Budding yeast, like other eukaryotes, carries its genetic information on chromosomes that are sequestered from other cellular constituents by a double membrane, which forms the nucleus. An elaborate molecular machinery forms large pores that span the double membrane and regulate the traffic of macromolecules into and out of the nucleus. In multicellular eukaryotes, an intermediate filament meshwork formed of lamin proteins bridges from pore to pore and helps the nucleus reform after mitosis. Yeast, however, lacks lamins, and the nuclear envelope is not disrupted during yeast mitosis. The mitotic spindle nucleates from the nucleoplasmic face of the spindle pole body, which is embedded in the nuclear envelope. Surprisingly, the kinetochores remain attached to short microtubules throughout interphase, influencing the position of centromeres in the interphase nucleus, and telomeres are found clustered in foci at the nuclear periphery. In addition to this chromosomal organization, the yeast nucleus is functionally compartmentalized to allow efficient gene expression, repression, RNA processing, genomic replication, and repair. The formation of functional subcompartments is achieved in the nucleus without intranuclear membranes and depends instead on sequence elements, protein–protein interactions, specific anchorage sites at the nuclear envelope or at pores, and long-range contacts between specific chromosomal loci, such as telomeres. Here we review the spatial organization of the budding yeast nucleus, the proteins involved in forming nuclear subcompartments, and evidence suggesting that the spatial organization of the nucleus is important for nuclear function.
THE cell nucleus not only harbors and expresses an organism’s essential genetic blueprint, but also ensures the proper expression, duplication, repair, and segregation of chromosomes while ensuring proper processing and export of messenger and ribosomal RNA (Spector 2003; Taddei et al. 2004b). The dense packing of highly charged molecules (DNA, RNA, histones, nonhistone proteins) in a limited nuclear space was once thought to constrain molecular dynamics, yet we now know that the nucleus is neither grid-locked nor a random jumble (Rouquette et al. 2010). Large rings of chromatin diffuse freely through the nuclear volume in a random diffusive walk (Gartenberg et al. 2004; Neumann et al. 2012), yet the nucleus can maintain functional subcompartments enriched for specific enzymes and chromatin states. Understanding this dichotomy is key to understanding how nuclear organization facilitates nuclear function (reviewed in Taddei et al. 2004b; Mekhail and Moazed 2010; Egecioglu and Brickner 2011; Rajapakse and Groudine 2011; Zimmer and Fabre 2011).
Chromosomes, and the nucleosomal fibers within them, can be thought of as basic structural elements of the nucleus. Long-range chromosome folding is constrained by the physics of polymer dynamics (Klenin et al. 1998; Dekker et al. 2002; Gehlen et al. 2006; Neumann et al. 2012), and the characteristics of the chromosome polymers themselves depend on the folding of the nucleosomal fiber (Rosa and Everaers 2008). Yet chromatin in interphase nuclei is not regularly compacted and is subject to reversible covalent modifications on both DNA and histones within the nucleosomal fiber. Therefore, crucial biophysical properties of long-range chromatin dynamics, such as persistence length, mass density, and diffusion rate, are variable and subject to changes induced by nucleosome remodelers and histone modifiers. Thus, the post-translational modification of histones contributes not only to local chromatin folding, but to the three-dimensional organization of the genome (van Steensel 2011).
A second major factor contributing to nuclear organization is the interaction between chromatin and stable structural elements of the nucleus. In budding yeast, the key structural elements are the nuclear envelope (NE), the nuclear pore complex (NPC), and the nucleolus. The NE encompasses different types of chromatin anchorage sites, including the spindle pole body (SPB), and unique protein components of the inner nuclear membrane that tether heterochromatin, the ribosomal DNA (rDNA), or different types of DNA damage (Akhtar and Gasser 2007; Mekhail and Moazed 2010). The NPC also plays a role in the transient anchoring of activated genes or of DNA damage that cannot be readily repaired by homologous recombination. Finally, long-range interaction of loci in trans, such as the clustering of telomeres or of transfer RNA (tRNA) genes, influences nuclear order. The combination of physical constraints on chromatin movement and protein–protein interactions helps generate nuclear subcompartments that are enriched for specific DNA sequences, factors, and enzymatic activities (Gasser et al. 2004; Rosa and Everaers 2008). How these subcompartments affect nuclear function remains a central topic of research.
The basic principles of nuclear organization can be observed in all eukaryotes from yeast to humans. This allows us to test the functional implications of nuclear organization in a single-celled organism, despite there being species- and tissue-specific nuclear features. With facile genetics, live microscopy, and genome-wide mapping approaches, budding yeast has proven to be extremely useful for testing the functional roles of nuclear structure, as reviewed below.
Features of the Yeast Nucleus
Unique and conserved characteristics
Yeast is, in many ways, a typical eukaryote, yet it has a few unique and defining features that deserve mention. As described above, the SPB is embedded in the nuclear envelope and nucleates both intranuclear microtubules in interphase and the mitotic spindle. The SPB has a position strictly determined by the site of new bud emergence (Lee et al. 1999), which itself is opposite the nucleolus (Yang et al. 1989; Bystricky et al. 2004, 2005). The position of bud emergence is determined by landmark proteins at the cell cortex and Cdc42, which are positioned through association with the previous bud site. These proteins nucleate actin filaments and determine the orientation of cytoplasmic actin. The nucleus, in turn, is oriented by the cooperative action of SPB-microtubule-Kar9-Myo2 and actin interactions in a manner dependent on the actin filament orientation (Gundersen and Bretscher 2003; Slaughter et al. 2009). In other organisms, the link between the cytoskeleton and the nucleus stems from cytoskeletal interactions with Nesprins, KASH-domain and SUN-domain proteins that span from the outer through the inner nuclear membrane (INM) (Fridkin et al. 2009; Razafsky and Hodzic 2009). There is one SUN-domain protein in yeast called monopolar spindle protein 3 (Mps3), which is an important anchor of the INM, but whose link to the cytoskeleton is not yet known.
As mentioned above, the centromeres of yeast chromosomes remain attached to the SPB by short interphase microtubules, and this interaction strongly orients chromosomes within the interphase nucleus by keeping centromeres clustered (Guacci et al. 1997; Jin et al. 1998; Heun et al. 2001b; Bystricky et al. 2004). Opposite this landmark is the nucleolus, which is generated around a single rDNA locus on chromosome XII (Chr XII) that contains ∼200 tandem copies of a 9.1-kb repeat. The positioning of this left arm of Chr XII, and the nucleolus that forms around it, strongly influences nuclear order. A subset of rDNA repeats are tethered to the NE to suppress recombination (Mekhail et al. 2008), keeping the nucleolus tightly associated with the nuclear periphery and lending a striking polarity to the yeast nucleus by restricting diffusion of this large chromosome (Figure 1, B and D).
Apart from the rDNA repeat unit, budding yeast chromosomes have little repetitive DNA, and—most notably—no simple satellite repeat DNA at centromeres. This eliminates centric heterochromatin, a major chromosomal structural feature that in other organisms impacts neighboring sequences both in cis and in trans (reviewed in Akhtar and Gasser 2007). The organizational role of the transcriptionally inert, compacted chromatin of mammalian centromeres is fulfilled, at least in part, by the TG-repeat DNA found at yeast telomeres. These TG repeats generate repressive subtelomeric chromatin domains, which spread for several kilobases from the chromosomal ends, silencing nearby promoters (Gottschling et al. 1990). This phenomenon is called telomere position effect (TPE), in analogy to the position effect variegation (PEV) that spreads from satellite-containing centromeres in other species. The repressive chromatin formed at telomeres requires the binding of the silent information regulatory (SIR) proteins to nucleosomes, which also occurs at the two silent mating-type loci, HML and HMR (reviewed in Rusche et al. 2003). Interestingly, native subtelomeric genes are transcriptionally inert under standard growth conditions independent of SIR factor binding, as they carry genes that are expressed only under restrictive nutrient conditions (Fabre et al. 2005).
Budding yeast lacks the histone H3 K9 methylation that typifies centromeric heterochromatin in most other eukaryotes, as well as the major protein ligand that recognizes this modification (heterochromatin protein 1, or HP1) and the RNA interference machinery that facilitates PEV in fission yeast (reviewed in Buhler and Gasser 2009). Instead, a trimeric complex of Sir2, Sir3, and Sir4 recognizes unmodified nucleosomes to repress transcription and reduce endonuclease accessibility. Importantly, like PEV, yeast silent chromatin can be propagated through mitosis in a heritable manner thanks to its continual nucleation by silencer elements or telomeric repeats (reviewed in Rusche et al. 2003). As in higher eukaryotes, this heterochromatic state is late replicating (Raghuraman et al. 2001) and is found adjacent to the NE, sequestered away from nuclear pores (Palladino et al. 1993; Taddei et al. 2004a) (Figure 1B).
In addition to lacking repressive methylation marks, the yeast genome is also unique in that it lacks canonical linker histones (e.g., H1 and H5) and has a shorter nucleosomal repeat length (165 bp rather than 200 bp; see Woodcock et al. 2006). While the yeast genome encodes an H1-related protein, Hho1, this protein binds nucleosomal linker DNA only in rare instances and is not a core component of yeast chromatin (Freidkin and Katcoff 2001). Yeast also lacks the histone H3 subvariant H3.1 and macroH2A, while yeast histone H3 is equivalent to vertebrate H3.3, and the yeast H2A serves as the damage-associated, phospho-accepting variant H2AX of other species (Kusch and Workman 2007). Finally, like most organisms with a closed mitosis, yeast lacks the nuclear intermediate filament protein lamin. Nonetheless, yeast expresses other structural proteins of the INM orthologous to Man1 and emerin (Mekhail and Moazed 2010), which are lamin-associated proteins that contribute to chromatin anchoring in higher eukaryotes (Fridkin et al. 2009). Importantly, yeast has allowed one to test the functional impact of chromatin sequestration by the NE by mutating these structural proteins and monitoring effects on transcription and genome stability.
Nuclear envelope and nuclear pore complex
A double membrane contiguous with the endoplasmic reticulum separates chromatin from the cytoplasm. Embedded in the inner membrane of this nuclear envelope one finds different structures and proteins that anchor chromosomes, including the SPB and nuclear pores (reviewed in Dieppois and Stutz 2010). Trafficking between the nucleoplasm and the cytoplasm occurs through ∼200 NPCs, which enable the free diffusion of small molecules as well as the regulated transport of macromolecules by the importin machinery (Alber et al. 2007; D’Angelo and Hetzer 2008; Aitchison and Rout 2012). Intriguingly, NPCs provide a platform for messenger RNA (mRNA) transcription and quality control, as well as its export, and tether a subpopulation of inducible genes after activation, both through the mRNA and a quality-control export complex called Tho-Trex (Dieppois and Stutz 2010).
NPCs are massive assemblies of ∼50 MDa containing 456 nucleoporins of 30 different types (D’Angelo and Hetzer 2008). They form a doughnut-shaped structure with an eightfold symmetry around a central channel, with flexible protein filaments emanating from the core into both the cytoplasm and the nucleoplasm. These provide binding sites for the transport of proteins, mRNA, and chromatin. A detailed map for the relative position of each nucleoporin was calculated on the basis of multiple molecular, biochemical, and structural data revealing a strongly modular structure (Alber et al. 2007).
Several conserved proteins, which in other species associate with lamins, are found at the NE in yeast (Huh et al. 2003). Of particular interest are the integral proteins of the INM (Lusk et al. 2007), including Doa10, a RING domain containing proteins that targets nuclear proteins for degradation; Mps3, a member of the SUN (Sad1, UNC-84) family that is a shared component of the INM and the SPB (Jaspersen et al. 2002); and helix–extension–helix-1 and -2 (Heh1 and Heh2) (M. C. King et al. 2006; Fridkin et al. 2009), which are orthologs of the mammalian lamin-associated protein MAN1 (Figure 2). While these proteins are thought to bridge from the INM or lamina to chromatin in multicellular organisms, in most cases the ligand on the chromatin side is still unknown (Fridkin et al. 2009; Razafsky and Hodzic 2009). The yeast SUN-domain protein Mps3 and the two MAN1 homologs (Heh1 or Src1 and Heh2) serve a similar function in yeast, despite the absence of lamins (Grund et al. 2008; Mekhail and Moazed 2010). In addition to these conserved components, a yeast-specific protein called Esc1 (establishes silent chromatin) (Andrulis et al. 2002) associates tightly with the inner face of the INM where it anchors silent chromatin through its high affinity for Sir4 (Andrulis et al. 2002; Gartenberg et al. 2004; Taddei et al. 2004a). Esc1 may have other functions, as its overexpression induces INM expansion (Hattier et al. 2007). Not surprisingly, functional cross talk exists between INM proteins (e.g., Esc1 and Mps3) and the nuclear pore basket proteins (e.g., Nup60, Mlp1, and Mlp2) (Therizols et al. 2006; Lewis et al. 2007; Palancade et al. 2007) in that loss of either a INM or a pore protein can interfere with the function of the other, even though they define spatially distinct domains of the NE when localized at high resolution (Taddei et al. 2004a; Horigome et al. 2011).
Long-range chromosome organization
Budding yeast chromosomes assume a Rabl-like conformation throughout the vegetative cell cycle (Figure 1F). The Rabl orientation reflects the spatial orientation of anaphase chromosomes, which means that yeast chromosome arms extend away from the centromeres that are held by the SPB (Rabl 1885; Yang et al. 1989; Dekker et al. 2002; Jin et al. 1998, 2000). Telomeres are generally found in clusters around the nuclear periphery (Figure 1B) (Palladino et al. 1993). This organization was demonstrated using fluorescence in situ hybridization (FISH), as well as by immunofluorescence (IF) for centromeric and telomeric proteins (Gotta et al. 1996; Jin et al. 1998). Later these perinuclear telomere foci were confirmed and tracked through the cell cycle using live imaging of GFP-tagged telomeres (Schober et al. 2008; Therizols et al. 2010). Whereas a polarized chromosomal organization exists transiently after telophase in most metazoan cells, it persists through interphase in budding yeast due to the persistent attachment of centromeres that remain linked by short intranuclear microtubules to the SPB (Guacci et al. 1997; Jin et al. 1998; Heun et al. 2001b; Bystricky et al. 2004). Treatment of interphase cells with nocodazole allows centromeres to move away from the SPB (Jin et al. 1998; Heun et al. 2001b; Bystricky et al. 2004).
In 2002, the Kleckner laboratory developed a molecular approach to monitor and model chromosome conformation [chromosome conformation capture (3C)] (Dekker et al. 2002) based on the detection of long-range interactions within and between chromosomes. A population-averaged three-dimensional model of Chr III was thus determined. In this model, Chr III appears to fold as a contorted ring, with a strong bend near the centromere and the telomeres in close proximity to each other (Dekker et al. 2002). This was confirmed by fluorescent tagging of right and left telomeres of Chr III and for a second short, metacentric chromosome (Chr VI) (Bystricky et al. 2005). Later, it was shown that telomere-mediated chromosome looping does not occur in chromosomes that have arms of unequal lengths (Schober et al. 2008; Therizols et al. 2010). The impact of chromosome arm length on selective telomere–telomere interaction was demonstrated in an elegant experiment in which long and short arms of different chromosomes were swapped. This converted a chromosome with unequal arm lengths to one with equal arm lengths, which enhanced intrachromosomal telomere–telomere interaction (Schober et al. 2008).
Genome-wide conformation capture approaches (Hi-C), inspired from the 3C method of Dekker et al. (2002), confirmed the clustering of centromeres in interphase cells and the widespread associations between pairs of telomeres on different chromosomes that had been observed by fluorescence microscopy (Duan et al. 2010). The Hi-C analysis also confirmed that two unlinked telomeres positioned at similar distances from their corresponding centromeres are more likely to interact than telomeres on arms of different lengths (Duan et al. 2010; Therizols et al. 2010), again illustrating the impact of the Rabl organization on long-range interactions (Figure 1). Interchromosomal contacts were also detected among clustered tRNA genes, early origins of DNA replication, and at sites of chromosomal breakage (Duan et al. 2010).
Importantly, interactions within a chromosome (such as those formed by right and left telomeres) occur far more frequently than interactions between different chromosomes (Rodley et al. 2009; Duan et al. 2010). This applies even to large chromosomes and chromosomes with unequal arms, arguing that a yeast chromosome defines a spatial unit or “territory.” Territory positioning is, of course, subject to architectural constraints imposed by centromere attachment to the SPB and telomere anchoring to the NE (Berger et al. 2008; Therizols et al. 2010), but territories could also be modeled in silico on the basis of polymer diffusion kinetics, without need for a proteinaceous scaffold or matrix (Rosa and Everaers 2008). In yeast, nonetheless, territories can be significantly remodeled by transcriptional activation (Berger et al. 2008), which is different from the situation in differentiated cells of multicellular organisms, where chromosome territories play a more dominant role in nuclear organization (Rouquette et al. 2010). This dominance can be explained both by the sheer size of mammalian chromosomes, each of which is larger than the entire yeast genome, and by the abundance of repetitive, noncoding DNA sequences in larger genomes.
Although chromosomes show a recognizable pattern of positioning, chromatin in all living cells is subjected to constant motion, which has been described as a constrained random walk (Marshall et al. 1997; Gasser 2002). Rapid time-lapse imaging led to the distinction of at least two types of motion in yeast: small random movements (<0.2 µm within 1.5 sec) that occur constantly, as well as larger steps over relatively short time intervals (i.e., >0.5 µm in a 10.5-sec interval) (Heun et al. 2001b). The movement of chromatin is ATP-dependent and varies between G1- and S-phase cells, yet can be quite accurately quantified by a mean squared displacement analysis as a constrained random walk (Figure 3). Not unexpectedly, different loci move within domains of different size within the nucleus, which are defined as radii of constraint (Rc) (Marshall et al. 1997; Heun et al. 2001b; Gasser 2002; Neumann et al. 2012). Indeed, telomeres, centromeres, and silent chromatin, which are tethered through protein–protein interactions at the NE, move within smaller radii of constraint than coding and noncoding regions along the longer chromosome arms (telomeric Rc < 0.4 vs. 0.6 μm for active loci) (Hediger et al. 2002; Gartenberg et al. 2004; Cabal et al. 2006; Dion et al. 2012; Neumann et al. 2012). Recent work shows that the targeting of ATP-dependent nucleosome remodelers leads to increased mobility, suggesting that the shifting or removal of nucleosomes changes the flexibility of the chromatin fiber and, in turn, its mobility (Gehlen et al. 2006; Neumann et al. 2012). This is consistent with the notion that a chromosome can be modeled as a flexible polymer chain subject to random forces and tethering effects (Klenin et al. 1998; Gehlen et al. 2006; Rosa and Everaers 2008; Neumann et al. 2012).
One of the least understood yet most pervasive features of nuclear organization is the existence of subnuclear compartments, in which specific DNA sequences and proteins accumulate. These are thought to create microenvironments that favor or impede a particular DNA- or RNA-based activity. The most obvious and well-characterized of such compartments is the nucleolus, the site of RNA polymerase I-mediated rDNA transcription and ribosome subunit assembly. We describe this in some detail as it illustrates the self-organizing character of such compartments.
In almost every eukaryotic nucleus, the most prominent subnuclear compartment is the nucleolus. In budding yeast, the nucleolus is a crescent-shaped structure occupying roughly one-third of the nuclear volume, abutting the NE and lying opposite the SPB (Yang et al. 1989; Bystricky et al. 2005). This compartment can be considered as a factory dedicated to ribosome biogenesis. Its morphology is strongly influenced by the cell growth rate, probably as a result of adapting the rate of ribosome production to the needs of the cell (Oakes et al. 1993; Powers and Walter 1999).
The budding yeast rRNA stems from a 9.1-kb repeat locus that is transcribed as a 35S precursor rRNA and a 5S rRNA by RNA polymerases I and III, respectively. Intriguingly, different yeast strains have different numbers of tandem repeats, extending from 100 to 200 (Pasero and Marilley 1993). Assembly of the nucleolus is thought to be a self-driven process, initiated by production of rRNA (Trumtel et al. 2000; Hernandez-Verdun et al. 2002). This argument stems in part from studies in which the rDNA repeat was transcribed by RNA Pol II, rather than by the endogenous RNA Pol I. This led to massive alterations in nucleolar structure, arguing that factors associated with RNA Pol I play a key role in proper nucleolar assembly (Oakes et al. 1993; Trumtel et al. 2000).
A recent study (Albert et al. 2011) showed that two Pol I-specific subunits, Rpa34 and Rpa49, are essential for the high polymerase-loading rate on active copies of rDNA and for nucleolar assembly. Intriguingly, nucleolar assembly can be restored in the absence of Rpa49 by decreasing the ribosomal gene copy number from 190 to 25, with a concomitant increase in the number of RNA Pol I complexes per transcribed unit. These authors argue that the spatial constraint of RNA Pol I transcription is a critical feature of nucleolar assembly (Albert et al. 2011). Exactly how the 2 Mb of rDNA is folded within the nucleolus is not known, although both cohesin, condensin, and Sir2 affect rDNA compaction during the mitotic cell cycle (Guacci et al. 1994; Lavoie et al. 2002, 2004).
The extended array of tandem repeats found in the rDNA locus serves as an ideal template for homologous recombination (HR), the favored pathway for double-strand break (DSB) repair in budding yeast. At the same time, it is clear that rDNA stability is critical for growth and survival of yeast, since the generation of extrachromosomal rDNA circles by inappropriate recombination provokes replicative senescence (Sinclair and Guarente 1997). Thus, budding yeast has evolved several mechanisms to suppress recombination within the rDNA array. Two mechanisms involve the silent information regulator, Sir2, which is a NAD-dependent deacetylase (Imai et al. 2000; Smith et al. 2000; Tanner et al. 2000). The first mechanism correlates with a Sir2-dependent local nucleosomal organization (Gottlieb and Esposito 1989; Bryk et al. 1997; Fritze et al. 1997; Smith and Boeke 1997) and the second involves longer-range chromatin tethering to the nuclear envelope through the CLIP (chromosome linkage INM proteins) complex. This complex includes two NE proteins, Heh1 (homologous to the human Man1 protein) and Nur1 (Figure 2). The rDNA is connected to the CLIP complex through cohibin, a V-shaped complex of two Lrs4 proteins and two Csm1 homodimers (Mekhail et al. 2008; Chan et al. 2011). Loss of either Sir2 or the cohibin-Heh1-anchoring pathway leads to instability of the rDNA repeat, followed by cell cycle arrest or premature senescence. Premature senescence can also be provoked by the loss of Sgs1 (Sinclair and Guarente 1997), a RecQ helicase that counteracts recombination between rDNA repeats.
Telomere foci—assemblies of repetitive DNA and silencing factors:
The clustering of the 32 yeast telomeres into three to six foci at the NE provides a second prominent feature of yeast nuclear organization (Palladino et al. 1993). At these clusters of telomeres, the silent regulatory factors Sir3 and Sir4 and the telomere repeat-binding factor repressor activator protein 1 (Rap1) accumulate (Gotta et al. 1996), while Sir2 is found both at telomeric foci and in the nucleolus (Gotta et al. 1997). These perinuclear telomeres are late replicating (Raghuraman et al. 2001) and generate a zone that favors SIR-mediated repression of silencer-flanked genes (Andrulis et al. 1998). The clustering of telomere repeats also ensures that SIR proteins do not bind promiscuously to repress other sites in the genome (Maillet et al. 1996; Taddei et al. 2009). Finally, telomere anchorage in S phase contributes to proper telomerase control and suppresses recombination among telomere repeats (Schober et al. 2009; Ferreira et al. 2011).
Intriguingly, telomere-containing foci are dynamic, moving with a constant random motion that is more constrained than that of a nontelomeric locus (Schober et al. 2008; Therizols et al. 2010). These foci fuse and divide in interphase and dissociate at least partially in metaphase (Laroche et al. 2000; Smith et al. 2003) to reform in early G1. With the exception of the intrachromosomal loops formed by short metacentric chromosomes, no strong preferences for telomere–telomere pairing were detected in interphase cells (Schober et al. 2008). Rather, telomeres on ends of chromosome arms of roughly equal length tend to interact, presumably due to the geometry imposed by the Rabl conformation in anaphase (Jin et al. 2000; Schober et al. 2008; Therizols et al. 2010), showing how that the linear architecture of a chromosome can impact its long-range contacts in the nucleus.
At the molecular level, budding yeast telomeres consist of 250–300 bp of irregular tandem repeats with the consensus sequence TG1–3 (Shampay et al. 1984). Rap1 (Rap1) (Shore and Nasmyth 1987) binds this repeat on average once every 18 bp (Gilson et al. 1993). The Rap1 C terminus carries a binding site for the silencing factors Sir3 and Sir4 (Moretti et al. 1994; Marcand et al. 1997; Wotton and Shore 1997), as well as for the telomerase-repressing factors Rif1 and Rif2 (Hardy et al. 1992; Wotton and Shore 1997). Rif1/Rif2 binding and Sir3/Sir4 binding are mutually antagonistic, and thus loss of the Rif proteins leads to enhanced subtelomeric repression (Moretti et al. 1994; Mishra and Shore 1999), and loss of Sir4 leads to a reduction in telomere length (Palladino et al. 1993), most likely due to increased Rif1/2 binding. Together with the end-binding complex Ku (Yku), telomeric Rap1 serves to recruit Sir3 and Sir4 to telomeres, and the large number of Rap1 sites generated by telomere clustering is thought to compete for limited amounts of SIR proteins. In this way, multiple weak, but closely juxtaposed, binding sites create an effective sink for this histone-binding complex (Gasser et al. 2004). Confirming this, the dispersion of SIR proteins from telomeres by the mutation of telomere anchorage sites reproducibly reduced the expression of ∼15 nontelomeric genes (Taddei et al. 2009).
SIR-mediated silencing is generated in a two-step process. First, the heterodimer formed between Sir4 and Sir2 catalyzes the NAD-dependent deacetylation of lysines in the histone H4 N-terminal tail. The deacetylation of histone H4 K16, in particular, generates a preferred binding site for Sir3 (Oppikofer et al. 2011), allowing Sir3 to bind and spread to the adjacent unacetylated nucleosomes (reviewed by Norris and Boeke 2010; Rusche et al. 2003). Sir3 appears to bind nucleosomal arrays in a stable, stoichiometric complex with Sir2 and Sir4 (Cubizolles et al. 2006), which renders linker DNA less accessible to transcription factors or other mediators of RNA pol II engagement at the transcriptional start site (Aparicio et al. 1991; Strahl-Bolsinger et al. 1997; Hecht and Grunstein 1999; Chen and Widom 2005; Martino et al. 2009).
As mentioned above, by disrupting telomere anchors but leaving SIR factors intact, it was shown that the sequestration of SIR proteins in telomere foci favors subtelomeric repression and prevents the promiscuous binding of SIR proteins at a distinct subset of promoters elsewhere in the genome. This mechanism may be usurped and regulated by environmental conditions to control physiological response. For example, the subtelomeric positioning of genes may help coordinate both the repression and the activation programs that permit growth under adverse conditions (Turakainen et al. 1993a,b; Halme et al. 2004; Fabre et al. 2005). Over time this could ensure an evolutionary advantage, which would explain why alternative carbon source genes accumulate in subtelomeric domains. As discussed below, the impaired tethering of telomeres also correlates with altered rates of recombinational repair in subtelomeric zones (Louis 1994; Therizols et al. 2006; Schober et al. 2009) and impaired control of telomerase (Hediger et al. 2006; Ferreira et al. 2011). We discuss how telomere clustering affects gene expression and genome stability in more detail below.
Although the 274 tRNA genes are found distributed along all 16 yeast chromosomes, many of these genes, when probed by FISH or Hi-C technologies, appear to be clustered in nuclear space (Thompson et al. 2003; Wang et al. 2005; Duan et al. 2010). Intriguingly, at least some tRNA clusters are found close to the nucleolus (Thompson et al. 2003; Wang et al. 2005), possibly reflecting the fact that tRNA genes and the 5S rDNA are coordinately transcribed by RNA pol III. This juxtaposition may favor coordinated activation of RNA polymerase I and polymerase III for expression of the protein synthesis machinery. Interestingly, the clustering of tRNA genes depends on condensin, while their association with the nucleolus depends on microtubule integrity (Haeusler et al. 2008). It is clear from Hi-C data and GFP-LacO tagging that not all RNA polymerase III genes show similar localization, and it is not yet clear why some tRNA genes shift toward the nucleolus, while others do not (E. Varela, personal communication; Duan et al. 2010). Nonetheless, the broad distribution of tRNA genes means that the association of even a subset of them with the nucleolus would impact global chromosome positioning.
It is noteworthy that RNA polymerase II genes found in the vicinity of actively transcribed tRNA genes become silenced through a phenomenon called tRNA-gene-mediated gene silencing (Wang et al. 2005). Although condensin mutations that affect tRNA gene clustering do not affect tRNA transcription, they do abolish the repression of nearby RNA pol II promoters, suggesting that tRNA clustering and tRNA-gene-mediated gene silencing are linked (Haeusler et al. 2008). Thus, tRNA genes can be transcribed away from the nucleolus, but interactions in trans and the long-range effect of this on chromatin status correlate with relocation.
Eukaryotic genomes initiate DNA replication at multiple sites or origins that are dispersed along each linear chromosomal arm. In budding yeast, origin function is coincident with short cis-acting sequences called autonomously replicating sequences (ARS), which support both autonomous plasmid replication and genomic initiation events. In budding yeast, DNA synthesis initiates uniquely at these sites, yet the ARS sequences are not sufficient to ensure efficient firing. Indeed, sites that are competent for initiation on plasmids do not necessary fire in the genome, and even the most efficient origins do not fire every cell cycle (Raghuraman et al. 2001).
The clustering of yeast replication forks in foci was first revealed in an in vitro replication assay that used isolated yeast nuclei as a template for the incorporation of fluorescently labeled nucleotide. Between 15 and 20 discrete foci were detected that were Origin Recognition Complex (ORC), S phase-, and origin-specific (Pasero et al. 1997). The same type of foci were later observed by visualizing replisome components such as PCNA and DNA polymerase α or ε by immunostaining or fluorescent tagging in living cells (Ohya et al. 2002; Hiraga et al. 2005; Kitamura et al. 2006). Because the number of these foci is far smaller than the number of replication forks, they are thought to correspond to replication factories in which several elongating forks are grouped. Consistent with this notion, neighboring origins along a chromosome initiate replication with similar timing and respond in a similar fashion to the loss of an S-phase cyclin (Yabuki et al. 2002; McCune et al. 2008).
In the highest resolution study of DNA replication to date, Tanaka and coworkers show that sister replication forks generated from the same origin of replication stay together within one focus, while the sequences on either side appear to be pulled through that same “replication factory” (Kitamura et al. 2006). Since replicons on different chromosomes might be coregulated by juxtaposition in a single focus, it is interesting to note that Hi-C data do score increased interaction between early firing origins (Duan et al. 2010). Such clustering could facilitate initiation by concentrating limiting factors (Mantiero et al. 2011) to favor firing in early S phase.
Although late-replicating regions are found mainly at the nuclear periphery, a subnuclear position is neither necessary nor sufficient for the control of late origin firing (Heun et al. 2001a; Ebrahimi et al. 2010). The timing of telomere replication is instead determined by telomere length (Bianchi and Shore 2007; Lian et al. 2011). The clustering of replication forks could nonetheless increase replication efficiency by concentrating replisome components and deoxy-nucleotides to ensure efficient elongation after recovery from replicative stress. There has been no genetic test of this model to date, since the proteins that ensure origin clustering are unknown. Indeed, the only protein known to influence replication focus formation in vertebrates is lamin (Spann et al. 1997), which yeast lack.
Sites of DNA repair:
DNA DSB repair by homologous recombination is also organized into foci called DNA damage response foci (Lisby and Rothstein 2009). These foci contain high local concentrations of repair factors and are called repair foci or factories. The spatially restricted structures may serve to restrict the modification of chromatin to the subdomain that is relevant for repair or to restrict end-joining to the proper chromosome end. Other models argue that different repair foci may prevent telomere addition at certain DSBs by sequestering the break away from telomerase. Alternatively, a focus found near other telomeres may facilitate TG addition at breaks. Intriguingly, multiple DSBs incurred on different chromosomes were shown to assemble into a single repair focus in yeast (Lisby et al. 2003). This topic will be reviewed elsewhere; thus, only damage mobility and its function are discussed here.
Mechanisms Underlying Nuclear Compartmentation
Redundancy in telomere-anchoring pathways
As described above, the positioning of chromatin within the nucleus depends on reversible interactions of chromosomal elements with structural components of the nuclear envelope, namely proteins of the INM, the SPB, and, in some cases, nuclear pores. The anchoring of telomeres is achieved through several different protein–protein contacts, which fluctuate during the cell cycle and which have different levels of importance for different telomeres (Figure 4) (Hediger et al. 2002; Taddei et al. 2004a; Hiraga et al. 2008; Schober et al. 2009). The fluctuation can be attributed in part to post-translational modifications; several pathways show a requirement for sumoylation by the E3 ligase Siz2 (Ferreira et al. 2011)
One major pathway of telomere anchoring requires Sir4 and is therefore amplified by the formation of silent chromatin (Hediger et al. 2002; Gartenberg et al. 2004; Taddei et al. 2004a). This was most clearly shown by tracking the position and mobility of an excised ring of yeast chromatin, which either did or did not contain silencers. The chromatin ring could bind autonomously to the NE when silenced, in a SIR-dependent manner, and did not associate when SIR protein could not bind (Gartenberg et al. 2004). Sir4 anchors repressed chromatin to the NE through a 312-aa domain in its C-terminal half that specifically binds an acidic protein associated with NE, Esc1 (Andrulis et al. 2002; Taddei et al. 2004a). In parallel, it was shown that Sir4 can mediate anchoring through the nucleoplasmic N-terminal acidic domain of Mps3 (Bupp et al. 2007). This may not be a direct interaction, however, and a recent study suggests that Sir4/Mps3 association is mediated by Lrs4, a component of the cohibin complex (Chan et al. 2011). Moreover, contrary to the Esc1–Sir4 interaction, Mps3-Sir4 anchoring requires an intact SIR complex and is completely dependent on silencing. Both Mps3 and Esc1 define NE-binding sites that can be distinguished by microscopy from nuclear pores, and their position around the NE is independent of nuclear pore distribution (Taddei et al. 2004a; Horigome et al. 2011).
Mps3 not only participates in SIR-dependent anchorage, but also mediates a silencing-independent pathway for telomere anchoring through its interaction with Est1 (Ever shorter telomeres) (Lundblad and Blackburn 1990), an accessory component of telomerase (Antoniacci et al. 2007; Schober et al. 2009). Tlc1, the RNA moiety of telomerase, is linked to telomeric chromatin through the yeast Ku70/Ku80 heterodimer, which also binds the end of chromosomes (Martin et al. 1999). It is thought to help recruit the telomerase either directly or indirectly (Peterson et al. 2001; Stellwagen et al. 2003; Fisher et al. 2004; Pfingsten et al. 2012) since the Ku-telomerase and Ku-DNA interaction may be mutually exclusive (Peterson et al. 2001; Stellwagen et al. 2003; Pfingsten et al. 2012). Whereas the intramembrane portions of Mps3 play a major role in SPB organization (Jaspersen et al. 2002; Nishikawa et al. 2003), its role in telomere anchoring requires its acidic N-terminal domain, which extends into the nucleoplasm (Bupp et al. 2007). The use of N-terminal deletions allows one to distinguish the roles of Mps3 in chromatin anchoring and in SPB formation and function. Nonetheless, the various telomere-anchoring pathways are extensively interlinked, since the yeast Ku dimer, which mediates Mps3 anchorage in S-phase cells, also directly binds Sir4 (Roy et al. 2004; Taddei et al. 2004a).
Both telomere and silent chromatin anchorage pathways show cell-cycle dependence and sensitivity to post-translational modification (Ebrahimi and Donaldson 2008; Ferreira et al. 2011). The Mps3-mediated anchoring pathways are S-phase-specific, while yeast Ku is able to anchor the telomere in an Mps3- and silencing-independent manner in G1-phase cells (Taddei et al. 2004a; Schober et al. 2009). Importantly, the S-phase-specific Yku80–Mps3 interaction was found to require the sumoylation of Yku80 by the E3 ligase Siz2 (Ferreira et al. 2011). This provides a plausible mechanism for switching off Yku-mediated peripheral positioning once the majority of the genome has been replicated, allowing telomeres to dislodge from the NE either during or just after replication (Ebrahimi and Donaldson 2008), possibly as a consequence of desumoylation. Sir4 is also phosphorylated in a cell-cycle-dependent manner (Kueng et al. 2012), which may contribute further to the release of telomeres observed during mitotic division.
Antagonism of the yeast Ku–telomerase–Mps3 interaction leads to a hyper-recombination phenotype among the shortened telomeres found in cells lacking Tel1, the ATM kinase homolog in yeast (Schober et al. 2009). This suggests that the sequestration of telomeres by the NE protein Mps3 could protect telomere ends from recombination and deleterious unequal strand exchange. Consistently, the release of telomeres by the deletion of SIZ2 in both G1- and S-phase cells correlates with abnormally long telomeres (Ferreira et al. 2011). Another telomere-bound protein, namely Cdc13, was also found to be sumoylated in a Siz2-dependent manner (Hang et al. 2011). Sumoylated Cdc13 contributes in an independent manner to telomerase activity, synergizing with the effects mediated by yeast Ku and Sir4. Thus, Siz2-mediated sumoylation limits telomere elongation by more than one mechanism, one of which correlates with telomere sequestration at the NE (Figures 4 and 5). Importantly, it was shown that critically short telomeres shift away from the nuclear periphery when they elongate in newly formed wild-type zygotes (Ferreira et al. 2011). If natural elongation coincides with telomere release, it follows that telomere sequestration by sumoylated forms of Sir4 and Yku might indeed attenuate telomerase activity.
Minimal anchoring assay and the validation of genetic screens
Given the redundancy in the pathways of telomere anchoring, it was difficult to dissect the requirements of anchoring by scoring the position of a LacO-tagged telomere in single mutants. To solve this problem, a “minimal anchoring assay” was developed in which proteins fused to LexA were targeted to an internal LacO-tagged locus (ARS607). The ability of the fusion protein to shift a randomly distributed locus to the periphery was scored and analyzed statistically (Taddei et al. 2004a). The assay is performed in a strain expressing GFP-Nup49 to allow an accurate quantitation of position in different genetic contexts. The minimal anchoring assay allows one to score for sufficiency and not only for dependence on a given factor, which is necessary if one wants to define the molecular machinery of chromatin anchoring.
In addition to these targeted approaches, a genetic screen for genes that influence telomere anchoring was carried out. This study identified the histone acetyl transferase Rtt109 and the acetylation of histone H3K56 as key modulators of telomere anchoring (Hiraga et al. 2008). This probably reflects impaired nucleosomal assembly after DNA replication. Other factors showing partial or cell-cycle-dependent loss of anchoring include Esc2, Ctf18, Tel1, Mre11, Elp4, and Rrm3 (Hiraga et al. 2008). Since these may affect telomere replication and the spread or assembly of silent chromatin, an effect on telomere tethering is not sufficient to consider them as direct anchors. Mutations in proteins of the nuclear pore such as Nup133 also affect telomere positioning, but these interfere with both mRNA export and cell cycle progression, suggesting that their effects are indirect (Doye et al. 1994; Therizols et al. 2006; Lewis et al. 2007; Palancade et al. 2007). Palmitoylation of telomere-bound proteins could also contribute to the anchoring of silent chromatin as recently shown for Rif1, although long telomeres in rif1Δ cells remain peripherally attached through Sir4 (Cockell et al. 1995; Park et al. 2011). Still, redundant roles of Rif1 and Rif2 in telomere anchoring remains are not excluded.
In addition to telomeres, persistent DNA damage, abruptly shortened telomeres, and a subset of highly transcribed genes are also found to shift to the nuclear periphery (reviewed in Oza and Peterson 2010; Nagai et al. 2011). In these cases, the NPC is clearly involved in the peripheral association, and for DSBs and short telomeres an additional pathway implicating Mps3 was described (Nagai et al. 2008; Kalocsay et al. 2009; Khadaroo et al. 2009; Oza et al. 2009). These anchoring pathways are less well understood than telomere-anchoring pathways, but have a clear impact on chromosome stability and are discussed in this context below.
Formation of chromatin compartments: association in trans
There are two basic models for the formation of nuclear compartments. The first reflects the localized tethering of similar sequences to a subnuclear structure, or a variant of this in which intermolecular interactions lead to domain clustering. In the second model, compartments arise passively due to volume exclusion effects. This is favored under conditions of macromolecular crowding, and the high concentration of macromolecules within the nucleoplasm (between 100 and 400 mg/ml) is thought to be sufficient to enhance attractive interactions between macromolecules (Richter et al. 2007). Most likely, subnuclear compartments arise from a combination of mechanisms. A few structural proteins appear to be involved in more than one mechanism, such as the cohesin, condensin, and the more recently characterized cohibin complexes (Poon and Mekhail 2011).
Cohesin, condensin, and cohibin are multi-subunit complexes that contain large coiled-coil-containing structural maintenance of chromosomes (SMC) proteins. These conserved SMC proteins form dimers through hinge domains that are found at the end of extended coiled-coil arms. They are able either to encompass nucleosomal fibers or to link chromatids to facilitate long-range contacts and chromosome condensation. The head domains of the SMC factors provide binding sites for non-SMC proteins, which are also important for the condensation and cohesion roles of these factors. Recently, yeast SMC complexes have been implicated in the clustering of tRNA genes, rDNA condensation, and rDNA stability (Freeman et al. 2000; Torres-Rosell et al. 2007; Tsang et al. 2007; Bermudez-Lopez et al. 2010). As mentioned above, cohibin connects the rDNA and silent chromatin to the NE by bridging between the chromatin-bound RENT/SIR complex and either CLIP or Mps3. The complex is proposed to link rDNA repeats to each other by clustering interspersed IGS1 regions (Mekhail et al. 2008; Chan et al. 2011), a mechanism thought to contribute to SIR-dependent telomere clustering as well.
Yeast telomere clusters and their associated concentration of silent information regulators provide a well-characterized example of a functional compartment generated by the juxtaposition of sequence-defined chromosomal elements. Interactions between subtelomeric zones were proposed to be governed by physical constraints, including chromosome arm length, centromere attachment to the SPB, and nuclear crowding (Therizols et al. 2010). On the other hand, interaction between HM loci was shown to depend on correctly assembled heterochromatin at these loci (Miele et al. 2009). Consistently, Sir3 has been shown to promote telomere–telomere association (Ruault et al. 2011). Sir3 overexpression triggered the clustering of telomeric foci into even larger aggregates positioned away from the NE, arguing that clustering can occur independently of anchoring.
Although this “hyper-clustering” phenotype correlates with more stable silencing in subtelomeric regions, silencing is not absolutely required for telomere cluster formation. Notably, the overexpression of a Sir3 mutant lacking silencing activity induces large telomere clusters, even in the absence of Sir2 and Sir4 (Ruault et al. 2011). The Sir3-mediated clustering is likely to be driven by Sir3 self-oligomerization, likely mediated by its C terminus (Georgel et al. 2001; D. A. King et al. 2006; Liaw and Lustig 2006; McBryant et al. 2008; Adkins et al. 2009).
Although silencing is not required for clustering, the involvement of Sir3 in both processes provides a mechanism for the self-propagation of a silent compartment. Specifically, given that Sir3 binding promotes telomere clustering, this leads inevitably to an increased local concentration of SIR factors. Given that Sir3 is limiting for the spread of silent chromatin, clustering would favor SIR complex spread into flanking chromatin, extending repression. With more Sir3 bound, more clustering would occur. Other factors also affect telomere clustering, namely the other SIR proteins, the yeast Ku heterodimer, Asf1, Rtt109, Esc2, the cohibin complex, and the two conserved factors Ebp2 and Rrs1 that moonlight in ribosome biogenesis (Gotta et al. 1996; Laroche et al. 2000; Gehlen et al. 2006; Hiraga et al. 2008; Miele et al. 2009; Chan et al. 2011; Horigome et al. 2011). However, because loss of these factors cause replicative stress and impair heterochromatin formation, a direct role in clustering is not established.
This raises a chicken-and-egg question: Do silent compartments persist simply because they are self-forming, or is there an initial triggering event that promotes the function that will be favored by the compartment? The bifunctional role of Sir4 as a mediator of repression and as a silencing-independent anchor for chromatin allows it to both participate in the self-organization of these compartments and trigger entry into the cycle of repression–sequestration–repression for a nonsilenced telomere. Importantly, chromatin immunoprecipitation showed that Sir4 can bind telomeres in the absence of Sir3, while the opposite was not true (Strahl-Bolsinger et al. 1997). Thus, a nonsilent telomere might be more likely to bind the NE through Esc1–Sir4 or telomere-bound Yku–Mps3 interactions than to cluster. This would shift a naive telomere into a zone enriched for telomere-bound Rap1 and its ligands Sir3 and Sir4. Proximity to pools of SIR proteins would enhance telomere repression, and, once silencing is established, perinuclear localization would be reinforced by the Sir4-anchoring pathway, while clustering would be promoted by Sir3. In this way, the formation of silent chromatin foci could be self-generating, and not only self-reinforcing. Such a mechanism may apply to other chromatin-based compartments with other functions and may also allow the nucleus to segregate factors away from sites at which they might do harm.
Mechanisms of chromatin movement
To generate compartments, chromatin must be able to move within the nucleus. An excised ring of chromatin explores the entire nuclear volume in random walk (Rc = 0.9 mm) (Figure 3) while a chromosomal locus moves in a constant, near-random walk within a radius of ∼0.5–0.7 μm (reviewed in Gasser 2002; Marshall 2002; Hubner and Spector 2010). For a yeast nucleus of ∼2 μm in diameter, this means that a chromosomal locus can explore ∼30% of the nuclear volume (Gartenberg et al. 2004), whereas the same movement in a mammalian nucleus corresponds to <1% of nuclear volume (Figure 3). In mammalian cells, the shift of a chromatin domain from a repressive to an active compartment may involve directional movement (Vazquez et al. 2001; Chuang et al. 2006). A similar shift was observed upon targeting of the viral transactivator VP16 to a yeast telomere: the activation displaced the telomeric locus from the NE (Taddei et al. 2004a; Hediger et al. 2006). This suggested that transcription might correlate with at least some degree of chromatin movement, consistent with the sensitivity of movement to glucose levels in the media as well as plasma and mitochondrial membrane potentials (Marshall et al. 1997; Heun et al. 2001b; Gartenberg et al. 2004).
By targeting transcriptional activators, chromatin modifiers, and nucleosome remodelers to a LacO-tagged locus, however, a recent study could differentiate the effects of transcription and of nucleosome remodeling on chromatin mobility (Neumann et al. 2012). No strict correlation was found between changes in mobility and transcriptional elongation, but the targeting of the INO80 remodeler complex clearly increased the radius of constraint of a tagged chromatin locus (Neumann et al. 2012). This required the ATPase activity of the remodeler and occurred without inducing transcription. The endogenous PHO5 promoter also showed increased mobility in an INO80-dependent manner, which was correlated with the INO80-dependent eviction of nucleosomes and rapid gene induction (Neumann et al. 2012). Together, these observations suggest that chromatin movement in the yeast interphase nucleus may reflect the nucleosome packing into a higher-order fiber and be influenced by remodeler-dependent removal of nucleosomes locally. On the level of a chromatin fiber, changes in movement may reflect changes in the persistence length of the polymer brought about by nucleosome displacement.
Both the nucleosome remodeler INO80 and the related complex SWR1 not only are recruited to promoters to enhance transcription, but also are recruited to DSBs to facilitate early steps in repair (reviewed in van Attikum and Gasser 2009; Lans et al. 2012). Intriguingly, DSBs also display enhanced mobility relative to the movement of undamaged loci (Dion et al. 2012; Miné-Hattab and Rothstein 2012). This increase in mobility required the strand-annealing factor Rad51 and the Snf2-related ATPase activity of Rad54 (Dion et al. 2012). These two proteins are necessary for repair by HR in an epistatic manner. The Rad9 protein and the ATR kinase, Mec1, which are both involved in the checkpoint response, also are required for increased movement of DSBs (Dion et al. 2012), and, consistently, only DSBs—and not protein-DNA adducts or nicks that do not activate the DNA damage checkpoint—increase locus mobility (Dion et al. 2012). It was also reported that undamaged sites increase mobility coincident with increased movement at the primary lesion, suggesting that there is a genome-wide response to DSBs (Miné-Hattab and Rothstein 2012). Together, these results suggest that locus movement not only is an inherent feature of chromatin, but also may play an active role in promoting the formation of compartments, shifting DNA from one compartment to another, or in facilitating DNA interactions during repair.
Functional Consequences of Nuclear Organization
The sequestration of SIR proteins at telomeres and silent HM loci due to telomere clustering has two functional consequences. First, the enrichment of SIR proteins favors subtelomeric repression, and second, it prevents promiscuous repression of a distinct subset of promoters elsewhere in the genome (Taddei et al. 2009). Sir3 protein abundance is relatively low (Ghaemmaghami et al. 2003) and is limiting for the spread of silent chromatin (Renauld et al. 1993). Therefore, the DNA sequence-dependent enrichment of SIR recruitment factors at telomeres probably accounts for the strong enrichment of SIR proteins in foci (Gotta et al. 1996). Indeed, the loss of telomere anchoring upon deletion of either subunit of yeast Ku leads to SIR protein delocalization throughout the nucleoplasm (Laroche et al. 1998).
The abundance of tandem RAP1-binding sites found in telomeric sequences competes with silencers for SIR factors, as monitored by functional assays. Early on, silencer-nucleated repression was found to be highly sensitive to the distance of the reporter from a telomere end (Renauld et al. 1993; Stavenhagen and Zakian 1994; Thompson et al. 1994; Maillet et al. 1996; Marcand et al. 1996), and telomere clustering favored the SIR-mediated silencing of endogenous subtelomeric genes (Mondoux et al. 2007; Taddei et al. 2009). Still, the association of subtelomeric loci with peripheral telomere clusters (visualized as Rap1-GFP foci) did not always correlate with silencing efficiency (Mondoux et al. 2007). This may be due to the dynamic behavior of individual telomeres, which move into and out of telomere foci, without loss of their epigenetic state (Schober et al. 2008). Most strikingly, however, was the demonstration that, by actively tethering a HMR reporter construct with a weakened silencer to the NE, transcriptional repression increased (Andrulis et al. 1998) in a manner strictly dependent on the presence of telomere clusters and SIR foci (Taddei et al. 2009). This argues that transcriptional repression does not reflect position per se, but rather access to local high concentrations of SIRs (Mondoux et al. 2007). Consistently, when telomere anchoring is impaired by deletion of YKU70 and ESC1, transcription of genes at internal loci (Maillet et al. 2001; Taddei et al. 2009) or at an excised ring of chromatin flanked by silencers (Gartenberg et al. 2004) showed enhanced repression, while telomere-proximal genes were derepressed.
Interestingly, changes in the spatial distribution of repetitive sequences can regulate patterns of gene expression genome-wide. Indeed, SIR protein dispersion from perinuclear foci was shown to specifically affect internal promoters carrying binding sites for Abf1 and the PAC-binding factors (Pbf1 and Pbf2), which regulate genes involved in ribosome biogenesis (Mondoux et al. 2007; Taddei et al. 2009). The cell may exploit such mechanisms by regulating SIR complex affinity at telomeres in a rapid response to environmental insult (Martin et al. 1999; Ai et al. 2002). Consistently, SIR dispersion, or modulation of TPE, has been observed under various forms of stress (Stone and Pillus 1996; Martin et al. 1999; McAinsh et al. 1999; Mills et al. 1999; Ray et al. 2003; Bi et al. 2004; Mercier et al. 2005; Smith et al. 2011). Stress-induced redistribution of SIR proteins may derepress subtelomeric genes required for growth on alternative carbon sources and simultaneously contribute to the down-regulation of genes involved in ribosome biogenesis as a global attempt to suppress growth and favor a survival pathway.
Over the past years NPCs have emerged as a major player in organizing gene activity. First, the nucleoporin Nup2 was shown to exhibit a strong boundary activity that can block the spread of heterochromatin when targeted on both sides of a reporter gene, possibly isolating chromatin by forming a loop (Ishii et al. 2002). Subsequently, a series of genome-wide studies suggested that nuclear pore components bind highly active genes (Casolari et al. 2004, 2005; Schmid et al. 2006). While some promoter-bound pore components are not stably bound to NPCs (Dilworth et al. 2005; Therizols et al. 2006; Ruben et al. 2011), the tagging and localization of specific inducible genes confirmed that some genes indeed shift to NPCs upon activation (for reviews, see Akhtar and Gasser 2007; Taddei 2007; Dieppois and Stutz 2010; Egecioglu and Brickner 2011) (Figure 6).
What moves active genes to the NPC is still unclear; indeed, each of the nuclear steps associated with gene expression, including initiation, elongation, mRNA processing, proofreading, and export, have been implicated. For example, the accumulation of a stalled intermediate in the mRNP export pathway in THO/sub2 mutants leads to the persistent NPC association of some active genes (Brickner and Walter 2004; Rougemaille et al. 2008), yet other data argue that promoter activation without transcription is sufficient for the shift (Casolari et al. 2004, 2005; Schmid et al. 2006). We suggest that multiple different steps of transcription contribute independently to the stable association of active genes with NPCs and that the importance of any individual step in this process may well be locus-specific. It is argued that the presence of multiple anchoring partners favors the formation of active gene loops, with promoters linking to 3′ sequences to improve the recycling of polymerases and elongation efficiency (O’Sullivan et al. 2004).
The targeting of genes to pores can also be mediated by sequence elements. Small DNA sequences called gene recruitment sequences (GRS I and II) were first identified in the promoter of INO1 and were proposed to function as a DNA “zip code” that confers perinuclear localization independently of the chromosomal context (Ahmed et al. 2010). Interestingly, the GRS I element is over-represented in the promoters of genes interacting with the NPC and in the promoters of stress-inducible genes.
NPC association is thought to be particularly important for inducible genes with galactose- or heat-shock-controlled promoters, as these genes require a rapid and high level of expression and efficient mRNA export. This could be facilitated by positioning the activated gene at pores. It was proposed in the “gene gating” hypothesis (Blobel 1985) that the NPC serves as a scaffold to build an assembly line that coordinates transcription, processing, and export of mRNA. Moreover, because of their eightfold symmetry, each pore could accommodate multiple genes, forming a factory that supports highly efficient transcription. Although “gene gating” might optimize the induction of some genes, pore association is clearly not obligatory for induced expression; mutations affecting gene-NPC association, for example, showed normal expression levels for induced GAL1 (Cabal et al. 2006). Moreover, gene repression did not generally lead to release from pores (Casolari et al. 2004). Finally, activation of the same promoter by different pathways or with different 3′ UTRs led to different positions of the gene within the nucleus (Abruzzi et al. 2006; Taddei et al. 2006). Nonetheless, it remains possible that the kinetics of induction was altered in these mutants, as association with the NPC was correlated with enhanced activation of at least some genes (Brickner and Walter 2004; Menon et al. 2005; Taddei et al. 2006). Thus, whether it acts by the formation of specialized transcription factories (Eskiw et al. 2010) or regulation of mRNA elongation, processing, and export, the NPC seems to efficiently fine-tune gene expression.
A second proposal suggests that the NPC–gene association is an epigenetic mark that enables a previous induction event to be “remembered” in subsequent generations, thereby facilitating re-induction (Brickner et al. 2007). Both GAL1 and INO1 genes are more rapidly reactivated after short-term repression, and this correlates with their persistence at the NPC after the inducing agent was removed. Although mechanisms involving cytoplasmic factors have also been implicated in rapid gene reactivation (Zacharioudakis et al. 2007), GAL gene transcriptional memory was nonetheless correlated with the formation of persistent loops between the 5′ and 3′ ends of the gene and association with the NPC (Laine et al. 2009; Tan-Wong et al. 2009). In the case of INO1, an 11-bp sequence appears to control both peripheral targeting and H2A.Z incorporation after repression (Brickner et al. 2007; Light et al. 2010). Finally, in cells lacking this sequence or a nuclear pore component, Nup100, transcriptional memory, and perinuclear association of INO1 were lost (Light et al. 2010). Thus, rapid re-induction may provide the rationale for NPC association of inducible genes.
Boundaries between active and inactive domains
Given the role of telomeric foci in the creation of repressive compartments and the ability of the NPC to support induced gene expression, one can conclude that the nuclear periphery has at least two functions in the control of gene expression. Consistent with this duality, proteins involved in gene activation or repression have distinct staining patterns at the nuclear periphery when localized at sufficiently high resolution (Taddei et al. 2004a; Horigome et al. 2011). Such juxtaposition of repressive (telomere foci) and activating compartments (NPCs) could favor efficient reversals of gene expression status, which may be especially relevant for subtelomeric genes, since, as mentioned above, subtelomeric zones are enriched for genes involved in alternative carbon source usage that are needed rapidly, but only under specific growth conditions (Fabre et al. 2005).
Intriguingly, mutations in nuclear pore constituents affect both gene activation and repression (Dilworth et al. 2005; Therizols et al. 2006; Ruben et al. 2011). This dual role does not appear to stem from the boundary activity of nuclear pore proteins, as reported by Ishii et al. (2002). Indeed, it was shown that, despite the fact that nuclear pore proteins can bind the native tDNA insulator adjacent to HMR, the loss of their binding did not compromise insulator activity, even though they contributed to silencing at this locus (Ruben et al. 2011). To reconcile the apparently conflicting role between NPCs and repressive telomeric foci, one might propose that the impact of NPCs on gene expression is not simply pore or position dependent, but depends on the binding of transactivators or repressors to cis-acting elements (Ruben et al. 2011). Consistent with this notion, increasing the association of the HXK1 subtelomeric gene with the nuclear periphery through a neutral anchor improved both its repression of glucose medium and its activation in the absence of glucose (Taddei et al. 2006).
Effects on genome stability
In contrast to the extensive studies on the function of nuclear organization in gene regulation and DNA replication, relatively little is known about how chromatin dynamics and higher-order nuclear organization contribute to genome stability. Until recently, many basic questions remained unaddressed, such as whether or not DSBs are mobile in the nucleus, whether DSBs are recruited to scaffolds for repair, or whether, instead, repair factors are recruited to sites of damage. It was also unclear whether specific pathways of repair occur preferentially in specific subcompartments of the nucleus, and if so, how chromatin dynamics contribute to the process of repair. With a combination of genetics and real-time live imaging, recent studies show that chromatin dynamics and nuclear organization do play a major role in the repair of DNA lesions and the maintenance of chromosome integrity in yeast.
Mobility of DSBs:
The introduction of two DSBs on different chromosomes can cause reciprocal translocations as frequently as they produce intrachromosomal deletions, arguing that DSBs are dynamic and can roam throughout the nucleus to find their appropriate template for repair (Lisby et al. 2003). Indeed, the subdiffusive movement of yeast chromatin increases four- or fivefold within 30 min after induction of a DSB, such that the lesion can explore ∼50% of the nuclear volume (Dion et al. 2012; Miné-Hattab and Rothstein 2012). Consistently, Houston and Broach have shown in living yeast cells that a homology search for a cleaved MAT locus is efficient, such that pairing with the homologous donor on the distal arm of the same chromosome can occur within 90 min after DSB induction (Houston and Broach 2006; Hicks et al. 2011). Given that a homology search takes longer for ectopic donor sequences, it can be predicted that DSBs require enhanced mobility (over that of an undamaged locus) to find the donor sequence for HR.
In mammalian cells, there are conflicting reports as to the mobility of DNA damage. In one case, an induced DSB appeared nearly immobile, and when break movement was enhanced, increased translocation frequencies with nearby chromosomes occurred (Soutoglou et al. 2007). In contrast to this, two reports show enhanced mobility for either a DSB in U2OS cells or an uncapped telomere in mouse embryonic fibroblasts (Dimitrova et al. 2008; Krawczyk et al. 2012). This latter was dependent on 53BP1, a repair factor that helps limit resection in favor of end-joining. Some of the discrepancy in the movement of DNA damage can probably be attributed to the fact that different types of DNA damage were induced and that the cells activate the ATR/ATM checkpoint response to different degrees, which was shown in yeast to contribute to DSB movement (Dion et al. 2012; Krawczyk et al. 2012).
In budding yeast, the rate of HR-mediated repair, scored in the presence of one or multiple dispersed copies of donor sequence, suggests that the homology search is indeed rate-limiting for recombination (Wilson et al. 1994). Specifically, the frequency of recombination was proportional to the number of targets in each strain tested. This dependency on copy number for both dispersed and clustered targets suggests that the homology search takes place by a random 3D collision, rather than by sliding the donor along DNA until the recipient is found. Intriguingly, Miné-Hattab and Rothstein (2012) report a general increase in chromatin mobility in response to elevated levels of γ-irradiation. This would indeed be likely to enhance the chances of random collision events between DSB and uncut donor.
Nuclear pores and Mps3 anchorage in noncanonical repair:
The random 3D search could also be facilitated by structures that recruit sites of damage and therefore promote recombination or alternative pathways of repair (Gehlen et al. 2011b). Two possibilities for this “repair scaffold” are nuclear pores and the above-mentioned SUN domain protein, Mps3. It was shown that certain types of DNA damage, such as persistent DSBs, collapsed replication forks, and eroded telomeres are recruited to nuclear pores, which are themselves associated with Ulp1, a Sumo protease, and a SUMO-dependent ubiquitin ligase complex, Slx5/Slx8 (Khadaroo et al. 2009). Namely, lesions that are impossible or difficult to repair by canonical HR may be brought to nuclear pores where alternative recombination and repair pathways can come into play. Genetic interactions suggest that the importance of the nuclear pore resides in its sequestration of Ulp1 and Slx5/Slx8, and its proximity to the proteosome (Therizols et al. 2006; Lewis et al. 2007; Palancade et al. 2007; Nagai et al. 2008). Thus, one scenario is that sumoylated proteins accumulate at collapsed forks or resected breaks that require Slx5/Slx8 ubiquitylation for proteasomal degradation, or alternatively, desumoylation by Ulp1, to enable appropriate repair. Exactly what the relevant protein may be is at present unknown, although Rad51, Rad52, Pol32, PCNA, and other proteins of the replication machinery are heavily sumoylated upon DNA damage (Melchior 2000; Nagai et al. 2011; Cremona et al. 2012).
Other studies have reported the association of irreparable DSBs with the INM protein Mps3 (Kalocsay et al. 2009; Oza et al. 2009). The tethering of a persistent DSB by Mps3 is thought to contribute to the suppression of ectopic HR. Similarly, the peripheral tethering of short telomeres by Mps3 contributes to the suppression of recombination between subtelomeric Y′ elements in cells lacking the ATM kinase, Tel1 (Schober et al. 2009). These studies are consistent with the observation that spontaneous Rad52 foci are strongly enriched in the nuclear lumen (Bystricky et al. 2009). Indeed, this study showed that a donor locus for HR can be shifted from a more peripheral location to an internal one, upon its recruitment for a recombination event. The enrichment for Rad52 foci in the nuclear interior further suggests that homology-driven recombination might even be suppressed at the NE. Whether sister-chromatid exchange is antagonized by peripheral tethering remains to be seen.
Clearly, however, the yeast NE presents more than one compartment for repair (Figure 5). The connection between pore-associated and Mps3-associated repair events is currently unknown, but it is likely that they are either mechanistically or kinetically linked. The involvement of the Nup84 complex in the repair of DSBs near telomere ends (Therizols et al. 2006; Lewis et al. 2007; Palancade et al. 2007) and the shift of eroded telomeres to nuclear pores (Khadaroo et al. 2009) argues that there may be a sequential treatment of damage and a passage from one site to another, depending on the repair required. Peripheral tethering in general appears to favor noncanonical pathways of repair, as both the pore association and the Slx5/Slx8 complex are important for the ALT mechanism of telomere maintenance (Azam et al. 2006; Khadaroo et al. 2009). The Slx5/Slx8 complex facilitates the amplification of telomere repeats to generate survivors in telomerase-deficient cells in a recombination event that requires Rad52 and Sgs1. Khadaroo et al. (2009) further suggest a link between recombination-mediated telomere elongation and nuclear pores, as they show that eroded telomeres are recruited to pores along with factors involved in checkpoint and recombination, namely RPA and Rad52 (Khadaroo et al. 2009). Thus, critically short telomeres, like persistent DSBs, may be delivered to pore-associated Slx5/Slx8 to enable the ubiquitylation and degradation of a protein that might otherwise prevent strand invasion.
Regulation of recombination at the rDNA locus:
The notion that tethering at the NE suppresses recombination is supported by studies of the rDNA. Whereas HR within the rDNA repeats can contribute to the maintenance of stable copy number (Smith 1974; Kobayashi et al. 1998), the frequency of HR in the rDNA is nonetheless significantly lower than would be expected from its repetitive nature. This suggests that recombination is actively suppressed at the rDNA locus (Petes 1980). Indeed, Torres-Rosell et al. (2007) showed that HR proteins, such as Rad52, Rad51, Rad55, and Rad59, are excluded from the nucleolus, although sensors of DSB, such as Mre11 and Rfa1, are present. Furthermore, they showed that a single DSB in the rDNA induced by I-SceI endonuclease shifts away from the nucleolus when Rad52 is recruited, suggesting that repair by HR must take place outside this compartment. Interestingly, nucleolar exclusion of Rad52 foci requires both the Smc5/Smc6 complex, which harbors E3 SUMO ligase activity, and the sumoylation of Rad52. The physiological importance of this nucleolar exclusion of HR is shown by the fact that the mutants that impair nucleolar exclusion, such as smc5 or smc6 mutants, show rDNA hyper-recombination by unequal sister-chromatid exchange and an accumulation of excised extrachromosomal rDNA circles (Burgess et al. 2007; Torres-Rosell et al. 2007).
Not only does the exclusion of Rad52 foci stabilize the rDNA, but also the juxtaposition of the rDNA to the NE itself appears to play a role in the repression of recombination between rDNA repeats (Mekhail et al. 2008). As described above, the anchoring of silent rDNA repeats is mediated by an INM, Heh1, which shares homology with human Man1 (M. C. King et al. 2006) and other members of the INM family of LEM (LAP-Emerin-Man1) domain proteins (Figures 2 and 5). Deletion of HEH1 releases the rDNA from the nuclear periphery, an event that correlates with an increase in unequal sister-chromatid exchange. Intriguingly, Heh1 is not required for rDNA silencing (Mekhail et al. 2008) and is thus unlike Sir2, which packages chromatin to reduce Pol II transcription within the rDNA and to prevent unequal exchange between repeats (Gottlieb and Esposito 1989). As mentioned above, Heh1 anchors the rDNA through Lrs4, which connects Sir2 to Heh1. Interestingly, by fusing Heh1 to Sir2, the requirement for Lrs4 in rDNA anchoring is bypassed, and the Sir2–Heh1 tether could suppress the rDNA instability that correlates with loss of Lrs4 (Mekhail et al. 2008). This suggests that proximity to the NE reduces rDNA recombination events, although Sir2 overexpression may have influenced the suppression of recombination in this experiment (Mekhail et al. 2008).
Whereas the sequestration of the rDNA and of telomeres at the NE probably reflects two different means to suppress Rad52-mediated events, it may nonetheless function both by excluding enzymatic activities or by favoring a processing event. One should note, however, that even though the rate of recombination between telomeric repeats increases when tethering is compromised in a yku70 mutant, this effect is not solely due to a loss of telomere tethering (Marvin et al. 2009a,b). It remains to be seen what other factors protect repeated sequences from unscheduled recombination events.
Despite rapid progress in our understanding of nuclear organization, much remains unknown. Recently, it has been shown that some pore components are highly mobile and remain bound at the pore for only short periods of time (Dilworth et al. 2005; Oka et al. 2010), yet the implications of these dynamics for chromatin-based functions are unclear. It is also unknown whether there are different classes of nuclear pore complexes that harbor different functions, perhaps with distinct distributions around the nuclear edge. This can perhaps be answered with specialized mass spectroscopy techniques. Clearly, modeling studies that incorporate dynamics combined with genetic assays, as performed for the segregation of plasmids (Gehlen et al. 2011a,b), will make important contributions to understanding nuclear dynamics.
Still unexplained is how different nuclear compartments signal to each other, how they detect the metabolic status of the cell or the environment, and whether certain events or compartments may be mutually exclusive within the same nucleus under specific kinds of stress. Clearly, there are changes in nuclear organization that correlate with the activation of DNA-damage-associated checkpoint kinases, such as the release of telomere-bound SIR proteins in response to DNA damage (Martin et al. 1999; McAinsh et al. 1999; Mills et al. 1999). In a more recent article it was shown that the anchoring of transcribed genes to the nuclear periphery is counteracted by checkpoint kinases through the phosphorylation of nucleoporins like Mlp1 (Bermejo et al. 2011). It was proposed that this release neutralizes the topological stress generated during the transcription of nuclear pore-associated genes. Efficient genome-wide position mapping techniques need to be developed to test the model that assigns function to perinuclear tethering sites or altered chromatin mobility.
Still unexplained is how nuclear compartments reform after mitotic division, or persist during the dynamic segregation of mitotic chromosomes. The contribution of organization to epigenetic inheritance is still poorly defined. Finally, it is still unclear whether noncoding RNA is a major structural element of the budding yeast nucleus, as it appears to be in other organisms. These and other questions about nuclear organization and its functional implications promise a fascinating future for this emerging field.
We thank S. Nagai for the repair figure and K. Bystricky for images in Figure 1. We thank V. Dion, C. Horigome, H. Ferreira, M. Oppikofer, and S. Kueng of the Gasser laboratory for a critical reading of the review. S.M.G. acknowledges support of the Novartis Research Foundation, the National Center for Competence in Research, Frontiers-in-Genetics, and the European Union Network Of Excellence Epigenome. A.T. is supported by the French Agence Nationale pour la Recherche and the European Research Council (ERC) under the European Community’s Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement no. 210508.
Communicating editor: J. Boeke
- Received March 19, 2012.
- Accepted May 29, 2012.
- Copyright © 2012 by the Genetics Society of America