In this study, we found that N-terminal deletions of 50, 100, and in some cases only 28 amino acids from the N‑terminus of Orc1 disturbed silencing at the HM loci and telomeres. Since this phenotype differed clearly from that of the missing N^α‑acetylation, different facets of Orc1’s function in silencing appear to be affected by the two types of mutations.

The N‑terminal 100 amino acids of Orc1 belong to the BAH domain. This domain is found also in a number of other chromatin-associated proteins, such as mammalian DNA (cytosine‑5) methyltransferases, components of the RISC chromatin-remodeling complex and histone deacetylase complexes (Callebaut et al. 1999), and was thus proposed to function as a protein-protein interacting module involved in transcriptional regulation and chromatin-mediated gene silencing.

By sequence homology analysis, the BAH domain of Orc1 had been originally located to amino acids 48 to189 (Callebaut et al. 1999). Crystallographic data of Zhang et al. (2002) redefined the region between amino acids 10 and 190 as the BAH core domain, whose secondary structure is mainly composed of β‑strands (Fig. 4.1). In addition, the so-called H‑domain, a small, non‑conserved helical subdomain between amino acids 100 and 129 (β6 and β7), was identified to be necessary and sufficient for the interaction with Sir1. This interaction is supported by the core domain, since a number of amino acid interactions stabilize the position of the H‑domain with respect to the core (Zhang et al. 2002). Notably, almost all of the amino acids participating in these interactions lie between amino acids 100 and 190, and the loop between amino acids 21 and 35 (connecting β1 and β2) was dispensable for Sir1 binding. This suggested that the region before amino acid 100 was not important for the function of the BAH domain. In contrast to this notion, we found that the two-hybrid interaction of Orc1 with Sir1 required the first 50 amino acids of Orc1, implicating that this region was also relevant for the BAH function.

Nevertheless, our data propose that the orc1 N‑terminal deletion mutants had further defects than the loss of Sir1 binding. Strikingly, truncated orc1 mutants did not resemble the α‑factor response phenotype of sir1Δ, and instead formed shmoo clusters, which indicate defects in the maintenance of silent chromatin rather than its establishment (Enomoto and Berman 1998). In contrast to that, an orc1 mutant missing the H‑domain responded to α‑factor similarly to sir1Δ (Zhang et al. 2002). Since the deletion of the N‑terminal 50 amino acids of Orc1 further decreased the α‑factor sensitivity of sir1Δ, we propose that the orc1 mutation caused a structural defect at the HML silencer, in addition to ineffective reestablishment of silencing due to the loss of SIR1. This structural defect might be based on the loss an interacting partner of Orc1 that is distinct from Sir1. This hypothesis is further supported by the telomeric phenotype of the orc1 truncation mutants, which was earlier characterized as a Sir1 independent effect.

Fig. 4.1: The crystal structure of the N-terminal domain of Orc1.
(A) Ribbon presentation. (B) Topology diagram showing the fold of the structure and deletion sites of the orc1 mutants investigated in this study. The BAH core structure is colored blue, the H-domain is shown in magenta, N- and C-terminal helices are shown in red. (adapted from Zhang et al. 2002)

Altogether, our data suggest that deletions within the first 100 amino acids of Orc1 not only have a destabilizing effect on the interaction with Sir1, but also impair the interaction of Orc1 with (a) further silencing partner(s). We hypothesize that this protein requires N^α‑acetylation as well as an intact N‑terminal structure of Orc1 for its binding (Fig. 4.2).

Interestingly, the deletion of the region between amino acids 28 and 52 of Orc1 also affected silencing, in contrast to the removal of amino acids 21 to 35 (Zhang et al. 2002). The telomeric phenotype of this mutant points to a special role of this region in the binding of the hypothesized silencing factor, which we propose to act in particular at the telomeres.

Why did the deletion of 10 amino acids from the N‑terminus of Orc1 not effect silencing? Whereas the BAH domain should be still functional in this mutant, N^α‑acetylation by NatA is expected to be interrupted. However, the penultimate amino acid tryptophan makes it to a possible substrate for the N^α‑acetyltransferase NatC. Thus, one conceivable scenario is that the N^α‑acetylation of Orc1(Δ1‑10) by NatC still enables it to interact with the hypothesized silencing factor and therefore to function like full-length Orc1. This could be tested in an IEF gel with Orc1(Δ1‑10) probes derived from wild-type and natC mutant strains.

According to our model, the region comprising the first 100 amino acids of Orc1 has a dual function in silencing: supporting the interaction of the H‑domain with Sir1 and providing the binding site for a further interacting factor. Further studies will be required to test our hypothesis. For example Co-IPs could test Sir1 binding to the truncated Orc1 versions, and ChIP assays can be used to test its association to silencers in the mutant strains. Moreover, it has to be specified what regions of Orc1 are required for the binding of new interacting partners.

Notably, the BAH domain is conserved in all known Orc1 homologs, posing the question of whether this motif mediates a role of Orc1 in transcripitional repression also in higher eukaryotes. Significantly, Orc1 is associated with heterochromatin in Drosophila, where its N‑terminus interacts physically with HP1, a central component of heterochromatin. In addition, the amino terminus of Xenopus Orc1 likewise interacts with HP1 homologs (Pak et al. 1997). Although it is not known at present whether the BAH module plays a direct role in this interaction, these data strongly suggest that not only the replication initiator function, but also the silencing function of ORC is evolutionary conserved.

Fig. 4.2: Model of protein interactions at the N-terminus of Orc1.
In addition to Sir1, the Orc1 N-terminus interacts with another, yet unknown protein. This protein requires the N^α‑acetylation of Orc1 for binding and is specifically recruited in the context of telomeric silent chromatin, when Sir1 is absent. In contrast, Sir1 is recruited via the H‑domain of Orc1 specifically to the HM silencers. This binding does not require the N^α‑acetylation of Orc1, but it depends on the integrity of the N‑terminal region of 100 amino acids.

In this study, we sought to reveal by which mechanism NatA functions at silenced loci. We identified Orc1 as a NatA substrate, and showed that its N^α‑acetylation was required particularly for telomeric silencing. Summarizing our data, we propose a novel silencing factor to bind to Orc1 in its N^α‑acetylated form.

In light of the high similarity of between Orc1 and Sir3 N‑termini, it is interesting that we also found Sir3 to be N^α‑acetylated by NatA. One possibility is that acetylated Orc1 and Sir3 both interact with the same hypothesized silencing factor, which is in agreement with the observation that double orc1 sir3 unacetylated mutants showed no additional silencing effects. However, these double mutants showed weaker derepression at the HM loci than nat1Δ, suggesting that further targets of NatA exist that require N^α‑acetylation for their silencing function.

We therefore propose a model, in which the effect of nat1∆ at the HM loci is the sum of several proteins lacking acetylation, among them Orc1 and Sir3. The cumulative effect on these proteins causes Sir3 and perhaps other Sir proteins to lose their ability to bind silenced chromatin, and thus causes derepression. Significantly, missing N^α‑acetylation caused only a partial loss of function in the two substrates here identified. In unacetylated Orc1, Sir1 recruitment was not affected, and unacetylated Sir3 did not completely disrupt silencing as seen in the sir3 null mutant (Rine and Herskowitz 1987).

Interestingly, the N-terminus of Sir3 interacts in two-hybrid assays with Abf1 (Gasser and Cockell 2001), and it is well conceivable that this interaction depends on N^α-acetylation of Sir3. This may provide an explanation for the missing effect of nat1Δ at HMR-E lacking the Abf1 binding site.

It would be in line with our model if some of the NatA targets in silencing acted not directly at the silenced loci, but rather would influence silencing indirectly by regulating silencing factors. To date, there exist only preliminary data on the regulation of members of the silencing complex. One example for such a regulation is the Sir3 hyperphosphorylation in response to mating pheromone, heat shock and starvation, which increases silencing (Stone and Pillus 1996). This modification requires an activated MAP kinase cascade, although the detailed mechanism of Sir3 regulation remains unclear (Ai et al. 2002; Ray et al. 2003).

How can further NatA substrates that require N^α‑acetylation for their silencing function be identified? Besides testing individual silencing components for N^α‑acetylation, genetic screens provide one possibility to reveal new genes linked to the silencing function of NatA. In light of the results of the multicopy-suppressor screen performed in this study, it is important to strictly focus the screen on those NatA substrates that act in silencing. In other words, when mating ability is used as a sensor for HM silencing, no genes involved in other aspects of mating should be isolated. This could be realized, for instance, by screening for mutants which reestablish silencing in a nat1Δ background. In addition, the screen has to be sensitive enough to detect slightest improvements of HM silencing, since single components have to be isolated out of an orchestra of components that constitute the nat1Δ phenotype.

To complete the picture of NatA’s silencing function, it is also necessary to uncover its involvement at the other silenced loci. Silencing of subtelomeric reporter genes was completely disrupted by the deletion of NAT1, and unacetylated orc1 and sir3 mutants displayed the same effect (Stone et al. 2000). At the moment it is not clear, whether the N^α‑acetylation of Orc1 and Sir3 alone constitutes the role of NatA at telomeres. Further experiments will be required to determine the structural integrity of the perinuclear foci in unacetylated orc1 and sir3 mutants and whether the association of other components, such as Rap1, is likewise affected.

In this study, we present evidence that NatA also has a function in rDNA silencing. Future work has to reveal which silencing components at the rDNA require N^α‑acetylation. Interestingly, both NatA substrates identified here are capable of binding to the rDNA array. Although Sir3 is not required for the repression of rDNA reporter genes (Bryk et al. 1997; Smith and Boeke 1997), it locates in the nucleolus of aging cells (Kennedy et al. 1997). Notably, the N‑terminus of Sir3 was shown to contain efficient information for nucleolar targeting (Gotta et al. 1998). Given its high homology to the N‑terminus of Orc1, both proteins might be recruited to the rDNA locus by the same mechanism, and a role of the already hypothesized common interaction partner in this process is conceivable. Alternatively, ORC can bind to the rDNA array via the ACS sites that are part of the NTS2 of each repeat (Huang and Moazed 2003). In order to verify a role of N^α‑acetylated Orc1 and Sir3 in rDNA silencing, the unacetylated mutants need to be tested for a rDNA phenotype. An interesting observation with regard to the mechanism of NatA’s role in rDNA silencing was that GFP-tagged Sir2 was still located in the nucleolus of nat1Δ cells (data not shown). This indicated that the integrity of the silencing-mediating RENT complex was not dependent on NatA.

In this study, we provide evidence that N^α‑acetylation participates in the regulation of chromatin structure in yeast, since the silencing proteins Orc1 and Sir3 depended on this modification to function properly. Thus, we propose that N^α‑acetylation can be classified as a chromatin regulatory mechanism comparable to acetylation or methylation of ε-N‑lysines.

However, in contrast to ε-N-acetylation, which readily can be removed by deacetylases (Dutnall and Pillus 2001), N^α‑acetylation is irreversible, and N^α‑deacetylases are hitherto unknown. This raises the question of how the modification can be removed in order to alter protein function upon demand. One possibility is that amino‑terminal proteolysis may remove N^α‑acetylation, as is proposed for histone methylation, which also is irreversible (Jenuwein and Allis 2001). Regulated ubiquitin‑based protein processing (Palombella et al. 1994) is a conceivable mechanism for the purpose, since the mutation of a putative ubiquitin‑specific protease was demonstrated to specifically enhance PEV in Drosophila (Henchoz et al. 1996). Another possibility is the removal of the N^α‑acetylated amino acid. Interestingly, acylamino acid‑releasing enzymes (AARE) have been identified in eukaryotes and an archeon (Ishikawa et al. 1998; Yamauchi et al. 2003), which catalyze the amino‑terminal hydrolysis of N^α‑acylpeptides to release N^α‑acetylated amino acids. Although AAREs act specifically on short nascent chains of 2-5 amino acids (Krishna and Wold 1992), related enzymes might perform this reaction on N^α‑acetylated proteins. A further alternative to modulate the function of N^α‑acetylated silencing proteins may be their removal from chromatin, much like preexisting histone modifications are eliminated by histone replacement during replication. Interestingly, methylated H3 histones are postulated to be exchanged during transcription for the unmethylated histone variant H3.3, which promotes the generation of active chromatin (Jenuwein and Allis 2001; Ahmad and Henikoff 2002).

We have not investigated whether any of these several theoretical mechanisms regulate the silencing activity of Orc1 or Sir3. In light of our data, we suggest that N^α‑acetylation may provide a stable, long-term epigenetic mark for maintaining chromatin states.

Intriguingly, N^α‑acetylation can be found in all kingdoms of life and especially in eukaryotes. The high level of evolutionary conservation suggests that this protein modification is capable of acting in fundamental cellular processes. In higher organisms, homologs of the NatA subunits have been linked to developmental and differentiation processes. Mouse mNAT1 is expressed in the developing brain and is regulated by physiological levels of functional N‑methyl‑D‑aspartate (NMDA) receptor in developing neurons (Sugiura et al. 2001). The mNAT1 homolog tubedown‑1 is expressed highly in developing tissues and down‑regulated upon differentiation (Gendron et al. 2000). Furthermore, a tubedown‑1 variant, Tbdn100, was isolated in a transcription regulatory complex, suggesting that it may be a transcriptional co‑regulator (Willis et al. 2002). Interestingly, the human homolog NATH also shows high expression in parts of the human brain and is overexpressed in malignant cells, for instance in papillary thyroid carcinomas and several leukemia and carcinoma cell lines (Fluge et al. 2002). It therefore has been hypothesized that NatA overexpression might simply correlate with high transcriptional activity. Even though this may be the case, overdosed N^α‑acetylation itself could result in deregulated gene expression and tumorgenesis, as overexpression of NAT1 impaired the stability of chromosomes in yeast (Ouspenski et al. 1999). In light of our findings, it is tempting to speculate that NatA acetylation regulates cell proliferation by modifying ORC function in replication or in the control of gene expression. Thus, it will be interesting to identify chromatin factors in higher eukaryotes whose function depend on N^α‑acetylation by NatA.

The implication of NatA homologs in cellular differentiation processes provokes the important question whether N^α‑acetylation plays a role in carcinogenesis. Interestingly, inappropriate regulation of chromatin structure (Singh et al. 2000) and notably mutated histone acetylating and deacetylating enzymes (Borrow et al. 1996; Vaziri et al. 2001) have been revealed to be tumor generating factors. In this respect, to enlighten the work of human N^α‑acetylating complexes is a challenging future task, which can be facilitated by a comprehensive knowledge of the mechanisms in yeast.