpsL (Fig. 3). However, we cannot formerly exclude the possibility that slightly reduced vgrG transcription might also contribute to this effect. Regardless, these in vivo complementation data, along with the full restoration of in vitro intramacrophage growth with expression of iglE in trans from attTn7, strongly support the notion that the loss of iglE is responsible for the pathogenesis defect observed.
Wild-type IglE is required for lethality of Schu S4 in C3H/HeN mice. Mice were infected i.n. with 447 CFU Schu S4 (closed circles), 356 or 35,600 CFU Schu S4 ΔiglE1 ΔiglE2::FRT (open circles), 183 CFU Schu S4 ΔiglE1 ΔiglE2::FRT-Tn7-iglE (filled triangles), 170 CFU Schu S4 ΔiglE1 ΔiglE2::FRT-Tn7-iglE (Cys22-Gly) (open triangles), or 183 CFU Schu S4 ΔiglE1 ΔiglE2::FRT-Tn7-iglE Δ(Asn96-Asp125) (filled diamonds) (n = 5 to 8 per group) and monitored for signs of morbidity for up to 3 weeks postinfection. The data are representative of two independent experiments. ***, P < 0.001 (log-rank test).
Mice infected with the Schu S4 ΔiglE1 ΔiglE2::FRT mutant show reduced bacterial burdens. Mice were infected i.n. with ∼102 CFU Schu S4 or derivatives. On days 3 and 5 postinfection, two to four mice were humanely sacrificed, and the bacterial burdens were determined following serial dilution and plating of organ homogenates for determination of the numbers of CFU. The limits of detection were 150 CFU/organ for spleens and lungs and 446 CFU/organ for livers. The data are presented by scatter dot plot, and the horizontal lines indicate the mean result. The bacterial burdens for mice infected with each strain were compared using two-way analysis of variance and are representative of two independent experiments. Filled circles, Schu S4; open circles, Schu S4 ΔiglE1 ΔiglE2::FRT; open triangles, Schu S4 ΔiglE1 ΔiglE2::FRT-Tn7-iglE.
IglE is a lipoprotein.iglE is predicted to encode a lipoprotein on the basis of the presence of a putative signal peptidase II (SPII) cleavage site and a conserved cysteine residue at position 22 in the iglE-coding sequence. However, there are limited published data to substantiate this directly (22). To test this, we first analyzed the behavior of IglE in the nonionic detergent TX-114. TX-114 has been shown to solubilize bacterial lipoproteins due to their amphipathic properties imparted by the three covalently attached long-chain fatty acids (50). Using this approach on the biosafety level 2 LVS surrogate, we found enrichment of IglE in the TX-114 detergent-soluble fraction, with no material remaining in the TX-114 detergent-insoluble fraction (Fig. 6A). Controls included two putative lipoprotein candidates, FTT0507 and FTT0825c (unpublished), or the OM protein (OMP) FopA (40). In each case, the candidate lipoproteins, like IglE, were solubilized into the TX-114 detergent-soluble phase, whereas FopA was found in both TX-114 detergent-soluble and -insoluble fractions (Fig. 6A). The basis for the partial fractionation of FopA to both the TX-114 detergent-soluble and -insoluble fractions is at present unknown but may reflect aggregation of internal hydrophobic patches resulting in the heightened solubility of this beta barrel-containing integral OMP. Therefore, to verify the TX-114 phase-partitioning results, LVS ΔiglE1 ΔiglE2::FRT (ΔiglE) or LVS ΔiglE1 ΔiglE2::FRT-Tn7-iglE (iglE+) was grown in CDM and then pulsed for ∼18 h with the radiolabeled long-chain fatty acid precursor [3H]palmitic acid, shown previously to be incorporated into F. tularensis lipoproteins (51, 52). Exposure to [3H]palmitate resulted in numerous labeled proteins (Fig. 6B), one of which was immunoprecipitated from LVS ΔiglE1 ΔiglE2::FRT-Tn7-iglE (iglE+) by anti-IglE serum but not by control preimmune serum (Fig. 6B and C). As such, the presence of multiple bands in the autoradiograph (Fig. 6B), but not in the corresponding immunoblot (Fig. 6C), most likely reflects differences in protein loading or poor resolution through SDS-PAGE, as the former strain contained 30 times more protein to ensure that a good signal-to-noise ratio was achieved. As expected, no such band was seen for LVS ΔiglE1 ΔiglE2::FRT, which is an iglE-null mutant and does not produce IglE protein. Taken together, these data are consistent with the findings of in silico analyses identifying IglE as a lipoprotein.
IglE encodes a bacterial lipoprotein. (A) TX-114 detergent-soluble and -insoluble phases were prepared from LVS, separated by SDS-PAGE, and analyzed by immunoblotting with the indicated monospecific polyclonal antisera. Abbreviations: DT, TX-114 detergent-soluble material; IN, TX-114 detergent-insoluble material; IglE, 0507, 0825, and FopA, blots probed with rat polyclonal antibodies monospecific for IglE, FTT0507, FTT0825c, and FopA, respectively. (B) Autoradiograph demonstrating in vivo incorporation of [3H]palmitic acid into polypeptides in the LVS iglE-null mutant (ΔiglE) and the iglE-null mutant complemented in trans from attTn7 (iglE+). Abbreviations: WCL, whole-cell lysate; IP-αIglE, immunoprecipitate obtained using rat anti-IglE antiserum-coupled Dynabeads; IP-NS, immunoprecipitate obtained using naive rat serum-coupled Dynabeads. (C) Analysis of IglE in immunoprecipitated fractions by SDS-PAGE and immunoblotting with rat polyclonal IglE antisera. (D) The N terminus of IglE contains a lipobox motif (bold) including a canonical cysteine at position 22 (underlined).
OM localization of IglE.Because IglE appears to encode a lipoprotein, we next asked whether this protein is retained at the IM or if it is shuttled to the OM. In most Gram-negative bacteria, lipoprotein export to the OM or retention at the IM is governed by the Lol sorting machinery and sequences embedded within the amino terminus of the mature lipoprotein (the so-called N + 2 rule [reviewed in reference 53]). Usually, this is the amino acid immediately following the canonical cysteine site of lipidation, wherein any amino acid other than aspartate at position N + 2 allows export to the OM (53). In IglE, the N + 2 amino acid is isoleucine, which should facilitate export to the OM. To verify this directly and compare the membrane distribution of IglE, strain TP569 constitutively expressing wild-type IglE from PrpsL in attTn7 was subjected to osmotic lysis and OM and IM fractionation experiments (40, 48). These experiments were initially performed using strain TP569 (LVS ΔiglE1 ΔiglE2::FRT-Tn7-iglE+) to increase the cellular content of IglE. The quality of separation was monitored by immunoblotting with serum specific for the IM protein SecY (found here in fractions with densities of 1.14 to 1.15 g/ml) and the OM protein Pal (found principally in fractions with densities of 1.16 to 1.19 g/ml) (Fig. 7). Using this technique, IglE was found predominantly within the OM fractions (densities, 1.16 to 1.19 g/ml) (Fig. 7). Some IglE was also observed in fractions with lower densities (1.13 to 1.11 g/ml), which do not correspond to either membrane fraction. Although we cannot fully explain this result on the basis of the available data, de Bruin and associates (36) reported similar findings for FPI-encoded IglA, IglB, and IglC proteins, which were speculated to form an as yet unidentified insoluble macromolecular structure. Finally, to exclude the possibility that the OM localization of IglE was influenced by constitutive expression of native IglE protein in trans from a distal site in the chromosome, we verified that native IglE from LVS had similar OM localization properties when grown under FPI-inducing (39) growth conditions in BHI broth (data not shown).
Subcellular localization of IglE in F. tularensis. The LVS iglE-null mutant expressing wild-type iglE in trans from attTn7 (iglE+) or the same strain expressing iglE Δ(Asn96-Asp125) was subjected to osmotic lysis and sucrose density fractionation as described in Materials and Methods. Sequential fractions were collected from sucrose gradients, separated by SDS-PAGE, and analyzed by immunoblotting using rat polyclonal IglE antisera. Rat polyclonal antisera specific for Pal, a known outer membrane protein, or SecY, a known inner membrane protein (IMP), were used as controls for localization. Sucrose gradient densities (g/ml) are shown at the top.
Mutant forms of IglE.The crystal structure of IglE has been solved (38), but there are few other experimental data to explain how this lipoprotein (Fig. 6), which lacks obvious orthologs outside Francisella species, might contribute to F. tularensis pathogenesis. In an effort to define the key regions and residues required for IglE function, we employed PCR-based mutagenesis approaches to generate a series of small deletion mutations or a point mutation in iglE using our PrpsL-iglE Tn7 clone (pTP418) as a template. After integration of the modified clones into the LVS iglE-null background, followed by sequence verification, we screened each for its ability to rescue the intramacrophage growth defect of the LVS iglE-null mutant when expressed under PrpsL control from attTn7. Because our data suggest that iglE encodes an OM-localized lipoprotein, we first constructed a mutation that altered the canonical cysteine 22 lipidation site to glycine [IglE (Cys22-Gly)]. The LVS iglE-null strain expressing iglE (Cys22-Gly) from attTn7 [LVS ΔiglE1 ΔiglE2::FRT-Tn7-iglE (C22-G)] was strongly impaired for intramacrophage growth (Fig. 8B), and a Schu S4 variant [Schu S4 ΔiglE1 ΔiglE2::FRT-Tn7-iglE (C22-G)] expressing the iglE Cys22-Gly allele in the iglE-null background was wholly avirulent following i.n. inoculation of ∼170 CFU into C3H/HeN mice (Fig. 4). Immunoblot analysis with IglE-specific polyclonal sera indicated that IglE (Cys22-Gly) is synthesized but appears to be less abundant than wild-type IglE expressed under the same conditions (Fig. 3A). The IglE (Cys22-Gly) variant also showed slower mobility in SDS-polyacrylamide gels than wild-type IglE, which ran with an observed molecular mass of ∼13.9 kDa (Fig. 3B). Although the exact molecular mass of the IglE (Cys22-Gly) protein could not be proven here, linear regression analysis indicates that this mobility shift correlates with a predicted ∼0.5- to 1-kDa change in overall mass. This result is consistent with the prediction that IglE (Cys22-Gly) should no longer be recognized for processing by SPII (predicted molecular mass of unprocessed IglE, 14.6 kDa) and, hence, should not be further modified through covalent fatty acid addition. Further studies would be necessary to prove this directly. By comparison, an IglE Δ(Asn96-Asp125) variant (Fig. 3A; see below) showed increased mobility in SDS-polyacrylamide gels, with a calculated mass change of ∼1.8 to 1.9 kDa, which is, again, in general agreement with a predicted molecular mass of 11.3 kDa (Fig. 3A). Taken as a whole, these data are interpreted to mean that cysteine 22 and, hence, the lipidation of IglE are required for intramacrophage growth and pathogenesis of F. tularensis.
Identification of IglE variants that fail to rescue intracellular growth. (A) Schematic depiction of regions of iglE mutated in this study. (B) Derivatives of the LVS iglE-null mutant or the same strain expressing wild-type or mutated forms of iglE from attTn7 were used to infect J774.A1 macrophage-like cells seeded into 12-well plates at an MOI of 20. The numbers of intracellular CFU were determined at 2 and 22 h p.i. The data are presented as the attenuation index calculated as the fold change in the number of CFU between 2 and 22 h p.i. for each test strain over the number for the complemented parent strain (iglE+) and represent the means of at least two independent determinations ± SDs. (C) Analysis of IglE abundance from calibrated protein lysates following separation by SDS-PAGE and immunoblotting with rat polyclonal IglE antisera. FTT0831c levels were monitored by immunoblotting with rat polyclonal FTT0831c antiserum as a loading control. IglE Δ(S120-D125*) is equivalent to IglE Δ(S120-D125) but lacks the nontemplated proline added as part of the inverse PCR mutagenesis strategy.
Because the IglE Cys22-Gly mutation likely exerts its effects by altering protein localization or other properties of this OM-localized lipoprotein (i.e., stability) but not IglE function per se, we next constructed a series of overlapping deletion mutations near the carboxyl-terminal region of this protein and assayed these for rescue of the intramacrophage growth defect of the LVS iglE-null mutant. This region was selected because IglE lacks obvious orthologs or functional domains (as found in the current Pfam database), and the carboxyl terminus of IglE likely extends into the periplasm or away from the cell surface and would be most readily accessible to facilitate possible protein-protein or other interactions. Three overlapping deletion mutations encompassing amino acids Asn96-Asp125, Ser110-Asp125, or Ser120-Asp125 (Fig. 8A) of IglE were constructed using the LVS iglE-null strain as a recipient host. These were then assayed for rescue of intramacrophage growth in J774 macrophages. As is shown in Fig. 8B, expression of wild-type iglE (iglE+) or a mutant allele of iglE lacking Ser120-Asp125 [IglE Δ(Ser120-Asp125) or IglE Δ(Ser120-Asp125*); the latter lacks the nontemplated proline added as part of the inverse PCR mutagenesis strategy] from attTn7 fully restored the intramacrophage growth of the iglE-null strain. This indicates that in trans expression of some relatively short carboxyl-terminal iglE deletion mutations is well tolerated and can fully complement the intramacrophage defect of the LVS iglE-null mutant in J774 cells. To this end, expression of full-length iglE bearing a carboxyl-terminal polyhistidine tag (strain TP814) also fully rescued intramacrophage growth of the iglE-null mutant (Fig. 8B). In contrast, strains expressing alleles of iglE bearing larger carboxyl-terminal deletions [IglE Δ(Asn96-Asp125) or IglE Δ(Ser110-Asp125)] were significantly impaired (Fig. 8B). To determine if altered protein was expressed, we performed an immunoblotting experiment with polyclonal rat antiserum specific for IglE. Both IglE Δ(Asn96-Asp125) and IglE Δ(Ser110-Asp125), which failed to rescue intramacrophage growth, were readily detected, albeit at a reduced abundance relative to that for the constitutively expressed wild-type form of IglE (iglE+). In contrast, we were unable to detect IglE Δ(Tyr2-Iso23) lacking the signal peptide (Fig. 8A) in the immunoblot, indicating that it is likely unstable (data not shown).
To address the possibility that a general inherent instability of some mutant forms of IglE was responsible for their inability to rescue the intracellular growth of an LVS iglE-null strain, TP569 expressing wild-type IglE (iglE+) and TP652 iglE Δ(Asn96-Asp125), TP809 iglE Δ(Ser120-Asp125), TP816 iglE Δ(Ser110-Asp125), and TP656 iglE (Cys22-Gly) were grown to mid-log phase (OD600, ∼0.1 to 0.19) and treated with a combination of rifampin and chloramphenicol at 1 and 4 mg/liter, respectively, to block transcription and translation, respectively. As expected, inhibition of cell growth was observed 60 min after drug addition, indicating that transcription and translation within the treated cells had indeed ceased (data not shown). We then performed an immunoblot using IglE-specific rat polyclonal antiserum to monitor the cellular stability of preexisting IglE over a defined period of time. FopA levels were monitored in parallel as an internal loading control. IglE was readily detected by immunoblotting from protein extracts prepared at multiple time points through 150 min after drug addition. We predicted that if mutant forms of IglE were unstable and subject to constitutive proteolysis, we would observe an appreciable depletion of the nonreplenished cellular pool of the IglE protein over time. Although we did observe a slight reduction in the cellular abundance of some mutant forms of IglE [e.g., IglE Δ(Ser110-D125) and IglE (Cys22-Gly)] over time, the net loss of protein over the course of 150 min of drug treatment was minimal and not likely to account for the appreciable difference in protein abundance in the cell (Fig. 9). In contrast and, again, in agreement with the findings from our earlier immunoblotting experiments, differences in the cellular abundance of some mutant forms of IglE were again apparent in the untreated samples (zero time point; compare the results for IglE to those for the internal loading controls in Fig. 9). This is interpreted to mean that the IglE variants that failed to rescue intracellular growth [i.e., IglE Δ(Asn96-Asp125), IglE Δ(Ser110-Asp125), and IglE (Cys22-Gly)] are not wholly unstable or subject to rapid general proteolysis per se. Further, as IglE is produced constitutively in the iglE-null background from the same site (attTn7) under PrpsL control, it seems unlikely that differences in transcription or translational efficiency could account for these differences. Rather, we argue that some mutant forms of IglE protein must be misfolded and degraded immediately following release from the ribosome, whereas other forms (that still remain in the cell) are protected. This could occur either through inefficient folding of some mutant forms of IglE into a stable (protease resistant) form or, possibly, via protection from degradation by some as yet unknown secondary stabilizing interaction.
Intracellular stability of IglE or IglE derivatives expressed from attTn7 in the F. tularensis subsp. holarctica LVS iglE-null background following treatment with rifampin and chloramphenicol. Calibrated protein lysates prepared at the indicated times after drug addition were separated by SDS-PAGE and assayed by immunoblotting with rat polyclonal IglE antisera. FopA levels were monitored by immunoblotting with rat polyclonal FopA antisera as a loading control.
Lastly, in order to verify that the results of our genetic screen with LVS in J774.A1 macrophages were indicative of the fact that key residues of IglE are required for function in virulent type A Schu S4, a Schu S4 iglE-null strain expressing iglE Δ(Asn96-Asp125) from attTn7 (Fig. 3A) was constructed and tested for systemic lethal infection of C3H/HeN mice following i.n. instillation of 273 CFU. Similar to what was observed with the Schu S4 iglE (Cys22-Gly) mutant, Schu S4 expressing iglE Δ(Asn96-Asp125) was attenuated, and none of the animals became ill during the course of the infection, indicating complete attenuation (Fig. 4). To eliminate the possibility that altered membrane trafficking was responsible for the failure of IglE Δ(Asn96-Asp125) to rescue the in vivo and intracellular growth of the Schu S4 and LVS iglE-null strains, respectively, we repeated cellular subfractionation experiments on TP652 expressing the iglE Δ(Asn96-Asp125) allele following osmotic lysis and separation of the OM and IM fractions. As is shown in Fig. 7, the IglE Δ(Asn96-Asp125) variant was found exclusively in the OM fraction. However, in contrast to wild-type IglE, we noted that this IglE variant was not also found in the nonmembrane fraction with a density of 1.11 to 1.13 g/ml (Fig. 7). These combined data are interpreted to mean that regions in the carboxyl-terminal one-third of the mature IglE protein, specifically, those encompassing asparagine 96 through serine 120, contribute to IglE function and, to some extent, IglE stability but are not required for proper membrane trafficking, whereas other regions, namely, serine 120 through aspartic acid 125, are dispensable for function.
DISCUSSION
The virulence of F. tularensis is thought to depend on its ability to invade, survive, and replicate within host macrophages. Many factors required for this process have been identified, principally on the basis of work with F. tularensis subsp. novicida and LVS. However, confirmation of the roles of many of these factors in fully pathogenic F. tularensis subsp. tularensis has been limited, and in some cases, the roles of these factors between these closely related subspecies have been inconsistent among studies (4–6). As such, there has been increased recognition of the need to validate key findings for the more pathogenic strains using relevant infection models. The role of IglE in Francisella pathogenesis has been evaluated previously. Weiss and associates identified all genes in the FPI, including iglE, during a transposon-based negative-selection screen of F. tularensis subsp. novicida U112 virulence using an intraperitoneal murine infection model (35). iglE was also found to be essential for intracellular replication of F. tularensis subsp. novicida U112 in murine RAW264.7 (34) and J774 (36) macrophage-like cells. In a set of unrelated studies, iglE was required by F. tularensis subsp. novicida U112 for wild-type colonization and full lethality of Drosophila melanogaster flies (33) but was not similarly identified in a second genetic screen employing cultured D. melanogaster-derived S2 cells (54). Importantly, none of these studies addressed the possibility of polar effects on downstream FPI genes, and efforts to complement the defects arising from the loss of iglE either were not reported (33, 34) or were not successful (36). As such, the absolute role of iglE in F. tularensis subsp. novicida pathogenesis has remained unclear. Here we show that iglE is required for intramacrophage replication of strain Schu S4 in BMMs and LVS in J774 macrophage-like cells and that this defect is fully reversed by constitutive expression of iglE in trans from PrpsL in attTn7. To our knowledge, this represents the first published report demonstrating a role for iglE in the intracellular survival of virulent type A Schu S4. Further, our ability to complement this defect shows that it is the loss of iglE and not polar effects caused by the iglE mutation that gives rise to the intramacrophage growth defect. This establishes, for the first time, a genotype-phenotype relationship for iglE, which is likely transcribed as part of a polycistronic operon that includes vgrG. Whereas iglE was found to be essential for the virulence of Schu S4 in C3H/HeN mice, prior intranasal immunization with Schu S4 ΔiglE1 ΔiglE2::FRT at doses well above those reported to protect mice with the current, unapproved LVS (55; G. T. Robertson, unpublished observation) failed to protect mice from death following secondary virulent Schu S4 pulmonary challenge 42 days after the initial immunization (data not shown). These observations are entirely consistent with those published previously, suggesting that highly attenuated F. tularensis strains fail to engender a significant or protective immune response (31).
Early studies suggested that certain FPI genes are required for intramacrophage growth, whereas others are not. Hierarchal clustering analysis of mutant phenotypes resulting from a genetic screen of candidate virulence factors contributing to F. tularensis subsp. novicida U112 virulence in D. melanogaster and J774 mouse macrophages indicated strong correlations among mutants lacking 14 of 18 FPI genes (including iglE), which was interpreted to mean that many of these genes likely contribute to a common function (33). However, this model is not supported by all of the results obtained to date. For example, whereas both LVS iglI and iglG mutants showed essentially wild-type growth in J774 macrophages, only the iglG mutant was proficient for growth in peritoneal exudates and BMMs (22). Differences in intracellular replication and induction of host cell death pathways were also observed following infection of J774 macrophages with pdpC, iglC, iglG, or iglI mutants of LVS (56). Consistent with this result, Long et al. found that while Schu S4 iglI and IglJ were essential for phagosomal escape and alteration of endosomal trafficking, a mutant lacking pdpC was only partially defective in this regard (30). This led to the conclusion that PdpC contributes to, but is not essential for, remodeling of the host phagosomal pathway (30). A similar result was reported for an LVS pdpC mutant (57). As pdpC was one of four genes (along with pdpE, pdpD, and anmK) not similarly required for lethality of F. tularensis subsp. novicida in the D. melanogaster fly model, it is tempting to speculate that a subset of FPI genes performs a common, nonredundant function and is essential for FPI activity and, hence, pathogenesis in multiple host species; another set may contribute to pathogenesis but is not absolutely essential for FPI activity per se. The extent of apparent phagosomal escape observed here for the Schu S4 iglE-null strain (i.e., 38% of bacteria were cytoplasmic) was higher than that reported for Schu S4 variants lacking either iglC (31), iglI, or iglJ or the FPI regulator fevR (30). Instead, the reduction in phagosomal escape observed in the Schu S4 iglE-null strain was more similar to that for Schu S4 lacking pdpC, which was only partially defective in this regard (30). In contrast to the latter study, however, our results indicate that iglE is essential for colonization and dissemination in the C3H/HeN mouse model following i.n. challenge. This appears to be inconsistent with the observations by Long and associates for a Schu S4 pdpC strain when administered to BALB/c mice at a much higher i.n. infection dose of ∼106 CFU (30). Further studies comparing the Schu S4 iglE-null strain to one or more of these mutants or an FPI deletion strain would be necessary to resolve this question.
Kinetic studies indicate that iglE expression is induced slightly during intracellular residence and peaks near the end of the cytosolic replication phase (58), but the function of this protein is unknown. IglE is annotated as a hypothetical protein but encodes several features that are typical of bacterial lipoproteins, including a short amino-terminal signal peptide bearing a positively charged N-terminal region, followed by a hydrophobic region and a conserved lipobox motif (Leu-Ser-Ser-Cys) which includes an invariant cysteine at position 22. Consistent with these features, IglE is predicted to encode a bacterial lipoprotein, on the basis of in silico analysis using the LipoP (version 1.0) lipoprotein prediction server (59), which gave a reliability score of 12.5. Our TX-114 phase-partitioning and [3H]palmitate incorporation studies support the prediction that IglE is a bacterial lipoprotein. Further, we predict that wild-type IglE is exclusively localized to the OM, based on osmotic lysis and separation of IM and OM fractions. Mutation of the invariant cysteine at position 22 to a glycine resulted in an IglE variant that was incapable of rescuing an iglE-null mutant of LVS for intramacrophage replication in J774 cells or a Schu S4 iglE-null strain for murine virulence when administered by the i.n. route. However, the molecular basis for this attenuation is not yet known. As measured by immunoblot, the IglE (Cys22-Gly) variant is present at lower levels and exhibits somewhat decreased stability in time course studies. However, the IglE (Cys22-Gly) variant is not wholly unstable. Further, whereas preliminary studies employing Sarkosyl enrichment of OMPs suggest differential fractionation of IglE (Cys22-Gly) to both the soluble and IM fractions, IglE (Cys22-Gly) was observed exclusively within the OM fraction following osmotic lysis and differential separation through discontinuous gradient centrifugation (G. T. Robertson, unpublished). At present, we do not have an explanation for these contradictory results. Regardless, our in vivo and intramacrophage survival data indicate that cysteine 22 and, by extension, processing by signal peptidase II and lipidation are essential for the biological function of IglE.
The presence of an N-terminal signal sequence and our combined biochemical and genetic data suggesting that IglE is an outer membrane-anchored lipoprotein are inconsistent with the results of a second study which showed that IglE coupled to a beta-lactamase reporter is secreted from the cell in a manner that is dependent on other core FPI proteins (37). The same authors also noted that PdpE (a second secreted FPI-encoded protein) also possesses a signal peptide (37). This raises the intriguing possibility either that the Sec and T6SS pathways are connected in Francisella or that some FPI-secreted substrates (i.e., IglE) have two biological functions, one which requires a membrane-anchoring lipid moiety (e.g., T6SS assembly) and another where soluble protein is required (e.g., as a T6SS substrate of unknown function).
Taken as a whole, the studies reported here provide strong evidence that iglE is a critical virulence factor for virulent Schu S4 and that the loss of iglE is indeed responsible for the overall defect in virulence observed. Given that the loss of iglE results in defects in both intracellular growth and phagosome escape with characteristics similar to those of other FPI mutants, we speculate that IglE is important for FPI function, assembly, and/or regulation, possibly as a structural or pilot protein. Consistent with this notion, ectopically expressed IglE was observed to localize to the OM of LVS, and its activity required amino acid sequences at or near its carboxyl terminus. The FPI is proposed to encode a putative secretion apparatus with structural and other distant similarities to the T6SS found in many diverse bacterial species (reviewed in reference 22). However, there is little direct evidence to date to suggest that a specific secretion apparatus exists. Several studies have demonstrated specific interactions between various FPI-encoded proteins (23, 28, 60, 61), and in the case of IglA, this interaction was shown to be critical for virulence and dependent on a conserved α-helical domain (28). A more recent biochemical study by de Bruin and associates posits that a non-membrane-associated multiprotein structure can be observed in immunoblots of protein fractions assayed following separation of the OM and IM through discontinuous sucrose gradients (36). Our results also suggest a correlation between the ability to successfully complement the loss of IglE in trans and the formation of a non-membrane-associated protein fraction that is observed in sucrose densities between 1.11 and 1.13 g/ml. Further studies on pooled fractions containing this material would be necessary to resolve whether this represents protein aggregation or some type of structured apparatus. However, given that wild-type IglE associates with these fractions, whereas a nonfunctional IglE Δ(Asn96-Asp125) variant does not, it is tempting to speculate that the regions of IglE that are required to facilitate an IglE protein-protein association(s) and, possibly, formation of an FPI-encoded structure are defined by this deletion. Precedence for this idea comes from studies of T6SS from other organisms. For example, TssJ (SciN) is required for type VI secretion and biofilm formation in enteroaggregative E. coli; inactivation of the gene encoding TssJ or sequestration at the IM (by altering the N + 2 residue for lipoprotein sorting) blocked TssJ translocation and protein function (62). Structural studies on TssJ (SciN) revealed a loop that was critical for protein-protein interactions with the IM protein TssM, thus defining a link between the OM and IM in a manner that was proposed to form a channel to accommodate the insertion of a secretion apparatus (63); a similar model has been proposed for Vibrio cholerae (64). Interaction of a T6SS-encoded putative lipoprotein, the N-terminal domain of an IcmF homologue, and the C-terminal domain of a third protein of unknown function (EvpA) is also required for T6SS function in Edwardsiella tarda, a pathogen of fish and humans (65). Other critical, but nonstructural, roles have also been reported for OM-localized lipoproteins in the regulation of T6SS function. For instance, TagQ, an OM-anchored lipoprotein, is required for transmembrane signaling that promotes H1-T6SS activity in Pseudomonas aeruginosa in response to environmental cues (66). Our genetic data suggest that IglE must reach a proper cellular location and that regions in the carboxyl terminus of IglE are necessary for intracellular growth and pathogenesis. As such, we propose that these effects are mediated by as yet undefined protein-protein interactions. However, it remains to be determined whether these effects are instead a result of altered cellular abundance (Fig. 8C) or other unforeseen properties resulting from deletion of the carboxyl-terminal region of this hypothetical protein, which can now be defined as a bona fide virulence factor for virulent Schu S4. Understanding how this protein functions at the biochemical level will provide further important insights toward unraveling the enigma of FPI function in Francisella pathogenesis.
ACKNOWLEDGMENTS
We thank Martin Pavelka (University of Rochester Medical Center), Larry Gallagher (University of Washington), and Herbert Schweizer (Colorado State University) for sharing genetic reagents and Nicole Dobbs (University of Texas Southwestern Medical Center) for technical assistance with animal studies.
This work was supported by grant number U54 AI057156 from the National Institute of Allergy and Infectious Diseases (NIAID), NIH.
The contents are solely the responsibility of the authors and do not necessarily represent the official views of the RCE Programs Office, NIAID, or NIH.
FOOTNOTES
- Received 10 May 2013.
- Returned for modification 22 June 2013.
- Accepted 3 August 2013.
- Accepted manuscript posted online 19 August 2013.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/IAI.00595-13.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
