Reactive Oxygen Species in Pathogen Clearance: The Killing Mechanisms, the Adaption Response, and the Side Effects

1Reactive oxygen species (ROS), including hydrogen peroxide (H2O2), hydroxyl radical (OH-), singlet oxygen (1O2), and superoxide anion (⋅O2–), are produced by the pathogen itself (endogenous) as a byproduct of aerobic respiration; they can also be encountered in the host environment (exogenous). ROS have been called “double-edged swords of life” in pathogen clearance. First, since ROS can directly damage DNA, lipids, and proteins, they are thought to be the weapon used by both antibiotics and the host immune system. However, controversies challenge the paradigm. Second, successful pathogens exploit ROS for their own adaption. This minireview aims to summarize the mechanisms of ROS-mediated killing by either antibiotics or the host, examples of pathogens that cope with the ROS conditions, the potential side effects of ROS in pathogen clearance, and novel antibacterial strategies centered around ROS.

2In general, antibiotics are thought to kill microbes through interaction with specific intracellular targets. However, in 2007, Collins group proposed a novel mechanism of quinolones-induced killing by the generation of endogenous hydroxyl radicals. Since then, ROS have been shown to play a central role in lethality of numerous classes of bactericidal antibiotics, regardless of their specific targets. However, the role of ROS in antibiotic lethality became controversial and has been challenged, since antibiotics can kill in the absence of ROS. The paradox contained in these statements can be solved by the idea that killing can derive from the primary damage of antibiotic or from a secondary, lethal stress response mediated by ROS. If primary damage with specific targets is severe enough, it can result in death directly. Otherwise, primary damage stimulates a pathway that leads to ROS accumulation as secondary damage. New evidence for this hypothesis was provided in Escherichia coli. described a novel experimental system in which the role of secondary ROS damage could be tested in isolation. In this case, even after complete removal of quinolones, ROS accumulation and cell death continued to occur, indicating that when secondary ROS damage exceeds a critical threshold, it becomes a self-amplifying process and the terminal stage when bacteria responds to antibiotics.

3However, how primary damage leads to ROS accumulation remained unclear. Since target-specific damage occurs in antibiotic killing, pathways leading to ROS accumulation most probably have drug-specific context as well. While different studies suggest that primary drug-target damage may activate processes such as envelopes stress response and programmed cell death, leading to ROS accumulation, most evidence is provided for aminoglycosides. Misfolded proteins induced by gentamicin first insert into the membrane in E. coli and lead the activation of the response regulator ArcA through CpxA, the envelope stress response sensor. ArcA then activates the tricarboxylic acid (TCA) cycle enzymes, leading to dysfunction of TCA cycle and the hyperactivation of respiration. Similar dysfunction of TCA cycle can also be observed in quinolones and β-lactams treatments, suggesting that such metabolic flux could be a shared mechanism pushing the cell into a state that provokes oxidative stress in antibiotic treatments. While β-lactams may directly activate envelope stress response by affecting membrane integrity, specific triggers for quinolones remain to be worked out. Several studies found that programmed cell death mediated by YihE kinase and MazF toxin was linked to a ROS cascade in quinolone treatment. Recently, additional evidence suggested that quinolones disrupted the nucleotide pool, leading to the increase of ATP demand. The increasing ATP demand elevates TCA cycle activity and cellular respiration and enhances antibiotic lethality.

4Exogenous ROS, as an antimicrobial weapon wielded by phagocytes, are generated from NADPH oxidase (NOX2) in response to microbe recognition. However, it has been surprisingly difficult to figure out exactly how phagocytic ROS production suppresses microbial growth. Some indirect evidence support the notion that phagocytic ROS directly kill pathogens. Chronic granulomatous disease (CGD) is a genetic disorder in which patients lack functional NOX2 protein and therefore are associated with impaired respiratory burst. Indeed, a severe neutrophil killing defect was reported in CGD patients. The host susceptibility to various pathogens including Salmonella enterica, Staphylococcus aureus, and Burkholderia cepacia in CGD patients highlights the importance of the respiratory burst to infectious diseases. However, whether host-derived ROS could directly kill pathogens is still a matter of debate, since several studies argued that intracellular ROS level in phagosomes was insufficient to kill pathogens.

5Alternatively, ROS generated by NOX, as a signal, promote pathogen elimination via activating a variety of innate and adaptive mechanisms in the host cell. For example, ROS are commonly believed to be required in autophagy. NOX-derived ROS are indispensable for the recruitment of the autophagy protein light chain 3 (LC3), promoting antibacterial autophagy. In addition, during infection, neutrophils and phagocytes can activate an additional antimicrobial mechanism, referred to as neutrophil extracellular traps (NETs). NETs, formed by chromatin and associated proteins, trap and kill various extracellular pathogens. NET formation requires the production of ROS and is impaired in NOX2-deficient neutrophils. Another study suggested that fosfomycin, a broad-spectrum antibacterial agent, could enhance NET-mediated killing of S. aureus via NOX2-dependent ROS accumulation in mouse model.

6In addition to the NOX complex, the mitochondrion is another cellular source of ROS in immune cells. Upon macrophage activation, mitochondrial conditions favor reverse electron transport in the electron transport chain and thus generate ROS. Mice deficient in proteins responsible for generation of mitoROS are highly susceptible to infections caused by Salmonella typhimurium and Listeria monocytogenes. During methicillin-resistant S. aureus (MRSA) infection, mitoROS were generated and delivered directly to the phagosome by mitochondrial-derived vesicles (MDVs) in a Toll-like receptor signaling-dependent manner. Of note, like NOX-mediated ROS, which can modulate IL-1β expression, mitoROS can also modulate the antimicrobial functions of innate immune cells by regulating production of multiple cytokines both in vitro and in vivo.

7The systems in bacteria that regulate the expression of antioxidant defense networks have been extensively reviewed. The ROS response is under the control of the master regulators. Transcription factors such as OxyR (i.e., in S. enterica, Francisella tularensis, and Porphyromonas gingivalis), PerR (i.e., in S. aureus and Bacillus subtilis), OhrR (i.e., in B. subtilis and Mycobacterium smegmatis), and SoxRS (i.e., in E. coli and S. aureus) can be activated by direct oxidation of their sensor proteins and then adjusting the bacterial response appropriately. Although the particulars vary among different species, in general, these regulons regulate genes required for antioxidant defense, including superoxide dismutase, catalase, thioredoxins, heme biosynthesis machinery, glutathione reductases, ferric uptake regulator (Fur), ferritin, and bacterioferritin. In addition, iron homeostasis is also critical to mitigate redox damage induced by Fenton reaction. Therefore, in pathogenic bacteria (i.e., E. coli, S. aureus, and Salmonella), the iron-sensing transcriptional repressors, such as Fur and DtxR (diphtheria toxin repressor), can also be utilized to sustain redox homeostasis by controlling the expression of genes encoding iron acquisition systems and iron-dependent enzymes.

8Apart from antioxidant defense systems, metabolism remodeling also plays a pivotal role in mitigating oxidative damage (Figure 1). Metabolic adaptions can reduce oxidative burden by retarding respiration. For example, the glyoxylate shunt (GS) is an anaplerotic reaction of the TCA cycle developed in numerous species, which bypasses two NADH-generating steps. In B. cepacia, GS genes were upregulated in cells surviving aminoglycoside treatment. Similarly, in Mycobacterium tuberculosis (MTB), GS enzyme isocitrate lyase deficient mutants were significantly more susceptible than wild-type strain toward isoniazid, rifampicin, and streptomycin. Consistent with these observations, the Collins group found that glyoxylate could serve as a direct biochemical inducer of aminoglycoside tolerance via reducing cellular respiration. Meanwhile, TCA cycle intermediates, such as fumarate, significantly potentiate tobramycin lethality.

9Replenishment of antioxidants can also be achieved by metabolic modulation. Pentose phosphate (PP) pathway is an important target to mitigate ROS damage, since NADPH is the cofactor for antioxidant systems. In E. coli, by increased abundance of glucose-6-phosphate dehydrogenase, the metabolic flux can be rerouted toward the PP pathway, leading to increased ROS tolerance.

10Ketoacids including α-ketoglutarate (KG) and pyruvate can undergo non-enzymatic decarboxylation in the presence of ROS, generate non-toxic byproducts, and thus alleviate the ROS damage. Bacteria like E. coli and Pseudomonas fluorescens tend to pool ketoacids both inside the cell and in the extracellular matrix under ROS conditions. For example, the increased generation of KG can be achieved by the modulation of TCA cycle enzymes like isocitrate dehydrogenase when using citrate, glucose, and malate as carbon sources. It can also be achieved by the deamination of glutamate when glutamate or histidine is supplied as the carbon source. When exposed to H2O2, P. fluorescens can go through a metabolic reconfiguration with enhanced activity of substrate-level phosphorylation as well as impaired activity of the TCA cycle, leading to evident pyruvate synthesis with glucose or glycerol as the carbon source.

11In addition to antioxidant defenses, activations of the oxidative defense regulators are also required for full virulence in pathogens. For example, OxyR contributes to the virulence of E. coli and Pseudomonas aeruginosa. Similarly, SoxRS has been shown to act as a positive regulator in Salmonella pathogenicity island (SPI)-2 mediated virulence in S. enterica. In S. aureus, after oxidation of the sensor protein by ROS, the redox-signaling regulator AirSR positively regulates the biosynthesis of staphyloxanthin (STX), an important virulence factor of S. aureus. MntR, a member of the DtxR family, is required for S. aureus pathogenesis via maintaining manganese homeostasis.

12In host conditions, virulence modulation can be achieved by ROS indirectly. During ROS generation, phagocytes simultaneously produce glutathione (GSH) to sustain intracellular redox homeostasis. Certain intracellular pathogens have even evolved to exploit the host GSH for adaptions. In Burkholderia pseudomallei, virulence is completely dependent on the expression of a type VI secretion system (T6SS). During the exit from phagosome, B. pseudomallei senses host GSH via sensor protein VirA, which then activates the expression of the T6SS. In L. monocytogenes, the extracellular host GSH can trigger the production of bacterial GSH. Both host GSH and bacterial GSH bind to the master virulence regulator PrfA and function as an allosteric activator.

13While pathogens can be eliminated by ROS, it is quite clear that certain pathogens exploit ROS to coordinate metabolism to thrive. For example, S. typhimurium evidently makes use of host-derived ROS during intestinal inflammation. ROS generated by phagocytes convert thiosulfate to tetrathionate, which in turn can be used as a respiratory electron sink by S. typhimurium, allowing it to outcompete the native microbiota. Another example comes from E. coli. In the intestinal lumen, oxygen from the epithelium collides with sulfide generated by luminal bacteria, potentially generating H2O2 through direct reaction. Also, lactic-acid bacteria excrete H2O2 as a direct metabolic product. Under such circumstances, cytochrome c peroxidase (Ccp) regulated by OxyR allows E. coli to employ H2O2 as a terminal oxidant for respiration. Helicobacter pylori is well known as a ROS-inducing gastric pathogen. It utilizes chemotaxis to seek sites optimal for efficient colonization. Recent studies showed that ROS could be sensed in H. pylori by the chemoreceptor TlpD. Host oxidants hypochlorous acid (HOCl) could act as a chemoattractant by reversibly oxidizing TlpD that inactivates the chemotransduction signaling complex. While H2O2 could act as a chemorepellent which initiates chemotaxis through TlpD to promote gastric gland colonization.

14Recently, the radical-based theory in pathogen clearance has been questioned for the protective role of ROS against antimicrobial killing. ROS can directly damage DNA, leading to genetic mutations. Indeed, ROS produced by non-lethal antibiotics induce mutations in specific antibiotic targets and promote the development of multi-drug resistance. If ROS is a shared mechanism in antibiotic lethality, then it is to be expected that protection against ROS can be one of the shared traits for bacterial resistance against antibiotics as well. Indeed, bactericidal antibiotic mediates de novo acquisition of resistance, which then provides protection against ROS accumulation upon exposure to a different type of antibiotics. In addition, it was recently demonstrated that non-lethal exposure to H2O2 boosted evolvability of bacterial populations by enhancing survival under oxidative stress.

15ROS also act as a signal for gene transformation. H2O2 is a metabolic product of certain oral streptococci such as Streptococcus gordonii under aerobic conditions. In H2O2-producing streptococci, the release of extracellular DNA (eDNA) is entirely dependent on H2O2. Also, numerous H2O2-non-producing species like Streptococcus mutans and Streptococcus pyogenes release eDNA in a H2O2-dependent manner, suggesting that ROS serve as an important environmental signal in oral biofilm. Intriguingly, a recent study shows that non-antibiotic pharmaceuticals including non-steroidal anti-inflammatory drugs (NSAIDs) and the lipid-lowering drug also induce significant ROS accumulation in Acinetobacter baylyi, which is closely related to the enhanced transformation of antibiotic resistance genes (ARGs).

16ROS are key weapons in host cells, while recent evidence suggests that host ROS induce antibiotic tolerance during infection. ROS generated by macrophages target TCA enzymes aconitase and succinate dehydrogenase and coerce S. aureus into a metabolic state with reduced respiration, which is incompatible with the killing mechanism of most bactericidal antibiotics. The persister is an extreme case of tolerance for the high levels of multi-drug tolerance. Host ROS can also modulate persister formation in E. coli. The activation of redox regulators SoxRS results in increasing expression of the AcrAB-TolC multidrug-resistant (MDR) pump, which in turn lowers the fluoroquinolones concentration and promotes antibiotic tolerance.

17Together, these examples indicate that ROS may potentiate the emergence of bacterial resistance and tolerance (Figure 2). Although there is currently no clinical evidence proving this hypothesis, contradictory evidence favoring these potential side effects have been indicated in several studies. Early study suggested that S. aureus internalized in leucocytes isolated from patients with CGD (with impaired respiratory burst) were more susceptible to rifampicin, when compared with wild-type leucocytes. Also, elevated levels of oxidation products are related to MTB infection in pulmonary tuberculosis (PTB) patients. While there is scarce demonstration that the use of antioxidants can increase pathogen burden, a study showed that the antioxidant resveratrol reduced Serratia marcescens burden in mice. Surprisingly, the use of antioxidant N-acetyl cysteine (NAC) as an adjuvant to directly observed treatment short course significantly caused early sputum negativity in PTB patients.

18The efficacy of antibiotics has been endangered by the rapid emergence of resistant pathogens. Therefore, new approaches to fight against pathogens, including antimicrobial photodynamic therapy (aPDT) and cold atmospheric plasma (CAP), have been suggested as efficient alternative approaches. Although their antibacterial efficacy also centered around ROS, the modes of action for these new techniques are distinct from conventional antibiotics, as ROS-mediated damage in such therapies is the primary stress-induced damage. For example, aPDT uses photosensitizers (PS) to generate ROS upon irradiation by visible light at specific wavelength, and CAPs are partly ionized gases producing a reactive mix by interacting with oxygen, generating a cocktail of ROS. Due to direct redox-active properties, aPDT and CAP can cause multi-target oxidative damage to pathogens as well as direct oxidation of polysaccharides in biofilm. Due to their multi-target mode of action and large quantities of ROS production which overwhelm the antioxidant defenses in pathogens, the additional advantage of these new techniques is the lack of development of resistance mechanisms. As novel ROS-inducing strategies, aPDT and CAP showed excellent potential to tackle difficult-to-eradicate infections as alternative treatments in wound healing, dental cure, and food decontamination.

19ROS are attractive weapons to kill pathogenic microbial cells. However, ROS, as a double-edged sword, should be regulated with care since under non-lethal ROS, certain pathogens have evolved delicate mechanisms to utilize ROS for their adaption to thrive. Also, numerous pathogens can use ROS as a stepstone for the evolution of antibiotic tolerance and resistance. Efforts to target bacterial adaptive pathways, as well as the use of novel ROS-inducing antibacterial strategies, will be promising approaches for antibacterial therapy.

20HL, XZ, YH, BL, LC, and BR: conception/design of the work, agreement to be accountable for all aspects of the work. HL, LC, and BR: drafting the work. XZ, LC, and BR: final approval of the manuscript to be published. All authors contributed to the article and approved the submitted version.

21The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.