INTRODUCTION
Streptococcus pneumoniae is a Gram-positive, catalase-negative bacterium which is classified as a facultative anaerobe. It colonizes the nasopharynx as do other pathogens such as
Moraxella catarrhalis,
Haemophilus influenzae, and
Neisseria meningitides. S. pneumoniae can spread from the upper respiratory tract to other parts of the human body, which leads to serious disease, such as pneumonia, otitis media, meningitis, and bacteremia. People susceptible to pneumococcal disease are young children, the elderly, and immunocompromised patients, especially in developing countries. For instance, pneumococcal septicemia is a major cause of infant death in developing countries. It causes about 25% of all preventable deaths in children under the age of 5 and more than 1.2 million infant deaths per year (
4,
24).
Reactive oxygen species (ROS), such as superoxide anion (O
2·−), hydrogen peroxide (H
2O
2), and hydroxyl radicals (OH
•), are a consequence of the use of oxygen (O
2) in fundamental enzymatic reactions and biochemical pathways. Accumulation of ROS has harmful effects on all living organisms; for instance, O
2·−and H
2O
2 damage proteins through oxidation, whereas OH
• targets DNA, leading to lesions and mutations, which also may be fatal (
22,
23,
63).
S. pneumoniae encounters external and internal oxidative stress during its life cycle. Major exogenous sources of ROS for
S. pneumoniae are neutrophils, macrophages, and other lactic acid bacteria in the nasopharynx. Neutrophils and macrophages release a diversity of ROS such as H
2O
2, OH
•, and O
2·−through the oxidative burst. Hydrogen peroxide is moderately stable and in the presence of metal ions can be converted via the Fenton reaction to OH
•, which is the only ROS that can directly harm most molecules and can be involved in the production of other ROS (
23).
In addition to these exogenous sources, the enzyme pyruvate oxidase (SpxB) internally generates significant amounts of hydrogen peroxide up to 2 mM (
46,
58), which may also be produced under
in vivo-like conditions such as biofilms (
31). This concentration is higher than that generated by many other species (
46,
48) and is sufficient to kill or inhibit other nasopharyngeal flora members such as
H. influenzae and
N. meningitides (
47) although it does not seem to interfere with
Staphylococcus aureus colonization (
33). Furthermore, these amounts of H
2O
2 have a cytotoxic effect both
in vivo and
in vitro on human epithelial (
12,
17) and endothelial (
6) cells. The action of SpxB is thought to play a central role in metabolism (
48) and influences competence via an as yet unknown mechanism (
2). The gene itself is regulated by SpxR (
52) and plays an important role in the virulence of pneumococci (
41,
53,
58). Thus, a major source of both endogenous and external ROS for
S. pneumoniae in the nasopharynx is the action of SpxB, and the levels of H
2O
2 produced are sufficient for the generation of OH
• via the Fenton reaction in the absence of exogenous H
2O
2 (
48,
49).
S. pneumoniae lacks canonical proteins identified in other bacteria that protect against oxidative stress, such as catalase, and homologues of global response regulators, such as OxyR, PerR, and SoxRS (
5,
48), and seems to lack an adaptive response to oxidative stress (
48). Interestingly, SpxB also plays an important role in resistance to H
2O
2 (
48). Other proteins implicated in the defense against oxidative stress in
S. pneumoniae are the manganese-dependent superoxide dismutase (SodA) (
62), an AhpD homologue (
44), the putative transcriptional regulator Rgg (
5), the pneumococcal manganese transporter PsaBCA that includes the pneumococcal surface antigen A (PsaA) (
36,
61), the serine protease HtrA (
21), the MerR/NmlR family transcription factor (
51), and the ClpP protease (
42,
54). Additionally the PsaD protein has been suggested to be a thiol peroxidase (
39). However, how
S. pneumoniae protects itself against oxidative stress, in particular of extracellular origin, is still not entirely clear.
To elucidate additional mechanisms, we performed a bioinformatic analysis of the genomes of
S. pneumoniae,
N. meningitides, and
H. influenzae, which identified a hypothetical open reading frame (ORF) belonging to the thiol-specific antioxidant (TlpA/TSA) branch of the thioredoxin super family (
32). Thioredoxins and thioredoxin-like proteins can protect bacteria against oxidative stress (
23). Therefore, we studied the role of this protein in oxidative stress survival. We demonstrated that the ORF and the operon in which it is located play an important role in the protection of
S. pneumoniae against external peroxide stress and the establishment of long-term infection and disease.
MATERIALS AND METHODS
Bacterial strains and growth conditions.
All strains used in the present study are listed in
Table 1 and were stored in 15% glycerol at −80°C.
S. pneumoniae was routinely grown as standing cultures in brain heart infusion (BHI) broth (Oxoid) at 37°C,
Lactococcus lactis NZ9000 was grown standing in M17 broth (Oxoid) (
60) containing 0.5% glucose (GM17) at 30°C, and
Escherichia coli was grown in LB broth in shaken cultures at 37°C.
L. lactis and
S. pneumoniae were grown, respectively, on GM17 and GM17 agar plates with 3% defibrinated sheep blood (Oxoid), whereas
E. coli was cultivated on LB agar. When necessary, chloramphenicol, spectinomycin, and trimethoprim were added to the growth medium in the following concentrations: 2.5, 150, and 15 μg/ml, respectively, for
S. pneumoniae; 4 μg/ml chloramphenicol for
L. lactis; and 20 μg/ml kanamycin for
E. coli. For
in vitro complementation analyses, all strains were exposed to nisin (3 ng/ml) during early log phase and at least 2 h before the application of oxidative stress.
Bioinformatic comparison of the genomes of respiratory pathogens.
The sequences of the predicted proteins in the genome of
S. pneumoniae R6 derived from the NCBI GenBank database (
http://www.ncbi.nlm.nih.gov/GenBank/) were sequentially compared to the predicted protein sequences in the genomes of
S. pneumoniae TIGR4,
Neisseria meningitidis MC58,
Neisseria meningitidis Z2491, and
Haemophilus influenzae Rd KW20 using BlastP. Proteins were considered homologous when the E value was <10
−13. Subsequently, the list of R6 proteins with a homologue in the other genomes was compared to the sequences of the predicted proteins in the genome of
L. lactis IL1403 using BlastP, and a protein was considered nonhomologous when the E value was >10
−19.
Generation of mutants in S. pneumoniae.
Strain R6
nisRK was generated by the introduction of the
nisRK genes into the
bgaA locus of R6 as described by Kloosterman et al. (
27). Deletion of
spd0572 was accomplished using primers SPD571for/SPD571rev and SPD573for/SPD573rev using plasmid pOri28 as described by Kloosterman et al. (
26). The
spd0571-0573 genes were deleted from the D39 genome by allelic replacement mutagenesis (
45). Briefly, primers SPD570for/571rev and 574for/574rev (
Table 2), respectively, were used to generate 5′ and 3′ flanking fragments of
tlpA and the operon
spd0571-0573, respectively. These fragments were fused to a spectinomycin resistance gene amplified with primers specfor and specrev (
3). Subsequently the PCR fragments were transformed into various
S. pneumoniae strains. Generation of the correct deletion mutant of
tlpA and the operon
spd0571-0573 was verified by PCR using the primers ctrldtlpAfor/ctrltlpArev and ctrl571-573for/ctrl571-573rev, which are located outside the original deletion region; additionally the
tlpA mutant was verified by Western blotting.
Isolation of RNA and RT-PCR.
RNA of
S. pneumoniae D39 was extracted as described before (
14). Briefly, cells were harvested in mid-exponential phase, immediately frozen in liquid nitrogen, and stored at −80°C. RNA extractions were performed as described previously (
13); to remove any possible DNA contamination, DNase I treatment was performed with TURBO DNA-free reagent (Ambion, Austin, TX), and the quality of the RNA was assessed using an Agilent RNA Nano Chip (Agilent Technologies, CA). A TaqMan reverse transcription-PCR (RT-PCR) kit (Life technologies) was used for the generation of cDNA from 100 ng of RNA. Control reactions were performed without reverse transcriptase to confirm the absence of contaminating genomic DNA. The resulting cDNA was used as a template for PCR.
Construction of pNZ8048 derivatives containing tlpA, the operon, and other derivatives.
For complementation analysis, a pNZ8048 (
11) derivative was constructed that contained the
tlpA gene or the
spd0571-0573 operon under the control of
nisA, a nisin-inducible promoter. Primers tlpAF and tlpaR or primers operonF and operonR, which contained NcoI and XbaI restriction sites, were used to amplify the
tlpA gene and the
spd0571-0573 operon from the chromosome of strain D39. The resulting PCR products were introduced into the pNZ8048 plasmid using the restriction sites mentioned above, resulting in pNZ8048::
tlpA and pNZ8048::
spd0571-0573, respectively. Subcloning of the latter plasmid using AvrII/XbaI, NcoI/BseYI, and NcoI/AvrII unique restriction sites in the operon and the plasmid resulted in pNZ8048::
spd0571-0572, pNZ8048::
spd0572-0573, and pNZ8048::
spd0573, respectively. The plasmids were transferred to
S. pneumoniae and
L. lactis for complementation studies.
Paraquat sensitivity assays.
Bacteria were grown until mid-log phase, and 200 μl of the cultures was added to 200 μl of medium with 60 mM paraquat (Supleco) or without paraquat (
61). Both samples were incubated at 37°C for 2 h. The percent survival was calculated by dividing the number of CFU of the cultures after exposure to paraquat by the number of CFU of the control without paraquat.
Hydrogen peroxide sensitivity assays.
The hydrogen peroxide sensitivity assay was performed essentially as described by Pericone et al. (
48). Briefly, bacteria were grown in BHI broth until mid-log phase (optical density at 600 nm [OD
600] of 0.2 to 0.3), and 100-μl aliquots of each culture were added to 100 μl of medium or 100 μl of medium containing 40 mM H
2O
2 (Merck), resulting in an exposure to 20 mM H
2O
2. The bacterial cultures were incubated at 37°C for 30 min. The reaction was stopped by the addition of 200 U of bovine liver catalase (Sigma); after serial dilutions from each well were prepared, aliquots were spotted onto blood agar plates. The percentage of survival of hydrogen stress was determined and calculated as described above.
In vitro competitive growth assay.
TIGR4 and the operon mutant, which contained a spectinomycin resistance cassette, were diluted 1:100 in a 1:1 ratio; as a control, the mutant was also diluted separately to 1:100 in BHI broth. Directly after cultures were mixed, CFU counts were determined at the start of and during exponential (OD600 of ∼0.2) and stationary (OD600 of ∼ 0.5) phases and 24 h later. All serial dilutions were plated on blood agar plates; in addition, the mixed cultures were also plated on plates with spectinomycin to determine the number of mutant bacteria. CFU counts were determined after overnight incubation. To calculate the ratio wild type to mutant, the number of spectinomycin-resistant colonies was subtracted from the total number of colonies. The number of CFU of the mutant derived from the mixed culture plated on spectinomycin plates corresponded with the number of mutant CFU derived from the individual culture plated on blood agar. This indicated that there were no negative effects of the spectinomycin and the coculturing with the wild type. Furthermore, the ratio of the wild type to mutant did not change more than 10-fold over the course of 24 h.
Competitive index assay in a mouse pneumonia model.
Experiments with female BALB/c mice 6 weeks of age were performed at the University of Texas Health Science Center at San Antonio, Texas, under Institutional Animal Care and Use Committee protocol 09022x-34. To minimize distress, mice were anesthetized with 2.5% vaporized isoflurane during all experimental procedures including inoculation and collection of nasal lavage and prior to euthanasia. Mice were infected with either TIGR4 and TIGR4 Δspd0571-0573 (Spr) at a ratio of 1:1 or TIGR4 and TIGR4 Δspd0571-0573 nisRK complemented with pNZ8048E or pNZ8048::spd0571-0573 at a ratio of 1:1. In the latter experiments mice received drinking water supplemented with nisin (2 μg/liter) to drive the expression of the operon from the nisA promoter on the plasmid. For intranasal challenge, mice were held up manually, and the left nostril was inoculated with 2.0 × 106 CFU in 25 μl of phosphate-buffered saline (PBS) in a dropwise fashion using a pipette. Immediately afterwards, mice were hung over their cage by their incisors on a wire until they awoke and removed themselves, typically within 20 to 30 s. For determination of bacterial titers in the nasopharynx, total numbers of pneumococci in nasal lavage fluids were quantified. Briefly, using the same protocol as challenge, 10 μl of PBS was inoculated into the left nostril, followed by its aspiration after 5 to 10 s. Typically, 2 to 4 μl was recovered per mouse. For determination of bacterial titers in the blood, 2 to 5 μl of blood was collected from the tail vein. Bacterial titers in nasal lavage fluid and blood samples were determined by serial dilution, plating on blood agar plates, and extrapolation from bacterial counts following overnight incubation. For determination of bacterial titers in the lungs, mice were sacrificed, and the lungs were excised, weighed, and homogenized. Bacterial titers were determined per gram of homogenized tissue. All dilutions were plated in replicate using plates with the appropriate antimicrobial to discriminate between strains. Statistical analysis of bacterial counts was done using a nonparametric Mann-Whitney rank sum test using SigmaStat statistical analysis software (Aspire Software, Ashburn, VA).
Production and purification of the TlpA protein and generation of anti-TlpA antibodies.
To avoid problems associated with the purification of proteins covalently attached to the membrane such as lipoproteins, tlpA without its signal sequence and its stop codon was amplified from the D39 genome using primers 572SQ and 572SC. The PCR product was cloned into the pET-26b(+) expression vector (Novagen, Inc.) using the NdeI and XhoI restriction sites. The resulting plasmid pET-26b::tlpA was subsequently transformed into E. coli strain BL21(DE3) for high-level TlpA production. Expression of tlpA including a C-terminal His6 tag, which is provided by the pET26b(+) vector, was induced by 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and its production was confirmed by Western blot analysis using an anti-His tag antibody (Zymed-Invitrogen). To purify the protein, a 1-liter culture of E. coli strain BL21(DE3) was induced with 1 mM IPTG; after 3 h the cells were harvested by centrifugation and resuspended in binding buffer (20 mM sodium phosphate buffer, 300 mM NaCl, 10% [vol/vol] glycerol, 5 mM imidazole, 3 mM dithiothreitol [DTT], pH 7.4). Cells were disrupted with a bead beater (Bertin) using three cycles of 30 s and after centrifugation at 14,000 rpm for 5 min; the clarified supernatant fraction was applied to a 5-ml His-Trap HP column (GE Healthcare). After the unbound proteins were washed off the column with binding buffer, the His-tagged TlpA protein was eluted from the column using elution buffer (20 mM sodium phosphate buffer, 300 mM NaCl, 10% [vol/vol] glycerol, 3 mM DTT, pH 7.4) with a gradient of increasing imidazole concentrations (up to 500 mM imidazole). The purity of the eluted protein fraction was determined using SDS-PAGE and Coomassie staining, and 1.14 mg of purified protein was used for the generation of a polyclonal antibody (Eurogentec, Seraing, Belgium).
Fractionation of S. pneumoniae cells.
Fractionation of S.
pneumoniae cells was performed essentially as described previously (
38).
S. pneumoniae was grown to the desired OD in BHI broth, after which the bacteria were spun down for 10 min at 7,000 rpm. The supernatant was precipitated using 50% trichloroacetic acid (TCA), and the cell pellet was taken up in 1/20 volume of lysis buffer (100 mM Tris-HCl, pH 8.0, 20% sucrose, 20 mM MgCl
2) to which fresh lysozyme (5 mg/ml), mutanolysin (200 U/ml), and EDTA-free protease inhibitor (Roche) were added, and the mixture was incubated for 30 min at 37°C. The protoplasts were spun down at 3,000 ×
g for 10 min at 4°C, resuspended in 0.5 ml of sucrose buffer (20% sucrose, 10 mM Tris-HCl, pH 8.0), and disrupted by the addition of 9.5 ml of 100 mM Tris-HCl with EDTA-free protease inhibitor and 1 mM EDTA; undisrupted protoplasts were removed by centrifugation (4,000 ×
g for 10 min at 4°C). The membrane and cytoplasm fractions were separated by centrifugation at 100,000 ×
g for 30 min at 4°C. Samples were taken up in loading buffer, incubated for 5 min at 95°C, and used for SDS-PAGE and Western blot analysis.
SDS-PAGE, Western blotting, and analysis of Western blot signals.
The presence of various proteins was detected by Western blot analysis. Protein fractions were separated by SDS-PAGE using precast NuPAGE gels from Invitrogen and then semidry blotted (1.25 h at 100 mA per gel) onto a nitrocellulose membrane (Protran; Schleicher and Schuell). Detection of antibodies was carried out with fluorescent IgG secondary antibodies (IRDye 800 CW goat anti-rabbit; LiCor Biosciences) in combination with the Odyssey Infrared Imaging System (LiCor Biosciences). For antibody detection, fluorescence at 800 nm was recorded using the linear range. Signals on the Western blots were quantified as follows: non-gamma-transformed images with nonsaturated signals, as determined with the plot profile function of ImageJ (rsb.info.nih.gov/ij), were analyzed using the gel function of the same program. Comparisons of the amount of signal for each type of sample were performed on the same Western blot using the “gel lane” analysis function of the same program. The only digital manipulation performed on the scans was the switching of lanes to obtain the desired order.
Thioredoxin activity assay.
Thioredoxin activity was determined as described by Holmgren (
18). Briefly, 2 μg of purified TlpA or 2 μg of thioredoxin from
E. coli (Sigma) and 100 mM DTT were mixed, and the reaction was started by the addition of 1 mg of insulin from bovine pancreas (Sigma) in 1 ml of 20 mM potassium phosphate buffer, pH 8.0; reactions were monitored for 1 h by measuring the reduction of disulfide bonds in insulin as determined by the precipitation of the protein, which is measured at 650 nm.
Statistical analysis.
Statistical significance of the hydrogen peroxide stress survival was determined using the independent sample t test using SPSS, version 16.0. For the analysis of the animal model a nonparametric Mann-Whitney rank sum test using SigmaStat statistical analysis software (Aspire Software, Ashburn, VA) comparing the ratios to a value of 1 was performed.
DISCUSSION
Throughout the human host,
S. pneumoniae encounters various sources of extracellular oxidative stress, which presents a major challenge for this facultative anaerobic bacterium. In this study we identified a novel component of the pneumococcal defense strategy against extracellular oxidative stress consisting of
tlpA and its neighboring genes
spd0571 and
msrAB. Furthermore, we demonstrated that these three genes play an important role in the establishment of long-term infection. Many of the genes identified in the bioinformatic analysis are involved in virulence, either in the three pathogens used for the comparison or others (
16,
27,
29,
43,
59), suggesting that all these genes may be involved in the virulence of
S. pneumoniae. Some of the genes identified in the bioinformatic analysis are involved in resistance to antibiotics and antimicrobial peptides. The only known niche for these pathogens is the human body, and they are thus constantly exposed to these compounds whereas
L. lactis is not. In the same line, there may be more or other amino acids or other nutrients available in the human body, which could explain the identification of genes involved in pyruvate and amino acid metabolism in this analysis.
The
tlpA gene is located in an operon with a
ccdA homologue and the gene encoding the pneumococcal MsrAB protein (
25). Deletion of both
tlpA alone and the whole operon resulted in a significant decrease in the survival of hydrogen peroxide in several strain backgrounds. This decrease was of similar magnitudes, suggesting that on a functional basis deletion of
tlpA is equivalent to deletion of the complete operon. Complementation by TlpA alone did not restore the phenotype in the single deletion mutant, indicative of polar effects and strongly suggesting that at least TlpA and MsrAB are needed for protection against oxidative stress. However, expression of
tlpA and
msrAB in the three-gene mutant did not increase the survival of oxidative stress, nor did expression of
tlpA and
spd0571 or of
msrAB alone. Only expression of all three genes was able to restore the phenotype. Thus, the effect of
tlpA deletion on the survival of the bacteria is not only due to polar effects on the
msrAB gene. Interestingly, the introduction of the three genes in
L. lactis dramatically increased the survival of this bacterium against oxidative stress, whereas expression of
tlpA,
spd0571-spd0572, or
spd0572-spd0573 had no effect (data not shown). Therefore, we concluded that as in
N. gonorrhoeae (
7), the function of all three proteins is needed for protection against extracellular oxidative stress.
The results of the current study are consistent with Brot et al. concerning the role of this operon in
Neisseria gonorrhoeae (
7). The TlpA protein is homologous (30% identity and 47% positive) to the N-terminal part of the PilB protein. In
N. gonorrhoeae the N-terminal domain of PilB accepts electrons from DsbD; in
S. pneumoniae the CcdA protein is homologous (28% identity, 47% positive) to the transmembrane gamma domain of DsbD. Thus, we hypothesize that the complex functions in the following way: TlpA is reduced by an electron derived from CcdA; subsequently TlpA reduces MsrA/B, which can then reduce oxidized methionines. We were unable to find any thioredoxin activity for TlpA (data not shown) using the method of Holmgren (
18), which may be due to an inability to interact with insulin or incorrect folding of the purified protein. CcdA is predicted to be a membrane protein, which would make it possible for the protein to interact with TlpA. Furthermore, in
N. gonorrhoeae MsrAB has been shown to be located in the outer membrane (
57), and Spd0573 is also predicted to be extracellular, strongly suggesting that these three proteins are indeed able to interact in the membrane of
S. pneumoniae.
Our study underscores the fact that this operon has a similar function in both Gram-positive and Gram-negative bacteria despite the absence of a periplasmic space in S. pneumoniae. In addition, overexpression of the operon in L. lactis resulted in a dramatic increase in oxidative stress survival despite the presence of an MsrA and an MsrB homologue in the lactococcal genome. This indicates that this is a robust and efficient three-partner system to combat extracellular oxidative stress, independent of the organism or conditions under which it functions. The overexpression experiments in L. lactis and wild-type S. pneumoniae and the induction of TlpA under oxidative stress conditions in strain G54 indicated that oxidative stress survival is correlated to the amounts of these proteins. Thus, differential regulation of this operon in S. pneumoniae may occur.
TlpA is a membrane lipoprotein, and deletion did not have an effect on survival in the presence of paraquat, which causes intracellular oxidative stress. Thus, we hypothesize that the operon is specifically involved in the repair of oxidized methionines in membrane or cell wall proteins and that other mechanisms are functioning inside the cell. The role of the operon during the establishment of infection indicates that damage resulting from extracellular hydrogen peroxide stress to noncytoplasmic proteins, in the form of oxidized methionines, is a major problem for
S. pneumoniae in the host as it is for
E. coli (
55). At the moment, it is not clear whether this is due to damage of a particular protein or multiple proteins. Many membrane, cell wall, and secreted proteins contain methionines, which makes it hard to address this question.
The role of ROS generated by cells of the immune system in containing pneumococcal infection is not entirely clear. It has been suggested that neutrophils kill pneumococci through the action of serine proteases instead of their oxidative burst (
35). On the other hand, pneumolysin induces the production of intracellular ROS in human neutrophils (
34), and
S. pneumoniae seems to modulate the oxidative burst (
1). Studies with mice lacking either the p47
phox or the p91
phox subunits of the NADPH-oxidase involved in ROS generation indicate that ROS generated by nonprofessional phagocytes play a role in the control of
S. pneumoniae in the host (
56). Interestingly, our mutant was able to establish infection but was cleared from the mice at a later stage, perhaps due to the influx of neutrophils and or macrophages at this time point. Both our
in vitro competition results and those described in the STM study by Lau et al. (
28), in combination with the fact that the operon is not essential, demonstrate that this protein complex has no obvious role in the survival of endogenously generated H
2O
2. Furthermore, we did not observe any obvious difference with the wild type when the mutant was grown on plates exposed to ambient air (data not shown). This suggests that it is the oxidative stress generated by the host which impairs the virulence of this mutant and that hydrogen peroxide stress resistance is an important virulence factor. Thus, in conclusion we have identified an operon that is involved in the protection of
S. pneumoniae against external peroxide stress and therefore plays an important role in virulence.