If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Immune cells kill invading microbes by producing reactive oxygen and nitrogen species, primarily hydrogen peroxide (H2O2) and nitric oxide (NO). We previously found that NO inhibits catalases in Escherichia coli, stabilizing H2O2 around treated cells and promoting catastrophic chromosome fragmentation via continuous Fenton reactions generating hydroxyl radicals. Indeed, H2O2-alone treatment kills catalase-deficient (katEG) mutants similar to H2O2+NO treatment. However, the Fenton reaction, in addition to H2O2, requires Fe(II), which H2O2 excess instantly converts into Fenton-inert Fe(III). For continuous Fenton when H2O2 is stable, a supply of reduced iron becomes necessary. We show here that this supply is ensured by Fe(II) recruitment from ferritins and Fe(III) reduction by flavin reductase. Our observations also concur with NO-mediated respiration inhibition that drives Fe(III) reduction. We modeled this NO-mediated inhibition via inactivation of ndh and nuo respiratory enzymes responsible for the step of NADH oxidation, which results in increased NADH pools driving flavin reduction. We found that, like the katEG mutant, the ndh nuo double mutant is similarly sensitive to H2O2-alone and H2O2+NO treatments. Moreover, the quadruple katEG ndh nuo mutant lacking both catalases and efficient respiration was rapidly killed by H2O2-alone, but this killing was delayed by NO, rather than potentiated by it. Taken together, we conclude that NO boosts the levels of both H2O2 and Fe(II) Fenton reactants, making continuous hydroxyl-radical production feasible and resulting in irreparable oxidative damage to the chromosome.
Mechanisms of bacterial killing by our immune cells are complicated and continue to attract experimental attention. Production of reactive oxygen species (ROS) superoxide and H2O2 ("HP" in figures) by the phagocyte NAD(P)H oxidase (phox) and reactive nitrogen intermediates (RNIs) by inducible nitric oxide synthase are important for killing endocytosed bacteria (
). Mice deficient in both gp91phox and nitric oxide synthase are susceptible to spontaneous internal infections, demonstrating that ROS and RNI are important elements of a synergistic macrophage antimicrobial response (
). These multiple systems in E. coli ensure that acute H2O2 doses up to 5 mM are only bacteriostatic in E. coli, although H2O2 concentrations ≥10 mM apparently self-potentiate to cause chromosomal fragmentation and loss of viability even in this quite resistant bacterium (
). Generally, in cells with intact H2O2 scavenging, low millimolar H2O2 concentrations kill by mode-one; mode-one killing is blocked by iron chelators and affects mostly the DNA repair mutants, defining mode-one as IF-iron–dependent killing via DNA damage. When WT E. coli is killed by 10 mM H2O2, it is still mode-one (
). In contrast, higher concentrations of H2O2, starting with 20 to 25 mM in E. coli, kill by mode-two, which is insensitive of iron chelation, active metabolism, or of DNA repair capacity—defining mode-two as killing of unknown nature that does not depend on IF-iron or DNA damage. The additional distinction between the two H2O2 killing modes relevant for this work is that, while mode-one killing is potentiated by NO, mode-two killing is inhibited by NO treatment (
Perhaps the biggest paradox of H2O2 toxicity is that inside our immune cells, bacteria are presumed to be killed by H2O2 concentrations that are at least 1,000× lower than the H2O2 concentrations that start killing them in growing lab cultures. While direct electrochemical detection of ROS and RNI in phagosomes within macrophages with nanoelectrodes is still being developed (
) (Fig. 1A, left). We have previously demonstrated that the nature of this phenomenon is not simply redundancy of two toxic treatments overwhelming antitoxicity mechanisms of the cells but potentiation of H2O2 toxicity by NO to cause catastrophic chromosome fragmentation (CCF) (
). Two general strategies around Fenton chemistry, by which NO could enhance the intracellular toxicity of H2O2, are as follows (Fig. S1): (1) increase in available Fe(II) and (2) stabilization of the intracellular H2O2.
The first strategy is confirmed in principle by the effect of iron chelators that completely block potentiated H2O2 toxicity (
). Consequently, the enzyme flavin reductase (Fre) uses the increased NADH pool to reduce free flavins, which in turn reduce ferric iron to ferrous iron, which then becomes available for the Fenton reaction (
). Since, in the presence of excessive H2O2, Fenton reaction rapidly turns all IF-iron from Fe(II) to Fe(III), NO should be able to potentiate Fenton by cycling Fe(III) back to Fe(II), continuously feeding Fenton reaction with the disappearing key ingredient, reduced iron (Fig. S1).
In another “iron-centric” model, we previously proposed that part of the CN potentiation of H2O2 toxicity was iron recruitment from cellular iron depots (
). In contrast, our recent study of the H2O2+NO toxicity found that NO inhibits H2O2 scavenging, by binding and inhibiting the heme-containing catalases (Fig. S1), to stabilize effective concentrations of H2O2 inside the treated cells, causing lethal densities of double-strand DNA breaks (
). We also found that in a catalase-deficient mutant, H2O2-alone exerts mode-two killing at H2O2 concentrations static for WT cells, while NO treatment blocks unknown targets of mode-two killing, offering temporary protection from lethal H2O2 concentrations (
In this study, we wanted to determine the contribution of various intracellular targets of NO potentiation to the overall H2O2+NO killing. Is the catalase inhibition by NO and resulting H2O2 stability the cause of death or does it simply correlate with a concurrent respiration inhibition in WT cells? Another objective was to find additional cellular targets of NO, if they exist, that could potentiate H2O2 toxicity. Also, there could be mutants/conditions, in which NO would alleviate H2O2 lethality, as described for other organisms (
). In short, this article addresses how NO affects the "iron" side of Fenton's reactants (Fig. 1A, inset, Fig. S1).
Parameters of the H2O2+NO treatment
To study NO potentiation of H2O2 toxicity, we treat E. coli cultures, growing in LB8 (LB buffered with 50 mM Tris HCl pH = 8.0) with 2.5 mM H2O2 and 0.6 mM DEA NONOate (H2O2 + NO henceforth), two treatments that are bacteriostatic by themselves yet kill within minutes in combination, by precipitating CCF (Fig. 1A) (
), and here by an H2O2 electrode, 2.5 mM H2O2 is completely degraded by WT cells in 20 min (Fig. 1B) but could be stabilized by genetic inactivation of both catalases (Fig. 1C). The katE or katG single mutants also show reduced capacity to degrade H2O2 (Fig. 1C); however, unlike the sensitive katEG double mutant, both single mutants are resistant to H2O2-alone treatment (
). NO in our treatment is produced by decomposition of NO-donor DEA NONOate; its slower release because of the higher pH of our LB8 explains the 2 to 6 min rise to a concave plateau around the maximal concentration, followed by 10 to 15 min decline to the half-maximal concentration transitioning into a decomposition tail (Fig. 1E). The unexpectedly rapid decline is because NO gas readily escapes from aqueous solutions, unless they are in bioreactors (
). We characterize NO exposure of the cultures at any given concentration of DEA NONOate with two parameters derived from the evolution curves of Figure 1E: the half-maximal [NO] and the period during which [NO] was above the half-maximal concentration (Fig. 1F). For example, starting with 600 μM DEA NONOate, NO exposure is 24 μM x 20 min [this exposure inhibits catalases in Fig. 1D by ∼95% (and see below)], while starting with 60 μM DEA NONOate, NO exposure is reduced to 6 μM x 15 min, and the catalase inhibition is only ∼50% (see below). NONOate donors are known to yield actual NO concentrations much lower than the initial donor concentrations (
) and therefore, unlike the WT, should be killed by NO-alone treatment and should be additionally sensitive to H2O2+NO treatment, if peroxynitrite is indeed so toxic. Yet, the sodAB mutant shows no sensitivity to either the standard NO dose or even 5× higher (3 mM) NO-alone treatment (Fig. 1G). Additionally, the mutant is significantly less sensitive than WT to H2O2+NO treatment (compare Fig. 1Aversus G), which we also observed before for a similar H2O2+CN treatment (
). It seems as if peroxynitrite, even if it could form in the cytoplasm, does not pose the same threat as the H2O2+NO combination.
Finally, since we addressed the role of iron in this work, we used electron paramagnetic resonance (EPR) to measure the concentration of IF-iron in our cells (Figs. 1, H and I and S3). WT cells provide the baseline of ∼75 μM of IF-iron, which is higher than 20 to 30 μM found in E. coli grown in minimal media (
). At the same time, the fur mutant, deficient in the cytoplasmic iron regulation, accumulates up to 200 μM IF-iron (Fig. 1, H and I), which is also well within the range for this mutant when grown in LB (
). Therefore, if in a given mutant, the level of IF-iron is the same as in the WT before starting the treatment, it is safe to interpret the change in the mutant's sensitivity to H2O2+NO in terms of either additional iron recruitment/reduction (cycling) by NO or as NO-inhibition of catalases. For example, the levels of IF-iron remain roughly the same in the katEG mutant (Fig. 1, H and I), showing that catalase inactivation does not upset iron homeostasis in E. coli (unlike, for example, does the inactivation of superoxide dismutases (
), but is it sufficient (that is, the only way NO could act)? If stabilizing H2O2via catalase inactivation is the only route of NO potentiation of H2O2 toxicity, then manipulating the other Fenton reactant, the IF-iron, should have no effect on killing by H2O2+NO. The levels of IF-iron are increased in the fur mutant (
); in our growth conditions, fur mutants have ∼2.5× more IF-iron than WT cells do (Fig. 1I). This increased iron makes the fur mutant only slightly sensitive to H2O2-alone treatment, yet significantly more sensitive than WT to H2O2+NO treatment (Fig. 2A). Thus, having more IF-iron available for Fenton makes cells vulnerable to H2O2, regardless of the presence or absence of NO.
The increased IF-iron in the fur mutant also accelerates the rate of chromosome fragmentation by H2O2+NO, leading to close to background levels of the remaining chromosomal DNA by 20 min of the treatment (Fig. 2, B and C). In other words, intact chromosomal DNA all but disappears in the fur mutant treated with H2O2+NO. H2O2-alone treatment is even more interesting; coincident with the small drop in survival (Fig. 2A), there is a 25% loss of the chromosomal DNA in the first 10 min, which is then followed by a surprising resumption of label incorporation after 20 min (Fig. 2, B and C). This fast ‘DNA recovery’ could be due to a faster scavenging of H2O2 in the fur mutant; however, contrary to this expectation, we found that the fur mutant actually scavenges H2O2 slower than WT cells (Fig. 2D), so that ∼1/3 of the original amount still remains after 20 min, at the time when essentially no H2O2 is detected in the medium of WT cells. A H2O2 scavenging defect has been reported previously in a fur mutant, wherein the sensitivity to H2O2 was attributed to low catalase activity rather than iron overload (
). Therefore, the nature of the resumption in chromosomal label incorporation in H2O2-alone–treated fur mutant cells, in spite of the higher [H2O2], remains unclear and likely reflects the significant changes in expression profile in this mutant (
) accelerating restart of a process inactivated by oxidative stress.
The ftnA and fre defects alleviate H2O2+NO toxicity, while dps and bfr defects potentiate it
Since having more iron in the cytoplasm sensitizes cells to H2O2 exposure, we tested more mutants with defects in iron handling. Defects in the iron depot proteins ferritins FtnA and Dps do not affect the level of IF-iron in a statistically significant way (Fig. 2E). However, regular ferritins (homologs of FtnA of E. coli) are known to release iron, if stimulated by chemicals like NO or CN (
). While ftnA and bfr mutants are not sensitive to H2O2-alone and dps mutant is only slightly sensitive, they do show differences from WT in their sensitivity to H2O2+NO (Fig. 2F). In particular, the ftnA inactivation alleviates H2O2+NO lethality (Fig. 2F), suggesting that FtnA depots provide a source of iron for the Fenton chemistry inside the cell. Contrary to this general idea and at the same time confirming the previous report (Woodmansee and Imlay 2003), we detected less IF-iron in WT cells treated with NO (Fig. 1, H and I). It could be that, in the presence of NO, the ferritin-released iron is immediately bound by DNA or by other big molecules, like stable RNA, to become undetectable by EPR. We have observed something like this before in vitro with CN-complexed IF-iron, when the added plasmid DNA effectively sequestered iron away from CN complexes (
In contrast to the ferritin deficiency, both the bfr and dps inactivation aggravate sensitivity to H2O2+NO treatment (Fig. 2F). Moreover, the double bfr dps mutant shows extreme sensitivity (Fig. 2F), suggesting that Dps and Bfr ferritins are redundant, and both are effective in sequestering iron when H2O2 is around. Thus, the H2O2+NO sensitivity of iron depot mutant confirms that in addition to H2O2 stabilization, NO potentiation does have a significant "iron dimension" to it, either releasing additional iron from iron depots FtnA or competing with IF-iron sequestration into Dps and Bfr.
Increased or decreased H2O2+NO sensitivity of ferritin mutants suggests a source of additional iron (FtnA) for promoting Fenton, but what is the mechanism of iron recruitment and keeping it reduced in the presence of H2O2? To state the problem differently, IF-iron will catalyze Fenton upon encounter with H2O2, but does it take multiple rounds of redox iron cycling to damage the chromosome beyond repair? To distinguish between the limited versus continuous Fenton, we inactivated the main Fe(III) → Fe(II) reduction enzyme, Fre. As the major siderophore reductase of E. coli, Fre catalyzes Fe(III) reduction to Fe(II) during iron acquisition from the environment (
). However, this metabolic activity becomes harmful during H2O2 exposure: by reducing the oxidized Fe(III) back to Fe(II), Fre keeps Fenton reaction going as long as reducing equivalents and H2O2 are available (
). The fre mutant has the same level of IF-iron as WT cells (Fig. 2E), indicating that (1) its decreased sensitivity to H2O2+NO is not due to a reduced levels of IF-iron and (2) no matter what the contribution of iron recruitment, continuous Fe(III) reduction back to Fenton-reactive Fe(II) is important for H2O2 toxicity in WT cells. Moreover, iron reduction by Fre seems to work in the same pathway as iron release from FtnA, as the H2O2+NO survival of the double ftnA fre mutant is not significantly different from those of single mutants (Fig. 2G). Overall, we conclude that the NO potentiation of the iron side of Fenton reaction significantly contributes to the overall lethality of the double treatment (Fig. 2H), whereas its complexity warrants further exploration, especially in comparison with the relatively straightforward NO stabilization of H2O2via inhibition of catalases.
Decreased iron reduction delays H2O2 toxicity in the katEG mutants
To probe the role of iron in the NO potentiation of H2O2 killing, H2O2-alone and H2O2+NO–treated cultures need to be compared. However, the two treatments are different in an important aspect: H2O2-alone is rapidly scavenged by catalases (Fig. 1B), while in the combined treatment, H2O2 is stable due to catalase inhibition by NO (Fig. 1D). To make H2O2-alone and H2O2+NO treatments comparable, we used the katEG catalase-deficient background to ensure stability of H2O2 concentrations for the duration of experiment (Fig. 1C). Indeed, the katEG double mutant is equally sensitive to H2O2-alone and H2O2+NO treatments (
); at the same time, since the katEG mutant has IF-iron levels comparable to that of WT cells (Fig. 1H), this sensitivity to H2O2-alone can be mostly attributed to the unscavenged H2O2.
Our first question was whether the IF-iron levels by themselves are lethal when H2O2 is stable. To address it, we measured H2O2-alone and H2O2+NO sensitivity of the katEG fre mutant, whose catalase deficiency makes H2O2 stable (Fig. 3A), while the fre defect in iron reduction should restrict repeated cycles of Fenton. When confronted with H2O2-alone, the katEG fre triple mutant showed a hybrid sensitivity pattern: until 30 min of the treatment, the mutant was almost as resistant as the completely resistant fre mutant, but by 45 min, it became almost as sensitive as the katEG mutant (Fig. 3B). Since H2O2 levels remain constant in the katEG fre mutant (Fig. 3A), the initial resistance of this mutant compared to the katEG (Fre+) strain must be due to the absence of iron cycling. Then, the eventual H2O2-alone toxicity in the katEG fre mutant must have a different nature—for example, because of a sudden availability of Fe(II) from an earlier unavailable source or a switch to a different mode of killing. In fact, the pattern of complete initial resistance with an eventual deep killing was reminiscent of the delayed mode-two killing of WT cells by 25 mM H2O2 in the presence of deferoxamine (DF)+NO (
These two possibilities could be distinguished by chelating iron: additional iron recruitment should be blocked by DF, while mode-two killing should be insensitive to DF. DF addition rescued, albeit partially, the late sensitivity of the katEG fre mutant to H2O2-alone (Fig. 3C), meaning that both explanations apply to the late toxicity of H2O2 in the katEG fre mutant. Another evidence consistent with mode-two killing comes from chromosomal fragmentation: in contrast to the katEG mutant, where H2O2-alone induces CCF, fragmentation in the katEG fre mutant is much reduced (Fig. 3, D and E). Between 10 and 30 min, there is a loss of 20% intact chromosomal DNA, but no associated loss in viability. However, between 30 and 45 min, another 20% loss in chromosomal DNA results in loss in viability by three orders of magnitude. Thus, the killing of the katEG fre mutant by H2O2-alone cannot be explained by only double-strand DNA breaks.
Similar to the fre single mutant, which is partially resistant to the H2O2+NO treatment, the katEG fre triple mutant is initially resistant to H2O2+NO, but after 30 min again loses viability to reach survival titers similar to the katEG mutant (Fig. 3F). The same timing of viability loss in the katEG fre mutant during H2O2-alone (Fig. 3B) versus H2O2+NO (Fig. 3F) treatments indicates a significant metabolic switch in this mutant after 30 min of H2O2 exposure. In general, the overall similarity of the two sensitivity patterns means no further NO potentiation in the katEG fre triple mutant. This not only confirms that catalases are targets of NO inhibition but also implies that Fre controls an independent NO-potentiation pathway. Iron chelators completely save the katEG fre mutant from H2O2+NO treatment (Fig. 3C), again showing that NO blocks an unknown target of mode-two killing by H2O2. The chromosomal fragmentation and DNA disappearance during H2O2+NO treatment is again reduced in the katEG fre mutant relative to the katEG mutant (Fig. 3, G and H). Moreover, at 45 min there is again no significant increase in either fragmentation or DNA loss to explain the precipitous loss in viability at this late time point (Fig. 3F).
Thus, iron reduction offers at least two candidate activities, FtnA and Fre, for NO potentiation of H2O2 toxicity (Fig. 3I), as illustrated by the fact that minimizing iron reduction with the fre defect decreases drastically the density of double-strand breaks in the katEG fre mutant cells, saving them during the first 30 min of H2O2-alone or H2O2+NO treatments. In addition, the two-fold reduction of IF-iron in the NO-treated cells (Fig. 1, H and I) further elevates the importance of procurement of Fe(II) for H2O2+NO killing.
Conditions for potentiation of H2O2 toxicity in the katEG mutants
In principle, NO pathways to potentiate H2O2 toxicity other than catalase inhibition could be revealed if lower bacteriostatic H2O2 concentrations for the katEG mutant could be again potentiated by NO. However, the static H2O2 concentrations for the katEG mutant in the range of 0.25 to 0.5 mM are not potentiated with 0.6 mM NO (Fig. S4A), implying that catalases are the only NO targets in our standard treatment conditions. Nevertheless, we found that lower NO concentrations, in the range of 0.06 to 0.15 mM, do sensitize the katEG mutant to 0.5 mM H2O2 treatment (Fig. S4, B and C), indicating existence of secondary NO targets. Moreover, blocking Fe(III)→Fe(II) reduction in the katEG mutant by the fre defect all but eliminates the sensitivity of the triple mutant to this milder 0.5 H2O2 + 0.06 NO treatment (Fig. S4C), suggesting that the potentiation of the secondary NO targets is still via Fe(II) generation fueling multiple Fenton cycles. In fact, in the previous paper, we reported that, in contrast to H2O2-alone treatment, H2O2+NO-treated katEG mutant is completely saved by iron chelation with DF (
). Potentiation by lower NO concentrations, but not by higher ones, show that in the katEG mutants, higher concentrations of NO inhibit, rather than potentiate, H2O2 toxicity, perhaps because excess NO acts as an iron chelator (Fig. 1, H and I).
Although we did find new NO-potentiating conditions for the katEG mutant (Fig. S4B), we decided to investigate the role of iron reduction in the chronic H2O2 toxicity with a gene candidate approach instead, using the standard conditions (2.5 mM H2O2 + 0.6 mM DEA NONOate), in order to align our readouts with previous results and to be able to also use catalase-proficient strains. Our objective was to verify the general metabolic process powering the FtnA/Fre-dependent IF-iron source that sustains Fenton in NO-treated cells. There was a strong candidate, as both iron recruitment from ferritins (
), specifically the heme-containing ubiquinol oxidases Cyo, Cyd, and App (Fig. 4A). The resulting NO-mediated respiratory block leads to accumulation of NADH, which in stably growing cells comprises a few percent of the total NAD pool (
), and this NADH excess facilitates reduction of FAD to FADH2 exactly by the same Fre, that also happens to reduce Fe(III) to the Fenton-catalyzing Fe(II), using reduced flavins as electron donors (Fig. 4A) (
). Indeed, we detected a 3-fold increase in the absolute intracellular NADH concentrations in WT cells upon NO treatment (Fig. 4B), which translates into 4.5× increase of NADH fraction in the total NAD pools (Fig. S5). In the in vitro respiration assay, while 2.5 mM H2O2 does not affect NADH oxidation by inverted membrane vesicles, 600 μM DEA NONOate completely inhibits NADH oxidation (= respiration) (Fig. 4C). Respiration in vivo is measured as cellular oxygen consumption (depletion of oxygen from the chamber containing the oxygen electrode), where plunging oxygen levels indicate normal respiration (Fig. 4D, the black curve), while stable oxygen levels indicate inhibited respiration. Respiration in WT E. coli is expectedly inhibited by 3 mM CN and, unexpectedly and transiently, by a mere 0.25 mM H2O2 (Fig. 4D), the latter being a peculiar artefact of oxygen production by the catalase reaction (Fig. S6A). Indeed, respiration is unaffected by 0.25 mM H2O2 in the katEG mutant, even though higher H2O2 concentrations start inhibiting it in this mutant (Fig. S6B), probably reflecting mode-two killing (
). This might also explain why the katEG fre mutant shows a period of resistance to H2O2-alone, as H2O2 mimics NO’s respiration inhibition in the katEG mutants. Although we could not challenge in vivo respiration with 600 μM DEO NONOate due to technical issues, we found that even 60 μM DEO NONOate inhibits this in vivo respiration completely (Fig. 4D), showing exquisite sensitivity of ubiquinol oxidases to NO (we will return to this point later).
As already mentioned, there are three ubiquinol oxidases in E. coli, Cyo, Cyd, and App, active during aerobic/oxic respiration (Fig. 4A). We found that eliminating any one of the three individual ubiquinol oxidases has no phenotype, in that the cyoB, or cydB, or appC single mutants are completely resistant to H2O2-alone treatment, while showing WT-like sensitivity to the H2O2+NO treatment (Fig. 4E). However, the double mutants in any two of the three cytochrome oxidases made the mutant growth impractically slow, so we decided to block the electron transport chain one step earlier, at the level of NADH dehydrogenases (NDH), Nuo and Ndh (Fig. 4A). Single ndh or nuo mutants again showed WT resistance to H2O2-alone and similar to WT sensitivity to H2O2+NO treatment (Fig. 4F) suggesting redundancy of the two enzymes under our growth conditions. It was the mutant lacking both NDHs that should be unable to respire and was expected to accumulate NADH, boosting the Fe(III)→Fe(II) reduction and thus sensitizing cells to H2O2. The ndh nuo double mutant is viable, even though slow-growing (Fig. 5A); the mutant indeed accumulates ∼10 times more NADH compared to the WT cells, and this NADH level does not further respond to the NO treatment (Figs. 5B and S5), suggesting that it is already at the maximum. If the NADH level is a critical indicator, the ndh nuo mutant should show significantly stronger H2O2 toxicity effects than those induced in WT cells by NO treatment.
The ndh nuo mutants are extremely sensitive to H2O2-alone
While WT cells consume O2 within 8 min, the ndh nuo mutant utilizes little O2 to respire aerobically, similar to the control respiration-inhibited WT cells treated with 3 mM CN (Fig. 5C). The IF-iron is increased about 2-fold in the ndh nuo mutant, to the levels between the WT and the fur mutant (Fig. 5D). The ndh nuo double mutant was extremely sensitive to H2O2-alone, dying ∼3 times faster than WT cells during H2O2+NO treatment, but eventually to the same final level of survival of ∼10−3 (Fig. 5E). The sudden plateauing of the survival by 10 min is because the ndh nuo mutant scavenges H2O2 faster than WT cells, within 3 min (Fig. 5F). The only catalase-proficient mutants known to be (slightly) sensitive to H2O2-alone were fur (Fig. 2A) and sodAB (Fig. 1G) (
), so the higher IF-iron in the ndh nuo mutant supports this correlation. However, a major part of its high sensitivity to H2O2-alone is likely due to the higher NADH levels (Fig. 5B) promoting iron reduction.
Addition of NO exacerbated the H2O2 lethality of ndh nuo double after 20 min of treatment so that, by 45 min, survival became 10-5 (Fig. 5E). At the same time, in the first 20 min, NO slowed down H2O2 killing of the ndh nuo double, although the rate was still faster than in the WT cells (Fig. 5E). These two opposite effects created a peculiar "hybrid" sensitivity curve, with "NO-protection" during the first 20 min switching to NO potentiation after 20 min (Fig. 5E). Since the ndh nuo double mutant is killed by both H2O2-alone and H2O2+NO treatments, this generally confirms the previous idea (
) that block of the respiratory chain is a way NO could potentiate H2O2 toxicity. However, the shapes of H2O2-alone versus H2O2+NO sensitivity curves in the ndh nuo mutant are sufficiently different (Fig. 5E) to suggest that NO changes the nature of H2O2 toxicity in the ndh nuo mutant, probably by complexing IF-iron, as reflected by its decrease in NO-treated cells (Fig. 1, H and I).
While 2.5 mM H2O2-alone fails to cause fragmentation in WT cells (
), it induces speedy CCF in the ndh nuo mutant (Fig. 5, G and H); moreover, the kinetics of CCF induced in the ndh nuo mutant by H2O2+NO appears slower (Fig. 5, G and H), again suggesting that NO actually retards the precipitous H2O2-alone toxicity in this mutant. Both the killing by H2O2-alone and by H2O2+NO in the ndh nuo mutant are blocked by iron chelation with DF (Fig. 5I), indicating exclusively (iron-dependent, chromosomal DNA-targeting) mode-one killing. The blocking effect of DF confirms that respiration inhibition poisons cells via increased iron reduction.
Limiting iron reduction with the fre defect makes the ndh nuo fre triple mutant significantly less sensitive to the H2O2-alone killing (Fig. S7), demonstrating that a significant part of the ndh nuo mutant sensitivity to H2O2-alone is due to iron cycling, rather than simply due to the elevated IF-iron. At the same time, ndh nuo fre mutant’s IF-iron levels are similarly elevated (Fig. 5D), implying that the residual sensitivity of these mutants to H2O2-alone is primarily due to the higher initial IF-iron levels. In short, our investigation reveals some correlation between initial IF-iron levels and sensitivity to H2O2 but also demonstrates that the rate of iron reduction and the H2O2 stability are more important for the in vivo effects of Fenton chemistry.
Interaction between the ndh nuo and katEG defects
Thus, we have characterized two double mutants for their chromosome fragmentation and viability: (1) the catalase-deficient katEG mutant (
) and (2) the NDH–deficient ndh nuo mutant (Fig. 5). Both combinations render E. coli cells sensitive to H2O2-alone treatments, but for opposite reasons in terms of the Fenton reactants: H2O2 stability in katEG versus Fe(II) flow in ndh nuo. In order to genetically test whether the ndh nuo versus the katEG defects sensitize cells to H2O2-alone by distinct pathways or the same pathway, we constructed a katEG ndh nuo quadruple mutant, in which increased iron cycling is combined with H2O2 stability. The quadruple mutant was not sensitive to NO-alone, showed a somewhat higher sensitivity to H2O2+NO, and was exquisitely sensitive to H2O2-alone treatment (Fig. 6A). This sensitivity of the quadruple mutant reflected the extreme rates of chromosome fragmentation after either treatment (Fig. 6, B and C). Iron chelation with DF completely suppressed the sensitivity of the quadruple mutant to both H2O2-alone or H2O2+NO (Fig. 6D), as well as the CCF induced by both treatments (Fig. 6, E and F), indicating that this fast killing is strictly mode-one. In other words, the unknown targets of H2O2 mode-two killing in the katEG mutant (
Comparison of the kinetics of H2O2-alone and H2O2+NO sensitivity curves of the four strains, WT, katEG, ndh nuo, and katEG ndh nuo, proved insightful. For the H2O2-alone treatment, to which WT cells were completely resistant, the effect of two pairs of H2O2-sensitizing mutations turned out to be additive. As a result, the quadruple mutant responded to the H2O2-alone treatment with a composite sensitivity curve, in which the early (fast) killing aspect of the ndh nuo mutant was combined with the late (deep) killing effect of the katEG mutant (Fig. 6G). Apparently, the early killing was due to the increased iron cycling and IF-iron levels (reflecting the ndh nuo defect), while the continuous later killing was due to the stability of H2O2 (reflecting the katEG defect). We conclude that ndh nuo and the katEG represent two independent pathways protecting E. coli against acute H2O2 toxicity.
This point was further corroborated by H2O2+NO treatment of the katEG ndh nuo quadruple mutant; compared with H2O2-alone, we observed ∼2.5-fold slower rate of killing (Fig. 6A), the effect that we have already observed with ndh nuo double mutant (Fig. 5B). In other words, even though NO potentiates H2O2 toxicity in WT cells, it clearly protects some H2O2-hypersensitive mutants from the same H2O2 concentrations. As a result of this "NO-buffering", in contrast to the different kinetics of killing of WT, katEG, ndh nuo, and katEG ndh nuo strains with H2O2-alone (Fig. 6G), the same four strains treated with H2O2+NO show, somewhat counterintuitively, similar initial rates ending with depth of killing differences within two orders of magnitude (Fig. 6H). We conclude that the nature of H2O2 toxicity, though masked by cellular resistance mechanisms in WT cells, is complex (mode-one + mode-two), while NO potentiation amplifies H2O2 toxicity even in WT cells, but also makes it mechanistically simpler, converting it to a slower mode-one in all mutant combinations (Fig. 6I).
Catalases and the respiratory chain are the major targets of NO in E. coli
In order to reveal the remaining pathways of NO potentiation of H2O2 toxicity, if any, we treated the katEG ndh nuo quadruple mutant with varying concentrations of H2O2-alone, looking for residual NO targets that potentiate H2O2 toxicity with two DEA NONOate concentrations, 0.06 mM and 0.6 mM (Fig. 7D). As controls, we tested WT, katEG double mutant, and ndh nuo double mutant with the same treatments as the quadruple mutant (Fig. 7, A–C). As shown previously (
), WT cells are not sensitive to any H2O2-alone concentration in this range, from 0.2 to 2.5 mM, while gradually increasing potentiation with NO from low [H2O2] to high [H2O2] and with higher [NO] potentiating better (Fig. 7A). In contrast, in the katEG mutant, 0.75 mM, 0.5 mM, and 0.2 mM H2O2 can be potentiated by 0.06 mM NO (corroborating Fig. S4 results), while the higher 0.6 mM NO concentration has no effect at these H2O2 concentrations and even starts protecting katEG mutants at the highest H2O2 concentrations (Fig. 7B). This was earlier interpreted to mean mode-two killing in the katEG mutant, with NO switching it to mode-one, slower killing (Fig. 6I) (
). In the ndh nuo mutant (Fig. 7C), H2O2 was lethal by itself and was further potentiated by both 0.6 mM and 0.06 mM NO, similar to WT cells (compared to Fig. 7A). Finally, in the katEG ndh nuo mutant, the lethality of the three intermediate H2O2 concentrations was reduced by both NO concentrations (Fig. 7D), suggesting no more targets for H2O2 potentiation by NO in this mutant of E. coli.
The lack of NO potentiation of the higher [H2O2] in the katEG mutant (Fig. 7B), the remaining ability to potentiate these [H2O2] with NO in the ndh nuo mutant (Fig. 7C) and finally, the disappearance of this potentiation with the removal of catalases in the katEG ndh nuo mutant (Fig. 7D) suggest that catalase inhibition by NO is a major cause of lethality during the 2.5 mM H2O2 + 0.6 mM DEO NONOate treatment. We suspected that this was because respiration, the other NO potentiation route, was more sensitive to lower NO concentrations that were still not enough to potentiate H2O2 toxicity. To test this idea, we measured respiration inhibition, both in vitro (Fig. 7E) and in vivo (Fig. 7F), by various low concentrations of NO and found that both processes are completely inhibited by 30 μM DEA NONOate (Fig. 7, E and F), the concentration that inhibits catalases by less than 50% and causes ∼one order of magnitude of H2O2 killing (
In other words, [NO]-dependence of respiration inhibition does not explain continued H2O2 killing at NO concentrations higher than 30 μM. To illustrate this, we plotted inhibition of respiration versus H2O2 scavenging at various concentrations of NO with the corresponding killing curve by H2O2+NO (again at various concentrations of the latter) (Fig. 7G). In general, the shapes of the two curves were different enough to conclude that respiration inhibition cannot be the main reason for H2O2+NO lethality at the millimolar H2O2 concentrations used in this study. In particular, 15 μM of DEA NONOate was sufficient to inhibit 90% of respiration in vivo, while this NO exposure still was not enough to cause lethality in H2O2-treated cells (Fig. 7G). At the same time, as shown previously, the shape of H2O2 scavenging inhibition by various [NO] coincided with the H2O2+NO killing curve (Fig. 7G), strongly suggesting that it is the former that drives the latter (
) that NO potentiates toxicity of otherwise static concentrations of H2O2, so the latter becomes lethal for WT E. coli by continually inducing multiple double-strand DNA breaks, which lead to catastrophic fragmentation of the chromosome the cells cannot repair (
). We also showed that a major NO potentiation pathway of H2O2 toxicity is by inhibition of heme-containing catalases, making H2O2 levels stable and thus enabling continuous Fenton chemistry with IF-iron. Since the katEG catalase–deficient mutant was equally sensitive to H2O2+NO and H2O2-alone treatments, catalase inhibition appeared to provide adequate explanation for NO potentiation (
). However, this answer could not be complete, as Fenton, in addition to H2O2, also requires Fe(II), yet H2O2 entry into the cytoplasm will (presumably) instantaneously oxidize all IF-Fe(II) to Fe(III), self-limiting Fenton's damage. In addition, Dps mini-ferritin will use H2O2 to sequester all remaining IF-iron; however, these two challenges notwithstanding, NO somehow ensures a flow of Fe(II) even in the presence of H2O2, making Fenton continuous.
This work explored the nature of this continuous source of Fe(II) in NO-treated cells. We found that a potential system maintaining the pool of reduced iron in H2O2-treated cells are the iron depots ferritins (FtnA), along with the previously identified Fre, and continuous free flavin reduction supported by the increased pools of NADH, resulting from the respiration block by NO. We modeled NO-inhibition of ubiquinol oxidases by genetically blocking the preceding step of NADH oxidation in the ndh nuo double mutant, deficient in NDH activity. The ndh nuo mutant not only accumulates NADH but also has increased IF-iron and is killed by H2O2-alone even faster than by H2O2+NO, confirming respiration inhibition as another route of NO potentiation of H2O2 toxicity. The katEG ndh nuo quadruple mutant, that keeps both the H2O2 and Fe(II) levels high, shows a remarkable sensitivity to H2O2-alone and instead of being potentiated by NO is actually saved by it, demonstrating that NO potentiation pathways are exhausted in the mutant. Finally, we show that respiration is inhibited at low NO concentrations, at which little lethality in the H2O2+NO treatments is observed. In contrast, lethality correlates well with catalase inhibition, which happens gradually and over a range of higher NO concentrations, elaborating our previous conclusion about H2O2 stability by complementing it with the nature of a continuous source of Fe(II).
Progress since previous studies
Imlay and Linn had shown some time ago that CN makes static concentrations of H2O2 lethal (
). Based on their previous work, Woodmansee and Imlay argued that CN has no effect on the intracellular H2O2 concentrations; they also showed that CN increases the IF-iron only two times and therefore argued that CN makes E. coli sensitive to low mM H2O2via inhibition of respiration, by producing electron donor that drives the Fenton reaction (
). Using semiquantitative PCR, they reported NO enhancement of in vivo DNA damage by H2O2. They also tested in vitro whether accumulation of NADH is directly responsible for Fe(III) reduction and had to reject this idea; eventually they found that the proximal Fe(III) reductant is a free flavin, FADH2, produced by Fre, in the reaction driven by NADH (
). In particular, they documented inactivation of Fe-S cluster enzymes, but no iron release from them; in fact, in their study, NO-alone treatment reduced IF-iron in the treated cells in half, just like in our case (Fig. 1, H and I). They showed that cyo cyd double mutant (the one with significant growth defects to be usable under our conditions) is extremely sensitive to H2O2-alone treatment and that NO fails to increase this sensitivity further (
) and the current one complement and extend their findings in several ways: (1) by demonstrating that catalases are also targets of NO inhibition, and their inactivation guarantees a stable presence of H2O2 for continuous Fenton; (2) by proposing that both CN and NO recruit additional iron from FtnA ferritin, via the same Fre-driven reduction dependent on the elevated NADH levels, while mini-ferritin Dps and bacterioferritin Bfr sequester iron during H2O2 treatment; (3) our ndh nuo double mutant behaves similar to their cyo cyd double mutant (extreme sensitivity to H2O2-alone, with no additional sensitization by NO), confirming the importance of respiratory chain in providing reduced iron for Fenton; (4) that no additional NO targets, beyond catalases and respiratory terminal oxidases, contribute to NO potentiation of H2O2 toxicity; and (5) last but not least, that DNA damage during the H2O2+NO or H2O2+CN treatments takes the form of CCF (breaking the chromosome into at least 100 pieces), explaining why the cells cannot repair it.
Separately, the H2O2+NO sensitivity of the bfr mutant, especially in the bfr dps double mutant combination, appears to be the first time that E. coli bfr mutant shows any phenotype, in contrast to the strong iron-accumulation defects of the bfr mutants in Pseudomonas, for example (
). One of the functions of bacterioferritin in E. coli, therefore, is to sequester IF-iron in conditions of oxidative stress—similar to the Dps function in E. coli or to the Bfr function in anaerobe Desufovibrio (
Preexisting Fe(II) IF-iron levels versus Fe(III) reduction during the treatment
The deep, CCF-based lethality of oxidative damage is remarkable if we consider that its nature is based on H2O2 simply rising to certain concentrations in the cytoplasm of the affected cells and interacting with IF-iron to produce a burst of hydroxyl radicals. Indeed (1) the known low concentration of (presumably) dispersed IF-iron should produce enough OH· for only limited damage to DNA, because the oxidative impact will be similarly dispersed around the cytoplasm; (2) the resulting DNA damage should be all single-stranded (mostly nicks, but also some base lesions) (
); (3) the time course of the DNA damage should be brief, restricted to the first few minutes of the treatment. Basically, Fenton driven only by IF-iron should be self-limiting. Some chromosomal damage is demonstrably driven by IF-iron levels during the first few minutes of H2O2-alone treatment, as illustrated by the initial dip in viability and detectable fragmentation after H2O2-alone treatment of the fur and sodAB mutants (Figs. 1G, 2, A and C) (
), which have higher IF-iron concentrations. However, if IF-iron level was the most important factor, then the ndh nuo mutant, with its IF-iron level lower than that of fur mutant (Fig. 5D), would show better survival and less fragmentation after H2O2-alone treatment. In contrast to this expectation, H2O2-alone treatment kills the ndh nuo mutant fast and deep (Fig. 5E), apparently because of the rapid fragmentation of its chromosomal DNA (Fig. 5, G and H). Thus, DNA damage is mostly driven by a reason other than the preexisting levels of Fe(II) IF-iron.
This reason, apparently, is continuous Fe(III) reduction to Fe(II), supported by high levels of NADH in the ndh nuo mutants and the elevated NADH levels in the NO-treated cells (Fig. 5B). This continuous source of Fe(II) Fenton reactant explains not only the observed massive DNA damage but also why Fenton in H2O2+NO-treated cells is not self-limiting. Indeed, in the H2O2+NO-treated cells, fragmentation still continues 1 hour later, indicating that it depends not only on the stability of H2O2 in the presence of NO but also on the continuous source of Fe(II). An additional evidence for the importance of iron cycling over IF-iron levels is offered by the ndh nuo fre mutant, which has the same levels of IF-iron as its ndh nuo progenitor (Fig. 5D) but is much less sensitive to H2O2-alone treatment (Fig. S7), because of the fre defect in Fe(III) reduction.
Accumulation of NADH in the ndh nuo mutant suggests that rapid NADH oxidation by the electron transport chain in WT cells provides an effective shield against oxidative damage, while its inactivation puts Fenton chemistry in overdrive. It is also interesting to note that genetic inactivation of the ndh nuo pathway does more than to simply phenocopy NO-treatment. This is apparent for both WT background (compare WT H2O2+NO versus ndh nuo H2O2-alone in Fig. 5E) and in the katEG background (compare katEG H2O2+NO of Fig. 6Hversus katEG ndh nuo H2O2-alone of Fig. 6G). The obvious explanation for the differences is that NO (at least at 0.6 mM DEA NONOate) does not cause the same level of NADH accumulation as the ndh nuo inactivation (Fig. 5B)—and thus, the lower expected level of Fe(III) reduction. Perhaps there are minor ubiquinol oxidases (not inhibited by NO?) in E. coli yet to be characterized?
The IF-iron versus Fenton-active iron
Since Fre and ferritin FtnA are both important for the lethality of the H2O2 + NO treatment in WT cells and since Fe(III), because of its higher charge, was expected to form complexes tighter than Fe(II) with big molecules like DNA (
). Therefore, in addition to iron release from ferritins and reduction by Fre, NO promotes iron removal from IF-iron pool, a complexing of a kind, that also enhances DNA damage. What is the nature of this removal?
The only way to increase DNA damage from Fenton without dramatically increasing the overall Fenton in the cell would be to run Fenton in the vicinity of DNA, preferably on DNA itself. Therefore, it was proposed that NO not only induces release of iron from ferritins and promotes its reduction by Fre but also recruits this iron to DNA, both removing it from IF-iron pool and increasing the DNA-damaging potential of subsequent Fenton (
). We have observed a similar scenario with CN in vitro; although CN forms stable complexes with free iron, when plasmid DNA is added, the iron–CN still binds this DNA, causing plasmid nicking in the presence of H2O2 (
). It would be interesting to repeat these in vitro experiments with NO.
Protection by NO
Not only does NO potentiate H2O2 toxicity but also its mode of action reverses to protection against H2O2 toxicity under certain conditions. For example, others showed that NO protected B. subtilis from H2O2 by limiting Fenton and recharging catalase (
). As explained in the introduction, there are two distinct modes of H2O2 toxicity, DNA-targeting iron-dependent mode-one versus iron-independent mode-two with unknown target. We observed that while iron chelators cannot save katEG mutants from mode II toxicity of H2O2, there is complete survival with the same H2O2 treatment when NO is additionally present (
). In other words, NO can function as an iron chelator and, in effect, helps other chelators to shield IF-iron from H2O2. In contrast to the katEG mutant, killed by mode-two with 2.5 mM H2O2, the same H2O2 concentration kills the katEG ndh nuo mutant by mode-one (Fig. 6D), implying the mode-two target is gone in the absence of NDHs. While the katEG mutants grow using aerobic respiration (Fig. 5C), the NDH mutants, ndh nuo and katEG ndh nuo, do not respire (Fig. 5C) and likely grow fermentatively. Thus, NO mimics the ndh nuo mutations and inhibits the fast mode-two killing, apparently by binding and protecting an undetermined target in the respiratory chain.
It isn’t only in the presence of chelators that NO shows its defensive side. Since NO targets both catalases and respiration, it could be expected conservatively that the loss in viability in the katEG ndh nuo mutant with H2O2 will be similar to that observed with H2O2+NO in WT cells (compare Fig. 7, A and D). However, the H2O2 lethality in the mutant is much higher than the H2O2+NO lethality in WT, revealing protective effects of NO in the WT cells. This is observed more clearly in the katEG ndh nuo survival of H2O2 challenge, where NO slows down cell killing considerably (Fig. 7, A and D). In general, NO's effects in the cell vary, explaining contrasting effects of its combination with H2O2 in various mutants. We posit the two general ways NO could protect against H2O2 toxicity: (i) NO blocks a mode II target related to aerobic respiration (for example by binding heme iron) and (ii) NO acts as a general weak iron chelator.
Since phagosomes produce H2O2 indirectly, via superoxide (
). However, since it is the protonated form of peroxynitrite that preferentially penetrates the bacterial cell envelope (while the charged ONOO– has to use anion channels) and with its pKa close to neutral, peroxynitrite becomes really poisonous for E. coli at pH significantly higher than physiological ones (
). Moreover, acute peroxynitrite treatment of E. coli fails to induce the SOS response, suggesting no significant DNA damage but instead induces transcriptional responses pointing to protein nitration and nitrosylation as the main cytoplasmic impact (
). Finally, the idea that H2O2+NO treatments works via generating peroxynitrite around bacterial cells is inconsistent with the lack of protection against the treatment by bicarbonate, which completely protects against bona fide peroxynitrite (
In our experimental system, peroxynitrite contribution to the overall H2O2+NO toxicity could be only minor, for the following reasons: (1) were peroxynitrite a major contributor, the superoxide dismutase-deficient sodAB mutant would be more sensitive to H2O2+NO treatment, but in fact it is more resistant than WT (Fig. 1G); (2) peroxynitrite is toxic independently of iron (
); (3) if inhibition of respiration by NO indeed generated enough superoxide, then NO-alone treatment via formation of peroxynitrite inside cells would at least affect WT cells and would kill the sodAB mutants—but it does not (Fig. 1, A and G). Further experiments are needed to clarify any potential role of peroxynitrite formation in NO-potentiated H2O2 toxicity and its underlying chromosome fragmentation.
NO potentiates the intracellular Fenton reaction, causing lethality via CCF. NO potentiation has two major routes, and both occur via its binding to heme-containing enzymes: (i) inhibition of catalases to make H2O2 stable and (ii) inhibition of respiration to boost iron recruitment and reduction in the presence of H2O2. In the future, it would be important to develop conditions with similar effects but utilizing more physiological low micromolar concentrations of H2O2 and NO. Due to its polyanionic nature, DNA binds iron avidly, creating a natural platform for Fenton chemistry. The resulting hydroxyl radicals should induce singly damaged sites including nicks in DNA, but their relationship to double-strand DNA breaks that fragment the chromosome is still unclear. Finally, it would be interesting to explore the interactions between ferritins and DNA in the presence of H2O2 and NO in vitro using plasmid-nicking assays.
Strains and plasmids
Our E. coli strains (Table S1) are all derivatives of K-12 BW25117 (
). The mutants were all deletion-replacements from the Keio collection, purchased from the E. coli Genetic Stock Center and all verified by PCR (and also phenotypically, whenever possible).
Enzymes and reagents
Catalase from bovine liver, H2O2, deferoxamine mesylate, horseradish peroxidase, and o-dianisidine:2HCl were all purchased from Sigma. DEA-NONOate was from Cayman Chemical. A 60 mM stock solution of DEA-NONOate was prepared fresh each time by dissolving several milligrams of the chemical in 0.1 M NaOH. NAD+/NADH assay kit was from Abcam (ab65348).
Growth conditions and viability assay
To generate killing kinetics, fresh overnight cultures were diluted 1000-fold into modified lysogeny broth (LB) [10 g tryptone, 5 g yeast extract, 5 g NaCl, 250 μl 4 M NaOH per liter (
). The stabilization of pH was required for reproducibility of NO delivery by DEA NONOate. Cultures were shaken at 37 °C for about 3 h or until they reached exponential phase (A600 ∼ 0.3). At this point, the cultures were made 0.6 mM for DEA NONOate and/or 2.5 mM for H2O2 (these two standard concentrations were used throughout the experiments; nonstandard concentrations are specified in a few experiments), and the shaking at 37 °C was continued. Viability of cultures was measured at the indicated time points by spotting 10 μl of serial dilutions in 1% NaCl on LB plates (LB medium above supplemented with 15 g of agar per liter). The plates were developed overnight at 28 °C, and the next morning, the pin-prick colonies in each spot with 10 to 200 colonies were counted under the stereomicroscope. All titers were normalized to the titer at time 0 (before addition of the treatment). For the iron chelator treatment, cultures grown as above were made 20 mM for deferoxamine mesylate 5 minutes before hydrogen peroxide treatment.
). The 40 mM o-dianisidine stock preparation: 318 mg of o-dianisidine:2HCl was added to 10 ml of 95% ethanol, then mixed with 25 ml of DI water. Assay cocktail: 60 μg/ml horseradish peroxidase, 150 μM o-dianisidine in potassium phosphate buffer (50 mM KPi, 0.1 mM EDTA, pH 7.8), kept ice-cold. Overnight cultures were diluted 1000-fold into LB8 medium and shaken for 3 h at 37 °C (A600 ∼ 0.3). At the desired timepoint after addition of H2O2 (±NO), 300 μl aliquots of cultures were withdrawn and cleared of the cells in a microcentrifuge for 1 min. Culture supernatant was diluted 1:10 into the potassium phosphate buffer. The diluted sample (667 μl) was mixed with 333 μl of the assay cocktail, and after 45 s at room temperature (∼20 °C), absorbance at 460 nm was measured.
Measuring chromosomal fragmentation by pulsed-field gel electrophoresis
). All strains were grown in LB8 medium; overnight cultures were diluted 1000-fold and grown with 1 to 10 μCi of 32P-orthophosphoric acid per milliliter of culture for 3 h at 37 °C (A600 ∼ 0.3) before addition of 0.6 mM NO + 2.5 mM H2O2 (or the indicated treatment). The reactions were stopped by addition of 312 μg of catalase (13 μl of 24 mg/ml stock), and aliquots of the culture were taken at the indicated times to make plugs. Cells of the aliquot were spun down, resuspended in 60 μl of TE buffer, and put at 37 °C. A total of 2.5 μl of proteinase K (5 mg/ml) was added, immediately followed by 65 μl of molten 1.2% agarose in the lysis buffer (1% sarcosine, 50 mM Tris HCl pH 8.0, and 25 mM EDTA) held at 70 °C. The mixture was pipetted a couple of times before being poured into a plug mold and let solidify for 2 min at room temperature. The plugs were then pushed out of the molds and incubated overnight at 60 °C in 1 ml of the lysis buffer. Half-plugs were loaded into a 1.0% agarose gel in 0.5× Tris–borate–EDTA buffer and run at 6.0 V/cm with the initial and the final switch times of 60 and 120 s, respectively, at 12 °C in CHEF-DR II PFGE system (Bio-Rad) for 20 to 22 h. The gel was vacuum dried at 80 °C (on Whatman paper) and then exposed to a PhosphorImager screen (Fujifilm) overnight. The resulting signals were quantified with a PhosphorImager (Fuji Film FLA-3000).
Electrochemical detection of NO and H2O2
Actual concentrations of NO and H2O2 were measured using NO sensor ISO-NOP (
) and H2O2 sensor ISO-HPO2 connected to the TBR4100 Free Radical Analyzer (World Precision Instruments). Before calibration, the sensors were polarized in PBS (137 mM NaCl, 2.7 mM KCl,10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) for over 12 h and 2 h, respectively. The NO sensor was calibrated by adding increasing concentrations of KNO2 to 0.1 M H2SO4+0.1 M KI. Changes in current (ΔpA) corresponding to increasing NO concentrations were measured to generate a standard curve. To measure NO released from DEA NONOate, LB or LB8 media were incubated at 37 °C on a temperature probe–controlled heated stir plate with stirring set at 170 rpm. The baseline current was recorded before DEA NONOate was added to the desired concentration, and the new current was recorded. The ΔpA calculated by subtracting baseline was used to determine the actual concentrations of NO. The H2O2 sensor was calibrated by adding increasing concentrations of H2O2 to PBS. Changes in current (ΔpA) corresponding to increasing H2O2 concentrations were measured to generate a standard curve (Fig. S2A). To measure H2O2 in cultures, 300 μl aliquots were withdrawn and cleared of the cells in a microcentrifuge for 1 min. The baseline current in PBS was recorded. Culture supernatant was diluted 1:10 into PBS, and the new current was recorded. The ΔpA calculated by subtracting baseline was used to determine the actual concentrations of H2O2.
O2 consumption assay
Cells were cultured to A600 = 0.2 in LB8 as described above. Respiration was measured with a Digital Model 10 Clark-type oxygen electrode (Rank Brothers, Ltd) at 37 °C, as described before (
). NO, H2O2, or CN were added to the desired concentrations, once the oxygen electrode chamber was filled and equilibrated with cell culture. The machine was calibrated by air-saturated LB medium and sodium dithionite.
). Overnight cultures in LB8 supplemented with 0.2% glucose were diluted to A600 = 0.010 in 1 L LB8 with 0.2% glucose and were grown to A600 = 0.3. Cells were collected by centrifugation for 5 min at 10,000g, resuspended in 10 ml ice-cold potassium phosphate buffer (50 mM, pH 7.8), and lysed with French press. After pelleting debris by spinning for 20 min at 20,000g, supernatant was spun further for 2 h at 140,000g to collect the membrane pellet. Membranes were resuspended in 3 ml ice-cold potassium phosphate buffer by repeated pipetting and stored on ice at 4 °C. The NADH consumption assay was performed with 200 μM NADH and 20 μl membranes. After addition of CN or nitric oxide, when needed, the total reaction volume was made 1000 μl with potassium phosphate buffer.
NAD+/NADH measurement assay
Extracts from E. coli cells were prepared and processed as described (
). Briefly, overnight LB8 cultures were diluted to A600 = 0.003 in 40 ml LB8 and grown to A600 = 0.25 to 0.3 at 37 °C with shaking. When appropriate, cells were treated with 0.6 mM DEA NONOate (final concentration) for 5 min. Cells were collected by filtration and resuspended in 700 μl ice-cold extraction buffer (from the Abcam kit). For NAD extraction, the resuspended cells were further diluted 50-fold in extraction buffer, lysed with 0.2 M HCl at 55 °C for 10 min, and neutralized to pH∼7.0 with NaOH. For NADH measurement, in the ndh nuo mutant, the filtered and resuspended cells were further diluted 10-fold in extraction buffer. For WT and NO-treated cells, the filtered and resuspended cells were used directly. Cells were lysed instantly with 0.2 M NaOH at 55 °C for 10 min and neutralized to pH∼7.0 with HCl. Cells were centrifuged at 12,000 rpm for 3 min to remove debris, and the collected supernatant was used directly in the Abcam NAD+/NADH assay protocol.
The procedure generally follows the protocol by Sen et al. (
) with some differences. Cells were grown in 500 ml LB8 to A600 between 0.1 and 0.25. When appropriate, cells were made 0.6 mM for DEA NONOate or 3 mM for CN and incubated for 10 min. Cells were harvested by centrifugation at 7000g for 5 min at 4 °C. The cell pellet was resuspended in 10 ml LB8 prewarmed to 37 °C. The medium also contained 10 mM DETAPAC (diethylenetriaminepentaacetic acid, pH 7.0) and 20 mM DF (pH 8.0). The cells were incubated at 37 °C for 15 min with shaking at 220 rpm. The cells were washed twice with 2 ml of ice-cold 20 mM Tris-HCl pH 7.4 and then resuspended in 300 μl of ice-cold 30% glycerol, 20 mM Tris-HCl pH 7.4, transferred into an EPR tube and frozen on dry ice with ethanol. The A600 of the final cell suspension was measured after a 1:1000 dilution. Ferric chloride standards were prepared in the same Tris buffer containing glycerol. The spectrometer settings were the following: microwave power, 10 mW; microwave frequency, 9.05 GHz; modulation amplitude, 12.5 G at 100 KHz; time constant, 0.032; temperature, 15°K.
All data described in the manuscript are contained within the manuscript itself.
The authors declare that they have no conflicts of interest with the contents of this article.
Special thanks are to Jim Imlay (Microbiology, UIUC) for his generous help and advice with this project and to Imlay lab members Sergey Korshunov and Maryam Khademian for help with respiration measurements and to Ananya Sen for help with the EPR protocol. We are grateful to the entire Kuzminov lab for their support and encouragement with this project.
P. A. and A. K. conceptualization; P. A. and A. K. methodology; P. A. investigation; P. A. formal analysis; P. A. and A. K. writing-original draft; P. A. and A. K. writing-review & editing; A. K. visualization; A. K. funding acquisition.
Funding and additional information
This work was funded by the NIH grant GM132484 . The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.