Review

Reactive Oxygen Species–Mediated Mitochondria-to-Nucleus Signaling: A Key to Aging and Radical-Caused Diseases

See allHide authors and affiliations

Science's STKE  25 Apr 2006:
Vol. 2006, Issue 332, pp. re3
DOI: 10.1126/stke.3322006re3

Abstract

Mitochondria-generated reactive oxygen species have been implicated as a common feature that connects aging of organisms and age-related diseases. Efficient elimination of these radicals by antioxidants correlates with increased life span. Understanding how the mitochondrion signals to the nucleus to regulate antioxidant proteins might be a key to aging processes and treatment of human diseases.

Introduction

Disruption of mitochondrial functions has been implicated in more than 40 known diseases, including atherosclerosis, ischemic heart disease, cancer, diabetes, and neurodegenerative diseases such as Alzheimer’s disease, Parkinson’s disease, Huntington’s disease, and amyotrophic lateral sclerosis (1). Most of these diseases are linked to an increase in mitochondrially generated reactive oxygen species (mROS), which can damage macromolecules such as mitochondrial or nuclear DNA, alter or activate cellular signaling pathways, or cause apoptotic or necrotic cell death (27). For example, in humans a strong correlation has been found between ROS-mediated accumulation of aggregated and altered proteins and the appearance of age-related neurodegenerative diseases (810). Oxidatively altered macromolecules including proteins, lipids, nucleic acids, and carbohydrates also accumulate during aging. Thus, both aging and the development of age-related diseases can be influenced by insufficient detoxification and with it a lack of ability to replace oxidatively damaged macromolecules over time. This may be due to inefficient sensors that detect increases in cellular ROS concentrations, or to an increased rate of ROS production that exceeds the scavenging capacity of cells (Fig. 1).

Fig. 1.

Sensors of ROS can lead to the expression of genes whose products are protective or can induce cell death or cellular senescence. ROS can induce DNA damage or lead to the oxidation of macromolecules, depending on radical, origin, and concentration. The accumulation of cellular defects can induce cell death or senescence and results in an aging phenotype or contributes to age-related diseases. ROS can be decreased by dietary antioxidants or by up-regulation of antioxidant proteins. The expression of protective genes is regulated by ROS sensors, which activate signaling pathways that lead to the detoxification of cells from ROS and as a result increase longevity.

The free radical theory of aging, as formulated by Denham Harman, is supported by the observation that the life span of most organisms is roughly proportional to their metabolic rate and thus to the rate at which the organism generates mROS (11, 12). Under normal conditions, the mitochondrial respiratory chain generates ROS as a by-product of cellular energy production. These ROS are rapidly detoxified by various antioxidant proteins. In a wide spectrum of animal species, dietary antioxidants or caloric restriction, as well as chemical antioxidants or increased expression of antioxidant proteins, can lower mROS production, which translates into an extension of life span (5, 1319).

On the cellular level, the role of ROS in the induction of cell death and its contribution to aging and age-related diseases has been intensively investigated (3, 20). However, little is known about the protective signaling cascades that are activated by oxidative stress under normal metabolic conditions. These signaling pathways are activated by mitochondrial or cytoplasmic proteins that detect an increase in ROS and lead to the increased expression of genes encoding proteins responsible for cellular and mitochondrial detoxification. Cellular antioxidant proteins are encoded by nuclear genes (21). Mitochondria primarily generate superoxide, a nondiffusible oxygen radical. This separates the source of radicals and the location of gene regulation; thus, mitochondria-to-nucleus signaling cascades must exist to detect mitochondrial superoxide generation and regulate cellular detoxification.

What are the mitochondrial sensors for oxidative stress? How might they relay signals from the mitochondria to the nucleus? Which protective genes are activated? It is also not known whether these sensor systems become insensitive or inactive over time or under pathological conditions, which could allow a subtle increase of ROS and damaged macromolecules to cause aging and age-related diseases. The identification of these sensor systems, and the development of activators for these signaling cascades or mimetics of their antioxidant products, might reveal strategies to extend life span in humans and to prevent neurodegenerative or other oxygen radical–caused diseases.

Generation and Regulation of Mitochondrial Reactive Oxygen Species

The mitochondrial respiratory chain is the major source of ROS. Other cellular sources for oxygen radicals are xanthine oxidase, located in the peroxisomal matrix or membranes, and NADPH (reduced nicotinamide adenine dinucleotide phosphate) oxidases, located at the plasma membrane (22).

Mitochondria-generated ROS have their origin at complex I and complex III of the electron transport chain, which are localized at the mitochondrial inner membrane (Fig. 2) (23). At these complexes, ubisemiquinones (UQH•) donate electrons to oxygen and generate superoxide (Fig. 3). From the sites of generation, superoxide is released into the mitochondrial matrix (70 to 80%) and into the intermembrane space (20 to 30%) (24). Superoxide produced into the intermembrane space passes through the outer membrane of the mitochondria through the mitochondrial permeability transition pore (MPTP) (25). Components of this pore such as the voltage-dependent mitochondrial anion channel (VDAC, mitochondrial porin), localized in the outer membrane, allow leakage of superoxide into the cytoplasm (25, 26). Detoxification of superoxide at both locations (the mitochondrial matrix and the cytosol) is mediated by superoxide dismutase (SOD) enzymes and leads to the generation of hydrogen peroxide (H2O2) (Fig. 3). The manganese SOD (MnSOD) is localized in the mitochondrial matrix, and the copper/zinc SOD (CuZnSOD) is primarily a cytosolic protein (27, 28).

Fig. 2.

Generation of mROS and mitochondria-to-nucleus signaling. Superoxide is generated at complexes I and III of the electron transport chain at the mitochondrial inner membrane and is released into the mitochondrial matrix or the intermembrane space. SOD enzymes degrade superoxide to H2O2. Both superoxide and H2O2 presumably act as second messengers and can lead to the activation of ROS sensors, which can regulate protective signaling to the nucleus to regulate antioxidant genes.

Fig. 3.

Superoxide generation and degradation. The oxygen molecule can accept an additional electron to generate superoxide (O2), which is a more reactive form of oxygen by virtue of its radical structure. Formation of superoxide within the cell occurs at the mitochondrial inner membrane as a by-product of the electron transport chain, linking generation of cellular energy to regulation of longevity. Superoxide contributes to cellular and organismal aging as well as to radical-mediated diseases such as neurodegenerative diseases and cancer. Nuclear genes encode for protective proteins that detoxify cells from superoxide. SOD enzymes (MnSOD, CuZnSOD, and extracellular SOD) reduce superoxide to H2O2. H2O2 is the substrate for peroxiredoxins (Prx), which reduce H2O2 to water, as well as for other enzymes such as catalase and glutathione peroxidase (GPX).

Besides their function as regulators of cell signaling cascades, both superoxide and H2O2 can cause nonspecific DNA damage. A direct DNA-damaging function for H2O2 has been shown through its reaction with metal-binding histones, which results in the oxidative damage of nucleotides. Most of the DNA-damaging effects, however, are due to the generation of the hydroxyl ion (•OH) from H2O2 by the Fenton reaction or from superoxide by the Haber-Weiss reaction (29, 30). Hydroxyl ions are highly reactive with mitochondrial and nuclear DNA. In this context, for example, damage of mitochondrial DNA has been implicated in the onset of diabetes type II, and damage of nuclear DNA has been implicated in cancer development and progression (3, 10, 22).

H2O2 is a small, uncharged, nonradical ROS that, like superoxide, reacts with proteins and lipids. In contrast to locally present and nondiffusible superoxide, H2O2 can freely diffuse through cellular membranes and represents a bona fide long-range second messenger for signaling (31). There is evidence that H2O2 can activate many components of intracellular signaling cascades that are involved in cell survival, proliferation, differentiation, and cell death (32). Several antioxidant proteins can detoxify cells from H2O2. These enzymes are localized in the cytoplasm or at distinct organelles such as mitochondria and peroxisomes. Most prominent are catalase, which is localized in the cytosol; glutathione peroxidase (GPX), which is localized in the mitochondrial matrix, the outer membrane, and the cytosol; and the peroxiredoxins (Prx), a family of proteins that are localized at various locations within the cell, such as the peroxisomes (PrxV), the mitochondria (PrxIII), or the cytosol (PrxI) (3335).

ROS are produced locally at the mitochondria and can damage mitochondrial structures and DNA. Thus, to be fully effective, detoxification of cells from mROS generated during normal metabolism must occur at the mitochondria. Antioxidant enzymes that act on mitochondria-generated ROS are encoded by nuclear DNA and localize to the mitochondrial matrix or the outer mitochondrial membrane. Some of these detoxifying proteins are constitutively expressed; others are synthesized in response to oxidative stress, and expression of the genes encoding them is tightly regulated by transcription factors. The mitochondrial matrix protein MnSOD is an example of such a regulated antioxidant. The SOD2 gene is tightly regulated by the transcription factors FOXO3a and NF-κB, whose activity is tightly regulated as well. FOXO3a regulates the activation of this gene in quiescent cells in response to H2O2, and NF-κB regulates it in normal growing cells in response to mitochondria-generated superoxide (36, 37). Increased SOD2 expression in response to increased ROS indicates that monitoring systems or sensor systems exist within cells to trigger mitochondria-to-nucleus signaling and induce the expression of protective genes. These sensor systems and their mechanisms of activation are not well characterized. However, knowledge and manipulation of these signaling cascades might enhance efforts to prevent age-related diseases and to extend human life span.

Mitochondria-to-Nucleus Signaling Cascades as Regulators of Protective Genes

The response to an increase in mROS depends on the molecular background of cells or tissues and the concentration and type of oxygen radicals that are produced (22). But how do cells sense an increase in mitochondria-generated oxygen radicals, and how do these radicals activate protective signaling cascades?

A sensor for ROS is expected to be modified by radicals in a way that alters its interaction with binding partners or its activity. Superoxide, for example, regulates the activity of proteins containing iron sulfur centers (38), and H2O2 inactivates dual-specificity and tyrosine phosphatases by reversible oxidation of the sulfhydryl group in the active-site cysteine residue (3941). Similar oxidation of cysteine residues by ROS might also affect transcription factors and kinases (41, 42). Thus, cysteine-containing kinases or phosphatases located at the mitochondrial membranes or in the matrix might serve as radical sensors to relay oxidative stress into a transcription-activating signaling cascade.

Several phosphatases are localized at the mitochondria. For example, the phosphotyrosine phosphatase PTP1B is present in brain mitochondria and can be reversely inactivated by oxidation in response to epidermal growth factor stimulation (43). Further, the dual-specificity phosphatases TIM50 and PTPM1 are localized at the inner membrane (44, 45), and the phosphotyrosine phosphatases SHP-2 and PTP-D1 are localized at the intermembrane space or the outer membrane of mitochondria or at both locations (4648). The protein tyrosine phosphatase PTEN (phosphatase and tensin homolog deleted on chromosome 10) has been described to be reversibly inactivated by H2O2 (49). All of these phosphatases are potential candidates to be direct sensors for oxygen radicals at distinct locations within the mitochondria. Their oxidation and inactivation might lead to the activation of mitochondrially localized tyrosine or serine-threonine kinases. Tyrosine kinases of the Src family (Abl, Src) and serine-threonine kinases such as protein kinases C and D (PKC and PKD), Raf, and Akt are localized at the mitochondria, and their activity might increase through oxidative stress–mediated inactivation of some of these phosphatases. Src, for example, is regulated by tyrosine phosphorylation, and the inactivation of regulatory phosphatases leads to receptor-independent activation of Src. Additionally, Src also has a cysteine residue motif that may allow activation through its direct oxidation (41).

Signaling from the mitochondria to the nucleus requires the shuttling of the activated signaling proteins from the mitochondrial compartments to the outer membrane, the cytosol, or the nucleus. Mitochondrial matrix proteins such as p32 have been suggested to play bridging roles in interactions between the mitochondrion and nucleus. p32 has a mitochondrial targeting sequence, but it can (under poorly defined circumstances) translocate to the nucleus and thus might serve as a mitochondria-to-nucleus shuttle (50). This protein has also been described to bind to ROS-activated kinases such as PKC and PKD, which both can be located at the mitochondria (51). Other mitochondrially located kinases such as PINK1 (PTEN-induced kinase–1) have a mitochondrial targeting sequence; however, it is not known whether they are activated by oxidative stress (52). c-Jun N-terminal kinase (JNK) is a mROS-induced mediator of signaling. Although JNK is mostly considered to be proapoptotic, recent publications have shown that JNK phosphorylates forkhead transcription factors and extends life span in C. elegans and D. melanogaster (53, 54). The kinases that have been identified to be activated by mROS and that may participate in protective mitochondria-to-nucleus signaling are Src, Abl, PKCδ, and PKD. None of these proteins has a known mitochondrial targeting sequence, but they all localize to the mitochondria. Except for Src, the mechanisms of interaction and the exact mitochondrial localization of these proteins have not been determined. Src can be localized in the intermembrane space and at the outer membrane (48, 55, 56). One way Src can be recruited to the mitochondria is through binding of the mitochondrial adapter protein Dok-4 (downstream of kinase–4). In this context, Src has been implicated in the activation of the transcription factor NF-κB (57). Another adapter protein that facilitates the localization of Src to the mitochondria is AKAP121 (58). AKAP121 is also the anchoring protein for the protein tyrosine phosphatase D1 (PTPD1) that can activate Src through dephosphorylation of the kinase’s inhibitory tyrosine in the C terminus (48).

ROS-activated protective signaling cascades may increase expression of nuclear genes encoding antioxidant proteins, DNA-repair proteins, stress-regulated chaperones, and antiapoptotic proteins. Indeed, increased life span correlates with increased stress resistance and expression of stress response genes such as those encoding SOD and Hsp (heat shock protein) (2). Several transcription factors that can be activated by oxygen radicals have been implicated in the activation of antioxidant genes. Most prominent are FOXO transcription factors of the Forkhead transcription factor family and NF-κB.

In growing cells, FOXO3a is negatively regulated by insulin signaling pathways through the serine-threonine kinase Akt (36, 59). However, after serum deprivation or in senescent cells in which Akt is not active, FOXO3a is active and its activity can be increased by H2O2-induced oxidative stress (36). Reduced activity of insulin signaling pathways in D. melanogaster correlates with increased stress resistance and increased CuZnSOD and MnSOD activity (36, 60, 61). Under these conditions, FOXO3a activates the SOD2 gene encoding MnSOD and other protective genes such as catalase and GADD45 (6264). Because FOXO transcription factors are inactive under normal growth conditions, FOXO-mediated regulation of SOD might only have a role in the protection of senescent cells from ROS. However, the overexpression of FOXO3a in many species increases longevity. It was recently shown that FOXO transcription factors can be activated by klotho, a circulating anti-aging hormone that binds to cell surface receptors, suppresses insulin signaling, and increases resistance to oxidative stress at the cellular and organismal level (65). Caloric restriction increases longevity in various organisms (66). FOXO transcription factors might regulate life span by responding to caloric restriction—that is, by activating the SIR2 gene, which encodes the histone acetylase SIRT1 (67).

In normal growing cells, the SOD2 gene is regulated independent of FOXO3a by the transcription factor NF-κB, which can be activated by the mROS superoxide and H2O2 (21). Mitochondrial oxidative stress regulates the activity of NF-κB through a kinase cascade initiated by the activation of Src. In this signaling cascade, Src activates Abl and PKCδ, which both directly phosphorylate and activate PKD. PKD then activates NF-κB through the canonical IKK complex (21, 37).

Besides the oxidative stress–mediated regulation of the SOD2 gene, which occurs in senescent cells through FOXO3a and in growing cells through NF-κB, a connection between both transcription factors in the activation of other antioxidant genes (such as the catalase gene) has not yet been shown. Depending on the cellular and tissue context as well as presence of serum growth factors, FOXO3a and NF-κB might have synergistic or independent roles in the regulation of these genes. It is also not known whether mROS-activated NF-κB, like FOXO3a, increases organismal life span. However, NF-κB–mediated SOD2 activation points to a function of this signaling pathway in regulating aging.

As is becoming evident, portions of signaling pathways are activated by intracellular oxidants and regulate protective genes in a redox-dependent manner. These studies have advanced our understanding of how ROS might participate in diverse processes such as aging and the initiation of age-related disease. In addition, different mitochondrially generated ROS lead to different cellular responses. The understanding of how they control cellular fate by altering macromolecules, inducing cell death, or regulating signaling cascades that lead to detoxification is likely to provide new opportunities for pharmacological and genetic intervention in aging and age-related diseases.

Dysregulation of Mitochondrial Detoxification and Its Contribution to Aging and Disease

The occurrence of many diseases such as cancer, diabetes type II, neurodegenerative diseases, and cardiovascular diseases increases exponentially with chronological age. This suggests that both aging and age-related diseases might share common mechanisms (68). The rates at which organisms age and their susceptibility to age-related diseases appear to be determined by cellular metabolism and subsequent generation of ROS and the accumulation of damaged macromolecules (10). The failure of effective detoxification of cells from mROS by nuclear-encoded antioxidant proteins has been discussed as a major cause of aging and age-related diseases (5, 6, 10, 11, 6872). However, the development of an age-related disease is likely to be facilitated by additional events, possibly including a genetic predisposition or an acquired age-dependent accumulation of defects caused by pathologically high generation of oxygen radicals or by defective detoxification systems.

Inherited mutations in antioxidant enzymes can predispose individuals to certain age-related diseases. For example, 25% of the inherited cases of amyotrophic lateral sclerosis result from mutations in the CuZnSOD enzyme, the cytosolic form of the antioxidant protein SOD, indicating inappropriate detoxification from ROS as a factor that predisposes individuals for this disease (73). In some cases of inherited Parkinson’s disease, mutations were detected in PINK1, a kinase that is required for mitochondrial function (52, 74). Inherited forms of Alzheimer’s disease have been linked to mutations in the amyloid precursor protein (APP) or in gene products that are involved in APP processing. Various hypotheses exist regarding how APP induces injury, but brains of Alzheimer’s disease patients do show markers of increased oxidative stress. Thus, one possibility is that APP can exert its destructive function by promoting generation of mROS (9, 75, 76). However, cause and effect remain unclear, and other evidence indicates that increased formation of amyloid plaques can be a result of increased mitochondrial oxidative stress (77).

Besides genetic predisposition, an increase in oxidative damage to DNA or oxidative modification of proteins might be a cause for some age-related diseases. Hydroxyl radicals (•OH) generated from superoxide or H2O2 are reactive with DNA in a nonspecific manner, effectively increasing the random mutation rate within cells and thus promoting oncogenic transformation (78). Mutations in mitochondrial DNA that encode components of the oxidative phosphorylation system are commonly found in various cancers, including ovarian, kidney, liver, lung, colon, leukemia, and breast cancers (79). Mitochondrial dysfunction is one of the most characteristic features of cancer cells. ROS-mediated damage of mitochondrial or nuclear DNA has also been described for other diseases such as the development of diabetes type II. DNA damage also directly contributes to aging, and mutations in DNA repair genes cause a number of premature aging syndromes. Mutations in the mitochondrial DNA of aging tissue result in increased apoptosis and in the development of an aging phenotype independent of oxidative damage of other macromolecules (10, 11, 20, 80, 81). Further, the detoxification of mitochondria from ROS, by targeting of catalase to the mitochondria, results in increased organismal longevity (82).

The oxidation of proteins can accelerate their rate of degradation but can also cause protein aggregation (83, 84). Dysfunction in protein processing or degradation and the accumulation of misfolded proteins have important roles in the onset of Parkinson’s disease and Alzheimer’s disease. In Parkinson’s disease patients, for example, oxidative modification of α-synuclein by ROS leads to synuclein aggregation (8587). There is also evidence for deficient mitochondrial function in the electron transport chain in these patients (71, 74, 88). Similarly, in animal models of Parkinson’s disease, the progression of the disease is accelerated in animals treated with chemical inhibitors of the mitochondrial electron transport chain, which generate oxidative stress. Oxidized proteins also have an important role in the development of a range of cardiovascular and metabolic diseases. It is thought that cardiovascular risk factors lead to an increase of oxidative stress within the vessel wall, due to increased production of ROS in mitochondria and consequent oxidation of low-density lipoprotein (LDL). The deposition of oxidized LDL in the vessel wall initiates the development of atherosclerotic plaques in cardiovascular diseases. Mitochondrial damage correlates with the extent of atherosclerotic disease in mice and humans (10, 89), and mice lacking one copy of the SOD2 gene encoding the mitochondrial SOD show an increased rate of atherosclerotic plaque formation (89).

Finally, defective signaling and lack of up-regulation of antioxidant proteins can be a cause for ROS-mediated diseases. The deletion of antioxidant enzymes in mice increases the risk for cancer. Mice deficient in the antioxidant protein peroxiredoxin-I or heterozygous deleted for SOD2 exhibit increased spontaneous tumor formation (90, 91). Consistent with this finding, cancer cells in culture tend to produce more H2O2 than do normal cells (92), and the antioxidants SOD, glutathione peroxidase, and catalase show altered expression or activity in many tumor cell lines and tumor tissue (93). ROS are bona fide second messengers (94) that modulate the tumor phenotype by redox-dependent regulation of numerous cellular processes such as angiogenesis, proliferation, cell survival or death, cellular adhesion, and migration of tumor cells (9598). Thus, a decrease of mROS such as superoxide, H2O2, and hydroxyl radicals may be beneficial as a treatment for cancer (22).

Given the importance of mROS in aging and age-related diseases, and the potential to prevent the onset of these diseases by regulating cellular ROS levels, interest has increased in identifying oxidative stress–regulated signaling pathways and the mediators of ROS-regulated cellular effects.

Perspective

Mitochondria are the most important physiological source of oxygen free radicals, and at the same time they are a cellular target for free radical–mediated damage. Evidence also indicates that mitochondrial sensors exist to detect and relay an increase in mROS to the nucleus, resulting in increased expression of genes that encode detoxifying enzymes. However, the role of these redox-activated protective proteins in intercellular communication and gene expression is just starting to be understood as a vital mechanism in health and disease. Pharmacological agents that directly target these pathways might provide new tools to treat human diseases and increase human life span.

Therapies that aim to decrease oxidative burden in humans might offer ways to attempt to slow aging and ameliorate age-related diseases (5, 10). However, because age-related diseases also have a genetic component, it is not clear whether new antioxidant drugs that directly target mitochondrial oxidative stress will have a general preventive function or whether they can be used only to relieve symptoms. Some antioxidant drugs will probably only be effective in a discrete subset of age-related diseases. Likewise, they may only be efficient to target age-related diseases if applied in combination with other drugs such as L-DOPA (L-3,4-dihydroxyphenylalanine) in Parkinson’s disease or chemotherapy in cancer.

Different strategies can be applied to reduce exuberant ROS production in cells. Caloric restriction decreases superoxide formation and hence decreases the rate of aging. Because caloric restriction also decreases ATP formation, its clinical application is limited (14, 99). Another strategy is to supplement the diet with agents that depress superoxide radical formation by competing with oxygen for electrons from the respiratory chain (100) or to reduce oxygen radicals with dietary or chemical antioxidants. Antioxidants such as coenzyme Q10 (ubiquinone), which target mitochondrial dysfunctions, are efficacious in some mouse models of Parkinson’s disease and are in clinical trials (88). Some antioxidant dietary ingredients such as resveratrol show promise in regulating aging. Resveratrol increases longevity in several organisms by regulating SIRT (101103). Resveratrol is also implicated in the prevention of age-related diseases such as cancer because it regulates transcription factors that control tumor cell survival (22). Further, mimetics of specific cellular antioxidants such as the SOD/catalase mimetic EUK-134 can be used to decrease oxygen radicals and are already in clinical trials (104).

However, with most of these approaches it is difficult to target radicals at defined cellular locations. The targeted detoxification of ROS produced at a specific location within the cell or tissue might be important. Thus, an approach to manipulate signaling cascades that lead to the expression of defined organelle-targeted antioxidant proteins (such as PrxIII or MnSOD for mROS degradation) might allow control of specific ROS at particular locations within cells. This will require fuller identification of ROS-activated signaling pathways and knowledge of their activation mechanisms and their target genes.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
View Abstract

Stay Connected to Science Signaling

Navigate This Article