DNA polymerase delta governs parental histone transfer to DNA replication lagging strand

Significance

As faithful DNA replication is a crucial process for genetic inheritance, the accurate copying of epigenetic information is vital for maintaining cell lineage. It is well known that DNA polymerases are responsible for genomic DNA duplication. In this study, we have identified that Pol δ, one of the DNA polymerases, serves as a key chaperone for transferring parental histones to the lagging strand during DNA replication, assuring the faithful inheritance of epigenetic information carried by parental histone posttranslational modifications to the daughter cells. This finding reveals an exquisite process that Pol δ facilitates parental histone inheritance while replicating the lagging DNA strands.

Abstract

Chromatin replication is intricately intertwined with the recycling of parental histones to the newly duplicated DNA strands for faithful genetic and epigenetic inheritance. The transfer of parental histones occurs through two distinct pathways: leading strand deposition, mediated by the DNA polymerase ε subunits Dpb3/Dpb4, and lagging strand deposition, facilitated by the MCM helicase subunit Mcm2. However, the mechanism of the facilitation of Mcm2 transferring parental histones to the lagging strand while moving along the leading strand remains unclear. Here, we show that the deletion of Pol32, a nonessential subunit of major lagging-strand DNA polymerase δ, results in a predominant transfer of parental histone H3–H4 to the leading strand during replication. Biochemical analyses further demonstrate that Pol32 can bind histone H3–H4 both in vivo and in vitro. The interaction of Pol32 with parental histone H3–H4 is disrupted through the mutation of the histone H3–H4 binding domain within Mcm2. Our findings identify the DNA polymerase δ subunit Pol32 as a critical histone chaperone downstream of Mcm2, mediating the transfer of parental histones to the lagging strand during DNA replication.

Recent studies have highlighted the crucial role of chromatin structure in gene regulation (1). In eukaryotic cells, the nucleosome serves as the basic unit of chromatin, comprising approximately 150 base pairs of DNA and a histone octamer with two copies of each of H2A, H2B, H3, and H4. Chromatin organization is not uniform along the chromosome; certain regions are densely packaged as heterochromatin, while others are loosely arranged as euchromatin (2, 3).

The replication of chromatin is intricately linked with DNA replication, yet a significant knowledge gap persists regarding the precise mechanisms governing chromatin inheritance during this process. Unlike the H2A–H2B dimers, which can be readily removed and turn over quickly during transcription, the H3–H4 tetramers are more stable and emerge as an ideal vector for preserving epigenetic memory across cell divisions (4, 5). Experimental evidence of the local recycling of histone H3–H4 tetramers during DNA replication supports this hypothesis (6). In line with this, the observation that the core histone component H3–H4 tetramer remains intact during DNA replication further reinforces this concept (7). As the replication fork advances, nucleosomes undergo the disassembly process to facilitate the passage of the replisome. Subsequently, the newly synthesized DNA daughter strands recycle parental histones from the parental strand and recruit newly synthesized histones, thereby restoring nucleosome structure. This nucleosome assembly process signifies the initial step in transferring epigenetic information from parent to daughter cells (8, 9). The faithful transfer and restoration of posttranslational histone modifications, carried by the parental histone H3–H4, are crucial for maintaining cell identity through cell division (3, 10).

Several newly synthesized H3–H4 chaperones, defined by both their binding to H3–H4 tetramers and roles in nucleosome assembly (8, 9), have been identified, including CAF1, RTT106, and ASF1. These chaperones not only play essential roles in chromatin replication, but also in the DNA damage repair process (11, 12). However, compared to the assembly of new H3–H4, the recycling process of parental histone poses greater challenges due to methodological difficulties. Biochemical analysis has shown that Mcm2, a subunit of the replicative helicase MCM, contains a histone-binding motif (HBM) allowing it to bind to the histone H3–H4 tetramer in yeast (13). Subsequent structural analysis revealed that the MCM–H3–H4 interaction is conserved among eukaryotic cells (14). Through genome-wide sequencing analysis tools, such as eSPAN (enrichment and sequencing of protein-associated nascent DNA) (15, 16), as depicted in SI Appendix, Fig. S1A, several mediators of parental histone H3–H4 transfer have been identified through the altered balance of leading and lagging strand inheritance while perturbed (15, 17, 18). The DNA polymerase ε subunit Dpb3/Dpb4 has been shown to bind to H3–H4 in vitro and mediates the transfer of parental histone H3–H4 tetramers to the leading strand. Meanwhile, the Mcm2-Ctf4-Polα axis facilitates the transfer of parental histones to the lagging strand. Mutations in the HBM of Mcm2 or mutations disrupting the CMG-Ctf4-Polα interaction result in defective transfer of parental histone H3–H4 tetramers to the lagging strand (15, 17–19). Moreover, Polα has also been shown to directly bind to histone H3–H4 tetramers and facilitates lagging strand transfer (19, 20). Although mutations altering these parental histones transfer only exhibit a minor phenotype in chromatin stability in budding yeast model (13, 17, 21), there are profound impacts on the differentiation of mouse embryonic stem cells and more importantly mouse embryonic development (22–24).

However, the disruption of both Dpb3/Dpb4 and Mcm2-Ctf4-Polα pathways does not impede cell growth (21), suggesting the presence of an additional regulatory mechanism compensating for these two canonical pathways. In this study, we found that the nonessential DNA polymerase δ subunit, Pol32, binds to histone H3–H4 both in vivo and in vitro. Depletion of Pol32 leads to a predominant transfer of parental histone H3–H4 to the replication leading strand. Furthermore, biochemical analysis shows that the binding of Pol32 to parental histones is dependent on Mcm2. In conclusion, we identify DNA polymerase δ as a key histone chaperone downstream of Mcm2, mediating the transfer of parental histone to the lagging strand.

Results

Pol32 Is a Parental H3–H4 Histone Chaperone.

Our previous work has shown that the strand bias effect is more pronounced when disrupting the Mcm2–H3–H4 interaction (mcm2-3A mutant) than eliminating the Pol1 (polymerase1)–H3–H4 interaction (pol1-2A2 mutant) (SI Appendix, Fig. S1 B and C) (18, 19). This observation implies the presence of additional receptors for parental H3–H4 on the replication lagging strand. In yeast, newly deposited H3–H4 is typically marked with a H3K56ac modification, which is gradually removed by histone deacetylase during later S/G2 phases (25, 26). Subsequently, histone methyltransferases add methylation modifications to the nucleosomes (27). To identify candidate factors that changed the parental histone transfer pattern, we deleted several nonessential components involved in lagging strand replication (SI Appendix, Fig. S1D) and used H3K4me3-specific antibodies to track parental histone H3–H4 with the eSPAN method (Procedure see SI Appendix, Fig. S1A) (16–18). Pol32, the human homolog of POLD3, emerged from this screening as a key regulator. Compared to a slightly lagging strand bias of the WT control, the H3K4me3 eSPAN of pol32Δ exhibited a strong leading strand bias in HU (hydroxyurea)-treated early S phase cells (Fig. 1 A–C). The magnitude of this strand bias is comparable to that observed in the mcm2-3A mutant and is considerably stronger than in the pol1-2A2 mutant (SI Appendix, Fig. S1 B and C). Meanwhile, the H3K56ac eSPAN of pol32Δ showed a pronounced lagging strand bias (Fig. 1 A–C), reinforcing the conclusion that Pol32 participates in the deposition of parental histone H3–H4 on the replication lagging strand. Further analysis of the H3K4me3 eSPAN without HU treatment still displayed a clear leading strand bias (SI Appendix, Fig. S1 E and F). Taken together, we found that Pol32 is involved in parental histone transfer.

Fig. 1.

Pol32 transfers parental histone H3–H4 to lagging strands. (A) The average bias ratio of H3K4me3 (Upper) and H3K56ac (Lower) eSPAN signal at each of the 10 nucleosomes flanking the early replication origins in WT and pol32Δ cells. For cell culture, alpha-factor arrested G1 cells were released into medium containing HU (200 mM) and BrdU for 45 min. The sample treatment following SI Appendix, Fig. S1A. (B) Heatmap showing the bias ratio and pattern of H3K4me3 (Left) and H3K56ac (Right) eSPAN signal at each nucleosome flanking the early replication origins in pol32Δ cells. (C) Snapshot of BrdU-IP-ssSeq, H3K4me3, and H3K56ac eSPAN peaks surrounding the ARS1309 origin in pol32Δ cells. (D) Experimental design to track the parental H3 with HA-tag (H3–HA) and newly synthesized H3 with T7-tag (H3-T7). The experimental procedure following previous publications (17, 28). (E) The average bias of H3–HA (Upper) and H3-T7 (Lower) eSPAN signal at the early replication origins in WT and pol32Δ cells. (F) In vivo interaction of Pol32 with H3–H4. TAP-tagged Pol32 (Pol32-TAP) was pulled down from yeast cell lysate with IgG beads. Copurified proteins were separated by SDS-PAGE and detected with specific antibodies. Cells without Pol32-TAP expression served as a negative control. (G) Immunoprecipitation (pull-down) recombinant hH3.1–H4 (human) tetramers by recombinant glutathione S-transferase (GST)-Pol32 or GST protein. Bound proteins were detected by Coomassie Brilliant Blue (CBB) staining.

To better understand the mechanisms of which Pol32 regulates parental histone transfer, without relying on the histone modification marks, we adapted the Recombination-Induced Tag Exchange system (6, 28) to mark all the parental histones with a HA (hemagglutinin) tag. Then recombination is induced prior to early S phase release to tag the new histones with T7 (Fig. 1D). With this system, we confirmed that H3–HA (parental) and H3–T7 (new) eSPAN revealed a distinct leading and lagging strand pattern across all of the genomic origins of replication (Fig. 1E). This pattern clearly contrasts with the lack of a clear bias in the WT control for both H3–HA and H3–T7 eSPAN (Fig. 1E). These findings collectively support the involvement of Pol32 in transferring parental histone H3–H4 to the replication lagging strand.

It has been shown that the Dpb3/Dpb4 of Pol ε and catalytic subunit of Polα can directly bind to histone H3–H4 tetramers (13, 19). However, it remains unknown whether Pol δ can bind to H3–H4. To test this, we then performed the Pol32 immunoprecipitation in DNase I digested chromatin and observed its interaction with H3 and H4 within the cell (Fig. 1F). Mcm2, which has been reported to bind H3–H4 (13, 14), served as the immunoprecipitation positive control (Fig. 1F). We further confirmed the Pol32–H3–H4 interaction is direct, by using in vitro purified Pol32 and H3–H4 tetramers (Fig. 1G). These findings suggest that Pol32 can act as a receptor or chaperone for parental H3–H4, to facilitate their assembly onto lagging strands.

Pol32 Receives Parental H3–H4 from Mcm2 and Transfers It to DNA Replication Lagging Strand.

Next, we asked whether the relationship between Mcm2 and Pol32 in parental histone H3–H4 transfer is parallel or sequential at the replication fork. To test this, we conducted Pol32 immunoprecipitation in both WT and mcm2-3A strains. Compared to WT cells, the amount of H3–H4 pulled down by Pol32 significantly decreased under the mcm2-3A mutant context (Fig. 2A). Importantly, the Pol32-coprecipitated H3K4me3 (likely parental histone) also decreased, while the Pol32-bound H3K56ac (likely free histone) remained unchanged in the mcm2-3A mutant (Fig. 2A). These data suggest that the binding of Pol32 to parental histone H3–H4 is dependent on Mcm2, supporting the idea that parental histone H3–H4 is transferred from Mcm2 to Pol32.

Fig. 2.

Pol32 receives parental histones from Mcm2. (A) Analysis of Pol32-TAP-associated proteins from WT, mcm2-3A, and pol1-2A2 mutant cells. (B) Analysis of the Pol32–Mcm2 interaction by Pol32-TAP purification in WT and mcm2-3A mutant cells. (C) H3–H4 tetramers promote Mcm2-Pol32 interaction. To assess the effect of H3–H4 on the interaction between Mcm2 and Pol32, Mcm2 was incubated with or without H3–H4, and then, the products were added into beads bearing GST-Pol32 or GST proteins (negative control). The pull-down assay was performed in buffer containing 500mM NaCl. (D) Interaction model of Mcm2, H3–H4, and Pol32 at the replication fork. (E and F) ctf4Δ reduces the interaction of parental histones (H3K4me3 as a surrogate) with Pol32 (E), but does not interfere with the Mcm2-Pol32 interaction (F). Plasmids (pRS423-Pol32-3×Flag or empty vector) were transfected into WT, ctf4Δ, and ctf4-4E strains. Pol32-Flag was isolated by immunoprecipitation, and its associated proteins were monitored by immunoblotting. (G) Average eSPAN bias of H3K4me3 (Upper) and H3K56ac (Lower) at each nucleosome flanking the early replication origins in mcm2-3A, pol32Δ, and mcm2-3A pol32Δ cells. The data displayed in the box filled with grids have been shown in Fig. 1A. (H) Box plots showing the ratio of sequencing reads at lagging over leading strands of eSPAN signal around each early replication origin in mcm2-3A, pol32Δ, and mcm2-3A pol32Δ cells. Two replicates were performed. (I) eSPAN bias pattern of H3K4me3 (Upper) and H3K56ac (Lower) in pol32Δ strongly correlated with that of mcm2-3A.

We then explored whether Pol32 could interact with Mcm2 and whether this interaction depends on H3–H4. Coimmunoprecipitation experiments unveiled a robust interaction between Pol32 and WT Mcm2, whereas the interaction was diminished with the histone-binding mutant mcm2-3A (Fig. 2B). These data indicated that the Mcm2-Pol32 interaction is partially dependent on their association with H3–H4. Consistently, our in vitro binding assay with purified proteins also showed that the interaction between Pol32 and Mcm2 is stronger in the presence of H3–H4 (Fig. 2C), suggesting that Mcm2 and Pol32 can directly bind to each other, and the H3–H4 tetramer can mediate and enhance their interaction. These findings support a parental H3–H4 tetramer handoff model from Mcm2 to Pol32 (Fig. 2D).

Next, we asked whether the replisome hub protein Ctf4 also plays a role in the Pol32–Mcm2 interaction. Co-IP experiments showed that the interaction of Pol32 to H3K4me3 was slightly reduced (Fig. 2E), whereas the interaction with Mcm2 remained unaffected in ctf4Δ cells (Fig. 2F). In the Co-IP experiment involving the ctf4-4E mutant (disrupting the Ctf4-Pol1 interaction) (29), we observed little changes in both Pol32–Mcm2 and Pol32–H3–H4 interactions (Fig. 2 E and F). Altogether, the replisome hub protein Ctf4 has an additional role in supporting Pol32 in parental histone transfer; however, the interaction between Mcm2 and Pol32 is independent of Ctf4.

In conjunction with the biochemical analyses, we also employed the parental and new histone eSPAN to investigate the Mcm2-Pol32 interaction. The H3K4me3 (parental histone) eSPAN profiles of mcm2-3A and pol32Δ single mutants exhibited a similar strong leading bias pattern (Fig. 2 G and H and SI Appendix, Fig. S2 A and B). Moreover, the strand bias exhibited in the mcm2-3A and pol32Δ cells at all the replication origins are highly correlated (Fig. 2I). Importantly, combinations of mcm2-3A and pol32Δ mutants displayed synergistic effect compared to the single mutants in the H3K4me3 (parental histone) eSPAN analysis (Fig. 2 G and H and SI Appendix, Fig. S2 A and B). Meanwhile, the H3K56ac (new histone) eSPAN revealed a reciprocal pattern to that of the parental histone (Fig. 2 G and H and SI Appendix, Fig. S2 A and B), further solidifying the H3K4me3 (parental histone) results. These eSPAN analyses together with the biochemical data support a model whereby Mcm2 initially receives parental histone H3–H4 and subsequently transfers parental histones to Pol32 on the replication lagging strand (Fig. 2D). However, a proportion of the parental histone receipt by Pol32 is independent of Mcm2.

Pol1 Is a Pol32-Downstream Chaperone in the Parental Histone H3–H4 Transfer Pathway.

We next sought to explore the relationship between Pol1 and Pol32 in the parental histone H3–H4 transfer pathway. Within the pol1-2A2 (histone-binding mutant) context, Pol32 exhibited similar binding ability to H3–H4, H3K4me3, and H3K56ac when compared to the WT (Fig. 2A). Furthermore, we found that in the absence of Pol32, the interaction of Pol1 with H3K4me3-H4 was reduced, while the interaction with H3K56ac-H4 remained unchanged (Fig. 3A). We also conducted an in vitro binding assay with purified proteins and found that histone H3–H4 promotes the interaction between Pol32 and Pol1 (Fig. 3B). This is further supported by the in vivo result that Pol1–Pol32 interaction was weakened by the pol1-2A2 mutation, which eliminates the H3–H4 binding capability of Pol1 (Fig. 3C). Previous studies have identified the domains responsible for the Pol1–Pol32 interaction: Pol1 278-532a.a. (in fission yeast) and Pol32 270 to 309a.a. (30, 31). Deletion of either of two domains dramatically decreased the interaction between Pol1 and Pol32 (we specifically deleted the 278 to 348 region of Pol1 due to essential catalytic function located in Pol1 349 to 532a.a. region), without affecting the interaction between Pol32 and H3–H4 (Fig. 3D and SI Appendix, Fig. S3A). Moreover, disrupting the interaction of Pol1 with Pol32 (pol32Δ270-309) does not affect its interaction with H3–H4 (Fig. 3D and SI Appendix, Fig. S3B). Consistently, the parental histone transfer remained affected, as the H3K4me3 eSPAN of pol32Δ270-309 cells resembled the WT exhibiting a slight lagging strand bias (Fig. 3E). These results together suggest that the H3–H4 tetramer bridging, rather than the Pol1–Pol32 direct interaction, is crucial for the parental histone transfer pathway.

Fig. 3.

Pol1 functions as a Pol32-downstream chaperone in the parental histone H3–H4 transfer pathway. (A) Analysis of Pol1-TAP-associated proteins from WT and pol32Δ cells. (B) Pol32–Pol1 (1 to 527a.a.) interaction was enhanced in the presence of H3–H4 tetramers by in vitro binding assay. The beads bearing GST-pol32 or GST (negative control) proteins were first incubated with H3–H4 tetramers, followed by adding Pol1 (1 to 527). The pull-down assay was performed in a buffer containing 300 mM NaCl. (C) In vivo analysis of the Pol32–Pol1 interaction by Pol32-TAP purification in WT and pol1-2A2 mutant cells. (D) pol1Δ278-348 and pol32Δ270-309 are partial defects in Pol1–Pol32 interaction, but these deletion mutants do not affect their interaction with histone H3. Left and Right panels are analysis of Pol32-TAP- and Pol1-TAP-associated proteins, respectively. The 278 to 527a.a. of Pol1 and 270 to 309a.a. of Pol32 are reported involved in Pol1–Pol32 interaction (30–32). As 278 to 527a.a. of Pol1 is overlapped with the catalytic core (Pol1-core; 349 to 1,259a.a.) and Pol1Δ278-527 is lethal, we used a nonessential pol1Δ278-348 in this assay. Input controls are shown in SI Appendix, Fig. S3 A and B. (E) The average bias ratio of H3K4me3 eSPAN signal at each of the nucleosomes flanking the early replication origins in WT, pol32Δ, and pol32Δ270-309 cells. The data displayed in the box filled with grids have been shown in Fig. 1A. (F) The average eSPAN bias of H3K4me3 (Upper) and H3K56ac (Lower) at each nucleosome flanking the early replication origins in pol1-2A2, pol32Δ, and pol1-2A2 pol32Δ cells. The data displayed in the box filled with grids have been shown in Fig. 1A. (G) The eSPAN bias pattern of H3K4me3 (Left) and H3K56ac (Right) in pol32Δ is correlated with that of pol1-2A2.

Additionally, the Pol1 histone-binding mutant (pol1-2A2) exhibited a much smaller strand bias on both H3K4me3 and H3K56ac eSPAN, than the pol32Δ single or pol1-2A2 pol32Δ double mutants (Fig. 3F). Meanwhile, the pol32Δ single and pol1-2A2 pol32Δ double mutants exhibit comparable histone transfer patterns, for both H3K4me3 and H3K56ac eSPAN (Fig. 3F). Moreover, the pol1-2A2 and pol32Δ cells exhibit a strong correlation for both parental and new histone eSPAN (Fig. 3G). Overall, our eSPAN results indicate that Pol1 receives parental histone H3–H4 from Pol32 and transfers it to the replication lagging strand.

To provide a more complete view of the lagging strand transfer pathway, we conducted Mcm2 immunoprecipitation in WT, pol32Δ and pol1-2A2 strains. The interaction between Mcm2 and H3–H4 remained unaffected in pol32Δ and pol1-2A2 strains (SI Appendix, Fig. S3C). Furthermore, we also performed Pol1 immunoprecipitation in both WT and mcm2-3A strains. Compared to WT cells, the amount of H3–H4 pulled down by Pol1 significantly decreased under the mcm2-3A mutant context (SI Appendix, Fig. S3D). These results further suggest that Mcm2 acts upstream of Pol32, with Pol1 functioning downstream in the lagging strand transfer pathway.

The Histone Chaperone Function of Pol32 Is Independent of PCNA Interaction and DNA Binding Capability.

Furthermore, as Pol32 contains a PIP (PCNA interaction protein-box) motif at the C-terminal (33), we conducted the H3K4me3 eSPAN analysis of pol32-2A (PIP motif mutation: Pol32-F344A, F345A), and observed no bias (SI Appendix, Fig. S3E), indicating that the histone chaperone function of Pol32 is independent of its interaction with PCNA.

It has been shown that Pol32 functions as the processivity factor for DNA polymerase δ and facilitates its localization to the nucleus (34). It is possible that Pol32 also harbors DNA binding ability. Indeed, we observed a notable Pol32–DNA shift in the DNA binding assay (SI Appendix, Fig. S4A). However, supplementing additional DNA did not affect the interaction between Pol32 and H3–H4 (SI Appendix, Fig. S4B). These analyses collectively suggest that Pol32 indeed possesses DNA binding capability, and importantly, this domain does not impede its histone chaperone functions.

Pol32 Deletion Leads to Chromatin Instability.

Finally, we sought to further dissect the phenotypic function of Pol32. When grown within a rich growth medium, neither the pol32Δ mutant nor its combination with other mutants defective in parental histone transfer exhibit clear growth defects (SI Appendix, Fig. S5A). However, under replication stress conditions, for example with HU or camptothecin (CPT) treatment, they exhibit enhanced sensitivity (SI Appendix, Fig. S5A). This observed phenotype aligns with the well-established role of Pol32 in the DNA repair process (30, 35, 36). Interestingly, we did not observe any additive effect when combining pol32Δ with mutants affecting lagging strand parental histone transfer chaperones. Chromatin instability is a typical outcome of mutations in histone chaperones, and it has been shown that pol32Δ induces chromatin instability at both HML and telomeric regions through URA3/FOA analysis (37). To gain a deeper insight into its impact, we employed a more sensitive CRASH (Cre-reported altered states of heterochromatin) system (38) to investigate whether pol32Δ influences gene silencing in the HML locus (SI Appendix, Fig. S5B). Briefly, Cre is inserted to the HML locus, usually silenced within yeast. Cre would be derepressed and recombine the red fluorescent protein, green fluorescent protein (RFP-GFP) cassette once the silencing at the HML locus was lost, resulting in GFP expression. Thus, the ratio of the loss of silencing can be measured by the percentage of the cells expressing GFP. We found that pol32Δ leads to a onefold increase in the loss of silencing (SI Appendix, Fig. S5C), suggesting that Pol32 deletion leads to chromatin instability.

Discussion

In this study, we have identified the DNA polymerase δ subunit Pol32 as a key regulator of parental histone H3–H4 transfer to the DNA replication lagging strand. Using both in vivo and in vitro biochemical analyses, we showed that Pol32 can directly bind to histone H3–H4. Furthermore, we gained insight into the functional mechanism by showing that Pol32 operates downstream of Mcm2 and upstream of DNA Pol1, in the parental histone transfer pathway (SI Appendix, Fig. S5D). Our study unveils the molecular mechanism underlying the parental histone transfer process, a pivotal step in epigenetic inheritance.

Parental histone H3–H4 transfer, a key step in epigenetic inheritance, is intricately regulated by replisome components. There are three DNA polymerases (Pol α, Pol ε, Pol δ) responsible for genomic DNA replication (39, 40). Previously, the leading strand Pol ε subunits (Dpb3/Dpb4) and lagging strand Pol α catalytic subunit (Pol1) have been identified as involved in parental histone H3–H4 recycling (15, 18, 19). In this study, we showed that the Polδ subunit Pol32 is also involved in the recycling of parental histones. Just as mutations in the catalytic domain of DNA polymerases are associated with alterations in DNA replication fidelity, mutations in the chaperones of parental histone transfer are linked to chromatin instability (3, 41). These findings highlighted the crucial role of DNA polymerase complexes in chromatin replication and epigenetic instability. In the unicellular yeast system, chromatin instability can be frequently observed via loss of silencing at normally repressed loci such as silenced mating-type loci or telomeric regions. Chromatin instability has also been shown to have a role in antifungal drug resistance phenotypes (42). In Drosophila, the partition of parental histones plays an important role for germ cell differentiation (43). In mammals, chromatin instability may lead to cell differentiation and embryonic development defects (22, 23). Taken together, the fundamental DNA replication machinery plays a regulatory role in epigenetics inheritance and cell fate determination.

A pivotal question arises regarding why defects in the parental histone transfer process lead to chromatin instability. A “copy and paste” mechanism for histone tail modifications has been proposed to explain epigenetic inheritance (21, 44). The PRC2 complex, a major gene silencer in mammalian cells, binds to the histone tail modification H3K27me3, promoting further local H3K27me3 modifications (45). Similarly, in the absence of demethylase, Clr4 in fission yeast can perform this reader-writer function to replicate H3K9me3 modifications (44). This “read and write” mechanism is executed by PRC2 and Clr4 enzymes. In budding yeast, where H3K27me3 and H3K9me3 modifications are undetectable, current observations do not align with the copy and paste model. In the dpb3Δ and mcm2-3A mutants, parental histones exhibit a strong leading and lagging strand bias, respectively. Both single mutants display increased chromatin instability compared to wild-type cells (17, 18). For instance, a dpb3Δ mcm2-3A double mutant demonstrated an elevated loss of silence frequency relative to single mutants (21), despite sharing a similar parental histone deposition pattern with the WT (GSE240331). In both the previous and current studies, the mcm2-3A mutant exhibits a loss of silencing at a level similar to the pol1-2A2 mutant, despite displaying a substantially higher parental histone bias pattern (19). Notably, pol32Δ exhibited a marginal loss of silence frequency but a pronounced parental histone bias pattern compared to the pol1-2A2 mutant. These observations cannot be easily explained by a simple copy and paste model of histone modification. The minor phenotypic effect of parental histone transfer mutants is understandable, considering the notion that the parental histone transfer pathway is expected to have a profound effect on DNA sequence-independent inheritance, a phenomenon not yet detected in budding yeast. In this study, the identification of a key histone H3–H4 chaperone Pol32, in sequential with Mcm2, provides a promising target for further exploration of the mechanisms underlying epigenetic inheritance by histone chaperones.

Materials and Methods

Yeast Strains.

All yeast strains used in this study were W303-1 background except the histone H3-tag switch strains (Cyc565 and cyc1276), which were S228C background (SI Appendix, Table S1). Pol1 and Pol32 mutagenesis was performed using the CRISPR/Cas9 plasmid pML104 or pML107 along with dsDNA donor (46), while gene deletion and TAP (Tandem affinity purification)/myc-tag were knock-in using PCR-based methods (47). The oligonucleotides used in this study are listed in SI Appendix, Table S2.

eSPAN.

We performed eSPAN according to the protocol in our previous studies (16, 17) with some modifications. Briefly, yeast cells were synchronized with alpha-factor and released in yeast extract peptone dextrose (YPD) medium with BrdU (400 μg/mL) following the standard protocol (18). The cells were digested with MNase (New England Biolabs (NEB), Cat# M0247S), after saving 1% for BrdU-IP, the rest of digested chromatin underwent ChIP against H3K4me3 (Abcam, ab8580), H3K56ac (25), HA-Tag (Sigma, Cat#12CA5), or T7-Tag (BETHYL, Cat#A190-117A), individually. Then DNA extracted from the ChIP products was end-repaired and ligated with adapters using the VAHTS^® Universal DNA Library Prep Kit for Illumina V3 (Vazyme, Cat#ND607). After saving 1% for ChIP-seq analysis, the remaining products were subjected to BrdU-IP as previously reported (16), and the resulting products were amplified with NEBNext High-Fidelity 2× PCR Master Mix (NEB, Cat# M0541L) and indexing primers. The obtained libraries were sequenced using the paired-end method by the Illumina Nova 6000 platform.

Tandem Affinity Purification Assay for Pol1, Pol32, and Mcm2.

Yeast cells were cultured in liquid YPD medium at 30 °C until the OD₆₀₀ reached 1.8 to 2.0, and were harvested by centrifugation. The cell pellets were washed with cold water containing 10% glycerol and resuspended in an equal volume of TAP IP buffer (25 mM Tris pH 8.0, 100 mM NaCl, 10 mM MgCl₂,1 mM ethylenediaminetetraacetic acid, 0.01% NP-40, 1 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride, 1 mM Pefabloc, 1 mM Benzamidine, 15 KU/mL DNase I). Cell resuspensions were lysed using the Mini-Beadbeater-16 (Biospec) at cryogenic conditions with four cycles (30 s on, 1-min off for each). The cell lysate was centrifuged at 20,000 g for 10 min, and the soluble portion was incubated on ice for 30 min with 75 μg/mL ethidium bromide. After centrifugation at 20,000 g for 30 min, the supernatant was mixed with prewashed IgG beads (Cytiva, Cat# 17096901) and incubated at 4 °C for 2 h. The IgG beads were washed three times with TAP IP buffer and then boiled in the SDS sample buffer. Western blot analysis was performed with antibodies against protein A (Proteintech, Cat# 66945-1-Ig), H3 (Abcam, Cat#ab1791), H4 (Proteintech, Cat# 16047-1-AP), H3K56ac (25), H3K4me3 (CST, Cat#9751), and myc-Tag (CST, Cat#2276). The detailed antibody information is displayed in SI Appendix, Table S3.

Pol32-Flag Immunoprecipitation Assay.

Yeast cells carrying pRS423-Pol32-3×Flag or an empty vector pRS423 control were cultured in synthetic complete medium minus histidine at 30 °C until the OD₆₀₀ reached 1.8 to 2.0. The samples were lysed following the TAP purification assay steps and the supernatant was incubated with prewashed anti-Flag Magnetic Beads (MCE, cat#HY-K0207) at 4 °C for 2 h. After three washes, the beads were boiled in the SDS loading buffer and subjected to western blot analysis with antibodies against Flag (CST, Cat#14793), H3 (Abcam, Cat#ab1791), H4 (Proteintech Cat# 16047-1-AP), H3K56ac (25), H3K4me3 (CST, Cat#9751), and myc-Tag (CST, Cat#2276).

Purification of Recombinant Mcm2, Pol1, and Pol32.

Yeast Pol32 and codon-optimized Mcm2 were cloned into the pGEX4T-1 vector to construct GST (glutathioneS-transferase)-tagged Pol32 and Mcm2. The codon-optimized N-terminal of yeast Pol1 (1 to 527 AA) was cloned into the pET28b vector to produce HIS-tagged Pol1. To express these fusion proteins, the plasmids were transformed into the BL21 (DE3) Escherichia coli strain (TransGen Biotech, Cat#CD601-02). The transformed cells were cultured in Luria-Bertani medium (Sangon biotech, Cat#A507002-0250) supplemented with 50 μg/mL ampicillin (for pGEX4T-1 vector expression) or 50 μg/mL kanamycin (for pET28b vector expression). The cells were cultured at 37 °C with shaking until OD600 reached 0.8. Expression of these fusion proteins was induced by 0.3 mM IPTG (Isopropyl β-D-1-thiogalactopyranoside) (Sangon biotech, Cat#A600168-0100) at 16 °C for 16 h. Bacterial cells were pelleted by centrifugation at 4,000 rpm for 10 min, resuspended in a lysis buffer containing 20 mM Tris-HCl and 500 mM NaCl, and lysed by a high-pressure homogenizer (ATS company, AH-1500). After centrifugation, the supernatant of the samples was subjected to affinity purification. GST-tagged proteins were purified by Glutathione Resin (GenScript, Cat#L00206-100) and HIS-tagged protein by NI-Charged Resin (GenScript, Cat#L00666-100). In the end, GST-Pol32 and HIS-Pol1 were purified with a gel filtration column (Cytiva, HiLoad 16/600 Superdex 200 column). For GST-Mcm2, purification was accomplished through gel filtration column chromatography following the removal of the GST tag by TEV (Beyotime Biotechnology, Cat#P2310M).

Purification of Recombinant Histone H3.1–H4 Tetramer.

The human histones H3.1–H4 tetramers were purified as previously reported (48). The pETDuet-1 vector and BL21 (DE3) were used as the expression vector and host strain, respectively. When the cell culture reached an OD600 of 1.5, 1 mM IPTG was added, and the cells were cultured at 37 °C for an additional 4 h to induce protein expression. The cells were lysed in lysis buffer (1 M NaCl, 20 mM Tris-HCl pH 7.5) and centrifuged to obtain the supernatant. Then, a heparin column was used to capture the H3–H4 tetramer from the supernatant, and a gel filtration column was used for further purification.

GST Pulldown Assays to Detect Pol32–H3–H4, Pol32–Mcm2, and Pol32–Pol1 Interaction.

For the pulldown of GST-Pol32 with H3–H4 tetramers, Glutathione Resin (50 μL) was suspended in 200 µL of binding buffer with three different salt concentrations (300, 500, or 750 mM NaCl in 20 mM Tris, pH 7.5). And 1 nmol of GST-Pol32 was added and incubated at 4 °C for 2 h. Then, 1 nmol of H3–H4 tetramer was added and incubated for another 1 h. After incubation, the beads were washed four times with 1 mL of washing buffer (0.5% Triton X-100 in binding buffer). Finally, the sample was prepared by adding 50 μL of protein loading buffer and analyzed with sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel and Coomassie Blue staining (Kyr Biotechnology, Cat#18.001.500).

To test the effect of H3–H4 tetramers on the Pol32–Mcm2 interaction, Glutathione Resin (50 μL) was suspended in 200 µL of binding buffer (20 mM Tris, pH 7.5, and 500 mM NaCl) and 1 nmol of GST-Pol32 was added and incubated at 4 °C for 2 h. At the same time, Mcm2 was incubated with H3–H4 tetramers at 500 mM NaCl for 1 h to form the Mcm2-H3–H4 complex. Then, 1 nmol of Mcm2 or the Mcm2-H3–H4 complex was added to the Pol32-beads mixture and incubated for an additional 1 h. After incubation, the beads were washed four times with 1 mL of washing buffer (0.5% Triton X-100 in binding buffer). Finally, the sample was prepared by adding 50 μL protein loading buffer and analyzed with SDS-PAGE gel and Coomassie Blue staining.

To test the effect of H3–H4 tetramer on the Pol32–Pol1 interaction, Glutathione Resin (50 μL) was suspended in 200 µL of binding buffer (20 mM Tris, pH 7.5, and 300 mM NaCl), and 1 nmol of GST-Pol32 was added and incubated at 4 °C for 2 h. Then, 1 nmol of H3–H4 tetramers was added to the experimental sample or not, and was incubated for another 1 h. After washing three times with 1 mL of washing buffer, 1 nmol Pol1 was added to the Pol32–H3–H4-beads mixture or Pol32-beads mixture, respectively, and incubated for 1 h. Unbound proteins in the system were washed again using washing buffer. Finally, the sample was prepared by adding 50 μL loading buffer and analyzed with SDS-PAGE gel and Coomassie Blue staining.

To analyze the effect of DNA on the interaction between Pol32 and H3–H4, two complementary ssDNA oligonucleotides were synthesized by Rui Biotech (Beijing, China) and annealed to form a 58 bp dsDNA. The oligos are listed in SI Appendix, Table S2. Then Glutathione Resin beads (50 μL) were suspended in 200 µL of binding buffer (20 mM Tris, pH 7.5, and 300 mM NaCl), and 1 nmol of GST-Pol32 was added and incubated at 4 °C for 2 h. Then the 58 bp dsDNA with gradient concentration was added and incubated for another 1 h. After washing three times with 1 mL of washing buffer, 1 nmol H3–H4 was added to each DNA-Pol32 mixture respectively and the mixture was incubated for 1 h. Unbound H3H4 in the system were washed again using washing buffer. Finally, the sample was prepared by adding 50 μL loading buffer and analyzed with SDS-PAGE gel and Coomassie Blue staining.

Electrophoretic Mobility Shift Assay to detect Pol32–DNA Interaction.

GST-Pol32 was incubated with a 58 bp dsDNA in assembly buffer (20 mM Tris pH 7.5, 150 mM NaCl, and 1 mM DTT) for 30 min at room temperature. Then, the Pol32–DNA mixtures were separated on a 5% Polyacrylamide gel, prepared as previously reported at 40 V for 120 min at 4 °C (49). The gel was visualized using SYBR Safe DNA stain (Thermo Fisher Scientific, Cat#S33102).

Yeast Spot Assay.

Freshly cultured yeast cells were diluted to an optical density at 600 nm (A600) of 0.6. Subsequently, a 10-fold serial dilution was executed, and the cells were spotted onto standard growth medium with or without the specified concentration of the drug agent (YPD+20 or 50 μM HU or 4 or 10 μg/mL CPT). Images were captured at various time points, and representative images are presented.

Analysis of Silencing-Loss at the HML Locus.

To measure the effect of Pol32 deletion on chromatin state, we analyzed the silencing-loss rate at the HML locus in both wild-type and pol32Δ strains using the CRASH (Cre-reported altered states of heterochromatin) assay according to a previously published protocol (17). Briefly, 10 colonies of wild-type and pol32Δ strains were individually cultivated in YPD medium until the stationary phase. Subsequently, they were diluted to OD600 = 0.01 in YPD, and cultured for another 5 h at 30 °C. We collected 50,000 events for each colony using a BD cytometer. The silencing-loss rate at the HML locus was indicated with the ratio of GFP+/Total cells.

Data Analysis.

The sequence mapping and analysis for BrdU-IP-ssSeq and eSPAN were conducted following procedures similar to our previously published work (16, 17). Sequenced reads were mapped to the reference genome of Saccharomyces cerevisiae (sacCer3) with Bowtie2 software (50). Only paired-end reads correctly mapped on both sides were selected for further analysis. Following the reference genome, the Watson and Crick strand reads were separated using in-house Perl programs. BEDTools (51) and self-developed Perl programs were used to calculate the genome-wide read coverage.

In order to calculate the average bias of each nucleosome, the eSPAN reads at each nucleosome surrounding the 134 early DNA replication origins were individually counted and assigned to their respective nucleosome positions (52). Following normalization against the corresponding BrdU-IP-ssSeq, the average bias pattern of eSPAN was calculated as log₂ (Watson strand reads/ Crick strand reads) at each nucleosome position. Similarly, log₂ (Lagging strand reads/ Leading strand reads) at each nucleosome position of each of the 134 early replication origins was used to analyze the bias pattern of eSPAN peaks at each origin.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The deep sequencing data generated in this study have been deposited in the Gene Expression Omnibus (GEO) database under accession code GSE252049 (53).