Glycolytic dependency of high-level nitric oxide resistance and virulence in Staphylococcus aureus.

UNLABELLED: Staphylococcus aureus is a prolific human pathogen capable of causing severe invasive disease with a myriad of presentations. The ability of S. aureus to cause infection is strongly linked with its capacity to overcome the effects of innate immunity, whether by directly killing immune cells or expressing factors that diminish the impact of immune effectors. One such scenario is the induction of lactic acid fermentation by S. aureus in response to host nitric oxide (NO·). This fermentative activity allows S. aureus to balance redox during NO·-induced respiration inhibition. However, little is known about the metabolic substrates and pathways that support this activity. Here, we identify glycolytic hexose catabolism as being essential for S. aureus growth in the presence of high levels of NO·. We determine that glycolysis supports S. aureus NO· resistance by allowing for ATP and precursor metabolite production in a redox-balanced and respiration-independent manner. We further demonstrate that glycolysis is required for NO· resistance during phagocytosis and that increased levels of extracellular glucose limit the effectiveness of phagocytic killing by enhancing NO· resistance. Finally, we demonstrate that S. aureus glycolysis is essential for virulence in both sepsis and skin/soft tissue models of infection in a time frame consistent with the induction of innate immunity and host NO· production. IMPORTANCE: Staphylococcus aureus is a leading human bacterial pathogen capable of causing a wide variety of diseases that, as a result of antibiotic resistance, are very difficult to treat. The frequency of S. aureus tissue invasion suggests that this bacterium has evolved to resist innate immunity and grow using the nutrients present in otherwise sterile host tissue. We have identified glycolysis as an essential component of S. aureus virulence and attribute its importance to promoting nitric oxide resistance and growth under low oxygen conditions. Our data suggest that diabetics, a patient population characterized by excess serum glucose, may be more susceptible to S. aureus as a result of increased glucose availability. Furthermore, the essential nature of S. aureus glycolysis indicates that a newly developed glycolysis inhibitor may be a highly effective treatment for S. aureus infections.

INTRODUCTION

Staphylococcus aureus is a Gram-positive bacterium capable of causing a wide variety of clinical manifestations in humans, including skin and soft tissue infections (SSTIs), pneumonia, bacteremia, endocarditis, and osteomyelitis. S. aureus infections are more common, invasive, and severe than those caused by coagulase-negative staphylococci (CNS), despite the shared colonization of human skin and frequent resistance to antibiotics (1, 2). The traditional explanation for this difference has been the acquisition of many unique virulence factors by S. aureus that promote colonization of fresh wounds and destruction of host tissues (3). Complementary to this concept is the hypothesis that an increased incidence of tissue invasion by S. aureus has led to metabolic adaptation that further distinguishes S. aureus from CNS. One instance where this is particularly evident is for the growth of S. aureus in the presence of the innate immune radical nitric oxide (NO·).

NO· is a small, membrane-permeable gas produced by activated phagocytes via an inducible nitric oxide synthase (iNOS or NOS2) (4). In this context, NO· acts as a broad-spectrum antimicrobial agent with a strong propensity for disrupting microbial metabolism, DNA replication, and lipid integrity (5). The effects of NO· are caused by direct binding of NO· to transition metals and/or organic radicals and indirectly via the formation of reactive nitrogen species (RNS), which themselves bind thiols, amines, and hydroxyl aromatic groups (6). Due to the high concentration of iron and thiol-containing enzymes in bacterial respiratory chains, a primary effect of NO· at levels of ≥500 nM is the disruption of cellular respiration (7, 8). In the context of infection, activated phagocytes can generate ≥1 µM NO· intracellularly (9, 10). While S. aureus strains can enzymatically detoxify NO· using the flavohemoprotein Hmp, mutants lacking this enzyme are still capable of replicating during nitrosative stress (11). Furthermore, despite possessing Hmp, CNS are incapable of growing in the presence of high levels of NO· (levels sufficient to inhibit cellular respiration) (8). Instead, what distinguishes S. aureus from CNS during growth under high levels of NO· is lactic acid fermentation.

Fermentation promotes S. aureus NO·-resistant growth by converting NADH into NAD+ (i.e., balancing redox), which allows for continued carbon catabolism via NAD+-dependent pathways. During NO· exposure, S. aureus performs heterolactic acid fermentation using three lactic acid dehydrogenases (Ldh1, Ldh2, and Ddh) (8). Interestingly, CNS lack the ldh1 allele and do not ferment lactate in response to NO· stress, despite possessing anaerobically responsive ddh and ldh2 alleles (8). The reason for this discrepancy is currently unknown. However, we have demonstrated that NO·-dependent induction of ldh1 in S. aureus is glucose dependent, leading us to consider the possibility that S. aureus growth in the presence of high NO· may itself be carbon source dependent (12).

Given the large contribution of NO· resistance to S. aureus infection and the metabolic basis for this unique resistance, we undertook a more extensive examination of the substrates that support S. aureus resistance to high levels of NO· (13). Our results indicate that S. aureus requires specific glycolytic substrates to grow in the presence of high levels of NO· in vitro and during phagocytosis. Furthermore, we demonstrate that glycolysis, but not gluconeogenesis, is required for S. aureus pathogenesis during murine infection. Together these results clearly establish the metabolic programming required for S. aureus to establish invasive infection, and in agreement with clinical observations, implicate the host glycemic state as an important factor in determining susceptibility to this pathogen.

RESULTS

Fermentative growth of S. aureus during NO· stress is glycolysis dependent.

Unlike other pathogenic bacteria, including closely related staphylococci, S. aureus is capable of growing in the presence of high levels of nitric oxide (0.2 to 1 mM NO·) (8, 11). The identification of ldh1 as a unique S. aureus NO· resistance factor implicates metabolic adaptation, specifically the induction of fermentative metabolism, as an important feature of high-level NO· resistance. However, little is known about the metabolic substrates and central metabolic pathways that support this resistance. Therefore, we examined the abilities of 10 different carbon sources (carbon balanced) to support growth of S. aureus under high NO· stress. High NO· stress was achieved using a mixture of NO· donors (1 mM diethylamine nitric oxide [DEANO] and 10 mM NOC-12), which achieve maximum [NO·] = ~1 mM by 1.5 h and maintain levels ≥500 nM for ~5 h. The carbon sources are loosely divided into two groups: glycolytic (glucose, sorbitol, mannose, and mannitol) and nonglycolytic (glycerol, gluconate, ribose, pyruvate, l-lactate, and Casamino Acids) (see Fig. S1 in the supplemental material). Only a subset of glycolytic carbon sources (hexoses) supported the growth of S. aureus in the presence of high NO· concentrations, while all of the nonglycolytic carbon sources failed to support growth (Fig. 1A and B). These phenotypes were confirmed using two additional S. aureus strains (Newman and SF8300) (data not shown).

FIG 1 

NO·-resistant growth of S. aureus is hexose dependent. (A) S. aureus COL grown at 37°C in CDM alone or with NO· mix with glucose (25 mM), pyruvate (50 mM), or Casamino Acids (0.5%) as the primary carbon source (n = 3; error bars show the standard errors of the means [SEM]). NO· mix was added at mid-log phase, and absorbance was normalized to the time of NO· addition (t0) (n = 3; error bars show SEM). Time (in hours) is shown on the x axis. (B) Average growth rates of S. aureus COL in CDM plus various primary carbon sources (carbon balanced to 25 mM glucose) following NO· exposure (NO· mix), normalized to the average mid-log aerobic growth rate in each carbon source (n = 3; error bars show SEM). Significance was calculated using a Student’s two-sided t test (*, P ≤ 0.05; **, P ≤ 0.01).

Given the glycolytic dependency of S. aureus NO· resistance, we examined the contribution of the glycolysis-responsive regulator CcpA to growth under high nitrosative stress. However, there was no significant difference in the normalized maximum growth rates of wild-type and ccpA::Spr S. aureus during high nitrosative stress (47.1% ± 5.5% and 51.9% ± 5.4%, respectively). These data suggest that the dependency on hexoses for NO· resistance does not stem from CcpA-mediated induction of an unknown NO· resistance factor.

Respiration inhibition partially explains the glycolysis dependence of S. aureus NO· resistance.

The primary effect of high NO· levels is the inhibition of cellular respiration. Thus, we postulated that anaerobic growth of S. aureus would also be glycolysis dependent. Consistent with this hypothesis, growth of S. aureus during anaerobiosis was restricted to the same set of carbon sources that support NO· resistance, with one exception, pyruvate (Fig. 2A). Addition of the alternative electron acceptor nitrate (50 mM) to the media improved the anaerobic growth of S. aureus on all carbon sources. These anaerobic growth phenotypes were confirmed in two additional S. aureus strains (Newman and SF8300 [data not shown]). Furthermore, the respiration-deficient S. aureus menD::Err mutant, which lacks the ability to synthesize menaquinone electron carriers, exhibited the same carbon source-dependent growth under aerobic conditions as wild-type S. aureus grown anaerobically (e.g., glucose and pyruvate supported growth, but amino acids did not) (Fig. 2B). These data are consistent with NO·-dependent respiration inhibition as the primary cause of the observed carbon source constraints for S. aureus NO·-resistant growth, albeit not for pyruvate.

FIG 2 

S. aureus growth on most carbon sources is respiration dependent. (A) Maximum absorbance (650 nm) of S. aureus COL grown at 37°C for 24 h anaerobically in CDM alone or with potassium nitrate (50 mM) with various primary carbon sources (carbon balanced to 25 mM glucose) (n = 3; error bars show SEM). (B) Aerobic growth of wild-type (WT) S. aureus Newman and menD::Err strain at 37°C for 32 h in CDM with glucose (0.5%), pyruvate (1%), or Casamino Acids (1%) as the primary carbon source (n = 3; error bars show pooled standard deviations [SD]).

Fermentative growth of S. aureus on pyruvate requires acetogenesis.

The observation that S. aureus growth on pyruvate is respiration independent suggests that NO· may directly inhibit enzymes necessary for fermentative growth on this substrate. To identify what these targets may be, we first compared the excretion of several key metabolic end products (acetate, formate, lactate, and ethanol) during anaerobic growth of S. aureus on either glucose or pyruvate (see Fig. S1 and S2 in the supplemental material). Growth on glucose resulted in a mixture of end product metabolites, a large proportion of which are important for balancing redox stress (l-lactate, d-lactate, formate, and ethanol [EtOH]). S. aureus grown anaerobically on pyruvate, on the other hand, generated significantly less lactate and ethanol and significantly more acetate and formate. This result is consistent with high-level flux of pyruvate to acetate for energy production concomitant with a shift from energy-producing glycolysis (2×ATP/glucose consumed) to energy-consuming gluconeogenesis (4×ATP/glucose produced) (Fig. S1).

To confirm that gluconeogenic catabolism of pyruvate proceeds via the predicted ATP-consuming pathway in S. aureus (see Fig. S1 in the supplemental material), we measured the growth of wild-type S. aureus, Δpyc mutant (pyc encodes pyruvate carboxylase), and ΔpckA mutant (pckA encodes phosphoenolpyruvate carboxykinase) anaerobically on pyruvate (Fig. S2B). Both mutants failed to grow under these conditions. Additionally, we confirmed that acetogenesis (generates 1×ATP/pyruvate [Fig. S1]) was required for anaerobic growth of S. aureus on pyruvate, but not on glucose, using an S. aureus ΔackA mutant (ackA encodes acetate kinase) (Fig. S2C). These data suggest that S. aureus requires acetogenesis in order to produce sufficient ATP to maintain cellular homeostasis and run gluconeogenesis while growing on pyruvate under nonrespiratory conditions.

Partial inhibition of PDHC by NO· limits acetogenesis and growth of S. aureus on pyruvate.

The requirement for acetogenesis during anaerobic growth on pyruvate suggests that NO· may directly inhibit acetogenic enzymes in S. aureus and thus limit growth. Indeed, previous work has shown that PDHC (pyruvate dehydrogenase complex) and Pfl (pyruvate formate lyase), enzymes that catalyze the first committed step in acetogenesis from pyruvate (see Fig. S1 in the supplemental material), are susceptible to NO· inactivation (8). Furthermore, oxygen also inhibits Pfl activity, and thus, PDHC is likely the dominant enzyme for converting pyruvate to acetyl coenzyme A (acetyl-CoA) during our in vitro NO· growth assays (14). However, NO· must not completely inactivate PDHC, as acetyl-CoA is an essential precursor metabolite for growth on any carbon source. Thus, we postulated that partial inhibition of PDHC by NO· might be growth limiting for S. aureus on pyruvate, but not hexoses, as a result of an increased need for carbon flux through acetyl-CoA to acetate for energy production.

In support of this hypothesis, the S. aureus ΔpflAB mutant (pflAB encodes Pfl) was more susceptible to the PDHC inhibitor triphenylbismuthdichloride (TPBC) when grown anaerobically on pyruvate (MIC = 1.5 µM) than glucose (MIC = 4 µM) (15). Additionally, NO· treatment during growth on pyruvate significantly reduced the intracellular concentration of acetyl-CoA and the “flux” of pyruvate to acetate in wild-type S. aureus compared to aerobic and anaerobic growth conditions (Fig. 3A and B). Finally, NO· treatment significantly reduced intracellular ATP levels for S. aureus on pyruvate but did not decrease ATP levels for S. aureus grown on glucose (Fig. 3C). Given the propensity of bacteria to maintain a constant intracellular level of ATP/cell and the increased demand for acetogenic flux in S. aureus grown on pyruvate during nonrespiratory conditions, these data suggest that partial inhibition of PDHC/Pfl by NO· inhibits S. aureus growth on pyruvate by limiting ATP production and gluconeogenic flux.

FIG 3 

NO· exposure limits acetogenic flux of pyruvate. (A) Average intracellular acetyl-CoA levels for S. aureus COL grown at 37°C in CDM with pyruvate (1.1%) as the primary carbon source (n = 3; error bars show SEM). (B) Average acetogenic “flux” of pyruvate (acetate produced/pyruvate consumed) for S. aureus COL grown at 37°C in CDM with pyruvate (50 mM) as the primary carbon source (n = 3; error bars show SEM). (C) Average intracellular ATP levels (in micromolar per OD650) of S. aureus COL grown aerobically with or without NO· mix in CDM with glucose (25 mM) or pyruvate (50 mM) as the primary carbon source (n = 3; error bars show SEM). Significance was calculated using a Student’s two-sided t test (*, P ≤ 0.05; **, P ≤ 0.01; N.S., not significant).

Intracellular S. aureus survival requires NO· resistance.

During infection, NO· levels within phagosomes of activated phagocytes surpass those known to completely inhibit S. aureus respiration (9, 16). Consistent with this, a lactate dehydrogenase-deficient mutant of S. aureus (Δldh1Δldh2Δddh mutant) was significantly attenuated for survival in RAW 264.7 macrophages compared to wild-type S. aureus (Fig. 4A) (the survival kinetics for wild-type S. aureus were the same for S. aureus strains COL, Newman, and SF8300 [data not shown]). Furthermore, the survival defect of the Δldh1Δldh2Δddh mutant was completely abrogated by treating the macrophages with the iNOS inhibitor N-iminoethyl-l-lysine (l-NIL). We validated the efficacy of l-NIL treatment by demonstrating reduced levels of extracellular nitrite in macrophages (see Fig. S3A and S3B in the supplemental material). Given that lactic acid fermentation in response to NO· is unique to S. aureus, we compared the survival of wild-type S. aureus to Staphylococcus haemolyticus and Staphylococcus saprophyticus. Both CNS exhibited significantly greater susceptibility to macrophage killing than S. aureus (Fig. 4B). l-NIL treatment largely reversed these differences, suggesting that S. aureus NO· resistance promotes survival during phagocytosis.

FIG 4 

Lactate fermentation enables S. aureus NO· resistance during phagocytosis. (A) Percent survival of wild-type S. aureus COL and Δldh1Δldh2Δddh mutant in activated RAW 264.7 macrophages with l-NIL (100 nM) or without l-NIL at 20 h postinoculation (MOI = 10:1) (n = 3; error bars show pooled SD). (B) Percent survival of wild-type S. aureus COL (SA), S. haemolyticus (SH), and S. saprophyticus (SS) in activated RAW 264.7 macrophages with or without l-NIL (100 nM) at 20 h postinoculation (MOI = 10:1) (3 ≤ n ≤ 4).The fold effect of l-NIL treatment on bacterial survival is indicated above the bars. Statistical significance was assessed using a Student’s two-sided t test (**, P ≤ 0.01; N.S., not significant).

S. aureus NO· resistance during phagocytosis is glycolysis dependent.

Given our in vitro data, we hypothesized that S. aureus survival during phagocytosis would be glycolysis dependent. To examine this possibility, we constructed S. aureus mutants with mutations in several predicted glycolytic genes (pfkA [encodes phosphofructokinase], pyk [encodes pyruvate kinase], and gapA [encodes glyceraldehyde-3-phosphate dehydrogenase A]) and gluconeogenic genes (pyc, pckA, and gapB [encodes glyceraldehyde-3-phosphate dehydrogenase B]) in several strain backgrounds (see Fig. S1 in the supplemental material). To confirm the predicted metabolic roles of these genes, we grew the mutants aerobically in media with either glucose or Casamino Acids as the primary carbon source. Both S. aureus ΔpfkA and Δpyk mutants were unable to grow on glucose (i.e., encode glycolytic enzymes), while S. aureus ΔpckA and ΔgapB mutants were unable to grow on Casamino acids (i.e., encode gluconeogenic enzymes) (Fig. S4A and S4B). All other mutants exhibited either no observable phenotype or very small growth defects (Fig. S4C and S4D). These phenotypes were confirmed using two additional S. aureus strains (Newman and SF8300 [data not shown]). Complementation restored growth of all S. aureus COL mutants on their respective carbon sources (Fig. S3E and  S3F).

Next, we examined the survival of the glycolytic (ΔpfkA and Δpyk) and gluconeogenic (ΔpckA and ΔgapB) mutants during phagocytosis by RAW 264.7 macrophages. Both glycolytic mutants (ΔpfkA and Δpyk mutants) exhibited significantly reduced survival at 12 and 20 h postinfection (hpi), while the gluconeogenic mutants (ΔpckA and ΔgapB mutants) displayed no survival defect (Fig. 5A). Attenuation of the glycolytic mutants was not a result of differential bacterial uptake (~75% for all strains) and was completely reversed by l-NIL treatment (Fig. 5B). Furthermore, survival of wild-type S. aureus was significantly improved by the addition of increasing amounts of glucose to the tissue culture medium (Fig. 5C). In the absence of NO·, excess glucose had no effect on bacterial survival. Furthermore, glucose treatment did not affect macrophage NO· production (Fig. S3A and S3B). Altogether, these data suggest that S. aureus has access to a variety of nutrient sources within macrophages, but host NO· necessitates the specific utilization of glycolytic carbon sources as a result of respiration inhibition.

FIG 5 

S. aureus requires glycolysis to resist NO· during phagocytosis. (A) Percent survival of S. aureus COL central metabolic mutants in activated RAW 264.7 macrophages at 12 and 20 hpi (3 ≤ n ≤ 6; error bars show pooled SD). (B) Percent survival of S. aureus COL glycolytic mutants in activated RAW 264.7 macrophages with or without 100 nM l-NIL at 20 hpi (5 ≤ n ≤ 6; error bars show pooled SD). The fold effect of l-NIL treatment on bacterial survival is indicated above the bars. (C) Percent survival of S. aureus COL in activated RAW 264.7 macrophages at 12 h with or without l-NIL treatment in RPMI 1640 with increasing amounts of glucose (n = 3; error bars show pooled SD). Statistical significance was calculated using a Student’s two-sided t test (*, P ≤ 0.05; **, P ≤ 0.01; NS, not significant).

S. aureus virulence in mice is glycolysis dependent.

To determine the greater contribution of glycolysis and gluconeogenesis to S. aureus pathogenesis, we infected C57BL/6 mice intravenously with wild-type S. aureus and S. aureus Δpyk, ΔpfkA, and ΔpckA mutants. Infection with wild-type and gluconeogenic-defective S. aureus ΔpckA mutant produced similar results: both strains colonized and proliferated in the kidney and liver (Fig. 6A and B), and both strains induced ~20% weight loss in infected mice by day 5 (data not shown). Organ burdens and weight loss in the wild type and ΔpckA mutant were not significantly different. The glycolytic mutants (Δpyk and ΔpfkA mutants), on the other hand, were significantly attenuated compared to the wild type and were essentially cleared from the kidney and liver by day 2 (Fig. 6A and B and see Fig. S6A and  S6B in the supplemental material). Additionally, the S. aureus Δpyk mutant exhibited a 6-log-unit decrease in tissue burden (day 7) and an inability to form an open lesion in a skin and soft tissue infection (SSTI) model of S. aureus infection (Fig. S5).

FIG 6 

Glycolysis is essential for S. aureus virulence during sepsis. C57BL/6 mice were infected via tail vein injection with 7.5 × 106 CFU of wild-type S. aureus Newman or ΔpckA or Δpyk mutant. Kidneys (A) and livers (B) were harvested on 1, 2, and 5 days postinfection, and bacteria were counted (8 ≤ n ≤ 12). Each symbol represents the value for an individual mouse. The lines indicate mean organ burden. The limit of detection was 100 CFU/organ.

To examine the contribution of host NO· production to the observed attenuation of the S. aureus glycolytic mutants, we infected both wild-type and iNOS−/− C57BL/6 mice with wild-type S. aureus and S. aureus Δpyk and ΔpfkA mutants intravenously. Loss of iNOS resulted in increased burden of wild-type S. aureus in the kidney but did not affect the burden of either glycolytic mutant in either organ (see Fig. S6A and  S6B in the supplemental material). The failure of iNOS deletion to reverse the attenuation of the glycolytic mutants during sepsis suggests that other selective factors maintain the need for bacterial glycolysis even in the absence of host NO· production. Staining of both uninfected tissue and S. aureus tissue abscesses revealed regions of severe tissue hypoxia, suggesting that low oxygen may also necessitate the need for hexose-derived fermentative metabolism independently from NO·-mediated respiration inhibition (Fig. 7 and Fig. S7).

S. aureus infected tissues display acute hypoxia. Representative kidney abscesses harvested from C57BL/6 mice on day 2 post-infection with 7.5 × 106 cfu of wild type S. aureus Newman were stained with either (A) hematoxylin and eosin (H&E) or (B) Hypoxyprobe (pink) and DAPI (blue). Tissues were imaged at 20× magnification using an Olympus B × 60 fluorescence microscope.

DISCUSSION

Unlike other pathogens, S. aureus is capable of growing during high nitrosative stress. S. aureus achieves this unique resistance via the induction of an NO·-resistant metabolic state (8). The substrates and central metabolic pathways that support these activities were previously poorly defined. Using a variety of defined medium conditions, we demonstrated that S. aureus resistance to high levels of NO· (i.e., levels sufficient to inhibit respiration) is glycolysis dependent both in vitro and during phagocytosis. We determined that nonglycolytic carbon sources fail to support NO·-resistant growth due to a combination of respiration-dependent catabolism and partial inhibition of PDHC/Pfl activity. Finally, we showed that glycolysis is required for S. aureus pathogenesis in a murine sepsis and SSTI model of infection.

Interestingly, several large S. aureus virulence screens have failed to identify glycolysis as an essential component of S. aureus virulence (17–19). This can be explained by the fact that the two glycolytic mutants we identified, the Δpyk and ΔpfkA mutants, grow very poorly in the glucose-rich media used to generate the transposon libraries (e.g., brain heart infusion [BHI] medium and tryptic soy broth [TSB] [data not shown]). By removing glucose as a substrate for growth and supplying abundant gluconeogenic substrates, we were able to grow S. aureus ΔpfkA and Δpyk mutants without observable growth phenotypes and identify this metabolic pathway as essential for S. aureus pathogenesis. One study previously noted the importance of glycolysis to S. aureus virulence, although they utilized an invertebrate model and an S. aureus ΔgapA mutant (20). We observed no glycolytic defect for this mutant in three S. aureus strains (COL, Newman, and SF8300), implying that GapB is reversible and can compensate for loss of GapA. This discrepancy could be the result of our ΔgapA mutants separately acquiring compensatory mutations or the fact that the previously reported mutant was constructed in a highly mutagenized strain of S. aureus (NCTC8325-4) (21, 22). Regardless, experimentation with the Δpyk and ΔpfkA mutants demonstrates that glycolysis is essential to S. aureus virulence.

Glycolysis supports S. aureus virulence by promoting growth during innate immune activation. Specifically, a subset of glycolytic carbon sources (hexoses) is compatible with nonrespiratory growth; a selective pressure is generated by host NO· production and abscess formation (i.e., low oxygen). Hexoses support nonrespiratory growth of S. aureus because they allow for redox-balanced ATP generation and precursor metabolite formation, while avoiding catabolic pathways that are directly inhibited by NO·. Other metabolic substrates may require quinone-dependent (and thus respiration-dependent) enzymes (e.g., amino acids and lactate), limit carbon availability via excessive fermentation (e.g., mannitol and sorbitol [Fig. S8]), or require NO·-sensitive enzymes for catabolism (e.g., pyruvate). However, hexose catabolism via glycolysis and heterolactic acid fermentation alone cannot fully explain the growth of S. aureus during NO· stress, as redox balancing under these conditions requires use of all available glucose for lactic acid production. While EtOH fermentation could alleviate this metabolic stress, and does under anaerobic conditions, we do not observe EtOH production by S. aureus following NO· exposure (8). These data suggest that additional, unidentified, redox-balancing reactions support NO·-resistant growth of S. aureus on hexoses.

Likewise, nonrespiratory growth induced by low oxygen and host NO· production is probably not the only selective pressure that necessitates S. aureus glycolytic metabolism. For instance, iron restriction, as experienced during infection, has been shown to induce expression of S. aureus glycolytic and lactic acid fermentation genes and restrict growth of S. aureus on gluconeogenic substrates (23, 24). This is a result of the high abundance of iron in respiratory and tricarboxylic acid (TCA) cycle proteins. Despite this, respiratory metabolism has been shown to be important for S. aureus virulence, as evidenced by the attenuation of terminal oxidase- and heme-deficient mutants in the hearts and livers of infected mice (25). This discrepancy can likely be explained by the spatiotemporal nature of infection. At certain times, S. aureus will have sufficient oxygen and iron to perform respiratory metabolism, but at other times, host iron sequestration, low oxygen, and host NO· production will necessitate glycolytic fermentation. This likely explains the lack of virulence restoration when infecting iNOS−/− mice with the Δpyk and ΔpfkA mutants. While host NO· production likely necessitates Pyk and PfkA activity within activated phagocytes (Fig. 5B), the iron-restricted and hypoxic environment of infected tissue also select for high glycolytic flux in S. aureus. Thus, the glycolytic requirement for full S. aureus virulence is multifaceted, making enzymes in this pathway attractive candidates for antimicrobial development.

Likewise, the concentrations of metabolic substrates available to S. aureus may vary according to tissue and stage of infection. A limited supply of gluconeogenic substrates and robust availability of glycolytic substrates during the early stages of infection fits well with our model of S. aureus metabolic adaptation to tissue invasion and innate immunity (Fig. 8). For instance, glycolytic substrates are essentially absent in sweat, the primary carbon source for skin-associated staphylococci (26, 27). However, glucose is highly abundant in human serum and certain tissues/organs (e.g., the liver) (28, 29). Furthermore, it is well-known that Streptococcus, another Gram-positive extracellular pathogen, relies extensively on carbohydrate metabolism to support growth and virulence (30). Alternatively, intracellular bacteria (e.g., Listeria, Legionella, Mycobacterium, etc.) often rely heavily on gluconeogenic substrates for growth (e.g., fatty acids, glycerol, and amino acids), likely due to the 10:1 nature of gluconeogenic to glycolytic substrates inside the host cell cytoplasm (31–35). While S. aureus lacks many of the predicted glycosidases utilized by streptococci for the catabolism of complex carbohydrates (e.g., mucins), the predominantly extracellular nature of S. aureus pathogenesis suggests that this bacterium may primarily be using mono- and disaccharides to support growth (the same mono- and disaccharides utilized by host cells to support growth).

FIG 8 

Model: tissue invasion necessitates S. aureus glycolytic metabolism. S. aureus skin invasion (1) is mediated by virulence factors and/or host tissue damage and is accompanied by decreased oxygen but increased glucose availability. Following tissue seeding, S. aureus is phagocytized (2) during which it must employ glycolytic-based fermentation in order to resist the effects of high host NO· production. Toxin expression allows for S. aureus escape from the intracellular environment (3). Immune activation results in further influx of cells, which surround S. aureus forming an abscess (4) that further limits both oxygen and glucose availability. S. aureus able to escape this enclosure disseminate to the blood stream (5), an environment rich in glucose and heme-bound oxygen. From the bloodstream S. aureus invades other sterile tissues including the kidney and liver (6). In these organs, a variety of factors necessitate S. aureus glycolysis including tissue hypoxia, host NO· production (7), iron limitation, and nutrient availability.

Interestingly, people with diabetes (a medical condition characterized by defects in total body glucose homeostasis) are highly susceptible to S. aureus infection (36–38). This phenomenon is generally attributed to immune dysfunction among this patient population, as diabetics exhibit increased susceptibility to many pathogens and have documented problems with phagocyte chemotaxis and respiratory burst, despite largely being characterized as hyperinflammatory (39). However, only uncontrolled diabetics (i.e., those exhibiting hyperglycemia) exhibit increased infection rates, suggesting that increased glucose availability may be a key determinant for S. aureus virulence following invasion of sterile tissue (40–42). Furthermore, if these patients are truly hyperinflammatory, then increased glucose availability will promote greater resistance of S. aureus to increased host NO· production. This is clear from our observation that S. aureus survival during macrophage phagocytosis was improved, in a NO·-dependent manner, when serum glucose levels were increased from normal concentrations (1 g/liter) to the concentrations found in diabetics (2.5 to 4 g/liter). This is especially intriguing given that NO· resistance is a unique feature of S. aureus pathogenesis, and infection rates among diabetics are disproportionately increased for S. aureus in comparison to other pathogens (38).

These data suggest that the development of bacterium-specific glycolysis inhibitors could be a promising strategy for treating invasive S. aureus infections. This is especially important given the increasing drug-resistant nature of many S. aureus infections (e.g., methicillin-resistant S. aureus [MRSA], vancomycin-resistant S. aureus [VRSA], and vancomycin-intermediate S. aureus [VISA]). However, glycolysis is a highly conserved metabolic pathway among prokaryotes and eukaryotes, making it especially difficult to selectively target S. aureus without inducing detrimental effects in the host. Nonetheless, Zoraghi et al. have identified a novel class of antimicrobials that selectively inhibit S. aureus Pyk without inhibiting human Pyk or inducing host cell toxicity (43, 44). They showed that these novel antimicrobial agents were cytotoxic to MRSA, and together we demonstrated that the cytotoxic activity of these compounds was dependent on the presence of Pyk (45). The data presented in this study suggest that this drug could be a powerful antistaphylococcal therapeutic, effective for treating both SSTI and sepsis. Furthermore, several clinically relevant small-colony variants (SCV) of S. aureus that are associated with chronic infection presentations are deficient in cellular respiration and thus might be particularly susceptible to a Pyk inhibitor (46).

It is often speculated that the requirement for a given carbon source/central metabolic pathway is simply a reflection of the substrates available to pathogens during growth in their respective infectious niches. However, we have implicated glycolysis as a requirement for S. aureus resistance to innate immunity. Given that NO· resistance is a unique feature of S. aureus virulence, further investigation into the glycolytic capabilities of the different Staphylococcus species is warranted and may in fact reveal a great deal about the disparities that exist in their ability to cause invasive infection.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

Staphylococci were cultivated in brain heart infusion (BHI) medium or in chemically defined medium (CDM) in which primary carbon sources could be modified (47). Wild-type (WT) and mutant strains, plasmids, and primers utilized in this study are listed in Table S1 in the supplemental material. Mutants were generated via allelic replacement using the Escherichia coli/S. aureus shuttle vectors pBT2ts, pBTK, pBTE, and pBTS as previously described (29). Complementation of mutants was carried out by cloning a desired gene under its original promoter into the BamHI site of pOS1 or cloning the gene into the NdeI site of pOS1-plgt wherein transcription is driven by the constitutive lgt promoter. Aerobic S. aureus growth curves were performed at 37°C in 96-well plates (200 µl/well) using a Tecan Infinite M200 Pro microplate reader with 1 mm orbital shaking. Growth was monitored via absorbance at 650 nm (A650) at 15-min intervals. NO· donors were suspended in 0.01 N NaOH and added to cultures at an optical density at 650 nm (OD650) of 0.15 (mid-log phase). The NO· donors utilized in this study are 2,2′-(hydroxynitrosohydrazono)bis-ethanimine (DETA)/NO (2 and 4 mM) or a mixture of NOC-12 and DEANO (10:1 molar ratio). Anaerobic cultures were grown in tubes (16 by 150 mm) containing 5 ml of medium with stirring at 37°C in a Coy anaerobic chamber.

Growth rate analysis.

Growth rate was calculated at 15-min intervals using the following equation: Δln A650/Δtime (in hours). Aerobic growth rate is the average growth rate during a 2-h window of peak logarithmic growth. NO· growth rate is the average growth rate during a 4-h window from 1 to 5 h after the addition of NO·.

Enzymatic determination of excreted metabolites.

Excreted metabolites were harvested from the supernatants of heat-inactivated (70°C for 5 min) culture aliquots. Samples were stored at −20°C and thawed on ice prior to analysis. l-Lactic acid, d-lactic acid, ethanol, formic acid, and acetic acid concentrations were measured enzymatically using commercially available reagents (Roche Yellow Line kits; R-Biopharm AG, Darmstadt, Germany). Pyruvate levels were measured enzymatically (A340) using the following reaction solution: 150 mM Tris-HCl (pH 8.5), NADH (333 µg/ml), and l-lactate dehydrogenase (approximately 14.5 U).

Quantification of intracellular acetyl-CoA levels.

S. aureus COL cells were grown in 55 ml CDM plus pyruvate (1.1%) in a 125-ml flask at 37°C with shaking at 250 rpm (aerobic and NO·-treated samples). At an OD660 of 0.5, DETA/NO (final concentration of 4 mM) was added to one flask, and an equal volume of 10 mM sodium hydroxide was added to the other flask. At 3 min (NaOH flask) and 15 min (DETA/NO flask) after the addition of DETA/NO or NaOH, the S. aureus cells were collected via vacuum filtration, washed with 100 ml of 4°C sterile saline, transferred into 600 µl of 4°C sterile phosphate-buffered saline (PBS), and then snap frozen using a dry ice/ethanol bath (total time of ≤5 min). Additionally, S. aureus COL cells were grown in 55-ml CDM plus pyruvate (1.1%) in a 125-ml flask in a Coy anaerobic chamber at 37°C with stirring. At an OD650 of 0.7, the flask was removed from the anaerobic chamber, and the cells were harvested according to the procedure outlined above for the aerobic and NO·-treated samples. The frozen cells were thawed on ice and then lysed via bead beating. The supernatants were then treated with perchloric acid (final concentration of 0.036 N) and neutralized using potassium bicarbonate (total protein concentrations were measured prior to perchloric acid treatment). We then measured acetyl-CoA concentrations in the samples using the PicoProbe acetyl-CoA assay kit (BioVision Inc.). Acetyl-CoA concentrations were normalized to total protein.

Acetogenic “flux” measurements.

The flux of pyruvate to acetate during growth of S. aureus was estimated by dividing the amount of extracellular pyruvate consumed by the amount of extracellular acetate produced over a set period of time and measured according to the enzymatic methods described above. For aerobic and anaerobic samples, this period was from the beginning to the end of exponential phase (as determined by a linear growth curve graphed on a log scale). For NO·-treated S. aureus, this period was from 1 to 5 h after the addition of NO· (the fermentative window, as determined by sustained lactate production).

ATP measurement.

S. aureus cells were grown aerobically in the Tecan M200 Pro microplate reader (as described above) in CDM with either glucose (25 mM) or pyruvate (50 mM) as the primary carbon source. At an OD650 of 0.15, NO· mix was added to a subset of wells. At 0, 1, 2, 3, 4, and 5 h after the addition of NO·, the intracellular ATP concentrations were determined using the BacTiter-Glo microbial cell viability assay (Promega). ATP levels were then normalized to OD650 values for each time point and then averaged over the following time periods: 1 to 3 h after NO· addition for untreated wells (aerobic) and 1 to 5 h after NO· addition in NO·-treated wells.

Phagocytosis assays.

RAW 264.7 macrophages were suspended in RPMI 1640 (catalog no. 11875-093; Gibco) supplemented with fetal bovine serum (FBS) (10%) and HEPES (25 mM) at a concentration of 1 × 106 cells/ml, seeded into the wells of a 48-well plate at 0.5 ml/well, and incubated for 18 h at 37°C. The cells were then activated via incubation in RPMI 1640 plus lipopolysaccharide (LPS) (100 ng/ml) and gamma interferon (IFN-γ) (20 ng/ml) for 6 h at 37°C and then spin inoculated in RPMI 1640 alone with S. aureus (opsonized in Hanks balanced salt solution [HBSS] plus 10% mouse serum with active complement at 37°C for 30 min) at a multiplicity of infection (MOI) of 10:1 (S. aureus RAW 264.7). Following incubation of the RAW 264.7 cells with S. aureus for 30 min at 37°C, the cells were washed twice with PBS and then incubated in RPMI 1640 plus gentamicin (100 µg/ml) for 1 h at 37°C. The infected RAW 264.7 cells were then washed twice with PBS, after which certain wells were treated with 0.01% Triton X-100 to induce RAW 264.7 cell lysis for bacterial enumeration (time zero), while the remaining infected RAW 264.7 cells were incubated in RPMI 1640 plus gentamicin (12 µg/ml) at 37°C for an additional 12 or 20 h. Experiments involving glucose titration utilized RPMI 1640 without glucose (catalog no. 11879-020; Gibco) that was supplemented with FBS (10%), HEPES (25 mM), and tissue culture-grade glucose (1, 2, 3, and 4 g/liter) that was introduced at the time of bacterial inoculation. To inhibit iNOS activity, l-NIL (100 nM) treatment was introduced simultaneously with activation of the RAW 264.7 cells with LPS and IFN-γ and continued for all the following incubations. Nitrite levels of untreated and l-NIL-treated culture supernatants (heat treated at 70°C for 5 min) were measured using the Greiss reaction.

Murine infection models.

All mutants utilized for animal infections were first transduced and then confirmed to still exhibit their respective in vitro growth phenotypes. For virulence assessment via sepsis, 6- to 8-week-old female WT and iNOS−/− C57BL/6 mice from Taconic Farms (Albany, NY, USA) were inoculated via tail vein with 7.5 × 106 CFU of WT S. aureus Newman or isogenic mutants in 100 µl of PBS. Weight loss was monitored daily for the period of infection, with mice exhibiting excessive weight loss being sacrificed per the IACUC-approved protocol. At days 1, 2, and 5 postinfection, the kidneys and livers were removed from the infected animals, homogenized in 500 µl sterile PBS, and dilution plated on BHI (WT S. aureus Newman and ΔpckA mutant) or tryptic soy agar (TSA) without glucose supplemented with 0.5% pyruvate (S. aureus Newman Δpyk and ΔpfkA mutants) to enumerate CFU. S. aureus strain Newman was utilized for these experiments, as weight loss exhibited by mice infected intravenously with this strain correlates reliably with bacterial burdens in the organs, allowing us to monitor disease progression. For virulence assessment in the SSTI model, 6- to 8-week-old female C57BL/6 mice from Jackson Laboratory (Bar Harbor, ME, USA) were injected with 1 × 107 CFU WT S. aureus LAC or Δpyk mutant in 20 µl of sterile PBS subcutaneously. The abscess size (length by width) was monitored daily. On day 7, the mice were euthanized, and the abscesses were removed, homogenized in 500 µl of PBS, and dilution plated on TSA without glucose supplemented with 0.5% pyruvate to enumerate CFU. S. aureus strain LAC was utilized for the SSTI model, as it belongs to a subtype of S. aureus strains (USA300) that are the primary cause of these infections in the United States.

Fluorescence immunohistochemistry.

Uninfected C57BL/6 mice and those inoculated with WT S. aureus intravenously (i.v.) (7.5 × 106 CFU) 2 days prior were injected intraperitoneally (i.p.) with Hypoxyprobe-1 (Hypoxyprobe Inc., Burlington, MA, USA) (60 mg/kg of body weight) suspended in sterile PBS. After 1 h, liver and kidney tissue was collected, fixed in 10% formalin, paraffin embedded, and sectioned (10 µm) by the Histopathology Core Facility at the University of North Carolina (UNC). Sections were hematoxylin and eosin (H&E) stained by the Histology Core, while unstained sections were prepared for immunohistochemistry as previously described (13). Briefly, unstained sections were deparaffinized using a series of xylene and ethanol washes and then incubated in hot 10 mM sodium citrate buffer (pH 6) for 20 min. Tissues were then blocked in 10% donkey serum (Jackson ImmunoResearch, West Grove, PA) and then stained with affinity-purified anti-Hypoxyprobe-1 rabbit antisera (PAb2627AP; Hypoxyprobe Inc., Burlington, MA, USA) diluted 1:200 in PBS plus 2% donkey serum. Staining was detected using biotinylated secondary antibodies followed by streptavidin-Dylight 594 conjugates (Jackson ImmunoResearch). Stained sections were mounted in ProLong antifade reagent gold with 4′,6′-diamidino-2-phenylindole (DAPI) (Invitrogen, Grand Island, NY, USA) and viewed with an Olympus BX60 fluorescence microscope. Images were captured using iVision software v.4.0.0 (BioVision Technologies, New Minas, Nova Scotia, Canada).

S. aureus glycolytic, gluconeogenic, and fermentative metabolic pathways. Metabolites are indicated by black dots, while metabolic reactions are indicated by black lines. For emphasis, the black dots representing several key metabolites discussed in the manuscript have been replaced with the names of the metabolites. From this group, the metabolites used as primary carbon sources are colored blue (glycolytic) or orange (nonglycolytic). Additionally, several reactions have also been colored to indicate their glycolytic (blue) or gluconeogenic (orange) direction. Protein names are included next to several important metabolic reactions and are colored to indicate their glycolytic (blue), gluconeogenic (orange), or miscellaneous (black) function(s). These proteins include PfkA (phosphofructokinase), GapA (glyceraldehyde-3-phosphate dehydrogenase A), GapB (glyceraldehyde-3-phosphate dehydrogenase B), Pyk (pyruvate kinase), Ldh1 (l-lactate dehydrogenase 1), Ldh2 (l-lactate dehydrogenase 2), Ddh (d-lactate dehydrogenase), PDHC (pyruvate dehydrogenase complex), Pfl (pyruvate formate lyase), AckA (acetate kinase), Pyc (pyruvate carboxylase), and PckA (phosphoenolpyruvate carboxykinase). Casamino Acids enter central metabolism either at pyruvate or through quinone-dependent oxidation by the TCA cycle. Download

Figure S1, TIF file, 2.4 MB

Anaerobic growth on pyruvate requires acetogenesis. (A) Mid-log-phase production of extracellular metabolites during anaerobic growth of S. aureus COL at 37°C in CDM with glucose (0.5%) or pyruvate (1%) as the primary carbon source (n = 3; error bars show SEM). (B) Anaerobic growth of wild-type S. aureus COL and Δpyc and ΔpckA mutants at 37°C in CDM with pyruvate (1%) as the primary carbon source (n = 2; error bars show SEM). (C) Anaerobic growth of wild-type S. aureus COL and ΔackA mutant at 37°C in CDM with glucose (0.5%) or pyruvate (1%) as the primary carbon source (n = 2; error bars show SEM). Statistical significance was calculated using a Student’s two-sided t test (**, P ≤ 0.01). Download

Figure S2, TIF file, 2.7 MB

Aerobic growth of predicted S. aureus glycolytic and gluconeogenic mutants. (A to D) Growth of predicted S. aureus COL central metabolic mutants at 37°C in CDM supplemented with glucose (25 mM) (A and C) or Casamino Acids (0.5%) (B and D) (n = 3; error bars show pooled SD). (E) Growth of complemented S. aureus COL glycolytic mutants (Δpyk and ΔpfkA) at 37°C in CDM with glucose (25 mM) (n = 1; error bars show SD). (F) Growth of complemented S. aureus COL gluconeogenic mutants (ΔpckA and ΔgapB) at 37°C in CDM with Casamino Acids (0.5%) (n = 1; error bars show SD). Download

Figure S3, TIF file, 2.4 MB

NO· production in RAW 264.7 cells is not affected by mutations in S. aureus central metabolism or by increased extracellular glucose. (A) Extracellular nitrite levels (in micromolar) for RAW 264.7 macrophages with or without l-NIL at 12 and 20 h postinfection with wild-type and central metabolic mutants of S. aureus COL (MOI of 10:1) (n = 1). (B) Extracellular nitrite levels for RAW 264.7 macrophages with or without l-NIL in RPMI 1640 with glucose at 1 and 4 g/liter, 12 hpi with wild-type S. aureus (n = 3; error bars show SEM). Download

Figure S4, TIF file, 2.4 MB

Glycolysis is essential for S. aureus virulence during skin and soft tissue infection. C57BL/6 mice were infected with 1 × 107 CFU of wild-type S. aureus LAC or Δpyk mutant via subcutaneous injection. (A and B) Infection was monitored by measuring lesion size (A) and day 7 abscess burden (B) (5 ≤ n ≤ 7; error bars show SEM). Significance was calculated using a Mann-Whitney test (**, P ≤ 0.01). Download

Figure S5, TIF file, 2.4 MB

Loss of iNOS does not reverse attenuation of the S. aureus glycolytic mutants during sepsis. Wild-type and iNOS−/− C57BL/6 mice were infected via tail vein injection with 7.5 × 106 CFU of WT S. aureus Newman or Δpyk or ΔpfkA mutant. (A and B) Kidneys (A) and livers (B) were harvested 24 h postinfection, and bacteria were counted (n = 4). Each symbol represents the value for an individual mouse. The lines indicate mean organ burden. The limit of detection is 100 CFU/organ. Download

Figure S6, TIF file, 2.8 MB

Uninfected murine kidneys and livers display areas of hypoxia. The kidneys and liver from a uninfected C57BL/6 mouse were harvested, fixed, paraffin embedded, and sliced. Following deparaffinization and antigen retrieval, tissue slices were blocked in 10% donkey serum and then stained with affinity-purified anti-Hypoxyprobe rabbit antisera, biotinylated goat anti-rabbit secondary antibody, and streptavidin-Dylight 594 conjugate (pink). Stained tissue slices were then mounted using Prolong Antifade gold containing DAPI (blue) and imaged at a magnification of ×20 using an Olympus BX60 fluorescence microscope. Download

Figure S7, TIF file, 2.5 MB

Excessive fermentation limits S. aureus growth on hexitols during NO· stress. Production of l-lactate by wild-type S. aureus COL grown at 37°C in CDM with mannose (25 mM) or mannitol (25 mM) following NO· (NO· mix) addition (n = 3; error bars show SEM). Change in extracellular lactate levels was assessed from 0 to 6 h after the addition of NO·, and concentrations were normalized to the absorbance at 650 nm. Statistical significance was calculated using a Student’s two-sided t test (*, P ≤ 0.05). Download

Figure S8, TIF file, 2.6 MB

Strains, plasmids, and primers.

Table S1, DOCX file, 0.2 MB