Emerging characterization of the role of SIRT3-mediated mitochondrial protein deacetylation in the heart

Michael N. Sack


Studies to quantify the protein acetylome show that lysine-residue acetylation rivals phosphorylation in prevalence as a posttranslational modification. Interesting, this posttranslational modification is modified by nutrient flux and by redox stress and targets the vast majority of metabolic pathway proteins in the mitochondria. Furthermore, the mitochondrial deacetylase enzyme SIRT3 appears to be regulated by exercise in skeletal muscle and in response to pressure overload in the heart. The alteration of protein lysine residues by acetylation and the enzymes controlling deacetylation are beginning to be explored as important regulatory events in the control of mitochondrial function and homeostasis. This review focuses on the mitochondrial targets of SIRT3 that are functionally implicated in heart biology and pathology and on the direct cardiac consequences of the genetic manipulation of SIRT3. As therapeutic modulators of other SIRT isoforms have been identified, the longer-term objective of our understanding of this biology would be to identify SIRT3 modulators as putative cardiac therapeutic agents.

  • cardiac hypertrophy
  • cylcophilin D
  • manganese superoxide dismutase
  • SIRT3

this article is part of a collection on Posttranslational Protein Modification in Metabolic Stress. Other articles appearing in this collection, as well as a full archive of all collections, can be found online at http://ajpheart.physiology.org/.

The continuing characterization of the role of mitochondria has increased our understanding regarding their function in modulating energy production, reactive oxygen and nitrogen species homeostasis, calcium regulation, heme metabolism, programmed cell death (apoptosis), retrograde signaling (50), and in the housekeeping functions of mitophagy and mitochondrial dynamics (23). In the heart specifically, mitochondria encompass between 23 and 32% of myocellular volume (63), which reflects, in part, the unrelenting energetic, redox stress, and calcium flux activity operational in this contractile organ. The composite of these diverse mitochondrial homeostatic functions and the density of these organelles in the myocardium support the concept that that the regulation of mitochondrial function may play a pivotal role in modifying cardiac biological responses to stressors including cardiac hypertrophy and heart failure, ischemia-reperfusion injury, the diabetic heart, and aging-associated decline in cardiac function (4, 18, 47, 53).

The understanding of the regulatory programs controlling mitochondrial function has also advanced exponentially during the last few decades as studies have delineated the control of mitochondrial function at the biochemical, transcriptional, and posttranscriptional regulatory levels (31, 37, 60, 74). These regulatory control mechanisms have been previously reviewed and are not the focus of this article. Nevertheless, their roles in fine tuning mitochondrial biology are proposed to modulate cardiac susceptibility to oxidative stress, ischemia-reperfusion injury, metabolic syndromes, and heart failure (11, 33, 47, 61).

A previously unrecognized regulatory pathway that plays an additional role in controlling mitochondrial function is the posttranslational modification (PTM) of mitochondrial proteins via the acetylation/deacetylation of protein lysine residues. The role of modifying the acetylation profile of mitochondrial proteins in altering protein function was posited by a study identifying acetylated mitochondrial proteins isolated from the mouse liver, comparing the fed and fasted state (40). Changes in caloric levels associate with alterations in the acetylation of mitochondrial proteins controlling fat oxidation, the tricarboxylic acid (TCA) cycle, the electron transport chain (ETC), and amino acid metabolism and in controlling redox stress (84). The number of mitochondrial proteins modified by acetylation has now been expanded to include almost all proteins in the major metabolic pathways within mitochondria (84). The enzymes that modulate these proteins and how these PTMs regulate protein function are actively being explored. The goal of this review is to describe the mitochondrial deacetylation enzymes identified sirtuins, to evaluate their roles in regulating mitochondrial target proteins, and to explore, where investigated, the role of these PTMs on cardiac biology and pathophysiology. Recent reviews explore the role of SIRT3 in biology beyond the heart, and these broader studies will not be discussed in this manuscript (7, 80, 85).

Historical Perspective on the Identification and Delineation of Mitochondrial Deacetylase Enzymes

The silence information regulator 2 (SIR2) or sirtuin family of class III deacetylases differs from the class I and II histone deacetylases by distinct protein sequences and in that they are NAD+- versus zinc-dependent deacetylases (25). The yeast SIR2 modulates replicative senescence and is considered a pivotal mediator of longevity and of DNA repair (25). Seven mammalian homologues of SIR2 have been identified and are designated as SIRT1 through SIRT7 (21). These sirtuins have distinct subcellular locations and are proposed to regulate specific biological functions unique to their subcellular locations including transcriptional silencing, diverse effects on cell growth, aging, stress tolerance, and metabolism (68).

Three sirtuins, namely SIRT3, −4, and −5, are mitochondria enriched (68). An analysis of mitochondrial protein acetylation modifications in SIRT3, −4, and −5 knockout mice show that SIRT3 is the most robust mitochondrial deacetylase (44), although SIRT5 has been shown to modulate the urea cycle enzyme carbamoyl phosphate synthetase 1 (48). Although amino acids are important intermediate metabolites in the heart, our understanding of the role of SIRT5 in the heart is limited to date and will therefore not be discussed further in this review. Additionally, the role of SIRT5 in the heart may be quite significant as emerging data are beginning to identify that SIRT5 may modulate TCA cycle intermediates via alternate lysine modifications including succinylation and malonylation (54, 83).

The earliest studies of SIRT3 resulted in some perplexing findings. The human SIRT3 sequence included a functional amino-terminal mitochondrial localization sequence (MLS) with an enriched mitochondrial localization (52, 67). In contrast, the initial cDNA sequence identified that encoded for the mouse SIRT3 generated a shorter protein lacking a MLS. In the absence of this sequence, the overexpression of SIRT3 resulted in both mitochondrial and nuclear SIRT3 expression (5). This discrepancy between the species resulted in questioning the subcellular location of the murine SIRT3 (15, 26). Three groups have recently identified the murine SIRT3 MLS and show that this sequence results from an intra-exon splice variant generating an alternate translation start site (5, 14, 36). This amino-terminal region has also been confirmed as a legitimate MLS (5). Why these alternative murine transcripts are generated has to date not been characterized. Furthermore, cognizance as to whether some of the earlier findings following overexpression of the MLS-deficient SIRT3 may have resulted in the identification of substrates and functions that may be artifactual needs to be considered. Nevertheless, the vast majority of studies exploring SIRT3 targets and mitochondrial function have been explored with either the full-length transcripts or by using SIRT3-deficient mice. The review of these studies, with specific reference to the heart, will form the foundation of this review.

The Investigation of Mitochondrial Protein Acetylation to Explore Sirtuin Biology

The technology to explore the acetylation PTM continues to be developed, and methodology reviews have been recently published (30). In brief, the bulk of methodologies rely on antibodies raised to recognize acetylated proteins. Investigators should be cognizant of the fact that different commercial antibodies can recognize distinct acetylated residues in the same sample. Also, as sirtuin deacetylation is nutrient sensing, the degree of target protein acetylation is robustly altered by the fed or fasted state (29, 40). Conjugated antibodies are also being employed for affinity enrichment of peptide fragments with differential lysine residue acetylation to then identify on mass spectroscopy. This immunoenrichment approach enables investigators to identify the potential lysine residues that are necessary for the functional effects of acetylation/deacetylation in modifying protein function. Interestingly, the acetyl modification is not ubiquitous or unidirectional in response to a specific manipulation, e.g., fasting and feeding and/or to the administration of an acetyltransferase inhibitor. These divergent PTM events could support either that enzyme-linked specific and random modifications occur concurrently or that additional regulatory systems are operational that have, to date, not been identified. The use of mass spectroscopy to identify specifically modified lysine residues is very robust as the modification is covalent and the addition of a acetyl group shifts the lysine residue by 42 Da. Once target proteins and target lysine residues operational in this PTM have been identified, the subsequent characterization depends on the function of the target substrate. Here, mutagenesis can be employed to either mimic deacetylation [substitution of the lysine with a similarly structured basic residue (arginine) or to mimic acetylation (substitution of the basic lysine residue with an uncharged glutamine)] (45, 78). This mutagenesis analysis then enables the investigators to prove the functional effect of the acetyl-lysine modification on substrate protein function. Hence, although the scientific toolbox to explore the role of protein acetylation status and the functional consequences in response to sirtuin function is not complete, the tools available are robust to enable rapid advances in this field.

The Biochemistry of Sirtuin Activation

Sirtuin activation is directly linked to levels of the intermediate metabolites, i.e., by the levels of NAD+, by the ratio of NAD+ to NADH, and by the NAD+ catabolite nicotinamide (2, 16, 43). NAD+ functions as a cofactor in the deacetylation reaction, and although this biochemistry is not the focus of this review, it is beginning to be explored (79). Interestingly, nicotinamide generated in the deacetylation reaction itself inhibits sirtuin activity, and nicotinamide depletion during NAD+ biosynthesis inversely activates sirtuins (9).

In vertebrates de novo NAD+ biosynthesis is the minor NAD+ generation pathway and uses tryptophan and nicotinic acid as metabolic precursors. This pathway does, however, appear to be induced by exercise and following the administration of peroxisome proliferator activated receptor-α agonists (34). The major pathway to generate NAD+ involves the salvage of NAD+ using nicotinamide as the precursor. In mammals there are two intermediary steps in NAD+ salvage, initiated by the conversion of nicotinamide to nicotinamide mononucleotide via the nicotinamide phosphoribosyltransferase (NAMPT) enzyme. Nicotinamide/nicotinic acid mononucleotide adenylyltransferase (NMNAT) then converts nicotinamide mononucleotide to NAD+. These biochemical pathways are most well characterized in the nucleus and are pivotal for the activity of the nuclear-enriched SIRT1 (57). Moreover, NAMPT has been identified as the rate-controlling step in NAD+ biosynthesis in that the overexpression of NAMPT but not NMNAT increased cellular NAD+ levels (57).

The investigation into the biology of NAD+ in the mitochondria has begun to be explored, and the identification of a mitochondrial-enriched NMNAT isoform implicates that subcellular compartment-specific functioning of NAD biosynthesis may be operational (8). This is further supported in that the metabolic stress of fasting increases mitochondrial NAMPT with the concomitantly increase in mitochondrial NAD+ levels with the resulting amelioration to genotoxic stressors (81).

The regulation and role of this metabolic pathway in directly controlling cardiac metabolism and contractile function have not been extensively explored, although the transcript and protein levels of NAMPT are downregulated in the murine heart in response to pressure overload-induced cardiac hypertrophy and the overexpression of NAMPT in cardiomyocytes increased cellular NAD+ and ATP levels (32). Taken together, these data would suggest an ameliorative role of this regulatory pathway and of sirtuin activation in cardiac metabolism and in the adaptation to pathological hypertrophic growth. Further support of this hypothesis is that the administration of NAD+ to mice or primary cardiomyocytes exposed to hypertrophic agonist simulation ameliorates the hypertrophic phenotype in parallel with the maintenance of cellular ATP levels (55). The activation of this NAD+-mediated antihypertrophic program appears to function via SIRT3-mediated activation of the AMPK program and the attenuation of redox stress (55). The direct metabolic consequences of this and their effect, if any, on the hypertrophic program have not been ascertained nor whether the manipulation of the NAD+ metabolic pathway effects cardiac contractile functioning.

The regulation of SIRT3 transcript and protein levels are beginning to be explored. At the transcriptional level, the SIRT3 promoter is shown to be upregulated by estrogen-related receptor-α and its cognate transcriptional coactivator peroxisome proliferator-activated receptor-γ coactivator 1α (41). The induction of mitochondrial production of reactive oxygen species similarly results in the induction of the SIRT3 transcript and protein levels in cultured cells (12). In skeletal muscle, SIRT3 protein levels diminish with aging, and exercise increases skeletal muscle SIRT3 levels in both young and old individuals (42). In the liver, SIRT3 activity is diminished in parallel with the reduction of NAD+ levels before a reduction in steady-state SIRT3 levels in response to a high-fat diet (38), and a chronic high-fat diet does deplete hepatic SIRT3 levels (6). In parallel, SIRT3 levels increase in the liver in response to progressive fasting. In the heart, SIRT3 levels are induced in response to hypertrophic agonists and by pressure overload (76, 77). Together, these data show that the activation of SIRT3 is modifiable by biomechanical, redox, and caloric load and that its function is integrally linked to NAD+ biology.

Target Proteins Modulated by SIRT3

The substrates of SIRT3 have been recently reviewed (46, 58). Hence, in this review the focus will exclusively be on those SIRT3 targets that conceptually integrate mitochondrial pathways known to either modulate cardiac function or are themselves targeted during the development of cardiac pathology. Here the role of SIRT3 in modulating fatty acid β-oxidation (FAO), the TCA cycle, the ETC, redox-stress modulation proteins, and the regulation of cyclophilin D are discussed.

SIRT3 Modulation of Energy Transduction

The role of energy deficits in the development and progression of heart failure is well established, and cardiac high-energy phosphate levels directly correlate with survival in cardiomyopathy patients (51). As fatty acids are a major cardiac fuel substrate, the downregulation of enzymes controlling FAO with the progression from cardiac hypertrophy to heart failure (62) has evoked the question as to whether the diminution of FAO contributes toward the development of heart failure (33). In contrast, the exquisite balance in substrate utilization is illustrated where the induction of regulatory programs to upregulate cardiac FAO results in the development of hypertrophy and diastolic dysfunction, mimicking the diabetic heart (19). Together, these studies support that the modulation of FAO in the heart must be tightly regulated to coordinately support energy requirements without excess flux that could otherwise result in substrate-driven pathology. As SIRT3 is regulated by metabolite intermediates, it is therefore not surprising that one of its target substrates is the FAO enzyme long-chain acyl-CoA dehydrogenase (29). In that study, Hirschey and colleagues show that SIRT3-mediated deacetylation of long-chain acyl-CoA dehydrogenase increases enzyme activity and that the SIRT3 knockout mice have blunted cardiac FAO. It is interesting that the effects of SIRT3 on bioenergetics comparing the wild-type and knockout mice are most robust and significant in the fasted state, i.e., when SIRT3 is upregulated and the activity of SIRT3 appears to be most robust (29). The role of SIRT3 in the upregulation FAO has recently been confirmed under caloric-restricted conditions (28). In isolation, these studies do not ascertain the role of SIRT3 in cardiac pathology but do suggest that the action of SIRT3 may support the augmentation of FAO under fasting conditions, a nutrient state when the relative contribution of fatty acids to cardiac energy supply is increased.

SIRT3 activity has also been shown to modulate the activity of enzymes supplying intermediates for the TCA cycle. The functional characterization of individual target proteins is now being actively investigated. The first substrate identified as a target of SIRT3 was the mitochondrial enzyme acetyl-CoA synthetase 2 (27, 66), which abundant in the murine heart (22), is inactivated by acetylation and rapidly reactivated by SIRT3-mediated deacetylation. This target of SIRT3 is activated under caloric-restricted conditions and functions during ketogenic states to convert acetate to acetyl-CoA for energy production. Three additional mitochondrial matrix proteins have been identified as substrates of SIRT3, whereby lysine deacetylation results in increased enzyme activity. These include glutamate dehydrogenase, which facilitates the oxidative deamination of glutamate to α-ketoglutarate (44, 64); the TCA cycle enzyme isocitrate dehydrogenase 2 (IDH2) (73); and urea cycle enzyme ornithine transcarbomylase (28). All three enzymes catabolize intermediary metabolites that can feed into the generation of reducing equivalents for energy production. As with the role of SIRT3 in modulating FAO, the direct evaluation of the role of SIRT3 in modulating the TCA cycle and the pathways supplying substrates to this cycle on cardiac function has not been explored. Interestingly, a role for anaplerosis, i.e., the recruitment of alternative intermediary pathways to enhance flux through the TCA cycle (59), is beginning to be appreciated in facilitating metabolic plasticity to maintain energy production in the hypertrophied heart (56). Whether the role of SIRT3 in regulating intermediary metabolites is operational is an intriguing concept that has not yet been investigated.

The final common pathway for the generation of ATP from the reducing equivalents is the ETC. To date, SIRT3 has been shown to deacetylate and activate multiple protein subunits in the ETC including proteins in complex I and II and in the ATP synthase ATPase complex V (1, 6, 13). As would be expected by SIRT3 activating these complexes, it has been shown that the genetic depletion of SIRT3 accordingly compromises complex activity, mitochondrial oxygen consumption, and ATP production (1, 5). In light of the high-energy demand of the heart and the multiple mitochondrial energy transduction proteins modified by SIRT3, these data would suggest that the absence or reduction of SIRT3 would predispose to significant cardiac deficits either at baseline or even more so in response to biomechanical and metabolic stressors. Data are emerging in this arena and will be discussed in Direct Cardiac Consequences of the Genetic Manipulation of SIRT3.

SIRT3 Modulation of Reactive Oxygen Species Biology

As discussed previously, early SIRT3 studies gave rise to discrepant data regarding its subcellular localization, where the overexpression of SIRT3 does not mirror the localization of the endogenous protein (5). This perturbed localization may result from incomplete or unphysiological import of exogenous proteins into the double-membrane mitochondrial matrix as has been found in other overexpression studies (75). Hence, in this review the investigations that show that SIRT3 controls antioxidant programs via nuclear and cytosolic regulatory mechanisms are not extensively discussed, although cited for completeness (35, 69, 76, 77). More recent studies have shown more direct mitochondrial effects of SIRT3 in regulating programs controlling redox stress (39, 41). These include data to show that SIRT3 directly deacetylates and activates MnSOD with increased protection against radiation-induced stress in cell cultures and in murine liver (78). Under caloric-restricted conditions, the known activation of IDH2 is shown to maintain higher levels of reduced glutathione in wild-type versus SIRT3 knockout mice (73). Here, the proposed mechanism involves IDH2-mediated increase in NADPH levels, which are then employed by glutathione reductase to increase the levels of reduced glutathione. In this study, the coexpression of SIRT3 and IDH2 enhanced cellular protection against redox stressors (73). Numerous other mitochondrial redox stress modulatory proteins such as thioredoxin, glutathione-S-reductase, aconitase, and aldehyde dehydrogenase have also been shown to be modified by lysine acetylation, although the functional characterization of these proteins in the heart do not appear to have been characterized at this time (40, 84). Table 1 shows the SIRT3 targets that have been functionally validated to date.

View this table:
Table 1.

SIRT3 targets identified and characterized

SIRT3 Regulation of Cyclophilin D

The mitochondrial matrix peptidyl-prolyl isomerase cyclophilin D (Ppif) has also been identified as a substrate for SIRT3-mediated deacetylation (24, 71, 72). As opposed to the other targets of SIRT3 deacetylation discussed in this review, here deacetylation results in the inhibition in the activity of cyclophilin D. Our understanding of the role of cyclophilin D in mitochondrial biology has increased in recent years with the generation of Ppif knockout mice. This genetic model shows that cyclophilin D functions to enhance acute mitochondrial susceptibility to increased membrane permeability transition (3, 49). Also, the Ppif knockout mice have impaired mitochondrial calcium efflux with elevated mitochondrial calcium levels, resulting in the induction of calcium-dependent mitochondrial enzyme activities (17). The cardiac phenotype in these mice show exaggerated cardiac hypertrophy in response to pressure overload with a more rapid decline in cardiac contractile function. From a reductionist perspective, if SIRT3 activation inhibits cyclophilin D, the SIRT3 knockout mice may be expected to retain cyclophilin D activity and possess an increased propensity to mitochondrial permeability transition and/or possibly greater susceptibility to pressure overload-induced cardiac hypertrophy (17). Although how this function of SIRT3 integrates with the other functions attributed to SIRT3 described above are yet to be delineated. Furthermore, how the maintenance of cyclophilin D activity following SIRT3 depletion may modulate mitochondrial calcium homeostasis is another area that appears not to have been resolved.

Direct Cardiac Consequences of the Genetic Manipulation of SIRT3

The initial characterization of SIRT3 in the heart was performed by Sundaresan and colleagues where primary cardiomyocytes were subjected to hypertrophic agonists by exposure to angiotensin II or phenylephrine (77) and in the heart in response to exercise and pressure overload (76). All of these pharmacological and mechanical stressors resulted in the upregulation of SIRT3. Interestingly, in the SIRT3 knockout mice, cardiac hypertrophy and the development of cardiac fibrosis were accelerated compared with SIRT3-competent mice. In parallel, the absence of SIRT3 resulted in the attenuation of antioxidant enzyme activities in response to the hypertrophic agonists (76). Together, these data suggest that the antioxidant effects of SIRT3 may play an important role in ameliorative hypertrophic agonist-induced cardiac hypertrophy. This phenotype was also shown where pressure overload was directly induced by thoracic aortic constriction (24). Here, the SIRT3 knockout mice had a higher postoperative mortality from the banding studies compared with the wild-type mice and the knockout mice develop excessive age-associated hypertrophy and myocardial fibrosis (24). The role of the prevention of cyclophilin D inhibition in the absence of SIRT3 may also be implicated in this progressive cardiac pathology (24).

As SIRT3 has robust effects attenuating redox stress pathology, it may be extrapolated to SIRT3 having an ameliorative role in response to ischemia-reperfusion injury. In a similar vein, prior studies have shown that the absence of SIRT3 or its downregulation is associated with a diminished capacity to adapt to caloric excess in the liver (6, 38). Similarly, as obesity and diabetes provoke mitochondrial perturbations in the heart (10, 11, 61), it could be postulated that SIRT3 would function accordingly in modulating the cardiac response to lipotoxic stressors. However, the role of SIRT3 in modulating adaptation to either ischemia-reperfusion of the diabetic heart does not appear to have been directly investigated to date.


In the last five years, a ubiquitous PTM, i.e., protein lysine residue acetylation/deacetylation, has been found to be operational in the mitochondria. Although this PTM appears to play a role in mitochondrial homeostasis, it should be cautioned that this modification rather fine tunes mitochondrial function and is not necessary for basal mitochondrial metabolic control. This statement reflects the successful birth and growth of the SIRT3 knockout mouse. Nevertheless, the fact that SIRT3 function is modified by caloric levels, redox stress, and biomechanical stimulation and plays a role in modifying mitochondrial protein activities implicates its function in fine tuning biology in the emerging cardiac pathologies associated with diabetes, obesity, and cardiac ischemia. To date, the biology of SIRT3 in these pathologies have not been published. However, direct investigations have shown that SIRT3 plays an important ameliorative role in preventing pathology form pressure overload and aging-associated decline in cardiac function. These effects and the SIRT3 target pathways implicated in this biology are shown in Fig. 1. Ongoing studies promise to add further important insights into the role of mitochondrial sirtuins and their control of cardiac mitochondrial and the consequences on heart pathology. Whether the direct pharmacological modulation of SIRT3 may confer greater benefit compared with the disappointing results to date of other antioxidant approaches in cardiac disease is an intriguing concept (65), although direct pharmacological activators of SIRT3 have not to my knowledge been developed at this time.

Fig. 1.

Cardiac pathology in SIRT3 knockout mice. Cardiac hypertrophy, chamber dilatation, contractile dysfunction, and increased fibrosis is evident following pressure overload, adrenergic agonists, and in response to aging in the SIRT3−/− mice. The target proteins identified in these defective phenotypes include the role of SIRT3 in augmenting antioxidant defenses and in inhibiting cyclophilin D. As the reduction of fatty acid oxidation accompanies cardiac dysfunction whether the role of SIRT3 in activating fatty acid oxidation proteins in this context is feasible but has not yet been directly measured. ROS, reactive oxygen species; LV, left ventricular; MPT, mitochondrial permeability transition; Ac, acetyl; LCAD, long-chain acyl-CoA dehydrogenase.


This work was supported by the Div. of Intramural Research of the National Heart, Lung, and Blood Institute.


No conflicts of interest, financial or otherwise, are declared by the author(s).


Author contributions: M.N.S. interpreted results of experiments; M.N.S. prepared figures; M.N.S. drafted manuscript; M.N.S. edited and revised manuscript; M.N.S. approved final version of manuscript.


  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.
View Abstract