Staphylococcus aureus is a pathogen that infects multiple anatomical sites leading to a diverse array of diseases. Although vertebrates can restrict the growth of invading pathogens by sequestering iron within haem, S. aureus surmounts this challenge by employing high-affinity haem uptake systems. However, the presence of excess haem is highly toxic, necessitating tight regulation of haem levels. To overcome haem stress, S. aureus expresses the detoxification system HrtAB. In this work, a transposon screen was performed in the background of a haem-susceptible, HrtAB-deficient S. aureus strain to identify the substrate transported by this putative pump and the source of haem toxicity. While a recent report indicates that HrtAB exports haem itself, the haem-resistant mutants uncovered by the transposon selection enabled us to elucidate the cellular factors contributing to haem toxicity. All mutants identified in this screen inactivated the menaquinone (MK) biosynthesis pathway. Deletion of the final steps of this pathway revealed that quinone molecules localizing to the cell membrane potentiate haem-associated superoxide production and subsequent oxidative damage. These data suggest a model in which membrane-associated haem and quinone molecules form a redox cycle that continuously generates semiquinones and reduced haem, both of which react with atmospheric oxygen to produce superoxide.
Staphylococcus aureus is an important human pathogen
Staphylococcus aureus is a Gram-positive coccus that is ubiquitous among the human population owing to the fact that the skin and anterior nares of more than one-quarter of all individuals are asymptomatically colonized by this organism (Wertheim et al., 2004; DeLeo et al., 2010). However, S. aureus is armed with an arsenal of toxins, host-modulating factors and other virulence determinants that enable this bacterium to infect virtually every organ and cause numerous diseases including pneumonia, endocarditis, toxic shock syndrome and sepsis (Gordon and Lowy, 2008). Because of these characteristics, staphylococcal infections can be either hospital- or community-acquired and result in high rates of mortality and morbidity (Wertheim et al., 2004; Klevens et al., 2007; DeLeo et al., 2010). Compounding this problem is the widespread existence of antibiotic-resistant S. aureus, necessitating the identification of new targets for therapeutic intervention.
S. aureus elaborates haem acquisition systems to overcome host-induced nutrient limitation
During infection, vertebrates sequester essential nutrients from invading pathogens in order to restrict microbial growth in a defence tactic known as nutritional immunity (Kehl-Fie and Skaar, 2010; Wakeman and Skaar, 2012). Bioavailable iron concentrations within the host are particularly limiting due to a number of factors: a physiological pH that results in low iron solubility, storage of iron within iron binding proteins and intracellular localization of iron (Bullen and Griffiths, 1999; Crichton, 2001). The net result of these mechanisms creates an environment virtually devoid of free iron with greater than 90% of iron residing within host cells. The bulk of host iron is contained within the tetrapyrrole ring of haem, which is further bound within the protein haemoglobin and sequestered inside erythrocytes (Hammer and Skaar, 2011). In order to gain access to this abundant iron source, S. aureus secretes haemolysins to lyse erythrocytes, liberating free haemoglobin. S. aureus is able to acquire haem iron from haemoglobin using the iron-regulated surface determinant (Isd) system (Skaar et al., 2004; Reniere et al., 2007).
Haem and menaquinone (MK) are required for S. aureus respiration
Haem biosynthesis and acquisition are important to the survival of S. aureus. In addition to being the most abundant iron source available within vertebrates, haem is also utilized as a cofactor in cytochromes and is thus required for bacterial respiration. During respiration, electrons are passed between various carriers to a terminal electron acceptor, which is oxygen in the case of aerobic respiration. The function of respiration is to create a proton motive force (PMF) that can be harnessed to generate ATP. Therefore, much of the electron transport cascade is carried out at the membrane. Electrons are extracted during the catabolism of carbon sources and passed along the electron transport chain via diffusible electron carriers such as NADH to membrane-associated quinones (Simon et al., 2008). While multiple types of quinone molecules exist in nature, the only quinone known to be utilized for respiration in S. aureus is MK (Bentley and Meganathan, 1982). Quinones subsequently donate electrons to the haem molecules located within cytochromes. S. aureus, like the Gram-positive model organism Bacillus subtilis, possesses two classes of these haem-containing protein complexes, the cytochrome bd and cytochrome aa3 quinol oxidases (Zamboni and Sauer, 2003).
Various genetic mutations can yield bacterial strains that are deficient in respiration. These strains exhibit a slowed growth phenotype and are thus known as small colony variants (SCVs). This type of mutation is of clinical interest because SCVs of S. aureus and various other pathogens are commonly isolated from patients experiencing chronic infections (Proctor et al., 2006). The respiration deficiency of SCV strains provides increased resistance to aminoglycoside antibiotics because this therapeutic class requires a membrane potential in order to enter the cell (Baumert et al., 2002). The most common disease-associated SCVs are defective in either MK or haem biosynthesis (Proctor et al., 2006).
Excess haem is toxic to bacteria
Despite the beneficial aspects of haem, this molecule becomes extremely toxic to S. aureus and other organisms when present in excess (Anzaldi and Skaar, 2010). The nature of this toxicity is not fully understood and possibly multifaceted. However, it is likely that this phenomenon is, at least in part, the result of cellular damage due to the reactivity of the iron atom within haem. Previous studies have demonstrated that haem can induce oxidative damage to DNA and proteins in vitro (Aft and Mueller, 1983; 1984). Haem-induced DNA damage has also been observed in vivo in both S. aureus and Escherichia coli (Nir et al., 1991). Additionally, recent studies have demonstrated that, while invading pathogens can utilize haem-containing host proteins such as haemoglobin as nutrient sources, the host can also employ these proteins as a form of innate immunity to generate reactive oxygen species (Jiang et al., 2007; Du et al., 2010). Interestingly, Gram-positive microorganisms tend to be more sensitive to haem toxicity than Gram-negative bacteria (Nitzan et al., 1994; Anzaldi and Skaar, 2010). S. aureus in particular is known to be exquisitely sensitive to haem stress (Ladan et al., 1993). This observation, combined with the fact that the hallmark changes in protein expression associated with oxidative damage (Gaupp et al., 2012) are absent in haem-exposed S. aureus, suggests that additional mechanisms of haem toxicity may exist in this pathogen (Friedman et al., 2006).
Haem toxicity can be overcome in S. aureus
To cope with haem stress, S. aureus expresses a putative efflux pump known as HrtAB (Torres et al., 2007; Stauff et al., 2008). HrtAB was identified by a proteomics analysis screening for staphylococcal proteins regulated in response to variations in the concentration of exogenous haem (Friedman et al., 2006). The two genes most highly upregulated in response to increasing levels of haem were named haem-regulated transport proteins HrtA and HrtB. hrtA and hrtB are located in an operon and encode a putative ATPase and permease respectively. The expression of the HrtAB putative efflux pump is positively regulated by a two-component system known as the haem sensor system HssRS (Stauff et al., 2007; Torres et al., 2007). Deletion of either hssRS or hrtAB results in increased haem sensitivity (Stauff et al., 2007; Torres et al., 2007; Stauff and Skaar, 2009). Orthologues of this haem detoxification system have been identified in a number of other Gram-positive microorganisms including Bacillus anthracis, Enterococcus faecalis, Corynebacterium diphtheriae, Lactococcus lactis, Listeria monocytogenes and Streptococcus agalactiae (Torres et al., 2007; Bibb and Schmitt, 2010; Fernandez et al., 2010; Lechardeur et al., 2012), indicating that the need to resist haem stress is widespread among bacteria. Work performed in L. lactis indicates that the HrtAB putative efflux pump detoxifies haem by exporting excess haem from the cell (Pedersen et al., 2008; Lechardeur et al., 2012), a mechanism that is likely employed by the HrtAB homologues of other microorganisms.
A genetic screen implicates MK biosynthesis in contributing to S. aureus haem stress
In an effort to uncover gene products required to facilitate haem toxicity, a ΔhrtA S. aureus strain that is highly susceptible to haem stress was subjected to random transposon mutagenesis via transformation with the pTV1 vector carrying the Tn917 transposon (Youngman et al., 1983). Haem-resistant mutants were selected based on growth upon medium infused with 30–50 μM haem. The resulting mutants were analysed to identify the site of the transposon insertion, and all of the identified sites mapped to genes encoding products that are either members of the MK biosynthesis pathway or are responsible for the production of substrates that feed into this pathway (Fig. 1). The exact sites of the transposon insertions and the operonic arrangements of the genes within the MK pathway can be visualized in Fig. S1. Mutants were isolated containing disruptions of aroB, aroK and aroC. These genes are part of the phenylalanine, tyrosine and tryptophan biosynthetic pathway and encode 3-dehydroquinate synthase, shikimate kinase and chorismate synthase respectively. AroB, AroK and AroC are required for the production of chorismate, the first precursor in the MK biosynthesis pathway. Additional mutants were isolated with insertions in menC (encoding 2-succinylbenzoate synthase), menE (encoding 2-succinylbenzoate-CoA ligase), menB (encoding dihydroxynaphthoic acid synthetase) and an unnamed gene denoted as NWMN_1381. NWMN_1381 is located within an operon containing the MK biosynthetic genes, ubiE and hepT. UbiE is predicted to be an S-adenosylmethionine:2-DMK methyltransferase responsible for catalysing the final step of MK biosynthesis in which a methyl group is attached to demethylmenaquinone (DMK) (Wissenbach et al., 1992). NWMN_1381 and HepT are homologues of the heptaprenyl diphosphate synthase subunits of B. subtilis responsible for the production of the isoprenyl lipid chain that allows MK to be inserted into the cell membrane (Zhang et al., 1997; Leatherbarrow et al., 1998). All of these gene products are predicted to be directly involved in MK biosynthesis, establishing the putative MK biosynthetic pathway of S. aureus. These data suggest that a component of MK biosynthesis contributes to haem stress in S. aureus.
Biochemical analyses of non-polar mutations in the MK biosynthesis pathway yield insight into previously uncharacterized aspects of S. aureus respiration
In order to assess the impact of MK biosynthesis on haem stress, we generated non-polar deletions in the genes encoding the final three annotated enzymes of this pathway: menB, menA and ubiE. These genes were deleted in both the background of wild-type (WT) S. aureus and the ΔhrtB S. aureus strain lacking a functional HrtAB system. Although the transposon screen was performed in a ΔhrtA background, the non-polar deletions were constructed in the ΔhrtB background because ΔhrtA cells have a defect in membrane stability due to dysregulated HrtB permease function, which is unrelated to the role that HrtA plays in protection against haem stress (Attia et al., 2010). If the MK biosynthetic pathway predicted by the genomic screen is correct, deletion of menB should result in a blockage of MK biosynthesis at the formation of 2-succinylbenzoyl-CoA, deletion of menA should yield 1,4-dihydroxy-2-naphthoate as the final MK biosynthetic product and deletion of ubiE should result in a build-up of DMK (Fig. 1). As expected, all strains containing deletions in menA and menB possessed a SCV phenotype when streaked onto agar plates, indicating that respiration was no longer occurring in these mutants (Fig. 2A). The respiration deficiency of these SCV strains was further confirmed by their inability to generate a PMF or to oxidize NADH to the levels of their MK-producing parental strains (Figs 3A and S2). Because SCV strains cannot respire, they resort to energy production via fermentation which results in an accumulation of l- and d-lactate (Fig. 3B). Of note, several disease-associated SCV isolates were recently found to have mutations mapping specifically to the menB locus (Lannergard et al., 2008). Interestingly, the ΔubiE strains, unlike the ΔmenA and ΔmenB strains, maintained normal colony morphology (Fig. 2A), indicating that both DMK and MK can function within the electron transport chain of S. aureus. This finding is not entirely surprising given that DMK possesses a redox potential midway between that of MK and ubiquinone and this molecule has previously been shown to function as an electron carrier in other bacteria (Holländer, 1976).
The respiration deficiency of the menB mutant can be rescued through the addition of menadione, a synthetic analogue of 1,4-dihydroxy-2-naphthoate (Fig. 2A). The SCV phenotype of menA mutants cannot be reversed by the presence of exogenous menadione but can be chemically complemented by the addition of MK4, a commercially available MK possessing a lipid tail comprised of four isoprenyl groups (Fig. 2A). These observations support the annotated function of menA as being the enzyme responsible for the addition of the polyprenyl lipid tail to the quinone molecule.
Mutants of the MK biosynthesis pathway were further verified to be deficient in MK production through the use of thin layer chromatography (TLC) and mass spectrometry. MK extracts from WT, ΔubiE, ΔmenA and ΔmenB were resolved adjacent to MK4 on TLC plates (Fig. 2B). The MK band from WT cells resolved at a slightly higher position on the plate than MK4, indicating that the polyprenyl tail of MK from S. aureus is longer than that of the commercially available MK. The DMK band observed in the ΔubiE extract resolved at a slightly lower position than the WT MK band possibly due to the lack of a methyl group in this molecule. No MK or DMK band could be observed in the ΔmenA or ΔmenB extracts.
In order to prove that the bands observed by TLC were MK and DMK and to determine the number of prenyl groups comprising the lipid tails of these molecules, the extracts were also analysed by mass spectrometry (Fig. 2C). Mass spectrometry of WT extracts revealed that the predominant MK species of these cells possessed eight isoprenyl groups (MK8) but MK with seven isoprenyl groups (MK7) and nine isoprenyl groups (MK9) were also present. These results support prior findings which indicated that MK8 is the major MK produced by S. aureus during aerobic growth (Jeffries et al., 1967; Hammond and White, 1969). Mass spectrometry of ΔubiE extracts demonstrates that these cells are incapable of producing MK and instead accumulate DMK. The bulk of this DMK has eight isoprenyl groups (DMK8) but DMK7 and DMK9 isoprenologues can also be detected. These results establish the S-adenosylmethionine:2-DMK methyltransferase function of UbiE in S. aureus.
Growth analyses confirm that MK biosynthesis is required to potentiate haem toxicity
To verify that disruption of MK biosynthesis protects S. aureus from haem toxicity, WT and mutant S. aureus strains were streaked onto agar plates in the absence or presence of 40 μM haem to mimic the conditions of the transposon mutagenesis selection (Fig. 4A). As expected, no growth was observed for the ΔhrtB strain upon exposure to high levels of exogenous haem. Haem toxicity was not eliminated in a ΔhrtB strain lacking ubiE, the gene encoding for the final enzyme in MK biosynthesis. However, ΔhrtB strains that were also deficient in menA and menB grew as well as the ΔmenA and ΔmenB strains with functional HrtAB systems, indicating that the haem toxicity associated with the lack of HrtAB could be completely reversed by eliminating biosynthetic steps prior to the formation of DMK. When respiration was restored in the ΔmenA and ΔmenB strains through the addition of MK4 to the growth media, the ΔhrtB ΔmenA and ΔhrtB ΔmenB mutants became susceptible to haem stress.
The fact that non-polar deletion of ubiE does not counter haem toxicity indicates that the protective effect of the transposon insertion disrupting the ubiE-containing operon results from the deletion of the NWMN_1381 and hepT polyprenyl biosynthetic genes. Deletion of NWMN_1381 and hepT yields a blockage in MK biosynthesis equivalent to the disruption of menA, the enzyme responsible for attaching the polyprenyl chain to the quinone molecule (Fig. 1).
We next sought to determine the effects of MK deficiency on haem toxicity in liquid culture (Figs 4B–E, S3A and B). Because haem toxicity increases in liquid media, these cells were grown in the presence or absence of 6 μM haem as compared with the 40 μM haem used in the agar plate assay. After 8 h of haem exposure, WT haem-treated cells grew almost as well as untreated cells (Fig. 4C). This result was in marked contrast to the haem-treated ΔhrtB cells which displayed virtually no growth. This haem-induced growth inhibition was almost completely reversed in the ΔhrtB ΔmenA and the ΔhrtB ΔmenB double mutant strains. However, as observed on solid medium, it was evident that the ΔhrtB ΔubiE mutant was as susceptible to haem stress as the ΔhrtB strain, indicating that strains containing DMK experience as much haem toxicity as strains producing MK. The protective effect of ΔmenA and ΔmenB deletion against haem toxicity was completely reversed upon restoration of aerobic respiration with the addition of MK4 to the growth medium (Figs 4D, E and S3B). Since only respiration-deficient MK biosynthesis mutants are protected against haem stress, we sought to test the possibility that any disruption of aerobic respiration would counteract haem toxicity.
In order to assess the contribution of aerobic respiration to haem-induced stress, we tested haem toxicity in a strain lacking both of the S. aureus cytochromes. This strain has a non-polar deletion of the qoxB gene and a transposon insertion in the cydB gene, which encode subunits of cytochrome aa3 and cytochrome bd respectively. The ΔqoxB ΔcydB strain is completely deficient in aerobic respiration (N.D. Hammer, unpublished). Because the ΔqoxB ΔcydB strain possesses an intact HrtAB haem detoxification system, we compared its haem susceptibility with that of WT S. aureus and the ΔmenB mutant. The growth of these strains was assessed in the presence or absence of 10 μM haem in order to induce sufficient haem toxicity in these HrtAB-expressing strains (Figs 5 and S3C). While deletion of menB increased resistance of S. aureus to haem stress, the ΔqoxB ΔcydB strain was not protected from haem toxicity and actually appeared to be more susceptible to haem exposure than WT cells.
Together, these data indicate that aerobic respiration is not directly potentiating haem stress. Instead, it appears that quinone molecules such as DMK and MK possessing lipid tails and thus the ability to associate with cell membranes are contributing to the toxic effects of haem. These data also reveal that deletion of the haem-containing cytochromes actually increases haem stress. It is possible that by eliminating natural cellular haem reservoirs such as cytochromes, haem toxicity is increased because displaced haem is being trafficked to inappropriate locations. Alternatively, deletion of the cytochromes may alter the overall redox status of cellular MK pools to a more toxic state.
Menaquinone deficiency does not impair haem uptake
While the growth analyses clearly demonstrate a link between MK biosynthesis and haem stress, these results could potentially be attributed to a general defect in haem uptake. In order to confirm that haem is still imported by MK-deficient strains, cell membranes were isolated from ΔhrtB and ΔhrtB ΔmenA S. aureus strains grown in the presence or absence of exogenous haem. Spectrophotometric analysis of the haem contents of these strains revealed that significant haem accumulation can be observed in membranes from cells with or without the ability to synthesize MK (Fig. 6). These data suggest that the MK-dependent toxic effects of haem can be attributed to a direct interaction between these two molecules.
Disruption of MK biosynthesis alleviates haem-associated oxidative stress
Previous reports have indicated that haem is capable of inducing DNA damage in S. aureus, which is likely a result of oxidative stress (Nir et al., 1991). Therefore, we assessed the presence of haem-induced oxidative protein damage using an OxyBlot™ kit (Millipore) (Fig. 7A). Oxidative damage to proteins results in the formation of carbonyl groups that can be chemically modified and detected through standard immunoblotting methods. Bacterial cultures were grown in the presence or absence of 5 μM haem and assessed for oxidative stress. WT S. aureus cells displayed virtually no haem-induced protein damage after 8 h of haem exposure. However, increased protein carbonylation could be observed for the ΔhrtB mutant grown in the presence of haem as evidenced by high levels of antibody staining throughout the entire haem-treated ΔhrtB lane. The increased haem-induced protein damage associated with the loss of hrtB was dramatically reversed by disruption of either menA or menB. The antibody staining for individual lanes was quantified and graphed as per cent staining in haem-treated culture relative to untreated samples (Fig. 7B). While significant increases in haem-induced protein damage were observed in ΔhrtB as compared with WT, no significant differences were detected between ΔmenA and ΔhrtB ΔmenA or ΔmenB and ΔhrtB ΔmenB strains, indicating that MK is contributing to haem-induced oxidative stress. The role of MK in haem-induced protein damage was further confirmed through chemical complementation. When grown in the presence of exogenous MK4, the ΔhrtB ΔmenA strain accumulated high levels of haem-associated protein carbonylation comparable to the damage observed in the respiration-competent ΔhrtB and ΔhrtB ΔubiE strains (Fig. 7C and D).
Haem-associated oxidative damage localizes to cell membranes
Exogenously acquired haem is known to accumulate in the cell membranes of S. aureus strains grown under iron-replete conditions (Skaar et al., 2004). Additionally, the haem stress-associated quinones, MK and DMK, are predicted to localize to the phospholipid bilayer due to the fact that these molecules possess lipophilic polyprenyl tails. Therefore, it seemed likely that haem-induced oxidative damage was originating within and possibly confined to the membrane. In order to determine which cellular components are most severely impacted by haem stress, cells were fractionated and these fractions analysed for the presence of protein carbonylation (Fig. 8). Significant antibody staining was observed in the haem-treated membrane fractions of ΔhrtB cells but virtually no haem-induced protein damage could be measured in cytoplasmic fractions. These data demonstrate that haem-induced oxidative damage is primarily localized to the cell membrane.
Endogenous MK biosynthesis and exogenous haem excess result in superoxide production
In order to elucidate the nature of the MK- and haem-associated oxidative damage, S. aureus strains were analysed by electron paramagnetic resonance (EPR) and spin trapping technique. Increased levels of various transient free radical species can be detected upon incubation with the spin trap molecule 5-ethoxycarbonyl-5-methyl-1-pyrroline N-oxide (EMPO) which results in the formation of stable EMPO–free radical adducts. These EMPO adducts can be identified based on their unique EPR spectra. EPR was used to assess ΔhrtB and ΔhrtB ΔmenA samples cultured in the presence or absence of exogenous haem (Fig. 9A). These analyses revealed EPR signals that were only present in the haem-treated ΔhrtB cultures. The hyperfine splitting constants of these spectra were indicative of the EMPO–OH adduct (Zhang et al., 2000). However, the EMPO–OH adduct can be generated either through the presence of hydroxyl radicals or through the decay of the superoxide EMPO adduct (Locigno et al., 2005) (Fig. 9B). In order to distinguish between these possibilities, samples were treated with either superoxide dismutase (SOD) or dimethyl sulfoxide (DMSO) to scavenge superoxide or hydroxyl radicals respectively (Fig. 9C). The specific inhibition of EMPO–OH signal by the superoxide scavenger SOD demonstrates that enhanced superoxide production is occurring in haem-treated ΔhrtB S. aureus.
A model for the role of MK in potentiating haem toxicity
An explanation for the role of MK in facilitating haem stress derives from the function of this molecule within the electron transport chain. During typical aerobic respiration, membrane-associated quinones such as MK and DMK donate electrons to cytochrome-associated haem to promote the generation of a PMF (Simon et al., 2008). MK and DMK can possess up to two electrons when fully reduced. Loss of a single electron results in the formation of a semiquinone (SQ) which can react with molecular oxygen to generate superoxide (Barja, 1999) (Fig. 10A). While SQs are produced to some extent during aerobic respiration, there are cellular mechanisms in place to limit their formation and toxicity (Osyczka et al., 2005; McMillan et al., 2012).
Upon exposure to excess exogenous haem, haem molecules accumulate in the bacterial membrane within an as yet unidentified haem reservoir (Skaar et al., 2004). Like SQs, reduced haem molecules are known for their ability to react with atmospheric oxygen to generate superoxide and haemin (Rifkind et al., 2004) (Fig. 10B). Haemin is thought to be relatively inert; however, just as MK is able to donate electrons to cytochrome-associated haemin, MK should also possess the ability to reduce other membrane-localized haemin molecules. The reduction of haemin by MK and DMK should allow the superoxide-generating haem autoxidation to continuously occur, resulting in high levels of superoxide production. Additionally, the reduction of haemin by MK and DMK should result in increased levels of superoxide-generating SQs, leading to even greater haem stress in MK-producing cells (Fig. 10C).
In WT S. aureus, haem-catalysed superoxide production is minimized due to the presence of the HrtAB haem detoxification system. The fact that the HrtAB system is minimizing haem-associated oxidative damage is likely the reason that no oxidative stress response proteins were identified in previous screens analysing haem-responsive protein expression (Friedman et al., 2006). In this work, we have shown that both ΔqoxB ΔcydB and ΔhrtB S. aureus experience increased haem stress as compared with WT cells. It is possible that this phenomenon is due to an increase in haem trafficking to an unknown membrane-associated reservoir. Because there are no cytochromes to appropriately house haem molecules in the ΔqoxB ΔcydB strain, cellular haem might be diverted to inappropriate sites which are more prone to haem autoxidation and/or SQ production. An alternative explanation for the increase in haem stress identified within the ΔqoxB ΔcydB strain is that the overall redox status of the cellular MK pool is altered, resulting in an overabundance of reduced MK. Greater levels of reduced MK might be capable of promoting more rapid rates of redox cycling within the membrane-associated haem molecules and, therefore, a more rapid accumulation of superoxide. The haem stress of the ΔhrtB mutant is even greater than that observed in the ΔqoxB ΔcydB strain because ΔhrtB lacks the HrtAB detoxification system. Without a detoxification system, the cell membrane becomes saturated with superoxide-generating haem. However, haem stress is significantly alleviated by eliminating MK biosynthesis in the ΔhrtB strain because, in the absence of MK or DMK, the membrane-associated haemin cannot be redox-cycled to its toxic form.
This hypothesis for the connection between MK production and haem toxicity is supported by previous literature in which quinone molecules were used as reducing agents to potentiate the formation of reactive oxygen species in vitro (Yu and Anderson, 1997; Duesterberg and Waite, 2007). Other researchers have shown that quinones can play an in vivo role in the generation of free radicals (Gutteridge et al., 1984; Stohs and Bagchi, 1995; Huycke et al., 2001; Korshunov and Imlay, 2006; Medina et al., 2006) and, in fact, deletion of MK biosynthesis can protect against copper-mediated oxidative stress in L. lactis (Rezaïki et al., 2008).
Because S. aureus and other pathogens encounter haem during infection of a vertebrate host, it is possible that these organisms experience haem stress within certain niches of the host environment. In support of this hypothesis, haem detoxification systems are present in a number of Gram-positive microorganisms that regularly come in contact with vertebrate blood such as B. anthracis, E. faecalis, C. diphtheriae, L. lactis, L. monocytogenes and S. agalactiae (Torres et al., 2007; Bibb and Schmitt, 2010; Fernandez et al., 2010; Lechardeur et al., 2012). Of these organisms, it appears evident that B. anthracis experiences haem toxicity during infection as demonstrated by the upregulation of the HrtAB haem detoxification system in a murine infection model (Stauff and Skaar, 2009). This finding likely applies to S. aureus, a related Gram-positive pathogen of the Bacillales order. While it is known that disruption of hssRS or hrtAB alters the virulence of S. aureus (Torres et al., 2007; Attia et al., 2010), the impact of haem toxicity during staphylococcal infection remains to be fully explored. If S. aureus does experience haem stress within the host environment, the haem susceptibility of this pathogen is an area of potential interest that could be exploited as a target for therapeutic intervention.
Prolonged staphylococcal infections can induce genetic mutations that result in a phenotype known as SCVs (Proctor et al., 2006). These strains are characterized by decreased growth rate due to an impaired respiratory chain as well as an ability to reside within eukaryotic cells (von Eiff et al., 1997a,b). SCVs are associated with osteomyelitis, device-related infections and cystic fibrosis, and are often isolated from patients experiencing persistent infections (von Eiff et al., 1997a,1997b; 2006; Seifert et al., 1999; Becker et al., 2004). SCV isolates are thought to arise because of their increased resistance to aminoglycoside antibiotics, a therapeutic class that requires a PMF to enter the cell (Musher et al., 1977; Baumert et al., 2002). SCVs isolated from patients often have a respiration defect associated with MK biosynthesis pathways (Proctor et al., 2006). While studies have shown that aminoglycoside exposure can induce the formation of SCVs in vitro (Musher et al., 1977), there have been cases in which SCVs arise without prior aminoglycoside treatment, indicating that there may be additional factors contributing to SCV selection (Proctor et al., 2006). Our research has demonstrated that mutations in MK biosynthesis protect against haem toxicity. Therefore, it is possible that resistance to haem stress could be a factor in the selection for SCV formation during infection.
Chemicals and oligonucleotides
All chemicals used in this study were purchased from Sigma-Aldrich unless otherwise specified. DNA oligonucleotides were purchased from Integrated DNA Technologies. The DNA oligonucleotides used in these studies are shown in Table S1.
Bacterial strains and growth conditions
All experiments in this study were performed in the S. aureus clinical isolate Newman (Duthie and Lorenz, 1952). Isogenic mutants lacking hrtA and hrtB were previously described (Torres et al., 2007; Attia et al., 2010). S. aureus cultures were grown on tryptic soy broth (TSB) solidified with 1.5% agar at 37°C or in TSB at 37°C with shaking at 180 r.p.m. Growth curves were performed in 96-well round-bottomed plates containing 100 μl of culture per well. Five millilitres of bacterial cultures were grown in 15 ml conical tubes, and 20 ml of bacterial cultures were grown in 50 ml conical tubes. All cultures were grown in the dark to avoid the compounding issue of singlet oxygen that can be generated by haem in the presence of ambient light.
Staphylococcus aureus ΔhrtA transposon mutagenesis and selection
Staphylococcus aureus ΔhrtA was transformed with the Tn917-containing plasmid pTV1 (Youngman et al., 1983) and plated onto TSA containing 10 μg ml−1 chloramphenicol at 30°C in order to promote replication of the temperature-sensitive plasmid. A single colony was inoculated into 5 ml of TSB containing 10 μg ml−1 chloramphenicol, and the culture was shaken at 30°C until turbid. Cells were centrifuged, washed twice in fresh TSB, suspended in 1 ml of TSB, and 1:10, 1:100 and 1:1000 serial dilutions were made. Four hundred microlitres of cell suspension from each dilution was spread onto separate haemin selection medium (TSA, 10 μg ml−1 erythromycin, 30–50 μM haemin) and warmed to 42.5°C, a non-permissive temperature for the pTV1 plasmid which promotes transposon insertion in order to maintain antibiotic resistance. Plates were incubated at 42.5°C for 3 to 5 days until colonies appeared. Colonies from the plate harbouring a low density of mutants were streaked for single colonies onto fresh haemin selection medium and grown at 37°C for 5 days.
Transposon integration site determination
Genomic DNA was prepared from 5 ml of cultures of ΔhrtA transposon mutants using a Wizard® Genomic DNA Purification Kit (Promega) according to the manufacturer's instructions. Two micrograms of genomic DNA was digested with DraI (New England Biolabs) overnight at 37°C followed by incubation at 65°C for 20 min to inactivate DraI. Intermolecular ligation reactions containing 200 ng of digested DNA and T4 DNA ligase (Promega) were assembled and incubated at room temperature for 20 min. Twenty nanograms of ligated DNA was used as a template in PCR reactions containing primers annealing within tn917 and amplifying in opposite directions, transposon primer 1 and transposon primer 2. PCR products were analysed by agarose gel electrophoresis and treated with exonuclease I (NEB) and SAP (Promega) according to manufacturer's instructions. PCR-amplified DNA was column-purified using a PCR Purification Kit (Qiagen) and sequenced using transposon primer 2. The sequence flanking the tn917 inverted repeat was used to interrogate the S. aureus Newman genome sequence to determine the tn917 integration site.
Construction of ΔubiE, ΔmenA and ΔmenB mutants
Established methods to inactivate genes in S. aureus were used to construct a marked in frame mutation in ubiE encoding the MK biosynthesis methyltransferase and the unmarked mutations in the menB and menA genes encoding dihydroxynaphthoic acid synthetase and 1,4-dihydroxy-2-naphthoate octaprenyltransferase respectively (Lukomski et al., 2000; Bae and Schneewind, 2006). Briefly, approximately 500–1000 bp regions upstream and downstream of the targeted genes were amplified. The flanking primers incorporated attB1 and attB2 recombinase sites. The downstream and upstream fragments were either fused through PCR sewing or sequential cloning into pCR2.1 (Invitrogen). The marked deletion for the ubiE gene was generated by inserting the non-polar spectinomycin resistance cassette from pSL60-1 (Lukomski et al., 2000) in frame in an XmaI site between the two flanking regions. The knockout constructs were then recombined in the pKOR1 vector (Bae and Schneewind, 2006). The resultant vectors (pKOR1-ΔmenA, pKOR1-ΔmenB and pKOR1-ΔubiE::spc) were transformed into the S. aureus restriction-negative modification-positive strain RN4220 (Novick, 1991) followed by electroporation into strain Newman or the ΔhrtB strain generated in a Newman background, and the mutants were obtained as previously described (Bae and Schneewind, 2006). Primers AA502, AA505, AA508 and AA509 were used to generate the ubiE deletion. Primers CAW430, CAW431, CAW437 and CAW438 were used to generate the menA deletion. Primers DLS365, DLS366, DLS367 and DLS368 were used to generate the menB deletion.
After 16 h of growth, 20 ml of bacterial cultures (OD600 = 3.2) were pelleted at 3200 g for 10 min at 4°C and washed in TSM (100 mM Tris, pH 7; 500 mM sucrose; 10 mM MgCl2). Cell pellets were stored frozen at −80°C prior to MK extraction. Quinone extraction was performed using cold acetone in a manner similar to previously described protocols (Rezaïki et al., 2008). Whenever possible, extraction procedures were performed at 4°C in the dark. Cell pellets were suspended in 5 ml of ice-cold acetone and 1 ml of acid-washed glass beads. Samples were vortexed for 2 min and sonicated in an ice-cooled water bath for 15 min. Cell debris was pelleted at 13 000 g and the supernatant was retained. Another 5 ml of cold acetone was added to the cell debris and the vortexing, sonication and pelleting steps were repeated. Finally, 10 ml of cold acetone were added to the remaining cell debris and the vortexing, sonication and pelleting steps were repeated. The supernatants from each extraction were combined and evaporated under vacuum at 30°C. Evaporated pellets were suspended in 1 ml of heptane and loaded onto a preparative silica gel column (SPE DSC-Si silica tube 1 ml, 100 mg, Supelco). The sample-loaded column was washed once with 1 ml of heptane and quinones were eluted with the addition of 1 ml of 3% diethyl ether/97% heptane. Samples were dried under nitrogen at 37°C. The dried quinone extracts were suspended in 20 μl of ethanol for subsequent analysis.
Thin layer chromatography
Thin layer chromatography was performed using 5 × 10 cm silica gel aluminum sheets (60 F254, EMD) in a manner similar to previously described methods (Rezaïki et al., 2008). Ten microlitres of the quinone extract was loaded adjacent to 10 μl of a migration standard consisting of 0.5 mM MK-4 and resolved using heptane–diethyl ether (85:15, v/v) as the mobile phase. Quinones were visualized by UV exposure, and the plates were photographed using an AlphaImager (Alpha Innotech).
Menaquinone extracts were subjected to high-pressure liquid chromatography (HPLC) mass spectrometry (MS) using an Acquity Ultra Performance LC system (Waters) interfaced with a Finnigan TSQ Quantum Ultra Triple Quadrupole mass spectrometer (Thermo Scientific). Ten microlitres of sample was injected and HPLC separation was achieved using an Xterra C18 column (Waters) using a gradient of (A) methanol : H2O (95:5, v/v) and (B) methanol : isopropanol (30:70, v/v). Gradient conditions were as follows: 0–1 min, A = 100%; 1–9 min, B = 0–100%; 9–10 min, B = 100%; 10–11 min, A = 0–100%; 11–14 min, A = 100%. The column temperature was maintained at 50°C. Eluted molecules were subjected to positive ion MS using APCI as the ionization interface. MK and DMK molecules were detected using parent ion scans for m/z 187 and m/z 173, respectively, as previously described (Geyer et al., 2004).
Measurement of PMF
Following the manufacturer's instructions, the Baclight Bacterial Membrane Potential Kit (Invitrogen) was used to monitor the PMF of S. aureus mutants. Briefly, bacterial cultures were grown to mid-exponential phase and added to a 1 ml PBS solution containing 30 μM 3,3-diethyloxacarbocyanine iodide (DiOC2). Suspensions were incubated for 30 min at 37°C prior to analysis by the Vanderbilt Flow Core. For each sample, the membrane decoupler, carbonyl cyanide m-chlorophenyl hydrazone (CCCP, 5 μM), was used as a positive control for loss of the PMF.
Measurement of l- and d-lactate production
Bacterial cultures grown overnight were diluted 1:100 into fresh TSB. These cultures were grown for 15 h at 37°C. The cells were centrifuged and the supernatant was collected. Following manufacturer's instructions, d- and l-lactate was monitored in culture supernatants using the Boehringer Mannheim enzymatic UV method (Roche).
Measurement of haem content
In order to assay haem uptake by S. aureus strains, cells were cultured overnight in plain TSB and diluted to an OD600 of 0.5 in the morning. The diluted cultures were grown in 5 ml of TSB for 6 h in the presence or absence of 5 μM haem. Cultures were pelleted at 3200 g for 10 min at 4°C, and pellets were washed in TSM (100 mM Tris, pH 7; 500 mM sucrose; 10 mM MgCl2). Pellets were suspended in 500 μl of TSM supplemented with 100 μg of lysostaphin and incubated for 30 min at 37°C. The resulting protoplasts were sedimented by centrifugation at 20 000 g for 5 min. Protoplasts were suspended in 200 μl of TKM (50 mM Tris HCl, pH 7; 60 mM KCl; 10 mM MgCl2) and lysed by sonication for 10 s. Cell membranes were sedimented by centrifugation at 100 000 g for 30 min. Membranes were suspended in TKM, and protein content was measured by absorbance at 280 nm. The haem content of concentration-matched samples was assayed similarly to previously described methods (Lombardo et al., 2005). Briefly, membrane fractions were diluted to a final volume of 550 μl in TKM containing 300 mM NaCl and 24% DMSO. Samples were then acidified through the addition of 0.2 ml of 50 mM glycine-HCl (pH 1) and haem was subsequently extracted by adding 0.2 ml of chloroform and vortexing several times. Spectral absorbance readings were taken of the chloroform phase at wavelengths ranging from 300 to 500 nm. Haem units were determined using the formula Ac = 2A388 − (A450 + A330) and concentration was graphed as haem units mg−1 protein.
Analysis of oxidative protein damage
Five millilitres of bacterial cultures were grown for 8 h in TSB in the presence or absence of 5 μM haem. These cultures were pelleted at 3200 g for 10 min at 4°C, and pellets were washed in TSM (100 mM Tris, pH 7; 500 mM sucrose; 10 mM MgCl2). Pellets were suspended in 500 μl of TSM supplemented with 100 μg of lysostaphin and incubated for 30 min at 37°C. The resulting protoplasts were sedimented by centrifugation at 20 000 g for 5 min. Protoplasts were suspended in 200 μl of lysis buffer (50 mM Tris HCl, pH 7; 60 mM KCl; 10 mM MgCl2; 2% β-mercaptoethanol) and lysed by sonication for 10 s. For experiments in which cells were fractionated, cell lysates were further processed by centrifugation at 100 000 g for 30 min. The supernatant was retained as the cytoplasmic fraction, and the pellet was suspended in 200 μl of lysis buffer and retained as the membrane fraction. Proteins affected by oxidative damage were detected using the OxyBlot™ Protein Oxidation Detection Kit (Millipore). Following manufacturer's instructions, the oxidatively damaged proteins were derivatized with 2,4-dinitrophenylhydrazine (DNP). DNP-derivatized proteins were detected via standard immunoblotting methods using a primary antibody specific to the DNP moiety on the damaged proteins (Millipore) and a goat anti-rabbit IgG (Alexa Fluor 680-conjugated) secondary antibody (Invitrogen). Immunoblots were visualized using an Odyssey® Imager (LI-COR Biosciences). Staining intensity of individual sample lanes was quantified using the Odyssey® software.
Electron paramagnetic resonance study
Bacterial cultures were grown overnight and diluted to OD600 = 0.1 in 20 ml of TSB. Diluted cultures were grown at 37°C for 30 min in the presence or absence of 5 μM haem. Cultures were pelleted at 3200 g for 10 min at 4°C, and pellets were washed in ice-cold KHB (99.01 mM NaCl, 4.69 mM KCl, 2.5 mM CaCl2, 1.2 mM MgSO4, 25 mM NaHCO3, 1.03 mM K2HPO4, 20 mM Na-HEPES and 5.6 mM d-glucose, pH 7.35). Bacterial pellets were suspended in KHB at a concentration of 6 × 108 cfu ml−1. Prior to EPR analysis, 30 μM diethylene triamine pentaacetic acid (DTPA) and 50 mM 5-ethoxycarbonyl-5-methyl-1-pyrroline N-oxide (EMPO, Enzo Life Sciences) were added to 200 μl of sample. Samples were incubated at 37°C for 5 min before being subjected to EPR analyses at the Vanderbilt Free Radicals in Medicine core facility. When applicable, pegylated superoxide dismutase (SOD, 25 units) or dimethyl sulfoxide (DMSO, 5% final volume) was included in the EPR samples.
We would like to thank members of the Skaar laboratory for critical reading of this manuscript. Mass spectrometry and flow cytometry core services were performed through Vanderbilt University Medical Center's Digestive Disease Research Center supported by NIH Grant P30DK058404. EPR analyses were performed at the Vanderbilt University Free Radicals in Medicine core facility. Work in the Skaar laboratory is supported by Grant Numbers AI069233, AI073843 and AI091771 from the National Institutes of Health. C. A. W. was supported by Grant Number T32-HL094296 from the National Heart, Lung, and Blood Institute and Grant Number F32-AI100535 from the National Institute of Allergy and Infectious Diseases. N. D. H. is supported by the National Institute of Health Ruth L. Kirschstein fellowship F32-AI091244-01. L. L. A. was supported by Grant T32-HL069765 from the National Institute of Allergy and Infectious Diseases. The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, and Blood Institute, the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.