Staphylococcus aureus Lactate- and Malate-quinone Oxidoreductases Contribute to Nitric Oxide Resistance and Virulence

Summary

Staphylococcus aureus is a Gram-positive pathogen that resists many facets of innate immunity including nitric oxide (NO·). S. aureus NO·-resistance stems from its ability to evoke a metabolic state that circumvents the negative effects of reactive nitrogen species. The combination of L-lactate and peptides promotes S. aureus growth at moderate NO·-levels, however neither nutrient alone suffices. Here we investigate the staphylococcal malate-quinone and L-lactate-quinone oxidoreductases (Mqo and Lqo), both of which are critical during NO·-stress for the combined utilization of peptides and L-lactate. We address the specific contributions of Lqo-mediated L-lactate utilization and Mqo-dependent amino acid consumption during NO·-stress. We show that Lqo conversion of L-lactate to pyruvate is required for the formation of ATP, an essential energy source for peptide utilization. Thus, both Lqo and Mqo are essential for growth under these conditions making them attractive candidates for targeted therapeutics. Accordingly, we exploited a modeled Mqo/Lqo structure to define the catalytic and substrate-binding residues. We also compare the S. aureus Mqo/Lqo enzymes to their close relatives throughout the staphylococci and explore the substrate specificities of each enzyme. This study provides the initial characterization of the mechanism of action and the immunometabolic roles for a newly defined staphylococcal enzyme family.

Graphical Abstract

Introduction

Staphylococcus aureus is one of the most important pathogens to human health. Illnesses caused by S. aureus manifest predominantly as skin/soft tissue infections that can lead to more severe diseases such as sepsis, pneumonia, endocarditis, myocarditis and osteomyelitis (Diekema et al., 2001; Wasi and Shuter, 2003; Klevens et al., 2007). S. aureus colonizes and thrives in multiple body sites and interferes with nearly every component of the host innate immune response, making the pathogen difficult to prevent and treat (Foster, 2005). Factors limiting the success of opsonophagocytosis, complement-mediated killing, cationic peptides and reactive oxygen and nitrogen species are all encoded by S. aureus, and these gene products contribute to disease severity and duration.

Nitric oxide (NO·) is an important effector of the innate immune response with both immunomodulatory and antibacterial activities. Inflammatory NO· is required for the clearance of a variety of pathogens (Fang, 2004; Richardson et al., 2006). NO· is generated by activated leukocytes through an inducible NO·-Synthase (iNOS) and can act directly on invading organisms or indirectly through oxidation to reactive nitrogen species (RNS). (Lewis et al., 1995; Nalwaya and Deen, 2005). NO· and RNS are responsible for nitrosation of protein thiols, nitration of tyrosine residues, nitrosylation of metal centers (notably heme in respiratory oxidases), peroxidation of lipids and deamination of DNA bases (Wink et al., 1991; Radi, 1996; Wink and Mitchell, 1998; Schopfer et al., 2003). Because of these reactions, NO· interferes with many bacterial metabolic pathways, such as the tricarboxylic acid (TCA) cycle, aerobic respiration, pyruvate metabolism, fatty acid metabolism and nucleic acid synthesis (Richardson et al., 2006; Richardson et al., 2008; Richardson et al., 2011). Despite the constraints imposed on pathogen metabolism by RNS, S. aureus is uniquely able to thrive in the presence of significant NO· levels. This trait distinguishes S. aureus from closely related coagulase negative staphylococci including S. epidermidis, S. haemolyticus and S. saprophyticus, which are not capable of significant growth in the presence of NO·. Many pathogens limit exposure to NO· by restricting iNOS induction, occluding iNOS from the phagosome or detoxifying NO· through flavohemoproteins like Hmp (Chakravortty et al., 2002; Bang et al., 2006; Das et al., 2009). While S. aureus expresses an Hmp homolog, the enzyme only partially contributes to S. aureus NO·-resistance (Richardson et al., 2006; Richardson et al., 2008). In contrast to other bacteria, the key element to S. aureus NO·-resistance is the capacity of the pathogen to adopt an alternative metabolic state that circumvents the effects of NO· (Richardson et al., 2008). The exact nature of this S. aureus NO·-resistant metabolic state has yet to be fully defined.

NO· attacks heme cofactors in respiratory oxidases, interfering with aerobic respiration (Brown et al., 1997; Stevanin et al., 2007). Accordingly, NO·-exposure induces fermentative metabolism in S. aureus resulting in L-lactate production and excretion that establishes redox balance (Stevanin et al., 2000; Hochgräfe et al., 2008). However, over time as the NO·-levels are enzymatically reduced in vitro, the excreted L-lactate is reassimilated in a process involving an NADH-independent lactate dehydrogenase (Fuller et al., 2011). NADH-independent LDH activity has been described for fifty years as associated with staphylococcal membranes, yet the gene encoding this enzyme was only recently characterized. Originally annotated as malate:quinone oxidoreductase 2 (Mqo2), we showed that this enzyme actually comprises the NADH-independent LDH activity of S. aureus and was therefore renamed Lqo. Δlqo mutants were attenuated in murine sepsis models of myocarditis and this defect in virulence was predicated on host NO·-production. However, the precise metabolic role of Lqo under NO·-stress is yet to be fully understood. In contrast to the fermentative production of L-lactate, the use of L-lactate as a carbon/energy source via Lqo requires residual respiration and therefore is only relevant when NO·-levels are at or below μM concentrations. While higher levels of NO· restrict S. aureus metabolism to the use of glycolytic carbohydrates, lower levels of NO· allow for growth on the combination of L-lactate and peptides, but not on either of these nutrients alone. Lqo was shown to be essential for growth on lactate/peptides, and therefore S. aureus requires Lqo-mediated NO·-resistance to thrive in host tissues with relatively low levels of free glucose and elevated concentrations of lactate (Fuller et al., 2011). Furthermore, Lqo may also be important for S. aureus skin colonization, where hydroxyacids are abundant and glucose levels are low (Gallagher et al., 2008). Still, the metabolic requirement for the combination of lactate and peptides and the necessity of Lqo for utilization of these nutrients are poorly understood.

In this report, we propose a nutrient partitioning hypothesis whereby Lqo-mediated lactate consumption is used in part for energy generation and Mqo-dependent amino acid utilization provides a carbon source for the majority of biomass. We show that Lqo is specifically required under moderate NO· stress to use L-lactate from the environment to generate acetate. Acetate formation yields ATP that is critical for amino acid import. Amino acid utilization requires an active TCA cycle, and in S. aureus, Mqo is required for the conversion of malate to oxaloacetate, a precursor for gluconeogenesis. Thus, both Lqo and Mqo are required for growth on peptides and L-lactate during moderate NO·-stress. These enzymes are highly similar (50% identity) and thus represent an attractive “double hit” target for anti-S. aureus therapeutic development. Moreover, other coagulase negative staphylococcal pathogens encode three to four of these enzymes underscoring their importance to the overall staphylococcal metabolic strategies outside the context of NO·-resistance. Here, we initiate studies directed at elucidating the reaction mechanisms of this staphylococcal enzyme family. Accordingly, we have confirmed the presence of important conserved residues in the Mqo/Lqo substrate-binding pockets and have identified the substrates of other staphylococcal Mqo enzymes. These findings mechanistically expand our knowledge of the unique metabolic scheme employed by S. aureus while under host NO· stress.

Results

Lqo provides S. aureus with energy during NO· stress

While the combination of L-lactate and peptides provides S. aureus with a means to thrive under low-level NO·-stress in vitro, neither alone will suffice (Fuller et al., 2011). Consequently, S. aureus requires both Mqo (SACOL2362) and Lqo (SACOL2623) to resist NO· under these conditions since Δlqo cannot utilize L-lactate during NO·-stress and Δmqo cannot efficiently catabolize peptides (Figure 1) (Fuller et al., 2011). Replacing L-lactate with pyruvate in combination with amino acids also supports NO·-resistant growth and eliminates the requirement for Lqo, but not Mqo (Figure 1C). This implies that Lqo-mediated pyruvate generation is specifically required under these conditions. However, other pyruvate precursors (e.g. D-lactate, alanine and serine) do not suffice in combination with amino acids under NO·-stress (Figure 1D). Thus, Lqo is essential for growth on gluconeogenic nutrients during NO· stress unless pyruvate is readily available.

Figure 1. Both Lqo and Mqo are necessary for NO·-resistance in S. aureus.

A. WT S. aureus COL and isogenic mutant derivatives were grown aerobically in PN medium with 0.5% Casamino acids plus 0.5% L-lactate as the primary carbon source. B. WT S. aureus COL and isogenic mutant derivatives were grown in PN medium with 0.5% Casamino acids plus 0.5% L-lactate and supplemented with15mM DETA/NO once cultures reached OD₆₅₀ = 0.15. C. Average growth rates of WT S. aureus COL and mutant derivatives following NO· exposure in PN medium with 0.5% Casamino acids plus 0.5% pyruvate as the primary carbon source. (growth rates were averaged over four hours of growth following NO· exposure, * significantly reduced growth rate compared with WT, p ≤ 0.05, Student’s t-test). D. L-lactate is the only pyruvate precursor that supports growth in combination with amino acids during NO· stress (* significantly enhanced growth yield compared to amino acids alone (none), p ≤ 0.05, Student’s t-test).

One possible fate of pyruvate is its conversion to acetate via acetyl phosphate, a reaction involving the AckA enzyme. This version of acetogenesis yields ATP in the process. S. aureus growing on lactate and peptides during NO·-stress excrete high concentrations of acetate into the media as L-lactate is consumed (Figure 2A and 2B). To determine the importance of acetate formation (and/or ATP production) during NO· stress, growth of a ΔackA mutant was assessed. Like Δlqo, the ΔackA mutant could not grow on a mixture of amino acids and L-lactate when exposed to NO· (Figure 2C). However, unlike Δlqo, growth of ΔackA could not be rescued by replacing L-lactate with pyruvate, showing that acetate/ATP formation through AckA is necessary for survival during NO· stress (data not shown). Accordingly, addition of exogenous acetate also failed to restore growth of the ΔackA mutant supporting the role of AckA-generated ATP, not acetate itself, as being required for survival in the presence of NO· (data not shown).

Figure 2. L-lactate serves as an acetogenic source of ATP during NO· stress.

A. Amount of acetate produced and L-lactate consumed (left axis) over time compared to S. aureus COL growth (right axis) in PN medium plus 0.5% Casamino acids and 0.5% L-lactate during NO· exposure. B. Ratio of acetate produced to L-lactate consumed after 20 hours of aerobic growth with or without NO· stress in PN medium plus 0.5% Casamino acids and 0.5% L-lactate. (* indicates significantly higher ratios under NO· stress, p ≤ 0.05, Student’s t-test). C. Growth of S. aureus COL ΔackA with or without NO· stress in PN medium plus 0.5% Casamino acids and 0.5% L-lactate. D. Comparison of intracellular ATP levels (normalized to OD₆₅₀) in WT S. aureus COL or isogenic Δlqo cells under 4 hours of NO· stress. Cells were grown in PN medium plus 0.5% Casamino acids with 0.5% of either L-lactate or pyruvate. Significance was calculated using a Student’s two-sided t test (*, P ≤ 0.05, n ≥ 3).

The requirement of Lqo and AckA for the conversion of L-lactate to acetate and ATP implies that L-lactate is serving as an important energy source under NO·-stress when S. aureus is catabolizing peptides and L-lactate. To examine the role of Lqo/AckA in ATP generation during NO·-stress, S. aureus wild type and Δlqo strains were grown under NO· stress and ATP levels were measured. Δlqo had significantly lower levels of ATP under NO· stress when compared to WT cells (Figure 2D). The requirement for Lqo to maximize ATP production was circumvented by providing pyruvate, which was consistent with the ability of pyruvate to rescue Δlqo during NO· stress (Figure 2D). Significant ATP levels are required for peptide and amino acid import whereas glucose import can proceed through the phosphotransferase (PTS) system. Nutrient acquisition via PTS is less sensitive to ATP levels since it requires phosphenolpyruvate for phosphotransfer. To verify that S. aureus growth on peptides and amino acids is highly dependent on ATP, wild type bacteria were grown in PN medium supplemented with either glucose or Casamino acids as the carbon source with increasing concentrations of azide, a pan-ATPase inhibitor. Bacteria grown with Casamino acids were significantly more sensitive to azide-mediated ATPase-inhibition than bacteria grown with glucose (Figure S1). These data suggest that assimilation of L-lactate by Lqo ultimately serves as a critical energy source allowing for the optimization of intracellular ATP pools necessary for growth in peptide-rich environments under NO· stress.

Proposed Staphylococcus aureus Lqo active site mimics that of MDH/LDH enzymes

Lqo and Mqo are membrane associated, NAD-independent lactate-/malate-quinone oxidoreductases that belong to a family of putative 2-hydroxyacid:quinone oxidoreductases. These enzymes use FAD cofactors to directly transfer electrons from their substrates to membrane associated quinone molecules in the electron transport chain. Donation of electrons to the quinone pool requires efficient respiration, as electrons must ultimately pass to terminal electron acceptors such as oxygen or nitrate. In contrast, the NAD-dependent dehydrogenase counterparts to Lqo/Mqo (Ldh/Mdh) donate electrons to cytosolic NAD⁺, forming NADH and do not necessarily require respiration if other electron sinks are available for regenerating NAD⁺. The reaction mechanism of these NAD-dependent dehydrogenases involves coordinating the substrate carboxylate with an arginine, while a histidine accepts a proton after hydride extraction (Figure 3A). To probe the reaction mechanism of these staphylococcal 2-hydroxyacid quinone oxidoreductases, we generated a homology model of Mqo/Lqo using the streptococcal α-glycerophosphate oxidoreductase (GlpO) crystal structure as a template (Figure 3B). Like Mqo/Lqo, GlpO has a FAD cofactor and is a 2-hydroxyacid quinone-oxidoreductase (Colussi et al., 2008). Our Mqo/Lqo model predicts that Arg397 and His58 are in the correct positions to coordinate the substrate carboxylate and accept a proton following hydride extraction (Figure 3) (Deng et al., 1994). Furthermore, these residues are conserved in all of the Mqo and Lqo alleles found among the staphylococci (Figure S2) supporting a role for these residues in the Mqo/Lqo catalytic mechanism. Accordingly, Arg397 and His58 were individually converted to Ala via site-directed mutagenesis, and the mutant Lqo alleles were assayed for enzyme activity. We found that mutation of the conserved His58 and Arg397 abolished Lqo activity entirely (Table 1), consistent with the proposed reaction mechanism (Figure 3A). To ensure that no major structural changes in protein conformation were caused by the mutations, CD spectra in the far-UV region were recorded for LqoH58A and LqoR397A. Spectral analysis did not show any significant differences in protein structure when wild type and mutant proteins were compared (Figure S2).

Figure 3. Mqo and Lqo belong to a unique class of staphylococcal flavoenzymes.

A. Proposed reaction mechanism of NAD-independent Lqo/Mqo enzymes. B. Homology model of Lqo based on Streptococcus GlpO. FAD-cofactor is depicted as red, white and blue space-filled molecule in the enzyme’s interior pocket. Similarly, conserved H58 and R397 involved in the reaction mechanism re shown as dark blue space-filled amino acid residues. Residue D285 involved in Lqo substrate specificity is modeled as the brown space-filled molecule. In Mqo, this position is occupied by H284, not depicted here. C. Neighbor-joining tree of staphylococcal Mqo/Lqo proteins. S. aureus Mqo and Lqo fall into separate and distinct clades. Proteins orthologous to S. aureus Mqo are in blue. Proteins orthologous to S. aureus Lqo are in red. Green and orange branches represent conserved clades that share high sequence similarity and genomic contexts. * denote proteins shown here to oxidize lactate, * denote proteins shown here to oxidize malate. Abbreviations: aur, S. aureus; cap, S. captis; car, S. carnosus; epi, S. epidermidis; hae, S. haemolyticus; hom, S. hominis; lug, S. lugdunensis; sap, S. saprophyticus; war, S. warneri; xyl, S. xylosus.

Table 1.

Affinities of Mqo and Lqo wild type and mutant proteins for their substrates

	KM L-lactate	KM malate
Lqo	330 μM	26,000 μM
LqoH58A	Undetectable	N/A
LqoD285H	217,600 μM	N/A
LqoR397A	670 μM	Undetectable
Mqo1	Undetectable	180 μM
Mqo1H284D	Undetectable	Undetectable

Lqo and Mqo share 50.2% amino acid identity and are highly specific for their cognate substrates. We probed the homology model for differences within the substrate-binding pocket that might explain the specificity of these enzymes. Within the predicted Mqo active site at the opposite end of the carboxylate-coordinating Arg397 is His284; a residue correctly positioned to coordinate the β-carboxylate of malate (Figures 3A & 3B). In contrast, the Lqo active site contains Asp285 at this position that the model predicts would exclude malate by electrostatic repulsion of its β-carboxylate (Figures 3A & 3B). In an attempt to discern residues involved in substrate discrimination, Mqo His284 was mutated to Asp to mimic the binding pocket of Lqo. This mutation completely abolished Mqo activity towards malate without any measurable changes to CD implying that the Asp carboxylate can occlude malate from the Lqo pocket (Table 1; Figure S2B). However, the Mqo His284→Asp did not confer L-lactate specificity implying that more residues within this pocket contribute to substrate specificity. Interestingly, mutation of Lqo Asp285 to His to mimic the Mqo binding pocket only slightly decreased the affinity of Lqo for L-lactate and failed to confer activity towards malate (Table 1). Together, these findings suggest that Asp285 can exclude malate from the binding pocket of Lqo, presumably by repelling the substrate’s β-carboxylate (Figure 3A). However, Lqo Asp285 is not necessarily required for activity with L-lactate. Furthermore, Mqo His284 does not alone confer substrate specificity towards malate, thus the exact identity of the amino acids that confer substrate specificity among these enzymes remains elusive.

Lqo belongs to a family of staphylococcal-specific enzymes

To generate more information regarding protein features that dictate substrate specificity, we sought to better characterize the staphylococcal Mqo-family of enzymes. We identified the substrates for S. epidermidis Mqo1-4 (SERP1955, SERP2168, SERP2312 and SERP2412, respectively) and S. saprophyticus Mqo1-3 (SSP0081, SSP0156 and SSP0539, respectively) using two methods. First, these orthologs were purified in the same manner as the S. aureus His₆-tagged proteins described above, and were subjected to in vitro enzyme assays to verify their substrate specificity. A number of α-hydroxyacids were tested as potential substrates. Like S. aureus, the substrates for S. epidermidis Mqo1 and Mqo2 are malate and L-lactate, respectively. Additionally, S. epidermidis Mqo3 and S. saprophyticus Mqo1 and Mqo2 reacted with L-lactate. In contrast, S. saprophyticus Mqo3 reacted with malate and not L-lactate (Table 2). The substrate for S. epidermidis Mqo4 could not be verified by this assay due to a lack of reactivity. As a second method to complement in vitro analyses, the S. epidermidis mqo alleles were cloned into the S. aureus pOS1 expression vector system to determine their ability to complement S. aureus Δmqo and Δlqo mutants in vivo. These plasmids were expressed in S. aureus COL strains Δmqo (a strain unable to use amino acids as a carbon source due to disruption of the TCA cycle) and Δldh1Δldh2Δlqo (a strain unable to utilize L-lactate as a carbon source). COL strains containing the pOS1 plasmids were grown in PN defined medium with either 1% Casamino acids or 1% L-lactate as sole carbon sources. Expression of S. epidermidis Mqo1 and S. saprophyticus Mqo3 in COL Δmqo rescued growth on casamino acids, suggesting that these orthologs are able to functionally replace S. aureus Mqo (Table 2; Figure S3). Likewise, Δldh1Δldh2Δlqo S. aureus COL expressing either S. epidermidis Mqo2, Mqo3 or Mqo4 as well as S. saprophyticus Mqo1 or Mqo2 were able to use L-lactate as a substrate in the place of the S. aureus Lqo enzyme. Grouping these staphylococcal Mqo/Lqo orthologs by function in a pairwise sequence comparison revealed that Mqo proteins were generally ≥ 60% identical to each other and only 50–55% identical to Lqo enzymes (Table 3). Similarly, Lqo proteins were ≥ 60% identical to each other and only 50–55% identical to Mqo enzymes (Table 3).

Table 2.

Ability of staphylococcal Mqo proteins to use L-lactate and malate as substrates

Enzyme	Reactivity in vivo	Reactivity in vitro
SE Mqo1	Malate	Malate
SE Mqo2	L-lactate	L-lactate
SE Mqo3	L-lactate	L-lactate
SE Mqo4	L-lactate	N/A
SS Mqo1	N/A	L-lactate
SS Mqo2	N/A	L-lactate
SS Mqo3	N/A	Malate

Table 3.

Percent identity of Staphylococcal Mqo orthologs

	Mqo aur	Mqo1 epi	Mqo3 sap	Mqo4 epi	Mqo2 epi	Lqo aur	Mqo2 sap	Mqo3 epi	Mqo1 sap
Mqo aur	100	79.5	72	49.1	50.9	50.2	50.1	49.1	49.1
Mqo1 epi	79.5	100	70.8	48.5	52.7	50.4	51.9	50.1	50.5
Mqo3 sap	72	70.8	100	50.9	52.3	51.8	54.6	51.7	52.1
Mqo4 epi	49.1	48.5	50.9	100	66.4	66.1	60.5	67.8	66.2
Mqo2 epi	50.9	52.7	52.3	66.4	100	86.1	69.3	83.3	80.1
Lqo aur	50.2	50.4	51.8	66.1	86.1	100	67.5	83.5	78.9
Mqo2 sap	50.1	51.9	54.6	60.5	69.3	67.5	100	70.9	68.5
Mqo3 epi	49.1	50.1	51.7	67.8	83.3	83.5	70.9	100	84.1
Mqo1 sap	49.1	50.5	52.1	66.2	80.1	78.9	68.5	84.1	100

^*aur = S. aureus; epi = S. epidermidis; sap = S. saprophyticus

The staphylococcal Mqo enzyme family uniquely clusters away from other Mqo enzymes in both Gram-positive and Gram-negative bacteria. Additionally, the predicted Lqo enzymes belong to a distinct clade separate from the predicted malate-using enzymes. A phylogenetic comparison of the staphylococcal Mqo enzymes was performed by creating neighbor-joining trees using a number of protein sequences from different staphylococcal species (Figure 3C). Using overall sequence homology combined with genomic context, we could assign an Mqo allele to each of the 17 surveyed species of Staphylococci and these alleles clustered distinctly from the other enzymes in the family. These Mqo encoding genes were universally associated with genes encoding a lactate permease and the HssRS two component system (Figure S4A). In contrast, alleles related to S. aureus lqo could be found among closely related species in similar genomic contexts more or less associated with fdaB and acs (Figure S4B). Other presumed lqo alleles were found in S. captis, S. xylosus and S. equis in related genomic contexts with S. epidermidis mqo3 and S. saprophyticus mqo2 (Figure 3C, green and orange clades, respectively). There were also many other mqo alleles with unknown substrate specificity and unique genomic contexts, implying that this enzyme family has a complex evolutionary history among the various species of staphylococci.

Mqo and Lqo are required for the virulence of extracellular bacteria

To determine the contribution of Mqo and Lqo to S. aureus virulence, we infected C57BL/6 mice intravenously with WT, Δmqo, Δlqo, and ΔmqoΔlqo S. aureus Newman. As previously reported, Δlqo mutants have a modest defect in organism burdens in murine kidneys 7 days post inoculation (Fuller et al., 2011). Similarly, Δmqo mutants have an equivalent defect and notably, the contribution of each of these enzymes to virulence is not additive in that the double ΔmqoΔlqo mutant was equivalently attenuated as each single mutant (Figure 4A). This trend was more prevalent in the liver where each single mutant and the double ΔmqoΔlqo mutant were all equivalently attenuated with a 3-log reduction in organism burdens 7 days post inoculation (Figure 4B). These results are consistent with Lqo and Mqo acting in a single metabolic pathway as outlined above. In contrast, overall weight loss was significantly decreased in the double ΔmqoΔlqo mutant compared to each single mutant implying that each enzyme must contribute independently to virulence in some aspect of sepsis (Figure 4C). However, intracellular survival cannot account for the additive nature in total body weight loss since neither mutant has a survival defect in activated RAW246.7 cells (Figure S5). This is consistent with the high level of NO· encountered within these cells precluding efficient respiration and thus the functionality of Mqo and Lqo. Given that the most common disease presentation for S. aureus is skin infections, we also tested the S. aureus LAC ΔmqoΔlqo in a subcutaneous murine skin abscess model. As in the sepsis model, the double ΔmqoΔlqo mutant exhibited > 1 log reduction in organism burden 7 days post inoculation (Figure 4D). Thus, the Lqo/Mqo family of enzymes in S. aureus contributes significantly to overall virulence, in both additive and non-additive manners, in multiple infection models.

Figure 4. Mqo and Lqo are both required for full S. aureus virulence.

A. & B. C57BL/6 mice were infected via tail vein injection with 1X10⁷ CFU of wild-type S. aureus Newman or isogenic mutant strains. Kidneys A and livers B were harvested 7 days post-infection and bacteria were enumerated. Mutants were attenuated with significantly reduced organism burdens, p ≤ 0.05, Mann-Whitney test. C. Weight of the mice was monitored and recorded every day and average % weight loss is shown for day 7. (* significantly reduced weightloss compared to WT, $ significantly reduced weight loss compared to Δmqo or Δlqo, p ≤ 0.05 Student’s t-test). D. C57BL/6 mice were inoculated subcutaneously with 1 X 10⁷cfu of WT S. aureus LAC or isogenic ΔmqoΔlqo mutants and bacterial burdens enumerated 7 days post inoculation. (n ≥ 10, * p < 0.05, Mann-Whitney test).

Discussion

The ability of S. aureus to thrive in the presence of host NO· directly contributes to disease severity in that NO·-sensitive mutants are attenuated in murine infection models (Richardson et al., 2006; Richardson et al., 2008). Furthermore, NO·-sensitive staphylococcal species (e.g. S. epidermidis) can replicate to levels similar to S. aureus if host NO· is eliminated (Thurlow et al., 2013). High levels of NO· (≥1 μM) abolish S. aureus respiration and necessitate a glycolytic/fermentative metabolic strategy (Vitko et al., 2015). This is most relevant to intracellular survival of phagocytized S. aureus, which are most closely associated with iNOS, the source of high-level NO·. However, clusters of extracellular S. aureus are commonly found at the centers of organ and skin abscesses with no viable phagocytes in close proximity. In this environment, NO· concentrations are predicted to be below 1 μM (Brovkovych et al., 1997; Chin et al., 2008), a level associated with residual respiratory activity in S. aureus (Richardson et al., 2008). Moreover, the rapid consumption of glucose observed in activated phagocytes combined with robust L-lactate excretion means that bacteria within organ/skin abscesses will likely have abundant access to L-lactate but limited access to glucose (Kelly and O'Neill, 2015). This combined with the rich proteolytic capabilities of S. aureus makes L-lactate and peptide nutrient sources in tissue with moderate NO· levels a physiologic environment likely encountered by extracellular S. aureus within organ/skin abscesses.

The requirement for combined L-lactate/amino acid nutrient sources for NO·-resistance in S. aureus likely stems from the reactivity of NO· with sensitive enzymatic motifs such as iron-sulfur clusters and redox active thiols. For instance, for L-lactate to serve as both an efficient energy and carbon source, L-lactate must be converted to acetate as well as to each of the 13 metabolic precursors required to synthesize all biomolecules within the cell. Many of these (e.g. 2-oxoglutarate and succinyl-CoA) lie downstream of NO·-sensitive enzymes (aconitase and succinyl-CoA synthase) precluding their synthesis from L-lactate (Richardson et al., 2011). Inclusion of amino acids in the growth medium diminishes the need for generating many of these precursor metabolites as amino acids are the primary requirement for precursor metabolites. Furthermore, the presence of exogenous amino acids allows the cell to direct L-lactate utilization into energy production. This is key to NO·-resistance because, while the level of NO· encountered by these cells allows for residual respiration, full respiratory flux is still not possible. Thus, energy production cannot rely entirely on oxidative phosphorylation but requires contribution from substrate level phosphorylation (e.g. acetogenesis). Utilizing amino acids exclusively as both carbon and energy sources under NO·-stress does not allow for efficient ATP production via acetogenesis. The catabolism of most amino acids does not result in a respiration-independent source of ATP. However, serine can be directly converted to pyruvate to serve as an energy source, but this reaction requires an essential iron-sulfur cluster in the Serine Dehydratase enzyme, which can be sensitive to the presence of oxygen (Burman et al., 2004). Similarly, arginine can serve as an ATP source via the arginine deiminase system, however activation of this operon by ArcR, an iron-sulfur containing transcriptional activator, is sensitive to the presence of oxygen (Makhlin et al., 2007). Thus, the combination of L-lactate and amino acids allows for metabolic partitioning whereby L-lactate is used as an energy source as well as a carbon source and amino acids can provide precursor metabolites that cannot be synthesized from L-lactate in the presence of NO·.

Not all of the L-lactate is converted to acetate for ATP production. In fact, during NO· stress S. aureus converts approximately 40% of consumed lactate into acetate (Figure 2B). The rest of the lactate is likely converted into biomass via two main routes; conversion of acetyl-CoA into fatty acids for membrane lipids and shuttling of carbon into gluconeogenesis. These two processes would bypass the flux of L-lactate carbon into the oxidative branch of the TCA cycle via citrate synthase (GltA). Indeed, S. aureus ΔgltA grows as well as WT on L-lactate/amino acids during NO· stress implying GltA has little use under these conditions (Figure S6). In contrast, shuttling L-lactate carbon into gluconeogenesis requires the activity of pyruvate carboxylase (Pyc), and a Δpyc mutant exhibits a growth defect under these conditions (Figure S6). Thus, in addition to using L-lactate for energy production during NO·-stress, this nutrient also serves as a valuable carbon source via gluconeogenesis. However, Δpyc is significantly more fit than ΔackA implying energy production via acetogenesis is the most important fate of L-lactate under moderate NO· stress.

While it is now clear that Lqo is important for the assimilation of L-lactate during NO· stress, it is not obvious why the Ldh enzymes (Ldh1 and Ldh2) cannot also perform this function. The lack of NAD-dependent dehydrogenase contribution to growth on lactate/amino acids is exemplified by the inability of D-lactate to support growth under NO· stress. The only method of assimilating D-lactate is via the NAD-dependent Ddh enzyme because Lqo does not react with the D-enantiomer of lactate. Thus, NAD-dependent LDHs in S. aureus are unable to synergize with Mqo and afford the pathogen NO·-resistance in the presence of L-lactate/peptides. NAD-dependent lactate dehydrogenases generally produce lactate in order to restore redox balance through the production of NAD⁺. Thus, these enzymes become highly expressed upon redox imbalance via the inactivation of the Rex repressor. Rex will bind DNA and repress transcription until the build-up of NADH reduces the affinity of Rex for DNA. Given that intermediate NO·-levels allow for residual respiration, redox imbalance is not severe enough for robust Ldh and Ddh expression. Indeed, Δrex mutants become able to grow on the combination of D-lactate/amino acids in the presence of NO· implying that the expression of the NAD-dependent dehydrogenases are not high enough to support growth under these conditions (data not shown). Furthermore, the highly active Ldh1 enzyme is not abundantly produced in the absence of glucose regardless of redox imbalance by as of yet unknown mechanisms (Crooke et al., 2013). These facts explain the absolute dependency on Lqo for energy production during NO· stress in high lactate/peptide environments. The requirement of both Lqo and Mqo for full virulence stems from their concerted use of L-lactate and amino acids during infection. This results in a non-additive contribution of both enzymes in that the double mutant exhibits attenuation to similar extents as each single mutant with respect to organism burdens in kidney and liver abscesses. This attenuation is most pronounced in the liver, an environment known to require residual respiration for full virulence (Hammer et al., 2013). Here we demonstrate that the previously reported requirement for the terminal oxidase (QoxABCD) for both NO·-resistance and liver infection can be accounted for by the concerted use of L-lactate and peptides during NO· stress in this tissue (Hammer et al., 2013). The additive nature of the two enzymes with respect to weight loss is still poorly understood. Blocking full TCA cycle activity and limiting energy production and gluconeogenesis via loss of Lqo and Mqo may limit virulence factor production to an extent that reduced virulence without reduced organism burdens. Indeed, Δmqo mutants exhibit reduced hemolysis (data not shown). Thus, the additive effect of deleting both lqo and mqo on virulence factor production and disease severity is still under investigation but could explain the reduced morbidity of the double mutant compared with the single mutants despite similar organism burdens. However, the non-additive nature of these mutants in the context of organism burden strengthens the potential benefit of developing inhibitors that block the activities of both Mqo and Lqo. Resistance to such an agent could require gain-of-function mutations in both genes, a very unlikely genetic event. Thus, understanding the commonalities in the enzymatic activities of these two enzymes will aid in developing inhibitors active against both Lqo and Mqo enzymes.

Initial characterization of the reaction mechanism inherent to these enzymes revealed typical acid/base dehydrogenase chemistry whereby the substrates are coordinated by an arginine residue (Arg397), then oxidized via a flavin cofactor with a histidine accepting the extracted proton (His58). These residues are completely conserved in all members of the staphylococcal Mqo/Lqo family (Figure S2). The reaction mechanism that describes the flow of electrons from reduced flavin to the quinone pool is still unknown as is whether these enzymes act as multimers or how they interface with the membrane. Given the relatedness of Lqo and Mqo family members, it is likely that these processes are also conserved across the staphylococci. Thus, targeting commonalities between these proteins could yield agents that would limit both enzymes resulting in the attenuation observed in ΔmqoΔlqo strains. A broadly acting compound should avoid targeting features of these enzymes that confer substrate specificity. In order to learn how such related enzymes distinguish between L-lactate and malate, we tested a divergent amino acid at the base of the predicted substrate-binding pocket that seemed poised to confer substrate specificity. Lqo possesses a negatively charged Asp residue at position 285 that could occlude malate from the binding pocket. Mqo, on the other hand, contains a positively charged His at this position that could aid in stabilizing the dicarboxylate malate during catalysis. Our data are consistent with the carboxylate of D285 in Lqo occluding the β-carboxylate of malate but having little affect on reactivity with L-lactate. The analogous H284 in Mqo is required for the oxidation of malate however we were unable to “switch” substrates between Mqo and Lqo by modifying this residue in the two respective enzymes. Because malate is simply a carboxylated L-lactate, one might expect that the substrate specificity of these enzymes involves only a few interacting active site residues. Indeed, the specificity of Bacillus stearothermophilus LDH was switched to malate by the substitution of a single amino acid (Gln102 to Arg) (Wilks et al., 1988). However, in other instances, substrate specificity in these enzymes is more complex. For instance, to alter Escherichia coli MDH to oxidize L-lactate instead of malate required a five-residue change, although a single change abolished enzyme activity toward malate (Arg81 to Gln) (Yin and Kirsch, 2007). Thus, more work is required to understand how these related enzymes recognize divergent substrates and how we can inhibit the whole Mqo/Lqo family a single small molecule. Given their role in S. aureus NO-resistance and virulence, these efforts may have measurable impact on human health.

Experimental Procedures

Bacterial strains and growth conditions

S. aureus strains (Table 4) were grown in either Brain Heart Infusion medium (BHI) or in Pattee/Neveln (PN) medium. PN is a chemically defined medium consisting of a phosphate buffer, nitrogen and sulfur sources [(NH₄)₂SO₄ and MgSO₄], amino acids, nucleic acid bases and vitamins (thiamine, niacin, biotin and pantothenic acid) (Pattee and Neveln, 1975; Vitko and Richardson, 2013). For growth experiments, PN was supplemented with a carbon source (0.5% of both Casamino acids and L-lactate or pyruvate when indicated). Cultures were shaken at 250 rpm at 37°C. Growth was monitored at an absorbance of 650 nm using a Tecan Infinite M200 plate reader in 200 μL cultures in a 96-well plate. Where applicable, 15mM DETA/NO was added at OD₆₆₀ 0.15. Where applicable, 0.625–20mM fresh sodium azide was added to the wells at inoculum. S. aureus locus tags come from the COL genome. S. epidermidis locus tags come from the RP62A genome. S. saprophyticus locus tags come from the ATCC 15305 genome.

Table 4.

Strains, plasmids and primers

Strain	Description	Reference
RN4220	Restriction Deficient Methicillin Sensitive S. aureus	W. Shafer
COL	Methicillin Resistant S. aureus Clinical Isolate	MRSA Clinical Isolate, 1961
Newman	Methicillin Sensitive S. aureus Clinical Isolate	Osteomyelitis Isolate, 1952
LAC	USA300 Methicillin Resistant S. aureus Clinical Isolate	NARSA
M15	E. coli pQE60 expression strain with pREP4 plasmid	Qiagen
AR0369	S. aureus COL Δlqo::SpR	(Fuller et al., 2011)
AR0370	S. aureus COL Δldh1::ErR Δldh2::KmR Δlqo::SpR	(Fuller et al., 2011)
AR0915	S. aureus COL Δmqo::SpR	(Fuller et al., 2011)
AR0960	S. aureus COL ΔackA::KmR	This Study
AR1052	S. aureus COL ΔgltA::ErR	This Study
AR0763	S. aureus COL Δpyc::ErR	(Vitko et al., 2015)
AR0902	S. aureus Newman Δmqo::SpR	(Fuller et al., 2011)
AR0907	S. aureus Newman Δlqo::SpR	(Fuller et al., 2011)
AR1259	S. aureus Newman Δmqo::ErR Δlqo::SpR	This study
AR1262	S. aureus LAC Δmqo::ErR Δlqo::SpR	This study
AR0922	AR0370 + pNAS37 (S. epidermidis mqo1)	This Study
AR0923	AR0370 + pNAS38 (S. epidermidis mqo2)	This Study
AR0924	AR0370 + pNAS39 (S. epidermidis mqo3)	This Study
AR1476	AR0370 + pNAS61 (S. epidermidis mqo4)	This Study
AR0919	AR0915 + pNAS37 (S. epidermidis mqo1)	This Study
AR0920	AR0915 + pNAS38 (S. epidermidis mqo2)	This Study
AR0921	AR0915 + pNAS39 (S. epidermidis mqo3)	This Study
AR1475	AR0915 + pNAS61 (S. epidermidis mqo4)	This Study
Plasmid	Description	Reference
pQE60	Vector for bacterial expression of C-terminally 6xHis tagged insert	Qiagen
pLasso	Vector for bacterial expression of a cleavable C-terminally 6xHis tagged insert	(Edwards et al., 2013)
pBT2ts	E. coli/S. aureus shuttle vector	(Brückner, 1997)
pBTK	1.5 kb aphA3 (KmR) allele cloned into SmaI site in pBT2ts	(Fuller et al., 2011)
pOS1	S. aureus shuttle vector, CmR	(Schneewind et al., 1992)
pOS1-pLgt	S. aureus shuttle vector with lgt promoter, CmR	(Bubeck Wardenburg et al., 2006)
pJF138	C-term His6-tagged S. aureus Mqo in pQE60	(Fuller et al., 2011)
pJF139	C-term His6-tagged S. aureus Lqo in pQE60	(Fuller et al., 2011)
pJF140	C-term His6-tagged S. epidermidis Mqo1 in pQE60	This study
pJF141	C-term His6-tagged S. epidermidis Mqo2 in pQE60	This study
pJF142	C-term His6-tagged S. epidermidis Mqo3 in pQE60	This study
pJF143	C-term His6-tagged S. epidermidis Mqo4 in pQE60	This study
pNAS1	C-term His6-tagged S. aureus MqoH284D in pQE60	This study
pNAS2	C-term His6-tagged S. aureus LqoH58A in pQE60	This study
pNAS3	C-term His6-tagged S. aureus LqoD285H in pQE60	This study
pNAS4	C-term His6-tagged S. aureus LqoR397A in pQE60	This study
pNAS11	S. aureus lqo in pOS1 for complementation; pLqo	This study
pNAS24	5’ and 3’ homology regions of ackA cloned into pBTK to yield ΔackA::KmR	This study
pNAS37	S. epidermidis mqo1 cloned into pOS1-plgt XhoI/BamHI	This study
pNAS38	S. epidermidis mqo2 cloned into pOS1-plgt XhoI/BamHI	This study
pNAS39	S. epidermidis mqo3 cloned into pOS1-plgt XhoI/BamHI	This study
pNAS61	S. epidermidis mqo4 cloned into pOS1-plgt XhoI/BamHI	This study
pNAS53	CPD C-term His6-tagged S. saprophyticus mqo1 in pLasso	This study
pNAS45	C-term His6-tagged S. saprophyticus mqo2 in pQE60	This study
pNAS52	CPD C-term His6-tagged S. saprophyticus mqo3 in pLasso	This study
pNAS54	S. aureus mqo1 in pOS1-pLgt for complementation; pMqo	This study
Primer	Sequence
ackA-5.1A	TGGTAGAATTCTTGCTAAGACCTATGGGCAC	This study
ackA-5.1B	TGGTAGAATTCTGAACTACCAGCATTGATAGC	This study
ackA-3.1A	TGGTAGGATCCTGATGTTATGACATTCGGTGG	This study
ackA-3.1B	TGGTAGGATCCTCAGCAACAATCTCAGCACC	This study
lqo H58A F	GCTGGTACGGGTGCTGCAGCATTATGTG	This study
lqo H58A R	CACATAATGCTGCAGCACCCGTACCAGC	This study
lqo D285H F	GTTATTGAACAACACCATGCCAAAGTTTATGG	This study
lqo D285H R	CCATAAACTTTGGCATGGTGTTGTTCAATAAC	This study
lqo R397A F	CACTGCTGGTAAAGCTGTACAAGTTATC	This study
lqo R397A R	GATAACTTGTACAGCTTTACCAGCAGTG	This study
mqo1 H284D F	GATTGATCGTCATGATGCTAAAGTGTACG	This study
mqo1 H284D R	CGTACACTTTAGCATCATGACGATCAATC	This study
mqo1 H284A F	GATTGATCGTCATGCTGCTAAAGTGTACG	This study
mqo1 H284A R	CGTACACTTTAGCAGCATGACGATCAATC	This study
SEmqo1.1A	GATCCTCGAGAATACTCAACATAGCAAAACAGATG	This study
SEmqo1.1B	GATCGGATCCCTACTTTACATTCAAGTATTTTTTTACTTCTTCG	This study
SEmqo2.1A	GATCCTCGAGGCTATGTCTGACAAAAAAGAC	This study
SEmqo2.1B	GATCGGATCCTTATTTTGCTTTATCATAGAACCCTAATTCAAG	This study
SEmqo3.1A	GATCCTCGAGGCTAATAAAGAGTCAAAAAATGTTG	This study
SEmqo3.2B	GATCGGATCCTAAAGAGTGAATGGGTTAAACTACG	This study
Mqo4 promoter_1A	GATCGTCGACGTTTGCGACAGGGCAACC	This study
SEmqo4.2B	GATCGGATCCCATTAGACTTTCTACCTAATGCTAC	This study
SSmqo1.1A	GATCGTCGACCCAAGCGTACTTTAC	This study
SSmqo1.1B	GATCGGATCCGTTAATTTTAAGATATTTCG	This study
SSmqo3.1A	GATCGTCGACGGAAGATTCTTATCAAGG	This study
SSmqo3.2B	GATCGGATCCCTATTTATGATTTAATTC	This study
SSmqo4.1A	GATCGTCGACCATACGCACGTGCGTG	This study
SSmqo4.1B	GATCGGATCCCTATTTACTATAATAG	This study
Mqo1_6His.1A	TTTACCATGGCTATGACAACACAACATAGCAAAACAG	(Fuller et al., 2011)
Mqo1_6His.1B	TTTAGGATCCTTTAACTTGTAAATACTTAGTTACTTCT TC	(Fuller et al., 2011)
Mqo2_6His.1A	TTTACCATGGCTAAGTCTAATAGTAAAGACATC	(Fuller et al., 2011)
Mqo2_6His.1B	TTTAGGATCCGTTTTCGTAGTAACCTAATTCTAAGTC	(Fuller et al., 2011)
SE_mqo1_6His.1A	TTTACCATGGCTATGAATACTCAACATAGCAAAACAGATG	This study
SE_mqo1_6His.1B	TTTAGGATCCCTTTACATTCAAGTATTTTTTTACTTCTTCG	This study
SE_mqo2_6His.1A	TTTACCATGGCTATGTCTGACAAAAAAGAC	This study
SE_mqo2_6His.1B	TTTAGGATCCTTTTGCTTTATCATAGAACCCTAATTCAAG	This study
SE_mqo3_6His.1A	TTTACCATGGCTAATAAAGAGTCAAAAAATGTTG	This study
SE_mqo3_6His.1B	TTTAGGATCCTTTAGATTCGTAATAATTTAATTCTAAATCTTTAGATG	This study
SE_mqo4_6His.1A	TTTACCATGGCTATGAGTGAAGCAAATCATAAAAACATCG	This study
SE_mqo4_6His.1B	TTTAGGATCCTCTATTTAAATGTAAGTTTTTAGAAGTTTCG	This study
SS mqo4 CPD F (SSmqo1)	GGAATTCCATATGGCTACTAACAAAGAGTC	This study
SS mqo4 CPD R (SSmqo1)	CGCGGATCCCAGTTTACTATAATAGTTTAGG	This study
SSmqo3_6His.1A (SSmqo2)	TTTACCATGGCTATGAGTGAAAAGAATTCTAAAG	This study
SSmqo3_6His.1B (SSmqo2)	TTTAGGATCCTTTATGATTTAATTCTAATTCAGC	This study
SS mqo1 CPD F (SSmqo3)	GGAATTCCATATGAGTACACAACATAGC	This study
SS mqo1 CPD R (SSmqo3)	CGCGGATCCCAGGTTAATTTTAAGATATTTCG	This study
Lqo.4A2	GATCGCTAGCCACATTACTATCACCGGCA	This study
Lqo.4B	GATCCCCGGG TTAGTTTTCGTAGTAACC	This study
Mqo1-SD.1A	GATCCTCGAGAAAAAGGGGGACTGTATTTG	This study
Mqo1.1B	GATCCTCGAGTATTTAACTTGTAAATACTTAG	This study

Cloning and mutagenesis

Inactivation of ackA was accomplished using a modified allelic exchange method by cloning flanking DNA sequence on either side of a selectable kanamycin marker in the S. aureus/E. coli shuttle vector pBT2ts (pBTK). This construct was passaged through S. aureus RN4220, then electroporated into S. aureus COL as previously described and grown with selection at 30°C. Cointegration was accomplished by growing overnight at 43°C followed by plating on media with antibiotics (Brückner, 1997). Single colonies were picked and grown at 30°C for three consecutive days without antibiotic. Cultures were then diluted 1:100 and grown at 37°C to mid-exponential phase before adding chloramphenicol to inhibit growth of resolved cointegrates. Cycloserine was added at 30 minutes to kill Cm^R-cointegrates. Cultures were incubated at 37°C for 4–6 hours before surviving cells were plated on selective media and screened for allelic replacement by PCR analysis. The mutation was transduced by Φ-80α transduction into a fresh S. aureus COL strain as previously described (Novick, 1991). Antibiotic selection in E. coli was carried out using ampicillin (100 μg/mL). Antibiotic selection in S. aureus was carried out using the following concentrations: chloramphenicol 10 μg/mL, kanamycin 50 μg/mL.

Staphylococcus epidermidis mqo genes were cloned into either the pOS1 shuttle vector (SalI-BamHI sites) or pOS1-plgt with a constitutive promoter (XhoI-BamHI sites). The resulting plasmids were passaged through RN4220 and then electroporated into COL strains for in vivo complementation analysis (Brückner, 1997). For His-tagged protein purification, Staphylococcal mqo genes were cloned into either pQE60 (Qiagen) as previously described (Fuller et al., 2011) or into pLasso (Edwards et al., 2013), which adds a C-terminal, cleavable CPD His₆ tag.

Lqo/Mqo site-directed mutagenesis was accomplished by overlapping PCR. LqoH58A was made using Lqo H58A F and Lqo H58A R internal primers. LqoD285H was made using Lqo D285H F and Lqo D285H R internal primers. LqoR397A was made using Lqo R397A F and Lqo R397A R internal primers. MqoH284D was made using Mqo H284D F and Mqo H284D R internal primers. MqoH284A was made using Mqo H284A F and Mqo H284A R internal primers. Mqo2_6His.1A and Mqo2_6His.1B were the external primers used to make the Lqo point mutants, while Mqo1_6His.1A and Mqo1_6His.1A were using to make the Mqo mutants. Fragments of lqo or mqo were amplified from S. aureus COL and the joined mutagenized genes were cloned into EcoRI-BamHI digested pQE60. pQE60 plasmids were electroporated into E. coli M15 containing the pREP-4 plasmid before purification.

Lqo purification and analyses

C-terminal His₆-tagged versions of S. aureus Lqo and Mqo alleles and S. epidermidis Mqo proteins were constructed as previously described in the pQE60 vector system (Fuller et al., 2011). For purification of wild type and mutant proteins, 1 L of terrific broth supplemented with B vitamins (10uM riboflavin, niacin and pantothenic acid) was seeded with 5 mL overnight cultures and grown at 37°C with selection (200 μg/mL ampicillin, 25 μg/mL kanamycin). Once cultures reached an OD₆₆₀ of 0.6, 0.1mM IPTG was added and cultures were grown at 18°C overnight with shaking. Cells were pelleted and resuspended in lysis buffer (20 mM Tris-HCl, 0.5 M NaCl, 10 mM imidazole; pH 7.9) prior to lysis by sonication. Enzymes were purified using Ni-NTA Superflow system (Qiagen) according to manufacturer instruction using wash (20 mM Tris-HCl, 0.5 M NaCl, 40 mM imidazole; pH 7.9) and elution buffers (20 mM Tris-HCl, 0.5 M NaCl, 250 mM imidazole; pH 7.9). Protein yields were quantified by nanodrop.

Enzymatic activity for Mqo and Lqo derivatives was assayed as previously described by addition of 250 ng purified protein (1 μg if using non-specific substrate) in a 200 μL reaction containing 1.4 mM menaquinone, 5 μM flavin adenine dinucleotide (FAD), 40 μg NBT, varying concentrations of substrate (malate/L-lactate) and a 50/50 mixture of cardiolipin and phosphotidylethanolamine vesicles (Fuller et al., 2011). Reactions were initiated by addition of substrate and reduction of NBT was monitored at 585 nm.

Enzymatic determination of metabolite levels

Acetate and L-lactate levels were quantified as previously described as per manufacturer instructions (R-Biopharm) (Richardson et al., 2008). Extracellular metabolite levels were determined using S. aureus supernatants from 200 μL cultures grown in the Tecan plate reader overnight. Cultures were heat inactivated at 70°C prior to isolation of the supernatant. Significance was determined using Student’s t-test (2-tailed).

Determination of intracellular ATP levels

ATP levels in bacteria grown under NO· stress were measured using the BacTiter-Glo reagent (Promega). S. aureus strains were grown in PN medium plus carbon source (0.5% Casamino acids, 0.5% L-lactate or pyruvate) in 96-well plates. 15mM DETA/NO was added to the wells once cells reached an OD₆₅₀ of 0.15. Fifteen minutes (T₀) after NO· addition, 100 μL of culture was added to an equal amount of BacTiter-Glo reagent and luminescence was recorded every minute for 10 minutes. For analysis, the maximum luminescence value was used. Following T₀, cultures continued to grow in the presence of NO· and luminescence was measured every hour for four hours. A standard curve of ATP in PN medium was generated to calculate ATP concentrations.

Structural Modeling

The crystallographic structure of GlpO from Streptococcus (PDB ID 2RGH) was identified as a template for modeling the structure of Lqo using HHpred (http://toolkit.tuebingen.mpg.de/hhpred) (Söding et al., 2005). Protein homology was detected by HMM-HMM comparison (Söding, 2005). The calculated probability was 100% and the expectation value was e⁻³⁸, indicating confident homology between the GlpO template and Lqo. A homology model of Lqo was built with 2RGH as a template using Modeller (Sali and Blundell, 1993).

Neighbor-joining trees for publically available Mqo/Lqo alleles across all bacterial species and within the staphylococci were generated using Geneious Pro software (v. 5.5.8) using a Blosum32 matrix with gap open/extension penalties of 12/3.

Circular dichroism

Circular dichroism was performed with a Chirascan Plus automatic spectrometer. Lqo, Mqo and mutant proteins (0.25 mg/mL) were analyzed in 10mM sodium phosphate buffer (pH 7.4) with 0.5M ammonium sulfate (1M for Mqo1 proteins) between 200 and 260 nm. The spectrum of buffer or buffer plus additives has been subtracted from the spectrum of the corresponding samples to perform baseline correction.

Macrophage assay

RAW 264.7 macrophages were suspended in RPMI 1640 (Gibco) supplemented with 10% FBS and 25mM HEPES at a concentration of 1x10⁶ cells/ml, seeded into the wells of a 48-well plate at 0.5mls/well, and incubated for 18hrs at 37°C. The cells were then activated via incubation in RPMI + 20ng/ml IFN-γ for 6hrs at 37°C, and then inoculated with opsonized S. aureus at an MOI of 10:1. After a 30min. incubation at 37°C, the cells were washed twice with PBS, and then incubated in RPMI + 100μg/ml Gentamycin for 1hr at 37°C. The infected RAW cells were then washed twice with PBS, after which certain wells were treated with 0.01% Triton X-100 to induce lysis for bacterial enumeration via dilution plating (t = 0) while the remaining infected RAW cells were incubated in RPMI + 12μg/ml Gentamycin at 37°C for an additional 20 hrs and then lysed.

Murine infection models

Four to six-week-old female C57BL/6 mice from Jackson Labs (Bar Harbor, ME USA) were inoculated via tail vein with 5x10⁷ cfu in 100μl of S. aureus strain Newman or isogenic mutants. Weight loss was monitored daily for 7 days as an indicator of disease progression with mice exhibiting greater than 30% weight loss being sacrificed as per IACUC approved protocol. Bacteria burdens from kidneys and livers were determined 7 days post-inoculation.

For virulence assessment in the skin abscess model, 6–8 week old female C57BL/6 were anesthetized (Avertin, i.p. 250 mg/kg), then shaved followed by subcutaneous inoculation with 1 x 10⁸ CFU WT or ΔmqoΔlqo S. aureus LAC in 20 μl of sterile PBS. On day 7, mice were euthanized and the abscesses were removed, homogenized in 500 μl of PBS, and dilution plated on BHI medium to enumerate CFU.

Supplementary Material

Supp Fig S1. Figure S1. ATPase inhibition by azide disproportionately impairs S. aureus growth on amino acids.

S. aureus wild type COL was grown in PN medium with either 0.5% glucose or Casamino acids as a carbon source. Increasing amounts of azide were added to the media at inoculum.

Supp Fig S2. Figure S2.

(Top) ClustalW2 sequence alignment of Mqo and Lqo proteins from S. aureus, S. epidermidis and S. saprophyticus. (Bottom) CD spectra comparison of wild type purified Mqo or Lqo proteins with selected point mutants.

Supp Fig S3. Figure S3. Complementation of S. aureus Δlqo or Δmqo mutants with alleles from S. epidermidis.

TOP LEFT: Δmqo S. aureus COL grown in PN medium with amino acids as a carbon source expressing Mqo1-4 from S. epidermidis. TOP RIGHT: Δldh1Δldh2Δlqo S. aureus COL grown in PN medium with L-lactate as a carbon source expressing Mqo1-4 from S. epidermidis. BOTTOM: Δmqo or Δldh1Δldh2Δlqo S. aureus COL each expressing Mqo1-4 from S. epidermidis grown in PN medium with combined amino acids and L-lactate carbon sources (0.5% each) exposed to NO· (15 mM DETA/NO).

Supp Fig S4. Figure S4. Genomic contexts of staphylococcal lqo and mqo genes.

A. Comparison of genomic contexts of staphylococcal mqo genes. B. Comparison of genomic contexts of staphylococcal lqo genes. Abbreviations: aur, S. aureus; cap, S. captis; car, S. carnosus; epi, S. epidermidis; hae, S. haemolyticus; hom, S. hominis; lug, S. lugdunensis; sap, S. saprophyticus; war, S. warneri; xyl, S. xylosus.

Supp Fig S5. Figure S5. Mqo and Lqo are dispensible for intracellular growth in macrophages.

RAW246.7 macrophage-like cells were infected with an MOI of 10:1 of S. aureus Newman wild type or ΔmqoΔlqo double mutant. After 20 hours, macrophages were lysed and bacterial cells were enumerated. Values represent two independent experiments, (p > 0.05 WT vs ΔmqoΔlqo).

Supp Fig S6. Figure S6. Proposed carbon flux during low level NO·-stress with residual respiratory activity.

Lqo-derived pyruvate carbon can enter the TCA cycle via citrate synthase (GltA) or via anapleurotic oxaloacetate production catalyzed by pyruvate carboxylase (Pyc). Mutants lacking pyc showed modest growth defects while ΔgltA mutants exhibit WT NO·-resistance under these conditions. Furthermore, Δacn mutants showed no growth defect under these conditions suggesting that flux through the non-oxidative branch of the TCA cycle is minimal. In contrast, loss of Mqo or PckA resulted in little growth in the presence of NO· suggesting that the oxidative branch of the TCA cycle and gluconeogenesis are critical in this environment. Finally, as expected, very little flux through Pyk to pyruvate is necessary given that lactate serves as the primary pyruvate source via Lqo. Growth rates were calculated over the first four hours following NO· exposure at OD₆₆₀ = 0.15 and are represented as green if rates are not significantly less than WT, orange if significant at a p ≤ 0.05 and red if significant at a p ≤ 0.01, Student’s t-test (NO·-donor = 15 mM DETA/NO).