Luận án tiến sĩ: Regulation of genomic stability by Saccharomyces cerevisiae sirtuins Hst3p and Hst4p

Table of Contents PageHst3 and Hst4 control histone H3 K56 acetylation during the cell cycle 26 The nicotinamide binding pocket of Hst3 is necessary for regulation of The phenotypes of

Regulation of genomic stability by S cerevisiae sirtuins

Hst3p and Hst4p

by

Ivana Celic

A dissertation submitted to The Johns Hopkins University in conformity with the

requirements for the degree of Doctor of Philosophy

Baltimore, MarylandJune, 2006

UMI Number: 3243300

3243300 2007

UMI Microform Copyright

All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company

300 North Zeeb Road P.O Box 1346 Ann Arbor, MI 48106-1346

by ProQuest Information and Learning Company

The baker’s yeast, Saccharomyces cerevisiae, has five members of this family, Sir2p and

Hst1-4p, important for regulation of transcriptional silencing and genomic stability

Hst3p and Hst4p are two redundant sirtuins with the major role in maintenance of

genomic stability They control genomic stability by regulating the level of acetylation of

histone H3 lysine 56 This residue, present in the core of the nucleosome surface, is

acetylated during the S phase of the cell cycle and contributes to the repair processes

active during DNA replication At the end of the S phase, K56 of histone H3 is

deacetylated in a Hst3p- and Hst4p-dependent manner Failure to deacetylate K56 leads

to a growth defect, sensitivity to DNA-damaging agents and chromosome loss in hst3

hst4 cells These phenotypes can be suppressed by mutation of K56 into the

nonacetylable residue arginine Failure to deacetylate K56 also leads to activation of the

DNA-damage response and renders hst3 hst4 cells sensitive to perturbations in DNA

replication, repair and checkpoint function as is evident from numerous synthetic lethality

interactions that hst3 hst4 cells display with mutations in the genes that regulate these

processes The growth defect of hst3 hst4 cells can be suppressed by overexpression of

RFC1, the large subunit of RFC, a “clamp loader” that loads PCNA, the “sliding clamp”

onto DNA during DNA replication Interestingly, the growth defect of hst3 hst4 cells can

also be suppressed by deletion of CTF18 and, somewhat less efficiently, RAD24 and

ELG1 CTF18, RAD24 and ELG1 all encode large subunits of alternative RFCs, each of

which shares four smaller subunits (Rfc2p-Rfc5p) with Rfc1p We propose that ongoing

cycles of K56 acetylation and deacetylation contribute to the regulation of the functional

equilibrium between different RFC complexes in the cell

Jef D Boeke, Ph D (Thesis Advisor and Reader)

Professor

Department of Molecular Biology and Genetics

Johns Hopkins University School of Medicine

Brendan Cormack, Ph D (Thesis Reader)

Associate Professor

Department of Molecular Biology and Genetics

Johns Hopkins University School of Medicine

Acknowledgements

This has been a very, very long ride and many people have helped me during this time in

different ways

I would like to thank my advisor, Jef Boeke, for giving me opportunity to work in his lab,

for his patience and support over all these years

I am grateful to my committee members, Carol Greider, Cynthia Wolberger and Brendan

Cormack for words of encouragement and letting me graduate after all

I am very grateful to Alain Verreault and his postdoctoral fellow Hiroshi Masumoto, who

originally discovered that Hst3p and Hst4p regulated histone K56 acetylation, for their

generosity in sharing results and willingness to collaborate Also, many thanks go to

Wendell Griffith, who has done some great MS analysis of K56 acetylation

I thank the Boeke lab members for making the Boeke lab nice place to work I had a great

opportunity to overlap with some exceptional graduate students, Eric Bolton, Jeffrey Han

and Siew-Loon Ooi, who thought me lot of little tricks of the trade

I thank to my friends, some close, some far away, who made these years more bearable

Special thanks goes to Jeff Han, for his support and many, many moments of happiness

he brought in my life

My deepest gratitude goes to my family, my parents, my sister and my grandmothers for

being always there for me no matter what My parents have sacrificed a lot to put me

through school and ultimately to get me here and I will be always indebted for that My

sister and my grandmothers helped as much as they could I am, also, very grateful to my

sister for taking good care of my parents It has given me great peace of mind over these

years

Table of Contents Page

Hst3 and Hst4 control histone H3 K56 acetylation during the cell cycle 26

The nicotinamide binding pocket of Hst3 is necessary for regulation of

The phenotypes of hst3 hst4 mutant cells are caused by high levels of

The histone chaperone Asf1p is needed for H3 K56 acetylation 32

viChapter 3: Cells lacking Hst3p and Hst4p activate the DNA damage response

due to the presence of chronic DNA damage and display genetic

hst3 hst4 cells slow down cell-cycle progression in response to

hst3 hst4 cells activate DNA damage checkpoint in the absence of

hst3 hst4 mutant depends on replication checkpoint for viability 69

Increased H2A serine 129 phosphorylation in hst3 hst4 cells 71

hst3 hst4 require subset of DNA repair proteins for viability 72Overexpression of Rfc1p, large subunit of DNA clamp loader,

Appendix B: Chemistry of gene silencing: the mechanism of

NAD+

Appendix C: Telomeric and rDNA silencing in Saccharomyces cerevisiae

are dependent on a nuclear NAD salvage pathway 147

Appendix D: SIRT3, a human SIR2 homologue, is an NAD+

-dependent

Appendix E: Structure of a Sir2 enzyme bound to an acetylated p53 peptide 168

Appendix F: Sir2-dependent activation of acetyl-CoA synthetase by

Figure 2.1 Hst3p and Hst4p control histone H3 K56 acetylation during

Figure 2.2 Effect of nicotinamide on the acetylation of histone H3 K56 40

Figure 2.3 Mutation in invariant residues of Hst3p affects K56 acetylation level

Figure 2.4 The phenotypes of hst3 hst4 double mutants are substantially

Figure 2.5 The hht1-K56R mutation suppresses the mitotic chromosome loss

Figure 2.7 Effects of rapid inactivation of Hst3p on H3 K56 deacetylation 47

Figure 2.8 Induction of Hst3p synthesis leads to H3 K56 deacetylation in mature

Figure 2.9 Abnormal distribution of histone H3 K56 acetylation with respect to

replication forks is likely responsible for the DNA damage sensitivity

Figure 3.1 Drugs sensitivities of hst3 hst4 mutants. 79

Figure 3.2 Like wild-type cells, hst3 hst4 cells slow down DNA replication

Figure 3.3 Like wild-type cells, hst3 hst4 cells do not elongate their mitotic

Figure 3.6 hst3 hst4 cells required functional DNA replication checkpoint

Figure 3.9 Overexpression of RFC1 suppresses growth defect, Ts phenotype

Figure 3.10 Overexpression of Rfc1p does not affect histone H3 K56

Figure 3.11 Suppression of hst3 hst4 growth defect and Ts phenotype by

Figure 3.12 Suppression of hst3 hst4 growth defect and Ts phenotype by

Table 3.1 List of the genes whose expression was changed in hst3 hst4 cells 92

1Chapter 1

Introduction

Protein acetylation and deacetylation have emerged over the past decade or two as

important posttranslational mechanisms regulating various aspects of cellular biology

This regulation is carried out through the action of protein acetylases, which modify the

ε-NH2 of lysine residues and their counterparts, the protein deacetylases, which remove

acetyl groups (Kouzarides, 2000; Kurdistani and Grunstein, 2003) The Sir2 family of

proteins is an integral part of this regulatory network Sir2 proteins deacetylate lysine

residues in histones and other proteins in a very unique enzymatic reaction that requires

nicotinamide adenine dinucleotide (NAD+

) (Imai et al., 2000; Landry et al., 2000b; Smith

et al., 2000) The deacetylation reaction catalyzed by Sir2 proteins is absolutely

dependent on NAD+

This is what differentiates Sir2 proteins as class III deacetylases,

distinct from class I and class II deacetylases that don’t require NAD+

and deacetylate

proteins through a more simple hydrolysis mechanism (Finnin et al., 1999) Presently, it

is not known why sirtuins deacetylate proteins in such an energetically costly mechanism,

but it has been suggested that sirtuins may represent a cellular NAD+

sensor linking

metabolism to other aspects of cellular physiology, like genome stability, aging,

apoptosis and differentiation

Sir2 proteins or sirtuins form a large and highly conserved protein family All three

kingdoms of life, archea, bacteria and eukaryotes, are represented within the sirtuin

family (Figure 1.1) (Frye, 2000) Some organisms have only one sirtuin, while others

have multiple paralogs, for example yeast with five (Brachmann et al., 1995) and

mammals with 7 paralogs (Frye, 1999; Frye, 2000) All sirtuins share a conserved

catalytic core domain necessary for the NAD+

-dependent deacetylation reaction Sir2

existent, for example in archeal sirtuins, to very long ones, like in S cerevisiae or

H.sapiens These extensions mediate member-specific functions like protein localization,

as in the case of human SirT3 which is targeted to mitochondria by its N-terminal

extension (Onyango et al., 2002; Schwer et al., 2002), formation of multiprotein

complexes with specialized function, as in the case of S.cerevisiae Sir2p, which exists in

the form of at least two functionally different protein complexes within a cell (Ghidelli et

al., 2001) or they have an autoregulatory function, as in the case of S cerevisiae Hst2p in

which N and C-terminal domain have inhibitory activity by binding to the catalytic core

intermolecularly and intramolecularly, respectively (Zhao et al., 2003)

Figure 1.1 Phylogenetic tree of sirtuins (published with permission by William Hawse

and Cynthia Wolberger)

5Sirtuins catalyze NAD -dependent deacetylation of lysine residues in proteins (Imai et

al., 2000; Landry et al., 2000b; Smith et al., 2000) During this reaction, the glycosidic

bond between the C1’ atom of ribose and nicotinamide moiety in NAD+

is cleaved to

generate, in addition to the deacetylated lysine residue, free nicotinamide and acetyl-ADP

ribose (APDR) (Sauve et al., 2001; Tanner et al., 2000; Tanny and Moazed, 2001)

NMR and mass-spectrophotometric analysis of the Sir2 deacetylation product revealed

that actual product is 2’- and 3’-acetyl-ADP ribose (Sauve et al., 2001) 2’-acetyl-ADP

ribose is the first product of the reaction followed by non-enzymatic conversion to

3’-acetyl-ADP ribose to a final equilibrium ratio of 47:67 for the 2’ and 3’ stereoisomers

Labeling experiments with H2

18

O, MS and NMR analysis led Sauve et al (Sauve et al.,

2001) to propose the mechanism (Figure 1.2), which involves the nucleophilic attack by

the carbonyl oxygen of the acetyl-lysine on the C1’ carbon of the nicotinamide ribose

(N-ribose) in NAD+

This step involves formation of the highly reactive riboxacarbenium ion

that captures the acyl oxygen of acetyl-lysine to generate a 1’-O-alkyl-amidate, followed

by the attack of the N-ribose 2’hydroxyl (2’OH) to form a 1’, 2’-acyloxonium structure,

which is the precursor for 2’-acetyl-ADP ribose formation According to the proposed

mechanism, nucleophilic attack by the N-ribose 2’OH to 1’-O-alkyl-amidate is facilitated

by a conserved histidine residue that acts either directly as a general base and activates

2’OH of ribose, or indirectly, through a proton relay mechanism involving 3’OH of

N-ribose Consistent with this, in the crystal structure of AfSir2 (Min et al., 2001), this

residue is positioned in the vicinity of the 3’OH of ADPR

In addition to NAD+

-dependent deacetylation catalyzed by Sir2 proteins, the

O-alkyl-amidate intermediate can be attacked by free nicotinamide and reverse the reaction to

reform NAD and acetylated lysine (Landry et al., 2000a; Sauve et al., 2001; Sauve and

Schramm, 2003) This reaction is termed nicotinamide exchange to reflect the fact that

nicotinamide, which makes an attack on the O-alkyl-intermediate, can be derived from

solvent and not (necessarily) as a product of deacetylation reaction The nicotinamide

reaction occurs on the opposite side of the N-ribose 2’OH attack on O-alkyl-intermediate

and explains why nicotinamide functions as noncompetitive inhibitor of Sir2 proteins

(Bitterman et al., 2002; Jackson et al., 2003; Sauve and Schramm, 2003)

Figure 1.2 Proposed catalytic mechanism of Sir2 proteins Histidine 118 from Af2Sir2

was depicted as a catalytic base (Af- Archeoglobus fulgidus).

7The ability of Sir2 proteins to catalyze nicotinamide exchange and the requirement for

the acetylated peptide in nicotinamide exchange reaction further support the existence of

O-alkyl-intermediate during the reaction and are consistent with findings that mutation of

the conserved histidine, that largely abolishes deacetylation activity in some sirtuins, still

preserves some nicotinamide exchange activity (Min et al., 2001)

Generally, Sir2 proteins don’t show high specificity towards acetylated substrates in vitro

(Avalos et al., 2002) Multiple members of the Sir2 family are able to deacetylate

p53-derived peptides A crystal structure of A fulgidus Sir2 in the complex with p53 peptide

(Avalos et al., 2002) revealed that the substrate is bound in the cleft between two

domains of the enzyme, the larger, an inverted Rossman fold and the smaller one

composed of the residues that form helical and zinc-binding modules (Avalos et al.,

2002; Finnin et al., 2001; Min et al., 2001) The bound peptide forms a β sheet-likeinteraction with the flanking strands of the protein, one belonging to the Rossman fold

and the other to the “FGE” loop (for the highly conserved FGEXL motif, which is a part

of zinc-binding module) Interaction between enzyme and the substrate are mediated

through atoms of the peptide main chain, rather than through the atoms of side chains of

peptide residues Acetylated lysine binds in the hydrophobic tunnel that contains residues

from Rossman fold and the “FGE” loop This binding positions the acetylated lysine in

the close proximity of the nicotinamide-ribose and conserved histidine (histidine 118 in

Af2Sir2) Comparison of the Af2Sir2 structure in the complex with p53 peptide (Avalos

et al., 2002) with the structure of apo-Sir2 (Finnin et al., 2001) and Af1Sir2 (Min et al.,

2001) in the complex with NAD+

revealed significant conformation differences which

suggested that the enzyme undergoes conformational change induced by substrate

binding As evident from the crystal structure of Af2Sir2-NAD+

complex (Avalos et al.,

2004), NAD+

binds in a manner that buries the nicotinamide moiety in a highly conserved

region of the protein, called the “C pocket” Positioning of the nicotinamide in the C

pocket is promoted by binding of the substrate into an acetyl-lysine binding tunnel The

nicotinamide moiety forms extensive contacts with conserved residues in the C pocket,

which accounts for the most highly conserved motifs in sirtuins (GAGXS, GIPXFR and

TQNIDXL) The observed conformation of NAD+

bound to Af2Sir2 differs from what

would be expected for the low-energy state of the molecule The planarity of the ribose

and nicotinamide in bound NAD+

is distorted In addition, the carboxamide group of

nicotinamide is rotated from its preferred position and forms an H-bond with the

invariant residue Asp103 These observations, together with the burial of the charge of

nicotinamide in a largely hydrophobic C pocket, suggest that NAD+

is destabilized upon

binding to the enzyme Energetic destabilization of NAD+

would allow the weak

nucleophile, like the carbonyl of acetyl-lysine to attack the C’1 of N-ribose leading to the

cleavage of the glycosidic bond and release of nicotinamide Crystal structure of Af2Sir2

bound to NAD+

has revealed the presence of the channel that connects the C pocket with

the solvent This channel potentially represents the site of nicotinamide release from the

enzyme upon the cleavage of the glycosidic bond, but is also a potential site of

nicotinamide entry from the solvent, providing the structural basis for nicotinamide

exchange reaction (Avalos et al., 2004) and noncompetitive inhibition of sirtuins by

nicotinamide (Bitterman et al., 2002; Jackson et al., 2003; Sauve and Schramm, 2003)

9Sir2 proteins have recently attracted lot of attention because of their role in aging In

yeast, S cerevisiae, deletion of SIR2 gene reduces, while extra copy of SIR2 gene extends

life span (Kaeberlein et al., 1999) Similarly, increase dosage of a SIR2 gene extends the

lifespan of C elegans (Tissenbaum and Guarente, 2001) and D melanogaster (Rogina

and Helfand, 2004) The requirement for NAD+

in Sir2-catalyzed reaction suggested a

possible molecular explanation for the longstanding observation that caloric restriction

(CR) extends lifespan in many organisms In yeast, growth under conditions of limited

glucose mimics CR and leads to the extension of lifespan (Lin et al., 2000) Deletion of

the genes in cAMP-dependent protein kinase A pathway (PKA), responsible for the

utilization of the glucose in the cell, results in a similar lifespan extension The effect on

lifespan extension in PKA pathway mutants is not affected by glucose availability

suggesting that glucose restriction and PKA pathway act in the same pathway affecting

yeast lifespan, therefore, manipulating PKA pathway provides a genetic mean to mimic

CR in yeast Using PKA pathway mutants, Lin et al (Lin et al., 2000) have shown that

the extension of lifespan in PKA pathway mutant is dependent on Sir2p and the product

of the NPT1 gene, which encodes a key enzyme in the NAD+

“salvage” biosynthetic

pathway (Rajavel et al., 1998) This led to the proposal that CR extends life span through

NAD+

-dependent activation of Sir2p (Lin et al., 2000) In addition, CR appears to extend

life span in yeast by shifting metabolism toward respiration (Lin et al., 2002)

Inactivation of electron transport resulted in the loss of life extension by CR, while its

activation led to Sir2p-dependent life extension Lin et al (Lin et al., 2002) have

proposed that shift toward respiration, induced by CR, increases NAD+

/NADH ratio

Consistent with that, growth on limited glucose or manipulating metabolism towards

respiration led to the decrease in intracellular NADH, but it had no effect on NAD levels

(Lin et al., 2004) This was further supported by the ability of NADH to act as a

competitive inhibitor of Sir2p and by manipulating intracellular levels of NADH through

the overexpression of NADH-reductase, which led to decrease in the cellular level of

NADH and extension of lifespan However, this theory has been challenged by reports

that regulation of Sir2p activity by nicotinamide, the noncompetitive inhibitor of Sir2

proteins, is the more relevant basis for lifespan extension through CR Deletion of PNC1

gene, which encodes for nicotinamide deaminase (Ghislain et al., 2002) and, presumably

reduces levels of nicotinamide in the cell, eliminates the effect of CR on life extension,

while increased copy number of PNC1 extends yeast life span up to 70% (Anderson et

al., 2003) in Sir2p-dependent manner Pnc1p is induced under condition of CR and other

stresses known to extend yeast life span Although PNC1 deletion was shown to reduce

NAD+

levels in the cell (Ghislain et al., 2002), several lines of evidence suggested that

Pnc1p affects life span by reducing intracellular nicotinamide concentration and thereby

its inhibitory effect on Sir2p (Anderson et al., 2003), rather than providing more NAD+

for Sir2p through salvage NAD+

biosynthetic step PNC1 overepression cannot be

mimicked by addition of extra nicotinic acid to the growing medium as would be

predicted by the latter hypothesis Provided they are grown in the presence of nicotinic

acid, yeast cells lacking PNC1 and de novo NAD+

biosynthetic pathway genes are viable,

in contrast to the cells lacking NPT1 and the genes of the de novo pathway,

demonstrating that the contribution of PNC1 to NAD+

biosynthesis in vivo is

nonessential Overexpression of PNC1 can restore silencing under conditions that

eliminate the “salvage” NAD+ biosynthetic pathway, in which Pnc1p normally

11participates Finally, overexpression of the nicotinamide N-methyl-transferase, which

leads to the excretion of nicotinamide from the cell, mimics the effect of PNC1

overexpression In support of a role for PNC1 in CR and regulation of Sir2p, Anderson et

al (Anderson et al., 2003) have shown, that, under conditions of CR, NAD+

concentration actually negatively correlates with Sir2p activity In addition, they have

been unable to observe inhibition of Sir2p and its human homolog, SirT1, by NADH The

attractive feature of Sir2 regulation by nicotinamide is that it does not need directly to

involve NAD+

concentration, as it is an important cofactor in so many enzymatic

reactions in the cell

Most recently, the role of Sir2p in CR-mediated life extention was challenged altogether

Work by Kaeberlein et al has shown that CR in yeast is independent of Sir2p and

mediated by the TOR signaling pathway (Kaeberlein et al., 2005) The discrepancy with

the previously published work (Anderson et al., 2003; Lin et al., 2000; Lin et al., 2004;

Lin et al., 2002) may stem from particular strain background or different experimental

conditions used to mimic CR in earlier work Sir2p remains a regulator of aging in yeast,

probably through a mechanism that controls rDNA recombination and extrachromosomal

circles of rDNA (ERCs) formation (Sinclair and Guarente, 1997)

At present, the targets of Sir2 proteins, important for CR-mediated life span extension in

higher organisms, are not fully understood Unlike S cerevisiae Sir2p, which

deacetylates histones, the mammalian ortholog, SirT1 targets several non-histone

proteins, with roles in apoptosis SirT1 deacetylates p53 (Langley et al., 2002; Luo et al.,

2001; Vaziri et al., 2001) and downregulates transcriptional activity and induction of

apoptosis by p53 Another family of proteins regulated by SirT1 is FOXO transcription

factors SirT1 deacetylates Foxo3 in response to oxidative stress (Brunet et al., 2004;

Motta et al., 2004), which in turn regulates its activity as a transcription factor, leading to

upregulation of the genes required for DNA repair and cell cycle arrest and to

downregulation of some proapoptotic genes SirT1 binds and deacetylates DNA repair

protein Ku70 Deacetylated Ku70 sequesters the proapoptotic protein Bax away from

mitochondria and promotes cell survival (Cohen et al., 2004) In addition, CR induces

SirT1 expression in rats and reduces Bax-mediated apoptosis (Cohen et al., 2004) An

exception to its anti-apoptotic role is the deacetylation of NF-κB by SirT1, which

sensitizes cell to TNFα-induced cell death (Yeung et al., 2004)

Negative regulation of PPARγ, transcription factor that regulates adipocytes

differentiation and fat storage, by SirT1 represents another link between SirT1 and CR

(Picard et al., 2004) SirT1 binds PPARγ and promoters of PPARγ responsive genes Thebinding of SirT1 to these genes in white adipose tissue is affected by food deprivation,

but it is not known whether PPARγ, histones in the PPARγ responsive genes or both areactually deacetylated by SirT1

Additional substrates of SirT1 include PGCα (Rodgers et al., 2005) and MyoD (Fulco etal., 2003) Deacetylation of PGCα by SirT1 regulates its transcriptional activity duringgluconeogenesis and glycolysis, while deacetylation of MyoD by SirT1 reduces muscle

gene expression and muscle differentiation In addition to deacetylating MyoD, SirT1

deacetylates histones in promoters of muscle-specific genes

In addition to SirT1, there are six additional Sir2 homologs in mammals SirT2 is a

cytoplasmic sirtuin that deacetylates tubulin (North et al., 2003) SirT3 is located in

mitochondria (Onyango et al., 2002; Schwer et al., 2002) It is proteolytically processed

in mitochondria, which leads to activation of its NAD+

-dependent deacetylation activity

(Schwer et al., 2002) and it is important for the regulation of membrane potential and

reactive oxygen species production (Shi et al., 2005) Additionally, SirT3 regulates

thermogenesis in brown adipocytes (Shi et al., 2005) SirT6 is a mammalian sirtuin with

strong ADP-ribosylation activity and no detectable deacetylation activity (Liszt et al.,

2005) The biological significance of SirT6 ADP-ribosylation activity is not known The

functions of SirT4, SirT5 and SirT7 remain to be identified

In the yeast S cerevisiae, in contrast to mammalian and bacterial systems, substrates of

sirtuins appear to be limited to histones S cereviae has five sirtuins, SIR2 and HST1-4

(Brachmann et al., 1995) Sir2p, first sirtuin discovered (Ivy et al., 1986; Rine and

Herskowitz, 1987; Shore et al., 1984), regulates transcriptional silencing, form of

epigenetic repression of large chromosomal regions Silenced regions in S cerevisiae

genome include the mating type loci (Rine et al., 1979), telomeres (Gottschling et al.,

1990) and ribosomal DNA (rDNA) (Bryk et al., 1997; Smith and Boeke, 1997) Silencing

at these loci is associated with Sir2p-mediated hypoacetylation of histones (Braunstein et

al., 1993; Braunstein et al., 1996; Huang and Moazed, 2003) Sir2p forms two complexes

in S cerevisiae Sir2p associates with Sir3p and Sir4p and regulates HM and telomeric

silencing (Ghidelli et al., 2001; Moazed et al., 1997; Strahl-Bolsinger et al., 1997) In the

nucleolus, Sir2p associates with Net1p and Cdc14p and regulates rDNA silencing (Shou

et al., 1999; Straight et al., 1999)

Assembly of silent chromatin at HM loci and telomeres is initiated by the recruitment of

silencing proteins to the HM loci and telomeres, followed by spreading of the silencing

proteins over larger chromosomal regions (Rusche et al., 2002) Interaction among

silencing proteins and between silencing proteins and histone tails underline the assembly

of silent chromatin HM loci contain binding sites for transcription factors Rap1p and

Abf1p and for ORC Sir proteins are recruited to HM loci through interaction with these

DNA-binding proteins Mutations in any two of the three binding sites at HMR-E result

in the loss of HM silencing (Brand et al., 1987) as well as mutations in the genes

encoding Rap1p, Abf1p and ORC (Foss et al., 1993; Loo et al., 1995a; Loo et al., 1995b;

Sussel and Shore, 1991), underlining the importance of these protein-DNA interaction in

silencing Rap1p, Abf1p and ORC bound to HM silencers provide docking sites for Sir

proteins Sir1, one of the four Sir proteins required for HM loci silencing (Ivy et al.,

1986; Rine and Herskowitz, 1987), is recruited to the HM silencers through its interaction

with ORC (Triolo and Sternglanz, 1996) and facilitates the recruitment of other Sir

proteins (Rusche et al., 2002), but it does not spread throughout HM loci (Rusche et al.,

2002), consistent with its role in the establishment of silencing, but not for its

maintenance (Pillus and Rine, 1989) Sir4p is recruited to silencers through its

interaction with Sir1p (Triolo and Sternglanz, 1996) and Rap1 (Moretti et al., 1994)

Sir4p can also bind silencers independent of Sir2p and Sir3p (Hoppe et al., 2002; Luo et

al., 2002; Rusche et al., 2002) Sir4p recruits Sir2p and Sir3p to the silencer Sir2p and

15Sir4p exist in a stable complex in the cell (Ghidelli et al., 2001; Hoppe et al., 2002), but

Sir3p is not typically associated in stoichiometric amounts with this complex, although

interaction between Sir3p and Sir4p are demonstrated by two-hybrid (Moretti et al.,

1994) or GST-pull down experiments (Moazed et al., 1997) Sir3p also interacts with

Rap1 (Moretti et al., 1994) Recruitment of Sir proteins does not require catalytic activity

of Sir2p (Hoppe et al., 2002; Rusche et al., 2002)

Similar hierarchy of Sir protein assembly exists at telomere Sir proteins are recruited to

telomere through interactions between Sir4p and Rap1p and yKu70p bound to the

telomere (Cockell et al., 1995; Conrad et al., 1990; Gotta et al., 1996; Laroche et al.,

1998; Luo et al., 2002; Mishra and Shore, 1999; Strahl-Bolsinger et al., 1997) An

exception is that Sir1 and ORC are not required for telomeric silencing (Aparicio et al.,

1991) However, Sir1p appears to be important for silencing at some natural telomeres

(Fourel et al., 1999; Pryde and Louis, 1999) Once Sir proteins are assembled on a

silencer, they spread outward from the initial binding site Protein interactions among Sir

proteins are important for the spreading of the Sir complex Although Sir4p binding to

Rap1p is not dependent on Sir2p and Sir3p, efficient spreading of Sir complex requires

all Sir proteins (Hoppe et al., 2002; Luo et al., 2002; Rusche et al., 2002) An important

characteristic of the Sir complex is its ability to bind histone tails and mutations that

abolish this interaction cause loss of silencing (Hecht et al., 1995) Sir proteins

preferentially bind hypoacetylated histones (Carmen et al., 2002; Liou et al., 2005) and

NAD+

-dependent deacetylation activity of Sir2p is required for spreading of the Sir

complex (Hoppe et al., 2002; Rusche et al., 2002) These finding support the model for

silent chromatin assembly according to which recruitment of Sir complex to the silencer

allows initial deacetylation of histone tails by Sir2p, creating a high affinity binding site

for the Sir complex adjacent to the silencer What follows is the next round of

deacetylation and binding of another Sir complex to the high affinity binding site newly

created by deacetylation The cycle of neighboring histone deacetylation followed by Sir

complex binding allows the Sir complex to spread over several kilobases of DNA and

also helps explain the maintenance of the silencing during DNA replication and new

histone deposition K16 of histone 4 is an important target of Sir2p Sir2p preferentially

deacetylates K16 in vitro (Imai et al., 2000) and Sir binding to H4 peptide is diminished

if K16 is acetylated (Liou et al., 2005) These results are supported by ealier mutational

analyses of H4 that revealed a complete loss of silencing when K16 of histone 4 was

mutated (Park and Szostak, 1990; Thompson et al., 1994) Recent finding revealed

another important feature of Sir complex assembly The product of NAD+

-dependent

deacetylation by Sir2p, O-acetyl-ADP-ribose, increases number of Sir3p molecules in Sir

complex and induces conformational change of Sir complex (Liou et al., 2005), which is

important for Sir complex assembly

An alternative complex of Sir2p, Net1p/Cfip, and Cdc14p, calld the RENT complex,

regulates rDNA silencing (Shou et al., 1999; Tanny et al., 1999) rDNA silencing is a

special form of silencing that affects Pol II-transcribed genes inserted in rDNA array

(Bryk et al., 1997; Smith and Boeke, 1997) This form of silencing does not require Sir3p

and Sir4p (Smith and Boeke, 1997), but requires enzymatic activity of Sir2p (Hoppe et

al., 2002) and all four histones (Bryk et al., 1997) Additionally, various chromatin

17modifying proteins and DNA replication factors are important for rDNA silencing (Smith

et al., 1999) The S cerevisiae rDNA consists of 9.1 kb sequence tandemly repeated

100-200 times Sir2p has been shown to preferentially associate with the NTS region relative

to the 35S coding region Consistent with Sir2p association with NTS, hypoacetylation of

histone is observed in this region (Huang and Moazed, 2003) It was initially proposed

that Sir2p controls ratio between the open repeats, associated with gene expression, and

closed repeats, that are refractive to gene expression But an alternative mechanism has

been proposed in which transcription from pol I promoter may be favored through

inhibition of pol II promoter The pol I promoter contains cryptic pol II promoter

(Conrad-Webb and Butow, 1995) that is specifically inhibited by pol I-specific

transcription factor (Oakes et al., 1999; Siddiqi et al., 2001) A similar mechanism may

regulate expression of pol II-driven transgenes, although it is not clear how Sir2p could

participate in this process

Regulation of recombination between rDNA repeats is another important function of

Sir2p in the nucleolus (Gottlieb and Esposito, 1989; Kaeberlein et al., 1999) Each rDNA

repeat contains origin of replication and excised repeat can replicate as a plasmid

Accumulation of extrachromosomal circles of rDNA, known as ERCs, is the feature of an

aging yeast cell Although they can replicate, ERCs don’t contain centromere and are not

efficiently distributed into emerging daughter cell As a result, aging mother cell retains

large number of ERCs (Sinclair and Guarente, 1997) Sir2p controls formation of ERCs

Cells lacking Sir2p accumulate large number of ERCs and have shortened life span

(Kaeberlein et al., 1999) Catalytic activity of Sir2p is required to control ERCs formation

(Armstrong et al., 2002) Although accumulation of ERCs correlates with aging, isolation

of mutants that affect life span without ERCs formation, demonstrates existence of

additional pathway/s regulating aging in S cerevisiae (Heo et al., 1999; Hoopes et al.,

2002; Merker and Klein, 2002)

In addition to Sir2p, S.cerevisiae has four additional members of sirtuin family

(Brachmann et al., 1995), Hst1-4p (Figure 1.3) Hst1p is closest relative of Sir2p Indeed,

increased dosage of Hst1p can compensate for lack of Sir2p function (Brachmann et al.,

1995) Under normal cellular conditions, Hst1p is a gene-specific, rather than

region-specific, repressor Hst1p is required for repression of middle-sporulation genes during

vegetative growth (Xie et al., 1999) However, in the context of the specific mutation,

sum1-1, Hst1p can also function as a region-specific repressor at HM loci Hst1p interacts

physically with transcriptional repressor Sum1p sum1-1 encodes specific form of Sum1p

that can interact with ORC Through this interaction Sum1p is recruited to HM loci and it

subsequently recruits Hst1p, which deacetylates histones and contributes to the formation

of silent chromatin (Rusche and Rine, 2001; Sutton et al., 2001) Hst1p, together with its

partner Sum1p represses expression of the genes in the de novo pathway of NAD+

biosynthesis and it is suggested that Hst1p functions as NAD+

sensor in the cell (Bedalov

et al., 2003) Hst1p has been shown to repress FLO10, gene for cell wall protein that

regulates flocculation properties of yeast (Halme et al., 2004)

Hst2p is a very abundant cytoplasmic sirtuin, but its function and substrates(s) in the

cytoplasm are not known Interestingly, its closest relative, the human sirtuin is a

cytoplasmic tubulin deacetylase (North et al., 2003) Small amounts of Hst2p are present

in the nucleus where Hst2p regulates, together with Hst1p (Halme et al., 2004),

expression of FLO10 Hst2p is also required for CR-dependent, Sir2p-independent life

extension in S cerevisiae (Lamming et al., 2005) Hst2p was isolated in a high-copy

suppression screen for genes that increase rDNA silencing and it was subsequently shown

that overexpression of Hst2p can decrease rate of rDNA recombination and extend life

span independently of Sir2p (Lamming et al., 2005)

Hst3p and Hst4p are two closely related and functionally redundant S cerevisiae sirtuins

involved in regulation of genomic stability (Brachmann et al., 1995) Yeast cells lacking

either Hst3p or Hst4p have very mild phenotypic defects, but deletion of both genes

results in a growth impairment that is exacerbated at elevated temperature and defects in

cell cycle progression (Brachmann et al., 1995), demonstrating the functional redundancy

of these sirtuins In addition, hst3 hst4 cells show sensitivity to UV irradiation and

hydroxyurea (HU), an inhibitor of ribonucleotide reductase and increased genomic

instability as judged by increase in mitotic recombination rates and plasmid loss rate

(Brachmann et al., 1995) It is not known mechanistically how Hst3p and Hst4p control

genomic stability and what other gene/proteins partners participate with Hst3p and Hst4p

in that process Hst3p and Hst4p share most of the sequence conservation with other

sirtuins (Figure 1.4) in the core region, which has been shown to be important for NAD+

-dependent deacetylation activity, but no such activity has been reported for Hst3p or

Hst4p using enzymatic assay that demonstrated enzymatic activity of other sirtuins

Figure 1.3 Schematic alignment of S cerevisiae sirtuins The squares represent

sequences belonging to the conserved catalytic core The lines represent nonconserved

sequences at the N- and C-termini and the sequences dividing different regions of the

core The style of lines was chosen to illustrate the similarity between particular pairs of

sirtuins in those regions

In the Chapter 2, I will present experiments demonstrating that Hst3p and Hst4p regulate

genomic stability through the control of acetylation of lysine K56 in histone H3 This

residue in present on the nucleosome surface and it has a critical role in DNA repair

(Hyland et al., 2005; Masumoto et al., 2005) Chapter 3 describes experiments aimed at

addressing the cellular responses to histone H3 K56 hyperacetylation due to the lack of

Hst3p and Hst4p

Chapter 3 also describes genetic interactions between HST3, HST4 and genes involved in

various aspects of DNA metabolism Finally, in the Appendix, I will present my

contribution to earlier studies aimed at characterizing members of the Sir2p family

Figure 1.4 Alignment of Hst3p, Hst4p and selected sirtuins The invariant residues in 10

or more sequences are shown in red The residues invariant in less then 9 or chemically

similar in more than 6 sequences are shown in yellow The secondary structure is shown

above the sequence alignment with the color scheme adapted from Avalos et al., 2002

(Af1Sir2, Af2Sir2-Archaeoglobus fulgidus sirtuins 1 and 2; TmSir2-Thermatoga

maritima sirtuin; SirT1, SirT2 and SirT5-Homo sapiens sirtuins 1,2 and 5;

CobB-Salmonella enterica sirtuin; Sir2, Hst1, Hst2, Hst3, Hst4-Saccharomyces cerevisiae

sirtuins)

23Chapter 2

The sirtuins Hst3p and Hst4p control histone H3 lysine 56

acetylation

In many and possibly all eukaryotes, the newly synthesized histones deposited throughout

the genome during DNA replication are acetylated at several lysine residues within their

amino-terminal tails (Jackson et al., 1976; Kuo et al., 1996; Ruiz-Carrillo et al., 1975;

Sobel et al., 1995) More recent work with the budding yeast Saccharomyces cerevisiae

has established that specific residues within the globular domains are also acetylated

These include lysine 56 of H3 (Hyland et al., 2005; Masumoto et al., 2005; Ozdemir et

al., 2005; Xu et al., 2005) and lysine 91 in newly synthesized histone H4 (Ye et al.,

2005) In higher eukaryotes, the bulk of the acetylation of newly synthesized histones is

rapidly removed (Jackson et al., 1976; Ruiz-Carrillo et al., 1975; Taddei et al., 1999) No

comparable study has been performed in yeast However, a large fraction of histone

acetylation is turned over in S cerevisiae (Waterborg, 2000; Takahashi, Irizarry and

Boeke 2006, in press) In addition, the majority of histone H3 K56 acetylation, which

predominantly occurs in newly synthesized histone molecules during S-phase, is

normally removed before the onset of mitosis (Masumoto et al., 2005) These results

suggest that the regulated turnover of histone H3 K56 acetylation may be important,

although its physiological function is currently unknown

Sirtuins constitute an evolutionarily conserved family of proteins with sequences similar

to the catalytic domain of S cerevisiae Sir2p Sir2p is an unusual histone deacetylase

(HDAC) that requires NAD+

as a co-substrate to remove acetyl-lysine from several

residues within the amino-terminal domains of histone H3 and H4 (Borra et al., 2004;

Imai et al., 2000; Landry et al., 2000b; Smith et al., 2000; Suka et al., 2001) The S.

cerevisiae genome encodes four additional sirtuins, encoded by the HST1 (Homologous

to Sir Two) to HST4 genes (Brachmann et al., 1995), while the human genome encodes at

least 7 family members, known as SIRT1 to SIRT7 (Frye, 2000) Some sirtuins clearly

act on non-histone proteins This is the case for Salmonella enterica CobB and human

SIRT2, which respectively deacetylate acetyl-coenzyme A synthetase and tubulin (North

et al., 2003; Starai et al., 2002) Among the five yeast sirtuins, Hst3p and Hst4p are most

closely related to each other and most distantly related to Sir2p (Brachmann et al., 1995)

Single mutants in hst3 and hst4 show mild phenotypes, but cells lacking both Hst3p and

Hst4p exhibit pronounced defects in mitotic chromosome transmission, DNA damage

susceptibility and thermosensitivity (Brachmann et al., 1995; Starai et al., 2003),

suggesting a considerable degree of functional overlap between these two sirtuins

Despite the fact that the roles of Hst3p/Hst4p in chromosome metabolism were

established many years ago, the substrates of these two sirtuins have remained elusive

Here, we demonstrate a role for Hst3p and Hst4p in the cell cycle regulation of histone

H3 K56 acetylation We show that the vast majority of histone H3 molecules are

K56-acetylated in cells lacking both Hst3p and Hst4p This very high level of H3 K56

acetylation is largely responsible for the phenotypes of hst3 hst4 mutant cells (DNA

damage sensitivity, mitotic chromosome loss and thermosensitivity) because those

phenotypes are strongly suppressed by a point mutation of histone H3 K56 into a

positively charged, but non-acetylatable arginine residue Thus, failure to deacetylate

histone H3 K56 impairs several important aspects of chromosome metabolism

Hst3p and Hst4p control histone H3 K56 acetylation during the cell cycle

In an effort to identify proteins that regulate histone H3 K56 acetylation during the cell

cycle, we initially tested yeast strains lacking multiple known HDAC enzymes by

immunoblotting with H3 K56 acetylation-specific antibodies (Masumoto et al., 2005) H3

K56 acetylation normally occurs transiently during S-phase (Masumoto et al., 2005) In

order to ensure that the changes in H3 K56 acetylation observed in HDAC mutant strains

were not due to indirect effects on cell cycle progression, we arrested yeast cells with

nocodazole prior to making extracts for immunoblotting Yeast strains lacking several

class I and class II HDACs had normal levels of histone H3 K56 acetylation (Figure

2.1A) In striking contrast, a strain lacking the five known sirtuins had substantially

higher levels of H3 K56 acetylation than wild-type cells Of the five sirtuin single

mutants, only hst3 single mutant exhibited higher histone H3 K56 acetylation than

wild-type cells and these levels were further increased in hst3 hst4 double mutant cells (Figure

2.1B and 2.1C) Except for strains with an hst3 mutation, none of the other double null

mutants tested in the sirtuin family had any detectable impact on H3 K56 acetylation

With the exception of hst4, none of the other sirtuin mutations increased H3 K56

acetylation in cells lacking hst3 Thus, we conclude that Hst3p and Hst4p function in a

partially redundant manner to control histone H3 K56 acetylation in vivo The functional

overlap between Hst3p and Hst4p was further explored by following the progression of

histone H3 K56 acetylation during the cell cycle Based on immunoblotting, wild-type

cells, hst3 and hst4 single mutants had no detectable H3 K56 acetylation in G1 (Figure

2.1C) In contrast to the single mutants, hst3 hst4 double mutant cells had very high

levels of acetylation, even in G1, and those levels did not fluctuate during the cell cycle

(Figure 2.1C) Based on quantitative mass spectrometry, the vast majority of H3

molecules were K56-acetylated in hst3 hst4 double mutants (Figure 2.1D), thus

explaining why the acetylation was not cell cycle-regulated in that strain One other

notable difference between the two single mutants was that hst3 single mutants retained

high levels of H3 K56 acetylation in G2/M phase, whereas a significant portion of the

acetylation disappeared in hst4 single mutants and wild-type cells (Figure 2.1C) These

observations are consistent with the fact that hst3, but not hst4 single mutants, have mild

phenotypes that are greatly exacerbated in double mutant cells (Brachmann et al., 1995),

and with the fact that HST3 transcription peaks in G2/M phase of the cell cycle (Spellman

et al., 1998) Our results suggest that elevated levels of histone H3 K56 acetylation

during inappropriate periods of the cell cycle could underlie the phenotypes of hst3 hst4

mutants (see below)

The nicotinamide binding pocket of Hst3p is necessary for regulation of histone H3 K56 acetylation

The HDAC activity of Sir2 proteins can be inhibited by nicotinamide (Bitterman et al.,

2002) Interestingly, nicotinamide treatment also led to a considerable increase in histone

H3 K56 acetylation (Figure 2.2A) Nicotinic acid had no effect on the level of histone H3

K56 acetylation (data not shown) despite the existence of a specific permease, known as

Tna1p, that facilitates nicotinic acid entry into yeast cells (Llorente and Dujon, 2000)

Nicotinamide did not exert its effect by arresting or slowing down cell progression

through S-phase because no accumulation of S-phase cells was detected by flow

cytometry (Figure 2.2A) Moreover, in the continuous presence of nicotinamide, the

increase in H3 K56 acetylation was progressive over the course of six hours (Figure

2.2A), during which the cells underwent a few divisions This was apparent both from

cell counting (data not shown) and by the increase in total histone H3 (Figure 2.2A) This

suggests that nicotinamide blocks the deacetylation of at least a fraction of newly

synthesized histones during each passage through S-phase

Hst3p and Hst4p exhibit a considerable degree of sequence similarity with the catalytic

core of Sir2p responsible for NAD+

-dependent protein deacetylation (Figure 2.2B) We

mutated three invariant residues within the Hst3p core region and examined the effects of

these changes on Hst3p function in vivo Two residues, N152 and D154, were within the

so-called C-pocket, a highly conserved region within which the nicotinamide moiety of

NAD+

binds in other sirtuins (Avalos et al., 2004) In the structure of Archaeoglobus

fulgidus Sir2 (AfSir2), N101 (equivalent to N152 in Hst3p) hydrogen bonds with the 2’

hydroxyl of the ribose attached to nicotinamide (the primary site of transfer of the histone

acetyl group) and also makes van der Waals contact with nicotinamide (Avalos et al.,

2004) AfSir2 D103 (corresponding to D154 in Hst3p) makes a direct hydrogen bond

with the amino group of nicotinamide (Avalos et al., 2004) Mutations in residues

corresponding to N101 and D103 abolish the enzymatic activity of generic sirtuins

(Armstrong et al., 2002; Chang et al., 2002; Imai et al., 2000) The third residue changed

was the highly conserved histidine, H184 in Hst3p This invariant histidine has been

proposed to act as a catalytic base that activates the 2’ hydroxyl of the ribose for the

29attack on the 1’-O-alkylamidate intermediate formed during the deacetylation reaction

(Sauve and Schramm, 2003; Smith and Denu, 2006) Mutation of this residue abolishes

deacetylation activity and Sir2p function in vivo (Armstrong et al., 2002; Imai et al.,

2000; Min et al., 2001) In order to assess their phenotypes in a sensitized genetic

background, the Hst3p missense mutants were introduced by plasmid shuffling into hst3

hst4 mutant cells The Hst3p C-pocket mutants, but not the H184A mutant, were

expressed at somewhat lower levels than wild-type Hst3p (Figure 2.3A and data not

shown) It is possible that the C-pocket mutations decrease the stability of Hst3p

However, because Hst3p C-pocket mutations are essentially null mutations (Figure 2.3C),

it is equally plausible that the CEN plasmids encoding the Hst3p C-pocket mutants are

present in fewer copies than those encoding wild-type Hst3p or the Hst3p H184A mutant

This is because hst3 hst4 double null mutants are partially defective in the maintenance of

CEN plasmids (Brachmann et al., 1995) In nocodazole-arrested cells, the two Hst3p

C-pocket mutations led to high levels of H3 K56 acetylation, comparable to those observed

in hst3 hst4 null mutants (Figure 2.2A) Using a 2µ plasmid, rather than a CEN plasmid,

increased the expression of the Hst3p D154 mutant roughly 3-fold, albeit not to the level

of wild-type Hst3p, but did not result in any detectable decrease in histone H3 K56

acetylation (data not shown), suggesting that the Hst3p D154A mutant was completely

inactive The Hst3p H184A mutation resulted in a substantial, but less pronounced

increase in acetylation (Figure 2.3A) These results were corroborated by quantitative

mass spectrometry of histones isolated from nocodazole-arrested cells In hst3-N152A

hst4 mutant cells, virtually all (97.7%) of H3 was K56-acetylated in G2/M (Figure 2.3B),

compared with 45.2% in hst3-H184 hst4 mutants and only 24.5% in HST3 hst4 cells

(Figure 2.3B) As was the case for hst3 hst4 double mutant cells, Hst3p C-pocket

mutations conferred thermosensitivity (Figure 2.3C) and other phenotypes (hydroxyurea

sensitivity and derepression of telomeric silencing, data not shown) Interestingly, none

of these phenotypes were observed in hst3-H184 hst4 mutants (Figure 2.3C and data not

shown), despite the significant increase in histone H3 K56 acetylation in these cells

(Figure 2.3A and 2.3B) This suggests that these well-established hst3 hst4 mutant

phenotypes are only manifested in cells with very high amounts of H3 K56 acetylation

The phenotypes of hst3 hst4 mutant cells are caused by high levels of H3 K56

acetylation

To test whether the phenotypes of hst3 hst4 mutant cells were a consequence of a failure

to remove histone H3 K56 acetylation, we mutated H3 lysine 56 into a non-acetylatable

arginine residue (hht1-K56R mutation) Remarkably, the hht1-K56R mutation rescued the

growth at 37°C of hst3 hst4 double mutant cells (Figure 2.4A) Cells lacking both Hst3p

and Hst4p are sensitive to genotoxic agents that interfere with DNA replication fork

progression (Figure 2.4B) These include hydroxyurea (HU), which inhibits

ribonucleotide reductase and depletes deoxyribonucleoside triphosphates (Koc et al.,

2004), as well as methyl methane sulfonate (MMS), which generates 3-methyladenine, an

alkylated base that stalls DNA polymerase (Sedgwick, 2004) In addition,

3-methyladenine can be removed by DNA glycosylases, and the resulting abasic sites

converted into nicks by apurinic (AP) site endonuclease (Sedgwick, 2004) The

persistence of nicks is potentially cytotoxic, particularly when they are converted into

DNA double-strand breaks during replication (Tercero et al., 2003; Xiao et al., 1996)