Early initiation of a replication origin tethered at the nuclear periphery

DNA replication in eukaryotic cells is normally initiated from a large number of origins located at intervals along the linear chromosomes. Individual origins initiate replication according to a temporal program, with some origins initiating early and others later in S phase (Friedman et al., 1995). In metazoan cells, replication timing correlates with subnuclear localization and gene activity of chromosomal domains. Inactive chromatin regions usually replicate late and are frequently located at the nuclear periphery or close to a nucleolus (Cimbora and Groudine, 2001). The establishment of the replication-timing program occurs in early G1 phase, coincident with the re-establishment of the spatial organization of chromatin following mitosis (Raghuraman et al., 1997; Dimitrova and Gilbert, 1999; Li et al., 2001). The established late replication context of a peripherally positioned chromosomal locus can be maintained during S phase, even if the sequence was released from the nuclear periphery in late G1 phase (Heun et al., 2001).

In the budding yeast Saccharomyces cerevisiae, origin activation time appears to depend on chromosomal context. Replication origins near telomeres are a typical class of late-replicating origins in yeast (Ferguson and Fangman, 1992) and the telomeres are localized to the nuclear periphery during most of interphase. Moreover, disruption of the telomere-binding protein complex Ku effects not only telomere localization (Laroche et al., 1998), but also replication-timing control (Cosgrove et al., 2002). Despite these observations, there is no clear evidence that peripheral localization directly causes late replication.

The histone modification state of chromatin surrounding an origin does influence replication timing of the origin. For example, establishment of silent chromatin has been shown to cause hydroxyurea sensitivity of replication initiation, which is characteristic of late origins (Zappulla et al., 2002). Acetylation of histones close to a normally late-replicating origin makes the origin initiate earlier (Vogelauer et al., 2002; Goren et al., 2008; Knott et al., 2009). However, manipulating histone acetylation causes fairly small changes in replication timing. There is moreover no clear correlation between origin initiation time in S. cerevisiae and acetylation level of the surrounding nucleosomes (Nieduszynski et al., 2006), highlighting our limited understanding of the controls over the replication-timing program.

These potential mechanisms that might influence replication timing are not mutually exclusive, and one mechanism could positively or negatively affect the others. To obtain a full understanding of the molecular control(s) over replication timing, it is necessary to dissect the effects of the potential control mechanisms. To test the possibility that localization at the nuclear envelope delays origin firing, we artificially tethered the early replicating origin ARS607 to the nuclear envelope and examined the replication time of the repositioned origin. Here, we show that peripheral positioning of ARS607 is not sufficient to delay firing of this origin. Therefore, peripheral positioning of an origin is not sufficient to establish a late-replicating chromosomal region.

To examine whether perinuclear localization affects the temporal program of replication origin activation, we used a system designed to allow ARS607, an early-replicating origin on chromosome VI, to be artificially tethered to the nuclear periphery. ARS607 is located on the right arm of chromosome VI, approximately 51 kb from CEN6 and 71 kb from the right telomere (Shirahige et al., 1993), and shows largely random subnuclear positioning (Taddei et al., 2004). ARS607 is efficiently active (initiating replication in >85% of cells) and replicates early in S phase (Friedman et al., 1997; Yamashita et al., 1997). To enable controlled tethering of ARS607 to the nuclear periphery, we used a strain in which four copies of the lexA operator sequence (lexA^op) are inserted 0.7 kb from ARS607 (Fig. 1A) (Taddei et al., 2004). This insertion allows the ARS607 locus to be directed to the nuclear periphery by expressing a ‘tethering construct’ in which the LexA DNA-binding domain is fused to a protein moiety that can mediate peripheral positioning [such as fragments of the Ku and Sir proteins implicated in telomere tethering (Taddei et al., 2004)]. The locus was visualized by inserting an array of lacO operator sequences (centered 5.9 kb from ARS607) and expressing GFP fused to LacI repressor protein (LacI-GFP) (Straight et al., 1996). In vivo, 5.9 kb corresponds to separation of less than 50 nm (Bystricky et al., 2004); because this distance is much shorter than the ~200 nm resolution limit of light microscopy, the position of the lacO array can be considered to reflect the position of ARS607. The strain additionally expresses the nuclear pore protein Nup49 fused to either GFP or mCherry protein (Iwase et al., 2006), allowing microscopic assessment of ARS607 location relative to the nuclear envelope.

We employed three different LexA fusion tethering constructs: LexA-Sir4^PAD, LexA–Yku80-9 and LexA-Yif1, all of which were previously shown to mediate nuclear peripheral localization (Taddei et al., 2004). LexA-Sir4^PAD contains a fragment of the Sir4 protein. Sir4 forms part of the telomeric transcriptional silencing machinery, but because the Sir4^PAD fragment lacks the domain required for interaction with other Sir proteins, it cannot nucleate silent heterochromatin when tethered to a chromosomal locus (Ansari and Gartenberg, 1997; Taddei et al., 2004). Another tethering construct, LexA–Yku80-9, is formed from LexA fused to an allele of the telomere-binding Yku80 protein (Taddei et al., 2004). We also tested a construct containing LexA fused to the inner nuclear membrane protein Yif1, which can tether ARS607 to the nuclear periphery independent of proteins involved in telomere clustering (Taddei et al., 2004). Importantly, none of these LexA fusions induces transcriptional silencing at the ARS607 locus (supplementary material Fig. S1), although LexA–Yku80-9 was previously shown to induce moderate silencing at a crippled silencer (Taddei et al., 2004). All three LexA fusions mediate tethering of the ARS607 locus to the nuclear periphery in G1 and early S phase cells (data not shown), as previously reported (Taddei et al., 2004; Hiraga et al., 2006; Ebrahimi and Donaldson, 2008). Yeast chromatin is highly mobile and it should be noted that the ARS607 locus remains dynamic, even when positioned at the periphery by one of these constructs. Therefore, we observe ARS607 at the periphery in 60-70% of cells in a population snapshot (compared to 33% of cells for a randomly positioned locus). This observation is consistent with previous data showing that endogenous perinuclear chromosome domains, such as telomeres, are peripheral in 60-70% of cells (Hediger et al., 2002; Brickner and Walter, 2004).

We first checked that peripheral tethering does not repress origin activity, using neutral/neutral two-dimensional gel electrophoresis (Fig. 1B) (Friedman and Brewer, 1995). A bubble arc indicative of ARS607 origin activation was detected in the LexA-expressing strain. The presence of a similar bubble arc in the strains expressing LexA-Sir4^PAD, LexA–Yku80-9 and LexA-Yif1 indicated that ARS607 remains active when tethered to the nuclear periphery.

Next, we examined the effect of perinuclear tethering on replication timing of ARS607 using the dense-isotope transfer method. In this technique, the replication kinetics of specific sequences are monitored in a synchronized culture, by tracking the shift in density of genomic DNA fragments caused by the incorporation of specific carbon and nitrogen isotopes into nascent DNA (McCarroll and Fangman, 1988). The graphs in Fig. 2A show the replication kinetics of three different sequences: the early-replication origin ARS306; a late marker sequence, ‘chrXIV-internal’, centered at approximately 223 kb on chromosome XIV; and the ARS607 locus. As expected, in the control strain (expressing LexA alone), ARS607 replicated early in S phase, shortly after ARS306, which is one of the earliest-replicating sequences in the genome (Fig. 2A). Expression of either the LexA-Sir4^PAD or LexA-Yif1 localization constructs did not result in any noticeable change in the replication kinetics of ARS607 (Fig. 2A, right hand graphs). Expression of LexA–Yku80-9 caused a very slight delay in the replication time of ARS607 relative to the early marker ARS306 (Fig. 2A, lower left panel), but this change might not be significant because it lies within the margins of error typically seen for this type of experiment (Friedman et al., 1996).

Replication time in these experiments is defined as the time at which a sequence has replicated in half of the cycling cells. The ‘replication index’ (RI) is the replication time of a sequence expressed relative to the early and late markers (whose replication times are assigned as 0 and 1, respectively). Calculating the RI normalizes the differences in the speed with which different cultures release from synchronization and proceed through S phase. Fig. 2B shows RI values for ARS607 in the four experiments, and confirms that expressing LexA-Sir4^PAD or LexA-Yif1 caused no significant change in replication time, whereas LexA–Yku80-9 caused only a very slight change.

The results in Fig. 2 do not support the idea that perinuclear positioning mediates late replication. However, the replication timing values measured by the density-transfer method represent an average (mean) for all the cells in the population. Because the LexA fusion proteins typically cause perinuclear tethering in only 60-70% of cells at any moment (Taddei et al., 2004; Hiraga et al., 2006; Ebrahimi and Donaldson, 2008), it is conceivable that a delay to replication time caused by perinuclear localization could be obscured because ARS607 is not localized in 30-40% of cells. To investigate this possibility, we examined the replication of the ARS607 locus microscopically in individual cells, by monitoring the doubling of GFP fluorescence intensity (Kitamura et al., 2006; Ebrahimi and Donaldson, 2008). Briefly, time-lapse experiments were carried out to measure three parameters in cells undergoing bud emergence: GFP fluorescence intensity of the lacO-lexA^op-ARS607 locus, position of the locus relative to the nuclear envelope and bud size. Representative images are shown in Fig. 3A. The plots in Fig. 3C-D show the results for a series of individual cells. In these plots, the open diamonds show bud size at successive time points and red triangles indicate time points at which the ARS607 locus was at the nuclear periphery. Measuring the midpoint of the increase in GFP fluorescence (filled green circles) allows assignment of the ARS607 replication time in each cell.

In the control strain expressing LexA, the ARS607 locus was consistently replicated 2-6 minutes after bud emergence (Fig. 3B, Fig. 4A). Replication time of the ARS607 locus was also measured in cells expressing LexA-Sir4^PAD (Fig. 3C) and LexA–Yku80-9 (Fig. 3D). In neither case did we observe any consistent delay in replication timing relative to bud emergence. Importantly, even when examining these individual cells, no relationship was observed between ARS607 replication time and its localization status at immediately preceding time points (Fig. 4A). Specifically, the ARS607 locus showed no tendency to replicate later when it was localized at the periphery immediately before replication (Fig. 4A, grey and black circles do not cluster towards the top of the chart). Tethering of ARS607 to the nuclear periphery by LexA-Sir4^PAD or LexA–Yku80-9 did result in replication timing that appears slightly more scattered relative to bud emergence than observed for the control strain (Fig. 4A), but the variation in replication timing relative to budding (s.d. less than 4 minutes) was much smaller than the length of S phase (15-20 minute difference in replication time between earliest and latest origins). Moreover, the ARS607 locus replicated at a very similar average bud size, whether or not it was tethered to the nuclear periphery (Fig. 4B). To summarize, the results of single-cell analysis also indicated that peripheral localization of ARS607 has no major impact on its replication time.

The density-transfer method allows precise, standardized comparison of average replication times in a population, whereas single-cell imaging enables the simultaneous recording of replication time and locus subnuclear position in individual living cells. Assessing replication time using either technique revealed that perinuclear positioning of the origin does not impact its initiation time. In summary, our findings suggest that subnuclear localization is not the main determinant of replication timing in budding yeast. This reveals the reason for previous observations that telomeric origins can replicate late even when the origin is not properly localized to the nuclear periphery (Heun et al., 2001; Hiraga et al., 2006).

Yeast strains are described in supplementary material Table S1. Primer sequences used in strain construction are available on request. Plasmids pAT4, pAT4-Sir2, pAT4-Sir4^PAD, pAT4–Yku80-9 and pAT4-Yif1 were described previously (Taddei et al., 2004).

Quantitative measurements of microscopic images were performed as described (Ebrahimi and Donaldson, 2008). To determine the time point at which ARS607 replicated, a curve is fitted over the intensity data points; the midpoint of the intensity increase on the fitted curve is assigned as the replication time.

The replication timing analyses shown in Fig. 2 were carried out using the dense-isotope transfer technique, as previously described (Donaldson et al., 1998).

Genomic DNA was prepared as described (Huberman et al., 1987; Brewer et al., 1992). DNA fragments digested using NcoI and EcoRI were separated by neutral/neutral two-dimensional agarose gel electrophoresis (Friedman and Brewer, 1995) and transferred to neutral membrane (Qbiogene) by Southern blotting. The 1633-bp fragment containing ARS607 was detected using a suitable ³²P-labeled probe.

We thank Ian Stansfield, Berndt Müller, M. K. Raghuraman and Bonny Brewer for helpful comments. This research was supported by the Wellcome Trust (grant 082377), the Association for International Cancer Research (grant 05-0445) and Cancer Research UK (grant A2571). Deposited in PMC for release after 6 months.