Dps and DpsL Mediate Survival In Vitro and In Vivo during the Prolonged Oxidative Stress Response in Bacteroides fragilis

ABSTRACT

Bacteroides fragilis is a Gram-negative anaerobe and member of the human intestinal tract microbiome, where it plays many beneficial roles. However, translocation of the organism to the peritoneal cavity can lead to peritonitis, intra-abdominal abscess formation, bacteremia, and sepsis. During translocation, B. fragilis is exposed to increased oxidative stress from the oxygenated tissues of the peritoneal cavity and the immune response. In order to survive, B. fragilis mounts a robust oxidative stress response consisting of an acute and a prolonged oxidative stress (POST) response. This report demonstrates that the ability to induce high levels of resistance to tert-butyl hydroperoxide (tBOOH) after extended exposure to air can be linked to the POST response. Disk diffusion assays comparing the wild type to a Δdps mutant and a Δdps Δbfr mutant showed greater sensitivity of the mutants to tBOOH after exposure to air, suggesting that Dps and DpsL play a role in the resistance phenotype. Complementation studies with dps or bfr (encoding DpsL) restored tBOOH resistance, suggesting a role for both of these ferritin-family proteins in the response. Additionally, cultures treated with the iron chelator dipyridyl were not killed by tBOOH, indicating Dps and DpsL function by sequestering iron to prevent cellular damage. An in vivo animal model showed that the Δdps Δbfr mutant was attenuated, indicating that management of iron is important for survival within the abscess. Together, these data demonstrate a role for Dps and DpsL in the POST response which mediates survival in vitro and in vivo.

IMPORTANCE B. fragilis is the anaerobe most frequently isolated from extraintestinal opportunistic infections, but there is a paucity of information about the factors that allow this organism to survive outside its normal intestinal environment. This report demonstrates that the iron storage proteins Dps and DpsL protect against oxidative stress and that they contribute to survival both in vitro and in vivo. Additionally, this work demonstrates an important role for the POST response in B. fragilis survival and provides insight into the complex regulation of this response.

INTRODUCTION

Bacteroides spp. are members of the normal intestinal microbiome of humans. The intestine is a consistent and favorable environment that provides continuous access to nutrient sources for these strictly anaerobic organisms. Bacteroides spp. play many important roles in maintaining a healthy intestinal tract, such as polysaccharide degradation, protection of the gut epithelia from colonization by pathogenic bacteria, development of the intestinal tract, maturation of the mucosal and systemic immune systems, and transformation of toxic and mutagenic compounds (1–4). However, when the integrity of the intestinal wall is breached due to trauma, abdominal surgery, or diseases such as appendicitis, perforated ulcer, diverticulitis, and colon cancer, translocation of the normal flora into the peritoneal cavity can result in peritonitis and establishment of an intra-abdominal abscess. The inability of the host immune system to resolve the abscess can lead to bacteremia, sepsis, and in certain instances death (5, 6). B. fragilis is the most common anaerobe isolated from intra-abdominal abscesses, and it has been demonstrated to possess many factors that promote its survival outside the intestinal tract, such as capsular polysaccharides, proteases, neuraminidase, iron acquisition, hemolysins, and resistance to oxidative stress (1, 2, 7, 8). Oxidative stress occurs immediately when B. fragilis translocates from the anaerobic intestine to the more oxygenated (6% O₂) peritoneal cavity, and there is additional oxidative stress resulting from the immune response and polymorphonucleocyte (PMN) recruitment to the site of infection (9–12). Thus, the oxidative stress response is needed for survival during abscess formation (13).

The B. fragilis oxidative stress response is a well-coordinated global response (13). Numerous studies have identified genes and proteins involved in the acute oxidative stress response, many of which are controlled by the LysR family transcriptional regulator, OxyR (13–16). This response occurs rapidly after exposure to H₂O₂ or oxygen and involves activation of OxyR followed by induction of its regulon, whose gene products are aimed at peroxide detoxification, such as catalase (katB), alkyl hydroperoxide reductase (ahpCF), the nonspecific DNA binding protein Dps (dps), and others (13, 15, 17). If oxidative stress and exposure to air are extended for an hour or more, a global shift in transcription occurs, referred to as the prolonged oxidative stress (POST) response, aimed at remodeling cell physiology. This shift alters transcription of nearly 45% of the genes within the genome, with significant changes in the expression of genes for carbohydrate utilization, central metabolism, transport, and transcriptional regulators (13). These changes allow B. fragilis to survive for extended periods in air (>100 h), but specific regulatory factors that control the response have not yet been identified.

Management of intracellular iron availability is a key component of the oxidative stress response. The ferritin family of proteins is responsible for removing excess ferrous iron (Fe²⁺) from the cytoplasm of cells to prevent generation of the damage-inducing hydroxyl radicals via the Fenton reaction (18). These proteins bind and convert Fe²⁺ to nonreactive insoluble ferric iron (Fe³⁺), thus preventing production of hydroxyl radicals (19–21). Members of this family include ferritin, bacterioferritin, Dps, and the recently discovered Dps-like (DpsL) proteins (18, 22). Dps protects cells from oxidative stress damage and shows a strong induction in response to oxidative stress in many organisms (13, 23–25). The B. fragilis dps gene has been shown to be rapidly induced by the oxidative stress regulator OxyR in the acute oxidative stress response; however, an OxyR-independent induction of dps transcription also has been reported (15). Those results demonstrated that activity of a dps::xylB transcriptional fusion was significantly induced during aerobic incubation of the ΔoxyR mutant. This result was seen only for dps expression and not with other members of the OxyR regulon, indicating that a second regulator was responsible (15). The DpsL protein has been shown to have a structure and function that are very similar to those of Dps in both the archaeon Sulfolobus solfataricus and B. fragilis (22, 26). B. fragilis DpsL, the first identified in bacteria, is induced by oxygen and has been shown to play a protective role during periods of oxidative stress (13, 26). The DpsL gene was originally incorrectly annotated as a bacterioferritin gene and was designated bfr; however, later structural studies determined that it actually encodes a DpsL protein (26). B. fragilis does not have a true bacterioferritin. In this study, we used an assay to examine the protective response induced by extended exposure to air. The results demonstrated a role for Dps and DpsL in the POST response which promotes survival both in vitro and in vivo. The protective role that Dps and DpsL play during the POST response is linked to the presence of ferrous iron, indicating that these proteins function to convert and store reactive ferrous iron to nonreactive ferric iron. Additionally, this work indicates that transcriptional control of dps is mediated by a second unknown regulator during the POST response, and these data are consistent with previous findings (13, 15).

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacteroides strains used in this study are listed in Table 1. All strains were grown anaerobically in brain heart infusion broth supplemented with hemin, cysteine, and NaHCO₃ (BHIS) unless otherwise noted (27). Rifampin (20 μg/ml), gentamicin (50 μg/ml), tetracycline (5 μg/ml), cefoxitin (25 μg/ml), and erythromycin (10 μg/ml) were added to the medium when needed.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Phenotype and/or genotypea	Reference or source
B. fragilis strains
IB-101	638R clinical isolate, Rifr	54
IB-260	IB-101 Δkat::tetQ Rifr Tetr	14
IB 298	IB-101 ΔoxyR::tetQ Rifr Tetr	15
IB 336	IB-101 Δdps::tetQ Rifr Tetr	29
IB-430	IB-101 Δaphc::tetQ Rifr Tetr	This study
IB-445	ADB77 (isogenic with IB101) reverted to thyA+ Δtps Rifr	38
IB-542	IB-336 Δbfr::cfx Rifr Tetr Cfxr	This study
IB-567	IB-542 pFD288::bfr Rifr Tetr Cfxr Ermr bfr+	This study
IB-572	IB-336 pFD288::dps Rifr Tetr Ermr dps+	This study
IB-573	IB-542 pFD288::dps Rif Tetr Cfxr Ermr dps+	This study
BER-74	IB-101 Δbfr::cfx Rifr Tetr	26
IB 114	ATCC 25285 clinical isolate, Rifr	13
Other Bacteroides strains
B. uniformis IB-102	VPI strain 006-1 (ATCC 8492) Rifr	55
B. ovatus IB-103	VPI 0038 (ATCC 8433)	56
B. thetaiotaomicron IB-116	VPI 5482 (ATCC 29148)	56
B. vulgatus IB-351	ATCC 8482	56
B. caccae IB-568	VPI 3452A (ATCC 43185)	57
Parabateroides strains
P. distasonis BER-37	Clinical isolate CLA 348	58
P. merdae BER-39	VPI T4-1 (ATCC 43184)	57
E. coli strains
DH10B	F− mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 recA1 endA1 araD139 Δ(ara-leu)7697 galU galK rpsL nupG λ−	Invitrogen
HB101::RK231	HB-101 containing RK231, Kmr Tcr Str	59
Plasmids
pFD288	(Spr) Ermr oriT pUC19::pBI143, 8.8-kb shuttle vector	60
pFD516	Bacteroides suicide vector derived by deletion of pBI143 from pFD288, (Spr) Ermr	60

^aErm^r, erythromycin resistance; Cfx^r, cefoxitin resistance; Rif^r, rifampin resistance; Tet^r, tetracycline resistance; Sp^r, spectinomycin resistance; Amp^r, ampicillin resistance. For Bacteroides-E. coli shuttle vectors, parentheses indicate antibiotic resistance expression in E. coli.

Construction of mutant strains.

All primer sequences used for genetic manipulations are listed in Table S1 in the supplemental material. Briefly, the ΔahpC mutant was constructed by PCR amplification of segment of ahpC coding for the N-terminal fragment using oligonucleotides containing EcoRI sites. The fragment was cloned into a suicide vector, pFD516. The sequence for the C-terminal fragment was amplified using the same approach except that oligonucleotides contained a BamHI recognition site at the 5′ end and an SphI site at the 3′ end. This was then cloned into pFD516 coding for the N-terminal fragment. A 2.2-kb tetracycline cassette (tetQ) was inserted between sequences coding for the N- and C-terminal fragments using the SacI site. This plasmid was then mobilized into B. fragilis IB-101, and exconjugants were selected on BHIS plates containing rifampin, gentamicin, and tetracycline (28). Sensitivity to erythromycin was determined, and PCR was performed to confirm the double-crossover allelic exchange of ahpC::tetQ mutation in strain IB-430.

Construction of the Δdps Δbfr mutant, IB-542, was performed by mobilizing the BER-74 (26) mutational construct into IB-336 (29). Mutants were selected on rifampin, gentamicin, and cefoxitin. PCR was performed to confirm the double-crossover allelic exchange of bfr::cfxA. Complementation of the Δdps mutation in IB-336 and IB-542 was done as follows. The full-length dps gene, including its native promoter, was excised with SphI and EcoRI from plasmid pFD750, cloned into pFD288, and then mobilized into IB-336 and IB-542, respectively (15). Complemented mutants were selected on erythromycin to create IB-572 and IB-573, respectively. Complementation of Δbfr (IB-542) was performed by PCR amplification of full-length bfr gene, including its promoter and insertion into pFD288. Complemented mutants were plated on erythromycin to confirm the presence of the plasmid.

Disk diffusion assays.

Disk diffusion assays were performed as previously described (26). In brief, 100 μl of overnight culture was spread on BHIS plates (without cysteine), and a 6-mm filter disk was placed in the center of the plate. The disks were then saturated with 10 μl of 55 mM tBOOH. Plates either were immediately incubated anaerobically at 37°C or received 3 h of aerobic incubation at 37°C prior to anaerobic incubation. Following overnight anaerobic incubation, the diameters of the zones of growth inhibition were measured, and the results were reported as the averages from at least two independent experiments performed in triplicate. A Student's two-tailed t test was performed to determine significant differences between populations when appropriate.

Cell viability assays.

Cell viability assays were performed by growing cultures to an optical density at 550 nm (OD₅₅₀) of 0.3 in BHIS without cysteine. Cultures were then split, and half of the culture was shaken at 250 rpm in air at 37°C for 3 h. The remaining half was kept under anaerobic conditions and challenged with 500 μM tBOOH. Samples were taken over time and washed three times with BHIS to remove tBOOH. These samples were then serially diluted and plated to determine the number of CFU per milliliter. After 3 h of aerobic shaking, the other half of the culture was challenged with 500 μM tBOOH. Samples were taken and processed as described above. Results are reported as averages from two independent experiments performed in triplicate.

Cell viability assays performed using 2,2′-dipyridyl (Sigma-Aldrich, St. Louis) were done as follows. Cultures were grown in BHIS without cysteine to an OD₅₅₀ of 0.3 and then split. All cultures were kept under anaerobic conditions, but half of the culture was treated with 2,2′-dipyridyl (300 μM) 30 min prior to challenge with 500 μM tBOOH. Samples were taken over time, and the number of CFU per milliliter was calculated as described above. Results are reported as averages from two independent experiments performed in triplicate.

In vivo competition assays.

The rat tissue cage infection model has been described previously (8, 30). Briefly, a perforated sterilized ping-pong ball is surgically implanted into the peritoneal cavity of an adult male Sprague-Dawley rat and allowed to encapsulate for 4 to 5 weeks. During this time, the ball becomes encapsulated in connective tissue, the tissue becomes vascularized, and the ball fills with sterile serous fluid (∼25 ml per ball). Competition assays were performed in this model as previously described (31). In brief, overnight cultures were diluted in PBS (50 mM sodium phosphate, 150 mM sodium chloride [pH 7.4]) and mixed in a 1:1 ratio of wild type (IB-101) to mutant (Δdps, Δbfr, or Δdps Δbfr) to a total of 1 × 10⁵ CFU/ml as a standard inoculum. Four milliliters of inoculum was injected into the tissue cage. Samples were taken at specific time points, serially diluted, and plated on rifampin and gentamicin. After 2 to 3 days of incubation, 200 colonies from each sample were tested for growth on BHIS plates with or without antibiotic (tetracycline for the Δdps and Δdps Δbfr strains and cefoxitin for the Δbfr strain) to check the resistance phenotype and determine the ratio of mutant to wild type.

Competitive indices were calculated for each rat by dividing the number of surviving mutants by the number of surviving wild-type organisms. This was then divided by the ratio of mutant to wild-type organisms in the inoculum. Student's t test was performed to compare the differences between the single and double mutants' abilities to compete. All procedures involving animals followed National Institutes of Health guidelines (32) and were approved by the Animal Care and Use Committee of East Carolina University. For each bacterial strain, two trials including at least 3 animals each were performed.

RESULTS

B. fragilis exhibits an oxygen-induced resistance to tBOOH.

To observe the protective effects of the POST response, disk diffusion assays were used to measure sensitivity to the organic peroxide tBOOH. tBOOH is not easily degraded by cells and can persist, allowing extended periods of oxidative stress, and also causes an Fe²⁺-dependent mechanism promoting DNA damage (33–37). B. fragilis was very sensitive to tBOOH under anaerobic conditions, but when cells were pre-exposed to aerobic conditions for 6 h, they were completely resistant to tBOOH (Fig. 1A). Similar results were seen in assays with two B. fragilis strains, IB-101 and ATCC 25285, as shown in the quantified results in Fig. 1B. Disk diffusion assays that received 6 h of aerobic exposure prior to anaerobic incubation demonstrated no zone of inhibition, whereas assays that received no aerobic incubation had a zone of inhibition of about 50 mm. In order to determine a time course for this induced response, assays were performed in which we varied the length of time of aerobic exposure. As seen in Fig. 1C, complete resistance to tBOOH was achieved only after 3 h of aerobic incubation. Interestingly, this air-induced response requires extended oxygen exposure, whereas the rapid peroxide resistance response mediated by OxyR requires less than 30 min to mediate protection (15).

FIG 1.

Sensitivity to tBOOH after oxygen exposure. (A) B. fragilis IB-101 was exposed to 55 mM tBOOH either under anaerobic incubation or 6 h of aerobic incubation prior to anaerobic incubation. No zone of inhibition is visible in the assays that received aerobic incubation. (B) B. fragilis strains IB-101 and ATCC 25285 were exposed to 55 mM tBOOH in a disk diffusion assay, and zones of inhibition were measured. Dark gray bars represent cells that were exposed to air for 6 h prior to anaerobic incubation. Light gray bars represent assays that were not exposed to air. (C) Sensitivity of IB-101 exposed to 55 mM tBOOH in disk diffusion assays where time of aerobic incubation was varied. Data are representative of triplicate assays performed over two independent experiments; averages and standard deviations are shown.

Dps mediates POST resistance to tBOOH.

To identify gene products that mediate the increased resistance to tBOOH, disk diffusion assays were performed with known oxidative stress mutants. We first looked at the thiol-peroxidase scavengase (Δtps) and the thioredoxin peroxidase (Ωtpx) mutants, because previous studies have shown the mutants to be sensitive to tBOOH under anaerobic conditions (16, 38). Figure 2A shows the results from these experiments. Both mutant strains (Δtps and Ωtpx mutants) had complete resistance to tBOOH after incubation in air, indicating that the mutated enzymes do not play a role in the air-induced resistance response. We additionally looked at alkyl-hydroperoxide reductase (ΔahpC mutant) and catalase (Δkat mutant) because of the role they play in the detoxification of peroxides but found that both mutants were similar to the wild type after aerobic incubation. Interestingly, we found that the Dps (Δdps) mutant had a zone of inhibition of 42 mm even after aerobic incubation. These results indicated that Dps was required for most of the air-induced resistance. Previous studies have shown that exposure of B. fragilis cells to air generates a rapid induction of dps in an OxyR-dependent manner (15). However, when the ΔoxyR mutant was screened in the tBOOH disk diffusion assays, complete resistance was demonstrated after aerobic incubation. This indicated that Dps mediated the response in an OxyR-independent manner, suggesting that a second regulator of dps was responsible for inducing POST dps expression.

FIG 2.

Dps mediates oxygen-induced resistance to tBOOH. (A) Oxidative stress mutants were tested for sensitivity to 55 mM tBOOH in disk diffusion assays. Dark gray bars represent assays exposed to air for 3 h prior to anaerobic incubation. Light gray bars represent assays that were maintained under anaerobic conditions. (B) Cell viability assays were performed using wild-type IB-101 and ΔoxyR and Δdps mutant strains. Cultures were grown to an OD of 0.3 and then split. Half of the culture was shaken in air for 3 h (solid lines), and the other half was incubated under anaerobic conditions (dashed lines). The cultures were then challenged with 500 μM tBOOH. Samples were taken over time, and the number of CFU per milliliter was determined. Data are means from three biological replicates performed over two independent experiments, with standard deviations.

To confirm the results seen in the disk diffusion assays, cell viability assays were performed. Cultures were grown to mid-logarithmic phase and then split. One half was immediately challenged under anaerobic conditions with 500 μM tBOOH, and the number of surviving cells was measured. The other half received aerobic shaking at 37°C for 3 h prior to tBOOH challenge. Results in Fig. 2B show that IB-101, the ΔoxyR strain, and the Δdps strain demonstrated a rapid loss in cell viability when exposed to tBOOH under anaerobic conditions (dashed lines). However, IB-101 cultures that received 3 h of aerobic incubation prior to challenge demonstrated no loss in cell viability. In contrast, the Δdps mutant demonstrated a significant, >4-log decrease in cell viability after aerobic induction, indicating that Dps is important for this resistance phenotype. Interestingly, the ΔoxyR mutant was more similar to IB-101 and showed a much smaller decrease in cell viability than the Δdps mutant, indicating that expression of dps was still induced in the ΔoxyR mutant. Together, the data from the disk diffusion assays and the cell viability assays show that Dps is largely responsible for the oxygen-induced resistance to tBOOH in an OxyR-independent manner.

DpsL contributes to tBOOH resistance.

Previous work demonstrated that the BfDPSL (DpsL) and Dps are similar in protein structure and function (26). Consequently DpsL was investigated to determine if it might account for some of the tBOOH resistance phenotype. Disk diffusion assays were performed with the Δbfr (DpsL) mutant, and the results in Fig. 3 show that the Δbfr mutant had the same phenotype as the wild type. We reasoned that the presence of Dps might mask the role of DpsL, so a Δdps Δbfr mutant was constructed. The Δdps Δbfr mutant had a greater sensitivity to tBOOH (72 mm of inhibition) than the Δdps mutant (53 mm of inhibition). This increased sensitivity to tBOOH also was observed after aerobic incubation, suggesting that the absence of both Dps and DpsL causes the cells to be more sensitive to tBOOH.

FIG 3.

Dps and DpsL both contribute to tBOOH resistance. Δdps, Δbfr(DpsL), and Δdps Δbfr mutants were exposed to 55 mM tBOOH in disk diffusion assays. Dark gray bars represent assays exposed to air for 3 h prior to anaerobic incubation, and light gray bars represent assays that received only anaerobic incubation. Strains were complemented with pFD288 carrying the natural promoter of dps(pFD288::dps) or bfr(pFD288::bfr) to restore function. Data are averages from triplicate assays performed over two independent experiments, with standard deviations. *, P < 0.01.

To confirm the roles of Dps and DpsL, the native genes with their native promoters were cloned on a multicopy plasmid (pFD288) and used to complement the double mutant strain. The bfr-complemented mutant demonstrated complete protection after aerobic incubation and a significant increase in resistance to tBOOH under anaerobic conditions (Fig. 3; also, see Fig. S1 in the supplemental material). This was interesting because it demonstrated that bfr alone was able to fully protect cells from tBOOH even in the absence of dps. Additionally, we were able to complement both the single Δdps mutant and the double Δdps Δbfr mutant with dps on pFD288 and restore the oxygen-induced resistance response. Together, these results indicate that overexpression of Dps and DpsL can mediate this oxygen-induced resistance to tBOOH.

Dps and DpsL mediate protection by sequestering iron.

In many organisms, Dps converts Fe²⁺ to nonreactive Fe³⁺ during periods of oxidative stress to prevent production of highly damaging hydroxyl radicals (19, 24, 39). Similarly it has been shown that the B. fragilis DpsL protects against oxidative stress and is structurally very similar to Dps, although it contains an iron binding site similar to a bacterioferritin (26). To determine if the oxygen-induced response to tBOOH was linked to available reactive iron in the cytoplasm, cell viability assays were performed on cultures treated with dipyridyl, an iron chelator that can enter the cell. As shown in Fig. 4, cultures that were treated with dipyridyl did not show a loss in cell viability when exposed to tBOOH, whereas cultures that were not treated showed significant killing. This result demonstrates that the chelation of intracellular iron by dipyridyl rescued the wild type and all mutants, suggesting that the mechanism of killing during exposure to tBOOH is linked to the presence of reactive iron. The ability of tBOOH to cause oxidative stress by destruction of iron sulfur clusters and DNA cleavage has been documented (35). Since chelation of iron prevents killing of cells by tBOOH, it is likely that the POST response results in the reduction of cytoplasmic iron availability by Dps and DpsL, protecting the cells from damage.

FIG 4.

Chelation of iron rescues all strains under anaerobic conditions. Cell viability assays were performed with cultures of IB-101 (red), the Δdps mutant (green), the Δbfr mutant (orange c), and the Δdps Δbfr mutant (purple). Cultures were grown to an OD of 0.3 and then split. Half was treated with 2,2′-dipyridyl (300 μM) (solid lines), and half was not treated (dashed lines). All cultures were then challenged with 500 μM tBOOH, and the number of CFU per milliliter was determined over time. Data are averages from three biological replicates performed over two independent experiments, with standard deviations.

Dps and DpsL promote survival within the abscess.

Factors that contribute to B. fragilis survival within the abscess are poorly understood. In a previous study using the rat tissue cage model, in vivo microarray analysis demonstrated that the don locus was highly expressed in the infected tissue cages and was required for maximum survival in vivo (31). We reexamined these microarray data and found that there was a 4- to 6-fold increase in expression of dps and bfr, suggesting that Dps and DpsL may promote survival within the abscess. Consequently competition assays were performed to determine if the Δdps, Δbfr, and Δdps Δbfr mutant strains could be outcompeted by the wild type.

Equal numbers of wild-type and mutant cells were used to coinfect rat tissue cages, and samples were taken over a time course. The surviving number of wild-type and mutant cells was determined for each sample, and a competitive index was calculated. These results are shown in Fig. 5, where a competitive index score of 1 indicates that the mutant and wild type compete equally. The mean competitive index scores for the Δbfr mutant were 0.82, 0.76, and 0.89 for days 1, 4, and 8, respectively, indicating that this mutant was able to compete with the wild type. The Δdps mutant showed slight attenuation, with mean competitive index scores of 0.64, 0.53, and 0.60 on days 1, 4, and 8. Although there was a decrease in competitive index score, the values were not statistically significant. Most interesting was the decreased ability of the Δdps Δbfr mutant to compete with the wild type. The Δdps Δbfr mutant had low competitive index scores of 0.21 and 0.17 on days 4 and 8. These competitive index scores were significantly lower than those seen with the single Δdps and Δbfr mutants, indicating that loss of both was necessary to significantly affect survival within the abscess. Overall, these results indicate that both Dps and DpsL may play compensatory roles that contribute to survival within the abscess.

FIG 5.

Dps and DpsL are important for survival in vivo. In vivo competition assays were performed in the rat tissue cage model. Rat tissue cages were inoculated with equal amounts of WT and either the Δdps, Δbfr, or Δdps Δbfr strain to a final inoculum of 1 × 10⁵ CFU/ml. Samples were taken on days 1, 4, and 8 and plated to determine total CFU per milliliter. Colonies were then tested for antibiotic resistance phenotypes to determine the ratio of WT to mutant. Competitive indexes were then calculated for each rat as the output ratio of mutant to WT divided by the input ratio of mutant to WT. Data represent trials of at least three rats from two independent experiments. Mean values for each day are represented by a horizontal line. Student t tests were performed to compare the single mutant competition assays to the double mutant. A P value of <0.05 (*) was found when the double mutant was compared to either the Δdps or Δbfr mutant. A P value of <0.01 (†) was found when the double mutant was compared to either of the single mutants. No difference was seen between the Δdps and Δbfr mutants in competition assays.

Oxygen-induced resistance to tBOOH is not conserved across all members of the genus Bacteroides.

Bacteroides spp. show large variability in the number of ferritin and ferritin-like proteins coded in their genomes, and this diversity is apparent by the many combinations of ferritins, DpsL, and Dps homologues present, as shown in Fig. 6 (see also Table S2 in the supplemental material) (40). For instance, Bacteroides vulgatus lacks dps but has the genes for DpsL and three ferritins. In contrast, B. fragilis contains homologues for Dps, DpsL, and one ferritin and Bacteroides thetaiotaomicron has one Dps, one DpsL, and three ferritins. In general the distribution of the genes for these proteins is conserved in each of the Bacteroides species. Based on available genome sequences for B. fragilis (83 strains), Bacteroides uniformis (3 strains), Parabacteroides merdae (3 strains), Parabacteroides distasonis (8 strains), Bacteroides caccae (2 strains), Bacteroides ovatus (7 strains), B. vulgatus (8 strains), and B. thetaiotaomicron (3 strains), we observed that all strains of species possessing dps and dpsL homologues were consistent. With respect to conservation of the ferritins (ftnA), ftnA1 was conserved in all strains, but the presence of ftnA2 and ftnA3 was variable in strains of B. thetaiotaomicron, B. uniformis, and P. merdae (40). Because of the great diversity seen in the ferritin family proteins and the different responses to oxidative stress in these organisms, we wanted to examine whether the oxygen-induced resistance to tBOOH is conserved across the genus.

FIG 6.

Oxygen-induced resistance to tBOOH is not conserved across the genus Bacteroides. (A) Disk diffusion assays were performed with 55 mM tBOOH with closely related strains of Bacteroides and Parabacteroides. Dark gray bars represent assays exposed to air for 3 h prior to anaerobic incubation. Light gray bars represent assays that were not exposed to aerobic incubation. Zones of inhibition were measured. (B) Graphic representation of numbers and types of ferritins (based on sequence homology). Only species that have one dps, one dpsL, and one ftnA gene demonstrate the oxygen-induced resistance to tBOOH.

Disk diffusion assays were performed using several members of the genus Bacteroides (Fig. 6). Interestingly, the species could be grouped based on the number and type of ferritin homologues found in each genome. Bacteroides caccae and Bacteroides ovatus were similar to B. fragilis, which contains the same number and type of ferritin homologues (1 Dps, 1 DpsL, and 1 ferritin), and they all demonstrated the same phenotype in the tBOOH sensitivity assays. None of the other species tested demonstrated the aerobically induced resistance to tBOOH. B. vulgatus, which has no Dps homologue, showed greater sensitivity to tBOOH after oxygen exposure than under anaerobic conditions. Additionally B. uniformis, B. thetaiotaomicron, and P. merdae showed little to no difference in resistance levels regardless of aerobic incubation, indicating that these species do not have the oxygen-induced resistance response. Most noteworthy was the complete resistance to tBOOH seen in P. distasonis under both conditions.

When the number and type of ferritins in each species and their resistance profiles are compared, some interesting trends can be seen. The first is that only species that contain one Dps, DpsL, and ferritin homologue have the oxygen-induced resistance response phenotype. Interestingly, B. vulgatus, which lacks Dps, had high sensitivity to tBOOH after aerobic exposure, whereas P. distasonis, which has two Dps homologues, was completely resistant to tBOOH, supporting the idea that Dps plays a central role in resistance to tBOOH. However, B. uniformis, B. thetaiotaomicron, and P. merdae, which do have a dps homologue, did not show the inducible resistance phenotype. One explanation for this is that these species have evolved different regulatory mechanism for dps and dpsL expression.

DISCUSSION

B. fragilis has an extensive network of iron storage proteins in the ferritin superfamily—Dps, DpsL, and ferritin—all of which are linked in some way to the oxidative stress response (13, 15, 29). The current report is focused on the role of Dps and DpsL in protection against extended exposure to oxidative stress as part of the POST response. The assay used for this work required a period of prolonged aerobic incubation to rescue cells from tBOOH killing, and the results showed that Dps and to a lesser extent DpsL contributed to protection (Fig. 3). Support for this is that the Δdps mutant was extremely sensitive to tBOOH and could not be rescued by aerobic induction. In addition, the Δdps Δbfr mutant had greater sensitivity to tBOOH than the single mutants, indicating that Dps and DpsL function in a similar manner to protect against this stress. Complementation of the Δdps Δbfr mutant with either bfr or dps resulted in complete resistance to tBOOH after oxygen exposure and provided greatly enhanced resistance under anaerobic conditions (Fig. 3). Finally the Δdps, Δbfr, and Δdps Δbfr mutants were equally protected from tBOOH killing by the iron-chelating agent dipyridyl, indicating that the protective mechanism provided by both of these proteins involves removal of reduced iron from the cytoplasm during periods of oxidative stress (Fig. 4). Taken together, these data are evidence that during the POST response, induction of Dps and DpsL protects cells from damage caused by cytoplasmic ferrous iron. Given that Dps and DpsL share similar functional properties and that either can rescue the POST phenotype, it appears that it is differential regulation of the dps and bfr genes that is key to understanding their varied contributions to protection.

The acute oxidative stress response is designed to rapidly detoxify and minimize the effects of a sudden exposure to oxidative stress. This occurs within minutes of exposure, and OxyR is the major regulator for this response (15). By comparison, previous work and the findings from this study show that the POST response is a more global shift in cellular regulation and physiology occurring after exposure to air for more than 1 h (13). Analysis of dps regulation in B. fragilis has shown that transcription is rapidly induced by exposure to either H₂O₂ or air during exponential growth (15). This is mediated by OxyR and is considered part of the acute oxidative stress response. However, as demonstrated by this study, as well as microarray analysis of gene transcription during prolonged air exposure, Dps also plays an important role in the POST response, and this regulation is independent of OxyR (13). Most interesting was that prolonged exposure to air was required for protection from tBOOH. In exponentially growing cells, OxyR rapidly induces dps expression during oxidative stress; however, as shown in E. coli, OxyR does not induce expression of dps during stationary-phase growth even when cells are exposed to hydrogen peroxide (41). This suggests that the POST response requires the second regulator to induce expression of dps, because OxyR does not function in the nongrowing cells, which is similar to how E. coli regulates dps expression during various growth phases (39, 41). In contrast, bfr gene expression is relatively insensitive to H₂O₂ but is strongly induced by exposure to air for more than 1 h or in anaerobic stationary-phase cultures (13, 26). Overall, these data demonstrate a role for Dps in both the acute and POST oxidative stress responses, whereas DpsL appears to have a role only in the POST response. The regulation of the POST response is of great interest because it leads to protection when cells are not rapidly growing and allows the high aerotolerance seen in B. fragilis. Dps and DpsL both have roles during prolonged oxidative stress, and further studies of their differential regulation should help identify the important POST regulator(s).

To investigate the roles of Dps and DpsL in survival in vivo, growth of wild-type and mutant strains was compared in a rat tissue cage model of infection. This model has effectively been used to show the attenuation of mutant strains of B. fragilis within an artificial abscess (8, 31). Experiments that compared the ability of the wild-type strain to outcompete single Δdps and Δbfr mutants showed only slight attenuation that was not statistically significant. However, the double Δdps Δbfr mutant was significantly attenuated, as shown in Fig. 5, indicating that Dps and DpsL are both required for maximum survival in the abscess model. These data also indicate that Dps and DpsL may play overlapping roles in protecting the cells from oxidative stress damage in vivo, because the absence of both was required to see the phenotype. As previously shown in this model, B. fragilis reaches high cell numbers and then enters a stationary-like phase where the high cell density is maintained (8, 31). Interestingly, on day 1 the Δdps Δbfr mutant was able to compete effectively with the wild type and was close to 50% of the 10⁸ to 10⁹ CFU/ml; however, on days 4 and 8, there was a decrease in the competitive ability of the double mutant (Fig. 5). Results from the in vitro growth analysis (see Fig. S2 in the supplemental material) indicate that there is no general growth defect in the double mutant. It is reasonable to suggest from these data that the double mutant may experience DNA and protein damage due to oxidative stress and higher levels of ferrous iron within the cytoplasm during days 4 to 8.

Oxidative stress occurs immediately upon bacterial translocation from the anaerobic intestinal tract to the oxygenated peritoneal cavity, with additional stress resulting from the immune system's response to bacterial presence in the peritoneum, making high levels of ferrous iron toxic to cells. To survive, B. fragilis requires an effective system for management of intracellular iron, which in part is provided for by Dps and DpsL. The contribution of Dps and other ferritins to virulence has been shown in other organisms. Mutations in ferritin family genes in Salmonella enterica, Haemophilus influenzae, and Streptococcus pyogenes were responsible for defects in survival in vivo and affected susceptibility to killing from oxidative stress (42–45). These results are similar to those seen here with B. fragilis. Though the majority of Bacteroides spp. have dps and bfr (DpsL), only three species, B. fragilis, B. caccae, and B. ovatus, demonstrate the air-inducible tBOOH resistance phenotype, suggesting that the regulation of these genes may provide some advantage in an extraintestinal site (Fig. 6).

The normal environment for B. fragilis is the large intestine, which is known to be highly anaerobic, yet B. fragilis has a robust oxidative stress response. It has been shown that resistance to oxidative stress is important for establishment of intra-abdominal abscesses, but this habitat is a dead end, leaving the question of what selects for resistance to oxidative stress in the colon. One thought is that this stress may occur during the inflammatory response. Inflammation of the intestinal tract caused by Campylobacter jejuni, Helicobacter pylori, and many other pathogens results in increased levels of oxidative stress within the epithelial layer and the intestinal tract (46–49). B. fragilis is closely associated with the intestinal epithelium, which has been shown to incur significant DNA damage from reactive oxygen species from the host immune response (47, 50, 51). Additionally, it has been shown that an oxygen concentration gradient exists extending out from the epithelial surface, so that B. fragilis may be exposed to an environment with as much as 8% oxygen, depending on the precise site of colonization (51, 52). It is reasonable to suggest that these conditions could cause significant oxidative stress to the organism. Being able to store and scavenge reactive ferrous iron would be essential for survival of Bacteroides spp. in this changing environment and during the inflammatory response. Therefore, the diversity and quantity of ferritin-like proteins used by B. fragilis would give it an advantage and promote survival during these times. Then, in the event of intestinal damage and the translocation of the natural flora into the peritoneal cavity, the organisms that are better suited to survive the oxidative burst of the immune response will be able to persist, promoting the establishment of an abscess. This may in part explain why B. fragilis is so frequently isolated from intra-abdominal abscesses. Additionally, transmission of B. fragilis from mother to child results in exposure to an aerobic environment, and an effective oxidative stress response would provide for more efficient transmission (1, 53).

This report demonstrates that Dps and DpsL are part of the POST response in B. fragilis. These proteins are responsible for storing ferrous iron and preventing it from producing hydroxyl radicals in the cytoplasm during periods of oxidative stress. Studies are needed to further elucidate the regulation of the POST response and transcriptional control of dps and bfr. Dps and DpsL provide protection for the cells during survival within the abscess and ultimately within the intestinal tract. Overall, these data indicate that B. fragilis and potentially other members of the genus Bacteroides must be able to efficiently manage iron in order to survive as members of the natural flora of the intestinal tract.