Reconstituting human and SV40 replisomes with Okazaki fragment maturation reveals synchronized leading- and lagging-strand synthesis Authors: Amani Al-Amodi, Muhammad Tehseen, Vlad-Stefan Raducanu, Masateru Takahashi, Yujing Ouyang, Ammar U. Danazumi, Alfredo De Biasio, Samir M. Hamdan Published: 2026 ABSTRACT Eukaryotic lagging-strand synthesis requires iterative priming of Okazaki fragments (OFs) by Pol α, extension by Pol δ, and subsequent primer removal and ligation through Pol δ-mediated maturation of OFs (MOF). How the replisome coordinates these time-consuming lagging-strand processes while maintaining rapid and synchronized DNA synthesis remains unclear. Here, we reconstitute and compare the human and Simian Virus 40 (SV40) replisomes in the presence of MOF. In the human replisome, Pol α is recruited distributively to the CMG helicase for priming and transiently slows leading-strand synthesis by Pol ε, thereby directly coupling lagging-strand initiation to replisome progression. In contrast, Pol δ functions independently of CMG and remains highly distributive during both OF synthesis and MOF. Nevertheless, efficient recycling mechanisms of Pol δ maintain synchronization by prioritizing OF synthesis over MOF, resulting in prolonged accumulation of unligated OF intermediates without affecting leading-strand synthesis. In the SV40 replisome, where Pol δ directly interacts with the L-Tag helicase and synthesizes both DNA strands, leading-strand synthesis is similarly slowed by Pol α-dependent priming. However, MOF additionally slows leading-strand synthesis, revealing an additional layer of coordination required to maintain synchronization. These findings suggest a model in which the human replisome has evolved to simplify synchronization while maintaining replication speed and pave the way for a better understanding of its communication with other cellular processes. Introduction DNA replication is a fundamental yet inherently asymmetric process. The leading strand is synthesized continuously, while the lagging strand is replicated discontinuously as short Okazaki fragments (OFs) due to the strict 5′–3′ directionality of replicative DNA polymerases1,2 (Fig. 1A). This asymmetry demands precise coordination to ensure that leading-strand synthesis does not outpace lagging-strand synthesis, thereby preventing abnormal accumulation of single-stranded (ss) DNA. Although ssDNA is transiently stabilized by Replication Protein A (RPA), prolonged exposure increases the risk of replication stress, fork collapse, and genome instability3,4. Despite the successful reconstitution of replication in human, yeast, and Simian Virus 40 (SV40) systems5–10, the intrinsic mechanisms that synchronize leading- and lagging-strand synthesis remain unclear. Understanding this coordination is not only critical for defining normal replisome function but also for deciphering how the replication machinery responds to obstacles such as DNA lesions, active transcription, and chromatin remodeling11,12.Fig. 1Pol α priming slows human leading-strand synthesis.A Schematic representation of the human replisome (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). The Cdc45-Mcm2-7-GINS helicase (CMG complex) unwinds the parental DNA, providing single-stranded templates. DNA polymerase ε (Pol ε) synthesizes the leading-strand continuously, while Pol α-primase (Pol α) synthesizes RNA-DNA primers on the lagging strand, which are extended by Pol δ to produce Okazaki fragments. PCNA enhances the processivity of Pol δ. Replication protein A (RPA) coats transiently exposed ssDNA on the lagging strand. Maturation of Okazaki Fragments (MOF) involves FEN1, which cleaves the RNA flap generated by Pol δ strand- displacement, and DNA ligase 1 (Lig1), which seals the resulting nick. B Schematic illustration of the reaction scheme for human replication assays (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). Full details are provided in the Methods section. C Denaturing agarose gel analysis of a time-course experiment performed as in (B). Reactions were carried out in the absence of Pol α and Pol δ (lanes 1–6), with Pol α alone (lanes 7–12), and with Pol α, Pol δ, and RFC (lanes 13–18). D Quantification of median DNA length versus reaction time from panel (C). Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 8.615 kb. Additional fit parameters are presented in Supplementary Table 2. E Quantification of total DNA synthesis from Supplementary Fig. 7 Sigmoidal fit by Eq. (1) is applied to the mean of two independent experiments. Source data are provided as a Source Data file. Midpoint synthesis rates are calculated as derivatives of the model at the time point t1/2 at which 50% of the total synthesis is achieved. All intensities are normalized to the intensity of the 3 kb ladder band from the respective gel. Additional fit parameters are presented in Supplementary Table 2. Pol α priming slows human leading-strand synthesis. A Schematic representation of the human replisome (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). The Cdc45-Mcm2-7-GINS helicase (CMG complex) unwinds the parental DNA, providing single-stranded templates. DNA polymerase ε (Pol ε) synthesizes the leading-strand continuously, while Pol α-primase (Pol α) synthesizes RNA-DNA primers on the lagging strand, which are extended by Pol δ to produce Okazaki fragments. PCNA enhances the processivity of Pol δ. Replication protein A (RPA) coats transiently exposed ssDNA on the lagging strand. Maturation of Okazaki Fragments (MOF) involves FEN1, which cleaves the RNA flap generated by Pol δ strand- displacement, and DNA ligase 1 (Lig1), which seals the resulting nick. B Schematic illustration of the reaction scheme for human replication assays (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). Full details are provided in the Methods section. C Denaturing agarose gel analysis of a time-course experiment performed as in (B). Reactions were carried out in the absence of Pol α and Pol δ (lanes 1–6), with Pol α alone (lanes 7–12), and with Pol α, Pol δ, and RFC (lanes 13–18). D Quantification of median DNA length versus reaction time from panel (C). Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 8.615 kb. Additional fit parameters are presented in Supplementary Table 2. E Quantification of total DNA synthesis from Supplementary Fig. 7 Sigmoidal fit by Eq. (1) is applied to the mean of two independent experiments. Source data are provided as a Source Data file. Midpoint synthesis rates are calculated as derivatives of the model at the time point t1/2 at which 50% of the total synthesis is achieved. All intensities are normalized to the intensity of the 3 kb ladder band from the respective gel. Additional fit parameters are presented in Supplementary Table 2. At the eukaryotic replication fork, the Cdc45-MCM2-7-GINS helicase complex (CMG) unwinds parental dsDNA13,14, enabling three essential DNA polymerases to drive replication: Polymerase ε (Pol ε) continuously synthesizes the leading strand, while Polymerase α-primase (Pol α) generates RNA–DNA primers for the lagging strand, which Polymerase δ (Pol δ) then extends15,16 (Fig. 1A). Unlike Pol ε and Pol α, which are associated with CMG17–19, Pol δ lacks a direct interaction with the helicase20. OF synthesis begins with the recruitment of Pol α to the replisome, where it synthesizes an RNA–DNA hybrid primer (~ 20–30 nucleotides)21,22. Pol δ then extends the primer, requiring the processivity clamp Proliferating Cell Nuclear Antigen (PCNA), which is loaded by Replication Factor C (RFC)20. Pol ε also depends on PCNA for processivity but uniquely requires the RFC variant CTF18-RFC, ensuring its efficient loading at the replication fork23,24. The discontinuous nature of lagging-strand synthesis imposes a major challenge to the replisome due to the need for frequent Pol α priming and the multi-step maturation of OFs (MOF), which requires gap filling, primer removal, and ligation25 (Fig. 1A). MOF is initiated when Pol δ displaces the RNA primer upon encountering the previous fragment, generating a 5′-flap structure that is cleaved by Flap Endonuclease 1 (FEN1) in a reaction termed nick translation (NT)25. PCNA then recruits DNA Ligase 1 (Lig1) to seal the nick, ensuring strand continuity25,26. Although MOF has been incorporated into a reconstituted yeast and human replisomes27,28, how Pol δ coordinates OF synthesis and MOF to ensure synchronization with leading-strand synthesis and the dynamics underlying this coordination remain poorly understood, especially given the distinct behaviors of human and yeast Pol δ25,29. Furthermore, primer synthesis is a key rate-limiting step in DNA replication across all domains of life, constrained by the kinetics of NTP binding and phosphodiester bond formation30. While helicase–primase interactions enhance this activity, their efficiency varies between systems19,31–34. Cryo-EM structures of the human and yeast CMG–Pol α complexes provide key insights into how these interactions facilitate primer synthesis19. However, it remains unclear whether this coupling influences leading-strand synthesis and its synchronization with Pol δ activities during OF synthesis and MOF. In the human system, a particular challenge in lagging-strand synthesis arises from the limited processivity of the Pol δ–PCNA complex, which extends only 0.4 kilonucleotides (knt)35 compared to several knt in yeast36. This is attributed to fewer contacts between human Pol δ and PCNA, resulting in ~ 50-fold lower affinity of the human Pol δ–PCNA complex for DNA compared to yeast37–39. Furthermore, single-molecule experiments in a reconstituted yeast replisome established that although Pol δ can undergo concentration-dependent exchange, the exchange kinetics are relatively slow, with an average exchange rate of one Pol δ molecule every 16 min, allowing it to contribute to the synthesis of ~ 140 OFs40. In contrast, in humans, Pol δ–PCNA may dissociate before completing a single OF36, suggesting that multiple Pol δ molecules may contribute to the synthesis of each OF. Moreover, the reduced affinity of human Pol δ for PCNA compromises its ability to perform iterative cycles of NT with FEN1 during RNA primer removal25, whereas FEN1 enhances NT by Pol δ in yeast29. As a result, MOF in humans is ~ 55-fold slower than in yeast25,29. How the human replisome adapts to these constraints on Pol δ processivity to sustain continuous fork progression remains an open question. It is possible that protein–protein interactions within the human replisome stabilize Pol δ and promote its retention on DNA during OF synthesis and/or MOF. Supporting this idea, in the SV40 system, human Pol δ directly interacts with the viral large T-antigen (L-Tag) helicase, increasing its lifetime on DNA from ~2 sec to ~ 3 min35. In addition, the Pol32 subunit of yeast Pol δ (homologous to p66 in human) has been shown to stabilize its interaction within the replisome, likely through Pol α40,41. To investigate how eukaryotic replisomes coordinate leading- and lagging-strand synthesis in the context of MOF, we reconstituted and compared two complementary systems: the human and SV40 DNA replication machineries supporting MOF. The human replisome captures the full complexity of eukaryotic replication, utilizing distinct polymerases for leading- (Pol ε) and lagging-strand (Pol δ) synthesis. In contrast, the SV40 system provides a simplified model in which a single polymerase, Pol δ, synthesizes both strands in conjunction with the viral large T-antigen (L-Tag) helicase42–44. Notably, SV40 more closely resembles prokaryotic replication, where the lagging-strand polymerase likely interacts directly with the helicase35. This fundamental difference between the two systems provides a unique opportunity to dissect the intrinsic mechanisms that ensure lagging-strand synthesis remains coordinated with the replisome in eukaryotes. Results Reconstitution of the human replisome in vitro To reconstitute the human replisome from purified proteins (Fig. 1A), we assembled the core components, including the 11-subunit CMG helicase, the three DNA polymerases (Pol ε, Pol δ, and Pol α), the processivity clamp PCNA, its loaders (CTF18-RFC and RFC), RPA, and accessory proteins such as the Fork Protection Complex (FPC) and AND-1 (Supplementary Fig. 1), and followed similar reaction conditions and substrate design to those described previously (Fig. 1B)8. Briefly, CMG was first loaded onto an ~ 8.6 kb forked DNA template containing a primer for leading-strand replication. Pol ε was then assembled with the necessary components in the presence of only dATP and dCTP, allowing the replisome to be held at the 3′ end of the primer. Following a 5 min incubation, replication was initiated by adding RPA, ATP, rNTPs, dTTP, dGTP and [³²P]-dCTP. Analysis of the reaction products by alkaline gel electrophoresis revealed two distinct DNA species: long labeled fragments corresponding to leading-strand products and shorter fragments corresponding to OFs on the lagging strand (Fig. 1C). We performed several control experiments to ensure compatibility between our reconstituted reactions and those described in the recently reconstituted human replisome8. Drop-out reactions confirmed activity of FPC and AND-1 components in leading-strand synthesis (Supplementary Fig. 2). Control reactions confirmed that CMG loading was essential for replication, as no synthesis was detected when CMG was omitted (Supplementary Fig. 3A, lane 20). In addition, we confirmed that in a minimal replisome lacking AND-1 and FPC, Pol α required direct interaction with CMG to initiate primer synthesis, as no priming activity was detected in the absence of CMG (Supplementary Fig. 4, lane 1), whereas priming and extension were observed when CMG was present (Supplementary Fig. 4, lanes 3 and 4). Finally, replacement of CMG with the strand-displacing ϕ29 DNA polymerase failed to support priming on the lagging strand (Supplementary Fig. 4, lane 5 and 6), consistent with prior observations in the yeast system10. Previous experiments using both yeast and human reconstituted replisomes showed that increasing Pol α concentrations led to an increase in priming frequency and a decrease in OF size8,10, suggesting that Pol α priming is distributive, with more frequent priming events at higher enzyme concentrations. Consistent with these results, we report similar correlations among Pol α concentrations, priming frequency and OF size in our reconstituted human replisome (Supplementary Fig. 5). Collectively, these experiments support a unifying model in which Pol α is recruited to CMG through direct interactions, in agreement with the Cryo-EM structures of the human and yeast CMG–Pol α complexes19, and with biochemical evidence of CMG–Pol α co-migration in human8. Pol α priming slows leading-strand synthesis in the human replisome To examine coordination between leading- and lagging-strand syntheses, we systematically added lagging-strand components to a reconstituted leading-strand synthesis system. In the absence of Pol α and lagging-strand components, leading-strand synthesis proceeded at a median rate of ~ 1.8 kb min⁻¹ (Fig. 1D). This median rate reflects an average value for the entire population of replisome replication speeds in our reconstituted human system, whereas the ~ 4.4 kb min⁻¹ rate reported previously8 represents only the fastest-moving replisomes. Strikingly, the addition of Pol α reduced the leading-strand median synthesis rate to ~ 1.4 kb min⁻¹ (Fig. 1C, lanes 7–12; Fig. 1D), mirroring average replication fork velocities measured in human cells under physiological conditions45–49. This reduction suggests that Pol α priming slows leading-strand elongation. Control experiments confirmed that the slow down depended on primer synthesis, as omitting rNTPs (rC/G/U) abolished the effect in a statistically significant manner (Supplementary Fig. 3C), resulting in a leading-strand median synthesis rate of ~ 1.9 kb min⁻¹ (Supplementary Fig. 3B), similar to that observed in the absence of Pol α (Fig. 1D). Also, a side-by-side comparison of reactions containing Pol α without rNTPs to those lacking Pol α entirely showed no difference in the leading-strand synthesis profile, ruling out any inhibitory effect from Pol α alone (Supplementary Fig. 3A, compare lanes 4 and 21). Next, we tested whether primer handoff from Pol α to RFC–PCNA influenced the replication dynamics. RFC-mediated loading of PCNA on the lagging strand did not affect the leading-strand synthesis profile, indicating that primer transfer does not contribute to fork-speed regulation (Supplementary Fig. 6). Similarly, the addition of Pol δ to initiate OF synthesis (Fig. 1C, lanes 13–18) resulted in robust lagging-strand synthesis without further slowing the leading strand (Fig. 1D). To quantitatively assess the synchronization between leading- and lagging-strand synthesis, we measured the integrated intensity of synthesized DNA products over time (Fig. 1E and Supplementary Fig. 7). Both strands reached similar plateau levels, showing nearly identical nucleotide incorporation efficiency and only very limited strand displacement (SD) activity by Pol δ on the lagging strand. In the pre-plateau phase, lagging-strand synthesis followed closely behind leading-strand synthesis with a delay of less than 2 min at the synthesis midpoint (~ 50% total DNA synthesis), supporting the idea that replication fork progression is intrinsically synchronized. In addition, at their synthesis midpoints, the instantaneous synthesis rates of the two strands were found to be within ~ 1.2-fold of each other (Fig. 1E). This finding aligns with studies in yeast, where a minimal replisome (CMG, Pol ε, Pol δ, Pol α, RFC, PCNA, and RPA) maintained similar synthesis rates for both strands10. While our system includes the fully reconstituted human replisome, both studies consistently support the model of synchronized leading- and lagging-strand synthesis. Collectively, these results demonstrate that Pol α’s interaction with CMG modulates leading-strand synthesis while all subsequent steps in OF synthesis, including primer handoff to PCNA and lagging-strand elongation by Pol δ, have no additional impact on leading-strand synthesis. Consequently, Pol δ must be able to rapidly extend the primer and recycle to compensate for the slow primer synthesis, keeping pace with leading-strand synthesis. Pol δ is distributive and uncoupled during OF synthesis within the human replisome Next, we investigated the processivity of Pol δ during lagging-strand synthesis within the human replisome. The distinct size distribution of leading- and lagging-strand products in our system provided a direct readout for Pol δ’s processivity. To assess polymerase retention on the DNA, we initiated replication for 1 min with pre-assembled wild-type Pol δ (Pol δWT) and subsequently introduced either a 1- or 2-fold excess of a polymerase-deficient mutant (Pol δᴰᵉᵃᵈ) to compete with dissociating Pol δWT (Fig. 2A). As a control, equivalent amounts of Pol δWT were added to parallel reactions (Fig. 2B).Fig. 2Pol δ is distributive within the human replisome.A Schematic representation of the assay (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). B Denaturing agarose gel showing replication reactions performed with either wild-type Pol δ (Pol δWT) or catalytically inactive Pol δ (Pol δDead) at 1:1 or 1:2 concentrations. C Quantification of median DNA length versus reaction time from panel (B). Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 8.615 kb. Additional fit parameters are presented in Supplementary Table 2. D Microscale thermophoresis (MST) analysis of interactions between CMG and Pol δ compared with Pol ε. Data are presented as mean, and the error bars represent the SEM from three independent experiments. Source data are provided as a Source Data file. Pol δ is distributive within the human replisome. A Schematic representation of the assay (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). B Denaturing agarose gel showing replication reactions performed with either wild-type Pol δ (Pol δWT) or catalytically inactive Pol δ (Pol δDead) at 1:1 or 1:2 concentrations. C Quantification of median DNA length versus reaction time from panel (B). Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 8.615 kb. Additional fit parameters are presented in Supplementary Table 2. D Microscale thermophoresis (MST) analysis of interactions between CMG and Pol δ compared with Pol ε. Data are presented as mean, and the error bars represent the SEM from three independent experiments. Source data are provided as a Source Data file. The leading-strand synthesis rate remained unaffected by Pol δᴰᵉᵃᵈ competition, further supporting the conclusion that primer handoff from Pol α to Pol δ and primer extension by Pol δ do not impact leading-strand progression (Fig. 2C). In other words, leading-strand synthesis is intrinsically uncoupled from Pol δ’s activities during OF synthesis within the replisome. Microscale thermophoresis (MST) binding assays further supported this interpretation, revealing that human Pol δ does not stably associate with CMG, whereas Pol ε binds CMG with a dissociation constant (Kᴅ) of 30.5 nM (Fig. 2D). Previous findings in yeast show that leading- and lagging-strand synthesis can be uncoupled under specific conditions, including lesion bypass, protein barriers, and nucleotide depletion50–52. Our results extend these findings by illustrating that OF synthesis by Pol δ is intrinsically uncoupled from leading-strand synthesis within the replisome under normal conditions. The length and abundance of OFs progressively decreased with increasing Pol δᴰᵉᵃᵈ concentration, and this reduction was observed within 1.5 min of adding Pol δᴰᵉᵃᵈ (Fig. 2B, compare lanes 13–18 with 19–24). This reduction strongly suggests that Pol δWT frequently dissociates and reassociates from the primer-template strand during lagging-strand synthesis, consistent with its intrinsic distributive mode of action36,53. There are two possible mechanisms that could account for the observed inhibition by the catalytically inactive Pol δᴰᵉᵃᵈ: (1) complete dissociation of the Pol δWT from the primer-template junction followed by rebinding from solution, or (2) concentration-driven exchange in which the replisome-associated Pol δ is replaced by Pol δᴰᵉᵃᵈ. Our assay cannot formally distinguish between these two scenarios, as both would result in similar inhibition profiles upon addition of Pol δᴰᵉᵃᵈ. We therefore cannot exclude the possibility that a fraction of Pol δ remains bound to the replisome via CMG-independent protein contacts, while the exchange occurs at the level of primer-template engagement. However, if Pol δ were retained by the replisome upon dissociation from the primer-template, its high local concentration would be expected to outcompete Pol δᴰᵉᵃᵈ from solution for rebinding to the primer-template or the newly synthesized primer by Pol α. In contrast, single-molecule Pol δ exchange experiments in yeast show slow exchange kinetics within the replisome of 1 Pol δ per 16 min40, which indicates a more stable replisome-associated Pol δ. These findings align with previous findings that showed human Pol δ is inherently distributive compared to yeast Pol δ during both primer extension35,39,53,54 and MOF25,29, likely due to its lower affinity for PCNA37–39. Despite its distributive nature, human Pol δ was able to efficiently support robust OF synthesis and keep pace with leading-strand synthesis, demonstrating that Pol δ undergoes rapid and efficient recycling during OF synthesis. MOF is slow and uncoupled from the human replisome Previous results showed that the reconstituted human MOF reaction is slow25,55, raising the questions of whether Pol δ would acquire CMG-independent interaction within the replisome and if such interaction would enhance the rate of MOF and influence leading-strand synthesis. To address these questions, we systematically reconstituted the human replisome in the presence of MOF factors. The bimodal distribution of the lengths (longer leading-strand fragments and shorter OFs) of nascent strands observed in the previous experiments (Fig. 1C) reflects incomplete OF maturation due to the absence of processing factors. MOF involves RNA primer removal from the 5′ end of nascent OFs via NT by the Pol δ–PCNA–FEN1 complex to create ligatable DNA-DNA nicks, since human Lig1 cannot ligate RNA-DNA nicks25,26. A previous study identified primer removal via NT as the rate-limiting step in human MOF25. To test whether NT impacts leading-strand synthesis, we performed a time-course experiment in which FEN1 was included to support NT, but Lig1 was omitted to prevent nick sealing (Fig. 3A). Under these conditions, leading-strand synthesis reached its midpoint vicinity (~ 50% total DNA synthesis) after ~ 6 min (Fig. 3B) with a mean synthesis rate of ~ 1.3 kb min⁻¹ (Supplementary Fig. 8), similar to rates observed in the absence of NT (Fig. 1D). Furthermore, lagging-strand product lengths remained unchanged over time (Supplementary Fig. 9A), indicating that NT is not processive in human.Fig. 3MOF is slow and uncoupled from the human replisome.A Denaturing alkaline agarose gel time-course of replication reactions in the presence of NT (FEN1 and absence of Lig1). B Quantification of the time-dependent formation of the full-length product band from panel A. Sigmoidal fit by Eq. (3) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. The mean leading-strand processing time (~ 50% full-length DNA synthesis) is calculated as described in the Methods section from the fit parameters. All fit parameters are presented within the panel. C Schematic representation of the strand-specific nicking reaction using Nb.BbvCI nicking enzyme (left) (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp) and denaturing alkaline agarose gel analysis of replication products following nicking (right). Reactions were incubated for 15 min before nicking enzyme treatment. The experiment was independently repeated three times with similar results, and a representative gel is shown. D Denaturing alkaline agarose gel time-course of replication reactions in the presence of MOF components (FEN1 and Lig1) following nicking enzyme treatment. E Quantification of the time-dependent formation of the nicked and intact product bands from panel D. The sum of the two nicked product bands was attributed to full-length leading-strand product, and the intact top band was attributed to completely ligated lagging-strand product via MOF. Both curves represent sigmoidal fits by Eq. (3) applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. The mean leading- and lagging-strand processing times (50% full-length DNA synthesis) are calculated as described in the Methods section from the fit parameters. All fit parameters are presented within the panel. F Quantification of MOF yield from panel (D). Ligation efficiency is calculated as the percentage of full-length product relative to the total lagging-strand signal (ligated and unligated). The data represent the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. MOF is slow and uncoupled from the human replisome. A Denaturing alkaline agarose gel time-course of replication reactions in the presence of NT (FEN1 and absence of Lig1). B Quantification of the time-dependent formation of the full-length product band from panel A. Sigmoidal fit by Eq. (3) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. The mean leading-strand processing time (~ 50% full-length DNA synthesis) is calculated as described in the Methods section from the fit parameters. All fit parameters are presented within the panel. C Schematic representation of the strand-specific nicking reaction using Nb.BbvCI nicking enzyme (left) (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp) and denaturing alkaline agarose gel analysis of replication products following nicking (right). Reactions were incubated for 15 min before nicking enzyme treatment. The experiment was independently repeated three times with similar results, and a representative gel is shown. D Denaturing alkaline agarose gel time-course of replication reactions in the presence of MOF components (FEN1 and Lig1) following nicking enzyme treatment. E Quantification of the time-dependent formation of the nicked and intact product bands from panel D. The sum of the two nicked product bands was attributed to full-length leading-strand product, and the intact top band was attributed to completely ligated lagging-strand product via MOF. Both curves represent sigmoidal fits by Eq. (3) applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. The mean leading- and lagging-strand processing times (50% full-length DNA synthesis) are calculated as described in the Methods section from the fit parameters. All fit parameters are presented within the panel. F Quantification of MOF yield from panel (D). Ligation efficiency is calculated as the percentage of full-length product relative to the total lagging-strand signal (ligated and unligated). The data represent the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. To reconstitute the maturation step, we added FEN1 and Lig1 to the replication reaction and monitored lagging-strand products over time (Supplementary Fig. 9B). This caused a progressive shift from short OFs to longer products and a steady increase in the full-length ~ 8.6 kb product, indicating ligation of OF intermediates into mature lagging strands. Control experiments showed that neither FEN1 nor Lig1 alone could support the formation of the full-length product, consistent with active MOF requiring both factors (Supplementary Fig. 10A–C). In the presence of MOF factors, accumulation of the full-length product visibly continued at much longer time points (Supplementary Fig. 9B, D), compared to NT alone (Fig. 3B). To further validate that the additional full-length product reflects a completely ligated lagging strand, we employed a strand-specific nicking strategy that selectively cleaves the labeled leading strand at defined positions while leaving the labeled lagging strand intact (Fig. 3C). The nicking strategy resulted in complete conversion of the full-length leading-strand product into the expected fragments, confirming efficient and specific cleavage (Fig. 3C). In reactions containing lagging-strand synthesis components, OFs remained evident, whereas upon addition of MOF factors (FEN1 and Lig1), a full-length product persisted despite nicking, consistent with the formation of a continuous, ligated lagging strand (Fig. 3C). A time-course analysis of the nicked reactions in the presence of MOF (Fig. 3D) showed a progressive increase in the full-length ligated product, with a mean processing time (~50% full-length DNA synthesis) of ~ 14.5 min (Fig. 3E). Notably, the overall MOF efficiency reached ~ 60% (Fig. 3F), consistent with previous reconstitutions of isolated human and yeast MOF and NT25,29,55–57. Interestingly, the mean processing time (~ 50% full-length DNA synthesis) for leading-strand synthesis, determined from the nicking reaction, was ~ 6.5 min (Fig. 3E). This further supports our conclusion that MOF does not affect leading-strand synthesis and remains uncoupled from the replisome. This conclusion is further supported by the similarity between the limited yield and slow processing time observed in isolated human MOF reconstitutions (< 70% yield, ~ 3 min/OF25,55) and the values obtained here in the presence of the complete human replisome, suggesting that replisome interactions do not substantially enhance MOF activity. In parallel experiments, we quantitatively compared NT-only (Fig. 3A) and MOF-containing replisomes (Supplementary Fig. 9B) by fitting the time-dependent accumulation of the full-length product (~8.6 kb) to an analytical kinetic model. NT data were well-fitted by a single incomplete gamma function (Fig. 3B), yielding a mean processing time (~ 50% full-length DNA synthesis) of ~ 6 min and a leading-strand median synthesis rate of ~ 1.4 kb min⁻¹, consistent with median product-length analysis (Supplementary Fig. 8). In contrast, MOF data required a bimodal fit with two incomplete gamma functions (Supplementary Fig. 9D), strongly supported by BIC analysis (> 10-unit decrease compared with the single-component model; Supplementary Fig. 9C, D). In the presence of MOF factors, a second, slower component emerged with a mean processing time of ~ 20 min. Since leading- and lagging-strand synthesis differ by < 2 min (Fig. 1E), this > 2-fold slower component likely reflects a slow MOF process. These analyses are in strong agreement with those obtained from the aforementioned nicking-reaction experiments. Collectively, these observations indicate that MOF remains a slow process within the replisome, does not affect leading-strand synthesis, and that Pol δ is unprocessive and uncoupled during MOF. Nonetheless, Pol δ can rapidly synthesize OFs even in the presence of MOF, indicating its ability to recycle efficiently to prioritize OF synthesis over MOF, providing a mechanism that enables it to keep pace with leading-strand synthesis. Pol α priming and MOF slow SV40 leading-strand synthesis To determine whether leading-strand synthesis slowdown during primer synthesis, uncoupled lagging-strand synthesis, and slow MOF are conserved features of eukaryotic replication, we examined the fully reconstituted SV40 replisome (Supplementary Fig. 1). In the SV40 replisome, as in its human counterpart, primer synthesis is carried out by Pol α recruited to the L-Tag helicase58. However, unlike the human system, the SV40 replisome relies solely on Pol δ for both leading- and lagging-strand synthesis (Fig. 4A). The direct interaction between Pol δ and L-Tag35 suggests that Pol δ remains associated with the replisome during replication, similar to polymerases in prokaryotic DNA replication.Fig. 4Pol α priming and MOF slow SV40 leading-strand synthesis.A Schematic representation of the SV40 replisome (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). Unlike the human replication fork, the large T-antigen (L-Tag) unwinds the parental DNA strands, and both leading and lagging strands are synthesized by Pol δ. B Schematic illustration of the reaction scheme for SV40 replication assays (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). The nucleotide bias of the forked DNA allows selective labeling of either the leading strand using [32P]-dCTP or the lagging strand using [32P]-dGTP. Full details are provided in the Methods section. C Denaturing agarose gel analysis of a time-course experiment performed as in (B). Reactions were carried out in the absence of Pol α and MOF (lanes 1–7), with Pol α alone (lanes 8–14), and with Pol α and MOF components (FEN1 and Lig1) (lanes 15–21). D Quantification of median DNA length versus reaction time from panel (C). Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 3.375 kb. Additional fit parameters are presented in Supplementary Table 2. E. Quantification of total DNA synthesis from Supplementary Fig. 14. Sigmoidal fit by Eq. (1) is applied to the mean of two independent experiments. Source data are provided as a Source Data file. Midpoint synthesis rates are calculated as derivatives of the model at the time point t1/2 at which 50% of the total synthesis is achieved. Leading- and lagging-strand intensities are integrated across their whole corresponding lanes. All intensities are normalized to the intensity of the 3 kb ladder band from the respective gel. Additional fit parameters are presented in Supplementary Table 2. Pol α priming and MOF slow SV40 leading-strand synthesis. A Schematic representation of the SV40 replisome (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). Unlike the human replication fork, the large T-antigen (L-Tag) unwinds the parental DNA strands, and both leading and lagging strands are synthesized by Pol δ. B Schematic illustration of the reaction scheme for SV40 replication assays (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). The nucleotide bias of the forked DNA allows selective labeling of either the leading strand using [32P]-dCTP or the lagging strand using [32P]-dGTP. Full details are provided in the Methods section. C Denaturing agarose gel analysis of a time-course experiment performed as in (B). Reactions were carried out in the absence of Pol α and MOF (lanes 1–7), with Pol α alone (lanes 8–14), and with Pol α and MOF components (FEN1 and Lig1) (lanes 15–21). D Quantification of median DNA length versus reaction time from panel (C). Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 3.375 kb. Additional fit parameters are presented in Supplementary Table 2. E. Quantification of total DNA synthesis from Supplementary Fig. 14. Sigmoidal fit by Eq. (1) is applied to the mean of two independent experiments. Source data are provided as a Source Data file. Midpoint synthesis rates are calculated as derivatives of the model at the time point t1/2 at which 50% of the total synthesis is achieved. Leading- and lagging-strand intensities are integrated across their whole corresponding lanes. All intensities are normalized to the intensity of the 3 kb ladder band from the respective gel. Additional fit parameters are presented in Supplementary Table 2. To track the replication dynamics, we used a 3.3 kb synthetic DNA template with a nucleotide-biased sequence10: one strand lacked dG, whereas the other lacked dC, allowing independent detection of leading-strand [³²P]-dCTP and lagging-strand [³²P]-dGTP synthesis (Fig. 4B). Attempts to generate a longer biased substrate for human replication were unsuccessful, but the use of separate polymerases in the human system still enabled direct synthesis coordination analysis across both systems. Replication was initiated by staging the reaction, first loading L-Tag, FPC, and Mcm10, followed by Pol δ, PCNA, RFC, RPA, Pol α, dNTPs, rNTPs, and ATP (Fig. 4B). FPC and Mcm10 were included based on recent single-molecule and bulk biochemical studies showing that these factors enhance fork progression in the SV40 system35. Time-course analysis confirmed the formation of both strand products (Supplementary Fig. 11A, B). Component drop-out experiments (Supplementary Fig. 11A) validated Pol δ’s strict dependence on ATP/rNTPs, L-Tag, PCNA, and RFC, consistent with its known PCNA requirement on primed ssDNA substrates59,60. Restriction enzyme digestion verified distributed replication across the template (Supplementary Fig. 12A), while primer synthesis by Pol α required L-Tag, as substitution with ϕ29 DNA Polymerase could not support Pol α priming (Supplementary Fig. 12B). This result mirrors our observations in the human system and is consistent with previous findings in yeast10. As in the human replisome, Pol α slowed SV40 leading-strand synthesis (Fig. 4C), decreasing the median rate from ~ 1.0 kb min⁻¹ to ~ 0.7 kb min⁻¹ (Fig. 4D). The rate in the absence of Pol α matches previous single-molecule SV40 measurements35. To determine whether this slowdown was due to primer synthesis by Pol α or competition with Pol δ for the L-Tag, we performed reactions with Pol α in the presence or absence of rNTPs (rC/G/U) across different time points (Supplementary Fig. 13A). Blocking priming by omitting rNTPs restored leading-strand synthesis to the faster median rate of ~ 1.0 kb min⁻¹ (Supplementary Fig. 13B), indicating that primer synthesis causes the slowdown. Second, we compared reactions containing Pol α without rNTPs to those completely lacking Pol α (as in Fig. 4C) at the 3 min time point. The replication profiles were indistinguishable (Supplementary Fig. 13C), confirming that Pol α alone does not inhibit synthesis in the absence of priming. Together, these results rule out polymerase competition and support a model in which primer synthesis by Pol α slows leading-strand progression in the SV40 system. To test whether leading- and lagging-strand synthesis remain synchronized, we quantified the integrated intensity of synthesized DNA products over time. Synthesis curves were correlated even more tightly than in the human system, with a midpoint difference (~ 50% total DNA synthesis) of only ~ 1 min, with similar midpoint rates and very similar upper plateau levels (Fig. 4E and Supplementary Fig. 14), indicating synchronized replication synthesis in the presence of Pol α. To examine the interplay between MOF and leading-strand synthesis, we assembled the SV40 replisome in the presence of FEN1 and Lig1 (Fig. 4C, lanes 15–21). Remarkably, MOF further slowed replication, reducing the leading-strand synthesis median rate from ~ 0.7 kb min⁻¹ to ~ 0.3 kb min⁻¹ (Fig. 4D). Control experiments confirmed that neither FEN1 nor Lig1 affected the leading-strand synthesis rate in the absence of Pol α at the concentrations used in our assays (15 nM FEN1 and 20 nM Lig1) (Supplementary Fig. 15A), ruling out competitive binding to PCNA as the cause of the slowdown. In addition, an M13-primer extension assay showed that FEN1 did not impair Pol δ’s primer extension activity, ruling out general inhibition through either catalytic interference or PCNA competition (Supplementary Fig. 16). To investigate the cause of the leading-strand slowdown during MOF, we enabled NT but omitted Lig1. The results showed that NT alone reduced the leading-strand synthesis rate to the same extent as full MOF (Fig. 5A and Supplementary Fig. 15B). Next, we tracked lagging-strand synthesis over time using [³²P]-dGTP incorporation in the presence of MOF factors (Fig. 5B). The median DNA length of lagging-strand products progressively increased, reaching the synthesis midpoint (half-length DNA products) within ~ 13 min, whereas leading-strand synthesis reached the midpoint (half-length DNA products) within ~ 7 min (Fig. 5B–D). Notably, MOF in the SV40 system reached ~ 60% efficiency (Supplementary Fig. 15C), comparable to that observed in the human system despite the additional interactions between the lagging strand Pol δ and the L-Tag helicase. This suggests that Pol δ, upon completing OF synthesis, may perform part of the NT reaction within the replisome, slowing leading-strand synthesis, while the remaining NT and ligation steps proceed uncoupled from replication.Fig. 5SV40 MOF reaction is slow and impacts leading-strand synthesis.A Quantification of median DNA length versus reaction time from Supplementary Fig. 15B. Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 3.375 kb. Additional fit parameters are presented in Supplementary Table 2. B Denaturing agarose gel analysis of a time-course experiment monitoring lagging-strand synthesis in the presence of MOF factors (FEN1 and Lig1) using [32P]-dGTP. C Denaturing agarose gel analysis of a time-course experiment monitoring leading-strand synthesis in the presence of MOF factors (FEN1 and Lig1) using [32P]-dCTP. The reactions were identical to (B), except the labeled nucleotide tracks the leading strand. D Quantification of median DNA length versus reaction time from panels (B, C). Sigmoidal fit by Eq. (1) is applied to the mean of two independent experiments. Source data are provided as a Source Data file. Median synthesis rates are calculated by Eq. (2). The amplitude parameters were fixed to 3.375 kb and 3.344 kb for leading-strand and lagging-strand synthesis, respectively. Additional fit parameters are presented in Supplementary Table 2. SV40 MOF reaction is slow and impacts leading-strand synthesis. A Quantification of median DNA length versus reaction time from Supplementary Fig. 15B. Sigmoidal fit by Eq. (1) is applied to the mean of three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Median synthesis rates are calculated by Eq. (2). The amplitude parameter was fixed to 3.375 kb. Additional fit parameters are presented in Supplementary Table 2. B Denaturing agarose gel analysis of a time-course experiment monitoring lagging-strand synthesis in the presence of MOF factors (FEN1 and Lig1) using [32P]-dGTP. C Denaturing agarose gel analysis of a time-course experiment monitoring leading-strand synthesis in the presence of MOF factors (FEN1 and Lig1) using [32P]-dCTP. The reactions were identical to (B), except the labeled nucleotide tracks the leading strand. D Quantification of median DNA length versus reaction time from panels (B, C). Sigmoidal fit by Eq. (1) is applied to the mean of two independent experiments. Source data are provided as a Source Data file. Median synthesis rates are calculated by Eq. (2). The amplitude parameters were fixed to 3.375 kb and 3.344 kb for leading-strand and lagging-strand synthesis, respectively. Additional fit parameters are presented in Supplementary Table 2. Together, these findings indicate that in SV40, Pol δ operates partially within the replisome during OF synthesis and early NT, requiring coordination with leading-strand synthesis to prevent strand asymmetry until Pol δ is recycled. Discussion Pol α priming couples leading- and lagging-strand synthesis in eukaryotic replication By reconstituting the human and SV40 replication systems side by side, we were able to examine how Pol α priming influences fork progression across distinct replisomes. In both systems, engagement of Pol α is associated with a measurable reduction in leading-strand synthesis rate, revealing a conserved kinetic consequence of primer synthesis at the fork (Fig. 6A). This slowing might provide a previously unrecognized regulatory handle on Okazaki fragment length: by transiently reducing the rate of CMG–Pol ε elongation during priming, the replisome limits the amount of parental DNA exposed before each new primer is laid, thereby constraining the maximum size of the ensuing OF. Although priming events establish when lagging-strand synthesis can begin, our results indicate that overall fork coordination reflects the intrinsic kinetic architecture of the replisome, including the ability of Pol δ to efficiently complete each fragment. Thus, Pol α-dependent slowing contributes to controlling the physical landscape in which OFs are generated, while synchronization emerges from the integrated enzymatic behavior of the replisome.Fig. 6Proposed model of human and SV40 replisome synchronization.A Human and SV40 leading-strand median synthesis rates derived from sigmoidal fitting of median DNA length versus time under the indicated reaction conditions. Bars represent mean values from three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Pairwise comparisons were performed for the three human groups and the three SV40 groups. Numbers indicate the absolute value of Cohen’s d effect sizes standardized by the pooled standard deviation, and asterisks between brackets denote adjusted statistical significance determined by two-tailed, unpaired Welch’s t tests, without additional adjustments (*p < 0.05; **p < 0.01; ns, not significant). B Left panel: cryo-EM structure of the human replisome bound to Pol α19. Although not modeled in the structure, an ssDNA loop is predicted to connect the 5’-end of the lagging-strand exiting CMG to the active site of PRIM1 for primer synthesis. Right panel: the model of human Pol α in an elongation state (8 ribonucleotides and 3 nucleotides in the primer) (PDB: 8D9D)62 was aligned to the Pol α PRIM2NTD region of the model in the left panel, with Pol α in the model removed. Bottom panel: the model of Xenopus laevis Pol α in a further elongation state (8 ribonucleotides and 11 nucleotides in the primer) (PDB: 8V6J)61 was aligned to the Pol α PRIM2NTD region of the model in the left panel, with Pol α in the model removed. In all models, AND-1 was omitted for clarity. In the right and bottom panels, the blue dashed line represents the lagging-strand loop formed between Pol α and CMG during priming. Molecular models were created with pymol.com. C Human and (D) SV40 cartoon models proposed for replisome synchronization (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). Proposed model of human and SV40 replisome synchronization. A Human and SV40 leading-strand median synthesis rates derived from sigmoidal fitting of median DNA length versus time under the indicated reaction conditions. Bars represent mean values from three independent experiments. Source data are provided as a Source Data file. The error bars represent the SEM. Pairwise comparisons were performed for the three human groups and the three SV40 groups. Numbers indicate the absolute value of Cohen’s d effect sizes standardized by the pooled standard deviation, and asterisks between brackets denote adjusted statistical significance determined by two-tailed, unpaired Welch’s t tests, without additional adjustments (*p < 0.05; **p < 0.01; ns, not significant). B Left panel: cryo-EM structure of the human replisome bound to Pol α19. Although not modeled in the structure, an ssDNA loop is predicted to connect the 5’-end of the lagging-strand exiting CMG to the active site of PRIM1 for primer synthesis. Right panel: the model of human Pol α in an elongation state (8 ribonucleotides and 3 nucleotides in the primer) (PDB: 8D9D)62 was aligned to the Pol α PRIM2NTD region of the model in the left panel, with Pol α in the model removed. Bottom panel: the model of Xenopus laevis Pol α in a further elongation state (8 ribonucleotides and 11 nucleotides in the primer) (PDB: 8V6J)61 was aligned to the Pol α PRIM2NTD region of the model in the left panel, with Pol α in the model removed. In all models, AND-1 was omitted for clarity. In the right and bottom panels, the blue dashed line represents the lagging-strand loop formed between Pol α and CMG during priming. Molecular models were created with pymol.com. C Human and (D) SV40 cartoon models proposed for replisome synchronization (Created in BioRender. Alamodi, A. (2026) https://BioRender.com/m2f9yfp). Recent structural analyses of the human replisome bound to Pol α19 provide insights into our biochemical observations. Pol α engages CMG at the leading edge, positioning Pol α’s catalytic PRIM1 domain near the lagging-strand exit channel of CMG (Fig. 6B). Tethering of Pol α to CMG is mediated by multiple interactions, primarily involving the N-terminal domain of PRIM2 of Pol α, and MCM3 and GINS of CMG (Fig. 6B). Given the slow RNA synthesis rate of Pol α30, a transient accumulation of ssDNA must occur between CMG and Pol α due to their opposing directional activities. Since the CMG–Pol ε structure remains unchanged upon Pol α binding19, we propose that fork slowing results from an allosteric effect on CMG translocation, potentially modulating the ATP-driven conformational cycling of the MCM C-tier motor domains. Alternatively, the entropic penalty associated with unwinding ssDNA under looped constraints may resist CMG translocation, providing a mechanistic basis for the regulation of leading-strand synthesis during replication. The limited interaction surface between CMG and Pol α likely allows Pol α to transition between an autoinhibited apo state and a DNA-engaged priming state19. Our structural predictions further suggest that Pol α may retain CMG binding beyond primer synthesis and into the primer elongation phase19,61,62 (Fig. 6B). However, the extensive rearrangements required for transitioning from RNA priming to primer extension, particularly involving POLA1, POLA2, and the primer-template DNA (Fig. 6B), could transiently disrupt the interaction with the CMG, leading to Pol α dissociation. This dynamic behavior is consistent with the partially distributive nature of Pol α observed here, which was previously reported in human8 and yeast10, and is also supported by single-molecule studies of the yeast replisome40. Given that OF length is dictated by priming frequency, which in turn depends on Pol α levels, tight regulation of Pol α concentration is essential to ensure proper lagging-strand synthesis and its synchronization with leading-strand synthesis. While the structural details of the SV40 L-Tag helicase bound to the full Pol α complex remain unknown, prior studies localize the interaction to the C-tier of L-Tag and the N-terminal domain of POLA263. This suggests that Pol α operates as an integral component of the SV40 replisome, likely enforcing a similar priming-dependent slowing mechanism as in the human system. In bacteriophage T7, primer synthesis transiently pauses leading-strand synthesis, ensuring synchronization between the two strands64. An alternative model proposes that leading-strand synthesis proceeds uninterrupted at a rate determined by helicase unwinding, while lagging-strand primers are synthesized independently and extended at accelerated rates65,66. Our findings indicate that eukaryotic replisomes employ elements of both strategies: Pol α-mediated fork slowing ensures proper lagging-strand priming, while Pol δ synthesizes OFs in a fast yet distributive manner. Together, these mechanisms reduce excessive ssDNA exposure, enforce proper OF lengths, and support the inherently non-processive Pol δ–PCNA complex to remain synchronized with leading-strand synthesis. Uncoupling OF synthesis and MOF governs human lagging-strand replication Our findings indicate that human Pol δ does not stably interact with CMG under the conditions tested. Consistent with this, yeast Pol δ also does not bind CMG20, suggesting that a direct polymerase-helicase association is not a conserved feature of eukaryotic lagging-strand DNA synthesis. The uncoupling of Pol δ’s activities during OF synthesis and MOF in the human replisome is consistent with the observation that neither process affects the leading-strand synthesis rate (Fig. 6A). Consequently, MOF occurs spatially and temporally separated from the human replisome (Fig. 6C). This model aligns with in vivo studies showing that MOF can be delayed until bulk DNA replication is complete67. Extending these observations to SV40, where Pol δ interacts with L-Tag, revealed an additional layer of coordination: MOF, particularly NT, significantly slowed leading-strand synthesis (Fig. 6A), suggesting that lagging-strand Pol δ is partially coupled to L-Tag during OF synthesis and FEN1-dependent NT (Fig. 6D). This slowing might provide a mechanism to reduce the accumulation of ssDNA and control the length of OFs until Pol δ dissociates from the nascent OF, thereby enabling L-Tag–Pol α to initiate primer synthesis. The slowing of leading-strand synthesis in SV40 likely results from an allosteric effect, where binding of the lagging-strand Pol δ, while performing the slow NT reaction, modulates L-Tag helicase activity (Fig. 6D). A recent study showed that the rate and processivity of SV40 leading-strand synthesis increased slightly upon the addition of catalytically inactive Pol δ, presumably through its direct interaction with L-Tag35, further supporting coordination between the helicase and bound DNA polymerases. Collectively, the comparison between human and SV40 replisomes in the absence and presence of MOF showed how uncoupling Pol δ activities during OF synthesis and MOF simplified the synchronization requirement in the human replisome to only accommodate the slow primer synthesis by Pol α (Fig. 6A). This simplicity ensures that leading-strand synthesis remains unimpeded while minimizing disruptions to lagging-strand synthesis, enhancing overall fork progression. The intrinsic uncoupling of OF synthesis from the leading-strand synthesis under normal replication is consistent with previous observations under stress conditions. For example, a lagging-strand lesion can block Pol δ while leading-strand synthesis continues unaffected50, and the telomere-binding protein Rap1 can act as a protein roadblock, inducing lagging-strand gaps52. These studies indicate that leading-strand synthesis can proceed unimpeded even if there are gaps on the lagging strand. A more recent study showed that dNTP depletion stalls leading-strand synthesis, while CMG continues dsDNA unwinding, resulting in continued initiation of OF synthesis and accumulation of incomplete lagging-strand products. These excess OFs sequester and deplete RFC–PCNA, which directly impairs processive leading-strand synthesis during restart51, underscoring the importance of maintaining synchronization and limiting gap accumulation on the lagging strand. Distributive and rapid recycling of Pol δ mediate synchronization in the human replisome Our findings, together with previous studies25,35,36,53,68 indicate that human Pol δ is inherently distributive, frequently dissociating from the lagging-strand template before completing OF synthesis. This contrasts sharply with the yeast case, where Pol δ is highly processive within each OF39, and it is even retained within the replisome to synthesize ~140 OFs40. This retention within the replisome is thought to depend on the Pol32 subunit of Pol δ40, which mediates contacts with other replisome components such as the Pol1 subunit of Pol α-primase41,60. However, the human Pol δ may have a weaker interaction with Pol α due to its frequent dissociation from the DNA. We propose that this may explain why replicating human Pol δ in our experiment was effectively exchanged with Pol δ from solution, as opposed to the case in yeast, where it displays high processivity40. Comparison of reconstituted MOF from human and yeast establishes the low processivity of human Pol δ during MOF25,29, likely due to weaker Pol δ–PCNA interactions37,38, which render NT by Pol δ and FEN1 more distributive25. This low processivity manifested itself in human MOF being ~ 55-fold slower than in yeast25. Here, we extend these findings to show that Pol δ remained unprocessive during MOF within the replisome, supported by the observations that the lagging-strand product lengths do not change in the presence of only the NT reaction and that MOF remained intrinsically a slow process, continuing well after leading- and lagging-strand synthesis have completed. However, it remains possible that additional regulatory pathways enhance maturation efficiency in vivo. For example, post-translational modifications of MOF factors may further regulate this process. Phosphorylation of human FEN1 has been reported to decrease its interaction with PCNA69, which may promote dissociation of FEN1 from the nick product and facilitate more efficient re-engagement of Pol δ. Finally, additional proteins may associate with the replisome and/or the MOF complex to enhance the rate and processivity of human MOF. Importantly, the yield of MOF within the replisome remained comparable to that observed in previously reported isolated human MOF systems25,55. This observation suggests that replisome-associated interactions do not substantially enhance MOF efficiency. Instead, MOF appears to remain intrinsically distributive in the human system and to proceed with limited overall yield. Notably, inefficient intermediate utilization has also been observed in yeast, where primer utilization and productive clamp loading were identified as limiting factors70. Furthermore, even in the absence of the ligation step, both reconstituted human and yeast MOF systems exhibit limited yield during the NT sub-process, resulting in incomplete primer utilization25,29. It is therefore possible that the overall yield of MOF reconstitution is limited by earlier steps, such as PCNA loading70 or handover to Pol δ. Nevertheless, while previous studies generally examined a single MOF step and reported < 70% primer utilization in reconstituted NT and MOF reactions, our reconstitution requires ~ 5–12 OFs to undergo NT-mediated RNA primer removal followed by ligation to achieve complete lagging-strand maturation and generate the final full-length product, as estimated from the final product size relative to the average OF size (Supplementary Fig. 9A). Thus, the ~ 60% MOF yield observed here in the presence of the replisome is remarkable given the multiplicative effect of the limited efficiencies at each individual MOF step. Importantly, it has previously been shown that a considerable amount of Pol α-derived DNA is retained in the mature genome71, raising the possibility that the limited MOF yield observed here reflects a biologically relevant feature of human lagging-strand maturation rather than solely a reconstitution artifact. The distributive nature of Pol δ during OF synthesis and MOF, combined with the slow MOF reaction, could challenge synchronization with leading-strand progression, even though Pol δ extends primers at a rapid rate of 240 nt s−1 29,35. The robust OF synthesis in the presence of MOF indicates that OF synthesis remained synchronized with leading-strand synthesis, supporting a model in which Pol δ–PCNA prioritizes the completion of OF synthesis over MOF, leaving OFs unligated for a prolonged time. This model aligns with in vivo findings showing that OF processing and ligation can occur post-replication, uncoupled from DNA synthesis67. While the reduced processivity of human Pol δ during MOF could be perceived as a kinetic bottleneck, it may serve an essential role in polymerase recycling. Human cells contain an estimated ~ 1.96 × 10⁴ copies of Pol δ per cell72, requiring each polymerase molecule to be recycled thousands of times per S-phase (~ 1220–3060 times)72. The fully uncoupled and distributive nature of MOF may thus ensure the rapid turnover of Pol δ, supporting continuous lagging-strand synthesis without depleting the active polymerase pool36. In fact, human Pol δ exhibits the highest binding affinity for PCNA when PCNA is bound to a primer-template junction, rather than to nicked or flap DNA intermediates associated with MOF25. Furthermore, previous results show that human FEN1–PCNA outcompetes Pol δ–PCNA for the flap and nick substrates during NT25, which would act as a molecular brake for Pol δ’s engagement in MOF and enhance its recycling. These findings highlight the kinetic basis underlying Pol δ–PCNA association, which prioritizes OF synthesis over MOF and thus ensures synchronized strand synthesis. In contrast, yeast Pol δ is significantly more processive, remaining engaged longer during both OF synthesis40 and MOF29, reducing its dependence on frequent polymerase turnover39. However, since FEN1 stimulates NT in yeast29 it is possible that NT would enhance yeast Pol δ’s engagement in MOF, reducing its recycling kinetics upon colliding with the previous OF. On the other hand, the direct interaction between Pol δ and L-Tag35 in SV40 may facilitate polymerase redistribution between primer hand-off, OF synthesis, and MOF, ensuring that lagging-strand processing remains coupled to replisome progression, but at the cost of reducing the speed of the leading strand. Biological implications of the distributive human Pol δ The non-processive nature of Pol δ’s SD and NT in humans may offer a functional advantage by minimizing unnecessary re-replication of OFs and maintaining short flap lengths. In a processive mechanism, Pol δ could excessively extend beyond the RNA-DNA junction, generating longer flaps that necessitate additional resection before ligation. In contrast, a distributive approach ensures that Pol δ disengages once sufficient DNA synthesis has occurred, preventing excessive extension and maintaining efficient lagging-strand maturation. On the other hand, since Pol α lacks proofreading activity, errors introduced within Pol α-derived segments are only partially corrected during maturation. The NT machinery in yeast can processively replace the entire α-segment in the absence of Lig129, whereas in humans, only the first few nucleotides are removed25, leading to partial retention of Pol α-derived DNA in the mature genome71. This balance likely optimizes replication speed while preserving genome stability, reducing the risk of misincorporation or excessive re-synthesis at lagging-strand junctions. This disparity in processivity may also reflect species-specific differences in OF length regulation. In yeast, nucleosome assembly constrains Pol δ’s SD activity, enforcing a uniform OF size that aligns with in vivo observations. However, on naked DNA templates, yeast Pol δ generates abnormally long and heterogeneous fragments12,73. In humans, where MOF is highly distributive, frequent Pol δ turnover and leading-strand regulation during primer synthesis might provide an alternative, nucleosome-independent mechanism for maintaining fragment size. Furthermore, the prolonged retention of nicks during MOF may serve as a regulatory checkpoint for chromatin assembly, ensuring that histone deposition and chromatin maturation occur efficiently before cell division74. This suggests that MOF is not merely a terminal processing step but an integrated mechanism linking replication to chromatin inheritance. In summary, our study reveals intrinsic coordination mechanisms within the eukaryotic replisome, highlighting how polymerase dynamics and MOF are finely tuned to sustain genome integrity. The complete uncoupling of lagging-strand maturation in humans, unlike its partial integration in SV40, underscores distinct evolutionary strategies balancing replication efficiency with genome stability. Given the strong link between replication stress and cancer, future studies should investigate how these mechanisms adapt to oncogenic transformation, transcription-replication conflicts, and chromatin remodeling disorders. Defining how replication integrates with broader nuclear processes will be key to understanding genome maintenance and its implications for disease and therapy. Methods Plasmid construction MultiBac™ expression system (Geneva Biotech) was used to express human CMG, FPC, Mcm10, Pol δ, Pol ε, AND-1, RFC, Pol α, L-Tag, and CTF18-RFC in insect cells. GenScript performed all gene synthesis and cloning. For human CMG, MCM2 (accession number: P49736), MCM4 (accession number: P33991), MCM5 (accession number: P33992), MCM6 (accession number: Q14566), MCM7 (accession number: P33993) and GINS1-4 with N-terminally 8x histidine-tagged GINS4 (accession numbers: Q14691, Q9Y248, Q9BRX5, and Q9BRT9, respectively) with independent promoters and terminators were cloned into pACEBac1 (Gent+). MCM3 (accession number: P25205-2) and N-terminally twin-Strep tagged with TEV protease cleavage site Cdc45 (accession number: O75419) with independent promoters and terminators were cloned separately into pACEBac1 (Gent+). For human full-length FPC, N-terminus 6x histidine-tagged Claspin (accession number: Q9HAW4), TIMELESS (accession number: Q9UNS1) and C-terminus Strep-tagged Tipin (accession number: Q9BVW5) were amplified and cloned into pACEBac1 (Gent+), pIDS (Spect+) and pIDC (Cm+), respectively, using the Gibson assembly protocol. For human full-length Mcm10 (accession number: Q7L590), N-terminal Strep- and 8 x histidine-tagged insect cell-optimized sequences were cloned into pFastBac1. To generate a polymerase-and-exonuclease-deficient mutant (hereafter referred to as Pol δDead), D515V, D602A, D757A, and E795A were mutated in the p125 subunit of wild-type Pol δ (Pol δWT)37 using polymerase chain reaction (PCR). For Pol ε, the full-length PolE1 subunit (accession number: Q07864) was cloned into pACEBac1 (Gent+). Pol2 (accession number: P56282), Pol3 (accession number: Q9NRF9) C-terminus TEV cleavage twin-Strep tagged, and Pol4 (accession number: Q9NR33) with N-terminally 8x histidine-tagged with TEV protease cleavage site, with independent promoters and terminators were cloned into pIDS (Spect+). Finally, the single transfer vector with different subunit assemblies of Pol ε was generated using Cre recombinase according to the MultiBac™ expression system user manual. Full-length human AND-1 (accession number: O75717) N-terminally twin-Strep with TEV protease cleavage site was cloned into pACEBac1 (Gent+). For human full-length RFC, N-terminus 6x histidine-tagged RFC3 and RFC5 with independent promoters and terminators were cloned into pIDS (Spect+). RFC2, RFC4, and C-terminally Strep-tagged RFC1 with independent promoters and terminators were cloned into pACEBac1 (Gent+). Finally, the single transfer vector with all the subunits of RFC was generated using Cre recombinase according to the MultiBac™ expression system user manual. For human Pol α, N-terminally twin-Strep tagged with TEV protease cleavage site was cloned into pACEBac1 (Gent+) with POLA2 with independent promoters and terminators. PRIM1 and PRIM2 were cloned into pIDS (Spect+) with independent promoters and terminators. Finally, the single transfer vector with all the subunits of Pol α was generated using Cre recombinase according to the MultiBac™ expression system user manual. For L-Tag N-terminally double 6x histidine-tagged insect cell-optimized sequence was cloned into pFastBac1 using GenScript. For CTF18-RFC, N-terminally twin-Strep with TEV protease cleavage site CTF18 (accession number: Q8WVB6), RFC2 (accession number: P35250), RFC4 (accession number: P35249) with independent promoters and terminators were cloned into pACEBac1(Gent+). N-terminally 8x histidine-tagged RFC3 (accession number: P40938), RFC5 (accession number: P40937), DCC1 (accession number: Q9BVC3) and CTF8 (accession number: P0CG13) with independent promoters and terminators were cloned into pIDS (Spect+). FEN1 and Lig1 were expressed and purified as previously described26. Finally, the single transfer vector with different subunit assemblies of CTF18-RFC was generated using cre recombinase according to the MultiBac™ expression system user manual. All recombinant human and SV40 proteins were expressed using the MultiBac™ expression system (Geneva Biotech) in insect cell lines via recombinant baculovirus, except for PCNA and RPA, which were expressed using Escherichia coli expression systems. Baculovirus preparation To prepare the baculovirus, bacmid DNA of CMG, FPC, Mcm10, Pol δ, Pol ε, AND-1, RFC, Pol α, L-Tag, and CTF18-RFC plasmids were transfected separately into Sf9 or Hi5 cells using FuGENE® HD (Promega) according to the manufacturer’s instructions. The resulting supernatant was obtained as the P1 virus stock, which was then amplified to obtain the P2 virus stock. P2 virus stock was then further amplified to obtain P3 virus stock for large-scale expression. CMG Human CMG was expressed by co-infecting 6 L of Hi5 suspension cells (2 × 10⁶ cells/mL) with P3 baculoviruses encoding MCM2,4–7, GINS complex, MCM3, and Cdc45, for 72 h. Cells were harvested (5500 × g for 10 min) and resuspended in buffer A [40 mM HEPES-NaOH (pH 7.5), 150 mM NaOAc, 20 mM imidazole, 0.5 mM TCEP, 0.005% Tween-20, 10% glycerol, and one EDTA-free protease inhibitor cocktail tablet/50 mL]. Cells were sonicated, and debris was clarified by centrifugation (95,834 × g, 1 h, 4 °C). The supernatant was loaded onto a HisTrap HP 5 mL column (Cytiva) pre-equilibrated with buffer A, washed with 10 CV of buffer A, and eluted with a 10 CV gradient of buffer B [40 mM HEPES-NaOH (pH 7.5), 150 mM sodium acetate, 400 mM imidazole, 0.5 mM TCEP, 0.005% Tween-20, and 10% glycerol]. Eluate was pooled and loaded onto a StrepTrap XT 1 mL column (Cytiva) pre-equilibrated with buffer C [40 mM HEPES-NaOH (pH 7.5), 150 mM NaOAc, 0.5 mM TCEP, 0.005% Tween-20, and 10% glycerol], washed with 10 CV each of buffer C and buffer D (containing 5 mM MgOAc and 0.5 mM ATP), and rewashed with 30 CV of buffer C. CMG was eluted with two 10 CV steps of buffer C (50 mM biotin), pooled, concentrated, and purified by size-exclusion chromatography (Superdex 200 Increase 10/300 GL) in buffer E [40 mM HEPES-KOH (pH 7.5), 100 mM KOAc, 1 mM DTT, 2 mM MgOAc, and 10% glycerol]. Final fractions were pooled, concentrated, flash-frozen, and stored at − 80 °C. Mcm10 and FPC Mcm10 and FPC were expressed and purified as described previously35. Briefly, a 4 L of Sf9 suspension culture at 2 × 106 cells/mL was infected with P3 viral stock for 48 and 72 h, respectively. Cells were harvested, and a clear lysate was loaded separately on a HisTrap HP 5 mL affinity column (Cytiva), and the bound protein was eluted with imidazole. Fractions that contain Mcm10 or FPC were combined and loaded onto a Strep-Trap XT 5 mL column (Cytiva), and bound protein was eluted with Biotin. The fractions containing Mcm10 or FPC were combined. The combined fractions were incubated overnight with TEV protease to remove tags. The fractions were concentrated to 1 mL and loaded onto HiLoad 16/600 Superdex 200 pg (Cytiva). Protein fractions were pooled, concentrated, flash-frozen, and stored at − 80 °C. Pol δ Human WT and mutant Pol δ were expressed and purified as described previously37. Briefly, a 4 L suspension of Sf9 cells at 2 × 10⁶ cells/mL was infected with P3 viral stock and harvested after 68–72 h. Clarified lysate was loaded onto a HisTrap column (Cytiva), and bound Pol δ was eluted with imidazole and 50 mM NaCl. Eluted fractions were further purified by Mono Q anion exchange, followed by size exclusion chromatography on a HiLoad 16/600 Superdex 200 pg column (Cytiva). Protein fractions were pooled, flash-frozen, and stored at − 80 °C. Pol ε Human Pol ε was expressed and purified as described previously75. Briefly, 4 L of Hi5 suspension culture at a density of 2 × 106 cells/mL was transfected with the Pol ε P3 virus. After 72 h, the cells were harvested, and the clear lysate was loaded onto a 5 mL HisTrap affinity column (Cytiva), and the bound protein was eluted with imidazole. The pooled fractions of Pol ε were directly loaded onto the Strep-Trap XT 5 mL affinity column (Cytiva), and the bound protein was eluted with Biotin. The fractions containing Pol ε were combined and incubated overnight with TEV protease to remove tags. The fractions were then loaded onto a MonoQ 5/50 GL 1 mL anion exchange column (Cytiva) and eluted with 1 M NaCl. Eluents were pooled and concentrated to 1 mL and loaded onto HiLoad 16/600 Superdex 200 pg (Cytiva). Protein fractions were pooled, flash-frozen, and stored at − 80 °C. AND-1 Full-length AND-1 was expressed in 4 L Sf9 suspension culture (2 × 10⁶ cells/mL) infected with P3 baculovirus for 72 h. Cells were harvested and resuspended in lysis buffer [50 mM Tris-HCl (pH 8), 300 mM NaCl, 1 mM EDTA, 5 mM BME, 0.1% NP-40, 10% glycerol, and one EDTA-free protease inhibitor cocktail tablet/50 mL]. Cells were sonicated, and debris was cleared by centrifugation (95,834 × g, 1 h, 4 °C). The supernatant was incubated with 5 mL Strep-Tactin XT resin (Cytiva) pre-equilibrated with buffer A [50 mM Tris-HCl (pH 8), 300 mM NaCl, 5 mM β-mercaptoethanol (BME), and 10% glycerol]. After 2 h, the beads were spun down, and the supernatant was removed. The beads were then washed twice with 25 mL of buffer A, and the bound protein was eluted by incubating the beads with 10 mL of buffer B [50 mM Tris-HCl (pH 8), 300 mM NaCl, 5 mM BME, 50 mM biotin and 10% glycerol] for 2 h. Eluted protein was concentrated to 1 mL and loaded onto HiLoad 16/600 Superdex 200 pg (Cytiva) column pre-equilibrated with gel filtration buffer [50 mM Tris-HCl (pH 8), 200 mM NaCl, 1 mM DTT, and 10% glycerol]. Protein fractions were pooled, concentrated, flash-frozen, and stored at − 80 °C. RFC RFC was expressed by transfecting 4 L of Hi5 suspension culture at 2 × 106 cells/mL with P3 virus for 48 h. Cells were harvested (5500 × g, 15 min, 4 °C) and resuspended in lysis buffer [50 mM HEPES (pH 7.5), 500 mM NaCl, 20 mM imidazole, 5 mM β-mercaptoethanol, 5% glycerol, and one EDTA-free protease inhibitor cocktail tablet/50 mL]. Cells were sonicated, and debris was clarified by centrifugation (95,834 × g, 1 h, 4 °C). The supernatant was loaded onto a HisTrap HP 5 mL affinity column (Cytiva) pre-equilibrated with buffer A [50 mM HEPES (pH 7.5), 500 mM NaCl, 20 mM imidazole, 10 mM β-mercaptoethanol, and 5% glycerol]. The column was washed with 10 CV of buffer A and eluted with a 10 CV gradient of buffer B [50 mM HEPES (pH 7.5), 250 mM NaCl, 300 mM imidazole, 10 mM β-mercaptoethanol, and 5% glycerol]. Eluate was pooled and loaded onto a StrepTrap XT 5 mL column (Cytiva) pre-equilibrated with buffer C [50 mM HEPES (pH 7.5), 250 mM NaCl, 1 mM EDTA, and 5% glycerol], washed with 10 CV of buffer C and eluted with 10 CV buffer D [50 mM HEPES (pH 7.5), 250 mM NaCl, 1 mM EDTA, 5% glycerol, and 50 mM Biotin]. Protein fractions were pooled, flash-frozen, and stored at −80 °C. Pol α Pol α was expressed by transfecting 4 L of Hi5 suspension culture at 2 × 106 cells/mL with P3 virus for 48 h. Cells were harvested (5500 × g, 15 min, 4 °C) and resuspended in buffer A [25 mM Tris-HCl (pH 7.2), 300 mM NaCl, 10% glycerol, 0.5 mM TCEP, 0.02% NP-40-S and one EDTA-free protease inhibitor cocktail tablet/50 mL]. Cells were sonicated, and debris was clarified by centrifugation (95,834 × g, 1 h, 4 °C). Strep-Tactin XT superflow high-capacity resin was added to the lysate and incubated for 1 hr at 4 °C. Resin was collected in a 20 ml column and was washed with 50 ml buffer A. Protein was eluted with 10 CV of buffer B [25 mM Tris-HCl (pH 7.2), 100 mM NaCl, 10% glycerol, 0.5 mM TCEP, 0.02% NP-40-S, and 30 mM Biotin]. The fractions containing Pol α were pooled and loaded onto 1 ml MonoQ column (Cytiva) pre-equilibrated with buffer C [25 mM Tris-HCl (pH 7.5), 100 mM NaCl, 10% glycerol, 0.5 mM TCEP, and 0.01% NP-40-S]. The protein was eluted with a 20 CV gradient from 100 to 1000 mM NaCl. Peak fractions were pooled, concentrated to approximately 500 μl in an Amicon Ultra-15 30 kDa MWCO concentrator and applied to a Superdex 200 Increase 10/300 gel-filtration column (Cytiva) pre-equilibrated in buffer D [25 mM Tris-HCl (pH 7.2), 10% glycerol, 0.005% Tween-20, 0.5 mM TCEP, and 150 mM NaCl]. Peak fractions were pooled, frozen in liquid nitrogen and stored at − 80 °C. L-Tag L-Tag was expressed and purified as described previously35. Briefly, 4 L of the Sf9 suspension culture at 2 × 106 cells/mL density was transfected with the L-Tag P3 virus. After 48 h, the cells were harvested, and the clear lysate was loaded onto a 5 mL HisTrap affinity column, and the bound protein was eluted with imidazole. The pooled fractions of L-Tag were treated with TEV protease to remove the His-tag. The cleaved protein was then loaded onto a HisTrap 1 mL affinity column (Cytiva). Flow-through fractions containing L-Tag were concentrated and loaded onto a HiLoad 16/600 Superdex 200 pg (Cytiva). The collected protein was flash-frozen and stored at − 80 °C. CTF18-RFC CTF18-RFC was expressed and purified as described previously75. Briefly, a 4 L of Hi5 suspension culture at a density of 2 × 106 cells/mL was transfected with the CTF18-RFC P3 virus. After 60-65 h, the cells were harvested, and the clear lysate was loaded onto a 5 mL HisTrap affinity column (Cytiva), and the bound protein was eluted with imidazole. The pooled fractions of CTF18-RFC were directly loaded onto the Strep-Trap XT 1 mL affinity column (Cytiva), and the bound protein was eluted with Biotin. Eluted fractions containing CTF18-RFC were concentrated and loaded onto a HiLoad 16/600 Superdex 200 pg (Cytiva). The collected protein was flash-frozen and stored at − 80 °C. RPA Human RPA was expressed and purified as described previously37. Briefly, the cloned plasmid was transfected into BL21 (DE3) E. coli. The cells were grown in 2YT media at 37 °C to an OD600 of 0.7, and protein expression was induced with 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) and further incubated for 4–6 h at 37 °C. The cells were collected by centrifugation, lysed with lysozyme, and sonicated. The supernatant was loaded onto the HisTrap HP 5 mL affinity column (Cytiva), followed by the HiTrap Blue affinity column (Cytiva). RPA fractions containing all subunits were concentrated and loaded onto the HiLoad 16/600 Superdex 200 pg column (Cytiva). RPA protein fractions were flash-frozen and stored at − 80 °C. PCNA Human PCNA was expressed and purified as described previously37. Briefly, BL21(DE3) cells containing the PCNA plasmid were grown in 2YT media with ampicillin at 37 °C. The protein expression was induced with 0.5 mM IPTG when OD600 reached 1.2 and incubated further for 19 h at 16 °C. The cells were harvested by centrifugation and lysed by sonication. The supernatant was purified using a HisTrap HP 5 mL affinity column, followed by a HiTrap Q column, an anion exchanger (Cytiva), and size-exclusion chromatography on a HiLoad 16/600 Superdex 200 pg (Cytiva). The collected fractions were flash-frozen and stored at − 80 °C. ϕ29 DNA Polymerase Codon-optimized phi29 ORF (IDT) was cloned into the pE-SUMO vector via Gibson assembly and transformed into E. coli BL21 (DE3). Cultures were grown at 37 °C to OD600 1.0 and then shifted to 16 °C and induced with 1 mM IPTG overnight. Cells were harvested (5000 × g, 15 min), and the cell pellet was stored at − 80 °C until purification. For protein extraction and purification, the frozen cell pellet was treated with lysozyme (1.0 mg/mL) followed by lysis using a cell disruptor at 2500 psi. The lysate was clarified by sequential centrifugation at 22,040 × g for 30 min followed by 142,032 × g for 40 min. The supernatant was applied to a HisTrap Excel column (Cytiva) pre-equilibrated with buffer A [25 mM Tris-HCl, 500 mM NaCl, and 1.5 mM BME]. The bound proteins were eluted with buffer A (500 mM imidazole). Fractions containing Phi29 were pooled, dialyzed overnight with SUMO protease to remove the tag, and reloaded onto a HisTrap column to collect flow-through containing untagged Phi29. After reducing NaCl to 50 mM, the sample was further purified on a HiTrap Heparin column using a NaCl gradient. Peak fractions were dialyzed into storage buffer [10 mM Tris-HCl (pH 7.4), 100 mM KCl, 1 mM DTT, 0.1 mM EDTA, 0.5% Tween-20, and 50% glycerol] and stored at − 20 °C. FEN1 Human FEN1 was expressed and purified as previously described26. Briefly, human FEN1 was expressed in E. coli BL21 (DE3) cells using the pE-SUMO-pro expression vector (LifeSensors), which introduces an N-terminal 6x histidine-SUMO tag. Cells were grown in 2YT medium at 37 °C to an OD600 of 1.0, and protein expression was induced with 0.3 mM IPTG followed by incubation for 19 h at 16 °C. Cells were collected by centrifugation, lysed with lysozyme, and sonicated. The clarified lysate was loaded onto a 5 mL HisTrap HP affinity column (Cytiva). Fractions containing FEN1 were pooled and dialyzed overnight in the presence of SUMO protease to remove the tag. The sample was reapplied to the HisTrap HP column, and untagged FEN1 was collected in the flow-through, concentrated, and further purified using a HiLoad 16/600 Superdex 75 pg column (Cytiva). Purified FEN1 fractions were pooled, concentrated, flash-frozen, and stored at − 80 °C. Lig1 Human Lig1 was expressed and purified as previously described26. Briefly, the plasmid encoding untagged full-length human Lig1 was transformed into BL21 (DE3) E. coli. Cells were grown in 2YT medium at 24 °C to an OD600 of approximately 0.8, and protein expression was induced with 0.2 mM IPTG followed by incubation for 19 h at 16 °C. Cells were collected by centrifugation, lysed with lysozyme, and sonicated. The clarified lysate was sequentially purified using a phosphocellulose cation-exchange column (Whatman), a 5 mL HiTrap Q HP anion-exchange column (Cytiva), and a 5 mL HiTrap Blue HP column (Cytiva). Lig1-containing fractions were concentrated and further purified using a HiLoad 16/600 Superdex 200 pg column. Purified Lig1 fractions were pooled, concentrated, flash-frozen, and stored at − 80 °C. Fork annealing The sequence of the replication fork is listed in Supplementary Table 1. To make the fork, oligo 1873 was mixed with 1871 in a 25 μl reaction containing 200 mM NaCl and 25 mM EDTA. The mixture was heated to 95 °C and naturally cooled to room temperature (RT). Annealing leaves a 5′-phosphorylated GCCA overhang for ligation to one end of the BsaI-linearized plasmid. The 5’end of 1873 and 3’end of 1871 has phosphorothioate bonds to protect against exonuclease activity. Human replication substrate ZN38 plasmid was digested with BsaI (20,000 units/ml, New England Biolabs) in a reaction containing 1 x CutSmart Buffer (New England Biolabs) for 2 h at 37 °C. The digested plasmid was purified from an agarose gel using the QIAquick Gel Extraction Kit (QIAGEN). The BsaI-linearized DNA was ligated to a 5-fold molar excess pre-annealed fork and incubated in a reaction containing T4 DNA ligase (60,000 units, New England Biolabs), 10x T4 ligase buffer at 16 °C for 17 h. The excess unligated fork was removed by gel purification. SV40 replication substrate To monitor leading- and lagging-strand synthesis, we designed a substrate consisting of a 3.3 kb nucleotide-biased linear DNA, enabling strand-specific labeling with [32P]-dCTP (leading) or [32P]-dGTP (lagging)16. The amplified plasmid was digested with BsaI (20,000 units/ml) and BtsCI (20,000 units/ml), gel-purified and ligated to a 5-fold molar excess forked junction (oligo 1871 and 1873) at one end, and 10-fold molar excess blocking oligos (oligo 1914 and 1915) at the other end to prevent end-labeling. Excess un-ligated oligos were removed by gel purification. A 37-mer primer (oligo 1874) was annealed to the leading strand, leaving a 30-nt ssDNA gap for L-Tag loading. Human replication assay Replication reactions were adopted from a previous study8. They were conducted at 37 °C in a replication buffer consisting of 25 mM HEPES-KOH (pH 7.6), 100 mM K glutamate, 1 mM DTT, 10 mM Mg acetate and 100 μg/mL BSA. Unless stated in the figure legends, the DNA, protein and nucleotide concentrations in the final replication reaction were: 1.5 nM DNA, 25 nM CMG, 40 nM Pol ε, 20 nM RFC, 100 nM PCNA, 30 nM AND-1, 20 nM Pol α, 100 nM RPA, 20 nM FPC, 30 nM CTF18–RFC, 4 mM ATP, 30 μM dC/dT/dG/dATP, 200 μM C/G/UTP and α-[32P]-dCTP. Reactions were staged as follows: CMG was pre-incubated with the forked DNA in replication buffer supplemented with 0.1 mM AMP-PNP for 10 min at 37 °C, followed by the addition of Pol ε, CTF18-RFC, PCNA, AND-1, FPC, Pol δ, Pol α with dATP, dCTP for 5 min at 37 °C. Replication was initiated by the addition of ATP, dTTP, dGTP, GTP, CTP, UTP, [32P]-dCTP, and RPA. In reactions containing Pol δ, RFC was added when replication was initiated. Reactions were quenched upon adding 40 mM EDTA and electrophoresed on a 0.7% alkaline agarose gel at 35 V for 17 h. The gel was backed with DE81 paper, compressed, and imaged in a Sapphire Biomolecular Imager (Azure Biosystems). For the Pol δ competition experiment, the replisome was assembled as described above in the presence of all protein components, and the indicated concentrations of Pol δDead or Pol δWT were added 1 min after initiation. The reactions were quenched and processed as described above. SV40 replication assays Replication assays contained 1.5 nM linear forked DNA, 30 nM L-Tag, 100 nM RPA, 100 nM PCNA, 30 nM RFC, 15 nM FPC, 15 nM Mcm10, 60 nM Pol δ, 10 nM Pol α in 25 mM HEPES (pH 7.6), 100 μg/mL BSA, 1 mM DTT, 10 mM Mg acetate, 100 mM K glutamate, 4 mM ATP, 200 µM rNTPs, 150 µM of each unlabeled dNTP, and 20 µM of labeled dNTP. Reactions were staged as follows: L-Tag was added first and pre-incubated with FPC, Mcm10, and forked DNA for 5 min at 37 °C, followed by the addition of Pol δ, RFC and PCNA, along with dATP, dTTP, dGTP, dCTP, rNTPs, RPA, and ATP. For monitoring the leading-strand replication products, 10 μCi [32P]-dCTP was used, whereas for lagging-strand replication reactions, 10 μCi [32P]-dGTP was used. Reactions were quenched upon adding 40 mM EDTA and electrophoresed on a 0.7% alkaline agarose gel at 35 V for 17 h. The gel was backed with DE81 paper, compressed, and imaged in a Sapphire Biomolecular Imager (Azure Biosystems). Human nicking enzyme reaction The human replisome was assembled as described above. For leading-strand synthesis, CMG, Pol ε, PCNA, AND-1, RPA, FPC, and CTF18–RFC were included. For combined leading- and lagging-strand synthesis, Pol δ, Pol α, and RFC were additionally included. For leading–lagging synthesis with MOF, all proteins were included together with 15 nM FEN1 and 20 nM Lig1. All protein concentrations were identical to those used in the standard human replication assay. Following incubation for the indicated time points, reactions were heat-inactivated at 65 °C for 10 min. Samples were then treated with the strand-specific nicking enzyme Nb.BbvCI (NEB) for 1 h at 37 °C. Reactions were quenched by the addition of 40 mM EDTA and analyzed by electrophoresis on a 0.7% alkaline agarose gel at 35 V for 17 h, as described above. Maturation of Okazaki Fragments assay Standard replication reactions for both the human and SV40 systems were performed as described above. At the initiation stage, 15 nM FEN1 and 20 nM Lig1 were added unless otherwise stated in the figure legends. M13-primer extension The primer extension reaction was conducted in a buffer containing 25 mM Tris-HCl (pH 7.6), 100 mM NaCl, 1 mM DTT, 10 mM Mg acetate, 100 μg/mL BSA, 5 mM ATP, 100 μM each of dTTP, dATP, dGTP, 20 μM dCTP, and 10 μCi of [32P]-dCTP. 2 nM of M13-primed template (oligo 562) was initially incubated at 37 °C for 10 min with 100 nM PCNA, 400 nM RPA, 30 nM RFC in the reaction buffer. The reaction was initiated by the addition of 20 nM Pol δ and FEN1 as indicated in the figure legend. The reactions were stopped 10 s after initiation by adding 50 mM EDTA. Samples were processed as described for standard DNA replication assays. Restriction enzyme analysis of SV40 lagging-strand synthesis distribution SV40 standard replication reactions were performed as described above. The reactions were quenched at different time points. Reactions were heat inactivated at 65 °C for 10 min to inactivate the L-Tag and polymerases. The reactions were then divided into three tubes: tube 1 was untreated, tube 2 was treated with PsiI (NEB), and tube 3 was treated with EarI (NEB). Reactions were analyzed on a 2% native agarose gel. The gel was processed as described for standard DNA replication assays. Microscale thermophoresis (MST) His6-tagged CMG was labeled with RED-Tris-NTA dye in the PBS-T buffer (NanoTemper Technology, München, Germany) with a dye-to-protein ratio of 1:2. Pol ε or Pol δ was added to an equal volume of labeled CMG (20 μL final; ∼ 1 nM CMG probe final) at the indicated concentrations in 1 × PBST buffer. All proteins were spun at 14,000 rpm for 30 min at 4 °C to remove insoluble aggregates. All reactions were allowed to equilibrate for 10 min at RT and were subsequently drawn into premium-coated capillaries (NanoTemper Technologies). The thermophoresis profile at each concentration point was quantified in a Monolith NT.115 instrument (NanoTemper Technologies) using a 640 nm excitation LED laser set at 60% and an MST power set at 40%. All experiments were performed in triplicate. Quantification and statistical analysis DNA synthesis rates were calculated following25,29,76. Briefly, gel images were analyzed in ImageJ for both the DNA ladder and experimental reaction lanes. The migration distance of each band of the DNA ladder was paired with its corresponding known size to generate a standard calibration curve. This calibration curve was then used for all experimental lanes to convert median or specific band migration distances of unknown products into DNA length. Background was accounted for by subtracting the scalar total integrated intensity of empty lanes (their average if multiple empty lanes were present and used). Each lane profile was corrected for multiple radiolabel incorporations by dividing each product row by its corresponding number of incorporated radiolabels as determined from the product sequences. Time points which show detector saturation were avoided during analysis. For reactions containing both leading- and lagging-strand products, the intensities of the leading- and lagging-strand products were quantified by integrating the signal above and below 3 kb, respectively. To quantify synthesis kinetics, median product lengths were determined for each time point and plotted against reaction time. This provides a clear mathematical advantage over using the mean product length, which would prematurely strongly curve and artificially depress the kinetic trajectory as soon as the fastest molecules reach the template boundary. As the reaction progresses and most molecules naturally converge to the maximum template length, the median correspondingly converges to the exact size of the full-length product with a natural width of the full-length product band. Excluding the top band would therefore artificially truncate the dataset and disregard completed DNA products. By using the median product length combined with sigmoidal fitting (described below), all synthesized DNA mass is incorporated into the analysis, thus maintaining the mathematical validity of the synthesis metrics as the reaction proceeds to completion. Both median length and total intensity DNA synthesis experimental datapoints were also fitted to 3-parameter logistic sigmoidal dependencies as:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S\left(t\right)=A\left(1-\frac{1}{1+{\left(\frac{t}{{t}_{1/2}}\right)}^{\alpha }}\right)$$\end{document}St=A1−11+tt1/2αwhere S(t) is the current median DNA products length or integrated total DNA synthesis products intensity at time point t, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${t}_{1/2}$$\end{document}t1/2 is the time point at which 50% of the maximum median length or total synthesis is achieved, and αα is a power parameter related to the steepness of the sigmoid. Prior to fitting by Eq. (1), all integrated intensities were normalized to the intensity of the 3 kb band from their corresponding gels. For intensity-based analysis, A is an amplitude parameter corresponding to the intensity of the upper plateau level of intensity of the sigmoid and left free during the fitting. For median length-based analysis, A is an amplitude parameter corresponding to the median length of the upper plateau level of median length of the sigmoid and was set to the full product length as determined from the sequence of the substrate strands. All sigmoidal parameters and goodness-of-fit information about the sigmoidal fits are presented in Supplementary Table 2. For all median length-based sigmoidal fits described by Eq. (1), the median rate was defined as the ratio between maximum product half-length (A/2) and the time needed to reach this point \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({t}_{1/2})$$\end{document}(t1/2) as:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${v}_{1/2}=\frac{A}{2{t}_{1/2}}$$\end{document}v1/2=A2t1/2 For human leading strand ± MOF experiments presented in Fig. 3B, E and Supplementary Fig. 9C, D, the time-course of the formation of the full-length product was also analyzed by using an analytical kinetic model for a multi-step molecular motor with identically distributed step duration. The form of the analytical dependence was adapted from refs. 25,77 by assuming infinite processivity, since no trap was present in the reaction and since all components could rebind freely. Although the experimental conditions do not strictly adhere to the single-turnover kinetics with n equal limiting steps from which this kinetic model was originally derived, this methodology was employed strictly for its numerical stability during model selection. Standard generalized sigmoidal models, as the one in Eq. (1), featuring a continuous real-valued αα coefficient are prone to parameter covariance between the αα and the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${t}_{1/2}$$\end{document}t1/2 coefficients. This trade-off can artificially inflate the log-likelihood and leak into the BIC penalty, destabilizing the selection process. By constraining the kinetic model to an integer n, as in Eqs. (3) and (4), the associated regularized lower incomplete gamma remains mathematically well-behaved, ensuring robust convergence to the global minimum. This numerical stability is critical for accurately generating a stable BIC and confidently establishing the presence of multiple sigmoidal components during deconvolution. Leading-strand synthesis, in the absence of MOF factors was fitted to a single sigmoidal dependence as:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I\left(t\right)=A*\gamma \left(k*t,n\right)$$\end{document}It=A*γk*t,nwhere I(t) is the intensity of the full-length product at time point t, A is an amplitude parameter corresponding to the intensity of the upper plateau level of the sigmoid, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma \left(x,n\right)$$\end{document}γx,n is the regularized lower incomplete gamma function defined as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma \left(x,n\right)=\frac{1}{\Gamma (n)}\underset{0}{\overset{x}{\int }}{y}^{n-1}{e}^{-y}{dy}$$\end{document}γx,n=1Γ(n)∫x0yn−1e−ydy where Γ (n)Γ(n) is the gamma function defined as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Gamma \left(n\right)={\int }_{0}^{\infty }{y}^{n-1}{e}^{-y}{dy}$$\end{document}Γn=∫0∞yn−1e−ydy, k is a rate parameter related to the rate of nucleotide incorporation, and n is an integer shape parameter related to the number of nucleotide incorporation bursts. The mean synthesis processing time is then defined as n/k.n/k. Total (leading- strand and completely ligated lagging-strand products) synthesis in the presence of MOF factors could not be fitted to a single sigmoidal dependence and required a two-sigmoidal component model as:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I\left(t\right)={A}_{1}*\gamma \left({k}_{1}*t,{n}_{1}\right)+{A}_{2}*\gamma \left({k}_{2}*t,{n}_{2}\right)$$\end{document}It=A1*γk1*t,n1+A2*γk2*t,n2where all the component-indexed parameters have the same meaning as above. For each fit the sum of squared residuals (SSR) was recorded. Assuming Gaussian noise around the true trend, a Bayesian information criterion (BIC) value can be calculated for each model as:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${BIC}\left(n\right)=N*{\mathrm{ln}}\left(\frac{{SSR}}{N}\right)+p*{\mathrm{ln}}(N)$$\end{document}BICn=N*lnSSRN+p*ln(N)where N is the number of experimental datapoints to which the model is fitted and p is the number of parameters in the model. Variance of the noise must also be considered as a parameter in the model. Thus, for fitting by the model described by Eqs. (3) and (4), p would be equal to 4 and 7, respectively. Ligation efficiency was calculated as the percentage of full-length, completely ligated product relative to the total lagging-strand signal (defined as the sum of ligated and unligated products). Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Source Data