{"paper_id":"092f7293-e122-4f4a-a672-c63ee93f19af","body_text":"1 \n \nMANUSCRIPT TITLE \nSingle-molecule sequencing maps replication dynamics across the fission yeast genome, \nincluding centromeres \n \nAUTHORS \nIsabel Díez-Santos1, Sathish Thiyagarajan1, Anna M. Rogers1, Adam T. Watson2, Antony M. \nCarr2 and Conrad A. Nieduszynski1,* \n1 Earlham Institute, Norwich Research Park, Norwich, NR4 7UZ, United Kingdom \n2 Genome Damage and Stability Centre, University of Sussex, Brighton, BN1 9RQ, United \nKingdom \n* To whom correspondence should be addressed. Tel: +44 1603 450946; Email: \nconrad.nieduszynski@earlham.ac.uk \n \nABSTRACT \nDNA replication dynamics in fission yeast remain incompletely understood, particularly in \nrepetitive regions such as the centromeres and the rDNA. Here, we establish DNAscent—a \nnanopore sequencing method that detects BrdU-labelled nascent DNA and infers replication \ndynamics—to map replication at single-molecule resolution in fission yeast. For the first \ntime, we have identified thousands of replication forks, as well as initiation, termination and \npause events, on single sequenced molecules across the whole genome. This high coverage \nallowed us to identify replication patterns in poorly characterised regions. In the rDNA, we \ndetect strand-specific pausing at replication fork barriers. At the mating-type locus, we find \nthe most frequent pause site outside the rDNA. At centromeres, we find that replication \ninitiation predominantly occurs in the outer repeats, while termination localises to central \nregions and that, only in centromere 2, there is an enrichment in pauses at the centromeric \ntRNAs. This work establishes a powerful single-molecule method for studying replication \ndynamics in fission yeast and provides insights into replication across repetitive regions that \nconstitute a significant portion of the genomes of more complex organisms. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n2 \n \nINTRODUCTION \nDNA replication is a fundamental process in all organisms by which the genome is accurately \nduplicated. In eukaryotes, DNA replication initiates at multiple sites—termed origins—across \neach chromosome. From these sites, bidirectional replication forks proceed until they \nterminate when meeting opposing forks. During this process, replication may pause due to \nobstacles such as protein barriers or DNA lesions. Perturbed forks or insufficient replication \ninitiation events can be a source of genome instability (1–3). Therefore, significant effort has \ngone into developing methods to study the sites of replication initiation and the kinetics of \nreplication fork progression.  \nMuch of our understanding of replication dynamics—particularly in yeasts such as budding \nyeast (Saccharomyces cerevisiae; S. cerevisiae) and fission yeast (Schizosaccharomyces \npombe; S. pombe)—has come from population-based, short-read genomic studies (4). In S. \npombe, early genomic studies identified replication origin locations via chromatin \nimmunoprecipitation (ChIP) of origin-associated proteins or the increase in DNA copy \nnumber at active origin sites (5, 6). More recently, a method termed polymerase usage \nsequencing (Pu-seq) identified not only replication initiation, but also termination and fork \ndirection with high resolution (7). Pu-seq locates ribonucleotides erroneously incorporated \ninto DNA by mutant replicative polymerases. Since polymerase ε and polymerase δ primarily \nsynthesise the leading and lagging strands, respectively, the relative usage of these \npolymerases can be used to infer fork direction, and by extension, sites of initiation and \ntermination. These methods rely on short-read sequencing and therefore cannot resolve the \nevents occurring in complex repetitive areas such as the rDNA and centromeres. Moreover, \nthey offer a population-level view of replication dynamics and fail to capture cell-to-cell \nvariation. \nSingle-molecule and single-cell methods address the issue of cellular heterogeneity, but \nsingle-cell approaches often lack the resolution to precisely define replication events (8, 9). In \nfission yeast, DNA combing is the main single-molecule method that has been used to \nvisualise major aspects of replication dynamics (10, 11). Advances in DNA combing allowed \nthe analysis of ultra-long DNA molecules (>1 Mb) and provided information on replication \ndynamics in complex, repetitive areas such as the rDNA, centromeres and telomeres. \nHowever, DNA combing relies on antibody-based detection of nucleotide analogues, which \nlimits its spatial and temporal resolution and makes it challenging to map replication events \nprecisely. Recently, an alternative single-molecule method, DNAscent, was developed to \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n3 \n \nprovide genome-wide, long single-molecule information on replication fork direction, \ninitiation and termination sites, and fork pausing events (12–14). This technique was \nestablished in budding yeast (12), and has been successfully applied in cultured human cells \n(15, 16), and protozoan parasites (17, 18). DNAscent detects synthetic nucleotide analogues, \nsuch as bromodeoxyuridine (BrdU), incorporated into nascent DNA strands during \nreplication (Figure 1A). Varying BrdU incorporation levels with respect to time allowed \nidentification of where replication initiates and terminates, fork direction and sites of fork \npausing.  \nIn this study, we established the experimental conditions to perform DNAscent in S. pombe to \nmap replication dynamics with high coverage and with single-molecule resolution. We \nidentify thousands of molecules with replication fork direction, initiation, termination, and \npause sites across the whole genome including the rDNA, the mating-type locus and the \ncentromeres. We observe clear clustering of replication pause events at previously reported \nreplication fork barriers in the rDNA and the mating-type locus. At the centromeres, we \nidentify termination zones located within the central regions, initiation sites in the \npericentromeric regions and an increase in pauses only in centromere 2. Overall, we show \nthat DNAscent is a high-throughput method to study all major replication dynamic features—\ninitiation, termination, fork direction and fork pauses—across the whole genome in fission \nyeast. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n4 \n \nMATERIAL AND METHODS \nYeast strains \nSchizosaccharomyces pombe (S. pombe) strain AW2224 (h-, smt-0, leu1::leu1:Padh1-\nhENT1, gde1::his7:Padh1-hsvTK, cdc2-asM17, ade6-704) was used in this study. Strain \nAW2224 was created by crossing YDP435 (h-, smt-0, leu1::leu1:Padh1-hENT1, \ngde1::his7:Padh1-hsvTK, ade6-704) and AW2181 (h+, cdc2-asM17, ade6-704, leu1-32, \nura4D18) and random spore analysis was used to screen for the desired genotype (19). The \ncdc2-asM17 allele was confirmed through sensitivity to growth on YE-agar media containing \n3mM 3BrPP1. Intact Padh1-hENT1 and Padh1-hsvTK constructs were confirmed using PCR. \nSaccharomyces cerevisiae (S. cerevisiae) strain ARY017 (MATa, RAD5, BUD4, leu2, ura3, \ntrp1, ade2, his3 ChrVIII:202751::pSAC6-hENT1-tENO2-pPOP2-hsvTK-tTDH1) was used in \nthis study. To construct ARY017, the W303 strain T7107 was transformed with NotI digested \npAR045 which contains hENT1 and hsvTK which were codon optimised for expression in \nbudding yeast. \nCell cycle synchronisation, BrdU/thymidine treatment and flow cytometry \nS. pombe cells were grown in YES medium (Formedium, PCM0310) at 30°C until reaching \nan OD600 = 0.3. Then, cells were arrested in the G2 phase by the addition of 3BrPP1 (stock \nconcentration of 2 mM, Abcam, ab143756) at a 1:1000 dilution, followed by incubation for 3 \nhours. To resume cell cycle progression, 3BrPP1 was removed by vacuum filtration, and cells \nwere resuspended in fresh YES medium at 30°C. Five minutes after filtration, BrdU (Sigma, \nB5002) was added to a final concentration of 0.5, 2, or 4 μM. For thymidine-treated samples, \nthymidine (Sigma, T9250) was added 30 minutes after media filtration at final concentrations \nof 40, 200, or 400 μM. Samples were harvested 60 minutes after media filtration, pelleted by \ncentrifugation at 3,000 × g for 10 minutes, washed with PBS, and stored at -80°C until DNA \nextraction. Flow cytometry samples were collected before G2 arrest (asynchronous culture \ntimepoint), immediately after release (0 min timepoint), and every 5 minutes thereafter until \nsample harvesting to monitor cell-cycle progression. Flow cytometry samples were treated \nwith RNase A (Sigma, R6513) and proteinase K (Sigma, P2308) before DNA staining with \nSYTOX Green (Thermo Fisher, S7020), followed by sonication and analysis using a BD \nFACSAria Fusion cytometer. Flow cytometry profiles were analysed using FlowJo v10.8.1. \nS. cerevisiae cells were grown in YPAD medium (Formedium, CCM1010) at 30°C until \nreaching OD600 = 0.3. Cells were arrested in the G1 phase by adding α-factor (Cambridge \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n5 \n \nBioscience, Y1001) to a final concentration of 0.5 μM. BrdU was added 95 minutes after α-\nfactor addition at final concentrations of 10, 30, or 100 μM. Cell cycle progression was \nresumed 2 hours after α-factor addition by treating cells with pronase (Sigma, 53702) at 200 \nμg/ml. To prevent re-entry into a second cell cycle, nocodazole (Merck, 487928) was added \nat 15 μg/ml 20 minutes after pronase addition. Samples were harvested 90 minutes after \npronase addition, pelleted by centrifugation at 3,000 × rpm for 5 minutes, washed with water, \nand stored at -80°C until DNA extraction. Flow cytometry samples were collected before G1 \narrest (asynchronous culture timepoint), immediately before release (0 min timepoint), every \n10 minutes thereafter for one hour, and before sample harvesting (90 min timepoint) to assess \ncell-cycle progression. Flow cytometry samples were processed and analysed as described for \nS. pombe. \nBiological samples and their treatments are described in Supplementary Table S1. \nDNA extraction, library preparation and nanopore sequencing \nHigh molecular weight DNA extraction from S. pombe and S. cerevisiae was performed \nusing the Nanobind Tissue Kit (PacBio, 102-302-100) with modifications. This product is \nnow the Nanobind PanDNA Kit (PacBio, 103-260-000). Briefly, 100 mg of cell pellets were \nresuspended in 1 ml of Spheroplasting Buffer 1 (100 mM Tris-HCl pH 9.5, 14 mM β-\nmercaptoethanol) followed by cell wall digestion by resuspending the cells in 400 µl of \nSpheroplasting Buffer 2 (1 M Sorbitol, 25 mM EDTA, 1.6 mM Citric Acid, 8.4 mM Sodium \nCitrate) and 100 µl of 50 mg/ml Zymolase 100 T (AMSBIO, 120493-1) in Spheroplasting \nBuffer 2. Cells were incubated for 1-1.5 h on a Thermomixer at 35 °C with 300 rpm shaking \nfor 10 s every 5 min. Following digestion, spheroplasts were washed in PBS and resuspended \nin 20 μl Proteinase K, 50 μl Buffer SB, and 150 μl Buffer BL3. Samples were incubated at 55 \n°C in a ThermoMixer, shaking at 300 rpm for 90 minutes. Next, 20 μl of RNase was added, \nfollowed by an additional 90-minute incubation under the same conditions. Cells were \ncentrifuged at 2,000 x g at room temperature for 3 min. The supernatant was transferred to a \nnew tube using a wide-bore pipette and DNA precipitation and clean-up were performed \naccording to the manufacturer’s instructions. For extraction of DNA from S. cerevisiae, \nZymolase 20 T (AMSBIO, 120491-1) was used instead of Zymolase 100 T, and the RNase \nincubation step was reduced to 30 min. DNA concentration and integrity were assessed using \nthe Qubit dsDNA HS assay kit (Invitrogen, Q33230) and TapeStation (Agilent \nTechnologies). \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n6 \n \nLibraries were prepared using the ligation sequencing kit SQK-LSK109 (Oxford Nanopore \nTechnologies; R9 chemistry) or SQK-LSK114 (Oxford Nanopore Technologies; R10 \nchemistry) as stated by the manufacturer. Libraries were loaded on MinION flow cells FLO-\nMIN106 (Oxford Nanopore Technologies; R9 chemistry) or FLO-MIN114 (Oxford \nNanopore Technologies; R10 chemistry) as stated by the manufacturer. \nSequencing datasets are described in Supplementary Table S2. \nDNAscent pipeline  \nThe detection of BrdU and the inference of replication fork direction, and initiation and \ntermination sites, were performed using a DNAscent/FORKscent pipeline \n(pipeline_fast5_pod5_to_dnascent_dorado.sh) from \nhttps://github.com/DNAReplicationLab/fork_arrest. Briefly, nanopore basecalling was \nperformed using Guppy version 5.0.7 (configuration file dna_r9.4.1_450bps_hac.cfg) or \nGuppy version 6.5.7 (configuration file dna_r10.4.1_e8.2_400bps_5khz_hac.cfg) for R9 or \nR10 data, respectively. Reads were aligned to references genome sequences using minimap2 \nversion 2.24 (20). For S. pombe data, reads were aligned to either the reference genome \nASM294v2 (https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_000002945.1/), a custom \nrDNA array with 22 copies of rDNA, or a de novo assembly of the AW2224 strain. For S. \ncerevisiae data, reads were aligned to the reference genome sacCer3 \n(https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_000146045.2/). For subsequent \nanalyses only primary alignments with a read length ≥1,000 bp, were retained. Except for \nrDNA aligned data, only those alignments with a quality score ≥20 were retained. Alignment \n(bam files) and raw nanopore (fast5 files) data were then used to determine the probability of \nBrdU substitution at each thymidine position using DNAscent v2 or DNAscent v4 \n(https://github.com/MBoemo/DNAscent), for R9 or R10 data, respectively. BrdU \nprobabilities were stored in the modbam file format to allow downstream analyses and \nvisualisation with software tools like samtools, bedtools or modkit. To call replication fork \ndirection, initiation and termination sites, we used the software forkSense from the DNAscent \nv2 package. ForkSense uses the BrdU probabilities to infer replication fork direction and \ninitiation and termination sites. \nBrdU fraction over replication time \nWe compared the BrdU fraction on individual reads with the population median replication \ntime. For S. pombe, the genome replication timing profile data, in 1 kb windows and in 1 \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n7 \n \nminute intervals, was obtained from (7), accession number GSE62108. For S. cerevisiae, the \ngenome replication timing profile data, in 1 kb windows and in 1 minute intervals, was \nobtained from (21), accession number GSE48212. The S. cerevisiae data was lifted from the \nsacCer1 to the sacCer3 genome assemblies using liftOver (22). For each 1 kb replication \ntiming window, the fraction of BrdU (mean BrdU density using a probability threshold of \n0.5) was determined for each read that overlapped for ≥100 thymidines. Levels of BrdU \nsubstitution with respect to replication time were then analysed in 5 minute intervals and \nvisualised as violin plots. \nBrdU probability and fraction over genomic coordinates in single molecules \nTo plot the BrdU probabilities and fractions (or levels of BrdU substitution) from single \nmolecules, we first extracted the modification probabilities from the mod.bam file containing \nthe read of interest. Then, the level of BrdU substitution was determined in 300-thymidine \nwindows using a BrdU probability threshold of ≥0.5. The modification probability and the \nBrdU substitution data for a single read were then plotted relative to genomic coordinates. \nBrdU probabilities and substitution level were plotted as grey dots and a black line, \nrespectively. \nHeatmap of BrdU fractions  \nTo produce a heatmap of BrdU substitution level along multiple reads, we extracted the \nmodification probabilities using modkit (https://github.com/nanoporetech/modkit). Then, we \napplied a threshold of 0.5, so that modification probabilities <0.5 were converted to 0 and \nprobabilities ≥0.5 were converted to 1. These values were averaged, in windows of 1000 \nnucleotides, to determine the fraction of BrdU substitution. \nFraction of leftward-moving forks \nTo calculate the fraction of leftward replication forks, we counted the number of left and right \nforks at every base using bedtools genomecov, averaged the counts per 1000 bins and \ncalculated the fraction of left forks by dividing the number of left forks by the total number of \nforks. At each genomic interval, we compared the fraction of leftward forks from all nascent \nsingle molecules to the equivalent value from Pu-seq data. To assess correlation, we \ncalculated the Pearson correlation coefficient. The Pu-seq fraction leftward replication forks \ndata (GSE62108_PU-seq_leftward_moving_fork.wig) was obtained from (7). \nReplication initiation and termination site count \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n8 \n \nTo visualise DNAscent initiation and termination site frequency across the genome, we first \ncalculated the midpoint of each DNAscent initiation/termination site. Then, the number of \ninitiation/termination site midpoints was determined in 2 kb windows across the genome. \nComparison of DNAscent replication initiation sites to published datasets \nTo calculate the closest distance from the DNAscent initiation sites to initiation sites from \nindependent published datasets, we used bedtools closest. Then, we grouped the distances in \n1 kb bins, counted the number of events per bin, and plotted the count as a histogram in 1 kb \nbins. DNAscent initiation sites from the mitochondrial genome were excluded from the \nanalysis. To evaluate the significance of the distribution of distances, we compared them to \nthe average distance from 1000 independent randomisations of the DNAscent initiation sites. \nFor each randomisation, the observed DNAscent initiation site genomic intervals were \nshuffled (bedtools shuffle) across the reference genome.  \nWe compare previously reported (Pu-seq) initiation site efficiencies with the number of \nobserved DNAscent initiation events. For each Pu-seq initiation site we determined the \nnumber of DNAscent initiation site midpoints within 1 kb (either side). We then plotted the \nDNAscent initiation site counts with respect to the Pu-seq initiation efficiency as jitter dots \nwith boxplots.  \nThe file with Pu-seq initiation sites (GSE62108_origin-effici.bedgraph) was obtained from \n(7). Initiation sites from (5, 23, 24) where obtained from OriDB (25). \nComparison of DNAscent and Pu-seq replication termination sites \nTo compare the DNAscent termination site midpoint counts with the Pu-seq termination \nevent frequencies, we used 1 kb windows (sliding by 300 bp) since this is how the Pu-seq \ntermination data was reported. The data were analysed and plotted in bins of 2% Pu-seq \ntermination frequency. The file with Pu-seq termination frequencies (GSE62108_PU-\nseq_terminaton-events.wig) was obtained from (7). \nPause site detection \nTo detect pause sites on single molecules, we used the rDNA_detect.py pipeline from \nhttps://github.com/DNAReplicationLab/fork_arrest. Briefly, we selected those molecules \nwith a mean BrdU fraction of at least 0.05 and measured the BrdU fraction difference \nbetween an upstream and downstream window at each thymidine per molecule (window size \n3x 290T or approximately 3 kb in the yeast genome). Then, we detected peaks in the density \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n9 \n \ndifference, requiring that peaks are at least 15x 290T apart on the genome (approximately 15 \nkb). Then, we collated peaks in density difference across all molecules, retaining only those \nwith step sizes greater than 3x the standard deviation of all step sizes across all molecules. \nWe filtered out those peaks which were closer than 5 kb to the ends of molecules, and those \nwith maximum and minimum BrdU densities ≤0.5 or ≤0.01, respectively, in their upstream or \ndownstream windows. \nVisualisation of replication dynamics \nTo visualise the location of left and right forks, initiation sites, termination sites and pause \nsites, we used custom R scripts that took the genomic coordinates of each feature in the reads \nof interest, and represented them as blue arrows (left forks), right arrows (right forks), black \nrectangles (initiation sites), grey rectangles (termination sites), green triangles (pause sites in \nleading strand) and pink triangles (pause sites in lagging strand).  \nTo visualise the fraction of leftward-moving forks, initiation event counts, origin efficiencies, \ntermination event counts, termination event frequencies and pause event counts we used \nDnaFeaturesViewer (https://edinburgh-genome-foundry.github.io/DnaFeaturesViewer/).  \nDesign of the rDNA sequence \nTo study the ribosomal DNA (rDNA), we generated an rDNA sequence with 22 copies of the \nrDNA unit with the replication fork barrier (RFB) region in the middle. Each rDNA unit \nconsists of a 10.9 kb sequence obtained from the reference genome ASM294v2, coordinates \nchrIII:5539-16411. The rDNA sequence was annotated using the gene coordinates from the \nreference genome ASM294v2, the origin of replication ARS3001 coordinates from (26) \n(GenBank: AF040270.1) and the locations of the RFB based on (27, 28). \nDe novo assembly  \nTo obtain a reference sequence for strain AW2224, we used high-quality long reads from the \nsequencing dataset IDS_65-R9 (see Supplementary Table S2). Only reads with a quality \nscore >9 and read length ≥75 kb were used. We then performed a de novo assembly using \nFlye (v2.9) with an estimated genome size of 14 Mb. The resulting contigs were evaluated by \ncomparing them to the previously published assembly (29) and the reference genome \nASM294v2 using MUMmer (v3.23). All S. pombe chromosomes were recovered as single \ncontigs, and all contigs had a sequencing coverage >30. \nAnnotation of the mating-type locus and the centromeres \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n10 \n \nTo annotate the mating-type locus in the reference sequence for strain AW2224, mat genes \nand flanking genes were identified using BLAST+ (v2.9.0) (30), with known gene sequences \nfrom ASM294v2 as queries. The coordinates of the replication fork barrier RTS1 were \ndetermined by aligning its partial sequence from (31), while those of replication fork barrier \nMPS1 were identified using the MPS1 sequence from (32). The 263-bp smt-0 deletion (33) \nwas confirmed by aligning the mating-type locus from strain AW2224 with the mating-type \nlocus from ASM294v2. \nCentromeric repeats, core elements and flanking genes were identified using BLAST+ \n(v2.9.0), with known repeat, core and flanking gene sequences from ASM294v2 as queries. \nThe locations of tRNA genes were predicted using tRNAscan-SE (34). \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n11 \n \nRESULTS \nBrdU incorporation in fission yeast serves as a proxy for replication time \nNanopore sequencing enables inference of DNA replication dynamics through the detection \nof synthetic analogues of thymidine, such as bromodeoxyuridine (BrdU), that are \nincorporated into nascent strands during S-phase in a time-dependent manner (Figure 1A) \n(12, 13). This method was established in S. cerevisiae, where BrdU levels incorporated into \nthe DNA naturally decrease as replication progresses (12, 35). To detect BrdU incorporation, \ncells are fed BrdU, harvested post-S phase, high-molecular weight genomic DNA extracted \nand subjected to nanopore sequencing. The resulting sequence data are basecalled, aligned to \nthe reference genome sequence and sites of BrdU incorporation are determined using the \nDNAscent software (see Methods). The DNAscent software assigns a probability of BrdU \nsubstitution (grey circles; Figure 1A) to each reference thymidine position across every \nsequencing read. Positions with a BrdU probability ≥0.5 are considered positive BrdU calls, \nand the fraction of BrdU incorporation is calculated in 300-thymidine windows across each \nsequencing read (black lines; Figure 1A).  \nTo assess the profiles of BrdU incorporation in S. pombe and S. cerevisiae, we treated \nsynchronised cells with a range of BrdU concentrations during a single S phase. None of the \nBrdU treatments affected cell cycle progression, as shown by flow cytometry profiles of \ncellular DNA content (Supplementary Figure 1). In S. pombe, we observed that at the single-\nmolecule level, BrdU incorporation was constant along the DNA and did not correlate with \npreviously published median replication time from a cell population (Figure 1B). In contrast, \nS. cerevisiae cells exhibited BrdU incorporation gradients that correlated with cell population \nmedian replication timing data (Figure 1C). To obtain BrdU gradients in S. pombe, we added \na thymidine chase 30 minutes after the BrdU treatment – a time that corresponds with mid-S \nphase (Figure 1D and Supplementary Figure 1). In the same genomic location as shown in \nFigure 1B, we observed that the single molecules had BrdU incorporation levels resembling \nthe median replication time. This result indicated that a thymidine chase led to BrdU decrease \nwith respect to replication time. To further assess the relationship between BrdU \nincorporation and replication time, we plotted all single-molecule BrdU fractions against \npopulation level median replication times. In S. cerevisiae, BrdU incorporation decreased \nwith later replication times at all tested concentrations (Figure 1E; top), supporting the use of \nBrdU levels as a replication timing proxy. These observations are consistent with previous \nreports from our group and others (12, 35). In contrast, S. pombe exhibited no decline in \nBrdU incorporation with replication time unless a thymidine chase was applied (Figure 1E; \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n12 \n \nmiddle and bottom). Equivalent results were obtained using the newest nanopore sequencing \nchemistry (R10), although we observe that BrdU detection in R10 data was noisier, especially \nat late replication times (Supplementary Figure 2). These results indicate that BrdU \nincorporation kinetics differ between S. cerevisiae and S. pombe, and that in S. pombe, \nreplication timing cannot be inferred from endogenous BrdU incorporation patterns alone. \nHowever, a thymidine chase produces BrdU gradients that correlate with replication time, \nindicating that single-molecule BrdU fractions can also serve as a proxy of replication time in \nS. pombe. \nBrdU incorporation levels indicate replication fork direction, initiation and termination \nsites \nIn S. pombe, the addition of a thymidine chase in mid-S-phase resulted in high levels of BrdU \nincorporation followed by gradients of declining BrdU levels as replication proceeded \n(Figure 1D). These gradients resemble those observed in S. cerevisiae (12, 13) (Figure 1C) \nwhich allowed us to computationally infer replication fork direction, and call initiation and \ntermination sites on individual sequencing reads using the forkSense software (13). Positive \nBrdU gradients are identified as leftward-moving forks (blue arrows), whereas negative BrdU \ngradients are identified as rightward-moving forks (red arrows) (Figure 2A). Diverging \nleftward- and rightward-moving forks indicate initiation sites (black rectangles) and \nconverging leftward- and rightward-moving forks indicate termination sites (grey rectangles). \nIn a dataset from S. pombe cells treated with 4 µM BrdU and 400 µM thymidine, we obtained \n462,052 nascent single molecules in which we identified 370,298 forks, 65,098 initiation sites \nand 39,157 termination sites. As an example, we show all the single molecules with ≥1 forks \nin a 100-kb region from chromosome I (Figure 2B and C). BrdU densities are represented as \na heatmap in Figure 2B with one single molecule per row. The aligned single molecules \nclearly show variable levels of BrdU incorporation across this genomic locus. Most single \nmolecule reads show a peak of high BrdU incorporation (green-yellow; Figure 2B) around \nposition 845 kb indicating early replication, in contrast to a region with low BrdU \nincorporation (blue; Figure 2B) at around 865 kb indicating later replication. Declining levels \nof BrdU incorporation to the left and right indicate progressively later replication times and \ntherefore replication fork direction. Transitions from left- to rightward forks are identified as \ninitiation sites and right- to leftward as termination sites. Across the example locus on \nchromosome I, we observe alternating regions of predominantly leftward (blue arrows) and \nrightward (red arrows) fork direction (Figure 2C). The position of transitions from leftward to \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n13 \n \nrightward fork direction are similar between single molecules, for example at ~845 kb where \nmost molecules have diverging forks indicating an initiation site. By contrast, there is more \nvariability between molecules in the position of converging forks (for example at ~865 kb) \nthat indicate termination sites.  \nTo validate the single molecule replication fork direction, and initiation and termination sites, \nwe compared them to those identified by previous cell-population studies (5, 7, 23, 24). To \nallow direct comparison with the cell population data from these studies, we calculated an \nensemble of our single-molecule data. For replication fork direction, we calculated the \nfraction of leftward-moving forks (1 kb window, 300 bp steps) across the genome using all \nour single-molecule data. For initiation and termination sites, we counted the number of \nevents in 2 kb windows. When comparing the ensemble single molecule and Pu-seq fraction \nof leftward replication forks, we observed a strong correlation (R = 0.82) between the datasets \n(Figure 2D). At the example genomic locus, we see close agreement between the ensemble \nfraction of leftward forks from our single-molecule data and the Pu-seq data (Figure 2E). We \nthen compared the single-molecule initiation and termination sites, to those inferred from Pu-\nseq data. Half of the single-molecule initiation events (33,428 out of 60,731 sites) were \nwithin 2 kb of a Pu-seq initiation site, and this intersection was highly statistically significant \n(Supplementary Figure 3A). Similar results were obtained when looking at the intersection of \nsingle-molecule initiation sites and other independent studies (Supplementary Figure 3B-D). \nThe Pu-seq methodology determines both the location and efficiency of origins (fraction of \ncells in which the origin initiates DNA replication). Our dataset has a ~100x coverage of \nreads with ≥1 initiation site, therefore, we reasoned that the number of times an initiation was \nobserved on single molecules would indicate the efficiency of the origin. To test this, we \ncompared the count of single-molecule initiations with the Pu-seq efficiency measurement at \neach Pu-seq origins (Figure 2F). We observe a positive correlation between the count of \nsingle-molecule initiation events and Pu-seq origin efficiency. We also found a positive \ncorrelation between the count of single-molecule termination events and frequency of Pu-seq \nterminations (Figure 2G). This is reflected in the example locus (Figure 2E), where there is a \nstriking similarity between the single molecule and the Pu-seq data for fraction of leftward-\nmoving forks, and location and efficiency of initiation and termination sites. Together these \nfindings validate that the DNAscent methodology provides single molecule information on \nreplication fork direction and the location of initiation and termination events. In addition, \nensembles of single molecules provide a measure of the frequency of replication initiation \nand termination across the genome.  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n14 \n \nBrdU incorporation reveals replication pause locations in leading and lagging strands at \nthe rDNA replication fork barrier \nWe have shown that BrdU incorporation levels in single DNA molecules can be used to infer \nreplication fork directions, as well as initiation and termination sites. BrdU densities can also \nindicate the location of replication pauses, as previously demonstrated in S. cerevisiae (12, \n14) where sudden drops in BrdU density denote replication pauses within the ribosomal DNA \n(rDNA). Fission yeast rDNA consists of 100–150 tandem repeats, each containing the \nribosomal RNA genes, a replication origin, and a region containing four replication fork \nbarriers (RFBs) (Figure 3A) (36). RFBs are located downstream of the largest ribosomal gene \nto arrest rightward-moving forks, thereby limiting replication in the opposite direction to \ntranscription (37). Thus, we anticipated that we would observe a sudden drop in BrdU density \nindicating replication pauses at the RFB. First, we aligned our single-molecule dataset (4 µM \nBrdU / 400 µM thymidine) to a reference sequence of 22 rDNA repeats. Second, we used \nDNAscent to detect BrdU and forkSense to infer the position of replication forks, initiation \nand termination events. Third, paused replication forks were computationally identified from \nlarge decreases in BrdU density (see Methods), thus prioritising a low false positives rate at \nthe expense of identifying shorter pauses that would present as a smaller decrease in BrdU \ndensity (14). Among the 8,165 nascent reads that aligned to rDNA, we identified 12,677 \nforks, 3,654 initiation sites, 2,079 termination sites, and 376 pauses. Finally, combining \ninformation about fork direction and sequenced strand, we labelled each pause site as from a \nleading or lagging strands. \nIn example single molecules, aligned to the custom rDNA sequence, we observe clear drops \nin BrdU density that align with RFBs (Figure 3B). Interestingly, these molecules show that \nonly a fraction of the origins of replication were active (~1 in 3), consistent with observations \nfrom DNA combing (10). Then, we visualised all reads containing ≥1 pause and observed \nthat anticipated drops in BrdU density aligned to the RFB (Supplementary Figure 4A). We \nalso observed that most initiation sites overlapped with the previously characterised \nreplication origin, and that rDNA units are predominantly replicated by leftward forks, thus \nensuring co-directionality of transcription and replication (Supplementary Figure 4B). Next, \nwe collapsed the replication pause site data onto a single representative rDNA repeat. The \nidentified sites of pausing were predominantly on rightward-moving forks and enriched at the \nRFB location (Figure 3C). We observe that the leading strand pauses are more tightly \nclustered and closer to the RFB than the lagging strand pauses. These results confirm that the \nDNAscent methodology can identify replication pause sites and reveal the strand of synthesis \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n15 \n \nand fork direction, giving a more detailed and quantitative view of replication dynamics at \nthe rDNA (Figure 3D). \nFrequent replication pausing in the mating-type locus \nThe ability to detect pauses in replication within the rDNA prompted us to investigate pauses \nthroughout the unique genome. To test this, we mapped the data to a de novo assembly of our \nstrain AW2224 (see Methods). This allowed us to analyse genomic regions that differ \nbetween this strain and the reference genome, including the centromeres and mating-type \nlocus. Across the whole unique genome, the most prominent identified replication pause site \nwas within the mating-type locus (Supplementary Figure 5). The mating type of S. pombe is \ndetermined by the mat1 locus located at chromosome II. Flanking the mat1 gene, there are \ntwo elements that control the direction of replication across the mat1 locus: RTS1 and MPS1 \n(38) (Figure 4A). The RTS1 element is downstream of mat1 and acts as a polar terminator of \nreplication, terminating rightward-moving forks (39). The MPS1 element is upstream of mat1 \nand acts as a polar pause, stalling leftward-moving forks (40). Thus, the mat1 locus is \npredominantly replicated by leftward-moving forks that occasionally pause at the MPS1 site. \nThis tightly controlled replication pattern is proposed to allow the formation of an imprint \n(ribonucleotides and/or a nick) at the MPS1 site, which can lead to mating-type switching \n(41). However, the frequency of the pausing at this site in vivo is not known. \nOur single-molecule dataset allowed us to observe two distinct patterns of replication at the \nmat1 locus: continuous replication by a leftward fork, or leftward-moving fork pauses at the \nMPS1 site (Figure 4B). Unperturbed fork progression was the most frequent scenario (~60 % \nof molecules) (data not shown). Ensemble of replication features showed a clear termination \narea upstream mat1 which aligns with the expectation of right forks terminated at the RTS1 \nelement (Figure 4C). All pauses were found in left forks, and mapped to the MPS1 element. \nThese results indicate that the mat1 locus is almost exclusively replicated by a leftward-\nmoving fork (94%) that may pause (~40% of molecules) at the MPS1 site or continue through \nmat1, and that most rightward-moving forks terminate at the RTS1 site. \nReplication dynamics across the highly repetitive centromeres \nThe high levels of repetitive sequence across S. pombe centromeres have precluded \ncomprehensive analysis of the replication dynamics by previous short-read based genomic \nstudies. Each of the three centromeres is comprised of a central core flanked by inner (imr) \nand outer (dg and dh) repeats containing tRNAs (Figure 5A) (42). The inner repeats are \nunique for each centromere, but the outer repeat sequences are shared among centromeres \nand are heterochromatic. The central core is the site of kinetochore assembly; the sequence of \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n16 \n \ncentromere 2 central core (cnt2) is unique, while those of centromeres 1 and 3 (cnt1 and cnt3) \nare highly similar. Origins of replication have been found in the repeats of centromere 2 but \nnot within the central core (43, 44), and it has been proposed that the centromeric tRNAs act \nas barriers to DNA replication (45). Therefore, we sought to use the single-molecule \nreplication dynamics dataset to comprehensively analyse the landscape of replication across \nall three S. pombe centromeres. \nAfter careful examination of published S. pombe genomes sequences we determined that the \ncentromeric sequences were either incomplete or showed significant structural differences \ncompared to our data (not shown). Therefore, we used our de novo genome assembly and \nannotated the centromeric regions for our strain (see Methods). Using this assembly as a \nreference for DNAscent and forkSense analyses, we identified 4,748 forks, 1,116 initiations, \n729 terminations, and 285 pauses intersecting with the three centromeres. As an example, we \nshow a single molecule that aligned to centromere 1 (Figure 5B). Across centromere 1, most \ninitiation sites were located outside the central core, particularly in the first dh repeat (Figure \n5C). Termination sites were associated with the inner repeat downstream of the central core, \nand replication pauses were distributed throughout the centromere. For centromere 2, \ninitiation sites were found throughout the centromere, especially in the outer repeats (Figure \n5D). A distinct enrichment in termination sites was observed across the central core, and \nnotably, replication pauses were most frequent in proximity to the tRNAs within the outer \nrepeat sequences. Centromere 3 DNA replication resembled centromere 1, with clustered \ninitiation sites in the outer repeats and termination sites both in the central domain and in the \nrepeats, but with no clear enrichment of pauses (Figure 5E). These findings indicate that all \ncentromeres have replication origins outside the central core, termination events within the \ncentral region, and that centromere 2 may be unique in showing replication pausing close to \ntRNAs. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n17 \n \nDISCUSSION \nIn this study, we demonstrated that direct nanopore sequencing and detection of the \nnucleotide analogue BrdU can determine replication dynamics across the S. pombe genome. \nThis methodology, which we named DNAscent, captures all major features of DNA \nreplication—initiation, fork progression, termination, and pausing—on single molecules. \nAdditionally, it has the advantage of reporting the strand of synthesis which allows pauses on \nleading and lagging strands to be distinguished. Moreover, nanopore sequencing enabled \n>200× coverage of long (>30 kb, up to ~120 kb) nascent molecules genome-wide in a single \nMinION run. Such high coverage facilitates detection of rare events that are challenging to \nresolve with population-level methods, and also makes it possible to detect replication \ndynamics in repetitive sequences that cannot be studied with short-read sequencing methods. \nTo validate our findings, we compared them against several independent published datasets \n(Figure 2 and Supplementary Figure 3). Initiation sites aligned with those identified in prior \nstudies, with higher counts correlating with more efficient initiation sites. Fork direction and \ntermination sites also correlated with published Pu-seq data (7). Finally, most frequent pauses \nin the rDNA and the rest of the genome correlated with well-characterised replication fork \nbarriers (Figure 3 and 4). Altogether, this approach provides a detailed view of replication \ndynamics across the whole genome that complements and extends previous studies (5–7, 10, \n11, 23, 24). \nTo map replication dynamics, we measured changes in BrdU concentration with respect to \ntime. In S. cerevisiae, endogenous mechanisms lead to decreases in BrdU levels over time, \nwhereas in S. pombe BrdU levels were constant throughout S-phase (Figure 1). Therefore, in \nthis regard S. pombe is like other eukaryotes (15) and suggests different mechanisms of \ndNTPs pool regulation in budding yeast compared to other eukaryotes. Although dTTP levels \nhave been reported to rise in fission yeast during S-phase (46), the magnitude of this increase \nis notably less than in budding yeast. This may reflect species-specific regulation of \nribonucleotide reductase, the enzyme responsible for deoxyribonucleotide biosynthesis (47). \nNonetheless, we demonstrate that in fission yeast a thymidine chase can be used to \nexperimentally reduce BrdU levels with respect to time. Similar approaches might be used for \nother organisms that, like fission yeast, do not naturally show BrdU gradients. \nThe repetitive rDNA locus, comprising 100–150 tandem 10.9 kb units, poses particular \nchallenges for replication analysis. Our long-read sequencing approach allowed us to resolve \nreplication patterns across multiple contiguous rDNA repeats, revealing two primary modes: \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n18 \n \nactive initiation within the rDNA repeat (~1 active origin in 3 rDNA repeats), and passive \nreplication by forks entering from upstream (Figure 3B). Replication fork barriers shaped \nthese patterns, and we observed a concentration of replication pausing events near a region \nwith four replication fork barriers (48). Notably, lagging-strand pauses showed greater \npositional variability than those on the leading strand, consistent with the discontinuous \nsynthesis of Okazaki fragments. Leading-strand pauses, by contrast, often occurred at four \ndistinct positions that likely correspond to the four replication barriers (Figure 3C).  \nA de novo assembly allowed us to look at single-molecule replication dynamics in under \nstudied areas, such as the mating-type locus and the centromeres. The mating-type locus is \ncomprised of a heterochromatic silenced region with the mat2 and mat3 genes, and an \nactively transcribed region with mat1 flanked by a replication termination site (RTS1) and a \nreplication pause site (MPS1). We observed replication fork pausing at the MPS1 site, leading \nto leftward replication and a downstream shift in termination (Figure 4). While this behaviour \naligns with previous models (31, 32), it is now directly visualised at the single-molecule level \nfor the first time and it allowed us to estimate that ~40% of the cells pause in this location. In \nall centromeres, we found replication initiation sites within the outer centromeric repeats, \nwhereas the central region functions as a termination zone (Figure 5). These findings are \nconsistent with models proposing that replication dynamics are spatially organised to \npreserve centromere function (49). We also observed a modest enrichment of replication \npausing at tRNA genes within centromere 2. This includes the tRNAAla gene in the imr2 \nrepeat, previously identified as a chromatin barrier (50). To our knowledge, this is the first \ndirect observation of increased fork pausing at this locus, linking replication dynamics to \nknown chromatin features in centromeric regions. \nWe also identified parallels between the organisation of the mating-type locus and \ncentromeres. The mat2 and mat3 region shares heterochromatic features with the outer \ncentromeric repeats, while mat1, like the centromere central cores, is euchromatic and \nassociated with late replication. These similarities suggest a conserved strategy in which \nreplication timing and pausing contribute to local chromatin structure. Fork pausing at the \nMPS1 site may facilitate lagging-strand processes such as primer placement or chromatin \nmodification (38). A comparable mechanism has been proposed for centromere 2, where \npausing near flanking tRNA genes appears to protect the central core from inappropriate \nhistone deposition (45). Interestingly, such pausing was absent at centromeres 1 and 3, \nindicating a different type of regulation. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n19 \n \nIn the future, DNAscent, alongside other nanopore-based methods such as FORK-seq (51), \noffer new opportunities for studying replication at single-molecule resolution. In particular, \nthe ability to calculate fork velocity across different chromatin states will allow testing of \nwhether replication is slowed in heterochromatin. Fission yeast offers a powerful model for \nthis, given its compact genome and presence of mammalian-like heterochromatin, which is \nabsent in budding yeast (52). Furthermore, the capacity to track multiple molecular signals—\nincluding DNA modifications like methylation—on the same single molecule paves the way \nfor integrated analyses of epigenetic and replication dynamics (53). Future studies could \nsystematically map replisome pausing sites (14) and extend these analyses to mutants \naffecting chromatin structure (e.g. sir2, clr3), chromosome organisation (e.g. swi6), or \nreplication fork stability (e.g. swi1, swi3). \nIn summary, our findings establish DNAscent as a powerful tool for dissecting replication \ndynamics at single-molecule resolution in S. pombe. These principles hold true in more \ncomplex eukaryotes and can be applied to any organism that can incorporate BrdU. \n \nDATA AVAILABILITY \nBrdU calls on all aligned reads (mod.bam format), inferred replication dynamics (bedgraph \nfiles), reference sequences, annotations and replication timing datasets are available from \nZenodo under the DOI 10.5281/zenodo.15365203. The deposited files are listed in \nSupplementary Table S3. Code described here is available from the GitHub repositories \n(https://github.com/DNAReplicationLab/pombe_replication and \nhttps://github.com/DNAReplicationLab/fork_arrest). \nData will be publicly available after the review process. Please contact the corresponding \nauthor for access before then. Scripts were written in shell script, python, and R. \n \nSUPPLEMENTARY DATA \nSupplementary Data are available online. \n \nAUTHOR CONTRIBUTIONS \nIsabel Díez-Santos: Conceptualization, Formal analysis, Investigation, Methodology, \nVisualization, Writing – original draft, Writing – review & editing. Sathish Thiyagarajan: \nMethodology, Software, Writing – review & editing. Anna M. Rogers: Investigation, Writing \n– review & editing. Adam T. Watson: Methodology. Antony M. Carr: Funding acquisition, \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n20 \n \nResources, Supervision, Writing – review & editing. Conrad A. Nieduszynski: \nConceptualization, Funding acquisition, Resources, Supervision, Writing – original draft, \nWriting – review & editing. \n \nACKNOWLEDGEMENTS \nWe are grateful to colleagues at the Earlham Institute for their support and expertise: Dr. \nAndrew Goldson for guidance on flow cytometry, Grant Bexson for laboratory assistance, \nand Dr. Darren Heavens for advice on library preparation. We also extend our thanks to Dr. \nMartin Ayling, Sam Gallop, and Kamil Hepak from the Norwich Biosciences Institutes \nResearch Computing team for their assistance with software installation and maintenance. \nWe acknowledge Dr. Emma Heron and Dr. Yasukazu Daigaku for their prior work on S. \npombe and for providing the YDP435 strain. The T7107 strain was kindly gifted by Prof. \nTomoyuki Tanaka. Finally, we sincerely thank Prof. Crisanto Gutiérrez, Dr. Bénédicte \nDesvoyes, and Nerea Murugarren for their critical reading of the manuscript and valuable \nfeedback. \n \nFUNDING \nThis work was supported by the Biotechnology and Biological Sciences Research Council \n(BBSRC), part of UK Research and Innovation, through the following response-mode project \ngrants: BB/W01520X/1 (CAN and IDS) and BB/W014793/1 (AMC and ATW) (Role of \nSenataxins in resolving transcription-replication conflicts) and BB/W006014/1 (CAN and \nAMR) (Single molecule detection of DNA replication errors). The work was also supported \nby Core Capability Grant BB/CCG2220/1 (CAN and ST) at the Earlham Institute and the \nEarlham Institute Strategic Programme Grant Cellular Genomics BBX011070/1 (CAN and \nST) and its constituent work packages BBS/E/ER/230001B (CellGen WP2 Consequences of \nsomatic genome variation on traits). This research was supported in part by NBI Research \nComputing through use of the High-Performance Computing system and Isilon storage. Part \nof this work was delivered via Transformative Genomics, the BBSRC funded National \nBioscience Research Infrastructure (BBS/E/23NB0006) at Earlham Institute, by members of \nthe Single-Cell and Spatial Analysis Group. Funding for open access charge: UKRI1532: \nOABG 2025. \n \nCONFLICT OF INTEREST \nNone. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n21 \n \n \nREFERENCES \n1. Lambert,S. and Carr,A.M. (2013) Impediments to replication fork movement: stabilisation, \nreactivation and genome instability. Chromosoma, 122, 33–45. \n2. Thakur,B.L., Ray,A., Redon,C.E. and Aladjem,M.I. (2022) Preventing excess replication \norigin activation to ensure genome stability. Trends in Genetics, 38, 169–181. \n3. Berkemeier,F., Cook,P.R. and Boemo,M.A. (2025) DNA replication timing reveals \ngenome-wide features of transcription and fragility. Nat Commun, 16, 4658. \n4. Hulke,M.L., Massey,D.J. and Koren,A. (2020) Genomic methods for measuring DNA \nreplication dynamics. Chromosome Res, 28, 49–67. \n5. Hayashi,M., Katou,Y., Itoh,T., Tazumi,M., Yamada,Y., Takahashi,T., Nakagawa,T., \nShirahige,K. and Masukata,H. (2007) Genome‐wide localization of pre‐RC sites and \nidentification of replication origins in fission yeast. The EMBO Journal, 26, 1327–\n1339. \n6. Xu,J., Yanagisawa,Y., Tsankov,A.M., Hart,C., Aoki,K., Kommajosyula,N., \nSteinmann,K.E., Bochicchio,J., Russ,C., Regev,A., et al. (2012) Genome-wide \nidentification and characterization of replication origins by deep sequencing. Genome \nBiology, 13, R27. \n7. Daigaku,Y., Keszthelyi,A., Müller,C.A., Miyabe,I., Brooks,T., Retkute,R., Hubank,M., \nNieduszynski,C.A. and Carr,A.M. (2015) A global profile of replicative polymerase \nusage. Nat Struct Mol Biol, 22, 192–198. \n8. Sakamoto,M., Hori,S., Yamamoto,A., Yoneda,T., Kuriya,K. and Takebayashi,S. (2022) \nscRepli-Seq: A Powerful Tool to Study Replication Timing and Genome Instability. \nCytogenetic and Genome Research, 162, 161–170. \n9. van den Berg,J., van Batenburg,V., Geisenberger,C., Tjeerdsma,R.B., de Jaime-\nSoguero,A., Acebrón,S.P., van Vugt,M.A.T.M. and van Oudenaarden,A. (2024) \nQuantifying DNA replication speeds in single cells by scEdU-seq. Nat Methods, 21, \n1175–1184. \n10. Kaykov,A. and Nurse,P. (2015) The spatial and temporal organization of origin firing \nduring the S-phase of fission yeast. Genome Res, 25, 391–401. \n11. Kaykov,A., Taillefumier,T., Bensimon,A. and Nurse,P. (2016) Molecular Combing of \nSingle DNA Molecules on the 10 Megabase Scale. Sci Rep, 6, 19636. \n12. Müller,C.A., Boemo,M.A., Spingardi,P., Kessler,B.M., Kriaucionis,S., Simpson,J.T. and \nNieduszynski,C.A. (2019) Capturing the dynamics of genome replication on \nindividual ultra-long nanopore sequence reads. Nat Methods, 16, 429–436. \n13. Boemo,M.A. (2021) DNAscent v2: detecting replication forks in nanopore sequencing \ndata with deep learning. BMC Genomics, 22, 430. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n22 \n \n14. Thiyagarajan,S., Rogers,A.M., Müller,C.A. and Nieduszynski,C.A. (2025) Single-\nmolecule landscape of DNA replication pausing. bioRxiv. \n15. Carrington,J.T., Wilson,R.H.C., de La Vega,E., Thiyagarajan,S., Barker,T., Catchpole,L., \nDurrant,A., Knitlhoffer,V., Watkins,C., Gharbi,K., et al. (2025) Most human DNA \nreplication initiation is dispersed throughout the genome with only a minority within \npreviously identified initiation zones. Genome Biology, 26, 122. \n16. Jaworski,J.J., Pfuderer,P.L., Czyz,P., Petris,G., Boemo,M.A. and Sale,J.E. (2025) ecDNA \nreplication is disorganised and vulnerable to replication stress. \n10.1101/2025.03.22.644567. \n17. Damasceno,J.D., Silva,G.L.A., Marques,C.A., Krasilnikova,M., Lapsley,C., Beraldi,D. \nand McCulloch,R. (2024) Nuclear DNA replication in Leishmania major relies on a \nsingle constitutive origin per chromosome supplemented by thousands of stochastic \ninitiation events. 10.1101/2024.11.14.623610. \n18. Totañes,F.I.G., Gockel,J., Chapman,S.E., Bártfai,R., Boemo,M.A. and Merrick,C.J. \n(2023) A genome-wide map of DNA replication at single-molecule resolution in the \nmalaria parasite Plasmodium falciparum. Nucleic Acids Research, 51, 2709–2724. \n19. Moreno,S., Klar,A. and Nurse,P. (1991) Molecular genetic analysis of fission yeast \nSchizosaccharomyces pombe. Methods Enzymol, 194, 795–823. \n20. Li,H. (2018) Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics, 34, \n3094–3100. \n21. Müller,C.A., Hawkins,M., Retkute,R., Malla,S., Wilson,R., Blythe,M.J., Nakato,R., \nKomata,M., Shirahige,K., de Moura,A.P.S., et al. (2014) The dynamics of genome \nreplication using deep sequencing. Nucleic Acids Res, 42, e3. \n22. Hinrichs,A.S., Karolchik,D., Baertsch,R., Barber,G.P., Bejerano,G., Clawson,H., \nDiekhans,M., Furey,T.S., Harte,R.A., Hsu,F., et al. (2006) The UCSC Genome \nBrowser Database: update 2006. Nucleic Acids Research, 34, D590–D598. \n23. Heichinger,C., Penkett,C.J., Bähler,J. and Nurse,P. (2006) Genome-wide characterization \nof fission yeast DNA replication origins. EMBO J, 25, 5171–5179. \n24. Segurado,M., de Luis,A. and Antequera,F. (2003) Genome-wide distribution of DNA \nreplication origins at A+T-rich islands in Schizosaccharomyces pombe. EMBO Rep, \n4, 1048–1053. \n25. Siow,C.C., Nieduszynska,S.R., Müller,C.A. and Nieduszynski,C.A. (2012) OriDB, the \nDNA replication origin database updated and extended. Nucleic Acids Res, 40, D682-\n686. \n26. Kim,S.-M. and Huberman,J.A. (1998) Multiple Orientation-Dependent, Synergistically \nInteracting, Similar Domains in the Ribosomal DNA Replication Origin  of the \nFission Yeast, Schizosaccharomyces pombe. Mol Cell Biol, 18, 7294–7303. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n23 \n \n27. Krings,G. and Bastia,D. (2005) Sap1p Binds to Ter1 at the Ribosomal DNA of \nSchizosaccharomyces pombe and Causes Polar Replication Fork Arrest. Journal of \nBiological Chemistry, 280, 39135–39142. \n28. Mejía-Ramírez,E., Sánchez-Gorostiaga,A., Krimer,D.B., Schvartzman,J.B. and \nHernández,P. (2005) The mating type switch-activating protein Sap1 Is required for \nreplication fork arrest at the rRNA genes of fission yeast. Mol Cell Biol, 25, 8755–\n8761. \n29. Tong,P., Pidoux,A.L., Toda,N.R.T., Ard,R., Berger,H., Shukla,M., Torres-Garcia,J., \nMüller,C.A., Nieduszynski,C.A. and Allshire,R.C. (2019) Interspecies conservation of \norganisation and function between nonhomologous regional centromeres. Nat \nCommun, 10, 2343. \n30. Camacho,C., Coulouris,G., Avagyan,V., Ma,N., Papadopoulos,J., Bealer,K. and \nMadden,T.L. (2009) BLAST+: architecture and applications. BMC Bioinformatics, \n10, 421. \n31. Codlin,S. and Dalgaard,J.Z. (2003) Complex mechanism of site-specific DNA replication \ntermination in fission yeast. EMBO J, 22, 3431–3440. \n32. Vengrova,S. and Dalgaard,J.Z. (2004) RNase-sensitive DNA modification(s) initiates S. \npombe mating-type switching. Genes Dev, 18, 794–804. \n33. Styrkársdóttir,U., Egel,R. and Nielsen,O. (1993) The smt-0 mutation which abolishes \nmating-type switching in fission yeast is a deletion. Curr Genet, 23, 184–186. \n34. Lowe,T.M. and Chan,P.P. (2016) tRNAscan-SE On-line: integrating search and context \nfor analysis of transfer RNA genes. Nucleic Acids Research, 44, W54–W57. \n35. Theulot,B., Tourancheau,A., Simonin Chavignier,E., Jean,E., Arbona,J.-M., Audit,B., \nHyrien,O., Lacroix,L. and Le Tallec,B. (2025) Telomere-to-telomere DNA replication \ntiming profiling using single-molecule sequencing with Nanotiming. Nat Commun, \n16, 242. \n36. Sanchez,J.A., Kim,S.M. and Huberman,J.A. (1998) Ribosomal DNA replication in the \nfission yeast, Schizosaccharomyces pombe. Exp Cell Res, 238, 220–230. \n37. Sánchez-Gorostiaga,A., López-Estraño,C., Krimer,D.B., Schvartzman,J.B. and \nHernández,P. (2004) Transcription Termination Factor reb1p Causes Two Replication \nFork Barriers at Its Cognate Sites in Fission Yeast Ribosomal DNA In Vivo. Mol Cell \nBiol, 24, 398–406. \n38. Dalgaard,J.Z. and Klar,A.J.S. (2000) swi1 and swi3 Perform Imprinting, Pausing, and \nTermination of DNA Replication in S. pombe. Cell, 102, 745–751. \n39. Dalgaard,J.Z. and Klar,A.J.S. (2001) A DNA replication-arrest site RTS1 regulates \nimprinting by determining the direction of replication at mat1 in S. pombe. Genes \nDev., 15, 2060–2068. \n40. Dalgaard,J.Z. and Klar,A.J.S. (1999) Orientation of DNA replication establishes mating-\ntype switching pattern in S. pombe. Nature, 400, 181–184. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n24 \n \n41. Klar,A.J.S., Ishikawa,K. and Moore,S. (2014) A Unique DNA Recombination \nMechanism of the Mating/Cell-type Switching of Fission Yeasts: a Review. \nMicrobiology Spectrum, 2, 10.1128/microbiolspec.mdna3-0003–2014. \n42. Wood,V., Gwilliam,R., Rajandream,M.-A., Lyne,M., Lyne,R., Stewart,A., Sgouros,J., \nPeat,N., Hayles,J., Baker,S., et al. (2002) The genome sequence of \nSchizosaccharomyces pombe. Nature, 415, 871–880. \n43. Smith,J.G., Caddle,M.S., Bulboaca,G.H., Wohlgemuth,J.G., Baum,M., Clarke,L. and \nCalos,M.P. (1995) Replication of Centromere II of Schizosaccharomyces pombe. \nMolecular and Cellular Biology, 15, 5165–5172. \n44. Kim,S. and Huberman,J.A. (2001) Regulation of replication timing in fission yeast. The \nEMBO Journal, 20, 6115–6126. \n45. Zaratiegui,M., Castel,S.E., Irvine,D.V., Kloc,A., Ren,J., Li,F., de Castro,E., Marín,L., \nChang,A.-Y., Goto,D., et al. (2011) RNAi promotes heterochromatic silencing \nthrough replication-coupled release of RNA Pol II. Nature, 479, 135–138. \n46. Håkansson,P., Dahl,L., Chilkova,O., Domkin,V. and Thelander,L. (2006) The \nSchizosaccharomyces pombe replication inhibitor Spd1 regulates ribonucleotide \nreductase activity and dNTPs by binding to the large Cdc22 subunit. J Biol Chem, \n281, 1778–1783. \n47. Fleck,O., Vejrup-Hansen,R., Watson,A., Carr,A.M., Nielsen,O. and Holmberg,C. (2013) \nSpd1 accumulation causes genome instability independently of ribonucleotide \nreductase activity but functions to protect the genome when deoxynucleotide pools \nare elevated. Journal of Cell Science, 126, 4985–4994. \n48. Krings,G. and Bastia,D. (2004) swi1- and swi3-dependent and independent replication \nfork arrest at the ribosomal DNA of Schizosaccharomyces pombe. Proceedings of the \nNational Academy of Sciences, 101, 14085–14090. \n49. Hayashi,M.T., Takahashi,T.S., Nakagawa,T., Nakayama,J. and Masukata,H. (2009) The \nheterochromatin protein Swi6/HP1 activates replication origins at the pericentromeric \nregion and silent mating-type locus. Nat Cell Biol, 11, 357–362. \n50. Scott,K.C., Merrett,S.L. and Willard,H.F. (2006) A heterochromatin barrier partitions the \nfission yeast centromere into discrete chromatin domains. Curr Biol, 16, 119–129. \n51. Hennion,M., Theulot,B., Arbona,J.-M., Audit,B. and Hyrien,O. (2022) FORK-seq: \nSingle-Molecule Profiling of DNA Replication. Methods Mol Biol, 2477, 107–128. \n52. Hayles,J. and Nurse,P. (2018) Introduction to Fission Yeast as a Model System. Cold \nSpring Harb Protoc, 2018, pdb.top079749. \n53. Ostrowski,M.S., Yang,M.G., McNally,C.P., Abdulhay,N.J., Wang,S., Renduchintala,K., \nIrkliyenko,I., Biran,A., Chew,B.T.L., Midha,A.D., et al. (2025) The single-molecule \naccessibility landscape of newly replicated mammalian chromatin. Cell, 188, 237-\n252.e19. \n \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n25 \n \nFIGURE LEGENDS  \nFigure 1: Levels of BrdU incorporation in fission and budding yeast. A) Overview of the \nmethodology for detection of BrdU and inference of DNA replication dynamics on single \nmolecules. Cells are grown in the presence of the nucleoside analogue BrdU, which is \nincorporated into DNA in place of canonical thymidine. Then, DNA is extracted and \nnanopore sequenced. The resulting reads are analysed with a model that detects BrdU \n(DNAscent; see Methods). The density of BrdU along the sequenced DNA molecules is \nindicative of the kinetics of DNA replication if the BrdU densities decrease as replication \nproceeds. Peaks indicate replication initiation events, valleys indicate termination sites, \npositive gradients correspond to leftward replication forks, negative gradients to rightward \nforks, and sudden drops mark replication fork pause sites. B) S. pombe cell cycle \nsynchronised cultures were grown in media with BrdU and harvested post S-phase at the \nindicated times. High-molecular weight DNA was then extracted, nanopore sequenced, and \nanalysed. As an example, we show the BrdU detected on a single molecule from a culture \ntreated with 2 µM BrdU and the corresponding population level median replication time from \n(7). In the BrdU plot, the grey dots indicate the probability of BrdU at each thymidine and the \nblack line indicates the fraction of BrdU in 300 thymidine windows. In the replication time \nplot, time is shown increasing down the y-axis. In both plots the x-axis indicates the genomic \nposition on chromosome III. C) Similarly to S. pombe cells, S. cerevisiae cells were grown in \nmedia with BrdU and harvested post S-phase at the indicated times. High-molecular weight \nDNA was then extracted, nanopore sequenced, and analysed. As an example, we show the \nBrdU detected in a single molecule from a culture treated with 30 µM BrdU and the \ncorresponding population level median replication time from (21) plotted as in panel B. D) \nCell cycle synchronised S. pombe were grown in the presence of 4 µM BrdU with a 400 µM \nthymidine 30 minutes after release. As an example, we show the BrdU detected on a single \nmolecule and the corresponding population level median replication time plotted as in panel \nB. E) Fraction of BrdU incorporated over time in S. cerevisiae and S. pombe cells treated \nwith a range of BrdU and thymidine concentrations. For each single molecule, the fraction of \nBrdU incorporated in every 1 kb window was determined and plotted in 5 min bins of \npopulation level median replication time from (21) and (7), for S. cerevisiae and S. pombe, \nrespectively. The dotted black line indicates that only molecules with a BrdU fraction ≥0.05 \nare plotted. The black diamonds indicate the median BrdU fraction for each replication time \ninterval. \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n26 \n \n \nFigure 2: BrdU incorporation levels reveal replication fork direction, location and \nefficiency of initiation sites and location of termination sites. Single molecules were \naligned to the reference genome ASM294v2. A) Example of a nascent single molecule read \nwith inferred replication dynamics. BrdU probability and fraction are plotted against genomic \ncoordinates (as in Fig. 1B). The lines above the BrdU probabilities indicate the position of \nright-moving forks (red arrow), left-moving forks (blue arrow), initiation sites (black \nrectangle) and termination sites (grey rectangle). The heatmap indicates the BrdU fraction. B) \nHeatmap of BrdU fractions in single molecules with ≥1 fork. Two example areas (1 and 2) \nare shown zoomed in and highlighted in black. C) Replication fork directions, initiations and \nterminations on all single molecules from panel B. Features are represented as in the example \nsingle molecule from panel A. As in panel B, two example areas (3 and 4) are zoomed in and \nhighlighted in black rectangles. D) Fraction of leftward-moving forks identified on single \nmolecules and Pu-seq from (1). E) Comparison of replication dynamics from single molecule \nsequencing and Pu-seq at the example locus. The panels show, from top to bottom: fraction \nof leftward-moving forks detected on single molecules, fraction of leftward-moving forks \ndetected with Pu-seq, count of single molecules initiation sites, efficiency of Pu-seq origins, \ncount of single molecules termination sites, and frequency of termination events detected \nwith Pu-seq. Pu-seq data from (1). F) Single-molecule initiation counts versus Pu-seq origin \nefficiencies. The count of single-molecule initiation site midpoints within +/- 1kb from a Pu-\nseq origin were grouped by Pu-seq efficiency. Only DNAscent initiation counts ≤60 are \nshown. G) Single-molecule termination counts versus Pu-seq termination frequencies. Only \nDNAscent termination counts ≤30 are shown. \n \nFigure 3. DNA replication dynamics at the S. pombe rDNA. Single molecules were \naligned to a custom rDNA array containing 22 rDNA repeats. A) Schematic of the S. pombe \nrDNA. Each repeat contains rRNA genes, a replication origin (ARS3001), and a region with \nfork barriers. B) Example of two single molecules that align to the rDNA. A sharp drop in \nBrdU incorporation indicates the location of a pause in replication. BrdU probability and \nfraction are plotted against the custom rDNA array coordinates. The rDNA array is \nrepresented on top of the molecules. C) Normalised pause count per rDNA unit. Pauses were \ncollapsed in one rDNA unit and separated depending on whether they were found in the \nlagging/leading strand or left/right fork. D) Schematic of replication at the rDNA. After an \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n27 \n \norigin fires, the left forks progresses while the right fork is paused at the barriers on both the \nleading and lagging strand. \n \nFigure 4. DNA replication dynamics at the S. pombe mating-type locus. Single molecules \nwere aligned to a de novo whole-genome assembly from strain AW2224. A) Schematic \noverview of the mating-type locus in S. pombe. The mating-type locus is comprised of three \ngenes: mat1, mat2 and mat3 in chromosome II. At the mat1 locus, the replication terminator \nRTS1 prevents rightward-moving forks across the mat1 genes, while the replication fork \nbarrier MPS1 occasionally pauses leftward-moving forks. B) Examples of single molecules \nshowing unperturbed left fork progression or stalled left fork at the MPS1 site. BrdU \nprobability and fraction are plotted on the coordinates for de novo genome assembly and \nvisualised as in Fig. 1B. C) Ensemble of single-molecule replication dynamics at the mating-\ntype locus. The panels show from top to bottom: fraction of leftward-moving forks, count of \ninitiation sites, count of termination sites, count of pause sites in the leading strand, count of \npause sites in the lagging strand, count of pause sites in left forks, and count of pause sites in \nright forks. Single-molecule initiation and termination were counted in 2 kb genomic \nwindows and pause sites were counted in 500 bp windows. \n \nFigure 5. DNA replication dynamics at the S. pombe centromeres. Single molecules were \naligned to a de novo whole-genome assembly from strain AW2224. A) S. pombe centromeres \nare comprised of outer repeats (dg and dh), flanking a central region composed of inner \nrepeats (imr) and a central core (cnt). tRNAs are found within and adjacent to each \ncentromere. B) Examples of a single molecule that aligns to centromere 1 plotted on the \ncoordinates for de novo genome assembly and visualised as in Fig. 1B. C) Ensemble of \nsingle-molecule replication dynamics across centromere 1. The panels show from top to \nbottom: centromere annotation track with the tRNAs on a separate y-axis (for clarity only two \nflanking genes are shown either side of the centromere), fraction of leftward forks, count of \ninitiation sites, count of termination sites and the count of pause sites. Single-molecule \ninitiation and termination sites were counted in 2 kb genomic windows whereas the pause \nsites were counted in 500 bp windows. The central cores are highlighted with dashed lines. \nD) Ensemble of single-molecule replication dynamics across centromere 2 plotted as in panel \nC. E) Ensemble of single-molecule replication dynamics across centromere 3 plotted as in \npanel C. \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n28 \n \nFigure 1 \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n29 \n \nFigure 2 \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n30 \n \nFigure 3 \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n31 \n \nFigure 4 \n \n  \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint \n\n32 \n \nFigure 5 \n \n \n.CC-BY 4.0 International licensemade available under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is \nThe copyright holder for this preprintthis version posted July 18, 2025. ; https://doi.org/10.1101/2025.07.16.665067doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}