The uroS and yifB Genes Conserved among Tetrapyrrole Synthesizing-Deficient Bacteroidales Are Involved in Bacteroides fragilis Heme Assimilation and Survival in Experimental Intra-abdominal Infection and Intestinal Colonization

The human intestinal anaerobic commensal and opportunistic pathogen Bacteroides fragilis does not synthesize the tetrapyrrole protoporphyrin IX in order to form heme that is required for growth stimulation and survival in vivo. Consequently, B. fragilis acquires essential heme from host tissues during extraintestinal infection. The absence of several genes necessary for de novo heme biosynthesis is a common characteristic of many anaerobic bacteria; however, the uroS gene, encoding a uroporphyrinogen III synthase for an early step of heme biosynthesis, is conserved among the heme-requiring Bacteroidales that inhabit the mammalian gastrointestinal tract.

ABSTRACT

The human intestinal anaerobic commensal and opportunistic pathogen Bacteroides fragilis does not synthesize the tetrapyrrole protoporphyrin IX in order to form heme that is required for growth stimulation and survival in vivo. Consequently, B. fragilis acquires essential heme from host tissues during extraintestinal infection. The absence of several genes necessary for de novo heme biosynthesis is a common characteristic of many anaerobic bacteria; however, the uroS gene, encoding a uroporphyrinogen III synthase for an early step of heme biosynthesis, is conserved among the heme-requiring Bacteroidales that inhabit the mammalian gastrointestinal tract. In this study, we show that the ability of B. fragilis to utilize heme or protoporphyrin IX for growth was greatly reduced in a ΔuroS mutant. This growth defect appears to be linked to the suppression of reverse chelatase and ferrochelatase activities in the absence of uroS. In addition, this ΔuroS suppressive effect was enhanced by the deletion of the yifB gene, which encodes an Mg²⁺-chelatase protein belonging to the ATPases associated with various cellular activities (AAA⁺) superfamily of proteins. Furthermore, the ΔuroS mutant and the ΔuroS ΔyifB double mutant had a severe survival defect compared to the parent strain in competitive infection assays using animal models of intra-abdominal infection and intestinal colonization. This shows that the presence of the uroS and yifB genes in B. fragilis seems to be linked to pathophysiological and nutritional competitive fitness for survival in host tissues. Genetic complementation studies and enzyme kinetics assays indicate that B. fragilis UroS is functionally different from canonical bacterial UroS proteins. Taken together, these findings show that heme assimilation and metabolism in the anaerobe B. fragilis have diverged from those of aerobic and facultative anaerobic pathogenic bacteria.

INTRODUCTION

The Bacteroides species are major components of the human lower intestinal tract, constituting approximately 20% to 40% of the total bacteria, with numbers reaching 10¹¹ to 10¹² CFU/g of colonic content (1–3). Using the anaerobic, opportunistic, human pathogen Bacteroides fragilis as a model organism, several studies have shown that it has an essential requirement for both exogenous heme and inorganic iron for growth stimulation in vitro and in vivo (4–9). This group of host-associated anaerobic bacteria is heme auxotrophic, lacking several genes in the tetrapyrrole biosynthetic pathway to form protoporphyrin IX (PpIX) as the final precursor for the synthesis of heme in this organism (6). In addition to host-associated Bacteroides, other members of the human gut microbiota, such as Firmicutes, Actinobacteria, and Euryarchaeota, also lack complete tetrapyrrole and essential heme biosynthetic pathways (6, 10).

Despite the absence of several genes needed to synthesize tetrapyrrole precursors for heme biosynthesis, a few intermediary genes, such as the uroporphyrinogen III synthase gene uroS (formerly named hemD) and the oxygen-independent coproporphyrinogen dehydrogenase gene cgdH (formerly hemN), are conserved in B. fragilis and in many closely related anaerobic bacteria within the Bacteroidales order (6). In contrast, free-living aerobic species within the Cytophagales and Flavobacteriales orders contain the complete de novo heme biosynthetic pathway found in classical Gram-negative bacteria (6). UroS activity is a strategic step in tetrapyrrole macrocycle biosynthesis as it catalyzes the cyclization of the linear tetrapyrrole preuroporphyrinogen (1-hydroxymethylbilane) by inversion of the D ring to form the asymmetric cyclic tetrapyrrole uroporphyrinogen III, an essential isomer for the synthesis of hemes, chlorophylls, and cobalamins (11–13). However, the role of the uroS gene in the genomes of tetrapyrrole synthesizing-deficient Bacteroidales is not well understood since an exogenous source of heme or protoporphyrin IX is sufficient to support their growth.

Moreover, the lack of a tetrapyrrole macrocycle synthesis pathway is not a competitive disadvantage relative to the heme-prototrophic bacteria in the nutritionally complex and highly populated intestinal environment (6, 14). In fact, it is heme-synthesizing bacteria such as proteobacteria that are outnumbered by heme-requiring bacterial species such as Bacteroides, which have the ability to scavenge intestinal luminal heme from host tissues, dietary sources, and bacterial sources (3, 6, 14, 15). Therefore, the conserved presence of the uroS gene in Bacteroides species may contribute to the ability of these organisms to regulate and utilize heme to compete effectively with other bacteria in extraintestinal infections and in the intestinal tract.

In this study, we investigated the participation of UroS in the utilization of heme and protoporphyrin IX by B. fragilis in vitro and survival in vivo in a rat tissue cage model of intraperitoneal infection and in a mouse intestinal colonization model. Moreover, during the course of this investigation, we initiated studies to show that the magnesium chelatase-like AAA⁺ (ATPases associated with diverse cellular activities) ATPase YifB acts in combination with UroS for survival in vivo. AAA⁺ proteins form a large and diverse superfamily found in all organisms, and they are involved in a wide variety of different functions (16–18). The YifB family is widespread in bacteria, and sequence analyses show that YifB is most closely related to the chelatase family, with which it shares distinct sequence signatures (19). Therefore, our objective is to elucidate the role of the uroS and yifB genes in the pathophysiology and ecological competitiveness of B. fragilis, which will help to advance our understanding of the mechanism developed by B. fragilis to adapt and survive as the opportunistic pathogen most frequently isolated from human infections caused by anaerobic bacteria.

RESULTS

Incomplete heme biosynthesis pathway genes.

The uroS (hemD) gene is present in nearly all host-associated Bacteroidales genomes available in the GenBank database. The presence of uroD (hemE), cgdH (hemN), and pgoX (hemY) is also found in several species although less frequently (Fig. 1). All the host-associated Bacteroidales genomes lack the first four genes required for the synthesis of the UroS precursor hydroxymethylbilane (HMB): gtrR (hemA), gsaM (hemL), pbgS (hemB), and hmbS (hemC). In contrast, aerobic free-living Bacteroidetes members from the Flavobacteriales and Cytophagales orders possess the complete heme biosynthetic pathway. The oxygen-dependent coproporphyrinogen III oxidase gene, cgdC (hemF), is found in major free-living, heme-prototrophic species such as Cytophaga hutchinsonii and Flavobacterium johnsoniae but is absent in the heme-requiring, host-associated, obligately anaerobic species of the Bacteroidales order. Except for Porphyromonas gingivalis, all other species of the host-associated Bacteroidales lack the classical ferrochelatase gene ppfC (hemH).

FIG 1.

Schematic diagram of the conserved heme biosynthesis pathway genes in representative species within the Bacteroidetes phylum. (A) The classical heme biosynthetic pathway in most Gram-negative bacteria was used as a model for comparison with the Bacteroidetes. The enzymes’ new designations, as described by Dailey et al. (10), are depicted in bold letters above the old abbreviations shown within parentheses. (B) The order of the arrows depicting the gene ORF locus is shown according to their respective product name in the heme biosynthetic pathway for the purpose of comparative analysis and does not represent their chromosomal genetic organization. Each gene is represented by an arrow with different colors and shades. Arrows positioned in a vertical column are predicted to encode gene products with homology to each respective enzyme on top of the column. (Republished from Heme and iron metabolism in Bacteroides [6].) The drawings are not to scale. GtrR, glutamyl-tRNA reductase (EC 1.2.1.70); GsaM, glutamate-1-semialdehyde 2,1-aminomutase (EC 5.4.3.8); PbgS, porphobilinogen synthase (EC 4.2.1.24); HmbS, hydroxymethylbilane synthase (EC 2.5.1.61); UroS, uroporphyrinogen III synthase (EC 4.2.1.75); UroD, uroporphyrinogen decarboxylase (EC 4.1.1.37); CgdC, oxygen-dependent coproporphyrinogen III oxidase (EC 1.3.3.3); CgdH, oxygen-independent coproporphyrinogen III dehydrogenase (EC 1.3.99.22); PgdH1, protoporphyrinogen oxidase (EC 1.3.3.4); PgoX, protoporphyrinogen oxidase (EC 1.3.3.4); PpfC, protoporphyrin ferrochelatase (EC 4.99.1.1). The alternative routes of heme biosynthesis via the intermediates percorrin-2, sirohydrochlorin, siroheme, 12,18-didecarboxysiroheme, and iron-coproporphyrin III (Fe-COPRO III) in Archaea and Desulfovibrio (48–51) and via coproporphyrinogen oxidase (CgoX) in Gram-positive bacteria (52) are not shown for clarity since they do not appear to be present in Bacteroides spp. Bacterial names and strain designations are shown for each comparative row. B., Bacteroides; P., Parabacteroides.

Bacteroides species UroS is phylogenetically distinct from those of other bacteria.

An unrooted phylogenetic tree (Fig. 2) was obtained from a multiple-sequence alignment of Bacteroidetes UroS amino acid sequences with UroS sequences from other prokaryotes, archaea, and eukaryotes (see Fig. S1 in the supplemental material). The clusters with the Bacteroides spp. are noticeably separated from other bacteria. Archaea, eukarya, and all other bacteria are distinctly branched from a monophyletic clade separated from the branch related to aerobic, free-living, and heme-prototrophic species belonging to the Flavobacteriales and Cytophagales orders within the Bacteroidetes phylum. A comparison of the amino acid sequence alignments revealed that 7 out of the 10 conserved residues in the proposed active-site cleft of human UroS (Thr62, Ser63, Y168, S197, Lys220, T227, and T228) (20) are conserved in most of the Bacteroides species UroS proteins (Fig. S1). The Arg65 residue (human UroS), defined as being nonconserved in the active site (20), is replaced mostly by His or Thr in the Bacteroidetes. In addition, the preceding residue, Pro64 (human UroS), is replaced by Arg in nearly all the examined Bacteroidetes UroS proteins (Fig. S1). Furthermore, the key conserved tyrosine residue Tyr168 (human UroS) or Tyr155 (Thermus thermophilus UroS), which may function in the removal of the C₂₀ hydroxyl of the substrate hydroxymethylbilane (13, 20), is conserved in all the Bacteroidetes phylum UroS proteins analyzed except for a Phe164 substitution in Parabacteroides distasonis ATCC 8503.

FIG 2.

Phylogenetic relationship of 72 uroporphyrinogen III synthase (UroS/HemD) homologue representatives from the Bacteroidetes phylum, aerobic and facultative bacteria (belonging to Firmicutes, cyanobacteria, proteobacteria, and deinococcus groups), Archaea, and Eukaryota. The unrooted phylogenetic tree was constructed from a multiple-amino-acid-sequence alignment based on the ClustalW algorithm in the AlignX program of Vector NTI 11.5.5. The neighbor-joining method of Saitou and Nei (53) was used to calculate the distances between all pairs, and the calculated relative distance values are depicted in parentheses. GenBank accession numbers for each protein entry are shown in square brackets following each organism’s species designation.

Phylogenetic analysis of Bacteroides species YifB.

An unrooted phylogenetic tree (Fig. S2) constructed from multiple-sequence alignments of Bacteroides species YifB amino acid sequences with YifB sequences from other prokaryotes (Fig. S3) revealed that the clusters containing YifB from alpha-, beta-, gamma-, delta-, and epsilonproteobacteria; cyanobacteria; Chlorobi bacteria; cyanobacteria; actinobacteria; and Clostridium spp. are all branched from a distinct monophyletic clade divided between aerobic free-living groups within the Bacteroidetes phylum such as Cytophaga and Flavobacterium.

Role of UroS and YifB in growth in vitro.

To test whether the lack of the uroS gene would affect B. fragilis physiology, the growths of the ΔuroS mutant (BER-184), the ΔyifB mutant (BER-185), and the ΔuroS ΔyifB double mutant (BER-186) strains were compared to that of the parent strain (Fig. 3A and B). When bacterial strains were grown in the presence of PpIX and inorganic ferrous iron or in the presence of heme as the sole source of iron, the ΔuroS and ΔuroS ΔyifB strains showed a strong growth defect compared to the parent strain. In contrast, the deletion of the yifB gene did not appear to affect growth compared to the parent strain under the same conditions (Fig. 3A and B). The genetic complementation of the ΔuroS mutant (BER-202) and ΔuroS ΔyifB double mutant (BER-203) strains with the uroS gene expressed under the control of the starch/maltose-inducible promoter of the osu operon in the pFD1274 plasmid restored growth to levels similar to the parent strain level in medium containing starch (Fig. 3C and D). Western blot analyses using rabbit polyclonal antibodies raised against the purified B. fragilis UroS (Bf-UroS)-6×His recombinant protein confirmed the expression of the UroS protein in the genetically complemented strain BER-202 (Fig. S4).

FIG 3.

Growth of B. fragilis strains in defined medium (DM) containing 0.5% glucose (A and B) or 0.3% starch (C and D). Media were supplemented with 5 μg/ml protoporphyrin IX (PpIX) plus 100 μM ammonium ferrous sulfate (Fe) (A and C) or 5 μg/ml hemin (He) plus 400 μM bathophenanthroline disulfonic acid (BPS) (B and D). Strain designations are depicted as follows: parent (BER-183), ΔuroS (BER-184), ΔyifB (BER-185), ΔuroS ΔyifB (BER-186), parent/pFD1045 (BER-201), ΔuroS/pFD1274 (BER-202), and ΔuroS ΔyifB/pFD1274 (BER-203). Growth on starch in panels C and D was performed to induce the expression of the uroS gene in the pFD1274 plasmid derived from pFD1045. Vertical bars denote standard deviations of the means from two independent experiments in duplicate.

Heterologous complementation of the B. fragilis uroS deletion mutant with the E. coli uroS gene does not restore the growth phenotype.

When the Escherichia coli uroS (Ec-uroS) gene was expressed in the B. fragilis ΔuroS deletion mutant carrying pER-323 (BER-234), it did not restore growth compared to the B. fragilis ΔuroS strain expressing the native Bf-uroS gene in pER-320 (BER-236) (Fig. 4A). In fact, the presence of the E. coli uroS gene caused an increase in the growth defect of the B. fragilis ΔuroS strain in defined medium (DM) containing heme as a sole source of iron. The growth of the B. fragilis strains in supplemented brain heart infusion broth (BHIS medium) was not significantly affected after 10 h and 24 h of incubation, except for the suppression of BER-234 strain growth at mid-log phase (Fig. 4B). The expression of the B. fragilis uroS gene in the SASZ31Rif strain carrying pER-320 (EC334) or pER-337 (EC339) did not stimulate growth on plates containing Luria-Bertani (LB) medium (Fig. 4C). The SASZ31Rif strain carrying an empty vector (EC333) did not grow on LB plates after a 24-h incubation aerobically, as expected. In contrast, the SASZ31Rif strain complemented with the E. coli uroS gene in the plasmid pER-323 (EC335) or pER-325 (EC336) grew on LB plates incubated aerobically. BHI plates incubated anaerobically were used as the growth control for the SASZ31Rif derivative strains used in this study (Fig. 4D).

FIG 4.

Heterologous genetic complementation of the B. fragilis uroS deletion mutant and E. coli uroS mutant strains shows that B. fragilis UroS and E. coli UroS are functionally different. (A) Growth of B. fragilis strains in defined medium (DM) containing 0.5% glucose supplemented with 5 μg/ml hemin (He) plus 400 μM bathophenanthroline disulfonic acid (BPS). (B) Growth of B. fragilis strains in BHIS medium. Strains and the respective plasmids are depicted on the panels as follows: parent/pFD340 empty vector (BER-237), ΔuroS/pFD340 empty vector (BER-238), ΔuroS/pER-320 carrying the Bf-uroS construct (BER-236), and ΔuroS/pER-323 carrying Ec-uroS_BfRBS (BER-234). (C and D) Growth of the E. coli SASZ31Rif strain carrying different plasmid constructs in Luria-Bertani (LB) medium incubated aerobically (C) or in BHI medium containing 0.5% yeast extract (YE) and 1% glucose incubated anaerobically (D). Plasmid designations are depicted on each panel as follows: pFD340 empty vector (EC333), pER-320 carrying Bf-uroS (EC334), pER-323 carrying Ec-uroS_BfRBS (EC335), pER-325 carrying Ec-uroS (EC336), and pER-337 carrying Bf-uroS_EcRBS (EC339). Vertical bars denote standard deviations of the means from three independent experiments in triplicate.

Ferrochelatase and reverse chelatase activities.

When crude extracts of bacteria grown on defined medium containing PpIX and iron were analyzed, there were statistically significant decreases in ferrochelatase activity of approximately 3.5-fold in the ΔuroS mutant (493 pmol/mg/h) and 3.2-fold (544 pmol/mg/h) in the ΔuroS ΔyifB double mutant compared to the parent strain (1,719 nmol/mg/h) (Fig. 5A). This indicated that the lack of the uroS gene led to a reduction in ferrochelatase activity. The genetic complementation of the ΔuroS strain (ΔuroS/pFD1274) restored ferrochelatase activity to approximately 2,390 pmol/mg/h, confirming the contribution of UroS to this process. Conversely, the lack of the yifB gene alone showed a significant increase in ferrochelatase activity of approximately 2.2-fold (3,876 pmol/mg/h) compared to the parent strain (Fig. 5A). The complementation of the yifB gene in the ΔyifB/pFD1272 strain did not alter the ferrochelatase activity (4,097 pmol/mg/h) seen in the ΔyifB strain in the presence of PpIX. However, in the presence of heme, the ΔyifB/pFD1272 strain showed enhanced ferrochelatase activity, as described below.

FIG 5.

Ferrochelatase and reverse chelatase (dechelatase) activity assays of crude extracts of Bacteroides fragilis strains. Bacteria were grown in defined medium containing 5 μg/ml protoporphyrin IX (PpIX) plus 100 μM ammonium ferrous sulfate or 5 μg/ml hemin (He) plus 400 μM bathophenanthroline disulfonic acid (BPS). Media were inoculated with a 1:50 dilution of a culture grown overnight in BHIS medium. Bacteria were then grown anaerobically for 24 to 48 h. Details of the ferrochelatase and reverse chelatase activity assay settings are described in Materials and Methods. Data are means ± standard deviations from two biologically independent experiments in triplicate. The significance (P value) by an unpaired t test (parametric and two tailed) between two experimental groups is shown above the horizontal bars. n.s., not significant.

When bacteria were grown in defined medium supplemented with hemin as a sole source of iron (Fig. 5A), there were significant decreases in ferrochelatase activity of approximately 4.7-fold (1,422 pmol/mg/h) in the ΔuroS mutant (BER-184) and 4.6-fold (1,439 pmol/mg/h) in the ΔuroS ΔyifB double mutant (BER-186) compared to the parent strain (6,672 pmol/mg/h). The genetic complementation of uroS in the ΔuroS/pFD1274 (BER-202) strain increased activity to values approximately 2-fold (15,500 pmol/mg/h) higher than that in the parent strain, demonstrating that the presence of the uroS gene indeed affects ferrochelatase activity. Although the ΔyifB single mutant strain did not show significant differences in specific activity (7,183 pmol/mg/h) compared to the parent strain, the genetically complemented ΔyifB/pFD1272 strain (BER-204) showed an increase of approximately 2.9-fold (19,581 pmol/mg/h) compared to the parent strain.

Figure 5B shows that the reverse chelatase activities were not significantly altered in either the ΔuroS mutant (2,502 pmol/mg/h), the ΔyifB mutant (1,065 pmol/mg/h), or the ΔuroS ΔyifB double mutant (2, 320 pmol/mg/h) strain compared to the parent strain (1,421 pmol/mg/h) when bacteria were grown in iron-replete medium containing PpIX. The genetic complementation in the ΔuroS/pFD1274 and ΔyifB/pFD1272 strains did not significantly alter the enzyme activities in the presence of PpIX. When the ΔuroS mutant strain was complemented with the entire putative polycistronic operon, containing the BF638R_0059, BF638R_0060, and BF638R_0061 (uroS) genes (Fig. S5) in the ΔuroS/pFD1276 strain (BER-212)(BF638R_0059 and BF638R_0060 are diploids), there was a significant increase in reverse chelatase activity of approximately 4-fold (5,228 pmol/mg/h) compared to the parent strain (1,421 pmol/h/mg).

When bacteria were grown in defined medium with heme as a sole source of iron, the mean reverse chelatase specific activities in the ΔuroS mutant (0.11 pmol/mg/h) and ΔyifB mutant (0.14 pmol/mg/h) strains were reduced by approximately 13-fold and 10-fold, respectively, compared to the parent strain (1.44 pmol/mg/h). However, these results were not statistically significant (Fig. 5C). In addition, the activity in the ΔuroS ΔyifB double mutant strain was reduced to an undetectable level (below the blank level) compared to the parent strain. The genetically complemented strains showed restored reverse chelatase activity. However, the recombinant UroS-6×His (rUroS-6×His) protein purified under aerobic conditions did not show detectable forward or reverse chelatase activities (Fig. S6). The uroporphyrinogen III synthase specific activity of the rUroS-6×His protein is 0.869 ± 0.042 nmol/mg/h, and the K_m for the substrate HMB is 0.064 ± 0.013 μM.

To demonstrate whether or not the uroS and yifB genes would have a significant contribution to pathophysiology and survival in a host environment in vivo, experimental intra-abdominal infection and intestinal colonization studies were carried out.

UroS and YifB contribute to survival in vivo.

(i) Intra-abdominal infection. Equal numbers of parent and mutant cells were used to coinfect rat tissue cages, and samples were taken over a time course. The surviving populations of the parent strain and the ΔuroS mutant strain revealed that the ΔuroS mutant was quickly outcompeted at 1 day postinfection, with an ∼2-log-fold decrease compared to the parent strain; i.e., ∼98% of the total population was the parent strain (Fig. 6A). The ΔuroS mutant population decreased by >3-log-fold by 15 days postinfection compared to the parent strain. This indicates that the lack of the uroS gene is a disadvantage regarding competitive fitness in an extraintestinal environment. When the ΔuroS ΔyifB double mutant was coinoculated with the parent strain, the ΔuroS ΔyifB population decreased by 2 logs at 1 day postinfection. Next, the ΔuroS ΔyifB double mutant population steadily decreased at day 4 (2.7 logs), day 8 (3.5 logs), and day 15 (4 logs) postinfection compared to the parent strain, indicating that the mutation of the yifB gene in the ΔuroS ΔyifB double mutant slightly enhanced the competitive disadvantage in growth compared to the parent strain (Fig. 6B). The ΔyifB single mutant was not tested for dual inoculation in intraperitoneal infection. Figure 6C shows that the parent/pNBU2-bla-tetQ and parent/pNBU2-bla-ermG strain populations compete equally, demonstrating that the insertion of the pNBU2-bla-ermG or pNBU2-bla-tetQ vector with the respective antibiotic markers did not affect bacterial recovery counts. We assume that single-copy insertions of the pNBU2-derived constructs were inserted into the att1 or att2 integration site in tRNA^Ser without disrupting genetic functions based on the nonreplicating properties of the NBU2 genetic element described in Bacteroides thetaiotaomicron (21).

FIG 6.

Competitive survival of B. fragilis strains inoculated into the intraperitoneal tissue cage. (A) Dual infection with parent Erm^r (BER-208) versus ΔuroS Tet^r (BER-210) strains. (B) Dual infection with parent Erm^r (BER-208) versus ΔuroS ΔyifB Tet^r (BER-217) strains. (C) Dual-infection control with parent Erm^r (BER-208) versus parent Tet^r (BER-223) strains. Bacteria grown overnight in BHIS medium were diluted in PBS and mixed at a 1:1 ratio to approximately 1 × 10⁵ CFU/ml of each strain. Four milliliters of the dual-bacterium suspension was inoculated into the intraperitoneal tissue cage. Fluid samples were aspirated at time points for CFU counts as described in Materials and Methods. Data are expressed as the mean CFU per milliliter of intra-abdominal tissue cage fluid from six rats. The standard errors of the means (SEM) are denoted by vertical error bars.

The surviving numbers of parent strain and mutant strain cells determined for each rat were also used to calculate competitive indices. These results are shown in Fig. S7, where a competitive index score of 1 indicates that the mutant and wild type compete equally. The mean competitive index scores for the ΔuroS mutant were 0.03, 0.006, 0.002, and 0.0003 and the mean competitive index scores for the ΔuroS ΔyifB mutant were 0.01, 0.002, 0.0005, and 0.0003 at days 1, 4, 8, and 15 postinfection, respectively. The competitive indices indicated that the ΔuroS ΔyifB double mutant was slightly less competitive than the ΔuroS single mutant at days 1, 4, and 8 postinfection, although at day 15, both mutants were similarly less competitive than the parent strain.

(ii) Intestinal colonization. Dual colonization of the mouse intestinal tract with the parent strain and the ΔuroS mutant strain showed a decrease of approximately 2.5 logs in the ΔuroS mutant population compared to the parent strain after 7 days of colonization (Fig. 7A). The ΔuroS ΔyifB double mutant displayed a robust growth defect in the cecum, with an ∼4-log decrease (Fig. 7B), while the ΔyifB single mutant did not show a significant growth defect during the same colonization period compared to the parent strain (Fig. 7C).

FIG 7.

Competitive dual bacterial colonization of the C57BL/6J mouse intestinal tract by B. fragilis strains. (A) Parent Erm^r (BER-208) versus ΔuroS Tet^r (BER-210) strains. (B) Parent Erm^r (BER-208) versus ΔuroS ΔyifB Tet^r (BER-217) strains. (C) Parent Erm^r (BER-208) versus ΔyifB Tet^r (BER-226) strains. (D) 638R FA^r (fusidic acid-resistant) (BER-180) versus ΔfeoAB Tet^r (BER-51) strains. Erm^r is the erythromycin resistance marker in strains carrying pNBU2-bla-ermG, and Tet^r is the tetracycline resistance marker in strains carrying pNBU2-bla-tetQ except for ΔfeoAB::tetQ strain BER-51. Two and four days after stopping antibiotic treatment, mice (n = 6) were inoculated with the dual bacterial suspension at a 1:1 ratio in 0.2 ml PBS containing an average of 1.1 × 10¹⁰ to 2.3 × 10¹⁰ CFU/ml of each strain by oral gavage as described in Materials and Methods. After 7 days, the cecal content was serially diluted and plated on BHIS medium containing appropriate antibiotics for the selection and enumeration of colonies. Colony counts were normalized to CFU per gram of cecal content. Mice were placed on a standard rodent chow diet. The whiskers denote the minimum and maximum values. The line in the middle of the box is plotted at the median. The boxes span the interquartile range. The significance (P value) following an unpaired t test (parametric and two tailed) for the two groups is shown above the horizontal bars. n.s., not significant.

For the purpose of comparison, we performed dual colonization with the BER-180 strain (B. fragilis 638R fusidic acid resistant) and the BER-51 strain (B. fragilis 638R ΔfeoAB::tetQ) in mice placed on the same rodent diet. This normal standard diet is iron replete, containing 360 ppm iron (Prolab IsoPro RMH 300, product codes 5P75 and 5P76; LabDiet). Under these conditions, there was no significant difference in the population of the ΔfeoAB mutant strain compared to the parent strain (Fig. 7D).

DISCUSSION

In this study, we show that a deletion of the B. fragilis uroS gene causes a severe growth defect compared to the parent strain both in vivo and in vitro. This indicates that the conservation of an early intermediary gene in the heme biosynthetic pathway is necessary to endure the ecological pressure facing heme-requiring bacteria. Although heme or its precursor protoporphyrin IX is sufficient to maintain optimal growth and can be obtained from host tissues or dietary sources (6), the role of UroS in B. fragilis heme metabolism remains undefined. It is possible that B. fragilis UroS may function in a noncanonical way. There is little evidence that B. fragilis UroS acts as a genuine, primary uroporphyrinogen III synthase. This assumption is supported by the fact that the absence of the uroS and yifB genes compromises survival during extraintestinal infection and intestinal colonization compared to the parent strain. This indicates that these mutants were unable to obtain heme from host sources. Therefore, it is unlikely that these phenotypic properties were in any way associated with the conversion of hydroxymethylbilane into uroporphyrinogen III. This is based on the findings that the canonical E. coli UroS protein does not restore the growth of the B. fragilis ΔuroS deletion mutant strain, nor does B. fragilis UroS restore the growth of the E. coli uroS gene-deficient strain. This indicates that B. fragilis UroS and E. coli UroS may be functionally different.

B. fragilis UroS has a quite distant phylogenetic relationship to other bacterial UroS proteins. This agrees with the findings that Bacteroides spp. generally share little homology to other Gram-negative species, with amino acid identities of <40% being the most common (22). In addition, the UroS proteins have some of the most highly divergent amino acid sequences of the heme biosynthetic enzymes (13), although certain domains are highly conserved in several organisms (23). Moreover, the B. fragilis purified rUroS-6×His protein uroporphyrinogen III synthase specific activity found in this study (0.869 ± 0.042 nmol/mg/h) is several orders of magnitude (6 × 10⁵ to 1.7 × 10⁶) lower than the specific activity of UroS reported for aerobic and facultative anaerobic bacteria such as Bacillus subtilis (565 μmol/mg/h) (24) and E. coli (1,500 μmol/mg/h) (25). In addition, B. fragilis rUroS-6×His has a K_m (0.064 ± 0.013 μM) approximately 5-fold and 78-fold lower than the K_m values of UroS from B. subtilis (K_m of 330 ± 30 nM) (24) and E. coli (K_m of 5 μM) (25), respectively. These findings indicate that Bf-UroS binds the substrate HMB very efficiently, but the enzyme activity is very low compared to those of UroS proteins from B. subtilis and E. coli. The low Bf-UroS enzymatic activity, which indicates slow turnover or breakdown of the enzyme-substrate complex, may explain the inability of B. fragilis UroS to complement the E. coli UroS deficiency.

Nonetheless, our findings show that B. fragilis UroS has biochemical and molecular properties associated with other proteins and cellular processes involved in the chelatase and reverse chelatase activities whose mechanisms have yet to be characterized. In this regard, the uroS operon (see Fig. S5 in the supplemental material) contains two putative coexpressed, upstream genes that seem to affect reverse chelatase activity when expressed in a multicopy plasmid in the presence of PpIX. The ygdH gene (BF638R_0059) encodes a protein with a predicted NAD(P)-binding Rossmann fold domain, and the BF638R_0060 gene encodes a conserved, putative, integral membrane protein of unknown function with six transmembrane domains. Homology of BF638R_0060 is widespread among the Bacteroides-Prevotella group of heme-auxotrophic, host-associated anaerobes, and it has no significant homology to genes of any of the other major intestinal colonizers such as the Firmicutes, Actinobacteria, and Proteobacteria. This indicates that the conserved genetic organization of the genes associated with uroS in B. fragilis performs metabolic activities that are yet to be defined. In addition, we cannot exclude from consideration that UroS may be involved in the molecular alterations of free heme to maintain it at nontoxic levels. Taken together, these findings indicate that tetrapyrrole metabolism, heme metabolism, and their regulation in heme-requiring, commensal bacteria and opportunistic, pathogenic, anaerobic bacteria have yet to be elucidated.

This study has also shown that the deletion of the yifB gene intensifies the survival defect of the ΔuroS mutant strain in extraintestinal infection and intestinal colonization. Unfortunately, very little is known about the function of the YifB family of proteins present in several bacterial families. YifB is a member of a subfamily of proteins of unknown function within the AAA⁺ ATPase superfamily. This class of proteins represents a distinct class of P-loop nucleoside triphosphatases (NTPases) that are involved in diverse functions in which the energy extracted from ATP hydrolysis is used for molecular remodeling events (16–19). Bf-YifB shares distinct amino acid sequence homology with Mg²⁺ chelatase subunit I (BchI/ChlI) (44% identity and 62% similarity) from Rhodobacter sphaeroides and with the ATP-dependent Lon protease (44% identity and 62% similarity) from E. coli, which belongs to the large AAA⁺ ATPase superfamily. We do not have evidence for dual enzyme functionality, but our findings indicate that Bf-YifB is at least involved in chelatase/dechelatase activities, possibly by carrying out ATP hydrolysis to provide energy for the iron chelation/dechelation of heme anaerobically. Therefore, we believe that Bf-YifB may be divergent from the BchI/ChlI subunit, which, when associated with BchD/ChlD, binds and hydrolyzes the ATP necessary for the incorporation of an Mg²⁺ ion into protoporphyrin IX by the catalytic BchH/ChlH subunit of the Mg²⁺ chelatase complex (26, 27). Further investigations will help us to understand how Bf-YifB may contribute to the chelatase/dechelatase activities in vitro as well as in host tissues. It is plausible to speculate that the Bf-YifB and Bf-UroS mechanisms of action were developed and adapted to function anaerobically under the very low redox potential required by these organisms to grow. This supposition is based on the statement that “except for a few metabolic processes in which dioxygen serves as an essential reactant, the whole of metabolic biosynthesis in bacteria is anaerobic in character. This is clearly evidenced in strictly anaerobic bacteria capable of growth in minimal media which testifies to this fact” (28).

At this point in the investigation, it is not possible to explain the ecological benefit that the lack of the genes for de novo tetrapyrrole and heme biosynthesis achieves for the nutritional and pathophysiological attributes of B. fragilis. B. fragilis lacks GtrR (HemA), GsaM (HemL), PgbS (HemB), and HmbS (HemC) homologues for the synthesis of hydroxymethylbilane as a substrate precursor for UroS to form uroporphyrinogen III. Therefore, the significance of the conservation of UroS as well as YifB remains largely unknown. Some progress has been made to show that the ferrochelatase and reverse chelatase activities are decreased in the absence of the uroS and yifB genes; conversely, these activities are increased in the absence of the CobN-like BtuS1 and BtuS2 proteins (9). Moreover, the colonization fitness defects of the ΔuroS mutant and the ΔuroS ΔyifB double mutant strains were significantly greater than those of the ΔfeoAB mutant strain, indicating that heme assimilation and metabolism are important factors for survival, even under iron-replete conditions. These findings point to the fact that B. fragilis has developed unique regulatory and metabolic mechanisms to utilize heme that are divergent from those of other aerobic and facultative anaerobic bacteria.

The exploration and elucidation of the molecular mechanisms and regulatory systems that control heme utilization in B. fragilis pathogenesis will allow us to understand how intestinal, opportunistic pathogens rapidly adapt and compete for heme in extraintestinal infections and intestinal colonization. Investigations into metabolic pathways in heme-auxotrophic bacteria may identify new targets as a basis for the development of novel therapeutic interventions for anaerobic bacterial abscess infections that are currently difficult to treat as the number of antibiotic-resistant Bacteroides strains grows (29–31).

MATERIALS AND METHODS

Strains, media, and growth conditions.

B. fragilis strains and plasmids used in this study are listed in Table 1. Strains were routinely grown on BHIS medium (brain heart infusion broth supplemented with l-cysteine [1 g/liter], hemin [5 mg/liter], and NaHCO₃ [20 ml of a 10% solution per liter]) or as otherwise stated in the text. Rifampin (20 μg/ml), gentamicin (100 μg/ml), tetracycline (5 μg/ml), and erythromycin (10 μg/ml) were added to the media when required. A modified defined medium (DM) (32) containing KH₂PO₄ (1.15 g/liter), NH₄SO₄ (0.4 g/liter), NaCl (0.9 g/liter), l-methionine (75 mg/liter), MgCl₂·6H₂O (20 mg/liter), CaCl₂·2H₂O (6.6 mg/liter), MnCl₂·4H₂O (1 mg/liter), CoCl₂·6H₂O (1 mg/liter), resazurin (1 mg/liter), l-cysteine (1 g/liter), hemin (5 mg/liter), and glucose (5 g/liter) was used. The final pH was 6.9. Twenty milliliters of sterile 10% NaHCO₃ was added per liter of medium inside the anaerobic chamber. For some experiments, soluble starch (0.3 g%) was added to replace glucose as an inducer of the osu promoter in strains carrying plasmids derived from pFD1045 (Table 1). Heme was replaced by protoporphyrin IX (5 μg/ml) as the source of the tetrapyrrole macrocycle when required (4). For iron restriction in DM, the ferrous iron chelator bathophenanthroline disulfonic acid (BPS) was added at a 400 μM final concentration. Ammonium ferrous sulfate at 100 μM was added for ferrous iron-replete growth conditions. Escherichia coli strains were grown routinely on Luria-Bertani medium supplemented with ampicillin (100 μg/ml), streptomycin (50 μg/ml), kanamycin (50 μg/ml), or tetracycline (10 μg/ml) when appropriate. The E. coli uroS-deficient strain SASZ31 (33) was grown anaerobically on BHI medium supplemented with yeast extract (0.5%) and d-glucose (1%). The SASZ31 strain forms dwarf colonies on BHI plates (33).

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description or genotypea	Reference or source
Strains
B. fragilis 638R	Clinical isolate; Rifr	54
BER-51	638R ΔfeoAB::tetQ; Rifr Tetr	7
BER-180	638R spontaneous fusidic acid-resistant isolate	This study
BER-183	638R Δtdk; Rifr FUdRr	This study
BER-184	638R Δtdk ΔuroS; Rifr FUdRr	This study
BER-185	638R Δtdk ΔyifB; Rifr FUdRr	This study
BER-186	638R Δtdk ΔuroS ΔyifB; Rifr FUdRr	This study
BER-201	BER-183 carrying pFD1045; Rifr FUdRr Ermr	This study
BER-202	BER-184 carrying pFD1274, Rifr FUdRr Ermr	This study
BER-203	BER-186 carrying pFD1274; Rifr FUdRr Ermr	This study
BER-204	BER-185 carrying pFD1272; Rifr FUdRr Ermr	This study
BER-208	BER-183/pNBU2-bla-ermG; Rifr FUdRr Ermr	This study
BER-210	BER-184/pNBU2-bla-tetQ; Rifr FUdRr Tetr	This study
BER-212	BER-184 carrying pFD1276; Rifr FUdRr Ermr	This study
BER-217	BER-186/pNBU2-bla-tetQ; Rifr FUdRr Tetr	This study
BER-223	BER-183/pNBU2-bla-tetQ; Rifr FUdRr Tetr	This study
BER-226	BER-185/pNBU2-bla-tetQ; Rifr FUdRr Tetr	This study
BER-234	BER-184 carrying pER-323; Rifr FUdRr Ermr	This study
BER-236	BER-184 carrying pER-320; Rifr FUdRr Ermr	This study
BER-237	BER-183 carrying pFD340; Rifr FUdRr Ermr	This study
BER-238	BER-184 carrying pFD340; Rifr FUdRr Ermr	This study
E. coli
DH10B	Cloning host strain	Invitrogen
HB101::RK231	HB101 containing RK231; (Kmr) (Tetr) (Strr)	55
S17-1 λpir	Strain with the RK2 tra genes for conjugative transfer integrated into the chromosome (RP4-2-Tc::Mu-Km::Tn7 Pro Res− Mod+ Tpr Smr) (λpir lysogen)	56
BL21(DE3)	F− ompT gal dcm lon hsdSB(rB− mB−) λ(DE3)	Novagen
EC308	BL21(DE3) carrying pER-307; Kmr	This study
SASZ31	F− hisG1 xyl-7 hemD31 metE46 mobA2 ΔargH1 rplL9(L?) thiE1; (Strr)	33b
EC326 (SASZ31Rif)	SASZ31 spontaneous rifampin-resistant isolate	This study
EC333	SASZ31Rif carrying pFD340; (Strr) (Ampr) (Rifr)	This study
EC334	SASZ31Rif carrying pER-320; (Strr) (Ampr) (Rifr)	This study
EC335	SASZ31Rif carrying pER-323; (Strr) (Ampr) (Rifr)	This study
EC336	SASZ31Rif carrying pER-325; (Strr) (Ampr) (Rifr)	This study
EC339	SASZ31Rif carrying pER-337; (Strr) (Ampr) (Rifr)	This study
Plasmids
pExchange-tdk	Derivative of pKNOCK-bla-ermGb carrying a cloned tdk gene for counterselection; (Ampr) Ermr	36
pNBU2-bla-ermG	NBU2 integrase (intN2)-based genomic insertion vector derived from pKNOCK-bla-ermGb inserts into the NBU2 att1 or att2 site of tRNASer; (Ampr) Ermr	36
pNBU2-bla-tetQ	NBU2 integrase (intN2)-based genomic insertion vector derived from pKNOCK-bla-tetQb inserts into NBU2 the att1 or att2 site of tRNASer; (Ampr) Tetr	21
pET26b(+)	Expression vector with T7 promoter; lacI; pBR322 replicon; Kmr	Novagen
pFD288	Bacteroides-E. coli shuttle vector; oriT; pUC19::pBI143 chimera; (Spr) Ermr	34
pFD340	Bacteroides-E. coli expression shuttle vector; (Ampr) Ermr	39
pFD516	Suicide vector derived from deletion of pBI143 in pFD288; (TetX) (Spr) Ermr	34
pFD1045	Shuttle vector derived from pFD340 containing the starch maltose- and oxygen-inducible promoter of the osu operon; (Ampr) Ermr	57
pFD1241	1,669-bp DNA fragment containing a 528-bp deletion of the tdk gene cloned into the SalI/EcoRI sites of pFD516; (Spr) Ermr	This study
pFD1265	3,810-bp DNA fragment containing an internal deletion of the yifB gene cloned into the XbaI/PstI sites of pExchange-tdk; (Ampr) Ermr	This study
pFD1272	1,518-bp DNA fragment containing the promoterless yifB gene cloned into the BamHI and SacI sites of pFD340; (Ampr) Ermr	This study
pFD1274	987-bp DNA fragment containing the promoterless uroS gene cloned into the BamHI and SacI sites of pFD1045; (Ampr) Ermr	This study
pFD1276	2,601-bp DNA fragment containing entire uroS operon (BF638R_0059–BF638R_0061) cloned into the BamHI/SacI sites of pFD288; (Spr) Ermr	This study
pER-302	3,210-bp DNA fragment containing the ΔuroS construct cloned into the XbaI/SalI sites of pExchange-tdk; (Ampr) Ermr	This study
pER-307	845-bp uroS DNA fragment cloned in frame into the NdeI and XhoI sites of pET26b(+) to construct a UroS fusion recombinant protein with a C-terminal 6×His tag; (Kmr)	This study
pER-320	987-bp DNA fragment containing the promoterless uroS gene cloned into the BamHI and SacI sites of pFD340; (Ampr) Ermr	This study
pER-323	854-bp DNA fragment containing the E. coli uroS ORF with 21 bp of the B. fragilis ahpC RBS region placed immediately upstream of the uroS ATG codon (Ec-uroSBfRBS) cloned into the BamHI/SacI sites of pFD340; (Ampr) Ermr	This study
pER-325	872-bp DNA fragment of the E. coli native uroS gene including 49 bp upstream of the ATG codon cloned into the BamHI/SacI sites of pFD340; (Ampr) Ermr	This study
pER-337	942-bp DNA fragment containing the B. fragilis uroS gene with 57 bp of the E. coli RBS region placed immediately upstream of the uroS ATG codon (Bf-uroSEcRBS) cloned into the BamHI/SacI sites of pFD340; (Ampr) Ermr	This study

^aErm^r, erythromycin resistance; Rif^r, rifampin resistance; Tet^r, tetracycline resistance; FUdR^r, 5-fluor-2′-deoxyuridine resistance; Amp, ampicillin resistance; Sp^r, spectinomycin resistance; Str^r, streptomycin resistance; Km^r, kanamycin resistance. Parentheses indicate antibiotic resistance expression in E. coli.

^bThe SASZ31 strain was obtained from the E. coli Genetic Stock Center at Yale University.

Construction of the B. fragilis 638R tdk deletion mutant strain.

An 819-bp DNA fragment containing the upstream region plus 39 nucleotides of the N-terminal open reading frame (ORF) of the thymidine kinase gene (tdk) (BF638R_0630) was amplified from the B. fragilis 638R chromosome by PCR using primers NF-tdk and NR-tdk (see Table S1 in the supplemental material). The amplified fragment was cloned into the EcoRI and BamHI sites of the suicide vector pFD516 (34). An 850-bp DNA fragment containing the downstream region including 33 nucleotides of the C-terminal ORF was amplified using primers CF-tdk and CR-tdk and cloned into the BamHI and SalI sites of pFD516 containing the 819-bp N-terminal fragment to construct a 1,670-bp DNA fragment with a 528-bp internal deletion of the tdk gene. The new construct, pFD1241, was mobilized from E. coli DH10B into B. fragilis 638R by triparental mating (35). The transconjugants were selected on BHIS plates containing rifampin (20 μg/ml), gentamicin (100 μg/ml), and erythromycin (10 μg/ml). Resolution of Δtdk mutants was obtained by growing four transconjugants in BHIS broth containing rifampin (20 μg/ml) and 5-fluor-2′-deoxyuridine (FUdR) (200 μg/ml). The use of FUdR was performed according to the concentration used for positive counterselection of the Bacteroides thetaiotaomicron Δtdk strain (36). The Δtdk strain-enriched cultures were plated out on BHIS plates containing rifampin (20 μg/ml) and FUdR (200 μg/ml). Colonies were tested for erythromycin sensitivity to confirm the loss of the suicide vector. The FUdR-resistant and erythromycin-sensitive strains were selected for PCR analysis using primers NF-tdk and CR-tdk to confirm the deletion of a 528-bp internal DNA fragment of the tdk gene. The BER-183 strain (BF638R Δtdk) was used as the parent strain for further studies.

Construction of the uroS deletion mutant.

DNA fragments of 1,581 bp and 1,629 bp containing the upstream N-terminal and downstream C-terminal DNA fragment regions of the uroS gene (BF638R_0061) were amplified by PCR from the B. fragilis 638R chromosome using the primer sets UroS-NT-XbaI/UroS-NT-SEW and UroS-CT-SEW/UroS-CT-SalI, respectively. The N-terminal and C-terminal DNA fragments were joined together using a “sewing” PCR methodology. The 3,210-bp DNA fragment product containing the 429-bp internal deletion of the uroS gene was reamplified by PCR using primers UroS-NT-XbaI and UroS-CT-SalI. The amplified fragment was cloned into the XbaI and SalI sites of the positive counterselection suicide vector pExchange-tdk (36). The new construct, pER-302, was mobilized from E. coli S17-1 λpir into BER-183 by biparental mating. Transconjugants were selected on BHIS plates containing rifamycin (20 μg/ml), gentamicin (100 μg/ml), and erythromycin (10 μg/ml). Transconjugants were grown on BHIS broth with rifampin and gentamicin overnight, anaerobically at 37°C, to enrich for double-crossover genetic exchange recombinants. The cultures were plated on BHIS plates containing rifamycin (20 μg/ml), gentamicin (100 μg/ml), and FUdR (200 μg/ml). Sensitivity to erythromycin was assessed to confirm the loss of the suicide vector. PCR amplification using primers UroS-mutcheck-F and UroS-mutcheck-R was performed to confirm the deletion of the uroS gene in the new strain BER-184. Construction of the ΔuroS ΔyifB mutant was carried out by conjugating pFD1265, described below, from E. coli S17-1 λpir into BER-184 to obtain ΔuroS ΔyifB strain BER-186.

Deletion of the yifB gene.

A 1,923-bp DNA fragment containing the upstream N-terminal region and a 1,887-bp DNA fragment containing the downstream C-terminal region were amplified by PCR using the primer sets 4364-NT-XbaI-FOR/4364-NT-BamHI-REV and 4364-CT-BamHI-FOR/4364-CT-PstI-REV, respectively. The N-terminal fragment was cloned into the XbaI/BamHI sites of the pExchange-tdk vector, followed by the insertion of the C-terminal fragment into the BamHI/PstI sites to construct a 3,810-bp fragment containing an internal deletion of 1,210 bp from the yifB gene. The new construct, pFD1265, was mobilized from E. coli S17-1 λpir into BER-183 by biparental mating as described above. Selection for FUdR resistance and erythromycin sensitivity was performed as described above. The deletion of the yifB gene by double-crossover genetic exchange in the BER-185 strain was confirmed by PCR analysis using primers 4364_mutcheck-FOR and 4364_mutcheck-REV.

Genetic complementation of the ΔuroS mutant strain.

A 987-bp DNA fragment containing the uroS ORF plus 46 bp upstream of the ATG start codon and 179 bp downstream of the stop codon was amplified by PCR from the B. fragilis 638R chromosome using primers UroScpmt-BamHI-F and UroScpmt-SalSac-R. The promoterless amplified fragment was cloned into the BamHI and SacI sites of pFD1045. The new construct, pFD1274, was mobilized from E. coli DH10B into the BER-184 or BER-186 strain by triparental mating as described above, to make BER-202 and BER-203, respectively. The 987-bp DNA fragment containing the promoterless uroS gene was also cloned into the BamHI/SacI sites of pFD340 to construct pER-320. The new plasmid, pER-320, was mobilized from E. coli S17-1 λpir into BER-184 by biparental mating to obtain the BER-236 strain. A 2,601-bp DNA fragment containing the entire uroS polycistronic operon was amplified by PCR using primers UroS_operon-FOR and UroS_operon-REV and cloned into the BamHI/SacI sites of pFD288. The new construct, pFD1276, was mobilized from E. coli DH10B into BER-184 by triparental mating to obtain BER-212.

For genetic complementation of the B. fragilis uroS deletion with the E. coli uroS gene, an 854-bp DNA fragment was amplified by PCR from the E. coli DH10 chromosome (GenBank accession number ACB04829) using primers EcUroS-BamHI-FOR and EcUroS-SacI-REV. The oligonucleotide primer EcUroS-BamHI-FOR was designed to place the ribosome-binding site (RBS) of the B. fragilis ahpC gene (37, 38) immediately upstream of the E. coli uroS ATG codon. This procedure was carried out to replace the E. coli RBS region and insert a native B. fragilis RBS chromosomal region to optimize the translation of the E. coli uroS gene in B. fragilis. This DNA fragment (Ec-uroS_BfRBS) was cloned into the BamHI/SacI sites of pFD340 to construct pER-323. The new construct was mobilized from E. coli S17-1 λpir into the BER-184 strain by biparental mating to obtain the BER-234 strain. The expression of the uroS gene is driven by the constitutive IS4351 promoter from the original expression vector pFD340 (39). Promoter sequences located within the insertion sequence (IS) element also function in E. coli (39).

For genetic complementation of the E. coli SASZ31 uroS-deficient strain with the B. fragilis uroS gene, 57 bp of the E. coli RBS region was placed immediately upstream of the B. fragilis uroS ATG codon by PCR amplification from the pER-307 plasmid using the primers pER-307-BamHI-FOR and pER-307-SacI-REV. The amplified 942-bp DNA fragment (Bf-uroS_EcRBS) was cloned into the BamHI/SacI sites of pFD340. The new construct, pER-337, was mobilized from E. coli S17-1 λpir into SAZ31Rif (EC326) to obtain the E. coli EC339 strain. pFD340, pER-320, and pER-323 were also mobilized into the SAZ31Rif (EC326) strain to obtain EC333, EC334, and EC335, respectively. As a control, the native E. coli uroS gene containing 49 bp upstream of the ATG codon was amplified by PCR from E. coli DH10B using primers EcUroS-NT-FOR and EcUroS-SacI-REV. The 872-bp DNA fragment was cloned into the BamHI/SacI sites of pFD340. The new construct, pER-325, was mobilized from E. coli S17-1 λpir into E. coli SASZ31Rif to obtain the E. coli EC336 strain.

Genetic complementation of yifB.

A 1,518-bp DNA fragment containing the yifB gene ORF plus 38 bp upstream of the ATG start codon was amplified from the B. fragilis 638R chromosome using the primers 4364-Pless-NT-FOR and 4364-Pless-CT-REV. The 1,518-bp DNA fragment containing the promoterless yifB gene was cloned into the BamHI and SacI sites of the Bacteroides expression vector pFD340. The new construct, pFD1272, was mobilized into BER-185 by triparental mating as described above, to make BER-204.

Growth of ΔuroS and ΔyifB mutants and determination of forward and reverse chelatase activities.

Bacteria were grown on DM containing 5 μg/ml heme plus 400 μM BPS or medium containing 5 μg/ml protoporphyrin IX plus 100 μM ammonium ferrous sulfate. Media were inoculated with a 1:50 dilution from a culture grown overnight in BHIS medium. Growth was determined by measuring the optical density at 550 nm (OD₅₅₀) at time points. Bacteria were grown for 24 h anaerobically at 37°C. Bacterial cultures were centrifuged at 12,000 × g for 10 min at 4°C. The pellets were washed twice with phosphate-buffered saline (PBS) (10 mM Na₂HPO₄, 1.7 mM KH₂PO₄, 145 mM NaCl, 2 mM KCl [pH 7.4]) and centrifuged as described above. The pellets were frozen at −70°C and submitted to the Iron and Heme Core facility, Department of Internal Medicine, Division of Hematology, University of Utah Health Sciences, Salt Lake City, UT, to assay for ferrochelatase and dechelatase specific activities in bacterial crude extracts. The ferrochelatase and dechelatase assays were performed as previously described (9), according to the protocols available at the Iron and Heme Core website (http://cihd.cores.utah.edu/ironheme/#1465838003097-f4fffc70-f824).

In vivo dual bacterial competitive survival assays.

All procedures involving animals were performed according to the guidelines given in the National Research Council’s Guide for the Care and Use of Laboratory Animals, 8th ed. (40), and approved by the Institutional Animal Care and Use Committee of East Carolina University.

(i) Intra-abdominal infection. The rat tissue cage model of intraperitoneal infection was used as described previously (41–43). First, a perforated sterilized ping-pong ball was surgically implanted into the peritoneal cavity of a 3- to 4-month-old male Sprague-Dawley rat weighing >400 g (purchased from Charles River Laboratories Inc., Raleigh, NC) and allowed to encapsulate for 4 to 5 weeks. pNBU2-bla-ermG or pNBU2-bla-tetQ was conjugated into bacterial strains to introduce erythromycin or tetracycline antibiotic resistance markers for bacterial selective isolation and colony counting from mixed infections. Dual bacterial competitive assays were performed as previously described (43). In brief, cultures grown overnight in BHIS medium were diluted in PBS and mixed in a 1:1 ratio of the parent strain/pNBU2_ermG (BER-208) to the ΔuroS/pNBU2-bla-tetQ (BER-210) or ΔuroS ΔyifB/pNBU2-bla-tetQ (BER-217) strain for a total of approximately 1 × 10⁵ CFU/ml of each strain as a standard inoculum. Dual colonization with the parent strain carrying pNBU2-bla-ermG (BER-208) versus the parent strain carrying pNBU2-bla-tetQ (BER-223) was also performed as a control for possible inherently intrinsic variables when using antibiotic resistance markers for differential CFU quantification of laboratory strains recovered directly from infectious specimens. An aliquot of the inoculum suspension was serially diluted in PBS and plated on erythromycin (10 μg/ml) or tetracycline (5 μg/ml) to determine inoculum CFU per milliliter. Four milliliters of the inoculum was injected into the intraperitoneal tissue cage of 3 rats per group in two independent experiments (n = 6). Fluid sample aspirates were taken at 1, 4, 8, and 15 days postinfection; serially diluted; and plated on medium containing rifampin (20 μg/ml), gentamicin (100 μg/ml), and FUdR (200 μg/ml) plus erythromycin (10 μg/ml) or tetracycline (5 μg/ml). After 3 to 4 days of incubation in an anaerobic chamber at 37°C, colonies from each sample time point were counted and normalized to the CFU per milliliter of tissue cage serous fluid. The limit of detection was 1 × 10¹ CFU/ml. Competitive indices were calculated for each rat by dividing the number of surviving mutant cells by the number of surviving parent strain cells. This was then divided by the ratio of the mutant to the parent strain cells in the inoculum input, as previously described (43).

(ii) Intestinal colonization. Specific-pathogen-free (SPF), 6- to 8-week-old, male C57BL/6J mice were purchased from Jackson Laboratories and housed under microisolation conditions (animal biological safety level 2 housing facility). These mice were used for dual colonization with the parent strain/pNBU2-bla-ermG (BER-208) versus the ΔuroS/pNBU2-bla-tetQ (BER-210) or ΔuroS ΔyifB/pNBU2-bla-tetQ (BER-217) strain. According to other established mouse enteric colonization models (44–47), mice were given drinking water with the antibiotics gentamicin (0.30 mg/ml), ciprofloxacin (0.66 mg/ml), and metronidazole (0.40 mg/ml). The sweetener Stevia extract (20 mg/ml) was added to the water to increase consumption since other reports have indicated occasional refusal of mice to drink water containing antibiotics (44). Antibiotic treatment was carried out for 7 days and subsequently replaced with sterile drinking water. Two days and four days following the withdrawal of antibiotics, an inoculum containing a 1:1 ratio of the parent strain to the mutant strain was mixed to a total of approximately 2 × 10⁹ CFU for each strain in 0.2 ml PBS and given by oral gavage to 3 mice per group in two independent experiments (n = 6). Mice were housed with sterile autoclaved water and sterile irradiated standard rodent chow ad libitum. After 7 days, animals were euthanized. The cecum was slit open, and cecum content specimens were weighed, serially diluted in sterile PBS, and plated on rifampin (20 μg/ml), gentamicin (100 μg/ml), and FUdR (200 μg/ml) plus erythromycin (10 μg/ml) or tetracycline (5 μg/ml), as described above. For dual colonization with BER-180 versus BER-51 strains, cecum content dilutions were plated onto medium containing rifampin (20 μg/ml) and gentamicin (100 μg/ml) plus fusidic acid (12 μg/ml) or tetracycline (5 μg/ml). The CFU counts were normalized to CFU per gram of cecum content.

Statistical analysis.

The data were analyzed with GraphPad Prism software version 8.3.0 using an unpaired, parametric, two-tailed t test. Student’s t test at a 95% confidence interval was employed for comparisons of the two groups. A P value of <0.05 was considered statistically significant.