Fork Reversal Safeguards Epigenetic Inheritance During Replication Stress
Qiong Wu, Caixian Zhou, Yue Dou, Nadia I. Martin, Maria Gwit, Tae-Hee Lee, Mark Hedglin, Roséa Chen, Ke Zhu, Tianpeng Zhang, Shangming Tang, Tian Zhang, Tyler M. Weaver, Wenpeng Liu
TL;DR
This study shows that replication fork reversal helps maintain epigenetic information during DNA replication stress, preventing loss of histones and ensuring genome stability.
Contribution
The study identifies replication fork reversal as a novel mechanism for safeguarding epigenetic inheritance during replication stress.
Findings
Cells lacking fork reversal show reduced nucleosome density and loss of parental histones at replication forks.
PrimPol activation causes single-stranded DNA gaps, leading to PARylation and nucleosome loss in fork reversal-deficient cells.
Replication fork reversal is essential for both genome and epigenetic stability under replication stress.
Abstract
During DNA replication, epigenetic information carried by histone modifications is faithfully propagated and re-established on sister chromatids, ensuring cell identity. Chromatin reassembly is tightly coupled to DNA replication, however, whether and how perturbations to DNA replication affects the fidelity of epigenetic inheritance remain unclear. In this study, we reveal a critical role for replication fork reversal in maintaining the transmission of epigenetic information under replication stress. Cells defective in fork reversal exhibit reduced nucleosome density at replication forks, accompanied by the loss of parental histones during their transfer onto nascent DNA. Mechanistically, we demonstrate that PrimPol activation leads to single-stranded DNA gaps in fork reversal deficient cells, and that subsequent PARylation (poly ADP-ribosylation) and DNA–protein crosslinking on these…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsPARP inhibition in cancer therapy · DNA Repair Mechanisms · Epigenetics and DNA Methylation
Introduction
In eukaryotic cells, DNA replication is not merely the duplication of genetic material; it is also a critical process for the faithful transmission of epigenetic information^1^. During each S phase of the cell cycle, DNA must be precisely duplicated, and chromatin structure must be accurately reassembled to ensure cell identity, proper gene expression, and genome stability^2,3^. This chromatin-based inheritance, referred to as epigenetics, encompasses histone modifications, DNA methylation, and regulation by non-coding RNAs^4^. Programmed regulation of epigenetic states is essential for cell differentiation and organismal development^5^, whereas aberrant epigenetic changes can drive diseases and tumorigenesis^6–10^. In cancer therapy, epigenetic reprogramming exacerbates tumor cell heterogeneity and promotes drug resistance^11^. Therefore, understanding how epigenetic information is transferred during DNA replication is of fundamental importance.
During DNA replication, nucleosomes are transiently disrupted ahead of forks, parental histones are recycled by the histone chaperones FACT (FAcilitates Chromatin Transcription)^1,12–20^, ASF1 (Anti-Silencing Function 1) and CAF1 (Chromatin Assembly Factor-1), together with newly synthesized histones to assemble nucleosomes on nascent DNA^1,21–32^. Post-translational modifications (PTMs) on parental histones serve as templates that are copied onto adjacent new histones through a “read–write” mechanism, thereby completing the replication of epigenetic information^1,33–37^. Replisome proteins such as MCM2 (Minichromosome Maintenance Complex Component 2), PCNA (Proliferating Cell Nuclear Antigen), DNA polymerases ε/α, and many other chaperones contribute to parental histone transfer^25–27,35,38–48^. As a result, DNA replication exerts a profound influence on epigenetic stability. For example, depletion of Pol E3/E4, subunits of leading-strand histone chaperone polymerase ε, disrupt H3K9me3 marked parental histones recycling, thereby activating expression of LINE-1 (Long Interspersed Element-1) retrotransposon^49^. Conversely, mutations in lagging-strand histone chaperones, such as MCM2-2A, lead to parental histone accumulation on the leading strand^38,39^. Simultaneous disruption of both leading- and lagging-strand histone chaperones results in the loss of parental histones and histone PTMs, impairing stem cell differentiation and viability^50^.
DNA replication frequently encounters obstacles such as transcription–replication conflicts and repetitive sequences induced secondary structures^3,51^. Replication stress response pathways are activated to mitigate the replication obstacles^3,51^. Among these pathways, the generation of single-stranded DNA (ssDNA) and fork reversal are the two most common structural intermediates, occurring in diverse cell types in response to various forms of replication stress^52,53^. We and others previously identified that ssDNA can be processed by RAD51-mediated strand invasion together with multiple fork remodelers, including SMARCAL1, ZRANB3, and HLTF, to form reversed forks^53–64^. This process stabilizes replication intermediates and prevents excessive ssDNA accumulation to promote proper DNA replication. When fork reversal is impaired, cells activate PrimPol to reinitiate DNA synthesis downstream of the leading strand in a process known as repriming^65,66^. Although replication can resume, this generates abundant ssDNA gaps on nascent DNA that pose a threat to chromatin structure and genome stability^67–69^.
Under stressed conditions, replication forks slow and stall intermittently rather than globally ceasing^70^. It is unclear how fork reversal or ssDNA gaps influence parental histone recycling and nucleosome assembly. In this study, we demonstrate that replication fork reversal promotes parental histone recycling and nucleosome assembly on sister chromatids by suppressing ssDNA formation, thereby preserving epigenetic stability. In fork reversal defective cells, activation of the PrimPol primase leads to repriming and the generation of ssDNA gaps, which ultimately impairs recycling of parental histones and histone PTMs. We further show that impaired histone recycling results from the accumulation of ssDNA that triggers PARP1-mediated PARylation and DNA–protein crosslink formation, both of which inhibit nucleosome assembly. Together, our findings reveal a critical role for replication fork reversal in safeguarding epigenetic stability under replication stress.
Results
Replication fork reversal deficiency reduces nucleosome density on nascent DNA under replication stress
To investigate the impact of replication fork reversal on replication-associated nucleosome assembly at replication forks, we performed SILAC-iPOND-MS (Stable Isotope Labeling by Amino acids in Cell Culture - Isolation of Proteins on Nascent DNA - Mass Spectrometry) analysis in wild-type (WT) U2OS cells and U2OS cells lacking the three major fork reversal enzymes, including SMARCAL1, HLTF, ZRANB3 (termed as FRKO) to examine protein composition on nascent DNA, upon hydroxyurea (HU)-induced replication stress (Figure 1A). The SILAC-iPOND-MS analysis identified that FRKO cells have a markedly decreased abundance of nearly all histones on nascent DNA compared to WT cells (Figure 1A). Notably, the levels of the MCM2–7 helicase complex remained unchanged between WT and FRKO cells, indicating that histone reduction was not due to a decrease in replication fork number (Figure 1A). In addition, compared to WT cells, FRKO cells showed a mild increase in the levels of EdU (5-Ethynyl-2′-deoxyuridine) incorporation (Figure S1A), indicating that histone reduction in FRKO cells is not the result of reduced replication progression. To determine if the reduction of histone is nascent DNA specific, we performed immunofluorescence analysis for total chromatin bound histone H3. The results showed no difference between fork reversal–deficient and wild-type cells (Figure S1B–E), indicating that histone loss was restricted to nascent DNA.
To validate the mass spectrometry results, we pulse labeled nascent DNA with EdU, and simultaneously induced replication stress with HU for 4 hours, and quantified histone H3 on nascent DNA using a proximity ligation assay (PLA)^71^ in WT and two clones of FRKO U2OS cells. Since FRKO cells incorporate EdU more rapidly than wild-type cells (Figure S1A), direct comparison of PLA signals would not accurately reflect histone abundance on nascent DNA (Figure 1B, 1C). Therefore, we normalized PLA signals to EdU intensity, yielding a value representing histone abundance per unit length of replicated nascent DNA (Figure 1D). Using this approach, we observed a significant decrease in H3 levels on nascent DNA in the two FRKO clones. Additional PLA analyses showed consistently decreased in H2A, H2B, and H4 abundance on nascent DNA, in FRKO cells compared to WT U2OS cells, consistent with the observations for H3 (Figure 1B–D), indicating the FRKO cells have a decreased abundance of all four core histones rather than loss of individual histone types (Figure S1F–H).
To further validate these results, we performed Chromatin Occupancy after Replication (ChOR)-seq for profiling histone (or histone modification) occupancy on newly replicated DNA^72^. ChOR-seq revealed a marked reduction of histone H3 signal on nascent DNA in FRKO cells (Fig. 1E–F; Fig. S1I–J). Together, PLA and ChOR-seq independently support the iPOND-SILAC-MS results, indicating reduced histone abundance—and therefore lower nucleosome density—on nascent DNA in FRKO cells under HU treatment (Fig. 1A).
These experiments were performed in U2OS cells with co-depletion of SMARCAL1, ZRANB3, and HLTF. To determine which fork-reversal factor contributes most to this phenotype, we performed PLA in two independent clones for each single knockout (SMARCAL1, ZRANB3, or HLTF) (Fig. S2A–C). Among the single knockouts, HLTF KO showed the strongest and most consistent reduction in nascent H3 signal (Fig. S2A, S2C), suggesting that loss of HLTF contributes prominently to the FRKO phenotype. SMARCAL1 KO and ZRANB3 KO also exhibited modest decreases in nascent H3 signal (Fig. S2A–C). The triple knockout (FRKO) combines these losses and displays the most pronounced reduction (Fig. S2C). Consistent with observations in fork reversal enzyme knockout cells, knockdown of RAD51, which is also an essential factor for fork reversal{Zellweger, 2015 #61}, led to low nucleosome density on nascent DNA (Figure 1G and Figure S2D).
Notably, PLA and ChOR-seq showed no reduction in H3 levels in FRKO cells without treatment with HU (Figure S2E–G), supporting that histone loss arises as a result of replication stress. Together, these data indicate that fork reversal machinery maintains nucleosome density on nascent DNA during conditions of replication stress (Figure S2H).
Replication fork reversal deficiency leads to parental histone loss
During DNA replication, parental histones positioned ahead of the fork are disassembled and subsequently recycled onto daughter DNA through highly coordinated mechanisms, where they combine with newly synthesized histones to form nucleosomes^1^. Under conditions of persistent replication stress, the reduced nucleosome density observed on nascent DNA in FRKO cells may arise from either impaired parental histone recycling or defective incorporation of new histones, both of which threaten faithful epigenetic inheritance. Importantly, the deposition of certain histone variants, such as CENP-A, macroH2A, and H3.3, occurs independently of DNA replication^73–75^. Thus, detection of these variants on nascent DNA supports inheritance of parental histones, and their loss can be attributed to defective parental histone recycling (Figure 1A).
Prior work identified that specific histone post-translational modifications (PTMs), such as di- and trimethylation of H3K9, H3K79, and H4K20, are established on new histones many hours after replication or during the subsequent cell cycle, making them markers of parental histones during replication^31,39,49,50,76–78^. To assess whether parental histone recycling was impaired in FRKO cells under replication stress, we combined iPOND-MS with chemical derivatization of histone treatment to examine histone modifications at replication forks (Figure 2A)^79^. Upon HU-induced replication stress, slowly re-established histone PTMs (defined as parental histone specific marks on nascent DNA) such as H4K20me2, H3K79me2, and H3K9me3 were markedly reduced on nascent DNA in fork reversal–deficient cells (Figure 2B, Figure S3A–C). In contrast, histone PTMs characteristic of newly synthesized histone modifications^49,77^, such as H4K12ac, H4K5ac, H3K56ac, H3K9ac were either increased or remained unchanged (Figure 2B, Figure S3D–G). To validate H3K9me3 reduction in iPOND-MS results, we performed PLA assays to examine H3K9me3 levels on nascent DNA. We found that H3K9me3 levels on nascent DNA are reduced, consistent with iPOND-MS results (Figure 2C).
H3K9me3 is a parental histone PTMs that are associated with constitutive heterochromatin, which is lost upon replication stress in FRKO cells. To distinguish whether the reduction of H3K9me3 arises from replication stress or from preexisting global changes in FRKO cells, we performed immunofluorescence analysis. A slight decrease in cellular H3K9me3 level was observed only in S phase FRKO cells following HU treatment, rather than unperturbed condition (Figure 2D–E), indicating that the loss of this repressive mark in fork reversal–deficient cells is specifically attributable to replication stress. These results indicate the loss of parental histone during DNA replication in FRKO cells under stressed conditions.
H3K9me3 and H3K27me3 are parental histone–specific PTMs that also indicate compact chromatin. We also detected a decrease in the abundance of another compaction marker, H3K9me2 on nascent DNA in fork reversal–deficient cells (Figure 2B, 2F, Figure S3H), while the cellular H3K9me2 levels decrease also depends on HU treatment (Figure S3I–J). These findings are consistent with our findings that nucleosome density is reduced on nascent DNA in FRKO cells upon replication stress.
PrimPol-mediated repriming causes nucleosome density reduction
Histone chaperones are responsible for the deposition of newly synthesized histones and the recycling of parental histones behind the replication fork. Impaired recruitment of histone chaperones to the replication fork could explain the robust decrease in nucleosome density and parental histone recycling observed in FRKO cells during replication stress. Analysis of histone chaperone localization from our SILAC-iPOND-MS analysis revealed that almost all known histone chaperones properly localize to the replication fork in both WT and FRKO cells (Figure 3A), suggesting that the loss of parental histones and reduced nucleosome density are not due to defects of histone chaperone recruitment.
We noted that PrimPol (Primase And DNA Directed Polymerase) levels at replication forks were elevated after HU treatment in FRKO cells (Figure 1A), consistent with previous reports of PrimPol activation under fork reversal defects^65,66,80^. We hypothesized that PrimPol activation may contribute to loss of nucleosome density upon replication stress. Following HU treatment, FRKO cells exhibited elevated ssDNA gaps on nascent DNA (Figure 3B), which is dependent on PrimPol (Figure 3C). The results suggests PrimPol repriming is activated and generates ssDNA gaps in FRKO cells. To test whether PrimPol activation was responsible for reduced nucleosome density under replication stress, we depleted PrimPol in FRKO cells by siRNA or CRISPR-Cas9 and measured nucleosome density on nascent DNA by PLA assay (Figure 3D–E). PrimPol depletion by both methods rescued low nucleosome density in FRKO cells. To determine if the reduction in nucleosome density requires PrimPol catalytic activities, we reintroduced either wild-type PrimPol or a catalytically inactive point mutant (AxA, both primase and polymerase inactive; CH, primase mutants) into FR/PrimPol-knockout (FR/PPKO) cells. Only wild-type PrimPol was sufficient to restore nucleosome density, demonstrating that PrimPol repriming activity drives the loss of nucleosome density (Figure 3F–G).
Next, to determine if the parental-specific histone modification loss in FRKO cells results from PrimPol repriming, we performed iPOND-MS for histone modifications in FR/PPKO cells (Figure S4A–D). Our results demonstrate that PrimPol depletion restores H4K20me2 and H3K79me2 levels (Figure S4A–B) and reverses the increase of H4K5Ac and H3K56Ac in FRKO cells (Figure S4C–D). While we did not detect H3K9me2/3 peptides in iPOND-MS, PLA assays for H3K9me2/3 on nascent DNA confirmed restoration of H3K9me2 and H3K9me3 levels (Figure 3H–I).
Collectively, these results demonstrate that when fork reversal is absent, PrimPol-mediated repriming leads to loss of parental histones and a reduction in nucleosome density on nascent DNA (Figure 3J).
Single-stranded DNA gaps cause low nucleosome density in cells
To determine if the low nucleosome density is attributed to the elevated levels of ssDNA gaps that mediated by PrimPol, or enriched PrimPol levels per se, we examined nucleosome density in other previously defined ssDNA gaps contexts, such as BRCA1 or BRCA2 deficient cells. BRCA1 and BRCA2 are key fork protection factors, and BRCA1- or BRCA2-deficiency cells combined with HU treatment leads to nucleolytic degradation of nascent DNA and ssDNA gap formation; PrimPol repriming is also activated in the absence of BRCA2^67,81–85^. We performed PLA assay for EdU and H3 in BRCA1 deficient UWB1.289 cells, and UWB1.289 cells complemented with wild-type BRCA1. Compared to complementation cells, we observed low nucleosome density in UWB1.289 cells (Figure 4A, Figure S5A). Similarly, we also observed low nucleosome density in other BRCA1- or BRCA2-deficient cells lines (Figure 4B–D, Figure S5B–D). These findings are consistent with prior work showing that histone H3 levels are decreased on nascent DNA in BRCA2 knockout HeLa cells^85^.
Defects in Okazaki fragment maturation, such as knockdown of DNA ligase 1 (LIG1) or Flap Endonuclease 1 (FEN1), combined with HU treatment, also resulted in ssDNA gaps on the lagging strand in previous study^85^. We depleted LIG1 and FEN1 with siRNA in U2OS WT cells, and consistently observed PLA signals of H3 and EdU reduction (Figure 4E–F, Figure S5E–F). Inhibition of RAD51 triggers nucleolytic degradation of nascent DNA by nucleases such as MRE11, producing ssDNA gaps^84,86^. Therefore, we treated WT U2OS cells with RAD51 inhibitor B02 and observed low nucleosome density, which was rescued by co-treatment with MRE11 nuclease inhibitor (Figure 4G, Figure S5G).
In summary, we conclude that ssDNA gaps present in post-replicative nascent DNA reduce the nucleosome density on nascent DNA.
Single strand DNA gaps do not block nucleosome assembly in the in vitro assay
We hypothesized that single-stranded gaps in nascent DNA impair the ability to form nucleosomes. The ability of nucleosomes to form on ssDNA remains controversial. During DSB resection, the generation of long ssDNA tails/overhangs have been shown to evict nucleosomes in some studies^87–93^, whereas others have demonstrated nucleosome formation on ssDNA in vitro^94,95^. Additionally, it remains unclear whether nucleosomes can assemble on short ssDNA gaps like those found at the replication fork and what impact these short ssDNA gaps have on nucleosome stability. To determine whether nucleosomes can be assembled on DNA in the presence of ssDNA gaps, we designed a series of Widom 601 strong positioning oligonucleotides containing ssDNA gaps of various lengths (from 1–31 nts) positioned at the nucleosome dyad or the nucleosome entry/exit site (Supplemental Table 1). We then reconstituted recombinant nucleosomes containing human histones on these oligonucleotides containing ssDNA gaps using an established saltdialysis method^96–98^ (Figure 5A). Strikingly, we observed robust nucleosome formation regardless of the ssDNA gap size or ssDNA gap position within the nucleosome (Figure 5B–D), indicating that small ssDNA gaps do not inhibit nucleosome formation in vitro.
Having established that short ssDNA gaps (1–31 nt) at either the entry/exit or dyad regions do not markedly impair nucleosome assembly (Fig. 5B–D), we next asked whether longer gaps approaching the footprint of a full nucleosome would influence histone wrapping or stability. To address this, we generated a 147-nt ssDNA gap substrate flanked by a 25bp duplex regions with Cy3 label, and a 15bp duplex regions with Cy5 label (Figure 5E). We performed a high-resolution Cy3-Cy5 FRET (Förster Resonance Energy Transfer) assay to monitor nucleosome formation and conformational dynamics in real time (Figure 5E). When a nucleosome assembles onto a 147-nt ssDNA gap, the Cy3 and Cy5 fluorophores are brought into close proximity, resulting in a FRET signal (Figure 5F–G). Thus, the FRET signal serves as a sensitive readout for nucleosome assembly on ssDNA gaps. The steady-state FRET efficiency (E_FRET_ ≈ 0.27) confirmed the proximity of the labeled duplex arms (Figure 5F–G), consistent with a compact, fully wrapped configuration. These results demonstrate that even extensive single-stranded gaps are compatible with nucleosome formation, revealing that the histone octamer has an intrinsic ability to engage and stabilize flexible ssDNA regions. Together, these experiments (Figure 5), consistent with previous reports^95^, demonstrate that ssDNA gaps do not inherently preclude nucleosome formation in vitro. The discrepancy between the in vivo (Figure 3 and Figure 4) and in vitro (Figure 5) results suggests that the reduced nucleosome density observed in cells is likely caused by ssDNA-binding proteins, rather than by the presence of ssDNA itself (Figure S6A).
RPA depletion does not rescue the low nucleosome density in cells
Nucleosomes reconstituted on ssDNA (ssDNA Nucs) are similar in size and morphology to those assembled on dsDNA (dsDNA Nucs) and the histone octamer core engages the wrapped DNA with similar affinities in both complexes at physiological ionic strength^94,95,99–102^. However, ssDNA nucleosomes are more disordered and dynamic^95,100^, suggesting that regions of the ssDNA wrapped around a histone octamer core are dynamically exposed. The major ssDNA-binding protein complex, Replication Protein A (RPA), is highly abundant (10 μM range) in the nucleoplasm ^103–105^ and has the highest ssDNA affinity (KD μ pM to nM range) of any ssDNA-binding protein/protein complex that has been identified in human cells. Each RPA complex engages ~20-33 nt of ssDNA, extending the bound sequence into a linear conformation and increasing its bending 2 to 3-fold ^106–117^.
To determine whether RPA can substitute for nucleosomes on ssDNA, we performed parallel in vitro and in vivo experiments. In vitro, we compared the FRET signal of RPA binding to the ssDNA gaps only, and confirmed that RPA binding on ssDNA gaps can’t cause FRET signal (ssDNA gaps with RPA E_FRET_=0.075±0.001, ssDNA gaps only E_FRET_=0.08±0.001, Figure S6B). Next, we incubate pre-assembled ssDNA nucleosomes with physiological concentrations of RPA, and we observed a time-dependent decline in FRET efficiency (E_FRET_ decreasing from 0.25±0.001 to 0.119±0.001; Figure S6C).
Furthermore, E_FRET_ values decrease upon the addition of RPA down to an equilibrium value (E_FRET_ = 0.119 + 0.001, Figure S6C) that is noticeably higher than the equilibrium E_FRET_ values observed for the ssDNA Gap DNA substrate in the absence (0.080 + 0.001) and presence of RPA (0.075 + 0.001) (Figure S6B). It cannot be discerned whether the elevated equilibrium value (Compare the E_FRET_ in Figure S6C to E_FRET_ in Figure S6B) is due to indirect effects of the fully unwrapped histone octamer on the RPA-ssDNA filament or the presence of a minor population of stable, fully wrapped ssDNA nucleosomes or partially wrapped ssDNA nucleosomes. Nonetheless, the clear decrease in E_FRET_ observed upon the addition of RPA suggests that ssDNA readily unwraps from around histone octamer cores and the exposed sequences are engaged and stabilized by RPA.
We performed immunofluorescence staining for RPA in FRKO cells to corroborate the in vitro findings, since we observed increased levels of RPA2 and its phosphorylation, consistent with elevated ssDNA accumulation in FRKO cells upon HU treatment (Figure S5D and S6E). However, depletion of RPA1 and/or RPA2 did not rescue the low nucleosome density in fork-reversal-deficient cells (Figure S6F). These results suggest that RPA does not directly evict nucleosomes from ssDNA in cells, or that other ssDNA-binding factors may perform this function when RPA is depleted.
Transcription reduces nucleosome density in cells
Since we observed reduced chromatin compaction marks, such as H3K9me2/3 (Figure 2) and lower nucleosome density in FRKO cells (Figure 1), we hypothesized that chromatin decompaction may increase DNA accessibility to enzymes such as RNA polymerase II (Pol II). Conversely, transcription elongation and R-loop formation could further disrupt nucleosomes on nascent DNA. Thus, we proposed that RNA Pol II–mediated transcription on nascent DNA and low nucleosome density may reinforce each other.
To test this hypothesis, we first examined RNA Pol II occupancy on nascent DNA using a PLA assay targeting EdU and the major RNA Pol II subunit RPB1 (Figure S7A). We observed a significant increase in PLA signal in FRKO cells, indicating elevated Pol II association with nascent DNA. As a control, treatment with Triptolide (TPL), a potent RNA Pol II inhibitor that induces proteasome-dependent degradation of Pol II, dramatically reduced the RPB1 PLA signal, confirming the signal specificity for RNA Pol II (Figure S7A).
To determine whether increased transcription contributes to the reduced nucleosome density, we performed a PLA for H3 and EdU in TPL-treated WT and FRKO cells. TPL treatment significantly increased nucleosome density in both WT and FRKO cells, suggesting that Pol II activity on nascent DNA can reduce nucleosome density (Figure S7B). However, TPL treatment did not fully restore nucleosome density in FRKO cells to the levels in WT U2OS cells, indicating that transcription alone cannot account for the nucleosome loss in FRKO cells, and additional fork reversal dependent mechanisms likely contribute to this phenotype.
PARylation suppresses nucleosome assembly on ssDNA gaps in fork reversal deficient cells
PARylation is known to alter chromatin structure and orchestrate DNA repair^118–124^. We observed the increase in PARylation in FRKO cells upon HU treatment (Figure 6A). To determine if PARP1-mediated PARylation impacts nucleosome density, we depleted PARP1 in FRKO cells via siRNA. PARP1 depletion reduced total level of PARylation and partially rescued nucleosome density despite persistent ssDNA gaps (Figure 6B–D). Inhibition of PARG, the enzyme that hydrolyzes PAR chains, increased PARylation and decreased nucleosome density in WT cells, further supporting the conclusion that PARylation reduces nucleosome density in cells (Figure 6E and Figure 6F).
Unlike PARP1 depletion, which removes PARP1 from damaged/gapped DNA, PARP inhibitors primarily block PARylation while retaining, and in some cases trapping PARP1 on DNA, potentially preserving a physical barrier that can still interfere with nucleosome assembly. Treatment with the PARP1/2 inhibitor Olaparib reduced PARylation in WT cells after HU treatment (Figure 6G), but nucleosome density still decreased (Figure 6H). This indicates that PARP inhibition alone can lower nucleosome density on nascent DNA. In FRKO cells, Olaparib similarly reduced PARylation but nucleosome density remains low (Figure 6G–H), reinforcing that PARP1 inhibition itself causes this effect. Together, these findings indicate that ssDNA activates PARP1-mediated PARylation, which in turn reduces nucleosome density on nascent strand DNA. PARP1 trapping by Olaparib can also reduce nucleosome density on nascent DNA.
HMCES–DPC inhibits nucleosome assembly on ssDNA
We sought additional factors that might bind ssDNA and inhibit nucleosome assembly. From our iPOND-MS data, we noticed that in fork reversal–deficient cells, APOBEC3B levels at replication forks increased by ~1.5-fold (Figure 7A), chromatin-bound APOBEC3B increased by 3–5-fold (Figure S6A). APOBEC3B can deaminate cytosine residues, which are subsequently processed by glycosylases to generate AP sites^125^. HMCES, a PCNA-binding protein, has been shown to form covalent DNA–protein crosslinks (DPCs) at AP sites on ssDNA^126–132^. We observed a modest increase in levels of HMCES in iPOND-MS (Figure 7A), and a significant increase in HMCES-DNA crosslink by RADAR assay (Figure 7B).
We hypothesized that in FRKO cells, the high abundance of ssDNA on nascent DNA permits excess APOBEC3B activity, leading to the generation of AP sites that subsequently form HMCES–DPCs. These DPCs could represent a steric hindrance that blocks nucleosome assembly. To test this hypothesis, we depleted HMCES or APOBEC3B using siRNA and measured nucleosome density on nascent DNA by PLA (Figure 7C, Figure S8B–C). HMCES depletion restored nucleosome density in fork reversal–deficient cells, whereas APOBEC3B depletion only mild rescue the low nucleosome density, but not statistically significant (Figure 7C), likely because AP sites can also form spontaneously on ssDNA independent of APOBEC3B. Co-deletion of PARP1 and HMCES additively rescued nucleosome density in fork reversal–deficient cells (Figure 7D). HMCES loss is also known to activate TLS (translesion synthesis) for post-replication repair^127^, which in turn repair ssDNA gaps^67,80^. We next asked whether the rescue of nucleosome density upon HMCES depletion was mediated by TLS-dependent gap repair. Inhibition of TLS using the REV1 inhibitor JH-RE-06 in HMCES-deficient cells did not reduce nucleosome density (Figure 7E), demonstrating that HMCES–DPCs themselves, rather than TLS activation, block nucleosome assembly.
HMCES-DPC at dyad position inhibits nucleosome formation in vitro
To determine if a HMCES-DPC can block nucleosome assembly in vitro, we designed two Widom 601 strong positioning oligonucleotides containing a 31 nt ssDNA gap positioned at the nucleosome dyad or the nucleosome entry/exit site, with an uracil located in the ssDNA region (Figure S8D–E). These DNA substrates were treated with uracil DNA glycosylase to generate an AP-site, and the HMCES-AP-site DPC stabilized using sodium borohydride (Figure S8F–G)^127,128,133^. We then attempted to reconstitute recombinant human nucleosomes on these oligonucleotides containing the HMCES-DPC using a salt-dialysis method (Figure 5A)^96–98^. Notably, we observed robust nucleosome formation when the HMCES-DPC was positioned at the nucleosome entry/exit site (Figure 7F). In contrast, we observed minimal nucleosome formation when the HMCES-DPC was positioned in the DNA near the nucleosome dyad, indicating that HMCES represents a steric hindrance to nucleosome formation (Figure 7G). We hypothesize that elevated nucleosomal DNA dynamics at the entry/exit site compared to the dyad enables the nucleosome to accommodate the HMCES-DPC at this location^134^. Together, this data shows that a HMCES-DPC can directly impair nucleosome formation, though the ability to accommodate a HMCES-DPC will likely be dictated by the location of the HMCES-DPC in relation to the histone octamer during histone deposition on ssDNA gaps.
Together, these findings demonstrate that in FRKO cells, PrimPol-mediated repriming generates abundant ssDNA gaps, which activate PARP1-dependent PARylation and promote HMCES–DPC formation. At least three mechanisms: Transcription, PARylation and HMCES–DPC formation act in concert to inhibit nucleosome assembly on ssDNA gaps, thereby reducing parental histone recycling and lowering nucleosome density on nascent DNA (Figure 7H).
Discussion
Fork reversal is essential for both genome stability and epigenome stability
Fork reversal and ssDNA are global responses to various replication stresses^53^; they occur frequently during DNA replication in both tumor cells and normal cells^53^. The functions of fork reversal and ssDNA in genome stability have been established^3,55,69^. However, how they affect epigenetic inheritance (i.e., parental histone recycling) remains elusive, although chromatin replication is an essential component of DNA replication^1^. In this study, we demonstrate that fork reversal is critical for epigenome inheritance during DNA replication under stress. We propose a mechanistic model in which fork reversal suppresses ssDNA gap formation, thereby promoting parental histone recycling and nucleosome assembly on nascent DNA to preserve chromatin states. In the absence of fork reversal, PrimPol-mediated repriming generates abundant ssDNA gaps, which trigger PARP1-dependent PARylation and HMCES–DPC formation (Figure 6 and Figure 7). These events reduce nucleosome density and lead to the loss of both parental histone and the ratio of parental specific histone modifications (Figure 2 and Figure 3). Together, our work establishes a mechanistic paradigm linking replication fork intermediates (fork reversal and ssDNA) to faithful epigenetic inheritance, reframing replication stress from a purely DNA-centric problem to a driver of epigenetic instability with broad implications for human disease.
Chromatin features that are often inherited through parental nucleosomes, such as H3K9me3, macroH2A variants, and centromeric CENP-A chromatin, are linked to stable chromatin states (e.g., heterochromatin or centromere identity). H3K9me3-marked heterochromatin, in particular, is enriched at repetitive genomic regions, where it suppresses transcription^135^. Thus, the loss of parental histones during DNA replication in FRKO cells likely results in the derepression of repetitive elements. Prior studies in mouse embryonic stem cells have shown that disrupting parental-histone recycling via depletion of histone chaperones during DNA replication activates transposon elements, thereby triggering inflammatory and immune signaling and impairing differentiation and cell viability^49,50^. SMARCAL1 loss has been implicated increasing the response to immunotherapy through activating cGAS-STING pathway, and suppressing PD-L1 express^136^. Our study provided additional rationale that fork reversal could be targeted for immunotherapy since its loss may activate repeat genes, therefore producing antigens for activating cytokines.
Loss of slowly re-established histone marks on nascent DNA in FRKO cells reflects impaired parental histone recycling
A previous study reported that HU promotes ATR-dependent enrichment of H3K9me2/3 at stalled forks, and this enrichment was observed under conditions where fork reversal was not required^137^. An important question is whether the reduced H3K9me2/3 signal on nascent DNA in FRKO cells primarily reflects loss of parental histones, an altered ability to induce/enrich these marks during HU treatment, or both. Our data support the conclusion that impaired parental histone recycling contributes substantially to the phenotype. First, we observed robust ATR activation in FRKO cells under HU, making a simple failure to engage ATR signaling an unlikely explanation; however, we cannot exclude the possibility that downstream chromatin-marking outputs of ATR are rewired under sustained stress. Second, the reduction is not limited to H3K9me2/3 but extends to multiple parental-histone–associated signatures (e.g., H4K20me2 and H3K79me2) and to the inheritance-linked presence of histone variants such as CENP-A and macroH2A on nascent DNA, arguing for a broad loss of parental histone information rather than a single mark-specific defect. Consistent with this, we interpret the nascent-mark decrease as at least partially reflecting defective parental histone recycling, while acknowledging that impaired mark induction/enrichment and parental-histone loss are not mutually exclusive and may both contribute under prolonged HU exposure.
In summary, these findings indicate that fork reversal deficiency leads to a loss of parental histone information on nascent DNA, consistent with defective parental histone recycling. However, we cannot fully exclude the possibility that FRKO cells also exhibit a reduced capacity to establish or enrich specific marks under HU compared with WT cells. These mechanisms are not mutually exclusive, and both impaired parental histone recycling and altered mark induction may contribute to the phenotype observed under sustained HU treatment.
How is parental histone lost in FRKO cells?
Our results suggest that PrimPol repriming–mediated ssDNA gaps contribute to reduced nucleosome density on nascent DNA. However, our data do not fully exclude an alternative possibility: that FRKO cells experience impaired histone delivery during DNA synthesis due to altered usage of histone-chaperone–coupled replicative polymerases. Current evidence does not support a direct histone chaperone activity for PrimPol. Thus, one speculative model is that, in FRKO cells, PrimPol extends nascent DNA in contexts where replicative polymerases with tighter histone-coupled synthesis (e.g., Pol ε) would normally operate, resulting in DNA extension that is relatively uncoupled from histone deposition. Although we did not detect an obvious change in Pol ε abundance by iPOND-MS (Figure 3A), we can’t fully exclude this possibility.
Based on our findings together with prior studies, we conclude that ssDNA is not intrinsically incompatible with nucleosome assembly: from a biophysical standpoint, ssDNA can support the formation of ssNUCs. The critical unresolved issue is the temporal and spatial order of ssDNA formation relative to ssDNA-associated adducts, particularly DPCs and PARylation. Because we cannot determine when and where DPCs and PARylation arise with respect to ssDNA gaps, our data are consistent with two alternative, non–mutually exclusive models: 1) Pre-existing ssDNA regions are present in the genome prior to fork passage and are already decorated with DPCs and/or PARylation. In this scenario, as replication proceeds through these regions, parental histones would be unable to efficiently recycle onto the ssDNA substrate, resulting in an immediate deficit in nucleosome density on nascent DNA. 2) Replication itself generates abundant ssDNA gaps; ssNUCs can initially form on or near these ssDNA regions, but subsequent accumulation of DPCs and/or PARylation destabilizes these structures and promotes their eviction, thereby reducing nucleosome density.
Although our experiments show that HU treatment is associated with decreased nucleosome density, they do not definitively exclude the first model. Notably, HU exposure in our assays is sustained for ~4 hours, which could allow the establishment of pre-existing ssDNA lesions and associated adducts before or during replication, making it difficult to unambiguously assign causality and order of events.
Previous work has elegantly shown that under replication stress, ASF1 can sequester parental histones during fork stalling and facilitate their recycling upon fork restart^138^. Here, we focused on parental histone recycling during ongoing replication under stress and observed a pronounced loss of parental histones in FRKO cells. One possibility is that excessive ssDNA gaps in FRKO cells increase the frequency of parental histone disassembly, which in turn may overwhelm the buffering capacity of ASF1, ultimately resulting in defective parental histone recycling and reduced nucleosome density on nascent DNA.
ssDNA gaps reduce the nucleosome density on nascent DNA
Notably, all our cellular nascent DNA assays are EdU labeling based. Although EdU only in dsDNA, potentially limits resolution to detect histones on ssDNA. However, the spatial resolution of PLA is approximately 40 nm (~100 bp/nt), while the DNA fragments analyzed by iPOND and ChOR-seq are typically 100–400 bp/nt in length^72,139^. Previously published electron microscopy images showed that most ssDNA gap sizes on nascent DNA are less than 300bp^53,66,67,80,140^, in very rare cases can reach up to 1000 nt. The spatial distance of ssDNA-nucleosomes are even closer to the EdU in adjacent dsDNA after nucleosome assembly (Figure 5E). Therefore, we hypothesize that cellular assays such as PLA, ChOR-seq, and iPOND should have enough resolution to detect most of the ssDNA-nucleosome occupancy, as long as it exists. However, developing higher-resolution assays will be necessary to further confirm whether nucleosomes bind ssDNA directly in cells in the future.
ssDNA gap events that inhibit histone assembly
Our findings revealed that PARylation, transcription, and DPC are three events that reduce nucleosome density. PARP1-mediated PARylation promotes chromatin relaxation via multiple mechanisms: 1) the negatively charged poly(ADP-ribose) (PAR) chains neutralize histone positive charges and weaken histone–DNA interactions^122^; and 2) PAR chains act as docking platforms to recruit DNA repair enzymes and chromatin remodelers, such as ALC1 (CHD1L)^119,120,141–143^. Together, these mechanisms enhance chromatin accessibility, exposing ssDNA regions to repair factors. Surprisedly, we found that PARP inhibitor Olaparib treatment reduce the nucleosome density on nascent DNA, which is distinct from PARP1 depletion. This phenotype consistent with the recent finding that PARP inhibition releases the histones{Moser, 2025 #111}. We hypothesize that, PARP1 depletion removes PARylation and leaves ssDNA gaps relatively “clean,” so the nucleosome can form on the ssDNA gaps. However, PARP inhibitor blocks PARylation and trap PARP1 on DNA, since PARylation normally promotes PARP1 release^144,145^. We therefore propose that trapped PARP1 directly displaces nucleosomes. This idea is consistent with recent findings showing that PARP1 inhibition leads to histone eviction from replication fork^146^.
We examined the role of HMCES-DPCs, which were significantly increased in FRKO cells (Figure 7B). HMCES depletion rescued the low nucleosome density in FRKO cells, as shown by PLA (Figure 7C), and this was further confirmed by in vitro reconstitution assays (Figure 7F–G). Surprisingly, HMCES-DPCs impaired nucleosome assembly only when positioned near the nucleosome dyad (Figure 7G), not at the entry or exit sites (Figure 7H). However, we cannot rule out inhibitory effects of entry/exit DPCs in cells. This is because the discrepancy between in vivo and in vitro system: (1) our reconstitution assays involve single nucleosomes, where ssDNA gaps at the entry/exit sites are flexible and may minimize DPC effects; (2) the Widom 601 sequence was specifically selected for nucleosome formation assay in previous studies, which has high histone-binding affinity, which could overcome HMCES-DPC–mediated inhibition; and (3) in cells, HMCES-DPCs can undergo additional processing, such as ubiquitination or proteolytic degradation, triggering downstream repair cascades that may collectively reduce nucleosome density^127,147^. These complex in vivo events are not captured by simplified in vitro systems.
We cannot exclude other ssDNA binding proteins, or any ssDNA-related events that also compromise nucleosome formation. Therefore, we conclude that PARylation, RNA Pol II, and DPC are the least events that inhibit histone assembly.
Materials and Methods
Cell lines
U2OS and MDA-MB-436 cells were cultured in DMEM with 9% fetal bovine serum (FBS) (Biotechne). PEO4, PEO1, and DLD-1 cells were cultured in RPMI-1640 with 9% FBS. UWB1.289 cells were cultured in a 1:1 mixture of RPMI-1640 medium and Mammary Epithelial Growth Medium (MEGM), supplemented with FBS and antibiotics. For iPOND-MS, U2OS cells were cultured in SILAC-compatible DMEM (Thermo Fisher Scientific A33822) with 10% dialyzed FBS (R&D Systems S12850H) and either isotopically light or heavy 13C, 15N lysine, and 13C,15N arginine (Cambridge Isotope Laboratories)^70^. Cells were cultured at 37°C and 5% CO2 with humidity. All cell lines were regularly tested for mycoplasma and verified using short tandem repeat profiling.
Genome editing
PrimPol knockout was generated using CRISPR-Cas9 with the SpCas9(BB)-2A-Puro (PX459) V2.0 plasmid (Addgene #62988) as described^148^. The single guide RNA (sgRNA) sequence was 5’- CTGCCAAGCAAGTCAAGAGC-3’. Transfected cells were selected with puromycin, and single-cell clones were isolated and validated by sequencing and immunoblotting.
SILAC and Label free iPOND-mass spectrometry
iPOND-SILAC MS was performed as described previously^70^. Briefly, EdU-labeled and indicated drug-treated cells were crosslinked with 1% formaldehyde, quenched with glycine, permeabilized in 0.25% Triton X-100. Heavy-labeled and light-labeled cells were combined 1:1 prior to the click reaction. DNA-protein complexes were captured with Dynabeads MyOne Streptavidin C1 (Thermo Fisher 65002), washed, and eluted with elution buffer (25mM biotin, 0.4% SDS, 50mM Tris-HCl (pH 8.5)) at 65°C. Supernatant, containing eluted fractions, were incubated at 95°C to reverse crosslinking. For histone modification analysis, lysine were derivatized with 2.5% propionic anhydride (Sigma-Aldrich 8.00608) in 50 mM ammonium bicarbonate (pH 7.5). Subsequently, proteins were digested with 200 ng trypsin (Thermo Fisher 90057) in 50 mM ammonium bicarbonate (pH7.5) overnight. Peptides were desalted using C18 Stagetips (reversed-phase resin) and carbon Top-Tips (Glygen TT1CAR), speed-vacuumed dried, and stored at −20°C until MS analysis.
PrimPol complementation
PrimPol WT, AxA, and CH mutants were subcloned from pcDNA3.1/nV5-DEST to pLNCX2 using forward primer: 5’-CAGATCTCGAGATGAATAGAAAATGGGAAGCAA-3’ and reverse primer: 5’-CTGCGGCCGCCTCTTGTAATACTTCTATAATTAG-3’. pLNCX2-Primpol constructs and pVSV-G were co-transfected into GP2-293 cells using Polyethylenimine (PEI). Twenty-four hours post-transfection, the culture medium was replaced with fresh medium, and viral supernatants were harvested at 48 and 72 hours. The pooled viral supernatants were subsequently used to infect U2OS Primpol knockout #206 cells. Stable cell lines were established by selection with 100 μg/ml G418.
DNA combing assays
Cells were labeled with the nucleoside analogs as indicated in the figures and figure legends. For ssDNA gap detection, S1 nuclease digestion assay was performed as previously described^66^. DNA combing was performed using a DNA combing instrument as previously described^54^.
Immunofluorescence
For analysis of proteins in the insoluble/chromatin fraction, cells were pre-extracted with 0.5% Triton X-100 and subsequently fixed in 4% paraformaldehyde. Slides were blocked with 1% goat serum in PBST (0.1% triton-X100) and incubated with primary antibodies. EdU incorporation was visualized by click chemistry with an Alexa Fluor 488-conjugated azide. Immunofluorescent images were obtained and analyzed by ImageXpress microscopy and software (Molecular Devices).
PLA assay
Proximity ligation assays (PLAs) were performed using the Duolink kit (Sigma DUO92002, DUO92004, DUO92008) according to the manufacturer’s instructions, with the following modifications as described previously ^71^. Cells were pre-extracted and fixed as described in the immunofluorescence method, then blocked with 3% BSA and processed by click chemistry with biotin-azide and Alexa Fluor 488-conjugated azide (9:1 ratio). The following primary antibody pairs were used: rabbit H2A (CST 12349) (1:100 dilution), H3 (Abcam ab1791) (1:200 dilution), H3K9ac (Millipore 07-352) (1:100 dilution), H3K9me3 (CST 13969) (1:100 dilution), H4 (CST 13919) (1:100 dilution); mouse biotin (Millipore B7653) (1:200 dilution) and mouse H2B (Abcam ab52484) (1:100 dilution), H3K9me2 (Abcam ab1220) (1:100dilution) and rabbit biotin (CST 5597) (1:100 dilution). Images were obtained by ImageXpress Pico and analyzed by MetaXpress (Molecular Devices).
CHOR-Seq
ChoR-seq was performed as previous described^72^. Briefly, WT and FRKO U2OS cells (n = 3 biological replicates each) were labeled with 10 μM EdU for 15 min for no HU sample. For HU treated sample, the cells are treated with 10uM EdU and 1 mM HU (Sigma H8627) for 4 h before fixation with the truChIP kit (Covaris 520127). Chromatin was sonicated to 100–500 bp, and EdU-labeled Drosophila S2 chromatin was added at 0.05% (w/w) as a spike-in for normalization. H3 ChIP was performed with anti-H3 (10 μg, Abcam ab1791). Immunoprecipitated DNA was processed with the NEBNext Ultra II DNA Library Prep Kit (NEB E7645S), indexed, and sequenced on an Illumina NovaSeq platform.
Data processing
All sequencing reads were mapped to the H. sapiens reference genome (hg38) and D. melanogaster reference genome (BDGP6) using Bowtie2 (v2.5.4) with the default parameters^149^. Only uniquely mapping reads were kept, and PCR duplicates were removed using Markup from Samtools (v1.17) (https://www.htslib.org/doc/samtools-markdup.html). Genome occupancy was calculated using bamCoverage from Deeptools (v3.5.5)^150^ with standard parameters in 100 bp intervals, normalized to relative reads per million (RRPM) using D. melanogaster reads as described previously^151^, and visualized by the Integrative Genomics Viewer (IGV) (v2.18)^152^. To generate a common H3 peak dataset, peak calling in each sample were preformed using MACS2 (v2.2.7)^153^ with parameters of “--broad-cutoff 0.1” and merged into a consensus region set using the Intersect function in Bedtools (v2.26.0)^154^. Subsequently, for H3 overlapped peaks in RRPM-normalized datasets were exported with the computeMatrix tool from Deeptools, and the enrichment patterns were visualized using the plotProfile and plotHeatmap function in DeepTools. Lastly, using the Intersect function in bedtools to get the H3 distribution on genome repeat elements and visualize in R.
Preparation of oligonucleotides for recombinant nucleosomes
All oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA). A complete list of oligonucleotides used to make recombinant nucleosomes can be found in Supplementary Table 1. Each oligonucleotide was resuspended to a final concentration of 100 μM in a buffer containing 10 mM Tris (pH-7.5) and 1 mM EDTA. Complimentary oligonucleotides (see Supplementary Table 1) were mixed to a final concentration of 10 μM and annealed by heating to 95°C for 2 minutes and cooling to 10 °C at a rate of −5 °C/min. The annealed oligonucleotides were stored short-term at 4 °C prior to nucleosome reconstitution.
Radar Assay
Cells were lysed in Radar buffer (4M Guanidine thiocyanate, 1% Sarkosyl, 2% Triton X-100, 1% 1,4-dithioerythritol, 100 mM Sodium Acetate pH5.0, 20 mM Tris pH 8.0, 20 mM EDTA pH8.0; adjusted to pH 6.5). Genomic DNA was sonicated and precipitated with ethanol, resuspended in 8 mM NaOH, and quantified using Qubit fluorometry. Samples were then subjected to slot-blot onto nitrocellulose membrane, which were subsequently immunoblotted with HMCES primary antibody (Cell Signaling #17636) and corresponding secondary antibody.)
RNA interference
All siRNA transfections were performed using DharmaFECT reagents (Horizon Discovery T-2001-03) according to the manufacturer’s instructions. Experiments were performed 2 days after transfection. Qiagen AllStars Negative Control Nontargeting (NT) siRNA was used as a non-targeting control. siRNA sequences:
RAD51: AAGUGCUGCAGCCUAAUGAGAGUGUU,
Primpol: GAGGAAACCGUUGUCCUCAGUGUAUUU,
FEN1: UACUCUCACAGUAGUCACUGCUU,
Ligase1 #1 GGCAUGAUCCUGAAGCAGAUU,
Ligase1 #2 AAGGGCAAGACAGCAGAGGCCUU,
PARP1 #1 GGUGAUCGGUAGCAACAAAUU,
PARP1 #4 CCAUCGAUGUCAACUAUGAUU,
HMCES#1 GCGAACAUCCUGUCACUUAUU,
HMCES#3 ACCAACUGUCGUAGUGAUAUU.
Purification of recombinant human histones
The genes encoding human histone H2A (UniProt identifier: P0C0S8), H2B (UniProt identifier: P62807), H3 C110A (UniProt identifier Q71DI3), and H4 (Uniprot identifier: P62805) were cloned into a pET3a expression vector. Histone H2A, H3, and H4 were individually transformed and expressed in BL21 (DE3) pLysS competent E. coli cells (New England BioLabs). Histone H2B was transformed and expressed in BL21-CodonPlus (DE3)-RIPL E. coli cells (Agilent). The histones were grown in M9 minimal media (33.5 mM Na_2_HPO_4_•7H2O, 220 mM KH_2_PO_4_, 100 mM NaCl, 200mM NH_4_Cl, pH 7.2) supplemented with 0.4% glucose (w/v), 2 mM MgSO_4_, 0.2 mM CaCl_2_ and a 1% vitamin cocktail. The histones were grown at 37 °C to an OD600 of 0.4, and histone expression induced with a 0.4 mM (histone H2A, H2B, and H3) or 0.3 mM (histone H4) isopropyl- β-D-thiogalactoside (IPTG) for 3 hours (histone H3 and H4) or 4 hours (histone H2A and H2B). The cells were harvested via centrifugation and cell pellets stored long term at −80 °C. The individual histones were purified from inclusion bodies using well-established methods^96–98^. In brief, the histone cell pellets were lysed via sonication in a buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM benzamidine, 1 mM DTT, and 1 mM EDTA and the cell lysate clarified via centrifugation. The resulting pellet was washed three times with a buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM benzamidine, 1 mM DTT, 1 mM EDTA, and 1% Triton X-100 and washed a final time with a buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM NaCl, 1 mM benzamidine, 1 mM DTT, and 1 mM EDTA. The histones were extracted from inclusion bodies under denaturing condition (6 M Guanidinium-HCl), dialyzed into 8 M Urea, and purified using anion-exchange and cation-exchange chromatography under gravity flow. The purified histones were dialyzed five times against H_2_O, lyophilized, and stored long-term at −20 °C.
Preparation of H2A/H2B Dimer and H3/H4 Tetramer
Prior to nucleosome assembly, H2A/H2B dimers and H3/H4 tetramers were prepared using established methods ^96–98^. Each individual lyophilized histone was resuspended in a buffer containing 20 mM Tris-HCl (pH 7.5), 6 M Guanidinium-HCl, and 10 mM DTT. To refold the H2A/H2B dimer, H2A and H2B were mixed in a 1:1 molar ratio and dialyzed three times against a buffer containing 20 mM Tris-HCl (pH 7.5), 2 M NaCl, and 1 mM EDTA at 4 °C. To refold the H3/H4 tetramer, H3 and H4 were mixed in a 1:1 molar ratio and dialyzed three times against a buffer containing 20 mM Tris-HCl (pH 7.5), 2 M NaCl, and 1 mM EDTA at 4 °C. The refolded H2A/H2B dimer and H3/H4 tetramer were subsequently purified using a HiPrep Sephacryl S-200 16/60 HR gel filtration column (Cytiva) in a buffer containing 20 mM Tris-HCl (pH 7.5), 2 M NaCl, and 1 mM EDTA. Pure fractions containing H2A/H2B dimers and H3/H4 tetramer were combined and stored long term at −20 °C in a 50% glycerol slurry.
Nucleosome assembly and purification
All nucleosomes were prepared using a well-established salt-dialysis method ^96–98^, with minor modifications. In brief, DNA, H2A/H2B dimer, and H3/H4 tetramer were mixed at a ratio of 1:2.2:1:1, respectively, in a buffer containing 20 mM Tris-HCl (pH 7.5), 2 M NaCl, 1 mM EDTA, and 1mM DTT. Nucleosomes were assembled via a stepwise decrease in NaCl concentration over a 24-hour period - from 2 M NaCl to 1.5 M NaCl, to 1.0 M NaCl, to 0.5 M NaCl, to 0.25 M NaCl, and to 0 M NaCl. The assembled nucleosomes were then purified via ultracentrifugation for 42 to 44 hours at 4 °C using a 10% - 40% sucrose gradient. Final nucleosome purity and homogeneity was assessed by native PAGE (5%, 59:1 acrylamide:bis-acrylamide ratio). The purified nucleosomes were stored in a buffer containing 10 mM Tris (pH-7.5) and 1 mM EDTA at 4 °C.
Purification of recombinant UDG and HMCES
A pET-His-GFP vector (N-terminal His-GFP tag) with the human uracil DNA glycosylase (UDG) gene was obtained from GenScript. The pET-His-GFP-UDG vector transformed into BL21-CodonPlus (DE3)-RIPL E. coli cells (Agilent). The transformed cells were grown at 37 °C to an OD600 of 0.6 and UDG expression induced using 1.0 mM IPTG for 18 h at 18 °C. The cells were lysed via sonication in a buffer containing 25 mM HEPES (pH 7.5), 200 mM NaCl, 0.1 mM TCEP, 20 mM Imidazole, and a cocktail of protease inhibitors (AEBSF, leupeptin, benzamidine, pepstatin A). The lysate was clarified via centrifugation and loaded onto a HisTrap HP column (Cytiva) equilibrated with 25 mM HEPES (pH 7.5), 200 mM NaCl, 0.1 mM TCEP, and 20 mM Imidazole, and eluted in a buffer 25 mM HEPES (pH 7.5), 200 mM NaCl, 0.1 mM TCEP, and 400 mM Imidazole. The UDG protein was then liberates from the His-GFP tag via TEV protease and further purified using a HiPrep SP HP column (Cytiva) equilibrated with 25 mM HEPES (pH 7.5), 50 mM NaCl, and 0.1 mM TCEP, and eluted in a buffer 25 mM HEPES (pH 7.5), 1 M NaCl, and 0.1 mM TCEP. The resulting UDG protein was combined and purified via size exclusion chromatography using a HiPrep Sephacryl S-200 16/60 HR (Cytiva) in a buffer containing 20 mM HEPES (pH 7.5), 150 mM NaCl, and 0.1 mM TCEP. The purity of the UDG proteins were confirmed via denaturing SDS-PAGE and the purified UDG protein stored long term at −80 °C.
Recombinant human HMCES protein was purified using a previously described protocol^155^. In brief, a pNIC-HMCES vector was transformed in E. coli cells. Cells were grown in terrific broth (TB) medium at 37 °C to an OD_600_ of 0.7 and HMCES expression induced by the addition of 0.5 mM IPTG for 4 hrs. Cell pellets were collected via centrifugation and resuspended in a buffer containing 20 mM HEPES/KOH (pH 7.5), 500 mM KCl, 5 mM MgCl_2_, 30 mM imidazole, 10% glycerol, 0.1% IGEPAL, 0.04 mg/ml Pefabloc SC, 1 mM TCEP and cOmplete EDTA-free protease inhibitor cocktail. The lysate was loaded onto a Strep-Tactin^®^ XT Superflow^®^ high-capacity cartridge, washed with a buffer containing 20 mM HEPES/KOH pH 7.5, 500 mM KCl, 5 mM MgCl_2_, 1 mM TCEP, and eluted in a buffer containing 20 mM HEPES/KOH pH 7.5, 500 mM KCl, 5 mM MgCl_2_, 1 mM TCEP, and 50 mM biotin. HMCES was further purified using a HiTrap Heparin HP columns and eluted with buffer containing 20 mM HEPES/KOH pH 7.5, 1 M KCl, 5 mM MgCl_2_, 1 mM TCEP. The HMCES protein was combined and further purified using a HiPrep Sephacryl S-200 16/60 HR (Cytiva) in a buffer containing 20 mM HEPES/KOH pH 7.8, 150 mM KCl, 5 mM MgCl_2_, 10% glycerol, 1 mM TCEP. The purity of the HMCES protein was confirmed via denaturing SDS-PAGE and purified HMCES protein stored long term at −80 °C.
Generation of HMCES-DPC and HMCES-DPC nucleosome formation
AP-site DNA was generated via enzymatic digestion of dU-containing oligos (see Supplementary Table 1. The dU-containing oligonucleotides (10 μM) were incubated with recombinant UDG (10 μM) for 2 hours at 25 °C to excise the uracil and generate an AP-site. The oligonucleotides were then incubated with recombinant HMCES (20 μM) for 2 hours at 25 °C, and the HMCES-DPC stabilized via sodium borohydride (NaBH_4_) crosslinking for an additional 2 hours at 25 °C. The HMCES-DPC-DNA was loaded onto a HiPrep Q HP column (Cytiva) equilibrated with 50 mM HEPES (pH 7.5) and 400 mM NaCl, and eluted in a buffer 50 mM HEPES (pH 7.5) and 2 M NaCl. Fractions containing the HMCES-DPC-DNA were combined, exchanged into a buffer containing 20 mM Tris-HCl (pH 7.5), 1 mM EDTA, 1 M NaCl, and 1mM DTT, and stored short-term at 4 °C prior to nucleosome reconstitution.
To generate HMCES-DPC-DNA nucleosomes, HMCES-DNA, H2A/H2B dimer, and H3/H4 tetramer were mixed at a ratio of 1:2.2:1, respectively, in a buffer containing 20 mM Tris-HCl (pH 7.5), 1 M NaCl, 1 mM EDTA, and 1mM DTT. Nucleosomes were assembly via a stepwise decrease in NaCl concentration over a 24-hour period - from 1 M NaCl to 0.8 M NaCl, to 0.6 M NaCl, to 0.4 M NaCl, to 0.2 M NaCl, and to 0 M NaCl. The formation of the HMCES-DPC nucleosomes was assessed via native PAGE gels (5%, 59:1 acrylamide:bis-acrylamide ratio).
FRET assay
Oligonucleotides in FRET assay
Oligonucleotides comprising the single strand DNA (ssDNA) gap substrate (Table 1) were synthesized by Integrated DNA Technologies (Coralville, IA) or Bio-Synthesis (Lewisville, TX) and purified on denaturing polyacrylamide gels. The concentrations of unlabeled DNAs were determined from the absorbance at 260 nm using the provided extinction coefficients. Concentrations of Cyanine-labeled DNAs were determined from the extinction coefficient of their respective label (Cy3 at 550 nm, ε_550_ = 150,000 M^−1^cm^−1^, Cy5 at 650 nM, ε_650_ = 250,000 M^−1^cm^−1^). For annealing the ssDNA gap substrate, the template strand and the cyanine-labeled primer strands were mixed in equimolar amounts in 1X annealing buffer (10 mM TrisHCl, pH 8.0, 100 mM NaCl, 1 mM EDTA), heated to 95 °C for 5 minutes, and allowed to slowly cool to room temperature.
Recombinant Human Proteins in FRET assay
Human RPA was obtained as previously described ^156^. The concentration of active RPA was determined via a FRET-based activity assay as described previously ^109^. Human recombinant histone H3.1 – H4 tetramers (Cat. # 14-1056) and histone H2A-H2B dimers (Cat. # 10-1052) were purchased from EMD Millipore (Temecula, CA) and utilized to reconstitute nucleosomes on the ssDNA gap substrate via stepwise salt dilution dialysis method, as described previously ^157^. Assembled nucleosomes were purified and verified by native PAGE, as described previously ^157^.
FRET Experiments
All experiments were performed in a 16.100F-Q-10/Z15 sub-micro fluorometer cell (Starna Cells) at room temperature (23 ± 2 °C) in 1X Replication Buffer (25 mM TrisHCl, pH 7.5, 125 mM KOAc, 10 mM Mg(OAc)2) supplemented with 1 mM DTT, 0.1 mg/mL BSA, and the ionic strength adjusted to physiological (200 mM) by addition of KOAc. Fluorescence intensities were monitored in a Horiba Scientific Duetta-Bio fluorescence/absorbance spectrometer with excitation and emission slit widths set at 10 nm. For all experiments described below the final concentrations of all reaction components are indicated. The concentration of RPA (6.76 μM) utilized in the experiments described below agrees with the concentration of RPA in the nucleoplasm of a human cell (10 μM range)^103–105^.
For steady state (equilibrium) FRET assays in Figure S5B, a DNA substrate (100 nM ssDNA Gap or ssDNA Gap Nuc) is pre-incubated for 5 minutes in the absence or presence of RPA (6.76 μM heterotrimer). Under these conditions, the concentration of RPA is ~11-fold higher than the concentration of RPA-binding sites within the the ssDNA Gap DNA substrate. The solution is then transferred to a fluorometer cell and the cell is placed in the instrument. Next, the solution is excited at 514 nm and the fluorescence emission intensities (I) are monitored essentially simultaneously (Δt = 0.118 ms) at the peak emission wavelengths for Cy3 (563 nm, I563) and Cy5 (665 nm, I665) over time until both stabilize for at least 1 min. The I665 and I_563_ values within this stable region are utilized to calculate the approximate FRET efficiencies (EFRET) from the equation and the resultant values are averaged to obtain the EFRET value observed for a given condition at equilibrium. Finally, the solution is excited at 514 nM and the fluorescence emission spectra is recorded from 530 to 750 nm.
For pre-steady state FRET assays (Figure S5C) ssDNA Gap Nuc (100 nM) is transferred to a fluorometer cell and the cell is placed in the instrument. The solution is excited at 514 nm and I563 and I665 are monitored over time (as described above) until both stabilize for at least 1 min. The I665 and I_563_ values within this stable region are utilized to EFRET and the resultant values are averaged to obtain the EFRET value observed prior to the addition of RPA. Next, RPA (6.76 μM heterotrimer) is added, the resultant solution is mixed, and I665 and I_563_ are monitored beginning 30 s after the addition of RPA. All recorded fluorescence emission intensities are corrected by a respective dilution factor. For all recordings of the fluorescence emission intensities (I_665_ and I_563_), EFRET values are calculated as described above and plotted as a function of time after RPA is added.
For all plots in all figures, each data point/column/spectra represent the average ± standard error of the mean (SEM). Error bars are present for all data points on all plots in all figures but may be smaller than the data point.
Quantification and statistical analysis
The statistical test was performed using GraphPad Prism and R software. The statistics for the high-content microscopy experiments was performed on the mean of individual replicates. Statistical analyses were completed using Prism. A Kruskal-Wallis test was used for experiments with more than two samples, and P values were calculated by Prism for the multiple comparisons. A two-tailed t test was used to compare two samples with normally distributed data. No statistical methods or criteria were used to estimate sample size or to include/exclude samples. Statistical details of individual experiments can be found in the figure legends and in Results. Experiments shown are representative of at least two biological replicates unless otherwise indicated in the figure legend.
Supplementary Material
1
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Stewart-Morgan K. R., Petryk N. & Groth A. Chromatin replication and epigenetic cell memory. Nat Cell Biol 22, 361–371 (2020). 10.1038/s 41556-020-0487-y 32231312 · doi ↗ · pubmed ↗
- 2Alabert C. & Groth A. Chromatin replication and epigenome maintenance. Nature reviews Molecular cell biology 13, 153–167 (2012).22358331 10.1038/nrm 3288 · doi ↗ · pubmed ↗
- 3Berti M., Cortez D. & Lopes M. The plasticity of DNA replication forks in response to clinically relevant genotoxic stress. Nature reviews Molecular cell biology 21, 633–651 (2020).32612242 10.1038/s 41580-020-0257-5 · doi ↗ · pubmed ↗
- 4Allis C. D. & Jenuwein T. The molecular hallmarks of epigenetic control. Nature reviews genetics 17, 487–500 (2016).
- 5Cantone I. & Fisher A. G. Epigenetic programming and reprogramming during development. Nat Struct Mol Biol 20, 282–289 (2013). 10.1038/nsmb.248923463313 · doi ↗ · pubmed ↗
- 6Suva M. L., Riggi N. & Bernstein B. E. Epigenetic reprogramming in cancer. Science 339, 1567–1570 (2013). 10.1126/science.123018423539597 PMC 3821556 · doi ↗ · pubmed ↗
- 7Iacobuzio-Donahue C. A. Epigenetic changes in cancer. Annual Review of Pathology: Mechanisms of Disease 4, 229–249 (2009).
- 8Feinberg A. P., Ohlsson R. & Henikoff S. The epigenetic progenitor origin of human cancer. Nature Reviews Genetics 7, 21–33 (2006). 10.1038/nrg 1748 · doi ↗