Published online before print January 23, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Biochemistry and Molecular Genetics, University of Virginia Health System, Charlottesville
1Correspondence: University of Virginia Health System, Department of Biochemistry and Molecular Genetics, Jordan Hall, Box 800733, Charlottesville, VA 22908. E-mail: jss5y{at}virginia.edu
Key Words: Sirtuin • NAD+ • S. cerevisiae • silencing
SIR2 was first identified as a gene that is required to maintain cell-mating type in S. cerevisiae by repressing the transcription of mating-type genes. This organism is able to grow vegetatively as three different cell types, the a and 
haploids and the a/ diploid. Cell mating-type determination has been extensively reviewed [1
]. In brief, the mating-type information encoded at the MAT locus determines the haploid cell type. Both haploid cell types contain additional cryptic copies of the genetic information required to exist as either cell type. However, the repositories for this cryptic mating-type information, the homothallic left (HML) and homothallic right (HMR)a loci, are held in check by a dynamic process called transcriptional silencing [2
]. SIR2, along with the other SIR genes (SIR1, -3, and -4), were identified as being essential for this form of silencing [3
].
These cryptic HM loci are flanked by cis-acting silencer DNA elements (Fig. 1 ). The silencers provide binding sites for the Rap1 and Abf1 proteins as well as the ORC complex, which act to recruit Sir2 along with the other Sir proteins to the HM loci. Several physical interactions between the Sir proteins and the silencer-binding proteins have been observed [4 ]. In addition, Sir proteins can homodimerize and heterodimerize, which together with the silencer-binding proteins, create a nucleoprotein structure necessary for the establishment of silencing [4 , 5 ]. This recruitment model is substantiated by studies showing that artificially targeting the Sir proteins to the HM loci in the absence of a silencer is sufficient to establish silencing [6 7 8 ]. In addition to HM silencing, Sir2 functions in the repression of genes located in subtelomeric regions [9 , 10 ] (Fig. 1) . This form of silencing, called telomere position effect (TPE), is considered very similar to HM silencing as a result of the shared requirement for Sir2, Sir3, and Sir4 [10 ]. The "silencer" for TPE is the telomeric repeat array that is bound by multiple copies of Rap1 (for review, see ref. [2 ]).
 View larger version (30K): [in a new window] |
Figure 1. Transcriptional silencing in S. cerevisiae. Sir2 histone deacetylase is incorporated into two distinct protein complexes: Sir and regulator of nucleolar silencing and telophase (RENT; center left). The Sir complex recognizes HM silencers bound by combinations of origin recognition complex (ORC), repressor activator protein 1 (Rap1), and autonomous replicating sequence-binding factor 1 (Abf1; top) and multiples of Rap1 bound to telomeric repeats (center right). The Sir complex spreads from the silencer by interacting with histone tails deacetylated by Sir2. Additional Sir complexes can potentially associate with the silent loci after spreading (indicated by the double arrow) [21
]. The RENT complex interacts with the polymerase I (Pol I) transcriptional machinery, which propagates spreading (bottom right). RENT also interacts with the replication fork-block protein (Fob1), which is necessary for silencing downstream of Pol I transcription.
|
The nature of the silent state was originally viewed as a static structure that denies access of transcription factors to the promoter. More recently, silencing has been recognized to be more dynamic. For example, the promoters of genes embedded in silent chromatin in S. cerevisiae are permissive to activator binding and preinitiation complex recruitment, suggesting that the silent chromatin represses gene expression downstream of Pol II binding [26 ]. Dynamics in maintaining the silent state are also suggested by the sensitivity of transcriptional silencing to levels of Sir proteins, including Sir2 [6 , 19 , 27 ]. Furthermore, the boundary between silenced and nonsilenced domains can be altered by competing histone H4 K16 acetylation and deacetylation by something about silencing (Sas2) and Sir2, respectively [28 , 29 ]. The residence time of Sir2 at the silent domains is not known, but heterochromatic proteins in mammalian cells are known to be dynamic [30 ].
Sir2 is observed in the nucleus within telomeric foci and the nucleolus [36 ]. Additionally, two distinct complexes arise following purification of Sir2 [5 , 37 , 38 ]: the telomeric SIR complex, discussed above, and the nucleolar RENT complex [37 , 39 ]. RENT contains Net1, which interacts with Sir2 and Cdc14 (a phosphatase involved in the exit from mitosis pathway) [37 , 39 ]. Net1 is required for the association of Sir2 with the rDNA locus and is required for rDNA silencing [37 ]. Net1 also interacts with the Pol I transcriptional machinery and is required for the normal levels of rRNA synthesis [40 ]. These results suggest that Net1 and Pol I act to tether Sir2 to the rDNA. This hypothesis is supported by recent findings that RNA polymerase I is required for rDNA silencing of Pol II reporter genes [41 , 42 ] and that Sir2 immunoprecipitates with RNA polymerase I [43 ]. Although Net1 influences Pol I transcription, it is worth noting that sir2 mutant cells have no significant defect in Pol I transcription [44 ].
The mechanism of rDNA silencing mediated by the RENT complex involves spreading, although Sir3 and Sir4 are not involved. Evidence for Sir2 spreading in the rDNA comes from two recent studies [42 , 43 ]. First, silencing was found to spread unidirectionally from the rDNA into the unique sequences flanking the centromeric side of the locus. Overexpression of Sir2 extended the size of the silent region [42 ], a finding similar to the effect of overexpressing Sir3 at telomeres [19 ]. rDNA silencing and Sir2 spread in the same direction as Pol I-mediated rRNA transcription. These findings are fully consistent with the interaction of the RENT complex with Pol I and suggest that Sir2/RENT may track along the rDNA in association with transcribing Pol I. Support for the transcription-coupled chromatin modification by Sir2 comes from the finding that deletion of a single Pol I promoter results in a specific loss of silencing at the mutated rDNA repeat [42 ]. In addition, chromatin immunoprecipitation (ChIP) analysis showed that Sir2/RENT is concentrated over the Pol I promoter and replication fork-block/Pol I-terminator regions and appears to spread in the direction of Pol I transcription [43 ]. An association with the Fob1 protein, which is also required for rDNA silencing [43 ] (S. W. Buck and J. S. Smith, unpublished), mediates recruitment of Sir2/RENT to the replication fork-block/terminator region. Sir2 may be recruited to the rDNA by two separate mechanisms, but spreading in both cases is dependent on Pol I transcription.
As mentioned above, Sir2 is also responsible for silencing Pol II reporter genes in the rDNA. However, it appears that the effect of Sir2 on longevity is more closely linked to its suppressive effects on recombination than to rDNA silencing. rDNA silencing and the suppression of recombination can be separated. Deletion of the SGS1 DNA helicase gene shortens the lifespan and induces the accumulation of ERCs [48 , 52 ], but rDNA silencing is largely unaffected [34 ]. In addition, deletion of FOB1 extends the lifespan and prevents ERC formation [50 ] but actually causes a loss of Pol II–reporter gene silencing [43 ]. Finally, the Sir2-activating compound, resveratrol, extends the yeast lifespan and reduces rDNA recombination frequency but has little effect on rDNA silencing [53 ]. These findings don’t rule out that silencing Pol II genes could influence longevity, but it appears that ERC-mediated senescence predominates. This type of aging mechanism is clearly not the only way cells age. For example, petite yeast cells (with defective mitochondria) have a high level of ERCs and a long lifespan [54 , 55 ], indicating that petites can somehow escape this mechanism of death.
Various genes and conditions involved in energy metabolism affect the ability of Sir2 to regulate the lifespan. Caloric restriction or the reduction of glycolysis by a hexokinase or protein kinase A pathway mutation extends the lifespan with a concomitant augmentation of Sir2-dependent rDNA silencing [56
, 57
]. Mutations that directly affect electron transport in the mitochondria demonstrate that respiration plays a key role in the function of Sir2 in lifespan control and silencing of rDNA [58
]. Activating genes involved in respiration (by overexpressing Hap4) increases lifespan and rDNA silencing [58
]. Eliminating electron transport (with a cytochrome c1 mutation) prevents the salutary effects of caloric restriction. It has been proposed that the increase in respiration that occurs during caloric restriction would lead to an elevation in the intracellular NAD+/reduced NAD (NADH) ratio, which would activate Sir2. Support for this hypothesis comes from the findings that mutations in the NAD+ salvage pathway gene, nicotinate phosphoribosyl transferase 1 (NPT1; see Fig. 5
), cause a loss of rDNA silencing and reduce the intracellular NAD+ concentration 
2.5-fold [59
]. Furthermore, NPT1 is required for the extension of the lifespan caused by caloric restriction [57
]. HAP4 overexpression does elevate the intracellular NAD+ concentration [60
]. However, calorie-restricted yeast cells actually have slightly lower nuclear concentrations of NAD+ than cells grown in normal growth medium [60
, 61
]. In addition, the in vitro deacetylase activities of yeast Sir2 and its human homologue SIRT1 are not significantly affected by alterations in the NAD+/NADH ratio [60
].
 View larger version (18K): [in a new window] |
Figure 5. The de novo NAD+ biosynthesis and NAD+ salvage pathways in S. cerevisiae. In eukaryotes, the de novo pathway begins with tryptophan (Trp), which is converted to NaMN by a series of enzymatic reactions performed by Bna1–Bna6. The Nma1 and Nma2 NAM mononuceotide adenylyltransferases then convert NaMN to NaAD, which is converted to NAD+ by NAD+ synthetase (Qns1). Sir2 and NAD+ glycohydrolases can hydrolyze NAD+ into NAM and ADP ribose (or the acetylated version). The NAM generated by Sir2 is recycled into NaMN by the Pnc1 nicotinamidase and the Npt1 nicotinic acid (NA) phosphoribosyltransferase, which make up the NAD+ salvage pathway. NA can also be taken into the cell through a NA permease called Tna1.
|
The role of Sirtuins in regulating the lifespan extends beyond yeast, as an extra copy of Caenorhabditis elegans sir-2.1 extends the lifespan [67 ]. Homozygous Drosophila Sir2 (dSir2) mutants have a moderately shorter lifespan in Drosophila, although a separate study found no decrease in longevity [68 , 69 ]. Overexpression of dSir2 in Drosophila has not been reported, making it is difficult to compare the deletion phenotypes with the overexpression experiments performed with the C. elegans study. It is still unclear if mammalian Sirtuins have a role in longevity, although the SIRT1 mutant mice are smaller and infrequently survive postnatally [70 , 71 ].
 View larger version (69K): [in a new window] |
Figure 2. ClustalW alignment of various Sirtuins. Af1_Sir2: Archaeoglobus fulgidus Af1 Sir2; Af2_Sir2: A. fulgidus Af2 Sir2; Hs_SIRT1: human SIRT1; Sc_Sir2: S. cerevisiae Sir2; Sc_Hst1: S. cerevisiae homologous to Sir2 (Hst)1. The secondary structure elements are based on the Af1 Sirtuin structure shown in Figure 5
[72
]. The color-coding is as follows: Rossman-fold, blue; helical domain, green; flexible loop region, gray/green dashes; zinc ribbon, red.
|
Three laboratories then independently reported that Sir2 (along with Hst1–4, murine Sir2, and CobB) could catalyze the removal of acetyl groups from lysines on histone tails [79
80
81
]. This histone deacetylation (HDAC) activity was NAD+-dependent, and Sir2 and Hst2 were found to cleave the glycosidic bond between the NAM and ADP-ribose portions of NAD+ [66
, 82
] (Fig. 3
). To further solidify the role of NAD+ in the function of Sir2, a mutation in NPT1, a gene involved in the NAD+ salvage pathway, resulted in a loss of telomeric and rDNA silencing (ref. [79
], see below). Additionally, it was found that the HDAC activity of Sir2 was much stronger than its ADP-ribosylation activity and is believed to be the major Sirtuin activity [80
, 81
]. Of interest is that S. cerevisiae Sir2 was found to prefer histone H3 acetylated at lysines 9/14 and histone H4 acetylated at lysine 16 as substrates [81
]. Deacetylated K16 on H4 happens to be critical for silencing [17
, 85
]. Overexpression of Sir2 also results in global histone hypoacetylation [16
]. The in vivo relevance of the ADP-ribosylation activity of Sir2 currently remains a subject of debate, although it has been speculated to be involved in DNA double-stranded break repair [81
].
 View larger version (15K): [in a new window] |
Figure 3. Scheme for the Sirtuin deacetylation reaction. The scenario proposed by Sauve et al. [83
] is shown in the top pathway. The scenario proposed by Jackson et al. [84
] is shown in the bottom pathway. Both lead to the production of 2'-O-acetyl-ADP-ribose (2'-O-AADPR). The position of the glycosidic bond that is cleaved to release NAM is indicated.
|
The reaction is believed to proceed in a manner in which the C1' hydroxyl of the ribose moiety captures the acyl oxygen of the acetyl lysine and generates a 1'-O-alkylamidate intermediate (Fig. 3) [83 , 84 ]. Alternatively, the 2'-hydroxyl of the ribose ring could become activated by interactions with a nearby basic residue, possibly the previously mentioned conserved histidine and attack the amide carbonyl of the acetyl lysine, resulting in an acetonide intermediate (Fig. 3) [84 ]. Consistent with this possible mechanism, an interaction between this histidine and the 3'-hydroxyl of the NAD+ molecule was observed in a crystal structure of the Archeal Sir2–Af1 [87 ]. In the first scenario, the 2' hydroxyl of the ribose attacks the 1'-O-alkylamidate intermediate to form a cyclic acyloxonium intermediate, which is then hydrolyzed to form 2'-O-AADPR. In the second scenario, the acetonide intermediate is hydrolyzed to form the 2'-O-AADPR product (Fig. 3) . Regardless of the initial reaction step, the action of the 2' hydroxyl is necessary for catalysis. As mentioned above, the formation of 3'-O-AADPR was also observed. This product was found to occur as a result of a Sirtuin-independent transesterification of the primary 2'-O-AADPR product to 3'-O-AADPR; these two regioisomers reach an equilibrium [83 , 84 ].
This proposed reaction mechanism is further supported by revelations from crystallization of Sir2–Af1 complexed with NAD+ [72 ]. NAD+ is located in a region devoid of residues that could perform a nucleophilic attack on the NAD+ and allow formation of the 1'-O-alkylamidate intermediate. Additionally, NAD+ is positioned in such a way that it could be accessed by the acetyl lysine, through the proposed substrate-binding site, to form the 1'-O-alkylamidate intermediate [72 , 83 , 84 ].
 View larger version (45K): [in a new window] |
Figure 4. Comparison of the Archeal Af1 and Af2 X-ray crystal structures. The Af2 structure complexed with an acetylated p53 peptide substrate (protein database coordinates 1MA3; ref. [89
]) and the Af1 structure complexed with NAD+ (protein database coordinates 1ICI; [72
]) were generated with RasMol software. The different regions of Sir2 are color-coded as follows: blue, Rossmann-fold; green, helical module; red, zinc-ribbon domain. A light-blue sphere indicates the zinc2+ atom. In the Af2 structure, the p53 peptide is shown in orange, and the lysine side-chain that is deacetylated is shown in yellow. In the Af1 structure, the NAD+ molecule is shown in orange. A black arrow points to the loop in Af1 that is unstructured in the Af2 structure.
|
The Wolberger and Boeke laboratories [89 ] published the structure of Sir2–Af2 in complex with an acetylated peptide corresponding to a region of p53, previously shown to be deacetylated by SIRT1 [90 ]. The p53 peptide was found to lie in the previously suggested substrate cleft between the Rossmann-fold and the zinc ribbon [72 , 89 ]. The peptide formed the middle strand of an antiparallel ß-sheet that was flanked by a ß-strand from the Rossmann-fold and a second strand from a loop between the Rossmann-fold and the zinc-binding region, termed the FGE loop, as it contains the conserved FGExL motif [89 ]. As the peptide completes a ß-sheet that brings the two regions of Sir2–Af2 together, they have termed this linkage a ß-staple. The acetyl-lysine was found to adopt an extended conformation that allows it to insert into a cleft and point toward the position of the NAD+-binding site. Additionally, the acetyl-lysine was found to lie on top of the highly conserved histidine mentioned previously, forming a hydrogen bond with the FGE loop, thus linking the FGE loop to the helical module that contains residues necessary for NAD+ binding. The authors therefore suggested that there could be some cooperation between NAD+ and substrate binding [89 ].
Most eukaryotic Sirtuins have N- and/or C-terminal regions that extend beyond the conserved core domain. The function of these protein regions is largely unknown. However, in a recent X-ray crystal structure of full-length yeast Hst2 protein, a C-terminal helix (13) binds in a cleft between the Rossman-fold and helical domains and makes multiple contacts with residues in both domains [91
]. When the Hst2 and Af1/NAD+ crystal structures were superimposed [72
, 91
], the 13 helix partially overlapped with the NAM and ribose ring of NAD+ within the Af1 structure. It appears that the C terminus of Hst2 is autoinhibitory, as deleting the C terminus of Hst2 significantly decreases the Michaelis constant (Km) value for NAD+ [91
]. The N terminus of native Hst2 makes an extended loop that interacts in the putative peptide-binding site of an adjacent symmetry-related Hst2 molecule and mediates the formation of a homotrimer [91
]. Deletion of the N-terminal seven amino acids of Hst2 causes a loss of trimer formation and also decreases the Km for an acetylated histone H4 peptide substrate by approximately threefold [91
]. Hst2 is therefore also autoregulated by its N terminus. It will be interesting to see if this type of autoregulation by the N- and C-terminal domains of eukaryotic Sirtuins is a common characteristic.
2.5-fold [59
, 79
]. These results suggest that the silencing defects are a result of a lack of available NAD+ for Sir2 to perform its deacetylation function. NPT1 is also required for the extension of lifespan induced by caloric restriction, which is a Sir2-dependent process [57
]. Furthermore, overexpression of NPT1 or another NAD+ salvage gene, PNC1, strengthens silencing and extends lifespan [62
, 63
], and deletion of PNC1 weakens rDNA and telomeric silencing [59
]. In contrast, mutations in the de novo NAD+ synthesis genes, BNA1–BNA6, have very little effect on silencing or longevity [57
, 59
, 63
]. One of the curiosities observed in studies about the NAD+ salvage pathway was that deletion of PNC1, a gene encoding a nicotinamidase, caused a loss of silencing but did not cause an accompanying decrease in cellular NAD+ concentration [59 ]. Pnc1 is responsible for deamidating the NAM produced by the Sir2 deacetylation reaction and converting it into NA, which is then converted to NaMN by Npt1 (Fig. 5) . A previous study had found that NAM was a noncompetitive inhibitor of Hst2 HDAC activity in vitro [66 ]. Following up on that observation, the Sinclair laboratory [25 ] found that inclusion of NAM in the growth media impaired the ability of Sir2 to silence and extend the lifespan. Additionally, they determined that NAM was a noncompetitive inhibitor of S. cerevisiae Sir2 and human SIRT1 in vitro [25 ]. They proposed that inhibition occurs because NAM occupies a subsite of the NAD+-binding pocket within the Sirtuin proteins in which the NAM moiety of NAD+ would normally be cleaved from the ADP-ribose moiety [25 ]. NAM binding to this subsite would prevent cleavage. More recently, it has been suggested that NAM inhibits Sirtuin function by reversing the forward reaction [92 ]. High concentrations of NAM would drive base reversal of the covalent intermediate, mentioned above, back into NAD+ and acetyl-lysine [92 ].
PNC1 is up-regulated during caloric restriction and other mild stress conditions [62 , 64 ] and is necessary for lifespan extension by caloric restriction [62 ]. Furthermore, in vivo depletion of NAM through the overexpression of NAM N-methyltransferase leads to the extension of the lifespan and stronger rDNA silencing [62 ]. Together, these results indicate that Pnc1 activates Sir2 through the conversion of NAM to NA. In support of this hypothesis, Pnc1 overexpression can suppress the inhibitory effects of exogenously added NAM on silencing and longevity [65 ]. Pnc1 also prevents the inhibition of in vivo Hst1 activity by NAM [65 ]. As Pnc1 is phylogenetically conserved, it is therefore reasonable to propose that Pnc1 acts as a general regulator of Sirtuin activity by removing NAM and preventing inhibition of protein deacetylation. Furthermore, during times of stress when Sirtuins may be acting to promote cell viability, Pnc1 could help maintain elevated Sirtuin activity.
Another known in vivo histone deacetylase in S. cerevisiae is Hst1, which is most similar of the four Hst proteins to Sir2 [73
, 74
]. Overexpression of HST1 can partially suppress the silencing defect of a sir2
mutant at the HMR locus [73
]. However, one of the normal functions of Hst1 is to assist in the repression of middle-sporulation genes during vegetative growth [94
]. Hst1 forms a complex with the Sum1 repressor [95
, 96
], which is recruited to the promoters of specific genes that are normally active during mid-meiosis but are repressed by Sum1/Hst1 during vegetative growth [94
]. More recently, Hst1 was found to be responsible for the repression of de novo NAD+ biosynthesis genes and the TNA1 NA importer gene as part of a feedback loop (ref. [97
], see below). In the cases of meiotic genes and NAD+ biosynthesis genes, alterations in H3 or H4 acetylation were not observed by chromatin immunoprecipitation assays in an hst1 mutant [96
, 97
]. These results suggest that the deacetylated chromatin regions are very small and undetectable by ChIP, or Hst1 deacetylates nonhistone substrates at these promoters. However, as SIR2 overexpression can partially suppress the lack of repression in an hst1 mutant, Hst1 likely acts as a HDAC [94
]. Further evidence for Hst1 being a bona fide histone deacetylase comes from experiments with sum1-1 mutants that suppress the silencing defects of sir mutants by recruiting Hst1 to the silent MAT loci. In this circumstance, Hst1 was clearly found to deacetylate histones H3 and H4 by chromatin IP [96
, 98
].
Another recent example of a possible histone deacetylase in this protein family comes from Schizosaccharomyces pombe, which has three separate SIR2-like genes [99 , 100 ]. One of these, hst4+, is similar to HST3 and HST4 of S. cerevisiae, and mutants for each of these genes have abnormal morphology, silencing defects, and elevated chromosome-loss rates [73 , 99 ]. However, despite the defects in silencing, it is still unclear if these Sirtuins actually deacetylate histones in vivo. In contrast, the sir2+ gene of S. pombe was found to be required for heterochromatic silencing and specifically to deacetylate K9 of histone H3 in vivo and in vitro [100 ]. The specificity for H3-K9 is important, as heterochromatin in S. pombe is characterized by deacetylation of K9, which is required for the methylation of this H3 residue by a SET domain-containing histone methyltransferase, termed Clr4 or Su(var)3-9 in Drosophila [101 ]. Heterochromatin protein 1, which is Swi6 in S. pombe, specifically binds to the methylated K9 residue to help establish silencing [102 103 104 ]. In fact, most eukaryotes establish heterochromatin through this mechanism of K9 methylation [105 ]. Conversely, S. cerevisiae does not have detectable methylation of H3-K9, indicating that the requirement for Sir2-mediated histone deacetylation in heterochromatin formation has been conserved between S. cerevisiae and S. pombe despite the nonconservation of the histone H3-K9 methylation.
One of the five Drosophila Sirtuins (dSir2) functions in heterochromatic position-effect variegation but not telomeric silencing [68 , 69 , 106 ]. dSir2 can also deacetylate a histone H4 N-terminal tail peptide in vitro [106 ]. Taken together, these results suggest that dSir2 may be a histone deacetylase in vivo that has some analogous functions with S. cerevisiae Sir2, which is its closest yeast homologue. As with S. pombe sir2+, dSir2 colocalizes with heterochromatic histone H3-K9 methylation, although not exclusively. It also associates with euchromatin as a result of its association with several basic helix-loop-helix repressors [106 , 107 ].
In human cells, the closest relative to yeast Sir2 is the SIRT1 protein [75
]. However, this protein has not been shown to be involved in any kind of heterochromatin-like gene-silencing. SIRT1, which is called SIR2 in mouse, is highly capable of deacetylating histones in vitro [25
], indicating that it has the potential to be a histone deacetylase in vivo. One group has deleted SIR2 from mouse embryonic stem cells and generated knockout (KO) mice but found no effect on global gene silencing or global histone acetylation [71
, 108
]. SIRT1 and SIR2 are nuclear proteins, but they do not obviously localize to heterochromatin [108
, 109
]. To more closely test if SIRT1 or SIR2 is involved in heterochromatin formation, it would be interesting to examine the effect of the SIR2 KO on the pattern or intensity of histone H3-K9 methylation and H3-K9 acetylation at known heterochromatic, genomic locations such as the inactive X chromosome or imprinted loci. All of the human Sirtuins have been tested directly for general histone deacetylation activity in vitro, and only SIRT1, SIRT2, SIRT3, and SIRT4 were found to have robust activity on a histone H4 peptide [110
]. However, this does not mean that the in vivo substrates for all four are actually histones. For example, the in vivo target of SIRT2 is actually -tubulin not histones (ref. [110
], see below).
in vitro and in vivo [90
, 109
, 111
]. Acetylation of p53 by cyclic AMP response element-binding protein-binding protein (CBP)/p300 or p300/CBP-associated factor (PCAF) stimulates its transcriptional activation activity through coactivator recruitment and possibly improved p53 stability (for review, see ref. [112
]). Deacetylation by SIRT1/SIR2 leads to inactivation of p53 and improved cell survival during times of cellular stress through inhibition of the p53-dependent apoptosis pathway [90
, 109
, 111
]. Consistent with this finding, mouse SIR2 KOs have defects in spermatogenesis with increased apoptosis in testes [71
]. It is interesting that multiple polyphenol compounds were recently identified as activators of SIRT1 in that they stimulate the deacetylation activity of SIRT1 on p53 in vivo, causing increased cell survival during exposure to UV light [53
]. Resveratrol, a specific polyphenol found in red wine, was the most potent activating compound identified, suggesting that some of the health benefits of moderate wine consumption could be a result of the effects of this compound on the Sirtuins [53
].
Another nonhistone substrate of SIR2 is the TBP-associated factor (TAF)I68 subunit of the transcription initiation factor IB/SL1 [113
]. This complex is responsible for the proper recruitment of RNA polymerase I to the rDNA promoter and transcription initiation. Acetylation of TAFI68 by PCAF enhances its binding to the rDNA promoter, thus stimulating rDNA transcription [113
]. Deacetylation of TAFI68 by SIR2 therefore results in repression of rDNA transcription in vitro. The effect of SIR2 on rDNA transcription in vivo has not yet been tested. In contrast, deletion of yeast SIR2 has very little effect on rDNA transcription [44
], but the Net1 subunit of the RENT complex does stimulate Pol I [40
].
CobB has histone deacetylase activity in vitro [79 ], but as bacteria do not have histones, the relevant substrate(s) were not going to be histones. Instead, CobB was found to be required for short-chain fatty acid activation through the regulation of acetyl-CoA synthetase [114 ]. Acetyl-CoA synthetase is post-translationally modified by acetylation at lysine-609 within the active site, which inhibits enzymatic activity [115 ]. The mechanism of acetylation remains unknown, but CobB is responsible for deacetylating K609 and activating the enzyme [115 ]. It remains to be determined if other acetyl-adenosine monophosphate (AMP) intermediate-forming enzymes are also regulated by acetylation/deacetylation, although this has been proposed to be the case [115 ]. CobB and human SIRT5 are class III Sirtuins [75 ]. SIRT5 also does not efficiently deacetylate histones [110 ], suggesting that it could potentially have a nonhistone substrate, perhaps acetyl-CoA synthetase. Most nongram-positive bacteria and Archea species have class III Sirtuins, but S. cerevisiae, Drosophila, and C. elegans are examples of eukaryotes that do not have this class of Sirtuin [75 ]. Therefore, if the acetylation of AMP-forming enzymes is phylogenetically conserved, then classes of Sirtuins other than class III will carry out those functions. In S. cerevisiae, for example, Hst3 and Hst4 have been implicated in the ability to grow on acetate or propionate medium, suggesting that they may target acetyl-CoA synthetase [114 ].
Human SIRT2, like its closest yeast homologue Hst2, is a largely cytoplasmic Sirtuin that has a role in the control of mitotic exit in the cell cycle [110
, 116
117
118
]. Overexpression of SIRT2 significantly prolongs mitosis [118
]. Furthermore, the abundance of endogenous SIRT2 dramatically increases during mitosis and becomes multiply phosphorylated at the G2/M transition. The CDC14B phosphatase appears to dephosphorylate SIRT2, which results in proteosome-dependent degradation and the promotion of mitotic progression [118
]. In a separate study, an in vivo target for SIRT2 deacetylation was found to be -tubulin [110
]. SIRT2 also colocalizes with microtubules [110
], consistent with its role in mitotic exit. The actual function of tubulin acetylation is not clear, but it is interesting that -tubulin is hyperacetylated within the microtubules of stable structures such as cilia [110
, 119
]. This suggests that deacetylation by SIRT2 could potentially control mitotic exit by modulating microtubule dynamics. It is surprising that although yeast Hst2 is also cytoplasmic [116
], it does not appear to be a tubulin deacetylase [110
], which is consistent with its ability to influence transcriptional silencing when overexpressed in yeast cells [116
]. Given the vast diversity of Sirtuin deacetylation targets discovered so far, more surprises are sure to be discovered.
Further correlative evidence for NAD+ sensing comes from SIRT1/SIR2, which was recently found to function in skeletal muscle differentiation in mice [71
, 120
]. In mouse cells, SIR2 forms a complex with the PCAF/GCN5 histone acetyltransferases (HATs) [120
]. Only when complexed with PCAF can SIR2 associate with and deacetylate MyoD, a key muscle cell-differentiation factor. PCAF is also deacetylated [120
]. Drosophila Sir2 has also been found to associate with the CBP acetyltransferase by yeast 2-hybrid analysis [68
], raising the strong possibility that Sirtuins may modulate the activity of HATs. If acetylation of PCAF stimulates its HAT activity, then some of the effects of SIR2 on histone acetylation status of MyoD-regulated gene promoters could be a result of the repression of PCAF-mediated acetylation of histones rather than direct histone deacetylation by SIR2. A very important aspect of this study was the observed correlation between the NAD+/NADH ratio in differentiating muscle cells and SIR2 function [120
]. The NAD+/NADH ratio normally decreases during muscle cell differentiation. Artificially elevated NAD+/NADH ratios inhibit muscle-cell gene expression, perhaps by up-regulating SIR2 deacetylase activity. Accordingly, cells with reduced SIR2 protein levels are less capable of the transcriptional repression normally triggered by the elevated NAD+/NADH ratio [120
]. By this model, high NAD+ concentrations activate SIR2, which leads to the repression of specific muscle-gene expression.
Human SIRT3 is actually in a cellular environment that is always bathed in NAD+: the mitochondia. More specifically, it is located in the mitochondrial matrix [121 , 122 ]. In contrast to the cytoplasm, where NAD+ levels can change in response to changes in adenosine 5'-triphosphate abundance, NAD+ concentrations in the mitochondrial matrix are relatively stable (see ref. [122 ]). Therefore, it remains unclear if SIRT3 acts as a bona fide NAD+ sensor. Perhaps the high concentration of NAD+ in the mitochondria simply ensures that SIRT3 remains in a highly active state [121 , 122 ].
Perhaps the most convincing support for the role of Sirtuins in sensing the intracellular NAD+ concentration comes from yeast Hst1, which can function directly in the repression of de novo NAD+ biosynthesis genes [97
]. When intracellular NAD+ concentrations are low (as in an npt1 mutant), the deacetylase activity of Hst1 is impaired, resulting in the up-regulation of the de novo NAD+ biosynthesis genes [97
]. In this case, Hst1 may act as an effective NAD+ sensor, as its Km for NAD+ is relatively high (94.2 µM) compared with a value of 29.7 µM for the Sir2 protein. Although the Km of SIR2 has not been reported, the results with Hst1 suggest that Sirtuins with the highest NAD+ Km values are likely candidates to be NAD+ sensors. This raises the question of what is the concentration of free cellular NAD+ that is available for interaction with Sir2. Unfortunately, there have been wide variations in this number. The concentration of total cellular NAD+ in yeast cells has been measured between 1.5 and 3 mM using enzymatic-based detection methods [61
, 97
]. The free NAD+ concentration was recently measured as 4 mM using 13C-nuclear magnetic resonance spectroscopy [60
]. However, a study in mammalian cells estimated the free nuclear NAD+ concentration to be 85 µM [123
], which if the same in yeast cells, would be consistent with Hst1 having the proper Km for being a good NAD+ sensor [97
]. As NADH does not appear to affect yeast Sir2 activity significantly [81
], the NAD+/NADH ratio may not be sensed by the Sirtuins as a redox state but perhaps just through NAD+ concentration.
In conclusion, the Sirtuins are well-positioned to link the metabolic status of the cell to the regulation of protein acetylation. As work on this and other classes of protein/histone deacetylases accelerates, this post-translational modification is becoming more generally distributed throughout biology as an important regulatory mark. It is possible that over time, it may rival phosphorylation in terms of functional significance in signal-transduction pathways. Having a set of deacetylases that can potentially respond to a change in cellular NAD+ concentration therefore adds another level of complexity to the already-complex, transcriptional regulatory networks and signaling cascades. Future work will undoubtedly extend the functional diversity of Sirtuins much further.
Received September 16, 2003; accepted December 17, 2003.
Mol. Cell. Biol. 18,5392-5403 protein has a role in embryogenesis and gametogenesis Mol. Cell. Biol. 23,38-54 protein has no effect on global gene silencing in mouse embryonic stem cells Mol. Cancer Res. 1,402-409 promotes cell survival under stress Cell 107,137-148[CrossRef][Medline]-tubulin in the microtubular cytoskeleton and primary cilia of hyaline cartilage chondrocytes J. Anat. 199,393-405[CrossRef][Medline]This article has been cited by other articles:
 |
|
 |
|
  N. Altman-Price and M. Mevarech Genetic Evidence for the Importance of Protein Acetylation and Protein Deacetylation in the Halophilic Archaeon Haloferax volcanii J. Bacteriol., March 1, 2009; 191(5): 1610 - 1617. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Pons, F. R. de Vries, P. J. van den Elsen, B. T. Heijmans, P. H.A. Quax, and J. W. Jukema Epigenetic histone acetylation modifiers in vascular remodelling: new targets for therapy in cardiovascular disease Eur. Heart J., February 1, 2009; 30(3): 266 - 277. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. G. Kazantsev and A. M. Kolchinsky Central Role of {alpha}-Synuclein Oligomers in Neurodegeneration in Parkinson Disease Arch Neurol, December 1, 2008; 65(12): 1577 - 1581. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Buelow, Y. Song, and A. M. Scharenberg The Poly(ADP-ribose) Polymerase PARP-1 Is Required for Oxidative Stress-induced TRPM2 Activation in Lymphocytes J. Biol. Chem., September 5, 2008; 283(36): 24571 - 24583. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Fournel, C. Bonfils, Y. Hou, P. T. Yan, M.-C. Trachy-Bourget, A. Kalita, J. Liu, A.-H. Lu, N. Z. Zhou, M.-F. Robert, et al. MGCD0103, a novel isotype-selective histone deacetylase inhibitor, has broad spectrum antitumor activity in vitro and in vivo Mol. Cancer Ther., April 1, 2008; 7(4): 759 - 768. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. K. Shwab, J. W. Bok, M. Tribus, J. Galehr, S. Graessle, and N. P. Keller Histone Deacetylase Activity Regulates Chemical Diversity in Aspergillus Eukaryot. Cell, September 1, 2007; 6(9): 1656 - 1664. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. S Parihar and G. J. Brewer Mitoenergetic failure in Alzheimer disease Am J Physiol Cell Physiol, January 1, 2007; 292(1): C8 - C23. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. E. Morrison, N. Majdzadeh, X. Zhang, A. Lyles, R. Bassel-Duby, E. N. Olson, and S. R. D'Mello Neuroprotection by histone deacetylase-related protein. Mol. Cell. Biol., May 1, 2006; 26(9): 3550 - 3564. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Tribus, J. Galehr, P. Trojer, G. Brosch, P. Loidl, F. Marx, H. Haas, and S. Graessle HdaA, a Major Class 2 Histone Deacetylase of Aspergillus nidulans, Affects Growth under Conditions of Oxidative Stress Eukaryot. Cell, October 1, 2005; 4(10): 1736 - 1745. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Kaeberlein, T. McDonagh, B. Heltweg, J. Hixon, E. A. Westman, S. D. Caldwell, A. Napper, R. Curtis, P. S. DiStefano, S. Fields, et al. Substrate-specific Activation of Sirtuins by Resveratrol J. Biol. Chem., April 29, 2005; 280(17): 17038 - 17045. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. A. Fox and K. H. McConnell Toward Biochemical Understanding of a Transcriptionally Silenced Chromosomal Domain in Saccharomyces cerevisiae J. Biol. Chem., March 11, 2005; 280(10): 8629 - 8632. [Full Text] [PDF]
|
|
|
|
 |
|
 P. J. Barnes, I. M. Adcock, and K. Ito Histone acetylation and deacetylation: importance in inflammatory lung diseases Eur. Respir. J., March 1, 2005; 25(3): 552 - 563. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. L. Freeman-Cook, E. B. Gomez, E. J. Spedale, J. Marlett, S. L. Forsburg, L. Pillus, and P. Laurenson Conserved Locus-Specific Silencing Functions of Schizosaccharomyces pombe sir2+ Genetics, March 1, 2005; 169(3): 1243 - 1260. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Araki, Y. Sasaki, and J. Milbrandt Increased Nuclear NAD Biosynthesis and SIRT1 Activation Prevent Axonal Degeneration Science, August 13, 2004; 305(5686): 1010 - 1013. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
YEAST SILENT INFORMATION...