A CRISPR/Cas9-induced blunt-end telomere system in S. pombe reveals RNase H2-dependent RNA primer removal at the terminal Okazaki fragment of lagging telomeres

Abstract

Studying the fine-scale dynamics of telomere replication has been hindered by the heterogeneity of native telomeres and the limitations of existing tools. Here, we report a highly efficient and inducible CRISPR/Cas9-mediated system for generating de novo telomeres with defined blunt ends in S. pombe. This fine setting allows for the precise dissection of post-replicative telomere end structures at near single-nucleotide resolution. Using this system, we show that the replicated leading-strand telomere is blunt-ended, while the lagging-strand counterpart contains an ∼10-nt 3′ overhang resulting from RNA primer removal. By analyzing mutants deficient in ribonucleases, we found that the removal of this terminal RNA primer is specifically dependent on RNase H2, but not RNase H1. This RNase H2-dependent mechanism is essential for defining the mature structure of the lagging-strand telomere with authentic telomeric sequences. Our findings reveal a fundamental asymmetry in telomere end processing after replication and establish RNase H2 as the key enzyme responsible for resolving the terminal RNA primer in lagging-strand telomeres. This mechanism, conserved from budding to fission yeast, underscores the critical and evolutionarily ancient role of RNase H2 in defining eukaryotic telomere architecture.

Graphical Abstract

Graphical Abstract.

Introduction

Telomeres are the physical ends of linear chromosomes in eukaryotic cells and consist of repetitive DNA sequences bound by specific proteins. Their special structure distinguishes chromosome ends from normal DNA double-strand breaks (DSBs) and prevents terminal fusion, repair, or degradation, thereby maintaining chromosome integrity and genome stability [1–3].

The telomeric DNA sequence consists of short, guanine-rich nucleotide repeats that vary between species [4]. In the fission yeast Schizosaccharomyces pombe, the consensus sequence of telomeric DNA is 5′-G_2–6TTAC(A)-3′/3′-C_2–6AATG(T)-5′ and it is ~300 bp in length [1, 5]. The guanine-rich strand, known as the G-strand, usually has a 3′ terminal protrusion (also called a 3′ overhang), which is bound and protected by a specific telomeric single-stranded binding protein (i.e. Pot1) [1, 4]. Fission yeast telomeres also contain sub-telomeric regions of ~100 kb in length. Part of this region is a 50-kb sub-telomeric homologous (SH) region. The remaining distal part consists of unique sequences that form a highly condensed chromatin structure called the “knob” structure [6].

Telomeres share many structural similarities with DNA DSBs, but the structurally and functionally conserved protein complexes that specifically bind to telomeres (also known as shelterin) maintain the stability of the telomere structure while effectively preventing nucleases from recognizing and degrading the telomere ends [2, 7]. In fission yeast, shelterin consists of six proteins: Taz1, Rap1, Poz1, Tpz1, Pot1, and Ccq1. Taz1 and Rap1 form one sub-complex bound to double-stranded DNA, while Tpz1 and Pot1 form another sub-complex bound to single-stranded DNA [8–11]. Poz1 bridges these two sub-complexes [11]. Ccq1 interacts with Tpz1 to recruit telomerase [11–13]. In addition to shelterin, the heterodimeric Pku complex (consisting of Pku70 and Pku80) exerts a protective function at telomeres. Inactivation of the Pku proteins results in shortened telomeres and frequent chromosomal rearrangements in the proximal telomere sequence [14–16]. Due to the structural similarities between telomeres and DNA DSBs, many proteins involved in DNA DSB repair also function at telomeres [17]. For instance, the Mre11-Rad50-Nbs1 complex interacts with Ctp1 to create a single-stranded DNA gap, which initiates the cleavage of the 5′ end of the CA strand by the nuclease Exo1, which in turn produces a 3′ overhang [18–20].

DNA replication relies on DNA polymerases to carry out semi-conservative and semi-continuous replication in a strict 5′ to 3′ direction. The same principle applies to telomeric DNA replication [21–24]. The leading strand of telomeric DNA is synthesized continuously in the 5′–3′ direction by DNA polymerase ε (Polε), which uses the C strand as a template. Meanwhile, the lagging strand is synthesized as Okazaki fragments using the G strand as a template (Supplementary Fig. S1). First, Polα-primase synthesizes an 8–12 nucleotide (nt) RNA primer and an ∼20-nt DNA primer, and then Polδ extends the Okazaki fragment synthesis until it displaces the RNA–DNA primer in the downstream Okazaki fragment, forming a flap structure (Supplementary Fig. S1). This structure is subsequently cleaved by the specific nucleases Fen1 (or Dna2). The single-stranded nicks are then sealed by DNA ligase I (Ligase I), completing lagging strand synthesis [25–27]. However, when the replication process reaches the very end of the chromosome, there is no upstream Okazaki fragment that displaces the RNA–DNA primer in the very terminal Okazaki fragment. Thus, how the terminal RNA primer is removed remains unclear [27, 28] (Supplementary Fig. S1).

Due to the highly repetitive nature of telomere DNA sequences, the heterogeneity of telomere lengths among different chromosomes and the limited resolution of existing biochemical approaches, it has long been difficult for researchers to monitor the dynamic changes in telomere structure during DNA replication [29–31]. Previous studies have shown that the de novo telomeres generated by I-SceI endonuclease cleavage can efficiently bind telomere-associated proteins such as Taz1, Pot1, and Ccq1, thereby distinguishing themselves from ordinary DSBs and fulfilling functional roles [32–34]. Recently, we used the de novo telomere formation system to analyze telomere end replication and protection in the budding yeast Saccharomyces cerevisiae, and found that the RNA primer at the terminal Okazaki fragment is removed by RNase H2 [35]. However, given the significant evolutionary divergence (~300–400 million years) [36] and notable differences in telomere-associated proteins between fission and budding yeast [11, 37–40], it remained an open question whether this RNase H2-dependent mechanism is conserved in S. pombe, whose shelterin complex more closely resembles that of mammals. Addressing this question would not only resolve the mechanism of terminal primer removal but also offers insights into the evolutionary constraints shaping telomere replication.

In this study, we developed a highly efficient system for CRISPR/Cas9-mediated chromosome engineering in S. pombe. We observed that the conditional induction of CRISPR/Cas9 targets an intrachromosomal telomeric sequence to generate a de novo telomere with a blunt end. Additionally, inactivation of RNase H2 results in the retention of the RNA primer in the terminal Okazaki fragment at replicated lagging-strand telomeres, demonstrating that RNase H2 is responsible for the removal of the last RNA primer at S. pombe telomeres.

Materials and methods

Reagents and tools table

See Table 1. 

Methods and protocols

Strain construction

All yeast strains (see details in Table 1) of de novo telomere system are derivatives of the S. pombe strain: ura4-D18 leu1-32 his3-D1 ade6-M210, which was obtained from Dr Li-Lin Du [National Institute of Biological Sciences (NIBS), China]. Cas9 expression cassette (TetR-NLS-Cas9-NLS-bsdMX-sgRNA) was inserted to the leu1-32 locus. De novo telomere cassette (ura4⁺-dnt-hphMX/kanMX) was inserted to the 2-kb region 3′ of the gal1⁺ gene 3′ untranslated region [32]. Gene knock-in was performed by a PCR (polymerase chain reaction)-amplified DNA product-mediated homologous recombination followed by lithium acetate transformation, and positive clones were verified by PCR, and the resulting strain hereafter served as the WT control strain. Gene deletion mutants were generated by PCR fragments-mediated gene knockout.

Table 1.

Reagents and tools used in this study

Reagent/resource	Reference or source	Identifier or catalog number
Experimental models: fission yeast strains
h+ ura4-D18 leu1-32 his3-D1 ade6-M210	Dr Li-Lin Du (NIBS)	DY1778
h− ura4-D18 leu1-32 his3-D1 ade6-M210	Dr Li-Lin Du (NIBS)	DY1781
DY1781 leu1-32::pFA-TetR-PenotetS-Cas9-bsdMX-sgRNA11 gal1-3′::ura4+-dnt-sgRNA11 site-hphMX	This study	DNTFY171 (DF254)
DF254 exo1Δ::natMX	This study	DNTFY236
DF254 pku70Δ::kanMX	This study	DNTFY323
DF254 pku70Δ::kanMX exo1Δ::natMX	This study	DNTFY327
DF254 taz1Δ::natMX	This study	DNTFY374
DF254 pku70Δ::kanMX taz1Δ::natMX	This study	DNTFY378
DF254 rnh201Δ::natMX	This study	DNTFY175
DF254 rnh1Δ::kanMX	This study	DNTFY258
DF254 rnh1Δ::kanMX rnh201Δ::natMX	This study	DNTFY260
DY1778 leu1-32::pFA-TetR-PenotetS-Cas9-bsdMX-sgRNA15	This study	DNTFY240
DY1778 leu1-32::pFA-TetR-PenotetS-Cas9-bsdMX-sgRNA15 gal1-3′::ura4+-dnt-sgRNA15 site-kanMX	This study	DNTFY297 (DA260)
DA260 rnh201Δ::natMX	This study	DNTFY252
DA260 rnh1Δ::hphMX	This study	DNTFY355
DA260 rnh1Δ::hphMX rnh201Δ::natMX	This study	DNTFY359
Recombinant DNA
Plasmid: pFA-TetR-PenotetS-Cas9-bsdMX-sgRNA11	This study	pDNT55
Plasmid: pRS314-ura4+-dnt-sgRNA11-hphMX	This study	pDNT69
Plasmid: pRS314-ura4+-dnt-sgRNA15-hphMX	This study	pDNT97
Plasmid: pFA-TetR-PenotetS-Cas9-bsdMX-sgRNA15	This study	pDNT107
Oligonucleotides and other sequence-based reagents
Oligonucleotides used in the study see details in Supplementary Tables S3–S5	Azenta	-
Chemicals, enzymes, and other reagents
Propidium Iodide	Sigma	P4170
Hydroxyurea	Sigma	H8627
Proteinase K	Sigma	P6556
Hygromycin B	Sigma	V900372
l-Glutamic acid	Sigma	G5889
Single-stranded template DNA	Sigma	D9156
Nylon membrane	Thermo Scientific	AM10104
RNase A	Thermo Scientific	EN0531
EcoRV	Thermo Scientific	FD0304
SuperSignal™ West Atto	Thermo Scientific	A38556
Chemiluminescent Nucleic Acid Detection Module Kit	Thermo Scientific	89 880
Exo I	NEB	M0293L
RNase H	NEB	M0297L
Electrophoresis mode	Cytiva	SE260
Blasticidin S	Abcam	ab141452
Nourseothricin sulfate	Solarbio	N9210
G418 sulfate	BBI	A600958-0005
Anhydrotetracycline	APExBIO	C4291
DNA extraction reagent	Psaitong	PS1013
Zymolyase	MP Biomedicals	08 320 921
Acryl/Bis 30% Solution (29:1)	Sangon Biotech	B546017
Ligation high Ver.2	TOYOBO	LGK-201
In-Fusion HD Cloning Kit	Takara	639 649
Software
sgRNA design software	zlab	https://www.benchling.com/crispr
Fiji/ImageJ	Fiji/ImageJ	https://imagej.net/ij/
R, 4.4.1	The R project	https://www.r-project.org/
FlowJo version 10.2	BD Biosciences	https://docs.flowjo.com/flowjo/gettingacquainted/10-2-release-notes/

De novo telomere induction and cell cycle monitoring

All strains were first inoculated in hygromycin- or G418-containing selection medium to monitor potential leaky expression and were then cultured to log phase.

For de novo telomere formation assay, after being cultured to log phase in hygromycin/G418-containing medium, cells were collected and washed three times in EMM-N medium (with low supplements) to remove residual nitrogen sources, and then resuspended in EMM-N medium. After 16 h of synchronization, ahTET was added to induce Cas9 expression. Cells were harvested at the indicated time points.

For cell replication experiment, cells were collected and washed with EMM-N medium and resuspended in EMM-N medium. After 24 h of nitrogen starvation to achieve G0 arrest, ahTET was added to induce Cas9 expression for 8 h during this period. Subsequently, cells were collected and washed twice with sterile water to remove residual ahTET, and then resuspended in YES medium to release cells into the cell cycle. Cells were harvested at different time points. Cell cycle progression was confirmed by FACS (fluorescence-activated cell sorting) analysis. All cell culture experiments were conducted at 32°C. See Table 1 for the reagents used.

Flow cytometry (FACS)

Cells were first washed once with sterile ddH₂O, then thoroughly mixed with 70% ethanol, and incubated for 1 h at room temperature or overnight at 4°C. Fixed cells were washed twice with 50 mM sodium citrate buffer (pH 7.2) and resuspended in the same buffer containing RNase A (final concentration of 200 µg/ml), followed by incubation at 37°C for 2–3 h. Proteinase K (final concentration of 200 µg/ml) was added, followed by incubation at 50°C for 1 h. Cells were washed and diluted to an appropriate concentration, followed by sonication, and stained with propidium iodide (10 µg/ml) for 1 h at room temperature. FACS analysis was then performed [41]. See Table 1 for the reagents used.

Genomic DNA extraction

Cells collected from each time point were stored at −80°C, and their genomic DNA was extracted in parallel using a mild protocol [42]. Cells were first digested with zymolyase at a final concentration of 1.5 mg/ml in sorbitol buffer, then resuspended in lysis buffer, followed by phenol/chloroform extraction and ethanol precipitation. Finally, genomic DNA was dissolved in an appropriate volume of ddH₂O. See Table 1 for the reagents used.

Southern blot analysis

The dissolved DNA was further purified by incubation with RNase A. Additional RNase H treatment was performed at 37°C for 2 h in the RNA primer detection assay. Extracted DNA was digested with EcoRV at 37°C for 2 h, then Exo I was added, followed by incubation at 37°C for 1 h to remove the 3′ overhang.

For native Southern blotting[35], the final digested DNA was separated on native 8% polyacrylamide gel electrophoresis (PAGE) and transferred to a nylon membrane (Thermo Fisher). Then crosslinked by UV and incubated with denature buffer (0.5 M NaOH, 0.5% sodium dodecyl sulfate) at 45°C for 30 min, washed three times with ddH₂O, 5 min each time, and finally hybridized with CA probe (5′-TGT AAC CCC TGT AAC CCC TGT AAC CCC-3′-biotin) at 50°C overnight. Detection of the signal was performed according to the manufacturer’s instructions (Chemiluminescent Nucleic Acid Detection Module, Thermo Fisher). Finally, the membrane was exposed to the CCD camera.

For denaturing Southern blotting[35], an equal volume of 2× denature loading buffer (95% formamide, 0.025% bromophenol blue, 0.025% xylene cyanol FF, 0.5 mM ethylenediaminetetraacetic acid) was added to the final digestion product and boiled at 100°C for 10 min, then immediately chilled into ice for 5 min. Single-stranded DNA was then separated on a denaturing urea (8 M)–6% PAGE, transferred to membrane, crosslinked by UV, and hybridized with the specific CA or TG probe as indicated. Signal detection was performed as described for native PAGE. The sequences of the 3′-biotin-labeled probes are as follows: CA probe, 5′-TGT AAC CCC TGT AAC CCC TGT AAC CCC-3′-biotin; TG probe, 5′-GGG GTT ACA GGG GTT ACA GGG GTT ACA-3′-biotin. Sequences of the size markers for denaturing and native PAGE are listed in Supplementary Tables S3–S5, respectively. All results were repeated three times independently. See Table 1 for the reagents used.

Double-stranded and single-stranded markers preparation

For denaturing PAGE markers, the oligonucleotides used in this study were synthesized by the Azenta company. The sequences of the oligonucleotides are listed in Supplementary Tables S3–S5, which closely matched E-TRF sequences.

For native PAGE markers, dsDNA fragments were generated by annealing the complementary oligonucleotides used for denaturing PAGE. The annealing reaction was heated to 95°C for 10 min, then cooled from 95°C to 65°C at a rate of 0.01°C per second, and held at 65°C for 5 min. The annealing products were separated on an agarose gel and extracted from the gel.

Quantification and statistical analysis

Quantification in Fig. 1D of the telomere signals was performed by the ImageJ. Data are shown as mean ± SEM and n represents the total number of independent biological replicates. For each lane, the signal intensities of the target bands and control bands were normalized to their respective probe-hybridizing lengths, and the normalized target signal was divided by the normalized control signal to obtain a normalized ratio. The bar plots were created using Microsoft Excel 2021.

Figure 1.

De novo telomere generated by CRISPR/Cas9 possesses a stable blunt end. (A) Schematic illustration of the CRISPR/Cas9-mediated de novo telomere induction system. Constitutive expression of TetR inhibits enotetS promoter activity. The addition of ahTET induces Cas9 expression and cuts the sgRNA target sequence (pink rectangle), exposing the de novo telomere (filled triangles), which ends with a 6-bp non-telomeric sequence [shown in red in panel (B)]. (B) A zoomed-in view of the de novo telomere cassette containing the ura4⁺ gene, followed by 254 bp of telomere repeats and the hygromycin resistance marker (hphMX). EcoRV recognition sites are present both within and flanking the de novo telomere sequence. Cas9 cleavage results in a 106-bp E-TRF with a blunt end. The sequence of the E-TRF is displayed at the bottom, with the Cas9 cleavage site highlighted in red (non-telomeric sequence). (C) Southern blot detection of CRISPR/Cas9-mediated de novo telomere formation in WT cells, with 253 bp “uncut,” 211 bp “control,” and 106 bp “cut” (E-TRF) bands. (D) Quantitative analysis of the signal ratios of uncut and cut bands before and after ahTET induction. Signal intensities were normalized to the corresponding control band after normalization for the length of the probe-hybridizing region. Data are presented as mean ± SEM, n = 3/group. (E) Flow chart of de novo telomere formation (see text for a detailed description). Cells are collected at the indicated time points for FACS and Southern blot analysis. (F) FACS analysis of the DNA content of nitrogen-starvation synchronized WT cells. asn: asynchronous. (G) Southern blot detection of the 3′ overhang of the de novo telomere (E-TRF) induced in WT cells. Denaturing Southern blot detection of the CA-strand (H) and TG-strand (I), which represent the same membrane hybridized with TG and CA probe, respectively. LC: loading control.

Quantification in Figs 3E and 4I of the telomere signals was performed by the ImageJ software. The net signal intensities of the Exo I-sensitive and Exo I-resistant daughter telomere bands were quantified after background subtraction and normalized to the length of the corresponding probe-hybridizing region. The percentage of each daughter telomere was determined by dividing the normalized signal intensity of the corresponding band by the total normalized signal intensity of the Exo I-sensitive and Exo I-resistant daughter telomere bands. Data are presented as mean ± SEM and n represents the total number of independent biological replicates. Statistical significance was determined by two-way ANOVA followed by Tukey’s multiple comparisons test. Significance levels were indicated as: *P < .05, **P < .01, ***P < .001, ns, not significant. The bar plots were created using Microsoft Excel 2021.

Figure 3.

Approximately half of the daughter telomeres are blunt-ended after DNA replication. (A) Flow chart of de novo telomere replication within one cell cycle (see text for a detailed description). Cells are collected at the indicated time points for FACS and Southern blot analysis. (B) Southern blot analysis of the 3′ overhang of oligonucleotide markers: a 106-bp blunt-ended DNA and a 106 bp DNA with a 10-nt overhang, as labeled. (C) FACS analysis of DNA content in WT cells before and after release into the cell cycle. asn: asynchronous. (D) Southern blot analysis of the 3′ overhangs at both telomere ends during one round of DNA replication in WT cells. OM: oligonucleotide marker, gray arrow: overlapped daughter telomeres, blue arrow: blunt-ended leading telomere, red square bracket: Exo I-trimmed lagging telomere. (E) Quantification of daughter telomeres with different end structures (E-TRFs in the lane of release 5 h, +Exo I). The percentages of daughter telomeres with different structures were calculated from the background-subtracted signal intensities, normalized to the length of the corresponding probe-hybridizing regions. Data are presented as mean ± SEM, n = 3/group. Denaturing Southern blot detection of the CA-strand (F) and TG-strand (G) after replication in WT cells, which represent the same membrane hybridized with TG and CA probe, respectively. LC: loading control. (H) Schematic illustration of the different structures at the leading and lagging daughter telomeres that are replicated from the parental de novo telomere: the leading telomere has a blunt end, whereas the lagging telomere has a 3′ overhang after the removal of the last RNA primer. The structures of the lagging strand vary slightly due to different positions of the RNA primer initiation at the terminus.

Figure 4.

The last RNA primer is retained on lagging-strand telomere in rnh201∆ cells. FACS analysis of DNA content in WT (A), rnh201∆ (C), rnh1∆ (E), and rnh1∆ rnh201∆ (G) cells. asn: asynchronous. Southern blot analysis of the 3′ overhang at both the leading- and lagging- strand telomere ends during one round of DNA replication in WT (B), rnh201∆ (D), rnh1∆ (F), and rnh1∆ rnh201∆ (H) cells. Green arrow: Exo I-trimmed lagging telomere. Genomic DNA was pretreated with or without RNase H in vitro, as labeled. (I) Quantification of the percentage of Exo I-sensitive daughter telomeres in WT, rnh1∆, rnh201∆, and rnh1∆ rnh201∆ cells when pretreated with or without RNase H in vitro (E-TRFs in the lane of release 5 h, +Exo I). The percentage of Exo I-sensitive daughter telomeres was calculated from the background-subtracted signal intensities, normalized to the length of the corresponding probe-hybridizing regions. Data are presented as mean ± SEM, n = 3/group. Statistical significance was determined by two-way ANOVA followed by Tukey’s multiple comparisons test. Significance levels are indicated as: *P < .05, **P < .01, ***P < .001, ns, not significant. (J) A model for telomere replication in fission yeast. A blunt-ended parental telomere produces two daughter telomeres: the replicated leading strand telomere undergoes minimal processing, and its blunt end is capped by Pku, whereas the replicated lagging strand telomere has a Pot1-protected overhang following the removal of the terminal RNA primer.

Results

A CRISPR/Cas9-induced de novo telomere system generates a stable blunt end

Endonucleases such as I-SceI and HO have been widely used to create DSBs for exposing de novo telomeres [32, 43]. However, these systems suffer from low cleavage efficiency and generate complex end structures. For instance, I-SceI cleavage produces DSBs containing ~9 bp of non-telomeric sequences with a 4-nt 3′ overhang, raising concerns about whether such ends can be processed as authentic telomeres. To overcome these limitations and establish a more robust and precise tool for fission yeast telomere biology, we developed a novel inducible CRISPR/Cas9 system. This system generates de novo telomeres with defined blunt ends, mirroring a potential native telomere state and enabling high-resolution downstream analyses.

We combined a previously reported Cas9 expression system with the Tet-On system [44–46] to construct a system for rapid induction of de novo telomere formation in S. pombe (Fig. 1A). The enotetSp/Cas9-rrk1/sgRNA expression cassette and a tetR gene (encoding TetR repressor) cassette (under the control of an adh1 promoter) were integrated into the leu1-32 locus on chromosome II (Fig. 1A). Additionally, a 254-bp telomere sequence flanked by ura4⁺ and the sgRNA recognition sequence-hphMX gene was inserted near the native right telomere on chromosome II (Fig. 1A). Presumably, the constitutively expressed TetR repressor binds to the enotetS promoter, blocking Cas9 expression. When ahTET is added into the medium, it competitively binds to TetR, which derepresses the enotetS promoter, and induces the expression of the Cas9 (flanked by NLS sequences at both ends). The expressed Cas9/sgRNA complex recognizes the target sequence proximal to the 254-bp telomere sequence and generates a double-stranded break, exposing a de novo telomere with a blunt end. It is worth noting that the optimized de novo telomeric sequence was 254 bp long; such a long telomere could circumvent telomerase extension [47–49], a conclusion also supported by our own data (Fig. 1C and Supplementary Fig. S2). The de novo telomeric sequence consists of regular 5′-GGGGTTACA-3′ repeats, with one EcoRV site in the middle (Supplementary Table S1). Two additional EcoRV restriction sites were designed at both ends of the telomeric sequence to facilitate the detection of the EcoRV-digested terminal restriction fragment (E-TRF) by Southern blotting (Fig. 1B). When Cas9 is not induced, EcoRV digestion yields a control band of 211 bp and an uncut band of 253 bp (Fig. 1B). Upon Cas9 induction, a cut band (E-TRF) of 106 bp appears (Fig. 1B).

After testing multiple sgRNA sequences and evaluating their cleavage efficiencies (Supplementary Fig. S2 and  Supplementary Table S2), we selected the DF254 strain for further experiments (Fig. 1C and Supplementary Fig. S3). Induction of Cas9 with ahTET resulted in efficient cutting of the target sequence, exposing the de novo telomere, which displayed the anticipated 106 bp E-TRF (Fig. 1C). Quantitative analysis showed that after 8 h of ahTET induction, nearly all uncut bands were cleaved and converted into the E-TRF band (Fig. 1D). These results demonstrate that we have established a novel inducible CRISPR/Cas9 system for rapid and highly efficient genome editing in S. pombe.

To determine the terminal structure of de novo telomeres generated by Cas9, we arrested cells in the G0 phase by nitrogen starvation for 16 h [46] (Fig. 1E and F), then induced Cas9 expression by adding ahTET and harvested the cells at various time points. The de novo telomere structure was analyzed by Southern blotting. Results from both native and denaturing gels showed that the E-TRF maintained a constant length of 106 bp/nt in the absence or presence of Exo I digestion (Fig. 1G–I). These results suggest that de novo telomeres induced by Cas9 have blunt ends.

The Pku complex and endogenous Exo1 regulate end resection at de novo telomeres

The S. pombe Pku complex binds to telomere ends and plays a role in maintaining telomere stability [14–16]. To investigate whether the Pku complex inhibits the resection of the CA strand of de novo telomeres, we constructed a DF254 pku70Δ strain and analyzed the structure of de novo telomere ends in cells arrested in the G0 phase (Fig. 2A). When the E-TRFs of the de novo telomeres were examined in a native gel by Southern blotting, they were 106 bp in length within 2–8 h of induction (Fig. 2B). However, a subset of E-TRFs were sensitive to Exo I digestion, displaying diffuse bands ranging from ~91 to 106 bp (Fig. 2B), suggesting that short-range resection had occurred at the de novo telomeres (Supplementary Fig. S4). Denaturing Southern blotting further revealed that the CA strand exhibited a prominent 106-nt band at the 8-h induction time point, both before and after Exo I treatment, along with several additional faint bands of ~100, 91, and 81 nt in length. This pattern was not significantly altered by Exo I digestion (Fig. 2C). In contrast, prior to Exo I treatment, the TG strand was predominantly 106 nt in length; after Exo I digestion, it displayed a similar band pattern to that of the CA strand (Fig. 2D). These results collectively indicate that in the absence of Pku complex protection, the 5′ end of the CA strand of the de novo telomere becomes more susceptible to nuclease resection.

Figure 2.

De novo telomere is protected by Pku complex, preventing from endogenous Exo1-mediated resection. FACS analysis of the DNA content of nitrogen-starvation synchronized pku70∆ (A) and pku70∆ exo1∆ (E) cells. Southern blot detection of the 3′ overhang of the de novo telomere induced in pku70∆ (B) and pku70∆ exo1∆ (F) cells. Denaturing Southern blot detection of the CA strand (C) and TG strand (D) in pku70∆ cells, which represent the same membrane hybridized with TG and CA probe, respectively. LC: loading control. Denaturing Southern blot detection of the CA strand (G) and TG strand (H) in pku70∆ exo1∆ cells, which represent the same membrane hybridized with TG and CA probe, respectively. LC: loading control. The WT lanes shown in panels (F)–(H) correspond to the same Southern blots as those shown in panels (B)–(D), respectively. They are presented separately for clarity. (I) Model for de novo telomere end protection and processing in WT and pku70∆ cells.

We next deleted the exo1 gene, which encodes the 5′–3′ exonuclease Exo1 that is involved in telomere 5′ resection [50, 51], in both the DF254 and DF254 pku70Δ strains, and examined the de novo telomere structure using native and denaturing Southern blot assays. Native Southern blot results showed that in exo1Δ cells (a negative control), the de novo telomere end structure was as expected (Supplementary Fig. S5A, C, D, and E), identical to that in WT cells (Fig. 1G–I). Importantly, in the pku70Δ exo1Δ cells, the 5′ resection of the CA strand observed in pku70Δ cells (Fig. 2B) was no longer detected (Fig. 2F). Consistently, denaturing gel results further showed that both the CA and TG strands in the pku70Δ exo1Δ strain remain 106-nt in length, indistinguishable from those in WT cells, indicating little or no resection (Fig. 2G and H). These results collectively support the model that, in the absence of the Pku complex protection, the CA strand of the exposed de novo telomere is susceptible to Exo1. In other words, the Pku complex protects de novo telomere ends from Exo1-mediated resection, thereby avoiding the generation of long 3′ overhang structures [14, 50–52] (Fig. 2I). Notably, compared to WT strain, faint signals below the 106 nt band were detected in the pku70Δ exo1Δ strain, suggesting that additional nucleases may also contribute to CA strand resection of the de novo telomere when Pku is absent.

Why is the resection of the CA strand in pku70Δ cells restricted to specific lengths, rather than occurring as a continuous size range? We reasoned that the telomeric double-stranded DNA-binding protein Taz1 [9, 53] acts as a physical barrier to protect the ends from resection (Supplementary Fig. S4). The ladder pattern of the CA strand in Fig. 2C correlates well with the sporadic presence of Taz1 binding sites in the de novo telomere sequence. For example, a terminal Taz1 binding site lies just eight nucleotides from the end of the CA strand (Supplementary Fig. S4). We constructed taz1Δ and pku70Δ taz1Δ mutants and examined de novo telomere structures. Unexpectedly, the de novo telomere signals in the taz1Δ mutant displayed a similar pattern to those in the WT cells (Supplementary Fig. S5A, C, D, and E). Additionally, resection of the de novo telomere was barely detectable in the pku70Δ taz1Δ cells compared with wild-type cells (Supplementary Fig. S5G–I). Furthermore, faint bands of intermediate size below cut bands were detected in taz1Δ strain (Supplementary Fig. S5D), but their origin could not be determined. They did not appear to be generated by resection, as no products of the same size were detected on the complementary TG strand (Supplementary Fig. S5E). Because Cas9 cleavage efficiency is markedly reduced in taz1Δ cells under our experimental conditions [53], these data do not allow us to directly determine whether Taz1 itself is responsible for the observed resection pattern. Additional experiments will be required to test the role of Taz1 in this process.

Replication generates asymmetric telomere ends: a blunt leading strand and an overhanging lagging strand

We next examined the structures of daughter telomeres derived from the de novo telomere after one round of replication. The experimental protocol was as follows. The cells were cultured in EMM-N medium for 16 h to synchronize them in the G0 phase. Cas9 expression was then induced by adding ahTET for 8 h to generate de novo telomeres. The cells were then transferred to YES medium to allow progression through cell cycle until completion of one round of cell cycle (Fig. 3A). To facilitate analysis of replicated daughter telomere structures, we prepared a positive control marker, designated “OM” by mixing two DNA fragments in equal amounts: a 106-bp blunt-ended DNA that is insensitive to Exo I; the other is a 106-bp DNA with a 10-nt 3′ overhang that is sensitive to Exo I (Fig. 3B). The sequences of both fragments closely match that of the E-TRF.

The cells were harvested at specific time points and their DNA content was monitored by FACS analysis. The “2C” peak appeared at 3 h and became dominant at 5 h, indicating the completion of DNA replication (Fig. 3C). The lengths and structures of the E-TRFs were then examined using Southern blotting. As shown in Fig. 3D, the E-TRF of the parental telomere measured 106 bp and was insensitive to Exo I treatment prior to replication (Fig. 3D, 0 h), confirming blunt end as mentioned above in Fig. 1. After replication, the E-TRFs of the daughter telomeres were 106 bp in length prior to Exo I digestion (Fig. 3D, 5 h, −Exo I). Strikingly, upon Exo I digestion, the population of replicated E-TRFs resolved into two distinct species: one maintaining the parental 106 bp size (blunt-ended), and the other represented the resection products, which were mainly present as an ~96-bp band, accompanied by a minor smear signal ranging from 90 to 95 bp (Fig. 3D, 5 h, +Exo I). This size difference is consistent with the in vitro trimming of an ∼10 nucleotide 3′ overhang from a subset of the telomeres. Quantitative analysis revealed the 106-bp signal accounted for 50.3% of the total, while the resection products collectively accounted for 49.7% (the 96-bp band comprising 33.4% and the 90–95 bp smear comprising 16.3%) (Fig. 3E). Given the inherent asymmetry of DNA replication, we reasoned that the two species likely correspond to the leading- and lagging-strand replication products (Fig. 3H). Thus, the E-TRF of lagging telomere contains a double strand region and a 10–15-nt 3′ overhang (Fig. 3H). This 10–15 nt overhang likely corresponds to the region previously occupied by the RNA primer.

The lengths of both the replicated TG and CA strands were further examined using denaturing Southern blotting with CA and TG probes, respectively. Prior to Exo I treatment, the TG-strand signal was predominantly 106 nt at each time point (Fig. 3G). After Exo I treatment, approximately half of the TG-strand signal converted to 96 nt or shorter after replication (Fig. 3G, 5 h, +Exo I). We speculate that the bands labeled “undefined” (Fig. 3G) result from a small fraction of de novo telomeres that escaped protection and were subsequently resected. The CA-strand signal measured 106 and 96 nt after replication (Fig. 3F, 5 h), regardless of Exo I digestion. Notably, faint but diffuse CA-strand signals below the 96-nt band were visible. We speculate that these signals originate from differences in RNA primer initiation during terminal Okazaki fragment synthesis, as RNA primer synthesis can begin from either the first thymidine or the cytosines proximal to the 3′ end due to the nucleotide preference of primases (Fig. 3H) [54, 55]. These results are consistent with the native Southern blot data, and suggest the following: (i) the replicated leading strand telomere undergoes little (or no) end processing and maintains a stable blunt-end structure. (ii) The replicated lagging strand telomere contains a 3′ overhang of ~10–15 nt. (iii) The RNA primer at the terminal Okazaki fragment can be synthesized from the end of the TG strand, and is removed after replication. (iv) The length of the RNA primer is ~10 nt. Taken together, this finding reveals a fundamental asymmetry in telomere end processing after replication in S. pombe.

The RNA primer in the terminal Okazaki fragment on the lagging strand telomere is removed after completion of DNA replication

During DNA replication, DNA Polα-primase synthesizes an RNA–DNA primer to initiate Okazaki fragment synthesis on the lagging strand [26, 56]. In the telomeric region, the RNA primer at the end of the Okazaki fragment is thought to be removed, exposing the 3′ overhang and allowing Pot1 to bind and protect the telomere end (Supplementary Fig. S1) [11, 57]. To address whether RNA primers are present in the very termini of replicated lagging strands, we prepared genomic DNA in the presence or absence of RNase H (Supplementary Fig. S6), an enzyme that digests the RNA moiety in RNA–DNA hybrids in vitro [58, 59]. Native Southern blot results showed that the hybridization signals of replicated daughter telomeres exhibited the same pattern regardless of RNase H treatment (Fig. 4A and B). The lagging strand signal slightly increased from 45.3% to 51.8% upon RNase H treatment, but the difference was not statistically significant (Fig. 4I). These data suggest that the RNA primers in the terminal Okazaki fragments are efficiently removed during or after telomere replication.

RNase H2 is specifically required for the removal of the terminal RNA primer

In the internal region of a chromosome, the RNA primer of an Okazaki fragment is typically displaced by an adjacent Okazaki fragment, resulting in the formation of a flap structure. This structure is subsequently cleaved or degraded by endonucleases, such as Fen1 (and/or Dna2) [27, 60–62] (Supplementary Fig. S1). However, the RNA primer at the very end of the lagging strand telomere cannot be removed by the same mechanism as an internal RNA primer, because there is no upstream Okazaki fragment to displace it (Supplementary Fig. S1). As endoribonucleases RNase H1 and RNase H2 can digest the RNA moiety in a RNA–DNA hybrid and play a role in the removal of RNA primers in lagging strand DNA maturation [63–65], it has long been assumed that the primers at the terminal ends of lagging telomeres are removed by RNase H1 and/or RNase H2. However, due to the rarity and short length of these terminal RNA primers, they have been extremely difficult to detect, and this issue has remained experimentally unverified.

To genetically identify the nuclease responsible for terminal primer removal in S. pombe, we analyzed strains deficient in RNase H1 (rnh1Δ) or the catalytic subunit of RNase H2 (rnh201Δ). In the absence of RNase H1 and without in vitro RNase H treatment, the replicated lagging-strand telomeres remained sensitive to Exo I with a slight reduction in signal intensity compared to wild-type cells (Fig. 4E, F, and I). In contrast, loss of RNase H2 resulted in a marked loss of Exo I sensitivity, indicating a failure to generate the 3′ overhang (Fig. 4C and D left, −RNase H). These results suggest that the RNA primer at the terminus of the lagging telomere is retained in the rnh201Δ mutant, but not in the rnh1Δ mutant. The definitive proof came from an in vitro RNase H rescue experiment: pretreatment of rnh201Δ samples with RNase H restored Exo I sensitivity to approximately half of the telomeres (Fig. 4D right, +RNase H), demonstrating that the obstruction to overhang formation was indeed the retained RNA primer. Thus, in S. pombe, RNase H2 is both necessary and sufficient for the removal of the terminal RNA primer. Further combinatorial deletion of rnh1 and rnh201 resulted in nearly complete retention of the RNA primer on the replicated lagging strand telomere (Fig. 4G and H). We therefore conclude that RNase H2 is the primary enzyme responsible for removing the terminal RNA primers from lagging strand telomere, while RNase H1 may play a minor auxiliary role (Fig. 4J).

A de novo telomere with an authentic telomeric sequence produces a blunt-ended daughter leading strand and a 3′ overhanging daughter lagging strand

The de novo telomere generated by Cas9 in the DF254 strain ends with the 6-bp sequence 5′-CCCCCT-3′/3′-GGGGGA-5′ (Supplementary Fig. S3), which differs greatly from the telomeric consensus sequence 5′-G_2-6TTAC(A)-3′/3′-C_2-6AATG(T)-5′. It could be argued that the replication of this non-telomeric sequence and the associated end processing might not recapitulate the events that occur on native telomeres. To unequivocally demonstrate that our findings are not an artifact of the initial non-canonical terminal sequence but rather reflect a general mechanism for authentic telomeres, we engineered a new strain (DA260) where Cas9 cleavage exposes a native S. pombe telomeric sequence 5′-GGTTACAGGGT-3′/3′-CCAATGTCCCA-5′ (TG strand/CA strand) (Fig. 5A), matching the consensus native telomeric sequence 5′-G_2-6TTAC(A)-3′/3′-C_2-6AATG(T)-5′. We then examined both the length and the structure of the de novo telomeres during one round of replication (Fig. 5B and C). The E-TRFs with a native telomeric sequence at the end measured 106 bp, and were insensitive to Exo I treatment prior to replication, indicating a blunt end (Fig. 5C, 0 h). After replication, smeared signals below 106 bp were detected after Exo I digestion (Fig. 5C, 5 h, +Exo I). These results were similar to those obtained from the DF254 strain (Fig. 4B). Further in vitro RNase H treatment of the E-TRF samples slightly enhanced the lagging strand band signals from 41.8% to 45.0% (Fig. 5C and Supplementary Fig. S7). The replication of ends with native telomeric sequences produced asymmetric daughter telomere structures, confirming that the generation of a blunt-ended leading strand and an overhanging lagging strand is a fundamental feature of telomere replication in fission yeast.

Figure 5.

The replication of the de novo telomere with an authentic telomeric tip produces a daughter leading telomere with a blunt-ended and a daughter lagging telomere with a 3′ overhang. (A) Schematic illustration of the CRISPR/Cas9-mediated formation of de novo telomere with an authentic telomeric sequence (displayed at the bottom). The sgRNA pairs with a typical telomere sequence immediately upstream of its PAM sequence (boxed in red at the bottom). Cas9 cleavage exposes the new telomere ending in the sequence 5′-GGTTACGGTTACAGGGT-3′/3′-CCAATGCCAATGTCCCA-5′ (in green rectangle). (B, D, F, and H) FACS analysis of the DNA content in DA260, DA260-rnh201∆, DA260-rnh1∆, and DA260-rnh1∆ rnh201∆ cells before and after release into the cell cycle. asn: asynchronous. Southern blot analysis of the 3′ overhang at both the leading and lagging telomere ends during one round of DNA replication in DA260 (C), DA260-rnh201∆ (E), DA260-rnh1∆ (G), and DA260-rnh1∆ rnh201∆ (H) cells. Green arrow: Exo I-trimmed lagging telomere. Genomic DNA was pretreated with or without RNase H, as labeled.

We also deleted rnh1 and/or rnh201 in the DA260 strain and examined the structures of the leading and lagging daughter telomeres replicated from Cas9-induced de novo telomeres (Fig. 5D, F, and H). The results from the rnh1Δ strain were nearly identical to those observed in WT DA260 strain (Fig. 5G). In both the DA260 rnh201Δ and rnh1Δ rnh201Δ strains, all of the E-TRFs of the replicated de novo telomeres were 106 bp regardless of Exo I treatment (Fig. 5E and I, 0 h), indicating that both the leading and lagging strand telomeres are blunt-ended. When the E-TRFs were pretreated with RNase H and then digested by Exo I in vitro, a lagging strand signal appeared in addition to a 106-bp band (Fig. 5E and I,5 h, +Exo I, +RNase H), indicating that the RNA primer is retained on the lagging telomere, i.e. the lagging strand ends with a blunt RNA–DNA primer hybrid, when RNase H2 is absent. These results consistently support the conclusion that RNase H2 is responsible for removing the terminal RNA primers from lagging strand telomeres (Fig. 4J).

Discussion

Telomeres play a critical role in maintaining genome stability [1, 7]. Studying the mechanism of telomere replication and protection is highly significant for understanding fundamental biological issues such as genome stability, cellular senescence, and cell fate regulation. However, the highly repetitive nature of telomere sequences, the heterogeneity of telomere lengths within and between cells, and the limitations of existing telomere length analysis methods have made it difficult for researchers to precisely observe the fine structure of telomeres, particularly during the complex end processing that occurs during DNA replication at telomeres [66–68]. Our current study for the first time presents a significant methodological advance for studying telomere biology in S. pombe—a highly efficient and inducible CRISPR/Cas9 system for generating de novo telomeres with defined blunt ends. This tool overcomes the limitations of previous endonuclease-based systems (such as I-SceI or HO) [32, 43] and provides unprecedented resolution for dissecting the fine-scale dynamics of telomere replication and end-processing. In addition, compared with previously reported CRISPR/Cas9 systems, such as those using the nmt41 promoter to regulate Cas9 expression, which has a complicated procedure and exists a 14–15-h long lag [44, 45], or the adh1 promoter-driven Cas9 expression [69], which causes cytotoxicity in fission yeast, our current optimized CRISPR/Cas9 system provides a powerful and generally applicable tool for the fission yeast community (Supplementary Fig. S8), enabling high-resolution analysis of telomere dynamics and other time-sensitive chromosomal processes.

Applying this system, we have delineated the molecular events that follow the replication of a telomere in fission yeast, revealing both conserved principles and unique protective mechanisms. A central finding of our work is the inherent structural asymmetry of replicated telomeres: the leading strand telomere is maintained as a stable blunt end, while the lagging strand telomere bears an ∼10 nt 3′ overhang (Figs 3–5). This challenges the long-held assumption that leading strand telomeres undergo extensive processing and fill-in synthesis to achieve a uniform overhanging structure [2, 70]. Crucially, we demonstrate that this blunt-end structure is not an artifact of our experimental system, as it is faithfully produced even when the de novo telomere terminates with authentic native telomeric sequences (Fig. 5). The stability of the leading strand blunt end is likely ensured by the protective capping function of the Pku complex, which we show acts as a critical barrier to Exo1-mediated resection (Fig. 2).

Perhaps the most mechanistically insightful finding is the identification of RNase H2 as the primary enzyme responsible for removing the terminal RNA primer on the lagging strand telomere. This role of RNase H2 is specifically conserved in S. pombe, as RNase H1 plays little detectable role [35] (Figs 4 and 5). The elegant genetic evidence from our mutant analyses, the RNA primer is retained in rnh201Δ cells and can be subsequently removed in vitro by RNase H (Figs 4 and 5), provides compelling support for this conclusion. Together with our parallel findings in S. cerevisiae [35], this establishes RNase H2-mediated RNA primer removal as a fundamental and evolutionarily conserved cause to the “end replication problem” across a vast evolutionary distance, separating budding and fission yeast by over 300 million years (Fig. 4J).

The conservation of this mechanism in S. pombe carries particular significance. Unlike S. cerevisiae, the fission yeast telomeric shelterin complex shares greater homology with that of mammals [2, 4, 38]. Therefore, our findings may offer a more direct and relevant model for understanding telomere replication in higher eukaryotes. The existence of blunt-ended leading strand telomeres has been reported in plants and mammals [71, 72], but the mechanisms governing their formation and protection have remained unclear. Our work in S. pombe provides a clear mechanistic framework: full replication yields a blunt end, which is then stabilized by capping complexes like Ku (Pku in yeast), obviating the need for further processing. It will be of great interest to determine whether RNase H2 performs the same indispensable role in resolving terminal Okazaki fragments in mammalian cells. It is worth noting that the length of RNA primers in fission yeast has long been extrapolated from in vitro studies [73, 74]. Our present study zooms in the length of the primer at the lagging strand telomere of fission yeast and provides in vivo evidence to confirm the widely accepted claim that RNA primers are approximately 10 nucleotides in length (Figs 3 and 4).

In summary, we propose a revised model for telomere replication (“the asymmetrical end replication model”) where the end replication problem is predominantly driven by the inevitable removal of the RNA primer from the lagging strand, with the leading strand being fully replicated and protected (Fig. 4J). The CRISPR/Cas9-based de novo telomere system we developed provides a powerful genetic tool for the fission yeast community, enabling future high-resolution studies not only of telomere biology but also of other chromosomal dynamics. Finally, given the links between RNase H2 dysfunction and human autoimmune disorders and genome instability [75, 76], our findings raise the intriguing possibility that defects in telomere replication may contribute to these disease pathologies.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was supported by grants from the National Key Research and Development Program of China (No. 2023YFA0913400) and the Shanghai Academy of Natural Sciences (SANS). Funding to pay the Open Access publication charges for this article was provided by the Grants from the National Key Research and Development Program of China and the Shanghai Academy of Natural Sciences.

Data availability

The data underlying this article are available in the article and in its online supplementary material.