Parental histone transfer caught at the replication fork – Nature

Yeast strain

All of the yeast strains that were used in this study were derivatives of the W303 (leu2-3, 112 ura3-1 his3-11, trp1-1, ade2-1 can1-100) genetic background (Supplementary Table 2). Strain Y277 for replisome purification was constructed using a one-step PCR-based approach with pTF272 (pFA6a-TEV-6xGly-3×Flag-HphMX, Addgene) as DNA template to generate the PSF2-3×Flag tagging modification in the W303-1a background strain. Strains for eSPAN analyses were generated using the CRISPR–Cas9 plasmid pML104 along with the primers described in Supplementary Table 3.

Replisome purification

To isolate the endogenous replisomes for structural determination, 100 liters of log-phase (optical density at 600 nm (OD₆₀₀) of 2.0) yeast cells (Y277, PSF2-3×Flag) were first synchronized at G1 phase with alpha factor (12.5 ng ml⁻¹), followed by washing once with fresh YPD medium before being released into fresh medium containing hydroxyurea (200 mM) for 1.5 h. Cells were collected and washed with buffer I (50 mM HEPES-KOH pH 7.5, 150 mM l-glutamic acid potassium salt, 8 mM MgCl₂, 1 mM EDTA, 0.02% NP-40, 3 mM ATP, 2 mM NaF, 1 mM phenylmethanesulfonylfluoride (PMSF) and 1× protease inhibitor cocktail (Roche). The cell pellets were resuspended in 0.5 volumes of buffer I. The cell suspension was frozen drop-wise in liquid nitrogen and then disrupted with a freezer mill (SPEX CertiPrep 6850 Freezer/Mill). The cell powder was thawed on ice by adding an equal volume of buffer I to obtain the insoluble crude chromatin, followed by washing once with buffer I. To solubilize chromatin-bound replisomes, the crude chromatin was digested in buffer I containing benzonase (0.5 U μl⁻¹; 7sea biotech, RPE002) for 10 min at 37 °C, and then 1 h on ice. The suspension was then centrifuged for 20 min at 38,900g. The clear phase was recovered and subjected to anti-Flag immunoprecipitation (IP) with anti-Flag M2 Affinity Gel (Sigma-Aldrich, A2220) at 4 °C for 3 h. The beads were recovered and washed extensively with buffer II (50 mM HEPES-KOH pH 7.5, 150 mM KOAc, 8 mM MgCl₂, 1 mM EDTA, 2 mM NaF, 1 mM PMSF). The precipitated replisomes were eluted with 0.5 mg ml⁻¹ of 3×Flag peptide (GenScript, U6320GJ210-1) in buffer II at 4 °C. The eluates were combined and concentrated. The replisome samples were then applied on top of a 20–40% glycerol gradient containing glutaraldehyde (0–0.16%) for cross-linking in buffer II. The glycerol gradients were centrifuged in a TLS-55 rotor (Beckman Optima MAX-XP Ultracentrifuge) at 77,100g for 9 h. The fractions were collected and the cross-linking reaction was quenched by addition of Tris-HCl (pH 7.5) buffer to a final concentration of 40 mM. The fractions containing the replisomes were pooled and processed for EM analyses. Ultrafiltration for the removal of glycerol, buffer exchange and sample concentration was performed with buffer II using a centrifugal filter (Amicon Ultra, 0.5 ml, 50 kDa) at 6,000g at 4 °C.

EM analysis

For negative staining, the samples were stained with 2% (w/v) uranyl acetate and examined using the Talos L120C microscope (Thermo Fisher Scientific) operated at 120 kV to determine the sample quality and estimate the relative concentration of samples used for cryo-grid preparation.

For cryo-grid preparation, 4 μl aliquots of samples were applied to a glow-discharged holey carbon grid (C-flat R 1.2/1.3 Au) and plunge frozen into liquid ethane cooled by liquid nitrogen using the Vitrobot VI (Thermo Fisher Scientific) after 3 s blotting with filter paper at 4 °C and under 100% humidity.

The grids were loaded onto an FEI Titan Krios G3i transmission electron microscope operated at 300 kV. Images were recorded on a Gatan K3 summit direct electron detector and a Bio Quantum energy filter with a 20 eV slit width. Images with a total dose of 50 e⁻ Å⁻² were acquired within 4 s at a nominal magnification 81,000× (EFTEM mode), corresponding to a calibrated pixel size of 1.06 Å. The dose was fractionalized to 40 frames equally. The defocus range was set between −1.0 and −2.5 μm. EPU (v.2.12) was used for data collection.

Data processing

In total, 9,537 movie stacks were collected for the samples of the endogenous replisome and were preprocessed in RELION (v.4.0)⁴¹. The super-resolution movie stacks underwent local drift correction, electron-dose weighting and twofold binning using MotionCor2 (v.1.4)⁴². This process generated both dose-weighted and unweighted summed micrographs. The unweighted version of micrographs was used for contrast transfer function (CTF) estimation, particle picking and coarse 2D/3D classification. On the other hand, the dose-weighted micrographs were used for fine 3D classification and map refinement. CTF estimation was performed using CTFFIND4⁴³. Approximately 600 micrographs with high contrast were selected and processed for multiple rounds of manual/auto particle picking as well as initial 2D/3D classification to prepare accurate templates for particle auto-picking across the whole dataset. The auto-picking process was meticulously optimized using both the RELION template-matching method and the Topaz (v.0.2.5) deep-learning method^44,45.

A total of 1,542,000 particles was auto-picked from all of the micrographs and underwent initial 3D classification to exclude noise and bad particles (Extended Data Fig. 2d). From this analysis, 524,000 qualified particles were retained and processed for global refinement, which resulted in a 3.7 Å global density map. On the basis of this global map, the local densities of Spt16-MD–histone, Spt16-DD–Pob3, Ctf4 trimer, MCM CTDs and Polε were further enhanced through a cascade of mask-based local 3D classification and refinement steps (Extended Data Fig. 2e–i). All of the local classification procedures were performed in RELION with the ‘–skip_alignment’ option enabled. For the region of Spt16-MD–histone, three subgroups (108,000 particles) showing high occupancy and improved structural features after local classification were combined and processed for multiple sequential refinement steps using either RELION or cryoSPARC (v.4.0) with gradually narrowed masks. This process resulted in a final local density map for the region of Spt16-MD–histone–Tof1 at a resolution of 3.5 Å (Extended Data Fig. 2e,i). Local 3D classification focused on the region of Spt16-DD–Pob3 was further applied on the optimized dataset of Spt16-MD–histone (108,000 particles), yielding two subgroups (15,000 and 16,000 particles) that showed improved structural features for the relevant density (Extended Data Fig. 2f). It is evident the state II Spt16-DD–Pob3 exhibits a minor shift away from the Tof1–Csm3 platform compared with state I. Notably, the structures of the state I Spt16-DD–Pob3 are shown in Fig. 1 and Extended Data Fig. 3. For the region of Ctf4, the particles showing low resolution or low Ctf4 occupancy were removed through local classification, leaving 232,000 particles for refining the local density map of Ctf4 using cryoSPARC, resulting in a map with a resolution at 3.5 Å (Extended Data Fig. 2g). Local 3D classification focused on the CTDs of the MCM ring led to the identification of two conformations, conformation-1 (72,000 particles) and conformation-2 (246,000 particles). After global refinement using cryoSPARC, the final resolutions of the maps of these two structures reached 3.8 Å (Extended Data Fig. 2h). As a stable Polε is associated with conformation-1, a local classification focused on the region of Polε was performed to improve the local resolution of Polε (Extended Data Fig. 2i). A composite global density map was generated using phenix.combine_focused_maps⁴⁶. It consists of the densities of CMG, the CMG-bound DNA region from the global structure of Conformation-1, Tof1–Csm3–Mrc1–Spt16-MD–Mcm2-NTE–histone, their bound DNA region from the 3.5 Å Spt16-MD–histone–Tof1 locally optimized map, and Spt16-DD–Pob3 from the locally optimized map of Spt16-DD–Pob3 (state I), Ctf4 from the Ctf4 trimer local map, and Polε–Mcm5-WHD from the Polε locally optimized map. The local resolutions of the composite map (Extended Data Fig. 2k) were calculated using the RELION’s own local resolution estimation tool, based on the two composite half maps generated by phenix.combine_focused_maps. Chimera⁴⁷, ChimeraX⁴⁸ and PyMOL (v.2.5 Schrödinger) were used for figure preparation.

Model building

The cryo-EM structures of yeast CMG–Ctf4–Tof1–Csm3–fork-DNA complex (PDB: 6SKL), yeast RNA polymerase II–Spt4/5–nucleosome–FACT (PDB: 7NKY), yeast intact nucleosome (PDB: 1ID3) and human MCM2-HBD–(H3–H4)₂ complex (PDB: 5BNV) were used as initial models for model building. Predicted initial models from the AlphaFold Protein Structure Database⁴⁹ were used for Polε and WH domain of Mcm5. The domains or segments of the initial models were rigid-body fitted into the cryo-EM density maps of the replisome using ChimeraX (v.1.3)⁴⁸ and manually adjusted against the locally optimized density maps in Coot (v.0.9.8.92)⁵⁰. For the region of Spt16-DD–Pob3, the atomic models of Spt16-DD–Pob3-DD and Pob3-MD from the reported structure (PDB: 7NKY) were rigid-body fitted into the locally optimized map of Spt16-DD–Pob3 (state I), separately, without manual adjustment. Given the low resolution, the side chains of the Spt16-DD–Pob3 regions were removed. The model of the N-terminal loop of Spt16 ND linker was de novo built using the main chain backbone without assigning any side chain due to the limit of the local resolution. The model of the Spt16 CTD was manually built by referencing the positioning of Spt16 CTD in previously reported FACT–histone structures (PDB: 6UPK, 7NKY, 7XTI). The model of most of its CTD sequence was also built using only the main chain backbone, except for residues 969–974, where the side chains can be properly assigned. The assignment of the relevant side chains was further confirmed by the structure prediction of the Spt16–H3–H4 complex using Alphafold2.

The merged atomic model was further refined against the composite global density map using phenix.real_space_refine⁴⁶ to optimize the overall geometry quality. The quality of the deposited model was evaluated using phenix.molprobity⁵¹.

MS sample preparation

For MS analysis, the eluted replisome sample was fractionated using 20–40% glycerol gradient without glutaraldehyde cross-linking. The fractions containing replisomes were then resolved on a 4–12% Bis-Tris gel, followed by fixation in a 50% methanol/7% acetic acid solution. The gel was stained by GelCode Blue stain (Pierce), diced into 1 mm³ cubes and destained by incubating with 50 mM ammonium bicarbonate/50% acetonitrile for 1 h. The destained gel cubes were dehydrated in acetonitrile for 10 min and rehydrated in 25 mM NH₄HCO₃ with trypsin for protein digestion at 37 °C overnight. The resultant peptides were enriched with StageTips. The eluted peptides were dried down using the SpeedVac and resuspended in 0.1% formic acid for analysis using liquid chromatography coupled with tandem MS (LC–MS/MS).

MS data analysis

The LC–MS/MS analysis of the replisome sample was performed on the Q Exactive Plus Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher Scientific). The peptide samples in 0.1% formic acid were pressure loaded, rinsed for 5 min with 0.1% formic acid and subsequently eluted with a linear gradient from 4% B (96% A) to 35% B (65% A) in 90 min (A = 0.1% formic acid; B = 0.1% formic acid in 80% acetonitrile; flow rate, ~300 nl min⁻¹) into the mass spectrometer. The instrument was operated in a data-dependent mode cycling through a full scan (300–1,800 m/z, single µscan) followed by 10 HCD MS/MS scans on the 10 most abundant ions from the preceding full scan. The cations were isolated with a 2 Da mass window and set on a dynamic exclusion list for 60 s after they were first selected for MS/MS. The raw data were processed and analysed using MaxQuant (v.1.6.1.0). A fasta file containing yeast proteomes was downloaded from UniProt and used as protein sequence searching the database. Default parameters were adapted for the protein identification and quantification. Parent peak MS tolerance was 4.5 ppm, MS/MS tolerance was 20 ppm, the minimum peptide length was 7 amino acids, the maximum number of missed cleavages was 2. The proteins quantified were supported by at least two quantification events.

eSPAN method and data analysis

The eSPAN method was modified from previous studies with some modifications^32,33. In brief, yeast G1 cells were released into a fresh medium containing BrdU, a thymidine analogue that can be incorporated into nascent DNA during DNA synthesis. The chromatin from early S phase cells was digested with micrococcal nuclease (MNase) and analysed using IP with antibodies against BrdU, followed by sequencing to identify DNA strands that contained BrdU (BrdU-IP-ssSeq). To identify nascent-DNA-associated nucleosomes, the MNase-digested chromatin was used for chromatin IP (ChIP) with antibodies against two different histone modifications: H3K56ac (a marker of newly formed histones) or H3K4me3 (a marker of parental histones), respectively. Subsequently, the chromatin immunoprecipitated DNA was denatured into single-stranded DNA and the newly synthesized DNA was enriched by BrdU IP, followed by strand-specific deep sequencing. The sequencing reads obtained from eSPAN were divided to distinguish between the Watson and Crick strands. The average ratio of Watson/Crick strands around early replication origins was then calculated. To minimize differences in BrdU incorporation, this ratio was normalized to the MNase-BrdU-IP-seq dataset. This calculation, known as eSPAN bias, provides information about the relative levels of a modified histone on the leading and lagging strands of DNA replication forks (Fig. 5a). To assess the efficiency of parental histone recycling on nascent strands, another ratio was calculated: the total eSPAN sequence reads surrounding the ACS (origin of replication) divided by the total MNase-BrdU-IP-seq reads in the same region. This measurement is referred to as eSPAN density. By separating eSPAN and BrdU-IP-seq sequence reads into Watson and Crick strands, the eSPAN density on nascent leading and lagging strand chromatin can be determined (Fig. 5a).

S. cerevisiae yeast cells were cultured in YPD medium at 30 °C to mid-log phase (OD₆₀₀ = 0.45–0.5), and arrested at G1 phase by α-factor (5 mg ml⁻¹, 1000×, Chinese Peptide Company) at 25 °C. Cells were collected, washed with ice-cold double-distilled H₂O three times at 2,500 rpm for 5 min at 4 °C and then released into fresh YPD medium with 0.4 mg ml⁻¹ BrdU (Sigma-Aldrich, B5002-5G) at 23 °C for 40 min to label nascent DNA. Cells were cross-linked with 1% (w/v) paraformaldehyde (Sigma-Aldrich, P6148-1KG) at 25 °C for 20 min and then quenched with 125 mM glycine (Amresco, 0167-5KG) at 25 °C for 5 min.

Cells (OD₆₀₀ of around 100) were pelleted, and then washed twice with ice-cold 1× TBS buffer (0.1 mM PMSF freshly added) and then washed once with ice-cold buffer Z (1.2 M sorbitol, 50 mM Tris-HCl pH 7.4). The pellets were resuspended with 8.7 ml buffer Z (10 mM β-mercaptoethanol freshly added), and digested with 214 μl 5 mg ml⁻¹ zymolase (Nacalai Tesque, 07665-84) at 28 °C for about 35 min to obtain spheroplasts. The efficiency of digestion was evaluated by measuring the OD₆₀₀ in 1% SDS, with a decrease of over 90%. Pellets were collected, resuspended with 1.5 ml ice-cold NP buffer (1 M sorbitol, 50 mM NaCl, 10 mM Tris-HCl pH 7.4, 5 mM MgCl₂, 1 mM CaCl₂, with 0.5 mM spermidine, 0.007% (v/v) β-mercaptoethanol and 0.075% (v/v) NP-40 (Thermo Fisher Scientific, 28324) added freshly) and divided into four parts equally in LoBind tubes. For each part, a suitable amount of MNase (Worthington, LS004797) was added and incubated at 37 °C for about 20 min to digest the chromatin into mainly mono- and di-nucleosomes. Then, 8 μl 0.5 M EDTA (pH 8.0) was added to the reaction tube to stop the reaction. Next, 100 μl 5×ChIP lysis buffer (250 mM HEPES-KOH pH 7.5, 700 mM NaCl, 5 mM EDTA pH 8.0, 5% (v/v) Triton X-100, 0.5% (w/v) sodium deoxycholate, with 5 mM PMSF, 1.25 mg ml⁻¹ pefabloc, 5 mg ml⁻¹ bacitracin and 5 mM benzamidine added freshly) was added, and the sample was then incubated for 30 min on ice. The lysate was then clarified by centrifugation twice at 10,800 rpm at 4 °C, with the first round lasting for 5 min and the second round lasting for 15 min. The supernatant was collected and ready for ChIP experiments.

For ChIP assays, 50 μl supernatant was taken as the input, and 800 μl supernatant was taken for ChIP using H3K4me3 (Abcam, ab8580) or H3K56ac⁵² (prepared in-house) antibodies, with incubation at 4 °C for 12–16 h. The protein–DNA bound to antibodies was enriched using 20 μl prewashed protein G Sepharose agarose beads (GE Healthcare, 17061801) with 2 h of incubation at 4 °C. The binding system was then stepwise washed with the reagents below, and centrifuged at 2,500 rpm for 1 min at 4 °C: 1× ChIP lysis buffer (with 0.1 mM PMSF), once; 1× ChIP lysis buffer, incubated at 4 °C for 5 min, twice; 1× ChIP lysis buffer (with 0.5 M NaCl), once; 1× ChIP lysis buffer (with 0.5 M NaCl), incubated at 4 °C for 5 min, once; Tris/LiCl buffer, once; Tris/LiCl buffer, incubated at 4 °C for 5 min, once; Tris/EDTA buffer, twice. After washing, the liquid was removed using fine syringe needles. For both input and ChIP samples, 50 μl 20% (w/v) Chelex-100 (Bio-Rad, 1422822) was added, followed by boiling for 10 min at 100 °C for reverse-cross-linking. After cooling down, 5 μl of 20 mg ml⁻¹ proteinase K (Invitrogen, 25530015) was added and the sample was incubated for 30 min at 55 °C. The sample was then boiled at 100 °C for another 10 min. The supernatant was saved after centrifuging at 14,000 rpm for 1 min, 75 μl for the input sample and 25 μl for the ChIP sample. Next, 50 μl 2× TE was added, followed by centrifuging at 14,000 rpm for 1 min. The 50 μl supernatant was saved and mixed with the previous one. For the ChIP sample, 35 μl 1×TE was also added, with 35 μl supernatant saved after centrifugation. For both input and ChIP samples, 90 μl of the total supernatant was taken for BrdU IP to get MNase-BrdU-IP and eSPAN samples, respectively.

For BrdU IP, the 90 μl sample was denatured by snap-cooling in an ice–water mixture for 5 min after 5 min of boiling at 100 °C. Then, 10 μl 10× PBS, 800 μl BrdU IP buffer (1× PBS, 0.0625% Triton X-100) with 0.3 μl Escherichia coli tRNA (Roche, 10109541001) and 0.36 μl anti-BrdU antibody (BD Biosciences, BD555627) were added into each sample, followed by 2 h of incubation at 4 °C. The BrdU-labelled nascent DNA bound to antibodies was then enriched using 15 μl prewashed protein G Sepharose agarose beads with another 2 h of incubation at 4 °C. The binding system was then washed with ice-cold BrdU IP buffer three times and with 1× TE once, with 4–5 min of incubation at 4 °C or room temperature, respectively. After washing, the remaining liquid was removed using fine syringe needles. Then, 100 μl elution buffer (1×TE, 1% (w/v) SDS) was added, and the sample was incubated at 65 °C for 15 min at 1,300 rpm on the Eppendorf Thermomixer C. Next, 85 μl supernatant was collected after centrifuging at 14,000 rpm for 1 min. Another 40 μl elution buffer was added, with incubation at 65 °C for 5 min at 1,300 rpm. Next, 35 μl supernatant was collected after centrifuging at 14,000 rpm for 1 min and mixed with the previous one. Together, six samples were generated: input, H3K4me3-ChIP, H3K56ac-ChIP, MNase-BrdU-IP, H3K4me3-eSPAN and H3K56ac-eSPAN. All of the samples were purified using the MinElute PCR Purification Kit (Qiagen) to prepare DNA for library construction. The quality of DNA was evaluated using quantitative PCR.

The single-stranded DNA libraries were constructed using the Accel-NGS 1S Plus DNA Library Kit for Illumina (Swift) and sequenced by the Novogene Genome Sequencing Company on the Illumina NovaSeq 6000 system. After quality control, the adapter and sequencing reads with low quality were removed using Trimmomatic⁵³. Clean reads were mapped to the S. cerevisiae reference genome (sacCer3) using Bowtie2 (v.1.2.0)⁵⁴. Only paired-end reads correctly mapped on both ends were selected for further analysis. On the basis of the flag in the SAM files, each read was assigned to the Watson or Crick strand using a custom Perl program. BrdU-enriched regions were called using MACS2⁵⁵. DANPOS (v.2.2.2) was used to call nucleosome positions and occupancy⁵⁶. The eSPAN bias was defined as the average log₂ ratio of the Watson strand reads over the Crick strand reads around 139 early DNA replication origins. To reduce the impact of the difference in the incorporation of BrdU among different strains, the eSPAN data were normalized to MNase-BrdU-IP-Seq data. The normalized eSPAN bias could represent the relative amount of histone modifications on the leading or lagging strand at the replication forks. The eSPAN density was calculated as the eSPAN signals at Watson or Crick strands around 139 early replication origins, after normalization to MNase-BrdU-IP-Seq data, which could measure the efficiency of parental histone recycling on leading or lagging strands. Statistical significance was tested using rank-sum Wilcoxon tests.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.