Sirt1’s Complex Roles in Neuroprotection

Abstract

The nicotinamide adenine dinucleotide (NAD)-activated protein deacetylase Sir2p/Sirt1 has been strongly implicated in the modulation of replicative lifespan and promotion of longevity. Part of Sirt1’s capacity for lifespan extension in complex organisms may be attributed to its protective activity against neuronal degeneration. Manipulation of Sirt1’s activity or levels by pharmacological and genetic means in several models of neurodegenerative diseases demonstrated its neuroprotective credentials. However, recent data have indicated that under certain contexts, Sirt1 inhibition, rather than activation, is neuroprotective. These inconsistencies highlight the complex nature of Sirt1-mediated effects. The enzyme has both histone and nonhistone targets, and could potentially act in both nuclear and cytoplasmic compartments. These activities intertwine in a manner depending on the context of a system under investigation. One needs to be cautious in extrapolating results derived from short-term observations to a longer-term context, and in assessing efficacies of Sirt1-based therapeutic approaches in treating neurodegenerative diseases.

Introduction

The yeast Sir2p-related family of proteins, or sirtuins, has been extensively investigated as longevity factors, as well as potential therapeutic targets in human diseases such as cancer, metabolic diseases, and neurodegenerative disorders. An evolutionarily conserved enzyme family, sirtuins, is nicotinamide adenine dinucleotide (NAD)-dependent class III histone deacetylases (Blander and Guarente 2004; Denu 2005). Saccharomyces cerevisiae’s Sir2p functions as a transcriptional repressor through histone deacetylation at the telomeres, mating-type loci, and the rDNA gene loci (reviewed in Blander and Guarente 2004). The enzyme received initial attention as a factor that modulates yeast replicative lifespan (Kaeberlein et al. 1999; Lin et al. 2000; Howitz et al. 2003). Yeast Sir2 inhibits rDNA recombination as well as the formation of extrachromosomal rDNA circles (ERCs), which promotes replicative aging (Sinclair and Guarente 1997). Over-expression of Sir2 extends replicative lifespan of yeast. Excitingly, it also increased lifespan in invertebrate models of Caenorhabditis elegans (Tissenbaum and Guarente 2001) and Drosophila melanogaster (Rogina and Helfand 2004). In multicellular organisms, however, the mode of Sir2 orthologue action on longevity is unlikely to involve equivalents of yeast ERCs. In C. elegans, the Sir2 orthologue sir-2.1 acts through daf-16, a member of the FOXO family of transcriptional factors (Tissenbaum and Guarente 2001). Mammals have seven sirtuin paralogues (Sirt1-7), with Sirt1 being the closest to the yeast Sir2 (Haigis and Guarente 2006). However, that mammalian Sirt1 may also be pro-longevity in vivo has not been conclusively demonstrated.

Sir2/Sirt1’s link to longevity has been closely associated with the notion that it may mediate the effect of caloric restriction (CR), whereby a dietary regimen which reduction in food/energy uptake has been shown to enhance the lifespan of models ranging from yeast to rodents (Masoro 2005). Increased longevity induced by yeast cultured in reduced glucose apparently requires Sir2 activation, and lifespan extension by CR is not observed in SIR2 deletion strains (Lin et al. 2000). CR effect and Sir2p activity are both regulated by nicotinamide and the enzyme pyrazinamidase/nicotinamidase 1 (PNC1) (Anderson et al. 2003). Sirtuin activators, such as resveratrol, could mimic the CR effect, resulting in increasing DNA stability and lifespan extension in yeast (Howitz et al. 2003). Sinclair’s group had shown that resveratrol and other sirtuin activators mimic CR and extended the lifespan of both worm and fly in a Sir2 orthologue-dependent manner (Wood et al. 2004). Guarente and coworkers have also shown that resveratrol extends worm lifespan largely through Sir-2.1-mediated repression of ER stress genes, such as abu-11 (Viswanathan et al. 2005).

The importance of Sir2 to the CR response in yeast has remained highly debatable (Haigis and Guarente 2006; Sinclair et al. 2006; Longo and Kennedy 2006). CR’s lifespan extension effect could indeed become independent of Sir2 provided that ERCs are kept at low levels (KaeberlEin et al. 2004), and nicotinamide may inhibit lifespan extension by CR through a Sir2-independent mechanism (Kaeberlein et al. 2005a). Although Sir-2-independent effect of CR could in part be mediated by Hst2, another Sir2 homologue in yeast (Lamming et al. 2005), there is an evidence that CR effect in yeast is more dependent on nutrient-dependent kinases, such as target of rapamycin (TOR), protein kinase A (PKA), and Sch9 (the yeast homologue of mammalian S6 and Akt kinase; Kaeberlein et al. 2005b; Longo and Kennedy 2006). It should also be noted that Sir2 deletion actually enhances the chronological lifespan (as opposed to replicative lifespan) of yeast, which is a measure of viability in the nonreplicative, and stationary growth phase (Fabrizio et al. 2005). In C. elegans, factors, such as heat shock transcription factor-1 (Hsf-1), are also important in mediating the longevity effect of CR (Steinkraus et al. 2008). In mammals, the increase in physical activity of CR mice was shown to be Sirt1 dependent (Chen et al. 2005a), and Sirt1 transgenic mice do show CR phenotypes (Bordone et al. 2007). However, CR regimen does not activate Sirt1 in all tissues, and liver Sirt1 activity is in fact reduced by CR and activated by a high caloric diet (Chen et al. 2008). Sirt1-null mice exhibited infrequent survival to adulthood, are smaller, and have developmental defects (McBurney et al. 2003; Cheng et al. 2003). Consistent with the propose role of Sirt1 in CR, Sirt1-null mice are not CR-responsive in terms of lifespan extension, but this phenomenon is complicated by the fact that these mice have a shortened lifespan on a normal diet to begin with. Interestingly, mouse embryonic fibroblasts that are Sirt1 deficient have an increased replicative lifespan, and Sirt1 appears to limit this in response to chronic genotoxic stress (Chua et al. 2005).

The mammalian Sirt1 has been linked to multiple physiological functions and pathological roles in line with recent discoveries of its nonhistone cellular substrates. Prominent amongst these nonhistone substrates are the transcription factors tumor suppressor p53 (Luo et al. 2001; Vaziri et al. 2001), members of the FOXO family (Brunet et al. 2004; Motta et al. 2004; Daitoku et al. 2004; Giannakou and Partridge 2004) and NF-κB (Yeung et al. 2004; Chen et al. 2005b). All these are key transcriptional regulators of proliferation and survival for many different cell types. Sirt1 also deacetylates the peroxisome-proliferator-activated receptor-γ (PPARγ; Picard et al. 2004) and its transcriptional coactivator PPARγ coactivator-1α (PGC-1α; Nemoto et al. 2005; Rodgers et al. 2005; Lagouge et al. 2006), which regulates a wide range of metabolic activities in skeletal muscle, adipose tissues, and the liver. These nonhistone SIRT1 substrates functionally link nutrient availability and energy metabolism to adaptive changes in transcriptional profiles that affect cell survival in multiple organ systems and impinging on multiple human diseases (Bordone and Guarente 2005; Longo and Kennedy 2006; Anastasiou and Krek 2006). Foremost in this regard would be age-associated metabolic diseases (Guarente 2006). Accordingly, resveratrol administration improves health and survival of animal on a high calorie diet (Baur et al. 2006), and Sirt1 activators are effective as experimental therapeutics in animal models of type II diabetes (Milne et al. 2007). Transgenic mice with moderate over-expression of Sirt1 have increased energy efficiency and glucose tolerance (Banks et al. 2008). Transgenic expression of Sirt1 also reduced artherosclerotic lesions in apolipoprotein E-null mice (Zhang et al. 2008).

Sirt1 interacts with several factors with known involvement in oncogenesis and malignancy (Saunders and Verdin 2007). Its deacetylation of p53 (Luo et al. 2001; Vaziri et al. 2001) inhibits p53-dependent apoptosis in response to DNA damage and oxidative stress. Cells from Sirt1-deficient mice exhibited p53 hyperacetylation after DNA damage (Cheng et al. 2003). Sirt1 also deacetylates the DNA repair factor Ku70, which sequesters the pro-apoptotic factor Bax away from the mitochondria (Sawada et al. 2003; Cohen et al. 2004a). The ability of Sirt1 to attenuate apoptosis would conceivably enhance the survival of transformed cells. A loss of the tumor suppressor hypermethylated in cancer 1 (HIC1) appears to promote tumorigenesis via activating Sirt1, resulting in an attenuation of p53 function (Chen et al. 2005c). In undifferentiated mouse prostate adenocarcinoma, Sirt1 was significantly elevated while HIC-1 was markedly reduced (Huffman et al. 2007). Sirt1 could also contribute to the development of prostate cancer via an inhibition of FOXO1 activation (Jung-Hynes et al. 2009). The cell-cycle and apoptosis regulator E2F1 also induces SIRT1 expression (Wang et al. 2006). DNA damage by genotoxic cancer drugs, like etoposide, causes E2F1-dependent induction of Sirt1 expression, and silencing of Sirt1 increases sensitivity to etoposide.

The above findings would seem to suggest a general pro-survival effect of Sirt1 that might in certain context augment cell transformation and malignancy. On the other hand, however, there is also evidence that suggests that Sirt1 has tumor-suppressive functions. Sirt1-null embryos that died in utero exhibited impaired DNA damage response and genome instability. Mice heterozygous for both Sirt1 and p53 deletion develop tumors in multiple tissues, whereas activation of Sirt1 by resveratrol treatment reduces tumorigenesis (Wang et al. 2008a). In embryonic stem cells, Sirt1 redistributes to DNA breaks, promotes repair, and increases Sirt1 promotes survival in a mouse model of genomic instability (Oberdoerffer et al. 2008). Sirt1 appears to play a role in BRCA-1-associated tumorigenesis through its regulation of survivin expression (Wang et al. 2008b). SIRT1 deacetylates β-catenin and promotes cytoplasmic localization of nuclear β-catenin, and could be tumor-suppressive in malignancies promoted by β-catenin signaling such as intestinal cancers (Firestein et al. 2008).

Another important group of age-associated disorders in which Sirt1 has been prominently associated with is neurodegenerative diseases (Gan and Mucke 2008; Outeiro et al. 2008). Many of Sirt1’s nonhistone substrates are important for neuronal survival, and part of Sirt1’s capacity for lifespan extension in complex organisms could potentially be attributed to its neuroprotective effect. From another perspective, Sirt1’s deacetylation of histones regulates neural progenitor cell development and also provides epigenetic modulation of adult brain cell functions. The activity of Sirt1 on its histone and nonhistone targets could be intertwined in a complex manner as Sirt1 has both a nuclear and a cytoplasmic localization in neurons. Sirt1’s role in neuroprotection and neural differentiation is the focus of this review and shall be discussed in more detail.

Sirt1 in the Brain and Neural Cells

Of the seven mammalian sirtuins, Sirt1, Sirt6, and Sirt7 are the first recognized to be nuclear proteins (Michishita et al. 2005). Sirt3, Sirt4, and Sirt5 are predominantly localized to the mitochondria (Michishita et al. 2005), and likely function in aspects of energy homeostasis (Ahn et al. 2008). Sirt2, which is homologous to yeast Hts2p, is largely cytoplasmic, where its major target is likely Lys 40 of α-tubulin (North et al. 2003). A recent report showed that Sirt2 is selectively enriched in oligodendrocytes and modulates their differentiation through tubulin deacetylation (Li et al. 2007). Sirt2 may also modulate neuronal migration through its effect on microtubules (Pandithage et al. 2008). As discussed, Sirt2 may play roles that are either synergistic or antagonistic to that of Sirt1 in terms of neurodegeneration and neuroprotection.

In mouse embryos, Sirt1 is expressed at high levels in the heart and the developing CNS (Sakamoto et al. 2004), which suggest that it might have a role in neuronal development. In the adult brain, Sirt1 transcripts are highly expressed in metabolically relevant neurons, and fasting induced Sirt1 protein levels specifically in the hypothalamus (Ramadori et al. 2008). Interestingly, Horio and colleagues showed that Sirt1 contains functional nuclear localization signals as well as nuclear export signals, and that subcellular localization of Sirt1 varies in different adult tissues (Tanno et al. 2007). In some neurons, Sirt1 in fact appears predominantly cytoplsmic. Longo and co-worker also showed that while Sirt1 can be found in both nuclear and cytoplasmic fractions of the brain of newborn mouse pubs, it is present predominantly in the cytoplasmic fraction of the adult brain (Li et al. 2008). Sirt1 has always been viewed as a nuclear protein, and its targets are presumably nuclear as well. Although whether Sirt1 retains its full enzymatic activity in the cytosol is unclear, the above findings indicate that, at least in neurons, Sirt1 may have nonnuclear targets. These would of course include transcriptional factors that shuttle between the nucleus and the cytoplasm.

Sirt1’s Regulation of Transcriptional Factors That Affect Neuronal Survival (and Death?)

As mentioned above, many nonhistone targets of Sirt1 activity are transcriptional factors involved in regulating survival of a wide range of cells, which would include postmitotic neurons. p53 acetylation activates transcription of downstream targets with pro-apoptotic properties, and its deacetylation by Sirt1 would, in general, attenuate p53-dependent neuronal death. Recent findings implicated necdin, a maternally imprinted melanoma antigen family protein, as a p53-interacting protein that modulates its Sirt1 deacetylation in post-mitotic neurons through the formation of a tripartite p53-necdin-Sirt1 complex (Hasegawa and Yoshikawa 2008). Protection against neurodegeneration in animal models by the Sirt1 activator resveratrol is often associated with decreased p53 acetylation.

Members of the FOXO family of transcriptional factors are important regulators of genes that are involved in stress response, cell-cycle arrest, and cell survival (Lam et al. 2006). The regulations and consequences of Sirt1 deacetylation of FOXO family members are rather complex and apparently promoter specific. In general, Sirt1 enhances the expression of FOXO target genes that are involved in resistance to oxidative stress, but inhibits the expression of those associated with cell death induction, homeostatically promoting cell survival (Brunet et al. 2004; Motta et al. 2004; Daitoku et al. 2004; Giannakou and Partridge 2004). Interestingly, recent evidence suggests that Sirt2 could also deacetylate FOXO proteins in response to CR and oxidative stress (Wang et al. 2007).

Sirt1 deacetylates RelA/p65 subunit of NF-κB at Lys 310, resulting in a loss of NF-κB-regulated gene expression. In some cells, resveratrol appears to heighten sensitivity to TNF-α-induced apoptosis through this pathway (Yeung et al. 2004). This is an apparent departure from the other apoptosis suppression activity of Sirt1 described earlier, and indicates that Sirt1 may promote neuronal death under certain circumstances. In the context of neurons, Sirt1’s suppression of NF-κB activity could be protective in a noncell autonomous manner. NF-κB signaling in microglia makes an important contribution to neuronal death caused by amyloid β (Aβ). Both overexpression of Sirt1 and resveratrol markedly reduced Aβ-induced NF-κB signaling and are protective of primary cortical neurons in culture (Chen et al. 2005b).

Although the Sirt1 activation is generally viewed as neuroprotective, resveratrol’s apoptosis-promoting activity in cancer cells is well known (Filomeni et al. 2007; Fan et al. 2008). It is therefore possible that Sirt1 may also be a context-dependent promoter of neuronal death. Inhibition of Sirt1 by its inhibitor, nicotinamide, has indeed been shown to promote neuronal survival in acute anoxic injury (Chong et al. 2005). Han et al. (2008) have shown that Sirt1 induced by reactive oxygen species (ROS; resulting from 2-mecaptoethanol ethanol withdrawal from the culture medium) in mouse embryonic stem cells could block nuclear translocation of p53, thereby preventing the expression of p53-dependent antioxidant enzymes. Worse, cytoplasmic p53 could be translocated to the mitochondria and directly induces apoptosis.

Sirt1 and Transgenic Models of Neurodegenerative Diseases With Toxic Aggregates

A large number of studies using pharmacological and/or genetic manipulation of Sirt1 activity have pointed to a general neuroprotective outcome of Sirt1 activation. There are some measurable beneficial effect of resveratrol in acute CNS injuries (Kiziltepe et al. 2004; Kaplan et al. 2005), and resveratrol pretreatment mimics ischemic preconditioning via Sirt1 in a hippocampal slice culture model (Raval et al. 2006). A notable exception, in which Sirt1 inhibition was found to be neuroprotective, is the study of Chong et al. (2005) mentioned above.

Other findings have also fueled the notion that Sirt1 activation could be beneficial to chronic neuronal degeneration. One of the first animal models tested is that of Huntington’s disease (HD), in which expansion of a polyglutamine (PolyQ) stretch resulted in a mutant huntingtin polypeptide that forms cytotoxic aggregates in neurons (Borrell-Pages et al. 2006). Resveratrol protected both C. elegans and mouse neurons against the cytotoxicity of the mutant huntingtin. Overexpression of a huntingtin (Htt) fragment containing expanded polyQ in C. elegans neurons generates a HD model with mechanosensory defect. Both the resveratrol and an increased sir-2.1 gene dosage alleviated the worm neuronal dysfunction in a daf16-dependent manner. Resveratrol also decreased cell death associated with neurons cultured from mutant huntingtin (109Q) knock-in mice, an effect that is attenuated by Sirt1 inhibitors. However, it should be noted that not all Sir2 orthologues are protective of transgenic Htt mutants. In Drosophila, transgenic overexpression of Sir2 orthologue actually promotes apoptosis (Griswold et al. 2008). Furthermore, genetic or pharmacological reduction of Sir2 activity actually improved neuronal survival in mutant Htt transgenic flies (Pallos et al. 2008).

Caenorhabditis elegans models expressing human transgenes are also useful in modeling other neurodegenerative disease with toxic aggregates, such as the α-synuclein-containing inclusions in Parkinson’s disease (PD). Interestingly, sir2.1 is amongst genes which silencing exacerbates aging-associated formation of α-synuclein-containing inclusions in transgenic worms (Van Ham et al. 2008). Resveratrol attenuated the loss of cell viability and apoptosis of cultured cerebellar granule neurons by 1-methyl-4-phenylpyridinium (MPP⁺), a chemical which is selectively toxic to dopaminergic neurons and widely used for the induction of hemiparkinsonism in rodents (Alvira et al. 2007). However, as the Sirt1 inhibitor sirtinol did not effectively counter resveratrol’s effects in this particular report, whether the latter acts through Sirt1 activation is debatable. Oxyresveratrol protected dopaminergic SH-SY5Y neuroblastoma cells from 6-hydroxydopamine toxicity, and increased the basal levels of Sirt1 (Chao et al. 2008). The actual mode of function of resveratrol, as well as whether Sirt1 activation is important for its neuroprotective effect, appears to vary with the nature of toxic insult. Another report also showed that resveratrol’s protective effect on organotypic mid-brain slice cultures is not reversed by Sirt1 inhibitors. However, resveratrol and other Sirt1 activators’ protection of an alkylating agent-induced cytotocity are associated with attenuation of p53 acetylation induced by the agent (Okawara et al. 2007).

An interesting recent finding with regard to sirtuins and Parkinsonism is that inhibition of Sirt2 protects against α-synuclein toxicity. Outeiro et al. (2007) found that Sirt2 silencing by RNA interference or its inactivation with a potent and selective inhibitor effectively reduces the toxic effect of α-synuclein in H4 cells, dopaminergic neurons from primary midbrain cultures, and transgenic Drosophila. It is not immediately clear why Sirt1 and Sirt2 would have the exactly opposite influence on α-synuclein toxicity. In transfected cerebellar granule neurons, Sirt1 confers protection against a low potassium treatment, while Sirt2 (as well as Sirt3 and Sirt6) induces apoptosis (Pfister et al. 2008). Interestingly, even Sirt1 mutants that are defective in their deacetylase activities are protective. These results collectively point towards a well-known but often underappreciated fact, as well as hinting at another possibility that is often ignored. First, the action of resveratrol is complex, and Sirt1 activation, even supposing that an effective concentration was achieved in vivo to allow this to occur, cannot fully account its neuroprotective endpoint. As discussed later, resveratrol also activates Sirt2, whose effects on tubulin deacetylation may confer an entirely opposite, antiprotective effect in certain contexts. Secondly, Sirt1 may have yet undefined, deacetylase-independent activities that are neuroprotective.

Sirt1 in Alzheimer’s and Other Neuroinflammatory Degenerative Diseases

Alzheimer’s disease (AD) is the most prominent cause of age-associated dementia, and the benefits of Sirt1 activators in models of AD have been extensively discussed (Anekonda and Reddy 2006; Anekonda 2006; Markus and Morris 2008). As mentioned earlier, Sirt1 attenuation of NF-κB signaling in microglia is protective against neuronal death induced by Aβ peptides (Chen et al. 2005b). AD has been known as a disease of neuroinflammation, and it is clear that the beneficial effect of resveratrol in AD would be multifaceted (i.e., both as an antioxidant and as an antiinflammatory agent through Sirt1 activation).

A point of association between Sirt1 activity and AD that appears to be particularly relevant is that of CR, which has been proposed to protect against AD. CR regimen prevents Aβ peptide generation and plaque deposition in the brain of transgenic AD mice models (Patel et al. 2005; Wang et al. 2005a). CR resulted in Aβ reduction in the temporal cortex of squirrel monkeys, which was inversely correlated with Sirt1 levels (Qin et al. 2006a). In mice, CR appears to enhance amyloid precursor protein (APP) processing via the nonamyloidogenic α-secretase-mediated pathway (Wang et al. 2005a), which is modulated by the Rho-Rho-associated coiled-coil containing protein kinase 1 (ROCK1) signaling axis (Zhou et al. 2003; Pedrini et al. 2005; Tang 2005). The question is whether CR acts through Sirt1 in this regard. Pasinetti and colleagues showed that CR increases both Sirt1 and NAD levels in Tg2576 mouse brain, and virally transduced Sirt1 in primary neuron cultures from Tg2576 mouse had elevated sAPPα (the α-secretase product of APP cleavage), but reduced Aβ secretion and ROCK1 expression. A constitutively active ROCK1 prevented Sirt1-mediated elevation of sAPPα, and transgenic elevation of Sirt1 levels in the brain correlated with a reduction in ROCK1 levels (Qin et al. 2006b). This reduction may be mediated through Sirt1 deacetylation of FOXO3a (Qin et al. 2008). Sirt1 may therefore enhance α-secretase-mediated nonamyloidogenic APP processing in a manner that could be negatively influenced by ROCK1 signaling.

In a more recent report, Sirt1 levels were shown to be upregulated in a mouse model of AD and amyotrophic lateral sclerosis (ALS). Resveratrol was shown to protect against toxicity induced by the Cdk5 activator p25 and mutant Cu/Zn superoxide dismutase 1 (SOD1) in culture neurons and in vivo (Kim et al. 2007). This protection corresponds to deacetylation of p53, and direct injection of Sirt1 expressing lentivirus into the CA1 region of p25 transgenic mice (a model for AD and taupathy) protected against neurodegeneration. These results provided perhaps the strongest direct in vivo evidence to date that Sirt1 is protective against neurodegeneration, although similar results have not yet been shown for other mouse AD models. In another experimental paradigm, Sirt1 activators protected against the loss of retinal ganglion cells from optic neuritis induced in experimental autoimmune encephalomyelitis (EAE), an animal model of multiple sclerosis (Shindler et al. 2007). Taken together, activation or elevation of Sirt1 does appear to exert protection on neurons in neuroinflammatory diseases, both indirectly through suppression of microglia-mediated inflammatory responses and through neuronal cell autonomous mechanisms.

Sirt1, Sirt2, and Wallerian Degeneration

Another link between Sirt1 activity and neuroprotection lies in its controversial metabolic connection with the gene mutated in Wallerian degeneration slow (wld ^s) mice (Perry et al. 1990), which exhibit a marked delay in axonal degeneration induced by physical or chemical injury (Coleman and Perry 2002). Wld ^s encodes a fusion protein between the N-terminal 70 amino acids of ubiquitin fusion degradation protein 2a (Uf2a) and the complete coding sequence of nicotinamide mononucleotide adenylyltransferase 1 (Nmnat1, a key enzyme in the NAD biosynthesis pathway). Exactly which portion of the fusion protein delays axonal degeneration has remained enigmatic and unresolved. Proteasome inhibitors could inhibit axonal degeneration (Zhai et al. 2003), implying that the Uf2a fragment could play a significant role. There is also some evidence that the increase in Nmnat1 activity is at least partially responsible for axonal protection by Wld^s through Sirt activation, as neuroprotection is blocked by sirtinol and silencing of Sirt1 (Araki et al. 2004). Overexpressing enzymes from multiple NAD biosynthesis pathways in dorsal root ganglion (DRG) cultures provided varying degrees of protection against axonal degeneration after axotomy (Sasaki et al. 2006).

Nmnat1 activity of the Wld^s could not, however, fully account for the neuroprotective phenotype. NAD could delay axonal regeneration by a local protective mechanism that may be SIRT1 independent (Wang et al. 2005b), as the degeneration of axonal segments that have been separated from their soma (and therefore, does not involve nuclear Sirt1) could be delayed by exogenous application of NAD or its precursor nicotinamide. Transgenic overexpression of Nmnat1 alone also resulted in inferior axonal degeneration delaying effect compared to the Wld^s mutant protein (Conforti et al. 2007). In DRG explants, Wld^s is more potent than Nmnat1 in delaying degeneration, and even an enzyme-dead mutant of Wld^s still displayed residual-protective activity. It is possible that the sequences linking or flanking the two fused open reading frames of Wld^s may have yet undefined functions in delaying axonal degeneration, such as inactivation of Draper-mediated glial clearance of severed axons (Fainzilber and Twiss 2006). Very recent findings have indeed shown that both the N-terminal sequence and intact Nmnat1 activity are required for full Wld^s activity (Conforti et al. 2009; Avery et al. 2009).

Another set of recent finding indicated that basal level of microtubule acetylation was increased in cultured cerebellar granule cells from Wld ^s mice, with a corresponding decrease in NAD and Sirt2 (Suzuki and Koike 2007a). Sirt2 overexpression diminishes degeneration resistance of these cells. Interestingly, the authors also showed that treatment of these cells with resveratrol diminished resistance of Wld ^s neurons to axonal degeneration induced by the microtubule-disrupting drug colchicines as a result of Sirt2 activation (Suzuki and Koike 2007b). These results, together with those discussed earlier on the effect of Sirt2 on α-synuclein toxicity, suggest that the use of general sirtuin activators for therapeutic purposes should be advanced with caution.

Sirt1, the IGF Signaling Axis, and Neuronal Survival

For a multicellular organism, prolong health of its nervous system must contribute towards longevity in a significant manner. If Sirt1 indeed plays a seminal role in mediating the effect of CR and lifespan extension, its neuroprotective property should feature prominently in this regard. It is therefore somewhat surprising that recent evidence suggests that Sirt1 inhibition could actually enhance oxidative stress resistance in neurons. Work from Longo and colleagues showed that Sirt1 inhibitors increased the survival of wild-type neurons to oxidative damage, and that brains of Sirt1-/-mice exhibited reduced levels of markers of oxidative damage, such as protein carbonylation and lipid peroxidation (Li et al. 2008). The underlying reason for this observation is a reduction of insulin-like growth factor-1(IGF-1) receptor-Ras-Extracellular signal-regulated kinase (Erk) pathway activation, which contributes to cell death from oxidative stress. Sirt1 could apparently deacetylate the cytoplasmic IGF-1 receptor adaptor protein insulin receptor substrate-2 (IRS-2), thereby increasing its phosphorylation and signaling capacity. One general way of prolonging lifespan in multicellular organisms appears to be the inhibition or attenuation of insulin/IGF-1 signaling through the IGF-1 receptor (Kimura et al. 1997; Tatar et al. 2001; Holzenberger et al. 2003). At least in daf-2 and age-1 mutant C. elegans, restoring the IGF signaling pathway in neurons alone resulted in a wild-type lifespan (Wolkow et al. 2000). The relationship between CR and the IGF-1 signaling pathway is not clear cut, and although CR does reduce serum IGF-1 and IGF-binding protein-3 (IGFBP-3) in rodents, this effect is not obvious in humans (Fontana et al. 2008). A hint of a connection between CR, Sirt1, and IGF-1 signaling came from the finding that the latter attenuates Sirt1 induction by CR (Cohen et al. 2004b).

Results from the work discussed previously do not rule out a role for Sirt1 in CR, but provide one plausible explanation as to why attenuation of the IGF-1 pathway is almost universally prolongevity, in spite of the fact that IGF-1 signaling itself has clear short-term neuroprotective effects (Tang 2006; Longo 2009). Sirt1 inhibitors are also known to attenuate Ras-Erk signaling in cell types other than neurons (Ota et al. 2006). Sirt1 activities may therefore be pro or antineuronal survival in different contexts. As Sirt1-/-mice have shorter lifespans compared to wild type, the benefits derived from the loss of Sirt1 in terms of neuroprotection against oxidative damage certainly did not translate to increase longevity.

Concluding Remarks

There is a wealth of data indicating that Sirt1 elevation or activation is neuroprotective. From all the discussions, there are at least three caveats to this generalized notion. First, there are experimental paradigms that indicate that the opposite is true in some contexts, i.e., Sirt1 inhibition, rather than activation, is neuroprotective. Secondly, the most commonly used Sirt1 activator, resveratrol, is likely to have neuroprotective properties that are independent of Sirt1 activation, such as its functioning simply as an antioxidant and ROS scavenger. Furthermore, the quantity of resveratrol administered through dietary means that could eventually become available to activate Sirt1 in CNS cells is unclear. Resveratrol may be protective through several cell signaling pathways, such as AMP-activated kinase (Hwang et al. 2008), and in fact activation of some of these pathways could result in apoptosis in some cell types. Importantly, as resveratrol also activates Sirt2, its effect on neurons is complicated by the effects downstream of both neuronal and oligodendroglia Sirt2 (Li et al. 2007). Finally, Sirt1’s neuroprotective activity may not be entirely dependent on its decetylase activity (Pfister et al. 2008).

The first caveat may simply reflect the complex nature of cellular events that could be modulated by Sirt1. As a nuclear-localized histone and transcription factor decetylase, Sirt1 modulates mid- to long-term changes in transcription profiles. As a cytoplasmic enzyme, Sirt1 could modulate signaling events (e.g., through deacetylation of IRS-2) or the activity of molecules that modulate cell survival (e.g., p53), which are more immediate in terms of time frame. The net effect of artificially increasing Sirt1 levels and activities in any particular cell type would therefore depend on the status, availability, and interplay between all its nuclear and cytoplasmic substrates, evolving over time. In the face of this complexity, conflicting, context-dependent outcomes from different experimental settings are only to be expected. Another pair of examples in the recent literature on Sirt1 aptly illustrates this point. Prozorovski et al. 2008 showed that oxidative stress and inflammation promote differentiation of neural progenitor cells (NPCs) towards the astroglia lineage through Sirt1 activation, which suppresses transcription of the proneuronal mammalian achate schute Homolog 1 (Mash1), through interaction with hairy and enhancer of split homolog-1 (Hes-1). On the other hand, Hisahara et al. 2008 found that a neuronal differentiation stimulus causes translocation of cytoplasmic Sirt1 into the nucleus to initiate neuronal differentiation, and that Sirt1 overexpression enhances neuronal differentiation by suppressing Hes-1. Such polarized conclusions with regard to Sirt1’s role on neural differentiation require further experimental reconciliation.

The second and the third caveats stated above indicate that in spite of the highly rewarding possibility of sirtuin activators (or perhaps inhibitors) like resveratrol as readily available therapeutic options in aging-associated neurodegenerative diseases, one needs to assess more carefully their pharmacodynamics and long-term effects in patients. This would also mean that short-term, crude neuronal cell survival-type of readouts to gauge drug efficacy would be inadequate. Changes in expression profiles of sirtuins (and indeed other relevant pro or antisurvival genes) in different parts of normal and diseased animal brain over time should be monitored in future experiments, and behavioral outcomes assessed in animal models whenever possible.