Bacteroides fragilis requires the ferrous‐iron transporter FeoAB and the CobN‐like proteins BtuS1 and BtuS2 for assimilation of iron released from heme

Abstract The intestinal commensal and opportunistic anaerobic pathogen Bacteroides fragilis has an essential requirement for both heme and free iron to support growth in extraintestinal infections. In the absence of free iron, B. fragilis can utilize heme as the sole source of iron. However, the mechanisms to remove iron from heme are not completely understood. In this study, we show that the inner membrane ferrous iron transporter ∆feoAB mutant strain is no longer able to grow with heme as the sole source of iron. Genetic complementation with the feoAB gene operon completely restored growth. Our data indicate that iron is removed from heme in the periplasmic space, and the released iron is transported by the FeoAB system. Interestingly, when B. fragilis utilizes iron from heme, it releases heme‐derived porphyrins by a dechelatase activity which is upregulated under low iron conditions. This is supported by the findings showing that formation of heme‐derived porphyrins in the ∆feoAB mutant and the parent strain increased 30‐fold and fivefold (respectively) under low iron conditions compared to iron replete conditions. Moreover, the btuS1 btuS2 double‐mutant strain (lacking the predicted periplasmic, membrane anchored CobN‐like proteins) also showed growth defect with heme as the sole source of iron, suggesting that BtuS1 and BtuS2 are involved in heme‐iron assimilation. Though the dechelatase mechanism remains uncharacterized, assays performed in bacterial crude extracts show that BtuS1 and BtuS2 affect the regulation of the dechelatase‐specific activities in an iron‐dependent manner. These findings suggest that the mechanism to extract iron from heme in Bacteroides requires a group of proteins, which spans the periplasmic space to make iron available for cellular functions.

However, B. fragilis emerges as the most prevalent anaerobic organism in human infections (Finegold & George, 1989;Mazuski & Solomkin, 2009;McClean, Sheehan, & Harding, 1994;Park, Choi, Yong, Lee, & Kim, 2009;Smith, Rocha, & Paster, 2006). B. fragilis opportunistic infections occur as a consequence of a disruption in the integrity of the intestinal mucosa wall. Most of the bacteria leaked into the peritoneal cavity from the lumen is rapidly cleared by host defenses, but the anaerobe B. fragilis and facultative bacteria such as Escherichia coli often escape clearance and are the predominant organisms found in resulting abscess infections. As consequence, the majority of intraabdominal infections are of polymicrobial nature (Edmiston, Krepel, Seabrook, & Jochimsen, 2002;Johnson, Baldessarre, & Levison, 1997). Bacteroides fragilis accounts for 50%-90% of all anaerobes isolated from infections of which peritonitis, intraabdominal abscesses, and bacteremia are the predominant morbidities. In the event of a rupture in the abscess wall, B. fragilis may gain access to the bloodstream often leading to septic shock and systemic organ failure. Bacteroides fragilis is a deadly pathogen as it accounts for 5% of all positive blood cultures with mortality rates of 16%-45%. Despite the high incidence of B. fragilis in intraabdominal infections, the full range of virulence factors that allow it to arise as a predominant opportunistic anaerobic pathogen remain to be understood (Blairon et al., 2006;Brook, 1989Brook, , 2010Brook & Frazier, 2000;Cheng et al., 2009;Finegold & George, 1989;Mazuski & Solomkin, 2009;Nguyen et al., 2000;Park et al., 2009;Salonen, Eerola, & Meurman, 1998;Wilson & Limaye, 2004;Yoshino et al., 2012).
A factor that plays an important role in survival of B. fragilis in extraintestinal infection is its ability to acquire heme and inorganic iron from host tissues (Otto, van Dooren, Dozois, Luirink, & Oudega, 2002;Veeranagouda et al., 2014). Moreover, when B. fragilis is coinfected with E. coli strains producing hemoglobin protease during experimental polymicrobial infection, E. coli facilitates B. fragilis to utilize iron from heme to overcome host iron restriction mechanisms (Otto et al., 2002). Bacteroides fragilis as well as all other Bacteroides spp. have an essential requirement for heme and inorganic iron, and growth is stimulated by heme in a dose-dependent manner. In the absence of exogenous inorganic iron, B. fragilis is able to utilize heme as the source of iron in vitro (Chen & Wolin, 1981;Fuller & Caldwell, 1982;Rocha, de Uzeda, & Brock, 1991;Rocha & Smith, 2010;Sperry, Appleman, & Wilkins, 1977;Varel & Bryant, 1974). The growth stimulation of Bacteroides by heme was shown to be due to activation of the fumarate reductase complex involved in the reduction of fumarate to succinate for energy generation during glucose fermentation (Baughn & Malamy, 2003;Caldwell, White, Bryant, & Doetsch, 1965;Chen & Wolin, 1981;Harris & Reddy, 1977;Macy, Probst, & Gottschalk, 1975;Rocha & Smith, 2010;Sperry et al., 1977).
The heme-dependence of Bacteroides spp. is due to their inability to synthesize precursors of the macrocycle tetrapyrrole ring protoporphyrin IX (PpIX). This is a distinctive characteristic of host-associated Bacteroidetes in the genera Bacteroides, Prevotella, and Porphyromonas which are unable to synthesize their own PpIX due to the lack of most genes required for the formation of the tetrapyrrole macrocycle, though they can synthesize heme in vitro if PpIX and inorganic iron are supplied (Olczak, Simpson, Liu, & Genco, 2005;Rocha & Smith, 2010). Interestingly, this is a common characteristic of many bacterial species that colonize the lower intestinal tract of humans and other animals such as Bifidobacterium, Enterococcus, Lactococcus, Clostridium, and euryarchaeota (Rocha & Smith, 2010). In contrast, free living Bacteroidetes such as Flavobacterium, Cytophaga, and Salinibacter contain a complete heme biosynthesis pathway (Rocha & Smith, 2010). The Bacteroidetes can synthesize heme in vitro if PpIX and inorganic iron are provided, however, Bacteroides and Prevotella, do not contain homologs to the classical ferrochelatase HemH (PpfC) (Rocha & Smith, 2010; for review on heme biosynthesis in prokaryotes, see Dailey et al., 2017). This suggests that they utilize a novel mechanism to incorporate iron into PpIX for the synthesis of heme (Rocha & Smith, 2010).
Another aspect of heme metabolism in B. fragilis that has received little attention is its ability to dechelate divalent metalporphyrins. Support for the presence of a dechelatase mechanism in B. fragilis is provided by several lines of evidence demonstrating that Bacteroides/Prevotella are versatile in the removal of nonferrous divalent metal-bound porphyrins and side chain-modified porphyrins (such as Mn-PpIX, Mg-PpIX, Mn-mesoporphyrin IX, Mg-mesoporphyrin IX, Mn-deuteroporphyrin IX, or Mgdeuteroporphyrin IX). After dechelation, ferrous iron is inserted into the porphyrin free metal pocket through an unidentified ferrochelatase to form heme, mesoheme or deuteroheme, respectively.
These activities seem to be relevant for cellular functions because heme itself or side chain-modified heme (mesoheme and deuteroheme) can be incorporated into the cytochrome b-type of the fumarate reductase complex and they are equally functional physiologically (Caldwell et al., 1965;Fuller & Caldwell, 1982;Gardner, Fuller, & Caldwell, 1983;McCall & Caldwell, 1977).
This resourceful tetrapyrrole utilization appears to be important for the metabolism of heme in extraintestinal infections as well as during intestinal tract colonization. The formation of hemederived porphyrin in abscess pus is associated with the presence of anaerobic bacteria and the porphyrin pattern is remarkably similar to heme-derived porphyrin in the intestinal tract (Brazier, 1990). The exogenous source of heme in abscesses may be provided by extravasated red blood cells and polymorphnuclear leukocytes (Brazier, 1990). In the GI tract, the unabsorbed heme that reaches the lower intestinal tract together with nonpathological sources of luminal heme (including physiological epithelial shedding of cells and microbleeding) are converted by bacteria to a range of heme-derived porphyrins (Beukeveld et al., 1987;Rose et al., 1989;Young, Rose, St John, & Blake, 1990). The appearance and fluctuation of deuteroheme, mesoheme, and pemptoheme contents and conversion of hemederived porphyrins into deutero-meso-, and pempto-porphyrins in feces of healthy human subjects depends entirely and exclusively on the anaerobic bacterial flora (Beukeveld et al., 1987;Rocha & Smith, 2010;Rose et al., 1989;Young et al., 1989Young et al., , 1990. The aerobic and facultative anaerobic microbiota component of the intestinal tract plays no or a negligible role in the production of heme-derived porphyrins in the gut (Beukeveld et al., 1987).
However, there is a paucity of information on the mechanisms of heme-iron assimilation and its contribution to the formation of heme-derived porphyrins in B. fragilis.
Though progress has been made to establish that heme is a major source of iron for Bacteroides both in vitro and in vivo, the mechanisms involved in heme-iron acquisition have not been fully characterized. In addition, there is a paucity of information regarding the identification of the lower intestinal anaerobic bacteria species involved in heme dechelation and porphyrin side-chain modifications. In this study we show that when B. fragilis is grown on heme as the sole source of iron, the ability to remove iron from heme releasing free PpIX is regulated by iron availability. We provide evidence that iron is removed from heme extracytoplasmically and assimilation of iron released from heme to stimulate bacterial growth is dependent on the presence of the inner-membrane ferrous iron transporter system, FeoAB. Moreover, this study shows that BtuS, a member of the CobN-like family of proteins (Rodionov, Vitreschak, Mironov, & Gelfand, 2003), is involved in heme-iron assimilation.
Therefore, this study demonstrates that heme and iron metabolism in B. fragilis differs from the classical aerobic and facultative anaerobic bacterial systems.

| Bacterial strains and growth conditions
Bacteroides fragilis strains used in this study are shown in Table 1.
Strains were routinely grown anaerobically in brain heart infusion medium (BHIS) containing l-cysteine (1 g/L) and supplemented with hemin (5 mg/L) or otherwise stated in the text. After autoclaving, 20 ml of 10% NaHCO 3 per liter was added into the BHIS medium. For some experiments hemin was replaced with PpIX as delineated in the text. In an aqueous solution, it is not always possible to define completely the modified forms of heme (strictly ferrous iron-protoporphyrin IX) macrocycle due to the variation in the iron valence and the salts formed (see Smith, 1990 for review).
In addition, heme in an aqueous solution for an extended time can lead to formation of oxoμ-dimers of heme (Smith, 1990). Thus, we hereafter use the term heme to refer to an iron-protoporphyrin IX complex without specifically referring to its structural form or valence. Twenty μg/ml rifamycin, 100 μg gentamicin/ml, 5 μg tetracycline/ml, and 10 μg erythromycin/ml were added to the media when required. Media were supplemented with the ferrous iron chelator bathophenanthroline disulfonic acid (BPS), which does not enter the cell, to obtain iron-limiting conditions as previously described (Rocha & Krykunivsky, 2017). Addition of ferrous iron sulfate or ammonium ferrous iron sulfate was used to obtain ironreplete conditions. In some experiments, defined medium (DM) was used as described previously (Rocha & Krykunivsky, 2017) to determine the stimulatory effect of heme and PpIX on the bacterial growth rate.

| Construction of btuS1 and btuS2 insertional mutants
A 528 nt fragment was amplified from the N-terminus region of B. fragilis 638R_2505 (btuS1) gene using primers GAGGCGGATCCTGCCGCATCG, and CGAATGAGCTCCAAGTCTT CC containing modified nucleotides (bold font) and restriction sites (underlined) for BamHI and SstI, respectively. The amplified fragment was cloned into the BamHI/SstI sites of the suicide vector pFD516 (Smith, Rollins, & Parker, 1995

| Construction of btuS1 btuS2 double insertional mutant
A 1,174 nt fragment was amplified from the N-terminus region of B. fragilis 638R_2718 (btuS2) gene using primers GTTGTGTGGATCCGGCAATACTCG, and CTCTCCGGAAGCTT TTCTACCCGG containing modified nucleotides (bold font) and restriction sites (underlined) for BamHI and HindIII, respectively. The amplified fragment was cloned into the BamHI/HindIII sites of the suicide vector pYT102 (Baughn & Malamy, 2002

| Determination of total heme and total porphyrin in whole cell dry extract
We used a modified spectrophotometric method described by Kufner, Schelegel, and Jager (2005) et al., 2005) and normalized to nmol/mg dry weight. Total heme content was determined by pyridine-NaOH as described previously (Rocha & Smith, 1995). Bovine hemin (Sigma) was used as the standard.

| Reverse phase HPLC analysis of porphyrin extract
Acid extract fractions obtained as described above were filtered in a 0.

| The role of feoAB in B. fragilis growth with heme as the sole source of iron
When B. fragilis strains were grown in DM supplemented with heme in the presence of the ferrous iron chelator BPS, the wild-type parent strain was able to grow in the presence of heme as the only source of iron. In contrast, the growth of the ∆feoAB strain was abolished indicating that the ∆feoAB strain could no longer obtain essential iron from heme (Figure 1a,b). Partial growth of the ∆feoAB occurred in the presence of 5 μg and 10 μg heme/ml compared to maximum growth rate observed for the parent stain in media where residual iron was not chelated (Figure 1a,b). When PpIX replaced heme as the tetrapyrrole macrocycle, neither the parent nor the ∆feoAB mutant strains were able to grow in media that was iron-limited by the addition of BPS (Figure 1c,d). In the absence of BPS, the parent but not the ∆feoAB strain, grew in basal medium containing PpIX without added iron, indicating that residual iron was sufficient to support the growth of the parent strain in the presence of PpIX (Figure 1c,d).
The addition of 2 μM FeSO 4 did not significantly stimulate the growth of the ∆feoAB strain in the presence of either heme or PpIX, but addition of 100 μM FeSO 4 restored growth to wild-type levels ( Figure 1b,d). No growth of either strain occurred in the absence of added heme or PpIX (Figure 1a,b,  During the course of this investigation it was found that B. fragilis cultures grown in the presence of heme and BPS, but not under iron-replete conditions, produced porphyrin-like pigment when exposed to longwave-UV light (Supporting Information Figure S1).
This observation was further analyzed by culturing strains on BHIS plates containing high amount of heme at 100 μg/ml, supplemented with either 100 μM ammonium ferrous sulfate, 300 μM BPS, or 1 mM BPS. After 7 days of incubation, the plates were exposed to longwave UV light at 365 nm ( Figure 2). There was no inhibition of growth or of fluorescent porphyrin production by the parent, ∆furA, ∆feoAB, ∆feoAB ∆furA, and ∆feoAB/feoAB + strains on ironreplete culture plates (Figure 2a). In contrast, under iron-limiting conditions, porphyrin fluorescence was clearly seen in the parent, ∆furA and ∆feoAB/feoAB + strains (Figure 2b,c). However, the growth of the ∆feoAB and ∆feoAB ∆furA double-mutant strains were inhibited under iron-limiting conditions, confirming that the formation of heme-derived porphyrins was directly linked to heme as a source of iron. Genetic complementation of the ∆feoAB mutation with feoAB gene completely restored growth when heme was the sole source of iron. Therefore, this confirms that iron is likely being removed from heme in the periplasmic space since the FeoAB system is required for its assimilation.
F I G U R E 2 Growth of Bacteroides fragilis strains on brain heart infusion (BHIS) media containing 100 μg/ml hemin. Plates were supplemented with (a) 100 μM FeSO 4 . (b) 300 μM bathophenanthroline disulfonic acid (BPS) and (c) 1 mM BPS. Bacteria were grown in an anaerobic chamber at 37°C for 7 days. Plates were illuminated with a 365 nm UV-long wave lamp (UVP; model UVLS-28, Upland, CA) and pictures were taken with an Olympus Camedia C-4000 digital camera. WT: B. fragilis 638R wild type. Strain designations are depicted on each panel F I G U R E 3 Total heme (a) and total porphyrin (b) determination in dried whole cell extracts of Bacteroides fragilis strains grown in brain heart infusion (BHIS) media containing 100 μg/ml hemin and supplemented with 100 μM (NH 4 ) 2 Fe(SO 4 ) 2 (Fe), 250 μM bathophenanthroline disulfonic acid (BPS) or 1 mM BPS. Extraction of heme (He) and protoporphyrins (Pp) from dried cells was performed by organic/acid aqueous phase separations as described in the materials and methods section. Growth of the ∆feoAB and ∆furA ∆feoAB strains was inhibited by 1 mM BPS. The data presented are the mean of five independent experiments. Standard deviation bars represent deviance (±) of the mean

| Determination of total heme and hemederived porphyrins in whole cell extracts
When the 638R parent and ∆furA mutant strains were grown in medium containing 100 μg/ml heme and 100 μM ammonium ferrous sulfate, the total cellular heme amount was found to be approximately 1.2 nmol/mg and 2.2 nmol/mg of dry weight, respectively. In the presence of 1 mM BPS, the amount of These findings clearly show that accumulation of heme-derived porphyrin occurs in the presence of heme when free iron is not available. They also demonstrate that B. fragilis is able to remove iron from heme releasing free-protoporphyrin in a Fur-independent manner, indicating the presence of a yet to be identified iron-regulated heme dechelatase mechanism.

| Iron inhibits the formation of the hemederived porphyrins
To confirm that excess iron had an inhibitory effect on the production of heme-derived porphyrin, B. fragilis strains were spread on BHIS agar supplemented with 100 μg/ml heme and 1 mM BPS.
Then a filter paper disk was placed on top of the agar and 10 μl Nonetheless, these findings clearly show that B. fragilis possesses a dechelatase mechanism which is regulated by inorganic iron availability, and we also show that removal of iron from heme does not break the tetrapyrrole macrocycle structure hence releasing free PpIX. Moreover, these findings demonstrate that B. fragilis is able to modify the PpIX side-chain. No detectable traces of porphyrins were found in uninoculated culture media (Supporting Information Figure S2). ChlH required for the bacteriochlorophyll biosynthesis (Rodionov et al., 2003). The BtuS family of proteins (pfam02514) contains a domain common to the CobN protein and to the magnesium protoporphyrin chelatase (https://www.ncbi.nlm.nih.gov/Structure/ cdd/cddsrv.cgi?uid=pfam02514). Though BtuS is predicted to be involved in the salvage of metalloporphyrins rather than cobalamine biosynthesis (Rodionov et al., 2003), its role in anaerobic heme as-  Figure S4). Most Bacteroides species contain a single homolog of btuS with the exception of nearly all B. fragilis strains which carry both the btuS1 and the btuS2 homologs.
These findings suggest that BtuS1 and BtuS2 might be involved in removing iron from heme in the periplasmic space. This correlates nicely with the findings shown above demonstrating that the inner membrane transporter FeoAB is required for the assimilation of iron released from heme outside of the cytoplasm. Therefore, btuS1 and btuS2 genes were chosen for further characterization.

| Role of the CobN-like proteins BtuS1 and BtuS2 during growth with heme as the sole source of iron
To test the role of BtuS proteins in the utilization of heme-iron, bacteria were grown in DM containing 5 μg/ml heme and supplemented with 100 μM ammonium ferrous sulfate or 500 μM BPS. In the presence of heme and inorganic iron replete conditions, no growth defect was observed for the btuS1, btuS2 or btuS1 btuS2 double mutant strains compared to the parent strain ( Figure 6a). However, when bacteria were grown with heme alone as a sole source of iron, the btuS1 btuS2 double mutant showed a growth defect compared to the parent and single mutant strains (Figure 6b). The btuS1 single mutant did not show a significant growth defect while btuS2 mutant showed a partial decrease in growth rate compared to parent strain.
Growth of complemented strains was restored in part compared to the parent strain. The lack of complete complementation in the mutant strain may have been due to a polar mutation effect disrupting the expression of downstream genes which appear to be organized in a polycistronic operon. Moreover, genetic complementation carried out with single btuS1 or btuS2 genes in a multicopy plasmid may also have contributed to a pleiotropic effect. Nonetheless, this indicates that BtuS proteins are involved in acquisition of heme-iron.
However, to our surprise, the btuS1 btuS2 double mutant strain did not abolish the ability of B. fragilis to remove iron from heme releasing free PpIX ( Figure 5). In fact, the relative intensity of the PpIX peak in the btuS1 btuS2 double mutant increased approximately twofold compared to the parent strain, as determined by reverse-phase HPLC analysis. This suggests that BtuS1 and BtuS2 had no or little dechelatase enzymatic activity, but in contrast, their disruption increased formation of free PpIX. This is intriguing because formation of free PpIX in the btuS1 btuS2 double mutant indicates that the iron released from heme does not appear to be efficiently assimilated to support growth as addition of exogenous inorganic iron can restore the growth defect phenotype. Overall these findings demonstrate that the ability to dechelate heme is upregulated by iron limitation and that BtuS1 and BtuS2 proteins seem to have a negative regulatory effect. However, in the presence of PpIX, the negative regulatory effect is higher under F I G U R E 6 Growth of Bacteroides fragilis strains in defined medium containing 5 μg heme/ml (He). Media was supplemented with 100 μM ammonium ferrous sulfate (Fe) (a) or 400 μM bathophenanthroline disulfonic acid (BPS) (b) to obtain iron replete and iron-limiting conditions respectively. ♦ B. fragilis 638R wild type; □ BER-107 (btuS1); ▲ BER-109 (btuS2); ■ BER-111 (btuS1 btuS2); ○ BER-114 (btuS1 btuS2 btuS1 + ); • BER-115 (btuS1 btuS2 btuS2 + ). The growth curves for BER-114 and BER-1115 strains are only shown in (b). Data presented are an average of two determinations in duplicate F I G U R E 7 Ferrochelatase (FeCH) and reverse ferrochelatase activity assays in crude extracts of Bacteroides fragilis 638R (WT) and btuS1 btuS2 double-mutant strains. Bacteria were grown in brain heart infusion (BHIS) medium containing 10 μg heme/ml (He) or 10 μg protoporphyrin IX (PpIX)/ml. Media was supplemented with 100 μM ammonium ferrous sulfate (Fe) or 500 μM bathophenanthroline disulfonic acid (BPS). Bacteria were grown anaerobically at 37°C for 24-48 hr under iron replete conditions and for 72-96 hr under ironlimiting conditions. Details of the forward ferrochelatase and reverse ferrochelatase reaction assay settings are described in the material and methods section iron replete conditions than under iron restricted conditions. This negative effect is also observed for the forward chelatase activity under iron replete conditions but in contrast to the reverse activity, the forward chelatase activity it is not regulated by iron restriction.

| BtuS modulates dechelatase and ferrochelatase activities in vitro
Though the biochemical and genetic control mechanisms for the forward and reverse chelatase activities remain to be characterized, we show in this study that the BtuS1 and BtuS2 participate in the negative regulation of the forward and reverse chelatase activities of the heme metabolism in B. fragilis. In many bacteria, heme homeostasis is controlled at the initial steps of the biosynthetic pathway by HemX, a transmembrane anchored protein that regulates the abundance of glutamyl-tRNA reductase, GtrR (Choby et al., 2018).
In other bacteria such as the alpha-proteobacteria, control of heme biosynthesis occurs through the direct interaction of the Irr regulator with the ferrochelatase to modulate gene expression under ironlimiting conditions (Small, Puri, & O'Brian, 2009).
Therefore, we believe that the mechanism to extract iron from heme and make it available for cellular functions in Bacteroides requires a group of proteins acting together, possibly forming a complex, which spans the periplasmic space. The BtuS1 and BtuS2 proteins described in this study are not the only components of this hypothetical protein machinery. At a minimum it also includes a dechelatase enzymatic activity, which is still undiscovered (Figure 8).

| D ISCUSS I ON
In this study, we show that the ability of B. fragilis to remove iron from heme releasing free PpIX is dependent on an unidentified dechelatase activity mechanism that is upregulated when bacteria are present in an iron-limiting environment. Moreover, we demonstrate that utilization of iron removed from heme requires the inner membrane ferrous iron transporter system, FeoAB, indicating that the dechelatase activity occurs in the periplasmic space. It is well documented that members of the Bacteroidetes phylum such as Porphyromonas and Prevotella are able to release iron and form free PpIX from heme (Brazier, 1986;Fyrestam, Bjurshammar, Paulsson, Johannsen, & Östman, 2015;Fyrestam et al., 2017;Shah, Bonnett, Mateen, & Williams, 1979;Slots & Reynolds, 1982;Smalley & Olczak, 2017;Soukos et al., 2005). The removal of iron from heme in Prevotella melaninogenica (formerly black-pigmented Bacteroides melaninogenicus) occurs by the action of an unidentified demetallase, which was elegantly demonstrated by Shah et al. (1979). Other studies have suggested that the ability of P. gingivalis to remove iron from heme could be due to the reverse activity of its ferrochelatase HemH (Olczak et al., 2005). It has also been suggested that the P. gingivalis outer membrane cobalt chelatase CbiK-like protein IhtB may remove iron from heme extracellularly (Dashper et al., 2000). However, the lack of ferrochelatase HemH in B. fragilis (Rocha & Smith, 2010) (Rocha & Krykunivsky, 2017). Secondly, the removal of iron from heme by dechelatase activity, modulated by BtuS1 and BtuS2 proteins, is regulated by iron availability in a Fur-independent manner. The transport of ferrous iron in intestinal anaerobes is still poorly understood. In aerobic and facultative anaerobic bacteria the Feo system is generally composed of the small cytoplasmic protein FeoA and the large transmembrane protein FeoB.
This stands in contrast to many upper and lower GI tract anaerobes such as Porphyromonas, Bacteroides Eubacterium, Ruminococcus, and Clostridium species which contain the feoA and feoB genes encoded in a single fused polypeptide (Dashper et al., 2005;Rocha & Smith, 2010;Veeranagouda et al., 2014). The translated interpeptide amino acid residues between the FeoA and FeoB domains are highly conserved among the Bacteroidetes, and among the Firmicutes and Actinobacteria phyla of intestinal colonizers (Rocha & Smith, 2010).
As far we are aware, fused FeoAB peptides are not found in nongastrointestinal organisms suggesting its unique importance for the physiology of intestinal anaerobic bacteria (Rocha & Smith, 2010).
We have previously shown that the lack of FeoAB impaired B. fragilis to form abscesses in a mouse model of infection (Veeranagouda et al., 2014). However, in light of this study, this growth deficiency is likely to have been due to the ∆feoAB strain inability to acquire both heme-iron and inorganic iron from host tissues. However, the fact that the iron released from heme in the periplasm seems to be less efficiently assimilated to support growth of the btuS1 btuS2 double-mutant, it points out a potential role for the BtuS1 and BtuS2 system in facilitating its utilization. It may be energetically cost effective to remove iron from heme or modify the porphyrin side-chain in the periplasmic space rather than transporting porphyrins across cellular membranes for export. It is possible that this mechanism might facilitate rapid exchange of porphyrins and side chain-modified porphyrins among intestinal bacteria, especially in view of the fact that B. fragilis can incorporate side chain-modified heme into cytochrome b-type (for review see Rocha & Smith, 2010).
Recent studies have demonstrated that the oral pathogens P. gingivalis, Aggregatibacter actinomycetemcomitans, Prevotella intermedia, Prevotella nigrescens, and P. melaninogenica produce endogenous porphyrins when grown on media containing animal blood (Fyrestam et al., 2015;Soukos et al., 2005). The Porphyromonas and Prevotella porphyrins were mostly PpIX, coproporphyrinogen I, and coproporphyrinogen III (Fyrestam et al., 2015(Fyrestam et al., , 2017Soukos et al., 2005). However, no side chain-modified porphyrin such as mesoporphyrin IX or deuteroporphyrin IX were reported to be formed by these oral bacteria (Fyrestam et al., 2015(Fyrestam et al., , 2017Soukos et al., 2005). Therefore, the presence of PpIX, mesoporphyrin IX and deuteroporphyrin IX found in B. fragilis suggests that this organism contributes to the appearance of heme-derived porphyrins and side chain-modified porphyrins in extraintestinal infections as well as in the intestinal tract. This is in agreement with previous studies demonstrating that intestinal anaerobic bacteria are exclusively responsible for the conversion of heme to PpIX lacking iron, and modifications of the vinyl side chains of heme and PpIX to their respective deutero, meso, and pempto side chain forms (Beukeveld et al., 1987;Young et al., 1989Young et al., , 1990. There is a wide range of variations in the amount and type of heme-derived porphyrins among individuals, but the types of heme-derived porphyrins seem to be consistent within a given individual (Young et al., 1990). Though fecal flora metabolism and redox potential may account for the differences in the amount and types of heme-derived porphyrins (Young et al., 1990), we demonstrate in this study that intestinal iron limitations may also contribute to the fluctuations in the amount of heme-derived porphyrins formed by intestinal anaerobic bacteria.
Though heme has been shown to be inhibitory to many Grampositive and Gram-negative bacteria in vitro (Nitzan, Wexler, & Finegold, 1994), dietary heme has a robust effect in enhancing the abundance of Bacteroidetes relative to the Firmicutes population in mouse intestinal colonization models (Ijssennagger et al., 2012(Ijssennagger et al., , 2015. In addition to this effect on bacterial colonization, hemeinduced hyperproliferation and hyperplasia in the mouse colon only occurs in the presence of the gut microbiota (Ijssennagger et al., 2015). The specific bacterial species involved in such mechanisms have not been identified, and whether there is a link between bacterial action on modifications of heme structure and heme-derived porphyrins in enhancing or weakening heme mucosa toxicity remains to be determined. In this study we identify B. fragilis as being one of these bacteria able to act on and modify labile heme molecules in vitro. Because formation of heme-derived porphyrin seems to be regulated by free iron availability, we think that heme-derived porphyrins might be relevant for B. fragilis pathophysiology in extraintestinal infections where iron is limited by the host. This assumption is based on the fact that protoporphyrin IX acts as a competitive inhibitor of the proinflammatory activity of labile heme in macrophages (Figueiredo et al., 2007;Soares & Bozza, 2016). Moreover, we speculate that by modifying or scavenging labile heme in the site of infection, B. fragilis may alter the signaling effect of labile heme on recruitment of neutrophil and oxidative burst (Figueiredo et al., 2007;Soares & Bozza, 2016). Further investigation on heme-iron utilization systems and regulation will advance our understanding on the role heme plays in Bacteroides pathophysiology.

ACK N OWLED G M ENTS
This work was supported in part by an NIH National Institute of Allergy and Infectious Diseases grant number AI125921 to ERR. The chelatase and dechelatase assays were performed at the Iron and Heme Core facility at the University of Utah, supported in part by a grant from the NIH National Institute of Diabetes and Digestive and Kidney Diseases, Grant number U54DK110858. We thank Harry A.
Dailey, University of Georgia, Athens, for helpful discussions. We are grateful to Samantha Palethorpe for critical reading of the manuscript.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.