Redox organization of living systems

Abstract

Redox organization governs an underlying simplicity in living systems. Critically, redox reactions enable the essential characteristics of life: extraction of energy from the environment, use of energy to support metabolic and structural organization, use of dynamic redox responses to defend against environmental threats, and use of redox mechanisms to direct differentiation of cells and organ systems essential for reproduction. These processes are sustained through a redox context in which electron donor/acceptor couples are poised at substantially different steady-state redox potentials, some with relatively reducing steady states and others with relatively oxidizing steady states. Redox-sensitive thiols of the redox proteome, as well as low molecular weight redox-active molecules, are maintained individually by the kinetics of oxidation-reduction within this redox system. Recent research has revealed opposing network interactions of the metallome, redox proteome, metabolome and transcriptome, which appear to be an evolved redox response structure to maintain stability of an organism in the presence of variable oxidative environments. Considerable opportunity exists to improve human health through detailed understanding of these redox networks so that targeted interventions can be developed to support new avenues for redox medicine.

Graphical Abstract

Iťs a strange situation, wild occupation,

Living my life like a song—Jimmy Buffett

The 2022 Society for Free Radical Research—Europe Award is a considerable honor for me, and I share this honor with my friends, colleagues and co-workers, who make redox research so lively and fulfilling. In this conceptual minireview, I summarize the central role of redox network interactions in biological organization and function. I specifically conclude that experimental approaches designed to control “artifacts” in redox biology research may have obscured understanding of critical mechanisms of redox organization which support stability of living systems within highly variable oxidative environments. Detailed articles by others should be consulted for more thorough reviews of mitochondrial function and bioenergetics [1–3], redox signaling [4–7], oxidative stress [8–11], antioxidants [12–14], selenium [15–18], redox proteomics [19–21], redox transcriptomics [22, 23], redox reactions in epigenetic control [24–27], disease mechanisms [28–31], aging [32–34], rejuvenation [35], inflammation [36, 37] and cancer. [38–41] Importantly, the joint 2022 SFRR-E meeting with the Plant Oxygen Group introduced me to Ronald Mittler, PhD (University of Missouri, USA) and the remarkable progress in plant redox biology. He and I recently extended the concepts of the Redox Code, which Helmut Sies and I developed for metazoans [42], to outline the Redox Code of Plants.[43] Remaining needs exist to address the principles of redox organization for archaebacteria, eubacteria and yeast.

In this review, I provide an operational description of redox organization synthesized from a spectrum of earlier redox biology and biochemistry research [42] along with recent studies of metallomics, redox proteomics, transcriptomics, metabolomics [44–50]. My transition from reductionist cause-effect models of oxidative stress to network and systems approaches was highly influenced by the failure of free radical scavenger trials to provide health benefit [51] and an early perspective on network medicine [52]. Most critical to this transition, studies of the exposome show that humans can experience a million or more exposures in a lifetime, and studies of transcriptomics and metabolomics show that single oxidative exposures can cause changes in expression of thousands of genes and thousands of metabolites. In practical terms, this complexity requires network and systems approaches to understand the redox organization. Using the “Golden Mean” of redox homeostasis [53] as a foundation, detailed knowledge of the steady-state redox status of the ensemble of redox couples will advance understanding of the mechanistic details which enable defense against oxidants and electrophiles while also maintaining bioenergetics, metabolic organization, macromolecular structure and function, redox signaling and many other functions.

Bioenergetics lies at the heart of redox organization, with essentially all cellular processes in aerobic organisms linked directly or indirectly to relatively high-flux O₂-dependent cellular bioenergetics reactions. At the same time, protein structures and functions are controlled by the epiproteome [54], a series of post-translation modifications (e.g., oxidation, phosphorylation, methylation, acetylation) dependent on redox systems. These include reversible and irreversible modifications of the proteome dependent upon O₂, H₂O₂ and other oxidants, as well as modifications secondarily dependent upon oxidative phosphorylation to maintain ATP, S-adenosylmethionine (SAM), acetyl-CoA and other key precursors. Most vitally, an operational description of the redox-dependent epiproteome includes a spectrum of enzyme-catalyzed “on-off” switching mechanisms alongside non-enzymatic reactions determined by protein-specific amino acid reactivities at localized, ambient oxidant and metal ion concentrations. I believe that the interface between evolved high-specificity redox signaling mechanisms and the “background” lower specificity reactions represents a major frontier for redox research. As outlined below, omics-scale molecular data suggest that the redox organization of living systems in variable oxidative environments occurs through opposing response structures of the transcriptome and metabolome[55] that evolved to maintain stability of the functional redox proteome. More detailed elucidation of the nature of these network structures and their responses to environmental exposures is needed to enable transformative approaches for redox medicine.

Redox organization provides a context for living systems to survive and thrive within a variable external oxidative environment

The Redox Code [42] complements the Genetic Code and Central Dogma in providing the molecular logic for evolution and development of complex organisms within an oxidative environment (Fig 1). The Genetic Code is mostly static in nature and used to store and transfer information; along with Central Dogma and other principles (e.g., Histone Code), it defines how information is preserved, replicated, and used to direct biosynthesis of proteins (Fig 1A). The Genetic Code and Central Dogma do not, however, provide the logic for how biological systems extract energy, maintain order, develop and mature to enable reproduction, and survive and thrive within a variable and sometimes hostile environment. Hence, additional steps are needed.

Figure 1. Molecular Logic of Redox Organization of Living Systems.

A. The central logic for fundamental characteristics of life is provided by the Genetic Code and Central Dogma for information storage, transmission to progeny and use to support protein synthesis. However, logic for other aspects of life, such as energy capture and use to maintain metabolic organization are not addressed. B. Evolution of complex aerobic metazoan life occurred in association with a dramatic rise in atmospheric O₂. Emerging redox theory connects this increased atmospheric O₂ to the evolution of dynamic redox response mechanisms in complex multicellular organisms. C. These redox response mechanisms, along with epigenetics, provide the essential logic for the dynamic responses needed for a static genetic code to deliver multiple cell types with divergent properties. D. The logic for redox organization of bioenergetics, metabolism and protein structures and activities is provided by the four principles of the Redox Code[42].

A dramatic increase in atmospheric O₂ appears to have been the driver and enabler of complex multicellular metazoan evolution (Fig 1B). Higher O₂ created a greater threat for oxidative damage and also provided opportunity to develop more efficient mitochondrial respiration. Even with recognition of this, however, the molecular logic for elaboration of complex multicellular development remains undefined (Fig 1C). Thus, the Redox Code (Fig 1D) was provided as a set of principles governing redox organization in metazoans. In this, a critical first step involves extraction of energy from the environment and use of this energy to support cellular needs. Although many components are involved, NADH/NAD⁺ and NADPH/NADP⁺ redox couples are of central importance because these are maintained at different steady-state redox potentials (E_h) to simultaneously support catabolic and biosynthetic processes. Relatively high-flux reactions maintain E_h for NADH/NAD⁺ and NADPH/NADP⁺ couples at near-equilibrium with respective metabolic fuels. Studies in rat liver show that the NADH/NAD⁺ couple is relatively stable in cytoplasm at about −241 mV and in mitochondria at about −318 mV [56]. NADPH/NADP⁺ is relatively stable in cytoplasm at about −393 mV and in mitochondria at about −415 mV [56]. These values are poised to enable use of NADH/NAD⁺ to maintain ATP/(ADP + Pi + H) while simultaneously using the more reducing NADPH/NADP⁺ steady states to sustain biosynthetic reactions, defense against oxidative stress, and dynamic steady states of the redox proteome.

The NADH/NAD⁺ couple supports both mitochondrial and glycolytic production of ATP, and both mitochondrial and non-mitochondrial pathways maintain NADPH/NADP⁺ to support biosynthetic and biochemical defense mechanisms (Fig 2). Mitochondria use diverse substrates to support ATP production and support maintenance of NADPH supply through nicotinamide nucleotide transhydrogenase (NNT), while glucose in the cytoplasm serves as a common precursor to support ATP production from glycolysis along with NADPH supply from the pentose phosphate pathway (Fig 2A). The fundamental nature of the relatively stable NADH/NAD⁺ and NADPH/NADP⁺ couples for redox organization is evident from the interconversion of steady-state redox potentials (E_h) between redox couples and chemical potentials from ATP hydrolysis (ΔG = −nFΔE_h) and ion distribution and movement across biological membranes, as expressed by the membrane potentials for H⁺, K⁺, Ca²⁺ and other species (ΔG = −nFΔC for chemical gradients; ΔG = −nFΔψ for transmembrane potential (Fig 2A). Redox interconversion with other major energy currencies provides a central mechanism to harmonize and stabilize cellular energetics (Fig 2A, center). Rapid electron transfer reactions to maintain NADH/NAD⁺ and NADPH/NADP⁺ pools also provide a critically important context for bioenergetic and metabolic organization of living systems. This redox organization is extensive, with redox reactions governing high energy intermediates as well as pyrimidine biosynthesis for nucleic acids, CDP-choline for complex lipids and UDP-sugars for complex carbohydrates; amino acid homeostasis and nitrogen balance through oxidative deamination and reductive amination; and many non-protein functions of amino acids such as support of 1-carbon metabolism, GSH antioxidant defenses and synthesis of biogenic amines (Fig 2A). In this organizational structure, maintenance of the ubiquinol/ubiquinone couple in the mitochondrial inner membrane at a more oxidized steady-state E_h than NADPH/NADP⁺ and NADH/NAD⁺ provides a central redox hub for oxidative phosphorylation while also maintaining a stable context for organization of fatty acid metabolism, amino acid metabolism, and a key upstream step in pyrimidine metabolism catalyzed by dihydroorotate dehydrogenase (Fig 2A).[57, 58]

Figure 2. Overview of Systems to Maintain Redox Context for Life. A. Redox organization of bioenergetics and metabolism.

The first principle of the Redox Code addresses the central need to maintain a stable bioenergetic and metabolic context for life. In aerobic metazoans, this is achieved through parallel mitochondrial and cytoplasmic NADPH/NADP⁺ and NADH/NAD⁺ systems to support central bioenergetic currencies and a global system to maintain organization and activities of the metabolome (Center). Redox organization is essential for most cellular functions, illustrated by some important examples: amino acid and nitrogen metabolism, nucleotide biosynthesis, non-protein functions of amino acids such 1-carbon metabolism, GSH detoxification mechanisms, biogenic amine biosynthesis, NO biosynthesis, and fatty acid, complex lipid and complex carbohydrate metabolism. B. Redox organization of the proteome. The second principle of the Redox Code addresses the central switching mechanisms linking bioenergetics and metabolism with protein structure and function. The redox proteome is maintained by a series of redox hubs with different steady-state redox potentials (E_h) in different cellular compartments. In black font, the poise of the NADPH/NADP⁺ couple is more reducing than that of the NADH/NAD⁺ couple, and this difference enables the former to support biosynthetic and defense systems while the latter supports catabolism and oxidation of metabolic fuels to maintain ATP and other high-energy currencies. In red font, the redox proteome and detoxification systems are supported by the NADPH/NADP⁺ couple balancing oxidation by H₂O₂-dependent systems. C. Pleiotropy of thiol systems in redox organization. Organizational networks of the redox proteome have been extensively elucidated through mass spectrometry and targeted molecular and cellular methods. Cysteine thiols are maintained by NADPH-dependent systems through GSH-dependent and thioredoxin-dependent systems (shown in blue), counteracting oxidation by H₂O₂ and other oxidants (shown in red). GSH is maintained by NADPH-dependent glutathione reductase (GR) and operates through glutaredoxins (Grx) and other GSH-dependent enzymes, such as GSH S-transferases (e.g., GSTpi). Thioredoxins (Trx) are maintained by NADPH-dependent thioredoxin reductases (TR). Examples of proteins with different subcellular localizations undergoing reversible oxidation at specific cysteine residues are provided for proteins with standard abbreviations for names (Based upon Go et al [101]). D. Pleiotropy of H₂O₂ signaling. Examples of biologic systems responding to oxidant signaling in different subcellular compartments were recently described in detail [98] and are summarized here. Based upon Sies and Jones [98].

Redox organization depends upon additional redox couples poised at more positive (oxidizing) steady state E_h

While considerable attention has been given to the central NADH/NAD⁺ and NADPH/NADP⁺ redox couples, less attention has been given to the existence of other important organizational redox couples that are maintained with more oxidizing steady-state E_h. Ubiquitous thiol/disulfide systems are among the most critical for structural organization. Bulk measurements of the E_h of the most abundant cytosolic systems show that the 2 GSH)/GSSG couple is poised at approximately −220 mV in differentiated cells and −260 mV in proliferating cells (Fig 2B).[59] The mitochondrial (2 GSH)/GSSG redox potential is more reducing, at about −300 mV. Mechanisms maintaining the steady-state E_h values for these redox systems are not fully understood and involve NADPH/NADP⁺ through a relatively small number of NADPH-dependent H₂O₂ producing systems along with thiol reductases (glutathione disulfide reductase, thioredoxin reductases) [60]. The steady state E_h values for the central thiol/disulfide couples are also relatively stable, with active site disulfide of mitochondrial thioredoxin-2 at about −360 mV while cytosolic thioredoxin-1 is about −280 mV and nuclear steady state is about −300 mV (Fig 2B). [59] The steady-state E_h for (2 cysteine)/cystine is about −150 mV. Estimates of the steady-state redox potential in the cisternae of the secretory pathway indicate that values are about −150 mV. Live imaging using luminescent or fluorescent redox probes enable measurements of dynamic responses in living cells.[10] The essential takeaway is that although specific thiol/disulfide systems are dynamic, they are organized with a remarkably simple overall design to provide a context to link bioenergetics and metabolism with protein organization and function (Fig 2B).[42]

Bioenergetic and metabolic organization establish a context for macromolecular organization and function

This linkage of bioenergetics and metabolism to the operations of the proteome provides a central logic to organize protein structure, trafficking, macromolecular interactions and activities by changing their physical and chemical properties (Fig 2C). Because there are more than 200,000 cysteine residues in proteins, and 20% or more are at sites accessible to reversible modification, this provides an abundant and versatile network for integration and coordination of complex biologic processes. Along with other specific post-translational modifications, this yields a tiered structure in which a redox context for organization of the epiproteome is maintained by energy-linked modifications (phosphorylation, acetylation, etc.) and a small number of thiol-coordinating centers balanced by a small number of oxidant coordinating centers (Fig 2C). Among these oxidants, molecular O₂ is almost universally present in the low micromolar range; and in the presence of O₂, H₂O₂ appears to always be present in the low nanomolar range, with increased generation for use in redox signaling (Fig 2D). Collectively, this constitutes an extensive network with remarkable simplicity in organizational structure.

In the hierarchy of these processes, the stability and order are maintained through an integrated redox network coupled to the redox environment (Fig 2C, 2D). As indicated above, an evolutionary perspective is especially important because multicellular animal life evolved in response to a dramatic rise in atmospheric O₂ (Fig 1B). In the presence of trace levels of redox active transition metals[61], such as Mn, Fe, Co, Ni, and Cu (Fig 3A), both non-protein and protein thiols reduce O₂ to superoxide anion radical, which rapidly undergoes dismutation to H₂O₂ (Fig 3B). Trace levels of metal ions are always present in the environment, food and water of animals so that in the presence of O₂, low concentration of H₂O₂ is part of the redox context of life. Because redox-active metals also catalyze reduction of H₂O₂ by thiols, a steady state H₂O₂ concentration exists in the presence of redox-active metals, thiols and O₂ (Fig 3C). With protein thiols, reactivity of a thiol is dependent upon proximal amino acids. Thus, in the presence of trace levels of redox-active metals, a steady-state concentration of H₂O₂ and E_h of the thiol/disulfide system will occur due to the collective reaction rates of each, balanced by the rates of metal ion-catalyzed thiol-dependent H₂O₂ reduction (Fig 3C). Within this background steady-state H₂O₂, evolution of controlled NADPH oxidases to generate H₂O₂ set the stage for a network of specific thiol switching mechanisms in which H₂O₂ generation rates are uncoupled from H₂O₂ elimination rates (Fig 3D). This enabled use H₂O₂ in protein activation/deactivation switching mechanisms, providing the logic for control of protein functions in time and space as needed for advanced metazoan evolution (see Fig 1C). Coupled with patterned changes in the epigenome, reproduction of complex organisms follows simple logic to use redox signaling, bioenergetics and metabolism to develop multicellular organization and specialized functions (see Fig 1C, 1D).

Figure 3. Redox System Stability with Transition Metal Ions.

A. Examples of transition metals with redox activities in the biologic range (Based upon Nies [61]). B. Reactions catalyzed by Fe illustrate transition metal-catalyzed thiol reduction of O₂ to H₂O₂ and H₂O₂ to water. C. In systems such as aerobic cells with capability to maintain thiols, steady-state concentration of H₂O₂ and E_h for the thiol/disulfide couples will be achieved based upon the reactions i and ii in Panel B. D. Presence of thiol systems for H₂O₂ elimination along with a family of NADPH oxidases for H₂O₂ generation resulted in an uncoupling of the H₂O₂ generation and elimination as in Panel C. With this evolution, redox organization of protein thiols was released from the inherent coupling provided by the non-enzymatic metal-ion catalyzed system. This enabled evolution of specific redox organization networks as illustrated in Fig 2C and 2D. Additional discussion of GSH-dependent systems is available [102].

Redox theory for genome-exposome interactions

A schematic connecting the genome, central dogma and the external concentrations of the exposome highlights the centrality of the redox proteome and redox metabolome (Fig 4A). To test this experimentally, we used untargeted redox proteomics and untargeted metabolomics of mice responding to the toxic metal, Cd.[62, 63] Although Cd is not redox active, Cd binds avidly to Se and inhibits antioxidant selenoenzymes, especially mitochondrial thioredoxin reductase-2. Network analyses of redox-sensitive Cys residues in proteins with metabolic signals showed that the most dominant community of interactions centered on proteins that function in β-oxidation of fatty acids (Fig 4B). Associated metabolites included multiple acyl-CoAs and acyl-carnitines, and stable isotopic palmitate tracer studies showed that Cd inhibited β-oxidation of palmitoyl-CoA and generation of shorter chain acyl-CoAs. The directionality of changes further showed that some metabolites increased while others decreased in association with oxidation of protein Cys residues (Fig 4B). Thus, within the mitochondrial redox response structure associated with Cd toxicity, changes in the redox proteome are associated with both increased and decreased concentrations of specific metabolites. In other studies mapping oxidation of protein cysteines in colon carcinoma HT29 cells, Ingenuity Pathway Analysis showed that protein Cys mapped to functional pathways according to steady-state levels of oxidation (Fig 4C,D).[64] Specifically, proteins functioning in lipid metabolism, small molecule metabolism and cancer had greater ratios of thiols to non-thiol forms (Fig 4C), while proteins functioning in cell maintenance had more oxidized steady-states (Fig 4D).

Figure 4. Central Role of the Redox Proteome in Redox Organization.

A. Effect of external oxidative exposures on central dogma places the redox proteome as a central “shock absorber”. B. Integration of redox proteomics and metabolomics data from livers of mice exposed to a toxic dose of Cd show that the most central hub centered on metabolites and proteins associated with fatty acid β-oxidation. Stable isotopic tracer studies showed that this network structure was associated with Cd-dependent inhibition of β-oxidation of palmitate.[63] C. Mass spectrometry-based redox proteomics of colon carcinoma HT29 cells under control cell culture conditions showed that about 14% of measured protein cysteine residues were partially oxidized, with functional pathways mapping according to percentage oxidation. The most reduced networks are illustrated by Ingenuity Pathway Analysis.[64] D. The most oxidized functional pathways under control aerobic conditions in HT29 cells were associated with cell maintenance activities.[64] Panels C and D are reproduced from Go et al.[64].

The opposing characteristics can be expected from toxicology studies, which show that protective responses occur along with adverse responses.[65] This is highly relevant to hormesis, wherein protective responses can be activated to decrease impact from subsequent toxic exposures. Such mechanisms appear to be especially important in mitochondria, with mitohormesis providing a major mechanism for long-term metabolic reprogramming in aging and age-related disease.[66, 67] Recent findings suggest that such responses could be important in metabolic reprogramming that can contribute to early onset cancers.[68] For instance, decreased serum levels of histidine, threonine and proline in the third trimester of pregnancy were associated with breast cancer diagnosis within 15 years.[69] These changes could reflect effects on nutrient sensing mechanisms that activate oncogenic cell survival pathways and are activated by global changes in the food chain.[70] In principle, other agents such as δ-valerobetaine, a metabolite of lysine that is produced by the intestinal microbiome, could also predispose or re-enforce oncogenesis by inhibiting mitochondrial activities and stimulating glycolysis.[70] A shift from mitochondrial to glycolytic energy metabolism (Warburg effect) often accompanies oncogenesis. Such redox-dependent mechanisms could thereby contribute to a concerning rise in early-onset cancers, now recognized to impact many cancer types, including breast, endometrial, prostate, thyroid, and renal cancers.[68] Such cancers are occurring globally at younger ages during recent decades, and this represents an important example of the specific need to elucidate underlying redox mechanisms to guide approaches to reverse this tragic development.

Network analysis of model system with systematic variation in manganese reveals opposing transcriptome-metabolome interactions

A critical challenge to understand complex redox mechanisms has become apparent through extension of the Cd-dependent redox proteome-metabolome approach to obtain redox proteome-transcriptome-metabolome interactions for manganese (Mn) dose response. Mn is an essential nutrient in mammals, and deficiency results in decreased mitochondrial superoxide dismutase-2 (SOD2) activity and increased oxidative stress. Oxidative toxicity also occurs occupationally from excess Mn exposure in welders, who can develop neurotoxicity symptoms similar to Parkinson’s disease.[71] Analyses of mitochondrial activities as a function of Mn dose in a neuroblastoma cell line (SH-SY5Y) revealed a distinct monotonic increase in oxidative stress along with a biphasic stimulation followed by inhibition of respiratory activity.[72] Network analyses showed transcriptome changes associated with each of the respiratory activities [73, 74], and when these were combined with non-targeted metabolomics, four communities emerged (Fig 5).[55] Importantly, each of these communities contained positive and negative associations linked individually to Mn, O₂ consumption rate, mitochondrial H₂O₂ production rate and cellular protein thiol content (Fig 5A, top).[55] When the positive and negative associations were visualized separately, the positive and negative associations appeared as mirror images (Fig 3), with each containing sets of positively correlated transcript-metabolite interactions and negatively correlated transcript-metabolite interactions (Fig 5A, bottom). The results directly demonstrate that with controlled, systematic variation of a redox active metal, the response includes a network composed of opposing transcriptome-metabolome correlations (Fig 5B). This leaves little doubt that simple unidirectional cause-effect models can be fundamentally wrong. None-the-less, functional pathways can be identified from both transcriptome (Fig 5C) and metabolome (Fig 5D) analyses. This reveals a critically important gap in knowledge of redox organization, i.e., while net electron flow must occur according to electrochemical gradients defined by thermodynamic laws, there must also be a hierarchy of interactions which enable an organism to respond to oxidative challenges, which can be highly variable in nature, while at the same time maintaining stable redox organization and function.

Figure 5. Redox Proteome-Transcriptome-Metabolome Interaction Reveals Network Structure with Mirrored, Opposing Correlations.

A. Application of a data-dependent community detection tool, xMWAS [103] to integrate omics data from a Mn dose response study in neuroblastoma SH-SY5Y cells showed that each of 4 communities had both positive and negative associations. See Fernandes et al [55] for details. B. Each community included specific transcriptome-metabolome correlations as illustrated for the community associated with Mn. C. Pathway enrichment analysis of transcripts showed diverse functions associated with Mn dose. D. Pathway enrichment analysis of metabolites showed diverse pathways associated with Mn dose. Figures reproduced with minor modifications from Fernandes et al.[55]

Antagonism between Cd and Se in human lung metallome and metabolome network interaction

To test for antagonistic redox interactions in human tissue, we examined metabolic associations with Cd and Se in human lungs. Cd inhibits selenoenzymes, and experimental studies in mice had previously shown that Se supplementation protects against Cd-dependent oxidative stress in the lungs.[75] Analyses of metabolic associations with Se and Cd in human lungs showed that hundreds of metabolites correlated with Cd and hundreds also correlated with Se (Fig 6A).[50] The importance of this is clear from the regional distributions of Cd and Se in surface soils (Fig 6B), which are considerable, and emphasizes that oxidant/antioxidant exposures must be included in efforts to understand human variations in disease. Most of the lungs with higher Cd concentrations had severe chronic obstructive lung disease and positive associations with inflammatory lipids, while lungs with higher Se were from non-diseased or young individuals with genetic causes of lung failure and had lower levels of inflammatory lipids.[50]

Figure 6. Divergent Mechanisms in Opposing Metal-Metabolome Interactions.

A. Analysis of Cd and Se interactions in human lungs showed that large numbers of metabolites were associated individually with Cd and Se, and a relatively small number of metabolites enriched in inflammatory lipids associated with both, but in opposite directions. Within each of the communities of metabolites associated with Cd or Se, large numbers of positive and negative associations were present. Reproduced from Smith et al [50] with minor modifications; see Smith et al [50] for details. B. Maps and statistical graphics available from the United States Geological Survey Data Series 801 show that large regional differences occur for Cd, Se and other elements. These emphasize the need to consider redox organization which enables maintenance of “homeostasis” despite such variable exposures in human redox biology. C. Phytochelatins are a large family of GSH-related metal ion chelators synthesized by plants to protect against toxic metals. D. Phytochelatins bind a range of metals and have diverse activities in metal transport, sequestration, and detoxification. Recent research shows that phytochelatins are present in humans, apparently absorbed from plant-derived human food. Associations of phytochelatin concentration with Se and other metals suggest unrecognized activities could contribute to human health. Panels C and D are reproduced with minor modifications from Jarrell et al.[76]

Pathway enrichment analysis revealed most pathway associations for Cd were distinct from those for Se.[50] Common pathways for Cd and Se were linked to inflammatory lipids with specific metabolites showing opposite directions of association (central region of Fig 6A connecting communities). The relatively small number of opposing associations with Cd and Se raises critical questions concerning the nature of redox network interactions when compared to the opposing transcriptome-metabolome interactions for the model system Mn dose-response described above (Fig 5). At one extreme, the Cd-Se interactions appear to be driven by strong and specific physicochemical interaction between Cd and Se, with most of the network structures for each occurring independently of the other (Fig 6A). At the other extreme in the experimental model with increasing Mn (Fig 5), large numbers of positive and negative associations were observed for both transcripts and metabolites. With this in mind, a closer look at the metabolic associations with Cd and Se shows that Cd had 379 positive and 806 negative metabolite associations while Se had 1443 positive and 1772 negative metabolite associations.[50]. Thus, even with very specific antagonistic interactions of Cd and Se, global associations occurring individually with Cd and Se included opposing redox network response structures. Hence, the available evidence indicates that within the ensemble of redox systems maintaining the Golden Mean of redox homeostasis, the redox organizational structure includes opposing subnetwork response structures.

Metal ion chelators, phytochelatins, are present in plant-derived human food

The recent discovery that phytochelatins are present in humans[76] raises important questions concerning the transport and distribution of trace levels of redox-active transition metals. Phytochelatins are a family of GSH-related γ-glutamyl-cysteine polymers (Fig 6C) with high affinities for metal ions. Root vegetables and fruits are particularly rich sources of phytochelatins, and a largely plant-based diet can result in 10 mg or more daily intake.[77] Mammals do not contain known phytochelatin synthetases but contain Cys-rich metallothioneins which bind toxic metals and are highly inducible in response to exposures. However, understanding of transport and distribution of trace levels of transition metals remains incomplete. Data show phytochelatin uptake by mammalian cells[78, 79], so the finding of phytochelatins in human urine, with correlations to Se and metals [76], indicates that additional mechanisms exist. Possible mechanisms include use of phytochelatins in a selective manner to bind, transport and control activities of diverse redox-active metals (Fig 6D).

An elaboration of this concept is that the entire proteome, with an abundance of ligands for transition metals (thiols, carboxylates, hydroxylates, peptide bonds, primary and secondary amines), could represent a variable affinity, high-capacity metal ion chelation system. Evolution of the Cys proteome is known to have resulted in lower Cys content in proteins than expected if Cys codons were used randomly.[80] With this in mind, the abundance and reactivity of oxidizable cysteine residues could have been selected through evolution to include structures that chelate and control different metal ion reactivities. If so, it may turn out that the extreme care that redox biologists use to prevent “artifacts” from trace metal contamination ironically could have impaired ability to detect such mechanisms. Indeed, if evolution enabled tolerance to highly variable metal ion exposures, there is also a possibility that diverse redox networks and redox relay systems could exist which have not been characterized. In line with this, the relative preponderance of Cys residues on protein surfaces, which has been proposed as a possible decoy system to protect critical catalytic and regulatory Cys residues [81], may be part of a system to assure redox stability. If present, there is an urgent need for characterization because network structures which maintain stability in the presence of environmental exposures can be expected to prevent detection of environmental exposures that potentiate disease mechanisms and/or interfere with cancer therapeutics.

Motivation for study of the proteome as a variable affinity metal ion chelation system is enhanced by recent dose-response studies of vanadium pentoxide (V⁺⁵), a common form of vanadium of environmental and occupational concern. Studies at low micromolar concentrations showed that V⁺⁵ is taken up by human lung fibroblasts and reduced to V⁺⁴ with oxidation of GSH and protein thiols.[82] This low-level exposure stimulated GSH biosynthesis, increased protein S-glutathionylation, and activated fibroblast cell senescence, a process associated with human pulmonary fibrosis.[83–85] Importantly, the experiments showed increased release of GSH and GSSG from cells. At the same time, V⁺⁴ was also released and non-enzymatically reoxidized to V⁺⁵, establishing a cellular-extracellular redox cycling system. Supply of V⁺⁵ to mice in drinking water similarly caused lung oxidative stress, mitochondrial dysfunction and collagen deposits characteristic of lung fibrosis. Of considerable interest, these reactions occurred with micromolar V⁺⁵, while evidence for mitochondrial oxidative stress and metabolic changes were evident at nanomolar V⁺⁵.[45, 82] The latter did not appear to be related to the profibrotic changes, but none-the-less, diverse hydration spheres of V⁺⁵ and stable ligand structures with thiols and hydroperoxides [86], as well as particulate forms[46], raise the possibility for different oxidative mechanisms in different subcellular compartments occurring at different orders of magnitude of exposure. Of critical importance for human disease, such mechanisms could be overlooked due to unanticipated opposing network interactions of adverse and protective responses.

A look to the future: Frontiers for redox organization research

Among the needs to understand redox network responses in redox organization, an acute need exists to define the nature of the opposing redox networks. The interpretation suggested from the Mn studies above, in line with contemporary molecular and cell biology, is that the proteome controls the transcriptome and the metabolome, and the transcriptome and metabolome each contribute to control of the proteome (Fig 7A). If these processes include specific and purposeful opposing reactions, then simultaneous changes in each direction could provide a mechanism to counterbalance responses to environmental changes. This means that a redox network absorbs oxidative challenges by having off-setting responses, i.e., a response in one direction is countered by a response in the opposite direction. This can be viewed similarly to control of glucose levels though opposing activities of insulin and glucagon. In recent examples given above, redox challenges imposed by either Cd, Se, Mn or V include opposing changes in transcription and metabolism.[45, 46, 48, 50, 55, 72, 74, 75, 82, 87, 88] Whether these are specific or non-specific within the context of a global redox network remains to be delineated. In either case, however, central hubs must exist, and identification of these hubs in association with disease response can be useful for identification of mechanisms and potential interventions. For instance, the finding that the Zdhhc11 transcript was a hub for Se antagonism of Cd toxicity [75] led to studies showing that increased protein S-palmitoylation occurs in Cd toxicity.[89] Mass spectrometry-based proteomics analysis further showed that a Cd transporter and several profibrotic proteins undergo S-palmitoylation in response to Cd, and pharmacologic inhibition of S-palmitoylation protected against profibrotic signaling [90]. Thus, even without knowledge of the specific opposing positive and negative elements of response, network analysis provided identification of a key mechanistic hub and led to an approach for therapeutic intervention.

Figure 7. Holistic View of Opposing Responses in Redox Organization.

A. Available data suggests that the redox proteome is a central component of a transcriptome-proteome-metabolome network with opposing response structures underlying the stability of redox organization in face of a variable redox environment. B. Diurnal variation of plasma cysteine/cystine E_h in humans could reflect the existence of opposing redox processes as part of mechanisms to maintain redox stability. See Blanco et al [91] for details. C. Adaptive redox control systems can account for cyclical variations in plasma cysteine/cystine E_h through 1) variation in dietary intake of cystine (CySS), 2) activation of Nrf-2-dependent increases in antioxidant systems, 3) ongoing reductive mechanisms in cells, and 4) extracellular signaling mechanisms which control cellular/extracellular redox communication. See Dennis et al [95] for details. D. Intrinsic proteomic mechanisms also appear to contribute to redox organization through structures such as PAS (Per-ARNT-Sim) domains. These domains were recognized as a structure enabling bacterial movement response to changes in O₂ or redox potential [97], and the presence of these structures could contribute to diurnal redox responses of peroxiredoxins to light, HIF-dependent responses to O₂, and diurnal responses to diet.

A second critical need is to revisit earlier studies showing diurnal redox oscillations.[91] An example of a 24-h redox oscillation in E_h of plasma cysteine/cystine is shown for humans (Fig 7B). This occurs through well-characterized redox control mechanisms in which an oxidative or electrophilic challenge modifies thiols in KEAP1, leading to release and translocation of the Nrf-2 transcription factor [14], increasing expression of genes related to GSH biosynthesis and other antioxidant defenses.[92–94] As previously discussed [95], this type of response can be considered an adaptive redox control mechanism (Fig 7C) because it serves to allow an organism to adapt to environmental exposures through changes in transcription and metabolism. Importantly, this includes both transcriptome and metabolic changes that vary in a temporal manner (Fig 7C). By having oscillations with off-set, time-dependent increase followed by decrease as reflected in E_h of the cysteine/cystine redox couple, this oscillation may reflect the adaptive system response to diurnal environmental stresses. In this example, the amplitude of oscillation differed for men and women and for age, reinforcing the concept that the redox oscillation could provide important information concerning the redox dynamics of health and disease.[42] A better understanding of this phenomenon could yield practical benefit as a measure of redox stability and adaptability, much in the same way as heart rate variability provides a measure of health.

A third critical need exists to revisit molecular O₂ as an oxidant in aerobic organisms. Specifically, the PAS domain (Per-Arnt-Sim domain)[96, 97] in hypoxia inducible factor (HIF) could reflect an intrinsic redox control mechanism linked to O₂ (Fig 7D) that complements the adaptive redox control mechanism (Fig 7C). Redox control through PAS domain proteins could ensure redox stability in the presence of changes in light, O₂ and variable diet and environmental exposures (Fig 7D). Thus, through PAS domain proteins, O₂ may serve more widely in signaling than currently recognized. This could complement the transition of O₂ to reactive oxygen species by trace levels of redox-active metals and yield pleiotropic O₂-dependent mechanisms much like that currently understood for H₂O₂.[98] Thus, even though the reactivity of O₂ can be several orders of magnitude less than H₂O₂, this could be relevant because the tissue O₂ concentration (10 to 20 μM) is much higher than H₂O₂ (low nanomolar). O₂-dependent enzymes are known to catalyze oxidation of proline and lysine in proteins, as well as demethylation of histone proteins, and Cu-dependent and flavin-dependent thiol oxidases are known.[60] The critical importance of O₂ signaling is emphasized by the role of HIF-1α in early embryogenesis.[99] HIF-1α has ongoing O₂-dependent degradation in the presence of prolyl hydroxylase activity, and during hypoxia, rapid accumulation, translocation to nuclei and binding to co-activator ARNT results in profound changes in gene expression. Given that so much of redox biology and pathology remains poorly understood, and Herculean efforts are used to protect against metal ion- and O₂-dependent artifacts[9, 10], perhaps it is time to question the meaning of O₂-dependent artifacts. After all, prior to the demonstration of H₂O₂ in mammalian systems 50 years ago [100], H₂O₂ in mammalian systems was thought to be an artifact.

In summary, recent findings show that opposing response structures of the proteome, transcriptome and metabolome are common in biologic systems and could support redox stability in the face of oxidative challenges. If this is generally true for redox biology, then approaches for hypothesis-driven research must be refined to avoid unidirectional cause-effect analyses which can miss these opposing responses. The observed network responses reviewed here were obtained using experimental models with relatively low environmental concentrations of transition metals. Because metal ions can enhance O₂- or H₂O₂-dependent rates of thiol oxidation by orders of magnitude, one must question whether important stabilizing mechanisms for redox organization with variable metal ion exposures have evaded detection because experimental protocols prevent their detection by eliminating or controlling activities related to low levels of transition metals. With comprehensive redox proteomics, transcriptomics and metabolomics tools now available, redox biology faces a new frontier: to elucidate details of network structures which maintain redox stability within our variable oxidative environments. The entire redox proteome could include variable-affinity metal chelation sites with evolved redox reactivities embedded within a complex redox-relay system serving to dampen environmental impacts, just as the air dampens the impact of a sonic boom or gentle waves dissipate the splash of a rock tossed in a peaceful garden pool.

Highlights.

• Opposing redox network interactions maintain stability in variable oxidative environments

• Network interactions exist for the metallome, proteome, transcriptome and metabolome

• Direct molecular antagonism can be discriminated from opposing network structures

• Diurnal redox oscillations could reflect opposing networks in redox stability

• Understanding of redox networks could support new avenues for redox medicine

Abbreviations:

Cd

cadmium

Cys

cysteine

CySS

cystine

HIF

hypoxia-inducible factor

Mn

manganese

SAM

S-adenosylmethionine

Se

selenium