The role of metals in hypothiocyanite resistance in Escherichia coli

Abstract

The innate immune system employs a variety of antimicrobial oxidants to control and kill host-associated bacteria. Hypothiocyanite/hypothiocyanous acid (⁻OSCN/HOSCN) is one such antimicrobial oxidant that is synthesized by lactoperoxidase, myeloperoxidase, and eosinophil peroxidase at sites throughout the human body. HOSCN has potent antibacterial activity while being largely non-toxic towards human cells. The molecular mechanisms by which bacteria sense and defend themselves against HOSCN have only recently begun to be elaborated, notably by the discovery of bacterial HOSCN reductase (RclA), an HOSCN⁻degrading enzyme widely conserved among bacteria that live on epithelial surfaces. In this paper, I show that Ni²⁺ sensitizes Escherichia coli to HOSCN by inhibiting glutathione reductase, and that inorganic polyphosphate protects E. coli against this effect, probably by chelating Ni²⁺ ions. I also found that RclA is very sensitive to inhibition by Cu²⁺ and Zn²⁺, metals that are accumulated to high levels by innate immune cells, and that, surprisingly, thioredoxin and thioredoxin reductase are not involved in HOSCN stress resistance in E. coli. These results advance our understanding of the contribution of different oxidative stress response and redox buffering pathways to HOSCN resistance in E. coli and illustrate important interactions between metal ions and the enzymes bacteria use to defend themselves against oxidative stress.

IMPORTANCE

Hypothiocyanite (HOSCN) is an antimicrobial oxidant produced by the innate immune system. The molecular mechanisms by which host-associated bacteria defend themselves against HOSCN have only recently begun to be understood. The results in this paper are significant because they show that the redox buffer glutathione and enzyme glutathione reductase are critical components of the Escherichia coli HOSCN response, working by a mechanism distinct from that of the HOSCN⁻specific defenses provided by the RclA, RclB, and RclC proteins, and that metal ions (including nickel, copper, and zinc) may impact the ability of bacteria to resist HOSCN by inhibiting specific defensive enzymes (e.g. glutathione reductase or RclA).

INTRODUCTION

Hypothiocyanite/hypothiocyanous acid (⁻OSCN/HOSCN) is an antimicrobial pseudohypohalous acid synthesized from hydrogen peroxide (H₂O₂) and thiocyanate (SCN⁻) by mammalian heme peroxidases, including lactoperoxidase (LPO), myeloperoxidase (MPO), and eosinophil peroxidase (EPO)(1, 2). HOSCN is less reactive than the other major antimicrobial products of heme peroxidases, particularly when compared to the extremely reactive hypochlorous acid (HOCl) produced by MPO (1, 3). HOSCN is a specific thiol-oxidizing agent (2, 3) and is essentially non-toxic to human cells, a fact which has been attributed to the ability of mammalian selenocysteine-containing thioredoxin reductase to reduce HOSCN to nontoxic SCN⁻ and H₂O (4). In contrast, bacterial thioredoxin reductase is inhibited by HOSCN (4), as are a number of central bacterial metabolic enzymes (5–7). HOSCN has therefore long been considered a specifically antimicrobial product which the innate immune system uses to control bacteria without causing damage to host tissues, a model which has generally assumed that bacteria lack effective defenses against HOSCN (8–11).

That assumption has recently been shown to be an oversimplification, as recent studies from a number of laboratories have identified HOSCN⁻specific stress responses in a variety of host-associated bacteria (11). We identified RclA as an HOSCN⁻induced, highly-active bacterial HOSCN reductase that protects Escherichia coli and some other intestinal bacteria from HOSCN stress (12), followed rapidly by other research groups showing that the RclA homologs from Streptococcus pneumoniae (Har) and Staphylococcus aureus (MerA) played similar roles in those species (13, 14). Pseudomonas aeruginosa lacks an RclA homolog, but mounts an HOSCN stress response that depends on inorganic polyphosphate and a peroxiredoxin-like protein of unknown function called RclX (15, 16). The HOSCN stress response of S. pneumoniae is perhaps the best characterized among these bacteria, and studies in that species have identified both HOSCN⁻specific defense mechanisms (i.e. HOSCN reductase) and overlaps with more general oxidative stress and redox homeostasis mechanisms (e.g. the glutathione and thioredoxin systems)(14, 17–21).

Older studies of the antimicrobial activity of LPO generally did not use purified HOSCN, as we and the other labs cited in the previous paragraph have done more recently. Instead, they typically added a combination of LPO, SCN⁻, and one of a variety of H₂O₂-generating enzymatic systems to synthesize HOSCN in situ (22–25). This is in some ways more representative of the in vivo conditions in which bacteria are exposed to HOSCN, but makes it more difficult to control HOSCN concentrations or to separate bacterial responses to HOSCN from those to H₂O₂ or to other products of LPO (1, 26–29). Roughly 20 years ago, the laboratory of C.W. Michiels carried out a series of experiments to characterize the response of E. coli to the LPO / SCN⁻ / H₂O₂ system, in which they found that high pressure sensitizes E. coli to killing by this system (30, 31), that outer membrane lipids and porins play an important role in sensitivity (32), and that LPO / SCN⁻ / H₂O₂ induces a different response than other oxidative stressors (33, 34). Curiously, they also observed that exogenous nickel (Ni²⁺) sensitized E. coli to LPO / SCN⁻ / H₂O₂ (35), but provided no mechanistic explanation for this phenomenon.

In this paper, I confirm that Ni²⁺ does indeed sensitize E. coli to purified HOSCN and show that it does so by inhibiting glutathione reductase (Gor) activity. Since HOSCN efficiently inhibits E. coli thioredoxin reductase (TrxB)(4), this means that the combination of HOSCN and Ni²⁺ simultaneously disrupts both of E. coli’s cytoplasmic redox buffering systems, rendering it very sensitive to oxidative damage (36). I also show that RclA is subject to inhibition by metals, including Ni²⁺ but much more sensitively by Cu²⁺ and Zn²⁺. However, Ni²⁺-sensitization of E. coli to HOSCN was not RclA-dependent. These results confirm that low molecular weight thiols like glutathione are important conserved HOSCN resistance elements in bacteria (19, 37), help connect the older literature on bacterial responses to LPO to current work on the molecular mechanisms of HOSCN responses, and identify new connections between metal ions and the enzymes bacteria use to defend themselves against oxidative stress.

RESULTS

Nickel enhances HOSCN toxicity in E. coli.

Sermon et al. (35) reported in 2005 that E. coli is sensitized to killing by the LPO / SCN⁻ / H₂O₂ system in the presence of 300 μM Ni²⁺, and that mutants lacking the Ni²⁺ / Co²⁺ / Mg²⁺ transport protein CorA (38, 39) are protected from this effect. They did not propose a mechanism to explain this unusual phenotype. Recent work from my lab and others (12–19) has now begun to identify the molecular mechanisms underlying HOSCN resistance in bacteria, and I therefore revisited this observation to see if I could gain new insights into how Ni²⁺ or other metal ions might impact that process. First, I tested whether Ni²⁺ could sensitize E. coli to growth inhibition by purified HOSCN in minimal medium, the method which has been used in most of the recent papers examining bacterial HOSCN response (12–14, 17–19), rather than by the more complex enzymatic LPO / SCN⁻ / H₂O₂ system (1, 11, 40). As shown in Fig 1, in this assay addition of 300 μM NiSO₄ greatly enhanced the toxicity of 400 μM HOSCN for the wild-type, but not for the ΔcorA mutant, consistent with the results of Sermon et al. (35). 300 μM NiSO₄ did not inhibit growth of E. coli under these conditions (Supplemental Fig S1A) and neither did SCN⁻ at concentrations up to 40 mM (Supplemental Fig S1B). Addition of SCN⁻ did not change the minimal inhibitory concentration of NiSO₄ (1.25 – 2.5 mM)(Supplemental Fig S1C) and Ni²⁺ did not react directly with HOSCN (Supplemental Fig S2), indicating that the phenotype seen in Fig 1 was due to a physiological impact of Ni²⁺ on E. coli that sensitizes the cells to HOSCN.

FIG 1. Nickel enhances HOSCN toxicity.

E. coli strains MG1655 (wild-type) and MJG2314 (MG1655 ΔcorA761::kan⁺) were grown in MOPS minimal medium without micronutrients containing 0.2% glucose, supplemented with 300 μM NiSO4 and/or 400 μM HOSCN, as indicated, incubating at 37°C with shaking and measuring A600 at 30-minute intervals for 24 hours (n=3 experimental replicates, with 3 technical replicates per experiment; error bars of 1 standard deviation).

Inorganic polyphosphate protects E. coli against nickel-dependent HOSCN sensitization.

Inorganic polyphosphate (polyP) is a bacterial stress response effector that protects against oxidative stress by chelating metal ions and stabilizing damaged proteins (41–43). Mutants of Pseudomonas aeruginosa lacking the ppk1 gene encoding polyP kinase, which are therefore defective in polyP synthesis (44), are sensitive to inhibition by HOSCN (16). In contrast, E. coli Δppk mutants, which completely lack polyP (41, 45), were not more sensitive to HOSCN except in the presence of NiSO₄ (Fig 2), indicating that the primary role of polyP in the E. coli HOSCN response, at least under these growth conditions, is to shield against the HOSCN toxicity-enhancing effect of metals. A mutant lacking ppx, encoding the polyP-degrading exopolyphosphatase PPX (46), was slightly more sensitive to HOSCN than the wild-type, but, like the wild-type, did not become more sensitive when NiSO₄ was added under these conditions (Fig 2). It is worth noting that I performed the experiments in Fig 2 in minimal medium supplemented with 0.1% casamino acids, since ppk mutants grow very poorly without amino acids (47, 48), and that the sensitivity of E. coli to both HOSCN and NiSO₄ was somewhat different under these conditions from that seen in unsupplemented minimal medium (e.g. Fig 1).

FIG 2. Polyphosphate protects E. coli against nickel-dependent HOSCN sensitization.

E. coli strains MG1655 (wild-type), MJG0224 (MG1655 Δppk-749), and MJG0315 (MG1655 Δppx-750) were grown in MOPS minimal medium without micronutrients containing 0.2% glucose and 0.1% casamino acids, supplemented with 500 μM NiSO4 and/or 400 μM HOSCN, as indicated, incubating at 37°C with shaking and measuring A600 at 30-minute intervals for 24 hours (n=3 experimental replicates, with 4 technical replicates per experiment; error bars of 1 standard deviation).

Copper, zinc, and nickel inhibit RclA activity in vitro.

The E.coli HOSCN reductase RclA (12) (homologs of which are known as Har or MerA in Streptococcus pneumoniae and Staphylococcus aureus, respectively)(13, 14), is a flavin-dependent oxidoreductase that efficiently degrades HOSCN and protects bacterial cells against HOSCN toxicity, but is also known to interact with a variety of metal cations, notably Cu²⁺ and Hg²⁺ (49–51). One possible explanation for sensitization of E. coli to HOSCN by Ni²⁺, therefore, is that the metal ions might inhibit RclA’s HOSCN reductase activity, presumably by binding to the active site cysteine residues of RclA (50). In vitro assays with purified RclA (12, 49) and a variety of biologically relevant divalent metals revealed that Cu²⁺ and Zn²⁺ were extremely potent RclA inhibitors, capable of significantly inhibiting HOSCN reductase activity at concentrations as low as 1 μM (Fig 3A). Neither Cu²⁺ nor Zn²⁺ reacted with HOSCN on their own (Supplemental Fig S2). Ni²⁺ also significantly inhibited RclA activity, but only at much higher concentrations (>100 μM)(Figs 3A,B). Ca²⁺, Co²⁺, Mg²⁺, and Mn²⁺ had no impact on RclA activity (Fig 3A).

FIG 3. Copper, zinc, and nickel inhibit RclA activity in vitro.

Reactions (500 μl) contained 10 mM HEPES-KOH buffer (pH 7.4), 160 μM NADH, 100 μM HOSCN, and the indicated concentrations of metal salts, and were started by addition of 10 nM RclA. I quantified NADH consumption at A340 for 1 min at 20°C and calculated specific activities as μmol NADH consumed min⁻¹ mg⁻¹ RclA (n=3–4 experimental replicates; error bars of 1 standard deviation). Asterisks indicate activities significantly different from that of RclA with no metals added in a given set of experiments (one-way ANOVA with Dunnett9s multiple comparisons test; ** = P<0.01. *** = P<0.001, **** = P<0.0001). X symbols indicate that RclA precipitated at 100 μM CuCl2.

Sensitization of E. coli to HOSCN by nickel is independent of RclA, RclB, and RclC.

Testing whether metals could potentiate HOSCN toxicity by inhibiting RclA in vivo presented some technical challenges. While NiSO₄ was soluble in MOPS minimal medium up to a concentration of at least 40 mM (Supplemental Fig S1A), both CuCl₂ and ZnSO₄ formed visible precipitates (presumably insoluble phosphate salt crystals) when added to MOPS medium at concentrations higher than 1 μM (data not shown). At this concentration, CuCl₂, the most potent inhibitor of RclA in vitro (Fig 3A), had no effect on the HOSCN sensitivity of either wild-type or ΔrclA E. coli (Fig 4). To try to address this, I deleted the copA gene, encoding the Cu export proteins CopA(Z) (38, 52, 53), which would be expected to increase the accumulation of Cu in the cytoplasm in E. coli. While this mutation did indeed confer a detectable Cu sensitivity phenotype at 1 μM CuCl₂, it did not have a large impact on HOSCN sensitivity, and notably, any such impact was also apparent in the ΔrclA ΔcopA double mutant (Fig 4), indicating that Cu was probably not affecting HOSCN sensitivity in this case by inhibiting RclA. I used a similar approach with the Zn and Ni exporters ZntA and RcnA (Supplemental Fig S3)(38, 54, 55), and found that deletion of rcnA, which did substantially sensitize E. coli to NiSO₄ (Supplemental Fig S4), led to a small increase in HOSCN toxicity in the presence of 100 μM NiSO₄, but not 1 μM CuCl₂ or ZnSO₄. None of the export mutants led to detectable increases in HOSCN sensitivity in the presence of 1 μM ZnSO₄ (Supplemental Fig S3).

FIG 4. Sensitization of E. coli to HOSCN in vivo by copper is slight, and not RclA-dependent.

E. coli strains MG1655 (wild-type), MJG1759 (MG1655 ΔcopA767::kan⁺), MJG1958 (MG1655 ΔrclA747), and MJG2358 (MG1655 ΔrclA747 ΔcopA767::kan⁺) were grown in MOPS minimal medium without micronutrients containing 0.2% glucose, supplemented with 1 μM CuCl2 and/or 400 μM HOSCN, as indicated, incubating at 37°C with shaking and measuring A600 at 30-minute intervals for 24 hours (n=4 experimental replicates, with 3 technical replicates per experiment; error bars of 1 standard deviation).

E. coli ΔrclA mutants are more sensitive to HOSCN inhibition than wild-type cells (12), a phenotype that is easiest to visualize at lower HOSCN concentrations than the 400 μM treatment used in Figs 1, 2, and 4. The concentration of HOSCN used did, however, impact the ability of NiSO₄ to enhance HOSCN toxicity. When exposed to 200 μM HOSCN, wild-type E. coli were only modestly sensitized by addition of 300 μM NiSO₄, unlike the ΔrclA mutant, which was sensitized quite strongly (Fig 5). This indicates that the mechanism by which Ni²⁺ enhances HOSCN toxicity cannot possibly depend on inhibition of RclA. The same was true of the ΔrclABC triple mutant (Fig 5), which additionally lacks the RclB and RclC proteins (49, 56), indicating that neither RclB nor RclC were the target of Ni²⁺ either. RclB and RclC protect E. coli against HOSCN (12) by currently unknown mechanisms, and while the ΔrclABC mutant was not more sensitive to inhibition by 200 μM HOSCN than the ΔrclA mutant under these growth conditions (Fig 5), it was completely inhibited by 300 μM HOSCN (Supplemental Fig S5), unlike the ΔrclA mutant, which was able to eventually recover growth even at 400 μM HOSCN (Fig 4).

FIG 5. Nickel enhancement of HOSCN toxicity is independent of RclABC.

E. coli strains MG1655 (wild-type), MJG1958 (MG1655 ΔrclA747), and MJG0901 (MG1655 ΔrclABC1) were grown in MOPS minimal medium without micronutrients containing 0.2% glucose, supplemented with 300 μM NiSO4 and/or 200 μM HOSCN, as indicated, incubating at 37°C with shaking and measuring A600 at 30-minute intervals for 24 hours (n=3 experimental replicates, with 3 technical replicates per experiment; error bars of 1 standard deviation).

Glutathione is a major HOSCN defense in E. coli.

Having eliminated the known HOSCN⁻specific defenses of E. coli (i.e. RclABC) as targets of Ni²⁺, I next moved on to testing whether other genes involved in related oxidative stress responses might be involved. Deletion of the nemA, hslO, and mgsA genes, encoding proteins involved in defense against HOCl in E. coli (41, 57, 58), had no impact on HOSCN resistance (Fig 6A). Similarly, while deletion of the genes encoding thioredoxin reductase (trxB)(20, 38) or either of the two thioredoxin homologs of E. coli (trxA and trxC)(20, 38, 59) modestly slowed the growth of E. coli under non-stress conditions, none of those mutants had substantial HOSCN sensitivity phenotypes (Fig 6B). The same was true of a ΔgrxA mutant (Fig 6C) lacking glutaredoxin (21, 38) However, a Δgor mutant lacking glutathione oxidoreductase (Gor)(38, 60) was extremely sensitive to HOSCN. The E. coli Δgor strain was completely inhibited by 400 μM HOSCN (Fig 6C) and both Δgor and ΔgshA mutants (which are unable to synthesize glutathione)(38, 61) were strongly inhibited by 100 μM HOSCN, a concentration which barely effected the growth of the wild-type (Fig 7). Glutathione is abundant in E. coli (62), is oxidized efficiently by HOSCN (63), and S. pneumoniae gor mutants are also very sensitive to HOSCN (19), so this phenotype was not an enormous surprise. The question, though, was whether Gor or glutathione are involved in the Ni²⁺-sensitization phenotype. Although the variability in the recovery time of HOSCN⁻stressed Δgor and ΔgshA mutants was relatively high (note the large error bars in Fig 7), neither mutant appeared to be notably further sensitized to HOSCN by 300 μM NiSO₄.

FIG 6. Glutathione reductase contributes to E. coli HOSCN resistance.

E. coli strains MG1655 (wild-type), MJG0044 (MG1655 ΔnemA728), MJG0081 (MG1655 hslO::kan⁺), MJG0176 (MG1655 ΔmgsA778), MJG1958 (MG1655 ΔrclA747), MJG2371 (MG1655 ΔtrxB786::kan⁺), MJG2385 (MG1655 ΔtrxA732::kan⁺), MJG2387 (MG1655 ΔtrxC750::kan⁺), MJG2388 (MG1655 ΔgrxA750::kan⁺), MJG2398 (MG1655 Δgor-756::kan⁺) were grown in MOPS minimal medium without micronutrients containing 0.2% glucose, with or without 400 μM HOSCN, as indicated, incubating at 37°C with shaking and measuring A600 at 30-minute intervals for 24 hours (n=3 experimental replicates, with 3–4 technical replicates per experiment; error bars of 1 standard deviation).

FIG 7. Glutathione mutants are extremely sensitive to HOSCN.

E. coli strains MG1655 (wild-type), MJG2398 (MG1655 Δgor-756::kan⁺), and MJG2485 (MG1655 ΔgshA769:kan⁺) were grown in MOPS minimal medium without micronutrients containing 0.2% glucose, with or without 100 μM HOSCN and/or 300 μM NiSO4, as indicated, incubating at 37°C with shaking and measuring A600 at 30-minute intervals for 24 hours (n=3 experimental replicates, with 4 technical replicates per experiment; error bars of 1 standard deviation).

Glutathione reductase is inhibited by nickel, but not by HOSCN.

The in vivo results in Fig 7 suggested that Gor might be the target of Ni²⁺ in HOSCN⁻stressed cells but were not conclusive. I therefore used cell-free lysates of E. coli to directly test the impact of both NiSO₄ and HOSCN on Gor activity in vitro (64, 65). NiSO₄ inhibited Gor (Fig 8A), but HOSCN did not (Fig 8B), and HOSCN⁻oxidized lysates were as sensitive to inhibition by NiSO₄ as lysates which had not been exposed to HOSCN (Fig 8B). Since polyP could counteract the HOSCN⁻sensitizing effect of NiSO₄ in vivo (Fig 2), I added purified polyP to lysates to see if it could also suppress Gor inhibition in vitro. As shown in Fig 8C, 5 mM polyP (concentration expressed in terms of individual phosphate units due to the heterogenous length of commercially-available polyP) significantly reduced the ability of 1 mM NiSO₄ to inhibit Gor activity, presumably due to its ability to chelate divalent metal cations (66–68). Addition of 5 mM of the chelator EDTA completely eliminated NiSO₄ inhibition of Gor activity, and 5 mM polyP had no effect on Gor activity in the absence of NiSO₄ (Fig 8C). These results support the model that Gor is the target that Ni²⁺ affects to sensitize E. coli to HOSCN. In contrast to their effects on Gor activity, 5 mM NiSO₄ did not inhibit thioredoxin reductase (TrxB) activity in lysates, but HOSCN completely inactivated it (Fig 8D), consistent with previous reports that E. coli TrxB is HOSCN-sensitive (4).

FIG 8. Glutathione reductase is inhibited by nickel but not by HOSCN.

Glutathione and thioredoxin reductase activity was measured in cell-free lysates of E. coli MG1655. Glutathione reductase assays (A-C) contained 50 mM HEPES-KOH buffer (pH 8), 1.2 mM oxidized glutathione, 600 μM DTNB, 350 μM NADPH, and thioredoxin reductase assays (D) contained 50 mM HEPES-KOH buffer (pH 8), 500 nM E. coli thioredoxin 1 (TrxA), 0.1 mg ml⁻¹ bovine serum albumin, 500 μM DTNB, and 240 μM NADPH. Reactions also contained the indicated concentrations of NiSO4, polyP, and/or EDTA, and were started by addition of cell lysate (1.63–16.3 μg total protein). DTNB oxidation to TNB was measured over time, using the TNB extinction coefficient at 412 nm of 14100 M⁻¹ cm⁻¹, then specific activities were calculated as nmol GSH or TrxA reduced min⁻¹ mg⁻¹ total protein (n=3–6 experimental replicates; error bars of 1 standard deviation). HOSCN-oxidized lysates (used as indicated in B and D) were prepared by incubating cell lysate (3.26 mg ml⁻¹ total protein) for 5 min at room temperature with 3 mM HOSCN, then quenching unreacted HOSCN with an equal volume of 2 mM TNB. Asterisks indicate activities significantly different from that of (A, B, and D) lysate with no NiSO4 added or (C) lysate with 1 mM NiSO4 added (filled circles, one-way ANOVA with Holm-Sidak9s multiple comparisons test or mixed model with Dunnett’s multiple comparisons test)(* = P<0.05, ** = P<0.01, *** = P<0.001).

DISCUSSION

The molecular details of the layered defenses that different bacteria employ to protect themselves against HOSCN are now beginning to be deciphered (11–19, 37), decades after HOSCN was first identified as an antimicrobial compound produced by the immune system (69). This paper now shows that glutathione is a key component in the E. coli HOSCN stress response, distinct from the protection provided by more HOSCN-specific defenses (e.g. RclABC)(12). HOSCN is a specific thiol oxidant (2) and glutathione is one of the most abundant metabolites in the E. coli cytoplasm (reaching a concentration of 17 mM in cells grown in minimal glucose medium)(62), so it is not surprising that HOSCN stress would result in glutathione oxidation, and indeed, low molecular weight thiols (e.g. glutathione and bacillithiol)(70, 71) appear to be important for defense against HOSCN in a variety of bacteria (19, 37). My results also explain a puzzling older result (35) by showing that Ni²⁺ inhibits E. coli Gor, thereby sensitizing E. coli to HOSCN. Ni²⁺ has also been reported to inhibit both yeast and bovine glutathione reductases (72, 73), so this may be a general property of these enzymes, and could potentially contribute to the oxidative stress observed when diverse organisms are exposed to toxic levels of nickel (74–76).

I was somewhat surprised to find that the thioredoxin system does not appear to play an important role in HOSCN response in E. coli (Fig 6), since trxA mutants of S. pneumoniae are sensitive to HOSCN (18) and TrxB’s sensitivity to HOSCN inactivation has been proposed to be partially responsible for the antimicrobial activity of HOSCN (4). Based on my results, this seems unlikely, at least for E. coli, although since trxB gor and trxB gshA double mutants are exceptionally sensitive to oxidative stress (36), concurrent inactivation of TrxB by HOSCN and Gor by Ni²⁺ or other metals (Figs 3, 8) could amplify the individual impacts of these compounds and contribute to the Ni²⁺-sensitization phenotype (Fig 1).

One question that remains unanswered is whether the sensitivity of HOSCN defense enzymes to inhibition by metals (Figs 3, 8) is relevant to the ability of bacteria to resist HOSCN produced by their mammalian hosts (11). Ni²⁺ is present in food and water at widely varying concentrations (76), and E. coli has genes encoding both Ni²⁺ import proteins (e.g. nikABCDE, corA)(38, 39, 77) and exporters (e.g. rcnA)(55), implying that it is exposed to varying Ni²⁺ concentrations in its natural environment in the large intestine, some of which can reach potentially toxic levels (74). Whether the 300–500 μM Ni²⁺ concentrations that sensitized E. coli to HOSCN in laboratory growth media occur under physiological conditions where E. coli is also exposed to HOSCN is difficult to determine. For one thing, to my knowledge no studies to date have attempted to quantify HOSCN in the large intestine, although the presence of dedicated HOSCN resistance genes (e.g. rclABC) in diverse intestinal bacteria suggests it is present in at least some situations (11, 49). Cu²⁺ and Zn²⁺ were potent inhibitors of RclA activity in vitro (Fig 3), and while solubility issues in the minimal media I used to grow E. coli prevented me from determining unambiguously whether either of these metals could inhibit RclA in vivo, that does not necessarily mean that it cannot happen in the host environment. Both Cu²⁺ and Zn²⁺ are concentrated in the phagosomes of immune cells as part of the antimicrobial arsenal of those cells (78) and can reach much higher concentrations than I could achieve here (e.g. > 100 μM Cu²⁺ in the phagosomes of macrophages that have phagocytosed Mycobacterium cells)(79). It is certainly possible that under those conditions, inhibition of RclA by Cu²⁺ and/or Zn²⁺ could have a physiologically relevant impact on the ability of bacteria to survive host-produced oxidative stress. The expression of the rclABC operon of E. coli is strongly induced upon phagocytosis by neutrophils (80) and rclA contributes to the ability of Salmonella to survive phagocytosis by macrophages (50), but more work will need to be done to determine whether and under what circumstances metals affect bacterial HOSCN defenses during interactions with the host immune system.

MATERIALS AND METHODS

Bacterial strains and growth conditions

All strains used in this study are listed in Table 1. I grew E. coli at 37°C in Lysogeny Broth (LB)(81) containing 5 g l⁻¹ NaCl or in MOPS minimal medium (82) containing 2 g l⁻¹ glucose, with kanamycin (25 or 50 μg ml⁻¹) added to LB medium when appropriate. For in vivo experiments to assess the impact of metal ions on HOSCN resistance, I prepared MOPS medium without adding micronutrients (i.e. (NH₄)₆Mo₇O₂₄, H₃BO₃, CoCl₂, CuSO₄, MnCl₂, ZnSO₄, and CaCl₂). Minimal medium was used for all in vivo tests of HOSCN sensitivity due to the known ability of the components of undefined rich media to react with antimicrobial oxidants, including HOSCN (83–85).

TABLE 1.

Strains used in this study. Unless otherwise indicated, strains were generated in the course of this work. Abbreviations: Kn^R, kanamycin resistance.

Strain	Marker(s)	Relevant Genotype	Source
E. coli strains:
MG1655		F−, λ−, rph-1 ilvG− rfb-50	(87)
JW138	KnR	hslO::kan +	(90)
MJG0044		MG1655 ΔnemA728	(57)
MJG0081	KnR	MG1655 hslO::kan+
MJG0176		MG1655 ΔmgsA778	(41)
MJG0224		MG1655 Δppk-749	(41)
MJG0315		MG1655 Δppx-750	(41)
MJG0901		MG1655 ΔrclABC1	(80)
MJG1759	KnR	MG1655 ΔcopA767::kan+	(49)
MJG1958		MG1655 ΔrclA747	(12)
MJG2314	KnR	MG1655 ΔcorA761::kan+
MJG2344	KnR	MG1655 ΔzntA724::kan+
MJG2346	KnR	MG1655 ΔrcnA731::kan+
MJG2358	KnR	MG1655 ΔrclA747 ΔcopA767::kan+
MJG2371	KnR	MG1655 ΔtrxB786::kan+
MJG2385	KnR	MG1655 ΔtrxA732::kan+
MJG2387	KnR	MG1655 ΔtrxC750::kan+
MJG2388	KnR	MG1655 ΔgrxA750::kan+
MJG2398	KnR	MG1655 Δgor-756::kan+
MJG2485	KnR	MG1655 ΔgshA769::kan+

Databases and primer design

I obtained gene and protein sequences and other information from the Integrated Microbial Genomes database (86) and from EcoCyc (38). I designed PCR and sequencing primers with Web Primer (www.candidagenome.org/cgi-bin/compute/web-primer).

Strain construction

All E. coli strains used in this study were derivatives of wild-type strain MG1655 (F⁻, λ⁻, rph-1 ilvG⁻ rfb-50) (87). I used P1vir phage transduction (88, 89) to move the hslO::kan⁺ allele from E. coli strain JW138 (90), which is a 737 bp internal deletion of hslO tagged with a kanamycin-resistance cassette (91), into MG1655, generating strain MJG0081 (MG1655 hslO::kan⁺). I used P1vir transduction to move the ΔcorA761::kan⁺, ΔcopA767::kan⁺, ΔzntA724::kan⁺, ΔrcnA731::kan⁺, ΔtrxB786::kan⁺, ΔtrxA732::kan⁺, ΔtrxC750::kan⁺, ΔgrxA750::kan⁺, Δgor-756::kan⁺, and ΔgshA769::kan⁺ alleles from the Keio collection (92) into MG1655 or MJG1958 (MG1655 ΔrclA747)(12), generating strains MJG2314 (MG1655 ΔcorA761::kan⁺), MJG2344 (MG1655 ΔzntA724::kan⁺), MJG2346 (MG1655 ΔrcnA731::kan⁺), MJG2358 (MG1655 ΔrclA747 ΔcopA767::kan⁺), MJG2371 (MG1655 ΔtrxB786::kan⁺), MJG2385 (MG1655 ΔtrxA732::kan⁺), MJG2387 (MG1655 ΔtrxC750::kan⁺), MJG2388 (MG1655 ΔgrxA750::kan⁺), MJG2398 (MG1655 Δgor-756::kan⁺), and MJG2485 (MG1655 ΔgshA769::kan⁺). E. coli corA mutants are calcium-sensitive, so when phage were grown on corA strains, I used media containing only 5 mM CaCl₂ and 15 mM MgSO₄ (93). I confirmed chromosomal mutations by PCR and whole-genome sequencing (SeqCenter, Philadelphia, PA).

Preparation of HOSCN

I prepared HOSCN with lactoperoxidase (LPO; Worthington Biochemical cat. # LS000151) as previously described (12). I combined 1.2 μM LPO and 7.5 mM NaSCN in 10 mM HEPES-KOH, pH 7.4 (for in vitro experiments) or MOPS minimal medium (82) containing 0.2% glucose but lacking micronutrients (for in vivo experiments), then added 0.8 mM H₂O₂ three times at one-minute intervals and incubated on ice for 10 minutes. I added 30 μg of catalase to degrade excess H₂O₂, then removed the proteins with a 10K MWCO protein concentrator (Millipore) and quantified the resulting HOSCN solution (typically 3.5 to 4 mM) with 5-thio-2-nitrobenzoic acid (TNB; extinction coefficient at 412 nm = 14100 M⁻¹ cm⁻¹) by measuring the loss of absorbance at 412 nm after adding HOSCN (63). Cu, but neither Zn nor Ni, was able to catalyze a small amount of TNB oxidation (Fig S6), which was taken into account when necessary.

In vivo HOSCN stress treatment assays

I grew E. coli strains overnight at 37°C with shaking in MOPS minimal medium containing 0.2% glucose then normalized the cultures to A₆₀₀ = 1 and rinsed twice with sterile MOPS minimal medium containing 0.2% glucose but lacking micronutrients. For experiments involving Δppk mutants (i.e. Fig 2), which grow very poorly in the absence of amino acids (47, 94), the resuspension solution also included 4% (w/v) casamino acids (Fisher Scientific cat. #BP1424). I diluted the resulting cell suspensions 1:40 into MOPS minimal medium without micronutrients containing 0.2% glucose, supplemented with HOSCN and/or metal salts, as indicated. I performed growth curves in 200-μl volumes in clear 96-well plates in a Tecan Spark plate reader, incubating at 37°C with shaking and measuring A₆₀₀ at 30-minute intervals for 24 hours.

In vitro HOSCN reductase activity assay

The purification of Strep-tagged RclA has been previously described (12, 49). I measured the HOSCN reductase activity of purified RclA aerobically in 10 mM HEPES-KOH buffer (pH 7.4) at 20°C. Each reaction (500 μl) contained 160 μM NADH and, where indicated, 100 μM HOSCN and the indicated concentrations of metal salts and was started by addition of 10 nM RclA. I quantified NADH consumption over time using a GENESYS 10S UV-Vis spectrophotometer (ThermoFisher), measuring A₃₄₀ and using the NADH extinction coefficient at 340 nm of 6220 M⁻¹ cm⁻¹ to calculate NADH concentrations, then fit slopes to the resulting linear plots and calculated specific activities as μmol NADH consumed min⁻¹ mg⁻¹ RclA.

Glutathione and thioredoxin reductase activity assays

I prepared cell-free lysates of E. coli MG1655 from 400-ml overnight cultures grown in LB as follows: I rinsed pelleted cells with 50 mM HEPES-KOH buffer (pH 8) and stored them at −80°C before thawing on ice and resuspending in 5 ml of the same buffer. I then lysed the cells by sonication on ice in a Fisher Model 120 Sonic Dismembrator (5 min @ 50% amplitude, 5 sec on, 5 sec off), removed debris by centrifugation (40 min @ 21,000 g @ 4°C), and dialyzed thoroughly into 50 mM HEPES-KOH buffer (pH 8) at 4°C using 12–14 kDa MWCO dialysis tubing before storing aliquots at −80°C. I measured the total protein content of the lysates using the Bradford assay (ThermoFisher). Where indicated, I oxidized lysates before use by diluting them 1:5 into buffer containing HOSCN (3 mM final concentration), incubating for 5 min at room temperature, then adding an equal volume of 2 mM TNB to quench unreacted HOSCN (4).

I measured glutathione reductase activity in E. coli lysates by a modification of a previously published method (64). Reactions (500 μl) contained 50 mM HEPES-KOH buffer (pH 8), 1.2 mM oxidized glutathione (GSSG; Fisher Scientific cat. # AAJ6371503), 600 μM 5,5’-dithiobis-(2-nitrobenzoic acid)(DTNB), 350 μM NADPH, and the indicated concentrations of NiSO₄ and/or polyP (Acros Organics cat. # 390932500), and were started by addition of cell lysate (16.3 μg total protein). PolyP concentrations are expressed in terms of phosphate monomers, due to the heterogeneity in chain length of commercial polyP. I measured thioredoxin reductase activity in lysates by a modification of a similar previously published method (65). Those reactions (500 μl) contained 50 mM HEPES-KOH buffer (pH 8), 500 nM E. coli thioredoxin 1 (TrxA; Sigma-Aldrich cat. # T0910), 0.1 mg ml⁻¹ bovine serum albumin, 500 μM DTNB, 240 μM NADPH, and the indicated concentrations of NiSO₄ and were also started by addition of cell lysate (1.63–16.3 μg total protein). For both assays, I quantified DTNB oxidation to TNB over time using a GENESYS 10S UV-Vis spectrophotometer (ThermoFisher), measuring A₄₁₂ and using the TNB extinction coefficient at 412 nm of 14100 M⁻¹ cm⁻¹ to calculate TNB concentrations, then fit slopes to the resulting linear plots and calculated specific activities as nmol GSSG or thioredoxin reduced min⁻¹ mg⁻¹ total protein.

Statistical analyses

I used GraphPad Prism version 10.2.0 for Macintosh (GraphPad Software) to perform all statistical analyses, linear regressions, and graph generation. I based the graph color schemes on the color blindness-friendly “Bright qualitative color scheme” recommended of Dr. Paul Tol (95).

Data availability

All strains generated in the course of this work are available from the author upon request. I deposited DNA sequencing data in the NIH Sequence Read Archive (accession # PRJNA943195), and all other raw data is available on FigShare (DOI: 10.6084/m9.figshare.c.7106581).