Shelterin promotes tethering of late replication origins to telomeres for replication‐timing control

Abstract

DNA replication initiates at many discrete loci on eukaryotic chromosomes, and individual replication origins are regulated under a spatiotemporal program. However, the underlying mechanisms of this regulation remain largely unknown. In the fission yeast Schizosaccharomyces pombe, the telomere‐binding protein Taz1, ortholog of human TRF1/TRF2, regulates a subset of late replication origins by binding to the telomere‐like sequence near the origins. Here, we showed using a lacO/LacI‐GFP system that Taz1‐dependent late origins were predominantly localized at the nuclear periphery throughout interphase, and were localized adjacent to the telomeres in the G1/S phase. The peripheral localization that depended on the nuclear membrane protein Bqt4 was not necessary for telomeric association and replication‐timing control of the replication origins. Interestingly, the shelterin components Rap1 and Poz1 were required for replication‐timing control and telomeric association of Taz1‐dependent late origins, and this requirement was bypassed by a minishelterin Tpz1‐Taz1 fusion protein. Our results suggest that Taz1 suppresses replication initiation through shelterin‐mediated telomeric association of the origins at the onset of S phase.

Introduction

To ensure complete duplication of the eukaryotic genome within the S phase of the cell cycle, DNA replication initiates from a number of discrete loci known as replication origins. Replication at these origins is not initiated simultaneously at the beginning of S phase, but rather at distinct origin‐specific times during S phase. Temporal control of replication is conserved from yeasts to metazoans (MacAlpine & Bell, 2005; Gilbert et al, 2010).

In mammalian cells, replication foci characteristic of actively replicating chromosomal regions are localized to the interior of the nucleus in early S phase, while they are localized to the nuclear periphery from mid to late S phase (Nakamura et al, 1986; Nakayasu & Berezney, 1989; O'Keefe et al, 1992). Prior studies suggest that a spatiotemporal replication program is established at a specific time point (timing decision point, TDP) in the G1 phase (Dimitrova & Gilbert, 1999). In budding yeast, late replication origins are localized proximal to the nuclear envelope in G1‐arrested cells, while early origins are randomly distributed within the nucleus (Heun et al, 2001). However, artificial tethering of an early replication origin to the nuclear membrane did not shift its replication timing to late S phase, suggesting that peripheral localization of origins per se is not the critical determinant for their activation timing (Ebrahimi et al, 2010). Therefore, the molecular mechanisms of spatiotemporal replication control remain largely elusive.

Telomeres are specialized chromatin structures composed of an array of short telomeric repeats and specific binding proteins at the end of linear chromosomes in eukaryotes. In Schizosaccharomyces pombe, the telomeric protein Taz1 directly binds to double‐stranded telomere DNA (Cooper et al, 1997; Spink et al, 2000) and recruits two proteins, Rap1 and Rif1, to the telomeres (Kanoh & Ishikawa, 2001). Taz1 and Rap1, together with other components including Poz1, Tpz1, Ccq1, and Pot1, form the conserved shelterin complex that protects telomere ends and controls telomere length (Miyoshi et al, 2008; Palm & de Lange, 2008). Telomeres are anchored to the nuclear periphery through redundant mechanisms in many organisms (Palladino et al, 1993; Chikashige et al, 2009; Crabbe et al, 2012; Ferreira et al, 2013). In S. pombe, telomeres are clustered together and anchored to the nuclear periphery through the interaction of Rap1 with the nuclear membrane proteins Bqt3 and Bqt4 (Chikashige et al, 2009). Telomere clustering is required for meiotic chromosome movement and homologous recombination (Chikashige et al, 2006), but not for telomere length control or heterochromatin formation in the mitotic cell cycle (Chikashige et al, 2010).

Recent studies have demonstrated that telomere‐binding proteins play crucial roles in replication‐timing control in S. pombe. A subset of late‐replicating origins are accompanied by two copies of a telomere‐like sequence, and binding of Taz1 to this sequence is essential for the late‐timing control of these origins (Tazumi et al, 2012). In addition, these Taz1‐dependent late origins require Rif1 for timing regulation, although Rif1 is not localized to the origins (Hayano et al, 2012; Tazumi et al, 2012; Kanoh et al, 2015). Both Taz1 and Rif1 are also required for regulation of the late origins in subtelomeres spanning as far as 100 kb from the chromosome ends (Tazumi et al, 2012). However, Rif1 regulates another type of late origin, where Rif1 is localized independently of Taz1 (Hayano et al, 2012; Tazumi et al, 2012; Kanoh et al, 2015).

Recently, an intriguing model has been proposed in which Rif1 recruits protein phosphatase 1 (PP1) that acts as a counter‐phosphatase against the phosphorylation of Mcm2‐7 subunits (Dave et al, 2014; Hiraga et al, 2014; Mattarocci et al, 2014). Because phosphorylation of Mcm2 and Mcm4 by Dbf4‐dependent kinase (DDK) is essential for the assembly of initiation factors including Sld3, GINS, and Cdc45, recruitment of PP1 to the replication origins prevents replication initiation at the onset of S phase. This model is consistent with regulation of the subtelomeric origins and Taz1‐independent late origins, where Rif1 is localized. However, it is unclear whether replication‐timing control of Taz1‐dependent late origins, where Rif1 is not localized, involves additional mechanisms such as localization to specific intra‐nuclear regions where Rif1 and PP1 are abundant.

Here, we used a fluorescence microscopy system to examine whether replication origins, especially the Taz1‐dependent late origins, were localized to specific intra‐nuclear compartments in living fission yeast cells. The Taz1‐dependent late origins were localized adjacent to telomeres at the nuclear periphery during G1/S phase. Interestingly, the shelterin components Rap1 and Poz1 were required for replication‐timing control as well as telomeric association of the Taz1‐dependent late origins located in chromosomal arm regions, although these components were dispensable for the timing control of subtelomeric late origins. Moreover, expression of a minishelterin Tpz1‐Taz1 fusion protein restored both replication‐timing control and telomeric association of the origins in rap1Δ poz1Δ cells. These results suggest that Taz1 regulates the late replication timing of inner arm origins through shelterin‐mediated telomere association in S. pombe.

Results

Distinct intra‐nuclear localizations of early and late replication origins

To visualize the subnuclear localization of early and late replication origins in S. pombe, the replication origins were tagged with a lacO array/GFP‐conjugated LacI repressor. A lacO array of approximately 10 kb was inserted at 0.1 kb from the well‐characterized early origin ars2004, or at 0.7 kb from the Taz1‐dependent late origin AT2088 located on the left or right arm, respectively, of chromosome II (Fig 1A). Insertion of the lacO array did not change the replication timing of either origin (Appendix Fig S1).

Figure 1. Taz1‐dependent late replication origins are localized at the nuclear periphery.

1. The locations of early and late replication origins relevant to this study are schematically shown on Schizosaccharomyces pombe chromosome II. The early origins ars2004, ori2024, and ori2123, the Taz1‐dependent late origins AT2088 and ori2100, and the Taz1‐independent late origin ars727 are indicated. The location of the centromere (cen2) is depicted by a yellow square, and the telomeres are indicated by black arrowheads.
2. Schematic images for the analysis of the subnuclear locations of LacI‐GFP‐tagged replication origins in comparison with the mCherry‐Ish1 at the nuclear envelope under a fluorescence microscope. Among 21 focal planes with 0.1‐μm focus intervals, a single plane that contained the strongest lacO/LacI‐GFP signal was chosen for quantitative analyses.
3. Representative images of lacO/LacI‐GFP (green) at the AT2088 late origin (upper 4 panels) and at the ars2004 early origin (lower 4 panels) merged with Ish1‐mCherry (red) are presented. The optical section images were processed as described in “Materials and Methods”. The scale bar indicates 5 μm.
4. Images were classified into two categories: Class 1 (peripheral), GFP signals that overlapped with Ish1‐mCherry; and Class 2 (non‐peripheral), GFP signals that did not overlap with or had separated from Ish1‐mCherry. The proportions of Class 1 (blue) and Class 2 (pale blue) of the AT2088 late origin and the ars2004 early origin are presented in pie charts.
5. The Class 1 and Class 2 proportions of ori2100 (Taz1‐dependent late origin), ars727 (Taz1‐independent late origin), ori2024 (early origin), and ori2123 (early origin) are shown in pie charts.
6. The effect of a base substitution (AT2088‐S2632) in the telomere‐like sequence near the AT2088 on the peripheral localization of AT2088 and the effects of deletion of taz1 ⁺ on the localizations of late origins AT2088 and ori2100, and the early origin ars2004 were examined. Representative images of lacO/LacI‐GFP (green) and Ish1‐mCherry (red) are presented with the proportions of Class 1 (peripheral, blue) and Class 2 (non‐peripheral, pale blue). The scale bar indicates 5 μm.

Localization of the LacI‐GFP‐tagged origin relative to mCherry‐fused Ish1, a nuclear envelope protein (Taricani et al, 2002), was analyzed in living cells under a fluorescence microscope. Among 21 focal planes, we selected a single plane that contained the strongest LacI‐GFP signal (Fig 1B). Interestingly, the AT2088‐LacI‐GFP focus colocalized with Ish1‐mCherry at the nuclear membrane, whereas ars2004‐LacI‐GFP was observed at various locations in the nucleus (Fig 1C). For a quantitative analysis of LacI‐GFP localization in the nucleus, the number of cells with Ish1‐mCherry colocalization (peripheral, Class 1) and those without colocalization (non‐peripheral, Class 2) were counted (Fig 1D). The proportion of Class 1 was 79.3% for AT2088‐LacI‐GFP, and 48.5% for ars2004‐LacI‐GFP (Fig 1D). The difference in intra‐nuclear localization of the two origins was statistically significant (P < 0.0001, Fisher's exact test). These results demonstrate that AT2088 is more predominantly located at the nuclear periphery than ars2004.

We then examined whether peripheral localization is a common property of late replication origins. Another Taz1‐dependent late origin, ori2100, which is located near the centromere on the right arm of chromosome II (Fig 1A), predominantly colocalized with Ish1‐mCherry on the nuclear membrane, similar to AT2088 (P > 0.9999; Fig 1E and Appendix Fig S2). In contrast, a different type of late origin, ars727, which is dependent on Rif1 but not Taz1 and located on the right arm of chromosome II (Fig 1A), had 57.4% Class 1 localization, which was statistically different from that of AT2088 and ori2100 (P‐values < 0.0001; Fig 1E and Appendix Fig S2). In addition, the early origins ori2024 and ori2123 (Fig 1A) did not exhibit preferential localization at the nuclear periphery compared to AT2088 (P‐values < 0.0001; Fig 1E and Appendix Fig S2). The localization pattern of each origin was not significantly altered between the G1/S and G2 phases (Fig EV1). However, the results of a time‐lapse analysis suggested that the Taz1‐dependent late origin AT2088 dissociated from the nuclear periphery during the M phase (Fig EV2). Therefore, our results show that Taz1‐dependent late origins localize at the nuclear periphery throughout the interphase of the cell cycle at a higher frequency than Taz1‐independent late origins and early origins.

Figure EV1. Intra‐nuclear locations of LacI‐GFP‐tagged early and late origins in comparison with Ish1‐mCherry during G1/S and G2 phases.

Class 1 (peripheral, blue) and Class 2 (non‐peripheral, pale blue) proportions of the Taz1‐dependent late origins AT2088 and ori2100 and the early origins ars2004, ori2024, and ori2123 were determined separately during G1/S phase (bi‐nuclear) and G2 phase (mono‐nuclear) cells, and are shown in pie charts. P‐values were calculated using Fisher's exact test.

Figure EV2. Time‐lapse images of AT2088‐LacI‐GFP merged with Ish1‐mCherry during M phase.

Representative images of AT2088‐LacI‐GFP (green) and Ish1‐mCherry (red) obtained in 3‐min intervals before and after the onset of anaphase. Optical section images were projected by the maximum intensity method using the SoftWoRx 5.5 software on the DeltaVision Elite system. The scale bar indicates 1 μm.

Late replication origins are localized to the nuclear periphery through a Taz1‐dependent mechanism

Because only Taz1‐dependent late origins exhibited preferential localization at the nuclear periphery, we sought to determine whether regulation of localization involved a Taz1‐dependent mechanism. We first examined whether the telomere‐like sequence, the Taz1‐binding site, which is located proximal to AT2088 and essential for replication‐timing control (Tazumi et al, 2012), was required for peripheral localization of the origin by generating a mutant for this sequence. The AT2088‐S2632 substitution in the telomere‐like sequence markedly decreased the Class 1 proportion from 79.3 to 39.6% (Fig 1F, P < 0.0001 compared to AT2088), indicating that the telomere‐like sequence is essential for the peripheral localization of the origin. Next, we examined whether Taz1 was required for the nuclear peripheral localization of Taz1‐dependent late origins. The Class 1 proportion of AT2088 decreased from 79.3% in wild‐type cells to 57.7% in taz1Δ cells (Fig 1F, P < 0.0001). The Class 1 proportion of another Taz1‐dependent late origin, ori2100, also decreased in taz1Δ cells (P < 0.0001 compared to ori2100 in wild‐type cells), whereas the localization of the early origin ars2004 did not significantly change (P = 0.91, compared to ars2004 in wild‐type cells; Fig 1F). These results indicate that Taz1 binding near the origin is required for the nuclear peripheral localization of Taz1‐dependent late origins.

AT2088 occupies a telomere‐adjacent location during G1/S phase

Because the telomeres are anchored to the inner nuclear membrane during the interphase of the mitotic cell cycle (Chikashige et al, 2009), we asked if Taz1‐dependent late origins could associate with telomeres in the nuclear periphery. To test this possibility, we analyzed the location of AT2088‐LacI‐GFP in comparison with that of Taz1‐mCherry (Fig 2A). In a single focal plane, which was selected based on the strongest LacI‐GFP signal, one to four Taz1‐mCherry foci, which corresponded to the telomeres of chromosomes I and II, were detected (Fig 2B). In some nuclei, the AT2088‐LacI‐GFP focus was close to or partially overlapped with one of the Taz1‐mCherry foci, generating yellow boundary in the merged image (Fig 2B and C).

Figure 2. G1/S phase‐specific telomeric association of late origin AT2088 .

1. Schematic images are shown for the analysis of the location of AT2088‐LacI‐GFP in comparison with Taz1‐mCherry at the telomeres. Among 21 vertical planes with 0.1‐μm intervals, a single plane that contained both a strong lacO/LacI‐GFP (green) signal and the closest Taz1‐mCherry (red) signal is chosen for analysis. For quantitative analysis of the locations of AT2088 locus in comparison with telomeres, the distance between the center of AT2088‐LacI‐GFP signal and that of the closest Taz1‐mCherry signal (telomere) in a single focal plane was measured.
2. The LacI‐GFP signal at AT2088 (left), the Taz1‐mCherry foci at the telomeres (middle), and the merged image (right) in a single focal plane are presented. The scale bar indicates 1 μm.
3. Merged images of LacI‐GFP and Taz1‐mCherry are presented as in (B). The scale bar indicates 1 μm.
4. The distances between the center of AT2088‐LacI‐GFP signal and that of the closest Taz1‐mCherry signal (telomere) were measured in a single focal plane in G1/S and G2 phase nuclei of wild‐type and AT2088‐S2632 cells (as in A). The results (μm) are presented in the scatter plot. Lines indicate means ± SD. The numbers of nuclei analyzed were n = 60 and n = 119 for G1/S and G2 phases in wild type, respectively, and n = 70 and n = 103 for G1/S and G2 phases in AT2088‐S2632 cells, respectively. P‐values were calculated using the Kruskal–Wallis test in GraphPad Prism 6. ****P < 0.0001.
5. Schematic drawings of the lacO/LacI‐GFP (AT2088) and Taz1‐mCherry (telomere) in G1/S and G2 phases based on the results of Figs 1 and 2 are presented.

For a more detailed investigation of how these localizations corresponded to the cell cycle, we took advantage of the S. pombe morphology, in which G1/S phase and G2 phase cells correspond to bi‐nuclear and single‐nuclear cells, respectively, because cytokinesis and septum formation coincide with the G1/S phase. We measured the distance between the LacI‐GFP focus and the closest Taz1‐mCherry focus in G1/S and G2 phase cells (Fig 2A and D). Interestingly, the distance was significantly smaller in G1/S phase cells than in G2 phase cells (P < 0.0001, Kruskal–Wallis test, Fig 2D). The results of time‐lapse analysis further suggested that AT2088‐LacI‐GFP locus began to associate with the telomere shortly after sister chromatids separated in the M phase, and it moved within the vicinity of the telomere through G1/S phase (Fig EV3). Taken together, the AT2088 locus associates with the telomere from late M phase to G1/S phase, and it separates from the telomere in G2 phase although both AT2088 and the telomeres remain at the nuclear periphery (Fig 2E).

Figure EV3. Time‐lapse images of AT2088‐LacI‐GFP merged with that of telomeric Taz1‐mCherry during late M phase.

The images of AT2088‐LacI‐GFP (green) and Taz1‐mCherry (telomeres, red) obtained in 3‐min intervals are presented together with bright‐field images (left panels). The right and left cells correspond to late M and G2 phase cells, respectively. Images in a single focal plane are presented. The scale bar indicates 5 μm. At 0, 9, and 12 min, AT2088‐LacI‐GFP signal (green) overlapped, at least partly, with the Taz1‐mCherry signal (red), generating yellow spot or boundary. They were found as adjacent but non‐overlapping signals at the other time points.

To examine whether the telomere‐adjacent localization of AT2088 was regulated by a Taz1‐dependent mechanism, the effect of a base substitution in the telomere‐like sequence near AT2088 was analyzed. The distance between AT2088‐S2632‐LacI‐GFP and the closest Taz1‐mCherry focus in G1/S phase increased compared to that in wild‐type cells (Fig 2D, P < 0.0001, Kruskal–Wallis test), and there was no significant difference between G1/S and G2 phases in AT2088‐S2632 cells (Fig 2D, P = 0.40). These results suggest that telomeric association of Taz1‐dependent late origins requires Taz1 binding to telomere‐like sequences near the origins.

Telomeric association and replication‐timing control of Taz1‐dependent late origins are independent of nuclear peripheral localization

Telomeres are anchored to the inner nuclear membrane in the mitotic cell cycle through the interaction of Rap1 with nuclear membrane proteins Bqt3/4 (Chikashige et al, 2009). To examine whether anchoring of telomeres to the nuclear membrane played a role in telomeric association of Taz1‐dependent late origin, we analyzed the nuclear localization of AT2088‐LacI‐GFP in comparison with that of Taz1‐mCherry in bqt4Δ cells. Taz1‐mCherry foci were found in the interior of the nucleus in bqt4Δ cells (Fig 3A), consistent with previous observations (Chikashige et al, 2009). Interestingly, AT2088‐LacI‐GFP was located in close proximity to one of the interiorly shifted telomeres in bqt4Δ cells (Fig 3A). The distance between the AT2088‐LacI‐GFP signal and the closest Taz1‐mCherry focus of bqt4Δ cells was significantly smaller in G1/S phase than in G2 phase (Fig 3B, P = 0.0027, Kruskal–Wallis test) and was similar to that in wild‐type G1/S phase cells (Fig 3B, P > 0.9999). These results suggest that the telomeric localization of Taz1‐dependent late origins is independent of telomere anchoring to the inner nuclear membrane.

Figure 3. Telomere anchoring to the nuclear membrane is not required for telomeric association of the Taz1‐dependent origin and late replication control.

1. Localization of AT2088‐LacI‐GFP in comparison with Taz1‐mCherry was analyzed in bqt4Δ cells. Representative images of a single focal section (top) with enlarged images for the regions “a” (middle) and “b” (bottom) in G1/S phase cells are shown. One of the two nuclei in the G1/S phase cell containing “a” is out of focus. The scale bar indicates 5 μm (top panels) or 1 μm (middle and bottom panels).
2. The distances between the AT2088‐LacI‐GFP focus and the closest Taz1‐mCherry in G1/S (n = 85) and G2 (n = 101) phases of wild‐type cells and G1/S (n = 68) and G2 (n = 112) phases of bqt4Δ cells are shown in the scatter plots. Lines indicate means ± SD. P‐values were calculated using the Kruskal–Wallis test. ****P < 0.0001 and **P < 0.01.
3. Representative images of lacO/LacI‐GFP (green) at the AT2088 merged with Ish1‐mCherry (red) in bqt4Δ cells are presented. The proportions of Class 1 (peripheral, blue) and Class 2 (non‐peripheral, pale blue) during G1/S and G2 phases of bqt4Δ cells together with the results of wild type (as shown in Fig EV1) are shown in the pie charts. P‐values were obtained using the Fisher's exact test. ****P < 0.0001.
4. Effect of bqt4 ⁺ deletion on replication timing of early and late origins. Wild‐type and bqt4Δ cells synchronously released from the G2/M block were labeled with BrdU for 90 min at 25°C in the presence of HU. Replication (%) was quantified as described in “Materials and Methods” by qPCR using primers for nonARS, early origin ars2004, late origins ori2100 and AT2088, and subtelomeric origin TAS59. The mean values obtained from three independent experiments are presented ±SEM.

Because AT2088 remained in close association with telomeres that changed location toward the interior of the nucleus in bqt4Δ cells, we examined whether the location of the origin relative to Ish1‐mCherry was altered (Fig 3C). The proportion of AT2088 Class 1 localization remarkably decreased in G1/S and G2 phases in bqt4Δ cells (Fig 3C, P < 0.0001 compared to wild‐type cells). Similarly, the Class 1 localization of ori2100 decreased in bqt4Δ cells (Appendix Fig S3), indicating that peripheral localization of Taz1‐dependent origins was dependent on Bqt4. Taken together, these results demonstrate that peripheral localization is not a prerequisite for the association between telomeres and Taz1‐dependent late origins.

To examine whether the peripheral localization of Taz1‐dependent origins is required for late replication‐timing control in S. pombe, we measured DNA synthesis in wild‐type and bqt4Δ cells in the presence of hydroxyurea (HU), an inhibitor of ribonucleotide reductase, which prevents replication fork progression and late origin activation. Wild‐type and bqt4Δ cells were arrested at the G2/M boundary and then synchronously released and labeled with 5‐bromo‐2′‐deoxyuridine (BrdU), a heavy‐density analog of thymidine. BrdU‐labeled heavy‐light density DNA was separated from non‐replicated light–light density DNA by CsCl density gradient centrifugation, and DNA replication (%) was quantified by qPCR as described previously (Hayashi et al, 2007). In HU‐treated wild‐type cells, the early origin ars2004 incorporated BrdU several times more than a non‐origin locus (nonARS), late replication origins ori2100 and AT2088, or the subtelomeric late origin TAS59 (Fig 3D), consistent with previous results (Tazumi et al, 2012). In bqt4Δ cells, the replication profile, in which early replication occurred only at ars2004 but not at ori2100, AT2088 or TAS59, was very similar to that in wild‐type cells (Fig 3D), demonstrating that Bqt4 does not play a role in replication‐timing control. Therefore, nuclear peripheral localization is not a crucial determinant for replication timing of Taz1‐dependent late origins and subtelomeric late origins.

Shelterin components play important roles in replication‐timing control and telomeric association of Taz1‐dependent late origins

We then investigated whether other telomere‐binding proteins, such as shelterin components, participated in replication‐timing control and telomeric association of late origins. To examine the role of Rap1 in replication‐timing control, BrdU incorporation was measured for the early origin ars2004, the Taz1‐dependent late origins AT2088 and ori2100, the Taz1‐independent late origin ars727, and the subtelomeric late origins TAS59 and tel‐0.3 (Fig 4A). In HU‐treated wild‐type cells, all of the late origins and the non‐origin region (nonARS) did not significantly replicate, while the early origin ars2004 replicated early (Fig 4A). Contrastingly, in HU‐treated taz1Δ cells, ori2100, AT2088, TAS59, and tel‐0.3 replicated to a similar extent of ars2004, whereas replication of ars727, the Taz1‐independent late origin, remained lower than that of the nonARS control (Fig 4A), consistent with previous findings (Tazumi et al, 2012). Replication at the nonARS in taz1Δ cells was higher than that in wild‐type cells, possibly due to passive replication from a weak Taz1‐dependent origin located about 15 kb from the locus. Interestingly, in HU‐treated rap1Δ cells, the Taz1‐dependent late origins AT2088 and ori2100 showed early replication, whereas subtelomeric TAS59 and tel‐0.3 and ars727 showed late replication as observed in wild‐type cells (Fig 4A). Furthermore, in poz1Δ cells lacking another component of the shelterin complex, the results were very similar to those in rap1Δ cells (Fig 4A), indicating that Rap1 and Poz1 are required for the timing control of the Taz1‐dependent origins. These results show that Rap1 and Poz1 are required specifically for replication‐timing control of Taz1‐dependent origins located in the chromosomal arm regions, but are dispensable for control of telomere‐proximal origins and the Taz1‐independent late origin.

Figure 4. Shelterin components Rap1 and Poz1 are required for replication‐timing control and telomeric association of Taz1‐dependent late origins.

1. Effects of shelterin component deletion on replication timing of late origins. Wild‐type, taz1Δ, rap1Δ, and poz1Δ cells were synchronously released from the G2/M block and labeled with BrdU for 90 min in the presence of HU. Replication (%) was analyzed as described in Fig 3. The mean values obtained from three independent experiments are presented ±SEM.
2. Localization of AT2088‐LacI‐GFP in comparison with Taz1‐mCherry was analyzed in rap1Δ cells. Representative images of a single focal section (top) with enlarged images for the regions “G1/S” (middle) and “G2” (bottom) are shown. The scale bar indicates 5 μm (top panels) or 1 μm (middle and bottom panels).
3. Distances between the AT2088‐LacI‐GFP focus and the closest Taz1‐mCherry (telomere) during G1/S (n = 60) and G2 (n = 108) phases of rap1Δ cells and G1/S (n = 64) and G2 (n = 86) phases of poz1Δ cells together with the results of wild‐type cells (as in Fig 2) are shown in the scatter plots. Lines indicate means ± SD. P‐values were obtained by the Kruskal–Wallis test. ****P < 0.0001, ***P < 0.001, and **P < 0.01.
4. Representative images of lacO/LacI‐GFP (green) at the AT2088 merged with Ish1‐mCherry (red) in rap1Δ cells. Class 1 (peripheral) and Class 2 (non‐peripheral) proportions in G1/S and G2 phases rap1Δ cells together with the results of wild type (as shown in Fig EV1) are shown in pie charts. P‐values were calculated using Fisher's exact test. ****P < 0.0001 and *P < 0.05.

Because Taz1 binding to the telomere‐like sequence near the internal late origins is essential for late‐timing control of the origins, we examined the localization of Taz1 to the AT2088 locus in wild‐type and rap1Δ cells. A chromatin immunoprecipitation (ChIP) assay using Flag‐Taz1 showed that the amount of Flag‐Taz1 localized to the AT2088 locus in rap1Δ cells was similar to that in wild‐type cells (Fig EV4). We then examined whether Rap1 was localized at the AT2088 locus to affect replication timing. The results of ChIP analysis with Rap1‐Flag showed that there was no significant localization of Rap1‐Flag at the AT2088 locus in comparison with the control locus (nonARS) at the G2/M boundary or in G1/S phase, although Rap1‐Flag was localized to the region close to the telomeres (Fig EV5). Thus, it is likely that Rap1 plays a role in replication‐timing control of AT2088 without stable localization at the locus.

Figure EV4. Localization of Flag‐Taz1 at Taz1‐dependent late origins in rap1Δ cells.

Chromatin immunoprecipitation with anti‐Flag antibody was conducted in logarithmically growing wild‐type and rap1Δ cells expressing Flag‐Taz1 as described previously (Tazumi et al, 2012). Localization of Flag‐Taz1 was analyzed by qPCR using sets of primers for nonARS (gray), AT2088 (cyan), tel‐3 (violet), and tel‐0.3 (purple). The mean values obtained from three independent experiments are presented ±SEM. P‐values in the magnified results were obtained by Sidak's multiple comparisons test in GraphPad Prism 6. **P < 0.01 and *P < 0.05.

Figure EV5. Localization of Rap1‐Flag at the proximity of the telomere but not at the Taz1‐dependent late origin.

Chromatin immunoprecipitation with anti‐Flag antibody was conducted in G2/M‐arrested rap1‐Flag cells (0 min) and those in G1/S phase (80 min after release). Localization of Rap1‐Flag was analyzed by qPCR using sets of primers for nonARS, AT2088, tel‐3, and tel‐0.3. The mean values obtained from three independent experiments are presented ±SEM. P‐values were calculated using Sidak's multiple comparisons test in GraphPad Prism 6. ****P < 0.0001.

We examined whether Rap1 or Poz1 are involved in the intra‐nuclear localization of Taz1‐dependent late origins. In the absence of Rap1, Taz1‐mCherry foci shifted toward the interior of the nucleus, as was observed in bqt4Δ cells (Fig 4B). This was consistent with the requirement of Rap1 for telomere anchoring to the inner nuclear membrane. In contrast to the results in bqt4Δ cells, the AT2088‐LacI‐GFP focus was located distant from Taz1‐mCherry foci (Fig 4B). The distance between AT2088‐LacI‐GFP and the closest Taz1‐mCherry focus in G1/S phase of rap1Δ cells was significantly larger than in wild‐type cells (Fig 4C, P = 0.0064, Kruskal–Wallis test), although the distance was smaller than in G2 phase rap1Δ cells (Fig 4C, P < 0.0001). These results suggest that Rap1 has a role, at least in part, in the close association of the origin with telomeres. Similarly, in poz1Δ cells, the distance between AT2088‐LacI‐GFP and the closest Taz1‐mCherry focus in G1/S phase cells increased compared to that in wild‐type cells (Fig 4C, P = 0.0004 and Appendix Fig S4). Intriguingly, the proportion of peripheral AT2088 (Class 1) localization in rap1Δ G1/S phase cells decreased compared to that in wild‐type cells (Fig 4D, P < 0.0001, Fisher's exact test) and was significantly smaller than in G2 phase rap1Δ cells (Fig 4D, P = 0.021). These results suggest that the shelterin components Rap1 and Poz1 have roles in the intra‐nuclear localization of Taz1‐dependent late origins primarily during G1/S phase.

A minishelterin Tpz1‐Taz1 fusion protein bypasses requirement of Rap1 and Poz1 for replication‐timing control and telomeric association of Taz1‐dependent late origins

Previous studies have demonstrated that a minishelterin, an artificial fusion of Taz1 and Tpz1, is able to regulate telomere length in the absence of Rap1 and Poz1 (Pan et al, 2015). We examined whether a minishelterin Tpz1‐Taz1 fusion protein restored replication‐timing control in rap1Δ poz1Δ cells. BrdU incorporation was measured in rap1Δ poz1Δ double‐mutant cells with or without expression of the Tpz1‐Taz1 fusion protein (Fig 5A). In HU‐treated rap1Δ poz1Δ cells, the Taz1‐dependent late origins AT2088 and ori2100 replicated efficiently as the early origin ars2004, whereas the Taz1‐independent late origin ars727 and subtelomeric late origins TAS59 and tel‐0.3 remained lower than non‐origin control (Fig 5A). In contrast, early replication of AT2088 and ori2100 was suppressed by expression of the Tpz1‐Taz1 fusion protein in rap1Δ poz1Δ cells (Fig 5A), indicating that the minishelterin restored replication‐timing control of Taz1‐dependent late origins. These results suggest that a link between Taz1 and Tpz1 is important for the replication‐timing control of Taz1‐dependent late origins located in the chromosome arm regions.

Figure 5. Minishelterin restores replication‐timing control and telomeric association of Taz1‐dependent late origins in the absence of Rap1 and Poz1.

1. Effect of minishelterin expression on the replication timing of late origins. A minishelterin containing the Tpz1‐Taz1 fusion protein is schematically presented. Wild‐type and rap1Δ poz1Δ double‐mutant cells with or without expression of the Tpz1‐Taz1 fusion protein were synchronously released from the G2/M block and labeled with BrdU for 90 min in the presence of HU. Replication (%) was analyzed as described in Fig 3D. The mean values obtained from three independent experiments are presented ±SEM.
2. Representative images are shown for AT2088‐LacI‐GFP in comparison with Taz1‐mCherry in rap1Δ poz1Δ double‐mutant cells with (bottom) or without (top) Tpz1‐Taz1 expression. The scale bar indicates 5 μm.
3. The distances between the AT2088‐LacI‐GFP focus and the closest Taz1‐mCherry (telomere) in G1/S (n = 76) and G2 (n = 103) phases of rap1Δ poz1Δ cells and in G1/S (n = 82) and G2 (n = 99) phases of cells expressing the Tpz1‐Taz1 fusion protein together with the results of wild‐type cells (as in Fig 2) are shown in scatter plots. Lines indicate means ± SD. P‐values were calculated using the Kruskal–Wallis test. ****P < 0.0001 and **P < 0.01.
4. Localization of AT2088‐LacI‐GFP in comparison with Ish1‐mCherry was analyzed in rap1Δ poz1Δ cells with or without expression of a Tpz1‐Taz1 minishelterin. Class 1 (peripheral) and Class 2 (non‐peripheral) proportions are shown in pie charts. P‐values were calculated using Fisher's exact test. ****P < 0.0001.

We examined whether the artificial fusion of Taz1 and Tpz1 also restored the telomeric association of internal late origins in the absence of Rap1 and Poz1. The distance between AT2088‐LacI‐GFP and the closest Taz1‐mCherry during G1/S phase in rap1Δ poz1Δ double‐mutant cells was significantly larger than in wild‐type cells (Fig 5B and C). In contrast, expression of the Tpz1‐Taz1 fusion protein decreased the distance between AT2088‐LacI‐GFP and the closest Taz1‐mCherry locus in rap1Δ poz1Δ cells, such that it was similar to that of wild‐type cells (Fig 5B and C). This indicated that the minishelterin restored telomeric localization of the Taz1‐dependent late origin in the absence of Rap1 and Poz1. Contrastingly, expression of the minishelterin did not restore peripheral localization of AT2088 in rap1Δ poz1Δ cells (Fig 5D), consistent with the notion that telomeric association of the internal Taz1‐dependent late origin is independent of peripheral localization (Fig 3). These results suggest that the roles of Rap1 and Poz1 are to link Taz1 and Tpz1 in the shelterin complex and that the linkage is important for telomeric localization and replication‐timing control of internal late origins.

Discussion

Replication origins on eukaryotic chromosomes are programed to initiate at specific timings during S phase (MacAlpine & Bell, 2005; Gilbert et al, 2010). In addition, replication takes place in distinct nuclear regions at specific timings, suggesting a link between temporal and spatial control of DNA replication. However, the biological significance of spatiotemporal replication and the molecular mechanisms behind its control are still unclear. In this study, we demonstrated that a subset of late replication origins regulated by Taz1 are confined to a specific intra‐nuclear space, namely in proximity to telomeres in the nuclear periphery, during G1/S phase in fission yeast. Binding of Taz1 to the telomere‐like sequences near the origins is essential for both the localization and the late‐timing control of these origins, indicating that Taz1 plays a key role in linking the temporal and spatial regulation of chromosome replication.

In wild‐type cells, Taz1‐dependent late origins in chromosomal arm regions are predominantly located at the nuclear periphery (Fig 1) and near one of the telomeres during G1/S phase (Fig 2). Interestingly, the results in bqt4Δ cells, in which telomere anchoring to the inner nuclear membrane is disrupted, indicate that the telomere‐adjacent localization and late‐timing control are independent of the nuclear peripheral localization (Fig 3). Therefore, the peripheral localization itself is not a crucial determinant of late replication control, which is consistent with previous findings in budding yeast (Ebrahimi et al, 2010). In contrast, our study showed a strong correlation between telomere‐adjacent localization and late‐timing control of Taz1‐dependent origins. Furthermore, the Taz1‐dependent origin associates with the telomere when the replication program is established (Dimitrova & Gilbert, 1999). Taken together, our results suggest that telomere‐adjacent localization plays an important role in replication‐timing control of Taz1‐dependent origins.

We have previously shown that subtelomeric late origins replicate very late in the S phase in S. pombe (Hayashi et al, 2009). This is consistent with the model that Rif1 recruits PP1, which acts as a counter‐phosphatase against the DDK‐dependent phosphorylation of Mcm2‐7 subunits and represses early replication (Hiraga et al, 2014). Thus, localization of Rif1 should be important for regulation of replication timing. Interestingly, Rif1‐EGFP exhibits a cloud‐like tailed localization around a defined focus of Taz1‐mCherry in G1/S cells (Appendix Fig S5), as reported previously (Zaaijer et al, 2016). It is possible that PP1 is enriched around the telomeres forming a phosphatase dominant “PP1‐zone” (Fig 6). Thus, tethering of Taz1‐dependent origins to telomeres during G1/S phase may inhibit replication initiation through PP1 until DDK and CDK activities increase in late S phase (Fig 6). This implies that adjacent regions to the telomere‐like sequence would be inhibited by the PP1‐zone. Indeed, the average distance of a Taz1‐dependent late origin to the nearest early origin (67 kb) is much larger than the average distance between early origins (27 kb), based on the previous results (Tazumi et al, 2012), supporting the idea.

Figure 6. Model for replication‐timing control by shelterin‐mediated telomeric association.

A model for control of Taz1‐dependent late origins by shelterin‐mediated localization near the telomeres is shown. Taz1 binds to telomeric repeats (red arrowheads) at the telomeres, recruiting Rap1 and Rif1 to the telomeres. Telomeres are anchored to the nuclear membrane through the interaction of Rap1 with nuclear membrane protein Bqt3/4 during the interphase. Rif1 recruits PP1 that acts as a counter‐phosphatase for DDK‐phosphorylation of Mcm2‐7 subunits. Subtelomeric late replication origins are strongly suppressed by PP1. Because Rif1 has a cloud‐like localization around the telomeres (Appendix Fig S5), PP1 is likely to be enriched around telomeres, forming a “PP1‐zone”. Taz1 also binds to the telomere‐like repeat (two arrowheads) near the replication origins in chromosomal arm regions throughout the cell cycle. During M phase, Taz1‐bound origins as well as telomeres dissociate from the nuclear membrane. At the end of M phase, Taz1‐bound origins associate with telomeres, possibly through interaction of Taz1 with the shelterin component Rap1. During G1/S phase, telomeres are anchored to the nuclear membrane, and Taz1‐dependent origins tethered around the telomeres are suppressed by PP1.

Shelterin components Rap1 and Poz1 are required for telomeric association and replication‐timing control of Taz1‐dependent late origins located in chromosomal arm regions (Fig 4). Interestingly, neither Rap1 nor Poz1 is required for the control of subtelomeric late origins near the telomeres (Fig 4), suggesting that Rap1 and Poz1 have a specific role in regulation of telomere‐distal origins. Because Rap1 and Poz1 are components of the shelterin complex that forms a physical bridge connecting Taz1 and Pot1 at the telomeres (Miyoshi et al, 2008; Palm & de Lange, 2008), it is possible that they contribute to tethering Taz1‐bound internal origins to the telomeres (Fig 6). Indeed, the requirement of Rap1 and Poz1 for both timing control and telomeric association of the internal late origin was bypassed by expression of a minishelterin Tpz1‐Taz1 fusion protein (Fig 5), consistent with the importance of the physical linkage of shelterin in replication‐timing control. Furthermore, timing control of the internal late origins under the expression of the minishelterin in the absence of Rap1 and Poz1 still requires Rif1 (Appendix Fig S6), suggesting that certain physical inaccessibility of initiation factors due to the Tpz1‐Taz1 fusion protein is unlikely. In addition, tethering of LacI‐GFP‐Sds21/PP1 phosphatase to the lacO repeat proximal to AT2088 in rap1Δ cells decreased early replication of AT2088 (Appendix Fig S7), suggesting that requirement of Rap1 in timing control was, at least partly, bypassed by localization of PP1 to the origin. These results are consistent with the model that telomere‐adjacent localization causes repression of Taz1‐dependent late origins, although the involvement of other mechanisms dependent on Rif1 and PP1 cannot be excluded.

Recent studies using chromosome conformation capture (3C) or Hi‐C techniques have revealed that chromatin units in spatial proximity, known as topologically associated domains (TADs), influence transcription and DNA replication (Dileep et al, 2015; Gonzalez‐Sandoval & Gasser, 2016). The close association of Taz1‐dependent late origins with the telomeres described in the present study may represent a novel type of TAD that is formed transiently during G1/S phase. It is possible that this type of TAD contributes to formation of facultative heterochromatin around Taz1‐dependent late origins, although deletion of heterochromatin protein HP1 only marginally affected replication‐timing control (Zofall et al, 2016). The actual topological association between the chromosome arm regions and telomeres was not detected in wild‐type cells, but was detected in the mutant cells of a chromatin‐associated protein Sap1 (Mizuguchi et al, 2017). This was probably due to a transient interaction between internal late origins and telomeres during G1/S phase. A cell cycle‐specific investigation may be needed to study this interaction.

In the present study, we demonstrated that telomere‐like sequences and telomere‐binding proteins play important roles in intra‐nuclear localization and replication‐timing control of a subset of late replication origins. In addition, the results obtained in this study uncovered novel functions of telomeres in higher order chromatin architectures and long‐distance control of DNA replication. Although the mechanisms involved in replication‐timing control might be diverse in various organisms, localization of origins to specific nuclear compartment may play roles in control of the initiation process. Because telomeric sequences and telomere‐binding proteins are widely conserved in eukaryotes, investigation on the functions of these features in various organisms would shed light on our understanding of roles of intra‐nuclear compartments in chromosomal functions, including gene expression, DNA replication, DNA damage repair, and genome maintenance.

Materials and Methods

Yeast strains, media, and transformations

The Schizosaccharomyces pombe strains used in this study are listed in Appendix Table S1. Yeast cells were grown in YE medium as a complete medium and EMM medium as a minimum medium (Moreno et al, 1991), supplemented with 2% glucose and appropriate nutrients. Transformation of S. pombe was performed using the lithium acetate method (Forsburg, 2003), and transformants were selected on solid YE medium containing 2% agar supplemented with appropriate antibiotics, such as geneticin (G418, Nacalai Tesque), hygromycin B (Nacalai Tesque), and nourseothricin (clonNAT, Werner BioAgents), at a final concentration of 100 μg/ml.

Visualization of replication origin loci

Replication origin loci were visualized using a lac operator (lacO)/GFP‐conjugated lac repressor (LacI) system (Robinett et al, 1996; Straight et al, 1996). The LacI protein constructed in Nabeshima et al (1998) was used in this study; 256 tandem arrays of lacO repeats were integrated at a site 100–700 bp from the replication origin using the two‐step integration method (Yamamoto & Hiraoka, 2003). First, a DNA fragment containing a partial ura4 gene was integrated at the targeted locus, and the lacO tandem repeats were subsequently integrated at the locus, thus generating the ura4 ⁺ sequence. To integrate the partial ura4 gene at the early or late origin, pCT33 h was constructed by replacing the MluI‐SacI fragment containing the kanamycin‐resistant (kanMX6) gene of pCT33‐6 with the MluI‐SacI fragment containing the hygromycin‐resistant (hphMX6) gene from pFA6a‐hphMX6. Subsequently, a DNA fragment containing both the partial ura4 gene and the hphMX6 gene was amplified by PCR using two DNA primers from pCT33 h. The primers used to amplify pCT33 h are listed in Appendix Table S2. Integration of the amplified DNA fragment was executed at a targeted origin locus of LacI‐GFP‐carrying cells (HM5833: h ⁺ his7 ⁺::lacI‐GFP lys1‐131 ura4‐D18), which was obtained by standard genetic crosses between HM643 and YHS119. pCT31‐13 containing the lacO tandem repeats was linearized by StuI and introduced into the cell line containing the partial ura4 gene at the target site. Ura⁺ transformants were selected, and the insertion of lacO repeats was confirmed by PCR.

A fission yeast strain expressing Ish1‐mCherry and LacI‐GFP was obtained using standard genetic crosses between HMP16 and HM5833, resulting in HM5632. To construct the strain expressing Taz1‐mCherry, two short fragments were amplified by PCR from the genomic DNA of S. pombe wild‐type strain 972 using two sets of primers, #1744 and #3069; and #3042 and #1776 (Appendix Table S2). The products were used as primers for the amplification of taz1‐mCherry‐kan from pFA6a‐mCherry‐kanMX6. The PCR products were used for integrating at the taz1 ⁺ locus, and G418‐resistant transformants were selected.

Construction of fission yeast strains

HM3488, HM5023, HM5189, HM5760, HM5790, and HM5990, derivatives carrying cdc25‐22 P _nmt1 ‐TK P _adh1 ‐hENT, were obtained by standard genetic crossing. The ish1‐mCherry taz1Δ derivatives were generated using standard genetic crossing of HM5731, HM5541, HM5726, or HM5688 with HM5027.

For construction of the bqt4Δ strain, a 3.9‐kb fragment containing the bqt4 ⁺ gene was amplified by PCR from the genomic DNA of wild‐type strain 972 using the primers #3542 and #3543. Then, the upstream region amplified from the 3.9‐kb fragment using primers #3548 and #3549 was digested by HindIII and SalI and cloned into the HindIII‐SalI sites of pFA6a‐natMX6 to generate pBQ5′. Next, the downstream region of the bqt4 ⁺ gene was amplified from the 3.9‐kb fragment using the primers #3550 and #3551, digested by SacI and EcoRV, and then cloned into the SacI‐EcoRV sites of pBQ5′, resulting in pFA6a‐Bqt4d‐natMX6. The SacI‐EcoRV fragment of pFA6a‐Bqt4d‐natMX6 was used for transformation of HM5805 and HM5541, resulting in HM5835 and HM6022, respectively.

For construction of the minishelterin tpz1‐taz1 rap1Δ poz1Δ strain, the artificial tpz1‐poz1 fusion gene was amplified by PCR from the genomic DNA of the PP993 strain using primers #3575 and #3576, and used for transformation of HM5790, resulting in HM5991. The derivatives of taz1‐mCherry carrying rap1Δ or poz1Δ were constructed by standard genetic crossing of HM5805 with HM5025 resulting in HM5860, and HM5805 with HM3578 resulting in HM5861.

Live cell imaging

Schizosaccharomyces pombe cells grown in YE medium at 30°C to 1 × 10⁷ cells/ml were immobilized on a glass‐bottom dish (MatTek Corp.) coated with lectin (0.2 mg/ml, Sigma) in EMM medium. A DeltaVision Elite fluorescence microscopy system (GE Healthcare Life Science), which is based on an Olympus wide‐field IX71 fluorescence microscope and equipped with a CoolSNAP HQ2 CCD camera (Photometrics), was used for live cell imaging. Live cell images were obtained using an oil‐immersion objective lens (Plan Apo 60×; NA = 1.4; Olympus) at 30°C in a temperature‐controlled room. Optical section images were processed with 3D deconvolution and analyzed using the SoftWoRx 5.5 software on the DeltaVision Elite system (GE Healthcare Life Science).

BrdU incorporation and replication timing analysis

HM1865, HM3488, HM5023, HM5189, HM5769, or HM 5932 cells carrying P _nmt1 ‐TK P _adh1 ‐hENT cdc25‐22 were grown in EMM medium lacking thiamine to induce transcription of the TK gene at 25°C (1 × 10⁷ cells/ml). Cells were further incubated at 36°C for 3 h to arrest at the G2/M boundary due to the cdc25‐22 mutation, and the cell cycle was synchronously restarted at 25°C in the presence of 200 μM BrdU and 10 mM HU for 90 min. The preparation of cellular DNA and separation of BrdU‐labeled DNA from unreplicated DNA by CsCl density gradient centrifugation were carried out as previously described (Hayashi et al, 2007). The amount of DNA in the heavy–light and light–light densities was analyzed using qPCR with the primers listed in Appendix Table S2.

Chromatin immunoprecipitation assay

ChIP assays were performed as described previously (Handa et al 2012). Immunoprecipitation was conducted using Dynal magnetic beads (Invitrogen) associated with mouse anti‐FLAG (1:500, M2 monoclonal; Sigma). The immunoprecipitated and total DNAs were quantitated by qPCR with SYBR green 1 in Applied Biosystems StepOne real‐time PCR system (Life technologies). The primers used for qPCR are listed in Appendix Table S2.

Statistical analysis

Statistical analysis was performed using the Fisher's exact test for independence of two nominal variables with small sample sizes to compare the nuclear peripheral localization, or using the Kruskal–Wallis test, a non‐parametric method for testing whether samples originate from the same distribution, to compare telomere association in “GraphPad Prism 6” (GraphPad Software, Inc.), with P < 0.05 considered as significant. The middle bar indicates the mean distance between the LacI‐GFP and the closest Taz1‐mCherry foci, and the top and bottom bars indicate SD.

Author contributions

SO, TST, TN, YH, and HM participated in the design of the study. SO, SK, TH, and HM participated in molecular genetic and molecular imaging studies. SO, HO, HA, and YH contributed to analysis and interpretation of the fluorescent microscope images in YH laboratory. SO and HM drafted the manuscript, and HM participated in coordination of the study. All authors read and approved the final manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Review Process File

The EMBO Journal (2018) 37: e98997