Abstract
Chromosome replication and transcription occur within a complex nuclear milieu whose functional subdomains are beginning to be mapped out. Here we delineate distinct domains of the fission yeast nuclear envelope (NE), focusing on regions enriched for the inner NE protein, Bqt4, or the lamin interacting domain protein, Lem2. Bqt4 is relatively mobile around the NE and acts in two capacities. First, Bqt4 tethers chromosome termini and the mat locus to the NE specifically while these regions are replicating. This positioning is required for robust chromatin replication. Second, Bqt4 mobilizes a subset of Lem2 molecules around the NE to promote pericentric heterochromatin maintenance. Opposing Bqt4-dependent Lem2 mobility are factors that stabilize Lem2 beneath the centrosome, where Lem2 plays a crucial role in kinetochore maintenance. Our data prompt a model in which Bqt4-rich nuclear subdomains are ‘safe zones’ in which collisions between transcription and replication are averted and heterochromatin is reassembled faithfully.
Introduction
Within the nucleus, specific chromosome regions tend to localize to the periphery where they contact components of the nuclear envelope (NE) (Bickmore, 2013). Peripherally positioned chromatin tends to be enriched with repressive histone modifications associated with transcriptional silencing, while actively transcribing gene-rich regions tend to reside at internal positions within the nucleus. However, nuclear pore complexes (NPCs) at the periphery transiently host transcriptionally active regions (Capelson et al., 2010; Ishii, Arib, Lin, Van Houwe, & Laemmli, 2002; Lemaitre & Bickmore, 2015). Therefore, the NE appears to harbor at least two functionally distinct domain types.
Heterochromatic regions generally contain hypoacetylated nucleosomes with relatively low turnover rates, and histone H3 that is di-or tri-methylated at Lysine 9 (H3K9me) (Aygun, Mehta, & Grewal, 2013; Rea et al., 2000). Chromatin regions enriched with H3K9Me recruit protein complexes that suppress gene transcription; crucial to this network are histone deacetylases (HDACs) that maintain nucleosome hypoacetylation. Studies in the fission yeast S. pombe have been seminal to our understanding of conserved mechanisms underlying heterochromatin formation and function (Allshire & Ekwall, 2015; Grewal & Jia, 2007; Martienssen & Moazed, 2015; Motamedi et al., 2008; Woolcock, Gaidatzis, Punga, & Buhler, 2011). In S. pombe, Clr4 is the sole histone H3K9 methyltransferase and is critical for establishment of heterochromatin (Nakayama, Rice, Strahl, Allis, & Grewal, 2001; Ragunathan, Jih, & Moazed, 2015). Chromatin enriched with H3K9Me-containing nucleosomes provides a binding platform for HP1 family chromodomain proteins, including Swi6. HP1-mediated recruitment of SHREC, which harbors both a histone deacetylase (HDAC) and a Snf2-like nucleosome remodeler, (Sugiyama et al., 2007) creates a positive feedback loop to ensure recursive H3K9 methylation, recruitment of HP1 and spreading of SHREC along the silenced DNA region (Motamedi et al., 2008; Zhang, Mosch, Fischle, & Grewal, 2008). The conserved JmjC protein Epe1 is a critical regulatory factor that prevents repressive marks from spreading beyond the limits of the heterochromatic domain (Shimada et al., 2009; Trewick, Minc, Antonelli, Urano, & Allshire, 2007; Zofall & Grewal, 2006).
The nuclear RNAi pathway triggers heterochromatin assembly at several loci. Transcripts of such loci are converted to dsRNA by the RNA dependent RNA polymerase or convergent transcription, creating substrates for the Dicer (Dcr1) ribonuclease, which cleaves dsRNA to produce siRNAs. Once loaded into the Argonaute complex (RITS), these siRNAs target RITS to the chromosome via homology to nascent transcripts, eliciting H3K9 methylation, HP1 binding and transcriptional silencing. Dicer also acts to evict RNA polymerase II (RNAP2) from highly transcribed regions and from repetitive DNA regions like the tDNA, rDNA and pericentric repeats. Dicer-mediated release of RNAP2 promotes genome stability by limiting collisions between the transcription and replication machineries (Kloc, Zaratiegui, Nora, & Martienssen, 2008; Zaratiegui et al., 2011).
Centromeres are among the most prominent heterochromatic regions. The AT-rich central core region of centromeres, upon which the kinetochore is assembled, is flanked by heterochromatic repeats (Takahashi et al. 1992; Steiner et al. 1993; Wood et al. 2002) comprising the innermost repeats (imr) and the outer repeats (otr). This organization, in which a core platform for kinetochore assembly is flanked by heterochromatic repeats, is conserved in vertebrates (Kniola et al., 2001). HP1-containing pericentric heterochromatin recruits cohesin to the outer repeats (Nonaka et al. 2002), thereby ensuring faithful chromosome segregation (Bernard et al., 2001). Pericentric heterochromatin also plays a crucial role in the de novo assembly of functional CenpA-containing centromeres capable of kinetochore assembly (Folco, Pidoux, Urano, & Allshire, 2008b). Once a functional centromere is assembled, however, heterochromatin becomes dispensable for kinetochore maintenance.
Heterochromatin also extends inwards from chromosomal termini, for ~30 kb in S. pombe. Reporter genes inserted at these terminal regions are transcriptionally silenced (Cooper, Nimmo, Allshire, & Cech, 1997; Kanoh, Sadaie, Urano, & Ishikawa, 2005). Telomeric DNA, consisting of double-stranded (ds) G-rich repeats with a terminal single-stranded (ss) G-rich 3’ overhang, provides the platform for a sextet of proteins known collectively as shelterin (de Lange, 2009), which regulates telomerase activity and prevents chromosome ends from being treated as DNA double strand breaks. Heterochromatin is not required for these crucial telomere functions (Khair, Subramanian, Moser, & Nakamura, 2010; Tuzon et al., 2004). However, in survivors of telomerase loss, heterochromatin is critical for either protecting chromosome ends via ‘HAATI’ (Jain, Hebden, Nakamura, Miller, & Cooper, 2010) or conversely, for inhibiting the ‘ALT’ pathway of continual break induced replication characterized in telomerase-negative cancers (Benetti et al., 2008; Benetti et al., 2007). The fission yeast shelterin component Taz1 (ortholog of human TRF1 and TRF2) binds directly to telomeric dsDNA; Rap1 binds Taz1 (and, at least in the absence of Taz1, weakly binds other shelterin proteins) and directly interacts with the inner NE protein Bqt4 (Chikashige et al., 2009), promoting telomere localization to the NE. Telomere tethering is also promoted in a Bqt4 independent manner by the conserved Fun30 chromatin remodeler Fft3 (Steglich et al., 2015). Nonetheless, telomeres are not always tethered; for instance, dislodgment of telomeres from the NE prior to mitosis appears to contribute to accurate chromosome segregation (Fujita et al., 2012).
In addition to Bqt4 and the NPCs, other key fission yeast NE complexes include the ‘linker of nucleoskeleton and cytoskeleton complex’ (LINC) and the Lap2/Emerin/Man1 (LEM) family of lamin-associated proteins (Gonzalez, Saito, & Sazer, 2012). LINC comprises the conserved SUN-(Sad1) and KASH-domain (Kms1/Kms2) proteins that form a trans-NE link connecting the nucleoplasm to the cytoplasm. Kms2 interacts with the centrosome (the spindle pole body or SPB), which resides on the outer surface of the NE. The nucleoplasmic extension of LINC (Sad1) interacts with centromeres, which are clustered at the NE beneath the SPB (Fernandez-Alvarez, Bez, O'Toole, Morphew, & Cooper, 2016; Hagan & Yanagida, 1995; Scherthan, Sfeir, & de Lange, 2011). Two LEM proteins, Lem2 and Man1, have been shown to fulfill lamin-like functions despite the absence of canonical nuclear lamina in fission yeast (Cai et al., 2001; Gonzalez et al., 2012; Lee et al., 2001). Man1 has been shown to interact with regions 50-100 Kb from telomeres (Steglich, Filion, van Steensel, & Ekwall, 2012), while Lem2 appears to be sequestered near centromeres (Hiraoka et al., 2011). Recent studies have identified Lem2 as a regulator of heterochromatin silencing at pericentric regions (Barrales, Forn, Georgescu, Sarkadi, & Braun, 2016; Tange et al., 2016).
Notwithstanding the tendencies of specific chromatin regions to associate with the NE, the functions of this association in most instances remain to be elucidated. We identified Bqt4 in a screen (to be described elsewhere) for genes involved in maintaining the HAATI mode of survival in the absence of telomerase (Jain et al, 2010), in which heterochromatic rDNA repeats jump to all chromosome ends and acquire end protection capacity. In exploring the function of Bqt4 in wild type (wt) cells, we uncovered two classes of roles for Bqt4, one in directly regulating a subset of chromatin regions specifically during S phase, and another in mobilizing Lem2, which in turn has distinct roles in maintenance of the pericentric heterochromatin and of the centromeric central core. Along with the S-phase specificity of Bqt4's role in chromatin localization, we report several bqt4Δ phenotypes and genetic interactions that implicate Bqt4-rich microdomains as specialized ‘safe zones’ that prevent collisions between transcription and replication machineries, and thereby confer accurate maintenance of the heterochromatic state through successive generations.
Results
NE proteins organize into distinct domains
To chart the organization of key proteins at the NE, we used super-resolution fluorescence imaging (OMX system) (Dobbie et al., 2011) to view endogenously tagged and functional Bqt4, Man1 and Lem2 (see Methods). Consistent with previously described electron microscopy images (Chikashige et al., 2009), we observe that Bqt4 does not uniformly distribute around the NE but rather appears in puncta, indicating spatial heterogeneity in NE composition (Fig 1A). To address the composition of regions that lack detectable Bqt4 (dark areas in Fig 1A), we simultaneously imaged Bqt4 and Man1 (Fig 1B) or Lem2 (Fig 1C). Occasionally, Bqt4 and Man1 signals overlap but more often, they do not (Fig 1C, Fig S1A). Domain partitioning is also seen when simultaneously viewing Bqt4 and Lem2 (Fig 1C, Fig S1A).
The most intense Lem2 pool appears just beneath the SPB, while a less prominent subset of Lem2 molecules localizes to a ring around NE (Fig 1D). Remarkably, the ring of Lem2 is diminished to near-invisibility in cells lacking Bqt4, while a single Lem2 focus remains (Fig 1E, Fig S2B, and Fig S3), colocalizing with the SPB component Pcp1 (Fig S2B) (Gonzalez et al., 2012). In contrast, the localization patterns of the NPC protein Nup107 (Bai et al., 2004) (Fig S4), the NE component Cut11 (West, Vaisberg, Ding, Nurse, & McIntosh, 1998) (Fig S4), and Man1 (Fig S4) are all independent of Bqt4. Hence, Bqt4 is specifically required for distribution of Lem2 around the NE.
Bqt4 is highly mobile within the NE
The distinct localization patterns for Bqt4, Man1 and Lem2, along with the observation that Bqt4 controls the mobility of Lem2 around the NE, suggest that protein dynamics differ among these NE components and in different NE regions. Indeed, variable levels of dynamicity have previously been reported for proteins in the NE (Ellenberg et al., 1997; Rabut, Doye, & Ellenberg, 2004). To explore this possibility, we used fluorescence recovery after photobleaching (FRAP) to assess turnover rates. Fig 2A shows a representative nucleus from cells harboring endogenously tagged Bqt4-GFP. After capturing a reference ‘pre-bleach’ image, we photobleached a region of interest (ROI) encompassing a segment of the NE (Fig 2A). Bqt4-GFP signal inside the photobleached window recovers within ~12 seconds (Fig 2B; half-life 5.029 sec). Moreover, the reappearance of Bqt4-GFP signal within the ROI consistently shows a distinct recovery pattern; rather than uniformly recovering throughout the bleached NE segment, the signal appears progressively from the flanking NE edges to the center of the bleached region (Fig 2A), suggesting that dissociated Bqt4-GFP is replenished by movement around the NE. While Bqt4 in the unbleached portion of the NE may be continually replenished from nucleoplasmic pools, the described recovery pattern cannot be explained by exchange with a non-membrane bound pool of Bqt4.
The substantial reduction in Lem2-GFP levels encircling the NE in bqt4Δ cells prevents the robust use of FRAP for measuring Lem2 turnover around the NE in this setting. However, we were able to measure the turnover rate of Lem2-GFP localized beneath the SPB in both wt and bqt4∆ cells. At this site, loss of Bqt4 does not affect the half life of Lem2 residency (Fig 2C; Half life, wt: 23.11 sec, bqt4∆: 23.99 sec). Hence, the two distinct pools of Lem2, beneath the SPB and around the NE, are controlled by different mechanisms, with Bqt4 controlling the pool of Lem2 that is distributed away from the SPB region around the NE.
Lem2 localization beneath the SPB is regulated by the LINC-interacting protein Csi1
As the most prominent fluorescent signal of Lem2 is observed beneath the SPB, we conjectured that the LINC complex controls Lem2 accumulation at this site. Csi1 (chromosome segregation impaired protein 1) associates with Sad1 and promotes the robust association of centromeres with the LINC complex (Hou et al., 2012), making Csi1 a candidate regulator of Lem2 localization to this region. While the intense Lem2 focus colocalizes with the SPB component Sid4 in wt cells (Chang & Gould, 2000), this colocalization is abolished in 85% of csi1∆ cells as evinced by the absence of a Lem2-GFP dot beneath the SPB (Fig 3A). Notably, however, the ring of Lem2-GFP around the NE remains in csi1∆ cells. Therefore, Csi1 either recruits or stabilizes Lem2 beneath the SPB.
As Bqt4 promotes the distribution of Lem2 around NE while Csi1 concentrates Lem2 beneath the SPB, we wondered where Lem2 would localize when both these factors are missing. In cells lacking both Bqt4 and Csi1, Lem2-GFP is greatly diminished from the ring around the NE but still strongly detectable beneath the SPB (Fig 3B). However, FRAP analysis revealed that the turnover rate of Lem2 beneath the SPB in csi1∆ bqt4∆ cells is substantially increased relative to that seen in wt or bqt4∆ cells (Fig 3C). Hence, while the initial Lem2 recruitment to the LINC region beneath the SPB is controlled independently of Bqt4 and Csi1, Csi1 is required to stabilize Lem2 at this region; this stabilization is required to counterbalance the Bqt4-dependent mobilization of Lem2 away from the LINC region.
Bqt4 tethers telomeres and the mat locus specifically while they are replicating
Bqt4 has been implicated in the interphase positioning of telomeres, which form two to three clusters at the NE (Funabiki, Hagan, Uzawa, & Yanagida, 1993) (Chikashige et al., 2009). To assess whether telomeres localize specifically to Bqt4-containing regions, we visualized telomeres in cells harboring the various tagged NE proteins (Fig 3D-F). Consistent with previous observations (Dehe, Rog, Ferreira, Greenwood, & Cooper, 2012; Funabiki et al., 1993), we observe two to three Taz1 foci in each nucleus. Figure 3D shows a single focal plane that contains a single Taz1 focus; this telomeric focus colocalizes with Bqt4-GFP (Fig 3D). In contrast, telomeres show markedly lower degrees of colocalization with Man1 or Lem2 (Figs 3E, 3F and Fig S1B). Therefore, telomeres have a higher propensity to localize to Bqt4-rich than Man1- or Lem2-rich NE domains.
Centromeres localize to the NE during interphase, forming a cluster beneath the SPB (Funabiki et al., 1993). The intense focus of Lem2 detectable in this region (Fig 4A) is consistent with ChIP experiments showing an interaction between Lem2 and pericentric heterochromatin (Barrales et al., 2016; Tange et al., 2016). To investigate the exclusivity of centromere-Lem2 colocalization, we visualized endogenously mCherry-tagged Mis6, an inner kinetochore component (Takahashi, Chen, & Yanagida, 2000). Consistent with previous observations (Hiraoka et al., 2011), Mis6 sequesters within a Lem2-enriched domain beneath the SPB (Fig 4A, Fig S1C; quantified in Fig S1D). In contrast, Mis6 does not colocalize with Man1 (Fig 4B, quantified in Fig S1D). To observe the centromeric DNA sequences immediately surrounding the kinetochore region, we imaged cells that contain mCherry-tagged tet repressor (tetR-mCherry) and an array of tetracycline operators (tetO) inserted within the imr of centromere I. The marked imr and the SPB (Sid4-GFP) colocalize in 100% of cells (Fig S1E). We also imaged cells that contain LacI-CFP and an array of Lac operators (LacO) integrated ~10 Kb away from the otr of centromere III (Ding, Yamamoto, Haraguchi, & Hiraoka, 2004). In contrast to the central imr, this distal otr region colocalizes with the SPB in only 12% of cells (Fig S1E). Hence, the central core region and imr localize exclusively to the SPB region, while the pericentric otr sequences associate with NE regions away from the SPB. Collectively, the observations reveal a NE with distinct microdomains that are exclusively enriched with Man1, Lem2 or Bqt4, and interact with distinct chromosomal regions.
The specificity of Bqt4-mediated NE localization for telomeres and mat might be envisioned to stem from their shared heterochromatic nature (Cooper et al., 1997). However, H3K9Me-containing heterochromatin is dispensable for telomeric tethering to the NE, as Tel1L is not dislodged from the NE in clr4Δ cells (Fig S2A). Hence, although clr4+ deletion was reported to delocalize the mat locus, telomeres are tethered to the NE in a H3K9Me-independent manner. This is consistent with the observation that taz1Δ telomeres, in which silencing and H3K9 methylation are vastly reduced, nonetheless retain NE localization (Zaaijer, Shaikh, Nageshan, & Cooper, 2016).
The distinct localization patterns of telomeres and centromeres with respect to Bqt4 and Lem2 prompted us to explore the functional significance of Bqt4- and Lem2-microdomains. First, we examined telomere positioning as a function of cell cycle stage using cell shape as indicator. Fission yeast cells grow during G2 and reach the size threshold for S-phase initiation by the time mitosis is complete. Hence, binucleate cells (at the stage between mitotic nuclear division and cytokinesis) are in early/mid S-phase. As cells progress from late S phase to G2, cytokinesis gives rise to two mononucleate daughter cells that complete G2 and enter mitosis. Accordingly, we scored interphase binucleate cells as early/mid-S phase and mononucleate cells as late S/G2. Telomere 1-left (Tel1L) was visualized by a LacO array integrated at the sod2+ locus (Ding et al., 2004) in the proximal subtelomeric region (~50 Kb from chromosome end) in cells expressing LacI-GFP (Ding et al., 2004), hereafter referred to as Tel1L lacO/I-GFP. The NE was viewed by tagging the NE protein Cut11 with mCherry to allow measurement of the distance between Tel1L-lacO/I-GFP and the NE. The peripheral zone was defined as the outer third of the nuclear volume; a randomly positioned sequence will localize to this zone in ~33% of cells (Meister, Gehlen, Varela, Kalck, & Gasser, 2010). In 80% of wt settings, Tel1L localizes to the nuclear periphery (Fig 4C) whether early/mid S-phase or late S/G2. In contrast, this localization is disrupted in a cell cycle dependent manner in bqt4∆ cells. While Tel1L is at the periphery in ~60% of early/mid-S phase bqt4∆ cells, dislodgement is substantial in late S/G2 phase, with only ~48% localization at the NE (Fig 4C). Hence, the role of Bqt4 in localizing telomeres to the NE is more pronounced during late S/G2 than in early S phase.
As Lem2 is lost from SPB-distal regions of the NE in a bqt4Δ setting, we examined telomere positioning in lem2∆ cells. In contrast to bqt4Δ cells, Tel1L remains at the periphery throughout the cell cycle in lem2Δ cells (Fig 4C). Hence, loss of tethering in a bqt4Δ setting is not a function of losing Lem2 from telomeric tethering sites. In contrast, centromeres remain localized beneath the SPB in the absence of either Bqt4 or Lem2 (data not shown). Therefore, any role of Lem2 in tethering chromosomal loci is redundant with other mechanisms (Barrales et al., 2016).
The mating type locus (mat) is another prominent genomic region that positions at the NE (Alfredsson-Timmins, Henningson, & Bjerling, 2007; Noma, Cam, Maraia, & Grewal, 2006). To determine whether this localization requires Bqt4, we utilized a strain in which the lacO/I array is inserted at mat. While peripheral localization was observed during both S and G2-phase in wt cells, mat is dislodged from the NE in early/mid-S phase in bqt4Δ cells (Fig 4D). This dislodgment is transient, as mat re-localizes to the NE in late S/G2. In contrast, a control strain harboring a lacO/I array adjacent to a euchromatic locus (cut3) on the left arm of chromosome II (Chr II) (Ding et al., 2004) shows random positioning throughout the nucleus in both wt and bqt4∆ settings (Fig 4D).
The mat locus replicates in early S phase, whilst telomeres replicate in late S phase (Hayano et al., 2012; Hayashi, Takahashi, Nakagawa, Nakayama, & Masukata, 2009; Kim, Dubey, & Huberman, 2003). This temporal discord in the replication of telomeres versus the mat locus correlates with the temporal discord in dislodgment of these loci from the nuclear periphery in a bqt4∆ setting. Hence, we postulated that Bqt4 is necessary for tethering these loci specifically while they undergo DNA replication. To assess this possibility, we abolished the late specificity of telomere replication by deleting the gene encoding Taz1; in a taz1Δ background, telomeres replicate throughout the cell cycle (Dehe et al., 2012; Tazumi et al., 2012). We visualized taz1Δ telomeres by GFP tagging the ss telomere binding protein Pot1, which retains telomere localization in the absence of Taz1 (Baumann & Cech, 2001; Miyoshi, Kanoh, Saito, & Ishikawa, 2008). In a taz1+ background, Pot1-GFP foci localizes to the NE in both early S and late S/G2 (Fig 4E) and becomes dislodged in late S/G2 upon bqt4 deletion. In contrast, in a bqt4∆ taz1∆ setting in which telomere replication timing is deregulated, telomere dislodgement occurs throughout all stages monitored (Fig 4E). Taz1 also promotes the passage of replication forks through the repetitive telomeric DNA (Miller, Rog, & Cooper, 2006); hence, whilst absence of Taz1 advances the initiation of telomere replication, it also hinders completion of replication (Miller et al., 2006); consistently, telomeres remain dislodged throughout late S/G2 phase in a bqt4∆ taz1∆ background.
Swi6 binds heterochromatic DNA at the mat locus, and causes early replication by promoting the recruitment of the replication initiation factor Sld3 (Hayashi et al., 2009). While Swi6 also binds telomeric heterochromatin, the replication timing of telomeres is set by Taz1 independently of heterochromatin/Swi6. The specific effect of Swi6 allowed us to examine the correlation between tethering of the mat locus and its replication status. In swi6∆ cells, or in clr4Δ cells that lack H3K9Me and therefore lack Swi6 binding sites, mat remains tethered to the NE (Fig 4F), eliminating the possibility that Swi6 itself may affect tethering. However, in swi6∆ bqt4∆ cells, the mat locus is dislodged later, during late S/G2 phase (Fig 4F). Thus, Bqt4 is necessary for tethering of telomeres and the mat locus while these tethered loci are replicated.
The subnuclear positioning of replicating DNA has been visualized by fluorescent tagging of replication factors, such as the catalytic subunit of DNA polymerase α (Pol α) (Meister, Taddei, Ponti, Baldacci, & Gasser, 2007; Natsume et al., 2008), which localize to foci proposed to be the sites of active DNA replication. One or two such replication foci are observed at the NE during late S/G2 (Meister et al., 2007; Natsume et al., 2008). To examine whether telomeres colocalize with these late replication foci, we used cells expressing both Taz1-mCherry and Pol α-GFP. Figure 4G shows a representative nucleus in late S/G2. Three Taz1 foci of similar intensity are visible in the shown z-plane; a fourth Taz1 focus appears relatively less intense as it represents an out-of-focus telomere in a different z-plane. Using the fluorescence background in the nucleus to approximate its boundary, we scored the position of the colocalized dots relative to the nuclear periphery (Fig 4H). Consistent with published data (Meister et al., 2007), we observe significant colocalization between telomeres and replication foci at the NE in wt cells. In contrast, while Taz1 and Pol α show colocalization in the absence of Bqt4, this colocalization does not occur at the NE (Fig 4H), again indicating that the tendency of telomeres to replicate at the periphery requires Bqt4-mediated tethering.
Bqt4 is required for resistance to DNA damage agents during S-phase
The role of Bqt4 in localizing replicating telomeres and mat to the NE raises the question of whether such localization is important for their replication. To address this, we examined the effects of bqt4+ deletion on DNA replication-related processes. Bqt4 is required for cellular resistance to agents that induce DNA damage during S-phase, as bqt4Δ cells are hypersensitive to both hydroxyurea (HU) and methyl methanesulfonate (MMS; Fig 5A); in contrast, Bqt4 is dispensible for resistance to bleomycin, an agent that induces DNA damage independently of DNA replication (Manolis et al., 2001) (Fig 5B). These results demonstrate a requirement for either Bqt4 per se and/or for Bqt4-mediated tethering to the NE for robust DNA replication. To delineate these possibilities, we utilized strains harboring a deletion of the transmembrane domain of Bqt4 (Bqt4-∆TM); Bqt4ΔTM has been shown to delocalize from the NE and diffuse throughout the nuclear interior (Chikashige et al., 2009). Bqt4-ΔTM confers similar HU and MMS hypersensitivity to those of bqt4+ deletion (Fig S5A), indicating that the NE localization of Bqt4 is required for resistance to agents that hinder DNA replication.
We wondered if the DNA damage sensitivity of bqt4Δ cells stems from the heterochromatic nature of Bqt4-tethered regions, in which case the abolition of H3K9 methylation conferred by clr4+ deletion would suppress the damage sensitivity. However, clr4Δ confers synthetic lethality with bqt4Δ on MMS (Fig 5C). Therefore, Bqt4 and Clr4 play separate, redundant roles in controling MMS resistance. This suggests a scenario in which localization becomes particularly crucial when Bqt4-interacting chromosome regions are transcriptionally derepressed; enhanced transcription at these difficult-to-replicate regions may lead to excessive transcription/replication conflicts, causing severe MMS hypersensitivity.
Bqt4 and Lem2 microenvironments regulate telomeric and centromeric heterochromatin
The hypersensitivity of bqt4Δ cells to agents that challenge DNA replication, along with the heterochromatic nature of the Bqt4-tethered chromosome regions, raised the question of whether heterochromatin is altered by Bqt4 loss. To assess this, we created strains in which chromatin was induced to position away from both the Lem2 and Bqt4 microdomains. Given our observation that Lem2 concentrates beneath the SPB while Man1 and Bqt4 occupy distinct and largely non-overlapping positions around the NE, we sought to move replicating chromatin to Man1-enriched regions. An ectopic tethering system was achieved by fusing the C-terminus of Man1 with the GFP-binding protein (GBP) (Dodgson et al., 2013), which was in turn fused to mCherry. Man1-GBP-mCherry localizes around the NE (Fig 6A), as does unmodified Man1. To tether replicating DNA to Man1-GBP-mCherry, we utilized Pol α-GFP (Damagnez, Tillit, de Recondo, & Baldacci, 1991), which fully complements wt Pol α function (Meister et al., 2007). In cells lacking GBP, Pol α-GFP shows diffuse localization throughout the nucleus during early S-phase (Fig 6B). In contrast, Pol α-GFP accumulates around the NE in S-phase cells harboring Man1-GBP (Fig 6A). To examine the location of active DNA replication, we visualized the single-strand DNA binding protein Rad11 tagged with mCherry, which localizes to replication forks (as well as DNA damage sites). In cells lacking GBP, multiple Rad11 foci are visible throughout the nucleus during S phase while Rad11 signal is diffuse in nonreplicating cells (Fig S6A-C) (Zaaijer et al., 2016). By contrast, in nearly all Man1-GBP/Pol α-GFP cells, the Rad11 foci colocalize with peripherally localized Pol α-GFP (Fig S6D), suggesting that the bulk of DNA replication takes place at the Man1 microdomains. Hence, this approach enabled us investigate the effect of inducing the bulk of DNA replication to occur at the periphery, but away from both Lem2 and Bqt4 microdomains.
To detect changes in heterochromatin enriched with H3K9Me, we used a modified ChIP-Seq protocol that entails spiking every sample with chromatin from a strain in which the endogenous ade6+ gene has been deleted and ectopically inserted within the pericentric heterochromatin of centromere I. As all the experimental strains used contain only the single endogenous copy of ade6+, which lies in a euchromatic region (Cam et al., 2005), the signal emanating from ectopic ade6+ serves as a constant reference signal. This approach to provide normalization of H3K9Me enrichment levels is similar to the previously described ChIP-Rx method (Orlando et al., 2014). We mixed a liquid culture of the reference (spike) strain at 1:4 ratio with each sample of the experimental cultures (Fig 6C). The resulting spiked samples were then cross-linked and processed for ChIP. Fig 6D shows the enrichment of H3K9Me over the ade6 open reading frame, confirming our ability to detect the spike reads at the indicated spike ratio. To validate this normalization approach, we compared three independent wt strains with three independent dcr1Δ strains known to have vastly reduced levels of pericentric H3K9Me (Hall et al., 2002; Volpe et al., 2002). After normalizing to the reads obtained at the ade6 locus (no normalization to input), the number of reads covering the heterochromatic pericentromeres showed no significant variation among the wt strains (Fig 6E). In contrast, the strains lacking Dicer (Hall et al., 2002; Volpe et al., 2002) are similar to each other but show substantially reduced read coverage relative to wt at the pericentromeric regions (Fig 6E). Hence, normalization to the spike is sufficient to eliminate technical variations introduced across multiple samples.
Cells harboring Man1-GBP and/or Pol α-GFP were subjected to ChIP-Seq as described above. The read coverage for all pericentromeric regions was substantially reduced in Man1-GBP Pol α-GFP cells relative to wt (Fig 6F). Hence, centromeric heterochromatin maintenance is compromised by inducing centromeres to replicate at Man1 regions, away from the Lem2 microdomains. In addition to reinforcing previous observations of a role for Lem2 in pericentric heterochromatin maintenance, these data are consistent with our observation that Man1 microdomains are spatially distinct from Lem2 microdomains.
To assess the effect of NE-proximal replication on subtelomeric heterochromatin, we performed ChIP-Seq on three independent bqt4Δ isolates. In a wt background, the 35 Kb region near the left end of Chr I is enriched with H3K9Me (Fig 7A). In the absence of Bqt4, H3K9Me levels are increased over this subtelomeric region (Fig 7A); subtelomere 2L also shows increased H3K9Me enrichment in the bqt4∆ setting (Fig 7B). Moreover, while a sharp decrease in H3K9Me enrichment is noticeable at the boundary between telomeric heterochromatin and the euchromatic chromosome arm in wt cells, this distinct H3K9Me transition zone is not observed in the absence of Bqt4. Instead, H3K9Me spreads beyond this boundary, with additional H3K9Me peaks extending into adjacent, normally euchromatic, regions. Telomere proximal regions at other chromosome ends also display defective H3K9Me organization (Fig S7A). However, non-telomeric heterochromatic regions, such as that surrounding the mei4 gene (1.4 Mb from Tel2L) (Zofall et al., 2012), show normal H3K9Me profiles in the absence of Bqt4 (Fig S7B). In contrast to subtelomeres, pericentric H3K9Me enrichment is slightly reduced by bqt4+ deletion (Fig S7A) and H3K9Me enrichment at the mat locus is substantially reduced (Fig. S8A).
To probe our ChIP-seq observations with a physiological assay, we utilized a strain harboring a ura4 reporter gene inserted 7 Kb from the chromosome end (Fig 7C). This ectopic ura4 gene is known to be transcriptionally repressed and bound by Swi6 in wt cells (Kanoh et al., 2005), conferring poor growth on media lacking uracil (-ura). The same wt strain shows robust growth on media containing uracil and 5-Fluoroorotic acid (FOA), which is toxic to cells expressing ura4. This is largely true in a bqt4Δ setting as well, while cells lacking Swi6 show the expected derepression of ura4. However, a subtle but noticeable difference emerges when comparing the growth of wt and bqt4∆ strains on -ura; the bqt4∆ strains grow more poorly, indicating stronger repression of ura4 expression. In a similar experiment using cells with ura4 inserted 11 Kb from the chromosome end (Fig 7D), we found ura4 repression, consistent with previous observations (Kanoh et al., 2005). However, substantial growth variability emerged among five bqt4∆ isolates, with one showing similar growth to wt (poor on -ura and robust on FOA) and three bqt4∆ isolates (Fig 7D, rows 4-6) showing robust growth on both -ura and FOA, indicating enhanced variegation. Hence, the more centromere-proximal ura4 reporter is repressed in the absence of Bqt4, but can be derepressed under selection for ura4 prototrophy. Therefore, Bqt4 is dispensable for subtelomeric silencing, but the extent and robustness of subtelomeric silencing are both limited by the presence of Bqt4. In contrast to the subtelomeres, Bqt4 does not appear to regulate silencing at the telomeric regions (~500 bp from chromosome end; Fig S8B).
To determine whether the effect of Bqt4 on telomeric heterochromatin is mediated by tethering to the Bqt4-adjacent microenvironment, we examined H3K9Me enrichment at telomeres in the induced tethering strains harboring Man1-GBP and Pol α-GFP. In these strains, Bqt4 is present but DNA replication is induced to take place away from the Bqt4 microdomain. At Tel2L, the amplitudes of H3K9Me peaks are elevated in the ‘induced tethering’ background relative to wt (Fig 7E); this held true for other subtelomeres as well. Furthermore, the boundary regions appear to be affected in a manner resembling that of the bqt4Δ setting. Therefore, telomere replication in Man1-adjacent, rather than Bqt4-adjacent, microdomains leads to mis-regulated H3K9Me maintenance.
As induced replication near Man1 affects both subtelomeric and centromeric H3K9Me, we sought to address the possibility that decreased levels of centromeric H3K9Me result from the increase at subtelomeres. To tether telomeric regions to Man1 without altering pericentric localization, we performed ChIP-seq in cells expressing Rap1-GFP and Man1-GBP; in this scenario, the Rap1-bound telomeres will localize to Man1 domains while centromere replication is unaffected. In Man1-GBP/Rap1-GFP cells, subtelomeric H3K9Me increases in Man1-GBP/Rap1-GFP cells to an extent similar to that seen in Man1-GBP/Polα-GFP cells (Fig 7F). Centromeric H3K9Me peak amplitudes are modestly reduced in this scenario, but this reduction is minor compared with the reduction seen in Man1-GBP/Polα-GFP scenarios (compare Fig 6F with Fig 7G). Hence, we suspect that while titration effects due to increased subtelomeric H3K9Me levels contribute modestly to the pericentric effect of inducing replication near Man1, the Bqt4-microdomain that is at least partially vacated in the Man1-GBP/Polα-GFP scenario regulates pericentric replication directly.
Lem2 and Bqt4 become crucial for viability in the absence of Dcr1
The foregoing results outline heterochromatin replication and assembly functions of Lem2- and Bqt4-adjacent microdomains, but also demonstrate that Bqt4 regulates Lem2 itself. To gain insight into the mechanisms underlying Lem2- and Bqt4-mediated chromatin regulation, we examined a series of genetic interactions by constructing doubly and triply mutated strains. The lem2∆ and dcr1∆ single mutant cells show moderately (~5-fold) reduced growth relative to wt (Fig 8A). Loss of the anti-silencing factor Epe1 restores wt levels of viability to dcr1Δ but not lem2Δ cells, indicating that Lem2 acts in a pathway independent of heterochromatin maintenance to promote viability. Moreover, the double deletion lem2Δ dcr1Δ shows markedly reduced viability, demonstrating that Lem2 acts in a pathway separate from (and redundant with) Dicermediated pathways to promote viability. Accordingly, deletion of epe1 confers suppression of the lem2Δ dcr1Δ inviability, but not to the level of an epe1Δ dcr1Δ strain (Fig 8A). As discussed below, these results, along with our ChIP-seq and localization data. suggest that Lem2 functions both in pericentric heterochromatin maintenance and in centromere/kinetochore maintenance, which when compromised will make Dicer essential (Folco, Pidoux, Urano, & Allshire, 2008a; Klutstein, Fennell, Fernandez-Alvarez, & Cooper, 2015). Consistent with this idea, we find that lem2+ deletion greatly reduces the viability of cells harboring a temperature sensitive allele of Cenp-A (cnp1-1) (Takahashi et al., 2000). Moreover, while overexpression of the Cenp-A loading factor Ams2 has been shown to rescue the temperature sensitivity of cnp1-1 (Fig 8B) (Chen, Saitoh, Yanagida, & Takahashi, 2003), lem2+ deletion abolishes this suppression (Fig 8B).
In contrast to Lem2 or Dcr1, the loss of Bqt4 alone does not compromise growth under unperturbed conditions (Fig 8C). However, bqt4Δ and dcr1Δ are synthetically sick and bqt4Δ lem2Δ cells are inviable (data not shown) (Tange et al., 2016). Hence, in the absence of Lem2 or Dcr1, Bqt4 becomes crucial for viability. Below we propose a model in which Bqt4 defines domains conferring viability in the face of transcription/replication collisions. while Lem2 plays roles in kinetochore maintenance; Dicer intersects with both of these processes by promoting RNAP2 dislodgement as well as heterochromatin formation.
Discussion
Here we delineate biochemically distinct microenvironments at the periphery of the nucleus, with distinct capacities to regulate centromeric and telomeric chromatin replication (Fig 9). In addition to the Lem2-rich domain beneath the SPB, these microenvironments comprise Man1-rich domains and separate Bqt4-rich domains. Bqt4 regulates the replication of specific chromosome regions via two separable modes: First, by tethering specific heterochromatic regions, Bqt4-microdomains promote the fidelity of DNA replication and the ensuing reassembly of heterochromatin in the wake of the replication fork. Second, Bqt4 mobilizes Lem2 to sites around the NE to promote pericentric heterochromatin maintenance. Our results suggest separable roles for Lem2 as well; the subset of Lem2 molecules sequestered beneath the SPB influences CenpA loading and therefore kinetochore maintenance, while the Bqt4-mobilized subset of Lem2 molecules promotes pericentric heterochromatin maintenance. Collectively, our data suggest a model in which the central function of Bqt4 is to facilitate duplication of difficult-to-replicate repeats as well as their underlying chromatin.
Although Bqt4 has been known to tether telomeres to the NE, we find that this tethering role is specific to telomeres undergoing replication; redundant pathways are clearly capable of tethering nonreplicating telomeres to the NE. The passage of the replisome and attendant disruption of DNA protein interactions likely represents a challenge to NE tethering mechanisms. Conceivably, the relatively high mobility of Bqt4 along the NE allows the flexibility to keep contact with telomeres as they unwind and accommodate the replisome. As Bqt4 also tethers the mat locus while it is replicating, we infer that multiple loci, perhaps all of them repetitive and/or heterochromatic, are regulated in this manner. Nevertheless, it is clear that not all loci come under such Bqt4 control, as the euchromatic arm locus we studied (cut3+) shows neither NE enrichment nor Bqt4 dependent localization. The specificity of Bqt4 interaction with non-telomeric loci may be determined via its N-terminal DNA-binding domain (~110 residue), which shows homology to a family of fungal transcription factors; alternatively, by analogy with the telomeric Taz1-Rap1-Bqt4 interaction, Bqt4 may contact the mat region via other DNA binding proteins.
In addition to tethering specific chromatin segments to the NE while they replicate, Bqt4-microdomains are important for cellular resistance to agents that cause DNA damage during replication; Bqt4 appears dispensable for resistance to agents that cause replication-independent DNA damage. We propose a model in which the HU and MMS sensitivity seen in the absence of Bqt4 stems from damage specifically to the chromatin regions (like telomeres) that are tethered by Bqt4 to the NE. Indeed, these repetitive regions are particularly challenging for the replication machinery and cause extra trouble in HU or MMS. In particular, we suggest that Bqt4-rich domains are specialized microenvironments in which collisions between replication and transcription machineries are robustly handled or avoided. Such collisions are a known danger at the telomere-adjacent rDNA regions, and become more likely at all regions undergoing high levels of transcription and/or stalled replication. The observation that double deletion of bqt4+ and clr4+ triggers a synthetic loss of viability on MMS demonstrates that Bqt4 and Clr4 act via separate pathways to promote robust replication. Hence, when H3K9Me-containing heterochromatin is abolished by Clr4 loss, the resulting increases in transcription may enhance the danger of replication/transcription collisions in MMS, heightening the need for Bqt4-mediated tethering. Indeed, Bqt4 becomes crucial for viability even in unperturbed conditions when Dicer, which has been shown to promote transcriptional termination at sites of replication stress via eviction of RNA polymerase II, is missing. The observation that epe1+ deletion suppresses the severe growth defects of bqt4Δdcr1Δ cells is consistent with the notion that reduced transcription at sensitive regions (like rDNA and telomeres), due to enhanced heterochromatin maintenance/spreading in the absence of Epe1, obviates the need for processes that limit the occurrence of transcription/replication collisions. The notion that Bqt4-microdomains are ‘safe zones’ for replication of repetitive regions would explain our initial observation that Bqt4 becomes important in HAATI cells, in which rDNA repeats form all chromosomal termini, and cells harbor on average six-fold more rDNA than wt cells.
The most parsimonious hypothesis for why subtelomeric heterochromatin levels are altered by bqt4+ deletion would tie such alterations with the function of Bqt4 in replication. The loss of Dicer-mediated protection from replication/transcription collisions at pericentromeres has been shown to compromise reassembly of heterochromatin following replication fork passage (Zaratiegui et al., 2011). Similarly, loss of localization to the Bqt4-microdomains may confer subtelomeric replication/transcription collisions, thereby altering heterochromatin reassembly; by releasing limiting heterochromatin assembly factors, compromised pericentric heterochromatin may also contribute to the increased H3K9Me levels seen at subtelomeric heterochromatin in bqt4Δ settings (Tadeo et al., 2013). We note that the effects of bqt4+ deletion on pericentric and subtelomeric H3K9Me levels, as well as on subtelomeric silencing, are subtle. In concert with this observation, Bqt4 loss fails to confer appreciable sensitivity to the microtubule depolymerizing agent thiabendazole (TBZ, Fig S5) in contrast with Dcr1 loss, which is known to confer TBZ sensitivity due to loss of pericentric heterochromatin (Provost et al., 2002). Moreover, although bqt4 and dcr1 act in separate capacities to promote general growth, they are epistatic with respect to TBZ sensitivity. We presume that while pericentric heterochromatin levels are reduced by loss of Bqt4 alone, H3K9Me levels are high enough to prevent pericentric cohesin levels from dropping beneath the critical threshold that governs TBZ sensitivity.
The properties of Bqt4-rich regions that make them suitable for averting replication/transcription collisions remain to be determined. Perhaps these regions tend to exclude RNA polymerase II, thus averting local replication/transcription collisions, or concentrate the Pif1 helicase, which allows replication to negotiate the transcription machinery. The accumulation of the telomeric transcripts, known as TERRA, is thought to be reduced specifically during telomere replication; conceivably, this reduction depends on localization to the Bqt4-rich domain. These ideas are currently under investigation.
The genetic interactions between Bqt4 and Lem2 presumably reflect two components, (1) a pathway jointly regulated by Bqt4 and Lem2, and (2) distinct pathways regulated uniquely by either protein, evinced by their synthetic lethality (data not shown) (Tange et al., 2016). Bqt4 is required for the mobility of Lem2 to SPB-distal NE regions, and is therefore likely epistatic with Lem2 for the functions it carries out at these sites. Our microscopy shows distinct localization patterns for centromere cores and pericentric regions, with the centromeric cores colocalizing with Lem2 beneath the SPB and the pericentromeres colocalizing with non-SPB-adjacent Lem2. Hence, we surmise that the effect of Bqt4 on pericentric heterochromatin maintenance stems from loss of Lem2 from these SPB-distal sites. The similar modest nature of pericentric heterochromatin loss in both lem2Δ (Barrales et al., 2016; Tange et al., 2016) and bqt4Δ cells supports this idea. Along with the localization of centromeric core regions with Lem2 beneath the SPB, our observation that lem2+ deletion abolishes the ability of overexpressed Ams2 to rescue the hypomorphic cnp1-1 allele suggests that Lem2 regulates Cnp1 maintenance at the centromeric core. Compromised Cnp1 maintenance would necessitate robust pericentric heterochromatin, which is required for establishment of Cnp1 chromatin on centromere sequences lacking pre-established Cnp1 (Folco et al., 2008b). While lem2+ deletion alone reduces pericentric heterochromatin, this effect is modest and clearly insufficient to abolish Cnp1-containing centromere identity in lem2Δ single mutants. However, competition between replication and transcription is a particular challenge not only at rDNA, but also at pericentromeres, which display Swi6/Clr4-dependent early replication (Hayashi et al., 2009) that coincides with local transcription (Kloc et al., 2008; Zaratiegui et al., 2011), resulting in the formation of branched recombination intermediates that necessitate Dicer-mediated RNAP2 release; in the absence of Dicer, homologous recombination-mediated repair at pericentromeres is thought to be essential for viability. Hence, we propose that in a lem2Δdcr1Δ double mutant, a ‘perfect storm’ of factors severely compromises viability; replication of the region is highly problematic due to rampant collisions between the replisome and RNAP2, Cnp1 maintenance is compromised, and pericentric heterochromatin is abolished, disallowing both kinetochore establishment and kinetochore maintenance. Such considerations may also underlie the complete inviability of lem2Δbqt4Δ cells (data not shown) (Tange et al., 2016).
While our work has revealed roles of a specific pair of microdomains in regulating distinct chromatin regions, we hypothesize that this reflects a wider role for distinct NE subdomains in influencing chromatin structure and function at other loci. The orderly formation of nuclear microenvironments is even more critical in mammalian genomes with their higher proportion of repetitive sequences and diverse range of chromatin types found along the length of a chromosome (Akhtar et al., 2013; Gerhardt et al., 2016; Simonis et al., 2006; Zullo et al., 2012). For difficult-to-replicate chromatin regions, the presence of distinct nuclear subdomains, with specific molecular compositions tailored to interactions with distinct chromatin subtypes, may coordinate the fidelity of replication of each subtype, ensuring that entire chromosomes are inherited with both genetic and epigenetic information intact.
Methods
Yeast strains and media
Strains used in this study are described in Table S1. All media used were previously established (Moreno, Klar, & Nurse, 1991). Gene deletions and fluorophore-tag insertions were performed by the one-step gene replacement method, using appropriate plasmid templates (pFA6a-kanMX6, -hygMX6, and -NatMX6 selection cassettes) as described previously (Klutstein et al., 2015). Functionality was tested for each tagged protein: Bqt4 function was checked by monitoring meiosis and telomere-NE localization, Lem2 by rescue of the thiabendazole sensitivity of lem2Δ cells (Banday, Farooq, Rashid, Abdullah, & Altaf, 2016), and Man1 by localization to the NE. Strain mating and random sporulation were used to combine multiple genetic mutations. Viability assays were performed by adding HU, MMS, and bleomycin at indicated concentrations to molten agar-containing rich yeast-extract media (YE5S; Sunrise Science Products, San Diego, CA) pre-cooled to 55 °C. Overnight mid-log phase cell cultures (OD600 ~0.5) were serially diluted 5-fold, and stamped onto the plates containing media with the DNA damaging agents. Gene silencing assays were performed by stamping serially diluted cultures onto plates containing Edinburgh minimal medium (EMM, Sunrise Science Products) supplemented with appropriate amino acids and lacking the indicated nutrients.
Live microscopy
Strain cultures were grown overnight at 32 °C in 50 ml rich media to mid-log phase. After washing with EMM, cultures were diluted to original volume and the pellet from 1 ml of sample was resuspended in 20 µl of the imaging medium (1:2 YE5S-EMM). The resuspended cells (3µl) were mounted on an agarose pad formed on a glass slide as described previously (Hiraga, Robertson, & Donaldson, 2006). Cells were imaged on a wide-field inverted DeltaVision (GE Healthcare) microscope with a 100X oil immersion objective (NA 1.42) and a Xenon lamp excitation source. Imaging was performed at 22-25 °C. Images were captured and analyzed using SoftWoRx (GE Healthcare). Cells were imaged by capturing single time-point Z-stacks consisting of 25 images, with 0.25 µm between each Z-plane. The approach for analyzing positioning of the lacO/I dot within the nuclear zones was described previously (Hediger, Neumann, Van Houwe, Dubrana, & Gasser, 2002). To capture time-lapse images, cells were adhered to the bottom of a 35 mm glass bottom culture dishes (MatTek) as described previously (Klutstein et al., 2015), except that imaging was performed at 22-25 °C and time-lapse intervals are indicated in the appropriate figure legends. The FRAP experiments were performed using a Zeiss LSM 710 Confocal system (Carl Zeiss Inc, Thornwood, NY) with a Zeiss Axiovert microscope, a 25 mW Argon 488 nm laser, and a 100x Plan-Neofluar 1.3 NA oil immersion objective. Digital images were 512 × 512 pixels and 12 bit. For FRAP, 3 laser lines of the Argon laser (458, 488, 514 nm) were used to bleach areas of the cell at 100% transmission and 10 iterations. The bleach was started after 1 acquisition scan and images were acquired at the indicated intervals.
Super-resolution microscopy
SIM images were obtained with an Applied Precision OMX (GE Healthcare) using a 60X 1.42 NA Olympus Plan Apo oil objective, and front illuminated sCMOS cameras (6.45 um pixel size). All SIM microscopy was performed at 22–23 °C. Excitation sources consisted of a 488-nm laser (for GFP) or a 561-nm laser (for mCherry), and images were captured by alternating excitation using standard filters (FITC/AF488 and AF568/Texas Red). SIM reconstruction was done with SoftWorX, with a Wiener filter of 0.003. SIM images shown are single Z slices, digitally zoomed with bilinear interpolation using Fiji ImageJ (National Institutes of Health). To prepare cells for imaging, overnight cell cultures were grown (10 mL at OD595 ~0.5) and treated with 3.2% formaldehyde for 1 min before adding 0.1% Glutaraldehyde and incubating for 12 min at 25 °C. The fixed cells were washed and treated with 0.6 mg.mL Zymolyase 100T at 36 °C for 45 min, before treating with 1% Triton X-100 for 5 min at room temperature. Cells were washed and resuspended in 1 ml wash buffer (100 mM PIPES, 1mM EGTA, 1mM MgSO4 pH 6.9). To minimize photo-bleaching and amplify fluorescent signals, GFP and RFP boosters (ChromoTek, Germany) were added to 200 µl of the cells resuspended in wash buffer with added 1% BSA and 100 mM lysine hydrochloride. After incubation for 1 hr at room temperature, cells were washed and resuspended in 10 µl Vectashield (Vector Labs, Burlingame, CA). The cells (1 µl) were placed on a glass slide, covered with an acid washed coverslip, and sealed by nail polish before imaging.
Chromatin Immunoprecipitation
Cultures were grown overnight in 120 ml YE5S medium to mid-log phase and treated with formaldehyde (1% final conc.) for 10 min at room temperature with occasional swirling. As described in Fig 6C, 37.5 ml of the spike culture was mixed with 112.5 ml of each sample before treating with formaldehyde. Treated cells were washed with 50 mL refrigerated Milli-Q (Millipore) water. Cells were centrifuged at 2000 rpm for 2 min, resuspended in 1 ml water and centrifuged before adding 1 ml Zirconium beads to the pellet. The pellet and beads mixture was flash frozen in liquid nitrogen and stored at −80 °C in 2 ml screw-cap shatter proof tubes. The frozen pellet was resuspended in 500 µl lysis buffer: 50 mM Hepes pH 7.4, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% Sodium Deoxycholate, 1% PMSF and EDTA-free protease inhibitor cocktail (Roch) and lysed using FastPrep (MP Biomedical; setting: 5x 30 Sec cycles at 4.5 m/s). Each lysate was recovered in a 5 ml snap-cap tube by centrifugation at 1500 rpm for 3 min and sonicated using the Covaris E210 instrument. Sonication settings used were: 20%, 8 intensity, 200 cycles per burst for 12 minutes. The sonicated lysates were clarified by centrifugation at 14000 rpm for 10 min. For input DNA, 20 µl of the clarified lysate was set aside. The lysates were incubated with 5 µg H3K9Me (abcam #1220) antibody overnight at 4 °C. Magnetic protein-G beads (Invitrogen) were added to the lysate/antibody mixture and incubated for 4 hours before washing and eluting the immunoprecipitate from the beads. Elution by boiling and reverse cross-linking was performed using the ChIP Elute Kit (Clontech) to obtain ssDNA templates for Next-Gen library preparation.
Next-Generation Sequencing
The ssDNA samples prepared above were quantified using the Quibit ssDNA Assay Kit (molecular probes, life technologies), and 5-10 ng of each sample was used for library preparation. Libraries were prepared using DNA SMART ChIP-Seq Kit (Clontech) and Illumina indexed oligomers. Following PCR amplifications (15 cycles), libraries were size selected and purified using Agencourt AMPure XP beads (200-400 bp range). Each library was validated using Quibit 2.0 Fluorometer for concentration, Agilent 2100 Bioanalyzer for fragment size, and NEBNext Library Quant Kit (New England BioLabs) for adaptor presence. Typical library yield was 10-40 ng with average size of 400 bp. Finally, libraries from 12 samples were pooled in equimolar ratio (10 nM) and submitted for sequencing using Illumina HiSeq2500 Rapid Run mode (GENEWIZ, South Plainfield, NJ). A total of ~140 million 50 bp reads were obtained per pooled sample. The reads for each sample in the pool were separated per the corresponding Illumina indexes (% of >= Q30 bases: 97.85; mean quality score: 38.63), before subjecting to bioinformatics analysis.
Bioinformatic analysis
As required by the library construction procedure, three bases at the 5’ ends of the reads were trimmed. The 47 bp post-trimmed reads were aligned to the reference genome using bowtie2, allowing for 50 alignments per repetitive region. The original reference sequence for Chr II contains a partially deleted sequence of the mat locus; therefore, we customized the reference genome by replacing the deleted mat region with a contig of the h90 configuration of the mat2P-mat3M region (as described on ‘http://www.pombase.org/status/mating-type-region’; accessed July.2017). Before normalization, the resulting aligned SAM files were converted to sorted and indexed BAM files. Normalization was performed using a Python script that sequentially executed: 1- Get the number of total aligned reads, reads covering ade6 locus (chr III: 1316291-1318035), and reads covering a euchromatin region adjacent to ade6 locus (chr III: 1304000,1380000); 2- Calculate a normalization scale factor: 10 / (median of reads covering each base within the ade6 locus – median of reads covering each base in the neighboring euchromatic region); 3- Apply the scale factor to the aligned genomic reads (BAM files) using the genomecov function of the BEDTools suit to produce a bedgraph file. To generate the ribbon-line coverage plots shown in the above figures, an R script was created that imports the bedgraph files and, using a 250 bp sliding window, calculates the median, maximum and minimum of reads from strains with identical genotypes. The scripts used for this study have been deposited at ‘https://www.elucidaid.com/repository/ebrahimih17’.
Acknowledgements
We thank Michael Lichten, Shiv Grewal and our laboratory members for discussions; Robin Allshire, Yasushi Hiraoka and Masamitsu Sato for the generous gift of strains and reagents; Yoshiyuki Wakabayashi for help with the design of the ChIP-Seq protocol; Susan Garfield for help with the confocal microscopy/FRAP experiments; and Patrina Pellett for help with OMX imaging setup.