Anaerobic bacterial response to nitric oxide stress: Widespread misconceptions and physiologically relevant responses

How anaerobic bacteria protect themselves against nitric oxide‐induced stress is controversial, not least because far higher levels of stress were used in the experiments on which most of the literature is based than bacteria experience in their natural environments. This results in chemical damage to enzymes that inactivates their physiological function. This review illustrates how transcription control mechanisms reveal physiological roles of the encoded gene products. Evidence that the hybrid cluster protein, Hcp, is a major high affinity NO reductase in anaerobic bacteria is reviewed: if so, its trans‐nitrosation activity is a nonspecific secondary consequence of chemical inactivation. Whether the flavorubredoxin, NorV, is equally effective at such low [NO] is unknown. YtfE is proposed to be an enzyme rather than a source of iron for the repair of iron‐sulfur proteins damaged by nitrosative stress. Any reaction catalyzed by YtfE needs to be revealed. The concentration of NO that accumulates in the cytoplasm of anaerobic bacteria is unknown, but indirect evidence indicates that it is in the pM to low nM range. Also unknown are the functions of the NO‐inducible cytoplasmic proteins YgbA, YeaR, or YoaG. Experiments to resolve some of these questions are proposed.

Hmp, or related proteins (most recently reviewed by Poole, 2020).
In contrast, there is significant disagreement about how anaerobic bacteria respond to NO or nitrosative stress. This mainly involves NO reduction to nitrous oxide, which minimizes trans-nitrosation of protein −SH groups and glutathione. Many natural environments are oxygen limited rather than totally anaerobic, but at [O 2 ] below 50 µM, Hmp activity declines sharply due to its far lower affinity for oxygen than cytochrome oxidases (Poole, 2020;Robinson & Brynildsen, 2016). Even in oxygen-limited environments, NO detoxification therefore requires reduction to N 2 O rather than oxidation to nitrate.
A major source of disagreement in the literature has been the interpretation of the physiological significance of the results of experiments in which bacteria were exposed to concentrations of NO orders of magnitude higher than those known to occur in any natural environment. Metalloproteins bind and react with NO with widely differing affinities, and exposure often causes chemical inactivation or destruction (Figure 1; range 3). Although this problem was a major focus of the review by Spiro (2007), the points made in that review are still being ignored. Chemical damage due to extreme experimental conditions has been recognized by others, for example, by Mukhopadhyay et al. (2004), who referred to the possibility of "collateral damage," and by Rowley et al. (2012). This review will attempt to identify the dividing lines between chemical artifacts and physiologically relevant biochemical functions. Two assumptions will be made: first, that microbial metabolism has evolved to help bacteria survive under conditions that they encounter naturally; second, that gene expression is regulated to enable bacteria to adapt to environmental changes. If these assumptions are correct, knowledge of how gene expression is regulated is a good guide to the function of the gene product. Conversely, genes encoding enzymes that are known to be part of a stress response are usually regulated by a transcription factor directly involved in sensing the stress (Spiro, 2007).

| Major and minor sources of nitrosative stress in oxygen-limited environments
In anaerobic environments, bacteria are exposed to NO from various sources. First, NO is generated in the periplasm as an obligate intermediate in denitrification. It is also generated in the cytoplasm from nitrite that accumulates during denitrification or nitrate reduction to ammonia (Satoh et al., 1981;Smith, 1982). Bacteria associated with animal hosts are exposed to exogenous NO formed from arginine by NO synthase, an enzyme that is also present in some bacteria (Adak et al., 2002). Although NO is polar, it is soluble in water (1.94 mM in a saturated solution at 25°C). It can diffuse across the epithelial cell layer from the aerobic blood stream where it is formed into the oxygen-limited gastro-intestinal tract. Other minor or indirect sources of NO include atmospheric production of NO; and release of NO from nitrosated and nitrosylated cellular components. Whatever the source of NO exposure, the critical question is the concentration of NO to which bacteria are exposed naturally.
F I G U R E 1 Response of anaerobic enteric bacteria to different levels of nitrosative stress. The top three arrows indicate the likely ranges of [NO] that accumulate in the anaerobic bacterial cytoplasm (labeled 1), the range in which NorR activates the synthesis of NorVW (2) and the range at which metalloproteins are damaged chemically (3). Note that insufficient data are available to define the upper and lower limits of these ranges, so the figure is a best guess cartoon. Arrow A indicates the range of [NO] sufficient to inactivate highly sensitive [4Fe-4S] proteins such as aconitase B and fumarase B. Arrow B indicates the range over which the high affinity NO reductase Hcp is active (Wang et al., 2016). Arrows C and D indicate the much higher concentrations of NO required to inactive the transcription factors FNR, Fur, and OxyR Much of the current information about how anaerobic nondenitrifying bacteria respond to NO is based upon experiments with E. coli and other Enterobacteriaceae in which NO is generated as a side product during the reduction of nitrite to ammonia. In these bacteria, NO production from nitrite is catalyzed mainly by the cytoplasmic nitrate reductase, NarG (Calmels et al., 1988;Ralt et al., 1988;Rowley et al., 2012;Seth et al., 2012Seth et al., , 2018Smith, 1982). However, there are also other minor sources (Balasiny et al., 2018;Corker & Poole, 2003;Weiss, 2006). Under normal growth conditions, this side reaction accounts for only a very small proportion of the nitrite reduced by the dedicated nitrite reductases, NirBD and NrfAB (Smith, 1982;Wang et al., 2016). However, under extreme conditions, up to 20% of the nitrite was reduced via nitric oxide to nitrous oxide rather than to ammonia (Rowley et al., 2012). As it is unlikely that the conditions used in these experiments ever occur naturally, the authors noted that "it merits reflection whether this has any physiological importance."

| Concentrations of NO and its derivatives in natural environments
There are surprisingly few data in the literature of direct measurements of NO concentrations in natural environments, and even fewer estimates of the NO concentration that accumulates within the bacterial cytoplasm. These data are essential to define what is referred to below as "the physiologically relevant range" (Figure 1; range 1). In contrast, there have been many contradictory reports that the concentrations of NO, nitrosylated or nitrosated sources of NO are either high, or so low that they are physiologically irrelevant.
Reports that macrophages generate micromolar concentrations of NO are repeatedly cited. However, the data on which these statements were based report only the concentration of nitrite that accumulated over periods of up to 48 hr. Nitrite accumulates gradually as the end-product of NO synthesis by iNOS followed by its chemical oxidation (see, e.g., Roy et al., 2004). Rarely were attempts made to determine the much lower steady state concentration of NO in or around mammalian cells, even under extreme conditions used to complete laboratory experiments.
The intracellular concentration of NO in mammalian cells was originally estimated to be of the order of 1 µM (reviewed by Hall & Garthwaite, 2009). Some of these estimates were based upon data from measurements with electrodes that were later shown to be subject to interference by other compounds. Subsequently, much lower estimates were obtained from various approaches that included the use of guanylyl cyclase as an endogenous NO biosensor in tissues subjected to a variety of challenges. To quote directly from Hall & Garthwaite (2009), "All these independent lines of evidence suggest the physiological NO concentration range to be 100 pM (or below) up to ∼5 nM, orders of magnitude lower than was once thought." Many pathogenic bacteria are highly resistant to antimicrobial therapy because they form biofilms (Mah & O'Toole, 2001).
Formation of NO by the host is an effective defense mechanism in part because it provokes biofilm dispersal and hence increases vulnerability to antibiotic therapy. Key players in this bacterial response to NO are the hemoproteins of the H-NOX and NosP families. In many bacteria the hemoprotein H-NOX is associated with and regulates the sensor kinase of a two-component regulatory system that is activated by NO. In contrast, Pseudomonas aeruginosa is typical of other biofilm forming bacteria that lack an hnoX gene. These bacteria instead rely upon another hemoprotein, NosP, to activate the NO response (Hossain et al., 2017). Both proteins have been implicated in mechanisms that switch the formation of cyclic-di-GMP to its removal by phosphodiesterases. Description of the complex mechanism involved are beyond the scope of this review, but the relevant point is that both H-NOX and NosP respond to pM concentrations of NO that occur naturally in the human body (reviewed by Williams & Boon, 2019).
Nitric oxide generated from nitrite as a key intermediate of denitrification by α-and β-proteobacteria is reduced to nitrous oxide as rapidly as it is formed. A mutant of Pseudomonas stutzeri defective in the NO reductase, NorBC, was unable to grow anaerobically by denitrification because NO production catalyzed by the nitrite reductase, NirK, was toxic (Braun & Zumft, 1991). The steady state concentration of NO during nitrate denitrification by Pseudomonas denitrificans was 15 nM (Bakken et al., 2012). Other bacteria also maintain tight control of NO accumulation, limiting it to the range of 5 to 35 nM. As NO is reduced in the periplasm of these bacteria, the intracellular NO concentration will be much lower than this. Only a cytoplasmic NO reductase with a very high affinity for NO would be effective in preventing metalloprotein damage by these very low concentrations.
Denitrifying bacteria vary significantly in their ability to prevent NO release into their environment (Hassan et al., 2016).
Nevertheless, even a relatively prolific source, Agrobacterium tumefaciens, transiently accumulated only 100 nM NO during the transition from aerobic growth to anaerobic denitrification. Consequently, bacteria that share their environment are exposed to low nM concentrations of NO. However, far higher concentrations have been reported in some laboratory experiments. For example, a strain of Agrobacterium tumefaciens was unable to adapt from oxygensufficient to nitrate-dependent growth because as soon as the oxygen had been depleted, the [NO] increased to 8 µM (Bakken et al., 2012). This was sufficient to inhibit growth completely until the NO had been removed by reduction to nitrous oxide, raising the question whether some bacteria accumulate much higher [NO] in their natural environments than implied above. This possibility will be discussed further at the end of this review.
Many cytoplasmic proteins in anaerobic bacteria are extremely sensitive to inactivation even by these low concentrations of NO (Gardner et al., 1997;Hyduke et al., 2007;Justino et al., 2007). Data derived from experiments with 10 or even 100 µM NO or its surrogates therefore need to be interpreted with caution because these values are at least three and possibly up to seven orders of magnitude above the physiological range and well into the range at which metalloproteins will be chemically damaged by nitrosylation and the subsequent trans-nitrosation of protein −SH groups (Table 1; Figure 1).
Transcription of these genes is induced by many different sources of NO that include exposure to nitrate, nitrite, NO gas, surrogates for NO such sodium nitroprusside, the NO-releasing NONOates, especially diethylamine NONOate (DEANO), or the trans-nitrosating agents S-nitrosothiol and S-nitrosoglutathione (Table 1; Crawford et al., 2016;Flatley et al., 2005;Hausladen et al., 1998;Rogstam et al., 2007;Seth et al., 2018). All of the encoded proteins except the periplasmic nitrite reductase, NrfA-NrfB, are located in the cytoplasm. The norVW genes encode the NO reductase, NorV, and its NADH-dependent reductase, NorW. The physiological roles of YeaR and YoaG are unknown.

| The controversial physiological role of the hybrid cluster protein
Throughout three decades of research that produced high-resolution structural and spectroscopic information, the physiological role of Hcp remained unknown. Four roles have been proposed, two of which can immediately be discarded (Hagen, 2019). A catalytically ineffective hydroxylamine reductase activity was demonstrated for the E. coli Hcp (Wolfe et al., 2002). Despite warnings by the authors that this was unlikely to be its physiological function, Hcp is still annotated in many genomes as a hydroxylamine reductase. For similar reasons, a proposal that it provides defense against peroxide stress can be discounted (Almeida et al., 2006;Hagen, 2019).
In macrophage experiments, Kim et al. (2003) reported that an hcp mutant of Salmonella enterica serovar Typhimurium is sensitive to NO generated from acidified nitrite. Similar results were obtained by Boutrin et al. (2012) for Porphyromonas gingivalis, and da Silva et al. (2015) showed that a Desulfovibrio gigas hcpR mutant that cannot produce Hcp is sensitive to nitrosative stress. Wang et al. (2016) were the first to report that the E. coli Hcp is a high affinity NO reductase that detoxifies the low concentrations of NO that accumulate in the bacterial cytoplasm during anaerobic growth. This function is consistent with the requirement to protect cytoplasmic enzymes, especially the dehydratase family, from NO generated by NarG during nitrate reduction (Constantinidou et al., 2006;Filenko et al., 2007). Wang et al. (2016) showed that a major role of Hcr (the NADH-dependent hybrid cluster protein reductase) is to protect the NO reductase activity of Hcp from inactivation by high concentrations of its substrate, NO. These results need to be confirmed not only by other laboratories, but also for the Hcp proteins of other bacteria, especially those that lack Hcr.
Despite this evidence that the primary role of Hcp is to detoxify the very low concentrations of NO released into the bacterial cytoplasm during nitrate or nitrite reduction, a conflicting role was proposed that Hcp is an enzyme that is first nitrosylated by NO and then catalyzes nitrosation of a wide range of other proteins (Seth et al., 2018). Seventy-four mutants were constructed and the total amount of protein nitrosation during anaerobic growth in the presence of nitrate was compared with that of the parent strain.
Nitrosation in the hcp mutant was less than 10% that of the parent, significantly less than in any of the other mutants in this study.
However, decreases in protein nitrosation of more than 50% were also reported for mutants defective in ymgA, ymgC, ybaY, and other genes tested. Despite this and the fact that none of these other  Effective protection against NO toxicity by reduction to N 2 O would require a cytoplasmic NO reductase with both a very low K m for NO and a high efficiency, kcat/K m . The weak NO reductase activity of the flavohemoglobin Hmp clearly fails this test . The only estimated catalytic efficiency of any Hcp was 2.4 x 10 9 M -1 s -1 , six times higher than that estimated for NorV (Wang et al., 2016). However, the application of the Michaelis-Menten model by Wang et al. (2016) has correctly been challenged.

TA B L E 1 Sources and surrogate sources of NO used in various studies
Hagen (2019) Wonderen et al. (2008) norVW NO reductase, NorV, and its reductase, NorW Transcribed from a σ 54 -dependent promoter: enhancer protein NorR Hutchings et al. (2002) clustered at cell poles even in an hcp mutant, so contrary to the proposal of Seth et al. (2018), complex formation was neither de-

| Transcription factors that regulate gene expression in response to NO
All of the well-characterized transcription factors that regulate the bacterial response to NO are metalloproteins. There have been many reports of activators and repressors that fall outside this generalization, but most of the original claims have subsequently been challenged. Until independent experimental confirmation has been published, the initial claims remain controversial. Transcription factors that respond directly and with high sensitivity to NO fall into two classes: hemoproteins; and nonheme iron or iron-sulfur proteins (Rodionov et al., 2005). Typical of the former group are the HcpR The SrrAB two-component system of Staphylococcus aureus undoubtedly plays an important role in the defense against nitrosative stress (Richardson et al., 2006). However, the SrrAB regulon extends well beyond genes specifically involved in the nitrosative stress response, so it is unlikely that the environmental sensor, SrrA, detects NO directly (Kinkel et al., 2013).
In Campylobacter jejuni, synthesis of the globins Cgb (a singledomain globin) and Ctb (a truncated globin) is induced in response to NO via the positively acting transcription factor, NssR. These are also indirect effects rather than a transcriptional response initiated by the binding of NO to NssR. Similar arguments apply to the thiol-based RNA polymerase regulatory protein, DksA, in Salmonella (Crawford et al., 2016). rophages (Pullan et al., 2007). These observations strongly support the proposal that Hcp rather than NorV provides the main protection mechanism for cytoplasmic proteins under normal levels of nitrosative stress. Pullan et al. (2007) also demonstrated that nitrosative stress is due to metalloprotein NO nitrosylation rather than to −SH nitrosation reactions.

| How transcription control mechanisms reveal the roles of gene products
In contrast to norVW, expression of other genes induced by NO are repressed by NsrR. They are transcribed from promoters dependent upon the housekeeping sigma factor, σ 70 . This reflects the contrasting roles of the two major E. coli transcription factors NsrR and NorR that respond to nitrosative stress. Genes regulated by the σ 70 -dependent NsrR provide a first line of defense against the very low concentrations of NO that occur in the bacterial cytoplasm ( Figure 1, range 1). The housekeeping function of NsrR-regulated gene products is to protect vulnerable cytoplasmic proteins such as the dehydratase family against the very low concentrations of NO encountered during normal growth (Duan et al., 2009;Hyduke et al., 2007;Ren et al., 2008;Varghese et al., 2003). When the intracellular [NO] increases toward the concentration that would overwhelm NsrR-regulated defenses, NorR is activated and the NorVW | 35 COLE system provides an effective stress response (Figure 1; range 2: see also Chismon et al., 2010;Karlinsey et al., 2012). Insufficient data are available to define the lower limits of cytoplasmic [NO] or external nitrosative stress required to inactivate NsrR or activate NorR by nitrosylation.
The hcp-hcr and nrfAB operons are expressed only during anaerobic growth and require a functional FNR protein, but expression of hmp is repressed by FNR. This is consistent with the oxygendependent role of Hmp as an NO oxygenase during aerobic growth, but not during anaerobic growth . The hcp operon is strongly induced during nitrate and nitrite reduction to ammonia (Constantinidou et al., 2006;Vine et al., 2011). This established two important points. First, NsrR is far more sensitive to NO than FNR. Second, FNR is almost fully active even during anaerobic growth in the presence of concentrations of nitrate or nitrite above those encountered in natural environments (Figure 1, ranges A and C). Nevertheless, expression from NsrR-regulated promoters is fully derepressed in an nsrR mutant, but only partially derepressed during anaerobic growth in the presence of nitrate or nitrite (Vine et al., 2011). NsrR from Streptomyces coelicolor binds to its DNA target sites with very high affinity (Crack et al., 2016). Assuming the same is true for E. coli NsrR, this is further evidence that the concentration of NO in the E. coli cytoplasm is extremely low during nitrate or nitrite reduction, well below the K D of NsrR for NO and consistent with a cytoplasmic NO concentration in the low nanomolar or picomolar range, as estimated by Wang et al. (2016) and others.
Excess nitrate induces expression of the cytoplasmic nitrate reductase operon narGHJI and the nitrate-nitrite transporter gene, narK. As nitrite reduction to NO by NarG is the major source of NO in the cytoplasm, it makes physiological sense for the nitrite reductases, NrfAB and NirBD, as well as Hcp, to be regulated by FNR coordinately with nitrate uptake and reduction. Despite its high Km for NO, the high activity of the periplasmic nitrite reductase provides a first line of defense against NO originating outside the bacteria (Wonderen et al., 2008).
NO reduction by α-and β-proteobacteria is largely regulated by proteins of the Fnr/Crp family (Rodionov et al., 2005). In contrast, the NO reductase genes of the truncated denitrification pathway of pathogenic Neisseria are regulated by NsrR (Heurlier et al., 2008;Overton et al., 2006). This again makes physiological sense because one source of nitrosative stress in the human body is NO generated in aerobic tissues from arginine by NO synthases.
NO is also a product of nitrite reduction in oxygen-limited tissues.
As nitrite reduction provides an electron acceptor for energy generation when oxygen is scarce, expression of the nitrite reductase gene, nirK, is regulated by both FNR and NsrR. These are therefore examples of how gene regulation has evolved to enable bacteria to exploit their anaerobic environment. They illustrate how physiological roles can be revealed by knowledge of transcription control mechanisms. Conversely, transcription factors that regulate genes known to be involved in NO metabolism are likely to be bona fide NO sensors (Spiro, 2007).

| Errors in the assignment of physiological roles due to protein damage by unnatural levels of stress
Many misunderstandings of how nitrosative stress is regulated have arisen because the bacterial transcription factors that regulate responses to a lack or excess of redox-active metals, oxygen and reactive oxygen species are iron or iron-sulfur proteins. At high concentrations of NO or related species, these other transcription factors are inactivated chemically and are therefore unable to function.
This results in derepression of some genes and failure to express others, but only under extreme conditions rarely encountered by the bacteria in which they have evolved (Spiro, 2007). A remaining source of controversy is whether there is a dividing line between irrelevant chemical damage that requires extreme conditions never encountered naturally, and a physiological response that confers a survival advantage under conditions of stress (Figure 1, ranges 2 and 3). The FNR protein provides an excellent example.  (Green et al., 1996;Sutton et al., 2004). Although the same is true of inactivation by NO of dehydratases such as the dihydroxy-acid dehydratase, IlvD, aconitase and fumarase, these iron-sulfur proteins are much more sensitive than were subsequently shown to be due to relief of repression by NsrR rather than to relief of FNR repression (Bodenmiller & Spiro., 2006;Pullan et al., 2007).
The primary role of the SoxRS two-component regulatory system is to regulate the response to superoxide stress, but SoxR also binds NO (Nunoshiba et al., 1993). However, SoxRS does not regulate genes required for NO defense mechanisms and SoxR is not an NO sensor.
In many bacteria iron uptake and metabolism are regulated by the iron protein, Fur. As Fur, like FNR, can be inactivated by NO nitrosylation, it has incorrectly been assigned a role in the response to nitrosative stress (D'Autreaux et al., 2002). This conclusion is again

| The physiological role of YtfE in the repair of nitrosative damage
Dinitrosylation of some iron-sulfur proteins, for example, the dehydratase family, results in chemical damage and the slow release of iron atoms (D'Autreaux et al., 2002;Duan et al., 2009;Hyduke et al., 2007).
This free iron then becomes available to bind to other iron-deficient proteins. The di-iron protein YtfE (also known as RIC, for the repair of iron-sulfur centers) has been shown in vitro to be able to donate iron to apo-ferredoxin and to the iron-sulfur assembly protein, IscU. It also exchanges protein-bound iron with free iron (Justino et al., 2007;Nobre et al., 2014). These processes were extremely slow, requiring incubation periods of 15 to 75 min and higher than stoichiometric concentrations of YtfE than the protein to be reconstituted. This raised the possibility that there is an alternative, physiologically more relevant function of YtfE. A crystal structure of YtfE revealed the presence of two solvent-accessible channels, both of which converge to the diiron center and might therefore be critical for capturing substrates (Lo et al., 2016). The authors demonstrated the ability of their purified protein to reduce NO to N 2 O, but the reaction was too slow to account for the rates of NO reduction detected with bacterial suspensions.
We recently reported that YtfE directly or indirectly releases NO from nitrosylated iron-sulfur centers (Balasiny et al., 2018). Two surprising results were reported. First, a YtfE + Hcpstrain was far more sensitive to growth inhibition by NO than the isogenic Hcp -YtfEstrain.
Second, neither aconitase nor fumarase were more active in strains in which YtfE was functional than inactive. This failure of YtfE to reactivate aconitase and fumarase in the mutant might be explained in either of two ways. One possibility, previously published by Balasiny et al. (2018), is that NO is released directly from nitrosylated iron-sulfur proteins by YtfE. In the absence of Hcp, the NO released immediately re-nitrosylates iron-sulfur centers of enzymes like aconitase and fumarase, effectively establishing a futile repair cycle. The alternative explanation might be that NO is released attached to the nitrosylated iron atom, leaving an inactive iron-deficient enzyme. If so, the iron deficiency could then be repaired by the Suf or Isc pathway, NO being released by dissociation from the Fe-NO or Fe-(NO) 2 produced ( Figure 2 and Vine et al., 2011with Constantinidou et al., 2006.

| CON CLUS I ON S AND SOME OF THE MANY UNANSWERED QUE S TIONS
As NO is freely diffusible, is this simply the consequence of extreme nitrosylative damage to proteins required for the adaptive response?
If not, the highly unlikely alternative is that there is a barrier that prevents external NO from equilibrating with the cytoplasmic concentration. Any such barrier remains undefined. induced. This would provide an excellent source of bacteria to assess the level of protein −SH nitrosation in bacteria in a natural oxygenlimited environment.
Other unresolved questions include the physiological roles of YgbA, YibI-YibH, and YeaR-YoaG, proteins in enteric bacteria that are synthesized in response to nitrosative stress. Do any of them provide links between NO-sensing, biofilm dispersal, repair of DNA damage, or quorum sensing and loss of motility? Can we correlate their induced synthesis with specific sources of stress, for example, NO generated internally during nitrate reduction or exposure to external NO?
It is highly likely that YtfE (RIC) is an enzyme, but the reaction it catalyzes remains to be established. Both the sensitivity of this activity to high concentrations of NO and the form in which the NO is released need to be identified. The proposal that it has a physiological role in repairing nitrosylated iron or iron-sulfur centers seems increasingly unlikely to be correct (Lo et al., 2016). If, however, it is correct, how substrate specific is YtfE repair? Both of the NsrRrepressed operons, ytfE and yeaR-yoaG are more highly expressed in an fnr mutant than in the fnr + parent (Constantinidou et al., 2006).
They are also regulated by NarL. No FNR binding site is present in their promoter regions, so the regulatory mechanisms remain to be revealed. in response to oxygen, peroxide, or iron availability. It also led to the misconceptions that Hcp provides defense to reactive oxygen species, that it is a hydroxylamine reductase, or essential for transnitrosation; or that YtfE (RIC) is primarily an Fe donor that repairs damaged iron-sulfur centers. Hopefully, it has set some basic principles that must be followed to resolve the many unanswered questions.

ACK N OWLED G M ENTS
The author is greatly indebted to many colleagues for helpful dis-