|Home | About | Journals | Submit | Contact Us | Français|
Haemophilus influenzae is an obligate human pathogen that persistently colonizes the nasopharynx and causes disease when it invades the bloodstream, lungs, or middle ear. Proteins that mediate critical interactions with the host during invasive disease are likely to be secreted. Many secreted proteins require addition of disulfide bonds by the DsbA disulfide oxidoreductase for activity or stability. In this study, we evaluated the role in H. influenzae pathogenesis of DsbA, as well as HbpA, a substrate of DsbA. Mutants of H. influenzae Rd and type b strain Eagan having nonpolar deletions of dsbA were attenuated for bacteremia in animal models, and complemented strains exhibited virulence equivalent to that of the parental strains. Comparison of predicted secreted proteins in H. influenzae to known DsbA substrates in other species revealed several proteins that could contribute to the role of dsbA in virulence. One candidate, the heme transport protein, HbpA, was examined because of the importance of exogenous heme for aerobic growth of H. influenzae. The presence of a dsbA-dependent disulfide bond in HbpA was verified by an alkylation protection assay, and HbpA was less abundant in a dsbA mutant. The hbpA mutant exhibited reduced bacteremia in the mouse model, and complementation restored its in vivo phenotype to that of the parental strain. These results indicate that dsbA is required in vivo and that HbpA and additional DsbA-dependent factors are likely to participate in H. influenzae pathogenesis.
Haemophilus influenzae efficiently colonizes the human nasopharyngeal mucosa in a primarily asymptomatic manner, and the carriage frequency is ~80% in healthy adults (48). However, it can disseminate to other anatomical sites and cause otitis media, upper and lower respiratory tract infections, septicemia, and meningitis in children (13, 22, 37, 44, 48-50, 63). The incidence of H. influenzae meningitis has dramatically declined in populations immunized with a vaccine against the type b capsular polysaccharide. The vaccine has not affected the incidence of infection with nontypeable H. influenzae (NTHi) strains, which lack the capsule. NTHi strains predominantly cause respiratory tract infections and otitis media but in rare cases can invade the bloodstream, leading to meningitis. This disease profile raises the possibility that genes promoting intravascular invasion could be present in NTHi strains (15, 18, 53, 54). However, the molecular basis for the invasive properties of H. influenzae that promote transmission from the nasopharynx to the bloodstream or middle ear is not fully understood.
Secreted bacterial proteins mediate critical aspects of pathogenesis, including attachment, nutrient utilization, and subversion of host defenses. Many secreted proteins of gram-negative bacteria acquire disulfide bonds in the periplasm that stabilize their mature, folded structures (9). Formation of such linkages has been most extensively studied in Escherichia coli, in which a series of disulfide oxidoreductases (Dsb) create and exchange disulfide bonds in periplasmic proteins (for reviews, see references 34 and 51). The soluble periplasmic disulfide oxidoreductase, DsbA, directly catalyzes this process by exchanging its disulfide bond with free thiol groups of cysteine residues in target proteins (21, 75). DsbA is efficiently reoxidized by DsbB, a membrane protein that transfers electrons to quinones for subsequent transfer to electron acceptors of the respiratory chain (8, 51). The soluble periplasmic DsbC and DsbG proteins mediate rearrangement of mispaired disulfides using electrons transferred via the membrane-bound DsbD protein from cytoplasmic thioredoxin (3, 10, 38, 58, 64, 76). Mutants defective in periplasmic disulfide bond formation are viable under standard culture conditions but exhibit a range of phenotypes as a result of defective maturation of secreted proteins. The effects vary depending on the repertoire of periplasmic and secreted substrates of DsbA in different bacteria. The deficiencies can involve single enzymes that require a disulfide bond for activity, such as the periplasmic alkaline phosphatase, PhoA, of E. coli, as well as defects in components of transporters, resulting in inappropriate localization of substrates.
DsbA homologs contribute to the pathogenesis of multiple bacterial species, in which they are required for maturation or export of major secreted virulence factors. DsbA activity is required for production of functional type IV pili (also called fimbriae) that mediate adherence to host surfaces in Vibrio cholerae, Neisseria meningitidis, enteropathogenic E. coli, and uropathogenic E. coli (33, 55, 66, 77). Toxin production or secretion is defective in many dsbA mutants; the toxins affected include cholera toxin in Vibrio cholerae, heat-labile and heat-stable E. coli enterotoxins, and pertussis toxin in Bordetella pertussis (55, 65, 74). Type III secretion systems consist of multisubunit protein conduits that inject effector proteins directly from the bacterial cytoplasm into host cells to subvert diverse host cell functions. Components of the type III secretion apparatus are defective in dsbA mutants of Yersinia pestis, Shigella flexneri, Pseudomonas spp., and Salmonella enterica serovar Typhimurium (26, 32, 42, 68). Furthermore, DsbA has been implicated in systemic infection by E. coli K1 and prolonged survival of adherent-invasive E. coli within macrophages (12, 23).
The dsbA gene of H. influenzae (HI0846, also called por) transcomplements a dsbA mutant of E. coli (67). Disruption of dsbA with a transposon insertion resulted in changes in secreted protein localization in a cellular fractionation experiment and dramatically reduced the natural transformation efficiency (67). The role of DsbA in H. influenzae pathogenesis has not been examined. However, a transposon-based “signature-tagged mutagenesis” screen detected the putative dsbB homolog as a virulence gene candidate in an infant rat model of bacteremia, suggesting a potential role for periplasmic disulfide bond formation in H. influenzae pathogenesis (28). In H. influenzae, the protein targets of DsbA and virulence factors dependent on its activity have not been identified. In this study, we demonstrated that dsbA is required for H. influenzae bacteremia caused by both unencapsulated strain Rd and a virulent encapsulated type b strain. Heme uptake is required for aerobic growth of H. influenzae, which cannot synthesize the porphyrin ring (24, 70), and several heme utilization pathways have been implicated in bloodstream infection by H. influenzae (47, 61). We demonstrate that the heme transport protein, HbpA, contains a DsbA-dependent disulfide bond. A nonpolar hbpA deletion mutant caused reduced bacteremia in mice, yet the defect was not as pronounced as that of the dsbA mutant. Based on these results, it is likely that dsbA is required in vivo for production of optimal levels of hbpA and that additional virulence factors that remain to be identified also participate in the critical role of DsbA in H. influenzae pathogenesis.
H. influenzae Rd, a capsule-deficient serotype d derivative (71), and a virulent streptomycin-resistant derivative of H. influenzae type b strain Eagan (Hib) (5) were grown in brain heart infusion broth (BHI) supplemented with 10 μg/ml hemin and 10 μg/ml NAD (sBHI), in MIc, a low-nutrient medium capable of supporting growth of H. influenzae (7), or on sBHI agar plates at 35°C. Development of competence for transformation of H. influenzae was accomplished as previously described (7). For selection of Rd- and Hib-derived strains, antibiotics were used at the following concentrations: 8 μg/ml tetracycline, 20 μg/ml kanamycin, 10 μg/ml gentamicin, and 100 μg/ml streptomycin.
Plasmids and PCR products were constructed using standard molecular biology techniques (6). For complementation of mutants, DNA fragments were amplified by PCR and cloned between adjacent SapI restriction sites of the chromosomal delivery vector pXT10, which does not replicate in H. influenzae (71). The pXT10-based plasmids contained upstream (xylF) and downstream (xylB) homologous regions flanking the SapI cloning sites that allowed precise fusion of genes of interest to the xylose-inducible xylA promoter, as previously described (71). Recombination at the xylose catabolic locus replaced the endogenous xylA gene with the cloned fragment and the tetAR tetracycline resistance cassette. Plasmids were linearized by digestion with PciI and SacI, and tetracycline-resistant (Tetr) recombinants were selected on sBHI agar plates. Double crossovers within xylF and xylB were confirmed by performing PCR with primers specific for sequences outside the inserted recombinant region.
To generate a dsbA mutant and a complemented strain of H. influenzae which requires DsbA for natural transformation, we first generated a strain containing an inducible copy of dsbA and sequentially introduced the dsbA deletion and the complementation construct or the “empty vector” construct into this background. Initially, an additional copy of dsbA under control of the xylose-inducible promoter of xylA was introduced into H. influenzae Rd to create strain RX. The coding sequence of DsbA lacking the translational termination codon was amplified by PCR with primers F-NTdsb (5′-AAAGATCTGCTCTTCAATGAAAAAAGTATTACTTGC-3′) and 3dsbAHA (5′-AAAGATCTGCTCTTCGTAATGCATAATCTGGCACATCATATGGATATTTTTGCAATAAACCTTTTACGGTT-3′), which introduced SapI sites in the termini of the fragment. The resulting fragment was cloned into pXT10 that had been digested previously with SapI. The resulting plasmid, pXyldsbA1.1, was linearized and used to transform H. influenzae to tetracycline resistance to create strain RX.
Next, the native copy of dsbA was deleted from RX by replacement with the aacC1 gentamicin resistance gene to create strain RdsbAX by PCR “stitching” as follows. Overlapping PCR fragments generated with primers representing the 951-bp region immediately 5′ of the dsbA translational start codon (primers 5844H [5′-TTTAAGCTTTTAGATGACTGTTTTCTTTAAATC-3′] and 3Dsbout [5′-TTCTTTCCTCTTATTTAATGATACCGCGAG-3′]), the 569-bp aacC1 gene encoding gentamicin resistance (primers 5GentD [5′-TAAATAAGAGGAAAGAAATGTTACGCAGCAGCAACGATGTT-3′] and 3GentD [5′-CATTAAACCAATTTTTCGTTAGGTGGCGGTACTTGGGTCGAT-3′]), and the 1,641-bp 3′ region starting at the dsbA termination codon (primers 5Dsbout [5′-CGAAAAATTGGTTTAATGCCAGCCC-3′] and 3848H [5′-TTTAAGCTTCTACTTGCGAATGAGCCATAGGC-3′]) were combined by overlap extension PCR with primers 5844H and 3848H to precisely replace the dsbA coding sequence with the coding sequence of aacC1. The resulting 3,126-bp DNA fragment was used to transform strain RX, and gentamicin-resistant (Gmr) recombinants were isolated to create strain RdsbAX, which contained a single copy of dsbA under control of the xylose-inducible xylA promoter.
To complement the dsbA knockout with a wild-type copy of dsbA under control of its own promoter, overlap extension PCR was performed as follows. Primers pXT10thyA-F (5′-AGGGCTTGAATCGCACCTCCA-3′) and 3dsbkan1 (5′-CATCAGAGATTTTGAGACACGGGCCTCTTATTTTTGCAATAAACCTTTTACGGT-3′) were used in PCR to amplify a 1,983-bp fragment containing dsbA from a pXT10-based plasmid carrying dsbA coding sequences, pDsbA1.2. A 2,716-bp PCR product was amplified from a kanamycin-marked derivative of pXT10 with primers 5pkan1 (5′-GAGGCCCGTGTCTCAAAATCTCTGATG-3′) and 3revRfaD1 (5′-AACAGGCTACGATAAACCATTCAAAACAGT-3′). The 1,983- and 2,716-bp fragments were joined via the 27 bp of overlapping sequence by PCR performed with primers pXT10thyA-F and 3revRfaD1, the resultant 4,672-bp PCR product was transformed into strain RdsbAX (grown in the presence of 1 mM d-xylose to induce expression of dsbA), and kanamycin-resistant (Kmr) transformants were isolated to create strain RdsbAC.
To control for effects of modification of the xyl locus, a dsbA mutant containing the integrated “empty vector” sequences was generated by transforming RdsbAX grown in the presence of 1 mM d-xylose with a 4,334-bp PCR product having a precise deletion of the dsbA coding sequences of the 4,672-bp construct described above in RdsbAX, except that primers 3xylF1 (5′-ACGTTTATCAACAGCGATAGGATCAAGT-3′) and 3pDsbAsapKan (5′-CATCAGAGATTTTGAGACACGGGCCTCTTACGAAGAGCGGCGCGCCGCTCTTCCCATTTCTTTCCTCTTATTTAATGATACCGCGA-3′) were used instead of primers pXT10thyA-F and 3dsbAkan. Selection for Kmr transformants resulted in isolation of strain RdsbAV. To construct a strain that contained the “empty vector” in a wild-type background, the same 4,334-bp PCR product was transformed into H. influenzae Rd, and Kmr transformants were isolated to create strain RXV.
Similarly, the same set of constructs was used to generate the dsbA mutant HdsbAV, a vector-only strain (HXV), and a complemented strain (HdsbAC) in the H. influenzae type b strain Eagan background. The wild-type and dsbA mutant phenotypes of all strains were verified using a dithiothreitol (DTT) sensitivity assay (described below), and all mutations were confirmed by sequence analysis of the recombinant loci.
hbpA mutant strain RhbpA was constructed by replacement of the coding sequence of hbpA with the kanamycin resistance gene, aphI. The exchange fragment was synthesized by overlap extension PCR between three regions: a 1,083-bp PCR product containing the 5′ flanking region of hbpA generated using primers 5hbp1 (5′-AGTCATTCACGCCAGTTGGCACTGGAT-3′) and 3hbp1 (5′-TTCCCGTTGAATATGGCTCATACCTCAATGTTAGGCAGGGAATGCCCTA-3′), an 816-bp PCR product containing the coding region for the kanamycin resistance gene generated with primers 5kan1.1 (5′-ATGAGCCATATTCAACGGGAA-3′) and 3kan1.1 (5′-TTAGAAAAACTCATCGAGCATCAAATG-3′), and a 1,020-bp PCR product containing the 3′ flanking region of hbpA generated with primers 5hbp3 (5′-CATTTGATGCTCGATGAGTTTTTCTAATTCATATTGATTTACTTATTTTAAGCCCT-3′) and 3hbp3 (5′-CAAAAGGGGTGAGTATAAATTTACACTCAA-3′). The 1,083-, 1,020, and 816-bp fragments were joined in a PCR via their complementary ends using primers 5hbp1 and 3hbp3. The resulting 2,871-bp fragment was introduced into H. influenzae Rd, and Kmr transformants were selected on sBHI containing kanamycin to create strain RhbpA. To construct an hbpA knockout mutant carrying the integrated empty exchange vector, strain RhbpA was transformed with linearized vector pXT10, and Tetr transformants were isolated to create strain RhbpAV.
To complement the hbpA mutation with a copy of hbpA expressed from the hbpA promoter, a 1,842-bp fragment containing the hbpA coding region and including 142 bp upstream of hbpA was amplified from Rd using primers 5hbpha (5′-AAAGCTCTTCAATGATTAATTTGTTATAATCCATAGA-3′) and 3hbpha (5′-TTTGCTCTTCTTTATGCATAATCTGGCACATCATATGGATATTTACCATCAACACTCACACCATA-3′). This set of primers also added a C-terminal hemagglutinin (HA) epitope tag to hbpA. The PCR product was cloned between the two SapI sites of pXT10 to generate plasmid pXhbp1.5, which was then introduced into strain RhbpA with selection for Tetr to create strain RhbpAC. To introduce a nonpolar, in-frame deletion of dsbA into strain RhbpAC, this strain was transformed with the 3,126-bp dsbA replacement fragment described above, and Gmr transformants were selected to create strain RhbpACΔdsbA.
Strain RdV carrying pXT10 “empty vector” sequences in the xyl locus and strain RdlacZ (H. influenzae Rd carrying lacZ at the xyl locus) were constructed as previously described (72). Strain RdgalU was constructed by replacement of galU with the aphI Kmr cassette. For all mutant strains, replacement of endogenous loci by double-crossover homologous recombination with mutant constructs was confirmed by PCR performed with primers specific for sequences flanking the inserted recombinant region.
To evaluate sensitivity to DTT, strains were inoculated in triplicate using inocula from overnight cultures into 25 ml of sBHI in 50-ml Erlenmeyer flasks to obtain an optical density at 600 nm (OD600) of 0.01 and incubated at 35°C with shaking at 250 rpm. When cultures reached the log phase, they were diluted in sBHI to obtain an OD600 of 0.02, and 100 μl was transferred to a 96-well flat-bottom dish. Each well in the dish was then treated with 100 μl of sBHI containing 10 mM DTT at a final concentration of 5 mM or with sBHI alone (control wells). The dish was then incubated at 35°C for 16 h in a Versamax microplate reader (Molecular Devices, Sunnyvale, CA) set to read the absorbance at 600 nm every 10 min. Sensitivity was assessed using the relative OD600 values at the end of the incubation period.
To determine the sensitivity of the dsbA deletion mutant to H2O2, strains Rd, RXV, RdsbAV, and RdsbAC were inoculated using inocula from overnight cultures in triplicate into 25 ml of sBHI in 50-ml Erlenmeyer flasks or into 5 ml of sBHI in culture tubes to obtain an OD600 of 0.01. The resulting cultures were incubated aerobically at 35°C with shaking at 250 rpm (flasks) or in an anaerobic chamber (culture tubes) with BBL GasPak Plus generators (Becton, Dickinson and Company, Sparks, MD). When cultures reached the log phase, they were diluted in sBHI to obtain an OD600 of 0.02, and 100 μl of each culture was seeded into a 96-well flat-bottom dish. Hydrogen peroxide (Sigma-Aldrich, St. Louis, MO) diluted in 100 μl of sBHI was then added to cultures grown in 25-ml flasks at final concentrations of 0, 62.5, 125, and 250 μM in sBHI and to anaerobically grown cultures at final concentrations of 0, 62.5, 125, and 500 μM. The dishes were then incubated at 35°C for 16 h in a microplate reader, and the absorbance at 600 nm was determined every 10 min to evaluate the growth rates and final culture densities.
To determine the growth rates in rich media and in defined media, strains were inoculated in triplicate to obtain an OD600 of 0.01 by using inocula from standing overnight cultures into 25-ml Erlenmeyer flasks containing 15 ml of sBHI and MIc, respectively. The resulting cultures were incubated at 35°C with shaking at 250 rpm, and aliquots were removed to determine the absorbance at 600 nm every 30 min for 6.5 h. Growth rates were calculated by nonlinear regression analysis.
To evaluate the growth of the dsbA mutant in comparison to the growth of the hbpA mutant under heme limitation conditions, strains RXV, RdsbAV, RhbpAV, and RhbpAC were grown in standing overnight cultures, washed once in sterile Hanks' balanced salt solution (HBSS) (Invitrogen, Carlsbad, CA), and diluted to obtain an OD600 of 0.01 in BHI broth supplemented with NAD and different concentrations (5, 0.5, 0.25, and 0.025 μg/ml) of heme (Sigma-Aldrich, St. Louis, MO) or hemoglobin (Becton, Dickinson and Company) in a 96-well microplate (final volume, 200 μl). The cultures were then incubated at 35°C in the microplate reader, and the absorbance at 600 nm was determined every 10 min for 16 h.
To compare the generation times obtained with different heme concentrations under anaerobic and aerobic conditions, overnight cultures of strains Rd, RdV, RhbpAV, and RhbpAC (pelleted and resuspended in HBSS) were used to inoculate 10 ml of BHI containing different concentrations of free heme (10, 0.5, 0.05, and 0 μg/ml). The cultures were then aliquoted into the wells of 11 96-well flat-bottom dishes. One dish was incubated at 35°C for 14 h in the microplate reader, and the absorbance at 600 nm was determined every 10 min (no aerobic growth was detected in wells not supplemented with heme). The other 10 dishes were sealed in individual BD GasPak EZ Anaerobe gas-generating pouches (Becton, Dickinson and Company) and incubated at 35°C. Dishes were removed from the pouches at appropriate intervals, and the absorbance at 600 nm was recorded. Growth rates were determined as described above.
Cultures were grown in triplicate as described above for the DTT sensitivity assay, and competent cells were prepared from these cultures as previously described (7). The competence of mutant and parental strains was measured by assessing the transformation frequencies with chromosomal DNA from a streptomycin-resistant (Smr) H. influenzae strain (1 μg) and selection on sBHI agar plates containing 100 μg/ml streptomycin. Transformation efficiencies were calculated by dividing the number of Smr colonies by the number of colonies on sBHI agar plates without antibiotic. Transformation frequencies were normalized by log10 transformation and analyzed with Prism 4.0c (GraphPad Software, San Diego, CA) using analysis of variance (ANOVA) with Bonferroni's multiple-comparison test to evaluate differences in frequency between RdsbAV and all other strains.
Standing overnight cultures of strains having an OD600 of 0.01 were inoculated into 10 ml of sBHI in culture tubes. The resulting cultures were incubated in an anaerobic chamber with shaking at 120 rpm and 35°C for 5 h, conditions that were permissive for growth of the hbpA mutant. For coinfection, each experimental strain was mixed with the RdlacZ reference strain at a 1:1 ratio. Prior to inoculation, bacteria were washed and diluted in HBSS to obtain a final concentration of 2 × 109 bacteria per ml. Female 6.5-week-old C57BL/6J mice (four or five mice per strain; The Jackson Laboratory, Bar Harbor, ME) were inoculated by intraperitoneal (i.p.) injection of 200 μl of a bacterial suspension. Twenty-four hours after inoculation, 5 μl of blood was recovered aseptically from each mouse via tail bleeding. The blood was diluted into BHI broth, plated on sBHI agar plates for single-strain infections or on sBHI agar plates containing S-Gal (3,4-cyclohexenoesculetin β-d-galactopyranoside; Sigma-Aldrich) for coinfections, and incubated overnight in an anaerobic chamber at 35°C to determine the number of CFU. For statistical analysis, the numbers of CFU/ml for single-strain infections were normalized by log10 transformation for ANOVA using Prism 4.0c. The coinfection CFU data were log10 transformed, and the ratio of each experimental strain to RdlacZ was calculated and analyzed using Prism 4.0c. Comparisons of two data sets were performed using the t test, and comparisons of more than two data sets were performed using ANOVA with Bonferroni's multiple-comparison test. All procedures with animals were conducted in accordance with NIH guidelines and with prior approval by the University of Massachusetts Medical School Institutional Animal Care and Use Committee.
H. influenzae type b-derived strains were inoculated using inocula from standing overnight cultures having an OD600 of 0.01 into 50 ml of sBHI in 50-ml Erlenmeyer flasks. Cultures were incubated with shaking at 120 rpm at 35°C to obtain an OD600 of 0.4. Cells were washed once and diluted in sterile HBSS to obtain a final concentration of 2 × 103 bacteria per ml. Five-day-old Sprague-Dawley rat pups (Charles River Laboratories, Boston, MA) were inoculated i.p. with 100 μl of strains HXV (n = 11) and HdsbAV (n = 11) or with HdsbAC (n = 12). Infants inoculated i.p. with each strain were returned to their mothers, and each group was housed separately. Blood (5 μl) was collected aseptically via tail bleeding at 12, 36, and 120 h postinoculation, diluted into BHI, and plated on sBHI agar plates for to determine the number of CFU as described above. For statistical analysis, ANOVA with Bonferroni's multiple-comparison test was used as described above.
For analysis of HbpA, strains were inoculated using inocula from standing overnight cultures into duplicate 50-ml sBHI cultures in 50-ml flasks to obtain a starting density of 0.01 OD600 and were incubated at 35°C with shaking at 250 rpm. When cultures reached the log phase, 1 ml was removed and pelleted by centrifugation (18,000 × g for 5 min) for immunoblot analysis, and the remaining culture was used for RNA isolation as described below. After removal of the supernatant, the pellets were normalized by resuspension in an appropriate volume of HBSS. Cells (0.3 OD600 equivalents per lane) were then boiled in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer, and proteins were separated by 8% SDS-PAGE, followed by electrotransfer to Immobilon-P (Millipore, Billerica, MA). HbpA-HA was visualized by Western blotting using the primary antibody anti-HA1.1 (1:1000; Covance, Berkeley, CA) and the secondary antibody goat anti-mouse immunoglobulin G-horseradish peroxidase conjugate (1:5,000; Upstate, Lake Placid, NY). Equal sample concentrations were verified by Coomassie blue staining. HbpA-HA was quantified by generating a 10% dilution series of each protein sample and resolving proteins by 8% SDS-PAGE. HbpA-HA was then visualized by Western blotting as described above. HbpA levels were quantified by densitometry using ImageJ (National Institutes of Health, Bethesda, MD).
To quantify hbpA mRNA, we isolated total RNA in parallel from the 50-ml cultures that were used for the HbpA Western blot analysis using the TRIzol reagent (Invitrogen, Carlsbad, CA). RNA was then treated with DNase I (Ambion, Austin, TX), extracted with acid phenol, extracted with chloroform, and concentrated by ethanol precipitation. The RNA samples (total amount, 5 μg) were used as templates for cDNA synthesis with random primers (New England Biolabs, Beverley, MA) and SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA). Quantitative reverse transcription PCR (qRT-PCR) was performed with iQ SYBR green Supermix (Bio-Rad Laboratories, Hercules, CA), and fluorescence was measured using the DNA Engine Opticon II system (MJ Research, Waltham, MA). One-tenth of each cDNA reaction mixture was used as a template for qRT-PCR performed with primers 5′hbpART (5′-ATGATTAATTTGTTATAATCCATAGA-3′) and 3′hbpART (5′-CAAGCTGCCAAAACAAGAGT-3′), which amplified the first 200 bp of hbpA. Primers RpoA5′ (5′-GTAGAAATTGATGGCGTATTG-3′) and RpoA3′ (5′-TCACCATCATAGGTAATGTCC-3′) were used to amplify the RNA polymerase alpha subunit gene, rpoA, as an internal reference. The real-time cycler conditions used have been described previously (72).
Western blotting for assessment of binding of complement C3 and C4 activation products was performed as previously described (19, 56). Briefly, cultures of strains RXV, RdsbAV, and RdsbAC were grown as described above for HbpA Western blotting and then washed and suspended in HBSS containing 0.15 mM CaCl2 and 1 mM MgCl2 (final reaction mixture volume, 0.5 ml). Normal human serum (NHS) pooled from 12 healthy individuals was added to a final concentration of 2% and incubated for 30 min at 37°C, which was followed by differential treatment with 1 M methylamine (pH 11), which dissociates complement ester linkages but not amide-linked complement from target structures (19, 56). Bacteria were lysed in 1× SDS-PAGE sample buffer and analyzed by immunoblotting using primary antibodies to human C3 (Sigma-Aldrich, St. Louis, MO) and C4 (Biodesign, Saco, ME) and alkaline phosphatase-conjugated secondary anti-human antibodies as described previously (19). No differences in the binding profiles of the strains to C3 or C4 subunits with and without methylamine treatment were observed.
The sensitivity of dsbA mutants to serum was determined as previously described (57). Briefly, triplicate cultures of strains RXV, RdsbAV, RdsbAC, and RdgalU were grown as described above for the DTT sensitivity assay. At log phase, 2,000 CFU from each culture was diluted in HBSS and incubated at 37°C for 30 min with or without 2% (final concentration) NHS in a 150-μl reaction mixture. To determine the number of CFU, 15 μl was plated on sBHI agar. Bacteria were also incubated in parallel with serum that had been previously inactivated by incubation at 56°C for 30 min.
Ten optical density units of cells grown as described above for anti-HA immunoblotting was harvested at log phase by centrifugation at 5,000 × g for 5 min. Before thiol modification of periplasmic proteins, the outer membrane was disrupted using the methods described in a PeriPreps periplasting kit (Epicenter, Madison, WI). Briefly, the cell pellets were resuspended in 2 ml of 200 mM Tris (pH 7.4), 1 mM EDTA, 20% sucrose, and 30 U of lysozyme (Sigma-Aldrich, St. Louis, MO) and incubated at room temperature for 5 min. After incubation, 3 ml of cold water was added, which was followed by 10 min of incubation on ice. Each 5-ml preparation was then divided in half; one half was treated with 5 mM EZ-Link maleimide-(ethylene oxide)2-biotin (MPB) (which added 525.23 Da per bond) (Pierce, Rockford, IL), and the other half was not treated. After incubation for 50 min at room temperature, the resulting spheroplasts and associated membranes were collected by centrifugation at 4,000 × g for 15 min and resuspended in 375 μl of Peripreps lysis buffer (10 mM Tris-HCl [pH 7.5], 50 mM KCl, 1 mM EDTA, 0.1% deoxycholate). After lysis, equivalent 0.30 OD600 of each sample was boiled for 5 min in SDS loading buffer, and proteins were separated by nonreducing 8% SDS-PAGE. HbpA-HA was then visualized by Western blotting as described above. The apparent levels of HbpA-HA in the spheroplasts were similar to those in whole-cell lysates of the same number of cells (data not shown), suggesting that HbpA-HA is localized primarily in this fraction, which is consistent with membrane localization of the predicted HbpA lipoprotein.
A series of strains were constructed to evaluate the potential role of DsbA in H. influenzae pathogenesis (Table (Table1).1). We first verified that the dsbA mutant exhibited the DTT sensitivity phenotype previously seen with dsbA mutants of other species (43), and we determined its growth properties. Wild-type parent strain Rd, Rd carrying the “empty vector” (RXV), a dsbA deletion mutant carrying the “empty vector” (RdsbAV), and the complemented strain (RdsbAC) were evaluated to determine their growth under a range of conditions and to determine defects in DTT resistance and transformation. The generation times under aerobic conditions in rich medium (sBHI) for RXV, RdsbAV, and RdsbAC were 32 ± 2, 32 ± 3, and 36 ± 4 min, respectively, and the generation times in defined medium (MIc) were 48 ± 4, 41 ± 2, and 38 ± 3 min, respectively. Similarly, the growth yields of these strains after 6.5 h of anaerobic growth in sBHI were indistinguishable. The growth yields of all DsbA+ strains (Rd, RXV, and RdsbAC) were equivalent after 16 h in the presence of 5 mM DTT, with final average densities of ~0.5 OD600, whereas the growth of the DsbA− strain, RdsbAV, was dramatically attenuated under these conditions and the density did not exceed 0.1 OD600, similar to results obtained with dsbA mutants of E. coli (43). Strain RdsbAX, which contains a d-xylose-inducible copy of dsbA and was used to construct strains RdsbAV and RdsbAC, was resistant to DTT in the presence of 1 mM d-xylose and sensitive to DTT in the absence of d-xylose.
H. influenzae dsbA (por) was previously implicated in natural transformation. Therefore, we evaluated the transformation efficiencies of our strains using H. influenzae DNA carrying a streptomycin resistance allele. The transformation efficiencies relative to the wild-type parental strain Rd for strains RXV, RdsbAV, and RdsbAC, were 1.12, 1.33 × 10−6, and 1.04, respectively, and the 6-log-lower transformation frequency of the dsbA mutant (RdsbAV) relative to the other strains was statistically significant (P < 0.0001). Therefore, previously reported phenotypes associated with dsbA mutants were observed with our in-frame H. influenzae dsbA deletion mutants, and complemented strains exhibited the wild-type phenotypes.
The strains generated as described above provided a well-defined set of mutants for investigation of the role of dsbA during infection. Although not a recent clinical isolate, Rd has virulence properties similar to those of clinically important NTHi strains in several models of infection and has provided a useful system for studies of H. influenzae biology and pathogenesis (16, 40). The mouse model was used to evaluate bloodstream survival of the dsbA mutant, RdsbAV, compared to that of the vector-only control strain, RXV. At 24 h postinoculation, 48-fold-fewer bacterial CFU were recovered from mice inoculated with the dsbA mutant than from mice inoculated with the control strain (Fig. (Fig.1A).1A). The level of bacteria recovered from most of the mice inoculated with RdsbAV was close to the limit of detection. An additional experiment was conducted to confirm this result with a complemented strain, RdsbAC. To further assess the level of attenuation, this experiment was performed as a competition between each strain and strain RdlacZ, which expresses E. coli lacZ at the xyl locus, using mixed infections. Consistent with results obtained from single-strain inoculations, the competitive index of RdsbAV was 100- to 170-fold less than that of Rd, RXV, or RdsbAC (Fig. (Fig.1B).1B). Therefore, infection with a mutant containing a nonpolar dsbA deletion resulted in reduced levels of bacteremia in mice, and complementation verified that this effect was specific to the dsbA mutation.
H. influenzae lacks previously implicated dsbA-dependent virulence factors found in other species, including exotoxins and type III secretion structures. Production of the type IV pilus found in some NTHi strains is likely to require dsbA; however, the pilus gene cluster is absent in Rd (20). Therefore, we examined several major virulence-associated phenotypes of H. influenzae to determine whether a defect in a known pathogenic mechanism could account for the survival defect of the dsbA mutant in vivo.
Resistance to oxidative stress generated by hydrogen peroxide exposure has been correlated with H. influenzae pathogenesis in several studies (17, 72). Therefore, we addressed the possibility that loss of DsbA confers sensitivity to this oxidant. After exposure to either anaerobic or aerobic pregrowth conditions, the mutant and the wild type exhibited equal levels of growth inhibition during exposure to hydrogen peroxide at a range of doses (data not shown).
Multiple structures of the lipooligosaccharide (LOS) outer core have been implicated in animal models of H. influenzae bacteremia (25, 31, 60), and resistance to complement has emerged as an important virulence mechanism mediated by these structures (19, 30). Therefore, we investigated whether the dsbA mutant exhibits major LOS structural alterations or increased susceptibility to killing by serum complement. We detected no apparent differences in LOS mobility on SDS-PAGE gels for the dsbA mutant and the wild type. Wild-type, dsbA mutant, and complemented strains were compared using a serum bactericidal assay with 2% pooled NHS. The levels of survival for strains RdV, RdsbAV, and RdsbAC were 12, 2.9, and 11.6%, respectively (P < 0.05). For comparison, a galU mutant deficient in synthesis of the LOS outer core, a structure predicted to be essential for complement resistance, was tested in parallel and exhibited 0% survival. No killing of H. influenzae was observed with heat-inactivated serum, consistent with an essential role for complement in this assay. Differences in levels of complement binding to strains RdV, RdsbAV, and RdsbAC were not detected, as assessed on anti-C3 and anti-C4 immunoblots containing lysates of cells that had been incubated with 2% pooled NHS (data not shown), although it is possible that a small difference in C3 or C4 binding not detected by immunoblotting could have mediated the moderate increase in serum sensitivity observed for the mutant.
We concluded that the hydrogen peroxide resistance and LOS production of the dsbA mutant were not markedly impaired under the conditions tested. An effect on serum resistance that could play a role was observed. However, this effect was moderate, and it seems likely that DsbA influences additional factors required for virulence. To address this hypothesis, we sought the identities of potential DsbA substrates in H. influenzae. Proteins containing DsbA-dependent disulfide bonds have been identified in E. coli (29, 35, 39). These proteins were compared by BLASTP (1) (http://www.ncbi.nlm.nih.gov) to the predicted proteins in the H. influenzae genome to derive a list of potential DsbA targets in H. influenzae (Table (Table2).2). The H. influenzae proteins identified by this search include a predicted periplasmic lipoprotein, HbpA, which is required for utilization of multiple heme sources (27, 45). Multiple systems participate in scavenging heme from sources in the host that include heme-hemopexin, hemoglobin, hemoglobin-haptoglobin, heme-albumin, and free heme (14, 46). HbpA appears to be required for scavenging low levels of heme regardless of the source or carrier protein, suggesting that it could be critical for growth in vivo. The link between DsbA and HbpA suggested a potential mechanism of attenuation of the dsbA mutant in the mouse model.
Based on comparison to the crystal structure of the highly related E. coli DppA protein, which has an intramolecular disulfide bond (52), HbpA has a predicted disulfide bond between cysteine residues Cys27 and Cys255 of the mature protein. A third cysteine located at the N terminus constitutes the predicted lipoprotein acylation site. Many proteins containing DsbA-dependent disulfide bonds are less stable in DsbA-deficient cells. Therefore, we examined the effect of the dsbA deletion mutation on levels of HbpA. To address this question, we developed a functional derivative of HbpA fused to an epitope tag from the influenza virus HA (HbpA-HA). We first constructed a nonpolar hbpA deletion mutant (RhbpAV) of H. influenzae. The mutant was defective for aerobic growth on medium containing low levels of heme, as previously reported for an independently derived hbpA insertional mutant (45). Furthermore, the hbpA mutant exhibited anaerobic growth equivalent to that of wild-type strain Rd and the isogenic “vector-only” strain (RdV), regardless of heme availability (Table (Table3).3). When expressed in the hbpA mutant background, HbpA-HA fully complemented the mutant for aerobic growth at all heme concentrations tested (strain RhbpAC), suggesting that the epitope tag does not impair its function (Table (Table33).
The resulting strains were used to assess the effect of a dsbA mutation on levels of HbpA-HA on Western blots, and transcript levels were assessed in parallel by qRT-PCR. The levels of HbpA-HA in the dsbA deletion mutant RhbpACΔdsbA were approximately 50% of the levels detected in the DsbA+ control strain, RhbpAC, as determined by densitometry (Fig. (Fig.2A).2A). qRT-PCR detected no differences in the levels of hbpA-specific transcripts in these cultures (data not shown). Together, these results suggest that the effect of DsbA on HbpA abundance is mediated at a posttranscriptional level, consistent with its role as a disulfide oxidoreductase.
To more directly assess the role of DsbA in formation of disulfide bonds in HbpA, the HbpA-HA protein was analyzed by nonreducing SDS-PAGE after isolation from the DsbA− and DsbA+ H. influenzae strains (Fig. (Fig.2B).2B). Whereas HbpA-HA from DsbA+ cells appeared as a single band, samples from DsbA− cells yielded an additional HbpA-HA band with lower electrophoretic mobility. Treatment of cells prior to protein isolation with a thiol reactive ligand, MPB, resulted in no change in HbpA-HA in the parental strain (Fig. (Fig.2B,2B, lanes 1 and 2), as expected if the two nonacylated cysteine residues were in the oxidized state as a disulfide bond. In contrast, the more slowly migrating species in the dsbA mutant (Fig. (Fig.2B,2B, lane 3) exhibited an additional decrease in mobility in samples from MPB-treated cells (Fig. (Fig.2B,2B, lane 4), consistent with addition of MPB to free thiols on cysteine residues of this protein. Relatively low levels of the reduced form of HbpA were detected, consistent with decreased stability of the reduced form relative to the oxidized form in the dsbA mutant, a characteristic property of many DsbA-dependent proteins (9). Therefore, a longer exposure time was used to clearly visualize the reduced form in Fig. Fig.2B,2B, masking the decrease in total HbpA levels that was detected in the dsbA mutant in the quantitative studies described above (Fig. (Fig.2A2A).
HbpA in the oxidized form was detected in the dsbA mutant, and it is likely that some HbpA activity was retained in this mutant. Consistent with this observation, we could not detect a growth defect of the dsbA mutant on low-heme media. The growth rates of DsbA+ and DsbA− strains (RXV and RdsbAV) were compared to those of the hbpA mutant and complemented strains (RhbpAV and RhbpAC) with 5, 0.5, 0.25, and 0.025 μg/ml of either heme or heme-hemoglobin (data not shown). RhbpAV exhibited progressively lower growth rates as the heme or hemoglobin concentration was decreased, and RhbpAC grew at the same rates as the wild type, similar to results shown in Table Table3.3. Conversely, no differences between RXV and RdsbAV were observed, suggesting that the residual levels of active HbpA in the dsbA mutant were sufficient for acquisition of these heme sources in vitro. Together, these data indicate that the DsbA disulfide oxidoreductase is required to maintain the complete oxidation of free thiols on HbpA and for wild-type levels of this protein in H. influenzae.
Heme is required for aerobic growth and is obtained by H. influenzae from sources within the host. The decreased levels of HbpA observed in the dsbA mutant could have contributed to decreased survival of this strain in the bloodstream by interfering with heme acquisition in vivo, where heme is efficiently sequestered by multiple systems of the host. To evaluate this hypothesis, we assessed the role of hbpA in the mouse model using the hbpA mutant RhbpAV, the isogenic HpbA+ parent strain RdV, and the complemented strain RhbpAC. Inocula were prepared from cultures grown anaerobically, conditions which were permissive for growth of the hbpA mutant (Table (Table3),3), and mice were inoculated by the i.p. route (Fig. (Fig.3).3). In single-strain infections there was a decrease in the number of bacterial CFU recovered from mice inoculated with RhbpAV compared to mice inoculated with RdV (17-fold) or RhbpAC, (~60-fold); however, the trend was not statistically significant (Fig. (Fig.3A).3A). To control for variation between animals, we repeated the experiment using the competition format. Each strain was coinoculated with an equal number of cells of strain RdlacZ, and competitive indices were evaluated. The mutant exhibited a ~27-fold defect in competition relative to the “vector-only” and complemented strains, and the differences were statistically significant (Fig. (Fig.3B).3B). We concluded that survival of the hbpA mutant is attenuated in the bacteremia model, but to a lesser extent than survival of the dsbA mutant. Therefore, a decreased level of HbpA could contribute to the defect in dsbA mutants during infection, yet additional factors, such as serum sensitivity and other mechanisms that remain to be identified, are likely involved.
We next addressed whether dsbA is required during infection by the highly virulent organism H. influenzae type b strain Eagan. The infant rat bacteremia model provides a well-characterized system for examining factors required for H. influenzae type b pathogenesis. Therefore, the mutations used to evaluate the role of dsbA in Rd were moved into the Hib background. Infant rats that were 5 days old were inoculated i.p. with wild-type “vector-only” strain HXV, dsbA mutant HdsbAV, and complemented strain HdsbAC, and bloodstream infection was monitored at 12, 36, and 120 h postinoculation (Fig. (Fig.4).4). At all sampling times, the number of H. influenzae CFU recovered from animals inoculated with the dsbA mutant was at least 100-fold lower than the number of H. influenzae CFU recovered from animals inoculated with the parental or complemented strain, and the level of attenuation was statistically significant in all cases. Furthermore, by 120 h postinoculation only 2 of 11 animals inoculated with the dsbA mutant had detectable bacteremia, whereas most of the animals infected with the wild-type strain (9/11) or the complemented strain (11/11) remained infected, with mean bacterial levels of 9.5 × 104 and 5.6 × 104 CFU/ml, respectively. These results indicate that dsbA is required for efficient production and persistence of a high level of bacteremia in the infant rat model with a virulent clinical isolate of H. influenzae type b.
We report a role for the H. influenzae disulfide oxidoreductase, DsbA, in bloodstream infection. Nonpolar dsbA deletion mutations in either the strain Rd or H. influenzae type b strain Eagan background resulted in equal levels of attenuation in animal models, and the virulence of complemented strains was equivalent to that of the parental strains. Because the in vivo defect was observed with dsbA mutants of both nonencapsulated strain Rd and an encapsulated H. influenzae type b strain, an effect on production of capsule would be unlikely to account for these observations. Therefore, we investigated several other potential mechanisms. The primary set of factors implicated in pathogenesis of nonencapsulated H. influenzae in animal models includes genes involved in LOS synthesis, evasion of complement deposition, and oxidative stress resistance. We detected no apparent role for dsbA in LOS synthesis or hydrogen peroxide resistance, although we cannot exclude the possibility that these phenotypes are influenced in a subtle way that our assays were not sufficiently sensitive to detect. A decrease in serum resistance was observed in the dsbA mutant, and it will be of interest to establish the mechanism by which DsbA contributes to this virulence-related trait. However, the effect on serum resistance was only moderate and does not seem to be sufficient to account for the full defect of the dsbA mutant in pathogenesis. The results suggest that an unrecognized factor(s) may account for the observed virulence defect of the dsbA mutants.
To expand our search to other factors that could participate in the defect of the dsbA mutant in vivo, we considered a set of potential secreted substrates of H. influenzae DsbA identified by comparison of amino acid sequences to reported DsbA targets in other species. The resulting list of potential DsbA substrates in H. influenzae includes a number of known or suspected nutrient transport proteins. Therefore, a nutritional deficiency could contribute to the defect of the dsbA mutant in the blood. We examined growth of the H. influenzae dsbA mutant under a range of in vitro conditions. The H. influenzae dsbA mutant grew normally under aerobic and anaerobic conditions and in a low-nutrient medium. The only in vitro condition under which we could detect a growth defect for the dsbA mutant involved the presence of a high concentration (5 mM) of DTT, a condition that H. influenzae is unlikely to encounter in vivo. Reducing agents are present in plasma and include glutathione, which occurs as a mixture of reduced and oxidized forms with a total concentration estimated to be ~5 to 30 μM (4). Growth in the presence of glutathione was tested using concentrations ranging from 0.005 to 1 mM, and growth of the dsbA mutant and growth of the parental strain were equivalent under each of these conditions (data not shown). A general growth defect or sensitivity to physiological levels of reducing agents does not appear to account for the decreased virulence of our dsbA deletion mutants. If the effect of dsbA on pathogenesis involves a defect in nutrient uptake or utilization, then it is likely to involve a nutrient that is selectively limiting in H. influenzae's environment within the host.
One essential factor that H. influenzae cannot synthesize and must obtain from the host is the porphyrin ring of heme. The results of an amino acid sequence comparison of H. influenzae proteins with known or probable DsbA substrates in other species identified the H. influenzae heme binding protein, HbpA, as a potential substrate of DsbA. Deletion of hbpA results in an aerobic growth defect on media containing low levels of exogenous heme sources (27, 45) and normal growth under anaerobic conditions (Table (Table3).3). We identified the presence of a DsbA-dependent disulfide bond in HbpA and decreased abundance of HbpA in the dsbA mutant. Therefore, we evaluated the phenotype of an hbpA mutant during infection. The hbpA mutant exhibited a defect in the murine bacteremia model, although it was not as pronounced as the defect of the dsbA mutant. Complementation restored the ability of the hbpA mutant to cause bacteremia at a level similar to that observed with the parental strain. These results suggest that decreased levels of HbpA could contribute to the in vivo defect of the dsbA mutants.
The decreased level of HbpA in the dsbA mutant would be expected to influence growth under heme limitation conditions; however, we were unable to detect such an effect in vitro. It is likely that residual HbpA activity in the dsbA mutant is capable of supporting in vitro growth with low levels of heme. Nevertheless, both DsbA and HbpA participate in bloodstream infection. For survival in vivo, where diverse host factors efficiently sequester free heme, there may be a more stringent requirement for wild-type levels of HbpA than there were the in vitro conditions tested here. Alternatively, it is possible that the in vivo growth defect of the dsbA mutant resulted from effects on a DsbA-dependent protein whose role in virulence remains to be identified. The partial attenuation of the hbpA mutant compared to the more dramatic virulence defect of the dsbA mutant supports the hypothesis that other factors are involved. In this regard, two potential DsbA targets in H. influenzae, Pzp1 (ZnuA) and outer membrane protein P5 (Table (Table2),2), are required for growth under zinc-limiting conditions in vitro (41) and adhesion to mucosal epithelium during colonization of the chinchilla nasopharynx (11), respectively. In addition, homologs of Pzp1 and P5 in several other bacterial species have been implicated in pathogenesis (2, 36, 59, 62, 69, 73). We did not observe a requirement for zinc supplementation for in vitro growth of the dsbA mutant (data not shown), potentially due to residual activity of Pzp1, yet is possible that the zinc levels available to bacteria within the mammalian host are lower than those in vitro. A defect in the level or activity of Pzp1 in the H. influenzae dsbA mutant could contribute to its virulence defect, and additional studies are required to evaluate this hypothesis. In addition, a role for P5 during H. influenzae bacteremia has not been reported, and it will be interesting to investigate this possibility. Related outer membrane proteins in other pathogens have been implicated in diverse aspects of pathogenesis, including complement resistance (59, 62, 69), and changes in the outer membrane protein profile of the dsbA mutant could account for similar effects in H. influenzae. The roles of the other potential DsbA substrates in H. influenzae (Table (Table2)2) have not been defined, and their putative homologs in E. coli do not appear to mediate virulence-related functions. Furthermore, the complete set of H. influenzae DsbA substrates remains to be determined experimentally. Investigation of the virulence properties conferred by proteins that contain DsbA-dependent disulfide bonds in H. influenzae will likely uncover important aspects of the pathogenesis of this bacterium and may lead to novel approaches to treatment or prevention of invasive disease.
We thank Sanjay Ram for assistance and advice and Jeffrey Gawronski for helpful comments.
This work was supported in part by a grant from the American Heart Association and by NIH NIAID grant 1RO1-AI49437 to B.J.A.
Editor: J. B. Bliska
Published ahead of print on 22 January 2008.