[page 1↓]

1  Introduction

1.1 N-terminal acetylation of proteins

In eukaryotic cells, proteins undergo a number of co- and posttranslational modifications that extend the range of their possible molecular structure beyond the limits of the encoding amino acids and thus amplify their functional potential. This provides the bases for complex cellular mechanisms such as differentiation and gene regulation.

Among protein modifications, the acetylation of the α-amino group at the initiating amino acid, referred to as N‑terminal acetylation or Nα‑acetylation, is a major modification type. 80-90% of the mammalian cytosolic proteins and 50% of those in yeast are estimated to be Nα-acetylated (Polevoda and Sherman 2003b). In a cotranslational process, N‑terminal acetyltransferases (NATs) transfer an acetyl group from acetyl coenzyme A to nascent polypeptides of 20 to 50 amino acids when they are just protruding from the ribosome (Driessen et al. 1985) (Fig. 1.1). In proteins with small penultimate amino acids (1.29 Å or less radii of gyration), acetylation is preceded by the removal of the initial methionine residue by means of specific amino peptidases (Bradshaw et al. 1998). Notably, Nα‑acetylation is irreversible and thus functionally distinct from the reversible posttranslational acetylation of ε-amino groups (Nε‑acetylation) of internal lysines in histones, transcription factors (Cheung et al. 2000b), nuclear receptors and import factors (Bannister et al. 2000; Soutoglou et al. 2000).

The substrate specificity of NATs is not determined by a simple consensus motif, but rather is supposed to emerge from degenerate signals within the N‑terminal 50 amino acids (Polevoda and Sherman 2003a) of the substrate. The penultimate amino acid has a profound, although no absolute, effect on Nα‑acetylation. Proteins with methionine, alanine or serine termini are the most frequently acetylated, the latter two contributing more than 74% of all Nα‑acetylated proteins in the budding yeast Saccharomyces cerevisiae (Polevoda and Sherman 2003b).

Whereas Nα‑acetylation is one of the most common protein modifications in eukaryotes, it occurs only rarely in prokaryotes and archea. In Escherichia coli, RimI, RimJ, and RimL specifically Nα‑acetylate ribosomal proteins, apparently in a posttranslational manner (Tanaka et al. 1989). In general, Nα‑acetylation in prokaryotes and archea is thought to differ fundamentally from the process in eukaryotes (Polevoda and Sherman 2003b).

In eukaryotic organisms, the same system of Nα‑acetylation may operate in all species, since sequence homologs to subunits of yeast NAT’s exist in the genomes of all model organisms, [page 2↓]e.g. Caenorhabditis elegans, Drosophila melanogaster, Arabidopsis thaliana, Xenopus laevis, Mus musculus, and in humans(Polevoda and Sherman 2003b). Moreover, the acetylation patterns of Nα‑acetylated proteins are very similar in yeast and mammals, suggesting that they are evolutionary conserved. Interestingly, Nα‑acetylation is more frequent in mammals compared to yeast, which may point to some form of selection for this modification during evolution.

Nevertheless, the number of the Nα‑acetylated proteins characterized so far is limited, and only a few examples demonstrate the biological significance of this modification. It was originally suggested that Nα‑acetylation generally acts as protection from degradation, but this hypothesis is no longer favored (Mayer et al. 1989). In the current model, the biological importance of Nα‑acetylation varies with the particular protein. Accordingly, some proteins require Nα‑acetylation for their function and stability, whereas others do not. Tropomyosin, for example, depends on Nα‑acetylation for normal binding and stabilization of filamentous actin in yeast and vertebrate muscle cells (Urbancikova and Hitchcock-DeGregori 1994; Singer and Shaw 2003). In addition, Nα‑acetylated rat α‑melanotropin induces increased pigment-producing effects and enhanced activity in behavioral tests compared to the unacetylated form (Smyth et al. 1979; O'Donohye et al. 1982). Nα‑acetylation of the major coat protein gag of the L‑A double-stranded RNA virus in S. cerevisiae is essential for the assembly of virus particles (Tercero and Wickner 1992). Recently, AtMAK3, a homolog of yeast Mak3 in A. thaliana, was found to acetylate core proteins of photosystem II, which was necessary for the formation of thylacoid complexes and plant growth (Pesaresi et al. 2003). Furthermore, Nα‑acetylation can also affect the thermal stability of proteins, as observed for the NADP-specific glutamate dehydrogenase of Neurospora crassa (Siddig et al. 1980).

Importantly, not only the lack of Nα‑acetylation can result in various defects, but abnormal acetylation can likewise prevent regular protein function. In yeast, the catalytic α‑amino groups of some 20S proteasome subunits have to be protected from Nα‑acetylation to preserve their peptidase activity (Arendt and Hochstrasser 1999). This is realized by N‑terminal propetides, which become removed from the subunits during proteasome assembly. As another example, hemoglobin Lyon-Bron displays decreased oxygen affinity due to Nα-acetylation, which consequently causes anemia. In this α2-globin variant, the penultimate amino acid is mutated from valine to alanine, which converts the protein into a NAT substrate (Lacan et al. 2002).


[page 3↓]

Altogether, these examples demonstrate the biological significance of Nα‑acetylation in diverse organisms. In spite of this, the knowledge of mechanisms and players of this frequent modification to date is only marginal.

1.2 Nα-acetyltransferases in S. cerevisiae

Insight into the function of Nα‑acetylation comes from the analysis of NATs in S. cerevisiae. Here, three NAT complexes exist that are known as NatA, NatB and NatC according to their substrate specificity (Tbl. 1.1). NatA accounts for the majority of Nα-acetylated proteins in yeast, and acetylates proteins with alanine or serine, and occasionally with glycine or threonine termini. The other two NATs act on the N-terminal methionine when the second residue either is glutamate or aspartate, asparagine or methionine (NatB substrates), or else isoleucine, leucine, tryptophan or phenylalanine (NatC substrates) (Polevoda and Sherman 2003b). Interestingly, whereas all observed proteins with Met-Asp and Met-Glu N‑termini were Nα‑acetylated, only half of the potential NatC substrates were actually modified in vivo.

In addition to the aforementioned substrate types, a special subclass of NatA substrates with Ser-Glu, Ser-Asp, Ala-Glu or Gly-Glu termini was designated NatD substrates (Arnold et al. 1999). In systematic analyses for substrate specificities of NATs, these proteins were found to require not only NatA activity for Nα-acetylation, but also the integrity of NatB and NatC. As a possible interpretation, it was suggested that the acetylation of NatD proteins requires auxiliary factors to NatA, which in turn are substrates of the other two NATs (Polevoda et al. 1999).

Recently, the novel GNAT (GCN5-related N -acetyltransferase) homolog Nat4 was identified to specifically acetylate histones H2A and H4 (Song et al. 2003). Interestingly, this protein is well conserved from yeast to mammals not only in the GNAT domain, indicating a role in histone acetylation also for its homologs. However, mechanistic details or interaction partners of Nat4 are currently not known. Given the importance of the charge of histones for their association with DNA (see below), it is surprising that nat4Δ mutants displayed no detectable phenotype, although each nucleosome contained four extra positive charges due to the missing Nα-acetylation. Thus, the relevance of Nα‑acetylation for histone function remains subject to further investigation.

The three NAT complexes, NatA, NatB and NatC, not only differ in substrate specificity, but also in subunit composition. Significantly, they all contain a catalytic subunit homologous to [page 4↓]the GNAT superfamily of acetyltransferases (Tbl. 1.1). Besides NATs, this superfamily contains several histone acetyltransferases (HATs) and proteins involved in gene regulation and diverse other functions, such as detoxification and drug resistance (Neuwald and Landsman 1997). The members of the GNAT superfamily are characterized by a remarkably conserved binding motif for the donor substrate acetyl CoA and exhibit a highly consistent protein topology (Dyda et al. 2000).

Table 1: Characteristics of the three NAT complexes in S. cerevisiae*

 

NatA

NatB

NatC

Catalytic subunit

Ard1

Nat3

Mak3

Auxiliary subunit

Nat1

Nat5

Mdm20

Mak10

Mak31

Substrate** termini

Ser

Ala

Gly

Thr

Met-Glu

Met-Asp

Met-Asn

Met-Met

Met-Ile

Met-Leu

Met-Trp

Met-Phe

Selected substrates

ribosomal subunits (SU)

S1,2,5,7, 11,14,15,16,18,20,24 and L1,4,11,16,33,36; 19S proteasomal SU: Rpt4,5,6 and Rpn2,3,5,6,8; 20S proteasomal SU: Scl1, Pup3 and Pre6,8,9,10

Tropomyosin, actin,

ribosomal SU S21 and S28, 19S proteasomal SU Rpt3 and Rpn11,

20S proteasomal SU Pre1

gag protein of L-A virus,

20S proteasomal SU Pup2 and Pre5

Deletion mutant phenotypes

Slow growth; temperature and osmotic sensitivity; deficiency in utilizing non-fermentable carbon sources; inability to enter G0; inability to sporulate; chromosomal instability;

derepression of silent loci

Slow growth; temperature and osmotic sensitivity; deficiency in utilizing non-fermentable carbon sources; defects in vacuolar and mitochondrial inheritance; random polarity in budding; reduced mating efficiency; sensitivity to antimitotic drugs and DNA damaging agents

Temperature sensitivity;

deficiency in utilizing non- fermentable carbon sources

Characterized homologs

Homologs of Ard1: ARD1 (T. brucei), TE2 (human), Xat‑1 (X. laevis); Homolog of Nat5: SAN (Drosophila); Homologs of Nat1: NATH (human), NARG1, tbdn‑1, Tbdn100 (all mouse),

No homologs have been

characterized at present

AtMAK3 (A. thaliana)

* References are given in chapters 1.2 and 1.3.
* * Acetylation occurs only on subclasses of proteins containing the indicated termini, except for Met-Glu and Met-Asp.


[page 5↓]

In addition to the catalytic subunit, each NAT contains auxiliary components that are required for enzymatic activity (Polevoda and Sherman 2003b) (Tbl. 1.1). The loss of NAT activity generally results in multiple effects in yeast, among them temperature sensitivity and diverse growth defects. Mutations of NatB subunits display the most severe phenotypes, apparently associated with the partial loss of function of unacetylated actin and tropomyosin in these mutants (Polevoda and Sherman 2001; Polevoda et al. 2003). The phenotypic characteristics of NatC were hypothesized to stem from affected mitochondrial substrates (Tercero and Wickner 1992; Polevoda and Sherman 2001), whereas NatA acetylation is important for growth and cell cycle control. Notably, nat mutations are not lethal, suggesting that among the various substrates there is no essential protein depending absolutely on Nα‑acetylation. The hitherto known substrates of the individual NATs were mainly identified in systematic analyses of ribosomal (Arnold et al. 1999) and proteasomal proteins (Kimura et al. 2000; Kimura et al. 2003). Predominant methods applied were mass spectrometry and 2D protein migration analysis.

1.3 NatA – the major Nα-acetyltransferase complex of S. cerevisiae

NatA is the major NAT in yeast, accounting for most of the Nα-acetylated proteins. Given the portion of NatA targets on the total NAT substrates known to date, NatA acetylates potentially 2500 yeast proteins. About 140 NatA substrates have been identified so far, the list including ribosomal and 26S proteasomal subunits as well as some abundant proteins (Polevoda and Sherman 2003b) (Tbl. 1.1). NatA has the most degenerate substrate specificity of all NATs. Approximately 90% and 30%, respectively, of the tested serine and alanine termini, and only one fourth of the glycine and threonine proteins tested, were actually acetylated by NatA (Polevoda and Sherman 2003b).

NatA is not only the predominant, but also the best characterized NAT. The trimeric complex consists of the subunits Ard1, Nat1 and Nat5, which are present in a 1:1:1 stoichiometric ratio (Gautschi et al. 2003). Interestingly, in a tandem affinity purification (TAP) analysis, several other proteins, namely Asc1, Eno1, Mis1, Myo1 and YYGR090w, were co-purified with Ard1, and the ribosomal protein Asc1 (Inada et al. 2002) also associated with Nat1 (http://yeast.cellzome.com/). However, the question whether these proteins were present in stoichiometric amounts was not answered. Therefore, it remains unclear whether Asc1 or the other co-purified proteins are required for NatA’s function.


[page 6↓]

According to the current model, the NatA complex resides at the ribosome, close to the polypeptide tunnel exit (Fig. 1.1). Nat1 (N-terminal acetyltransferase 1) mediates the stable interaction of NatA with the large ribosomal subunit. In addition, it can be crosslinked to nascent polypeptides, and is thus predicted to contact the nascent chains in order to present them to the catalytically active Ard1. This interaction is probably mediated by five to eight tetratricopeptide repeats (TPR) clustered in the first third of the 854 amino-acid protein Nat1 (Gautschi et al. 2003). TPR motifs are evolutionarily ancient protein-protein interaction modules consisting of two antiparallel α‑helices that generate a super-helix with an amphipathic channel (Blatch and Lassle 1999). TPR clusters exist in a number of functionally distinct proteins and are important for the function of e.g. chaperones or protein transport complexes (Gatto et al. 2000). In addition to TPR repeats, Nat1 contains highly charged regions between amino acids 550 and 670 with predicted coiled‑coil structures. These are proposed to mediate the interaction of Nat1 to other subunits of the complex. Interestingly, a nuclear localization signal (NLS) is likewise predicted for Nat1 between amino acids 648-665. Its functional significance remains unclear, since Nat1 acts in the cytoplasm (Polevoda and Sherman 2003a).

The catalytic subunit Ard1 (arrest defective) is a protein of 238 amino acids and carries the acetyl CoA binding GNAT domain in the N-terminal part between amino acids 3-175 (Neuwald and Landsman 1997). Ard1 contacts the ribosome not directly, but probably binds to Nat1 via a C‑terminal coiled‑coil (Park and Szostak 1992; Gautschi et al. 2003) (Fig. 1.1).

Recently, the GNAT-family acetyltransferase Nat5 (also termed Rog2), another putative catalytic subunit, was found to be associated with NatA. Currently, it is unclear how Nat5 is bound. In addition, its function within the complex remains to be characterized, since nat5Δ mutants display no obvious phenotype (Gautschi et al. 2003). The question remains why the complex should contain a second catalytic subunit. Gautschi et al. (2003) propose that Nat5 may be responsible for the Nα‑acetylation of a small subset of proteins that are not involved in mating-type silencing or affected at elevated temperature. In an earlier study, the deletion of Nat5 suppressed the temperature sensitivity of the double mutants involved in ubiquitin-dependent protein degradation, mck1 mds1 and bul1 bul2 (Andoh et al. 2000). Moreover, the Nat5 homolog SAN in Drosophila (Accession GI 6980078) was found to play a role in sister chromatid cohesion. The interaction of SAN with homologs of Nat1 and Ard1 suggests that a NAT complex similar to NatA exists in Drosophila (Polevoda and Sherman 2003a).

So far, Nat1 and Ard1 homologs have been implicated in development and cellular proliferation of higher eukaryotes. For instance, Xat‑1, a Nat1 homolog in X. laevis, was [page 7↓]isolated in a screen for stage-specific gene expression during early embryogenesis (Choi et al. 2001). Likewise, NARG1 and tubedown-1, two mouse Nat1 homologs, were found highly expressed in certain embryonic tissues (Gendron et al. 2000; Sugiura et al. 2001). In Trypanosoma brucei, ARD1 was essential in mammalian and insect-stage cells (Ingram et al.

2000).


Fig. 1.1: The NatA complex is associated with the ribosome.
In the current model, the non-catalytic subunit Nat1 mediates the stable contact of NatA with the large ribosomal subunit. Nat1 interacts with the nascent polypeptide chain that emerges from the tunnel exit and guides it to the catalytic subunit Ard1, which transfers an acetyl moiety from acetyl coenzyme A to the N-terminal amino acid of NatA substrates. The putative catalytic subunit Nat5 is also associated with the complex. (adapted from Gautschi et al. 2003)

Interestingly, there are data suggesting that the function of at least some of the NatA homologs in mammals has diverged from that of their yeast counterparts. As an example, mouse ARD1, which is 57% homologous to the yeast protein, acetylates lysine 532 of the hypoxia‑inducible factor HIF‑1α, a protein involved in adaptation to changes in oxygen availability (Jeong et al. 2002). Thereby, mARD1 regulates the protein stability of HIF‑1α, since the acetylation is critical to its proteasomal degradation. Notably, in contrast to yeast Ard1, mARD1 acts alone and appears not to require a complex. As a second striking difference, mARD1 shows ε‑N‑acetyltransferase activity, which is functionally distinct from Nα-acetylation. It remains to be determined whether mARD1 can acetylate Nα-termini as well. In a possible scenario, mARD1 may perform both modifications in conjunction with different [page 8↓]partner proteins. Intriguingly, provided that yeast NATs are comparable to the evolutionary ancestors of NATs of higher organisms, the substrate shift of mammalian NAT proteins may serve as an example for evolutionary processes on conserved protein modifications.

Overexpression of both Nat1 and Ard1 is required to increase the in vivo activity of NatA (Park and Szostak 1992). Interestingly, similarly to nat1Δ, the overexpression of Nat1 results in chromosome loss, presumably due to a dominant negative effect on NatA integrity (Ouspenski et al. 1999). Deletions of NAT1 or ARD1 display the same pleiotropic phenotypes of slow growth, temperature sensitivity, chromosomal instability, inability to enter G0 and failure to sporulate as homozygous diploids (Tbl. 1.1) (Whiteway and Szostak 1985; Mullen et al. 1989). Importantly, NatA also functions in transcriptional repression, since nat1Δ and ard1Δ cause strong derepression of the HML silent mating-type locus and subtelomeric reporter genes (Mullen et al. 1989; Aparicio et al. 1991). This suggests that one or several proteins require Nαacetylation by NatA in order to function in transcriptional silencing.

1.4 Chromatin and gene regulation

Eukaryotic DNA is packed into a nucleoprotein structure called chromatin. This facilitates the compaction as well as the regulation of genetic information. Compaction is necessary in order to adapt the size of the DNA molecule to the nuclear dimensions. For instance, the human genome comprises about 12 000 Mbp, which corresponds to a molecule length of four meters. This size is scaled down to 10 μm by complex packaging mechanisms, resulting in a compact higher-order structure of chromatin. In addition to this spacial role, the dynamic nature of chromatin plays a crucial role in central genetic processes such as transcription, replication, recombination and repair.

The fundamental chromatin unit is the nucleosome, which is composed of two copies each of the four core histones H2A, H2B, H3 and H4 and approximately 146 bp of DNA wrapped in two turns around the histone octamer (Luger et al. 1997) (Fig. 1.2). This complex is repeated every 200±40 bp, thereby creating a “pearls on a string” structure of 11 nm width. With the aid of additional proteins, including histone H1 in mammals (Contreras et al. 2003), the nucleosomal array is further packaged into a 30 nm fiber of a spiral, or solenoid, structure with six nucleosomes per turn. This structure has to be unfolded to allow the access of regulatory proteins to the DNA. Thus, the dynamic feature of chromatin is a prerequisite for various processes on the genetic material.


[page 9↓]

In eukaryotes, chromatin is organized in two types of domains, namely euchromatin and heterochromatin. Euchromatic domains define transcriptionally active portions of the genome, whereas heterochromatin is largely inactive for gene expression (Grewal and Moazed 2003). The repressive character of heterochromatin is accompanied by several other features, such as a highly ordered nucleosomal array, reduced accessibility to restriction nucleases and other DNA altering enzymes (Wallrath and Elgin 1995), replication late in S-phase (Ferguson et al. 1991) and the tendency to localize to perinuclear regions (Andrulis et al. 1998; Feuerbach et al. 2002). Originally, heterochromatin was defined in cytological experiments as chromosomal blocks that remained condensed throughout the cell cycle (Heitz 1928).



Fig. 1.2: The basic structure of chromatin.
The 11 nm fiber consists of DNA wrapped in two turns around histone octamers (nucleosomes) at intervals of about 200 bp along the DNA. Further folding creates a spiral structure, the 30nm fiber. Positively charged (deacetylated) histone tails (arrows) facilitate higher-order folding, whereas the acetylation of histone tails (bars) promotes the unfolded state corresponding to active chromatin. The two chromatin states are well-defined in electron micrographic images.
(adapted from http://sgi.bls.umkc.edu/waterborg/chromat/chroma09.html)

Transcriptional repression in heterochromatin occurs in a sequence-independent fashion, making the chromosomal context in which a gene is located crucial for its transcriptional activity. In this context, a phenomenon called position-effect variegation (PEV) was revealed by pioneering experiments in Drosophila about 70 years ago. Muller (1930) described radiation-induced translocations that displaced the white + (w +) eye color gene from its normal euchromatic location in the vicinity of heterochromatin (Muller 1930). This resulted in a clonally inherited pattern of (w +) expression in some cells but not in others, thus causing a mosaic eye color phenotype.


[page 10↓]

Variegated position effects on translocated euchromatic genes are based on the ability of heterochromatin to spread in cis from a nucleation site into adjoining regions, which is probably aided by trans-interactions between different heterochromatic blocks (Wakimoto 1998).

Large heterochromatic blocks generally surround centromeres and telomeres of eukaryotic chromosomes (Perrod and Gasser 2003). Centromeres consist of large arrays of unspecific, often repetitive sequences in higher eukaryotes and facilitate proper sister-chromatid cohesion and chromosome segregation (Karpen and Allshire 1997). There is growing evidence that this function depends on the heterochromatic structure of centromeres, since mutations in heterochromatic components, like the histone lysine methyltransferase Clr4 in Schizosaccharomyces pombe, also interfere with chromosome segregation (Ekwall et al. 1996).

Heterochromatin generally stabilizes repetitive chromosomal regions by inhibition of recombination between homologous repeats. Besides centromeres and telomeres, this is especially important in mammalian genomes, which consist to a great portion of repetitive, non-coding sequences (Wichman et al. 1992). Presumably, over 90% of the mammalian genome is transcriptionally silent in differentiated tissues (Perrod and Gasser 2003). However, this high percentage is not only due to repetitive regions, but also reflects a role of heterochromatin in gene regulation during development and cellular differentiation.

Since the heterochromatic state is stably inherited through many cell divisions, it is suitable for long-term inactivation of large regions of the genome. One example is the stable inactivation of developmental regulators, such as the homeotic gene clusters in Drosophila. These genes are expressed only in precise spatially restricted patterns during development and are silenced in other parts of the embryo by means of Polycomb-Group proteins (see below) (Bienz and Muller 1995; Jones et al. 2000). Another prominent example for long-term gene repression by heterochromatin is the X chromosome dosage compensation in female mammals. Here, one of the two X chromosomes is inactivated in somatic cells in order to ensure equivalent levels of gene expression from sex chromosomes in males and females (Avner and Heard 2001).

Altogether, heterochromatic domains regulate gene expression in an epigenetic manner, meaning that heritable changes in gene activation occur without a corresponding change of the primary DNA sequence. Notably, epigenetic gene regulation is not only important to maintain certain expression states in differentiated cells, it also has a tremendous impact on processes that change global patterns of gene expression during development. For instance, [page 11↓]the genome-wide DNA‑methylation status undergoes dynamic changes during early embryogenesis of mammals (Li 2002), and thereby influences the organization and compartmentalization of the genome during tissue development. The disruption of this patterning may result in global genomic deregulation, since the inactivation of responsible DNA methyltransferases causes early embryonic lethality (Okano et al. 1999).

In contrast to heterochromatin, the less densely packaged structure of euchromatin makes it more accessible to proteins (Fig. 1.2). This may underlie the early time point of euchromatic replication during S-phase, which may in turn propagate the continuity of the open chromatin structure (Gilbert 2002). Accessibility is also the basis for the transcriptional activity of euchromatin. However, transient local chromatin remodeling processes are also required in euchromatin to deal with the general repressive character of nucleosomes. For instance, nucleosomes can occupy binding sites for transcription factors within promoter regions. In this case, nucleosomal repositioning is facilitated by specialized ATP-dependent chromatin remodeling complexes (Becker and Horz 2002). A prominent example is the SWI/SNF complex in yeast, which mediates the sliding of nuleosomes along the DNA template and also recruits further transcription-activating factors (Whitehouse et al. 1999; Krebs et al. 2000).

1.5 Chromatin modifying processes

In light of the above, the dynamic chromatin structure is a prerequisite for both global and local regulation of gene expression. Correspondingly, principle chromatin modulating mechanisms are conserved among eukaryotes, although there are differences in the components between the species (Moazed 2001).

One such mechanism specific to higher eukaryotes is DNA methylation. In mammals, DNA is methylated predominantly at cytosines of CpG dinucleotides and occurs in temporally and spatially variable patterns (Bird 2002). In essence, DNA methylation induces transcriptional repression, either by blocking transcription activators from binding or by recruiting repressive chromatin-remodeling proteins such as histone deacetylases (HDACs) or histone methyltransferases (HMTs) (Li 2002). DNA methylation is further involved in stable X chromosome inactivation in female mammals, along with a non-coding RNA (Xist) (Plath et al. 2002). Interestingly, non-coding RNA is also involved in dosage compensation in Drosophila. Here, rather than inactivating one X chromosome in females, the single X chromosome in males is hypertranscribed by the association of MSL (male-specific lethal) complexes that contain at least two roX RNAs (Park and Kuroda 2001).


[page 12↓]

Besides these mechanisms, a pivotal role in chromatin regulation in eukaryotes is played by histones and their posttranslational modifications. Each core histone is a small, highly basic protein composed of a globular histone fold domain, which interacts with the other histones of the nucleosome, and a N‑terminal “tail” that extends outwards from the nucleosome. Conserved histone tails, particularly those of H3 and H4, are subject to a variety of posttranslational modifications (Fig. 1.3). In the current model, these modifications not only modulate the strength of interaction between the histones and the DNA, and thereby the packaging state of chromatin, but they also serve as markers for specific histone binding proteins that further regulate the chromatin structure.


Fig. 1.3: Histone tail modifications.
The amino termini of core histones contain diverse posttranslational modifications. The diagram indicates known modifications at specific residues of human histones H3 and H4. M = methylation, A = acetylation, P = phosphorylation. (adapted from Lachner et al. (2003))

Histone modifications identified so far include methylation, phosphorylation, ubiquitination and ADP ribosylation (Rea et al. 2000; Sun and Allis 2002; Garcia-Salcedo et al. 2003). Among these, the ε-N-acetylation of lysine residues is the most prominent modification and a universal epigenetic mark in eukaryotes. Different conserved HAT and HDAC complexes change the acetylation state of histones in a dynamic manner (Dutnall and Pillus 2001; Carrozza et al. 2003). Generally, histone acetylation correlates with transcriptional activity, whereas histone hypoacetylation is a conserved hallmark of heterochromatin. In theory, acetylation affects transcription by neutralizing the histone charges, which weakens histone-DNA and internucleosomal contacts, thereby reducing chromatin compaction (Workman and Kingston 1998). In addition, the presence or absence of acetylation at specific lysines provides recognition sites for factors involved in activation or repression of gene expression. For example, deacetylation of lysine 16 on histone H4 (H4K16) by the NAD+‑dependent [page 13↓]HDAC Sir2 facilitates the binding of the silencing proteins Sir3 and Sir4 to nucleosomes and the spreading of the silencing complex in S. cerevisiae (see chapter 1.6) (Hoppe et al. 2002). In addition, X‑chromosome upregulation in males of Drosophila requires an increased level of H4K16 acetylation on this chromosome (Akhtar and Becker 2000).

Another conserved histone modification is lysine methylation (Lachner and Jenuwein 2002). Like hypoacetylation, H3K9 methylation generally correlates with heterochromatin in higher eukaryotes. The responsible HMT Su(var)3‑9 was identified to suppress PEV in Drosophila (Tschiersch et al. 1994), and its homologs in human (SUV39H1) and fission yeast (Clr4) act likewise in transcriptional repression (Nakayama et al. 2001; Peters et al. 2001). As in Drosophila, this occurs by the association of the structural heterochromatin protein HP1, which specifically interacts with methylated H3K9 and Su(var)3‑9 via conserved chromo- and chromoshadow domains (Bannister et al. 2001; Yamamoto and Sonoda 2003). Similarly, SUV39H1 and Clr4 recruit HP1 homologs and build SUV39H1/HP1 and Clr4/Swi6 methylation systems (Lachner et al. 2001).

Budding yeast also owns HMTs with a catalytic SET domain as in their metazoan counterparts. Moreover, methylation is equally associated with transcriptional repression, for instance in case of H3K4 and H3K36 methylation by Set1 and Set2, respectively (Briggs et al. 2001; Strahl et al. 2002). Recent studies demonstrated that Set2 methylates H3K36 in the coding region of actively transcribed genes (Krogan et al. 2003). In addition, its interaction with RNA polymerase (Pol) II implicates a role for Set2 in transcription elongation. The involvement of HMTs in a dynamic process such as transcription is particularly interesting given the present model that histone methylation is an irreversible modification.

Multiple covalent modifications occur at the same time on histone tails, and there are several examples for their interplay. For instance, on mammlian histone H3, the methylation of lysine 9 interferes with the phosphorylation of serine 10 (Rea et al. 2000), which is in turn synergistically coupled to the acetylation of lysines 4 and 9 (Cheung et al. 2000a; Clayton et al. 2000). Moreover, other findings suggest a link between two different epigenetic marks, namely DNA methylation and histone methylation (Tamaru et al. 2003).

The fact that posttranslational histone modifications can influence each other either positively or negatively gave rise to the hypothesis that they create a combinatorial code. This specific histone code may induce the recruitment of a certain set of chromatin-associated proteins that eventually dictate the particular state of gene expression (Strahl and Allis 2000). The hypothesis further states that protein motifs may have evolved that recognize histone modifications. In line with this, several proteins associated in gene regulation or [page 14↓]heterochromatic silencing were shown to share conserved domains, such as the acetyl-lysine binding bromodomain or the methyl-lysine binding chromodomain (Owen et al. 2000; Bannister et al. 2001). The histone code may also explain why certain types of histone modifications, e.g. methylation, can be involved in transcriptional activation as well as repression. It might further have the potential to reveal the mechanisms behind the maintenance of epigenetic chromatin marks throughout the cell cycle. To date, this process is poorly understood, although some first data exist. For instance, interactions between histone modifying enzymes and proteins of the replication machinery, e.g. the HAT Sas2 and the chromatin assembly factor subunit Cac1 in yeast (Meijsing and Ehrenhofer-Murray 2001) may contribute to the reestablishment of a given histone code on freshly replicated DNA. As another example, the Drosophila Fab-7 chromosomal element, the binding site for Polycomb and trithorax proteins to regulate homeotic genes, could convey the maintenance of an active chromatin state during mitosis and meiosis, possibly with H4 hyperacetylation as a heritable tag of the activated element (Cavalli and Paro 1999). Still, these are only first insights into different aspects of the complex, yet important, field of epigenetic inheritance.

1.6 Silencing in S. cerevisiae

Silenced chromatin in S. cerevisiae is akin to heterochromatin in higher organisms and shares main characteristics, such as general inaccessibility of DNA, hypoacetylated nucleosomes and late replication (Loo and Rine 1995; Lustig 1998). In addition, the overall pathway of assembly of silent chromatin appears to be similar in yeast and multicellular eukaryotes, since silencing proteins in yeast are functionally related to heterochromatic components in metazoans, such as Sir1 and HP1. Similar to higher organisms, budding yeast exhibits histone modifications, which are in parts species specific, such as H4K12 acetylation (Lachner et al. 2003). In light of these similarities, silenced loci in S. cerevisiae provide an excellent system to study the mechanisms of heterochromatin in eukaryotes.

In S. cerevisiae, silencing is facilitated by cis-acting elements and trans-acting proteins. There are three silenced loci known within the yeast genome, namely the silent mating-type loci HML and HMR, the rDNA array and the telomeres. Among these, the silent mating-type loci are the best characterized.

[page 15↓]

The silent mating-type loci HML and HMR

Haploid yeast cells exist as either a or α mating-type, which is determined by the MAT locus located near the centromere of chromosome III (Fig. 1.4). In MAT a cells, the MAT locus encodes the proteins Mata1/Mata2, and in MATα cells it encodes Matα1/Matα2, which are proteins that regulate the transcription of mating-type specific genes and therewith enable the cell to mate (Herskowitz et al. 1992). During mating, cells of opposite mating-type fuse to form a /α diploids, which in turn can undergo meiosis and sporulation to generate haploid progeny. Under certain conditions, haploid cells can switch their mating-type. This is possible, since two additional copies of MAT are present on the left and on the right arm of chromosome III, namely HML ( h omothallic m ating l eft) containing α‑information, and HMR ( h omothallic m ating r ight) containing the a‑information. Mating-type switches occur via specific recombination events between MAT and the HM loci, but are inhibited in laboratory strains due to the deletion of the responsible HO-endonuclease (Strathern et al. 1982).

Given the existence of genes of both mating-types in the yeast genome, cells must ensure that only the MAT locus is employed in order to preserve their mating ability. Thus, the silent mating-type or homothallic (HM) loci are transcriptionally silenced. Silencing defects at HML or HMR cause the derepression of these loci and subsequently a pseudodiploid state, which prevents the cells from mating. Thus, the silencing state of a single HM locus can be determined by testing the mating ability of a strain of opposite mating-type.

HM silencing is achieved by the presence of silencer elements on each side of the loci, the so-called E and I silencers. These are cis-acting, regulatory elements consisting of binding sites for the DNA binding proteins Rap1, Abf1 and the origin recognition complex, ORC (Fig. 1.4). Whereas number and orders of the different binding sites vary, an ORC binding site is present in all silencers.

Besides its role in silencing, the ORC complex has a well-conserved function in replication initiation (see chapter 1.7). However, this function appears not to play a role in silencing, since although its establishment requires S-phase passage, silencing does not require replication initiation or replication fork passage through the silencers (Kirchmaier and Rine 2001). In addition, the ORC binding sites do not need to be active replication origins to act in silencing. For example, the ORC binding sites within the HML silencers are no active origins, although they can serve as origins on plasmids (Sharma et al. 2001). In contrast, HMR-I and HMR-E are chromosomal origins of replication (Rivier et al. 1999), though the latter is inefficient. [page 16↓]Presumably, it is the tight binding of ORC to HMR-E that enhances its silencer activity but decreases its origin potential (Palacios DeBeer et al. 2003).

The four HM silencers do not only vary in their composition, but they are also of different importance for the silencing state of the respective locus. At HML, either E or I individually are sufficient to maintain the silencing state (Mahoney and Broach 1989). In contrast, at HMR, E is essential but I is dispensable for silencing (Rivier et al. 1999). The elements of HMR-E are functionally redundant, meaning that the absence of at least two of them is necessary to cause the loss of silencing. This redundancy is lost at the synthetic HMR-E silencer, which is engineered of minimal binding sequences for Rap1, Abf1 and ORC (McNally and Rine 1991).

Interestingly, all three binding proteins act individually also elsewhere in the genome without initiating transcriptional repression. For instance, Rap1 (Repressor/Activator Protein 1) and Abf1 (ARS Binding Factor 1) have essential functions as transcriptional activators of diverse genes (Lieb et al. 2001; Miyake et al. 2002). Thus, the ability of HM silencers to initialize silencing appears not just to be the sum of the silencing abilities of their single elements. It may rather be the combination of the elements and their close proximity that allows the recruitment of further silencing components, the Sir proteins, which eventually induce the formation of silent chromatin (Lustig 1998).

The four Sir (silent information regulator) proteins are trans-acting silencing factors. Sir2, Sir3 and Sir4 are the structural components of silenced chromatin and essential for silencing, but non-essential for growth (Rine and Herskowitz 1987). The formation of silenced chromatin is associated with the polymerizing Sir complex that interacts with nucleosomes and thereby spreads outward from its nucleation site at the silencers. The assembly of this complex is hypothesized to occur stepwise (Hoppe et al. 2002). At first, Sir2/Sir4 heterodimers bind via interactions of Sir4 with Sir1 and Rap1 to the silencer. At the same time, Sir3 binds independently via interactions with Rap1 and Sir4. The next step requires the enzymatic activity of Sir2, a NAD+-dependent histone deacetylase specific to H3K9, H3K14 and H4K16 (Imai et al. 2000). Histone deacetylation by Sir2 facilitates the binding of Sir3 and Sir4 to hypoacetylated histones and the recruitment of new Sir2/Sir4 (Rusche et al. 2002). Repetitions of these modification/binding cycles eventually result in multimerization and spreading of the complex along the chromosome (Fig. 1.4). This model is based on several individual observations of genetic or physical interactions between the different components (Hecht et al. 1995; Gasser and Cockell 2001; Rusche et al. 2002).

Sir1 is no part of the multimeric Sir complex, but is rather proposed to facilitate the establishment of silencing due to its ability to interact with ORC and Sir4 (Triolo and [page 17↓]Sternglanz 1996). This is supported by ChIP data of Rusche et al. (2002), who found Sir1 located primarily to the silencers and not distributed over the whole HM loci as the other Sir’s. In contrast to them, Sir1 is not vital to silencing (Rusche et al. 2002).

Although the silencing complex spreads in both directions along the chromosome, it is stopped from propagation into adjacent regions that are kept transcriptionally active by DNA elements interposed between silenced and active chromatin domains (Dhillon and Kamakaka 2002). Such so-called boundary elements have been identified at either side of HMR, as well as in subtelomeric regions, and are also known in other species (Gombert et al. 2003; Parnell et al. 2003). The boundary at the telomeric proximal side of HMR is a tRNA gene, which requires an intact transcriptional potential for its barrier capacity (Donze and Kamakaka 2001). This implicates that it functions passively by a stably bound protein complex (e.g. the RNA Pol III pre-initiation complex), which interferes physically with the spread of the silenced complex. Alternatively, boundaries may be active enzymatic barriers on the bases of associated HATs and chromatin remodeling enzymes, which oppose the propagation of hypoacetylated silenced chromatin (Suka et al. 2002).

Abb. 1.4: Mating-type loci and HM silencers.
The mating-type loci MAT, HML and HMR are localized on chromosome III of S. cerevisiae. HML and HMR are repressed due to the nearby silencers E and I, which consist of binding sites for ORC, Rap1 and Abf1. The silencers are nucleation sites for silencing complexes, as depicted for HMR-E. The Sir complex interacts with nucleosomes and spreads into the HMR locus thereby creating a silenct chromatin structure.


[page 18↓]

The spread of the silencing complex from HML-I is stopped by the CHA1 promoter, which locates about two kb downstream of the silencer (Donze and Kamakaka 2001), whereas the YCL069w locus is probably the boundary of HML-E (Lieb et al. 2001).

How does silenced chromatin inhibit transcriptional activity? Its compact structure was formerly proposed to prevent the access of transcription enzymes to promoters (Loo and Rine 1994). However, more recent data suggest that silenced chromatin prevents the elongation step rather than the recruitment of RNA Pol II, since factors of the transcriptional machinery cohabitate with Sir proteins at promoters of silenced chromatin (Sekinger and Gross 2001).

From a mechanistic point of view, chromatin silencing can be subdivided into three distinct processes, namely establishment, maintenance and inheritance. Establishment refers to the de novo generation of repression at active loci. Besides S‑phase passage, this requires Sir1, since sir1Δ strains are mixed populations of cells whose HM loci are either completely repressed or completely derepressed (Pillus and Rine 1989). To maintain the silenced state throughout one cell cycle, structural components of the Sir complex as well as intact histone tails are required (Cheng and Gartenberg 2000). In addition, mutations in subunits of CAF-I give rise to unstable HML repression (Enomoto and Berman 1998). Thus, CAF-I may not only assemble newly synthesized histones onto freshly replicated DNA, but also help to reassociate the Sir complex. As aforementioned, inheritance refers to the propagation of silencing to subsequent cell cycles. It requires the silencers as epigenetic markers (Rusche et al. 2002).

The telomeres

Telomeres are protected from exonucleolytic degradation, end-to-end fusion and recombination by their heterochromatic structure (Grunstein 1998; Stevenson and Gottschling 1999). In S. cerevisiae, telomeres consist of approximately 300 bp of C1‑3A/TG1‑3 repeats, which build nucleosome-free areas with multiple binding sites for Rap1 (Fig. 1.5) (Sandell and Zakian 1993). In the current view, Rap1 recruits Sir2/Sir4 and Sir3, which then form a complex in a similar mode as at the HM loci: after getting in contact with nucleosomes of the adjacent chromatin region, Sir3 and Sir4 interact with histone tails deacetylated by Sir2 and the polymerizing complex spreads inwards the chromosome thereby silencing subtelomeric genes (Luo et al. 2002).

Notably, telomeric silencing occurs discontinuously and is enhanced around subtelomeric CoreX-elements (Fourel et al. 1999; Pryde and Louis 1999). These elements are part of all [page 19↓]chromosomes and contain an ORC binding site (ACS; ARS consensus sequence), often coupled with an Abf1 binding site. The silencing maximum around CoreX depends on the ACS and, additionally, on SIR2, SIR3 and SIR4. Notably, the deletion of SIR1 also causes partial derepression at native telomeres, whereas mutations in the ORC subunits ORC2 and ORC5 have no effect (Pryde and Louis 1999). In the current model, silenced chromatin at telomeres is organized by interactions of telomeric Rap1‑Sir complexes with Sir proteins bound to CoreX under formation of a loop structure, which may further stabilize the heterochromatin-like complex (Strahl-Bolsinger et al. 1997; Pryde and Louis 1999).

Interestingly, Sir3 was found to be limiting for the propagation of the silencing complex (Renauld et al. 1993), and its overexpression extended the silent domain from 2-4 kb to up to 16 kb away from the telomeric repeats (Hecht et al. 1996). This extension coincided with the spread of Sir3, whereas the amount of Sir2 and Sir4 was reduced in telomere-distal chromatin (Strahl-Bolsinger et al. 1997).


Fig. 1.5: Silent chromatin at a yeast telomere.
The telomeric (TG1-3) repeats provide binding sites for Rap1, which recruits the Sir complex. The subtelomeric CoreX element contains a binding site for ORC and acts likewise as a nucleation site for the Sir complex. Due to interactions of the silencing proteins the telomere folds back and forms a loop, which further stabilizes the chromatin structure.

Rap1, Sir3, Sir4 and clusters of telomeric DNA were observed to colocalize in foci at the nuclear periphery (Gotta et al. 1996). These foci may be tethered to the nuclear envelope through interactions with the nuclear pore complex (Galy et al. 2000). To date it is not clear whether this perinuclear position is the cause or the consequence of telomeric silencing. Feuerbach et al. (2002) have demonstrated a repression-dependent, physical relocation of telomeres from variable intranuclear positions to perinuclear silent domains (Feuerbach et al. 2002). Hence, they hypothesized that the expression state of telomeres is determined by spatial positioning. In contrast, Tham and co-workers found no correlation between [page 20↓]transcriptional silencing of telomeres and their localization to the nuclear periphery (Tham et al. 2001)

Telomeric silencing is usually investigated in subtelomeric reporter strains. For this purpose, reporter genes were inserted at artificially truncated chromosome ends that lack the CoreX element (Gottschling et al. 1990). In these constructs, heterochromatin spreads continuously from the (TG1‑3) repeats towards the centromere (Renauld et al. 1993). Subtelomeric reporter genes are subject to epigenetic switches between transcriptional repression and expression (Chien et al. 1993). Since this variegation resembles position effects in Drosophila, it is referred to as telomeric position effect (TPE) (Gottschling et al. 1990). TPE may originate from a weaker establishment potential of silencing at the truncated ends due to the lack of Sir1, since sir1Δ does not affect the silencing of subtelomeric reporter strains (Fox et al. 1997). Interestingly, the conditional mutant alleles orc2-1 and orc5-1 caused silencing defects the truncated reporter constructs, in contrast to the missing effects at native telomeres (Fox et al. 1997).

The common components of silent chromatin at telomeres and the HM loci suggest a competition for limiting factors between them. Consistent with this idea, increased telomeric silencing goes along with decreased silencing at HMR (Buck and Shore 1995).

The rDNA locus

In the nucleolus, ribosomal DNA (rDNA) sequences are present in a tandem array of 100-200 copies of a 9.1 kb repeat (Fig.1.6). Each repeat encodes a 5S RNA, transcribed by RNA Pol III, and a 35S precursor RNA, transcribed by RNA Pol I and subsequently processed to 18S, 5.8S, and 25S RNA. The 35S coding regions are separated by nontranscribed spacers, NTS1 and NTS2 (Smith and Boeke 1997).

The highly repetitive nature of the rDNA array necessitates the formation of silenced chromatin to avoid recombination events. Consequently, only about half of the repeats are active at a given time point, whereas the other half is transcriptionally silent (Warner 1989). In addition, Pol II reporter genes inserted into the rDNA become also metastably repressed (Smith and Boeke 1997). Although the mechanism behind this repression is currently not well understood, it is known that rDNA silencing is mediated by a protein complex called RENT (regulator of nuceolar silencing and telophase). RENT contains the subunits Net1, Sir2, and Cdc14 (Shou et al. 1999), and was recently shown to localize to two distinct regions within the rDNA repeats (Huang and Moazed 2003). It binds to NTS1 via Fob1, which surprisingly is also [page 21↓]required for rDNA recombination (Kobayashi and Horiuchi 1996), and to NTS2 around the Pol I promoter.

Each repeat also contains an ACS site in NTS2. However, only about 20% of them are active origins. These are clustered along the rDNA array and separated by large regions where replication initiation is suppressed in a SIR2-dependent manner (Pasero et al. 2002). Therefore, like transcription, rDNA replication is under epigenetic control. Deletions of SIR2 shorten the life span of yeast cells, whereas its overexpression causes cells to live longer (Kaeberlein et al. 1999). This role of SIR2 as an anti-aging factor was found to be connected with its function in rDNA silencing. Loss of Sir2 results in reduced rDNA silencing and hence in increased recombination between the repeats, which eventually causes the accumulation of extrachromosomal rDNA circles (ERCs). These ERCs cause aging presumably because they titrate components of the replication or transcription machinery from the genomic DNA (Sinclair and Guarente 1997). Interestingly, calorie restriction also leads to life span extension on the basis of reduced rDNA recombination. Here, the activity of Sir2 may be increased due to the higher concentration of NAD+ in calorie restricted cells (Lin et al. 2000).


Fig. 1.6: Schematic structure of the rDNA array in S. cerevisiae.
The rDNA locus is an array of tandemly repeating units containing the coding regions for ribosomal RNA seperated by non-transcribed spacer regions NTS1 and NTS2. The latter holds a binding site for ORC. Binding sites for the silencing RENT complex are depicted by arrows. (adapted from Huang and Moazed (2003))

Althogether, the mode of rDNA silencing is different from that of HM loci and telomeres, since it requires only Sir2, but not the other Sir proteins. Nevertheless, all silenced loci may be linked by competition for limiting amounts of Sir’s. In line with this model, rDNA silencing is negatively regulated by the telomeres, which titrate Sir2 out of the nucleolus and sequester it via interactions with Sir4 (Smith et al. 1998).


[page 22↓]

1.7  Silencing proteins investigated in this thesis

The following section provides additional information on those silencing proteins that were in the focus of this study.

ORC and its largest subunit Orc1

The ORC complex consists of six subunits named Orc1 to Orc6 in order of their decreasing mass (Li and Herskowitz 1993; Bell et al. 1995; Loo et al. 1995a). All subunits are essential for the conserved function of ORC as the eukaryotic replication initiator complex (Bell et al. 1993; Gavin et al. 1995). Homologs of ORC subunits have been found implicated in replication also in S. pombe, D. melanogaster, X. laevis, and human cells (Carpenter et al. 1996; Grallert and Nurse 1996; Landis et al. 1997; Dhar et al. 2001).

In yeast, ORC binds to ACS sites of origin sequences, which are evenly distributed in the genome, and where ORC remains bound throughout the cell cycle (Tanaka et al. 1997). DNA binding requires the coordinate action of all ORC subunits except Orc6 (Lee and Bell 1997). In addition, it requires the binding, but not the hydrolysis, of ATP by Orc1 (Klemm and Bell 2001). To initiate replication, ORC recruits a multifactor prereplicative complex (pre-RC) during G1. Thereby, the direct binding of ORC to Cdc6 is the first and a key step, and is presumably mediated by ATP-bound Orc1 (Saha et al. 1998; Mizushima et al. 2000).

Due to their vital function, no deletion mutants of ORC subunits are available. Instead, the conditional mutant alleles orc2‑1 and orc5‑1 are frequently used for genetic analysis. Both mutants share phenotypes of impaired replication, including temperature sensitivity, elevated plasmid loss rate and reduced replication initiation. Moreover, double mutants are inviable (Liang et al. 1995; Loo et al. 1995a). It was furthermore demonstrated that the ORC complex is unstable and affected in DNA binding in these mutants (Bell et al. 1993). In addition to the replication defect, telomeric and HM silencing was also affected in orc2‑1 and orc5‑1 mutants (Loo et al. 1995a; Fox et al. 1997).

In contrast, a N‑terminally truncated orc1 mutant displayed no combined replication/silencing phenotype, but was impaired in silencing. Replication appeared unaffected, since orc1Δ1-235 could still complement an orc1Δ strain for growth and displayed only a 2-fold reduction in plasmid stability, compared to a 20 to 40-fold reduction in orc2‑1 and orc5‑1 strains (Bell et al. 1995). In this orc1 mutant, silencing was affected at a HMR sensitized to defects in ORC function by the lack of the Rap1 binding site in the E silencer (HMR-E ΔRAP).


[page 23↓]

In the current model, the N-terminal domain of Orc1 is responsible for the silencing function of ORC, which is the recruitment of Sir1 to silencers (Gardner et al. 1999). Recently, an ORC interaction region (OIR) was identified in the C-terminal part of Sir1 to be necessary and sufficient for the Sir1-ORC interaction (Bose et al. 2004). However, stable silencer association of Sir1 required the additional interaction with Sir4, which may consequently confine Sir1-ORC interactions to origins within silencers.

In the 914 amino acid protein Orc1, Sir1 binding occurs via a small non-conserved domain between amino acids 100 and 129 (Zhang et al. 2002). This so‑called H‑domain is part of the BAH (bromo-adjacent homology) domain, which is a conserved protein-protein interaction module (Callebaut et al. 1999). In addition, Orc1 has an AAA+ (ATPases associated with a variety of cellular activities)domain between amino acids 443 and 738. This highly conserved module contains ATP binding and hydrolysis-mediating Walker homology motifs (Neuwald et al. 1999). Hence, the AAA+ domain is essential for replication initiation by ORC, and mutations in this domain are lethal (Klemm and Bell 2001)

Most likely, the role of ORC in repressive chromatin is conserved in all metazoans. In Drosophila, ORC was localized to heterochromatin and interacted directly with HP1, probably via DmOrc1 (Pak et al. 1997). Recessive lethal mutations in DmORC2 are PEV suppressors and disrupt the localization of HP1 to heterochromatin. Likewise, HP1 point mutations that diminish ORC binding, also suppress PEV. In light of these data, Pak et al. (1997) suggested a conserved role of ORC to target non-DNA-binding-factors, such as Sir1 in yeast and HP1 in Drosophila, to sites destined to be heterochromatic.

Recently, it was demonstrated that human ORC1 directly interacted with HBO1 (histone acetyltransferase binding to ORC). HBO1 has HAT activity towards free and nucleosomal H3 and H4 (Iizuka and Stillman 1999) and belongs to the same HAT family (MYST) as MOF and Sas2, which affects silencing in yeast (Ehrenhofer-Murray et al. 1997). Thus, although the ORC-HBO1 interaction may be associated with the replication role of ORC (Burke et al. 2001), it may alternatively be another link between ORC and heterochromatin.

Sir3

Sir3 is a key player in TPE and HM silencing, but not in rDNA repression (Aparicio et al. 1991; Smith and Boeke 1997; Stone et al. 2000). As a component of the Sir complex, Sir3 contacts several other silencing proteins. Direct interactions have been demonstrated between Sir3 and Sir2, Sir4, Rap1, Abf1, Zds1, Zds2, Rad7, Sir3 itself and the N-termini of histones H3 and [page 24↓]H4 (Gasser and Cockell 2001). Physical interactions with Sir4, Rap1 and hypoacetylated histones occur via the C‑terminal half of Sir3 (Hecht et al. 1995; Park et al. 1998; Moretti and Shore 2001), whereas the N‑terminus of the 978 amino acid protein modulates the interactions. Notably, the simultaneous expression of both halves of SIR3 in trans partially complemented the sir3Δ mating defect, suggesting that the two domains can function independently (Gotta et al. 1998).

The N-terminal 214 amino acids of Sir3 are very similar to Orc1 (50% identity, 63% similarity) (Bell et al. 1995) and also contain a BAH domain between amino acids 48 and 189, although the H‑domain is missing (Zhang et al. 2002). Point mutations within the BAH domain lead to eso (enhancers of the s ir o ne mutant mating defect) phenotypes (Stone et al. 2000). They disrupted HM silencing in sir1Δ strains and additionally disrupted TPE as single mutants. Interestingly, nat1Δ also enhanced the sir1Δ mating defect, and this effect was epistatic with some of the sir3-eso mutations.

In line with the high degree of sequence similarity, the N-terminal domains of Sir3 and Orc1 were functionally interchangeable for mating-type silencing when tethered to the C-terminus of the other protein (Bell et al. 1995). Notably, they did not substitute each other in telomeric silencing, pointing to distinct functions of the proteins at the telomeres (Stone et al. 2000). Given the potential of the Sir3 N‑terminus to replace the Orc1 N‑terminus in HM silencing, it appears paradox that Sir3 cannot interact with Sir1 because of the missing H‑domain. Thus, Stone et al. (2000) proposed that the BAH domain of Sir3, when tethered to Orc1, may promote silencing in a Sir1 independent manner.

Consistent with the view that the Sir3 N‑terminus is a regulatory domain, there is evidence that its phosphorylation enhances TPE (Stone and Pillus 1996).

Sum1-1

Due to a single missense mutation in the C‑terminal part, SUM1‑1 is a dominant altered function allele, which can bypass the need for Sir proteins in HM silencing and increase telomeric repression in SIR wild‑type strains (Laurenson and Rine 1991).

Although SUM1‑1 is a suppressor of HM silencing defects, the wild-type gene product of SUM1 appears not to be a direct silencing component. Instead, Sum1 acts as transcriptional repressor of middle sporulation genes during mitosis and vegetative growth and binds specifically to MSE (middle sporulation element) sites in the promoter regions of its target genes (Xie et al. 1999). For this, Sum1 recruits Hst1 (Homologous of Sir Two), a NAD+-[page 25↓]dependent deacetylase with sequence homology to Sir2. The deletion of SUM1 had only minor effects on HM silencing and did not restore silencing in the absence of Sir proteins (Chi and Shore 1996).

For a long time it remained unclear how the Sum1‑1mediates silencing in the absence of the Sir complex, which is normally essential for silencing. (Sutton et al. 2001) found that Sum1‑1 also requires Hst1 and its NAD+-dependent deacetylase activity as well as ORC for its silencing function. In fact, the orc1Δ1‑235 allele eliminated SUM1-1 mediated silencing (Rusche and Rine 2001). In the present model, Sum1‑1 is bound by ORC (Orc1?) to the silencers and recruits Hst1, whose deacetylase activity leads to hypoacetylated nucleosomes and consequently to a condensed, silenced chromatin structure at the HM loci (Rusche and Rine 2001).

1.8 Outline of this thesis

The aim of this study was to determine the role of the Nα‑acetyltransferase complex NatA in transcriptional silencing in S. cerevisiae. Deletions of the NatA subunits NAT1 or ARD1 both result equally in impaired HML silencing and reduced TPE, suggesting the functional dependence of a silencing protein on Nα‑acetylation by NatA (Mullen et al. 1989; Aparicio et al. 1991).

So far, some genetic interactions between NAT1 and genes that encode silencing components have been identified. For instance, the nat1Δ mutant displayed an eso‑phenotype, which was not enhanced in combination with certain sir3‑eso alleles. Likewise, overexpressed Sir1 suppressed the nat1Δard1Δ silencing defect at the HMR‑E ΔRAP silencer (Stone et al. 1991 and 2000). However, neither Sir3 nor Sir1 have been directly implicated in NatA‑dependent silencing. In addition, histone H2B is a known NatA substrate but the deletion of its N‑terminus has no effect in silencing (Kayne et al. 1988).

To date, no significant silencing substrates of NatA have been found and the mechanism by which NatA is involved in silencing remains unclear. It is further not known whether NatA plays a role in rDNA silencing.

In this study, we found that NatA was required for all forms of silencing in S. cerevisiae. We further obtained evidence that Orc1 is a NatA substrate and its Nα‑acetylation is required for telomeric silencing. Genetically, nat1∆ functioned through the ORC binding site of the HMR‑E silencer. The requirement for NatA in silencing could be bypassed by artificially tethering [page 26↓]Orc1, but not the other Orc proteins, to the silencer, thus suggesting that Orc1 was a silencing-relevant NatA target. We found Orc1 to be fully Nα‑acetylated in wild-type and completely unacetylated in nat1∆ strains. Mutations in the penultimate residue of Orc1 that abrogated its ability to be acetylated by NatA caused a severe loss of telomeric silencing, as does the deletion of NAT1. The lack of acetylation did not affect the interaction of Orc1 with Sir1, since HM silencing was not impaired in the orc1 mutants and still depended on functional SIR1 in nat1Δ strains.

Genetic interactions further supported a functional link between NatA and ORC in replication, since nat1∆ was synthetically lethal with the replication-defective orc2-1 mutation. Notably, unacetylated orc1 mutants grew normally suggesting that another subunit of ORC requires Nα‑acetylation for its function in silencing. Furthermore, nat1∆ displayed synthetic lethality with SUM1‑1. Intriguingly, this lethality was suppressed by a deletion in the N‑terminus of Orc1, thus suggesting that Nα‑acetylation regulated the interaction of Orc1 with Sum1-1.

Furthermore, we found that the N-terminal 100 amino acid region of Orc1 was dispensable for growth, but had a function in silencing. Increasing deletions within this region disrupted silencing at the synthetic HMR locus and telomeres, and also reduced the α‑factor sensitivity of the mutants. In contrast to earlier proposals (Zhang et al. 2002), we found that the N‑terminal 50 amino acids of Orc1 were required for the interaction with Sir1, since the two-hybrid interaction with Sir1 was interrupted in the orc1Δ1‑51 mutant. However, this mutant further affected silencing in sir1Δ. Therefore, we suggest that the N-terminal 100 amino acids of Orc1 are not only required for Sir1 interaction, but also for the recuitment of another, yet unknown silencing factor.

Furthermore, we present evidence that Sir3 is also Nα‑acetylated by NatA. Since previous work (Stone et al. 2000) showed that the mutation of the penultimate amino acid of Sir3 causes silencing defects, we likewise propose that Sir3’s silencing function is regulated by NatA-dependent Nα‑acetylation and we further demonstrate that the localization of Sir3 to perinuclear foci depends on NAT1.

In addition, we report on a screen for a multicopy suppressor of the nat1Δ mating defect. This unbiased approach proved to be ineffective, since we isolated only indirect suppressors of the mating defect, but not of the HM silencing defect of nat1Δ.

In summary, our data further specify the role of NatA in transcriptional silencing. For the first time, we provide evidence of the functional dependence of two silencing proteins, Orc1 and Sir3, on Nα‑acetylation by NatA. We propose a model, by which Nα‑acetylation regulates the [page 27↓]binding of silencing factors to the N‑terminus of Orc1 and Sir3 to recruit hetrochromatic factors and establish repression. Thus, Nα‑acetylation represents a protein modification that modulates chromatin function in S. cerevisiae.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 4.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
08.06.2005