The Potential for Redox-Active Metabolites To Enhance or Unlock Anaerobic Survival Metabolisms in Aerobes

Classifying microorganisms as “obligate” aerobes has colloquially implied death without air, leading to the erroneous assumption that, without oxygen, they are unable to survive. However, over the past few decades, more than a few obligate aerobes have been found to possess anaerobic energy conservation strategies that sustain metabolic activity in the absence of growth or at very low growth rates. Similarly, studies emphasizing the aerobic prowess of certain facultative aerobes have sometimes led to underrecognition of their anaerobic capabilities.

ABSTRACT

Classifying microorganisms as “obligate” aerobes has colloquially implied death without air, leading to the erroneous assumption that, without oxygen, they are unable to survive. However, over the past few decades, more than a few obligate aerobes have been found to possess anaerobic energy conservation strategies that sustain metabolic activity in the absence of growth or at very low growth rates. Similarly, studies emphasizing the aerobic prowess of certain facultative aerobes have sometimes led to underrecognition of their anaerobic capabilities. Yet an inescapable consequence of the affinity both obligate and facultative aerobes have for oxygen is that the metabolism of these organisms may drive this substrate to scarcity, making anoxic survival an essential skill. To illustrate this, we highlight the importance of anaerobic survival strategies for Pseudomonas aeruginosa and Streptomyces coelicolor, representative facultative and obligate aerobes, respectively. Included among these strategies, we describe a role for redox-active secondary metabolites (RAMs), such as phenazines made by P. aeruginosa, in enhancing substrate-level phosphorylation. Importantly, RAMs are made by diverse bacteria, often during stationary phase in the absence of oxygen, and can sustain anoxic survival. We present a hypothesis for how RAMs may enhance or even unlock energy conservation pathways that facilitate the anaerobic survival of both RAM producers and nonproducers.

INTRODUCTION

The human heart pumps approximately three and a half billion times in the average lifetime. It has evolved to pump dependably, one beat after the other, because the consistent flow of blood throughout the body is necessary to sustain a human life. Bacteria do not contain a circulatory system per se. However, like the flow of blood, all cells maintain a flow of electrons through their metabolic pathways and a flow of ions across their membrane to remain alive. Without a redox-balanced cytosol, ATP generation coupled to electron transfer reactions becomes stalled along with biosynthesis, and without an electrochemical potential at the membrane, the cell can neither power its flagella (1, 2) nor transport nutrients into and waste out of the cytosol (3). Failure to uphold these physiological requirements results in rapid loss of viability in a population, as cell after cell struggles to conserve the energy needed to stay alive.

What does being labeled an aerobe imply? Bacteria classified as “obligate” aerobes are defined by their lack of growth in the absence of oxygen. It is important to note that this does not necessarily mean the cells die in the absence of oxygen: indeed, anaerobic survival mechanisms in obligate aerobes are known, though they are understudied relative to their likely relevance in the wild. To survive anoxia, these aerobes are faced with a metabolic dilemma: they must maintain redox balance and the proton motive force (PMF) despite interrupted flux through the electron transport chain (ETC) due to the absence of oxygen. While facultative aerobes are well understood to be able to conserve energy and even grow rapidly in the absence of oxygen, through either the use of alternative electron acceptors or fermentation, aerobic respiration is commonly viewed as their preferred metabolic mode. We contend, however, that this presumed physiological bias may better reflect what fosters fast growth in the laboratory rather than what promotes fitness in nature. The ability to efficiently survive until nutrients return is essential for any microbe’s evolutionary success and, in many natural scenarios, may be even more important than growth.

The bioenergetic basis of bacterial survival strategies in the real world is poorly understood. Key to gaining relevant physiological insights is to couple microenvironmental characterization with the design of reductionist laboratory experiments that bear fidelity to those conditions. It is well appreciated that a major form of bacterial life in the wild is within slowly growing biofilms (4) and that cells in the interior of these biofilms are often oxygen limited, because oxygen consumption outpaces its diffusion inwards (5). Yet these cells are very much alive: they are able not only to maintain their viability but also to mount transcriptional responses (6) and to produce a steady stream of new proteins (7). Furthermore, for any microorganism living in the soil, periods of rainfall are a common stressor, because wetted soil quickly becomes depleted in oxygen (8), and even in the human body, counterintuitively, aerobes that infect the lung often become embedded in mucus that is largely hypoxic or anoxic (9, 10). Therefore, researching anaerobic metabolisms in aerobes is to work toward understanding an aspect of bacterial physiology that is important in nature.

Anaerobic metabolisms within the facultative aerobic genera Pseudomonas (11) and Bacillus (12) and the obligate aerobic genera Arthrobacter (13, 14), Azotobacter (15), Streptomyces (16), and Mycobacteria (17, 18) have been studied in some detail; in this review, we will discuss Pseudomonas aeruginosa and Streptomyces coelicolor as illustrative examples. Even though streptomycetes undergo a spore stage in their life cycle, we focus our review only on the survival mechanisms of vegetative cells. While P. aeruginosa is still sometimes described as an obligate aerobe, it has long been known to be a facultative aerobe capable of denitrification (i.e., utilizing nitrate rather than oxygen as the terminal electron acceptor for respiration) (19–21). P. aeruginosa also grows anaerobically via arginine fermentation (22) and can survive anaerobically via pyruvate fermentation (11); interestingly, anaerobic survival powered by substrate-level phosphorylation can also be achieved by the reduction of self-produced extracellular electron shuttles called phenazines (23). We end this review by discussing the potential for functionally similar redox-active metabolites (RAMs) made by other aerobes to also promote anaerobic survival. Such multispecies use of these secondary metabolites for energy conservation would raise exciting new interpretations of their producers’ impact on microbial communities in soil, infection, and even industrial fermentation settings.

THE RELEVANCE OF ENERGY-LIMITED SURVIVAL

While many things can limit bacterial growth, from nutrient scarcity to chemical warfare to predation, from a catabolic perspective, electron donor (e.g., carbon) and electron acceptor (e.g., oxygen) limitations are the most obvious drivers of energy stress for heterotrophic aerobes and have been the best studied. There are significant differences in how the cell handles oxygen and carbon limitation (24), but they both diminish flux through the ETC by halting the flow of electrons in or out. This lack of flux can threaten survival by interrupting the PMF, ATP synthesis, and cellular redox-balancing systems. Given the paucity of mechanistic research on survival physiology (though this is beginning to be rectified [6, 24, 25]), as a point of departure for our discussion of how oxygen limitation may be overcome via unconventional catabolic pathways, we first remind the reader of the broader challenges that must be solved by cells that are energy limited and the ecological relevance of this state.

Irrespective of how energy limitation is imposed in the lab, a cell that is so challenged must transition from a conventional stationary state to a distinct starvation survival state, involving a (sometimes dramatic) decrease in cell size (26, 27) often accompanied by sustained changes in lipid content (26, 28), the proteome (26, 29), and the transcriptome (4, 26). Interestingly, the transition from growth into the starvation survival state seems to require proteolysis (30, 31), presumably to supply the amino acids needed to produce new proteins. This physiological state describes the vast majority of microbes in nature (4, 32, 33), as most bacterial environments are oligotrophic or energy limited (32). The amounts of dissolved organic carbon in soil vary but are estimated to be quite small: as an example, one comprehensive study in the United Kingdom estimated it to range between 1 and 100 mg liter⁻¹ (34). In fact, the methods used to measure dissolved organic carbon likely release labile carbon from complexes; therefore, the actual bioavailable carbon is probably even lower (32). Nonetheless, isotope-tracing experiments made with Vibrio environmental isolates demonstrate rapid consumption of labile glutamate added in picomolar concentrations (10⁻¹², the lower detection limit of the method) confirming that microbial carbon consumption can capture exceedingly dilute substrates (35).

These studies and others (32, 33) strongly suggest two things: (i) environments are kept energy limited by the microbes that inhabit them and (ii) nutrient flux and bacterial growth in the wild are quite low relative to growth in the lab. Given natural environmental variability, a cell can be induced into a growth-arrested state for many reasons, and it is often challenging (if not impossible) to know which particular limitation or stress is responsible for growth arrest at any given time. Regardless, division frequency estimates within environmental samples predict that soil heterotrophs replicate between fewer than 1 to 36 times per year, with the majority of estimates falling below one (36). To put that in perspective, growing one cell in a 5-ml culture to an optical density (OD) of approximately 1.0 (∼10¹⁰ cells) takes around 33 generations. This means growth typically made in a day in the lab simulates the equivalent of at least, and likely more than, a year’s worth of reproduction in the soil. While it is not always the case, if these soils become waterlogged, which can quickly render them anoxic (8), it stands to reason that the aerobes in these environments with the ability to survive in the absence of oxygen will have a fitness advantage over those that do not. But what does the ability to survive really mean?

When cells enter growth arrest, the majority of their available energy budget is used for non-growth-related maintenance processes (Fig. 1). The concept of maintenance energy arose from chemostat studies, which revealed that the bacterial growth yield falls below expectations given the ATP yield. This difference implied that a fraction of the ATP energy produced was being diverted away from growth processes. Since then, many studies have quantified this difference as the maintenance energy and found it to vary depending on both the species and conditions (37–39). One can create a linear fit showing how energy consumption changes with the growth rate in a chemostat and extrapolate this line to a hypothetical growth rate of zero in order to estimate the maintenance energy. The line does not pass through the origin, which fits the intuitive expectation that there must be some minimal power consumption necessary to sustain even nongrowing biomass against death and degradation.

FIG 1.

Proposed nongrowth components involved in microbial maintenance energy, adapted from van Bodegom (115).

However, the maintenance energy calculated this way assumes that the metabolic state of fast-growing cells is the same as that of starved cells, when in reality, starving cells have made a large variety of physiological adjustments. Indeed, chemostat-based maintenance energy inferences are estimated to be around 3 orders of magnitude higher than those measured in environmental samples (38). The interpretation of this discrepancy is difficult, as the relative contributions of processes contributing to chemostat versus environmental maintenance energy are not well defined. They, again, likely vary by species and conditions and may not all be essential to survival.

This all begs the question of what are the actual minimal energy requirements for a cell to survive? To differentiate between the concepts of chemostat-derived maintenance energy and the minimal energy needed to sustain an organism, Hoehler and Jørgensen suggested the term basal power requirement (BPR) (33). The BPR is defined as the minimum energy turnover rate per cell required to sustain a metabolically active state (33). While careful measurements of the BPR are rare, Lever et al. made one estimate from bulk oxygen consumption rates in aerobic marine sediments that approached an asymptote with sampling depth (26). They reason that this value might reflect the BPR in this environment and estimated it to be between 3 × 10⁻¹⁴ and 3 × 10⁻¹³kJ cell⁻¹year⁻¹, depending on whether one assumes the average Gibbs free energy yield of those cells’ aerobic metabolism to be 100 or 1,000 kJ mol O₂⁻¹. For comparison, Escherichia coli power consumption during exponential growth (also based on its bulk oxygen consumption rate and the same free energy yields of aerobic metabolism [40]) is extrapolated to be between 8 × 10⁻⁹ and 8 × 10⁻⁸kJ cell⁻¹year⁻¹, 4 to 6 orders of magnitude higher. It is important to note, however, that the BPR for organisms surviving in the absence of oxygen may be even less, as the burden of repairing cellular components damaged by oxidative species generated via aerobic respiration may inflate these values (41).

METABOLIC REQUIREMENTS FOR SURVIVAL

Because there are few mechanistic studies related to energy-limited survival, any description of minimal metabolic requirements is limited to our understanding of metabolism during fast growth. One primary question is when energy is limited, what are the cell’s metabolic priorities? Intuitively, preservation of a minimum PMF would seem essential to transport nutrients in and wastes out (42). This intuition relies on the assumption that active transport of some kind (either powered by the PMF or ATP) is necessary for homeostasis to be achieved, yet that may not always be the case. Intriguingly, recent evidence suggests that some obligate aerobes in their vegetative state, such as Bacillus subtilis, may take an entirely different survival strategy, benefiting from attenuation of the PMF (43). An exciting frontier for bacterial physiology lies in determining the level of PMF actually needed to sustain a cell and what mechanisms underpin its maintenance under different conditions. Innovations in tools that enable quantitative measurement of the PMF in single cells, together with tools that quantitatively measure ATP or other metabolites that permit the assessment of metabolic state, are a priority for development in bacteria.

Putting these existential unknowns and corresponding research opportunities aside, it is helpful to recall some basic facts about energy conservation in bacteria. To start, it is axiomatic that metabolically active, aerobically respiring cells charge the PMF through the electron transport chain (ETC), enabling ATP generation by dissipation of the chemiosmotic gradient through the F₁F_o-ATP(synth)ase. Yet a fact that is perhaps less well remembered is that when the ETC flux is stalled due to the lack of a terminal electron acceptor, cells have many other options for PMF maintenance. For instance, anaerobic proton extrusion can be accomplished by protonated carbon compound antiport (44), and driving the F₁F_o-ATP(synth)ase in reverse (which can be regulated [45]) can also charge the PMF (46). Intriguingly, oxygen-limited P. aeruginosa cells lower their ATP pools more quickly than carbon-limited cells, and treating such cells with PMF inhibitors reduces the number of survivors (24, 47). These results are consistent with energy limitation being more extreme for oxygen-limited cells due to their dependency on the low-energy-yielding acetate kinase-phosphate acetyltransferase (AckA-Pta) pathway, a pathway that generates ATP via substrate-level phosphorylation (11, 46). It is in this context that we believe much stands to be gained from identifying what constrains flux through substrate-level phosphorylation pathways in aerobes.

Directly related to this challenge is to ask what facilitates the maintenance of intracellular redox pools (the homeostasis of which is critical for both catabolism and anabolism)? These pools include species such as NADH/NAD⁺ (predominantly involved in energy conservation), NADPH/NADP⁺ (predominantly involved in biomass generation), and glutathione (predominantly involved in protein disulfide bridge homeostasis). When the NAD(P)H/NAD(P)⁺ balance is disrupted, reactions that rely on electron flow through these pools become stalled, and the growth rate of the cell diminishes. For example, Richardson et al. found that when the photoheterotroph Rhodobacter capsulatus is grown anaerobically on reduced carbon compounds, it displays a growth deficit in the absence of CO₂ (48). This growth defect can be recovered upon addition of alternative terminal oxidants, suggesting that the Calvin cycle functions as an electron sink necessary for proper redox balance during growth (48). Using C¹³ metabolic flux analysis, McKinlay and Harwood confirmed that in the photoheterotroph Rhodopseudomonas palustris, the Calvin cycle indeed functions as a crucial electron sink, accounting for almost 50% of NADH oxidation during conversion of acetate to biomass (49). Rao et al. found that in nonreplicating hypoxic Mycobacterium tuberculosis, the NADH dehydrogenase NDH-2 is essential to cells’ survival because it replenished the NAD⁺ pool of cells that were otherwise stalled in a predominantly reduced state (47). The importance of redox balancing is also well appreciated in the study of fermentation and metabolic engineering, as shifting the NADH/NAD⁺ ratio of the cell, either genetically or through providing differently oxidized carbon sources, can significantly alter not only growth but also the relative yields of fermentation products (50).

While both the PMF Δp (the potential energy stored across the membrane from differences in ion concentration) and NADH/NAD⁺ ratio vary depending on the organism and the environment (51, 52), as discussed above, it seems likely that there are limits to this variation beyond which most cells cannot live. At the extremes, without a PMF, the cell would be relying on passive import of substrates in a likely dilute environment. Without a properly maintained NADH/NAD⁺ ratio, flux through key redox enzymes would cease. The result of either is that the flux through ATP-producing pathways would be reduced along with the BPR processes necessary to maintain metabolic activity. Therefore, the PMF and intracellular redox state are valuable integrative metabolic outputs to track when assessing a survival phenotype, regardless of which stress triggers the growth-arrested state.

CONSEQUENCES OF LIFE WITHOUT OXYGEN

Focusing on oxygen as a key physiological variable, it is tempting to assume that the net energy available to respiring aerobes is always much greater than that available to fermenting anaerobes. In reality, there are trade-offs to each metabolic strategy given its respective environmental context. Understanding those trade-offs provides a framework for how energy budgets and relative fitness shift under changing conditions. Cells under anoxic conditions may produce energy from the respiration of alternative terminal electron acceptors such as fumarate, sulfate, and nitrate. Each such alternative electron acceptor has a lower redox potential than oxygen, and so the amount of energy the cell is capable of conserving from its reduction is lower than that which would result from the reduction of an equivalent amount of oxygen. Cells can also ferment by maintaining a balance of their NADH/NAD⁺ pool, permitting flux through pathways that generate ATP by substrate-level phosphorylation. Though the energetic efficiency of fermentation is approximately one order of magnitude less than that for aerobic respiration (for one mole of glucose, fermentation generates 1 to 5 mol of ATP depending on the pathway, much less than the approximately 32 mol generated from aerobic respiration), the fitness benefit of efficiency depends on context.

Aerobically respiring cells constantly deal with oxidative stress that consumes significant energy. Oxidative stress is characterized by oxidation of redox pools, interfering with homeostasis between anabolism and catabolism. While oxidative stress generically can be caused by a lack of electron donors relative to acceptors, the generation of superoxide radicals and hydrogen peroxide resulting from oxygen excess adds insult to injury. These toxic reactive oxygen species are generated when oxygen autoxidizes redox enzymes (especially flavoproteins) (41), making oxidative stress an unavoidable consequence of aerobic metabolism (53). If left unchecked, these species will destroy the metal centers of key metabolic proteins, oxidize the cysteine residues of proteins that warp their optimal structure, and even react with iron in the cell through the Fenton reaction to generate hydroxyl radicals that can cause DNA mutations (41). Cells ameliorate this stress with processes that require energy, including the production of superoxide dismutase and catalase proteins, among others, and by shunting some of their electrons back onto misoxidized proteins through the glutathione and/or thioredoxin pools (41).

Despite the acute challenges presented by transitioning from an oxic to an anoxic environment (54), the energy budget of aerobic cells that settle into an anaerobic survival state contracts as these cells no longer have to pay the costs of repairing oxidative damage. Inefficient fermentations may, under these circumstances, be perfectly adequate, particularly if the flux through these pathways is sufficient to meet metabolic demands. Furthermore, the number of proteins required for fermentation is significantly lower than for respiration; in E. coli, the protein cost is estimated to be around one-half (55), a significant fitness advantage given the high energetic cost of producing and maintaining proteins. From this viewpoint, the relative energetic deficit of fermentation begins to look more like a trade-off. It is similar conceptually to the proposed trade-off associated with rRNA copy number, where having multiple copies of the rrn operon could allow for increased rRNA production, reduced lag, and increased maximal growth rate but imposes an energy burden that becomes a detriment to fitness under energy-limited conditions (56, 57). Be that as it may, our ability to assess long-term fitness trade-offs between aerobic and fermentative metabolism is limited by our lack of understanding of how cells allocate energy when it is limited, which may have more to do with maximizing survival than growth. Indeed, recent modeling suggests that the relative fitness of a respiration versus fermentation strategy in an oxygen-variable niche is nuanced (58).

Beyond needing to sustain sufficient catabolic output under oxygen-limited growth arrest, there is the issue of rewiring anabolic pathways in the absence of oxygen. For example, in E. coli, the aspartate oxidase NadB, an essential enzyme in pyridine nucleotide biosynthesis [i.e., NAD(P)⁺] is known to reduce molecular oxygen to redox-cycle its solvent-exposed flavin but is also capable of using fumarate for this purpose during anaerobic growth (59). A similar example can be found in pyrimidine biosynthesis under anoxic conditions, specifically in the membrane-bound dihydroorotate dehydrogenase PyrD. This enzyme catalyzes a non-NAD(P)H-mediated redox reaction (the oxidation of dihydroorotate to orotate), which, under oxic conditions, proceeds by shuttling electrons into the quinone pool and therefore ultimately onto oxygen (60). Under anoxic conditions, E. coli transfers these electrons from the quinol pool onto fumarate through either fumarate reductase or the reverse reaction of succinate dehydrogenase (60). Despite having the ability to anaerobically respire, Campylobacter jejuni cannot grow in the absence of at least small amounts of oxygen because its ribonucleotide reductase, which catalyzes the essential conversion of ribonucleotides to deoxyribonucleotides, requires oxygen (61). These examples underscore that energy may not always be the limiting factor for growth when an organism fails to grow anaerobically; essential anabolic pathways can sometimes require oxygen. While we know enough to ask increasingly mechanistic questions about how oxygen-limited cells survive growth arrest, much remains to be learned about both catabolic and anabolic fitness determinants.

ANAEROBIC SURVIVAL MECHANISMS IN P. AERUGINOSA

P. aeruginosa is a ubiquitous bacterium that can cause devastating chronic lung infections in cystic fibrosis patients and chronic wounds in immunocompromised patients. These infection environments can experience prolonged states of hypoxia or anoxia (9, 10), motivating research into how P. aeruginosa survives when oxygen is scarce. Here, we focus on fermentative survival mechanisms rather than denitrification, as denitrification in P. aeruginosa has been well reviewed elsewhere (62, 63). After it was discovered that P. aeruginosa could generate ATP anaerobically through arginine catabolism (22), Eschbach et al. (11) showed that P. aeruginosa can survive, but not grow, under anoxic conditions through a mixed-acid fermentation of pyruvate to lactate, acetate, and succinate (Fig. 2A). P. aeruginosa was able to survive anoxic conditions in minimal pyruvate medium for up to 18 days before any loss in viability was detected, while those under anoxic conditions without a carbon source began to lose viability within 3 days (11). These fermentation and survival phenotypes were dependent on the genes ldhA, pta, and ackA, which comprise the reductive and oxidative arms of the fermentation. Interestingly, the cells did not survive anoxia when provided with glucose as a carbon source, suggesting the fermentation might be oxidatively limited and therefore unable to support flux through glycolysis under these conditions. Yet survival on glucose is sometimes possible, provided flux through ATP-yielding pathways is sufficiently high (23). How might this be achieved?

FIG 2.

Illustration of how RAMs can facilitate oxidative metabolism. (A) Working model of phenazine-based anaerobic survival catabolism in Pseudomonas aeruginosa (23). (B) A general way RAMs (e.g., phenazines) may serve as an oxidant to permit substrate-level phosphorylation in diverse aerobes. (C) In a microbial community, RAMs have the potential to benefit both producers and nonproducers by serving as shared oxidants. The bold black curved arrow represents RAM production. Acetyl-CoA, acetyl coenzyme A; PHZ, phenazine; RAM, redox-active metabolite; OX, oxidized; red, reduced.

Separately, P. aeruginosa is known for its production of colorful redox-active metabolites called phenazines (64). With the exception of early work by Friedheim on the potential for phenazines to serve as accessory respiratory pigments (65), the bulk of the literature has portrayed phenazines as antibiotics (66). This view of their biological activity stems from studies that showed that the phenazine pyocyanin (PYO) reduces oxygen, producing superoxide and hydrogen peroxide in vitro and in E. coli cultures (67, 68), where an aerobic growth inhibition was observed at concentrations as low as 1 μM PYO. These studies found that phenazines are reduced by E. coli and therefore generate a massive amount of oxidative stress in the presence of oxygen.

In contrast, P. aeruginosa can tolerate up to at least 100 μM PYO (69). An early hypothesis by our group was that such an alternative electron acceptor would be most beneficial when access to oxidants is limited, such as in the anoxic core of biofilms (70). To test this hypothesis, Wang et al. developed an anoxic survival assay, demonstrating that electrodes poised to oxidize phenazines could facilitate phenazine extracellular electron transfer (EET) that promoted cells’ survival (71). Under these conditions, at least 2 orders of magnitude more cells survived in the presence of phenazines after 7 days of anoxia. A prior study found that phenazines significantly rebalance the reduced NADH/NAD⁺ pool in the early stationary phase of dense cultures (72). This hinted that the metabolic mechanism behind the survival phenotype might be connected to pyruvate fermentation through fermentative redox balance. Data supporting this hypothesis were provided by Glasser et al., who showed the ackA and pta genes necessary for pyruvate fermentation were also necessary for the phenazine-mediated survival phenotype (23). They further demonstrated that this phenotype was unaffected in an ldhA mutant and that the cells could survive by fermenting glucose, overcoming the metabolic limitation identified for this mutant by Eschbach et al. (11) and concluding that phenazine-mediated EET was sufficient as the oxidative arm of fermentation (Fig. 2A). Importantly, phenazine-mediated EET enabled the production of a PMF, which correlated with survival (23).

An outstanding question is whether this mechanism enhances survival in the center of a biofilm. Like most secondary metabolites, phenazines are secreted by cells during the transition into stationary phase and afterward (72), a state that would be induced as a biofilm grows and cells are suffocated by their outer neighbors (5, 73). Under such conditions, oxygen would be available at a distance, similar to the electrode at a distance in the survival assay, and phenazine-mediated EET could close the gap. This hypothesis is supported by the observation that a Δphz strain that cannot produce phenazines forms highly wrinkled biofilms, which maximizes their access to oxygen (74). Indeed, phenazines and/or the redox state of the cell is directly linked to biofilm morphogenesis through the regulatory protein RmcA that modulates the production of matrix components responsible for these wrinkles (75, 76). Intriguingly, phenazines are retained in biofilms by binding to extracellular DNA, which helps facilitate the EET system as a whole (77). Deletion of certain membrane-bound cytochrome oxidases curtails phenazine reduction at depth in the colony biofilm system, impacting both the extracellular redox potential within the colony and its morphology (78). Whether the redox benefits achieved by recycling phenazine using these enzymes and/or others in the cytosol (79) permit energy generation via oxidative or substrate-level phosphorylation remains to be determined.

ANAEROBIC SURVIVAL IN STREPTOMYCES COELICOLOR

P. aeruginosa is not the only bacterium that might derive a catabolic benefit from its secondary metabolites. Indeed, such metabolites are made by diverse bacterial phyla, many of which reside in the soil (80). Of these, the Streptomyces genus is the best studied. Streptomycetes exhibit a three-stage life cycle, which includes outgrowth as a mycelial mat (which, like a more conventional biofilm, becomes anoxic in deeper layers [81]), development of aerial hyphae, and the release of spores from these hyphae. In the absence of oxygen, streptomycetes must rely on alternative energy conservation strategies besides aerobic respiration to survive. As mentioned above, one such alternative is nitrate. S. coelicolor has three distinct nitrate reductase operons (nar1, nar2, and nar3), each active in different stages of its life cycle (82). Interestingly, none of them can sustain growth in the absence of oxygen (82).

While spores are typically assumed to be metabolically inactive, some minimal amount of metabolic activity must be maintained—or at least, the potential for metabolic activity—so as to permit revival. While spore metabolism is outside the scope of this review, we simply note that in S. coelicolor, during the spore stage, the electron transport chain is poised to respire nitrate only when oxygen is absent, as nitrite production by spores is reversibly and rapidly inhibited by the presence of oxygen (83). Notably, this does not apply to the other life stages of S. coelicolor, where nitrate reduction can occur simultaneously with oxygen (82). While it is unclear why nitrate reduction is insufficient to support anaerobic growth of S. coelicolor, it has been inferred that nitrate reduction contributes to charging the PMF and/or redox balancing the cytosol in the absence of oxygen to contribute to cell survival (84). For a detailed summary and analysis of denitrification in S. coelicolor, please see the recent review by Sawers et al. (84).

However, nitrate respiration cannot be the only anaerobic energy-conserving mechanism for S. coelicolor, as a Δnar123 mutant has no anoxic survival deficit compared to survival of the wild type (WT) in the presence of nitrate (82). This is in contrast to another actinomycete, Mycobacterium tuberculosis, where nitrate respiration enhances survival of nonreplicating cells under hypoxic conditions (85). Indeed, even germinated S. coelicolor spores can survive at least 3 weeks of anoxia in nitrate-free medium, indicating that anaerobic survival is robust in the absence of nitrate and not limited to the spore state (16). What other metabolic mechanisms could S. coelicolor be using to provide the energy needed to survive anoxia?

Acid secretion has been observed in various Streptomyces species. As one example, S. coelicolor is known to secrete pyruvate and alpha-ketoglutarate during the transition from exponential to stationary phase (86). Acid secretion during aerobic growth often suggests rectifying redox or carbon imbalance and is sometimes referred to as overflow metabolism (55, 87). To our knowledge, only a single report suggests a canonical fermentation via lactate secretion in the species Streptomyces griseus growing under restricted aeration (88). This is interesting considering the S. coelicolor genome codes for a gene whose protein shares 44% amino acid identity to the E. coli lactate dehydrogenase (LdhA) (16). Although there are not yet any published studies that tested for evidence of fermentation in S. coelicolor under anaerobic survival conditions, these data hint at the possibility.

Furthermore, there are curious correlations between S. coelicolor’s secondary metabolites and phenazines. Just as phenazine-null P. aeruginosa mutants develop into wrinkled colonies, a similar morphology is observed in S. coelicolor colonies when actinorhodin or undecylprodigiosin production is interrupted (75, 89). In S. coelicolor, production of actinorhodin and undecylprodigiosin is upregulated by iron, phosphate, and also ammonium limitation (90, 91). While in this review we have focused on phenazines’ anaerobic survival benefits, phenazines are also involved in liberating iron from minerals in the form of ferrous iron (92, 93), and phenazine biosynthesis is correspondingly also upregulated by iron limitation (94) and by phosphate limitation (95). These correlations loosely hint at the possibility that actinorhodin and undecylprodigiosin share some of the same physiological functions as phenazines. Consistent with this notion, both phenazines and actinorhodin strongly activate SoxR (96), an iron-sulfur cluster transcription factor specifically activated by redox-cycling compounds (97, 98). Recently, it was demonstrated that actinorhodin is also a redox-active antibiotic (99), suggesting that actinorhodin might also confer a survival benefit to S. coelicolor like phenazines do for P. aeruginosa.

A GENERAL HYPOTHESIS FOR RAM-MEDIATED SURVIVAL

Phenazines and actinorhodin are examples of a broad class of secreted redox-active small molecules produced by many thousands of microbial species (64). For example, flavins can stimulate EET by Shewanella oneidensis and Listeria monocytogenes (100, 101). Similarly, diverse quinones can stimulate EET by Klebsiella pneumoniae, Lactococcus lactis, Shewanella oneidensis, and Sphingomonas xenophaga (102–105). Bioinformatic analysis indicates that hundreds of species from the phyla Actinobacteria and Proteobacteria contain phenazine biosynthetic clusters (64, 106). However, not all of these putative molecules are necessarily extracellular electron shuttles. Until they are proven to function as such, we prefer the label “redox-active metabolites,” or RAMs, to describe these types of metabolites.

The phylogenetic ubiquity of RAMs makes us wonder whether there could be an underlying unity to their physiological functions, such as helping diverse aerobes achieve redox balance during oxygen limitation. Many of these bacteria contain the genes needed for fermentative metabolisms, though flux through these energy-conserving pathways appears to be limited by the aerobes’ ability to balance the NADH/NAD⁺ pool. RAM reduction presents a potential mechanism by which microbes could relieve themselves of the excess reducing equivalents that would accumulate under oxygen-limited conditions, thereby increasing flux through redox-limited pathways (Fig. 2B and C). If flux were stimulated in this way through pathways that were otherwise inactive, we could consider those pathways to be “unlocked.” Such unlocking could be achieved by endogenous RAMs, as is the case for P. aeruginosa (23), or by exogenous RAMs for an organism that cannot produce them. Given that this mode of survival would be possible only for a non-RAM producer in the presence of another organism, what is metabolically possible for a single species may be fully understood only by studying its community.

However, it is well established that to many species, RAMs are toxic in the presence of ambient oxygen. In large part this is due to their ability to divert electrons from the cell and shuttle them in an energetically uncoupled fashion to oxygen, generating toxic superoxide radicals in the process (107). This interpretation is consistent with the fact that phenazines become toxic even to P. aeruginosa under certain conditions. Meirelles and Newman recently likened phenazines to a double-edged sword, with the relative dominance of benefit or detriment conferred by phenazines depending on the type of energy limitation the cell is experiencing (69). For example, under carbon-limited aerobic conditions, where the cell is predominantly oxidized, phenazines cause up to 75% of P. aeruginosa cells to lose viability within 26 h (69). As discussed earlier, this negative effect contrasts with oxygen-limited conditions, where the cell is predominantly reduced and phenazines contribute to energy conservation by redox balancing the cytosol (Fig. 2A). Our working model is that if the redox pool of the cell is predominantly oxidized, further oxidation by phenazines will be toxic, whereas if the cell is predominantly reduced, oxidation by phenazines will be largely beneficial.

We suspect this “rule” for phenazines holds even for other aerobes. The lack of any species-specific reduction of phenazines is an old observation (65, 108), and we have also observed reduction of phenazines by non-phenazine-producing species, including Staphylococcus aureus, Agrobacterium tumefaciens, Burkholderia cepacia, and Achromobacter spp. While the metabolic consequences of promiscuous phenazine reduction under anoxic and hypoxic conditions remain incompletely characterized, we have yet to encounter a species that cannot reduce phenazines, which is perhaps not surprising given their broad reactivity with redox-active cofactors in common enzymes (79, 107). For example, the pyruvate and alpha-ketoglutarate dehydrogenases have been implicated in vitro as phenazine reductases, which is consistent with studies of other redox-active compounds that can oxidize flavoproteins at rates equal to or higher than those for their physiological acceptors (109). In E. coli, multiple groups have found that synthetic phenazines such as phenazine methosulfate (PMS) and neutral red react with the quinone pool in the inner membrane (98, 110). Genetically, a cytochrome oxidase subunit has been shown to contribute to phenazine reduction in P. aeruginosa biofilms (78). Furthermore, phenazines are transported through well-conserved pumps (111) and specifically activate the well-conserved SoxR transcription factor, which contains an iron-sulfur cluster (98, 112). Generalizing from these examples, it stands to reason that RAMs likely react with conserved redox cofactors and therefore have the potential to be coupled to NADH oxidation through multiple pathways.

As mentioned above, exactly how phenazine reduction is accomplished while minimizing damage is not fully understood in P. aeruginosa. While phenazines are made intracellularly, they are rapidly pumped out of the cell via conserved efflux pumps (111). Upon phenazine reentry into the cell, if the cytochrome oxidase subunit previously mentioned (78), or an another enzyme, were able to reduce them on the periplasmic side of the inner membrane, this would provide a means to spatially circumvent undesirable reactions in the cytosol without sacrificing the benefits of promoting intracellular redox balance. Exploring the mechanisms and subcellular localization of RAM cycling in any species that benefits from RAMs is a priority for future research.

OPPORTUNITIES FOR DISCOVERY

Theoretically, all microorganisms must fall somewhere along a continuum of being unable to tolerate any oxygen exposure to being entirely reliant on oxygen for survival. In this minireview, we have shared a few examples of the gray areas along this spectrum to argue that they are the norm in the microbial world and represent opportunities for discovery. We respectfully contend that the classical growth classifications of “obligate” and “facultative” can effectively, albeit unintentionally, skew attention away from and even contradict how these organisms survive when they are not growing. Recognizing that many bacteria, from P. aeruginosa to S. coelicolor, are still sometimes classified as “obligate” aerobes despite their ability to conserve energy anaerobically (11, 17, 18, 82) exemplifies this problem.

As more discoveries are made of anaerobic metabolisms that support survival or slow growth within aerobes, the physiology underpinning these microbes’ ecological success will become clearer. We hope to have convinced the reader that an underappreciated aspect of their success is their ability to endure periods of energetic famine, such as when oxygen is limiting. Some vegetative cells can survive with little energy for years (113, 114), a natural phenomenon whose wonder should not go overlooked. Which genes, metabolic modules, community compositions, and environmental factors determine survival potential? How common are they across species and niches? Exploring the mechanisms of survival physiology not only will fill gaps in our understanding of the minimal energy requirements of life but may also reveal new targets to combat chronic infections and improve the engineering of metabolism decoupled from growth. On the whole, it will bring us steps closer to understanding the heart of microbial resiliency.