|Home | About | Journals | Submit | Contact Us | Français|
Host nitric oxide (NO·) production is important for controlling intracellular bacterial pathogens including Salmonella enterica serovar Typhimurium but the underlying mechanisms are incompletely understood. S. Typhmurium 14028s is prototrophic for all amino acids but cannot synthesize Methionine (M) or Lysine (K) during nitrosative stress. Here we show that NO·-induced MK auxotrophy results from reduced succinyl-CoA availability as a consequence of NO· targeting of lipoamide-dependent lipoamide dehydrogenase (LpdA) activity. LpdA is an essential component of the pyruvate and α-ketoglutarate dehydrogenase complexes. Additional effects of NO· on gene regulation prevent compensatory pathways of succinyl-CoA production. Microarray analysis indicates that over 50% of the transcriptional response of S. Typhimurium to nitrosative stress is attributable to LpdA inhibition. Bacterial methionine transport is essential for virulence in NO·-producing mice, demonstrating that NO·-induced MK auxotrophy occurs in vivo. These observations underscore the importance of metabolic targets for antimicrobial actions of NO·.
Nitric oxide (NO·) production is a critical component of the mammalian innate immune response (Fang, 2004). Host organisms incapable of producing inflammatory NO· via the inducible NO· synthase (iNOS) exhibit increased susceptibility to viral, fungal, parasitic and bacterial microorganisms (DeGroote et al., 1999). Host NO·-production is known to be particularly important for controlling the proliferation of intracellular bacterial pathogens such as Salmonella enterica serovar Typhimurium (Mastroeni et al., 2000). Survival in the intraphagosomal compartment of activated macrophages is a critical facet of S. Typhimurium infection and requires a type III secretion system that excludes both the phagocyte oxidase and iNOS from the Salmonella-containing vacuolar membrane (Chakravortty et al., 2002; Fang and Vazquez-Torres, 2002; Vazquez-Torres et al., 2000b). While this altered trafficking lowers Salmonella exposure to reactive oxygen species, the overall effect on the level of the membrane permeable NO· radical is uncertain. Indeed, NO· production is critical for the control of an S. Typhimurium infection, detoxification of host NO· via the Salmonella flavohemoprotein (Hmp) is required for virulence, and direct NO·-mediated bacterial DNA damage can be detected during murine infection (Bang et al., 2006; Mastroeni et al., 2000; Richardson et al., 2009; Vazquez-Torres et al., 2000a). Thus, despite altered trafficking of the iNOS enzyme in Salmonella-infected macrophages, Salmonella must still contend with significant amounts of NO· during infection.
Administering authentic NO· to cultured bacteria results in periods of growth arrest that are proportional to the duration of NO·-exposure. However, a mechanistic understanding of the bacteriostatic effects of host NO· has yet to be fully elucidated. The ability of NO· to inhibit bacterial growth likely revolves around the chemistry of this small, uncharged lipophilic radical. For instance, NO· has a strong predilection for interacting with heme iron and can also react with non-heme protein iron to form dinitrosyl-iron-complexes (DNIC) (Radi, 1996). In aerobic environments, NO· is oxidized to reactive nitrogen species (RNS) including NO2· and N2O3 (Lewis and Deen, 1994; Wink et al., 1993). N2O3 is capable of N- and S-nitrosylation by modifying nucleophilic amines and sulfhydryls with an NO+ equivalent, resulting in detectible DNA base deamination and protein thiol-nitrosation (Brandes et al., 2007; Lewis, 1995; Wink et al., 1991; Wink et al., 1994). NO2· reacts readily with protein thiols and aromatic hydrocarbons to generate thiyl radical species and nitrated tyrosine residues, respectively (Schrammel et al., 2003; van der Vliet et al., 1995). Moreover, the phagosome of an activated macrophage is replete with both O2−· and NO· that can combine to generate highly reactive peroxynitrite, ONOO− (Ischiropoulos et al., 1992). This ion readily decomposes to NO2· and HO· radical, an exceedingly potent oxidant that reacts at diffusion limits with a variety of biomolecules (Beckman et al., 1990). Thus, a living cell is comprised of a myriad of potential targets for NO·- and other RNS, some of which may be particularly consequential for cell growth and contribute directly to NO·-mediated bacteriostasis.
Given the reactivity of NO·, metabolic enzymes with essential redox centers comprised of protein thiols, Fe-S clusters, or hemes likely represent the major targets responsible for NO·-mediated growth inhibition. For instance, the strong reactivity of NO· towards heme iron is known to result in respiratory arrest in nitrosatively-stressed cells (Brown et al., 1997; Richardson et al., 2008; Stevanin et al., 2000). This limits a cell’s ability to maintain redox balance and generate energy, both of which may contribute directly to NO·-mediated cytostasis. Furthermore, the reactivity of 4Fe-4S clusters in the active sites of aconitase and dihydroxyacid dehydratase (IlvD) are known to be sensitive to NO·, affecting the TCA cycle and branched-chain amino acid biosynthesis, respectively (Duan et al., 2009; Gardner et al., 1997; Hyduke et al., 2007). NO· is known to modify the cysteine thiol active site of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), limiting glycolytic flux during nitrosative stress (Messmer and Brune, 1996). Furthermore, assessing global protein S-nitrosylation in bacteria and mammals reveals that enzymes involved in glycolysis, glutamate biosynthesis, fatty acid metabolism and pyruvate catabolism consistently undergo thiol modification during nitrosative stress (Brandes et al., 2007; Lefievre et al., 2007; Rhee et al., 2005). Glycyl-radical dependent enzymes such as pyruvate formate-lyase can also be inhibited by NO·, thereby limiting growth under anaerobic conditions (Richardson et al., 2008). The metabolic perturbations associated with NO·-exposure could represent the proximal causes of transient growth arrest observed during periods of nitrosative stress.
Transcriptional analyses of cells undergoing nitrosative stress further support the hypothesis that growth arrest results from metabolic constraints imposed by exogenous NO·. For example the transcriptional responses to nitrosative stress in both Mycobacterium tuberculosis and Staphylococcus aureus resemble the expression profiles observed upon shifting these bacteria to hypoxic environments (Ohno et al., 2003; Richardson et al., 2006). This underscores the tendency of NO· to inhibit aerobic respiration, necessitating the expression of anaerobic metabolic genes. In many bacterial and fungal species, genes involved in methionine, glutamine and branched-chain amino acid biosynthesis are strongly affected by NO· exposure (Flatley et al., 2005; Hromatka et al., 2005; Hyduke et al., 2007; Moore et al., 2004; Nittler et al., 2005; Ohno et al., 2003; Pullan et al., 2007; Richardson et al., 2006). Additionally, the bacterial stringent response is often induced in cells undergoing nitrosative stress, indicative of amino acid and/or other nutrient starvation (Hyduke et al., 2007). The SigB general stress response of B. subtilis is induced by NO·-exposure in an RsbP-dependent manner, suggesting that the SigB stimulon specifically responds to NO·-induced energy stress (Moore et al., 2004). Collectively, these findings demonstrate that diverse bacterial species respond in kind to nitrosative stress by altering the expression of discrete metabolic pathways. The identification of NO·-affected pathways yields insights into the mechanisms underlying NO·-mediated bacteriostasis.
The present work extends from an observation that Salmonella enterica serovar Typhimurium cultured in chemically defined medium replete with amino acids fares better during NO·-stress than cells in M9 minimal medium alone. This phenomenon has been exploited to probe the nature of metabolic inhibition imposed by exogenous NO·. In contrast to phenotypes observed in E. coli, S. Typhimurium does not exhibit a requirement for branched-chain amino acids (ILV) during nitrosative stress, but rather requires supplementation with both methionine (M) and lysine (K). The molecular basis for this NO·-induced MK auxotrophy is explored and has been determined to involve multiple interactions between NO· and specific components of the TCA cycle.
A lag phase of 7 to 8 h consistently accompanies the shift of LB grown S. Typhimurium strain 14028s into M9 Glucose (M9 Glc) medium (lag phase defined as the time to reach 50% maximal absorbance at 600 nm when diluted into fresh M9 medium to a density of OD600 = 0.05). Addition of the NO·-donor SperNO at 2 mM results in a lag phase four times longer than that observed in M9 Glc alone (Figure 1A). Given the kinetics of NO· mass transfer in this system, 2 mM SperNO results in a peak NO· concentration of ~100 μM after 40 min. This NO· concentration wanes over time, and minimal growth is detectable at 18 h, whereas normal growth resumes only after 24 h (Figure 1A). However, supplementing M9 Glucose with an amino acid mixture (M9 Glc AA) reduces the NO·-associated lag phase to less than half that of unsupplemented medium (Figure 1A). The amino acid mixture is comprised of all 20 essential L-amino acids, with the substitution of cystine for L-cysteine to minimize the reactivity of the amino acid supplement with NO·. Indeed, the stability of NO· in aerated M9 Glc is unaffected by amino acid supplementation (Figure S1). Thus, the abbreviated lag phase observed when NO·-stressed S. Typhimurium is grown in M9 Glc AA is not due to amino acid-mediated NO· scavenging. Rather, we tested whether NO· induces a transient auxotrophy for one or more amino acids in Salmonella. Accordingly, individual “dropout” AA mixtures were all shown to equally suppress the NO·-induced lag phase in S. Typhimurium except those lacking Methionine (M), Lysine (K) or Threonine (T) (Figure 1B). Moreover, mixtures lacking both M and K (−MK) did not offer any measurable protection to NO·-exposed cells despite the presence of 18 other amino acids (Figure S2). These data show that NO·-exposure in S. Typhimurium strain 14028s induces a transient MK auxotrophy and, to a lesser extent, limits the synthesis of threonine.
M, K, and T are all synthesized from oxaloacetate via aspartate (Figure 2A). It is conceivable that individual enzymes in all three amino acid biosynthetic pathways could be targeted by NO·. Alternatively, both M and K biosynthesis require the TCA cycle intermediate succinyl-CoA, and thus perturbation of the TCA cycle could account for the observed NO·-induced MK auxotrophy. Indeed, supplementation with succinyl-CoA relieved both the M and K auxotrophies during nitrosative stress, but had no effect on T auxotrophy (Figure 2B). Therefore, the inability of nitrosatively-stressed S. Typhimurium to synthesize M and K results from TCA-cycle inhibition and the limited availability of succinyl-CoA (Figure 3A). Providing the TCA-cycle intermediate succinate also relieved the NO·-dependent MK auxotrophy, while addition of malate, fumarate, α-ketoglutarate, isocitrate or citrate did not allow nitrosatively-stressed cells to synthesize M or K (Figure 3B and data not shown). All of the above carboxylate metabolites can be actively transported by S. Typhimurium 14028s, as they were able to replace glucose as a carbon source in M9 medium (data not shown). The production of succinyl-CoA from succinate during nitrosative stress in S. Typhimurium implies that succinyl-CoA synthase (SucCD, SUC in Figure 3A) must be resistant to the effects of NO·. In contrast, the inability to synthesize succinyl-CoA, and consequently MK, from α-ketoglutarate indicates that the α-ketoglutarate dehydrogenase (SucABLpdA, α-KDH) complex is inhibited by nitrosative stress. Similarly, the inability of fumarate to rescue the NO·-induced MK auxotrophy implies that the expression level of the fumarate reductase (FrdABCD, FRD) complex is inadequate in the presence of NO· for succinate synthesis. Finally, succinate can be produced via the glyoxylate bypass route involving isocitrate lyase (ICL) and malate synthase (MS) (Figure 3A). However, supplementation with isocitrate was unable to restore MK prototrophy (data not shown). Furthermore, inactivation of aceA encoding the ICL enzyme of S. Typhimurium did not exacerbate NO·-sensitivity (not shown). Therefore, under laboratory conditions, the glyoxylate pathway does not supply sufficient levels of succinate in NO·-stressed Salmonella to allow MK biosynthesis.
The inability of NO·-exposed Salmonella to synthesize succinyl-CoA from α-ketoglutarate suggests that the activity of the α-KDH complex is limited by nitrosative stress. The enzymatic activities of several TCA cycle enzymes were assessed in cell-free extracts of S. Typhimurium 14028s grown in M9 Glc medium (Figure 4). Increasing concentrations of NO· had little effect on the activities of malate dehydrogenase (MDH) or isocitrate dehydrogenase (IDH) (Figure 4A). In contrast, the activity of α-KDH could be inhibited in a dose dependent fashion by NO· (Figure 4A). α-KDH is an enzyme complex made up of a thiamine-dependent E1 component, α-ketoglutarate decarboxylase (SucA), as well as an E2 component, dihydrolipoyltranssuccinylase (SucB). The dithiol-containing cofactor, lipoic acid, is amide-linked to the E2 component and redox cycles between dihydrolipoamide and oxidized lipoamide during catalysis (Figure S3). E2-bound dihydrolipoamide is reoxidized by interacting with two cysteine residues in the E3 component of the complex, LpdA. Thus, the oxidative decarboxylation of α-ketoglutarate requires the coordinate action of four redox-active thiols, motifs known to be sensitive to modification by NO·. α-KDH belongs to a family of α-keto acid dehydrogenases including pyruvate dehydrogenase (PDH) and the glycine cleavage system (GCV). Indeed, as with α-KDH, exogenous NO· inhibited PDH activity in cell free extracts (Figure 4A). To specifically assess whether NO· interferes with LpdA/lipoamide interactions, the LpdA-dependent oxidation of free dihydrolipoic acid was monitored in cell-free extracts (Figure 4B). The exquisite NO·-sensitivity of the LpdA reaction suggests that the inhibition of α-KDH and PDH during nitrosative stress is a result of NO·-mediated interference of the LpdA/lipoamide reaction. These data show that lipoamide-dependent enzymes represent an important class of NO·-sensitive targets in bacteria. In fact, over 60% of the global transcriptional response to NO·-stress in S. Typhimurium may be attributed to NO·-inactivation of LpdA, as a ΔlpdA mutant exhibits a transcriptional profile similar to that of NO·-treated cells (Figure 5). Moreover, mutants lacking LpdA (Figure S4) or the LpdA-dependent α-ketoglutarate dehydrogenase (Tchawa Yimga et al., 2006; Bowden et al, 2010) are completely attenuated for virulence, underscoring the critical importance of functional LpdA during infection (Figure S4).
Under hypoxic conditions, the reactions of the TCA cycle between oxaloacetate and succinyl-CoA run in reverse in what is known as the reductive branch of the TCA cycle (Figure 3A). The reactions between citrate and α-ketoglutarate run normally, but α-KDH is not expressed. This provides the fermenting cell with a means to produce oxaloacetate, α-ketoglutarate and succinyl-CoA for amino acid biosynthesis, while still maintaining redox balance. In the reductive TCA branch, the conversion of fumarate to succinate is a critical step, and anaerobically-cultured cells rely on fumarate reductase (FRD) for the reductive TCA branch to proceed (Figure 3A). In theory, the reductive branch of the TCA cycle could provide NO·-exposed cells with a source of succinyl-CoA. However, this is not the case given the NO·-induced MK auxotrophy in S. Typhimurium. To explain this apparent inconsistency, the expression of sdh and frd was monitored in the presence and absence of NO·. Under routine laboratory conditions, S. Typhimurium expresses both sdh and frd transcripts at similar levels (Figure 6A). However, NO·-exposure drastically reduces frd expression so that cells are only able to oxidize succinate to fumarate (Figure 3A). Hypoxic expression of the frdABCD operon requires functional Fnr, a transcriptional activator with an oxygen labile iron-sulfur cluster. Low oxygen conditions stabilize the Fe-S cluster, allowing the Fnr-dependent expression of anaerobic metabolic genes including frdABCD. Under our culture conditions and cell densities, we observed moderate hypoxia, resulting in Fnr-dependent frd expression. Indeed, Δfnr S. Typhimurium expressed 20-fold less frd transcript than WT, and the supressive effects of NO·-exposure on frd expression were absent in this mutant (Figure 6A). Given the propensity of NO· to modify iron-sulfur cluster-containing proteins including Fnr (Cruz-Ramos et al., 2002), these data suggest that nitrosative stress limits frd expression through inactivation of Fnr. An alternative hypothesis is that NO·-dependent inhibition of LpdA function also limits Fnr activity, as many of the genes repressed in a ΔlpdA mutant compared to WT are known to be Fnr-activated. In either case, low Fnr-activity resulting in deficient Frd-levels drives carbon flux away from succinyl-CoA, thereby exacerbating the NO·-induced MK auxotrophy. Indeed, by ectopically expressing FRD in a Δsdh mutant, the NO·-dependent MK auxotrophy can be completely relieved, showing that the reductive branch of the TCA cycle could serve as a source of succinyl-CoA during nitrosative stress if it were expressed (Figure 6B). However, the effect of NO· on Frd expression represents a “regulatory block,” driving carbon-flux away from succinyl-CoA and contributing to the observed MK auxotrophy.
Previous reports have shown that NO·-exposure induces transient branched-chain amino acid auxotrophies (ILV) in E. coli K-12 strain MG1655 (Hyduke et al., 2007). Our observations in S. Typhimurium strain 14028s did not identify branched-chain amino acids, either individually or in combination, as being required for suppression of the NO·-mediated lag phase (Figure 1B and Figure S2). To define the prevalence of MK and ILV auxotrophies among enteric bacteria (E. coli and S. enterica), we surveyed three E. coli isolates and three S. Typhimurium isolates for ILV or MK requirements during NO·-stress. ILV supplementation was critical in all three E. coli isolates, whereas none of the S. Typhimurium isolates required branched-chain amino acids (Supplemental Table 1). Additionally, pathogenic E. coli isolates CFT073 and EC1 required M and K as well as ILV during NO·-exposure. Curiously, the laboratory-passaged strains S. Typhimurium LT2 and E. coli MG1655 did not display MK auxotrophies, and consistent with the lack of an ILV requirement in NO·-stressed Salmonella, LT2 exhibited no requirement for any single amino acid during NO·-exposure. A broader survey of 25 natural Salmonella enterica isolates showed that twenty-three require M and K during NO·-stress, with LT2 and a group IIIa isolate representing the only two MK prototrophs (Supplemental Table 1). Collectively, these data suggest that the inability to synthesize succinyl-CoA during nitrosative stress is an attribute of natural enteric isolates, but that laboratory passage may select for unidentified mutations that alleviate NO·-induced MK auxotrophy. Furthermore, the contrast between E. coli and S. Typhimurium regarding ILV requirements during NO·-exposure highlights a fundamental physiological difference between the two species that has yet to be characterized.
To determine whether host NO· limits de novo methionine biosynthesis in a model infection, the virulence of S. Typhimurium lacking the high-affinity D,L-methionine transporter was assessed in female C3H/HeN mice. The Salmonella metD locus encodes an ABC type transporter (metNIQ) that acquires free extracellular D- or L-methionine (Cottam and Ayling, 1989). Enteric bacteria are also believed to encode an as yet unidentified low-affinity transporter, MetP that is specific for L-methionine (Cottam and Ayling, 1989; Merlin et al., 2002). However, a metD mutant exhibits NO·-sensitivity in medium replete with amino acids including L-methionine, suggesting that this ABC transporter is necessary for efficient methionine uptake during NO·-stress (not shown). Indeed, ΔmetD S. Typhimurium is significantly attenuated for virulence during systemic murine infection (Figure 7). This indicates that de novo methionine synthesis is insufficient for growth in the host, and Salmonella relies on exogenous methionine pools during infection (WT vs. ΔmetD in untreated mice, p ≤ 0.001). Inhibiting the murine production of inflammatory NO· via administration of L-NIL (L-N6-(1-iminoethyl)-lysine) largely alleviates the effects of a ΔmetD mutation on virulence, implying that host NO· inhibits bacterial methionine synthesis in vivo, necessitating transport (ΔmetD in treated vs. untreated mice, p = 0.0078). However, the inability of L-NIL treatment to restore full virulence to a ΔmetD mutant also suggests the possibility of additional causes of methionine biosynthesis inhibition during infection or that L-NIL treatment failed to fully suppress iNOS activity (WT vs. ΔmetD in L-NIL treated mice, p = 0.023).
The connection between MK auxotrophy and lipoamide-dependent enzyme activity was originally observed more than 50 years ago (Davis et al., 1959). Mutant 309-1 of E. coli required either succinate supplementation or the addition of M, K (and interestingly T) for growth on glucose minimal medium. While the requirement for T has never been explained, it was determined that the inability to produce succinyl-CoA in this mutant was responsible for the MK auxotrophy (Kaplan and Flavin, 1965). The current study investigates the effects of exogenous NO· on enteric bacteria and consequently revisits the amino acid auxotrophies associated with reduced LpdA activity. NO·-exposed S. Typhimurium transiently exhibits the same MKT auxotrophy as lpdA mutant E. coli (Davis et al., 1959; Guest and Creaghan, 1973) and α-ketoglutarate dehydrogenase mutants of S. Typhimurium (Carrillo-Castaneda and Ortega, 1970; Langley and Guest, 1974). Furthermore, the NO·-induced MK auxotrophy can be eliminated by supplementation with succinate or succinyl-CoA. Assaying enzymatic activities in cell-free extracts clearly demonstrates the NO·-sensitivity of the LpdA/lipoamide reaction. Thus, enzymes requiring the lipoamide cofactor comprise a class of NO·-targets that limit bacterial growth during nitrosative stress. These results explain observations from other independent studies. For instance, NO·-exposed S. aureus continues to grow without producing measurable amounts of acetate. This correlates with the ability of exogenous NO· to inhibit staphylococcal pyruvate dehydrogenase, an observation explained by the results described herein (Richardson et al., 2008). Furthermore, the NO·-sensitivity of the LpdA/lipoamide reaction is consistent with the identification of AceF, the lipoamide-linked E2 component of PDH, as an S-nitrosylated protein in NO·-challenged E. coli (Brandes et al., 2007). Collectively, these observations suggest that NO· can target the LpdA/lipoamide reaction in various organisms during nitrosative stress.
In E. coli, inactivation of lpdA results in slowly growing cells that require succinate or MK supplementation (Davis et al., 1959; Guest and Creaghan, 1973). Strains carrying mutations in lpdA route carbon flux around the lipoamide-dependent enzymes by making use of PoxB for acetate production and the glyoxylate shunt to complete the TCA cycle (Li et al., 2006). Consistently, poxB mutants show heightened susceptibility to NO·, implicating this pathway as critical for routing glycolytic intermediates into the TCA cycle (not shown). In contrast, an aceA mutation that eliminates the glyoxylate bypass has no effect on S. Typhimurium NO·-sensitivity. Furthermore, inactivation of aceA in S. Typhimurium LT2 did not eliminate the MK prototrophy of this strain, indicating that altered glyoxylate flux cannot account for the ability of laboratory strains to synthesize MK in the presence of NO· (not shown). Thus, while the glyoxylate shunt is essential for ΔlpdA mutants, it seems to be underutilized during nitrosative stress. This disparity between the phenotypes of genetic (ΔlpdA mutant) and biochemical (NO· treatment) LpdA deficiency may reflect the NO·-sensitivity of aconitase (Figure 4A) and isocitrate lyase. The 4Fe-4S cluster of aconitase is known to be sensitive to both oxidative and nitrosative stress (Figure 4A) as a result of its solvent-exposed Fe atom (Gardner et al., 1997). Isocitrate lyase may also be susceptible to NO· inhibition, as a result of an essential Cys residue at its active site (Rehman and McFadden, 1997), which could account for the inability of isocitrate supplementation to restore MK prototrophy during nitrosative stress.
While the data presented in Figure 4B do not unequivocally define the NO·-sensitive target in LpdA (i.e., NO· may react with either thiol in lipoamide itself, with either catalytic cysteine in LpdA or any combination thereof) the exquisite NO·-sensitivity of the LpdA/lipoamide reaction is readily apparent. The reactivity of NO· with LpdA/lipoamide raises an interesting paradox regarding the LpdA enzyme of M. tuberculosis (MTB), which lacks an αKDH equivalent (Tian et al., 2005). While the M. tuberculosis LpdA homolog (LpdC in MTB) does participate in PDH activity, it is also required for alkyl peroxidase activity in this organism (Bryk et al., 2002). LpdC acts coordinately with AhpC, AhpD and DlaT (formerly annotated as SucB) to mediate M. tuberculosis resistance to peroxides and peroxynitrite. However, a proteomic analysis of NO·-exposed M. tuberculosis revealed that the LpdC protein is directly modified by NO· (Rhee et al., 2005). Whether this nitrosothiol-LpdC intermediate is capable of catalyzing pyruvate decarboxylation and/or alkylperoxide detoxification has yet to be determined. However, the present work demonstrates that in enteric bacteria, the interaction of NO· with LpdA/lipoamide results in decreased enzymatic activity.
Mutants in lpdA are only auxotrophic for MK during aerobic growth when the reductive branch of the TCA cycle is inactive. Given the reactivity of Fnr with NO· (Cruz-Ramos et al., 2002), the expression of frdABCD is further limited by nitrosative stress, thereby contributing to the NO·-induced MK auxotrophy (Figure 6). The reactivity of Fnr with NO· highlights an interesting challenge for species that control the expression of key fermentative enzymes by directly sensing the presence of oxygen. For instance, S. aureus and M. tuberculosis lack discernible Fnr homologs, and both organisms mount transcriptional responses to NO· that are very similar to those observed in anaerobically-cultured cells (Ohno et al., 2003; Richardson et al., 2006). This is not the case in E. coli or B. subtilis, both of which harbor Fe-S cluster-containing Fnr homologs (Hochgrafe et al., 2008; Hyduke et al., 2007; Moore et al., 2004). Furthermore, S. aureus ferments available carbon sources in order to replicate in the presence of NO·, contrasting with the lack of metabolic activity exhibited by NO·-stressed B. subtilis (Hochgrafe et al., 2008; Richardson et al., 2008). The present study describes another consequence of dependence on Fnr for expression of fermentative enzymes. By eliminating the reductive branch of the TCA cycle via direct or indirect effects on Fnr, NO· creates a “regulatory block” that limits synthesis of M or K under nitrosative stress. The expression of SDH during nitrosative stress results in the flux of carbon away from succinyl-CoA, and the inactivation of sdhCDAB was able to partially relieve the NO·-induced MK auxotrophy (not shown). However, full MK prototrophy during NO· stress is only achieved by increasing expression of frdCDAB (Figure 6). While SDH can exhibit “fumarate reductase” activity under certain anaerobic conditions, the redox potential associated with our experimental conditions (both oxygen and NO· present) is most likely too high to support reverse-SDH activity (Maklashina et al., 1998).
Not all Fe-S cluster-containing proteins are sensitive to attack by NO·. For instance, ectopic Frd expression bypasses the NO·-imposed regulatory block and allows the synthesis of MK during NO· stress (Figure 4). This implies that, in contrast to the clusters of aconitase and Fnr, the three Fe-S clusters of FrdB (3Fe-4S, 4Fe-4S, and 2Fe-2S) are resistant to NO·. Curiously, our studies did not identify a requirement for ILV in Salmonella exposed to NO·. This differs from observations in E. coli K-12, for which medium supplemented with ILV is protective during nitrosative stress (Hyduke et al., 2007). This has been attributed to the reactivity of NO· with the 4Fe-4S cluster of the dihydroxyacid dehydratase, IlvD. In keeping with this previous work, we readily detected the NO·-dependent ILV auxotrophy in E. coli K-12 strain MG1655 as well as in E. coli clinical isolates. The mechanism responsible for the ILV prototrophy of Salmonella during NO·-stress is not clear, but may represent a fundamental difference in Fe-S cluster homeostasis between E. coli and Salmonella.
Collectively, this work demonstrates that NO· imposes substantial metabolic restrictions on bacteria. Nitrosative stress results from the interaction of NO· with metal- and thiol-containing catalytic sites in a variety of metabolic enzymes. During infection, NO·-induced auxotrophies must be compensated for by nutrient acquisition systems, as illustrated by the requirement of Salmonella for the MetD transporter during nitrosative stress in vitro (not shown) and in vivo (Figure 7). Such systems might be exploited as therapeutic targets given their importance in virulence. The reactivity of NO· combined with numerous conserved bacterial metabolic targets helps to account for the broad-spectrum nature of this key host innate immune effector.
Bacteria used in this study are derivatives of S. Typhimurium 14028s and are listed in Table 1. Additional strains of Salmonella enterica used to survey NO·-induced MK- or ILV-auxotrophy were obtained from the SARB and SARC reference collections (Boyd et al., 1993; Boyd et al., 1996). Strains were grown in minimal (M9) medium supplemented with 0.2% glucose. To assess NO·-sensitivity, M9 medium was supplemented with the NO· donor Spermine/NONOate (SperNO, 2mM) (Cal Biochem) at the time of inoculation and subsequent growth was monitored by measuring optical density at 600nm at 37°C using a Bioscreen C incubator/reader (Growthcurves USA). M9 medium was further supplemented with amino acid mixtures at the following final concentrations: Ala (0.5 mM), Arg (0.6 mM), Asn (0.3 mM), Asp (0.3 mM), Cystine (0.3 mM), Glu (5 mM), Gln (5 mM), Gly (0.1 mM), His (0.1 mM), Iso (0.3 mM), Leu (0.3 mM), Lys (0.3 mM), Met (0.3 mM), Phe (0.3 mM), Pro (2 mM), Ser (4 mM), Thr (0.3 mM), Trp (0.1 mM), Tyr (0.1 mM), and Val (0.3 mM). Drop out media were formulated by omitting one or more of the above supplements. Further supplementation with TCA cycle intermediates was achieved by addition of Na2·succinate, Na2·fumarate, Na2·malate, Na3·citrate, Na3·isocitrate, Na2·α-ketoglutarate, Na·acetate or succinyl-CoA to 0.15% final volume (w:v). NO·-consuming activity of M9 medium with/without supplementation was determined by monitoring the fate of exogenously added NO· (10 μM via addition of the NO·-donor Proline-NONOate, AG Scientific, San Diego, CA) using an ISO-NOP electrode (WPI Instruments).
For mutant construction and cloning, the following antibiotics were added to Luria-Bertani (LB) medium when appropriate: penicillin G (250 μg·ml−1) kanamycin (50 μg·ml−1), and chloramphenicol (40 μg·ml−1). Mutant S. Typhimurium strains were constructed via the 3red3 method (Datsenko and Wanner, 2000) with oligonucleotides listed in Table 1. Each mutation was confirmed by PCR analysis using gene-flanking primers (Table 1), then transduced via phage P22 into wild type 14028s. Complementation of the ΔlpdA and metD mutations was achieved by cloning the WT lpdA allele and the WT metNIQ genes (Merlin et al., 2002) into the stable low-copy replicon, pRB3-237C (Berggren et al., 1995). Overexpression of FRD was accomplished by cloning a promoterless copy of the frdABCD operon from WT 14028s into pTrc99a, conferring IPTG-inducible control.
RNA was isolated from 5 ml mid-log (OD600 = ~1.0) cultures of S. Typhimurium (14028s) and derivatives cultivated in M9 medium using Qiagen RNA-Mini Kit (Cat. # 74104) as per manufacturer instructions. RNA was stabilized by the addition of 5 ml RNAProtect (Qiagen). RNA was spectrophotometrically quantified and 50 ng of total RNA was analyzed per Q RT-PCR reaction using the QuantiTect™ SYBR® Green RT-PCR kit (Qiagen). Reaction conditions were as specified by Qiagen and reactions were performed and analyzed using a Rotor Gene™ 2000 Real Time Cycler (Corbett Research, Sydney, Australia).
Overnight cultures were diluted 1:100 in fresh BHI medium, incubated at 37°C with constant agitation, and allowed to reach an OD600 of 1.0 before treatment with the NO·-donor SperNO (1mM) or no treatment (control) for 15 min. S. Typhimurium FLS186 (lpdA::CmR) or its isogenic WT parent were grown under identical conditions but not treated with SperNO. RNA from 1 mL culture aliquots of treated and untreated cells was stabilized by the addition of 5 ml RNAProtect (Qiagen). Total RNA was isolated using the Qiagen RNeasy midi kit according to the manufacturer’s guidelines.
For synthesis of cDNA, total RNA was first annealed with N6 random hexamers at 70°C for 10 min and allowed to cool on ice. The cDNA was synthesized in a reaction containing a nucleotide mix with a 2:3 ration of aminoallyl-dUTP:dTTP (Sigma-Aldrich, St. Louis, MO), Super Script- II reverse transcriptase (Invitrogen) and 50 μg total RNA incubated at 42°C overnight. Residual RNA was hydrolyzed in a final concentration of 0.3M NaOH and 0.125M EDTA at 65°C for 10 min, then neutralized with 1M Tris-HCL pH8.0. Unincorporated nucleotides were removed using a Qiaquick PCR purification kit with modified wash buffer: 5mM KPO4 pH8.0 and 80% ethanol. Aminoallyl-labled cDNA was eluted from the columns in water and concentrated on a SpeedVac. Cy5 and Cy3 dyes were coupled to the aminoallyl-labeled cDNA by resuspending the dried cDNA in 0.1M sodium carbonate (pH9.3) with either Cy5 or Cy3 dyes (GE HealthCare) dissolved in DMSO for 1 hr at room temperature. The reaction was neutralized by the addition of 3M sodium acetate (pH5.2) and excess dye removed by purification on a Qiaquick PCR column using the manufacturer’s wash buffer. Equal amounts of differentially-labeled cDNA (Cy5/Cy3) from control and SperNO-treated samples or control and FLS186 (lpdA::CmR) samples were hybridized to the Salmonella cDNA array, scanned using a Packard Biosciences ScanArray 5000 and quantified using DigitalGENOME (molecularware) spot-finding software at the Institute of Systems Biology (Seattle, WA).
Total RNA from three biologically independent replicates was analyzed. Genes showing differential expression from at least two of the three replicates were considered for further analyses. The levels of induction/repression were determined by averaging fold-change across all three biological replicates. Any gene with a significant mean expression change that was ≥ 2 induced/repressed was considered to be differentially expressed. Full data sets from microarray results have been submitted to GEO database (reference number GSE29735).
Six to eight week old CH3/HeN mice (Charles River Laboratories) were inoculated intraperitoneally with 2.5×103 cfu of wild type (14028s) and an isogenic ΔmetD mutant. Inocula were verified by plating on LB agar. The mice were monitored twice daily for 28 days and moribund mice sacrificed per IACUC protocol. To selectively inhibit iNOS-activity, mice were administered L-NIL (L-N6-(1-iminoethyl)-lysine, Calbiochem, La Jolla, CA) at 500 mg·ml−1 via drinking water 24h prior to inoculation and throughout the experiment.
For competitive infections, 8-week old female C3H/HeN mice (Taconic Farms) were inoculated i.p. with 2000 cfu comprised of a 1:1 mixture of WT S. Typhimurium 14028s and either an isogenic ΔmetD mutant (n = 5) or the complemented ΔmetD mutant strain carrying plasmid pJK688 (metNIQ) (n = 5). 516 Eight days post-infection, the spleens and livers were harvested and homogenized in PBS and plated 518 on LB agar plates to determine CFU. Colonies were additionally screened on 519 appropriate antibiotics to determine the ratios of mutant/WT. Competitive indices 520 were determined as the ratio of (mutant/WT)OUT to (mutant/WT)IN. P values were 521 determined using the Wilcoxon Rank Sum Test.
One liter of S. Typhimurium strain 14028s cultured in M9 medium was harvested at mid-exponential phase (OD600 = 0.3–0.5) by centrifugation and washed twice with sterile PBS. Cells were resuspended in 1 ml PBS and disrupted by sonication. Cell debris and unlysed bacteria were removed via centrifugation and total protein quantified using the Pierce bicinchoninic acid (BCA) protein assay reagent (Thermo Scientific, Rockford, IL).
LpdA activity was detected in cell free extracts (100 μg total protein per reaction) by spectrophotometrically monitoring the dihydrolipoamide-dependent reduction of 3-acetyl-NAD+ at 375 nm as previously described (Creaghan and Guest, 1972). Similarly, pyruvate dehydrogenase and α-ketoglutarate dehydrogenase activities were determined by monitoring the pyruvate- and α-ketoglutarate-dependent formation of 3-acetyl-NADH in cell free extracts. One hundred μg of protein from extracts of M9-grown bacteria were used to assay pyruvate dehydrogenase activity, whereas 1 mg of protein from LB-grown cell extracts was used to assay α-ketoglutarate dehydrogenase activity. A ΔlpdA mutant-extract served as a negative control for background activity, and 1 U of recombinant porcine LpdA (Calzymes, San Luis Obispo, CA) was used as a positive control. LB-grown cell extracts were also used to quantify malate dehydrogenase and isocitrate dehydrogenase activities as well as aconitase activity. Malate dehydrogenase activity was determined by spectrophotometrically monitoring the oxaloacetate-dependent oxidation of NADH at 340 nm in PBS (pH 7.4) in reactions with 100 μg total protein as described (Corp, 2009). Similarly, isocitrate dehydrogenase activity was quantified from extracts containing 100 μg total protein by spectrophotometrically monitoring the isocitrate-dependent reduction of NADP+ at 340 nm as previously described (Hirsch, 1952). Aconitase activity was assayed in reactions containing 100 μg total protein immediately after extract preparation using previously described methods (Gardner, 2002).
To assess the NO·-sensitivity of given enzymes, reactions were run identically with or without increasing concentrations (from 15.625 μM to 500 μM) of the NO·-donor ProliNO (Proline-NONOate, Calbiochem, La Jolla, CA). ProliNO was administered 2 min prior to initiating the reaction and NO·-mediated inhibition was expressed as percent activity compared with untreated extracts.
We would like to acknowledge Drs. M. McClelland and S. Porwollik for microarray analyses as well as A. S. Richardson for statistical support and analyses. This project was supported by grants from the National Institutes of Health (AI055396 to A.R.R., and AI39557 and AI77629 to F.C.F.).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.